MicroRNA-625-3p Increases Chemosensitivity in Ovarian Cancer Cells Through Decreasing SSX2IP-Mediated Cisplatin Export in Extracellular Vesicles

Simple Summary

Around thirty percent of high-grade serous ovarian cancer patients are not responsive to first-line platinum-based chemotherapy. Their diseases typically do not respond or progress during treatment or recur within 6 months of completing treatment, suggesting the intrinsic characteristics of the tumor. There are currently no biomarkers that can predict intrinsic resistance. MicroRNAs are known to provide a master layer of regulation of gene expression, however, the molecular mechanisms by which microRNAs confer a chemoresistant phenotype in ovarian cancer have not been elucidated. We found that miR-625-3p is significantly downregulated in chemoresistant cases and is associated with poorer overall and progression-free survivals. Further functional studies showed that the overexpression of miR-625-3p significantly decreased cisplatin resistance in ovarian cancer cells. Moreover, we identified SSX2IP as a direct target of miR-625-3p. SSX2IP plays a critical role in controlling microtubule length and orientation. We demonstrated that SSX2IP confers cisplatin resistance through the facilitation of the export of cisplatin in extracellular vesicles by promoting microtubule-dependent vesical trafficking. This study is crucial for developing new predictive biomarkers for intrinsic chemoresistance, and new treatment strategies for high-grade serous ovarian cancer based on upregulating miR-625-3p or downregulating SSX2IP expression in ovarian cancer cells, which will enhance cisplatin sensitivity and improve patient survival rates.

Abstract

Introduction: Advanced-stage high-grade serous ovarian cancer (HGSC) is a disease that is difficult to manage due to its heterogeneous clinical behavior. No reliable prediction of response to chemotherapy is currently available and the overall survival rate remains poor. Herein, we sought to determine the molecular mechanisms by which microRNAs (miRNAs) confer chemoresistance in ovarian cancer and demonstrate the efficacy of targeting miRNAs to sensitize HGSC to cisplatin treatment. Methods: Next-generation miRNA sequencing was performed using microdissected HGSC specimens to identify an miRNA signature for intrinsic chemoresistance, and miR-625-3p was selected for further study. The effects of miR-625-3p on cisplatin sensitivity were evaluated using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays and cell death enzyme-linked immunosorbent assay. Transcriptome profiling analysis, online prediction algorithms, and reporter assays were used to demonstrate SSX2IP as the direct gene target of miR-625-3p. Cell death enzyme-linked immunosorbent assays, mass spectrometry, and high-speed confocal microscopy were used to determine the roles of SSX2IP in mediating the effects of miR-625-3p in cisplatin sensitivity via the extracellular vesicle (EV) secretion of cisplatin. Results: An miRNA signature for intrinsic chemoresistance was identified. Amongst all the downregulated miRNAs in the chemo-refractory samples, only miR-625-3p was associated with poorer overall survival and progression-free survival rates. Further functional studies showed that the overexpression of miR-625-3p significantly decreased cisplatin resistance in ovarian cancer cells both in vitro and in vivo. SSX2IP (Synovial Sarcoma, X Breakpoint 2 Interacting Protein) was confirmed to be the direct gene target of miR-625-3p and its upregulation abrogated miR-625-3p-mediated cisplatin resistance by enhancing the EV export of cisplatin in ovarian cancer cells. Conclusions: These findings provide a new paradigm for intrinsic cisplatin resistance acquisition by HGSC cells, which will be crucial for developing new treatment strategies for ovarian cancer based on the upregulation of miR-625-3p or downregulation of SSX2IP to enhance cisplatin sensitivity and improve patient survival rates.

1. Introduction

Epithelial ovarian cancer is the fifth leading cause of cancer-related death among women and has the highest case-fatality rate among gynecologic cancers [1]. In the United States, about 20,000 new cases of ovarian cancer and 14,000 ovarian cancer-related deaths were estimated for 2023 [1]. Epithelial ovarian cancer is a histologically, clinically, and molecularly diverse disease for which there is a paucity of effective screening and therapeutic approaches [2,3,4,5,6]. High-grade serous ovarian cancer (HGSC) is the most common form of ovarian cancer. The vast majority of HGSC cases are diagnosed at late disease stages, limiting the opportunity for clinical intervention and resulting in a 10-year survival rate lower than 20%. The current standard of care includes primary debulking surgery followed by at least 6 cycles of platinum- and taxane-based combination chemotherapy or neoadjuvant chemotherapy followed by interval debulking surgery. Although upfront platinum or paclitaxel chemotherapy achieves a complete response, defined as progression-free survival 6 months after the initial treatment, for most women with HGSC who undergo the treatment, around 30% of patients do not have a complete response to initial treatment with surgery and chemotherapy and die within 1 year of diagnosis. Among these patients with incomplete response, HGSC either does not respond or progresses during treatment (i.e., refractory disease) or recurs within 6 months of treatment completion (i.e., resistant disease) [7]. Because most of the patients with incomplete response have undergone optimal therapy (i.e., therapy with optimal cytoreductive surgery and at least 6 cycles of platinum-based chemotherapy), the lack of response to upfront chemotherapy is due to the inherit or intrinsic characteristics of the tumor rather than incomplete clinical treatment.

Recently, gene expression profiling and integrated genome-wide DNA screening using microarray, DNA-Seq, and RNA-Seq platforms have accelerated the discovery of genes with altered expression in HGSC [2,3,4,8]. Nevertheless, research on the biological and clinical relevance of the altered expression of certain genes or biomarkers to ovarian cancer chemoresistance remains scanty. Furthermore, a clinically useful interpretation of a transcriptome-based signature that guides chemoresistance in HGSC is still needed.

