Knocking down AR promotes osteoblasts to recruit prostate cancer cells by altering exosomal circ-DHPS/miR-214-3p/CCL5 pathway

Abstract

Tumor-derived exosomes have been shown to play a key role in organ-specific metastasis, and the androgen receptor regulates prostate cancer (PCa) progression. It is unclear whether the androgen receptor regulates the recruitment of prostate cancer cells to the bone microenvironment, even bone metastases, through exosomes. Here, we found that exosomes isolated from PCa cells after knocking down androgen receptor (AR) or enzalutamide treatment can facilitate the migration of prostate cancer cells to osteoblasts. In addition, AR silencing or treatment with the AR antagonist enzalutamide may increase the expression of circular RNA-deoxyhypusine synthase (circ-DHPS) in PCa cells, which can be transported to osteoblasts by exosomes. Circ-DHPS acts as a competitive endogenous RNA (ceRNA) against endogenous miR-214-3p to promote C-C chemokine ligand 5 (CCL5) levels in osteoblasts. Increasing the level of CCL5 in osteoblasts could recruit more PCa cells into the bone microenvironment. Thus, blocking the circ-DHPS/miR-214-3p/CCL5 signal may decrease exosome-mediated migration of prostate cancer cells to osteoblasts.

INTRODUCTION

Prostate cancer (PCa) is the second leading cause of cancer-related death among men in Western countries and the sixth most common cause of cancer-related death worldwide. Besides, the incidence of PCa is increasing in developing countries.1,2 Bone metastasis is a common and serious situation that patients with castration-resistant prostate cancer (CRPC) face. Men with metastatic PCa suffer from an incurable disease with poor survival; the 5-year survival rate is only approximately 30%.3 The consequences of bone metastasis are often devastating.4 The development of PCa is known to be strongly associated with the androgen receptor (AR).5 AR can be a suppressor in some cases then to suppress metastasis in PCa.6 However, the role of AR in bone metastasis of PCa is rarely reported.

Exosomes are small membrane vesicles capable of playing important roles between different cells;7 many types of cells, such as T-cells,8 epithelial cells,9 and tumor cells,10 can secrete exosomes. It has been reported that exosomes are involved not only in cancer formation and progression but also in drug resistance by transferring proteins and nucleic acids.11,12 Growing amounts of evidence indicate that exosomes function a lot in the tumor microenvironment and the content or function of exosomes is also altered when tumors become more aggressive. In addition to playing a vital role in the tumor–tumor communication, exosomes can also regulate tumor–stromal communication to affect tumor development and metastasis.13,14 Moreover, tumor exosomal integrins played an important role in tumor-specific organ metastasis.15

Recently, circular RNAs (circRNAs) have been defined as a new group of endogenous RNAs. CircRNAs are widely expressed in animal cells, including human and mouse cells, and can regulate downstream gene activity at the posttranscriptional level.16 A large body of evidence suggests that circRNAs can bind to microRNAs (miRNAs) to affect the expression of downstream gene.17 The interaction between circRNAs and miRNAs has been widely observed to play a key role in many different cancers and has led to the elucidation of pathways that provide predictive biomarkers of cancers.18 Moreover, overexpression or underexpression of circRNAs has been reported to affect tumor development,19 but their role in PCa bone metastasis remains unclear. In this work, we explore the role of circular RNA-deoxyhypusine synthase (circ-DHPS; circBase ID: hsa_circ_0000893; chr19: 12790270–12791139) in PCa.

In this study, we found that exosomes isolated from PCa cells after knocking down AR or enzalutamide treatment can promote osteoblasts to recruit prostate cancer cells into the bone microenvironment. Furthermore, a mechanistic study showed that AR inhibition can activate the PCa cell exosomal circ-DHPS/miR-214-3p/CCL5 signal. Thus, targeting the newly identified AR/circ-DHPS/miR-214-3p/C-C motif chemokine ligand 5 (CCL5) signal may decrease exosome-mediated migration from PCa cells to osteoblasts.

PATIENTS AND METHODS

Patient blood samples and ethical considerations

Blood samples and clinical information were acquired from patients with PCa at The First Affiliated Hospital of Xi’an Jiaotong University (Xi’an, China) between August 2017 and August 2019. All participants gave written consent for the use of their clinical information and blood samples for academic research. This research was approved by the Ethics Committee of Xi’an Jiaotong University (Approval No. 2022-675). The patients’ characteristics are presented in Supplementary Table 1.

