LGALS3 Promotes Treatment Resistance in Glioblastoma and Is Associated with Tumor Risk and Prognosis

Gliomas represent the most frequent primary tumors in the central nervous system with glioblastoma (WHO grade 4), which is the most common and malignant histologic subtype (1). Despite the use of multimodal treatments, consisting of maximum surgical resection followed by radiotherapy and temozolomide chemotherapy, patients with glioblastoma have to follow a fatal course with the median survival of approximately 15 months (1). Moreover, targeted therapies, such as temsirolimus and bevacizumab (2, 3), against molecules that drive the growth of the bulk of glioblastoma have so far also been unsuccessful in clinical trials. The failure of tumor control at the primary site is caused, in great part, by the infiltrative nature of glioblastoma and acquired resistance, which makes tumor cells survive initial therapies and contribute to inevitable tumor relapse (4). Therefore, it is imperative to better elucidate the complicated pathobiology of glioblastoma and understand the molecular details underlying the response of tumor cells to genotoxins, both of which will promote to design more effective therapeutic strategies.

Galectins are a family of proteins characterized by their ability to bind β-galactoside–containing glycoconjugates through a conserved carbohydrate-recognition domain (CRD) and involved in many physiologic events, such as cell proliferation and differentiation, apoptosis, and immune response (5, 6). Moreover, dysfunction of galectins has been found to promote cancer development and progression (7). LGALS3 (also known as galectin-3) is the only chimera-type galectin with a single CRD connected to a long and flexible N-terminal domain (8), and also is one of the most extensively studied galectins in cancer. For instance, LGALS3 expression in gastric cancer increases tumor cell motility and confers an aggressive phenotype (9, 10), and contributes to metastasis in lung cancer (11), whereas the removal of LGALS3 increases the susceptibility of diffuse large B-cell lymphoma cells to cell death (12), and LGALS3 deficiency reduces cell proliferation and migration in hepatocellular carcinoma (13). However, the loss of galectin-3 has been shown to be associated with the epithelial–mesenchymal transition, increased sphere-formation ability, and drug resistance in some breast cancers (14), which is in contradiction with previous results that LGALS3 could enhance the metastatic potential of breast cancer cells (15). All indicate the multifaceted functions of LGALS3 in tumor biology. In addition, the alterations of LGALS3 expression also appear heterogeneous in cancers that most tumors of the digestive tract and urinary system display upregulation of LGALS3, but tumors of the reproductive system present decreased LGALS3 expression (5). In glioma, most of the studies related to LGALS3 have shown its high levels in astrocytic tumors, whose pathologic grade was significantly associated with the expression level of LGALS3 (16–18). Moreover, LGALS3 is also being reported to influence the glioma cell migration (19) and promote cell survival under hypoxic and nutrient-deprived conditions (20). However, the involvement of LGALS3 in the chemo- and/or radioresistance in glioblastoma has not yet been demonstrated.

Genome-wide association studies (GWAS) offer a powerful means to search for genes that confer susceptibility to complex diseases (21). More recently, this tool has identified SNPs at several loci with increased risk for glioma, such as 17p13.1 (TP53), 5p15.33 (TERT), 7p11.2 (EGFR), 9p21.3 (CDKN2B-AS1), and 20q13.33 (RTEL1; ref. 22), which has advanced our understandings of etiology in the malignancy. Previously, two genetic variants at LGALS3 SNP sites (rs4644 and rs4652) were reported to contribute to the susceptibility of glioma, and the patients carrying AA genotype of LGALS3 + 292A>C (rs4652) had marked shorter overall survival period than those did not (23, 24). However, other SNPs located in the LGALS3 gene were not included in those studies and the effect of SNPs on glioblastoma progression had not yet been assessed.

