Inhibition of PIM1 attenuates the stem cell–like traits of breast cancer cells by promoting RUNX3 nuclear retention

Abstract Finding out the driver gene critical for the maintenance of breast cancer stem cells (BrCSCs) is important for designing a new strategy to eradicate these cells to improve patient's prognosis. Here, in our study, we revealed that PIM1, an oncogenic serine‐threonine kinase and a well‐proven contributor to the tumorigenesis of breast cancer, was involved in BrCSCs regulation and promised to be a new target for eradicating BrCSCs. In brief, PIM1 could enhance the stem cell–like traits of breast cancer cells by promoting the phosphorylation and cytoplasmic localization of RUNX3. The nuclear dislocation of RUNX3 disabled this tumour suppressor and led to breast cancer cells gaining stem cell–like traits. Inhibition of PIM1 significantly induced the nuclear retention of RUNX3, recovered its transcriptional function and attenuated the stem cell–like properties of breast cancer cells. Those findings deepened our understanding of PIM1's oncogenic effect, underlining the significance of PIM1 in designing a new strategy aimed at BrCSCs.

to efficiently eradicate these cells, thus improving patient's prognosis.
PIM1 belongs to the PIM family of serine-threonine kinases, which consists of three proto-oncoproteins, PIM1, PIM2 and PIM3. 4 The pim-1 proto-oncogene was first identified as a common proviral insertion site associated with murine leukaemia virus-induced lymphomagenesis, and its oncogenic activity was verified with transgenic mice overexpressing Pim1 in the lymphoid compartment. 5 The oncogenic roles of PIM1 were verified in solid tumours as colorectal cancer, 6 hepatoma 7 and gastric cancer. 8 Knocking out all three PIM isoforms had limited side effects on mice, 9 which suggested targeting at PIM kinases could be a new safe anti-tumour strategy. PIM1 was reported to phosphorylate a variety of cell cycle-controlling proteins thus enhancing cancer cell proliferation. 10 In TNBC, PIM1 was shown to counteract the increased sensitivity to apoptosis induced by MYC activation. 7,11 However, the in-depth oncogenic mechanism of PIM1 is not well-elucidated, especially concerning its effect on breast cancer stem cells (BrCSCs).
RUNX3 belongs to the family of Runt-related transcription factors (RUNX), and the RUNX family was identified to play a pivotal role in both normal development and neoplasia. 12 RUNX3 was well identified to function as a tumour suppressor, and its inactivation was associated with tumorigenesis in lung adenocarcinoma, intestinal adenocarcinoma, colorectal cancer and gastric cancer. [12][13][14][15] In breast cancer, RUNX3 inactivation was reported to be related to tumorigenesis 16 and YAP-mediated stem cell-like traits. 17 Cytoplasmic mislocation is an important mechanism by which RUNX3 loses its antitumour activity. RUNX3 can be phosphorylated by a spectrum of oncogenic kinases, like Pin1, Src, Pak1, to translocate from nucleus to cytoplasm, thus leading to its subcellular mislocation in human breast, pancreatic and gastric cancer. [18][19][20] However in breast cancer, whether PIM1 acts as an upstream regulator of RUNX3 to phosphorylate it and promote its subcellular dislocation remains unclear and whether this mechanism plays a part in BrCSC-regulating effect of RUNX3 is hardly referred before.
In this study, we revealed that inhibition of PIM1 kinase could attenuate the stem cell-like traits in breast cancer by rescuing the nuclear expression of RUNX3. We demonstrated that PIM1 could phosphorylate RUNX3 to facilitate its cytoplasmic retention, thus suppressing the transcriptional activity of RUNX3 and promoting breast cancer to gain BrCSC-like traits. After PIM1 inhibition, RUNX3 could re-localize to the nucleus and regain its anti-BrCSC function. Moreover, RUNX3 was indispensable for the anti-BrCSC effects of PIM1 inhibition. This finding suggested the important role of PIM1/RUNX3 axis in the regulation of BrCSC biology and offered new targets for eradicating BrCSC population.

| Tissue microarrays
Tissue microarray (TMA) blocks consisting of 213 breast cancer cases were obtained from Department of Pathology, The Affiliated Hospital of Xuzhou Medical University. TMA blocks were constructed following the clinical ethic guidelines. Ethics approval to perform this study was obtained from the Human Research Ethics Committee of the Xuzhou Medical Affiliated Hospital.

