microRNA‐128‐3p overexpression inhibits breast cancer stem cell characteristics through suppression of Wnt signalling pathway by down‐regulating NEK2

Abstract Emerging evidence has reported that dysregulation of microRNAs (miRNAs) participated in the development of diverse types of cancers. Our initial microarray‐based analysis identified differentially expressed NEK2 related to breast cancer and predicted the regulatory microRNA‐128‐3p (miR‐128‐3p). Herein, this study aimed to characterize the tumour‐suppressive role of miR‐128‐3p in regulating the biological characteristics of breast cancer stem cells (BCSCs). CD44＋CD24−/low cells were selected for subsequent experiments. After verification of the target relationship between miR‐128‐3p and NEK2, the relationship among miR‐128‐3p, NEK2 and BCSCs was further investigated with the involvement of the Wnt signalling pathway. The regulatory effects of miR‐128‐3p on proliferation, migration, invasion and self‐renewal in vitro as well as tumorigenicity in vivo of BCSCs were examined via gain‐ and loss‐of‐function approaches. Highly expressed NEK2 was found in breast cancer based on GSE61304 expression profile. Breast cancer stem cells and breast cancer cells showed a down‐regulation of miR‐128‐3p. Overexpression of miR‐128‐3p was found to inhibit proliferation, migration, invasion, self‐renewal in vitro and tumorigenicity in vivo of BCSCs, which was further validated to be achieved through inhibition of Wnt signalling pathway by down‐regulating NEK2. In summary, this study indicates that miR‐128‐3p inhibits the stem‐like cell features of BCSCs via inhibition of the Wnt signalling pathway by down‐regulating NEK2, which provides a new target for breast cancer treatment.

factors, such as the tumour complication level and patient-related factors, like the patients' preferences. 4,5 Mitotic kinesins, whose overexpression has been found in diverse cancers, are demonstrated to serve as markers for the prognosis of breast cancer offering a potential treatment for this disease. 6 However, treatment of breast cancer still remains a great challenge owing to poor diagnosis and prognosis. 7 Therefore, it is very important to find new therapeutic targets for breast cancer treatment.
As non-coding RNAs, microRNAs (miRNAs) regulate the expression of genes at the post-transcriptional level, including genes that are involved in the development of various types of cancers, including breast cancer. 8,9 This is clearly exemplified in microRNA-128-3p (miR-128-3p), which is encoded in miR-128. miR-128-3p is enriched in central nervous system 10 and has been reported to play an essential role in the prognosis of breast cancer. 11 As Wnts are associated with proliferation, self-renewal, and migration of stem cells, the Wnt signalling pathway is usually regarded as therapeutic targets in different kinds of tumours. 12 Activating the Wnt signalling pathway can promote the malignant proliferation of breast cancer cells. 13 NIMArelated kinase 2 (NEK2) is a kind of mitotic kinases involved in carcinogenesis and development of various cancers. The up-regulation of NEK2 in several types of tumours indicates that it might serve as a potential target for cancer therapy. 14 In addition, a previous study has shown that the expression of NEK2 is always up-regulated in breast cancer. 15 Although both miR-128-3p and NEK2 have been studied in the growth of breast cancer, 16,17 their exact roles in breast cancer remain unclear. Therefore, this study was designed to investigate the specific relationship between miR-128-3p and NEK2 and their effects on the development of breast cancer. In the present study, we hypothesized that overexpression of miR-128-3p may regulate the biological characteristics including proliferation, migration and invasion of breast cancer stem cells (BCSCs) by mediating the expression of NEK2.

| Ethics statement
The study was approved by the Institutional Review Board of Chongqing Renji Hospital, University of Chinese Academy of Science. Written informed consent was obtained from each participant. Nude mice were used for in vivo studies and were cared for in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health.

