We recently reported that restoring the CYP27A1–27hydroxycholesterol axis had antitumor properties. Thus, we sought to determine the mechanism by which 27HC exerts its anti–prostate cancer effects. As cholesterol is a major component of membrane microdomains known as lipid rafts, which localize receptors and facilitate cellular signaling, we hypothesized 27HC would impair lipid rafts, using the IL6–JAK–STAT3 axis as a model given its prominent role in prostate cancer. As revealed by single molecule imaging of DU145 prostate cancer cells, 27HC treatment significantly reduced detected cholesterol density on the plasma membranes. Further, 27HC treatment of constitutively active STAT3 DU145 prostate cancer cells reduced STAT3 activation and slowed tumor growth in vitro and in vivo. 27HC also blocked IL6-mediated STAT3 phosphorylation in nonconstitutively active STAT3 cells. Mechanistically, 27HC reduced STAT3 homodimerization, nuclear translocation, and decreased STAT3 DNA occupancy at target gene promoters. Combined treatment with 27HC and STAT3 targeting molecules had additive and synergistic effects on proliferation and migration, respectively. Hallmark IL6–JAK–STAT gene signatures positively correlated with CYP27A1 gene expression in a large set of human metastatic castrate-resistant prostate cancers and in an aggressive prostate cancer subtype. This suggests STAT3 activation may be a resistance mechanism for aggressive prostate cancers that retain CYP27A1 expression. In summary, our study establishes a key mechanism by which 27HC inhibits prostate cancer by disrupting lipid rafts and blocking STAT3 activation.

Implications:

Collectively, these data show that modulation of intracellular cholesterol by 27HC can inhibit IL6–JAK–STAT signaling and may synergize with STAT3-targeted compounds.

This article is featured in Highlights of This Issue, p. 669

Prostate cancer is the most common noncutaneous cancer in American men (1). Recent reports have emphasized the role of diet and lifestyle as important modifiable predictors of prostate cancer risk (2–4). Of all dietary factors, cholesterol has received much attention for its role in prostate cancer (5–8). Importantly, epidemiologic and preclinical studies have found a link between higher serum cholesterol and greater prostate cancer incidence and/or progression (7, 9). Furthermore, cholesterol-lowering drugs (primarily HMG-CoA reductase inhibitors, a.k.a. statins) are inversely linked with prostate cancer incidence and/or progression, including reduced risk of advanced disease (10). Collectively, these data suggest the potential of altering cholesterol homeostasis to treat and/or prevent prostate cancer.

Prostate cells maintain intracellular cholesterol homeostasis via intricate feedback mechanisms (11–13). During prostate cancer tumorigenesis, the mechanisms that govern cholesterol homeostasis are often dysregulated, leading to increased intracellular cholesterol accumulation and promoting cell survival and growth (14–16). Our group recently published a report linking p450 sterol hydroxylase CYP27A1 loss and prostate cancer (8). In normal nonneoplastic cells, when cholesterol accumulates intracellularly, CYP27A1 hydroxylates cholesterol to create 27-hydxroycholesterol (27HC), which binds to and activates Liver-X-Receptor transcription factors (LXR; ref. 17). Activated LXRs modulate cholesterol metabolism by increasing reverse cholesterol transport, thereby reducing intracellular cholesterol levels (18). We found that restoration of CYP27A1 reduced intracellular cholesterol and decreased prostate cancer tumor growth in vitro and in vivo (8). Similarly, 27HC treatment reduced intracellular cholesterol and reduced prostate cancer cell, growth in vitro. Overall, we identified the CYP27A1/27HC axis as a cholesterol biosensor that is lost in prostate cancer cells, and restoration of this axis inhibits prostate cancer growth (8). However, the exact mechanism by which 27HC inhibits prostate cancer growth remains unclear.

Cholesterol is a crucial building block in the cell, with a majority localized to the lipid bilayer within specific membrane-associated microdomains called lipid rafts (14, 19). Many studies showed that intact membrane lipid rafts are important to facilitate signal transduction. Pathways initiated by cytokine stimulation, such as the IL6–JAK–STAT pathway, are activated via raft-dependent signaling and are disrupted after membrane-cholesterol depletion in prostate cancer cells (20). In sum, these reports suggest that intact lipid rafts are essential for sustaining pro-oncogenic signaling.

The IL6–JAK–STAT3 pathway is composed of extracellular IL6 ligand activating IL6 receptors, which phosphorylate JAK, which in turn phosphorylates STAT3. Phosphorylated STAT3 dimerizes and translocates into the nucleus to induce expression of genes with various protumorigenic properties (21). In addition, IL6–JAK–STAT3 is known to be an important signaling pathway in prostate cancer (21, 22). Multiple studies have shown that inhibiting this pathway can control prostate cancer growth in preclinical models (23–25) and analysis of human metastatic castrate-resistant prostate cancer (mCRPC) rapid autopsy biopsies shows elevated levels of activated p-STAT3 and IL6 receptor in bone metastases compared with lymph node and visceral metastases (26), validating the importance and therapeutic potential of this pathway in prostate cancer.

Given that 27HC exhibits anti–prostate cancer properties, we sought to determine the mechanism by which this occurs. We hypothesized that 27HC impairs lipid raft–mediated signaling via cholesterol depletion, which then alters pro-oncogenic signaling pathways in prostate cancer. We tested this hypothesis using IL6–JAK–STAT3 as a model signaling pathway given it requires functional lipid rafts and is important in prostate cancer. This is the first study to provide a mechanistic basis for the antitumor activity of 27HC in prostate cancer.

CYP27A1 expression correlation with disease stage and the IL6–JAK–STAT3 pathway

For gene correlation studies, we used a web-based software named the Prostate Cancer Transcriptome Atlas (PCTA; www.thepcta.org). This software currently provides two large sets of prostate cancer transcriptome profiles including the PCTA consisting of 2,115 prostate cancer samples described in You and colleagues (27) and The Cancer Genome Atlas (TCGA) prostate adenocarcinoma cohort consisting of 551 prostate cancer samples. Pathway activation score was computed by using Z-score method (28). Correlations were computed by the Spearman rank correlation method.

Cell lines and culture conditions

All cell lines were obtained and authenticated by ATCC. Cell lines were authenticated by STR profiling and tested for mycoplasma using a DNA-based PCR test for the detection of 19 species of mycoplasma at the start of the study by DDC Medical and most recently at Duke University cell culture facility by MycoAlert PLUS test Lonza. DU145 cells were grown in Dulbecco's Modified Eagle Medium (DMEM) + 10% fetal bovine serum (FBS). PC3, 22RV1, and LNCaP cells were cultured in RPMI + 10% FBS. 27HC was purchased from Enzo, and the LXR agonists GW3965 and TO1317 were purchased from Sigma-Aldrich. 27HC and LXR agonists were resuspended using DMSO in a 10 mmol/L stock. STAT3 inhibitors SH-4-54 and C188-9 were also purchased from Selleck Chemicals and resuspended at 50 mmol/L in DMSO.

Western blots

Cells were collected in RIPA lysis buffer with protease cocktail inhibitor 1 and phosphatase cocktail inhibitor 2 and 3 (Sigma). Protein was quantified using DC Bio-Rad protein assay (Bio-Rad) using BSA standard. For each blot, 30 to 60 μg of protein was loaded and run on SDS-PAGE gels and transferred onto 0.45 μmol/L nitrocellulose. p-STAT3Y705, total STAT3, p-JAK2Y1007/1008, and total JAK2 (Cell Signaling Technology) were all used at 1:1,000. For nuclear/cytoplasmic extracts from DU145 cells, the NE-PER kit (Thermo Fisher) was used according to the manufacturer's instructions. For Western blot quantification, ImageJ 1.52a software was utilized to quantify band densitometry for p-STAT3, total STAT3, and Actin or GAPDH loading controls for normalization. For normalized intensity calculations, first p-STAT3 and STAT3 values were divided by their respective loading control values. The resulting normalized values were then divided (p-STAT3/STAT) to generate a normalized intensity value. When shown as percentage, treatment group ratios were divided by control group ratio (100%) and multiplied by 100.

Coimmunoprecipitation

STAT3 monomer and dimers were assessed by coimmunoprecipitation Western blot as previously described (29). Briefly, LNCaP cells were treated with vehicle control or IL6 recombinant protein for 1 hours and lysed with 0.5% NP-40 lysis buffer on an end-over-end rotor at 4°C. Protein lysates were incubated with mouse anti-STAT3 (Cell Signaling Technology) on an end-over-end rotor at 4°C for 2 hours and followed by incubation with protein A beads for 1 hours. Pulldowns were washed with NP-40 lysis buffer and heated for 5 minutes in Laemmli loading buffer for elution. p-STAT3Y705 monomers and dimers were then probed for with a rabbit polyclonal p-STAT3Y705 antibody.