Here, we applied microarrays to detect the expression profile of radioresistant glioblastoma cells and identified LGALS3 as a potential effector. We further demonstrated LGALS3-mediating glioblastoma resistance to radio- and chemotherapy and promoting tumor cell proliferation. Meanwhile, we confirmed that LGALS3 expression was upregulated in glioblastoma and associated with poor prognosis of patients, and illustrated that SNP sites (rs4644 and rs4652) were related to increased risk in glioblastoma, and the C allele of rs4652 and the A allele of rs4644 could promote glioblastoma resistance to radio-chemotherapy. These findings highlight the importance of LGALS3 in glioblastoma treatment resistance and its potential to be a therapeutic target for this tumor.

The study protocol and acquisition of tissue specimens and blood samples were approved by both of the Ethical Review Committee of Second Military Medical University (Shanghai, China) and the Ethics Board of Fudan University (Shanghai, China). Written informed consent was supplied by all the participants. For the IHC-staining experiment, glioma specimens were obtained from the patients who underwent surgical treatment at Changzheng Hospital (Shanghai, China) from January 2000 to December 2010. Tumor histology of glioma was confirmed independently by two neuropathologists. Normal brain tissues were taken from severe trauma patients, on whom partial resection of the brain was required to reduce the increased intracranial pressure. For the GWAS, 961 newly diagnosed glioma patients and 1,351 healthy controls were consecutively enrolled between October 2004 and July 2009 from Huashan Hospital (Shanghai, China). Detailed recruitment criteria were defined previously (25). Each participant voluntarily donated peripheral whole blood sample of 3–5 mL for DNA extraction and genotyping.

Hek293T cells, the human glioblastoma cell lines (T98G and U251), and the human lung cancer cell line (A549) were obtained from the ATCC in 2013. These cell lines were last authenticated in the laboratory by short tandem repeat profiling based on eight markers in October 2017. Cells were cryopreserved and used within 6 months after thawing, not exceeding 10 passages. All cells were maintained in a humidified atmosphere of 5% CO₂ at 37°C and cultured in DMEM supplemented with 10% FBS. The procedures of packaging and infecting with lentivirus were done according to our previous study (26).

High-fidelity enzyme KOD Plus Neo (Toyobo) was used for PCR amplification for the construction. With the primers (F, GCTCTAGAGCCACCATGGCAGACAATTTTTCGCTC; R, CGGGATCCTTAAGTAGATGACTTACAAGAGGGATAT), PCR-amplified human wild-type LGALS3 (+191C and +292A) was cloned into the XbaI-BamHI sites of the CD513B vector. Mutant LGALS3 (+191A and +292C) was made using the QuickChange site-directed mutagenesis Kit (Stratagene).

To efficiently knock down the LGALS3 expression, shRNA-4 and shRNA-5, generated with 5′-GGAAGAAAGACAGTCGGT-3′ and 5′-GCTCACTTGTTGCAGTACAAT-3′ targeting the LGALS3 transcript respectively, were selected (Fig. 1A). Corresponding sense and antisense oligonucleotides were synthesized, annealed, and cloned into the HpaI-XhoI sites of the PLL3.7 vector.

Radioresistant T98G and A549 cells were established and maintained in our laboratory. Briefly, the radioresistant sublines (T98G-R and A549-R) were generated by repetitive pulse exposure of T98G glioblastoma cells and A549 lung cancer cells to ionizing irradiation using Gammacell-40¹³⁷Cs γ (MDS Nordion) for 2 Gy (0.8 Gy/minute) each time. Every other day, the undead T98G-R and A549-R cells were passaged and irradiated. The cycle was repeated for 2 months. When exposed to a total radiation dose of 60 Gy, the live T98G-R cells were seeded at 6-cm dish for monoclonal growth, all of which, as well as the rest survival T98G-R cells and 549-R cells, would then be harvested for RNA extraction. Control T98G and A549 cells were given the same treatments without ionizing irradiation. For experiments verifying the function of LGALS3 in glioblastoma cell radioresistance, T98G and U251 received irradiation with 5 Gy or 10 Gy one time. To detect the effect of LGALS3 on temozolomide resistance, T98G and U251 cells were cultured in the presence of 0, 100, 500, 1,000, 1,500, or 2,000 μmol/L of temozolomide (provided by Merck Sharp & Dohme Ltd) and growing for 30 hours prior to assays.