| Microarray analysis
The Gene Expression Omnibus (GEO) database (https://www.ncbi. nlm.nih.gov/geo/) was used to search for breast cancer expression profiles, and "limma" package in the R language was used for differential expression analysis with |logFC| > 2 and P < 0.05 as screening threshold of differential genes. The heat map of differentially expressed genes was constructed using the "pheatmap" package.
Known breast cancer-related genes were retrieved in DisGeNET (http://www.disge net.org/web/DisGe NET/menu), a database with substantial human disease-related genes and variants that are publicly available. The STRING database (https://strin g-db.org/) was employed for correlation analysis of the known genes and differentially expressed genes in breast cancer, and a genetic interaction network was constructed using Cytoscape software.

| Sorting and identification of BCSCs
Sorting of BCSCs: Human breast cancer cell line MCF-7 was cultured in an incubator using RPMI-1640 (Gibco) culture medium containing 10% foetal bovine serum (FBS) at 37°C with 5% CO 2 and saturated humidity. Collected cells in logarithmic growth phase were made into single-cell suspension, counted and centrifuged at 503.1 (g) for 10 minutes (r = 11 cm) with the supernatant discarded. Then, the cells were resuspended and 1 × 10 7 cells were mixed with 10 μL CD24-biotin and incubated at low temperature for 15 minutes.
After being rinsed, centrifuged, and resuspended, the cells were mixed and incubated with 20 μL of anti-biotin MicroBeads at low temperature for 15 minutes. After being rinsed, the cells were centrifuged at 503.1 (g) for 10 minutes (r = 11 cm) and resuspended in 500 μL buffer. CD24 −/low cells were obtained by negative sorting using magnetic-activated cell sorting (MACS). CD44 + CD24 −/low cells were obtained through positive sorting from CD24 −/low cells using the same method. The detection results were analysed by EXPO32 ADC Analysis software when the excitation light wavelength was 88 nm. 2. Detection by microsphere culture: Cells with good growth state were resuspended in Dulbecco's modified Eagle's medium (DMEM)/F12 medium containing epidermal growth factor (EGF) (20 ng/mL), basic fibroblast growth factor (bFGF) (10 ng/mL), insulin (5 mg/L) and B27. The cells were seeded into a plastic corning at 2 × 10 4 mL and cultured at 37°C, 5% CO 2 and saturated humidity with the culture solution changed once every 2-3 days. On the 7th day, the number and volume of the microspheres were observed.
The cDNA fragment of the binding site-mutant. NEK2 3′-UTR was constructed by point mutation and inserted into the pmirGLO vector. The inserted sequence was verified by sequencing. The pmirGLO-NEK2 or pmirGLO-mutant (MUT) homeobox C6 (HOXC6) recombinant vector was cotransfected into HEK293T with miR-128-3p mimic (miR-128-3p overexpression sequence) or miR-NC (negative control sequence) by lipofection. The cells were incubated for 48 hours, harvested and lysed. The lysate supernatant (100 μL) was taken and added to 100 μL of Renilla luciferase assay solution to detect Renilla luciferase activity.
In addition, 100 μL of the lysate supernatant was mixed with 100 μL of firefly luciferase detection reagent to detect the firefly luciferase activity. The SpectraMax M5 multi-mode plate reader (Molecular Devices) was used to detect Renilla luciferase activity and firefly luciferase activity, respectively with an interval of 2 seconds and a measurement time of 10 seconds.

| Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
The total RNA was extracted from cells and tissues using an RNA extraction kit (Invitrogen). Designed primers were synthesized by Takara (Table 1). The RNA concentration was measured, and cDNA was synthesized by reverse transcription according to the instructions of the reverse transcription kit (DRR047S, TaKaRa Biotechnology Co. Ltd). The relative expression of miR-128-3p was determined with U6 as the internal reference, whereas the other genes were detected with GAPDH as the internal reference. 18 Each experiment was repeated at least 3 times, and the expression of miR-128-3p was calculated by the 2 −ΔΔCt method.
Using the 2 −△△CT method, we calculated the relative transcriptional level of target gene mRNA: △Ct = △CT (experimental group) − △CT (control group); △Ct = Ct (target gene) − Ct (internal reference). Relative transcriptional level of target gene mRNA = 2 −△△CT . 19 Each specimen was divided into three parallel tubes and the experiment was conducted three times independently.

