Arrestin Domain Containing 3 Reverses Epithelial to Mesenchymal Transition and Chemo-Resistance of TNBC Cells by Up-Regulating Expression of miR-200b

Our previous studies demonstrated the importance of arrestin domain containing 3 (ARRDC3), a metastasis suppressor, in inhibiting invasive and metastatic potential of triple negative breast cancer (TNBC) in vitro and in vivo. However, little is known about ARRDC3 mediated transcriptional control and its target genes that are implicated in its metastatic suppressing activity. In this study, we used miRNA array and subsequent functional analyses to identify miRNAs whose expression are significantly regulated by ARRDC3 in TNBC cells. We identified miR-200b as a major target gene of ARRDC3. miR-200b played an essential role in mediating ARRDC3 dependent reversal of EMT phenotypes and chemo-resistance to DNA damaging agents in TNBC cells. Expression of miR-200b also increased the expression of ARRDC3 as well in TNBC cells, suggesting a positive feedback loop between these two molecules. In addition, we combined the therapeutic powers of miR-200b and 5-fluorourancil (5-FU) into a single compound (5-FU-miR-200b) to maximize the synergistic effects of these compounds. Chemically modified miR-200b (5-FU-miR-200b mimic) was more effective in inhibiting metastatic potentials of TNBC cells than unmodified miR-200b and does not require transfection reagents, implying its therapeutic potential in TNBC. Our studies showed the importance of therapeutic targeting ARRDC3/miR-200b pathway in TNBC.


Introduction
Breast cancer is the most common type of cancer in women globally [1]. Although significant advances in the treatment and detection of breast cancer have been made in recent years, the survival rate of patients with metastatic breast cancer has been dropping [1,2]. Accounting for approximately 15-20% of all total breast cancer cases, triple negative breast cancer (TNBC) is one of most aggressive breast cancer subtypes, with poorer prognosis than ER/PR positive or Her2 overexpressed breast cancer due to a lack of targeted therapy [2][3][4]. The treatment of TNBC patients is limited to cytotoxic drugs such as 5-fluorourancil (5-FU), paclitaxel, doxorubicin, but chemoresistance is the major reason for failure of chemotherapy and mortality in TNBC [5][6][7]. Further, the molecular mechanism(s) involved in chemoresistance and strategies to resensitize chemoresistant cancer cells to chemotherapy remain to be determined in TNBC.
We previously demonstrated that expression of arrestin domain containing 3 (ARRDC3), a potential metastatic suppressor, is suppressed in metastatic TNBC cells due to epigenetic silencing [8]. Our subsequent study showed that either a forced expression of ARRDC3 or treatment of small molecule

miRNA Preparation and Transfection
Pre-miR precursor miR-200b-3p and Pre-miR negative control precursor (NC #2, miR-NC) were purchased from Ambion (Invitrogen). miRIDIAN miRNA-200b-3p mimics were purchased from Dharmacon (Lafayette, CO, USA). To generate chemically modified 5-FU-miR-200b mimics, single-strand RNAs modified with 5-fluorouridine in place of internal U were synthesized in Dharmacon. The 2-ACE protecting groups of RNA oligonucleotides were removed by deprotection reaction according to the manufacturer's protocol. The equimolar amounts of sense and antisense strand were mixed and annealed to form RNA duplex (called mimics). For miRNA transfection, cells were plated into 6-well plate one day before transfection. 100 nM miRNAs were transfected into MDA-MB-231 cells by using Oligofectamine (Invitrogen-Life Technologies, Carlsbad, CA, USA), according to the manufacturer's instructions. For oligofectamine free transfection, 100 nM miRNAs were diluted in Opti-MEM and added to cells.

