Effect of p21-activated kinase 1 inhibition on proliferation, migration and invasion of glioma

Background: Abnormally expressed the p21-activated kinases (PAKs) are implicated in the development and treatment of glioma. Previous study has reported that PAK1 is expressed in glioma. However, the role and mechanism of PAK1 in glioma progression remain unclear. Methods: Western blotting was employed to detect the expression of PAK1 in human glioma tissues. CCK-8, EdU and colony formation assay were applied to evaluate the effect of PAK1 inhibition on cell proliferation of glioma. The flow cytometry was utilized to examine the cell cycle distribution and apoptotic rate of glioma. Wound healing and transwell assay were exploited to investigate the effect of PAK1 inhibition on cell migration and invasion of glioma. The orthotopic xenograft glioma model was established to probe the effect of PAK1 silencing on tumor formation of U87 cell. Results: It was showed that PAK1 was significantly upregulated in glioma tissues. Besides, high level of PAK1 was found to be associated with poor prognosis in glioma patients in TCGA database. PAK1 inhibition restrained cell proliferation, arrested cells at G1 phase, and induced cell apoptosis of glioma. PAK1 inhibition suppressed glioma cell migration and invasion. Moreover, knockdown of PAK1 dramatically decreased the protein expressions including MDM2, p38, p-p38, CyclinD1, CDK4, Bcl-2, MMP2, MMP9, Cofilin, while increased the protein levels of p53, Bax, p21 and Cleaved Caspase-3. Finally, orthotopic xenograft glioma model confirmed that silencing of PAK1 repressed the tumor formation of U87 cell transplantation. Conclusion: Our study described that PAK1 inhibition impedes proliferation, migration and invasion of glioma cells and thus probably as a novel target for glioma therapy. examine cell cycle (3i-3k) and apoptotic rate (3l-3n) of U87 and U251 after treatment with PKA1 shRNA, respectively. The percentage of apoptosis cells was exhibited. o-p. Western blotting analysis showed knockdown of PAK1 on expression of classical markers both cell cycle and apoptosis including CyclinD1, Bcl-2, Bax and cleaved caspas-3. Actin regarded as loading control. All experimental data were carried out in triplicate independently. PAK1, p21-activated kinase 1.

4 glioma cells. Most importantly, an intracranial glioma xenograft model in nude mice was stablished to observe the influences of silencing PAK1 on tumor growth in vivo. Consistently, we found that knockdown of PAK1 significantly retarded glioma formation. To sum up, these results suggested that PAK1 expression is elevated in glioma patients, inhibition of PAK1 suppresses cell proliferation, migration and invasion of glioma cells and prevents tumor growth in vivo, indicating that PAK1 serves as a novel biomarker and might be a promising target for malignant glioma therapy.

Glioma tissue specimens
All of the primary human glioma specimens included grade II (12 samples), grade III (12 samples) and grade IV (9 samples) were collected from the Department of Neurosurgery, the Affiliated Hospital of Xuzhou Medical University ranged from September 2016 to September 2019. Besides, no patients underwent antitumor therapies including radiotherapy and chemotherapy before the operation. In addition, 9 cases of normal brain tissue specimens were selected as control. These fresh samples were harvested and immediately frozen in liquid nitrogen. All the glioma samples were histologically  Table 1. This study was written informed consent obtained from all patients and approved by the Medical Ethics and Human Clinical Trial Committee of the Affiliated Hospital of Xuzhou Medical University.

