microRNA-200c/141 upregulates SerpinB2 to promote breast cancer cell metastasis and reduce patient survival

The microRNA-200 (miR-200) family is associated with tumor metastasis and poor patient prognosis. We found that miR-200c/141 cluster overexpression upregulated SerpinB2 in the MDA-MB-231 triple-negative (TN) breast cancer cell line. We observed transcription factor (c-Jun, c-Fos, and FosB) upregulation, nuclear localization of c-Jun, and increased SerpinB2 promoter-directed chloramphenicol acetyltransferase activity in miR-200c/141 cluster-overexpressing cells relative to controls. Additionally, miR-124a and miR-26b, which directly target SepinB2, were downregulated compared to controls. In mouse xenograft models, miR-200c/141 cluster overexpression promoted lymph node and lung metastasis, and siRNA-mediated SerpinB2 knockdown decreased lung metastasis, suggesting that SerpinB2 mediates miR-200c/141-induced lung metastasis. We also explored the clinical significance of SerpinB2 protein status through analysis of primary breast tumor samples and The Cancer Genome Atlas (TCGA) data. High SerpinB2 levels were associated with reduced survival and increased lymph node metastasis in breast cancer patients. SerpinB2 was overexpressed in the TN breast cancer subtype as compared to the luminal subtype. The present study demonstrates that SerpinB2 promotes miR-200c/141 cluster overexpression-induced breast cancer cell metastasis, and SerpinB2 overexpression correlates with increased metastatic potential and unfavorable outcomes in breast cancer patients. SerpinB2 may be a useful biomarker for assessing metastasis risk in breast cancer patients.

Individual miR-200 family member target genes appear to be cancer type-and context-dependent [3]. In many cancers, miR-200 member expression status serves as a surrogate marker for metastasis, response to drug treatments, and patient outcomes [4][5][6][7][8][9]. miR-200c overexpression predicts poor outcome in patients with hormone receptor-negative breast cancer [9]. miR-141 overexpression is observed in Research Paper metastatic cases and is associated with poor outcome in breast cancer (BC) patients [10,11]. Despite advances in our understanding of BC-related miR-200 members, their precise roles in breast cancer cell (BCC) metastasis are largely unknown, in part because each miR-200 has several putative targets with disparate functions. The underlying mechanisms by which miR-200 overexpression promotes metastasis require further study, and genes that serve as intermediaries in miR-200-associated metastasis are yet be identified.
The purpose of this study was to clarify whether SerpinB2 is involved in metastasis promoted by miR-200c/141 cluster overexpression in mouse xenograft models. Additionally, we investigated whether SerpinB2 expression is associated with patient clinicopathological features and outcomes using primary BC tissues and The Cancer Genome Atlas (TCGA) data.

SerpinB2 protein is overexpressed in the TN subtype relative to other subtypes
We investigated relationships between SerpinB2 expression and clinicopathological features (Table 1). High SerpinB2 expression was detected in 216 cases (64.1%), and low expression was found in 121 cases (35.9%). SerpinB2 was more likely to be overexpressed in tumors ≥2 cm in size (69.2%) compared with tumors <2 cm (52.0%; P<0.01). With regard to TNM clinical stage, SerpinB2 was overexpressed in 76.9% of advanced stage (III-IV) cases, but only 58.1% of early stage (I-II) cases (P<0.001). 75.0% of cases with LN metastasis had high SerpinB2 levels, as compared to only 55.6% of cases with no LN metastasis (P<0.001). Figure 4A shows SerpinB2 IHC scores based on a predetermined cutoff. TN subtypes exhibited strong SerpinB2 expression ( Figures 4B and 4C). SerpinB2 IHC scores were higher in the TN subtype (4.75 ± 0.38) than in the LA (3.29 ± 0.29) or LB (2.93 ± 0.26) subtypes, but not the HER2 subtype (3.76 ± 0.51) (LA vs. TN, P=0.0027, LB vs. TN, P<0.0001, LB vs. TN, P=0.1535; Figure 4D).  TCGA data analysis confirmed that SerpinB2 expression was highest in the TN subtype ( Figure 4E).
High SerpinB2 expression was observed in the TN subtype with LN metastasis, but SerpinB2 was only rarely detected (and at low levels) in the TN subtype without LN metastasis (Figures 4F and 4G). In xenografts and TN patient tumor samples, high SerpinB2 expression was also observed in macrophages and fibroblasts in adjacent stromal tissue (Supplementary Figure 5). MDA-MB-231 miR-200c/141 cell xenograft tumors displayed extensive macrophage infiltration (stained with F4/80 as a pan macrophage marker) into lung tissues, and SerpinB2 knockdown decreased macrophage infiltration into these tumors (Supplementary Figure 6).