| Western blot analysis
The total protein of tissues and cells was extracted. After ice bath- with Tris-buffered saline-Tween (TBST) and visualized using HRP electroluminescence (ECL). Grey value analysis of the target bands was performed using ImageJ software, and the experiment was repeated three times independently.

| Sphere formation assay
A total of 1 × 10 4 BCSCs were seeded into the low-adsorption 96-well plates and cultured in serum-free DMEM/F12 medium containing 20 ng/mL EGF and 20 ng/mL FGF-β with the solution semi-quantitatively changed every 2 days. After continuous culture for 10 days, the BCSCs were photographed under the inverted microscope and counted.

| Soft agar colony formation assay
Agarose at 0.7% low melting point was prepared in fresh DMEM and stored at 4°C for subsequent use. The 0.7% agarose was heated and melted, and 2 mL agarose was transferred to a petri dish with the

| Limiting dilution assay (LDA) in vivo and subcutaneous tumour formation in NOD-SCID mice
The cells were seeded in a low-adherence culture plate, and after 7 days of culture, BCSC spheres in each group were collected, cen-

| Statistical analysis
Statistical analysis was performed using the SPSS 21.0 software (IBM Corp.). Normal distribution and homogeneity of variance tests were conducted concerning all data. The enumeration data were expressed using cases or percentage (%), and the measurement data were presented as the mean ± standard deviation. Comparisons between two groups were analysed using t test, and Welch's correction was used for unequal variances. Data analysis among multiple groups was performed by one-way analysis of variance. The data analyses at different time-points were performed using repeatedmeasures analysis of variance. The data of skewed distribution were analysed by rank-sum test. All experiments were repeated three times. A P < 0.05 was considered statistically significant.

| NEK2 and miR-128-3p are involved in the development of breast cancer
From the GEO database, the breast cancer expression profile GSE61304 was retrieved. Through differential expression analysis  of the breast cancer samples and normal control samples in the expression profile, 177 differentially expressed genes were obtained, of which 58 differential genes were highly expressed, whereas 119 differential genes were down-regulated in breast cancer. The heat map of expression of the first 30 differentially expressed genes was constructed ( Figure 1A). To further screen for breast cancer-associated genes, the known genes for breast cancer were searched in the DisGeNET database and the top 10 genes with the highest scores were selected for subsequent analysis (Table 2). Correlation analysis of the first 10 significant differential genes in the GSE61304 expression profile and the known genes of breast cancer was conducted and the gene interaction network map was constructed ( Figure 1B), finding that among the top 10 differential genes, NEK2 gene was at the core of the network map and had an interaction with genes known to be related to breast cancer such as AKT1 and BRCA1. The pathway. 22 In order to further understand the upstream regulation mechanism of NEK2 gene, the regulatory miRNAs of NEK2 were predicted through DIANA, miRDB and other databases ( Figure 1D).
Three miRNAs mutually emerged from the four databases, among which miR-128-3p has the highest scoring. In summary, in breast cancer, miR-128-3p is likely to influence the development of breast cancer via the Wnt signalling pathway by regulating the NEK2 gene.

| CD44 + CD24 −/low exhibits high selfrenewal capacity
The number of cells was counted before and after cell sorting by MACS method. The flow cytometric data showed that the ratio of CD44 + CD24 −/low cells in MCF-7 cells was 3.41%. After purification of CD44 + CD24 −/low cells in the sorted CD44 + CD24 −/low cell subset, we obtained a CD44 + CD24 −/low cell ratio of over 90% (Figure 2A), which was qualified for the requirements of further experiments. To further identify the sorted cells, the sorted CD44 + CD24 −/low cells and non-CD44 + CD24 −/low cells were cultured in serum-free DMEM/F12 medium containing EGF, bFGF, B27 and insulin, respectively. After 7 days, the non-CD44 + CD24 −/low cell subset had fewer formed microspheres, and the formed microspheres were smaller in volume. The CD44 + CD24 −/low cell subset grew in a suspended state with more microspheres formed, and the microspheres were larger ( Figure 2B). These observations demonstrated that the CD44 + CD24 −/low group cells were consistent with the high self-renewal ability of BCSCs, indicating that we successfully isolated BCSCs.