Real-Time qRT-PCR of miR-200b Expression
miR-200b specific primer and the internal control RNU44 and RNU48 gene were purchased from Ambion (Thermo Scientific). cDNA synthesis was performed by High-Capacity cDNA synthesis kit (Applied Biosystems, Foster City, CA, USA), according to manufacturer's protocol.
Real-time qRT-PCR was carried out using TaqMan Gene Expression Assay (Applied Biosystems) for miR-200b primer on an Applied Biosystems 7500 Real-Time PCR machine. Fold change in expression was determined using the ddCT method after normalizing to control gene.
Those putative miRNA target sites resulting from at least three databases were considered positive. miR-200b target genes selected with these approach were then analyzed using mRNA microarray expression profiles (Affymetrix U133 plus 2.0) to detect gene down and up-regulated genes in TNBC cells as compared to ARRDC3 overexpressing cells. Significance analysis of microarrays (SAM) was conducted to determine differential mRNA expression. The data from Affymetrix microarray expression were analyzed by Mann-Whitney U Test and p < 0.05 was considered as significant.

Cell Viability by MTT Assay and Crystal Violet Stain
Cells (3.5 × 10 3 ) were seeded in 96-well plates with 100 µL media in triplicate and allowed to adhere overnight. The cells were treated with 5-FU at the concentrations indicated. After the treatment for 24, 48 or 72 h, viability was evaluated using the Kit-8 (Dojundo Molecular Technologies, Rockville, MD, USA) according to the manufacturer's instructions. Absorption at 450 nm was determined using an iMark Microplate Reader (Bio Rad). IC 50 values, representing the drug concentration causing 50% growth inhibition, were calculated using https://www.aatbio.com/tools/ic50-calculator.
For viability staining, cells were transfected with GFP and GFP-ARRDC3. After 24 h, cells were treated with drugs for indicated times. The cells were rinsed with PBS, fixed in methanol and stained with crystal violet. After rinse, plates were air dried overnight. Stained cells were counted and the images were captured by microscope and digital camera (Nikon, Melville, NY, USA)

Cell Cycle Assay
Cells plated on 6-well plate were transfected with miRNAs. After incubation for 48 h, cells were collected by trypsinization, washed with PBS and fixed with 66% ethanol at 4 • C for 1h. Cell were washed with PBS and stained in 200 µL of propidium iodide (PI) and RNase stating solution (PI Flow Cytometry kit for cell cycle analysis, Abcam), followed by incubation at 37 • C for 20 min. For Flow Cytometry, Cells were detached by trypsinization and washed with PBS containing 1% BSA. A total of 1 × 10 6 cells were incubated with APC-CD44 and PE-CD24 for 30 min on ice. CD24 − /CD44 + populations were detected on a BD FACSAria IIu flow cytometer (BD Bioscience, San Jose, CA, USA). Cell cycle distribution was performed using BD FACSCalibur flow cytometer and analyzed using ModFit LT v3.3 software.

Cell Motility Assay
Cell motility assays were performed by a transwell cell culture chamber of 8 µm pore size (Costar-Falcon, Corning Life Science, Tewksbury, MA, USA) according to the standard procedure. Transwell inserts were coated with collagen I (15 µg/mL) overnight at 4 • C. After washing the inserts with PBS next day, cells were added to the upper chamber of each well. Lysophosphatidic acid (LPA; 100 ng/mL) was added to the lower chambers as a chemoattractant. The chambers were incubated for 2 h at 37 • C with 10% CO 2 . The cells that did not migrate through the pores were mechanically removed by cotton swab. The migrated cells on the lower surface of the membrane were fixed and stained with 0.2% crystal violet and counted. Assays were performed in triplicate and repeated three times.

Colony Formation Assay
MDA-MB-231 cells transfected with control (NC), miR-200b and 5-Fu_miR200b were suspended in the top layer of DMEM (1 mL) containing 0.35% low-melt agarose (ISC Bioexpress, Kaysville, UT, USA) and then the top layer was overlaid on DMEM (2 mL) containing 0.75% agar in six-well plates. The cells were fed twice per week with 0.5 mL DMEM. After 3 weeks, colonies larger than 0.1 mm in diameter were counted per well by using bright-field optics. The images of colonies were assessed by a microscope and digital camera (Nikon). The average number of colonies was obtained from counting triplicate wells.

