Loss of PLK2 induces acquired resistance to temozolomide in GBM via activation of notch signaling

Expression profile of GSE68029 [28] and GSE16011 [29] were extracted from Gene expression omnibus (GEO). These transcriptome profiles were preprocessed as previously described [30]. Gene expression datasets of TCGA GBM was extracted from GDC Data Portal (https://​portal.​gdc.​cancer.​gov/​). The limma package [31] was used for identifying the differentially expressed genes in these datasets. The expression difference of individual gene was defined by log₂(Fold change) and adjusted P value, in which log₂FC < − 1 with an adjusted P value < 0.05 was defined as a down-regulated gene. Venn diagram was carried out to illustrate the overlapping down-regulated genes in these datasets.

This study was approved by the Scientific Ethics Committee at the First Affiliated Hospital of Xi’an Jiaotong University (approval no. 2016–18). All the patients’ samples used in this study have obtained necessary consent, and all samples were embedded in paraffin blocks as previously described [30].

Immunohistochemistry was performed as previously described [32, 33]. Anti-PLK2 primary antibodies were purchased from Abcam (cat. no. ab71311). Secondary antibodies including Goat anti-rabbit IgG (cat. no. ab97051, Abcam) and goat anti-mouse IgG (cat. no. ab205719, Abcam) were used in this study. Tissues embedded with paraffin were cut into 4-mm sections followed by deparaffinized, rehydrated, and stained with primary antibodies at 4 °C overnight. The slides were then incubated with secondary antibodies and stained with DAB. Consequently, the slides were counterstained with hematoxylin and images were taken under the light microscope.

The global immunohistochemistry score (GIS) was used to measure the relative expression of PLK2 in different samples. GIS was calculated by the following equation: immunoreactivity score = staining intensity score × positive cell score. The staining intensity score was scored as: 0, negative; 1, weakly positive; 2, moderately positive; and 3, strongly positive. The positive cell score was scored as: 0, negative; 1, < 10% positive; 2, 11–50% positive; 3, 51–80% positive; and 4, > 80% positive. Consequently, an immunoreactivity score > 5 was considered as high expression, and ≤ 5 was defined as low expression.

Lentivirus production and transduction were performed as described previously [30]. The lentiviral overexpression/knockdown PLK2 was designed and synthesized by Genechem (Shanghai, China). These lentiviruses were introduced into U87 and U251 cell lines according to manufacturer’s instruction. Stable clones transduced with PLK2 and shPLK2 were selected for 4 weeks by puromycin. Lentiviral particles packaging the shRNA are targeting PLK2 (5′-TAGTCAAGTGACGGTGCTG-3′) and the scramble control (5′-TTCTCCGAACGTGTCACGT-3′).

GBM cell lines including U87 and U251 were purchased from Cell bank, Type culture collection, Chinese Academy of Sciences (Xi’an China) as previously described [30]. Cells were cultured in DMEM-F12 containing 10% FBS and antibiotics (1% penicillin and streptomycin). U87 and U251 cells were cultured in a humidified condition containing 5% CO₂ at 37 °C.

For in vitro cell proliferation assay, cells were suspended adequately before seeded into 96-well plates at a density of 1 × 10³ cells/well. Cell number was counted by using alamarBlue following the manufacturer’s instruction after indicated treatments.

Cell viability assays were conducted as previously described [30]. In brief, cell number was calculated by cell counter with trypan blue, cells were then seeded into 96-cell plates after suspended adequately at a density of 2 × 10³ cells/100 uL per well. After 24 h of culturing, U87 and U251 cell lines received indicated in vitro chemotherapy using corresponding concentration of TMZ. At last, cell number was counted by alamarBlue and IC50 was calculated by SPSS 22.0.