Supplementary Table 1.

Patient characteristics

Characteristics	Bone metastases			P
	No (n=18)	Yes (n=13)	Total (n=31)
Age (year), median (range)	73.5 (56.0–89.0)	80.0 (58.0–86.0)	77.0 (56.0–89.0)	0.070
Staging, n (%)
T2	7 (22.6)	5 (16.1)	12 (38.7)
T3	7 (22.6)	2 (6.5)	9 (29.0)
T4	4 (12.9)	6 (19.4)	10 (32.3)	0.249
N0	9 (29.0)	4 (12.9)	13 (41.9)
N1	9 (29.0)	9 (29.0)	18 (58.1)	0.738
Gleason score, median (range)	7 (6–10)	8 (6–8)	7 (6–10)	0.793
PSA level at diagnosis (ng ml−1), median (range)	38.2 (7.3–78.2)	102.2 (38.1–300.0)	58.5 (7.3–300.0)	<0.001
During ADT (month), median (range)	26.5 (8.0–70.0)	32.0 (8.0–70.0)	27.0 (8.0–70.0)	0.756

Patient information such as age, PSA levels, Gleason scores, and staging. ADT: androgen deprivation therapy; PSA: prostate-specific antigen

Exosome isolation, characterization, and analyses

Exosomes were isolated using the Total Exosome Isolation Reagent (Thermo Fisher Scientific, Rochester, NY, USA) or an ultracentrifuge. In the ultracentrifuge process, the media collected from PCa cells were centrifuged (SL4R Plus, Thermo Fisher Scientific) at 300g and 2000g for 10 min to remove contaminating cells and dead cells, respectively, and finally ultracentrifuged (Optima XPN-80 ultracentrifuge, Beckman, Brea, CA, USA) at 100 000g for 70 min to isolate exosomes. After washing with phosphate-buffered saline (PBS) and a second ultracentrifugation, the collected pellets were dissolved in the indicated buffer (PBS, radioimmunoprecipitation assay [RIPA] buffer, or RNA lysis buffer). The exosome preparations were verified by transmission electron microscopy (Hitachi 7650 III [2008] analytical transmission electron microscope, Hitachi, Tokyo, Japan). Exosome size and particle number were analyzed using the LM10 nanoparticle characterization system (NanoSight LM10, Malvern Panalytical, Malvern, UK). The protein concentration of exosome was calculated by the bicinchoninic acid (BCA) kit (Thermo Fisher Scientific), and 30 μg ml⁻¹ of the exosome suspension was used for all experiments.

Reagents and materials

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), androgen receptor (AR) antibodies, and normal rabbit IgG were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). CCL5-neutralizing antibody was purchased from Genetech (San Francisco, CA, USA). Anti-mouse/rabbit antibody for western blot was obtained from Invitrogen (Grand Island, NY, USA).

Cell culture

The PCa C4-2, PC-3, and CWR22Rv1 cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The 2T3 osteoblastic cells were provided by Dr. Chawn-Shang Chang (University of Rochester Medical Center, Rochester, NY, USA). Cells were cultured in RPMI 1640 or Dulbecco’s Modified Eagle’s Medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS). All cell lines were grown at 37ºC in a humidified incubator with 5% (v/v) CO₂.

Lentivirus and packaging

The plasmids (pLKO.1, pLVTHM, pLKO.1-shAR, pWPI, pWPI-oeAR, pMD2G, psPAX2, pLVTHM-sh-circ-DHPS, and pLVTHM-miR-214-3p) were co-transfected into 293T cells to obtain the lentivirus according to the standard calcium chloride transfection protocol. The lentivirus supernatant was collected for concentration with density gradient centrifugation (Sorvall™ WX+, Thermo Fisher Scientific) and then aliquoted at −80°C for future use.

RNA extraction and quantitative real-time PCR (qRT-PCR) analysis

Total RNA was extracted by TRIzol reagent (Invitrogen) according to the protocol. Complementary deoxyribonucleic acid (cDNA) was obtained using Superscript III transcriptase (Invitrogen). The messenger ribonucleic acid (mRNA) expression level of the gene of interest was determined by qRT-PCR using a Bio-Rad CFX96 system with SYBR Green. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control.

