EphA2-mediatedM1-like polarization of microglia attenuatesglioblastoma metastasis

Xiang Liao (  liaoxiang025@126.com ) Nanjing Jinling Hospital Ru Fang Jinling Hospital Ying Tian Nanjing Jinling Hospital: East Region Military Command General Hospital Chen Chen Changzhou First People's Hospital Zhijun Wu Southeast University Zhongda Hospital Jing Sun Nanjing Jinling Hospital: East Region Military Command General Hospital Jianrui Li Nanjing Jinling Hospital: East Region Military Command General Hospital Xiaoxue Liu Nanjing Jinling Hospital: East Region Military Command General Hospital Ming Li Ma'anshan People's Hospital Guangming Lu Nanjing Jinling Hospital: East Region Military Command General Hospital

subtype has been reported to block GBM progression [5,6]. Despite the convincing functional of M1-like macrophage, the precise mechanisms for the polarization of TAM to M1-like subtype remain largely unclear.
As the most important class of receptor tyrosine kinases, Ephrin(Eph) kinases family attract increasing attention in carcinogenesis and tumor progression [7]. Different from the majority of Eph kinases that are mostly synthesized during the developmental process, EphA2 is mainly restricted to proliferating epithelial cells in adults and expressed high level on in human tumor tissue specimens and established cancer cell lines [8,9]. Recently, more and more researches con rmed that EphA2 expression in cancer cell highly associated with poor prognosis and reduced survival of tumor patients [10,11]. EphA2 overexpression in epithelial cells can promote oncogenic transformation [12]. As a direct transcriptional target of Ras-Raf-ERK signaling, EphA2 is overexpressed in Ras-transformed cells and prompted breast cancer progression [13]. EphA2 regulates glioblastoma cell proliferation via the MEK/ERK/RSK pathway [14]. Under normal conditions, by binding several receptors, EphA2 could also induce diverse signaling networks in the neighbor cell and following cell-to-cell contact [15]. However, the exact role of EphA2 in microglia polarization and the underlying molecular mechanism remains unclear. Therefore, we are interested if EphA2 induced the M1-like polarization of microglia and further inhibit the metastasis of GBM.
To test this hypothesis, we used the overexpressing and silencing methods and the results suggested that EphA2 induced the M1-like polarization of microglia and PI3K/ AKT signaling axis was involved in the process. Finally, we shown that EphA2 induced the M1-like polarization of microglia attenuated the migration and invasion of GBM cells in vitro and in vivo. Taken together, our ndings demonstrate that, distinct from its role in cancer cells, EphA2 expressed in microglia is a negative regulator of GBM progression.

Materials And Methods
Cell culture and treatment C6 cells were maintained in Dulbecco's modi ed Eagle's medium (DMEM) supplied with 10% fetal bovine serum (FBS) and 1% antibiotics. GL261 and BV2 cells were cultured in high glucose DMEM medium with 10% FBS and 1% antibiotics. All cells were incubated in 5% CO 2 at 37 °C. To inhibit or activate PI3K-AKT pathway, cells were treated by Honokiol (HY-N0003, MCE, China) or sc-79 (GC11645, Glpbio, USA) for 24 h.

TCGA analysis
The gene expression of glioma patients in TCGA (The Cancer Genome Atlas) were downloaded by Cancer Brower site (https://xenabrowser.net/datapages/), and the correlation between CD80 and MMP9, BIRC5, MYC and EphA2 was analyzed. High and low groups were de ned as above and below the mean respectively RNA isolation and quantitative real-time PCR (qPCR) Total RNA was isolated from cells with Total RNA Extraction Reagent (R401-01, Vazyme, Nanjing) according to the manufacturer's instructions. One microgram of RNA was reverse transcribed to cDNA with Hiscript Q RT SuperMix for qPCR (R122-01, Vazyme, Nanjing) as described by the manufacturer's instructions. Quantitative real-time PCR was performed using ChamQ SYBR Master Mix (Q311-02, Vazyme, Nanjing) and a LightCycler 96 (Roche, Risch-Rotkreuz, Switzerland) with the following conditions: 5 min at 95 °C, followed by 40 cycles at 95 °C for 30 s, 60 °C for 40 s and 72 °C for 60 s. All primers used for qPCR are shown in Table 1.

