Exosomal miR-500a-5p derived from cancer-associated fibroblasts promotes breast cancer cell proliferation and metastasis through targeting USP28

1As the most common cancer in women, breast cancer is the second leading cause of cancer-related mortality worldwide among women and represents a serious threat to women's health. Tumors are encircled by an extracellular matrix and stromal cells. These cellular and non-cellular components of the tumoral niche comprise the tumor microenvironment. The crosstalk between cancer cells and the tumor microenvironment is essential for tumor progression and metastasis,. Cancer-associated fibroblasts (CAFs) are a major cellular component of the tumor microenvironment in most solid cancers. Accumulating evidence indicates that CAFs play an important role in cancer initiation, progression, epithelial-to-mesenchymal transition (EMT), metastasis and therapy resistance by communication with cancer cells via exosomes -.

2Exosomes are small membranous vesicles with a particle size of 30-150 nm. These small extracellular vesicles are critical cellular communicators encapsulating a variety of proteins, lipid, mRNAs, microRNAs, and lncRNAs,. Among the exosome-carrying molecules, miRNAs have drawn much attention due to their basic features. MiRNAs in exosomes, can be taken up by neighboring or distant cells and subsequently promote oncogenic signaling by suppressing target mRNAs in recipient cells -. To verify the potential of a CAF-derived exosomal miRNA for use as a diagnostic marker or a therapeutic target for cancer, a comprehensive understanding of its mechanistic role in cancer progression is required.

3In this study, we isolated CAFs and normal fibroblasts (NFs) from patient tissues and identified the differential miRNA profiles of cancer cells treated with CAF-derived exosomes using next-generation sequencing. Our results demonstrated that miR-500a-5p was directly transferred from CAFs to breast cancer cells via exosomes. We also identified the molecular mechanisms by which exosomal miR-500a-5p modulated the malignant phenotypes in breast cancer cells.

4CAFs and NFs were isolated from the breast carcinomas and adjacent normal breast tissue, respectively. The isolated CAFs and NFs possessed a spindle-shaped morphology and were both vimentin-positive. However, compared with NFs, CAFs expressed higher levels of the specific fibroblast markers α-smooth muscle actin (α-SMA), fibroblast activation protein (FAP), and fibroblast-specific protein 1 (FSP1) (Figure 1A-B). Exosomes, which are nanometric membrane vesicles secreted by both CAFs and NFs, play an important role in the communication between cells in the tumor microenvironment. We isolated exosomes from the conditioned media (CM) collected from cultures of CAFs and NFs by ultracentrifugation. Exosomes were characterized and quantified by Western blot analysis, qNano and electron microscopy. Western blot analysis confirmed the expression of the exosomal markers CD63 and HSP70, while GM130 (cis-Golgi compartment-specific marker) and calnexin (endoplasmic reticulum membrane marker) expression was absent (Figure 1C), which was consistent with previous reports on exosomes. Electron microscopy and qNano analysis confirmed the double-layer membrane nature and size of the exosomes (10-210 nm diameter) (Figure 1D, E). Furthermore, we evaluated uptake of these CAF-derived exosomes by the breast cancer cells using the fluorescent dye PKH26. MDA-MB-231 or MCF7 cells were cultured with CAF-derived exosomes labeled with PKH26. After 3 h, the presence of red fluorescence in the breast cancer cells indicated efficient uptake of the CAF-derived exosomes (Figure 1F).

5To confirm the impact of our patient-derived CAFs on tumor cell proliferation, MDA-MB-231 or MCF7 cells were cultured with CAF- or NF-derived exosomes and tumor cell proliferation was evaluated by MTT assay. Compared to the NF-derived exosomes, CAF-derived exosomes significantly promoted the proliferation of the breast cancer cells (Figure 2A). Meanwhile, by using the EV-free supernatant after ultracentrifugation, we found that CAF-mediated promotion of breast cancer cell proliferation and migration is in an exosome-dependent manner or, alternatively, the effect is not the result of co-purifying molecular complexes in the ultracentrifuged pellet (Figure S1). We further performed the three-dimensional (3D) embedded assays to further investigate the impact of CAFs on tumor cells. Green fluorescent protein (GFP)-labeled MDA-MB-231 or MCF7 cells and mCherry-labeled NFs or CAFs were mixed and micro-injected into semi-solidified Matrigel to culture 3D spheroids for 7 days. The observation of both red and green fluorescence suggested the spheroids consisted of fibroblasts and breast cancer cells. Under co-culture conditions, we measured the size of the spheroids and found that the spheroids containing CAFs exhibited a larger volume (Figure 2B).