MicroRNAs (miRNAs) provide a novel master layer of regulation for gene expression and are expected to be a powerful tool to obtain more representative molecular portraits of a tumor’s specific characteristics and behaviors at diagnosis [9]. Recently, studies identified miRNA signatures for different subtypes of HGSC [4,8,10,11] and for chemoresistance in HGSC [11,12,13]. However, most of these studies used bulk ovarian tumor tissue samples with various amounts of stromal tissue contamination, which can decrease the accuracy of the cancer-cell specific miRNA profiles. Additionally, the molecular mechanisms by which these prognostic and predictive miRNAs confer intrinsic or induced chemoresistance have not been elucidated. Furthermore, the clinical relevance and usefulness of most of these miRNAs for the prediction of HGSC chemoresistance before treatment have not been defined.

In the current study, we used next generation sequencing and RNAs isolated from microdissected tumor tissue samples to identify an miRNA signature associated with intrinsic chemoresistance in HGSC. We characterized the clinical significance of one of the miRNAs (miR-625-3p) and delineated the molecular mechanism by which its direct target (SSX2IP) mediated the effect of the miRNA on cisplatin resistance.

2. Methods

2.1. Microdissection of Tissue Samples

Treatment-naïve HGSC tissue samples (n = 94), including 12 chemo-refractory (i.e., tumors not responsive to initial chemotherapy), 30 chemo-resistant (i.e., tumors with relapsed disease within 6 months after the last cycle of chemotherapy), and 52 chemo-sensitive (i.e., tumors with relapsed HGSC 6 months after the last cycle of chemotherapy) samples were obtained from The University of Texas MD Anderson Cancer Center Department of Gynecologic Oncology and Reproductive Medicine’s ovarian cancer repository under protocols approved by MD Anderson’s institutional review board. Informed consent was obtained from all patients. Clinical characteristics of patients are listed in Supplementary Table S1. Epithelial components of ovarian tumor tissues were obtained by microdissection by fixing tissue sections in 70% ethanol and staining them with 1% methyl green to visualize the histologic features. During microdissection, the areas of interest in the sections were carefully outlined, and areas with immune cell and blood vessel infiltration were excluded to minimize contamination.

2.2. miRNA Sequencing Using Ion Torrent

Total RNA was extracted from microdissected chemo-refractory (n = 10) and chemo-sensitive (n = 15) HGSC tumor tissues using the RNAqueous-Micro Kit (Thermo Fisher Scientific, Waltham, MA, USA), and the amount and quality of small RNA in the total RNA samples were determined using an Agilent Small RNA Kit (Agilent Technologies, Santa Clara, CA, USA) on an Agilent 2100 Bioanalyzer (Agilent Technologies). Then, 5 ng of small RNAs were subjected to library construction using the Ion Total RNA-Seq Kit v2 (Life Technologies Corp., Carlsbad, CA, USA) as described previously [14]. Emulsion PCR and ion sphere particle enrichment was performed using an Ion Xpress Template Kit (Life Technologies Corp.) as described previously [14], and samples were prepared for sequencing using the Life Technologies Corp.’s Ion Sequencing Kit protocol. The complete sample was loaded on an Ion 314 chip (Thermo Fisher Scientific) and sequenced on the Ion Torrent Personal Genomic Machine (Thermo Fisher Scientific) for 65 cycles.

Ion Torrent reads were collected and sorted according to barcodes using Ion Torrent Suite software (Thermo Fisher Scientific). The software scores the quality of the reads by assigning Q17 and Q20 scores according to the quality scoring computation. Fastq files from the Ion Torrent server were imported to the CLC Genomics Workbench (version 5.1; Life Technologies Corp.). Sequence reads were trimmed to remove reads with ambiguous nucleotides and to limit sequence lengths to 15 to 55 nucleotides. Trimmed sequence reads were aligned to a miRNA database (miRBase, release 18; The University of Manchester, Manchester, UK). Mapped reads were annotated and counted. The median number of reads was 182,000 for each sample. Ion Torrent reads were then normalized to the number of cells in each sample and presented as the total reads per 1000 cells.

2.3. Quantitative Reverse Transcription-Polymerase Chain Reaction (RT-PCR) Analysis

Total RNA was extracted from cultured cells using TRI reagent (Molecular Research Center, Cincinnati, OH, USA) or from tissue samples using the RNAqueous-Micro Kit (Thermo Fisher Scientific).

For the miR-625-3p expression analysis, first, 5 ng of total RNA was reverse transcribed using the TaqMan microRNA Reverse Transcription Kit (Thermo Fisher Scientific) and miR-625-3p (#002432) and U6 (#001973) snRNA-specific primers and probes (Thermo Fisher Scientific) and then quantified using real-time PCR on a CFX96 Touch real-time PCR detection system (Bio-Rad Laboratories Inc., Hercules, CA, USA). Then, the relative standard curve method (2^−ΔΔCt) was used to determine the relative miR-625-3p expression using U6 snRNA as the reference.

For the mRNA expression analysis, 0.5 μg of total RNA was used to synthesize the first strand of cDNA using the ImProm-II Reverse Transcription System (Promega Corp., Madison, WI, USA). The expression of SSX2IP was determined by multiplexing quantitative PCR (TaqMan Gene Expression Assay, Thermo Fisher Scientific) using FAM-labeled SSX2IP (Hs00907875_m1) and VIC-labeled GAPDH-specific TaqMan probes and primers (Thermo Fisher Scientific). The relative standard curve method (2^−ΔΔCt) was used to determine the relative mRNA expression, using GAPDH as the reference.