The total RNA was extracted from cell lines using TRIzol Reagent (Invitrogen) following the manufacturer's protocol. First-strand cDNA was synthesized using ReverTra Ace (Toyobo). The qRT-PCR was conducted by ABI PRISM 7900HT Sequence Detection system in the presence of SYBRGreen dye (Toyobo). Each sample was tested in triplicate. GAPDH was used as an internal control. All PCR primers used are as follow: LGALS3 forward primer, GGCCACTGATTGTGCCTTAT and reverse primer, GAAGCGTGGGTTAAAGTGGA; GAPDH forward primer, GCGAGATCCCTCCAAAATCAA and reverse primer, GTTCACACCCATGACGAACAT.

The cell samples were homogenized in lysis buffer containing a cocktail of 1% protease and 1% protease inhibitors (Sigma). Protein lysates were separated by 10% SDS-PAGE and transferred onto a PVDF membrane (Invitrogen), which were blocked in 5% milk for 1 hour and then treated with the appropriate antibody at 4°C overnight. The blots were visualized by an enhanced chemiluminescence detection system (Thermo Fisher Scientific). GAPDH was used as an internal control.

T98G and A549 cells treated with ionizing irradiation (60 Gy) and their respective control cells were harvested for RNA sample preparation. Then, the microarray hybridization was performed at KangChen Biotech Corporation by a HumanHT-12 v4 Expression BeadChip (Illumina). Data acquisition and analysis followed the standard Illumina Custom protocol. mRNAs with diffscore value >13 or <−13 were considered to be differentially expressed between radioresistance group and control.

The human specimens were fixed in formalin and then embedded in paraffin wax. The IHC assay was performed on these tissues to detect the LGALS3 expression and IDH1 mutation by methods described previously (26). LGALS3 antibody for IHC was obtained from Cell Signaling Technology (1:600, #87985). An anti-IDH1 R132H mAb (1:20, Dianova, DIA-H09) was used to detect the IDH1 status in glioma specimens that were recruited in the IHC assays. The intensity of positive-staining cytoplasm or nucleus of LGALS3 was scored on a scale of 0 to 3 (0: negative; 1: light brown; 2: medium brown; 3: dark brown), which was multiplied by the value of positive percentage in cytoplasm and nucleus, respectively. Then, the two scores of cellular cytoplasm and nucleus were added to estimate the expression level of LGALS3, which was divided into the high or low cut by the median total score 3. The status of IDH1 was directly interpreted as null or mutation, based on whether having a positive IHC staining.

The Cancer Genome Atlas (TCGA; www.cancergenome.nih.gov) provides multimodal data of more than 500 glioblastoma cases. Level 3 mRNA expression data, based on the Affymetrix microarrays (Human gene U133A) and Agilent microarrays (AgilentG4502A), clinical follow-up information, and molecular pathologic characteristics, were collected for further analysis. The expression of LGALS3 expression was classified as either high (expression value ≥11.595 or −0.70) or low (expression value <11.595 or −0.70), according to the data from the two types of microarrays, respectively.

Cells in the logarithmic phase of growth were seeded in 96-well plates (2,000 cells/well) in sextuple and allowed to grow for 1, 2, 3, 4, and 5 days. Cell proliferation assay was analyzed by the Cell Counting Kit-8 (CCK-8; Dojindo) following the manufacturer's instructions. The optical density was measured at 450 nm wavelength using a microplate reader (EL × 800, BIO-TEK). Data were obtained from three independent assays performed in triplicate.

