Background
Prostate cancer (PCa) is the most commonly diagnosed cancer and the third leading cause of cancer-related death in European men [
1]. Radical prostatectomy (RP) and radiotherapy are the standard of care to treat primary prostate adenocarcinoma and 70% of patients will be cured [
2,
3]. Androgen deprivation therapy (ADT) is often applied as adjuvant treatment for radiotherapy or to treat advanced, hormone sensitive tumors [
4] but most patients will develop castration-resistant PCa (CRPC) and progress to metastatic disease, which comprises the most lethal phase of PCa [
2]. To date, treatment options for this stage of the disease are limited [
5,
6]. New therapeutic strategies are therefore urgently required. miRNAs have been suggested as diagnostic and prognostic biomarkers and their application as new molecular therapeutic strategies in cancer and other diseases is currently under investigation [
7‐
10].
MicroRNAs (miRNAs) are 19-25 nt long small non-coding RNAs that modulate a multitude of biological processes by negative regulation of their target mRNAs [
11,
12]. miRNAs are encoded in introns or separate genes [
13‐
15] and transcribed into a long primary transcript (pri-miR), which is cut by Drosha/DGCR8 complex into the ~ 70 nucleotide long precursor miRNA (pre-miRNA). In the cytoplasm, the pre-miRNA is processed by Dicer into the mature miRNA and incorporated into the Argonaut complex to regulate mRNA stability and translation via binding in the 3’UTR of its target mRNA [
16,
17]. miRNAs have been found to be dysregulated in all types of cancer [
18,
19] and some exert oncogenic or tumor suppressive function (oncomiRs) [
20]. miR-221 is a cancer-associated miRNA [
21]. It is encoded in a gene cluster together with miR-222 on the short arm of the X chromosome. miR-221/− 222 are transcribed together into the pri-miRNA, which is further processed to mature miR-221-3p and the passenger miR-221-5p to regulate gene expression [
22].
miR-221 is overexpressed in a variety of epithelial cancers including breast, liver, bladder, pancreas, gastric, colorectal cancer, melanoma, papillary thyroid carcinoma and glioblastoma [
21,
23]. In PCa miR-221 upregulation [
24,
25] as well as miR-221 downregulation has been reported [
26‐
30]. Due to the strong downregulation of miR-221 in PCa tissue and in the blood of PCa patients, miR-221 has been suggested as prognostic and predictive biomarker in PCa [
31‐
33]. However, the role of miR-221 in tumorigenesis and cancer progression is controversial. miR-221-3p has been investigated extensively compared to miR-221-5p. In vivo studies demonstrated that miR-221/222 downregulation impairs the growth of PCa xenografts implying that miR-221-3p acts as oncogenic miRNA in PCa [
34]. Moreover, miR-221-3p promotes proliferation of PCa cells by direct interaction and downregulation of p27/Kip1 [
35]. Other tumor suppressors targeted by miR-221-3p include p57/Kip2, PTEN, TIMP3 and PUMA [
36‐
38] . In contrast, miR-221-3p can act as tumor suppressor via downregulation of oncogenic c-kit, thereby regulating angiogenesis [
39] and inhibiting the growth of erythroleukemic cells [
40]. Moreover, it has been shown to reduce proliferation of PCa cells via downregulation of Bmi-1 [
41], SOCS3 and IRF2 [
42] and inhibit migration and invasion by targeting Ecm29 [
28].
miR-221-5p exerts tumor suppressive function in colorectal cancer, reducing metastatic burden in vivo [
43]. However, miR-221-5p shows oncogenic function by targeting SOCS1 in PCa cells. Consequently, silencing of miR-221-5p reduced tumor growth in vivo in a xenograft mouse model [
44]. These studies indicate that despite relatively consistent reports of miR-221 downregulation in PCa its role in tumorigenesis is still unclear.