2.4. Cell Lines and Culture Conditions

Human ovarian cancer lines A224 and ALST (which were gifts from Dr. Michael Birrer’s laboratory at Massachusetts General Hospital), CaOV3, OV90, OVCAR3, OVCAR8 (American Type Culture Collection, Manassas, VA, USA), OVCA420, OVCA432, and OVCA433 (gifts from Dr. Robert Bast’s laboratory at MD Anderson Cancer Center) were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and penicillin-streptomycin (Thermo Fisher Scientific). Human ovarian cancer cell lines PEO1 (established before the development of clinical resistance), PEO4 (established after the development of clinical resistance to chemotherapy), and PEO6 (established after the development of clinical resistance to chemotherapy at a more advanced stage) (European Collection of Authenticated Cell Cultures, Salisbury, UK) were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 nM glutamine, penicillin/streptomycin, and 1% sodium pyruvate (Sigma-Aldrich Co., St. Louis, MO, USA). Human ovarian surface epithelial cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM glutamine, penicillin-streptomycin, and 1 ng/mL epidermal growth factor (Thermo Fisher Scientific). All cell lines had negative testing results for mycoplasma contamination and were authenticated by short tandem repeat profiling in the Cytogenetics and Cell Authentication Core at The University of Texas MD Anderson Cancer Center.

2.5. Establishment of Stable Transfectants

OVCA433 and A224 stable cell lines overexpressing miR-625-3p and SSX2IP were established by transfection with the miR-625 stem-loop and SSX2IP open reading frame cDNA clone (GeneCopoeia Inc., Rockville, MD, USA), respectively, and selected using puromycin. The positive clones were identified using quantitative RT-PCR and Western blot analyses. All stable transfectants were cultured in complete RPMI medium supplemented with 200 μg/mL puromycin (MilliporeSigma, Burlington, MA, USA).

2.6. 3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide (MTT) Assay

OVCA433 and A224 stable transfectants were seeded in a 96-well plate and treated with cisplatin for 72 h. After that, the cells were incubated with 50 μL of 1 mg/mL MTT (Sigma-Aldrich Co.) in phosphate-buffered saline (PBS) for 3 h. The formazan that formed was then solubilized by adding 150 μL of dimethyl sulfoxide. The absorbance was read at 570 nm using a FLUOstar Galaxy plate reader (BMG Labtech, Offenburg, Germany).

2.7. Apoptosis Assay

Apoptosis in OVCA433 and A224 parental and SSX2IP-overexpressed stable cells treated with cisplatin was measured using the Cell Death Detection ELISA^PLUS Kit (Sigma-Aldrich Co.) as described previously [14]. The data were expressed as the relative percentage of apoptosis.

2.8. Effect of miR-625-3p on Cisplatin Sensitivity In Vivo

Luciferase-labeled A224 cells (2 × 10⁶) were injected once intraperitoneally into 6-week-old female BALB/c athymic nude mice to establish tumors. After randomization, miR-625-3p mimics and control miR mimics were delivered into mice intraperitoneally using in vivo-jetPEI as a delivery vehicle twice weekly followed by cisplatin treatment once weekly for 6 weeks. Tumor volumes were measured and quantified using the IVIS-Lumina XR in vivo imaging system (Caliper Life Sciences, Inc., Hopkinton, MA, USA). Mice were euthanized after the 6-week treatment using a carbon dioxide chamber followed by cervical dislocation. Tumor tissues were fixed in formalin and processed for immunohistochemistry.

The described animal procedures were reviewed and approved by the institutional animal care and use committee of MD Anderson. All experiments were performed in accordance with the relevant guidelines and regulations.

2.9. Microarray Analysis

Total RNAs from mock or miR-625-3p-transfected A224 cells were extracted using a TRI reagent (Molecular Research Center). Purified RNA samples were amplified, labeled, and hybridized onto Affymetrix GeneChip Human Genome U133 Plus 2.0 microarrays (Affymetrix Inc., Santa Clara, CA, USA) according to the manufacturer’s protocol. The arrays were then washed and stained using the Affymetrix Fluidics Station 450 (Affymetrix Inc.). The arrays were scanned by the Cancer Genomics Laboratory at MD Anderson using a GeneChip Scanner 3000 7G (Affymetrix Inc.). A heat map was generated using dChip software (Affymetrix Inc.).

2.10. Western Blot Analysis

Cell extracts were prepared in radioimmunoprecipitation assay buffer (20 mM sodium phosphate, 150 mM NaCl pH 7.4, 1% Nonidet P-40, 0.1% SDS, and 0.5% deoxycholic acid) containing a complete protease inhibitor cocktail (Roche Diagnostics Corp., Indianapolis, IN, USA). Proteins were separated on SDS–polyacrylamide gels and electrophoretically transferred to an immobilon polyvinylidene fluoride membrane (EMD Millipore, Billerica, MA, USA). The membranes were incubated with Anti-SSX2IP (1:100; HPA027306; Sigma-Aldrich Co.), anti-CD63 (1:1000; NBP2-42225; Novus Biologicals, Centennial, CO, USA), and β-actin (loading control, 1:5000; clone AC-15, Sigma-Aldrich Co.) overnight at 4 °C and then incubated with the appropriate horseradish peroxidase-conjugated secondary antibodies at a 1:10,000 dilution (Thermo Fisher Scientific) for 1 h at ambient temperature. Signals were developed using enhanced chemiluminescence detection reagents (Denville Scientific Inc., Holliston, MA, USA) and visualized on X-ray film (Fujifilm, Tokyo, Japan).

2.11. Immunohistochemical Analysis

Immunolocalization of SSX2IP was performed using formalin-fixed, paraffin-embedded ovarian tumor sections obtained from mice treated with miR-625-3p mimics or control miR mimics with cisplatin or from patients with chemo-sensitive or chemo-refractory HGSC. Slides containing the sections were stained with commercially available anti-SSX2IP (1:100; HPA027306; Sigma-Aldrich Co.). Target protein expression in the mice and human tumor sections was visualized using a Betazoid 3,3′-diaminobenzidine and Warp Red chromogen kit (Biocare Medical, Concord, CA, USA), respectively. To quantify target protein expression, slides were scored according to the staining intensity and percentage of SSX2IP-positive cells. SSX2IP tumor expression scores were calculated as the products of staining intensity and percentage of SSX2IP-positive staining.