Tag SNPs at the LGALS3 locus were selected from the genotype data for Han Chinese in Beijing, China (CHB) from the International HapMap Project (http://www.hapmap.ncbi.nlm.nih.gov) using the software Haploview V.4.0 (http://www.broad.mit.edu/mpg/haploview) to capture SNPs with a minimum minor allele frequency (MAF) of 0.05 in LGALS3 gene ± 2 kb with an r² of 0.80 or greater.

Genomic DNA was extracted from venous blood using the Qiagen Blood Kit (Qiagen), and diluted to a final concentration of 15–20 ng/μL. Primers for amplification and extension reactions were designed using the Mass-ARRAY Assay Design Version 3.1 software (Sequenom). Genotyping was performed with the Mass-ARRAY iPLEX platform (Sequenom) using an allele-specific MALDI-TOF mass spectrometry assay (27). Then, genotyping quality was examined via a detailed quality control procedure that ensured a >99% successful call rate with duplicate calling of genotypes, internal positive control samples, and Hardy–Weinberg equilibrium (HWE) testing.

The significance of the differences between groups with continuous data was evaluated by the Student t test. Data were presented as the mean ± SD unless otherwise indicated. Differentially expressed genes (DEG) detected in microarray were calculated by the Illumina Custom algorism, as the Diffscore of each DEG was more than 13 or less than −13 between radioresistant T98G cells and controls with P < 0.05. Wilcoxon rank-sum test was used to compare the difference of counting data between two groups. Spearman analysis was used to examine the relationship between LGALS3 expression and glioma grade. Progression-free survival (PFS) and overall survival (OS) curves were plotted by the Kaplan–Meier method and compared by log-rank test. The identification of relevant prognostic factors was performed by univariate analysis, multivariate analysis, and stepwise backward Cox regression model. Variables with a value of P ≤ 0.2 on univariate analysis were added into the multivariate analysis. OS and PFS were calculated in days from the date of diagnosis to the time of death and to the time of tumor progression or recurrence, or death of the patient from glioblastoma, respectively. HWE in control subjects was assessed by the standard χ² test (df = 1) for each SNP. Unconditional logistic regression was conducted to calculate ORs and 95% confidence intervals (95% CI) as estimates of relative risk for the single locus genotypes, adjusting for age and sex. SPSS software (SPSS) was used for all the statistical analyses. All statistical tests were two-tailed, the results were considered statistically significant when the P values were less than 0.05.

To investigate the genes involved in resistance of radiotherapy of glioblastoma, T98G cell line was chosen to be irradiated with 60 Gy. After irradiation, viable T98G cells were monoclonally selected into eight positive groups for culture, which was named as T98-1a, -1n, -2d, -2h, -3c, -4b, -4d, and -4p, respectively. The rest and the unselected T98G cells were mixed cultured and named as T98G-4th. Irradiation of lung cancer cell A549 (A549R) with the same dose was performed as a parallel experiment. As indicated by microarray analysis, a total of 4307 DEGs (Supplementary Table S1) were found in all eight monoclonal groups comparing with the T98G control (Fig. 2A). Only 3 DEGs occurred in all eight groups, among which five and more groups contained 68 same DEGs.

Furthermore, TCGA dataset was introduced to help filter out both the radiation and patient prognosis–related DEGs in the 68 ones, of which S100A, BMP2, LGALS3 (Fig. 2B and C), and NNMT genes stand out by a preliminary screening analysis. Moreover, LGALS3 was then found included in both T98G-4th and A549R DEGs. qPCR further confirmed the consistency of LGALS3 expression detected in the microarray (Fig. 2D). These results revealed that glioblastoma cells might not be resistant to radiation through some common mechanisms, but also escape the radiation-induced death in some specific manners. LGALS3 was potentially involved in a common pathway related to irradiation resistance in glioblastoma cell, as well as other types of cancer cell.

