A microRNA profiling microarray expression dataset consisting of three pairs of parental EOC cells and their PTX-resistant sublines (GSE148251) (20) was downloaded from Gene Expression Omnibus (GEO platform). The miRNA expression levels were analyzed using GEO2R bioconductor project and provided as log2 ratios (21). We explored three mRNA target-predicting algorithms including mirSVR, Target scan and miRDB to screen the potential downstream targets of hsa-miR-105 (22).

The human EOC cell lines (SKOV3, TOV21G and OVCAR8) and 293T cells, obtained from the American Type Culture Collection (ATCC), were maintained in RPMI-1640 medium supplemented with 10% (v/v) fetal bovine serum (FBS), penicillin (100 U/ml), streptomycin (100 U/ml) and amphotericin B (25 µg/ml) at 37°C in a 5% CO₂ atmosphere. The PTX-resistant cell lines (SKOV3/PTX and TOV21G/PTX) were generated by exposing parental cells to gradually increased doses of PTX (Sigma-Aldrich; Merck KGaA), and were finally maintained in complete medium containing 50 nM PTX (23). All cell lines used were recently authenticated using short tandem repeat analysis performed by Suzhou Genetic Testing Biotechnology, and were tested for mycoplasma contamination using Universal Mycoplasma Detection Kit (ATCC) prior to the experiments.

Hsa-miR-105 mimics and inhibitors for in vitro transfection and hsa-miR-105 micrON™ agomir for in vivo transfection, along with their negative controls, were purchased from Ribobio. Zinc And Ring Finger 2 (ZNRF2) lentiviral shRNA particle was obtained from GeneCopoeia. To manipulate the hsa-miR-105 expression, cells were transfected with 100 nM mimics or inhibitors with the aid of Lipofectamine™ 3000 (Thermo Fisher Scientific, Inc.) for 48 h. For lentiviral transduction, the viral particles were prepared by co-transfection of expression plasmid and packaging plasmids into 293T cells, followed by harvest of virus-containing supernatants 48 h after transfection. SKOV3 and TOV21G cells were then transduced with the lentivirus, followed by selection of positive clones using 1 µg/ml of puromycin (Sigma-Aldrich; Merck KGaA) for ~14 days.

Forty-eight hours after transfection with the indicated oligonucleotides, EOC cells were seeded in 96-well plates at the density of 5×10³ cells/well. Cells were then exposed to different doses of PTX as indicated or DMSO for 24 h after adhering to the plate. Upon completion of PTX treatment, fresh medium was added and cell viability was determined using Cell Counting Kit-8 (MedChemExpress), as per the manufacturers' instructions. Each concentration was conducted in triplicate and repeated at least twice. The half maximal inhibitory concentration (IC₅₀) of cell viability was calculated as the level that caused 50% reduction in cell viability vs. the parental cells (24).

About 500 cells were seeded into the 6-well plates. After a 48-h culture, PTX (20 nM) was added into the cultures. After a 3-day treatment with PTX, cells were transferred to fresh medium and cultured for another 12 days until colonies were large enough to be clearly discerned. Cells were then fixed using 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) for 20 min, followed by staining with crystal violet stain solution (0.5%) (Sangon Biotech). Counting colonies was finally achieved by using Oracle Java SE software (25), with the average number of colonies being plotted (mean ± SD, n=3).

The invasiveness of EOC cells in PTX-containing medium was determined by Transwell chamber (26). Briefly, 1×10⁵ cells were seeded in the upper chamber of a Transwell with Matrigel-coated membrane (Corning, Inc.). After a 24-h culture, cells were then treated with 20 nM of PTX for another 24 h, followed by removal of cells from the upper chamber. The cells that had migrated to the lower surface were fixed and stained with 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific, Inc.) for 5 min and then counted under a Leica M205 FCA fluorescence microscope (Leica Microsystems, Inc.). Ten randomly selected fields were chosen for cell number quantification.

