Identification of prognostic values defined by copy number variation, mRNA and protein expression of LANCL2 and EGFR in IDH1/2-wild-type glioblastoma


 Background

Epidermal growth factor receptor (EGFR) and lanthionine synthetase C-like 2 (LanCL2) genes locate in the same amplicon, and co-amplification of EGFR and LANCL2 is frequent in glioblastoma. However, the prognostic value of LANCL2 and EGFR co-amplification, and their mRNA and protein expression in glioblastoma remain unclear yet.
Methods

This study analyzed the prognostic values of the copy number variations (CNVs), mRNA and protein expression of LANCL2 and EGFR in glioblastoma specimens from TCGA database or our tumor banks.
Results

The amplification of LANCL2 or EGFR, and their co-amplification were frequent in glioblastoma of TCGA database and our tumor banks. CNVs of LANCL2 or EGFR were significantly correlated with IDH1/2 mutation but not MGMT promoter methylation status. LANCL2 or EGFR amplification, and their co-amplification were significantly associated with reduced overall survival (OS) of glioblastoma patients, rather than IDH1/2-wild-type glioblastoma patients. mRNA and protein overexpression of LANCL2 and EGFR was also frequently found in glioblastoma. LANCL2, rather than EGFR, was overexpressed in relapsing glioblastoma, compared with newly diagnosed glioblastoma. However, mRNA or protein expression of EGFR and LANCL2 was not significantly correlated with OS of glioblastoma patients. In addition, the intracellular localization of LanCL2, not EGFR, was associated with the grade of gliomas.
Conclusions

Taken together, amplification and mRNA overexpression of LANCL2 and EGFR, and their co-amplification and co-expression were frequent in glioblastoma patients. Our findings suggest that CNVs of LANCL2 and EGFR were the independent diagnostic and prognostic biomarkers for histological glioblastoma patients, but not for IDH1/2-wild-type glioblastoma patients.


Background
Glioblastoma multiforme (glioblastoma, GBM), belonging to the highest World Health Organization (WHO) grade glioma (grade IV), is the most common malignant and aggressive primary brain tumor (47.7%) with a high mortality rate (1,2). According to the 2018 CBTRUS report, the incidence rate of GBM is 3.21 per 100,000 population, which is the highest in malignant brain tumors (2,3). Under the Stupp's therapeutic protocol (maximal surgical resection followed by adjuvant radiotherapy and chemotherapy with temozolomide), the median overall survival of GBM patients is 14.6 months, while the 2-year overall survival (OS) rate and progression-free survival (PFS) rate is 26.5% and 10.7%, respectively (4). Aberrations of molecular markers such as O(6)-methylguanine-DNA methyltransferase (MGMT) promoter methylation, codeletion of 1p and 19q, isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) mutation, telomerase reverse transcriptase (TERT) promoter mutation, TP53 mutation, and epidermal growth factor receptor (EGFR) overexpression show prognostic signi cance to guide treatment decisions of GBM patients. In particular, IDH1/2 gene mutations are found in more than 70% of grade II-III glioma and secondary GBM that arises from low-grade glioma. GBM patients with IDH1/2 gene mutations often have a better clinical outcome than those with wild-type IDH (5). TERT promoter mutation is found in approximately 80% of patients with primary GBM that develops rapidly without any clinical or histologic evidence of a less malignant precursor lesion. GBM patients with TERT mutation often have poor survival and a high risk of death (6).
EGFR, a member of receptor tyrosine kinases (RTKs), is essential to the pathological process in various cancers via activation of PI3K/Akt signaling pathway. Analyzed by The Cancer Genome Atlas (TCGA) database, overall alterations including ampli cation, mutation, rearrangement and altered splicing of EGFR in GBM are highly frequent (57.4%) (7). Compared with secondary GBM, EGFR ampli cation (36%) and overexpression (more than 60%) are more common in primary GBM (8,9). Evidence shows that EGFR gene ampli cation has a strong correlation with EGFR overexpression. Approximate 98% of primary GBM with EGFR ampli cation also exhibit EGFR overexpression, while 70%-90% of those with EGFR overexpression show EGFR ampli cation (8,10). A number of studies demonstrate that the ampli cation and overexpression of EGFR are associated with poor prognosis of GBM patients, especially young people (10)(11)(12).
However, a retrospective study shows that EGFR ampli cation is not a prognostic factor for GBM patients treated with surgery. And a meta-analysis also shows that EGFR ampli cation is not signi cantly associated with OS of GBM patients, indicating a heterogeneity of signi cance among difference studies and subjects (13).
Lanthionine synthetase C-like 2 (LanC Like 2, LanCL2), a member of eukaryotic LanC-like protein family, is a homologue of prokaryotic LanC involved in the synthesis of the antibiotic named as lantibiotics (14). LanCL2 is a receptor of abscisic acid (ABA) which is not only a plant hormone but also an endogenous mammalian hormone involved in glycemic control (15). It is also known as testis adriamycin sensitivity protein (TASP) that is able to increase sensitivity of tumor cells to adriamycin via reduction of P-glycoprotein (16). Accumulating evidence show that LanCL2 plays important roles in the regulation of stress response, in ammation and glycometabolism, providing potential target for the treatment of chronic in ammatory, metabolic and immune-related diseases (17,18). LANCL2, along with SEC61G and ECOP genes, are located in the ank of EGFR gene at chromosomal 7p11.2. These genes are in the same amplicon, and their co-ampli cation with EGFR is common in GBM patients (19,20). However, it is not clear that whether co-ampli cation of EGFR and LANCL2 has prognostic value for GBM patients, and what are their mRNA and protein expression patterns.
Here, this study analyzed the copy number variations (CNVs), mRNA and protein expression pro les, and their prognostic values of LANCL2 and EGFR in GBM specimens from TCGA database or from the tumor banks of Shenzhen Second People's Hospital and Sun Yatsen University Cancer Center. We showed that ampli cation and mRNA overexpression of LANCL2 and EGFR, and their co-ampli cation and co-expression were frequent in GBM patients. CNVs of LANCL2 and EGFR were the independent diagnostic and prognostic biomarkers for histological GBM patients, but not for IDH1/2-wild-type GBM patients. The protein expression pattern and role of LanCL2 were independent to EGFR. LanCL2 overexpression was correlated with glioblastoma recurrence, and its activation may trigger its translocation into the nucleus.