6Furthermore, in Transwell assays to verify the impact on the migration of breast cancer cells, we found that both CAF-derived exosomes and co-culture with CAF cells increased metastasis of MDA-MB-231 and MCF7 cells (Figure 2C). Thus, these data indicated that exosomes released by patient-derived CAFs accelerate the proliferation and migration of breast cancer cells.

7MiRNAs are frequently encapsulated in exosomes and involved in intercellular communication. To identify miRNAs that are transferred to breast cancer cells by exosomes secreted from CAFs, we incubated MDA-MB-231 and MCF7 cells with NF- or CAF-derived exosomes and further analyzed the miRNA profiles using miRNA microarrays. Data analyses identified that 31 miRNAs were up-regulated at least 1.5-fold. Among the 31 differentially-expressed miRNAs, miR-500a-5p was most increased (more than 4-fold) both in MDA-MB-231 cells and MCF7 cells incubated with CAF-derived exosomes (Figure 3A). These results were confirmed by qRT-PCR (Figure 3B). MiR-500a-5p is a less well-studied miRNA in breast cancer. By excavating high-throughput RNA-seq data deposited in The Cancer Genome Atlas (TCGA), a trend towards increased miR-500a-5p expression in breast cancer tissues was observed (Figure 3C). Moreover, the higher expression of miR-500a-5p contributed to a trend of poorer overall survival outcome in all types of breast cancer (Figure 3D). These data suggested that miR-500a-5p may have a certain cancer-promoting effect.

8To explore the origin of miR-500a-5p in breast cancer, we assessed its expression in frozen tissue sections from breast cancer patients by in situ hybridization (ISH). Intriguingly, although miR-500a-5p expression was detected in both the cancer nest and tumor stroma, it was more highly expressed in the stroma (Figure 3E). To further confirm this pattern, we detected miR-500a-5p levels in isolated CAFs and NFs. As shown in Figure 3F, a significant increase in miR-500a-5p expression was detected in CAFs (Figure 3F). Furthermore, miR-500a-5p was upregulated in exosomes isolated from CAFs compared to those derived from NFs (Figure 3G). Importantly, prior addition (24 h) of inhibitor GW4869 to blocked CAFs exosome production decreased miR-500a-5p expression in MDA-MB-231 cells treated with CAFs-exosome (Figure 3H), suggesting that CAFs secrete extracellular miRNAs predominantly in an exosome-dependent manner.

9We next investigated the ability of breast cancer cells to take up miR-500a-5p from exosomes secreted by CAFs. Exosomes from CAFs that were transiently transfected with FAM-tagged miR-500a-5p mimics were incubated with MDA-MB-231 and MCF7 cells. Evaluation by fluorescence microscopy revealed green fluorescence in MDA-MB-231 and MCF7 cells (Figure 4A). To determine whether the high level of miR-500a-5p in CAFs can be transferred to breast cancer cells, we incubated MDA-MB-231 and MCF7 cells with exosomes isolated from CAFs transfected with miR-500a-5p mimics or negative control (NC). qRT-PCR analysis indicated that miR-500a-5p expression was significantly upregulated in both cell lines following incubation with the CAF-derived exosomes (Figure 4B).