| Immunohistochemistry (IHC) assay
Rehydrated slides taped from TMA block were boiled in antigen retrieval solution at 96°C for 40 minutes, then treated with serumfree blocking solution (Beyotime) and incubated overnight at 4°C in a diluent solution (Beyotime) supplemented with monoclonal antibody targeting at RUNX3 (D236-3, MBL, Japan) or PIM1 (sc-374116, Santa Cruz, USA). A peroxidase-3, 3′-diaminobenzidine-based detection system (Zsbio) was used to detect the immunoreactivity. H-score was calculated by multiplying the staining intensity (ranged from 0 to 3) with 100× percentage of positively stained area to obtain a number scaled 0-300. The scoring was performed by a single pathologist

| Immunofluorescence (IF) assay
The immunofluorescence assay was conducted as described. 22 In brief, slides were fixed in 4% paraformaldehyde and blocked with 5% BSA, followed by incubation with anti-PIM1 or anti-RUNX3 antibody in blocking solution at 4°C overnight. Wash the slides using 1 × PBS (0.1% Tween-20) for 3 times and incubate them in blocking solution with goat anti-rabbit IgG 488 or goat antimouse IgG 549 for 30 minutes, followed by DAPI for another 5 minutes. Slides were mounted and imaged with a fluorescence microscope (ZEISS AXIO, Germany). The quantification of immunofluorescence intensity was conducted in Image J_v1.8.0 by measuring the staining intensity of cytoplasm or nucleus for every single cell and calculating the ratio cytoplasm/(cytoplasm + nucleus) or nucleus/(cytoplasm + nucleus).
Data were presented in triplicate by measuring three images randomly captured from different fields.

| Cell culture
The breast cancer cell lines T47D, MCF-7, SK-BR-3, BT-549 and MDA-MB-231, as well as non-malignant MCF-10A cells, were stored in our laboratory. All cell lines were maintained as recommended by the suppliers. The cell lines were authenticated by short-tandem-repeat (STR) analysis every year. All cell lines were cultured following standard procedure.

| Cell transfection
Transfection of plasmids or siRNA was performed using Lipofectamine 2000 (Invitrogen) according to the product guidance.

| RNA isolation and quantitative real-time PCR (qRT-PCR) analysis
Total RNA was extracted by RNA isolator reagent (Vazyme) according to the manufacturer's instructions. cDNA was generated using All-In-One RT MasterMix (ABM) according to the manufacturer's instructions. Real-time PCR was carried out with ABI Step One Plus Real-Time PCR system (ABI) using primers specific for the detection of Cdkn1a (p21), and the values were normalized relative to the expression of endogenous control gene GAPDH. Sequences of primers used for qRT-PCR in this study are listed as below:

| cDNA expression constructs
The complete PIM1 protein coding sequence (NCBI reference sequence NM_001243186.1) was subcloned into the pCMV-

| Western blot (WB) analysis and Immunoprecipitation (IP)
Cells were lysed by scraping them into RIPA lysis buffers supplemented with 1% PMSF. The cell lysates were maintained on ice with vortex every 10 minutes and then subjected to centrifuge at 4°C, 11 000 g. The supernatant was harvested for BCA protein assay (Beyotime, China). In some cases, nuclear and cytoplasmic proteins were separated using Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime).
Primary antibodies used in this study were listed below:

| ALDH1 activity assay
ALDH1 activity of breast cancer cells was assessed by ALDEFLUOR kit (STEMCELL) according to the instruction provided by the manufacturer. In brief, adjust the cell sample to a final concentration of 10 6 cells/mL. 5 μL activated ALDEFLUOR™ Reagent was added to 1 mL cell suspension (test tube) and immediately transfer 0.5 mL to another tube containing 5 μL ALDEFLUOR™ DEAB Reagent (Control tube). Incubate "Test" and "Control" tubes for 30-60 minutes at 37°C. Following incubation, centrifuge all tubes for 5 minutes at 250 g and remove the supernatant. Resuspend the cell pellet in 0.5 mL ALDEFLUOR™ Assay Buffer and subject the cell suspension to flow cytometric analysis.

| Luciferase reporter assay
The promoter region of Cdkn1a (the sequence was from Genecopoeia) was cloned into pGL3-basic vector. Nonsense sequence control siRNA (NC) or siPIM1 was co-transfected with pGL3-Cdkn1a and

| 3D Mammosphere formation assay
Cells were detached from plates by Accutase (Invitrogen) and resuspended in MammoCult medium (Stem Cell Technologies). The single-cell suspension prepared was mixed with methylcellulose and re-plated in 24-well ultra-low attachment plate (Corning).
The mammosphere colony formed after 2-week culture in a CO 2 incubator. The colony numbers were counted, and the diameters of the colony were measured under an inverted microscope (Olympus).