In this study we characterized the effects of miR-221-5p overexpression and downregulation in vitro in two human PCa cell lines (PC-3M-Pro4luc2 cells and C4–2 cells) and investigated the function of miR-221-5p in two in vivo models. Our results demonstrate tumor suppressive function of miR-221-5p in the context of PCa and support the inverse correlation between miR-221-5p expression and disease progression in microRNA expression dataset [
45]. We show that miR-221-5p overexpression decreased proliferation and migration of PCa cell lines. Consistently with our in vitro data, miR-221-5p overexpression reduced tumor burden in vivo in a zebrafish model of cancer cell extravasation and in a xenograft mouse model of tumor growth.
Methods
Analysis of microRNA dataset
miR-221 (3p and 5p) expression in PCa patients was analysed in a publicly available dataset of 28 normal prostate samples and 113 tumor samples (99 primary and 14 metastatic samples; GSE21036) [
45]. miR-221-5p and -3p expression data from GSE21036 dataset was extracted with R using shinyGEO [
46].
Cell culture
Normal prostatic epithelial cell line, Ep156T, was derived from normal tissue of a PCa patient undergoing radical prostatectomy [
47] and cultured in MCDB-153 medium (Sigma Aldrich) containing 1.2 g/L NaHCO
3 (Sigma Aldrich) and supplemented with 1% FBS (Seraglob, Bioswisstec Ltd), 1% Penicillin/Streptomycin (100 U/L/100 μg/ml; Sigma Aldrich), 1% non-essential amino acids (without L-Glutamine; Gibco®, ThermoFisher Scientific), 200 nM Hydrocortisone (Sigma Aldrich), 10 nM Dihydrotestosterone (DHT; Fluka Chemica), 10 nM Triiodothyronine (T3; Sigma Aldrich), 1% Insulin-transferrin-selenium (Gibco®, ThermoFisher Scientific), 5 ng/ml EGF (Peprotech), 50 μg/ml bovine pituitary extract (Gibco®, ThermoFisher Scientific) and 10 nM R1881 (LCG Group). PC-3M-Pro4luc2 cells originate from serial passage of PC-3M cells in the prostate of athymic mice [
48] and were maintained in DMEM medium: DMEM containing 4.5 g/L D-Glucose and L-Glutamine (Gibco® LifeTechnologies™) supplemented with 10% FetalClone® II (HyClone™, GE Healthcare), 1% Penicillin/Streptomycin (100 U/L/100 μg/ml) and 0.8 mg/ml G418 Geneticin® (Gibco®, LifeTechnologies™) for the selection of luciferase positive clonal cell pools. The androgen-independent DU145 cell line was first isolated from a brain metastasis of a PCa patient [
49] and was maintained in DMEM (4.5 g/L D-Glucose) medium supplemented with 10% FBS and 1% Penicillin/Streptomycin (100 U/L/100 μg/ml). VCaP cells [
50] were derived from a bone metastasis of hormone refractory PCa and were maintained in RPMI 1640 medium with 2 g/L NaHCO
3 and L-glutamine (Biochrom) supplemented with 10% FBS and 1% Penicillin/Streptomycin (100 U/L/100 μg/ml). LNCaP cells were originally derived from a lymph node metastasis [
50] and grown in T-medium: DMEM containing 1 mg/L glucose (Sigma Aldrich) supplemented with 20% F12K (Gibco®, LifeTechnologies™), 10% FBS, 1% Insulin-transferrin-selenium, 13.6 pg/ml T3, 0.25μg/ml Biotin, 25 μg/ml Adenine and 1% Penicillin/Streptomycin (100 U/L/100 μg/ml). An androgen-independent cell line, C4–2 cells, was derived from LNCaP cells by passaging in castrated mice [
50]. C4–2 cells were cultured in T-medium. All cell lines were maintained at 37 °C in an atmosphere containing 5% CO
2 and were passaged when they reached a confluency of 70–90%. For the rescue experiment, cells were seeded and the total amount of cells was passaged in bigger cell culture flasks when they reached 80–90% confluency. The human cell lines employed for functional characterisation in this study have been authenticated using highly-polymorphic short tandem repeat loci (STRs).