SerpinE1 and uPA levels are the same in LA, LB, HER2, and TN subtypes
Consistent with TCGA data analyses, SerpinE1 staining did not differ among LA (3.29 ± 0.29), LB (2.93 ± 0.26), HER2 (3.76 ± 0.51) and TN (3.67 ± 0.35) subtypes (Figures 5A, 5B and 5C). A slightly increased uPA IHC score was observed in the TN subtype relative to the LB subtype, but was not significant ( Figure 5D). uPA levels in the TCGA data analysis did not different among the four subtypes ( Figure 5E).

SerpinB2 overexpression is associated with reduced overall survival (OS) in BC patients
Median OS in 143 BC patients was 77.59 months. OS was reduced for patients with SerpinB2overexpressing tumors ( Figure 6A, P=0.027). SerpinB2 overexpression was also associated with reduced OS in patients with LN metastasis ( Figure 6B, P=0.035), but not in those without LN metastasis ( Figure 6C, P=0.105).

DISCUSSION
In BC patients and animal models, high miR-200c and miR-141 levels have been associated with enhanced metastatic colonization and poor clinical outcomes [7,9,10,22,23]. Consistent with these observations, we found that stable miR-  These genes were not predicted targets of miR-200c/141 cluster members. We also found that miR-200c/141 overexpression increased expression and nuclear localization of SerpinB2 transcription factors (c-Jun, FosB and c-Fos), increased SerpinB2 promoter activity, and downregulated miR-26b and miR-124a, suggesting that miR-200c/141 directly or indirectly affects transcription factors and/or miRNAs to upregulate SeripinB2.
SerpinB2 is upregulated during inflammation, and is expressed in diverse cell types, including cancer cells, macrophages, and fibroblasts [13]. While SerpinB2 has been studied with respect to extracellular matrix remodeling, tumor growth, and metastasis in diverse cancers, its association with patient prognosis is unclear [16,18,20,21,25,26]. SerpinB2 exists in two forms, an extracellular form (>60 kDa) and a predominant intracellular form (47 kDa), which can be upregulated by inflammation or stress-related factors, and released into the extracellular space [13,14]. Secreted SerpinB2 inhibits extracellular uPA, which is associated pericellular and extracellular proteolysis and intracellular SerpinB2 interacts with a variety of cytosolic proteins, which is involved in intracellular signaling pathways to regulate cell survival, but its intracellular activities are not yet wellunderstood [15]. SerpinB2 also defends against tumoral invasion by preventing plasminogen activation [16,26].
TNBC molecular subtypes were subclassified as two basal-like (BL1 and BL2), an immunomodulatory  The c-Jun, c-Fos and FosB which are SerpinB2 transcription factors [28] increased in MDA-MB-231 miR-200c/141 cells compared to control. miR-124a and miR-26b which are able to target SepinB2 [29] decreased in MDA-MB-231 miR-200c/141 cells compared to control. miRNAs have the capability of activating gene expression through posttranscriptional regulation for example, increased mRNA stability and miRNA-mediated decoy [30]. There is, as yet, no report of underlying mechanism by which miR-200c/141 increase c-Jun, c-Fos, and FosB or decrease miR-124a and miR-26b. Del Vecchio G et al. reported that miR-200a could increases the c-Jun amount through a microRNA-mediated stabilization of its mRNA [31]. A common mechanism for miR-124 downregulation is the methylation of the promoter region of its gene through transcriptional regulators [32]. We speculate that miR200c/141 may involve in posttranscriptional regulation for upregulation of c-Jun and c-Fos via miRNAmediated mRNA stability and miRNA-mediated decoy. Decreases of miR-124a and miR-26b in MDA-MB-231 miR-200c/141 cells may be due to the overexpression of genes to regulate hypermethylation. We will further investigate underlying mechanisms by which miR-200c/141 increases c-Jun and c-Fos and decreases miR-124a and miR-26b. With respect to the role of SerpinB2 in lung metastasis, we demonstrated that SerpinB2 knockdown suppressed miR-200c/141 overexpression-induced migration and lung metastasis by MDA-MB-231 cells.
Mouse xenograft results showed that upregulation of SerpinB2 by miR-200c/141 overexpression might be a main mechanism for increased lung metastasis in MDA-MB-231 cells. Thus, SerpinB2 likely represents an effector or mediator of miR-200c/141 cluster-induced metastatic behavior in TNBCs. Recently, SerpinB2 in BCCs was shown to promote cancer cell survival and metastatic outgrowth by converting astrocytic FasL into a paracrine death signal for cancer cells, and by inactivating L1 cell adhesion molecule (L1CAM) to spread metastatic cells along brain capillaries [21]. Many studies have linked SerpinB2 to macrophage survival as well as the tumorassociated M2 phenotype in lung metastasis [21,[33][34][35]. We observed macrophage infiltration within lung metastatic foci of MDA-MB-231 miR-200c/141 xenograft tumors; SerpinB2 knockdown decreased this infiltration, suggesting that SerpinB2 promotes TNBC cell lung metastasis via macrophages recruitmentto tumors.
Overexpression of plasminogen activation system members, including uPA, SerpinE1 and SerpinB2, is associated with malignancy, tumor progression, and metastasis [17,19]. SerpinB2 inhibition of uPA in tumors is of significant interest in prognosis and clinical outcome prediction in various cancers [16,20,21,36]. Although others have explored the prognostic value of SerpinB2 in BC patients, the status of SerpinB2 from tumor extracts was evaluated using RT-PCR and immunoenzymatic assays, which do not discriminate between cancer and stromal cells that also express SerpinB2 [16,20,[36][37][38][39][40]. However, IHC analysis can reveal specific SerpinB2 locations in tumor tissues. The present study incorporated TCGA datasets, the University of North Carolina's RNASeqV2, and IHC score analysis for SerpinE1, SerpinB2 and uPA. We observed high SerpinB2 levels in TNBC relative to other BC subtypes, and SerpinE1 and uPA levels differed between LA, LB, HER2, and TN subtypes. Additionally, SerpinB2 expression was higher in primary tumors with LN metastasis than in non-LN metastasis cases, indicating that high SerpinB2 status may be a prognostic factor for predicting LN metastasis.
Bouchet, et al. and Foekens, et al. suggest that SerpinB2 overexpression may be a marker for prolonged survival in BC patients with low uPA and SerpinE1 levels [26,36]. They evaluated SerpinB2 levels from primary tumor lysates using immunoenzymatic assays, which do not reflect the expression site of SerpinB2 in tumor cells. In contrast to these studies, our IHC results showed that high levels of SerpinB2 in cancer cells within the primary tumor tissue were associated with reduced OS for all BC patients. In our study, tumor cells overexpressing SerpinB2 did not display SerpinE1 or uPA downregulation. Our results suggest SerpinB2 overexpression in BCCs with concomitant SerpinEl or uPA expression might contribute to poor prognosis and unfavorable clinical outcomes. Univariate analysis confirmed that a poor outcome was associated with SerpinB2 status, HER2, TNM stage, and LN metastasis in all BC patients. In multivariate analysis, SerpinB2 is not independent prognostic factor, but contribute to LN status or HER2 expression associated with poor outcome.
Korpal M, et al. reported high miR-200 family levels in lung-pleural metastases with reduced BC patient survival and poor prognosis [7]. This is consistent with our xenograft results, and supports the potential role of the miR-200 family in promoting metastasis. However, to our knowledge our study was the first to investigate relationships between miR-200c, miR-141, and SerpinB2 in BC tumor tissues. In TNBC patients, we found a positive correlation between miR-200c, miR-141 and SerpinB2 mRNA. Together, our findings suggest that miR-200c, miR-141 and SerpinB2 are poor prognostic factors in TNBC.
In conclusion, SerpinB2 upregulation was shown to be involved in miR-200c/141 cluster-promoted lung and LN metastasis in BC xenograft models. In patient tumor samples, higher SerpinB2 levels were detected in TNBC as compared to other BC subtypes, and were associated with LN metastasis and reduced survival. Taken together, our findings provide new insights into the role of SerpinB2, which was upregulated by miR-200c/141 overexpression, in promoting BCC lung and LN metastasis, and suggest that SerpinB2 could be used to assess metastasis risk in BC patients.