ARRDC3 Reverses Chemo-Resistance to DNA Damaging Agents and EMT Phenotypes of Mesenchymal Subtype TNBC Cells
Our previous studies demonstrated that a metastasis suppressor, ARRDC3 is epigenetically silenced in TNBC cells [8], and that restoring ARRDC3 expression represents an important anti-cancer mechanism of selective inhibitors of nuclear exporters (SINEs) that effectively inhibits TNBC functions in vitro and in vivo [9]. In the current studies, we tested whether the metastasis-suppressing function of ARRDC3 is linked to reduction of chemoresistant nature and epithelial to mesenchymal transition (EMT) phenotypes of TNBC cells. We chose MDA-MB-231 cell line as they represents metastatic mesenchymal subtype of TNBC cell line whose ARRDC3 expression is epigenetically silenced [8] and generally resistant to DNA damaging chemo-agents [18]. We measured and compared cytotoxicity of DNA damaging agent, 5-FU in MDA-MB-231 cells with or without ARRDC3 over-expression ( Figure 1A). Overexpression of ARRDC3 in MDA-MB-231 cells sensitizes them to 5-FU as shown in microscopic images and reduction in IC 50 value to 5-FU ( Figure 1A). We then assessed the effects of ARRDC3 expression in phenotypic changes and stemness in MDA-MB-231 cells. As shown in Figure 1B, transient transfection of GFP tagged ARRDC3 yields 28.9% efficiency of GFP-ARRDC3 cDNA expression in MDA-MB-231 cells. GFP-ARRDC3 transfected cells become more round shape (epithelial like) whereas GFP-null vectors transfected cells maintain MDA-MB-231 cells' typical elongated shape (mesenchymal like) ( Figure 1B). We then measured CD24 -/CD44 + population (stemness like) upon transfection of either GFP-null or GFP-ARRDC3 vectors into MDA-MB-231 cells ( Figure 1C). Expression of ARRDC3 reduced CD24 − /CD44 + population roughly by 20%, which correspond to transfection efficiency of GFP-ARRDC3 in MDA-MB-231 cells ( Figure 1C). These outcomes suggest that ARRDC3 plays a role in reversing epithelial to mesenchymal phenotypes (EMT) phenotypes, stemness and chemo-resistance in TNBC cells. metastatic mesenchymal subtype of TNBC cell line whose ARRDC3 expression is epigenetically silenced [8] and generally resistant to DNA damaging chemo-agents [18]. We measured and compared cytotoxicity of DNA damaging agent, 5-FU in MDA-MB-231 cells with or without ARRDC3 over-expression ( Figure 1A). Overexpression of ARRDC3 in MDA-MB-231 cells sensitizes them to 5-FU as shown in microscopic images and reduction in IC50 value to 5-FU ( Figure 1A). We then assessed the effects of ARRDC3 expression in phenotypic changes and stemness in MDA-MB-231 cells. As shown in Figure 1B, transient transfection of GFP tagged ARRDC3 yields 28.9% efficiency of GFP-ARRDC3 cDNA expression in MDA-MB-231 cells. GFP-ARRDC3 transfected cells become more round shape (epithelial like) whereas GFP-null vectors transfected cells maintain MDA-MB-231 cells' typical elongated shape (mesenchymal like) ( Figure 1B). We then measured CD24 -/CD44 + population (stemness like) upon transfection of either GFP-null or GFP-ARRDC3 vectors into MDA-MB-231 cells ( Figure 1C). Expression of ARRDC3 reduced CD24 − /CD44 + population roughly by 20%, which correspond to transfection efficiency of GFP-ARRDC3 in MDA-MB-231 cells ( Figure  1C). These outcomes suggest that ARRDC3 plays a role in reversing epithelial to mesenchymal phenotypes (EMT) phenotypes, stemness and chemo-resistance in TNBC cells.  To further define relationship between ARRDC3 levels and EMT phenotypes, we selected a panel of breast cancer cell lines with different ARRDC3 levels ( Figure 2). As shown in Figure   To further define relationship between ARRDC3 levels and EMT phenotypes, we selected a panel of breast cancer cell lines with different ARRDC3 levels ( Figure 2). As shown in Figure 2A, ARRDC3 levels were very low to undetectable in mesenchymal subtype (Basal B) of TNBC cell lines in comparison to luminal subtype of breast cancer cell lines and epithelial subtype (Basal A) of TNBC cell lines. The loss of ARRDC3 expression in mesenchymal subtype of TNBC cell lines is correlated with loss of E-cadherin (epithelial marker) and presence of vimentin (mesenchymal marker), which are established EMT markers ( Figure 2A). The forced expression of ARRDC3 restores the expression of E-cadherin and decreased the levels of vimentin ( Figure 2B). The results support the role of ARRDC3 in reversing EMT phenotypes in TNBC cells.