Fluorescent cholesterol probe design, purification, and direct stochastic optical reconstruction microscopy (dSTORM) imaging

Perfringolysin O Domain 4 (PFO-D4 C459A; refs. 30, 31) was used as a cholesterol probe for direct stochastic optical reconstruction microscopy (dSTORM) imaging. The single engineered N-terminal cysteine was used for attachment of an Alexa Fluor 647-C2-Maleimide label. The expression plasmid was custom made by GenScript and contained a cleavable N-terminal His6-Smt3 tag (32) for affinity purification. The expression plasmid was transfected into BL12 cells (NEB) and induced with IPTG. After incubation, the cells were pelleted and resuspended in buffer A (50 mmol/L Phosphate pH 7.9, 300 mmol/L NaCl) with protease inhibitor (Sigma-Aldrich S8830). Cells were then lysed by French press and cell debris pelleted by ultra-centrifugation at 45,000 RPM for 45 minutes at 4°C. Protein was purified with HisPur Cobalt Resin (Thermo Pierce) and dialyzed into PBS. The His6-SMT3 tag was cleaved with SUMO protease Ulp1, and both the tag and the protease were removed by passing through additional cobalt-affinity resin. Alexa Fluor 647 (AF647) maleimide was used for labeling of protein according to the manufacturer's protocol. Excess label was removed by a Bio-Gel P-4 gel filtration column (Bio-Rad). Sample purity and monomeric state were verified by PAGE gels and size exclusion chromatography. We obtained degree of labeling equal to one, indicating single Cys residue was labeled with AF647. We refer to the resulting construct as PFO-D4-AF647.

dSTORM imaging and analysis

Samples were imaged on a Nikon 3D N-STORM microscope using a 100× objective in total internal reflection fluorescence (TIRF) mode with 647 nm excitation. Frames (20,000) were acquired at 100 Hz using NIS-Elements 4.3 software (Nikon). These raw images were processed by NIS-Elements to extract the x–y locations for each fluorophore appearance and the results were exported for analysis with custom MATLAB code. Pair-correlation analysis (33) was used to quantify clustering. This method calculates the autocorrelation curves of regions of interest (ROI) to provide cluster size, increased local density (cluster density divided by average density), and the number of detected molecules per cluster. Pair-correlation analysis considers multiple localizations from the same fluorescent reporter. We used DU145 cells sparsely labeled with PFO-D4-AF647 (random monomers) to calculate the average number of localizations for each reporter. This parameter was found to be 2 and was incorporated into pair-correlation analysis and density calculations.

Lipid raft FACS analysis

For determination of lipid raft integrity by flow cytometry analyses, cells were treated as indicated for 48 hours. Subsequently, media was removed, and cells were washed with RPMI base media (without serum), and 0.5 μmol/L of di-4-ANEPPDHQ (in 2 mL of RPMI base media) was added to each well for 30 minutes. Cells were then washed with PBS, trypsinized, and resuspended in 500 μL of PBS + 1% BSA. Samples were analyzed on a BD LSR II Flow Cytometer. Generalized polarization (GP) values were calculated using the following equation GP = (LO + LD)/(LO − LD), using FlowJo. GP values are a measure of membrane order, with values ranging from −1 (denoting that all the emission is collected in the disordered, long-wavelength channel) to +1 (denoting that all the fluorescence is collected in the ordered channel; ref. 34).

Proliferation assays

DU145 or PC3 cells were plated in a 96-well black walled, clear bottom plate, and after 36 hours, cells were treated as specified for 72 hours using DMSO as a vehicle. CyQUANT proliferation assay (Thermo Fisher). Drug combination studies were conducted on an IncuCyte S3 live-cell imaging system (Essen Bioscience) on 96 whole-well scan mode. Plates were imaged every 24 hours and quantified with bundled image analysis software and normalized cell confluence to treatment day 0.

Kinetic cell migration and Matrigel invasion

Scratch wound migration and matrigel invasion assays were conducted on an IncuCyte S3 live-cell imaging system (Essen Bioscience). DU145 cells (4.5 × 105) were plated on imagelock 96-well plates and treated for 24 hours with indicated drug. Cells were then washed twice with PBS, and a 96-pin wound making tool was used to make a uniform scratch in all 96 wells simultaneously (IncuCyte 96-well WoundMaker Tool, cat. # 4563). For cell migration assays, complete DMEM was added with indicated treatments and plate imaged. For matrigel invasion assays, scratch wound was overlaid with 3 μg/mL of matrigel and incubated for 1 hour at 37°C, and then overlaid with 100 μL of DMEM with indicated treatments. Images were acquired every 2 hours. Quantification was conducted using the IncuCyte Scratch Wound software module.

RNA and quantitative real-time PCR

DU145 cells were treated in 6 wells with DMSO or 27HC for 48 hours. Cells were rinsed with 1× PBS, and RNA was collected from cells using the Qiagen RNeasy kit with DNAse digestion. Following isolation, RNA was quantified using the Nanodrop (Thermo Scientific) and reverse transcribed using Bio-Rad iScript cDNA synthesis kit using 1 μg of RNA. For RT-qPCR, 10 ng of cDNA was used per reaction using Superscript SYBR green (Bio-Rad) and assays were performed on ABI Viia7 (Thermo Scientific). Primer sequences are available in Supplementary Materials.

ChIP-qPCR

ChIP-qPCR experiments were performed with the ChIP-IT High Sensitivity kit (Active Motif) according to the manufacturer's instructions. Briefly, 80% confluent DU145 cells treated for 48 hours with vehicle (50% ethanol and 50% DMSO) or 27HC (10 μmol/L) were cross-linked with 1% formaldehyde at room temperature for 10 minutes. Chromatin was sonicated to 100–500 bp in a Bioruptor UCD-200 (Diagenode) and 30 μg of chromatin per reaction was incubated overnight with the pertinent amount of antibody recommended by manufacturer p-STAT3Y705 (D3A7) XP antibody (cat. #9145), STAT3 (79D7) Rabbit mAb (cat. #4904) Cell Signaling Technology or IgG control. Following RNase A and Proteinase K treatment, ChIP DNA was purified with the column system provided with the kit and quantitative PCR was conducted. A negative primer set that amplifies a 78 base-pair fragment from a gene desert on human chromosome 12 (Human Negative Control Primer Set 1, Active Motif) was used as a control. Primer sequences are available in Supplementary Materials.

In vivo xenograft study

Mouse experiments were conducted as approved by the Cedars-Sinai Medical Center IACUC board protocol number IACUC006565. Male severe combined immune-deficient (SCID) mice were purchased from Taconic Biosciences and acclimated for 48 hours according to IACUC policy. DU145 cells (1.0 × 106) cells were injected in a 1:1 solution of base DMEM and matrigel in the lower right flank of mice and tumors were allowed to grow to approximately 200 mm3 prior to randomization to cyclodextrin vehicle control or 27HC (40 mg/kg) treatment group, and injected subcutaneously daily as previously described (35).

Statistical analysis

GraphPad Prism 7.04 was used for statistical analysis of in vitro studies. For time-course live-cell spheroid assays, multiple t tests were conducted, and multiple comparisons correction adjusted P value was calculated by the Holm–Sidak method with alpha = 0.05. For time-course live-cell proliferation assays and migration/invasion assays, two-way repeated-measures ANOVA multiple comparison analysis was conducted with the Bonferroni multiple comparison test for adjusted P value both with alpha = 0.05. To evaluate the combined effects of 27HC and STAT3 inhibitors, a modified coefficient of drug interaction (CDI) equation was used (36). Briefly, the CDI is calculated as follows: CDI = AB/(A × B). AB is the effect of the combination groups compared with control group; A or B is the percent inhibition of the single-agent group compared with control group. CDI <1 indicates that the drugs are synergistic, CDI = 1 indicates drugs are additive, and >1 antagonistic. A CDI <0.7 indicates that the drug combination is significantly synergistic (36). To compare tumor growth kinetics in vivo, a generalized estimating equation with exchangeable correlation was used to test whether tumor volume growth varied over time between the two treatment arms. Time was treated as a categorical variable to not assume linear growth. P < 0.05 was considered statistically significant.

CYP27A1 expression in prostate cancer disease states and subtypes

We previously observed that CYP27A1 mRNA was downregulated in prostate cancer versus normal benign prostate in multiple independent gene-expression data sets (8). To reassess our previous findings, we queried a virtual pooled cohort of 2,115 patient samples consisting of 1,321 prostate cancers and 794 benign prostate tissues and found that CYP27A1 mRNA expression is progressively downregulated from benign to metastatic castrate-resistant prostate cancer (Fig. 1A). One-way ANOVA test revealed significant differences across different grades and disease states (F = 84.774, P < 0.001). Binary comparisons of primary versus benign (log2 fold change = −0.612, P < 0.001) and mCRPC versus primary (log2 fold change = −0.171, P < 0.001) represent statistical significance by two-tailed rank-sum test. In addition, applying a previously described novel prostate cancer subtyping algorithm (27), we found that CYP27A1 expression is lowest in prostate cancer subtype 1 (PCS1), the most lethal prostate cancer subtype (Fig. 1B; F = 88.854, P < 0.001). Similarly, one-way ANOVA test of TCGA samples showing significantly lower expression levels of CYP27A1 in GS > 7 versus other Gleason subsets (GS = 7 and GS < 7; Fig. 1C; F = 36.768, P < 0.001) and PCS1 versus other subtypes (PSC2 and PCS3; F = 97.764, P < 0.001; Fig. 1D).

Figure 1.

CYP27A1 gene expression is progressively downregulated during progression and in lethal PCS1 subtype. Lollipop plot, box plot, and line plot display expression profiles of individual samples, distribution of gene expression, and average expression and trend lines of the gene expression in each disease category, respectively. Plots show CYP27A1 gene expression in 5 disease categories from (A) PCTA samples (F = 84.774, P < 0.001) and (C) TCGA samples (F = 36.768, P < 0.001). Plots show CYP27A1 expression in 3 PCS categories for (B) PCTA samples (F = 88.854, P < 0.001) and (D) TCGA samples (F = 97.764, P < 0.001). GS, Gleason score; mCRPC, metastatic castration resistant prostate cancer; PCS, prostate cancer subtype.

Figure 1.