Protein extraction and western blot assay

Cells were collected and lysed in ice-cold RIPA buffer. Equal amounts of proteins (30 µg) were separated by a 4%–15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel. After being transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, USA), it was blocked for about 1 h, and then, the membranes were incubated with primary antibodies overnight in a cold room. Then, the blots were incubated with secondary antibodies for about 1 h and then visualized with the ECL system (Thermo Fisher Scientific).

Luciferase assay

The human promoter sequence of miR-214-3p was obtained from Ensembl and then inserted into the pGL3-basic vector (Promega, Madison, WI, USA). CCL5 3’ UTRs involving wild-type or mutant miRNA-responsive elements were cloned into the psiCHECK2 vector construct (Promega). Cells were seeded in a 24-well plate, and plasmids were co-transfected into cells according to the protocol. Luciferase activity was measured by the Dual-Luciferase Assay Kit (Promega) based on the manufacturer’s instructions.

RNA pull-down assay

The cells were collected, and the pellets were resuspended in 1 ml of cell lysis buffer for 72 h. The quantified cells were added with 1.5 µl RNase inhibitor, 500 pmol l⁻¹ antisense oligos, and 10 µl of streptavidin agarose beads to spin in the cold room overnight. Total RNA was extracted by TRIzol (Invitrogen) and subjected to qRT-PCR analysis.

Chromatin immunoprecipitation assay (ChIP)

The assay was conducted according to our previous protocol. Briefly, cell lysates were cross-linked and then ultrasonicated for chromatin fragments, which were then incubated with anti-AR antibody (2.0 µg) overnight in cold room. IgG was used as a negative control. Agarose gel electrophoresis was used to test the PCR products.

Statistical analyses

SPSS version 19.0 (SPSS Inc., Chicago, IL, USA) was used for all statistical analysis data. All experimental data are expressed as the mean ± standard deviation (s.d.). Differences in mean values of two groups were performed using the two-tailed Student’s t-test, and means of more than two groups were determined by one-way analysis of variance (ANOVA). P < 0.05 was defined as statistically significant.

RESULTS

Exosomes from PCa cells enable osteoblasts to attract more PCa cells

To establish a stable exosome co-culture system, we first isolated exosomes from PCa cells (Figure 1a). Next, the shape and particle size distribution of the isolated exosomes were detected using a transmission electron microscope (Figure 1b) and a nanoparticle characterization system15 (Figure 1c). The expression levels of exosomal markers, tumor susceptibility 101 (TSG101), CD9, and CD63, were validated by western blot (Figure 1d). All data above confirmed the characteristics of exosomes and are consistent with previous studies.20

Figure 1.

Exosomes from PCa cells can enable osteoblasts to attract more PCa cells. (a) The schematic shows the process of isolating exosomes from PCa cells. (b) An electron microscope was used to detect the exosomes. (c) The exosome size and particle distribution were analyzed. (d) Expression of TSG101, CD9, and CD63 was analyzed by western blot. (e) The schematic shows that a Transwell chamber assay was used to identify the ability of PCa cells to migrate onto exosome-treated osteoblasts. Transwell chamber assay was performed to identify the migration capacity of (f) C4-2 cells and (g) PC-3 cells after addition of conditional medium from osteoblast 2T3 with/without treated exosomes. Transwell chamber assay to identify the migration capacity of (h) C4-2 cells and (i) PC-3 cells using GW. Data were presented as the mean±standard deviation of three independent experiments. Statistical analysis was performed using the two-tailed Student’s t-test. ^*P < 0.05. PCa: prostate cancer; CM: conditional medium; exo: exosomes; GW: GW4869 (exosome generation inhibitor); FBS: fetal bovine serum; a.u.: arbitrary unit.

Exosomes have been shown to be a very important regulator of cell–cell communication and a promising clinical biomarker in PCa and tumor metastasis.15,21 Thus, we want to determine whether exosomes are related to the process of PCa cell migration to bone. We used a Transwell chamber assay to identify the migration capacity of PCa cells after co-culture of PCa exosomes with osteoblasts in vitro. We began treating osteoblast 2T3 cells with exosomes from PCa cells and then used conditional medium (CM) from osteoblast 2T3 cells to attract PCa cells (Figure 1e). Results showed that exosomes from C4-2 and PC-3 cells may allow 2T3 cells to attract more C4-2 cells than that from untreated 2T3 cells (Figure 1f). After replacing the upper chamber C4-2 cells with PC-3 cells, a similar result was obtained (Figure 1g). Importantly, through the blocking assay, we found that the ability of osteoblast 2T3 cells to recruit C4-2 cells (Figure 1h) and PC-3 cells (Figure 1i) could be abolished by adding doses of 5 μmol l⁻¹ and 10 μmol l⁻¹ of exosome generation inhibitor, GW4869. Taken together, the results showed that exosomes from PCa cells may allow osteoblasts to attract more PCa cells (Figure 1).