To confirm the function of LGALS3 in glioblastoma radioresistance, we introduced exogenous LGALS3 expression or interference vectors in T98G and U251 cells to detect the cell survival rate after short-term exposure of ionizing irradiation. As analyzed by CCK-8, we found that both glioblastoma cells with LGALS3 overexpression showed higher survival rate in 5 Gy and 10 Gy groups than that of the control, however, the downregulation of LGALS3 was related to the decreased survival ability of T98G and U251 cells (Fig. 1B and C).

We next examined whether LGALS3 played an important role in resistance to temozolomide therapy. T98G and U251 cells transfected with LGALS3-overexpressing, -depleting, or control vector were treated with a broad range of temozolomide concentrations, and the cell survival was also determined by CCK-8. As shown in Fig. 1D and E, less glioblastoma cell death induced by temozolomide at levels of 500, 1,000, 1,500, and 2,000 μmol/L were observed in LGALS3-overexpressing groups. But, stronger cytotoxic effects on the cell survival were found in T98G and U251 cells with depleted LGALS3 compared with the control cells.

To further address the biological importance of LGALS3 in glioblastoma, we detected whether other malignant phenotypes changed in T98G and U251 cells with stably increased or depleted LGALS3 expression. As determined by CCK-8 assay, both of the LGALS3-overexpressed T98G and U251 cells displayed higher proliferation rate than controls, whereas targeting LGALS3 by shRNA-4 and shRNA-5 could markedly inhibit cell proliferation (Fig. 1F and G). Collectively, these results suggest that LGALS3 may be a critical factor in regulating glioblastoma cell growth.

LGALS3 has been shown to be potentially associated with glioblastoma patient prognosis in TCGA database (Fig. 2; Supplementary Table S2). Thus, we then analyzed the LGALS3 expression and its correlation with patient survival in our cohort including 120 glioblastoma cases (Supplementary Table S3). As seen in Fig. 3, LGALS3 expression was noticeably increased in glioblastoma tissues when comparing with normal brains (3.10 ± 0.12 vs. 1.36 ± 0.19, P < 0.001), and positively correlated with pathologic grade of glioma (P < 0.001). To confirm the value of LGALS3 in the prognostic prediction of patients, univariate survival analysis (Supplementary Table S4) was first performed and found that patients with glioblastoma with low total LGALS3 expression score had a markedly longer OS and PFS than those in the high-score group in our cohort (OS, 10 vs. 18 months, P = 0.001; PFS, 7 vs. 15 months, P < 0.001; Fig. 3E and F). Furthermore, Cox proportional hazards regression model (Table 1) in the absence of MGMT methylation data (due to insufficient tissue for the analysis) identified LGALS3 expression as an independent prognostic factor of both OS (HR) = 1.946, P = 0.001) and PFS (HR = 2.294, P = 0.001) for patients with glioblastoma.

As known, LGALS3 facilitated glioblastoma development and progression, we next expanded our exploration of the association of LGALS3 gene with glioblastoma risk. A total of 961 glioma cases and 1,351 controls were recruited, and the characteristics of all participants were summarized in Supplementary Table S5. Five tag SNPs, rs4652, rs4644, rs7157768, rs8013027, and rs17128230, at LGALS3 locus, were discovered and the genotyping rate for all of them were over 99%. Among the healthy subjects, all genotypic frequencies were in HWE (P > 0.05). Notably, the allele frequencies of two SNPs (rs4644 and rs4652) were significantly different between the cases and the controls (P = 00.037 and 0.028, respectively; Supplementary Table S6). The genotype distribution of the five SNPs in the LGALS3 gene in case patients and control subjects and their associations with glioma risk are summarized in Supplementary Table S7. Logistic regression analyses revealed that CA+CC homozygotes of rs4652, after adjusting for age and sex, were significantly associated with an increased susceptibility to glioma compared with wild-type carriers in the dominant-effect model (adjusted OR = 1.19; 95% CI = 1.00–1.42). The homozygous variant genotype of rs4644 was individually associated with an increased risk of gliomas in the recessive-effect model (AA vs. AC+CC: adjusted OR = 2.10; 95% CI = 1.22–3.62). Furthermore, stratification analyses were performed using the histology subtypes (glioblastoma and other gliomas) for the two LGALS3 variants (rs4652 and rs4644). Analyses using codominant models showed both CC genotype of rs4652 and AA genotype of rs4644 associated with significantly increased risk of glioblastoma (OR = 1.44; 95% CI = 1.01–2.06 and OR = 2.81; 95% CI = 1.46–5.38, respectively; Table 2). Consistently, the AA genotype of rs4644 also displayed an increased association with risk in patients with glioblastoma in a recessive model (adjusted OR = 2.81; 95% CI = 1.47–5.36). However, no significant association was observed in the other glioma subgroup.