Cell cycle analysis
PC-3M-Pro4luc2 cells (20′000 cells/ml, 34′000 cells/ml and 40′000 cells/ml) and C4–2 cells (20′000 cells/ml, 34′000 cells/ml and 44′000 cells/ml) were seeded in 5 ml complete medium. The next day, a mild cell cycle arrest was induced by starvation for 48 h in the respective standard medium containing 0.3% FC II or 0.3% FBS, respectively. Starvation medium was replaced by complete medium containing 10% FC II or 10% FBS, respectively, and cell cycle progression was analysed at 0 h, 16 h and 48 h after cell cycle release. 250′000 cells were stained by Propidium iodide (PI) staining in Nicoletti buffer (0.1% Na3C6H5O7, 50 μg/ml PI and 0.1% Triton-X in water) at 4 °C for 30 min and DNA content measured by flow cytometry (BD™ LSR II). The percentage of cells in G1, S and G2/M phase was analysed in FlowJo™ 10 (FlowJo LLC) by applying the Dean-Jett-Fox model.
Transfection
For transfection, 90′000 PC-3M-Pro4luc2 cells/ml and 150′000 C4–2 cells/ml were transfected with Lipofectamine® 2000 Reagent (Invitrogen) in 2 ml Opti-MEM® (Gibco®, LifeTechnologies™) according to the manufacturer’s protocol as described before [
51]. Cells were transfected with 10 nM hsa-miR-221-5p pre-miR™ miRNA Precursor (miR-221-5p; assay ID: PM12613, Ambion), 10 nM Pre-miR™ Negative Control #1 (scrambled; assay ID: AM17110, Ambion), 10 nM Anti-miR™ miRNA Inhibitor hsa-miR-221-5p (anti-miR-221-5p; assay ID: AM12613, Ambion) or 10 nM Anti-miR™ miRNA Inhibitor Negative Control #1 (anti-scrambled; assay ID: AM17010, Ambion).
MTS assay
Transfected PC-3M-Pro4luc2 cells and C4–2 cells were seeded at a density of 10′000 cells/ml or 20′000 cells/ml, respectively, in 150 μl complete medium. 20 μl 3-(4,5 dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- sulfophenyl)-2H-tetrazolium (MTS; CellTiter 96® AQueous One Solution Cell Proliferation Assay, Promega) were added to each well at indicated time points (0 h, 24 h, 48 h, and 72 h after seeding) and absorbance measured at 490 nm wavelength after 2 h incubation at 37 °C.
Clonogenicity assay
PC-3M-pro4luc2 cells were seeded at a density of 50 cells/ml and C4–2 cells at 400 cells/ml in 2 ml complete DMEM or T-medium, respectively, and incubated at 37 °C for 14 days. The medium was re-freshed twice a week and after 2 weeks the cells were fixed in 4% paraformaldehyde (PFA) and stained with crystal violet (Merck) 1:20 diluted in distilled water. Pictures of plates were taken by ChemiDoc (BioRad) and colonies counted by ImageJ 1.51j8 [
52].
Migration assay
PC-3 M-Pro4luc2 cells and C4–2 cells were starved overnight in their respective standard medium containing 0.3% FC II or 0.3% FBS, respectively. The next day, 60′000 PC-3M-Pro4luc2 cells or 100′000 C4–2 cells were seeded in 500ul starvation medium in Boyden chambers with 8 μm pore size (Corning). Complete medium containing 10% FC II or 10% FBS, respectively, was added to the bottom of the well. PC-3M-Pro4luc2 cells and C4–2 cells were allowed to migrate for 22 h and 27 h, respectively, and then fixed in 4% PFA and stained with crystal violet as described by Zoni et al. [
51]. Five images per chamber were taken by bright field microscope (Olympus) and migrated cells were counted.