2.12. Luciferase Reporter Assay

OVCA433 and A224 cells were seeded onto 24-well plates 1 day before transfection. The cells were transfected with the SSX2IP 3′ untranslated region (UTR) reporter construct containing the predicted miR-625-3p binding sites and 5 nM miR-625-3p mimics using Lipofectamine 2000 (Thermo Fisher Scientific). The cells were washed and lysed with the passive lysis buffer from the Dual-Luciferase Reporter Assay System (Promega Corp.) 24 h after transfection. Luciferase activity was measured in each cell lysate using a FLUOstar Galaxy plate reader. Relative luciferase activity was first normalized with Renilla luciferase activity and then compared to the luciferase activities of the respective control.

2.13. Measurement of Intracellular Platinum Level

Platinum accumulation in SSX2IP-overexpressed OVCA433 and A224 stable transfectants was quantified by flameless atomic absorption spectrometry, essentially as described previously [15]. Briefly, cells were first exposed to 100 µM cisplatin for 4 h and washed twice in ice-cold PBS. These were then harvested using trypsin, and the cell pellets were resuspended in 1 mL ice-cold PBS and split into two fractions, 0.1 mL of which was used for protein quantification, and the remaining 0.9 mL was digested overnight in benzethonium hydroxide (B2156, Sigma-Aldrich Co.) at 55 °C, acidified with hydrochloric acid, and subjected to platinum content quantification by flameless atomic absorption spectrometry. The data were normalized to the protein concentration and expressed as the amount of platinum per mg protein (ng Pt/mg protein).

2.14. Isolation of EVs

SSX2IP-overexpressed OVCA433 and A224 stable transfectants were washed with PBS and incubated with freshly prepared complete medium containing EV-free fetal bovine serum for 48 h. EVs were isolated from the conditioned medium by differential centrifugation. Briefly, the conditioned medium was centrifuged at 300 g for 10 min and then at 2000 g for 20 min at 4 °C to remove cells, followed by filtration through a 0.22-μm filter to remove cell debris. EVs were pelleted by ultracentrifugation at 100,000 g for 90 min. These were resuspended in PBS and collected by ultracentrifugation again at 100,000 g for 90 min. The presence of EVs was confirmed with the anti-CD63 antibody (Supplementary Figure S1A). The size and concentration of the EVs (Supplementary Table S2) were quantified by qNano (Izon Science, Christchurch, New Zealand).

2.15. Measurement of EV Platinum Content

SSX2IP-overexpressed OVCA433 and A224 stable transfectants, with or without nocodazole, an inhibitor of the tubulin polymerization process, were treated with 2 μM cisplatin for 1 h at 37 °C to load cells with the drug. These were then washed with PBS before being incubated with complete medium containing EV-free fetal bovine serum for 2 h at 37 °C to collect EVs. EVs were isolated from the conditioned medium by differential centrifugation as described above. Resuspended EVs were first mixed with aqua regia (1:3 molar ratio solution of nitric acid to hydrochloric acid) and then diluted with a 2% hydrochloric acid solution and incubated at room temperature overnight. Agilent 7900 inductively coupled plasma-mass spectrometry instrument (Agilent Technologies, Inc.) was used to measure the platinum concentration. The platinum content was normalized to the EV protein content and expressed as parts per billion platinum per µg protein (ppb Pt/µg protein).

2.16. Isolation of Microtubule

To stabilize the microtubules, SSX2IP-overexpressed OVCA433 and A224 stable transfectants were treated with 20 mM sodium azide in Dulbecco PBS (Thermo Fisher Scientific) at 37 °C for 2 h to. Cells were then incubated with microtubule-stabilizing buffer (0.1 M PIPES, pH 7.1, 1 mM MgSO₄, 1 mM EGTA, and 10% glycerol) with 0.5% Triton X-100 and complete protease inhibitor cocktail (Roche Diagnostics Corp.) to extract the monomer fraction in the supernatant. The polymer fraction was solubilized and collected in microtubule-stabilizing buffer with 6× Laemmli-reducing SDS buffer for Western blot analysis.

2.17. Visualization of EV Trafficking

CD63-pEGFP and CellLight tubulin-RFP (Thermo Fisher Scientific) were co-expressed in SSX2IP-overexpressed OVCA433 stable transfectants before live cell imaging. High-speed 4-dimensional confocal images (XYZT, sampling every 10 s) of the SSX2IP-overexpressed cells and control stable cells were collected with an Andor Dragonfly 505 confocal system equipped with a super-resolution module (Oxford Instruments, Abingdon, UK). Trafficking paths of 274 CD63⁺ vesicles were extracted from 61 SSX2IP-overexpressed OVCA433 stable transfectants, and those of 367 CD63⁺ vesicles were extracted from 43 control stable cells. The traveling speed and displacement of individual vehicles, total volume of vehicles, and microtubules per cell were determined using Imaris image analysis software (Oxford Instruments) from the time series of reconstructed 3-dimensional images.

2.18. Statistical Analysis

SPSS software, version 23 (IBM Corp. Armonk, NY, USA), was used to perform statistical tests. Data are presented as the mean ± the standard deviation unless otherwise specified. A two-tailed Student t-test was used to test differences in sample means for data with normally distributed means. The Mann–Whitney U test was used for non-parametric data as appropriate. Correlation between variables was determined using a Spearman correlation analysis. A p-value of <0.05 was considered to be statistically significant.

3. Results

3.1. An miRNA Signature Predictive for Chemoresistance in HGSC Patients

To identify miRNAs associated with intrinsic chemoresistance, we performed next-generation Ion Torrent miRNA-sequencing analysis on RNA isolated from microdissected chemo-refractory samples (n = 10) and microdissected chemo-sensitive samples (n = 15) of treatment-naïve advanced HGSC. We identified an miRNA signature with a set of miRNAs and their variants, which showed significant differential copy numbers in the chemo-refractory samples compared to the chemo-sensitive samples (Figure 1A,

Table 1. Top differentially expressed miRNAs identified in chemo-sensitive and chemo-refractory HGSC patients by Ion Torrent next generation sequencing.