We then detected the impact of +191A and +292C haplotype on glioblastoma progression and treatment resistance. As shown in Fig. 4A and D, both T98G and U251 cells with overexpression of +191A and +292C LGALS3 displayed similar abilities of proliferation as those of LGALS3 wild-type (+191C and +292A) cells. However, LGALS3 with +191A and +292C haplotype increased the survival rate of T98G and U251 cells under different DNA damage stresses (Fig. 4B, C, E, and F). Collectively, these findings indicated that SNPs rs4644 and rs4652 were involved in the regulation of LGALS3 function in glioblastoma.

In this study, we have addressed the oncologic functions of LGALS3 in treatment resistance and glioblastoma progression, LGALS3 gene expression in predicting patient prognosis, and the role of LGALS3 gene susceptibility in tumorigenesis. Our results revealed that LGALS3 promotes glioblastoma cell resistance to irradiation and temozolomide chemotherapy, improved the ability of proliferation, and acted as an independent risk factor for OS. And we also identified two SNPs (rs4652 and rs4644) in LGALS3 associated with glioblastoma risk and found that the C allele of rs4652 and the A allele of rs4644 contributed to more resistance to chemo-radiotherapy. These data suggest that LGALS3 may be a driver of glioblastoma development and recurrence, and a potential target for cancer treatment.

Radiotherapy is a mainstay treatment for many types of cancer including glioblastoma, and the temozolomide agent has been approved for treating this tumor for over one decade (28). The goal of chemo-radiotherapy is particularly to achieve direct or indirect DNA damage, which results in the termination of cancer cell division and proliferation and ultimately triggers cell apoptosis (29, 30). However, DNA repair mechanisms removing DNA lesions induced by ionizing irradiation or temozolomide and the activation of pathways blocking apoptosis are found to facilitate cancer cell survival and, in great part, to drive glioblastoma resistance towards the standardized therapeutic regimen (31, 32). Previous studies have shown that LGALS3 contains an Asp-Trp-Gly-Arg (NWGR) motif in the C-terminal domain, which is highly conserved in the BH1 domain of Bcl-2 gene family including Bcl-2, Bcl-xL, and Bax, and critical for the antiapoptotic function of Bcl-2 (33). Subsequently, LGALS3 has been demonstrated as a vital antiapoptotic molecule, which can act like Bcl-2, and to promote treatment resistance in multitype cancers (33). Besides, LGALS3 is also found to interact with Bcl-2 in a lactose-inhibitable manner (34), whereas Bcl-2 is linked to chemoresistance and aggressive recurrence in glioblastoma (35). Another Bcl-2 family member, Bcl-xL/Bcl-2–associated death promoter (Bad), accounting for the intrinsic pathway of apoptosis, can be downregulated by LGALS3 after treatment with chemotherapeutic drug (36), and the expression of LGALS3 also stimulates the phosphorylation of Bad (ph-Bad), which is in a highly expressed and inactive state in glioblastoma (37). In addition, as responding to cancer-killing effects induced by agents, mitochondria will release cytochrome c to upregulate Bax delivering the death signal to form Bax/Bak homodimers or antagonize Bcl-2 antiapoptosis effect through forming Bax/Bcl-2 heterodimers at outer membrane of mitochondria, which results in disruption of mitochondrial membrane potential and activation of caspases in glioblastoma (38). However, LGALS3 can inhibit mitochondrial depolarization and damage, and then contribute to the prevention of cytochrome c release, fewer caspases activity, suppression of apoptosis, and anticancer drug resistance (36, 39). Thus, it is indicated that LGALS3 may function through a cell death inhibition pathway that may involve Bcl-2 family. Moreover, cancer stem cell in glioblastoma has been demonstrated as a core contributor in chemo- and radioresistance through preferential activation of the DNA damage checkpoint response and extensive DNA repair aberrations (40, 41). Recently, studies in other cancers showed that LGALS3 expression could identify a subset of cancer stem cells with high levels of stem cell characteristics, including chemotherapy resistance (42). It is speculated that the increased level of LGALS3 in glioblastoma stem cell (43) may drive the therapeutic resistance by maintaining the stemness of cancer stem cells in the malignancy, which needs further investigations.