Zebrafish model
Tg(mpo:GFP)i114 zebrafish line [
53,
54] was maintained according to standard protocols (
www.ZFIN.org). PC-3 M-Pro4 cells fluorescently labelled with mCherry (PC-3M-Pro4_mCherry cells) were transfected with 10 nM miR-221-5p or 10 nM scrambled 48 h before inoculation. Zebrafish embryos were dechorionized at 2 days post fertilisation (dpf), anaesthetized with 0.003% tricaine (Sigma) and placed on 10 cm petri dish coated with 3% agarose. A single cell suspension of transfected PC-3M-Pro4_mCherry cells in PBS was kept at room temperature and was loaded into borosilicate glass capillary needles (1 mm O.D. × 0.78 mm I.D.; Harvard Apparatus). About 400 cells were injected above the ventral end of the Duct of Cuvier, where the Duct of Cuvier opens into the heart, using a Pneumatic Picopump and a manipulator (WPI). Data are representative of at least ≥20 embryos per group. Survival rate of control group lower than 80% was used as discard cut-off. After implantation, zebrafish embryos were maintained at 33 °C [
55]. Images were taken at 1 day post injection (dpi) and 4dpi and red pixels were quantified by imageJ.
Orthotopic mouse model
Male, immunocompromised Balb/cByJ nude mice were housed at the Central Animal Facility of the Medical Faculty of the University of Bern in a specific pathogen-free environment at a 12 h light/dark cycle. Mice had free access to autoclaved chow and water and were housed at a maximum of 5 animals/cage. Mice were maintained and checked daily according to animal license. Animal experiments were approved by the Canton of Bern, Switzerland (Permit Number: BE55/16), and carried out in accordance with the Swiss Guidelines for the Care and Use of Laboratory Animals and the ARRIVE guidelines. Mice were handled exclusively during day time at the Central Animal Facility of the Medical Faculty of the University of Bern. At the beginning of the experiment, mice were 8-weeks old and weighed 22 g ± 2 g. To conduct this study, a total of 10 mice was used to assess orthotopic growth of miR-221-5p overexpressing PC-3M-Pro4luc2 cells. PC-3M-Pro4luc2 cells were transfected with 10 nM miR-221-5p and 10 nM scrambled 48 h prior to inoculation. 8-weeks old male Balb/cByJ nude mice (Charles River France) were allocated to control group (scrambled:
n = 5 mice) and experimental group (miR-221-5p = 5 mice). The mean body weight was comparable in both groups at beginning of the experiment. Mice were anaesthetized with Domitor® (0.5 mg/kg)/Dormicum® (5 mg/kg)/Fentanyl (0.05 mg/kg) triple narcosis mix subcutaneous (s.c.) injection, which is well tolerated by the mouse strain, and the eyes were protected by Vitamin A cream during surgery. Approximately 50′000 single cells were reconstituted in 10 μl PBS and injected into the right lobe of the anterior prostate. Surgeries were started in the morning with implantation of miR-221-5p transfected PC-3M-Pro4luc2 cells. Mice of the second cage (
n = 5) were injected with scrambled transfected PC-3 M-Pro4luc2 cells. Immediately after surgery, the mice were s. c. injected with analgesic mix of Alzane® (1.1 mg/kg)/Anexate® (0.45 mg/kg)/Temgesic® (0.075 mg/kg) and received analgesia (Temgesic® 0.1 mg/kg) by s. c. injection twice a day for 3 days after surgery. Tumor growth was screened weekly by assessing bioluminescence with s. c. injection of 30 μl/mouse 250 mM D-Luciferin sodium salt (Synchem) and measuring by NightOWL II LB983 in vivo imaging system (Berthold Technologies). For BLI measurements, mice were anaesthetized by s. c. injection of Domitor® (0.5 mg/kg)/Dormicum® (5 mg/kg)/Fentanyl (0.05 mg/kg) triple narcosis mix and received an antidote mix of Alzane® (1.02 mg/kg)/Anexate® (0.42 mg/kg)/Naloxon (0.6 mg/kg) after measurement. Animals without signal along the course of the experiment were excluded from the analysis. At 28dpi the bioluminescence signal reached more than 10
6cps and tumors were dissected. Tumor volume was calculated as π/6 x length x width
2 [
56] and normalised to the body weight (Additional file
1: Table S1). Tumor architecture and molecular markers were histologically characterised (see next paragraph). After tumor dissection, the experiment was terminated and all mice were euthanized by CO
2 according to the recommendations of the Federal Veterinary Office.
