The Cholesterol Metabolite 27HC Increases Secretion of Extracellular Vesicles Which Promote Breast Cancer Progression

Abstract

Cholesterol has been implicated in the clinical progression of breast cancer, a disease that continues to be the most commonly diagnosed cancer in women. Previous work has identified the cholesterol metabolite 27-hydroxycholesterol (27HC) as a major mediator of the effects of cholesterol on breast tumor growth and progression. 27HC can act as an estrogen receptor (ER) modulator to promote the growth of ERα+ tumors, and as a liver X receptor (LXR) ligand in myeloid immune cells to establish an immune-suppressive program. In fact, the metastatic properties of 27HC require the presence of myeloid cells with neutrophils (polymorphonuclear neutrophils; PMNs) being essential for the increase in lung metastasis in murine models. In an effort to further elucidate the mechanisms by which 27HC alters breast cancer progression, we made the striking finding that 27HC promoted the secretion of extracellular vesicles (EVs), a diverse assortment of membrane bound particles that includes exosomes. The resulting EVs had a size distribution that was skewed slightly larger than EVs generated by treating cells with vehicle. The increase in EV secretion and size was consistent across 3 different subtypes: primary murine PMNs, RAW264.7 monocytic cells, and 4T1 murine mammary cancer cells. Label-free analysis of 27HC-EVs indicated that they had a different metabolite composition to those from vehicle-treated cells. Importantly, 27HC-EVs from primary PMNs promoted tumor growth and metastasis in 2 different syngeneic models, demonstrating the potential role of 27HC-induced EVs in the progression of breast cancer. EVs from PMNs were taken up by cancer cells, macrophages, and PMNs, but not T cells. Since EVs did not alter proliferation of cancer cells, it is likely that their protumor effects are mediated through interactions with myeloid cells. Interestingly, RNA-seq analysis of tumors from 27HC-EV-treated mice do not display significantly altered transcriptomes, suggesting that the effects of 27HC-EVs occur early on in tumor establishment and growth. Future work will be required to elucidate the mechanisms by which 27HC increases EV secretion, and how these EVs promote breast cancer progression. Collectively, however, our data indicate that EV secretion and content can be regulated by a cholesterol metabolite, which may have detrimental effects in terms of disease progression, important findings given the prevalence of both breast cancer and hypercholesterolemia.

Elevated circulating total cholesterol has been identified as a poor prognostic for breast cancer patients, while the use of cholesterol-lowering drugs such as statins is associated with increased recurrence-free survival (1–3). Previous work in preclinical models has demonstrated that elevated dietary cholesterol increases breast tumor growth as well as metastasis (4, 5). While the underlying mechanisms behind these observations are undoubtedly multifactorial, it is now well established in preclinical models that the primary cholesterol metabolite, 27-hydroxycholesterol (27HC), is a major mediator. Specifically, 27HC can activate the estrogen receptor (ER) to directly promote the proliferation and growth of ERα-positive tumors (4, 6, 7), while working through the liver X receptors (LXRs) in myeloid cells to suppress the anticancer immune response and promote metastasis (5, 8). Intriguingly, using murine models, we have found that the prometastatic effects of 27HC to the lung requires the presence of neutrophils (polymorphonuclear neutrophils; PMNs) (5), and that 27HC can act through any type of myeloid cell to suppress subsequent T cell expansion and activity, and promote the progression of breast and ovary tumors (8, 9).

PMNs are present in the tumor microenvironment and are known to be important for cancer progression, largely being described for their protumor capacity (10), although they may also have antitumor effects (11). While they have long been described as short-lived effector cells involved in acute immunity, there is now significant evidence supporting more nuanced physiological roles for neutrophils. For example, it has been shown that neutrophils can differentiate to have longer life spans once they undergo transendothelial migration into their target tissue (12, 13). Additional roles for neutrophils include their ability to facilitate cross-talk between other cells of innate and adaptive immunity (12, 14), including being able to suppress T-cell expansion and activity when treated with 27HC (8).

Myeloid cells such as PMNs have been documented to secrete extracellular vesicles (EVs) that are involved in modulating immune function. For example, activated PMNs release EVs that promote proteolytic extracellular matrix remodeling, thereby furthering pathology associated with inflammation (15). PMN-derived microparticles including exosomes can protect from inflammatory arthritis (16). PMNs can also release functional ectosomes (17, 18), which are able to inhibit the maturation of monocyte-derived dendritic cells (19), important antigen-presenting cells. Finally, there are a wealth of studies implicating EVs secreted from various sources in the pathophysiology of cancer (20-24).

Originating from a wide range of cell types, EVs are membrane-bound particles that contain microRNA (25), DNA (26), proteins (27), and lipids (28). When secreted, they are selectively taken up by “target” cells where they are then are involved in modulating downstream signaling, gene regulation, or cellular function (29). EVs broadly encompass heterogeneous populations of small particles released by cells, and this term includes groups such as microvesicles/microparticles (30), oncosomes (31), ectosomes (32), exosomes (33), and exomeres (34). While these groups are differentiated by form, function, and general characteristics, entities such as microparticles, ectosomes, and oncosomes bleb off the plasma membrane of cells, while exosomes are formed by internal budding of the plasma membrane to form multivesicular bodies within endosomes (32, 35–37). Exosomes are defined as between 40 and 100 nm in diameter and are often characterized by their expression of tetraspannin proteins, CD63, CD81, and CD9, remnants of the endosomal origin. It is important to note that the definition of “exosomes” continues to evolve as our understanding of EV heterogeneity expands (38). For the work described herein, we refer to the populations described as EVs, recognizing the wide heterogeneity of the specific groups this encompasses.

There is emerging evidence that EVs, including exosomes, are important for regulating cholesterol and lipid homeostasis (39–42). For example, cholesterol can be sequestered in EVs under hypercholesterolemic conditions (43), while statins (inhibitors of HMGCoA-reductase, the rate-limiting step in cholesterol synthesis) can reduce EV secretion (44). Interestingly, 27HC too can be detected in EVs (45). However, whether or how EV secretion is regulated by hormones and metabolites is poorly understood.

