Integrated network analysis identifies hsa-miR-4756-3p as a regulator of FOXM1 in Triple Negative Breast Cancer

Both aberrantly expressed mRNAs and micro(mi)RNAs play important roles in cancer cell function, which makes integration analysis difficult. In this study, we first applied master regulator analysisalgorithm and confirmed hsa-miR-4756-3p as a candidate miRNA in triple negative breast cancer (TNBC) patients; hsa-miR-4756-3p could regulate TNBC cell line apoptosis, proliferation, migration, and cell cycle as well as suppress TGF-β1 signalling andtumour growth. In TNBC, forkhead box protein M1 (FOXM1)was found to be an hsa-miR-4756-3p target gene, and FOXM1 knockout completely inhibited hsa-miR-4756-3p-induced cell migration and metastasis, TGF-β1 signalling, and epithelial mesenchymal signal activation, which indicated that hsa-miR-4756-3p functions via the FOXM1-TGFβ1-EMT axis.

Breast cancer is a heterogeneous cancer. Based on the molecule subtype technology, breast cancer is divided into 5 subtypes, viz., Luminal A, Luminal B, HER2+, Normal like, and Triple Negative 1 . Among them, triple negative breast cancer(TNBC) is characterized by its poor prognosis 2 , high aggressiveness 3 after chemotherapies, and insensitivity towards target therapy 4 . Identifying TNBC-specific aberrant expression of genes or micro (mi) RNAs would not only help inthediagnosis of TNBC patients but can also revealnovel druggable targets in TNBC therapy 5 . miRNAsare a class of non-coding endogenous RNA molecules, whichrangefrom 22 to 25 nucleotidesin length 6 ; miRNAsfunction by binding to 3′-UTRs of their target genes and inducing target mRNA degradation 7 . In normal cells, miRNA expression and function are constrained by space and time 8 ; they also follow cell-and tissue-specific patterns 9 . However, in tumour cells such as in TNBC, specific miRNAsare critical for tumorigenesis and development 10 . Although many studies have already confirmed that multiple core miRNAs can induce epithelial mesenchymal transition (EMT) 11 , stemness 12 , migration, and invasion 13 , thus, facilitating tumour cell growth and invasion, a more comprehensive study is still needed to investigate the miRNA regulatory network in TNBC.
In this study, we appliedmaster regulator analysis (MMRA) algorithm to identifythe core regulatory miRNAs in TNBC 14 . hsa-miR-4756-3p was confirmed as the top candidate miRNA. We have also elucidatedthe role of hsa-miR-4756-3p in tumour cell function in vivoas well as in vitro, which is related with TGFβ1 and forkhead box protein M1 (FOXM1).
Increased expression of hsa-miR-4756-3p induces apoptosis and cell cycle arrest andinhibitscell proliferationand migration in vitro. Expression level of hsa-miR-4756 in TNBC patientshas been found to be low. Further, whether hsa-miR-4756 expression level can affect cell characteristics remains unclear. To address this, we chose the TNBC cell line MDA-MB-231 and transfected it with control miRNA as well ashsa-miR-4756-3p mimic for 48 h. First, we confirmed the cell apoptosis rate in control miRNA andhsa-miR-4756-3p groups, which were 5.03 ± 0.18 and 23.47 ± 1.46, respectively;cell apoptosis was increased in the case of transfection with hsa-miR-4756-3p ( Fig. 2A). Cell proliferation in the hsa-miR-4756-3p mimic group was suppressed (Fig. 2B). Forassessing cell migration, we used wound healing assay. After culture in 37 °C for 12 h, migration velocity in the control and hsa-miR-4756-3p mimic groups was found to be 4.81 ± 0.17 and 1.19 ± 0.09, respectively; cell migration in the hsa-miR-4756-3p group was suppressed (Fig. 2C). Regarding the cell cycle, the proportion of cells in the G2 phasein the control and hsa-miR-4756-3p mimic group was found to be 18.33 ± 1.13 and 4.35 ± 0.39, respectively (Fig. 2D). These results indicated that the increased hsa-miR-4756-3p expression can trigger inhibition of cell proliferation and migration and induction of cell apoptosis as well as cell cycle progression. As hsa-miR-4756-3p widely takes part in cell functions such as invasion, migration and proliferation, and TGFβ-1 signal is an important pathway in tumour development 8 . First, we detected the changes in the TGFβ-1 pathway in control and hsa-miR-4756-3p mimic groups;the protein levels of TGFβ-1 full length, TGFβ-1 type1 receptor, and TGFβ-1 type 2 receptor were decreased as was the related SMAD3 phosphorylation, which indicated down-regulation ofthe TGFβ-1 pathway after increase in hsa-miR-4756-3p expression (Fig. 2E), all of these western blot were repeated by 3 times. We then injected MDA-MB-231 cells into themammary glandsof nude mice, nude mice were treated withcontrol miRNAand hsa-miR-4756-3p inhibitor. After 1 month, the tumour sizes in the hsa-miR-4756-3p group were found to be significantly smaller (Fig. 2F).