CYP27A1 gene expression is progressively downregulated during progression and in lethal PCS1 subtype. Lollipop plot, box plot, and line plot display expression profiles of individual samples, distribution of gene expression, and average expression and trend lines of the gene expression in each disease category, respectively. Plots show CYP27A1 gene expression in 5 disease categories from (A) PCTA samples (F = 84.774, P < 0.001) and (C) TCGA samples (F = 36.768, P < 0.001). Plots show CYP27A1 expression in 3 PCS categories for (B) PCTA samples (F = 88.854, P < 0.001) and (D) TCGA samples (F = 97.764, P < 0.001). GS, Gleason score; mCRPC, metastatic castration resistant prostate cancer; PCS, prostate cancer subtype.

Close modal

Visualization of 27HC-induced cholesterol depletion

To visualize the effects of 27HC on membrane cholesterol, we optimized labeling of cholesterol with PFO-D4-AF647, which binds cholesterol at residues T490 and L491 (Supplementary Fig. S1A). We found that the fluorescent signal saturated at approximately 300 nmol/L PFO-D4-AF647 and we therefore used this concentration for all experiments (Supplementary Fig. S1B). DU145 cell image and detected density of cholesterol in the steady state are shown in Fig. 2A (left) and Fig. 2B, respectively. To test the effect of cholesterol depletion, we treated DU145 cells with 10 mmol/L Methyl-β-cyclodextrin (MβCD) as previously described (33). MβCD treated cells that were labeled with PFO-D4-AF647 had negligible fluorescent signal (Fig. 2B), consistent with the known effects of MβCD to significantly strip cells of cholesterol. To determine the effect of 27HC treatment on membrane cholesterol, we imaged DU145 cells treated for 2, 6, 12, and 48 hours. The density of detected cholesterol molecules decreased significantly by 2 hours (P < 0.001) and was negligible after 24 and 48 hours (Fig. 2A and C). The negligible fluorescent signal after 24 to 48 hours treatment suggests that cholesterol content was below the mole percent threshold for this probe (31). As a control, 24-hour treatment with DMSO had no significant effect on detected cholesterol density (Fig. 2B).

Figure 2.

Cholesterol density and distribution is altered in DU145 cell membranes upon treatment with 27HC. A, dSTORM images of DU145 cells after varying 27HC incubation times. Scale bars, 5 μm. B, Detected cholesterol densities in steady state, after 24 hours incubation with 0.1% DMSO, and after 40 minutes of cholesterol depletion with 10 mmol/L MβCD. C, Densities of detected molecules vs. 27HC incubation times. The blue line shows best exponential fit to the data (time constant 4.8 hours). D–F, Clustering metrics averaged over all ROIs as calculated by pair-correlation analysis for varying 27HC incubation times: steady state, 27HC 2-hour treatment, and 27HC 6-hour treatment. D, Increased local density. E, Detracted molecules per cluster. F, Cluster radius. Error bars, SEM. For each condition, minimum of 12 cells and 47 ROI's were analyzed. *, P < 0.05; **, P < 0.01; ***, P < 0.001. G, GP values of flow cytometry analyzed di-4-ANEPPDHQ for lipid raft organization in DU145 and LNCaP cells in cells treated as denoted (10 μmol/L) for 48 hours; H, 27HC-treated cells ± cholesterol (CHO; 10 μg/mL) supplementation (n = 3; *, P = 0.001; #, P < 0.01; $, P < 0.05).

Figure 2.

Cholesterol density and distribution is altered in DU145 cell membranes upon treatment with 27HC. A, dSTORM images of DU145 cells after varying 27HC incubation times. Scale bars, 5 μm. B, Detected cholesterol densities in steady state, after 24 hours incubation with 0.1% DMSO, and after 40 minutes of cholesterol depletion with 10 mmol/L MβCD. C, Densities of detected molecules vs. 27HC incubation times. The blue line shows best exponential fit to the data (time constant 4.8 hours). D–F, Clustering metrics averaged over all ROIs as calculated by pair-correlation analysis for varying 27HC incubation times: steady state, 27HC 2-hour treatment, and 27HC 6-hour treatment. D, Increased local density. E, Detracted molecules per cluster. F, Cluster radius. Error bars, SEM. For each condition, minimum of 12 cells and 47 ROI's were analyzed. *, P < 0.05; **, P < 0.01; ***, P < 0.001. G, GP values of flow cytometry analyzed di-4-ANEPPDHQ for lipid raft organization in DU145 and LNCaP cells in cells treated as denoted (10 μmol/L) for 48 hours; H, 27HC-treated cells ± cholesterol (CHO; 10 μg/mL) supplementation (n = 3; *, P = 0.001; #, P < 0.01; $, P < 0.05).

Close modal

Pair-correlation analysis (33) was applied to characterize cholesterol-enriched domains (clusters) in steady state, after 2 hours of 27HC treatment, and after 6 hours of 27HC treatment. The resulting autocorrelation curves and clustering metric histograms are shown in Supplementary Fig. S2. Cluster properties showed very small change between steady state and 2 hours. After 6 hours, smaller cluster radius, fewer molecules per cluster, and a larger increased local density were detected, on average (Fig. 2DF). This indicates that after 6-hour treatment with 27HC, cholesterol-enriched domains are largely reorganized. Overall, our dSTORM data suggest that 27HC treatment (i) reduces cholesterol density in a time-dependent manner and (ii) leads to reorganization of cholesterol-enriched domains after 6-hour treatment. Though 27HC leads to even further declines in membrane cholesterol at 24 hours (Fig 2C), suggesting 27HC has lasting effects on membrane cholesterol, its effects on cluster properties at these later time points were tested.

27HC causes lipid raft disorganization and depletes membrane cholesterol

Given 27HC reduced plasma membrane cholesterol, which is crucial for lipid raft formation, we assessed lipid raft microdomains using the fluorescent based probe di-4-ANEPPDHQ as previously described (37). Flow cytometry analysis of di-4-ANEPPDHQ showed that 27HC, but not the LXR agonists GW3965 and TO1317, reduced GP values, marking disorganization and loss of rigidity of lipid raft microdomains in DU145 and LNCaP cells (Fig. 2G). Moreover, supplementation of exogenous cholesterol rescued lipid raft microdomain GP values (Fig. 2H), confirming the direct importance of cholesterol depletion for these effects.

27HC reduces p-STAT3Y705 levels in DU145 cells

To examine the importance of 27HC's antiproliferative effects via IL6–JAK–STAT3 signaling, we utilized two cell lines with altered JAK–STAT3 signaling: DU145 with constitutively active STAT3 (p-STAT3Y705) and PC3 that are STAT3 null (38, 39). As shown in Fig. 3A and B, DU145 and PC3 cell proliferation was not affected by LXR agonists GW3965 and TO1317 after 72 hours. However, 27HC at 5 μmol/L and 10 μmol/L elicited a growth inhibitory response at 72 hours in DU145 cells, but not the STAT3-null PC3 cells. Interestingly, we found that 27HC and GW3965, but not TO1317, altered p-STAT3Y705 levels at 48 hours (Fig 3C). However, only 27HC treatment induced cell death as denoted by cleaved PARP1. Relative to DMSO control at each time point, Western blot analysis of p-STAT3Y705 showed downregulation of STAT3 activity to ∼66% at 48 hours and further to ∼33% at 72 hours post 27HC (10 μmol/L) treatment (Fig. 3D and E). These data further demonstrate the importance of disrupting lipid membranes as evidenced by inhibition of the IL6–JAK–STAT3 pathway in mediating 27HC's antiproliferative effects.

Figure 3.

27HC effects on prostate cancer cell growth and STAT3 signaling in vitro and in vivo. A, PC3 and (B) DU145 cell growth response to LXR agonists 27HC, GW3965, and TO1317 at indicated doses after 72 hours of treatment by MTT assay. C, STAT3 activation and cell death assessed by p-STAT3 and PARP1 cleavage in DU145 cell lysates treated with various doses of LXR agonists (0–20 μmol/L) for 48 hours as indicated. D, Activation of STAT3 assessed in DU145 cells at 48 and 72 hours after treatment with 27HC (10 μmol/L) treated in triplicate. E, Percent active STAT3 (p-STAT3/STAT3) normalized to GAPDH by densitometry. **, P = 0.0049; ****, P < 0.0001. F, Upstream JAK2 activation was assessed in LNCaP cells pretreated with 27HC for 48 hours and then stimulated with rIL6 (10 ng/mL) and cell lysates collected at indicated times (minutes). G, DU145 cells were supplemented with cholesterol (CHO) alone or in 27HC-treated cells for 72 hours and cell lysates probed as indicated showing that CHO rescues 27HC-mediated STAT3 inactivation. H, DU145 xenograft tumor growth kinetics in vivo–injected subQ with vehicle (cyclodextrin) or 27HC (40 mg/kg) daily. I, Endpoint tumor lysates (n = 5/group) immunoblotted as indicated (individually and pooled) and (J) blot pixel densitometry of bands normalized to actin bands and plotted.

Figure 3.

27HC effects on prostate cancer cell growth and STAT3 signaling in vitro and in vivo. A, PC3 and (B) DU145 cell growth response to LXR agonists 27HC, GW3965, and TO1317 at indicated doses after 72 hours of treatment by MTT assay. C, STAT3 activation and cell death assessed by p-STAT3 and PARP1 cleavage in DU145 cell lysates treated with various doses of LXR agonists (0–20 μmol/L) for 48 hours as indicated. D, Activation of STAT3 assessed in DU145 cells at 48 and 72 hours after treatment with 27HC (10 μmol/L) treated in triplicate. E, Percent active STAT3 (p-STAT3/STAT3) normalized to GAPDH by densitometry. **, P = 0.0049; ****, P < 0.0001. F, Upstream JAK2 activation was assessed in LNCaP cells pretreated with 27HC for 48 hours and then stimulated with rIL6 (10 ng/mL) and cell lysates collected at indicated times (minutes). G, DU145 cells were supplemented with cholesterol (CHO) alone or in 27HC-treated cells for 72 hours and cell lysates probed as indicated showing that CHO rescues 27HC-mediated STAT3 inactivation. H, DU145 xenograft tumor growth kinetics in vivo–injected subQ with vehicle (cyclodextrin) or 27HC (40 mg/kg) daily. I, Endpoint tumor lysates (n = 5/group) immunoblotted as indicated (individually and pooled) and (J) blot pixel densitometry of bands normalized to actin bands and plotted.

