GBP2 facilitates the progression of glioma via regulation of KIF22/EGFR signaling

Identifying the mechanism of glioma progression is critical for diagnosis and treatment. Although studies have shown that guanylate-binding protein 2(GBP2) has critical roles in various cancers, its function in glioma is unclear. In this work, we demonstrate that GBP2 has high expression levels in glioma tissues. In glioma cells, depletion of GBP2 impairs proliferation and migration, whereas overexpression of GBP2 enhances proliferation and migration. Regarding the mechanism, we clarify that epidermal growth factor receptor (EGFR) signaling is regulated by GBP2, and also demonstrate that GBP2 interacts directly with kinesin family member 22(KIF22) and regulates glioma progression through KIF22/EGFR signaling in vitro and in vivo. Therefore, our study provides new insight into glioma progression and paves the way for advances in glioma treatment.


INTRODUCTION
Glioma is among the most frequently occurring types of adult primary brain cancer, with a very high mortality rate owing to its localization and strong invasion capacity [1][2][3]. Despite treatment with surgical resection, chemotherapy, and radiation therapy, the median survival time of glioma patients is only 14-16 months [4], death from glioma accounts for the major death of primary brain cancers [1].
At least four subtypes of human gliomas have been described: pro-neural, neural, classical, and mesenchymal subtypes [1,5]. Recently, substantial progress has been made regarding our understanding of the molecular pathogenesis of gliomas. Distinct proteins have been identified in different glioma subtypes, including YKL40 [5,6], fibronectin 1 (FN1) [7], EGFR [5][6][7], and Stat3 [8]. These advances have been used both to improve diagnostics and to generate more effective therapeutic strategies, such as targeted therapies, for gliomas. However, it is essential to identify even more potential target pathways.
GBPs (guanylate-binding proteins), which were first found to be induced by IFN-γ in human fibroblasts, belong to the superfamily of interferon-inducible GTPases [9]. GBP2, a member of the GTPase superfamily, has critical roles in carcinoma. Studies have demonstrated that GBP2 suppresses cell proliferation in colorectal cancer via WNT signaling [10]. In breast cancer, GBP2 inhibits mitochondrial fission and cell invasion [11], and GBP2 overexpression is correlated with a better prognosis [12]. However, high expression of GBP2 is an indicator of esophageal squamous cell carcinoma (ESCC) [13]. It was also reported that GBP2 promotes invasion of GBM via the Stat3/FN1 cascade [14]. GBP2 has also been shown to be a prognostic biomarker of pancreatic adenocarcinoma [15]. Thus, GBP2 also has different functions in different tissues.
KIF22 belongs to the kinesin-10 family and plays diverse parts in cytoskeleton dynamics and during synaptic development [16][17][18][19]. Recently, studies have provided insight into the role of KIF22 during cancer progression. For example, KIF22 regulates EGFR signaling and attenuates the internalization of lung cancer cells [17]. In prostate cancer patients, high KIF22 expression is correlated with tumor development and poor prognosis [20]. Studies of KIF22 functions in breast cancer, gastric cancer, and tongue SCC have also been reported [21][22][23].
This work showed that GBP2 promoted proliferation and migration in glioma cells by regulating EGFR signaling pathway through interactions with KIF22. In vivo, depletion of GBP2 reduced tumor sizes. In summary, we clarified the oncogenic role of GBP2 in glioma and elucidated for the first time the GBP2/ KIF22/EGFR pathway in glioma progression.