Cell lines and reagents
The glioma cell lines U87, U251 and human embryonic kidney cell line HEK293T were purchased from the Shanghai Cell Bank, Type Culture Collection Committee, Chinese Academy of Sciences. All cell lines were cultured in Dulbecco's Modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (both purchased from Hyclone (Logan, UT, USA)) and supplemented with penicillin The oligonucleotide (100 µM) were annealed and diluted with sterile water 100 times for following use. Subsequently, the annealed oligomers were subcloned into the pLV-shRNA-EGFP(2A)Puro vector by means of both EcoR I and BamH I cloning sites. The recombinant plasmids were confirmed whether the sequence of shRNA was inserted into the plasmid vector by agarose gel electrophoresis and DNA sequence analysis. The lentiviruses were packaged by three-plasmid-based expression system, namely co-transfecting the core plasmid and the packaging plasmids (psPAX2 and pMD2.G plasmid) in 6 HEK293T cells, which was carried out by PolyJet (SignaGen, Gaithersburg, MD, USA) according to the manufacturer's instructions. The lentiviruses were collected and used for glioma cell infection. 48 h after infection, the lentivirus-infected cells were cultured in the medium containing 2.5 μg/mL puromycin selection to establish stable shPAK1 cell lines (Sigma, St. Louis, MO, USA).

Cell viability assay
For the experiment of IPA-3 treatment, 8*10 3 cells/mL U87 and U251 cells were treated with DMSO vehicle and the different concentrations of IPA-3 ranged from 0 to 25 μM for 24 h and then subjected to CCK-8 assay to determine the optimal efficiency concentration of IPA-3 on cell proliferation. We found that U87 and U251 cells displayed more sensitive to 25

EdU incorporation assay
This assay aimed to detect cell proliferation using Cell-Light™ EdU Cell Proliferation Detection Kit (Ruibo Biotech, Guangzhou, China) according to the manufacturer's instructions. For the experiment treated with IPA-3, 8*10 3 cells/mL U87 and U251 cells were placed in 96-well plate per well for 20 h, then treated with DMSO vehicle and IPA-3, respectively. Then treated with 50 μM 5-ethynyl-20deoxyuridine (EdU) in 37 °C for 2 h. Afterwards, the cells were fixed by 4% paraformaldehydrate for 30 min and treated with 0.5% Triton X-100 for 10 min, separately. Thereafter, the cells were incubated with 1×Apollo® reaction solution for 30 min and stained with 1X Hoechst33342 reaction cocktail for 30 min in the dark, respectively. Finally, the cells were washed by phosphate buffered 7 saline (PBS) and observed with Olympus IX-71 inverted microscope (Olympus, Japan). The percentage of EdU-positive cells was calculated by the ratio of EdU-positive cells to total Hoechst-stained cells.
For the experiment treated with PAK1 shRNA, a total of 8 *10 3 cells/mL U87 and U251 cells after infection PAK1 shRNA were seeded into 96-well plate, the following disposals were similar to the above descriptions. All experiments were performed in triplicate independently.

Colony formation assay
For the experiment of IPA-3 treatment, 1*10 3 cells/mL U87 and U251 cells after treatment with or without IPA-3 per well were seeded in 6-well plate, each group cells were repeated in three wells, then placed in 37 °C, 5% CO 2 incubator, after 14 days, the plates were washed by PBS, then fixed with 4% paraformaldehyde for 30 min, finally stained with 0.3% crystal violet diluted in methanol for 15 min, and visualized immediately on an Olympus IX-71 inverted microscope. Colonies containing more than 50 cells were counted. For the experiment of PAK1 shRNA treatment. A total of 1*10 3 cells/mL U87 and U251 cells after infection PAK1 shRNA were seeded in 6-well plate, each group cells were repeated in three wells, the following disposals were conducted as previously described. The data was analyzed by the Image J software (National Institutes of Health, Bethesda, MD). All experiments were performed in triplicate independently.