miR-200b-3p Regulates ARRDC3 Expression and Forms a Positive Feedback Loop with ARRDC3 to Reverse EMT Phenotypes and Chemo-Resistance of TNBC Cells
To identify miRNA(s) that mediate the metastasis suppressing function of ARRDC3, MDA-MB-231 cells that express GFP (transfection control) or ARRDC3 were used for RNA extraction and miRNA array processing. Purified RNAs from each sample were labeled with fluorescent dye and hybridized to the miRNA microarrays overnight. The microarrays were scanned on an Axon Gepenix 4000B scanner, and the data was extracted from images using GenePix software ( Figure 3A). We selected a group of mi-RNAs (12 in total) whose roles have been implicated as tumor suppressor, and their levels are up-regulated 1.5 fold or higher by ARRDC3 expression in MDA-MB-231 cells ( Figure 3A). We initially screened these mi-RNAs for their ability to reduce the IC 50 values to 5-FU (in a similar level that ARRDC3 reduces IC 50 to 5-FU) and found that miR-200b-3p (miR-200b from now on) was most effective among those tumor suppressing miRNAs listed in Figure 3A  (HCC1419, MCF7) by qRT-PCR showed that miR-200b expression is down-regulated in TNBC cells ( Figure 3D). The outcome once again suggests the correlation between ARRDC3 and miR-200b levels.
To further define relationship between ARRDC3 levels and EMT phenotypes, we selected a panel of breast cancer cell lines with different ARRDC3 levels ( Figure 2). As shown in Figure 2A, ARRDC3 levels were very low to undetectable in mesenchymal subtype (Basal B) of TNBC cell lines in comparison to luminal subtype of breast cancer cell lines and epithelial subtype (Basal A) of TNBC cell lines. The loss of ARRDC3 expression in mesenchymal subtype of TNBC cell lines is correlated with loss of E-cadherin (epithelial marker) and presence of vimentin (mesenchymal marker), which are established EMT markers (Figure 2A). The forced expression of ARRDC3 restores the expression of E-cadherin and decreased the levels of vimentin ( Figure 2B). The results support the role of ARRDC3 in reversing EMT phenotypes in TNBC cells.

miR-200b-3p Regulates ARRDC3 Expression and Forms a Positive Feedback Loop with ARRDC3 to Reverse EMT Phenotypes and Chemo-Resistance of TNBC Cells
To identify miRNA(s) that mediate the metastasis suppressing function of ARRDC3, MDA-MB-231 cells that express GFP (transfection control) or ARRDC3 were used for RNA extraction and miRNA array processing. Purified RNAs from each sample were labeled with fluorescent dye and hybridized to the miRNA microarrays overnight. The microarrays were scanned on an Axon Gepenix 4000B scanner, and the data was extracted from images using GenePix software ( Figure 3A). We selected a group of mi-RNAs (12 in total) whose roles have been implicated as tumor suppressor, and their levels are up-regulated 1.5 fold or higher by ARRDC3 expression in MDA-MB-231 cells ( Figure  3A). We initially screened these mi-RNAs for their ability to reduce the IC50 values to 5-FU (in a similar level that ARRDC3 reduces IC50 to 5-FU) and found that miR-200b-3p (miR-200b from now on) was most effective among those tumor suppressing miRNAs listed in Figure 3A     (D) RNAs were isolated from the indicated cells and prepared as described in the Material and Method section. miR-200b expression level was quantitated by qRT-PCR. Column, mean from at least three independent experiments; bars, SD. The statistical analysis was done using Student's t test. *, p < 0.05, **, p < 0.01.