The Quantitative RT-PCR assays were conducted as previously described [30]. In brief, total RNA was extracted using RNeasy mini kits according to the manufacturer’s instruction, and the concentration of RNA was determined by Nanodrop 2000. qRT-PCR was performed following the synthesis of cDNA according to the standard protocols. GAPDH was used as an internal control. Relative mRNA expressions were calculated by 2^-ΔΔt method. The primer sequences were listed as below:

PLK2: forward, 5′-GCTGATGTCTGGCTGTTCATCAG-3′ and reverse, 5′-CTTCCCTGTAGATCTCACA GTG-3′;

HES1: forward, 5′-AGTGAAGCACCTCCGGAAC-3′ and reverse, 5′-TCACCTCGTTCATGCACTC-3′.

c-MYC: forward, 5′- AAACACAAACTTGAACAGCTAC-3′ and reverse, 5′- ATTTGAGGCAGTTTACATTATGG-3′.

p21: forward, 5′- GGCATAGAAGAGGCTGGTGG-3′ and reverse, 5′- ATGGCGCCTGAACAGAAGAA-3′.

Cyclin D3: forward, 5′- AAACTTGGCTGAGCAGAGCA-3′ and reverse, 5′- GAACAGAGCCAGTCTCCACC-3′.

GAPDH: forward, 5′-ACCCAGAAGACTGTGGATGG-3′ and reverse, 5′-TTCAGC TCAGGGATGACCTT-3′.

Western blot analyses were performed as previously described [32]. Antibodies used in this study were shown as below: Anti-PLK2 primary antibody was purchased from Abcam (cat. no. ab71311). Anti-β-actin antibody was purchased from Abcam (cat. no. ab115777). Anti-NOTCH1, anti-NOTCH2, anti-HES1, anti-c-MYC and anti-cyclin D3 primary antibodies were bought from Cell Signaling Technology (CST) with the catalog number of #4147, #5732, #11988, #18583 and #2936, respectively. Additionally, Anti-Rabbit IgG and Anti-Mouse-IgG were purchased from CST (cat. no. #7074 and #7076, respectively).

Colony formation assays were conducted to detect the tumorigenic potential of U87 and U251 cell lines. Different numbers of stable U87 and U251 cells transduced with lentiviral PLK2 or vector were seeded into 6-well plates under indicating conditions. After cultured for 14 days to form colonies, these cells were fixed with methanol and stained with methylene blue.

The wound healing assay is used to study the migratory ability of U87 and U251 GBM cell lines. Cells were seeded in 6-well plates after adequately suspended. Afterwards, a sterile pipette tip was used to produce a wound line when cells reached the appropriate confluence. These cells were then washed at least 3 times to be cultured in FBS-free medium. Images were taken under inverted microscope at the same location. The leading edges were marked by white lines, and the relative distance of the borders was measure by Image J software.

To construct TMZ resistant cell lines, U87 and U251 were gradually treated with increasing concentration of TMZ. In brief, the beginning concentration of TMZ was set according to the IC50 of U87 and U251 naïve cells. The medium was changed every 2 days to keep the concentration of TMZ at a stable level for 2 weeks. Afterwards, new IC50 was calculated using the pre-treated cell lines and the process was continued until the IC50 of TMZ reached the amount of 500uM. These cells were then used for subsequent experiments.

The usage of experimental animals in this study was approved by the Ethics Committee of the School of Medicine, Xi’an Jiaotong University (approval no. 2016–085). In vivo xenograft model was constructed by using 6-week-old female nude mice. U87 and U251 cells transduced with indicating lentiviruses was suspended and diluted to the density of 1 × 10⁵ cells in 2 uL PBS then slowly injected into the nude mice brains. Each group of treatment contains five mice, and they were monitored until the following symptoms were observed: unsteady gait, arched back, more than 10% weight loss or leg paralysis.

Flow cytometry assays were performed as previously described [30]. U87 TMZ resistant cells were transfected with either empty vector or plk2 overexpression lentivirus then treated with 200uM TMZ for 24 h. DMSO were used as a negative control. The Alexa Fluor® 488 Annexin V/Dead Cell Apoptosis kit was used strictly following the manufacturer’s protocol to measure the cell apoptotic rate.