Cell cycle and cell apoptosis detection
For the cell cycle experiment, this assay aimed to detect cell cycle distribution using kFlour647 Click-iT EdU Flow Detection Kit (KeyGEN BioTECH, Nanjing, China) according to the manufacturer's instructions. Cell fixation and permeabilization using the eBioscience TM Intracellular Fixation &Permeabilization Buffer Set (Invitrogen). 2*10 5 cells/mL U87 and U251 cells which were pre-treated with or without IPA-3 were seeded into 6-well plate for 20 h, once attached to the plate, treated with Finally, the cell suspension was filtered by a 50-mm nylon mesh and immediately analyzed by flow cytometry (BD, Franklin Lakes, NJ, USA). Cell cycle analysis was performed by flowjo 7.6 software. For the experiment of PAK1 shRNA treatment, 2*10 5 cells/mL U87 and U251 cells pre-transfected PAK1 shRNA were seeded into 6-well plate, the following disposals was conducted as the above protocol.
For cell apoptosis assay, 2*10 5 cells/mL U87 and U251 cells pre-treated with or without IPA-3 were performed using PE Annexin V Apoptosis Detection Kit I (BD, Biosciences, USA). The cells were harvested and resuspended in cold PBS, then centrifuged and discarded the supernatants thoroughly.
Subsequently, resuspended cells in 1×Binding Buffer, transfered 100 µL of the solution to a 5 ml culture tube, and added 5 µL phycoerythrin (PE) Annexin V and 5 µL 7-amino-actinomycin-D (7-AAD), respectively. Vortexed the cells and incubated for 15 min at room temperature (RT) in the dark.
Eventually, added 400 µl of 1X Binding Buffer to each tube, the cell suspensions were filtered by a 50mm nylon mesh and immediately analyzed by flow cytometry, wherein 10,000 cells were counted from each sample. For the experiment treated with PAK1 shRNA was conducted as previously described. All experiments were performed in triplicate independently.

Wound healing assay
The experiment aimed to evaluate the migration behavior of glioma cells. U87 and U251 cells pretreated with or without IPA-3 were seeded in 6-well plates. When the cells were at 80% confluence, the scratches were performed along the straight line with a sterile pipette tip, rinsed with PBS three times to remove detached cells completely. Subsequently, changed serum-free medium and incubated in a condition of 37 °C, 5% CO 2 and saturated humidity conditions, at the designated time 0 and 24 h, randomly chose six 100х magnification fields per sample around the lesion border by Olympus IX-71 inverted microscope. Finally, counted the migrating cells across the wound to quantify the relative cell migration rates. For the experiment treated with PAK1 shRNA was conducted as previously described. All experiments were performed in triplicate independently.

Transwell migration and invasion assays
The two experiments were performed using a chemotaxis chambers with polycarbonate nucleopore membrane. The filters with a diameter of 6.5 mm and pore size of 8 μm (Invitrogen). To detect cell invasion ability, firstly, the filters were pre-coated with 50 uL Matrigel (1 mg/mL) (BD, Biosciences, USA), then rehydrated with 100 μL FBS-free medium in 37 °C incubator for 2 h. Subsequently, a density of 6*10 3 cells/mL U87 and U251 cells pre-treated with or without IPA-3 suspended with serumfree medium were plated in the upper layer part of each chamber, 600 μL culture medium containing 10% FBS as a source of chemoattractants was added to the each lower chamber and incubated in 37°C incubator. 36 h later, removed the non-invasive cells of the upper chamber with cotton swabs.
Afterwards, invasive cells on the lower surface of the filter were fixed with 4% paraformaldehyde for 30 min and stained with 0.3% crystal violet (w/v) diluted with methanol for 15 min. Finally, five fields of adherent cells in each well were randomly photographed to quantify the cell invasion rates under an Olympus IX71 inverted microscope (Olympus, Japan). The assessment of cell migration ability was the same conduction except that filters were not pre-coated with Matrigel. For the experiment treated with PAK1 shRNA was conducted as previously described. All experiments were performed in triplicate independently.

Western Blotting analysis
The cells were lysed with RIPA lysis buffer (50 mM Tris/HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.25% Na-deoxycholate) supplemented with protease and phosphatase inhibitor cocktail, and lysates were incubated on ice for 30 min to yield a homogenate, subsequently,