Histology and immunofluorescence stainings
Dissected tumors were formalin-fixed, paraffin-embedded (FFPE) and cut in 4 μm sections for staining by haematoxylin and eosin (H&E). Images were taken by Pannoramic 250 Flash II Scanner (3D Histech Ltd). For immunofluorescence, antigens of FFPE sections were retrieved in boiling citrate buffer and sections blocked in 1%BSA (Fluka)/PBS-Tween (Merck) for 30 min. The sections were stained for monoclonal anti-human proliferating cell nuclear antigen (PCNA; Cat.#: P8825, Sigma Aldrich) 1:500, polyclonal anti-human cleaved caspase-3 (cl. casp-3; Cat.#: 9661, Cell Signaling) 1:500, polyclonal anti-human pan-cytokeratin (panCK; Cat.#: A0575, DAKO) 1:250 and monoclonal anti-mouse α-smooth muscle actin (α-SMA; Cat.#: A2547, Sigma Aldrich) 1:400 in 1%BSA/PBS-Tween at 4 °C overnight. Subsequently, sections were incubated with secondary antibodies AlexaFluor 488 donkey anti-mouse (Cat.#: A-21202, LifeTechnologies™) 1:250 and AlexaFluor 555 donkey anti-rabbit (Cat.#: A-31572, LifeTechnologies™) 1:250 in PBS at room temperature for 1½ h. Images were taken by LCI Leica DMI4000 B microscope (Leica Microsystems).
For phalloidin staining, PC-3M-pro4luc2 cells transfected with 10 nM miR-221-5p or 10 nM scrambled were stained with Acti-stain 488 Phalloidin (Cat.#: 176753, abcam) according to manufacturer’s protocol. Briefly, 8′000 cells/well were seeded on 8 chamber polystyrene vessel tissue culture treated glass slides (Falcon) 24 h post transfection and let attach for 48 h. Cells were fixed in 4% PFA, permeabilised in 0.1% Triton-X in PBS and stained with Acti-stain 488 Phalloidin for 40 min at room temperature. Images were taken by LCI Leica DMI4000 B microscope and analysed by imageJ for mean phalloidin intensity normalised to the number of nuclei.
RNA isolation and RT-qPCR
Total RNA was isolated from wild-type cell lines (Ep156T, LNCaP, C4–2, VCaP, PC-3M-Pro4luc2 and DU145 cells) or transfected PC-3M-pro4luc2 cells and C4–2 cells 72 h post transfection by TriPure Isolation Reagent (Roche) according to manufacturer’s protocol with an additional wash with 90% ethanol after RNA precipitation in isopropanol. For mRNA detection, cDNA was synthesised from 500 ng total RNA by M-MLV reverse transcriptase (Promega) and qPCR was performed by FastStart Universal SYBR Green Master Mix (Roche #04913850001) for EMT markers and housekeeping genes HPRT and β-actin (see Additional file
1: Table S2 for primer sequences). miRNA expression was analysed by cDNA synthesis from 2 ng total RNA by TaqMan™ Advanced miRNA cDNA Synthesis Kit (applied biosystems) and qPCR for hsa-miR-221-5p TaqMan™ Advanced miRNA Assays (assay ID: 478778_mir, applied biosystems) and housekeeping miRNAs hsa-miR-103a-3p TaqMan® Advanced miRNA Assays (assay ID: 478253_mir, applied biosystems) and hsa-miR-186-5p TaqMan® Advanced miRNA Assays (assay ID: 477940_mir, applied biosystems). Expression was normalised to housekeeping genes or miRNAs, respectively, and results are shown as relative expression calculated by 2
-ΔCt or as LOG difference calculated by LOG(2
-ΔΔCt) method. Ct values > 35 were excluded from the analysis.