Cell culture and lentiviral transduction
TNBC MDA-MB-231 cells were purchased from the Korean Cell Line Bank. Cells were cultured in RPMI-1640 (WelGENE, Seoul, Korea) containing 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1% penicillin/ streptomycin, grown in a 5% CO 2 incubator at 37°C, and subcultured weekly. Lentiviral vectors containing the miR-200c/141 cluster (GenBank ID: 406985 406933) and GFP constructs were kindly supplied by Dr. Gregory J. Goodall [41].  Table 2. Real-time PCR reactions were run on an ABI PRISM® 7900 using a SYBR Green PCR master mix (Applied Biosystems, Foster City, CA, USA). Results were analyzed by the ΔCt method, which reflects the threshold difference between a target gene and β-actin in each sample. TaqMan MicroRNA Assays (Applied Biosystems, South San Francisco, CA, USA) were used to quantify mature miRNA levels, following the manufacturer's instructions. miRNAs were isolated from cells using the mirVana miRNA isolation kit (Applied Biosystems), and primers for detecting miR-200c, miR-141, miR-124a and miR-26b were purchased from Applied Biosystems. Reverse transcription was performed using the TaqMan microRNA reverse transcription kit (Applied Biosystems) according to the manufacturer's instructions. RUN48 was used as the endogenous control.