GBP2 was highly expressed in glioma
The expression of GBP2 was detected in GBM and LGG samples; the results showed that mRNA levels of GBP2 in GBM and LGG were higher than those in normal controls (three times higher in GBM and double in LGG, Fig. 1A). With respect to overall survival and disease-free survival, lower GBP2 expression was associated with a higher survival percentage in LGG combined with GBM samples (Fig. 1B, C).
Knockdown of GBP2 impaired proliferation and migration of glioma cells Western blotting confirmed the high depletion efficiency of GBP2 in both glioma cell types ( Fig. 2A). Cell proliferation was analyzed by CCK8 assay. With GBP2 depletion, cell viability was significantly reduced in both cells over 4 days (Fig. 2B). In the siGBP2 group, colony numbers decreased by at least half compared with those of the control group (Fig. 2C). We also carried out flow cytometry assays. The results showed that the percentage of apoptosis cells was nearly doubled by GBP2 depletion (Fig. 2D). Knockdown of GBP2 increased the proportion of G 0 /G 1 -phase cells in the cell cycle (Fig. 2E). Depletion of GBP2 apparently impaired cell migration by at least 50% (Fig. 2F). All the above results confirmed the oncogenic role of GBP2.
Overexpression of GBP2 enhanced proliferation and migration of glioma cells The RT-qPCR and western blotting results showed that GBP2 was highly expressed in U87 cells (Fig. 3A, B). The cell viability assay showed that overexpression of GBP2 enhanced cell viability (Fig.  3C). Colony formation improved by a factor of at least two in GBP2-overexpressing cells (Fig. 3D). Flow cytometry analysis demonstrated that GBP2 overexpression accelerated the cell cycle (the proportions of S-and G 2 /M-phase cells increased, while those of G 0 /G 1 -phase cells decreased) (Fig. 3E) and attenuated cell  Depletion of GBP2 impairs proliferation and migration in glioma cells. A GBP2 was depleted efficiently. Western blotting results for GBP2 knockdown in U87 and U251 cells. B CCK8 assay results for cell proliferation in U87 and U251 cells transfected with siCtrl, siGBP2#1, or siGBP2#2. C Colony formation assay results of U87 and U251 cells transfected with siCtrl, siGBP2#1, or siGBP2#2. Clusters of more than 50 cells were identified as colonies. D Apoptosis results for U87 and U251 cells transfected with siCtrl, siGBP2#1, or siGBP2#2 using annexin V and PI. E Cell cycle results for U87 and U251 cells transfected with siCtrl, siGBP2#1, or siGBP2#2 using PI staining. F Transwell assay results for U87 and U251 cells transfected with siCtrl, siGBP2#1, or siGBP2#2. Each experiment repeated as least 3 times. Magnification: ×10. *P < 0.05, ** P < 0.01, *** P < 0.001. apoptosis (by nearly 80%) (Fig. 3F). The numbers of migration cells nearly doubled in GBP2-overexpressing U87 cells (Fig. 3G). Therefore, the results of the overexpression experiments were highly consistent with those of the knockdown experiments.

GBP2 regulated the EGFR signaling pathway
To further clarify the mechanism, gene microarray was performed to search for differentially expressed genes after GBP2 depletion. The results showed that 243 mRNAs were upregulated, while 560 mRNAs were downregulated (Fig. 4A). To identify the role of GBP2 in glioma progression, IPA analysis was performed based on the microarray data. Canonical_Pathway analysis showed that the at least 10 pathways were significantly enriched, including Signaling by Rho Family GTPases pathway and RhoGDI Signaling et al (Fig. 4B). While the network analysis indicated that EGFR as the target of GBP2. EGFR and its related genes were upregulated after GBP2 knockdown, which included APBB2, C16orf58, CAMLG, CANT1, CEND1, CSRP1, DOCK10, DYRK3, EML6, FAM69A, FBLIM1, FOXN3, GALNT10, GSDME, LMBR1, MEGF6, MTX3, NCEH1, NUCKS1, and PCDHGA3 (Fig. 4C). To further validate the results, we checked the levels of key proteins in the EGFR pathway from two Glioma cell lines through western blotting. Although both the p-EGFR and EGFR were downregulated upon GBP2 depletion, the ratio of pEGFR/EGFR was also reduced. Consistent results were observed in GBP2 overexpression experiments ( Fig. 4D and Supplementary Fig. 1). Therefore, GBP2 regulates the EGFR signaling pathway.

GBP2 interacted with KIF22 and regulated KIF22 expression
In order to explore the molecular mechanism of GBP2 regulating EGFR activity, we employed mass spectrometry to search for proteins directly interacting with GBP2. We found that 9 proteins were particularly interacted with GBP2 (Fig. 5A). Among the 9 proteins, we found that KIF22, DDX31 and YTHDF2 could regulated the expression and activity of EGFR [17,24,25]. In turn, we selected them for further investigation. As the co-IP results indicated, GBP2 could interact with KIF22 in glioma cells, but not DDX31 and YTHDF2 ( Fig. 5B and Supplementary Fig. 2). Previously study showed that KIF22 delayed EGFR internalization and enhanced EGFR signaling. We then analyzed the expression of KIF22 in GBM and LGG, and found identical results to those obtained for GBP2 (Fig. 5C). In the overall survival analysis, lower KIF22 expression was associated with a higher survival percentage (Fig. 5D). In both cell types, knockdown of GBP2 significantly downregulated the expression of KIF22. Consistently, GBP2 overexpression upregulated KIF22 protein levels ( Fig. 5E).