MicroRNA	Type	p-Value	Ratio (Refractory/Sensitive)
miR-451a	mature 5′ super variant	0.01	6.25
miR-451a	mature 5′ variant	0.002	3.7
miR-451a	mature 5′ super	0.035	3.45
miR-126	mature 3′ sub/super variant	0.031	2.7
miR-145	mature 5′ super variant	0.035	2.44
miR-345	mature 5′ variant	0.04	0.62
miR-93	mature 5′ super variant	0.035	0.6
miR-590	mature 5′	0.031	0.57
miR-625	mature 3′	0.046	0.5
miR-106b	mature 3′ sub variant	0.035	0.49
miR-135b	mature 5′	0.006	0.4
miR-24-1/miR-24-2	precursor	0.035	0.36

). We selected miR-625-3p for further studies, as quantitative RT-PCR analysis on the same set of microdissected specimens confirmed significantly lower levels of miR-625-3p in the chemo-refractory samples than in the chemo-sensitive samples (Figure 1B,C). Further validation was performed using RNA isolated from an independent set of 94 microdissected chemo-refractory (n = 12), chemo-resistant (n = 30), and chemo-sensitive (n = 52) samples. There was a significant decrease in miR-625-3p in the refractory samples compared to those in the resistant and sensitive samples (p = 0.022 and 0.005, respectively; Figure 1D, Supplementary Table S1).

Moreover, using the median miR-625-3p expression level as a cutoff, lower miR-625-3p levels were significantly associated with poorer overall and progression-free survival rates in such patients (p = 0.0001 and 0.007, respectively; Figure 1E,F), as determined by Kaplan–Meier analyses and log-rank tests. In addition, multivariate Cox regression analysis confirmed the significant association between miR-625-3p levels and overall (hazard ratio [HR] = 2.69, 95% confidence interval [CI] = 1.15–2.85, p = 0.001) and progression-free survival rates (HR = 1.82, 95% CI = 1.15–2.85, p = 0.01) after adjusting for age and debulking status (Supplementary Table S3).

Compared to that of cell line PEO1, which was established from a patient with ovarian cancer before the development of clinical resistance, the miR-625-3p levels in the chemo-resistant cell lines PEO4 and PEO6, which were established from a patient with ovarian cancer after they developed clinical resistance to chemotherapy at different time points, were significantly lower (Figure 1G).

3.2. MiR-625-3p Enhances Cisplatin Sensitivity in Ovarian Cancer Cells In Vitro

The results from the clinical correlation study suggested that miR-625-3p might be a favorable predictive marker for chemoresistance and cancer progression. To evaluate the functional roles of miR-625-3p in modulating chemoresistance in ovarian cancer cells OVCA433 and A224, HGSC cell lines with low expression levels of endogenous miR-625-3p (Supplementary Figure S1B) were stably transfected with miR-625-3p mimics or control miR mimics (Supplementary Figure S1C). Their effects on cisplatin resistance as well as cisplatin-induced apoptosis were evaluated.

While the results showed that increased miR-625-3p levels did not affect cell growth and apoptosis in both cell lines (Supplementary Figure S1D,E), upregulation of miR-625-3p significantly decreased the number of viable cells and increased the percentage of apoptosis when they were treated with cisplatin, which suggests that miR-625-3p enhances cisplatin sensitivity in ovarian cancer cells (Figure 2A,B). Furthermore, the downregulation of miR-625-3p by transfection with the anti-miR-625-3p inhibitor in the HGSC cell lines OVCA432 and PEO1 (Supplementary Figure S1F), which express high levels of endogenous miR-625-3p, resulted in reduced cisplatin sensitivity (Supplementary Figure S1G). Again, decreased miR-625-3p levels did not affect cell growth in both cell lines (Supplementary Figure S1H). This confirmed the biological role of miR-625-3p in enhancing cisplatin sensitivity.

3.3. MiR-625-3p Potentiates the Effect of Cisplatin in Suppressing Ovarian Cancer Growth In Vivo

To evaluate the effect of miR-625-3p on cisplatin resistance in vivo, miR-625-3p mimics or control miR mimics were delivered into luciferase-labeled A224-bearing aythmic mice intraperitoneally using in vivo-jetPEI as a delivery vehicle followed by cisplatin treatment. Tumor growth was monitored using the IVIS-Lumina XR in vivo imaging system. Our results showed that the luciferase activity was significantly lower in mice treated with both miR-625-3p mimics and cisplatin compared to those treated with control miR mimics and cisplatin (p = 0.021, Mann-Whitney U test; Figure 2C–E). Treatment was continued until one group of mice became moribund, at which time all the animals were euthanized and necropsied. Tumor weight was determined at the time of dissection. Tumor weight was significantly smaller in mice treated with both miR-625-3p mimics and cisplatin compared to those treated with control miR mimics and cisplatin (p = 0.0001; Supplementary Figure S1I). These data suggest that upon cisplatin treatment, miR-625-3p may sensitize ovarian cancer cells.

3.4. SSX2IP Is a Direct Gene Target of miR-625-3p

To delineate the molecular mechanisms by which miR-625-3p modulates chemoresistance in ovarian cancer cells, A224 cells transiently transfected with miR-625-3p mimics or control miR mimics were analyzed using transcriptome profiling. A total of 10 genes, which demonstrated more than a 5-fold decrease in the expression in miR-625-3p-transfected cells compared to those transfected with control mimics, were identified (Figure 3A,

Table 2. Significantly downregulated mRNAs identified in miR-625-3p overexpressed A224 cells by transcriptome profiling.

