MEOX2 expression is positively correlated with glioma malignancy.
To evaluate the clinical association between MEOX2 with glioma, we firstly analyzed the expression pattern of MEOX2 in the TCGA glioma dataset. Results showed that MEOX2 expression was increased in parallel with WHO grades and histological stages (Fig. 1A). It’s observed that elevated MEOX2 was linked to molecular biomarkers, including wildtype IDH1, 1p/19q non-codeletion and MGMT promotor unmethylation, which represented increased tumor malignancy and poor prognosis (Fig. 1A). MEOX2 expression was significantly upregulated in GBM (WHO grade IV) compared with either normal brain or WHO grade II/III gliomas (including oligodendroglioma, astrocytoma and anaplastic oligodendroglioma) in the TCGA cohort (n = 681) (Fig. 1B). Moreover, MEOX2 was upregulated in the IDH wildtype gliomas, particularly in anaplastic astrocytoma and GBM (Fig. 1C).
We next quantified MEOX2 expression in the TCGA dataset for specific tumor anatomic structures from Ivy Glioblastoma Atlas Project. Results showed that high MEOX2 expression was enriched in regions of cellular tumor and infiltrating tumor (Fig. 1D), which are crucial in the progression of GBM. Finally, we assessed the prognostic value of MEOX2 expression in human gliomas by Kaplan-Meier survival analysis. The Kaplan-Meier survival curves were generated based on median values of MEOX2 expression in TCGA glioma datasets. Results showed that high MEOX2 patients exhibited significantly shorter overall survival (OS) than those with low MEOX2 (n = 681, P < 0.0001) (Fig. 1E).
To validate the correlation of MEOX2 with glioma malignancy, we further analyzed MEOX2 expression in the CGGA dataset (the Chinese Glioma Genome Atlas). Similarly, the mRNA level of MEOX2 was upregulated in GBM (WHO IV) compared with WHO grade II/III gliomas (Fig. 1F-G). In accordance with the expression patterns, Kaplan–Meier curves showed that high MEOX2 expression predicted poor OS (Fig. 1H). Altogether, we conclude that MEOX2 expression is increased with glioma malignancy, which can predict the prognosis of glioma patients.
MEOX2 is associated with distinct genomic alterations in gliomas.
To determine whether MEOX2 expression was correlated with specific genomic characteristics in glioma, we performed somatic mutation analysis and copy number variation (CNV) using the TCGA datasets. Analysis of somatic mutation profiles based on MEOX2 expression levels revealed a high frequency of mutations in TTN (25%), TP53 (20%), NF1 (13%), MUC16 (11%) and PIK3CA (10%) in the MEOX2-high group (n = 166), whereas mutations in IDH1 (88%), TP53 (61%), ATRX (46%), and CIC (16%) were more common in MEOX2-low group (n = 165) (Fig. 2A-B).
Furthermore, the CNAs profiles showed that amplification of Chr 7 accompanied by deletion of Chr 10 was enriched in gliomas with high MEOX2 expression, which are both common genomic events in GBM (Fig. 2C). Meanwhile, the incidence of the 1p/19q codeletion was more frequent in low MEOX2 expression samples (Fig. 2C).
Moreover, GISTIC analysis identified significantly recurrent amplification and deletion events. In MEOX2-high samples, frequently amplified genomic regions contained oncogenic driver genes, including PIK3C2B (1q32.1), PDGFRA (4q12), EGFR (7p11.2) and CDK4 (12q14.1), whereas deleted regions contained tumor suppressor gene CDKN2A/CDKN2B (9p21.3) (Fig. 2D). Conversely, in MEOX2-low samples, significant amplifications showed peaks in 4q12, 8q23, 8q24.13, and 12p32.32, whereas the frequently deleted genomic regions were 2q37.3, 4q34.3, 9p21.3, 10q26.13 and 19q13.42 (Fig. 2E). All these findings uncover the genomic alternations correlated with distinct MEOX2 expression in gliomas.
Validation of MEOX2 expression pattern and survival prognosis.
To validate the above findings from the public database, we examined the MEOX2 expression profiles in glioma cohort of our institution. We found that the mRNA level of MEOX2 was increased in tumor tissues compared with paired adjacent peritumors (n = 20) (Fig. 3A). The ROC curves further confirmed that MEOX2 expression was a predictive biomarker of glioma (AUC = 0.840) (Fig. 3B). Consistently, the protein level of MEOX2 was also upregulated in tumor tissues compared with paired adjacent peritumors (Fig. 3C-D).
Based on the bioinformatics and experimental evidence, we next asked whether altered MEOX2 affected the malignant behaviors of glioma cells. As a first step toward assessing the biological functions of MEOX2 in glioma, we conducted a lentivirus-based strategy to manipulate MEOX2 expression in the KNS89 glioma cell line. Western blot results showed that the MEOX2 protein level was dramatically increased by MEOX2 overexpression (OE) compared with control (Ctrl), and decreased in MEOX2 knockdown (KD) group compared with matched control (NC) (Fig. 3E-G).