| miR-128-3p is poorly expressed in BCSCs and breast cancer cells
The present study selected 52 breast cancer patients to detect the expression of miR-128-3p in breast cancer tissues and adjacent tissues using RT-qPCR, whose results showed that the expression of miR-128-3p was lower in breast cancer tissues than in adjacent tissues has lower expression in the CD44 + CD24 − cell subset compared to CD44 − CD24 − subset ( Figure 3D).

| Overexpressed miR-128-3p inhibits proliferation, migration and invasion of BCSCs
EdU assay was applied to analyse the effect of miR-128-3p on the proliferation of BCSCs and the results ( Figure 4A) showed that after inhibition of miR-128-3p, the proportion of EdU-positive cells was significantly higher than that in response to inhibition of miR-128-3p-

F I G U R E 3 miR-128-3p is down-regulated in BCSCs and breast cancer cells.
A, Relative expression of miR-128-3p in breast cancer tissues and adjacent tissues (*P < 0.05 compared with adjacent tissue); B, Relative expression of miR-128-3p in breast cell lines (*P < 0.05 compared with MCF-10A); C, Relative expression of miR-128-3p in non-spheroids and spheroids (*P < 0.05 compared with non-spheroids); D, Relative expression of miR-128-3p in CD44 − CD24 − and CD44 + CD24 − (*P < 0.05 compared with CD44 − CD24 − ); U6 was used as the internal reference of the expression level of miR-128-3p and GAPDH was as the internal reference of NEK2, Wnt, β-catenin, CD24, CD44, Nanog, Oct-4 and Pygo2. The measurement data are expressed in the form of mean ± standard deviation. Difference analysis between two groups was performed by t test, and the data analysis among multiple groups was performed by one-way analysis of variance  Figure 5C). In addition, we also performed soft agar colony formation to detect the colonyforming ability of BCSCs. The results showed that the number of clones formed by cells has increased when miR-128-3p was inhibited, and the number of clones formed by cells has significantly decreased when miR-128-3p was overexpressed, revealing that overexpression of miR-128-3p inhibited the amplification of BCSCs ( Figure 5D). The aforementioned results provide evidence that self-renewal of BCSCs could be suppressed in response to the up-regulation of miR-128-3p.

| Overexpression of miR-128-3p inhibits tumorigenicity and tumour growth of BCSCs in vivo
To further validate the regulation of miR-128-3p on BCSCs, we investigated the effects of miR-128-3p in vivo. Initially, LDA in vivo was applied to examine the effect of miR-128-3p on tumorigenicity and stem cell ratio in BCSCs. The results showed that the number of tumours and cancer stem cell proportion was significantly higher when miR-128-3p was inhibited than that of the corresponding NC group (Table 3, Figure 6A). In addition, we further explored the effect of miR-128-3p on tumour growth using a mouse subcutaneous

| NEK2 is a direct target gene of miR-128-3p
According to an analysis by online analysis software, a specific binding region was identified between the 3′-UTR of the NEK2 gene and the miR-128-3p sequence ( Figure 7A). The target relationship between NEK2 and miR-128-3p was verified using dual-luciferase reporter assay ( Figure 7B). The results showed that overexpression of miR-128-3p has significantly inhibited the luciferase activity of 3′-UTR in NEK2 wild-type (WT) compared to the NC group (P < 0.05), whereas miR-128-3p had no significant effect on the luciferase activity of 3′-UTR in the NEK2-MUT (P > 0.05). The results of RT-qPCR ( Figure 7C) and Western blot analysis ( Figure 7D,E) showed that compared to the corresponding NC group, the mRNA and protein expression of NEK2 has significantly decreased after overexpressing