The transcriptome profiles of previously mentioned datasets were read into R programming and preprocessed by background correction, gene ID transformation and normalization. These data were then ordered by the expression level of PLK2 to divide all samples into two groups (PLK2^high and PLK2^low) by quartile cutoff. Afterwards, these profiles were uploaded into the GSEA software [34] strictly following the guideline of the software to elucidate the enriched KEGG pathways that are significantly enriched in PLK2^low groups with the number of permutations set at 1000.

The ubiquitination level of Notch1 was determined by GST-mediated pull-down assays (Thermo Scientific, Rockford, IL). Briefly, GST-tagged Notch1 was transfected into indicated cell lines (isogenic PLK2 overexpressed cells or their negative control) by lipofectamine according to manufacturer’s instruction. After 24 h, cell lysates were collected and incubated with GST beads for 1 h. Then, GST complexes were washed extensively with lysate buffer and used to perform western blot assays.

All the results in this study are exhibited as mean ± SD (Standard deviation) as described previously [35]. The number of independent replications is presented in the figure legends. Two-tailed t tests were conducted to evaluate statistical differences and One-way ANOVA following Dunnett’s post-test was used to compare the statistics in more than two groups. Kaplan-Meier survival analyses were performed using log-rank test. In addition, all statistical analysis was performed by GraphPad Prism 7 or SPSS 22.0 software, and statistical significance was defined as a two-sided P value less than .05 unless specifically indicated.

As well known, aberrant expression and activation of kinases result in oncogenesis of a wide range of cancers including GBM [36]. Therefore, to investigate the differentially expressed kinase-encoding genes in TMZ resistant GBMs, hierarchical bi-clustering was performed by using previously published GEO database (GSE68029) [28]. The differentially expressed genes were then aligned with the 668-known kinase-encoding gene symbols. As is shown in Fig. 1a and Supplementary Table 1, PLK2 was among the most significantly suppressed kinase-encoding genes in TMZ resistant group. In addition, the differentially expressed genes of GBM versus non-tumor tissue were analyzed by using GSE16011 [29] and TCGA datasets. The results showed that PLK2 was consistently down-regulated in these 2 databases (Fig. 1b, c; Supplementary Table 2, 3). As is illustrated in Fig. 1d, Venn diagram was drawn to find out the overlapping down-regulated genes. Totally, 4 genes were negatively correlated with both TMZ resistance and GBM tissues, including PRKACB, DGKZ, PLK2 and RPS6KA5. Taken together, PLK2 showed significantly differential expression in correlation with GBM and TMZ resistance. Therefore, we picked PLK2 as a candidate gene in this study.

To gain more insight into the clinical relevance of PLK2 expression in GBM, TCGA GBM database were analyzed by using the online server GEPIA (http://​gepia.​cancer-pku.​cn/​) [37]. As is shown in Fig. 2a, PLK2 was significantly down-regulated in GBM compared with non-tumor tissues. Furthermore, Rembrandt database was used to investigate the expression pattern of PLK2 in different pathological subtypes of glioma including astrocytoma, GBM and oligodendroglioma. It is obvious that non-tumor tissues exhibited the highest expression of PLK2 compared with other types of glioma (Fig. 2b). Moreover, expression of PLK2 were found to be remarkably suppressed in all 4 molecular subtypes of GBM including classical (CL), mesenchymal (MES), neural and proneural (PN) when compared to non-tumor tissues (Fig. 2c, Supplementary Fig. 1C). These data suggested that loss of PLK2 was a significant biomarker for glioma/GBM, which lead us to investigate more deeply into the clinical relevance of PLK2 in GBM. Herein, IHC staining was performed to evaluate the expression of PLK2 in samples derived from patients underwent surgical resection in the Department of Neurosurgery, The First Affiliated Hospital of Xi’an Jiaotong University from 2009 to 2015. The results indicated that PLK2 was expressed predominantly in the cytoplasm cytomembrane of low-grade glioma and non-tumor samples, however, the staining was significantly lower in GBM samples (Supplementary Fig. 1A). Kaplan-Meier survival analysis showed that glioma patients with lower expression of PLK2 represented a shorter overall survival when compared with higher PLK2 expression (Fig. 2d). To confirm these findings, 676 patients’ samples from Rembrandt database were analyzed and similar results were observed (Fig. 2e). Although underwent post-surgery radiotherapy or TMZ chemotherapy, the overall survival was unfavorable in PLK2 low expressed patients (Supplementary Fig. 1D and E). Furthermore, the glioma patients were divided into 2 groups according to the GIS of PLK2 expression. The results indicated that PLK2 was likely to be enriched in low grade gliomas compared with GBM (Supplementary Fig. 1B). Taken together, these data demonstrated that knock-down of PLK2 is associated with pathological malignancy and revealed severe outcomes in glioma.