10Next, we incubated MDA-MB-231 and MCF7 cells with exosomes isolated from CAFs overexpressing miR-500a-5p. MTT and Transwell assays showed that CAF-derived exosomal miR-500a-5p significantly promoted cell proliferation and migration (Figure 4C-D). Furthermore, mCherry-tagged CAFs transfected with miR-NC, miR-500a-5p mimics or miR-500a-5p inhibitor were mixed with GFP-labeled breast cancer cells and micro-injected into the semi-solidified Matrigel to culture 3D spheroids. Under co-culture conditions, the spheroids formed in the miR-500a-5p overexpressing group were significantly larger than those in the NC group, whereas smaller spheroids were formed in the miR-500a-5p inhibition group (Figure 4E).

11We next investigated the biological functions of miR-500a-5p in breast cancer cells. MDA-MB-231 cells and MCF7 cells were transiently transfected with miR-500a-5p mimics or miR-500a-5p inhibitor. MTT and the colony formation assays revealed that miR-500a-5p promoted the proliferation and colony formation ability of both cell types compared with the NC group, while these effects were reversed following miR-500a-5p inhibition (Figure 5A-B). Furthermore, GFP-tagged MDA-MB-231 cells transfected with miR-NC, miR-500a-5p mimics or inhibitor were used to culture 3D spheroids. Evaluation of green fluorescence showed that cells transfected with miR-500a-5p mimics formed larger spheroids compared with the control group, whereas those transfected with miR-500a-5p inhibitor formed small spheroids (Figure 5C). These findings suggested the positive effect of miR-500a-5p on breast cancer cell proliferation. Furthermore, we performed Transwell and scratch wound healing assays to analyze the effect of miR-500a-5p on tumor metastasis (Figure 5D, S2A-C). The results indicated that miR-500a-5p enhanced both the migration and invasive abilities of MDA-MB-231 and MCF7 cells, while both functions were significantly reduced after miR-500a-5p inhibition. Given that EMT plays a vital role in cancer metastasis, we further investigated the expression of EMT-related markers after ectopic expression and depletion of miR-500a-5p. Western blot analysis demonstrated that ectopic expression of miR-500a-5p decreased the expression of epithelial markers and increased the expression of mesenchymal markers (Figure 5E, S4A), highlighting the significant role of miR-500a-5p in regulating EMT in breast cancer cells. Therefore, these results demonstrated that miR-500a-5p promotes cell proliferation, migration, invasion, and EMT in breast cancer cells.

12To explore the mechanism by which miR-500a-5p promotes cancer progression in breast cancer cells, three miRNA target gene databases (miRDB, TargetScan and miRWalk) were searched to identify the potential target of miR-500a-5p. In total, 151 genes were listed in all three algorithms (Figure 6A). Among all the predicted genes, we chose the 14 most highly ranked genes as candidates. As the most downregulated gene according to the qRT-PCR analysis, USP28 was selected for further investigation (Figure 6B). USP28 expression in MDA-MB-231 cells was reduced following miR-500a-5p mimics transfection in a dose-dependent manner at both the mRNA and protein levels (Figure 6C-D). MRE analysis suggested that miR-500a-5p has two potential binding sites within the 3'-UTR of USP28 (Figure 6E). To verify USP28 as a direct target of miR-500a-5p, wild-type or mutated binding sites in the 3'-UTR of USP28 were cloned into the luciferase reporter plasmid to generate the wild-type (WT) and mutant luciferase reporters (MUT1 or MUT2 or MUT1+2). After co-transfection with miR-500a-5p mimics, we observed reduction in luciferase activity in both the WT and MUT2 groups, while there was no diminution in luciferase activity in the MUT1 and MUT1+2 groups (Figure 6F). These findings suggested that MUT1 contained the affirmative mutant binding sites in the 3'-UTR of USP28, thus indicating that USP28 is a direct target of miR-500a-5p.

13Furthermore, gene expression profiles of breast cancer from TCGA showed that the expression of USP28 had a downward trend in tumor tissues (Figure 6G). Moreover, USP28 also showed a trend towards reduced expression in breast cancer patients with lymph node metastasis (Figure 6H). These findings suggested that USP28 may function as a breast cancer suppressor which was in contrast to the function of miR-500a-5p.