All animal research was conducted in accordance with Guide for the Care and Use of Laboratory Animals from NIH, and was approved by the Animal Research Ethics Board of First Affiliated Hospital of Xi'an Jiaotong University (approval no. XMU-2012-1682). Animals were maintained under a constant 12-h light:12-h darkness cycle (lights on at 08:00 h) and controlled conditions of humidity (between 70 and 80%) and temperature (22±1°C). They were allowed free access to experimental diets and water, and were allowed to acclimatize for at least 1 week before the experiment. For establishment of the tumor xenograft model, 2×10⁶ SKOV3/PTX or TOV21G/PTX cells were subcutaneously injected into flanks of female BALB/cA-nu mice (n=125, 6-weeks-old, weighing 19–23 g, obtained from the Animal Facility of The First Affiliated Hospital of Xi'an Medical University). On day 13 post cell inoculation, mice received intraperitoneal injections of 5 mg/kg of PTX (dissolved in PBS)/kg body weight every three days. On the following day of the first PTX injection, 10 nmol of hsa-miR-105 micrON^TM agomir or negative control (dissolved in 0.1 ml PBS) was locally injected into the tumor mass once every 3 days for two weeks. Mice receiving PBS injection served as the negative controls. The animals were checked daily, and tumor volume (V) was recorded and calculated using the formula: V=(length × width²)/2. Humane endpoints used in this study included tumor ulceration, infection, or necrosis, loss of ability to ambulate (inability to access food or water) and labored respiration. Four weeks after cell inoculation, mice were sacrificed by cervical dislocation following sodium pentobarbital anesthesia (200 mg/kg, intraperitoneally). Some xenograft tissues were snap frozen in liquid nitrogen, followed by being stored at −80°C. The other xenografts were immediately fixed in 10% formalin and processed for histological examination.

Upon receipt of the written informed consent, a total of 115 female patients with EOC and known clinical follow-up were recruited from the Department of Obstetrics and Gynecology, First Affiliated Hospital of Xi'an Medical University during August 2012 and June 2019. All patients with EOC underwent initial cytoreductive surgery followed by 6–8 cycles of standard paclitaxel-based chemotherapy (27). Patients with progressive disease during primary chemotherapy according to medical image analysis or with recurrent disease within 6 months of completing primary chemotherapy were termed PTX-resistant (28). Some tumor samples were snap frozen in liquid nitrogen, followed by being stored at −80°C until use, whereas other tumor samples were fixed in 10% formalin and processed for histological examination. The protocols involved in the human study strictly conformed to the 2008 Revised Declaration of Helsinki, and were approved by the Ethics Committee of the First Affiliated Hospital of Xi'an Medical University (approval no. XMU-2012-1683).

Real-time RT-qPCR was carried out as described elsewhere (29,30). Total RNA was isolated and purified using PureLink™ miRNA Isolation Kit (Thermo Fisher Scientific, Inc.). RT-qPCR analysis of miRNA expression was performed using Applied Biosystems TaqMan microRNA assay system (Applied Biosystems), with the aid of microRNA-specific primers (5′-AAAAGCUGGGUUGAGAGGGCGA-3′) and TaqMan™ MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific, Inc.). To quantify the expression levels of other target genes, 500 ng of total RNA was used to synthesize the first strand cDNA using iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories, Inc.). Subsequent qPCR was carried out on the Applied Biosystems^® 7500 Fast Real-Time PCR System. Relative expression levels of target miRNA and mRNAs were determined using the standard ^ΔΔCq method (31), with human U6 snRNA or 18S RNA serving as the internal controls (32). Primers used in the current study are listed in Table I.

Western blot analysis was conducted according to published protocols (33). Total protein was prepared using ReadyPrep™ Protein Extraction Kit (Bio-Rad Laboratories, Inc.). Protein samples (~30 µg) were separated on 10% SDS-PAGE and transferred to PDVF membranes (Millipore Corp.). After being blocked by 5% nonfat dry milk in TBST, the membranes were incubated with primary antibodies (Table II) at 4°C overnight. Positive signals were developed using an EZ-ECL Chemiluminescence Detection kit (Biological Industries) on a GOBlot™ Western Blot Processor (Cytoskeleton) after incubation with proper secondary antibodies.

Localization of hsa-miR-105 in EOC cells was revealed by RNA fluorescence in situ hybridization (FISH) using a locked nucleic acid (LNA) probe for human mature hsa-miR-105 (5′-UCAAAUGCUCAGACUCCUGUGGU-3′) and scramble miRNA control probe (5′-GTGTAACACGTCTATACGCCCA-3′; Exiqon), both of which were labeled with digoxigenin at the 5′end with the aid of the DIG RNA Labeling Kit (Roche). EOC cells were fixed in 4% paraformaldehyde, followed by hybridization at 37°C overnight in a dark moist chamber. After thorough rinse, cells were incubated with TSA fluorescent signal reaction solution (PerkinElmer) at room temperature for 30 min and were then sealed using Fluoroshield™ with DAPI histology mounting medium (Sigma-Aldrich; Merck KGaA). Positive staining was observed under a Leica M205 FCA fluorescence microscope.