Knockdown of KIF22 suppressed proliferation and migration of glioma cells
Several studies have identified KIF22 as an oncogene [17,[20][21][22][23]. Here, we explored the function of KIF22 in glioma cells. The knockdown efficiency of siKIF22 was confirmed by western blotting. KIF22 is required for the EGFR signaling pathway. Levels of proteins related to the EGFR signaling pathway were significantly decreased after KIF22 depletion in the U87 and U251 cell lines, while GBP2 expression not altered after KIF22 knockdown (Fig. 6A), and cell viability was reduced by at least half in both cell types (Fig. 6B). Colony formation numbers decreased significantly in KIF22-knockdown cells (Fig. 6C, D). The cell cycle analysis showed that knockdown of KIF22 blocked glioma cells at G0/G1 phase (Fig. 6E). The percentage of apoptotic cells nearly doubled following KIF22 depletion (Fig. 6F). Invasion ability was also suppressed, decreasing by at least 60% compared with that of control cells (Fig. 6G). All the above results were consistent with those obtained for GBP2 and indicated a role for KIF22 as an oncogene in glioma cells.

Fig. 3
Overexpression of GBP2 enhances proliferation and migration in glioma cells. A RT-qPCR results confirming overexpression of GBP2 at the mRNA level. B Western blotting results confirming overexpression of GBP2 at the protein level. C CCK8 analysis of cell proliferation in GFP-and GFP-GBP2-overexpressing U87 cells. D Colony formation analysis in GFP-and GFP-GBP2-overexpressing U87 cells. E Cell cycle analysis of U87 cells overexpressing GFP or GFP-GBP2 using PI staining. F Apoptosis analysis of GFP-and GFP-GBP2-overexpressing U87 cells using annexin V and PI. G Transwell invasion results of U87 cells overexpressing GFP or GFP-GBP2. Each experiment repeated as least 3 times. Magnification: ×10. * P < 0.05. ** P < 0.01, *** P < 0.001.

GBP2 regulated EGFR signaling and glioma progression dependent on KIF22
The western blotting results demonstrated that the downregulation of EGFR and p-EGFR caused by GBP2 depletion was rescued by KIF22 overexpression (Fig. 7A). Cell viability and colony formation numbers suppressed by GBP2 knockdown returned nearly to normal levels when overexpression of KIF22 was introduced ( Fig. 7B-D). Impaired migration ability was also rescued by KIF22 overexpression in both glioma cell lines (Fig.  7E). We also measured the in vivo tumorigenicity. Consistent with the above results, whereas tumor size was reduced by GBP2 knockdown, it was restored to control levels by KIF22 overexpression (Fig. 7F). Thus, GBP2 regulated glioma progression dependent on KIF22.