14Given that our findings indicated that USP28 is regulated by miR-500a-5p, gain- and loss- of function experiments were conducted to determine the biological characteristics of USP28 in breast cancer cells. MTT assays showed that USP28 overexpression reduced cell proliferation in MDA-MB-231 and MCF7 cells, and these effects were partially reversed by transfection with miR-500a-5p mimics (Figure 6I). We also conducted 3D cultures to investigate the roles of miR-500a-5p and USP28 in cancer cell proliferation. Ectopic USP28 expression in MDA-MB-231 and MCF7 cells resulted in the formation of small spheroids and this effect was antagonized by miR-500a-5p overexpression (Figure 6J). Transwell assays showed that USP28 overexpression reduced both cell migration and invasion in MDA-MB-231 and MCF7 cells, and these effects were partially reversed by transfection with miR-500a-5p mimics (Figure 6K). Next, we designed three small interfering RNAs (siRNAs) to inhibit USP28 expression and chose the siRNA with the best interfering effect for the MTT and Transwell assays (Figure S3A). The results showed that cell proliferation, migration and invasion were enhanced after USP28 inhibition (Figure S3B-C).

15We then detected the expression of EMT-related markers in breast cancer cells after upregulation and downregulation of USP28. Following USP28 overexpression in MCF7 and MDA-MB-231 cells, E-cadherin expression was significantly increased, while expression of N-cad, FN1, ZEB1, snail and slug was reduced. USP28 knockdown showed converse patterns of expression of these EMT markers (Figure 6L, S4B). Therefore, these data indicated that miR-500a-5p promotes breast cancer progression and metastasis by sponging USP28.

16To evaluate the effect of miR-500a-5p on breast cancer in vivo, we generated CAFs stably expressing miR-500a-5p by lentivirus infection. Then, MDA-MB-231 cells were co-injected nude mice with miR-500a-5p-expressing CAFs (CAF/miR-500a-5p) or negative control CAFs (CAF/NC). We found that tumor volume and size were significantly larger in the CAF/miR-500a-5p group (Figure 7A-B). qRT-PCR revealed that miR-500a-5p expression was higher in the CAF/miR-500a-5p group, whereas USP28 showed the opposite expression pattern (Figure 7C-D). Next, IHC analysis confirmed α-SMA expression in both groups and stronger Ki67 expression in the CAF/miR-500a-5p group (Figure 7E). These results suggested the involvement of CAFs in the xenograft tumor composition and indicated that miR-500a-5p increases breast cancer cell proliferation in vivo. Moreover, examination of EMT markers in the tumors by Western blot analysis showed a positive correlation of high miR-500a-5p expression with low levels of USP28, E-cad, and high levels of N-cad, FN, slug and vimentin (Figure 7F). To determine the effects of miR-500a-5p expression on breast cancer cell metastasis in vivo, MDA-MB-231 cells stably expressing miR-320a or NC were injected into nude mice intravenously. Lungs were harvested 6 weeks after the cell injections, and we found that miR-500a-5p significantly increased surface lung metastases and the lung metastatic burden (Figure 7G-H). Collectively, our data suggested that high levels of miR-500a-5p in CAFs are transferred into tumor cells and downregulates USP28 expression, which promotes cell proliferation, metastasis and EMT in breast cancer (Figure 8).

17In recent decades, the roles of the microenvironment in cancer progression, metastasis and therapeutic outcome have gained increasing attention,. CAFs have emerged as key players among stromal cells, owing to their abundance in most solid tumors and their diverse tumor-restraining/promoting roles -. In this study, we isolated patient-derived breast CAFs and confirmed their tumor-promoting roles by comparison with NFs. Previous studies demonstrated that CAFs and cancer cells communicate not only through classical paracrine signaling mechanisms (cytokines, chemokines) -, but also through exosomes,. Exosomes are endosome-derived microvesicles that contain lipids, proteins and nucleic acids (DNA, mRNA or non-coding RNAs, such as miRNAs and lncRNAs). As the major RNA component of exosomes, miRNAs are dysregulated in several types of cancers and modulate cancer progression and metastasis,. Exosomal transfer of several miRNAs, such as miR-196a, miR-21, and miR-1228, have been shown to play an important role in communication between cancer cells and the tumor microenvironment,,,.