The distribution of hsa-miR-105 in EOC tissues was revealed using chromogenic in situ hybridization (CISH) (29). Briefly, after a routine deparaffinization and rehydration, 4% paraformaldehyde-fixed paraffin-embedded EOC tissue sections were treated with 10 µg/ml of Proteinase K (Sigma-Aldrich; Merck KGaA) to expose RNA. Sections were then incubated sequentially with hybridization buffer without probes at 21°C for 2 h and with hybridization buffer containing 40 nM of the above-mentioned DIG-labeled LNA-modified probes overnight. After incubation with the blocking solution (PBST containing 2% goat serum and 2 mg/ml bovine albumin) for 1 h, sections were treated with anti-digoxigenin-AP antibody (Sigma-Aldrich; Merck KGaA) at 4°C overnight, followed by signal development using the BCIP/NBT Substrate Kit (Vector Laboratories, Inc.) according to the manufacturer's instruction.

Localization of the zinc and ring finger (ZNRF2) protein in EOC tissues was detected using immunohistochemistry as described elsewhere (34). In brief, 4% paraformaldehyde-fixed paraffin-embedded EOC tissue sections were routinely deparaffinized and rehydrated, followed by antigen retrieval by incubating slides in 0.01% citrate acid in a heated water bath (95°C) for 30 min. After elimination of endogenous peroxides activity by treatment for 10 min with 0.3% hydrogen peroxide in methanol, the sections were incubated with the primary antibody (Table II) at 4°C overnight. Positive signals were finally developed with the aid of the VECTASTAIN^® Elite^® ABC system (Vector Laboratories, Inc.) according to the protocol recommended by the manufacturer.

The 3′UTR segment of human ZNRF2 was amplified from ovarian cell genomic DNA and cloned into psiCHECK-2 luciferase reporter vector using NotI/XhoI sites. Site-directed mutagenesis of the hsa-miR-105 binding site on the ZNRF2 3′UTR was achieved using QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies, Inc.). Co-transfection of reporter plasmids and hsa-miR-105 mimics into 293T cells was carried out using Lipofectamine™ 3000. Forty-eight hours after transfection, relative luciferase activity was measured by the Dual-Luciferase Reporter Assay System (Promega Corp.).

FLAG-Ago2 was a gift from Edward Chan (Addgene plasmid no. 21538; Addgene) (35). The FLAG-Ago2 construct and hsa-miR-105 mimics were con-transfected into 293T cells for 48 h using Lipofectamine™ 3000, followed by cell harvest using RIPA buffer supplemented with RNase A, DNase and protease-free from Thermo Fisher Scientific, Inc. The FLAG^® Tag antibody (Thermo Fisher Scientific, Inc.) was preincubated with Protein G-agarose beads (Sigma-Aldrich; Merck KGaA) at 4°C overnight. Subsequently, the resultant cell lysates were incubated with preconjugated antibody at 4°C overnight. Pellets from immunoprecipitation were then dissolved in TRIsure™ (BioLine) and subjected to PCR analysis.

All experiments were conducted in triplicate and repeated at least twice. Quantitative values are expressed as the mean ± SD. Differences among/between experimental groups were analyzed by one-way analysis of variance (ANOVA) followed by Tukey's test or Student's t-test, wherever appropriate. The association between hsa-miR-105 expression and overall survival was assessed using the log-rank Kaplan-Meier analysis. The receiver operating characteristic (ROC) curve was used to examine the predictive power. P<0.05 was considered to be statistically significant.