To further study the function of PLK2, exogenous PLK2 was introduced into U87 and U251 cell lines through lentiviral transduction. The transduction efficiency was confirmed by immunofluorescence imaging (Fig. 3a). qRT-PCR (Fig. 3b) and western blot assays (Fig. 3c; Supplementary Fig. 2A) showed that PLK2 expression was significantly increased in both U87 and U251 cells transduced with PLK2 overexpression lentivirus compared with negative control. To explore the effect of PLK2 overexpression on tumor malignancy in GBM cells, in vitro cell proliferation assays as well as colony formation assays were performed. The results indicated that PLK2 overexpression impaired cell growth and self-renewal ability (Fig. 3d-f, Supplementary Fig. 2B-D). Moreover, the apoptotic rate of glioma cells transduced with PLK2 overexpression lentivirus increased significantly compared with its negative control (Fig. 3g, Supplementary Fig. 2E). According to the cell migration assays, the migratory ability of U87 and U251 cell lines were significantly decreased by PLK2 overexpression (Fig. 3h and i, Supplementary Fig. 2F and G). Additionally, the intracranial xenograft mouse model was constructed to investigate the function of PLK2 on tumorigenesis. As a result, most of the mice transplanted with PLK2 overexpressed U87 cell line failed to form tumors (Fig. 3j). Kaplan-Meier analysis showed elevated PLK2 suppressed the in vivo tumor growth thus prolonged the survival of xenograft mice (Fig. 3j, Supplementary Fig. 2H). Altogether, these data showed that PLK2 overexpression reduced the malignancy including proliferation, migration, self-renewal, and tumorigenesis of GBM cells both in vitro and in vivo.

To further investigate the correlation between PLK2 and acquired TMZ resistance, TMZ resistant U87 and U251 cell lines were constructed before cell viability assays. As is shown in Fig. 4a, resistant cell lines showed less vulnerability to TMZ treatment compared with non-treated naïve cell lines. Additionally, down-regulated PLK2 expression was found in both U87 and U251 resistant cell lines (Fig. 4b, c). We then introduced PLK2 overexpression lentivirus into TMZ resistant cell lines and measure the transduction efficiency by immunofluorescence (Supplementary Fig. 3A). qRT-PCR analysis illustrated that the PLK2 mRNA expression was significantly increased in both U87 and U251 transfected with PLK2 overexpression lentivirus compared with the control cells (Fig. 4d). Consistently, the protein levels of PLK2 were also up-regulated by exogenous lentiviral overexpression (Fig. 4e). In vitro cell proliferation assays were then performed to investigate the function of PLK2 overexpression on acquired TMZ resistance. The results showed that enhanced PLK2 expression significantly attenuated the proliferation of resistant U87 and U251 cell lines (Fig. 4f, Supplementary Fig. 3B). Furthermore, the flowcytometry assays were used to investigate the apoptosis of U87 and U251 resistant cell lines when treated with a combination of TMZ and PLK2 overexpression. Notably, PLK2 overexpression significantly elevated the apoptosis rate of resistant glioma cell lines under the treatment of TMZ (Fig. 4g, Supplementary Fig. 3C). Also, the in vivo xenograft mouse model was constructed using TMZ resistant U87 and U251 cell lines, and mice transplanted with PLK2 overexpressed U87 resistant cell lines became more sensitive to TMZ treatment compared with the control groups. Consistently, Kaplan-Meier analysis showed prolonged survival time in PLK2 overexpression mice group when treated with TMZ (Fig. 4h, Supplementary Fig. 3D). Taken together, these findings indicated that loss of PLK2 promotes acquired TMZ resistance of GBM while increased PLK2 expression enhances TMZ-sensitivity of GBM.