Intracranial mouse model
All mice experiments were conducted in accordance with protocols approved by the Xuzhou Medical University Animal Care and Use Committee and followed guidelines for animal welfare. For the purpose of establishing intracranial GBM xenografts, 6 to 8-week-old female BALB/c-nude mice were purchased from Beijing Weitong Lihua Experimental Animal Technology Company and housed in Individually Ventilated Caging (IVC) systems in a 12 h light/dark cycle. A total of 3*10 5 U87 cells/mL expressing mCherry routinely treated by mycoplasma per well were seeded in 6-well plate. When the cell confluence reached 50%, the cells were infected with PAK1 shRNAs and control lentivirus, respectively. After 48 h of infection, the cells were trypsinized and resuspended with sterile PBS. A total of 5*10 5 shPAK1 and control cells (in 5 μL PBS) were collected and injected into the intracranial striatum of nude mice with a stereotactic instrument and a microsyringe. Before injection, mice were anesthetized with 3% chloral hydrate, subsequently, a hole about 5 mm in diameter was burred located in 1 cm anterior and 2 cm lateral to the bregma, then with the stereotactic guidance, the microsyringe needle penetrated 3 cm from the calvaria to the striatum and injected cells at a rate of 1 μL/min [13][14][15][16]. The animals were randomly divided into 3 groups, with 8 mice in each group, including shControl, shPAK1-2, shPAK1-3, respectively. Starting on day 28 d after tumor cell implantation, sacrificed 2 mice by cervical dislocation from each group every week, and the brain samples were fixed in 4% paraformaldehyde. Afterwards, performed frozen slices for hematoxylin and eosin (HE) staining to obtain tumor growth status from different groups.

Statistical analysis
All experiments were performed in triplicate independently. Data were represented as the mean ± SD as indicated. The Student's t-test (two-tailed) was used for comparison of two groups, comparisons among multiple groups were analyzed with One-way analysis of variance (ANOVA). P<0.05 was considered statistically significant. All statistical analyses were performed using the SPSS 16.0 software (SPSS Inc., Chicago, IL, USA).

PAK1 is frequently overexpressed and predicts poor prognosis in glioma patients
Accumulating evidence has indicated that PAK1 aberrantly amplificates and is also involved in tumorigenesis in breast cancer, ovarian cancer and pancreatic cancer and so on [17][18][19]. Nevertheless, the study on whether and how PAK1 contributes to carcinogenesis of glioma are still poorly known. In the present study, to investigate the expression of PAK1 in glioma samples, we first examined the differential protein levels in glioma tissues. 33 primary human glioma specimens (12 were grade II, 12 were grade III, and 9 were grade IV) and 9 nontumoral brain tissues from human as control were detected by western blotting. We found that the expressions of PAK1 in gliomas were higher than those in the control. Furthermore, the protein expression of PAK1 was found to be markedly upregulated in high-grade gliomas (HGG) compared to low-grade gliomas (LGG) patients ( Fig. 1a-b).
Additionally, to understand the clinical significance of PAK1, an independent cohort of 153 human glioma specimens from The Cancer Genome Atlas (TCGA) database (https://cancergenome.nih.gov/) was used to validate the correlations between PAK1 expressions and survival time of glioma patients.
The outcomes provided a provocative evidence that PAK1 expression was positively correlated with glioma grade, and the higher protein level of PAK1, the worse prognosis of glioma patients (Fig. 1c).
Taken together, these results provided us the preliminary evidence that PAK1 might be involved in glioma carcinogenesis and progression.