18In this study, we identified miR-500a-5p as a CAF-derived exosomal cargo that promoted the proliferation and metastasis of breast cancer cells. Previous studies have shown that miR-500a-5p (known as miR-500a) was upregulated in hepatocellular carcinoma and gastric cancer tissues and promoted tumor progression and metastasis -. In contrast, miR-500a-5p was downregulated in colorectal cancer tissues and remarkably inhibited colorectal cancer cell proliferation and migration by targeting HDAC2. MiR-500a-5p is a less well-studied miRNA in breast cancer. Although Esposti et al. suggested that miR-500a-5p regulated oxidative stress response genes in breast cancer and predicted cancer survival, the roles of miR-500a-5p in breast cancer and the underlying mechanisms have not yet been elucidated. In our study, ISH revealed that miR-500a-5p is significantly overexpressed in the stroma compared with the levels detected in the tumor parenchyma. By incubating breast cancer cells with exosomes isolated from patient-derived CAFs, we confirmed that miR-500a-5p was transferred to the cancer cells and increased their proliferation and metastasis in vitro and in vivo.

19MiRNAs are a family of short (20-24 nt) non-coding RNAs that post-transcriptionally regulate gene expression by affecting both the stability and translation of mRNAs in multicellular organisms. As one of the most prevalent and important post-translational modifications, ubiquitination is engaged in multiple cancer-related pathways. The deubiquitylating enzymes (DUBs) are involved in cancer-modulated process by regulating the ubiquitination process. In this study, the DUB USP28 was verified as a target of miR-500a-5p. USP28 has been shown to play complex and versatile modulatory roles in various cancers, through its involvement in multiple pathways that can have opposing functions. Previous studies revealed that USP28 is required for the stability of oncoproteins such as c-MYC in colon cancer -, and c-JUN and NOTCH in colorectal cancer. In addition, USP28 was found to stabilize the tumor suppressor Chk2. Notably, USP28 performs dual regulation of the proto-oncogenic FBW7. USP28 can counteract the activity of FBW7 and promote the stability of FBW7-substrates in cancer cells. In contrast, USP28 was also found to stabilize FBW7 directly, leading to a reduction in FBW7 substrate proteins. Thus, USP28 functions as either a tumor promotor or suppressor depending on the status of autocatalytic ubiquitination of FBW7. In breast cancer, Cao et al. found that HDAC5 positively regulated LSD1 by promoting the stability of USP28 protein, which is a confirmed deubiquitinase of LSD1. In addition, HDAC5 promoted breast cancer development and progression in a LSD1-dependent manner, however, there were no direct evidence that USP28 was involved in breast cancer progression in the study. In contrast, in a detailed study on the function of USP28 in breast cancer, Richter et al. found that USP28 deficiency in breast cancer cells enhances conversion toward a more aggressive phenotype by promoting EMT, proliferation, migration, angiogenesis, and decreased adhesion. Furthermore, a positive correlation between USP28 expression and survival was identified in patients with ductal breast carcinomas. However, the underlying mechanisms of breast cancer suppression by USP28 are not clear. In accordance with these findings, our study revealed that USP28, as the target gene of miR-500a-5p, reversed the ability of miR-500a-5p to promoted breast cancer cell proliferation and metastasis, indicating that USP28 functions as a breast cancer suppressor. Given the complexity of the various functions of USP28, specific mechanism for inhibiting breast cancer will be further studied in our future work.

20In conclusion, our results show that patient-derived CAF exosomal miR500a-5p can confer an aggressive phenotype in breast cancer cells through its transfer from neighboring CAF cells. Thus, our findings suggest that inhibiting exosomal miR500a-5p is a new strategy for suppressing breast cancer growth and metastasis.