The AR plays a negative role in the release of exosomes to facilitate the attraction of PCa cells by osteoblasts

Previous studies have shown that AR signaling plays an important role in tumor formation and PCa development.22 Some studies also indicate that downregulation of AR with siRNA may enhance PCa cell invasion.23,24 Thus, we asked whether AR was related to exosome-mediated migration of PCa cells to bone. First, we verified the mRNA and protein level of AR in PC-3 and C4-2 cells after AR overexpression and knockdown (Supplementary Figure 1 ^{(60.6KB, tif)} ). Our results showed that pretreatment of 2T3 cells with exosomes from C4-2 sh-AR cells attracted more C4-2 cells than with C4-2 control exosomes (Figure 2a). When substituting for PC-3 cells, we obtained the same result, which means that the exosomes of the C4-2 sh-AR cells caused the osteoblast 2T3 cells to attract more PC-3 cells (Figure 2b). In addition, we also determined that exosomes from PC-3 overexpressing AR cells led osteoblastic 2T3 cells to attract less C4-2 cells (Figure 2c) and PC-3 cells (Figure 2d) than 2T3 cells treated with PC-3 control exosomes. Notably, previous report confirmed that androgen deprivation therapy (ADT) with enzalutamide significantly increased PCa metastasis.25 Therefore, we assessed whether ADT could increase tumor migration through tumor-released exosomes. We treated PCa cells with enzalutamide for 2 days and then collected exosomes. Exosomes from enzalutamide-treated C4-2 cells were added to osteoblast 2T3 cells, and the results showed that the exosomes from enzalutamide-treated cells significantly increased the ability of osteoblast 2T3 cells to attract C4-2 cells (Figure 2e). Results were similar when used with PC-3 cells (Figure 2f). The results illustrate that the AR plays a negative role in facilitating the osteoblastic attraction of PCa cells (Figure 2).

Figure 2.

AR plays a negative role in exosomes released to facilitate osteoblastic attraction of PCa cells. Transwell chamber assay to identify the ability of (a) C4-2 cells and (b) PC-3 cells to migrate when treating 2T3 cells with exosomes derived from C4-2 control/sh-AR. Transwell chamber assay to identify the ability of (c) C4-2 cells and (d) PC-3 cells to migrate when treating 2T3 cells with exosomes derived from PC-3 control/oe-AR. Transwell chamber assay to identify the ability of (e) C4-2 cells and (f) PC-3 cells to migrate when treating 2T3 cells with C4-2 control or pretreated 5 μmol l⁻¹ and 10 μmol l⁻¹ enzalutamide-derived exosomes. Data were presented as the mean±standard deviation of three independent experiments. Statistical analysis was performed using the two-tailed Student’s t-test. ^*P < 0.05. AR: androgen receptor; exo: exosomes; Enz: enzalutamide; PCa: prostate cancer; sh-AR: knocking down AR expression; oe-AR: overexpressing AR expression.

CCL5 facilitates the attraction of osteoblasts to PCa cells

We examined how exosomes from AR knockdown or enzalutamide-treated cells could increase osteoblast attraction to PCa cells. As several studies have shown that bone can release some chemokines or growth factors which, in turn, can attract PCa cells to bone tissue,26 we used a qRT-PCR array to screen these chemokine candidates. The results show that CCL5 was significantly increased in 2T3 cells after treatment with exosomes from PCa C4-2, PC-3, and CW22Rv1 cells (Figure 3a). Importantly, we also determined that CCL5 was significantly elevated in 2T3 cells when treated with exosomes from C4-2 sh-AR cells compared to control cells, and the opposite results were obtained in 2T3 cells treated with exosomes from PC-3 cells overexpressing AR compared to 2T3 cells treated with exosomes from control cells (Figure 3b). A blocking assay using neutralizing antibody showed that blocking CCL5 in 2T3 cells could partially suppress PCa cell migration into bone microenvironments (Figure 3c and 3d). Furthermore, exosomes from C4-2 sh-AR cells caused osteoblastic 2T3 cells to attract more PCa cells, which could be abolished by a CCL5-neutralizing antibody (Figure 3e and 3f). Importantly, after analysis in the GEO database, we found that patients with metastatic PCa have very high expression of CCL5 (Figure 3g). We found that upregulation of CCL5 could facilitate osteoblasts to attract PCa cells.