Inhibition of PAK1 suppresses glioma cell proliferation
Given the high expression of PAK1 in glioma, and PAK1 probably plays an important role in the initiation and progression of glioma. Thus, to explore the role of PAK1 in glioma, we used IPA-3, a potent ATP-competitive small molecule inhibitor of PAK1, to block the function of PAK1. Firstly, glioma cells U87 and U251 were treated with IPA-3, in a dose and time-dependent manner and subjected to CCK-8 assay. The IPA-3 concentrations ranged from 0 to 25 μM for 0, 12, 24, 36, 48 and 60 h. DMSO treated cells were regarded as vehicle control. We found that treatment with IPA-3 retarded the growth of glioma cells both in a dose-dependent (Fig. 2a-2b) and time-dependent manners compared with vehicle control (Fig. 2c-2d), additionally, the efficacy treated with 25 μM for 24 h appeared markedly (P <0.05 for both U87 and U251). Similarly, inhibiting PAK1 through treatment of IPA-3 decreased the ratio of EdU-positive cells in U87 (P<0.002) and U251 (P< 0.004) (Fig. 2e-2g). Likewise, colony formation assay exposed to IPA-3 further verified the previous conclusion (P<0.01 for both U87 and U251, Fig. 2h-2j). In addition, we employed PAK1 shRNA to further confirm the prior results. Three shRNA targets including shPAK1-1, shPAK1-2 and shPAK1-3 were subcloned into a lentiviral vector pLV-shRNA-EGFP(2A) Puro and used to produce lentivirus for glioma cell infection. The cell lines were identified by green fluorescent protein (GFP) images and western blotting. As shown in Suppl. Fig. 1, when 48 h of lentivirus infection, more than 80% of lentivirus-infected U87 and U251 cells exerted GFP fluorescence. The protein level of PAK1 was further confirmed a significant reduction in the shPAK1-2 cells by western blotting (Fig. 2k-2l). Therefore, we chose shPAK1-2 for the subsequent further investigation. In addition, shControl was used regarded as control and shPAK1-3 as well.
Furthermore, similar results were also observed in colony formation assay. Compared with the shPAK1-3 and shControl cells, the number of colonies formed in shPAK1-2 cells was significantly reduced (P < 0.001 for both U87 and U251, Fig. 2s-2u).
Recently, Fang et al. reported that PAK1 promoted acute pancreatitis via p38 and NF-κB pathways [20]. Moreover, Qing et al. found that PAK1-dependent mitogen activated protein kinase (MAPK) pathway activation was required for colorectal cancer cell proliferation [21]. In addition, PAK1 was also involved in regulating expression of relative genes in the DNA damage response pathway [22]. Therefore, the proteins including tumor suppressor proteins p53 and p21 (Cyclin Dependent Kinase Inhibitor-p21) and their downstream target proteins mouse double minute-2 (MDM2) and cyclin-dependent kinase 4 (CDK4) were detected. Giving PAK1 can integrate multiple signal pathways and be involved in the proliferation. Therefore, western blotting was performed to further investigate the mechanisms of how PAK1 affected the proliferation of glioma cell. After infection with shPAK1 for 72 h, U87 and U251 cells samples were collected and lysed for western blotting. As shown in Fig. 2v-2w, we observed the marked downregulation in p38, p-p38, as well as MDM2 and CDK4 in shPAK1-2 relative to shPAK1-3 and shControl. However, p53 and p21 protein levels were dramatically elevated. In short, treated with inhibitor IPA-3 or shRNA-mediated PAK1 gene silencing suppresses cell proliferation both in U87 and U251 cells.