Immunohistochemistry
Tissues were xed in formalin, dehydrated with alcohol, and para n-embedded. Five-micron sections of tissues were cut and stained with eosin and hematoxylin. For IHC staining, 5 µm sections of tissues were stained with antibodies at 4 °C overnight, followed by incubation with HRP-labeled secondary antibodies for 1 h at RT and visualization with a DAB Substrate Kit (Cell Signaling Technology, USA).

Isolation of microglia from neonatal rat
Microglia of neonatal rat were prepared as described elsewhere [16]. Brie y, brains of neonatal SD rat were dissected and dissociated. The cells were seeded in DMEM-F12 medium containing 10% FBS with 75 cm 2 culture ask. On the 14th day, ask was agitated on rotary shaker at 240 rpm for 3 h at 37℃. Collected microglia from the supernatant and suspended the pellet after centrifugation.

Transwell assay
Tumor cell migration and invasion were analyzed by Transwell assays without or with Matrigel. Brie y, 10 4 C6 or GL261 cells were seeded in the upper chamber of a 24-well Transwell or Matrigel chamber with 8 µm pores (Corning Inc., Corning, NY, USA) and serum-free medium. The bottom chamber contained complete medium with 10% FBS and 10 4 BV2-Ad-con or BV2-Ad-EphA2. Following incubation for 24 h or 36 h, the cells on the upper membrane surface were removed with cotton swabs, and cells that migrated or invaded through the lter were xed with 4% paraformaldehyde for 15 min and stained with 0.5% crystal violet for 5 min. The number of migrated or invaded cells was calculated in 6 randomly selected elds, and each condition was analyzed in triplicate.

Protein digestion and iTRAQ labeling
Protein digestion was performed according to the FASP procedure described in Wisniewski, et al. [17], and

Enrichment of phosphorylated peptides by TiO2 beads
Labeled peptides were mixed, concentrated with a vacuum concentrator and resuspended in 500 µl of loading buffer (2% glutamic acid/65% ACN/2% TFA). Then, TiO2 beads were added and agitated for 40 min. Centrifugation was performed for 1 min at 5000 g, resulting in the rst beads. The supernatant from the rst centrifugation was mixed with additional TiO2 beads, resulting in the second beads, which were collected as before. Both beads were combined and washed with 50 µL of wash buffer I (30% ACN/3% TFA) three times and then with 50 µL of wash buffer II (80% ACN/0.3% TFA) three times to remove the remaining non-adsorbed material. Finally, the phosphopeptides were eluted with 50 µL of elution buffer (40% ACN/15% NH4OH) [18], followed by lyophilization and MS analysis.
Five microliters of phosphopeptide solution was mixed with 15 µl of 0.1% (v/v) tri uoroacetic acid, and 10 µl of the solution mixture was then injected for nanoLC-MS/MS analysis using a Q Exactive MS (Thermo Scienti c) equipped with Easy nLC (ProxeonBiosystems, now Thermo Scienti c). The peptide mixture was loaded onto a C18 reversed-phase column ( For MS analysis, peptides were analyzed in positive ion mode. MS spectra were acquired using a datadependent top10 method dynamically choosing the most abundant precursor ions from the survey scan oxidation(M), phosphorylation (S/T/Y). The score threshold for peptide identi cation was set at a 1% false discovery rate (FDR) and a phosphoRS site probability cutoff of 0.75 [19].

Stereotactic implantation of glioma cells
Rats were iso urane-anesthetized and placed in a stereotaxic instrument (Anhui Zhenghua Co., LT., Huaibei, China). The stereotaxic coordinates of the intracranial injection were AP = + 1.6 mm to the bregma, ML = − 2 mm lateral to the bregma, and DV = − 6 mm ventral. The animals received a single microinjection of 1 × 106 C6 rat glioma cells (total 10 µl) or DMEM (10 µl) into the right stratum using a 25 µl syringe (Gaoge, Shanghai, China). The rats were sacri ced on day 14 after injection.