TCGA database analysis
CNVs and mRNA expression data analyzed using the GISTIC2 algorithm in the TCGA database were achieved in the cBio Cancer Genomics Portal (http://www.cbioportal.org) (21,22). The TCGA Pan-Cancer Atlas dataset involving more than 11,000 human tumors across 33 different cancer types was selected (23)(24)(25)(26). The clinical data of GBM and low-grade glioma (LGG) from the TCGA Pan-Cancer database were downloaded to analyze the OS and PFS of the cohort using Kaplan-Meier survival analysis and log-rank test. Chi-square test was carried out to estimate the correlation of CNVs, while the correlation of mRNA expression (RNASeq V2 RSEM) were calculated by Pearson's correlation. CNVs including shallow deletion (possibly heterozygous deletion), diploid, low-level gain and high-level ampli cation were de ned as the putative copy number values of -1, 0, 1 and 2, respectively.

Tumor specimens
Tumor specimens were retrospectively obtained from the tumor banks in the Shenzhen Second People's Hospital and the Sun Yat-sen University Cancer Center. All tumor samples were histologically diagnosed as GBM (WHO grade IV). Four human normal brain tissues (including two craniocerebral trauma, one para-carcinoma and one epilepsy) and four grade I gliomas were used as the negative controls.
Identi cation of all tumor samples or normal brain tissues were con rmed by an experienced pathologist. This study was approved by the Research Ethics Committee of Shenzhen Second People's Hospital and Sun Yat-sen University Cancer Center. All patients were given written informed consent.
DNA extraction and copy number assay Genomic DNA (gDNA) was extracted using QIAamp DNA Mini Kit (QIAGEN) and copy number variations were evaluated by TaqMan Copy Number Assays (Thermo Scienti c) following the manufacturer's instructions. The Taqman Copy Number Assay probes for LANCL2 (Hs04953915_cn) and EGFR (Hs04983302_cn) genes were used for copy number quantitation, while TaqMan Copy Number Reference Assay RNase P was served as the reference. TaqMan Genotyping Master Mix was employed for the PCR ampli cation procedure, and 20 ng/well of gDNA was added in each PCR reaction, which was performed in ABI Quantstudio™ DX. Each reaction was duplicated.

Western blotting
Total proteins were extracted by RIPA lysis buffer and protein concentrations were determined using the BCA protein assay (Thermo Scienti c). Proteins were then separated by 8% SDS-PAGE and transferred to PVDF membranes (Millipore). After blocking with 5% non-fat milk or 5% BSA, membranes were incubated with gentle agitation in primary antibodies (1:1000) overnight at 4°C and then in HRP-conjugated secondary antibodies (1:5000) for 1 hr at room temperature. Positive signals were visualized by ECL chemiluminescence using ChemiDoc MP Imaging System (Bio-Rad).