Inhibition of PAK1 arrests glioma cell cycle progression and facilitates cell apoptosis
Considering that cell cycle and apoptosis play essential parts in regulating cell growth, and may contribute to proliferation function. Thus, to elucidate whether the PAK1 inhibition-mediated growth restraint was derived from cell cycle arrest and/or apoptosis, firstly, we analyzed the number and distribution of cells in the phases of the cell cycle by flow cytometry. U87 and U251 cells were treated with or without IPA-3 for 24 h and then subjected to cell cycle assay. The outcome showed that IPA-3 treatment contributed to a rise in the fraction of cells in G1 phase (P= 0.0042 and P=0.036 for U87 and U251, respectively), however, the number of cells in S phase was less than the group treated with vehicle control (P=0.0037 and P=0.025 for U87 and U251, respectively, Fig. 3a-3d). Subsequently, to determine the apoptotic rate of U87 and U251 after treatment with IPA-3, flow cytometry assay was conducted. The data showed that IPA-3 treatment increased the proportions both Annexin V and 7-AAD-positive cell (P=0.0046 and P=0.0052 for U87 and U251, respectively, Fig. 3e-3h). After that, the experiment treated with PAK1 shRNA was also conducted. U87 and U251 cells were infected with PAK1 shRNA and control lentivirus, after infection for 48 h, flow cytometry displayed that U87 and U251 cells infected with shPAK1-2 decreased the percentage of cells in S phase (P=0.026 and P=0.035 for U87 and U251, respectively) whereas arrested the cells in G1 phase (P=0.005 and P=0.016 for U87 and U251, respectively) compared with shPAK1-3 and shControl (Fig. 3i-3k).
Similarly, U87 and U251 infected with PAK1 shRNA and subjected to apoptosis assay. As shown in Fig.   3l-3n, U87 and U251 cells of shPAK1-2 enhanced Annexin V percentage and population in 7-AADpositive cells in comparison to shPAK1-3 and shControl (P=0.005 and P=0.006 for U87 and U251, respectively). In brief, these findings indicated that inhibition of PAK1 blocks cycle progression and promots apoptosis of glioma cells. Previous studies have indicated that PAK1 participated in Wnt/βcatenin signaling pathway, leading to upregulation of Wnt target genes such as cyclin D1 [23]. The activated PAK1 in cancer cells brought about a defective mitosis and multipolar spindle orientations.
Thus, to explore the underlying mechanism of cell cycle arrest and apoptosis in glioma cells, the expressions of classical markers related to these processes both cell cycle and apoptosis including cyclinD1, B-cell leukemia/lymphoma 2 (Bcl-2), cleaved caspas-3 and Bax (Bcl2 Associated-X Protein) were estimated in response to PAK1 shRNA treatment by western blotting. As shown in Fig. 3o-3p, as expected, the expressions of cyclinD1 and anti-apoptotic protein Bcl-2 in shPAK1-2 were significantly lower than shPAK1-3 and shControl, in contrast, pro-apoptotic protein cleaved caspas-3, as well as Bax in shPAK1-2 were significantly higher.
Together, these findings above suggested that suppressing PAK1 resultes in S arrest and enhances apoptotic rate of glioma cells through regulating multiple related proteins both cell cycle and apoptosis.

Inhibition of PAK1 suppresses glioma cell migration and invasion
Due to the high infiltration behaviors of glioma mainly composed of tumor cell migration and invasion.
Thus, we next investigate whether inhibition of PAK1 exerts an effect on migration and invasion of glioma cell. Wound healing assay exhibited that the number of cells migrated to the central area at 24 h was obviously decreased in the group treated with IPA-3 compared to the group of vehicle control (P=0.0046 and P=0.0057 for U87 and U251, respectively, Fig. 4a-4d). Moreover, transwell migration assay without matrigel further showed that IPA-3 treament significantly degraded the number of cells migrated to the chamber in U87 (P<0.001) and U251 (P=0.0005) cells (Fig. 4e-4g).
Thereafter, the transwell assay containing matrigel was also performed to observe invasion behaviors of U87 and U251 cells treated by IPA-3. The results were consistent with the migration experiment, namely, inhibiting PAK1 notably decreased the number of cells that crossed the matrigel in U87 (P=0.00061) and U251 (P=0.00075) cells (Fig. 4h-4j).
Furthermore, U87 and U251 cells after infection with PAK1 shRNA were also performed by wound healing and transwell assays. Compared with the shPAK1-3 and shControl groups, the number of both migration and invasion in shPAK1-2 group was significantly reduced (Fig. 4k-4r), suggesting that silencing PAK1 markedly repressed glioma cell migration and invasion.
As well known, the epithelial to mesenchymal transition (EMT) exerts an important role in tumor progression. Which has dispalyed in many metastatic cancers such as breast cancer, hepatocellular carcinoma, malignant glioma, and so forth [24]. Indicating that tumor cells probably utilize the process of EMT to obtain their metastatic dissemination. Considering that metalloproteinase 2 (MMP2) and metalloproteinase 9 (MMP9) are crucial enzymes for the degradation of the extracellular matrix and may contribute to tumor cell migration and invasion. Therefore, we evaluated the protein levels of MMP2 and MMP9 after infection PAK1 shRNA by western blotting. Correspondingly, one of the representative biomarkers of cell invasion and migration Cofilin was also evaluated. As we expected, silencing PAK1 notably decreased the MMP2 and MMP9 protein levels in U87 and U251 cells. Likewise, Cofilin was well (Fig. 4s-4t). These results suggested that PAK1 is involved in the migration and invasion of glioma cells and knockdown of PAK1 inhibits the migration and invasion of glioma cells.