M1-like polarization of microglia negatively correlation with GBM progression
To investigate the role of microglia polarization in GBM progression, we rstly assessed the M1-like and M2-like microglia related markers expression in GBM orthotopic tumor tissue. The expression level of CD80, M1-like microglia marker, was lower in the tumor tissues than the normal brain tissues, while the expression level of CD206, M2-like microglia marker was in verse as shown immunohistochemistry staining (Fig. 1A).
To connect this results to clinical application, using the GBM patient tissue specimens (patients information in Table 2), we performed the immuno uorescence staining and the results shown that CD80 was signi cantly downregulated in recurrence GBM tissues than the primary GBM tissues, while CD206 was in verse (Fig. 1B). Due to the limitations of the tissue samples, we evaluated the correlation of CD80 expression level and patients' survival based on the information in TCGA database. The heat map revealed a linear reverse relationship between CD80 and tumor proliferation and metastasis indicators (BIRC5,MYC, and MMP9, Fig. 1C), which suggested the negatively relationship between with GBM progression and M1-like polarization of microglia.
Interestingly, we noticed a similar trend of expression level between CD80 and EphA2 in the TCGA database (Fig. 1C). Using the GBM patient tissue specimens, we found the expression level of EphA2 was also higher in the primary GBM tissues than the recurrence GBM tissues, indicating the potential correlation between EphA2 and M1-like polarization of microglia (Fig. 1D).

GBM cells prevented EphA2 expression and M1-like polarization of microglia
The polarization of microglia and protein expression is regulated by both instinct and external factors. We next assessed the role of GBM cells in determining EphA2 expression and polarization of microglia using the co-culture system ( Fig. 2A). Co-cultured BV2 with the GBM cells (GL261 or C6) for the indicated time, qPCR shown that the expression level of CD80 in BV2 was downregulated in a time dependent manner in the presence of GL261 or C6, while the Arg-1 and CD206 were upregulated (Fig. 2B), which suggested GBM cells prevented M1-like polarization of microglia.
Moreover, co-cultured BV2 with GL261 or C6, the expression level of EphA2 in BV2 was signi cantly down-regulated in a time dependent manner with the peak inhibition at 24 h ( Fig. 2C and 2D). The similar results can also be acquired by immuno uorescence staining (Fig. 2E), indicating GBM cells inhibited EphA2 in microglia.

Epha2 Promoted The M1-like Polarization Of Bv2
To investigate the role of EphA2 in microglia polarization, we rst transfected EphA2 cDNA plasmid into BV2 cells and con rmed the overexpression of EphA2 proteins in these cells by western blot and qPCR ( Fig. 3A and 3B). qPCR shown that iNOS signi cantly upregulated, while CD206 and Arg-1 downregulated in EphA2 overexpression BV2 cell (Fig. 3C). The similar results can also be acquired in the sorted primary microglia (Fig. 3D). Western blot shown the expression level of iNOS, TNF-α and IL-1β in EphA2 overexpression BV2 cell increased (Fig. 3E). These results indicated that EphA2 induce M1-like polarization of microglia.
To further elucidate the functional role EphA2 in BV2 polarization, two stable EphA2-knockdown cell clones (EphA2 shRNA1 and shRNA2) were established, and high silencing e ciency (90% silencing) was con rmed by western blot ( Figure S1). Compared to the parental cancer cells, EphA2 knockdown cells showed the upregulation of CD206 and Arg-1, and downregulation of iNOS at the RNA level (Fig. 3F). Western blot shown the expression level of iNOS, TNF-α and IL-1β in EphA2 knockdown BV2 cell decreased (Fig. 3G). These results demonstrated knockdown EphA2 prevented M1-like polarization of microglia. Moreover, NLRP3 expression level increased in EphA2 overexpression BV2 cells, and decreased in EphA2-knockdown BV2 cells, which indicated that EphA2 was also involved in the in ammation response (Fig. 3E and G).
To further con rmed the role of EphA2 induced M1-like polarization of microglia in GBM cells metastasis, we performed Transwell migration and invasion assays. Co-cultured with EphA2 overexpression BV2 cells, the number of living C6 and GL261 GBM cells that passed through the membrane was much lower than the number of the cells for the control group (Fig. 3H-K). These results supported the hypothesis that EphA2 induced M1-like polarization of microglia prevents the invasion and metastasis potential of living GBM cells.