| Clonogenic assay
Detach cells from the tissue culture plate using 0.25% Trypsin-EDTA solution and resuspend cells in serum-free cell culture media to a final concentration of 10 6 cells/mL. An appropriate number of cells (3000-5000) were seeded per chamber on a 24-well plate.
Incubate the plates in a CO 2 incubator at 37°C for 1 week until cells in control group have formed colonies with substantially good size. Add 1 mL 4% paraformaldehyde fix solution per well and leave the plates at room temperature (RT) for 5 minutes. Remove fixation solution, and add 1 mL 0.5% crystal violet solution per well and incubate the plate at RT for 2 hours. Remove crystal violet carefully and immerse the plates in distilled water to rinse off extra crystal violet. Air-dry the plates and photograph the colonies by a Zeiss microscope.

| Proliferation assay
After transfection for 24 hours, cells were detached from plates by 0.25% Trypsin-EDTA solution and resuspend cells in serum-free cell culture media to 106 cells/mL. 3000-5000 cells were seeded per chamber on a 96-well plate. Incubate the plates in a CO 2 incubator at 37°C for another 48 hours. 10 μL CCK8 was added to each cell and incubate the plate for 3 hours.
Microplate reader (Thermofisher) detected the absorbance at 450 nm. After deducting the blank, the proliferation rate of cells was calculated relative to control group.

| Statistical analysis
All data were obtained from at least three independent experiments (n ≥ 3) and presented as the mean ± SD (standard deviation). Statistical analyses (two-tailed t test, Pearson correlation, chi-square test, survival analyses, log-rank test and Tukey's multiple-comparison test) were performed using Prism 6 (version 6.0f) from GraphPad Software, Inc Contingency table and progressionfree survival (PFS) were also created and analysed using Prism 6 (version 6.0f). P < .05 was considered to indicate statistical significance throughout this study.

| PIM1 facilitated the stem cell-like traits in breast cancer cells
It was well mentioned that PIM1 played pivotal roles in the tumorigenesis of breast cancer. As expected, PIM1 was pervasively expressed across breast cancer cell lines irrespective of hormone receptor status with the exception of MCF-10A, a normal human breast epithelial cell line, which barely expressed PIM1 ( Figure S1A).
We also performed IHC staining on a TMA consisting of 213 breast cancer cases. The expression of PIM1 was not correlated with conventional clinicopathological parameters such as age, tumour subtype, histological grade, clinical stage or Ki67 expression. However, high PIM1 expression was related to worse PFS and a high proportion of BrCSC population (CD44+/CD24−) (Table 1, Figure 1A,B).
To elucidate the contribution of PIM1 to BrCSCs, we used clonogenic assay to evaluate the impact of PIM1 inhibition on the ability of breast cancer cells to form colonies. As shown in Figure 1C,D, the clonogenic ability was reduced when knocking down PIM1 using siPIM1. However, knocking down PIM1 did not affect the proliferation rate, indicating the impaired colonies formation ability after PIM1 knockdown was not due to inhibited cells growth ( Figure 1E). We then assessed the ALDH1 activity following PIM1 knockdown using flow cytometric assay. As expected, knockdown of PIM1 significantly down-regulated ALDH1+ population in MCF-7 ( Figure 1F,H) and MDA-MB-231 ( Figure 1G,I) cells.

| RUNX3 mediated the anti-BrCSC effects of PIM1 inhibition
PIM1 was revealed to be able to phosphorylate RUNX3 at Runt region of RUNX3, promoting RUNX3 to translocate from nucleus to cytoplasm, 23,24 and RUNX3 was recently reported as negative regulator of BrCSCs. 17 We then hypothesized that RUNX3 mediated

| RUNX3 attenuated the epithelialmesenchymal transition (EMT) mediated by PIM1 overexpression
Cell invasion and EMT were often intricately related to stem cell-like and reversed by RUNX3(4A). And this was in accordance with the anti-invasion effect of PIM1 knockdown and also the further proinvasion effect of RUNX3 knockdown ( Figure 2F). We also noted that RUNX3(4A), which could not be phosphorylated by PIM1, more profoundly reversed the pro-EMT effects of PIM1. And this implied us that PIM1 might exert both its pro-BrCSC and pro-EMT effects by phosphorylating RUNX3 due to the fact that breast cancer cells could gain stem cell-like traits by undergoing EMT process. 25-27

| PIM1 phosphorylated RUNX3 to promote its nuclear exportation
Next, we sought to confirm whether PIM1 could regulate RUNX3