Close modal

27HC inhibits recombinant IL6-mediated JAK2 and STAT3 phosphorylation

To determine the effects of 27HC on STAT3 upstream JAK2 activation, LNCaP cells, which have low basal levels of p-STAT3Y705, were pretreated with vehicle control or 27HC (10 μmol/L) for 48 hours, then treated with recombinant IL6 (rIL6; 10 μg/mL) and cells were collected at denoted time points. Western blot analysis for p-JAK2Y1002/1007 (Fig. 3E) showed increased p-JAK2 within 30 to 60 minutes following IL6 treatment in vehicle pretreated LNCaP cells with essentially no p-JAK2 increase in 27HC (10 μmol/L) pretreated cells.

To test if downregulation of p-STAT3Y705 is mediated by 27HC-induced cholesterol loss, cell culture media were supplemented with exogenous cholesterol. As previously demonstrated, 27HC treatment downregulated p-STAT3 levels, whereas cotreatment of 27HC and cholesterol rescued 27HC induced p-STAT3Y705 downregulation (Fig. 3F), suggesting the effects of 27HC on p-STAT are mediated via cholesterol depletion. Overall, these data demonstrate that 27HC decreases JAK2 phosphorylation, which in turn reduces p-STAT3Y705 activation, and these effects may be rescued by exogenous cholesterol supplementation.

27HC reduces DU145 xenograft tumor growth

To test if 27HC could slow STAT3-driven tumors in vivo, we grafted 1 × 106 DU145 cells to male SCID mice. When tumors were ∼200 mm3, mice were randomized to control or 27HC treatment groups. Mice were injected daily (s.c.) with vehicle (cyclodextrin) or 27HC in cyclodextrin (40 mg/kg) for 37 days as previously described (35). There was a statistically significant difference in tumor growth between the two treatment arms (time × treatment P interaction <0.001), with slower tumor growth in mice treated with 27HC (Fig. 3G). Western blot analysis of tumor lysates showed reduced p-STAT3Y705 and total STAT3 in tumors from 27HC treatment groups compared with vehicle control (Fig. 3H and I), but no significant difference in the ratio of p-STAT3/STAT3 (Fig. 3J). Our in vivo experiment demonstrated that 27HC treatment of the aggressive DU145 xenografts reduces STAT3 levels, STAT3 activation, and slowed tumor growth in vivo. STAT3 has an autoregulatory role; thus, long-term treatment in vivo with 27HC (i.e., 37 days) versus 72 hours in vitro may account for reduced total STAT3 levels in vivo, but not in vitro (40, 41).

27HC inhibits STAT3 dimer formation, nuclear localization, and target gene expression

STAT3Y705 phosphorylation is necessary for STAT3 dimer formation, nuclear translocation, and transcriptional activity. Given that 27HC inhibited STAT3 phosphorylation, we conducted cell nuclear-cytoplasmic fractionation experiments, which showed 27HC reduced total and p-STAT3Y705 levels in nuclear fractions of DU145 cells (Fig. 4A). This was confirmed by immunofluorescence showing increased levels of total STAT3 in the cytoplasm of DU145 cells treated with 27HC versus DMSO for 72 hours versus nuclear localization in DMSO-treated cells (Fig. 4B). p-STAT3Y705 dimer formation, both basal and IL6 induced, was also inhibited by 27HC treatment in LNCaP cells (Fig. 4C). Well-known STAT3 target genes, IL6 and Survivin, were significantly downregulated in DU145 cells upon 27HC treatment, but not with treatment using the LXR agonists GW3965 or TO1317 (Fig. 4D). Consistent with previous results, the LXR target gene ABCA1 was significantly upregulated by the LXR agonists and 27HC (Fig. 4D), demonstrating that the LXR agonists were bioactive, although they did not elicit a reduction in STAT3 gene targets. Moreover, low-density lipoprotein receptor (LDLR) was also downregulated only in response to 27HC (Fig. 4D).

Figure 4.

27HC inhibits STAT3 nuclear localization, dimerization, and transcriptional activity. A, DU145 whole-cell (WCL), nuclear (NUC), and cytoplasmic (CYTO) cell lysate fractions probed as indicated with a-tubulin as cytoplasmic and lamin A/C as nuclear loading controls. B, Immunofluorescence imaging of total STAT3 (red) in DU145 cells in DMSO or 27HC (10 μmol/L) treated cells at 72 hours. C, Immunoprecipitation of total STAT3 and immunoblot for p-STAT3 in control LNCaP cells, 27HC pretreated (72 hours, 10 μmol/L), rIL6 (30 minutes, 10 μg/mL), or both. Arrows denote STAT3 monomers and dimers. Input 10% of immunoprecipitation input protein lysates. D, qRT-PCR analysis of denoted genes in DU145 cells treated as indicated (10 μmol/L) for 48 hours (data show mean + SD from triplicates.). *, P < 0.05; $ and # = P < 0.001. E, Immunoblots of lysates from LNCaP cells treated for 0.5 or 24 hours with conditioned media from DU145 vehicle control– or 27HC-treated cells (DU145-CM and DU145-CM + 27HC, respectively) in a 1:1 ratio with fresh 10% RPMI-FBS. F, Individual day 13 22RV1 spheroid brightfield images at endpoint treated as above. G, Spheres were imaged periodically using IncuCyte S3 and spheroid size quantified with spheroid module software. (Multiplicity-adjusted t test P values day 9: P = 0.01; day 13: P = 0.0001; n = 5/group.) H, ChIP-qPCR for p-STAT3 and total STAT3 in DU145 cells treated for 48 hours as denoted (data show mean + SD from triplicates). p-STAT3 ChIP-qPCR normalized by a factor of 1.5 to account for reduction of p-STAT3 upon 27HC treatment at 48 hours. RT-qPCR and ChIP results are representative of two independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

27HC inhibits STAT3 nuclear localization, dimerization, and transcriptional activity. A, DU145 whole-cell (WCL), nuclear (NUC), and cytoplasmic (CYTO) cell lysate fractions probed as indicated with a-tubulin as cytoplasmic and lamin A/C as nuclear loading controls. B, Immunofluorescence imaging of total STAT3 (red) in DU145 cells in DMSO or 27HC (10 μmol/L) treated cells at 72 hours. C, Immunoprecipitation of total STAT3 and immunoblot for p-STAT3 in control LNCaP cells, 27HC pretreated (72 hours, 10 μmol/L), rIL6 (30 minutes, 10 μg/mL), or both. Arrows denote STAT3 monomers and dimers. Input 10% of immunoprecipitation input protein lysates. D, qRT-PCR analysis of denoted genes in DU145 cells treated as indicated (10 μmol/L) for 48 hours (data show mean + SD from triplicates.). *, P < 0.05; $ and # = P < 0.001. E, Immunoblots of lysates from LNCaP cells treated for 0.5 or 24 hours with conditioned media from DU145 vehicle control– or 27HC-treated cells (DU145-CM and DU145-CM + 27HC, respectively) in a 1:1 ratio with fresh 10% RPMI-FBS. F, Individual day 13 22RV1 spheroid brightfield images at endpoint treated as above. G, Spheres were imaged periodically using IncuCyte S3 and spheroid size quantified with spheroid module software. (Multiplicity-adjusted t test P values day 9: P = 0.01; day 13: P = 0.0001; n = 5/group.) H, ChIP-qPCR for p-STAT3 and total STAT3 in DU145 cells treated for 48 hours as denoted (data show mean + SD from triplicates). p-STAT3 ChIP-qPCR normalized by a factor of 1.5 to account for reduction of p-STAT3 upon 27HC treatment at 48 hours. RT-qPCR and ChIP results are representative of two independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

To confirm the effects of IL6 downregulation and loss of IL6 autocrine effects, we collected 24-hour conditioned media (CM) from DU145 control cells (DU145-CM), which produce high levels of IL6 (42), or cells pretreated with 27HC (DU145-27HC-CM). After 48 hours of 27HC pretreatment, media were removed, cells washed with PBS, and fresh media without 27HC were added for 24 hours of conditioning. LNCaP cells, which exhibit significantly lower levels of endogenous p-STAT3 (38), were treated with 1:1 ratio of fresh RPMI media and CM for 0.5 or 24 hours. Immunoblots of LNCaP protein lysates show DU145-CM induced p-STATY705; however, this was significantly blunted in LNCaP cells treated with DU145-27HC-CM (Fig. 4E). In addition, time lapse imaging 22RV1 spheroids over 13 days (Fig. 4F) cultured in 1:1 RPMI to DU145-27HC-CM grew significantly slower starting at day 6 compared with DU145-CM 22RV1 spheroids (Fig. 4G) and became smaller after day 9, suggesting cell death. Collectively, these data are consistent with 27HC inhibiting IL6 production by DU145 cells and consequently loss of IL6 auto- and paracrine effects.