Vesicle Free Delivery of 5-FU-miR-200b Mimic
miRNA delivery into cells generally requires transfection reagents. We found that delivery of 5-FU-miR-200b mimic into cells does not require any transfection agents. MDA-MB-231 cells were transfected with miR-NC (negative control), miR-200b, or 5-FU-miR-200b mimic by using oligofectamine or incubated with these miRNAs without delivery vehicle. The effects of expression of these miRs on TNBC cell proliferation were monitored by MTT assay for 6 days upon transfection or incubation ( Figure 5A). While miR-200b reduced the rate of cell proliferation of MDA-MB-231 cells in comparison to miR-NC, 5-FU-miR-200b mimic dramatically reduced the cell growth in a way that cell numbers started to decrease from day 3 ( Figure 5A).
Interestingly, 5-FU-miR-200b mimic is still able to block proliferation of MDA-MB-231 cells even without oligofectamine treatment (without delivery vehicle) whereas miR-NC and miR-200b had no effects on proliferation without oligofectamine ( Figure 5A). miR-200b has no inhibitory effects on proliferation of luminal subtype of breast cancer cells, MCF7, suggesting that its inhibitory effects can be TNBC specific ( Figure 5B). However, 5-FU-miR-200b mimic reduces proliferation of MCF-7 cells with or without oligofectamine ( Figure 5B). This outcome indicates that 5-FU-miR-200b mimic has

Discussion
Our studies have demonstrated the mechanism by which ARRDC3 reverses EMT phenotypes and chemo-resistance to 5-FU in TNBC cells. miR-200b is an important target gene of ARRDC3 and mediates chemo-sensitizing function of ARRDC3. Based on our outcome that expression of ARRDC3 in TNBC cell is particularly effective in sensitizing TNBC cells to 5-FU over other conventional chemo-drugs, and that miR-200b also sensitize TNBC cells to 5-FU, we anticipate that ARRDC3/miR-200b feedback loop is an important therapeutic target for TNBC. This rationale led to development of novel strategy that combine therapeutic powers of 5-FU and miR-200b into a single compound (5-FU-miR-200b mimic) that can be characterized as a new treatment option for TNBC.
Based on the heterogenous nature of TNBC, it remains to be determined whether therapeutic targeting of ARRDC3/miR-200b pathway is effective in breast cancer in general, or more effective on specific subtypes of TNBC. Our result that ARRDC3 and miR-200b levels are particularly low to absent in the mesenchymal subtype of TNBC cell lines suggest that restoration of ARRDC3 or miR-200b can be more effective in mesenchymal subtype TNBC. Our previous report that IC50 of selective inhibitors of nuclear exporter (SINEs) is inversely correlated with the levels of ARRDC3 further supports this hypothesis as restoration of ARRDC3 represents anti-cancer mechanism of SINEs [9]. Our data that miR-200b effectively reduces proliferation of TNBC cell line (MDA-MB-231), but not luminal subtype breast cancer cells (MCF-7) ( Figure 5) suggest that targeting ARRDC3/miR-200b pathway can be more effective in TNBC. It is possible that reversal of EMT phenotypes is linked to sensitization of mesenchymal subtype of TNBC cells to DNA damaging agents such as 5-FU whereas non-mesenchymal breast cancer cells are less susceptible by increasing ARRDC3 or miR-200b expression.
We have created a new miRNA mimic by incorporating 5-FU to the therapeutic miRNA, in this case, miR-200b. Interestingly, 5-FU-miR-200b mimic is still able to block proliferation of MDA-MB-231 cells even without oligofectamine treatment (without delivery vehicle) whereas miR-NC and miR-200b had no effects on proliferation without oligofectamine ( Figure 5A). miR-200b has no inhibitory effects on proliferation of luminal subtype of breast cancer cells, MCF7, suggesting that its inhibitory effects can be TNBC specific ( Figure 5B). However, 5-FU-miR-200b mimic reduces proliferation of MCF-7 cells with or without oligofectamine ( Figure 5B). This outcome indicates that 5-FU-miR-200b mimic has broader range of specificity in terms of its inhibitory functions. Reduction of fibronectin (miR-200b target gene) expression was only observed by 5-FU-miR-200b mimic, but not by miR-NC or miR-200b without oligofectamine ( Figure 5C). The results suggest that 5-FU-miR-200b mimic can get into the cells and inhibit TNBC cell proliferation and target gene expression without transfection agents.