PI3K/Akt pathway is involved in EphA2-mediated M1-like polarization of BV2
To further understand the underlying mechanism of EphA2 regulating microglia polarization, iTRAQ labeling was rstly used to identify differentially expressed phosphorylated proteins. iTRAQ-LC-MS/MS analysis results demonstrated that a total of 2,232 phosphorylated proteins, 4,941 phosphorylated peptides and 7,136 phosphorylated sites were identi ed (Fig. 4A). A total of 443 quanti ed proteins with P < 0.05 and an expression change greater than 2-fold or less than 0.5-fold between the KO and control groups were manually selected. Compared with the control group, 370 proteins were upregulated and 73 were downregulated (Fig. 4B). To better understand the involvement of EphA2 in physiological functions in BV2 cells, Blast2 go software was used to determine all different phosphorylated peptides, and a GO functional enrichment analysis of proteins corresponding to the phosphorylated peptides was then performed by Fisher. The results demonstrated that EphA2 involved nucleic acid binding, intracellular organelles, nucleoplasm and organelles (data not shown). KEGG pathway enrichment analysis revealed that 21 phosphorylated peptides were related to PI3K-AKT pathway ( Fig. 4C and 4D).
As the PI3K-AKT is an important regulator of macrophage polarization, we hypothesized that it may function downstream of EphA2 to mediate the M1-like polarization of BV2. To test this hypothesis, we assessed the PI3K-AKT pathway related proteins in the EphA2-knockdown, EphA2-overexpression and parental BV2 cells. Compared to the parental BV2 cells, the expression level of p-PI3K and p-AKT was signi cantly increased in EphA2-overexpression BV2 cells (Fig. 4E-F), while the signi cantly decreased expression of p-PI3K and p-AKT in EphA2-knockdown BV2 cells was observed (Fig. 4G-H).
To further con rm the role of PI3K-AKT pathway in EphA2 mediated the M1-like polarization of BV2, PI3K-AKT pathway activator, sc-79, was also used. As expected, the expression level of p-PI3K and p-AKT was increased in BV2 cells in the presence of sc-79 (Fig. S2). Treated EphA2-knockdown BV2 cells with sc-79, the expression level of CD80 and IL-6 was signi cantly upregulated, while Arg-1 was downregulated as assessed by RT-PCR (Fig. 4I). Moreover, co-cultured with EphA2-knockdown BV2 cells in the presence of sc-79, the ability of C6 and GL261 metastasis and invasion was signi cantly inhibited (Fig. 4J).

EphA2-mediated M1-like polarization of BV2 inhibited GBM cells proliferation and metastasis in vivo
To validate our in vitro observation EphA2-mediated M1-like polarization of microglia inhibiting GBM cells metastasis, we performed tumor growth and metastasis assays in vivo. We rstly tested examined the metastasis of GMB cells in nude mice by injecting 1 × 10 6 C6 cells co-cultured with BV2 cells (control group) or EphA2 overexpression BV2 cells into the mouse tail vein. From the HE staining, we observed that C6 cells exposed to EphA2 overexpression BV2 cells showed signi cantly more and larger metastatic foci than the control group (Fig. 5A). Moreover, we also assessed the proliferation ability of GMB cells in nude mice by injecting 1 × 10 6 C6 cells co-cultured with BV cells (control group) or EphA2 overexpression BV2 cells. As shown in Fig. 5B-D, tumors from EphA2 overexpression BV2 cells co-culture group grew slower and smaller than the control group, and larger necrotic area was also observed from the HE staining. Moreover, there was a signi cant reduction of proliferation (survivin and c-myc) and metastasis (ZEB1 and β-catenin) related markers in protein expression level (Fig. 5E). These in vivo results further supported the in vitro evidence that EphA2-mediated M1-like polarization of microglia prevented GBM cells metastasis.