Knockdown of PAK1 inhibits glioma formation in vivo
To further clarify whether down-regulation PAK1 also inhibits tumor growth in vivo, we evaluated the effect in situ by establishing xenograft tumor model in nude mice and analyzed the size of gliomabearing mice. The mice were treated with intracranial injection. When the xenograft model was established, starting on 28 d, sacrificed 2 mice from each group every week, then the brain samples were subjected to frozen slices for HE staining to compare tumor growth status of the same section among different groups.
As shown in Fig. 5a and Table 2, no mice loaded tumor in shPAK1-2 group, this was significantly less than those of shPAK1-3 (3 mice) and the shControl (4 mice). Furthermore, the size of xenograft tumor in shPAK1-3 group was obvisouly smaller than that of shControl group under the macroscopic observation. To sum up, the result demonstrated that knockdown of PAK1 tremendously suppresses tumor growth in mice bearing xenograft glioma.

Discussion
Malignant glioma is the most common type of primary intracranial tumors in adults, and accounts for about 70 percent of human brain tumors [25]. Specially, characterized by its high heterogeneous, widespread consistent infiltration, quick malignant progression. Despite effective treatments including surgical removal, chemotherapy and radiotherapy delay the development and progression to some extent [26][27][28]. Its current clinical prognosis remains badly poor. Therefore, it is urgent and necessary to explore a novel effective therapeutic target.
Accumulating evidence has indicated that PAK1 performs a crucial part in cytoskeletal dynamics, cell survival, proliferation and EMT of various cancer cells, such as gastric cancer and non-small cell lung cancer [29,30]. However, its functions in glioma remain unclear. In the present study, we first demonstrated that the protein level of PAK1 was significantly upregulated in various grades of glioma tissues. High expression of PAK1 was closely associated with poor prognosis. Our findings are consistent with the previous report that phosphorylated PAK1 level in the cytoplasm correlates with shorter survival time in patients with glioblastoma [17]. We also proved that inhibition of PAK1 by IPA-3, small molecular inhibitor of PAK1 and PAK1 shRNA suppressed the proliferation, cycle progression, migration, invasion as well as induced apoptosis of glioma cells in vitro. Furthermore, the xenograft inoculated shRNA-mediated PAK1 silencing suppressed the growth of U87 cell in vivo. Totally, these preliminary findings manifested that high levels of PAK1 in glioma tissues may accelerate tumor progression. Therefore, PAK1 may be a potential prognostic biomarker for glioma therapy.
As the previous study, PAK1 is believed to be involved in several survival signaling pathways, for instance, activated PAK1 can stimulate PI3K/AKT and NF-κB (nuclear factor κB) pathway, which are important in tumor progression [31,32]. Therefore, in this research, a series of associated protein levels were also detected. We found that knockdown of PAK1 decreased p38 and p-p38 expressions.
As well known, p38 is the major subfamily of MAPK cascades and has been widely markedly aberrantly amplified in glioma and involved in the regulation of cell cycle progression and proliferation [33]. It was reported that PAK1-dependent MAPK pathway activation was required for colorectal cancer cell proliferation [34]. In short, these results suggested that besides regulating PI3K/AKT and NF-κB pathway, knockdown of PAK1 could also regulate proliferation by inactivating MAPK pathway. Previous study demonstrated that PAK1 gene silencing might decrease proliferation and increase apoptosis of hepatocellular carcinoma cells by regulating the p53/p21 pathway [35]. In this study, we showed that knockdown of PAK1 increased the expressions of p53 and p21, and decreased the expressions of CyclinD1, MDM2 and CDK4. Guardian of the genome p53 is a tumour suppressor that regulates the expression of multifarious genes involved in apoptosis, growth arrest, inhibition of cell cycle progression and accelerated DNA repair in response to genotoxic or cellular stress [36]. Nonphosphorylated p53 is ubiquitylated degradation by the MDM2, which acts as a ubiquitin ligase and major regulator of p53. p53 is involved in cell-cycle inhibition by stimulating the expression of p21 [37]. Additionally, recent studies revealed that PAK1 overexpression promotes cell proliferation in cutaneous T cell lymphoma via suppression of p21 [38]. p21 is an inhibitor of CDKs (Cyclin-Dependent Kinases) that regulates the cell cycle via perturbation of its partner cyclins. cyclins associate with CDKs to regulate their activity and the progression of the cell cycle through specific checkpoints. Disruption of Cyclin action leads to either cell cycle arrest, or to uncontrolled cell cycle proliferation [39]. p53 stimulates two major apoptotic pathways: extrinsic pathways and intrinsic pathways, resulting in a cascade of activation of caspases and high expression of the Bcl-2 family proteins, respectively, which in turn induce apoptosis [40]. Our findings demonstrated that knockdown of PAK1 increased the expressions of pro-apoptotic proteins Cleaved caspase-3 and Bax, while decreased the expression of the anti-apoptotic protein Bcl-2.
An increasing number of studies have shown that the EMT is considered as a crucial mechanism which tumor cells utilize this process to acquire migration and invasion properties in metastastic cancers [24]. It is reported that Cofilin induces the epithelial-mesenchymal transition of gastric cancer cells by promoting cytoskeletal rearrangement. Cofilin is an actin-binding protein by promoting cell migration and invasion, facilitating tumor metastasis [41]. Besides, the recent study also find that CXCL10 accelerates EMT and metastasis by MMP-2 in hepatocellular carcinoma [42]. Matrix metalloproteinases (MMPs) have been regarded as proteolytic enzymes which play an important role in tumor invasion and migration through degrading basement membrane and extracellular matrix (ECM) components [43]. In the study, we found that knockdown of PAK1 decreased the expressions of Cofilin, MMP2 and MMP9. Thus, we speculate that silencing PAK1 inhibited EMT programs may be via targeting decreasing MMP2, MMP9 and Cofilin.