Figure 3.

CCL5 facilitates the attraction of osteoblasts to PCa cells. (a) qRT-PCR cytokine array in 2T3 cells treated with the indicated exosomes. (b) qRT-PCR cytokine array in 2T3 cells treated with exosomes from C4-2 control/sh-AR cells and exosomes from PC-3 control/oe-AR cells. (c) Transwell chamber assay to identify the ability of C4-2 cells to migrate using CCL5-neutralizing antibody when treating 2T3 cells with/without exosomes derived from PCa cells. (d) Quantification of the ability of C4-2 cells to migrate from c. (e) Transwell chamber assay to identify the ability of C4-2 cells to migrate using CCL5-neutralizing antibody when treating 2T3 cells with exosome derived from C4-2 control/sh-AR. (f) Quantification of the ability of C4-2 cells to migrate from e. (g) CCL5 expression was analyzed in primary versus metastatic PCa tumors using the GEO database GSE6919. Data were presented as the mean± standard deviation of three independent experiments. Statistical analysis was performed using the two-tailed Student’s t-test. ^*P < 0.05. CCL5: C-C motif chemokine ligand 5; CCL5 Ab: CCL5-neutralizing antibody; exo: exosomes; ctrl: control; AR: androgen receptor; PCa: prostate cancer; sh-AR: knocking down AR expression; oe-AR: overexpressing AR expression; qRT-PCR: quantitative real-time polymerase chain reaction; IL: interleukin; CXCL: C-X-C motif chemokine ligand; IGF: insulin like growth factor; FGF2: fibroblast growth factor 2; TGF-β: transforming growth factor beta.

Figure 4.

The exosome-derived circ-DHPS facilitates CCL5 expression. NanoSight was used to detect the exosome number of (a) C4-2 control/sh-AR and (b) PC-3 control/oe-AR cells. (c) qRT-PCR showing expression of circRNA candidates in the exosomes of C4-2 control/sh-AR and PC-3 control/oe-AR cells. (d) qRT-PCR showing expression of circ-DHPS in 2T3 cells treated with the indicated exosomes. (e) qRT-PCR showing expression of circ-DHPS in C4-2 control/sh-AR and PC-3 control/oe-AR cells. (f) qRT-PCR showing expression of circ-DHPS in C4-2 cells and exosomes derived from C4-2 cell after treatment with 10 μmol l⁻¹ enzalutamide. (g) qRT-PCR showing CCL5 expression after knocking down circ-DHPS in 2T3 cells. (h) qRT-PCR showing expression of circ-DHPS and CCL5 in 2T3 cells treated with control or exosomes from C4-2 with/without GW4869 treatment. (i) Blocking assay knocking down circ-DHPS in 2T3 cells revealed expression of circ-DHPS and CCL5 after treatment with exosomes derived from C4-2 control/sh-AR. Blocking assay knocking down circ-DHPS in 2T3 cells to identify the ability of (j) C4-2 cells and (k) PC-3 cells to migrate when treating 2T3 cells with exosome derived from C4-2 control/sh-AR. (l) qRT-PCR analysis of circ-DHPS in plasma from patients with primary versus bone-metastatic PCa. Data were presented as the mean± standard deviation of three independent experiments. Statistical analysis was performed using the two-tailed Student’s t-test. ^*P < 0.05. DHPS: deoxyhypusine synthase; GW: GW4869 (exosome generation inhibitor); exo: exosomes; NS: nonsignificance. CCL5: C-C motif chemokine ligand 5; sh-AR: knocking down AR expression; oe-AR: overexpressing AR expression; qRT-PCR: quantitative real-time polymerase chain reaction; circ-DHPS:circular RNA deoxyhypusine synthase.