Chloramphenicol acetyltransferase (CAT) reporter assay
Cells (1×107) were transfected with 20 μg of plasmid reporter DNA using Lipofectamine-2000 (Thermo Fisher Scientific, Inc.). CAT assays were performed with 50 μg of whole cell lysate using the CAT reporter gene activity detection kit (Sigma-Aldrich). SerpinB2 promotermediated CAT activity was normalized to the positive control plasmid, pCAT-Control (Promega), to allow comparison of the results of independent experiments.

Western blot analysis
Proteins from cells lysed in RIPA buffer were separated using SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked using 5% skim milk in Tris-buffered saline containing Tween-20, incubated overnight at 4°C with primary antibodies directed against SerpinB2 (Abcam, Cambridge, UK), SerpinE1 (Abcam), uPA (Abcam), c-Jun (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or phosphoc-Jun (Santa Cruz Biotechnology), and β-actin (Sigma-Aldrich, St. Louis, MA, USA), and then incubated with HRP-conjugated secondary antibodies (Santa Cruz Biotechnology). Blotted membranes were visualized using enhanced chemiluminescence reagents (GE Healthcare, Piscataway, NJ, USA) and protein expression was depicted as intensity of each protein relative to that of β-actin using ImageJ software.

Bioluminescence imaging (BLI)
BLI was conducted using the IVIS luminal II system (Caliper Life Sciences, Hopkinton, MA, USA) with living image acquisition and analysis software. After administration of D-luciferin (150 μg/ml), bioluminescent signal in 1×10 3 to 1×10 5 cells per well was quantified. Mice were intraperitoneally injected with D-luciferin (150 mg/kg). BLI of mice was performed weekly postinjection, and the bioluminescent signal of the thorax area was quantified. www.impactjournals.com/oncotarget

Magnetic resonance imaging (MRI) and fluorescence imaging
At 58 days post-injection, in vivo MRI studies were performed on a 9.4 Tesla animal MR scanner (Agilent Technologies, Palo Alto, CA, USA) using a one-channel volume Millipede Coil. In vivo T 2 -weighted mouse thorax images were acquired in the coronal plane using fast spin echo sequence. Ex vivo GFP fluorescence images of excised lungs were obtained and analyzed using Maestro imaging system (CRI Inc, Woburn, MA, USA).

Clinical samples and ethics statement
A total of 480 tissue microarrays of primary tumors in patients who were not treated before surgery, along with medical records, were collected from Outdo Biotech Co. Ltd. (Shanghai, China) between Aug 2004 and Dec 2008. BC subtypes were defined as luminal A (LA, n=96), luminal B (LB, n=108), HER2 (n=37) and TN (n=96), regarding ER, PR and HER2 status. Followup survival data collected from 143 patients for more than 10 years were used. The follow-up deadline was Jul 2014, at which point 42 patients had died, and 101 remained alive. Survival time was counted from the date of surgery to the follow-up deadline or date of death (all patients eventually died of cancer recurrence or metastasis). Total RNAs were isolated from breast cancer tissue of 21 patients with TNBC and real-time RT-PCR was performed to analyze miR-200c, miR-141 and SerpinB2 mRNA. The Institutional Review Board (IRB) of Seoul National University Hospital (IRB No. 1604-063-754 and 1608-131-787) reviewed and approved the study protocol, and exempted the study from the obligation to obtain informed consent. This study was performed in observance of the World Medical Association's Declaration of Helsinki.

Analysis of the cancer genome atlas (TCGA) dataset
From the TCGA data portal, we downloaded clinical information and RNA-seq expression profiles for 917 primary BC tumors. For RNA-seq expression profiles, we used normalized Level-3 gene expression data produced by University of North Carolina's RNASeqV2 pipeline using the Illumina HiSeq platform. Based on the clinical information, average expression values were calculated for SerpinB2, SerpinE1, and uPA in LA (n=474), LB (n=258), HER2 (n=36) and TN (n=149) subtypes.

Statistical analyses
For both in vitro and in vivo data, means ± standard deviations were calculated from at least five independent experiments and statistical analysis of data was performed using analysis of variance (ANOVA) and Student-Newman-Keuls test. For clinical data analysis, the Chi-square test was used to compare baseline characteristic distributions among groups. Survival analyses were performed using the Kaplan Meier method. Cox proportional hazard regression models were used to estimate hazard ratios (HR) with 95% confidence intervals (CI). A multivariate Cox regression model was fitted based on all characteristics that had P≤0.10 in the univariate analysis. For all tests, a P<0.05 was considered significant.