MEOX2 promoted the proliferation, migration, and invasion of glioma cells.
Next, we tested the hypothesis that MEOX2 regulates malignant behaviors of glioma in established cell models. First, we examined whether MEOX2 regulated cell proliferation of glioma cells. Results of CCK8 assay showed that overexpression of MEOX2 increased cell proliferation remarkably (Fig. 4A), whereas silencing MEOX2 decreased the cell growth of glioma cells (Fig. 4B). Consistently, colony formation assay showed similar results that MEOX2 positively regulated the growth of glioma cells (Fig. 4C-F). Moreover, results of the EdU assay revealed that the ratio of EdU-positive cells was dramatically increased in MEOX2-OE cells (Fig. 4G-H), and decreased by MEOX2-KD (Fig. 4I-J).
Then, we investigated the effect of MEOX2 on the migration and invasion of glioma cells. Results of the wound healing assay showed that MEOX2 OE accelerated the migration rate of glioma cells (Fig. 5A-B), while MEOX2 KD decelerated it (Fig. 5C-D). Additionally, results of transwell assay confirmed that MEOX2 OE increased both cell migration and invasion of glioma cells, and MEOX2 KD displayed opposite phenotypes (Fig. 5E-H). All data reveal that MEOX2 promotes proliferation and migration/invasion of glioma cells.
MEOX2 inhibited TMZ-induced cell death in glioma cells.
Afterwards, we determined whether MEOX2 was involved in cell death/survival under basal and chemotherapeutic conditions. MEOX2 OE decreased the apoptotic rate of glioma cells by TMZ treatment (400 and 600 µM) indicated by Hoechst staining (Fig. 6A-B). Conversely, MEOX2 KD increased the tendency of TMZ-induced cell death (particularly by 600 µM TMZ treatment) (Fig. 6C-D). Furthermore, data from flow cytometry by Annexin-V flour 594/7-AAD staining showed similar results that MEOX2-OE cells were more resistant to the apoptosis induced by TMZ, whereas MEOX2-KD increased the sensitivity to TMZ (Fig. 6E-F). All these results highlight the role of MEOX2 in glioma cell survival, suggesting that MEOX2 is a potential therapeutic target for glioma intervention.
MEOX2 promoted glioma progression in vivo.
To further verify the pro-tumor functions of MEOX2 in vitro, we examined whether MEOX2 regulated glioma malignancy in vivo, by transplanting MEOX2-OE and KD cells into the right frontal cortex of immunocompromised nude mice. As anticipated, MEOX2 OE significantly increased tumor progression indicated by the maximal coronal sectional area of the tumor after orthotopic xenograft (Fig. 7A-B). Reciprocally, MEOX2 KD decreased the volume of glioma xenograft (Fig. 7C-D). In compliance with the intracranial tumor progression, body weight curves showed that mice inoculated with MEOX2-OE cells exhibited decreased body weight compared with Ctrl, and mice with MEOX2-KD cells had increased body weight (Fig. 7E-F). All these in vivo data support the notion that MEOX2 promotes glioma progression.
Mechanistic exploration of MEOX2 on glioma malignancy.
Next, we wonder the molecular mechanisms by which MEOX2 promotes glioma malignancy. Given the fact that MEOX2 is a homeobox transcriptional factor, we first examined the global transcriptome profiling of KNS89 cells infected with lentiviruses expressing either MEOX2 OE or KD by RNA-seq. Results showed that there were 373 genes altered by MEOX2 OE, including 125 downregulated and 248 upregulated genes (Fig. 8A, left panel). Gene ontology (GO) analysis showed that the differential expression genes (DEGs) were enriched in oncogenic pathways (e.g., PI3K/AKT), extracellular matrix modeling and focal adhesion (Fig. 8A, right panel). Similarly, there were 1,364 genes altered by MEOX2 KD, which were enriched in PI3K/AKT, transcriptional regulation in cancer pathways (Fig. 8B).
The transcriptional repertoire suggests that MEOX2 may regulate glioma malignancy through PI3K/AKT related signaling network. To test this hypothesis, we examined related pathways and found that MEOX2 OE indeed activated AKT signaling, indicated by the increased phosphorylations of AKT kinase at both Ser473 and Thr308. It’s also noted that MEOX2 increased mTORC1 activity, indicated by increased pS6. MEOX2 OE decreased the pro-apoptotic factor Bax, which may contribute to the decreased chemotherapeutic sensitivity (Fig. 8C). Meanwhile, MEOX2 KD reciprocally decreased AKT and mTORC1 activity, and CDK1 levels, which may impair glioma cell proliferation (Fig. 8D).