Exosome-derived circ-DHPS facilitates CCL5 expression

To further dissect the mechanisms by which exosomes from PCa cells with altered AR increase CCL5 expression in 2T3 cells, we first determine whether AR affects the number of exosomes released. However, no significant difference was observed in exosome secretion and quantity after altering the AR (Figure 4a and 4b). Thus, we focused on the components inside the exosomes. Previously, some studies have shown that circRNAs play crucial roles in gene expression and have been enriched in exosomes;27,28 thus, we first assessed whether circRNAs changed after AR alteration. Based on analyses of multiple databases (circBase, starBase v2.0, and TargetScan) and later on our validation using the qRT-PCR assay, circ-DHPS was identified as a potential candidate, because it was downregulated in exosomes from AR-overexpressing PC-3 cells and upregulated in exosomes from AR knockdown C4-2 cells (Figure 4c).

Next, we observed that circ-DHPS levels are increased in 2T3 cells treated with AR knockdown exosomes compared to control exosomes. The same trend was seen in 2T3 cells treated with AR-overexpressing exosomes (Figure 4d). Moreover, in exosome-producing PCa cells, circ-DHPS expression was increased in AR knockdown C4-2 cells and decreased in AR-overexpressing PC-3 cells (Figure 4e). Similarly, circ-DHPS increased in C4-2 cells and exosomes after treatment with enzalutamide (Figure 4f). Then, after knocking down circ-DHPS by siRNA, CCL5 expression was decreased in 2T3 cells (Figure 4g). Using GW4869 to block exosome generation in C4-2 cells, we determined that CCL5 expression was consistent with circ-DHPS expression in 2T3 cells (Figure 4h). Importantly, when we reduced circ-DHPS expression in 2T3 cells, we increased CCL5 expression (Figure 4i) and the recruitment abilities observed after treatment with sh-AR exosomes were abolished (Figure 4j and 4k). From clinical data, we also found that circ-DHPS was significantly increased in patients with bone-metastatic PCa (Figure 4l). The above results lead to the conclusion that knocking down AR can increase circ-DHPS/CCL5 expression and can lead osteoblasts to attract more PCa cells to bone (Figure 3 and 4).

Mechanism by which AR manipulates circ-DHPS/CCL5 expression

Most circRNAs are directly or inversely proportional to host gene expression due to host gene transcription and splicing;29 therefore, we identified the DHPS gene as the host gene to further dissect the mechanism by which AR regulates circ-DHPS expression. The Ensembl and PROMO 3.0 website were conducted to search for possible androgen response elements (AREs) in the 5 Kb 5’ promoter region of the DHPS gene, and ultimately, 6 potential AREs were found (Figure 5a). We used C4-2 cells to perform ChIP, the results of which indicate that ARE3 and ARE4 are two positive binding sites (Figure 5b). Luciferase reporter assay results showed that its activity in C4-2 cells increased after knocking down the AR (Figure 5c). Importantly, there were no significant differences in luciferase reporter activity after ARE3 mutation (Figure 5d).

Figure 5.

Mechanism by which AR manipulates circ-DHPS/CCL5 expression. (a) The predicted androgen response element (ARE) in the 5 Kb region of the DHPS promoter. (b) A ChIP assay revealed ARE-binding sites in C4-2 cells. (c) Luciferase activity of the circ-DHPS promoter was evaluated by luciferase assay in C4-2 cells after knocking down AR. (d) The luciferase activity of the circ-DHPS mutant promoter was evaluated by luciferase assay in C4-2 cells after knocking down AR. (e) RNA pull-down assay to show that miR-214-3p interacts with circ-DHPS. (f) qRT-PCR assay was performed to show miR-214-3p and CCL5 expression. (g) The potential plot of predicted CCL5 3′ UTR sequence alignment with wild-type (WT) versus mutant (mut) miR-214-3p targeting site. (h) CCL5 3’ UTR luciferase activities were detected by transfection with PGL3-CCL5 WT/mutant 3’ UTR plus vector control/miR-214-5p overexpression plasmids into 2T3 cells. (i) Blocking assay by overexpressing miR-214-3p in 2T3 cells showing CCL5 expression after treatment with exosomes derived from C4-2 control/sh-AR. Blocking assay by overexpressing miR-214-3p in 2T3 cells to identify the ability of the (j) C4-2 and (k) PC-3 cells to migrate when treating 2T3 cells with exosomes derived from C4-2 control/sh-AR. (l) miR-214-3p expression was analyzed in primary versus metastatic PCa tumors using the GEO database GSE31568. Data were presented as the mean ± standard deviation of three independent experiments. Statistical analysis was performed using the two-tailed Student’s t-test. ^*P < 0.05. ARE: androgen response element; 3’ UTR: 3’ untranslated region; wt: wild type; mut: mutant; qRT-PCR: quantitative real-time polymerase chain reaction; circ-DHPS: circular RNA deoxyhypusine synthase; CCL5: C-C motif chemokine ligand 5; AR: androgen receptor; sh-AR: knocking down AR expression; oe-AR: overexpressing AR expression; exo: exosome; NS: nonsignificance.