| PIM1 inhibition rescued RUNX3 nuclear expression
Then, we sought to interrogate whether PIM1 inhibition could rescue RUNX3 nuclear expression. PIM1 knockdown was shown to promote endogenous RUNX3 to relocate in the nucleus of MCF-7 and T47D ( Figure 6A Data were expressed as means ± SE of three independent experiments. Statistical significance was assessed using unpaired Student's t test and one-way ANOVA (***P < .001; **P < .01; and *P < .05) could promote RUNX3 target gene transcription ( Figure 6G). In line with this, RUNX3(4A) was shown to recapitulate the effects exerted by PIM1 knockdown on Cdkn1a (p21) expression ( Figure 6H), which suggested that PIM1 inhibition could promote RUNX3 target genes expression by rescuing RUNX3 nuclear expression.
AZD1208, well-acknowledged PIM1 inhibitors, was shown to dose dependently promote RUNX3 nuclear expression and inhibit RUNX3 cytoplasmic expression in T47D, MCF-7, BT-549 and MDA-MB-231 cells ( Figure 7A,B), without impairing the cell growth ability within 48 hours ( Figure 7C,D). And this further corroborated our hypothesis that inhibiting PIM1 kinase could facilitate RUNX3 nuclear redistribution.

| PIM1 facilitates RUNX3 cytoplasmic retention in breast cancer tissues
To validate that PIM1 mediates the cytoplasmic translocation of with different hormone receptor status ( Figure S3). In accordance with these results, both cytoplasmic RUNX3 expression and PIM1 expression were higher in TNBC cases when compared with that in cases of hormone receptor-positive types ( Figure 8B, upper-panel); and nuclear RUNX3 and PIM1 expression were lower in TNBC when compared with that in hormone receptor-positive types ( Figure 8B, lower panel). All these results further suggested that PIM1 could facilitate RUNX3 cytoplasmic retention, and this was also confirmed in the representative images of PIM1/RUNX3 IHC staining.
RUNX3 was a well-acknowledged tumour suppressor as a transcriptional factor. And this was in accordance with our results that cytoplasmic RUNX3 retention and nuclear RUNX3 dislocation was associated with higher histological grade ( Figure 8D). Higher nuclear RUNX3 and lower cytoplasmic/higher nuclear PIM1 expression indicated lower Ki67 staining intensity ( Figure 8E). All these clinicopathological results suggested that RUNX3 dislocation may mediate the tumorigenic effects of PIM1 in breast cancer.

| D ISCUSS I ON
Pim oncogenes are overexpressed in a wide range of tumours from a haematological or epithelial origin. 4 The mechanisms by which  RUNX3 was referred by other groups to mediate the epithelial-mesenchymal transition induced by TGF-β and play a part in the stem cell-like traits induced by aberrantly activated Wnt signalling in gastric cancer 35 and to reduce cancer stem cells in hepatocellular carcinoma by suppressing Jagged1-Notch signalling. 36 However, the role of RUNX3 in breast cancer stem cell was barely mentioned.
We confirmed in our work that RUNX3 could attenuate the stem cell-like traits in breast cancer cells and mediate the antitumour effects of PIM1 inhibition in this way. We also firstly reported the "pro-stemness" effects of PIM1 in breast cancer cells. In consideration of the fact that stem-like traits were intricately correlated with tumour metastasis, relapse and resistance to chemical agents, 37,38 we suggested that PIM1 could promote tumour metastasis by facilitating RUNX3 nuclear dislocation. More work needed to be conducted to interrogate whether PIM1 inhibition could suppress tumour metastasis in breast cancer. Targeting cancer stem cells (CSCs) population was for a long time being a research hot spot. 37 However, due to the ever-changing tumour micro-environment (TME) and intrinsic tumour biology, CSC population within tumour bulk is hardly druggable. To find out the "driver gene" in CSC population is important for developing new breast cancer treatment strategies. And in this paper, we revealed PIM1 to be promising for targeting at to wipe out CSC population specifically due to its regulation on RUNX3 cytoplasmic translocation, and inhibition of PIM1 could attenuate stem cell-like traits of breast cancer cells.
In conclusion, our work demonstrated that PIM1 could site-specifically phosphorylate RUNX3 to promote its translocation from nucleus to cytoplasm by protein-protein interaction. And we also revealed PIM1 inhibition as a new strategy to target at cancer stem cells population and RUNX3 was essential to the anti-BrCSC effect of PIM1 inhibition.

This work was supported by a grant from Special Foundation of
President of Xuzhou Medical University: the clinicopathological and molecular biological study on basal-like breast carcinoma (grant number: 09KJZ05).

CO N FLI C T O F I NTE R E S T
The authors declared that they have no conflicts of interest to this work.

AUTH O R CO NTR I B UTI O N
CX and HL designed and conducted most of the experiments. QY, YL and CC conducted the TMA construction and the IHC experiments.
CX, CC, QL and YL conducted the plasmid constructions. CX, HL and DM analysed the data and wrote the manuscript.

E TH I C A L A PPROVA L
The study was approved by the Ethics Committee of the First Affiliated Hospital of Xuzhou Medical University; informed consent was waived for this study.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.