DISCUSSION
To the best of our knowledge, this study was the first time to elucidate the GBP2/KIF22/EGFR pathway in the progression of glioma carcinoma. We employed knockdown and overexpression strategies, and the results of in vitro and in vivo assays were highly consistent in demonstrating the function of GBP2 and KIF22 as oncogenes in glioma.
Despite great progress over the past decades in determining the molecular pathogenesis of gliomas [5][6][7][8], the mortality rate of glioma patients has not been significantly reduced. To improve the prognosis of glioma patients, early diagnosis, reliable prognostic biomarkers, and targeted therapy strategies are needed. Potential targets including YKL40, FN1, EGFR, and Stat3 have been discovered [5][6][7][8]. However, more targets are needed. In this work, we identified GBP2/KIF22/EGFR as a potential therapeutic target for glioma. Since several other KIF superfamily members have been reported to be involved in the development of glioma with low and high grades [16,26], and KIF proteins targeting drugs were in clinical trials in other pathologies and could give novel therapeutic possibilities in the future. Nevertheless, the detailed mechanisms, as well as the potential biomarker need to be further investigated. Thus, our study will pave the way for further advances in glioma diagnosis and treatment.
GBP2, a member of the GTPase superfamily, has functions in protective immunity [27]. Recently, studies have focused on the functions of GBPs in cancer progression. GBP2 is highly expressed in esophageal SCC, where it represents a potential biomarker [13]. Yu et al. demonstrated that GBP2 was highly expressed in human glioma and promoted the invasion of glioma cells [14], which was consistent with our study. Besides, we also indicated that GBP2 played a critical role in glioma proliferation. Otherwise, Yu et al. also suggested that Stat3/FN1 signaling pathway was a key point in the glioma invasion regulated by GBP2. Silencing FN1 or blocking Stat3 activity by WP1066 could reversed the invasiveness induced by GBP2. In our study, microarray assay and subsequent experiment indicated that GBP2 knockdown could significantly depress the expression and the phosphorylation of EGFR. In glioblastoma, classical subtype is characterized with EGFR amplification. EGFR plays critical roles in many cellular processes, such as proliferation, migration, adhesion [28], as well as tumor growth [29]. Further work will include determining relationship between these two pathways, and whether they exist in the same GBM stage or at different stages.
To explain the role of EGFR activity regulated by GBP2, the mass spectrometry was performed. The result showed that KIF22 was more abundant in the GBP2-Flag group compared than IgG group. The following CO-IP assay and Western blot suggested that the GBP2 interacted with KIF22 and regulated the expression of KIF22. In addition, our results indicated that the protein level of KIF22 was decreased after GBP2 knockdown, while no significant changes were found in the mRNA level, as indicated by RNAseq, which might be owing to that GBP2 regulated KIF22 expression at post-transcriptional level. KIF22 has been reported essentially for kinds of cancer progression. Importantly, Pike R et al. reported that high expression of KIF22 maintained the location of EGFR at cytomembrane, leading the continuous activation of EGFR [17]. Similarly, our study indicated that silencing the KIF22 significantly suppressed the progression of glioma progression, and also impaired the protein expression and phosphorylation of EGFR. These results suggested that KIF22 is essentially for EGFR activation in glioma cells.
Furthermore, to confirm whether GBP2 promoted the glioma progression by KIF22, we overexpressed KIF22 in GBP2 knockdown cells. As expected, KIF22 overexpression markedly reversed the inhibited malignant phenotype induced by GBP2 silencing in vitro and in vivo. These results confirmed that GBP2 regulated glioma progression by KIF22/EGFR signaling pathway. However, how does GBP2 regulate KIF22 protein level under the condition of binding to KIF22 needs further studies.
In summary, we demonstrate that GBP2 was overexpressed in glioma tissues. Depletion of GBP2 impaired the proliferation and migration, whereas overexpression of GBP2 showed the opposite effect. Regarding the mechanism, we clarify that EGFR signaling was regulated by GBP2, and also demonstrated that GBP2 directly interacted with KIF22 and regulated glioma progression through KIF22/EGFR signaling. Therefore, our study provides new insight into glioma progression and paves the way for advances in glioma treatment.

MATERIALS AND METHODS Clinical samples form TCGA
The expression of GBP2 and KIF22 were analyzed in GBM or LGG (TCGA data) and normal tissues (Match TCGA normal and GTEx data) in the website (http://gepia.cancer-pku.cn). Moreover, the overall survival and disease-free survival was also analyzed in the website (http://gepia.cancerpku.cn).

Cell viability assay
Cell viability was evaluated using a Cell Counting Kit-8 (CCK-8) purchased from Sigma. Cells were digested with trypsin and seeded in triplicate into 96-well plates at a density of 2000 cells with 100 µL culture medium per well. After incubation for the appropriate number of days, 10 µL CCK8 reagent was added to cells with continuous incubation at 37°C for 2 h. The absorbance of each well was measured at 450 nm. Each treatment groups included at least five wells.

Colony formation
U87 or U251 cells (1 × 10 3 cells per well) were seeded into 6-well plates. The medium was changed every 3 days. Colonies were formed after 14 days. Cells were washed with phosphate-buffered saline (PBS) three times, fixed with 4% paraformaldehyde for 20 min, and stained with GIEMSA staining solution for 20 min. More than 50 cells were considered to constitute a colony. The colonies were counted under a optical microscope (AxioObserverZ1) to capture digital images from four random microscope fields. Then, the colonies were counted using ImageJ software (version1.47, USA), each treatment groups included at least triplicate wells.