Fold Change	Gene Symbol
−8.193928	GALNT1
−7.1412263	MYCBP
−6.738094	CDH6
−6.501197	KCNJ16
−6.2234616	PHTF2
−5.813403	COMMD8
−5.8041797	TGFBR1
−5.7514067	SSX2IP
−5.6314034	TMEM64

). By integrating the results from the transcriptome profiling analysis and online miRNA target prediction algorithms, including TargetScan, miRanda, and RNA22 [16,17], we identified Synovial Sarcoma X breakpoint 2 interacting protein (SSX2IP) as a potential direct gene target of miR-625-3p. A consensus miR-625-3p binding site was identified within the 3′ UTR of SSX2IP mRNA with seed sequence spans from 2666 to 2671 base pairs. Using immunofluorescence analysis, we found that SSX2IP was co-localized with PCM1, a protein that is associated with the centrosome complex (Supplementary Figure S2A).

To validate SSX2IP as a direct target of miR-625-3p, we first performed quantitative RT-PCR and Western blot analyses on miR-625-3p-transfected OVCA433 and A224 cells. The results showed a significant decrease in SSX2IP mRNA and protein expressions in both cell lines transfected with miR-625-3p compared to those transfected with control miR mimics (Figure 3B,C). However, OVCA432 and PEO1 cells transfected with an miR-625-3p inhibitor showed a significant increase in SSX2IP mRNA and protein expression compared to those transfected with control miR inhibitors (Supplementary Figure S2B,C). Immunolocalization of SSX2IP on tumor tissues collected from mice treated with both miR-625-3p mimics and cisplatin, and from those treated with control miR mimics and cisplatin, showed that SSX2IP protein expression was significantly lower in the presence of miR-625-3p mimics (Figure 3D). In addition, miR-625-3p and SSX2IP mRNA levels in HGSC patient samples as measured by quantitative RT-PCR analysis were significantly inversely correlated (p = 0.043), suggesting that SSX2IP is regulated by miR-625-3p (Supplementary Figure S2D).

Moreover, we examined whether miR-625-3p regulates SSX2IP mRNA expression through direct binding to the predicted binding site on the 3′ UTR of SSX2IP. The 3′ UTR of SSX2IP containing the predicted miR-625-3p recognition site (5′-TCACACAACCATAAGTATAGTA-3′) was cloned and placed in the pMIR-REPORT miRNA expression reporter vector. OVCA433 and A224 cells transfected with the reporter construct were co-transfected with either miR-625-3p mimics or control miR mimics, as well as the Renilla luciferase construct for normalization. Luciferase activity was measured by the dual-luciferase reporter assay. The results showed that the relative luciferase activity of the reporter construct was significantly reduced in cells transfected with miR-625-3p mimics compared to those transfected with control miR mimics (Figure 3E), suggesting that SSX2IP is the direct downstream gene target of miR-625-3p.

3.5. SSX2IP Confers Chemoresistance in Ovarian Cancer Cells

To evaluate the roles of SSX2IP in conferring chemoresistance in ovarian cancer cells, the effects of SSX2IP on cisplatin resistance and cisplatin-induced apoptosis on ovarian cancer cells were examined. OVCA433 and A224 cells were stably transfected with a construct containing full-length SSX2IP cDNA or a control vector before cisplatin treatment. SSX2IP overexpression was confirmed by quantitative RT-PCR and Western blot analyses (Supplementary Figure S2E,F). Compared to control cells, the number of viable SSX2IP-overexpressed OVCA433 and A224 cells was significantly higher and apoptosis was significantly lower in such cells, suggesting that SSX2IP confers chemoresistance in ovarian cancer cells (Figure 4A,B). In addition, SSX2IP expression was significantly higher (p = 0.019) in the chemo-refractory (n = 10) HGSC samples than in the chemo-sensitive (n = 10) HGSC samples, as indicated by the immunolocalization of SSX2IP (Figure 4C,D).

To evaluate the clinical significance of SSX2IP, SSX2IP mRNA expression levels in microdissected HGSCs were determined using quantitative RT-PCR analysis, and a survival analysis was performed. High SSX2IP levels (>mean SSX2IP level) were associated with a significant decrease in the overall survival rates for patients with HGSC (p = 0.028; Figure 4E). Similar to the increased miR-625-3p expression levels, increased SSX2IP expression levels did not affect cell growth and apoptosis in both cell lines (Supplementary Figure S2G).

3.6. SSX2IP Mediates miR-625-3p-Modulated Chemoresistance in Ovarian Cancer Cells

To determine whether the upregulation of SSX2IP can abrogate miR-625-3p-mediated cisplatin resistance in ovarian cancer cells, we transfected OVCA433 cells with miR-625-3p mimics and full-length SSX2IP cDNA or an empty vector. SSX2IP abrogated cisplatin resistance (Figure 4F). Additionally, the upregulation of SSX2IP also rescued OVCA433 cells from cisplatin-induced apoptosis (Supplementary Figure S3A). Together, these data suggest that SSX2IP mediates the role of miR-625-3p in cisplatin resistance in ovarian cancer cells.

3.7. SSX2IP Enhances the EV Export of Cisplatin in Ovarian Cancer Cells

Since chemotherapeutic agents such as cisplatin are present in EVs secreted from cancer cells [18,19], we sought to determine whether SSX2IP-overexpressed ovarian cancer cells had a significantly lower amount of therapeutic agent compared to mock transfectants. According to atomic absorption spectroscopy, SSX2IP-overexpressed OVCA433 and A224 cells had a significantly lower amount of intracellular cisplatin than the mock transfectants after 2 h of cisplatin treatment (Figure 5A). Meanwhile, there was a significantly higher number of EVs secreted in SSX2IP-overexpressed OVCA433 and A224 cells compared to the mock transfectants as measured using qNano analysis (Figure 5B). We further examined the total cisplatin amount in EVs secreted from SSX2IP-overexpressed OVCA433 and A224 cells and found that EV cisplatin in SSX2IP-overexpressed OVCA433 and A224 cells was significantly higher than in the mock transfectants, as measured using inductively coupled plasma-mass spectrometry (Figure 5C). These results suggest that SSX2IP confers cisplatin resistance in ovarian cancer cells via enhanced EV export of cisplatin.