Discussion
In the present study, we demonstrated that EphA2 expressed on microglia from the GMB patients and orthotopic tumor tissue. Elevated EphA2 promoted the M1-like polarization and prevented M2-like polarization of microglia. EphA2 activates PI3K/ AKT in microglia, which is important signaling events underlying M1-like polarization of microglia. Furthermore, EphA2 induced the M1-like polarization of microglia attenuated the migration and invasion of GBM cells in vitro and in vivo (Fig. 6); therefore, activating EphA2 may contribute to the GBM inhibition tumor microenvironment.
As the main innate immune cells of the CNS, the primary goal of microglia is to maintain homeostasis [20]. In the GBM microenvironment, microglia and macrophages have been designated as M1-or M2-polarized cells in response to pro-in ammatory and anti-in ammatory cytokines [21,22]. Although this de nition was over-simpli ed, they are actually histologically indistinguishable from each other and still widely used in the research. In this study, we used the common accepted marker to de nition M1 polarized cells according to the previously publication [23,24]. It was con rmed that M2-like polarization of microglia contributed to thetumor progression and reduces survival, while M1-like polarization of microglia in verse. The polarization of microglia is determined by the external and instinct factors, but the exact underlying mechanisms are still unclear. In our study, M1-like polarization of microglia was in uenced by the surrounding and the expression level of EphA2. However, the exact avenue of GBM cells regulated the polarization of microglia still need further deeper investigation and exosome may play a critical role in the process. In the ischemic stroke model, LPS-exo exhibited potent anti-in ammatory and neuroprotective roles by skewing microglia polarity from the M1 phenotype to the M2 phenotype [25]. In multiple sclerosis (MS), BMSC-derived exosomes attenuated in ammation and demyelination of the CNS by regulating microglia polarization in an EAE animal model [26].
Under normal condition, EphA2 expression is mainly restricted to proliferating epithelial cells [27]. While EphA2 is upregulated at the gene and protein levels in human tumor tissue specimens, especially the highest malignancy tumors [28]. Therefore, overexpression and aggressive features of EphA2 in tumor cells and relatively low expression in most normal adult tissues make this protein a potential therapeutic target in cancer. Several EphA2 target therapy strategies had been developed in cancer treatment, such as blocking its expression and activation, promoting its degradation, or EphA2-based immunotherapy [29][30][31]. Suppression EphA2 in GBM cells displayed multiple malignant features inhibition, including impaired anchorage independent growth, proliferation, and migration [32].Yamaguchi et al. developed a Dendritic cell (DC)-based vaccine pulsed with EphA2-derived peptide (Eph-DCs) which shown a promise colorectal cancer inhibition effect [33].In our study, EphA2 expressed on microglia induced M1-like polarization of this subpopulation. Moreover, EphA2-metdiated the M1-like of microglia inhibited the GBM metastasis and growth. Therefore, the exact role of EphA2 in tumor is context dependent and its target therapy still need further evaluation before clinical translation.
In summary, we identi ed EphA2-PI3K-Akt signaling pathway as a driver for M1-like polarization of microglia, which further attenuated the migration and metastasis of GMB cells in vitro and in vivo. As the widely accepted concept of target EphA2 in cancer therapeutic intervention, our research provided a new information to rationale for targeting EphA2 to improve treatment outcomes in glioblastoma cancer.

Conclusion
M2 microglial polarization was associated with decreased EphA2 expression in microglia cocultured with GBM cells. BV2 cells displayed an M1 phenotype when EphA2 was overexpressed and an M2 phenotype when EphA2 was knocked down via PI3K-AKT pathway. Furthermore, EphA2 overexpression in BV2 cells inhibited GBM cell migration and invasion. In conclusion, our study con rmed that EphA2 promoted M1 microglia polarization, which blocked to glioma progression. This study further explains the mechanism of microglia polarization and provides an effective therapeutic target for the clinical treatment of GBM.
Abbreviations glioblastoma multiforme GBM The animal experiments conducted strictly in line with the Animal StudyGuidelines of Nanjing University.

Consent for publication
All authors have agreed to the publication of the article.

Availability of data and materials
The data supporting the ndings of this study are included in this paper andits additional les

Competing interests
The authors have declared that no con ict of interest exists.