F I G U R E 4
Overexpressing miR-128-3p decreases proliferation, migration and invasion of BCSCs. A, Cell proliferation ability after transfection detected by EdU assay (×200); B, Cell migration ability detected by Transwell assay (×200); C, cell invasion ability detected by Transwell assay (×200); *P < 0.05 compared with the corresponding NC group; the experiment was repeated three times, the measurement data are expressed in the form of mean ± standard deviation, and the data analysis among multiple groups was performed by one-way analysis of variance

| miR-128-3p inhibits proliferation, migration and invasion of BCSCs by silencing NEK2
The results of the EdU assay ( Figure 8A) showed that the propor-

| miR-128-3p inhibits self-renewal of BCSCs by down-regulating NEK2
RT-qPCR and Western blot analyses were conducted for the determination of mRNA and protein expression of surface markers (CD24 and CD44) and stemness transcription factors (Nanog and Oct-4) of BCSCs after transfection. In addition, sphere formation and soft agar colony formation were conducted to detect the ability of sphere formation and colony formation respectively.
The results showed that the mRNA and protein expression of CD24, CD44, Nanog and Oct-4 was decreased in the si-NEK2 group compared to the corresponding NC group ( Figure 9A,B), and sphere formation, colony formation and proliferative capacity were decreased significantly ( Figure 9C,D). Compared with the miR-128-3p inhibitor + si-NEK2-NC group, the mRNA and protein expression of CD24, CD44, Nanog and Oct-4 (Figure 9 A,B), and the sphere-forming ability, colony-forming ability and cell proliferation ( Figure 9C,D) were also declined in the miR-128-3p inhibitor + si-NEK2 group. These results suggest that miR-128-3p inhibits self-renewal of BCSCs by down-regulating NEK2.

| miR-128-3p inhibits tumorigenicity and tumour growth of BCSCs by down-regulating NEK2
Limiting dilution assay in vivo was used to detect the number of tumours and the proportion of tumour stem cells. The results showed that the number of tumours and the proportion of tumour stem cells were lower in the si-NEK2 group compared with those in the corresponding NC group; compared with those in the miR-128-3p inhibitor + si-NEK2-NC group, the number of tumours and the proportion of tumour stem cells were low in the miR-128-3p inhibitor + si-NEK2 group (Table 4, Figure 10A). The subcutaneous tumour-bearing model of mice showed that compared with the blank group and the NC group, the si-NEK2 group exhibited shorter tumour formation time with the slow growth rate and smaller tumours. Compared with the miR-128-3p inhibitor + si-NEK2-NC group, the miR-128-3p inhibitor + si-NEK2 group had a shorter tumour formation time, slower growth rate, and smaller tumours ( Figure 10B-D). These results revealed that up-regulated miR-128-3p could repress tumorigenicity and tumour growth of BCSCs by down-regulating NEK2.

| miR-128-3p inhibits the Wnt signalling pathway by down-regulating NEK2
The TOPFlash plasmid contains a firefly luciferase reporter gene, and the upstream promoter region of luciferase contains three repeat TCF-binding sequences, which can regulate the expression of downstream luciferase according to the activity of β-catenin.
The TCF-binding sequence in the TOPFlash plasmid is mutated, were decreased in the DKK-1 + NEK2 group, and Wnt signalling pathway activity was inhibited; compared with the miR-128-3p inhibitor + si-NEK2-NC group, the levels of TCF-4 and Pygo2 were decreased and the activity of the Wnt signalling pathway was decreased in the miR-128-3p inhibitor + si-NEK2 group ( Figure 11D).
All the results suggest that miR-128-3p can inhibit the Wnt signalling pathway activation via silencing of NEK2.