Protein isolation and Western blot
Proteins were isolated 72 h post transfection with RIPA buffer (150 mM NaCl, 1% Triton-X, 1% sodium deoxycholate, 1% sodium dodecyl sulphate (SDS) in 25 mM Tris, pH 7.6) containing proteinase inhibitor (Roche) and phosphatase inhibitor (Roche). Protein was quantified by Pierce™ BCA protein assay kit (Thermo Scientific) according to manufacturer’s protocol for microplate procedure. 20 μg of protein were denatured in Lämmli buffer (BioRad) containing β-mercaptoethanol (Sigma Aldrich) by boiling at 95 °C for 10 min. Proteins were separated on a 12% SDS PAGE gel and transferred to a PVDF transfer membrane (Thermo Scientific). After blocking in 5% skimmed milk (Carl Roth GmbH + Co KG) in TBS-Tween the membranes were incubated with polyclonal goat anti-E-CAD (Cat.#: AF648, R&D Systems) 1:1000 or monoclonal mouse anti-VIM (Cat.#: ab8979, abcam) 1:1000 in 5% BSA in TBS-Tween at 4 °C overnight. The next day, membranes were incubated with mouse HRP-conjugated anti-β-actin (Cat.#: A3854, Sigma Aldrich) 1:20′000 in 5% skimmed milk and secondary antibodies HRP donkey anti-goat IgG (Cat.#: ab97110, abcam) 1:10′000 and ECL™ peroxidase labelled anti-mouse (Cat.#: NA931VS, GE Healthcare) 1:10′000 in TBS-Tween at room temperature for 1 h. WesternBright™ ECL-HRP substrate (Witec) was added to the membranes and images taken by Fusion-FX7–820 (Witec). Protein bands were quantified by imageJ as described by Luke
Miller.org (
https://lukemiller.org/index.php/2010/11/analyzing-gels-and-western-blots-with-image-j/).
Statistics
The data was analysed by t-test, one-way ANOVA or two-way ANOVA as indicated in the figure legends with a confidence interval α ≤ 0.05. Data are shown as mean ± SD. P values smaller than 0.05 were considered as statistically significant (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). For BLI data of mouse experiment, normality plot of residuals was checked to make sure it approximately follows a straight line. All statistical tests were performed in GraphPad Prism 7.01 (GraphPad Software) or R (The R Foundation).
Discussion
In this in vitro and in vivo preclinical study, we demonstrated that miR-221-5p acts as tumor suppressor in PCa. miR-221 is generally overexpressed in epithelial cancers and plays an oncogenic role [
21]. However, in prostate cancer, oncogenic [
25,
34,
35,
44,
58] and tumor suppressive function for miR-221 have been described [
27,
28,
41,
42]. Here we reported downregulation of miR-221-5p in PCa cell lines compared to non-cancerous prostatic epithelial Ep156T cells implying an important role for miR-221-5p in the maintenance of a normal epithelial phenotype. Proliferation was decreased by miR-221-5p overexpression in PC-3M-Pro4luc2 cells and C4–2 cells. Notably, we observed faster cell cycle progression in PC-3M-Pro4luc2 cells than in C4–2 cells despite higher baseline expression of miR-221-5p in PC-3M-Pro4luc2 cells. The high proliferation rate of PC-3M-Pro4luc2 cells can be attributed to their high expression of TGF-α and EGF-R as well as p53 mutation and PTEN deficiency which might compensate for intrinsically higher miR-221-5p levels [
59]. However, miR-221-5p overexpression significantly reduced proliferation of PC-3M-Pro4luc2 cells. Knock-down of miR-221-5p had no functional effects on PC-3M-Pro4luc2 cells. Despite specific action of anti-miR-221-5p, we achieved a miR-221-5p downregulation of 40%. Given the redundant action of miRNAs and the endogenously low miR-221-5p expression levels in PCa cell lines compared to normal prostatic epithelial cells, the in vitro miR-221-5p downregulation might be too weak to see a positive functional effect. Previously published work demonstrated miR-221-5p overexpression to reduce proliferation of PCa cells by downregulation of cell cycle regulatory proteins [