21All the clinical samples were obtained from patients with invasive breast carcinoma without previous radiotherapy or neoadjuvant chemotherapy treated at the Qilu Hospital, Shandong University (Shandong, China). Our study was approved by the Ethics Committee of Qilu hospital, Shandong University (the ethical permission number: KYLL-2017(KS)-084). Written informed consent was provided by all participants prior to enrollment. All experimental methods were conducted in accordance with the Helsinki Declaration. All animal studies were undertaken in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

22Primary CAFs and NFs were isolated from invasive ductal carcinoma samples obtained from patients undergoing breast ablation or resection. CAFs were derived from the area at the epicenter of the tumor with a maximum diameter of less than 10 mm, whereas NFs were obtained from the zone of normal tissue, more than 20 mm away from the tumor margin. Tissue samples were minced and digested for 2 h at 37 ºC with shaking in collagenase/hyaluronidase (#07912, Stem Cell Technologies, Durham, NC, USA) diluted in Dulbecco's modified Eagle's medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; Sigma, St. Louis, MO, USA). After centrifugation, cell pellets were resuspended in DMEM supplemented with 10% FBS and filtered through a 100 μm cell strainer. The cells were then seeded into 60 mm tissue culture dishes and cultured in DMEM supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin and 10% FBS at 37 ºC in a humidified atmosphere containing 5% CO2. NFs and CAFs were further identified by detection of the CAF-specific markers α-SMA, FAP and FSP1.

23For conditioned medium preparation, the fibroblasts were incubated with DMEM containing 10% exosome-free FBS for 2 days. The conditioned medium was collected, centrifuged at 1,000 ×g for 5 min and 10,000 ×g for 30 min before being filtered (0.22-μm pore size).

24For co-culture experiments, CAFs or NFs and cancer cells were mixed at a ratio of 3:1 by seeding cancer cells in six-well plates and then seeding CAFs or NFs in the upper chambers of Boyden chambers (4 μm pore size; Millipore, Darmstadt, Germany).

25Exosomes were isolated from the conditioned medium by differential centrifugation as previously described. In brief, conditioned medium was centrifuged at 300 ×g for 10 min and then at 2,000 ×g for 20 min at 4 ºC to remove cells. The supernatant was then centrifuged at 10,000 ×g for 30 min at 4 ºC to remove cell debris. The exosomes in the supernatant were pelleted by ultracentrifugation at 100,000 ×g for 90 min and resuspended in PBS before collection by ultracentrifugation again at 100,000 ×g for 90 min. The exosome particle size was measured by nanoparticle tracking analysis (NTA) using ZetaView PMX 110 (Particle Metrix, Meerbusch, Germany). The exosomes were used for RNA/protein extraction or cell treatment.

26Purified exosomes were labeled using the PKH26 red fluorescent labeling kit (Sigma) according to the manufacturer's protocol as previous studies,. In brief, the exosomes enriched by ultracentrifugation were resuspended in 100 μL of diluent C and mixed with 100 μL PKH26 dye solution (4 × 10-6 M). After incubated for 5 min, 200 μL of serum was added to stop. The labeled exosomes were then washed twice with PBS and coincubated with breast cancer cells. Uptake of labeled exosomes by breast cancer cells over 3 h was observed under a microscope.

27Exosomes were dropped onto the copper grid and negatively stained with 2% phosphotungstic acid for 2 min. After air-drying, the samples were observed using a transmission electron microscope (FEI Tecnai G2 Spirit, Thermo Scientific, USA) at 80 kV.

28ISH was performed to determine the expression of miR-500a-5p using the Enhanced Sensitive ISH Detection Kit (POD) (MK1030, Boster Biological Technology, Wuhan, China) according to the manufacturer's instructions. In brief, frozen sections (10 mm thick) prepared from formalin-fixed tissues were digested with proteinase K at 37 ºC for 2 min. After prehybridization, slides were hybridized with miR500a-5p or scramble DIG-labeled probes at 37 ºC overnight followed by washing with SSC buffers. Subsequently, the slides were blocked and incubated with biotinylated mouse anti-DIG antibody and streptavidin-biotin-peroxidase complex (SABC) at 37 ºC, followed by three rinses with 0.5M phosphate-buffered saline (PBS). The sections were then incubated with avidin-biotin peroxidase complex. The results were visualized by development of a colored substrate 3,3'-diaminobenzidine (DAB) and counterstaining with hematoxylin. Finally, the sections were dehydrated in a graded ethanol series, cleared in xylene, and mounted with coverslips.