We focused on miRNAs since circRNAs may compete with miRNAs to regulate their target genes,17,30 to dissect the mechanism by which AR-regulated circ-DHPS can increase PCa migration. As a previous study proved that CCL5 is a direct target of miR-214-3p,31 we performed an RNA pull-down assay to show that circ-DHPS can interact with miR-214-3p (Figure 5e). We then overexpressed miR-214-3p in 2T3 cells, and the results showed that miR-214-3p decreased CCL5 expression (Figure 5f). We then applied the reporter assay, and the results confirmed that miR-214-3p could not decrease luciferase reporter activity after mutation of binding sites in the 3’ UTR of CCL5 (Figure 5g and 5h), which means that miR-214-3p can directly regulate CCL5. Importantly, the results of a rescue assay further revealed that exosomes from C4-2 sh-AR cells induced CCL5 expression (Figure 5i) and that recruitment abilities were abolished (Figure 5j and 5k) after we overexpressed miR-214-3p in 2T3 cells. Clinical data also showed significantly decreased miR-214-3p in patients with metastatic PCa (Figure 5l), which was consistent with the findings of this study.

DISCUSSION

Previously, AR was known to play a vital role in the progression and metastasis of many hormone-related cancers,32,33 and in PCa, AR has been shown to be a metastasis suppressor.34 In the present work, we targeted the AR with AR-shRNA or using anti-androgen enzalutamide, which led to increased recruitment of PCa cells to osteoblasts by altering circ-DHPS expression through modulation of endogenous AR transcription. circ-DHPS can then sponge and alter miR-214-3p to affect CCL5 signals. A previous study reported that miR-214-3p was involved in the PI3K-Akt pathway in the treatment of castration-resistant prostate cancer.35 Likewise, in our study, we elucidated that miR-214-3p played an important role in the progression of PCa cells to migrate to bone, and clinical data showed that miR-214-3p expression was reduced in patients with metastatic PCa compared to primary patients.

Here, we found that circ-DHPS was elevated in patients with bone metastasis from prostate cancer. Due to their high level of conserved sequences and stability, and tissue- and stage-specific expression in mammalian cells, circRNAs have the potential to become ideal biomarkers for tumor diagnosis.16 Growing evidence show that circRNAs are closely associated with cancers. Compared to adjacent nontumor tissues, gastric cancer tissues have a much lower expression of hsa_circ_002059, which indicates that hsa_circ_002059 may be a promising stable biomarker for the diagnosis of gastric cancer.36 In PCa, circular RNA ATP-binding cassette subfamily C member 4 (circABCC4) expression is associated with poor survival and circABCC4 enhances PCa progression through the miR‐1182‐forkhead box P4 signal.37 In our study, circ-DHPS acts as a competing endogenous RNA (ceRNA) for miR-214-3p sponge, playing a vital role of PCa cells in the process of migration to the bone microenvironment. Thus far, it is unclear whether other characteristics of cancer cells, such as proliferation and invasion, are also regulated by this circ-DHPS. In addition, it is also unclear whether this regulation of circ-DHPS and miR-214-3p is functionally connected to other mechanisms that led to ADT-enhanced PCa metastasis, such as altered AR-coactivator nuclear receptor coactivator 2 (NCOA2),34 transforming growth factor beta 1 (TGF-β1) signaling, and glutathione reductase (GR) signaling. Moreover, our results revealed that exosomes, as a new role of intercellular communication, can also transport circRNA, which is consistent with previous findings.27