As an initial effort to understand the potential involvement of hsa-miR-105 in paclitaxel resistance, we examined the expression of hsa-miR-105 in a microRNA profiling microarray expression dataset consisting of three pairs of parental EOC cells and their PTX-resistant sublines (GSE148251) (20). Student's t-test demonstrated that the expression of hsa-miR-105 was significantly downregulated in PTX-resistant sublines compared to their parental cell lines (percentage decrease >78.8%, P<0.0001, Fig. 1A). To validate these findings, we generated the PTX-resistant SKOV3/PTX and TOV21G/PTX cell lines using a previously validated protocol. SKOV3/PTX and TOV21G/PTX cells were −31.4 and −18.1 times more resistant to PTX than their parental SKOV3 and TOV21G cells, respectively (Fig. 1B). Compared to the expression level of hsa-miR-105 in the parental cells which was set at 100%, the relative percentage of hsa-miR-105 expression in SKOV3/PTX and TOV21G/PTX cells was 34.2 and 28.6%, respectively (Fig. 1C). Moreover, FISH analysis revealed a highly enriched expression of hsa-miR-105 in the parental SKOV3 cells compared to the PTX-resistant SKOV3/PTX subline (Fig. 1D). To further provide the in vivo evidence for the association between deregulated expression of hsa-miR-105 and paclitaxel resistance, we established a xenograft model to mimic the clinical treatment with PTX. At two weeks after PTX treatment, xenografts derived from SKOV3/PTX cells were clearly larger than those derived from parental SKOV3 cells, indicative of a successful resistance of SKOV3/PTX cells to PTX. Subsequent qPCR analysis showed that the expression of hsa-miR-105 in the SKOV3/PTX xenografts was significantly reduced by ~53.6% compared to the SKOV3 ×enografts (P<0.01, Fig. 1E). These data together suggest that hsa-miR-105 deficiency correlates with paclitaxel resistance in EOC cells.

To explore the oncogenic roles of the hsa-miR-105 deficiency, we silenced the expression of hsa-miR-105 in SKOV3 and TOV21G cells using specific inhibitors (anti-miR-105) (Fig. 2A). Following PTX treatment, knockdown of hsa-miR-105 expression conferred a stronger pro-survival effect in a dose-dependent manner, as revealed by viability assays (Fig. 2B). Consistently, higher efficiency of colony formation upon PTX co-incubation was observed in the hsa-miR-105 inhibitor (anti-miR-105)-transfected SKOV3 and TOV21G cells (Fig. 2C). Additionally, ablation of endogenous expression of hsa-miR-105 significantly increased cell invasiveness by ~3.6- and ~3.2-fold in PTX-challenged SKOV3 and TOV21G cells, respectively (Fig. 2D). We also used another serous EOC cell line (OVCAR8) to elucidate the potential involvement of hsa-miR-105 in the development of resistance to PTX (Fig. S1). These findings suggest that hsa-miR-105 deficiency may confer resistance to paclitaxel in EOC cells.

In order to investigate the antitumor effects of hsa-miR-105 at the in vivo level, we employed the micrON^TM agomir system which has been shown to be promising for therapeutic gene delivery in vivo (36). We evaluated the therapeutic effects of micrON^TM agomir-mediated hsa-miR-105 restoration on tumor growth in a xenograft model derived from either SKOV3/PTX or TOV21G/PTX cells (Fig. 3A). Intratumoral injection of hsa-miR-105 micrON^TM agomir resulted in a significant increase in hsa-miR-105 expression in xenografts at 28 days after cell inoculation (Fig. 3B), as well as an inhibition of tumor growth, especially during the late phase of PTX treatment (Fig. 3C). In accordance, the hsa-miR-105-overexpressing and PTX-treated tumors were smaller at 28 days after cell inoculation (Fig. 3D), and expressed lower levels of Ki67 (a well-known proliferation marker for human tumor cells, Fig. 3E), compared to the Ctrl group. These results indicate that in vivo inhibition of hsa-miR-105 expression by using the micrON^TM agomir may have the potential to be used as an alternative therapeutic option for EOC.