27HC inhibits STAT3 DNA binding activity

Given 27HC reduces STAT3 nuclear localization and downregulates known STAT3 target genes, we assessed STAT3 DNA occupancy by a STAT3-ChIP-qPCR assay with pulldowns for p-STAT3Y705 and total STAT3 at the IL6 promoter. In addition, our analysis of publicly available STAT3 ENCODE ChIP-Seq data from breast cell lines shows a strong signal at the LDLR promoter, thus we also probed for STAT3 binding at the LDLR promoter. As shown in Fig. 4H, 27HC significantly reduced total STAT3 and p-STAT3Y705 binding at IL6 and LDLR promoters compared with control cells. To our knowledge, physical binding of STAT3 at the LDLR promoter has not been previously shown. These data show that reduced levels of total STAT3 and p-STAT3Y705 are bound to transcriptional target gene sequences; however, we cannot speculate as to the effects of 27HC on STAT3 transcriptional complexes or chromatin structure.

Synergistic antiproliferative effects of 27HC–STAT3 inhibitor combinations

Although 27HC lowers p-STAT3Y705 levels, a moderate level of active STAT3 remains (Fig. 3D). Thus, we tested if 27HC treatment may synergize with STAT3 inhibitors to more fully inhibit STAT3 signaling and thereby reduce proliferation. In this regard, we utilized two STAT3 inhibitors that target the phospho-Y-binding pocket of the STAT3 SH2 domain SH-4-54 (43) and C188-9 (a.k.a. TTI-101, ClinicalTrials.Gov Identifier NCT03195699) and further disrupt JAK-mediated STAT3 phosphorylation (44). Preliminary dose response studies were conducted with SH-4-54 and C188-9 individually in DU145 and PC3 to determine combination doses (Supplementary Fig. S3) when combined with 27HC 10 μmol/L. Single treatment of DU145 cells with SH-4-54 at 2.5 and 5 μmol/L alone had no significant effects versus control, whereas significant differences in proliferation were detected at 10 μmol/L at each time point (P < 0.0001; Fig. 5A). Interestingly, the combination of SH-4-54 at 5 μmol/L and 27HC had significant synergistic effects (Fig. 5B) 24 hours after treatment (CDI = 0.60) and onward (Table 1; ref. 36). Synergism (CDI <1 but >0.7) was observed between 27HC and SH-4-54 10 μmol/L dose in all time points. Multiple t test–adjusted P values and CDI values for SH-4-54 and 27HC are summarized in Table 1. For C188-9, at the 72-hour time point, all standalone C188-9 dose treatments (10, 20, and 40 μmol/L) had significant antiproliferative effects versus control (Fig. 5C and Table 1). Synergism was observed between all combinations of C188-9 and 27HC beginning at 24 hours, and 20 μmol/L C188-9 + 27HC combination became significantly synergistic at 48 and 72 hours after treatment (Fig. 5D). Complete multiple comparison t test P values between groups are listed in Supplementary Tables S1 and S2.

Figure 5.

Combined effects of 27HC (10 μmol/L) and STAT3 inhibitors SH-4-54 and C-1889 on proliferation, cell migration, and matrigel invasion. A, 27HC and SH-4-54 combination DU145 cell proliferation studies and (B) CDI value bar graph for drug interactions. C, 27HC and C-1889 combination DU145 cell proliferation studies and (D) CDI value bar graph for drug interactions. DU145 cells were treated as indicated and imaged every 24 hours on the IncuCyte S3 system. Cell proliferation plotted as a percent normalized to day 0 at time of treatment (n = 5/group), #, Significant difference compared with DMSO control at denoted time point and bracketed; ***, statistically significant difference between single agent at indicated dose and 27HC (10 μmol/L) combination. C and D, Above gray dashed line signifies antagonistic effects; *below gray dashed line indicates synergism and #below black dashed line indicates significantly synergistic interaction. Representative images of DU145 (E) scratch wound migration and (F) plotted wound-closure kinetics. G, Scratch wound matrigel invasion and (H) plotted matrigel invasion wound-closure kinetics. Plotted time course of % wound confluence, vertical dashed lines indicate time point in which DMSO control group reached 100% wound closure.

Figure 5.

Combined effects of 27HC (10 μmol/L) and STAT3 inhibitors SH-4-54 and C-1889 on proliferation, cell migration, and matrigel invasion. A, 27HC and SH-4-54 combination DU145 cell proliferation studies and (B) CDI value bar graph for drug interactions. C, 27HC and C-1889 combination DU145 cell proliferation studies and (D) CDI value bar graph for drug interactions. DU145 cells were treated as indicated and imaged every 24 hours on the IncuCyte S3 system. Cell proliferation plotted as a percent normalized to day 0 at time of treatment (n = 5/group), #, Significant difference compared with DMSO control at denoted time point and bracketed; ***, statistically significant difference between single agent at indicated dose and 27HC (10 μmol/L) combination. C and D, Above gray dashed line signifies antagonistic effects; *below gray dashed line indicates synergism and #below black dashed line indicates significantly synergistic interaction. Representative images of DU145 (E) scratch wound migration and (F) plotted wound-closure kinetics. G, Scratch wound matrigel invasion and (H) plotted matrigel invasion wound-closure kinetics. Plotted time course of % wound confluence, vertical dashed lines indicate time point in which DMSO control group reached 100% wound closure.

Close modal
Table 1.

Combination study summary table.

SH-4-54 and 27HC comb.Adjusted PCDI valueCombination effect
24-h time point 
 Control vs. 27HC 0.9955   
 Control vs. 2.5 μmol/L 0.9857   
 Control vs. 5 μmol/L >0.9999   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC 0.8285 1.03 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.60 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC 0.8934 0.92 Synergism 
48-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 2.5 μmol/L >0.9999   
 Control vs. 5 μmol/L 0.5826   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC <0.0001 1.02 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.53 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC 0.001 0.89 Synergism 
72-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 2.5 μmol/L 0.946   
 Control vs. 5 μmol/L 0.0006   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC <0.0001 1.04 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.54 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.72 Synergism 
C188-9 and 27HC comb. Adjusted P CDI value Combination effect 
24-h time point 
 Control vs. 27HC 0.5282   
 Control vs. 10 μmol/L 0.7551   
 Control vs. 20 μmol/L 0.2687   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC 0.167 0.85 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.86 Synergism 
 40 μmol/L vs. 40 μmol/L + 27HC 0.167 0.91 Synergism 
48-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 10 μmol/L 0.4243   
 Control vs. 20 μmol/L <0.0001   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.77 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.67 Significantly synergistic 
 40 μmol/L vs. 40 μmol/L + 27HC <0.0001 0.78 Synergism 
72-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 10 μmol/L 0.0339   
 Control vs. 20 μmol/L <0.0001   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.72 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.61 Significantly synergistic 
 40 μmol/L vs. 40 μmol/L + 27HC <0.0001 0.83 Synergism 
SH-4-54 and 27HC comb.Adjusted PCDI valueCombination effect
24-h time point 
 Control vs. 27HC 0.9955   
 Control vs. 2.5 μmol/L 0.9857   
 Control vs. 5 μmol/L >0.9999   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC 0.8285 1.03 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.60 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC 0.8934 0.92 Synergism 
48-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 2.5 μmol/L >0.9999   
 Control vs. 5 μmol/L 0.5826   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC <0.0001 1.02 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.53 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC 0.001 0.89 Synergism 
72-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 2.5 μmol/L 0.946   
 Control vs. 5 μmol/L 0.0006   
 Control vs. 10 μmol/L <0.0001   
 2.5 μmol/L vs. 2.5 μmol/L + 27HC <0.0001 1.04 Antagonistic 
 5 μmol/L vs. 5 μmol/L + 27HC <0.0001 0.54 Significantly synergistic 
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.72 Synergism 
C188-9 and 27HC comb. Adjusted P CDI value Combination effect 
24-h time point 
 Control vs. 27HC 0.5282   
 Control vs. 10 μmol/L 0.7551   
 Control vs. 20 μmol/L 0.2687   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC 0.167 0.85 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.86 Synergism 
 40 μmol/L vs. 40 μmol/L + 27HC 0.167 0.91 Synergism 
48-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 10 μmol/L 0.4243   
 Control vs. 20 μmol/L <0.0001   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.77 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.67 Significantly synergistic 
 40 μmol/L vs. 40 μmol/L + 27HC <0.0001 0.78 Synergism 
72-h time point 
 Control vs. 27HC <0.0001   
 Control vs. 10 μmol/L 0.0339   
 Control vs. 20 μmol/L <0.0001   
 Control vs. 40 μmol/L <0.0001   
 10 μmol/L vs. 10 μmol/L + 27HC <0.0001 0.72 Synergism 
 20 μmol/L vs. 20 μmol/L + 27HC <0.0001 0.61 Significantly synergistic 
 40 μmol/L vs. 40 μmol/L + 27HC <0.0001 0.83 Synergism 

Effects of 27HC–STAT3 inhibitor combination on cell migration and Matrigel invasion

Lipids rafts have been shown to be essential for lamellipodia formation and cancer cell migration (45). Likewise, STAT3 overexpression in DU145 and addback to PC3 induced the formation of lamellipodia and increasing tumor metastasis, supporting a role for STAT3-driven migratory phenotype in prostate cancer (23). Boyden chamber invasion assays showed 27HC significantly slowed matrigel invasion of DU145 cells (Supplementary Fig. S4). Thus, we tested 27HC alone or in combination with STAT3 inhibitors on DU145 cell migration (Fig. 5E). At the 32-hour time point (Fig. 5F, vertical dashed line), which is when the DMSO control group reached 100% wound closure, statistically significant differences in migration rates were observed for 27HC (P < 0.0001), SH-4-54 + 27HC combination (P < 0.0001), and C188-9 + 27HC (P < 0.0001) treated groups versus control (Fig. 5F). No significant effects on cell migration were detected in SH-454 or C188-9 treatments versus DMSO control at any time point. In addition, synergism was detected among 27HC + SH-4-54 combination (CDI = 0.93), whereas antagonistic effects were observed among 27HC + C188-9 (CDI = 1.19).