27]. However, miR-221-5p overexpression has also been reported to promote proliferation of PCa cell lines by activating Ras/Raf/MEK/ERK pathway and consequently stable silencing of miR-221-5p reduced tumor growth of PCa cell xenografts [
44]. These studies imply that miR-221-5p targets a multitude of pathways and acts as tumor suppressor or oncogenic miRNA depending on the cellular and experimental conditions. For a better understanding of miR-221-5p function during PCa growth, we studied miR-221-5p in a transient overexpression orthotopic in vivo model. miR-221-5p overexpression significantly reduced tumor growth and tumor volume compared to control. Tumors did not display differences in proliferation or apoptosis rates as determined by PCNA and cl. casp-3 staining, which is likely due to the loss of overexpressed miR-221-5p in the course of the in vivo experiment. Indeed, miR-221-5p diminished significantly over 2 weeks after overexpression in vitro, which was associated with loss of effect on proliferation. This supports the hypothesis that the observed tumor suppressive effects are dependent on miR-221-5p (over)expression levels and might also account for the lack of detectable direct effects of miR-221-5p overexpression in the dissected tumors. While there is still a significant difference in tumor growth in vivo after 28 days, the effect of miR-221-5p overexpression on cell growth was lost after 12 days in vitro (Fig.
5c). This is due to the fact that control cells reached confluency earlier than miR-221-5p overexpressing cells and were not proliferating anymore. In contrast, scrambled transfected cells proliferated at a high rate at 2 weeks post injection in vivo and did not experience any spatial limitations for growth. Despite transient transfection, miR-221-5p overexpression reduced tumor burden significantly in vivo suggesting an initially large inhibitory effect of miR-221-5p overexpression on PCa growth. Our data imply that transient miR-221-5p overexpression reduces PCa cell proliferation in vitro and influences the growth kinetic of orthotropic tumors in vivo. Analysis of publicly available microRNA dataset (GSE21036) [
45] led us to the finding that miR-221-5p expression is progressively decreased in PCa tumor and further decreases in metastasis. Low miR-221-5p expression correlated with lower histological differentiation as assessed by Gleason score and with disease progression as assessed by TNM staging, indicating that miR-221-5p downregulation is a pre-requisite for local and advanced disease in PCa patients. The significance of miR-221 as tumor suppressor in PCa is supported by the notion, that miR-221 is downregulated in clinical specimens of TMPRMSS2:ERG fusion positive PCa, which comprise over 50% of all diagnosed tumors [
60]. Moreover, low miR-221 levels in PCa tissue are associated with earlier recurrence after radical prostatectomy [
61].
The ability to undergo unlimited division and to form clones is an essential feature of aggressive cancer cells. miR-221-5p overexpression significantly reduced colony formation suggesting that miR-221-5p might interfere with self-renewal mechanisms. Interestingly, murine miR-221-5p has been shown to directly target and downregulate Oct4, Nanog, Sox2, Klf4 and PRMT7 in mouse embryonic stem cells thereby functioning as “anti-stemness” miRNA [
62]. In human PCa cells, miR-221-3p has been shown to target the stem cell factor Bmi-1 [
41]. Oct4, Nanog, Sox2 and Bmi-1 are upregulated in PCa tissue and tumor-initiating PCa cells [
63,
64]. However, additional studies are required to elucidate the mechanistic effect of miR-221 and its interaction with transcription factors implicated in self-renewal and cell dedifferentiation, leading to highly aggressive, metastatic disease.