29Breast cancer cell lines (MDA-MB-231 and MCF7) were purchased from American Type Culture Collection (ATCC, USA) and characterized by Genetic Testing Biotechnology Corporation (Suzhou, China). Cells were routinely cultured in the DMEM supplemented with 1% penicillin/streptomycin and 10% FBS at 37 ºC under 5% CO2. Cells were transfected using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions.

30The cDNA of USP28 were cloned into the pcDNA3.1-Myc/His vector (Invitrogen) and verified by sequencing. For luciferase reporter assays, PCR-derived fragments from the 3′-UTR of USP28 (WT) was cloned into the pmiRGLO vector (Promega, Madison, WI, USA). The mutant luciferase reporters (MUT1 or MUT2 or MUT1+2) were generated by PCR-based site-directed mutagenesis. MiR-500a-5p mimics, inhibitors, USP28 siRNA and negative controls were purchased from GenePharma (Shanghai, China). The sequences were as follows (sense strand): miR-500a-5p mimics, 5'-UAAUCCUUGCUACCUGGGUGAGA-3'; NC, 5'-UUCUCCGAACGUGUCACGU-3'; miR-500a-5p inhibitors, 5'-UCUCACCCAGGUAGCAAGGAUUA-3'; Inhibitor NC: 5'-CAGUACUUUUGUGUAGUACAA-3'; USP28 siRNA, 5'-GGCCUAGAACUCUAUCAAA-3'.

31Fibroblast cells grown on cover slips were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, blocked in 3% goat serum, and incubated overnight at 4 ºC with primary antibodies (α-SMA, CST, #19245, 1:500; FAP, Affinity Biosciences #AF5344, 1:1000; FSP1, Affinity Biosciences, #DF6516, 1:500; vimentin, Millipore, MAB3400, 1:1000). Thereafter, the samples were incubated with an Alexa Fluor 488-conjugated anti-Rabbit IgG (Invitrogen, USA; 1:500) at room temperature for 60 min in the dark. Nuclei were detected by treatment with 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen, USA, 1:300).

32Cells were harvested and proteins were extracted from cells as previously described. The protein concentration was determined using a protein assay kit (Bio-Rad, Hercules, CA, USA). Equal amounts of proteins (30 µg) were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, Bedford, MA, USA). After blocking with 5% non-fat milk, the membrane was incubated overnight at 4 ºC with the primary antibody. The primary antibodies used in our study were as follows: α-SMA, 1:1000; FAP, 1:1000; FSP1, 1;300; vimentin, 1:1000; E-cad, abcam, #ab216833, 1:500; N-cad, abcam, #ab216833, 1:500; FN1, proteintech, 15613-I-AP, 1:500; Slug, Santa Cruz, sc-166476, 1:200; Snail, Santa Cruz, sc-393172, 1:200, and ZEB1, CST, #70512, 1:1000. Then HRP-conjugated secondary antibodies (1:3000) purchased from ZhongShanJinQiao (BeiJing, China) were incubated and the signal was detected with enhanced chemiluminescence.

33Total RNA was extracted using TRIzol reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. The purified RNA was reverse-transcribed to form cDNA using a reverse transcriptase kit (TaKaRa, Dalian, China). Quantitative real-time PCR (qRT-PCR) was performed with SYBR Premix ExTaqTM II (TaKaRa) and β-actin as an internal control. The relative expression level of mRNA was normalized to β-actin expression and was calculated using the 2-ΔΔCt method. The primer sequences are shown in Table 1. For exosomal RNA analysis, the exosome concentration was determined by both NTA and BCA assay. 10 µg (2 × 108) exosomes were used for total RNA extraction. 0.5 µg of total RNA was first reverse transcribed with Mir-X™ miRNA First-Strand Synthesis kit and performed qRT-PCR analysis. The U6 small nuclear RNA was used as an internal control.