To test the effects on DU145 Matrigel invasion, scratch wounds were immediately overlaid with matrigel, and wound closure through matrigel was monitored in real time (Fig. 5G). At the 64-hour time point, which is when the DMSO control group reached 100% wound closure (Fig. 5H, vertical dashed line), significant effects on matrigel invasion were observed for 27HC (P < 0.0001), C188-9 (P < 0.01), but not SH-4-54. Significant effects were observed for both combination groups (P < 0.0001) and control. In addition, there was synergism between 27HC and SH-4-54 (CDI = 0.74) and significantly synergistic effects between 27HC and C188-9 (CDI = 0.67) on DU145 matrigel invasion. Overall, our functional assays demonstrate that 27HC, alone or in combination with STAT3 inhibitors, reduces migration and invasion of DU145 cells.

IL6–JAK–STAT3 correlates with retained CYP27A1 in advanced and aggressive prostate cancers

We assessed the correlation between Hallmark IL6–JAK–STAT3 pathway (GSEA M5897; ref. 46) and CYP27A1 expression in prostate cancer patients. We used the PCTA web software, which enables easy calculation of a pathway activation score based on the given pathway gene set using the Z-score method (28). We found that the Hallmark IL6–JAK–STAT3 pathway in PCTA and TCGA cohorts exhibits significant positive correlation with CYP27A1 expression across all prostate cancer disease states with increasingly higher correlation with disease progression. In this regard, in the PCTA data set, no correlation was observed in benign tissues between IL6–JAK–STAT3 and CYP27A1; however, Spearman rho values increased with disease progression in GS < 7 (rho = 0.205, P < 0.001), GS = 7 (rho = 0.20, P < 0.001), GS > 7 (rho = 0.42, P < 0.0001), and mCRPC (rho = 0.389, P < 0.001; Fig. 6A). Likewise, in the TCGA data set, IL6–JAK–STAT3 and CYP27A1 correlation rho values increased from GS < 7 (rho = 0.383, P = 0.005), GS = 7 (rho = 0.267, P < 0.0001), and GS > 7 (rho = 0.571, P < 0.001). Importantly, Hallmark IL6–JAK–STAT3 and CYP27A1-positive correlation pattern was observed in AR-high (rho = 0.27, P = 0.0001) and AR-low tumors (rho = 0.24, P = 0.0001; Supplementary Fig. S5). Among the three prostate cancer subtype categories (27), the highest correlation rho values were observed in the lethal PCS1 subtype category in both PCTA (rho = 0.341, P < 0.001) and TCGA (rho = 0.660, P < 0.001) data sets (Fig. 6B). These clinical data suggest that, although counterintuitive to the mechanistic data, CYP27A1 expression is positively associated with IL6–JAK–STAT3 signatures in prostate cancer tumors and have distinct correlations based on prostate cancer subtype and aggressiveness independent of AR status.

Figure 6.

Activation of the IL6–JAK–STAT pathway is highly correlated with CYP27A1 expression in high-grade, mCRPC and lethal prostate cancer subtype. A and B, Scatter plots and regression lines depicts correlations between IL6–JAK–STAT3 hallmark pathway activation and CYP27A1 gene expression in prostate cancer samples from the PCTA virtual cohort and the TCGA cohort. Individual dots on the plot represent individual prostate cancer samples from the cohorts. The correlation of all the samples and samples in different disease categories by Gleason score were displayed in both PCTA and TCGA cohorts (A). Correlation pattern in the PCS categories were seen in both PCTA and TCGA cohorts (B). C, Working model of 27HC action on IL6–JAK–STAT3 signaling. Under normal condition increased levels intracellular cholesterol due deregulation of cholesterol homeostasis facilitates lipid raft IL6–JAK–STAT3 signaling for cell autocrine effects and induction of IL6 autocrine effects, LDLR, and survival genes (i.e., Survivin). Addition of exogenous 27HC binds and activates LXRs inducing transcriptional program that reduces intracellular and lipid raft cholesterol thus impairing IL6–JAK–STAT3 signaling and reducing STAT3 target gene expression including IL6 and LDLR. Addition of STAT3 SH2 domain inhibitors may inhibit remaining STAT3 activation by blocking binding to JAK2 SH2 domains. STAT RE, STAT3 response element; LXRE, LXR response element.

Figure 6.

Activation of the IL6–JAK–STAT pathway is highly correlated with CYP27A1 expression in high-grade, mCRPC and lethal prostate cancer subtype. A and B, Scatter plots and regression lines depicts correlations between IL6–JAK–STAT3 hallmark pathway activation and CYP27A1 gene expression in prostate cancer samples from the PCTA virtual cohort and the TCGA cohort. Individual dots on the plot represent individual prostate cancer samples from the cohorts. The correlation of all the samples and samples in different disease categories by Gleason score were displayed in both PCTA and TCGA cohorts (A). Correlation pattern in the PCS categories were seen in both PCTA and TCGA cohorts (B). C, Working model of 27HC action on IL6–JAK–STAT3 signaling. Under normal condition increased levels intracellular cholesterol due deregulation of cholesterol homeostasis facilitates lipid raft IL6–JAK–STAT3 signaling for cell autocrine effects and induction of IL6 autocrine effects, LDLR, and survival genes (i.e., Survivin). Addition of exogenous 27HC binds and activates LXRs inducing transcriptional program that reduces intracellular and lipid raft cholesterol thus impairing IL6–JAK–STAT3 signaling and reducing STAT3 target gene expression including IL6 and LDLR. Addition of STAT3 SH2 domain inhibitors may inhibit remaining STAT3 activation by blocking binding to JAK2 SH2 domains. STAT RE, STAT3 response element; LXRE, LXR response element.

Close modal

Although we have previously shown that 27HC exerts anticancer effects in prostate cancer, the mechanistic basis for these observations has not been previously described. In this study, we show that 27HC treatment leads to a rapid and profound depletion of plasma membrane-cholesterol levels and disruption of lipid raft size and architecture. This, in turn, impairs oncogenic signaling through the well-described IL6–JAK–STAT3 signaling axis both in vitro and in vivo and delayed in vivo tumor growth in a constitutively active STAT3 model of prostate cancer growth. Intriguingly, 27HC appeared to sensitize cells to STAT3 inhibitors, a finding that requires further study as a novel treatment paradigm. Our data suggest that 27HC-mediated lipid raft perturbation is a key mechanism in part explaining the anti–prostate cancer activity of 27HC as evidenced by altered signaling of the IL6–JAK–STAT3 pathway. Importantly, we believe that 27HC, by blocking lipid raft signaling, would block multiple key pathways mediated by lipid rafts. Herein we used STAT3 as our model, but our data do not exclude the high likelihood that other lipid raft–mediated pathways are also disrupted.

Following 27HC treatment, we saw a rapid reduction in detected plasma membrane cholesterol (Fig. 2C). As the majority of intracellular cholesterol is localized in the lipid rafts, it is not surprising that the profound cholesterol depletion after 6 hours of 27HC treatment disrupted the steady-state organization of cholesterol enriched domains (Fig. 2DF). Given the disruption of the lipid rafts did not occur when exogenous cholesterol was added (Fig. 2H), it is likely the cholesterol depletion directly interfered with lipid raft function. Impairment of lipid rafts by cholesterol depletion inhibits membrane-bound AKT and IL6-mediated signaling activity in prostate cancer cells (14). Herein, we extended these findings to show that impaired lipid raft function, due to cholesterol depletion by 27HC, inhibited the IL6–JAK–STAT3 pathway. Moreover, 27HC also impaired STAT3 signaling with phenotypic outcomes on cell proliferation, migration/invasion, and blocked IL6-induced STAT3 activation. These data suggest that cholesterol depletion not only blocks intracellular oncogenic signaling (AKT) as previously described, but also blocks prostate cancer cells' ability to respond to external stimuli (IL6). Our data are consistent with a previous study in which filipin, a cholesterol-binding compound that disrupts plasma membrane lipid rafts, inhibited IL6-mediated STAT3 activation in LNCaP cells (20).Together, this helps explain the anti–prostate cancer activity of 27HC-mediated cholesterol depletion and provides a clear insight into a key mechanism by which 27HC inhibits prostate cancer growth.

Clinical trials of STAT3 inhibitors in cancer have not been fruitful in part due to weak on-target activity or potency and unfavorable pharmacokinetics (47). Moreover, we were not able to identify any current trials testing STAT3 inhibitors in prostate cancer. As such, the finding that 27HC works, in part, via inhibiting STAT3 is significant and suggests that STAT3 may indeed be a viable target in prostate cancer with proper adjuvant therapy. However, the overall modest growth inhibition of 27HC monotherapy seen in the DU145 xenograft model, although significant, argues that 27HC alone is unlikely to be a clinical tool to block STAT3. However, our in vitro data highlight a potential mechanism of sensitization to STAT3-targeted therapies (Fig. 6C). Lastly, it is important to reiterate that non–STAT3-driven prostate cancer cell lines and xenograft tumors (i.e., 22RV1) are perturbed by increased levels of 27HC (8). Thus, 27HC-mediated effects likely involve other signaling pathways in addition to IL6–JAK–STAT3. Nonetheless, our findings provide crucial mechanistic insight into how 27HC in part exerts its anti–prostate cancer activity.