The main physiologic functions of EVs are still being revealed, and their role in pathologic states is the subject of intense ongoing research. Given the role of oxysterols such as 27HC in the regulation of cholesterol homeostasis (3), the likely roles of EVs in cholesterol homeostasis (28, 40), and the tumor-promoting roles of both 27HC and EVs (4, 5, 22), we sought to determine whether 27HC regulates EV secretion. Our findings demonstrate that 27HC can induce EV secretion and uncover a novel pathway by which cholesterol promotes breast cancer progression.

Materials and Methods

Animals

All protocols involving the use of animals were approved by the University of Illinois Institutional Animal Care and Use Committee (IACUC). Female mice between 8 and 10 weeks of age were purchased from Charles River Laboratories and housed at the University of Illinois in individually vented cages at 3 to 5 mice per cage. Food and water were provided ad libitum.

Cell Culture

Murine mammary cancer cell lines (4T1 and Met1) and RAW 264.7 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 1% penicillin/streptomycin, 1% nonessential amino acid, 1% sodium pyruvate, and 10% fetal bovine serum (FBS). In experiments requiring the harvest of EVs, 10% EV-depleted FBS was added in place of normal serum.

Enrichment and Culture of Primary Bone Marrow–derived PMNs

We adapted our protocol from Swamydas et al. (46). Briefly, femurs and tibia from treatment-naïve mice were isolated and briefly soaked in 70% ethanol, rinsed in phosphate-buffered saline (PBS). Sterile medium was used to flush bone marrow over a 70-μM filter. Collected cells were centrifuged at 200g for 5 minutes and resuspended in 1 mL of medium. A fresh density gradient was prepared with Histopaque 1077 and Histopaque 1119, and cells were overlaid above Histopaque 1077. Cells were centrifuged at 800g for 30 minutes at 4°C. At the end of this centrifugation, cells distributed into 2 distinct layers between interfaces and a pellet containing red blood cells. The upper layer was gently removed using a transfer pipette and discarded. The lower layer of cells containing mature neutrophils with condensed nuclei was collected and gently pipetted into a fresh tube containing 10 mL of medium. Cells were washed at 200g for 5 minutes, and the resulting pellet was resuspended in complete RPMI medium containing 10% EV-depleted FBS.

Preparation of Bone Marrow–derived Macrophages

Bone marrow cells were collected from the tibia and femur of female BALB/C mice. Cells were passed through a 70-μm cell strainer and cultured in complete Dulbecco’s modified Eagle’s medium supplemented with Macrophage colony-stimulating factor (25 ng/mL, Novoprotein). Medium was replaced every 3 to 4 days. On day 10, bone marrow–derived macrophages (BMDMs) were harvested for further experiments (EVs uptake).

Preparation of Pan T Cells

Untouched, pan T cells were isolated form the spleen and lymph nodes of BALB/c mice. Cells were passed through a 40-μm cell strainer and isolated using Dynbeads Untouched Mouse T cells kit (Invitrogen) according to the manufacturer’s protocol. Resulting cells were used for further experiments (EVs uptake).

Quantitative Polymerase Chain Reaction

RNA was extracted using GeneJet RNA Purification kit (Thermo Fisher) and iScript Reverse Transcription Supermix (Bio-Rad) was applied for cDNA synthesis form 1 µg of RNA, according to the manufacturers’ protocols. Quantitative polymerase chain reaction (qPCR) primers were designed via Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). A list of all used primers is presented elsewhere (Table 1 (47)). mRNA expression was investigated with qPCR using iTaq Universal SYBR Green Supermix (Bio-Rad) on a CFX384 Touch real-time PCR detection system (Bio-Rad). Relative expression was determined via the 2^–ΔΔCT method and normalized to the housekeeping gene (TATA box binding protein).

Light Microscopy

Bone marrow cells were harvested and cultured as described above, and cultured on sterile glass coverslips coated with poly L-lysine for 4 hours. Cells were then fixed in 4% formalin. Cells were washed twice in PBS and then coverslips were mounted onto glass slides with Prolong Gold mounting medium, and cured for 24 hours. Coverslip edges were sealed and then the slides imaged.

EV harvesting

EVs were enriched using 1 of 2 different commercial kits: Total Exosome Isolation reagent (ThermoFisher) or ExoQuict-TC ULTRA (SBI). For characterization of EVs using simultaneous label-free autofluorescence multiharmonic (SLAM) imaging, differential centrifugation was used as described (48).

EV characterization by Nanoparticle analysis

The concentration and size of isolated EVs were quantified using the dynamic light-scattering particle-sizing system by NanoSight NS300 (Malvern Panalytical). For each measurement, 3 30-second videos were captured and analyzed using the built-in NanoSight Software NTA3.1

Electron Microscopy

Scanning electron microscopy (SEM) was performed on bone marrow derived PMNs. PMNs were treated for 17 hours with dimethyl sulfoxide (DMSO) or 27HC. Cells were then washed, fixed (glutaraldehyde, sodium cacodylate) and prepared for imaging by SEM. Transmission electron microscopy (TEM) was performed on isolated EVs as follows. Isolated EVs were fixed with an equal volume of 4% paraformaldehyde solution for 5 minutes, and then loaded on the EM grid. TEM images were obtained using JEOL 2100 TEM at 200 kV. For EVs from PMNs (Fig. 2C (47)), EVs were fixed with glutaraldehyde and negatively stained with ammonium molybdate.

Label-free Imaging With SLAM Microscopy

Enriched EVs from conditioned media of Raw264.7, PMNs or 4T1 cells were imaged with SLAM microscopy as previously described (48, 49). Briefly, 2-photon autofluorescence and 3-photon autofluorescence were used to detect the concentrations of flavin adenine dinucleotide (FAD) and Nicotinamide adenine dinucleotide (phosphate) (NAD(P)H). Emission beams at 580 to 640 nm and 420 to 480 nm respectively were used for each channel and they were used for segmenting out the EVs and calculating the optical redox ratio.