Discussion
Our studies have demonstrated the mechanism by which ARRDC3 reverses EMT phenotypes and chemo-resistance to 5-FU in TNBC cells. miR-200b is an important target gene of ARRDC3 and mediates chemo-sensitizing function of ARRDC3. Based on our outcome that expression of ARRDC3 in TNBC cell is particularly effective in sensitizing TNBC cells to 5-FU over other conventional chemo-drugs, and that miR-200b also sensitize TNBC cells to 5-FU, we anticipate that ARRDC3/miR-200b feedback loop is an important therapeutic target for TNBC. This rationale led to development of novel strategy that combine therapeutic powers of 5-FU and miR-200b into a single compound (5-FU-miR-200b mimic) that can be characterized as a new treatment option for TNBC.
Based on the heterogenous nature of TNBC, it remains to be determined whether therapeutic targeting of ARRDC3/miR-200b pathway is effective in breast cancer in general, or more effective on specific subtypes of TNBC. Our result that ARRDC3 and miR-200b levels are particularly low to absent in the mesenchymal subtype of TNBC cell lines suggest that restoration of ARRDC3 or miR-200b can be more effective in mesenchymal subtype TNBC. Our previous report that IC 50 of selective inhibitors of nuclear exporter (SINEs) is inversely correlated with the levels of ARRDC3 further supports this hypothesis as restoration of ARRDC3 represents anti-cancer mechanism of SINEs [9]. Our data that miR-200b effectively reduces proliferation of TNBC cell line (MDA-MB-231), but not luminal subtype breast cancer cells (MCF-7) ( Figure 5) suggest that targeting ARRDC3/miR-200b pathway can be more effective in TNBC. It is possible that reversal of EMT phenotypes is linked to sensitization of mesenchymal subtype of TNBC cells to DNA damaging agents such as 5-FU whereas non-mesenchymal breast cancer cells are less susceptible by increasing ARRDC3 or miR-200b expression.
We have created a new miRNA mimic by incorporating 5-FU to the therapeutic miRNA, in this case, miR-200b. While 5-FU-miR-200b mimic is technically a single compound, it already possesses the combination therapy component (5-FU as a DNA damaging component and miR-200b as a sensitizing component) in it and this synergistic effect was obvious (Figure 4). Although double stranded 5-FU-miR-200b is stable, it will eventually break down to release 5-FU and other smaller fragments [19]. We have demonstrated that 5-FU released from 5-FU-miR-200b will be able to form the suicide ternary complex 5-FdUMP-TS-THF, which will result a super band shift on the western blot analysis compared to 5-FU control (Figure 4). This clearly demonstrated that 5-FU released from the intact 5-FU-miR-200b will be able to function as a therapeutic 5-FU molecule, and it will be incorporated in to RNA/DNA to influence RNA or DNA synthesis. It is interesting to note that 5-FU-miR-200b mimic seems to possess broader specificity of inhibiting breast cancer subtypes than wild type miR-200b. miR-200b is mainly effective in inhibiting proliferation of TNBC cell line, but not luminal breast cancer cell line whereas 5-FU-miR-200b mimic is effective in inhibiting proliferation of both cell lines ( Figure 5). This outcome suggests that 5-FU-miR-200b mimic can be further characterized as pan-breast cancer therapeutic agent including TNBC. The mechanism by which 5-FU-miR-489 mimic enters into the cells without transfection reagent is not defined yet. However, there are multiple evidence supporting that chemical modifications of oligonucleotides confer their ability to get into the cells without transfection agents (called "gymnosis") [20,21]. While defining the mechanism of cell entry of 5-FU-miR-200b mimic is outside the scope of our current studies, this represents a significant advancement in miRNA based therapeutic development, as delivery technology is a major bottleneck.

Conclusions
In conclusion, we identified ARRDC3/miR-200b pathway as a key target to reverse EMT phenotypes and chemo-resistance of TNBC cells in the current studies. Our strategy that combines therapeutic powers of 5-FU and miR-200b into a single compound could provide the basis for future trial of 5-FU-miR-200b mimic with other conventional chemotherapy compounds as unique strategy for future breast cancer treatment.