Regarding developing a therapeutic strategy to treat prostate cancer, it is noteworthy that our combination of 27HC and STAT3 inhibitors worked together to both inhibit STAT3 signaling and slow tumor cell growth in vitro with synergistic activity. If validated in future studies, this has important clinical implications. First, it suggests that combination therapies targeting the same pathway may provide better efficacy than single-agent drugs—an idea already proven true in prostate cancer with dual targeting of the androgen receptor via castration plus the androgen synthesis bioinhibitor, abiraterone, or androgen receptor blocker, enzalutamide. Second, it suggests that novel approaches to targeting STAT3 may succeed where prior single-agent STAT3 inhibitors have failed (48). We suggest further research should explore the possibility that 27HC in combination with STAT3 inhibitors may have in vivo anti–prostate cancer activity, though this was beyond the scope of the current study, which was focused on the mechanistic understanding of how 27HC exerted its anti–prostate cancer activity.

It should be noted that like prostate cancer, breast cancers exhibit dysregulation of cholesterol and CYP27A1 expression. However, unlike prostate cancer, where CYP27A1 and thereby 27HC is frequently lost (8), breast cancers show upregulation of CYP27A1 (35) and thereby increased levels of 27HC. The likely explanation is that 27HC, beyond being an LXR agonist, is also an estrogen receptor (ER) agonist. In this respect, 27HC was described to be an ER ligand that promotes ER+ breast tumor growth (49). Therefore, higher 27HC directly stimulates ER to drive breast cancer growth. This highlights that our findings are likely prostate cancer specific and may not apply universally to other cancer types, particularly breast cancer.

Given 27HC suppresses STAT3 activity, one would expect high CYP27A1 (i.e., presumably high intracellular 27HC) tumors to be associated with lower STAT3 activity. However, in a large set of human prostate cancer transcriptomes, we found that tumors with high CYP27A1 expression had high STAT3 activation signatures with associations being stronger in high-grade tumors, mCRPC tumors, and PCS1 subtype tumors, a subtype of prostate cancer that You and colleagues previously showed was very aggressive (27). This suggests that in these aggressive tumors, STAT3 activation may be a resistance mechanism to retained CYP27A1 expression. Ultimately, more studies using human prostate cancer specimens are needed to better understand these interrelationships and importantly to understand which subset of prostate cancers may be sensitive to 27HC and STAT3 inhibition and which may not be.

Our study is not without limitations. Although our results were shown in multiple cell lines, we were not able to model the full heterogeneity of prostate cancer biology. Thus, it is possible that other prostate cancer cell lines and ultimately human prostate cancers behave differently. In addition, alternative formulations and in vivo delivery strategies for improved bioavailability of 27HC would need to be explored given the lipophilic nature of 27HC. Understanding this heterogeneity and improving 27HC bioavailability will be important for translating these findings to the clinic. Specifically, for future studies combining STAT3 inhibitors and 27HC, identifying biomarkers of sensitivity will be crucial. Finally, though we identified that inhibiting lipid rafts as evidenced by decreased STAT3 activation is one of the key mechanisms by which 27HC exerts its anti–prostate cancer activity, there are likely other lipid raft–mediated pathways as well as perhaps nonlipid raft–mediated effects too. Thus, focused research efforts are needed to fully understand all the mechanisms by which 27HC works in prostate cancer cells.

In summary, we demonstrate that 27HC results in rapid cholesterol depletion leading to disruption of lipid raft signaling and specifically inhibits the IL6–JAK–STAT3 signaling axis. This provides key mechanistic insight into our prior findings that 27HC inhibits prostate cancer cell growth. Future studies are required to explore the clinical implications of these findings for prostate cancer therapy.

No potential conflicts of interest were disclosed.

Conception and design: S. Dambal, M. Alfaqih, M.R. Freeman, E. Macias, S.J. Freedland

Development of methodology: S. Dambal, M. Alfaqih, S. Sanders, E. Maravilla, M. Reis-Sobreiro, M. Rotinen, T.J. Talisman, S.J. Freedland

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Dambal, G.C. Galvan, L.M. Driver, T.J. Talisman, E. Macias, S.J. Freedland

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Dambal, E. Maravilla, G.C. Galvan, M. Rotinen, M.S. Behrove, T.J. Talisman, J. Yoon, S. You, J.-T. Chi, M.R. Freeman, E. Macias, S.J. Freedland

Writing, review, and/or revision of the manuscript: S. Dambal, M. Alfaqih, S. Sanders, E. Maravilla, A. Ramirez-Torres, G.C. Galvan, M. Reis-Sobreiro, M. Rotinen, T.J. Talisman, S. You, J. Turkson, J.-T. Chi, M.R. Freeman, E. Macias, S.J. Freedland

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Maravilla, E. Macias, S.J. Freedland