In addition, the typical oncogenic pathway RAS signaling, which drives tumorigenesis and progression in glioblastoma (44), was also reported in the downstream of LGALS3, and the overexpression of LGALS3 activated RAS and induced cancer cell proliferation (45). Taken together, both mentioned may explain why the high expression level of LGALS3 is related to short survival time for patients with glioblastoma.

Our study also underscored the functional single nucleotide variants of LGALS3 in glioblastoma progression. Previous findings revealed that rs4652 and rs4644 could contribute to glioma risk and the SNPs at rs4652 (+292 A>C) were associated with tumor grade and prognosis of patients with glioma (23, 24).Though lacking sufficient tissue for enough DNA to validate the relationship between the two SNPs and clinical and pathologic characteristics of the patients in our cohort, we first determined the positive impact of the C allele of rs4652 combined with the A allele of rs4644 on resistance to ionizing irradiation and chemotherapeutic drug in glioblastoma. As reported previously, amino acids Ala62-Tyr63 of the LGALS3 harbor the cleavage site for matrix metalloproteinases (MMP) and rs4644 rendering residue 64 of LGALS3 changing from histidine to proline may alter the pattern of MMPs cleavage, which may result in elevated LGALS3 level (46, 47). However, different from the enhanced phenotypes of treatment resistance, the difference of cell proliferation abilities between +191 A - +292 C and +191 C - +292 A groups was not observed in T98G and U251 cells, which indicated diverse signaling pathways underlying LGALS3 in glioblastoma progression. In addition, it is also possible that SNPs change the configuration of LGALS3, which facilitates the interaction of LGALS3 with MUC1 on the cell surface to promote EGFR dimerization and activation, and activated EGFR has been found to promote glioblastoma cell therapeutic resistance (48–50). Besides, the two SNPs (rs4652 and rs4644) were identified to contribute to glioma risk, which needs additional studies to confirm the association between gliomagenesis and these SNPS, the genes, and the pathways underlying LGALS3. Especially, LGALS3 pathway–based approaches may have a potential in the identification of novel genes conferring tumor susceptibility.

In conclusion, this study demonstrates the oncogenic role of LGALS3, along with its functional SNPs, in glioblastoma chemo- and radioresistance. Further understanding of the mechanism involved by LGALS3 may accelerate to develop a useful therapeutic intervention for this malignancy.

No potential conflicts of interest were disclosed.

Conception and design: H. Wang, Y. Yan, H. Chen, J. Chen

Development of methodology: H. Wang, L. Hu, Y. Yan, J. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Wang, Q. Huang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Wang, X. Song, Q. Huang, D. Yun

Writing, review, and/or revision of the manuscript: H. Wang, X. Song, T. Xu, D. Yun

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Wang, T. Xu, Y. Wang, H. Chen, D. Lu, J. Chen

Study supervision: D. Lu, J. Chen

This project was supported by the National Natural Science Foundation of China (grant numbers 81572501 and 81272781). We thank Dr. Mei Jiang for her help in English language editing.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.