34Cell viability was assessed by MTT and colony formation assays as previously described. For MTT assays, 20 µL 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was added to the medium and the cells were incubated at 37 ºC for 4-6 h. After aspirating the supernatant, 100 mL DMSO was added to each well and absorbance values at 490 nm were measured using a microplate reader (Bio-Rad). For colony formation assays, cell colonies were fixed with methanol and stained with 0.1% crystal violet (Sigma) before the colonies were counted under a microscope.

35Cell migration and Matrigel invasion assays were performed as previously described. For migration assays, after infiltrating the Transwell for 24-48 h, cells were fixed and stained with 0.1% crystal violet. For invasion assays, the Transwell membrane was coated with Matrigel. The cells on the lower surface were photographed and counted by ImageJ software.

36Lentiviral packaging was carried out in HEK-293T cells using the 3rd-generation system as described previously. Lentiviral particles (Lv-mCherry or Lv-EGFP) were harvested at 48 h after transfection. CAFs were labeled with red fluorescent protein by lentivirus infection with Lv-mCherry, while MDA-MB-231 cells were labeled with green fluorescent protein by lentivirus infection with Lv-EGFP. After 3 days, the labeled CAFs and MDA-MB-231 cells were mixed and injected into the semi-solidified Matrigel (BD Biosciences) using a microsyringe. The mixed cell masses were cultured in Matrigel for 7 days and imaged using a fluorescence microscope. The size of the spheroids under a light microscope was calculated according to the following formula: volume = 0.5 × length × width2.

37Cells were co-transfected with pmirGLO plasmids containing the wild-type or mutant USP28 3′-UTR and miRNA mimics using Lipofectamine 2000 (Invitrogen). After 48 h, luciferase activity was determined using the Dual-Luciferase Reporter Assay Kit (Promega, Madison, WI, USA) according to the manufacturer's instructions. Firefly luciferase activity was normalized against Renilla luciferase activity.

38Female BALB/c nude mice (aged 4 weeks) were randomly divided into two groups (10 mice per group) and the in vivo studies were performed as previously described,. We established CAFs stably expressing miR-500a-5p by lentivirus infection and selection with geneticin. A total of 3 × 106 MDA-MB-231 cells mixed (1:1) with CAF/miR-500a-5p or CAF/Ctrl cells in 200 μl of PBS: Matrigel at a 3:1 ratio were subcutaneously injected into the flank of nude mice. After the tumors were readily visualized, the tumor growth rate was monitored by measuring tumor diameters every 4 days using a Vernier caliper and the tumor growth curve was recorded accordingly. The tumor volume was calculated according to the following formula: volume = 0.5 × length × width2.

39A total of 8 × 104 MDA-MB-231 cells stably expressing miR-500a-5p or NC were injected into the lateral tail veins of female BALB/c nude mice (aged 4 weeks) to produce lung metastasis. Six weeks later, lungs were collected and stained with hematoxylin and eosin.

40Tumors from mouse flanks were fixed in formalin for at least 24 h and embedded in paraffin. The paraffin-embedded tissue sections (6 μm thick) were evaluated for the expression of α-SMA and Ki-67 by immunohistochemical (IHC) staining using a streptavidin-peroxidase-biotin reagent kit (Zhongshan Biotechnology, Beijing, China) according to the manufacturer's protocol. Tissue sections were then incubated with streptavidin-HRP complex and then hematoxylin. For the negative control, the antibody solution was replaced with PBS.

41Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, San Diego, CA, USA) and PASW Statistics 18.0 (IBM, SPSS, Chicago, IL, USA). Differences between two groups of data and statistical significance were analyzed by Student's t-test or one-way analysis of variance (ANOVA). Survival curves were generated using the Kaplan-Meier method, and the differences were assessed by a log-rank test. P < 0.05 was considered to indicate a statistical significance, and P < 0.01 and P < 0.001 were considered to indicate a high degree of statistical significance.