Study supervision: E. Macias, S.J. Freedland

This work was supported by the Department of Defense Prostate Cancer Research Program award to S.J. Freedland (PC160648) and institutional funds from the Department of Pathology, Duke University School of Medicine to E. Macias.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Siegel
RL
,
Miller
KD
,
Jemal
A
. 
Cancer statistics, 2019
.
CA Cancer J Clin
2019
;
69
:
7
34
.
2.
Peisch
SF
,
Van Blarigan
EL
,
Chan
JM
,
Stampfer
MJ
,
Kenfield
SA
. 
Prostate cancer progression and mortality: a review of diet and lifestyle factors
.
World J Urol
2017
;
35
:
867
74
.
3.
Richard
A
,
Faeh
D
,
Bopp
M
,
Rohrmann
S
. 
Diet and other lifestyle factors associated with prostate cancer differ between the German and Italian region of Switzerland
.
Int J Vitam Nutr Res
2016
;
86
:
1
8
.
4.
Wolk
A
. 
Diet, lifestyle and risk of prostate cancer
.
Acta Oncol
2005
;
44
:
277
81
.
5.
Allott
EH
,
Farnan
L
,
Steck
SE
,
Song
L
,
Arab
L
,
Su
LJ
, et al
Statin use, high cholesterol and prostate cancer progression: results from HCaP-NC
.
Prostate
2018
;
78
:
857
64
.
6.
Allott
EH
,
Masko
EM
,
Freedland
AR
,
Macias
E
,
Pelton
K
,
Solomon
KR
, et al
Serum cholesterol levels and tumor growth in a PTEN-null transgenic mouse model of prostate cancer
.
Prostate Cancer Prostatic Dis
2018
;
21
:
196
203
.
7.
Platz
EA
,
Till
C
,
Goodman
PJ
,
Parnes
HL
,
Figg
WD
,
Albanes
D
, et al
Men with low serum cholesterol have a lower risk of high-grade prostate cancer in the placebo arm of the prostate cancer prevention trial
.
Cancer Epidemiol Biomarkers Prev
2009
;
18
:
2807
13
.
8.
Alfaqih
MA
,
Nelson
ER
,
Liu
W
,
Safi
R
,
Jasper
JS
,
Macias
E
, et al
CYP27A1 loss dysregulates cholesterol homeostasis in prostate cancer
.
Cancer Res
2017
;
77
:
1662
73
.
9.
Jamnagerwalla
J
,
Howard
LE
,
Allott
EH
,
Vidal
AC
,
Moreira
DM
,
Castro-Santamaria
R
, et al
Serum cholesterol and risk of high-grade prostate cancer: results from the REDUCE study
.
Prostate Cancer Prostatic Dis
2018
;
21
:
252
9
.
10.
Bansal
D
,
Undela
K
,
D'Cruz
S
,
Schifano
F
. 
Statin use and risk of prostate cancer: a meta-analysis of observational studies
.
PLoS One
2012
;
7
:
e46691
.
11.
de Boussac
H
,
Pommier
AJC
,
Dufour
J
,
Trousson
A
,
Caira
F
,
Volle
DH
, et al
LXR, prostate cancer and cholesterol: the good, the bad and the ugly
.
Am J Cancer Res
2013
;
3
:
58
69
.
12.
Krycer
JR
,
Phan
L
,
Brown
AJ
. 
A key regulator of cholesterol homoeostasis, SREBP-2, can be targeted in prostate cancer cells with natural products
.
Biochem J
2012
;
446
:
191
201
.
13.
Pommier
AJC
,
Dufour
J
,
Alves
G
,
Viennois
E
,
De Boussac
H
,
Trousson
A
, et al
Liver X receptors protect from development of prostatic Intra-Epithelial Neoplasia in Mice
.
PLoS Genet
2013
;
9
:e1003483.
14.
Zhuang
L
,
Kim
J
,
Adam
RM
,
Solomon
KR
,
Freeman
MR
. 
Cholesterol targeting alters lipid raft composition and cell survival in prostate cancer cells and xenografts
.
J Clin Invest
2005
;
115
:
959
68
.
15.
Yue
S
,
Li
J
,
Lee
SY
,
Lee
HJ
,
Shao
T
,
Song
B
, et al
Cholesteryl ester accumulation induced by PTEN loss and PI3K/AKT activation underlies human prostate cancer aggressiveness
.
Cell Metab
2014
;
19
:
393
406
.
16.
Stopsack
KH
,
Gerke
TA
,
Sinnott
JA
,
Penney
KL
,
Tyekucheva
S
,
Sesso
HD
, et al
Cholesterol metabolism and prostate cancer lethality
.
Cancer Res
2016
;
76
:
4785
90
.
17.
Nelson
ER
,
DuSell
CD
,
Wang
X
,
Howe
MK
,
Evans
G
,
Michalek
RD
, et al
The oxysterol, 27-hydroxycholesterol, links cholesterol metabolism to bone homeostasis through its actions on the estrogen and liver X receptors
.
Endocrinology
2011
;
152
:
4691
705
.
18.
Zhu
R
,
Ou
Z
,
Ruan
X
,
Gong
J
. 
Role of liver X receptors in cholesterol efflux and inflammatory signaling (review)
.
Mol Med Rep
2012
;
5
:
895
900
.
19.
Kusumi
A
,
Fujiwara
TK
,
Tsunoyama
TA
,
Kasai
RS
,
Liu
AA
,
Hirosawa
KM
, et al
Defining raft domains in the plasma membrane
.
Traffic
2020
;
21
:
106
37
.
20.
Kim
J
,
Adam
RM
,
Solomon
KR
,
Freeman
MR
. 
Involvement of cholesterol-rich lipid rafts in interleukin-6-induced neuroendocrine differentiation of LNCaP prostate cancer cells
.
Endocrinology
2004
;
145
:
613
9
.
21.
Loh
CY
,
Arya
A
,
Naema
AF
,
Wong
WF
,
Sethi
G
,
Looi
CY
. 
Signal transducer and activator of transcription (STATs) proteins in cancer and inflammation: functions and therapeutic implication
.
Front Oncol
2019
;
9
:
48
.
22.
Santoni
M
,
Conti
A
,
Piva
F
,
Massari
F
,
Ciccarese
C
,
Burattini
L
, et al
Role of STAT3 pathway in genitourinary tumors
.
Future Sci OA
2015
;
1
:
FSO15
.
23.
Abdulghani
J
,
Gu
L
,
Dagvadorj
A
,
Lutz
J
,
Leiby
B
,
Bonuccelli
G
, et al
Stat3 promotes metastatic progression of prostate cancer
.
Am J Pathol
2008
;
172
:
1717
28
.
24.
Blando
JM
,
Carbajal
S
,
Abel
E
,
Beltran
L
,
Conti
C
,
Fischer
S
, et al
Cooperation between Stat3 and Akt signaling leads to prostate tumor development in transgenic mice
.
Neoplasia
2011
;
13
:
254
65
.
25.
Hafeez
BB
,
Fischer
JW
,
Singh
A
,
Zhong
W
,
Mustafa
A
,
Meske
L
, et al
Plumbagin inhibits prostate carcinogenesis in intact and castrated PTEN knockout mice via targeting PKCepsilon, Stat3, and epithelial-to-mesenchymal transition markers
.
Cancer Prev Res
2015
;
8
:
375
86
.
26.
Don-Doncow
N
,
Marginean
F
,
Coleman
I
,
Nelson
PS
,
Ehrnstrom
R
,
Krzyzanowska
A
, et al
Expression of STAT3 in prostate cancer metastases
.
Eur Urol
2017
;
71
:
313
6
.
27.
You
S
,
Knudsen
KB
,
Erho
N
,
Alshalalfa
M
,
Takhar
M
,
Ashab
HA
, et al
Integrated classification of prostate cancer reveals a novel luminal subtype with poor outcome
.
Cancer Res
2016
;
76
:
4948
58
.
28.
Levine
DM
,
Haynor
DR
,
Castle
JC
,
Stepaniants
SB
,
Pellegrini
M
,
Mao
M
, et al
Pathway and gene-set activation measurement from mRNA expression data: the tissue distribution of human pathways
.
Genome Biol
2006
;
7
:
R93
.
29.
Macias
E
,
Rao
D
,
Carbajal
S
,
Kiguchi
K
,
DiGiovanni
J
. 
Stat3 binds to mtDNA and regulates mitochondrial gene expression in keratinocytes
.
J Invest Dermatol
2014
;
134
:
1971
80
.
30.
Johnson
BB
,
Moe
PC
,
Wang
D
,
Rossi
K
,
Trigatti
BL
,
Heuck
AP
. 
Modifications in perfringolysin O domain 4 alter the cholesterol concentration threshold required for binding
.
Biochemistry
2012
;
51
:
3373
82
.
31.
Liu
SL
,
Sheng
R
,
Jung
JH
,
Wang
L
,
Stec
E
,
O'Connor
MJ
, et al
Orthogonal lipid sensors identify transbilayer asymmetry of plasma membrane cholesterol
.
Nat Chem Biol
2017
;
13
:
268
74
.
32.
Lee
CD
,
Sun
HC
,
Hu
SM
,
Chiu
CF
,
Homhuan
A
,
Liang
SM
, et al
An improved SUMO fusion protein system for effective production of native proteins
.
Protein Sci
2008
;
17
:
1241
8
.
33.
Sengupta
P
,
Jovanovic-Talisman
T
,
Skoko
D
,
Renz
M
,
Veatch
SL
,
Lippincott-Schwartz
J
. 
Probing protein heterogeneity in the plasma membrane using PALM and pair correlation analysis
.
Nat Methods
2011
;
8
:
969
75
.
34.
Kwiatek
JM
,
Owen
DM
,
Abu-Siniyeh
A
,
Yan
P
,
Loew
LM
,
Gaus
K
. 
Characterization of a new series of fluorescent probes for imaging membrane order
.
PLoS One
2013
;
8
:
e52960
.
35.
Nelson
ER
,
Wardell
SE
,
Jasper
JS
,
Park
S
,
Suchindran
S
,
Howe
MK
, et al
27-Hydroxycholesterol links hypercholesterolemia and breast cancer pathophysiology
.
Science
2013
;
342
:
1094
8
.
36.
Zhao
Y
,
Gao
JL
,
Ji
JW
,
Gao
M
,
Yin
QS
,
Qiu
QL
, et al
Cytotoxicity enhancement in MDA-MB-231 cells by the combination treatment of tetrahydropalmatine and berberine derived from Corydalis yanhusuo W. T. Wang
.
J Intercult Ethnopharmacol
2014
;
3
:
68
72
.
37.
Jin
L
,
Millard
AC
,
Wuskell
JP
,
Dong
X
,
Wu
D
,
Clark
HA
, et al
Characterization and application of a new optical probe for membrane lipid domains
.
Biophys J
2006
;
90
:
2563
75
.
38.
Mora
LB
,
Buettner
R
,
Seigne
J
,
Diaz
J
,
Ahmad
N
,
Garcia
R
, et al
Constitutive activation of Stat3 in human prostate tumors and cell lines: direct inhibition of Stat3 signaling induces apoptosis of prostate cancer cells
.
Cancer Res
2002
;
62
:
6659
66
.
39.
Clark
J
,
Edwards
S
,
Feber
A
,
Flohr
P
,
John
M
,
Giddings
I
, et al
Genome-wide screening for complete genetic loss in prostate cancer by comparative hybridization onto cDNA microarrays
.
Oncogene
2003
;
22
:
1247
52
.
40.
Ichiba
M
,
Nakajima
K
,
Yamanaka
Y
,
Kiuchi
N
,
Hirano
T
. 
Autoregulation of the Stat3 gene through cooperation with a cAMP-responsive element-binding protein
.
J Biol Chem
1998
;
273
:
6132
8
.
41.
Narimatsu
M
,
Maeda
H
,
Itoh
S
,
Atsumi
T
,
Ohtani
T
,
Nishida
K
, et al
Tissue-specific autoregulation of the stat3 gene and its role in interleukin-6-induced survival signals in T cells
.
Mol Cell Biol
2001
;
21
:
6615
25
.
42.
Okamoto
M
,
Lee
C
,
Oyasu
R
. 
Interleukin-6 as a paracrine and autocrine growth factor in human prostatic carcinoma cells in vitro
.
Cancer Res
1997
;
57
:
141
6
.
43.
Haftchenary
S
,
Luchman
HA
,
Jouk
AO
,
Veloso
AJ
,
Page
BD
,
Cheng
XR
, et al
Potent targeting of the STAT3 protein in brain cancer stem cells: a promising route for treating glioblastoma
.
ACS Med Chem Lett
2013
;
4
:
1102
7
.
44.
Bharadwaj
U
,
Eckols
TK
,
Xu
X
,
Kasembeli
MM
,
Chen
Y
,
Adachi
M
, et al
Small-molecule inhibition of STAT3 in radioresistant head and neck squamous cell carcinoma
.
Oncotarget
2016
;
7
:
26307
30
.
45.
Murai
T
. 
The role of lipid rafts in cancer cell adhesion and migration
.
Int J Cell Biol
2012
;
2012
:
763283
.
46.
Liberzon
A
,
Birger
C
,
Thorvaldsdottir
H
,
Ghandi
M
,
Mesirov
JP
,
Tamayo
P
. 
The molecular signatures database (MSigDB) hallmark gene set collection
.
Cell Syst
2015
;
1
:
417
25
.
47.
Wong
ALA
,
Hirpara
JL
,
Pervaiz
S
,
Eu
JQ
,
Sethi
G
,
Goh
BC
. 
Do STAT3 inhibitors have potential in the future for cancer therapy?
Expert Opin Investig Drugs
2017
;
26
:
883
7
.
48.
Fizazi
K
,
Tran
N
,
Fein
L
,
Matsubara
N
,
Rodriguez-Antolin
A
,
Alekseev
BY
, et al
Abiraterone plus prednisone in metastatic, castration-sensitive prostate cancer
.
N Engl J Med
2017
;
377
:
352
60
.
49.
Wu
Q
,
Ishikawa
T
,
Sirianni
R
,
Tang
H
,
McDonald
JG
,
Yuhanna
IS
, et al
27-Hydroxycholesterol promotes cell-autonomous, ER-positive breast cancer growth
.
Cell Rep
2013
;
5
:
637
45
.