Cell cycle analysis
Following the manufacturer's manual, the cell cycle was analyzed using FxCycle PI/RNase Staining Solution (Invitrogen). First, cells grown to about 80% confluence were digested with trypsin and centrifuged for 5 min at Fig. 7 GBP2 regulates glioma progression through KIF22. A Western blotting results confirming GBP2 knockdown efficiency and showing expression of EGFR, p-EGFR and KIF22 in U87 and U251 cells transfected with siCtrl, siGBP2, or siGBP2 + KIF22. B Cell viability assay results for U87 and U251 cells transfected with siCtrl, siGBP2, or siGBP2 + KIF22. C, D Colony formation assay results for U87 and U251 cells transfected with siCtrl, siGBP2, or siGBP2 + KIF22. Clusters of more than 50 cells were identified as colonies. E Transwell assay results for U87 and U251 cells transfected with siCtrl, siGBP2, or siGBP2 + KIF22. F U87 cells exogenously expressed shCtrl, shGBP2, and shGBP2 + KIF22 were injected into athymic nude mice (n = 5), and tumor formation was analyzed. Representative images and weight of xenografted tumors are shown. Each experiment repeated as least 3 times. Magnification: ×10. **P < 0.01, ***P < 0.001. pellets, and iced 75% ethanol was used to fix the cells for at least 2 h. After centrifugation and washing with D-Hanks, 0.5 mL of FxCycle™ PI/RNase staining solution was added, and the cells were incubated at room temperature for 15-30 min. Finally, samples were analyzed with 488 nm excitation, and emission was collected with a 585/42 bandpass filter or equivalent. Images are representative of three different experiments.

Cell apoptosis assay
Cell apoptosis of U87 and U251 cells was evaluated by Annexin V-FITC and propidium iodide (PI). Cell suspensions were collected and centrifuged at 13,000 rpm for 5 min, then washed with 200 µL 1× binding buffer and resuspended in 190 µL 1×binding buffer. PI (10 µL, 20 µg/mL) was added to the samples and they were analyzed by flow cytometry. Each treatment groups included at least triplicate wells.

Xenograft model
PBS (0.1 ml) containing 1 × 10 7 U87 cells transfected with shRNA, shGBP2 or shGBP2 + KIF22 overexpression were injected subcutaneously into the left posterior flank area of each mouse, and the tumor size was evaluated twice a week using a caliper. Athymic nude mice, 5-6 weeks old, were randomly divided into three groups, with five mice per group. The mice were sacrificed after 30 days. Vernier calipers were used to measure the length and width of the tumor, and the tumor volume was calculated according to the following formula: V = 1/2*ab 2 (a, length; b, width). This study was approved by The First Hospital of Shanxi Medical University Animal Ethics Committee.
Co-immunoprecipitation experiment and mass spectrometry FLAG-tagged GBP2 or KIF22 was overexpressed in 293T cells and lysed with a low-salt lysis buffer supplemented with proteinase inhibitors and phosphatase inhibitors. Immunoprecipitations were carried out using 1 mg of whole-cell extract and 3 µL of α-FLAG antibody. Immunoprecipitations were incubated for 2 h at 4°C. Samples were washed with 1 mL immunoprecipitation buffer three times, and detected by mass spectrometry analysis or SDS-PAGE. The protein derived from co-IP assay were detected using mass spectrometry assay. Briefly, equal amounts of proteins were loaded and separated in 12% SDS-PAGE gels. The gels were stained with Coomassie blue (SimplyBlue SafeStain, Invitrogen) overnight before excised into slices. Then, the protein bands of interest were excised and the gels were de-stained and digested with trypsin at 37°C overnight. The digested proteins were then desalted for LC-MS/MS analysis (Genechem, Shanghai) according to previous studies indicated [30]. The proteins were identified and quantitated by Protein Pilot 4.0TM software (AB Sciex, USA).

Microarray
TRIzol reagent (Invitrogen) was used to extract total RNA, and agarose electrophoresis and a NanoDrop instrument (Thermo) were used for quality control and quantification. Microarray analysis was performed by Shanghai GeneChem Co., Ltd. (Shanghai, China). The dysregulated gene was identified by p < 0.05 and fold change >2. Moreover, the gene expression network was performed by Ingenuity Pathway Analysis (IPA).

Statistical analysis
Data were expressed as mean ± standard deviation, and analyzed using GraphPad Prism 8.0. Differences between two groups were analyzed by two-tailed Student's t test. Differences among more than two groups were analyzed by one-way ANOVA followed by Tukey's post hoc test. P < 0.05 was considered to indicate statistically significant differences.

DATA AVAILABILITY
All data generated or analyzed during this study are included in this published article.