| D ISCUSS I ON
Breast cancer is the most common tumour among women worldwide, accompanied by high recurrence and mortality. 23  In addition, our findings suggest that overexpression of miR-128-3p can inhibit proliferation, migration and invasion of BCSCs F I G U R E 8 miR-128-3p reduces proliferation, migration and invasion of BCSCs by down-regulating NEK2. A, Cell proliferation ability after transfection detected by EdU assay (×200); B, Cell migration ability detected by Transwell assay (×200); C, Cell invasion ability detected by Transwell assay (×200); *P < 0.05 compared with the corresponding NC group; the measurement data are expressed in the form of mean ± standard deviation, and the data analysis among multiple groups was performed by one-way analysis of variance. The experiment was repeated three times  to inhibit migration and proliferation of breast cancer cells. 32 In line with our study, it has been proved that increased expression of NEK2 can activate the Wnt signalling pathway. 22 A previous study suggests that the β-catenin/Wnt signalling pathway which is closely related to stem cells also play a vital role in tumours. 33 In addition, it has been proven that the expression level of β-catenin in cancer F I G U R E 9 miR-128-3p decreases self-renewal of BCSCs by down-regulating NEK2. A, mRNA expression of BCSC-related markers after transfection detected by RT-qPCR; B, Protein expression of BCSC-related marker after transfection detected by Western blot analysis; C, Sphere-forming ability of BCSCs after transfection by sphere formation assay (×400); D, the colony-forming ability of BCSCs after transfection detected by colony formation assay; *P < 0.05 compared with the corresponding NC group; the measurement data are expressed as mean ± standard deviation, and the data analysis among multiple groups was performed by one-way analysis of variance. The experiment was repeated three times

TA B L E 4 Tumorigenicity of BCSC spheroids
F I G U R E 1 0 miR-128-3p represses tumorigenicity and tumour growth of BCSCs via silencing of NEK2. A, The ratio of stem cells in mice measured by LDA in vivo; B, The size of tumour cells in nude mice after transfection; C, tumour weight in nude mice; D, Volume of tumour formation in nude mice; *P < 0.05 compared with the corresponding NC group; the measurement data are expressed in the form of mean ± standard deviation, and the data analysis among multiple groups was performed by one-way analysis of variance. The data analysis at different time-points was performed using repeated-measures analysis of variance. The experiment was repeated three times stem cells can be down-regulated by miR-128. 34 Then, another study indicates that CD24 and CD44 are cancer stem cells which can promote the development of breast cancer. 35 Furthermore, it has been proven that quercetin 3-methyl ether can down-regulate the expression of Nanog, which is a gene associated with BCSCs. 36 Oct-4 and Nanog play vital roles in breast cancer, as decreased expression of Oct-4 and Nanog can inhibit migration of BCSCs. 37 A previous study has confirmed that when miR-128-3p is overexpressed, migration of invasive breast cancer cells can be inhibited. 38 In addition, NEK2, which could alter cell migration, has been proven to be one of the most predictive genes that are related to metastasis-free survival in breast cancer. 39 Accumulating evidence suggests that decreased expression of NEK2 can repress amplification and proliferation of BCSCs. 40,41 In our study, when miR-128-3p was overexpressed or NEK2 was down-regulated, the Wnt signalling pathway could be inactivated, thus suppressing cell migration, proliferation, and self-renewal of BCSCs.
Based on the previous researches, the present study confirmed that overexpressed miR-128-3p inhibited the proliferation, migration and invasion of BCSCs by suppressing the Wnt signalling pathway by down-regulating NEK2 ( Figure 12). Collectively, this study defines the potential role of miR-128-3p and NEK2 as a therapeutic target in breast cancer treatment by regulating the Wnt signalling pathway. However, the research is still at the preclinical stage, and future studies on the mechanism of action are required.

ACK N OWLED G EM ENT
We acknowledge and appreciate our colleagues for their valuable efforts and comments on this paper.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

AUTH O R CO NTR I B UTI O N S
Yuanwen Chen designed the study. Nian Wu collated the data and designed and developed the database. Lei Liu and Huaying Dong carried out data analyses and produced the initial draft of the manuscript. Yuanwen Chen and Xinao Liu contributed to drafting the manuscript. All authors have read and approved the final submitted manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets generated/analysed during the current study are available.