Primary Tumor Growth Model

Mice were ovariectomized to model a low estrogen state. Mice were then grafted with either 2 × 10⁵ 4T1 or Met1 cells into their axillary fat pads to allow primary tumors to develop. Twenty-four hours after grafting, EVs were injected daily via retro-orbital injection. EVs from 27HC- or DMSO-treated PMNs were isolated and then quantified by BCA protein assay, and between 3 and 4 μg in PBS was delivered to mice daily for 8 days (4T1 tumor–bearing mice) or 18 days (Met1-iRFP tumor–bearing mice). Sham injections were performed with PBS.

Metastatic Colonization Model

EVs from 27HC- or DMSO-treated PMNs were isolated and then quantified by BCA protein assay, and between 3 and 4 μg in PBS was delivered to mice daily for 14 days Retro-orbital injection sham injections were performed with PBS. At the end of 14 days, EV treatments were stopped and mice were injected with 1 × 10⁶ 4T1 or Met1 cells via the tail vein. Metastases were allowed to establish over 3 weeks, and quantified by in vivo imaging (IVIS; 4T1-luc) or ex vivo (quantitative PCR for specific markers such as luciferase for 4T1, and PyMT and iRFP for Met1 cells). In vivo bioluminescent imaging (IVIS) was performed as previously described (50).

Proliferation

Cells were seeded and treated with EVs at indicated concentrations and treated every other day with EVs at indicated concentrations. Proliferation was assessed indirectly by DNA concentration as described previously (9).

Flow Cytometry and EV Uptake

For EV characterization, DMSO- and 27HC-EVs were stained for known EV markers. Antibodies were purchased from BD Pharmingen: CD9 (Alexa Fluor 647, BD:564233 (51)), CD63 (PE, BD:564222 (52)), CD81 (BV421, BD:740060 (53)), and granulocyte (PMN) marker Ly6G (FITC, BD: 127606 (54)). Antibodies were used at a working dilution of 1:100 in fluorescence-activated cell sorting (FACS) buffer. EV were isolated with ExoQuict-TC ULTRA (SBI) and stained with diluted antibodies for 20 minutes at 4°C. After the designated time, in order to wash unbound antibody, ExoQuick-TC reagent was added for second round of EV isolation. The resulting EVs were analyzed by flow cytometry. The same protocol was adapted for analysis of EV uptake. During EV staining and isolating procedure, Diluent C with PKH 67 (Sigma-Aldrich) were added for 5 minutes of incubation at room temperature. The staining process was stopped by adding an equal volume of EV depleted FBS (Sigma-Aldrich). All other steps were performed as described above. During the staining procedure the negative control (unstained EVs) sample was processed as any other sample. Stained EVs were then incubated with different cells within 4 hours after preparation.

EV uptake analysis was conducted using Met-1, 4T1, RAW 264.7, BMDMs, PMNs, and T cells (activated with αCD3 and αCD28 or nonactivated (55, 56)). Prepared cells were seeded on 6-well plates (500 000 cells/well) and incubated with stained EVs in 3 biological replicates for each EV treatment: DMSO, 27HC, and negative control EV staining. After a 21-hour incubation, cells were fixed in 4% formaldehyde and diluted in FACS buffer for flow cytometry using a BD LSR Fortessa analyzer.

RNA-seq

RNA was isolated from Met1 tumors and processed for RNA-seq. Reads were 100 nt in length, and libraries prepared with Illumina’s TruSeq Stranded mRNA Sample Prep Kit. RNAseq libraries were quantitated by qPCR and sequenced for 101 cycles from each end of the fragments using a HiSeq 4000 sequencing kit version 1. The resulting reads were quality checked using MultiQC (v1.2). Kallisto (v0.44.0) was used to pseudoalign the reads to transcriptome and quantify the abundance of the transcripts, using transcriptome reference index built from the transcriptome fasta of Mus Musculus build GRCm38. Transcript abundance was subsequently summarized to gene level using R package tximport. Gene counts were log transformed to compare the gene expression profile of each sample based on multidimensional scaling (MDS): A distance matrix was generated by calculating the Euclidean distance between each sample based on gene expressions and classical MDS was performed using the matrix. For downstream analysis, genes with counts per million above 0 in at least 1 sample were retained, leaving 26796 genes for differential expression gene analysis. Voom was subsequently used to transform count data to log2-counts per million (logCPM), estimate the mean–variance relationship and obtain appropriate weights for each gene and sample. Differential gene expression analysis was performed using R packages limma-voom and EdgeR on the logCPM values. Corrections for multiple testing were made using the false discovery rate method (57). RNA-seq data has been submitted to the NCBI Gene Expression Omnibus (accession number: GSE171949 (58)).

Results

Elevated 27HC Induces Release of Unique EVs

There is an increasing abundance of studies implicating EVs in the pathological progression of cancer—both for EVs derived from cancer cells and for EVs derived from stromal cells, such as immune cells (24). However, the regulation of EV biogenesis, content, and secretion is less clear. Cholesterol and statins have recently been implicated in EV biogenesis (44), although the precise mechanisms behind this regulation remain elusive. Since 27HC is a primary metabolite of cholesterol, a regulator of cholesterol homeostasis (3), and we and others have shown that 27HC promotes the progression of breast and ovarian cancers (2, 4, 5, 7–9), we hypothesized that 27HC may also regulate EV secretion. Therefore, we treated a variety cell types with 27HC and assessed EV concentration using different approaches. We focused our efforts on 2 different myeloid cell types, primary PMNs and RAW 264.7 cells (a model of monocytes/macrophages), as myeloid cells have been strongly implicated in the 27HC-mediated progression of breast and ovarian cancer (5, 8, 9). We also assessed 4T1 cells, which are a murine model of triple negative breast cancer. Cells were treated with vehicle (DMSO) or 27HC. After 24 hours, EVs were isolated from the conditioned media using a conventional kit. Subsequent analysis of total protein content indicated that the concentration of EVs secreted from cells treated with 27HC was consistently higher than vehicle-treated cells (Fig. 1A).

Figure 1.