FOXM1 is a well-studied gene in TNBC. It is known to be highly expressed primarily in TNBC as compared with that in other subtypes. After treatment with FOXM1 inhibitor thiostrepton, the EMT markers ZEB1 andvimentin, and cell cycle markers CDK1 and PLK1 were found to bedecreased 15 . The underlying mechanism of action of FOXM1 in breast cancer involves sustained activation of SMAD3/SMAD4 activity, which further induces TGF-β-dependent EMT and promotes metastasis 16 . On the other hand, overexpressed FOXM1 can recruit tumour associated macrophages (TAM), which can also secrete TGF-β1 and promote tumour cell invasion 17 . Using FOXM1 KO 231 cells, we found that cell migration induced by hsa-miR-4756-3p was mediated byFOXM1. Besides, FOXM1 KO eliminated metastatic cells from trained 231-injected nude mice, which further indicated the importance of FOXM1 in hsa-miR-4756-3p function. By western blotting, we also showedthat the levels of the TGF-β1 signalsincluding TGF-β1, TGF-β1 type1 and 2 receptors, the downstream Smad3 pathway, the EMT pathway, which included Snail, Twist, Slug, N-cadherin, and vimentin, were all increased in hsa-miR-4756-3p inhibitor-treated mice.
Based on these results, we proved the lower expression of hsa-miR-4756-3p in TNBC patients' primary tumours. We also confirmed that hsa-miR-4756-3p could regulate TNBC cell line function via the FOXM1-TGFβ1-EMT axis.
Cell apoptosis assay, proliferation assay, and wound healing assay. (1) In the apoptosis assay, control and hsa-miR-4756-3p mimic group cells were transfected with control miRNA and hsa-miR-4756-3p mimic for 48 h, washed with PBS solution 2 times, and subsequently labelled with APC-conjugated Annexin-V (BD Biosciences, San Jose, CA)/propidiumiodide(PI). Then cells were detected using FACS Calibur (BD Biosciences, San Jose, CA);apoptotic cells were quantified by APC + cell ratio. (2) In the cell proliferation assay, after transfection of miRNA, cells were dispersed in 96-wells plate. Each well had 1500 cells, and totally we set 0, 24, 48, 72, 96 h time point. When reach the time point, 10% volume of Cell Counting Kit-8 (CCK-8)(Dojindo molecular technologies, Rockville, MD, USA) was added to it and the cells were cultured for 2 h at 37 °C. The cell proliferation was then detected by Microplate Reader for OD at 450 nm; cell proliferation index was quantified as (reads from time points)/(reads from 0 h). (3) In the wound healing assay, after control miRNA and hsa-miR-4756-3p mimic cells attached to the plate, a scratch was made on cell monolayer and migration velocity was quantified as (cell movement distance in 12 h quantified by scale on the picture)/ hours. All the experiments were repeated 3 times and select median as criteria.
Cell cycle assay. For the cell cycle assay, cells were harvested and resuspended in 75% ethanol + 5%FBS, and then incubated at −20 °C for overnight. The cells were then stained with high concentration PI (Sigma-Aldrich, St Louis, MO, USA). The cell cycle was detected by FACS caliber and analysed by FlowJo 7.6.1 software. All the experiments were repeated 3 times and select median as criteria.
Luciferase reporter assay. In the luciferase report assay (1) Cells were transfected with control miRNA and hsa-miR-4756-3p mimic. At the same time, cells were transfected with empty vector, original FOXM1 3′-UTR, and mutation type of FOXM1 3′-UTR (Switchgear Genomics, Menlo Park, CA, USA); all the cells were also transfected with control vector. (2) After 48 h, cells were lysed and luciferase intensity was detected by LightSwitch Dual Luciferase assay kit (Biotek, Winooski, VT, USA) using the microplate reader. (3) Luciferase intensities were first normalized by Cypridina TK control, and then by using empty vector sample reads as control. Finally the changesin the Luc intensity were quantified as (sample normalized reads)/(empty vector normalized reads). All the experiments were repeated 3 times and select median as criteria.
Nude mice experiment. In the mice experiment, all the experiment complied with National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978). Female nude mice (Beijing Wei-tong Li-hua Laboratory Animals and Technology, Beijing, China) were first injected with 0.5 million MDA-MB-231 cell in the mammary gland fat pad; the mice were divided into control miRNA and hsa-miR-4756-3p mimic groups, each group had 5 mice. The mice were then treated with control miRNA and hsa-miR-4756-3p inhibitor intraperitoneally using DOPC liposomes for 2 times a week. After 1 month, mice were dissected and primary tumour diameters were calculated. The immunohistochemistry (IHC) experiment was carried out by Xue bang company (Beijing, China).

Construction of highly metastatic MDA-MB-231 cells.
Normally metastasis node of MDA-MB-231 cell is diffused and unstable. To get highly metastasis ability 231, we first injected labelled MDA-MB-231 cell with GFP, and then injected it in the mammary gland of nude mice. After 2 months, we extracted GFP + cells from lung using FACS sorting (Aria 3, BD Biosciences, San Jose, CA). We then injected the selected cells in nude mice mammary gland again. After 3 such rounds, we got 231 with high metastasis ability; we termed these cells as trained 231 cells.