As we previously demonstrated that PLK2 functions as a tumor-suppressor gene and attenuated chemoresistance in GBM, it is essential to deeply investigate the underlying signaling pathways regulated by PLK2. To this end, bioinformatics analysis with TCGA dataset and GSE16011 dataset were performed. The transcriptome profiles of GSE16011 and TCGA datasets were respectively grouped into 2 groups according to the expression of PLK2. Hierarchical bi-clustering analysis indicated significant gene signatures in PLK2^high and PLK2^low GBM (Fig. 5a, b). GSEA analysis were then performed to figure out the negatively correlated pathways of PLK2. As is shown in Fig. 5c-f, Notch signaling pathway was the only pathway that was simultaneously enriched in PLK2^low group in both GSE16011 and TCGA databases. Therefore, we choose Notch signaling pathway as a candidate downstream pathway of PLK2.

To further confirm the results of the potential mechanism underlying PLK2 and Notch, qRT-PCR and Western blot assays were then carried out. The results showed that PLK2 overexpression significantly reduced the downstream targets of Notch signaling pathway including HES1, cMYC, p21, Cyclin D3 on transcriptome and translation levels (Fig. 6a, b, Supplementary Fig. 4A, B). Additionally, PLK2 knock-down considerably enhanced these responding targets of Notch signaling pathways (Fig. 6c, Supplementary Fig. 4C). To further investigate the potential mechanism by which PLK2 overexpression suppresses Notch signaling pathway, we overexpressed HES1 in lentiviral PLK2 overexpressed U87 cells (Fig. 6d, Supplementary Fig. 4D). Notably, the suppression of cell growth (Fig. 6e), colony formation (Fig. 6f, g, Supplementary Fig. 4F, G) and migratory ability (Fig. 6h, Supplementary Fig. 4H) by PLK2 overexpression can be partially rescued when combined with HES1 overexpression, indicating that PLK2 reduces the malignancy of GBM cells by inactivation of Notch signaling pathway via transcriptional regulation of HES1.

To further elucidate the effect of PLK2 on Notch signaling activation, Western blot analyses were performed to investigate the expression of activated Notch1 and Notch2 in U87 cell line. The result demonstrated that Notch 1 expression was remarkably reduced after exogenous overexpression of PLK2, however, the expression of Notch2 was not significantly affected (Fig. 6i, Supplementary Fig. 4I). Consistently, shRNA-induced suppression of PLK2 increased Notch1 protein in GBM, indicating that Notch1 was an essential regulator through which PLK2 could inactivate Notch pathway (Fig. 6j, Supplementary Fig. 4 J). To validate this, protein stability assays were performed with U87 cells transduced with or without PLK2 overexpression lentivirus by using MG132, which was used to block the protein proteolysis. The results showed that protein stability of Notch1 was significantly reduced after PLK2 overexpression in U87 cell (Fig. 6k, Supplementary Fig. 4 K). Consistent with these results, ubiquitinated Notch levels were clearly increased in PLK2 overexpressed samples compared with levels in control groups (Fig. 6l and Supplementary Fig. 4 L), indicating that loss of PLK2 stabilized Notch1 through suppressing degradation of Notch1 protein then activated Notch signaling, which finally leaded to acquired TMZ resistance and tumor recurrence.