27HC induces the secretion of EVs from myeloid cells and cancer cells. Indicated cell types were treated with vehicle (DMSO) or 27HC for 24 hours. EVs were enriched from the conditioned media. (A) Total protein concentrations of EVs enriched from the conditioned media of PMNs, RAW 264.7 or 4T1 cells were determined by the BCA method. Asterisks (*) denote statistical significance (Student’s t test, P < .05, from left to right N = 8, 8, 4, 4, 3, 3). (B) Resulting EV concentrations from conditioned media were determined by nanoparticle tracking analysis (NTA). For each experiment particle counts were normalized to that of DMSO. 27HC consistently increased EVs secreted by all 3 cell types (Fisher exact test, P < .05. N = 8, 7 or 4 in each treatment group for PMN, 4T1 and RAW 264.7 respectively). (C) NTA analysis of EVs enriched from conditioned media of PMNs indicates that 27HC shifts their size distribution. Left panel: frequency histogram. Right panel: mean size ± standard error of the mean (asterisk indicates significant difference, P < .05, N = 2530 and 2536). Frequency histograms for EVs from 4T1 and RAW264.7 cells are available elsewhere (Fig. 1 (47)). (D) Representative scanning electron micrographs of bone marrow-isolated PMNs treated with either DMSO or 27HC. (E) Representative transmission electron micrographs of EVs enriched from the conditioned media of PMNs or 4T1 cells treated with either DMSO or 27HC. Additional micrographs from different experiments are shown elsewhere (Fig. 2C (47)). (F) Flow cytometric analysis of EVs released by PMNs treated with either DMSO or 27HC indicate that they express Ly6G (granulocyte marker), CD81 (tetraspanin), CD63 (tetraspanin), and CD9 (tetraspanin). CD81, CD63, and CD9 are reported markers of EVs. Left panels: unstained controls to observe baseline fluorescence. Right panels: stained samples. Single stained controls are shown elsewhere (Fig. 3 (47)).

To confirm these findings, we utilized nanoparticle tracking analysis (NTA; NanoSight NS300, Malvern Panalytic) to quantify particle concentration and size distribution. Intriguingly, regardless of cell type assessed, 27HC increased EV secretion (Fig. 1B). In addition to increasing secretion rate, 27HC also shifted the size distribution of secreted EVs, with the net result being an increase in the mean size of particle (Fig. 1C; Fig. 1A,B (47)).

Cultured PMNs isolated from murine bone marrow had characteristic features including the polymorphic nucleus, as assessed with Wright’s stain (Fig. 2A (47)). We have previously found that cultured PMNs remain viable for at least 72 hours (5), and routinely confirm this. For these studies, SEM also revealed that the characteristic structure of cultured PMNs remained intact, and that there were no obvious differences between DMSO- or 27HC-treated PMNs. However, more “debris” was noted for 27HC-treated PMNs (Fig. 1D; Fig. 2B (47)). In order to evaluate the resulting EVs at the microscopic level, we utilized TEM to visualize EVs secreted from PMNs and 4T1 cells. TEM images and sizes of EVs isolated from the different cell types resembled previously reported images of EVs, validating our isolation approach (Fig. 1E; Fig. 2C (47)) (59-63). Some morphological heterogeneity was observed for EVs isolated from both DMSO- or 27HC-treated cells, and thus it is difficult to ascertain whether 27HC altered morphology using this low-throughput technique.

EVs have been described to express certain tetraspanins, which are commonly used as confirmatory surface markers. We used flow cytometry to assess whether the EVs we isolated also expressed these markers, and thus verify their identity as EVs. As can been seen in Fig. 1F (also Fig. 3 (47)), EVs isolated from primary PMNs expressed the granulocytic marker Ly6G as expected. They also exhibited positive staining for several characteristic tetraspanins: CD63, CD9, and CD81 (Fig. 1F; Fig. 3 (47)). 27HC treatment of PMNs did not significantly alter the relative number of EVs expressing CD63, CD9, or CD81 (Fig. 4A (47)). Furthermore, 27HC did not consistently alter the mRNA expression of these tetraspanins across the parental cells (PMNs, 4T1, or RAW 264.7), indicating that it was indeed regulating assembly and secretion, not just expression of EV components (Fig. 4B (47)).

In order to further characterize EVs derived from cells treated with 27HC, we made use of SLAM microscopy (49). This label-free nonlinear optical imaging technique uses a single-excitation source and shaped ultrafast pulses permitting simultaneous and efficient acquisition of multiphoton autofluorescence (FAD and NAD(P)H) and second/third harmonic generation from a wide array of cellular and extracellular components. It has been successfully adapted to image EVs (48). Intriguingly, in previous studies, SLAM microscopy found that a low FAD:(FAD+NAD(P)H) ratio was indicative of EVs from invasive ductal breast carcinoma compared with those EVs isolated from tissue from women with no history of cancer undergoing breast reduction surgery (48). Furthermore, tumor stage was correlated with a decreased FAD:(FAD+NAD(P)H) ratio (48). For these assays, EVs were isolated by ultracentrifugation and imaged by SLAM. Interestingly, EVs from 27HC-treated PMNs, RAW 264.7, or 4T1 cells had a decreased FAD:(FAD+NAD(P)H) ratio compared with those from DMSO-treated cells (Fig. 2). The difference in distribution was more pronounced in PMNs (Fig. 2A). Also of interest, EVs from RAW 264.7 cells or 4T1 cells tended to have a higher ratio at baseline than PMNs, with 27HC reducing this (Fig. 2B). Thus, the metabolic signature of EVs derived from 27HC-treated cells (PMNs, RAW 264.7, or 4T1) more closely mirrored the metabolic ratio observed in higher stage breast cancers. This observation gives reason to speculate that 27HC may be 1 factor responsible for the lower metabolic ratios observed in EVs from higher grade tumors or between invasive breast cancer compared with healthy tissue (48); this is based on the observations that the expression of the enzyme CYP27A1 has been positively associated with breast cancer grade (4) and lethal prognosis (64), suggesting that local 27HC may be a driving force behind the EV signature change from normal through to stage 4 breast cancer.

Figure 2.

Simultaneous label-free autofluorescence-multiharmonic (SLAM) microscopy demonstrates that 27HC induced EVs have unique metabolic signature. SLAM microscopy was used to evaluate EVs from media conditioned by (A) PMNs, (B) RAW 264.7 cells, or (C) 4T1 cells that were treated with either vehicle (DMSO) or 27HC. Left panel: frequency distribution histogram. Right panel: mean size ± standard error of the mean (asterisk indicates significant difference, P < .05) (A: N = 524 and 1040. B: N = 635 and 906. C: N = 295 and 234).