To confirm the role of the EV export of cisplatin in chemoresistance in ovarian cancer cells, we treated SSX2IP-overexpressed A224 cells with GW4869, which is a neutral sphingomyelinase inhibitor for blocking EV generation. We found that GW4869 better counteracted the effects of SSX2IP on cisplatin-induced apoptosis than the mock transfectants (Supplementary Figure S3B).

3.8. SSX2IP Facilitates EV Export of Cisplatin Through the Promotion of Microtubule Formation

Next, we determined whether increased SSX2IP expression promoted microtubule formation in ovarian cancer cells. Using Western blotting, we demonstrated that SSX2IP-overexpressed OVCA433 and A224 cells had significantly more polymerized tubulin in the microtubule fraction than the mock transfectants (Figure 6A). Also, microtubule volume was significantly higher in the SSX2IP-overexpressed OVCA433 cells than in the mock transfected cells in the 3-dimentional cell model reconstructed from optical sections using Imaris (n = 61; Supplementary Figure S3C). Additionally, the overexpression of SSX2IP (n = 25) resulted in the increased accumulation of SSX2IP in the microtubule-organizing center (MTOC), as identified by γ-tubulin in OVCA433 cells compared to the mock transfectants (n = 29; Supplementary Figure S3D). These results suggest that SSX2IP promotes microtubule formation in ovarian cancer cells.

To investigate whether SSX2IP facilitates the EV export of cisplatin through the promotion of microtubule formation, SSX2IP-overexpressed OVCA433 and A224 cells were treated with cisplatin with or without nocodazole and the total amount of cisplatin in the EVs secreted from the cells was measured using inductively coupled plasma-mass spectrometry. The addition of nocodazole reduced the amount of EV cisplatin to a greater extent in the SSX2IP-overexpressed OVCA433 and A224 cells than in the mock transfectants (Figure 6B), suggesting that the inhibition of microtubule formation suppresses the EV export of cisplatin. Nocodazole treatment also counteracted the effects of SSX2IP on cisplatin-induced apoptosis in OVCA433 and A224 cells (Supplementary Figure S3E).

To further determine whether SSX2IP facilitates EV export of cisplatin through the promotion of microtubule-dependent vesicle trafficking, we examined whether CD63-labeled multivesicular bodies move along microtubules from the MTOC toward the periphery of OVCA433 cells using 4-dimentional confocal live cell imaging. We found that CD63-positive vesicles moved along the microtubules (Figure 6C), and such vesicles in the SSX2IP-overexpressed OVCA433 cells (n = 274) were significantly more displaced (p = 0.0001; Figure 6D) and traveled significantly faster (p = 0.0001; Figure 6E) than those in the control cells (n = 367) in a 5-minute sampling period. These findings suggest that enhanced microtubule-dependent vesicular trafficking mediates the effect of SSX2IP in facilitating the EV export of cisplatin in HGSC cells.

4. Discussion

In the current study, we identified an miRNA signature for intrinsic chemoresistance, demonstrated that the downregulation of miR-625-3p correlates with intrinsic cisplatin resistance, and identified poor patient survival rates. We also showed that miR-625-3p’s novel direct target, SSX2IP, mediated the effect of miR-625-3p in enhancing cisplatin sensitivity by decreasing microtubule-dependent microvesicle trafficking and the EV secretion of cisplatin, subsequently promoting cisplatin-induced apoptosis. The expression and roles of miR-625-3p, however, are conflicting in different cancers. It is upregulated in thyroid and colorectal cancers, while downregulated in non-small cell lung and gastric cancers [20]. In gastric cancer, miR-625-3p inhibits metastasis and reverse multidrug resistance [21,22], but in colorectal cancer, miR-625-3p suppresses apoptosis and promotes resistance to oxaliplatin [23].

Meanwhile, deregulated miRNAs and their target genes have profound impacts on various malignant phenotypes of ovarian cancer cells [24]. Multiple miRNAs have been shown to modulate chemoresistance in ovarian cancer cells and affect patient survival rates. For example, the miR-200 family, the miR-214/199a cluster, the let-7 family, miR-149-3p, and miR-874-3p have been shown to modulate chemosensitivity through targeting multiple signaling molecules associated with epithelial–mesenchymal transition or cisplatin-induced apoptosis [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44]. However, these miRNAs were not identified in our study, likely because most of these miRNAs are associated with adaptive resistance to cisplatin because they were previously identified using ovarian cancer cells treated with drugs in vitro rather than using treatment-naïve microdissected HGSC samples.

Unlike previously identified chemoresistance-associated miRNAs, which target genes associated with the signaling networks altered by cisplatin treatment in ovarian cancer, the novel miR-625-3p/SSX2IP axis confers chemoresistance by physically modulating cisplatin levels inside HGSC cells. Previous studies demonstrated that SSX2IP is a microtubule anchoring factor that promotes centrosome maturation and microtubule nucleation [45,46]. The depletion of SSX2IP leads to disorganized, shorter, and less intense interphase microtubules [45,47,48,49]. Microtubules are usually organized with their minus ends associated with the MTOC near the nucleus and radiate outward so that the plus ends are at the periphery of the cells. Dynamic instability, which is the transition between catastrophe and rescue, is the most characteristic dynamic behavior of microtubules and is involved in multiple cellular functions, including mitosis and cell signaling. In addition, microtubules and their associated motor proteins (i.e., dyneins and kinesins) are responsible for long-range intracellular transport and the positioning of membrane vesicles and organelles [50,51], and are involved in multivesicular endosome trafficking and EV release [52]. However, alterations in different components of the microtubules and their associated proteins have been shown to confer resistance to microtubule-targeting agents. These include the expression of different α- and β-tubulin isotypes, point mutations in tubulin, post-translational modifications to tubulins, altered expressions of tubulin regulatory proteins and microtubule-associated proteins, interactions with actin cytoskeleton, and activations of signaling pathways that interact with microtubules including JNK, PI3K/Akt, and MAPK [53,54,55]. Our data demonstrate that SSX2IP significantly increased the volume of the microtubules compared to mock transfectants in the 3-dimenional cell model reconstructed from optical sections (Imaris surface model). Western blot analysis also showed an increase in polymerized tubulin in the microtubule fraction of SSX2IP-overexpressed HGSC cells compared to that in mock transfectants. Overexpression of SSX2IP resulted in the increased accumulation of SSX2IP in MTOC (identified by γ-tubulin) in HGSC cells as expected. These findings suggest that SSX2IP enhances microtubule formation, which might mediate the effect of SSX2IP in conferring chemoresistance in HGSC cells, because microtubules are responsible for the intracellular transport and positioning of membrane vesicles and organelles toward the periphery of a cell [47].