Epithelial-to-mesenchymal transition (EMT) is an established characteristic of highly aggressive cancer cells. Differential expression of E-CAD, N-CAD or VIM upon miR-221-5p overexpression has already been reported in PCa cell lines [
27,
44]. We showed that miR-221-5p overexpression induced an increase of the E-CAD/VIM ratio on mRNA and protein level in PC-3M-Pro4luc2 cells leading to a more epithelial phenotype, which can be mainly attributed to a decrease of VIM protein expression. In contrast to the highly mesenchymal PC-3M-Pro4luc2 cells, C4–2 cells express intrinsically lower VIM levels, which are not affected by miR-221-5p overexpression leading to a decreased E-CAD/VIM ratio in miR-221-5p overexpressing C4–2 cells. Our data support the notion that the EMT expression profile of PC-3M-Pro4luc2 cells and C4–2 cells are differentially affected by miR-221-5p overexpression. Additionally, miR-221-5p overexpression reduced migration of both cell lines dramatically. PC-3M-Pro4luc2 cells display elongated morphology and migrate in a “mesenchymal manner” [
65]. In contrast, C4–2 cells show a less mesenchymal phenotype and migrate in an “amoeboid manner” [
65,
66]. We have previously shown that different molecular mechanisms support motility in PC-3M-Pro4luc2 and C4–2B cells [
51]. Interestingly, with the diminished miR-221-5p overexpression levels 2 weeks post transfection the migration of PC-3 M-Pro4luc2 cells was significantly increased compared to control (Fig.
5f). It has previously been shown for miR-430 that miRNAs can balance the expression of agonist/antagonist pairs [
67]. Similarly, the moderately increased miR-221-5p levels at 2 weeks post transfection in combination with cellular compensatory mechanisms [
12,
68] might have led to an imbalance between positive and negative regulators of migration resulting in increased migration. The ability to migrate is a pre-requisite for tumor cell extravasation and metastasis formation. This highly dynamic process requires engagement of adhesion molecules including selectins, cadherins, integrins, CD44 and immunoglobulin superfamily receptors [
69]. In line with the modulation of E-CAD and VIM expression and reduced migration in vitro by miR-221-5p overexpression, extravasation of highly aggressive PC-3M-Pro4_mCherry cells was reduced by miR-221-5p in vivo, in a zebrafish model of tumor cell extravasation. miR-221-5p transfected PC-3M-Pro4_mCherry cells formed less and smaller metastatic foci at the caudal hematopoietic tissue (CHT) of zebrafish embryos, suggesting a critical role of miR-221-5p to prevent extravasation at distant sites. These results matched our in vitro observations and similarly as in the mouse experiment, miR-221-5p overexpression reduced tumor burden in zebrafish embryos.
Taken together, our findings and the loss of miR-221 during PCa progression [
27,
33], might suggest a role for miR-221 in counteracting tumor cell migration and metastasis formation. Accordingly, we found lower miR-221-5p levels in the more advanced, androgen-independent C4–2 cells compared to the parental, androgen-dependent LNCaP cell line [
50]. VCaP cells were derived from a vertebral metastasis and represent a model of advanced, metastatic PCa. miR-221-5p levels were nearly undetectable in VCaP cells. This finding aligns with downregulation of miR-221-5p in metastatic disease compared to primary tumor as we found by analysis of a publicly available patient dataset. Increased methylation of the miR-221/− 222 locus and miR-221-5p suppression has been associated with metastatic PCa and might contribute to disease progression [
27]. We observed higher miR-221-5p expression in AR
− cell lines than in AR
+ PCa cell lines, which is possibly due to the absence of AR, a transcriptional repressor of the miR-221/− 222 gene cluster [
70]. ADT resistance and progression to metastasis is often linked to restored AR activity [
71], which potentially leads to repression of miR-221/− 222 in CRPC and could further promote PCa progression in this critical phase. However, additional experiments are necessary to elucidate the relation between miR-221-5p and AR status.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.