EVs derived from 27HC-treated PMNs promote tumor growth and metastatic colonization in syngeneic murine models of mammary cancer.

EV secretion was stimulated by 27HC with resulting EVs having altered biophysical properties similar to EVs isolated from high-stage breast tumors. Therefore, it was of interest as to whether EVs resulting from 27HC treatment directly impacted tumor progression. Since we have previously implicated myeloid cells in mediating many protumorigenic properties of 27HC (5, 8, 9), and in particular PMNs in promoting the colonization of lung by metastatic breast cancer cells (5), we focused on EVs derived from this cell type.

We first assessed the impact of acute supplementation with EVs on the growth of 4T1 tumors. One day after orthotopic graft with 4T1 cells, daily administration of EVs was initiated which continued through day 8 postgraft. In support of our hypothesis, tumors in mice treated with EVs derived from 27HC-treated PMNs were consistently larger than those in mice treated with EVs derived from DMSO-treated or untreated PMNs (Fig. 3A). Interestingly, tumors in mice treated with EVs isolated from DMSO-treated PMNs grew at the same rate as those in mice that were untreated, indicating that 27HC reprogramed seemingly innocuous EVs into having protumor properties (Fig. 3A).

Figure 3.

EVs derived from 27HC-treated PMNs promote the growth of murine mammary tumors. PMNs were isolated from the bone marrow of naïve, wildtype female mice, and cultured in the presence of DMSO (vehicle) or 27HC. (A) 4T1 cells were orthotopically grafted into the mammary fat pad of female mice. Daily placebo or EV treatment commenced between days 1 and 8 (inclusive, represented by dashed lines). (B) Met1 cells were orthotopically grafted into the mammary fat pad of female mice. Daily placebo or EV treatment commenced on day 1 (dashed line) and continued through to the end of the experiment. Data were fit to an exponential growth regression model (y-intercept constrained to 0) to generate the curves pictured. Asterisk (*) represents significant difference between 27HC-EV and both DMSO-EV and placebo treatments, on the final day of measurement (P < .05, 2-way ANOVA followed by Tukey’s multiple comparison test) (for A: N = 9, 9, and 10. For B: N = 9, 0, and 10 for placebo, DMSO-EVs, and 27HC EVs respectively).

To complement the 4T1 tumor growth study, and evaluate the robustness of this response across models, we evaluated the impact of 27HC-EVs on the growth Met1 tumors. This line was originally derived from a lung metastatic lesion from an MMTV-PyMT mouse, and forms rapidly growing tumors with metastatic potential when grafted orthotopically (4, 5). EV treatment commenced 1 day postgraft and continued throughout the experiment. Similar to 4T1 tumors, EVs from 27HC-treated PMNs promoted the growth of Met1 tumors over those grown in mice that were treated with DMSO-EVs or placebo (PBS, no EVs) (Fig. 3B).

In order to assess the impact of 27HC-EVs on metastasis, we first used qPCR to quantify the presence of markers indicative of metastasis within the lungs of mice bearing orthotopic tumors (as in the lungs of those mice from Fig. 3A and 3B). Since our 4T1 cells had been engineered to express luciferase, we quantified the presence of luciferase transcript, as a measure of metastatic burden. For Met1 tumors, we quantified the presence of PyMT transcript, since the expression of this transgene would be restricted to Met1 cells. Interestingly, EVs derived from 27HC-treated PMNs resulted in increased luciferase or PyMT transcript in the lungs of 4T1- or Met1-bearing tumors respectively, indicating that in addition to promoting tumor growth, these EVs also promoted metastasis from the primary tumor (Fig. 4A and 4B). This was supported by the finding that 27HC-EVs also increased the proliferation marker, Ki-67, for both models (Fig. 4A and 4B), as well as iRFP in the lungs of mice bearing Met1 tumors, Met1 cells having been engineered to express this protein (4) (Fig. 5 (47)).

Figure 4.

EVs derived from 27HC-treated PMNs promote the metastatic colonization and outgrowth of mammary cancer. (A) Lungs from mice where 4T1 grafts were grown orthotopically (from Fig. 3A) were harvested and assessed for luciferase or Ki-67 mRNA by qPCR (N = 9-10/group). (B) Lungs from mice where Met1 grafts were grown orthotopically (from Fig. 3B; N = 9-10/group) were harvested and assessed for luciferase or PyMT mRNA by qPCR. (C) Lungs from mice where Met1 cells were grafted intravenously (tail vein) were harvested and assessed for PyMT, iRFP, or Ki-67 mRNA by qPCR (N = 10/group). (D) 4T1 cells were grafted intravenously and mice treated daily with placebo or EVs. Left panels: imaging of luciferase signal indicating metastatic outgrowth in the lungs and bones. Right panels: quantification of luciferase signal indicating that EVs from 27HC-treated PMNs promote metastatic colonization and growth in both the lungs (upper body) and bones (lower body) (N = 5/group). Asterisks (*) represent significant differences between DMSO-EVs and 27HC-EVs (A-C; P < .05, 2-tailed unpaired t test), or significant differences between placebo or DMSO-EVs and 27HC EVs (D: 1-way ANOVA followed by Student Newman–Keuls multiple comparison test).

To further investigate the impact of EVs derived from 27HC-treated PMNs on metastatic outgrowth, we utilized 2 models where mammary cells were introduced intravenously: Met1 and 4T1. The intravenous model of metastatic colonization, including the 4T1 model, is a well-established and accepted model of colonization and an appropriate model of late-stage metastasis. It continues to be standard in the field, has recently been shown to be a good tool in discovering new pathways and targets (65), and importantly, 27HC promotes colonization of this model (5). It is important to note that by introducing cells intravenously, any effects observed are restricted to the later stages of metastasis (extravasation, colonization, and growth). Since previous work has found that the procolonization effects of 27HC were largely due to its effects on the host (5), we pretreated mice with EVs for 2 weeks prior to engraftment with cells. When lungs from Met1-bearing mice were assessed, it was found that 27HC-EVs resulted in increased PyMT and iRFP expression, 2 markers expected to be expressed only within Met1 cells, and thus indicative of increased metastatic burden (Fig. 4C). Expression of the proliferative marker, Ki-67 was also increased in Met1 metastasis–bearing lungs from mice treated with 27HC-EVs. Similarly, pretreatment of mice with EVs isolated from PMNs treated with 27HC promoted the metastatic colonization and outgrowth of 4T1 lesions, as assessed by in vivo imaging for luciferase activity (Fig. 4D). Intriguingly, however, in addition to increasing metastatic outgrowth in the lungs, which is a common site for this model, it also significantly increased metastatic burden in the bones (Fig. 4D). Collectively, these data support the hypothesis that EVs generated from PMNs contribute to the protumor growth and metastatic properties of 27HC.