While many studies, including our previous study [14], have demonstrated the role of EVs derived from stromal cells in conferring chemoresistance in cancer cells, the role of cancer cell-derived EVs in conferring resistance to therapeutic agents in cancer cells has not been reported. Safaei et al. previously demonstrated abnormal lysosomal trafficking and enhanced the EV export of cisplatin in drug-resistant human ovarian carcinoma cells [18,56]. The acidic microenvironment, which is a result of high extracellular lactate content, inadequate neovascularization in tumors, and increases in H⁺-ATPases, pumps protons across the plasma membrane of tumor cells, facilitating the sequestration and transportation of cytotoxic agents such as cisplatin into EVs [57]. In addition, EVs can also reduce extracellular drug levels by displaying bait targets for therapeutic antibodies on their surface, thus facilitating cancer progression by limiting drug availability. For example, EVs carry the cluster of differentiation 20 (CD20) receptor, which acts as a bait for therapeutic anti-CD20 antibodies such as rituximab. In breast cancer cells, the human epidermal growth factor receptor-2 (HER2) is found on the surface of EVs, resulting in sequestering of the therapeutic monoclonal antibody trastuzumab [57]. However, the roles of SSX2IP in facilitating the EV export of chemotherapeutic agents in cancer cells, and whether microtubules mediate the effect of SSX2IP EV secretion in HGSC cells, have not been previously explored.

Our data showed that HGSC cells transfected with the full-length SSX2IP construct had a significantly lower amount of intracellular cisplatin than the mock transfectant after drug treatment. In addition, the number of EVs secreted in SSX2IP-overexpressed HGSC cells was significantly higher than in the mock transfectants. These findings suggest that SSX2IP confers chemoresistance in ovarian cancer cells via enhanced EV export of cisplatin. Safaei et al. [56] also previously demonstrated enhanced EV export of cisplatin with more cisplatin loss from the cisplatin-resistant cancer cells compared to cisplatin-sensitive cancer cells. They also measured cellular and EV platinum content released from the cells.

Recent studies also showed that the growth of the microtubular structure is not limited to the plus end. Rather, it also depends on the attachment of the minus end to the centriole [58]. SSX2IP is a microtubule anchoring and centriolar satellite protein, and our data indicate that polymerized tubulin and microtubule volume were increased in cells overexpressing SSX2IP, therefore, SSX2IP may stabilize the peri-centriole structure, leading to increased tubulin nucleation (i.e., more microtubule arrays arising from the MTOC) and altered microtubule dynamics (i.e., a bias towards growth and stabilization) [48]. Increases in these two parameters can explain the increases in polymerized tubulin and microtubule volume and provides the structural base to enhance microtubule-dependent vesicular transport.

Our study, however, is limited to exploring only the role of EV export of cisplatin. Other mechanisms such as the role of microtubule dynamics and microtubule-associated signaling networks altered by chemotherapeutic agents, which may also enhance anti-apoptotic pathways and confer chemoresistance in HGSC cells, were not examined in the current study. Furthermore, the following are of interest: SSX2IP overexpression, which may increase cisplatin resistance in ovarian cancer cells by decreasing mitochondrial apoptosis via Bim-LC8 complex formation, and LC8, which binds to tubulin and promotes microtubule assembly and bundling [59], dimerizes Bim to form a large complex of Bim on the mitochondria outer membrane [60]. Anti-apoptotic proteins such as Bcl-2 and Mcl-1 recruited by Bim to these complexes are stabilized and protected from proteosomal degradation [60,61]. In addition to forming the Bim/LC8 complex, increased microtubule nucleation and abundance induced by SSX2IP may also modulate the activity of signaling molecules including JNK, P13K/Akt, and MAPK [53,54,55], which directly interact with microtubules and abrogate the pro-apoptotic effect of both cisplatin and paclitaxel.

Recently, Zhong et al. reported the involvement of miR-625-3p in chemoresistance in HGSC, and there are some discrepancies between their findings and ours [62]. They observed a significant increase in cell growth after miR-625-3p mimic transfection and a significant decrease after miR-625-3p inhibitor transfection without cisplatin treatment in both OVCAR3 and OVCAR4 cells. Without normalization, these changes in cell growth after cisplatin treatment may be misleading. In our study, cell proliferation was not significantly different after either miR-625-3p overexpression or miR-625-3p inhibition in the four HGSC cell lines that we examined. Moreover, they performed overexpression and silencing of miR-625-3p in the same cell lines without considering the intrinsic expression of miR-625-3p in those cells. These differences may potentially explain the conflicts between our studies.

5. Conclusions

We demonstrated that the downregulation of miR-625-3p is associated with the development of refractory HGSC and is a prognostic marker for patients with advanced HGSC. Moreover, SSX2IP is a direct target of miR-625-3p, which confers cisplatin resistance through the facilitation of the EV export of cisplatin by promoting microtubule-dependent vesicle trafficking. Further studies to delineate the molecular mechanism by which SSX2IP modulates chemoresistance in HGSC are warranted.