Conclusion
In conclusion, our study elucidated that PAK1 significantly overexpressed in various grades of glioma tissues. High expression of PAK1 was closely associated with poor prognosis. Furthermore, PAK1 inhibition restrained the proliferation, cycle progression, migration, invasion as well as induced apoptosis of glioma cells. However, the whole underlying mechanism is still unclear. Therefore, the potential mechanism of how PAK1 promotes glioma development needs to further explore. Targeting PAK1 may be a novel therapeutic strategy for malignant glioma.

Supplementary information
Additional file 1: Figure S1. Cell morphology, green fluorescent with three shRNAs after infection for 48 h. BF, bright field. GFP, fluorescence. Scale bar, 100 μm.   Figure 1 PAK1 is frequently overexpressed and predicts poor prognosis in glioma patients a-b.

Figures
Western blotting analysis and quantitation of PAK1 expressions in 33 primary human glioma tissues (12 were grade II, 12 were grade III, and 9 were grade IV) and 9 normal brain Inhibition of PAK1 arrests glioma cell cycle progression and facilitates cell apoptosis a-h. Figure 4 32 Inhibition of PAK1 suppresses glioma cell migration and invasion a-j. The wound healing and transwell assays were performed to evaluate the U87 and U251 cells migration (Fig. 4a-4g) and invasion (Fig. 4h-4j) after treatment by inhibitor IPA-3 (25 μM). IPA-3, p21-activated kinase inhibitor III. k-r. The wound healing and transwell assays aimed to evaluate the U87 and U251 cells migration (Fig. 4k-4p) and invasion (Fig. 4q-4r)   Knockdown of PAK1 inhibits glioma formation in vivo a. Representative HE staining images of mice brain bearing tumor. Scale bar, 100 μm. T, tumor.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.