EVs derived from 27HC-treated PMNs do not alter cancer cell proliferation nor mRNA expression of key genes.

Since we found that 27HC-EVs from PMNs promoted mammary tumor growth and metastasis (Figs. 3 and 4), we were interested in elucidating the mechanism by which this occurs. Therefore, we evaluated the most logical hypothesis that these EVs might directly alter the proliferation of mammary cancer cells. We evaluated this in 3 different cell lines, using different approaches with respect to EV “dose.” First, E0771 cells were treated twice with 20 ng of EVs isolated from PMNs treated with either DMSO or 27HC, and no difference in proliferation was noted 4 days after seeding (Fig. 5A). Second, 4T1 cells were treated every 2 days with 5, 10, or 15 million EVs (as determined by NTA) every other day for 9 days, and no differences in proliferation were noted (Fig. 5B). Likewise Met1 cells were treated with EVs at concentrations every other day, and again no differences in proliferation were observed after 7 days (Fig. 5C). Comprehensive qPCR analysis of several cancer-related genes, spanning cytokines, angiogenesis, proliferation, and epithelial mesenchymal transition found that 27HC-EVs did not consistently change expression between 4T1 or Met1 cells (Fig. 6A,B (47)). Further, the expressions of classic ERα target genes Tff1 (PS2), PR, or CXCL12 were not altered in RAW 264.7 cells treated with 27HC-EVs (Fig. 6C (47)). The gene expression of enzymes responsible for 27HC metabolism (CYP27A1 and CYP7B1) were likewise unchanged when either 4T1, Met1, or RAS 2647 cells were incubated in the presence of 27HC-EVs (Fig. 6A-C (47)). Therefore, despite their clear role in promoting tumor growth and metastasis, it is unlikely that 27HC-EVs work directly on cancer cells, at least in terms of promoting cellular proliferation.

Figure 5.

EVs derived from 27HC-treated PMNs do not alter proliferation or consistent changes in gene expression across different murine mammary cancer cell lines. (A) E0771 cells plated on day 0 and were treated with 20 ng (total protein) of EVs from PMNs treated with either DMSO or 27HC on days 1 and 3. Proliferation was assessed (DNA content) on day 4. Data are with respect to DMSO EVs. (B) 4T1 cells were plated on day 0 and were treated on day 1 and every 2 days thereafter with either 5, 10, or 15 million EV particles derived from PMNs treated with DMSO or 27HC. On day 9, proliferation was assessed. Data is with respect to DMSO-EVs (5 million particles). (C) Met1 cells were plated on day 0 and treated on day 1 and every 2 days thereafter with indicated numbers of EV particles derived from PMNs treated with either DMSO or 27HC. On day 7, proliferation was assessed. Data are with respect to DMSO-EVs (5 million particles). (N = 3/group).

EVs From PMNs Are Preferentially Taken Up by Certain Cell Types

Since 27HC-EVs failed to influence cellular proliferation, it was not outwardly clear which cells these EVs actually target to exert their protumor effects. Therefore, in order to further investigate the potential mechanisms by which 27HC-EVs from PMNs promote tumor progression, we screened for potential target cells. EVs from PMNs were isolated and labeled with PKH67. We were able to analyze the EVs themselves using flow cytometry, confirming positive staining (Fig. 7A (47)). Labeled EVs were then incubated with different cell types for 21 hours prior to analysis by flow cytometry. Although 27HC-EVs did not promote proliferation, (Fig. 5), it was still possible that they impact cancer cells directly. Using a low stringent gating strategy (Fig. 6A; Fig. 7B-F (47)) based on cells treated with nonlabeled EVs, we found that nearly all Met1 cells take up EVs (Fig. 6B). Met1 cells took in EVs equally well for both EVs from DMSO or 27HC-treated PMNs (Fig. 6B). 4T1 cells also took up EVs (Fig. 6C; Fig. 7B (47)). However, EVs from 27HC-treated PMNs were taken up at greater efficiency in 4T1 cells than those from DMSO-treated PMNs, based on total number of positively stained cells (Fig. 6B and 6C).

Figure 6.

PMN derived EVs target select cell types. EVs were isolated from the conditioned media of PMNs treated with either vehicle (DMSO) or 27HC. They were then stained and incubated with indicated cell types. After 21 hours, cells were washed, fixed and analyzed by flow cytometry. Flow cytometry of stained EVs as well as gating strategies are elsewhere (Fig. 7 (47)). (A) Example gating strategy used for Met1 cells: (i) cells that were not treated with EVs, (ii) cells that were stained with unstained EVs (referred to as no stain in B-H), (iii) cells that were treated with DMSO-PMNs, and (iv) cells that were treated with 27HC-EVs. (B) Quantified data for Met1, indicating percentage of cells that stained positive for labeled EVs. Likewise, quantified data for (C) 4T1 cells, (D) RAW 264.7 cells, (E) primary murine bone marrow–derived macrophages (F) primary PMNs with 2 y-axis scales to allow for comparison, and either (G) nonstimulated T cells or (H) T cells activated with CD3 and CD28 antibodies. Please note different y-axis scale for (G) and (H) compared with (B-F). Different letters denote statistical significance (1-way ANOVA followed by the Student Newman–Keuls multiple comparison test, P < .05, N = 3/group).