To further dissect the molecular basis underlying the effects of hsa-miR-105 on PTX responsiveness, we explored three mRNAtarget-predicting algorithms including mirSVR, Target scan and miRDB (22). A total of 17 genes were identified to be potential downstream targets of hsa-miR-105, as these genes were observed to be overlapped in all three databases (Fig. 4A). We then evaluated the transcriptional expression levels of these 17 genes in four pair of PTX-responsive and PTX-resistant EOC tissues using RT-qPCR. ZNRF2 was the only candidate gene whose expression was found to be increased by more than 3-fold in the PTX-resistant tissues compared with the PTX-responsive EOC tissues (Fig. 4B). Transfection with hsa-miR-105 mimics caused a significant decrease in the expression of ZNRF2 in SKOV3/PTX and TOV21G/PTX cells, at both mRNA (Fig. 4C) and protein levels (Fig. 4D). Consistently, immunostaining and ISH analyses revealed that the expression of hsa-miR-105 was negatively correlated to the expression of ZNRF2 protein in EOC specimens from our clinical cohort (Fig. 4E). To provide the direct evidence for the direct regulation of ZNRF2 expression by hsa-miR-105, we performed luciferase reporter assay. Co-transfection of reporter plasmids and hsa-miR-105 mimics into 293T cells resulted in a ~66.4% reduction in the transcriptional activity of ZNRF2 3′UTR, while this inhibitory effect was totally abolished when the binding site of hsa-miR-105 on the 3′UTR of ZNRF2 was mutated (Fig. 4F). To better confirm the interaction between hsa-miR-105 and ZNRF2, we conducted Ago2-mediated RNA-IP assay. In the cells that were co-overexpressed with hsa-miR-105 and FLAG-Ago2, ZNRF2 mRNA was found to bind hsa-miR-105 (Fig. 4G). These results collectively suggest that hsa-miR-105 may regulate cellular responsiveness to PTX by directly targeting ZNRF2 in EOC cells.

Since ZNRF2 has been shown to be involved in chemoresistance (37), we next investigated the biological effects of the manipulation of ZNRF2 expression in hsa-miR-105-mediated chemoresistance by generating SKOV3 and TOV21G cells that were stably transfected with ZNRF2 shRNA. Transfection with hsa-miR-105 inhibitors (anti-miR-105) in control cells resulted in a significant increase in the expression level of ZNRF2, whereas this induction was totally abolished in the SKOV3 and TOV21G cells that were stably deprived of endogenous ZNRF2 (Fig. 5A). Hsa-miR-105 inhibition-induced cell survival upon PTX challenge was attenuated in the SKOV3 and TOV21G cells that were stably transfected with ZNRF2 shRNA (Fig. 5B). Furthermore, ZNRF2 depletion enhanced the ability of PTX to decrease colony formation efficiency (Fig. 5C), in the presence of hsa-miR-105 inhibition in both SKOV3 and TOV21G cells. Thus, we identified ZNRF2 as a main downstream effector mediating the action of hsa-miR-105 in the promotion of resistance to PTX.

In order to broaden the translational research enterprise of the current study, we measured the expression levels of hsa-miR-105 in tumor samples from a total of 115 EOC patients that had received standard paclitaxel-based chemotherapy. The association between hsa-miR-105 expression and clinicopathological parameters of the patients with EOC is summarized in Table SI. Patients with high levels of hsa-miR-105 were more likely to have clinical response to PTX chemotherapy (P<0.0042). Tumor stage and grade did not significantly differ between patients with high levels of hsa-miR-105 and patients with low levels of hsa-miR-105 (Table SI). RT-qPCR revealed that hsa-miR-105 was significantly downregulated in the PTX-resistant EOC specimens compared with that in the PTX-responsive EOC specimens (P<0.0001, Fig. 6A). A ROC curve was then generated to assess the predictive power of hsa-miR-105. The corresponding AUC was 0.8649 (95% CI, 0.8024–0.9275) (Fig. 6B), suggesting that hsa-miR-105 may serve as a valuable diagnostic biomarker for EOC chemotherapy. Importantly, the patients with lower levels of hsa-miR-105 exhibited poorer survival following PTX-based chemotherapy (P=0.0014; Fig. 6C). Previous studies have identified hsa-miR-105 as a circulating miRNA (19). Given that circulating miRNAs are potent and convenient biomarkers for the evaluation of tumor development and progression (38), we sought to determine the clinical implication of hsa-miR-105 expression in the peripheral circulation using our 115-patient cohort. Expression levels of hsa-miR-105 were significantly decreased in the plasma of patients with PTX-resistant EOC, compared with those from patients with PTX-responsive EOC (P<0.0001; Fig. 6D). The contingency plot indicated that high levels of plasma hsa-miR-105 were associated with improved responsiveness to PTX (Fig. 6E). Subsequent ROC curve analysis revealed that the corresponding AUC was 0.8797 (95% CI, 0.8169–0.9424) (Fig. 6F), verifying the predictive power of the circulating hsa-miR-105. These clinical data support our pre-clinical findings and show that downregulation of hsa-miR-105 occurs in both tumor and peripheral circulation in the EOC patients with acquired PTX resistance.