Exosomes are taking on new roles in cell–cell communication, as they can deliver many bioactive molecules, including proteins, DNA, mRNA, and noncoding RNAs from one cell to another, which can exchange signals between donor cells and recipient cells.38 Furthermore, the results from previous and current studies further identified that tumor-derived exosomes were involved in several important steps in the spread of metastasis from a primary tumor, such as increased migration and invasion of cancer cell and establishment of a premetastatic niche (PMN).39,40 These exosomes can direct tumor-neighboring stromal cells to specific target tissues or organs to prepare a suitable soil for tumor implantation.41 In metastatic melanoma, it has been reported that exosomes can transfer the MET proto-oncogene (MET) to bone marrow receptor cells, making them prone to a pro-vasculogenic phenotype, which will eventually increase tumor metastasis.42 More importantly, evidence has shown that PCa-derived exosomes, through phospholipase D (PLD), could interact with the bone microenvironment, especially osteoblasts, during the metastatic process.43 Here, we used tumor-derived exosomes to treat 2T3 osteoblast cells and found that these osteoblasts could recruit more PCa cells, and this effect could be abolished by the exosome generation inhibitor GW4869. These results confirmed that exosomes contribute to PCa bone metastasis through intercellular communication. Thus, we speculate that PCa cells transfer substances, such as some of the previously mentioned bioactive molecules, to osteoblasts through exosome secretion, and alter their tumor microenvironment.

Previous studies have shown that the osteoblastic microenvironment acts as a PMN, attracting bone-metastasizing tumor cells.44,45 In our research, results from preclinical studies using multiple cell lines revealed that exosomes from PCa cells promoted osteoblasts to express CCL5 through circ-DHPS transport. As an important part of the tumor microenvironment, CCL5 has been shown, through different methods, to exert tumor-promoting roles, especially during metastasis. Several studies have shown that CCL5/C-C motif chemokine receptor 5 (CCR5) interactions can promote tumor development in several ways, such as stimulating angiogenesis, inducing stromal cell recruitment, and participating in immune escape mechanisms.46 More importantly, previous studies in breast cancer have shown that lymphatic endothelial cells within premetastatic organs are upregulated by tumor secretory factors and begin to express CCL5 to facilitate recruitment, extravasation, and colonization of tumor cells. These observations indicate that CCL5 contributes to the formation of premetastasis niche and is highly related to tumor metastasis,47 which was consistent with our findings. Furthermore, we also found that CCL5 was highly expressed in the samples from patients with metastatic PCa.

CONCLUSION

We determined that AR could regulate circ-DHPS expression by binding to the host gene promoter, and the release of exosomes containing circ-DHPS into bone cells allows circ-DHPS to act as a sponge of miR-214-3p, upregulates CCL5 expression, and stimulates more PCa cells to migrate to bone (Figure 6). Thus, these data indicated that exosomes and circ-DHPS may be potential prognostic biomarkers to predict metastases and targets for individualized drug therapy, which may prevent PCa cells from migrating into the bone microenvironment and improve the prognosis of cancer patients.

Figure 6.

Mechanism underlying how the AR regulates PCa cells that migrate to osteoblasts. AR decreases circ-DHPS expression by binding to the promoter of its host gene. Without AR downregulation, exosomes containing circ-DHPS are released towards bone cells, where circ-DHPS acts as a miR-214-3p sponge and upregulates CCL5 expression, thereby stimulating more PCa cells to migrate into the bone microenvironment. AR: androgen receptor; CCL5: C-C motif chemokine ligand 5; DHPS: deoxyhypusine synthase; PCa: prostate cancer; circ-DHPS: circular RNAdeoxyhypusine synthase.

AUTHOR CONTRIBUTIONS

ZY, JQC, and HJX conceived and designed the study. TJL collected individual and clinical data. YLC, ZKM, and ZXW analyzed the data. ZY, JQC, HJX, YZF, KW, and SX performed experiments. ZY, JQC, XYW, and LL wrote and revised the manuscript. ZY and JQC acquired the financing. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declare no competing interests.

Supplementary Figure 1

RNA and protein level of PCa cells after overexpressing and AR knockdown. (a) RNA level of PC-3 and C4-2 cells after overexpression and knocking down of AR. (b) Protein level of PC-3 and C4-2 cells after overexpression and knocking down of AR. AR: androgen receptor; sh-AR: knocking down AR expression; oe-AR: overexpressing AR expression; GAPDH: glyceraldehyde-3-phosphate dehydrogenase.