As mentioned earlier though, it is likely that the protumor effects of 27HC-EVs are mediated through cells other than the cancer cells themselves. We therefore assessed EV uptake in the RAW 264.7 cells, a model of monocytic myeloid immune cell. Similar to 4T1 cells, EVs from 27HC-treated PMNs were more efficiently taken up by RAW 264.7 cells than EVs from DMSO-treated PMNs (Fig. 6D; Fig 7C (47)), and the overall uptake in terms of number of positive cells was similar to 4T1 cells. Likewise, primary BMDMs were also susceptible to EV internalization (Fig. 6E; Fig. 7D (47)). Again, a larger percentage of macrophages were positive for EV staining after treatment with 27HC-EVs than with EVs from DMSO-treated PMNs (~3-fold increase; Fig. 6E).

It is possible that EVs secreted from PMNs influence other PMNs in an autocrine manner. Indeed, as illustrated in Fig. 6F, primary PMNs were able to take in EVs, and particularly EVs secreted from 27HC-treated PMNs, although at a far reduced efficiency than the cancer cell lines or macrophages (gating strategy shown in Fig. 7E (47)). Interestingly, however, primary T cells did not display evidence of EV uptake, regardless of whether they were derived from PMNs treated with DMSO or 27HC, or whether the T cells were unstimulated or activated with CD3 and CD28 antibodies (Fig. 6G and 6H; Fig. 7F (47)). Various cell surface proteins such as integrins have been implicated in EV docking and being taken up by cells. It is possible that T cells lack these proteins—an aspect that remains to be explored. It is also important to note, that being able to take up EVs is only 1 aspect of how EVs work, and that cells such as T cells may still be influenced through direct contact with EVs (66-68).

In summary, EVs from primary PMNs were taken up by both of the mammary cancer cell lines that exhibited increased tumor growth when mice were treated with EVs from 27HC-treated PMNs (Met1 and 4T1). 4T1 cells appear to have an increased capacity to intake 27HC-EVs compared with DMSO-EVs, while Met1 cells showed no preference. RAW 264.7 cells, primary macrophages and primary PMNs all took in EVs, with a preference for 27HC-EVs. In contrast, T cells did display signs of EV internalization (Fig. 6).

Transcriptome From Late-stage Tumors Indicate no Significant Difference Between Those Treated With DMSO-EVs or 27HC-EVs

Since it was clear that 27HC-EVs derived from PMNs promoted tumor growth and metastatic colonization and outgrowth, but did not alter cellular proliferation of cancer cells, we decided to adopt a nonbiased approach to glean insight as to the mechanisms by which these EVs promote cancer progression. Met1 tumors from mice treated with DMSO-EVs or 27HC-EVs were used for RNA-seq (tumors from Fig. 5B). Somewhat surprisingly, no significant changes in gene expression were observed between these tumors (Fig. 7; Fig. 8 (47)). When assessed in isolation, there were some genes with seemingly altered expression, with an outlier, such as IL2, CD177, or Ch25h (Fig. 8A (47)). Expression of genes associated with 27HC metabolism, ER signaling, or LXR signaling were not significantly altered (Fig. 8B (47)). However, the outlier sample was not consistent between other genes and none of these genes had a false discovery rate of P < .05. Therefore, despite differences in tumor growth, the transcriptome of tumors taken at late stage (17 days postgraft) were not significantly different, indicating that the effects of EVs occur either (1) in a small subset of cells where gene expression changes were diluted in the larger tumor, (2) early on in tumor growth, or (3) do not involve regulation of gene transcription, but rather influence resource availability (such as a source for lipid) or alternate signaling.

Figure 7.

RNA-seq analysis of tumors from mice treated with 27HC-EVs indicates impact is likely at early stages of tumor growth. Met1 tumors from Fig. 3B were processed for RNA-seq. (A) A Multigene scaling plot indicating mRNA expression across genes for each sample. Gene counts were log transformed and underwent MDS. Each point represents 1 tumor sample. (B) After filtering lowly expressed genes, differential gene expression analysis was performed using Limma-voom package, and no differentially expressed gene was detected by a threshold of false discovery rate <0.05. (C) Mean difference (MD) plot shows the average log expression against the log2-fold change of genes from 27HC-EV-treated samples compared to the controls. Each point represents a gene. N = 9 for DMSO-EVs and 10 for 27HC-EVs.

Discussion

Cholesterol and its metabolites have been implicated in the pathophysiology of breast cancer onset and progression. It is becoming increasingly apparent that cholesterol has multifactorial effects on cancer biology, with a major contributor being its metabolite, 27HC (2, 3). It is unlikely that 27HC is in and of itself carcinogenic, and, indeed, circulating levels are associated with either no risk or decreased risk depending on the menopausal status of women (69). However, there is now emerging clinical evidence supporting a role for 27HC in promoting the progression of breast cancer. First, the concentration of 27HC was found to be elevated in breast tumors and normal adjacent tissue compared with tissue from healthy volunteers (7). Second, protein expression of the enzyme that synthesizes 27HC, CYP27A1, was associated with a higher breast tumor grade (4). In a recent study, it was found that higher CYP27A1 expression was correlated with higher breast tumor grade, size, and risk of late fatal disease, most prominently in postmenopausal ERα+ patients (64). Finally, elevated mRNA expression of the catabolic enzyme of 27HC (CYP7B1) within breast tumors is a good prognostic factor; elevated tumoral levels of CYP7B1 being expected to result in lower 27HC concentrations, as demonstrated for thyroid cancer (70). Likewise, EVs have been implicated in many facets of cancer biology, and are quickly emerging as biomarkers that are both diagnostic and prognostic. The basic pathways involved in the biogenesis of EVs have been described, and our understanding of these processes is rapidly evolving (33). However, the (patho)physiological factors that regulate biogenesis and secretion remain obscure.

Hypoxia has been identified as 1 such factor that regulates EV biogenesis, promoting secretion in renal proximal tubular cells of the kidney via HIF-1α (71). In ovarian cancer, hypoxia increases exosome secretion in a STAT3-dependent manner by inducing a more secretory lysosomal phenotype, upregulating Rab27a and downregulating Rab7, ultimately leading to increased docking of multivesicular bodies at the plasma membrane. Knockdown of STAT3 reversed these effects (72). Hypoxia-induced EV secretion was also observed in breast cancer lines MCF7, MDA-MB 231, MDA-MB 435, and 4T1, although different Rab proteins were implicated (73). Various oncogenic signals have also been implicated in EV biogenesis, such as Src and Ras (74–76). Metabolic factors have also been implicated in EV regulation, such as the sirtuins (77, 78). Interestingly, treatment of the macrophage-like THP1 cell line with an inhibitor of HMGCR (Simvastatin) resulted in decreased EV secretion (44), suggesting that cholesterol may regulate EV biogenesis. Given that many of the pathogenic effects of cholesterol are in part mediated by the metabolite 27HC, we evaluated the effect of this oxysterol on EV secretion and related downstream pathophysiology.

27HC resulted in increased EV secretion in several different cell types, including primary PMNs, RAW 264.7 monocytic cells, and 4T1 mammary cancer cells. The secreted EVs expressed several tetraspanins that are characteristic of EVs, CD81, CD63, and CD9. 27HC did not alter the parental cell mRNA expression of these tetraspanins in a meaningful way, indicating that it was indeed regulating assembly and secretion, not just expression of EV components. Interestingly, the size distribution of secreted EVs was significantly different from PMNs treated with DMSO compared with 27HC, with 27HC resulting in increased mean and median sizes. The increase in size distribution was consistent across all 3 cell types investigated. The biological consequences, if any, of this altered size distribution remain to be determined. Regardless, however, this would suggest that in addition to increasing EV secretion, 27HC also results in altered composition of EVs, and that the effects of 27HC are mediated through a pathway that is ubiquitously active across cell types.

Importantly, we demonstrate that exposing PMNs to a 27HC-rich environment produces EVs which promoted the growth and metastasis of mammary cancer in mice. Syngeneic models with complete immune systems were used, since we have previously shown that many of the effects of 27HC are mediated through the immune system (5, 8). Two different syngeneic models of murine mammary cancer exhibited increased tumor growth when mice were treated with EVs secreted by PMNs treated with 27HC. In these mice, there was also evidence of increased spontaneous metastasis to the lungs, as quantified by increased expression of markers specific to the cancer cells (4T1: luciferase, and Met1: PyMT or iRFP), as well as the proliferation marker Ki-67. In order to better assess colonization of distal tissues and subsequent outgrowth in a more direct manner, we used the well characterized intravenous graft model, and found that 27HC-EVs from PMNs also promoted metastatic burden in this model. Intriguingly, in vivo imaging suggests that 27HC-EVs promoted metastatic colonization of the bones in addition to the more common lung metastatic site (Fig. 4D). This is of interest as it has previously been documented that EVs play an active role in establishing the pre-metastatic niche, and can dictate the tissue preference of circulating cancer cells (79). The notion that 27HC-EVs modulate the microenvironment is reinforced by our findings that 27HC-EVs from PMNs fail to alter the cellular proliferation of 3 different murine mammary cancer lines, and failed to regulate the expression of several genes associated with various aspects of cancer biology (Fig. 5).

Interestingly, EVs from PMNs were taken up by several cell types including mammary cancer cell lines, macrophages, and PMNs, while T cells did not. The lack of evidence for T cell uptake of EVs lends support to previous findings that EVs isolated from macrophages do not impact T cell expansion (8). Quite intriguingly, several cell types, including macrophages showed increased uptake of EVs from 27HC-treated PMNs compared with vehicle-treated PMNs (Fig. 6). Although not tested yet, it is possible that this differential uptake may be 1 mechanism by which 27HC-EVs promote breast tumor progression. Future work will be required to delineate what if any effects EVs from PMNs have on those cells that do interact with them.

In order to start determining the mechanisms by which 27HC-EVs promote breast cancer progression, tumors from mice treated with EVs were sent for nonbiased RNA-seq (tumors from the mice in Fig. 5B). However, no significant changes in gene expression were observed between these tumors (Fig. 7), despite a sizable replicate number (N = 9 for DMSO-EV treatment, N = 10 for 27HC-EV treatment). Thus, treatment with 27HC-EVs did not impact the transcriptome of these tumors, at least at the late stages that they were sampled. This lack of regulation, despite differences in tumor growth, could be interpreted in several different ways. First, since bulk RNA-seq was utilized, it is possible that the impact of 27HC-EVs was restricted to a small subset of cells within the microenvironment, and their transcriptomic changes were masked by the rest of the tumor mass. Secondly, it is possible that the effects of 27HC-EVs were restricted to early tumor growth or metastatic colonization. This is supported by our data indicating that pretreatment of mice with 27HC-EVs increases the metastatic colonization (Fig. 4C and 4D), and previous work indicating that myeloid cells are required for the metastatic properties of 27HC (5). Finally, it is possible that 27HC-EVs do not result in changes in gene regulation per se, but instead alter other aspects of tumor biology, such as resource availability, altered metabolic signaling, or signal transduction cascades that do not result in altered gene expression. In terms of resource delivery, EVs are known to carry various lipids and other metabolites, and might even be a delivery vessel for 27HC itself (45).

Collectively, our results indicate that a cholesterol metabolite promotes the secretion of EVs across several different cell types, and that 27HC-induced EVs from PMNs increase mammary tumor growth and metastasis. Given the prevalence of hypercholesterolemia among women, and the correlation between circulating cholesterol and 27HC (80, 81), it will be of importance to determine the mechanisms by which 27HC stimulates EV secretion, and how 27HC-induced EVs promote breast cancer progression. This knowledge would fuel the therapeutic exploitation of EV regulation.

Glossary

Abbreviations

27HC

27-hydroxycholesterol

BMDM

bone marrow–derived macrophage

DMSO

dimethyl sulfoxide

ER

estrogen receptor

EV

extracellular vesicle

FBS

fetal bovine serum

LXR

liver X receptor

MDS

multidimensional scaling

NTA

nanoparticle tracking analysis

PBS

phosphate-buffered saline

PMN

polymorphonuclear neutrophil

qPCR

quantitative polymerase chain reaction

SEM

scanning electron microscopy

SLAM

simultaneous label-free autofluorescence multiharmonic

TEM

transmission electron microscopy

Data Availability

Some data generated or analyzed during this study are included in this published article or in the data repositories listed in References. Some datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.