Statistical analysis
Data were presented as mean ± S.E.M and all statistical analyses were carried out using GraphPad Prism 8 software. Relative protein expression was evaluated by measurement of density of Western blotting bands using Image J software. Difference among groups which did not follow a normal distribution was compared using the Mann-Whitney U test or Kruskal-Wallis One-way ANOVA with Dunn's multiple comparisons test. Patients' survival was analyzed using the Kaplan-Meier method and statistical comparisons were evaluated using the log-rank test. The distribution of categorical values within two groups was analyzed by the chi-square test (Fisher's exact test).
The difference was considered to be signi cant at p <0.05. The licenses of software are available under any requirement for permission for use.

Results
Ampli cation and co-ampli cation of LANCL2 and EGFR were prevalent in glioblastoma, but were not the independent prognostic markers for IDH1/2-wild-type glioblastoma patients.
Firstly, to investigate the CNVs of LANCL2 and EGFR genes in a panel of cancers, 32 studies of different cancer types in TCGA Pan-Cancer Atlas database (n=10967) were selected. Results showed that the dominant genomic alterations of LANCL2 and EGFR in cancers were ampli cation and mutation, while gene fusion and deep deletion were rare. Glioblastoma, head and neck squamous cell carcinoma, esophagogastric adenocarcinoma and non-small cell lung cancer were the top four tumors with the highest alteration frequencies of LANCL2 and EGFR ( Fig.1A-B). Subsequently, two studies Glioblastoma Multiforme (n=592) and Brain Lower Grade Glioma (n=514) were further analyzed. The ampli cation frequencies of LANCL2 and EGFR in GBM were up to 27.65% (159 of 575 cases) and 44.35% (255 of 575 cases), whereas those in LGG were only 3.91% (20 of 511 cases) and 7.63% (39 of 511 cases), respectively (Fig.1C). The data of LGG contained grade II and III gliomas, including oligodendroglioma, oligoastrocytoma and astrocytoma. Among the LGG data, the ampli cation frequencies of LANCL2 and EGFR in astrocytoma were the highest (7.33% and 13.92%, respectively), while those in oligoastrocytoma were the lowest (1.07% and 1.60%, respectively) (Fig.1D). Co-ampli cation of LANCL2 and EGFR was common in GBM, but it was rare in LGG. LANCL2 ampli cation was found in 61.96% of GBM samples and 51.28% of LGG samples containing EGFR ampli cation. Furthermore, nearly all GBM and LGG samples containing LANCL2 ampli cation displayed EGFR ampli cation (Fig.1E). The main types of LANCL2 and EGFR CNVs in GBM were copy number gain and ampli cation, whereas shallow deletion and diploid were infrequent. Chi-square test demonstrated a signi cant correlation between the CNVs of LANCL2 and EGFR (p<0.001) (Fig.1F). We analyzed the top ten genes which had the highest co-ampli cation frequencies with LANCL2 or EGFR. Results indicated that the ampli cation frequencies of EGFR, SEC61G and VOPP1 genes were the top three highest in LANCL2-ampli ed GBM samples, while SEC61G, LANCL2 and VSTM2A were the top three genes co-ampli ed with EGFR ( Fig.1G-H). Kaplan-Meier survival analysis was performed to evaluate the prognostic values of LANCL2 and EGFR ampli cation in GBM patients. Results showed that LANCL2 and EGFR ampli cation was signi cantly associated with reduced OS but not PFS of patients with GBM ( Fig.1I-L).
The relationship between CNVs of LANCL2/EGFR and molecular pathology of GBM samples was analyzed. Wild-type IDH1/2 was mainly found in GBM samples with LANCL2/EGFR gain or ampli cation. Chi-square test found that CNVs of LANCL2 or EGFR were signi cantly correlated with IDH1/2 mutation but not MGMT methylation status ( Fig.2A-D). To observe whether the prognostic value of LANCL2/EGFR ampli cation was affected by IDH1/2 mutation status, Kaplan-Meier survival analysis was performed in IDH1/2-wild-type GBM patients (n=354). Results found that LANCL2 or EGFR ampli cation was not signi cantly correlated with OS and PFS of IDH1/2-wild-type GBM patients (Fig.2E). Similarly, co-ampli cation of LANCL2 and EGFR was also signi cantly associated with reduced OS but not PFS of GBM patients. However, in GBM patients with wild-type IDH1/2, co-ampli cation of LANCL2 and EGFR was not correlated with OS and PFS (Fig.2F). mRNA overexpression of LANCL2 and EGFR were frequent in glioblastoma, but were not associated with the prognosis of glioblastoma patients.
The mRNA expression pro les of LANCL2 and EGFR were investigated in 32 different cancers of TCGA database. In the histograms, the average mRNA expression of LANCL2 and EGFR was organized from lowest to highest priority. Among them, LGG, testicular germ cell carcinoma, GBM and uveal melanoma were the top four tumors with the highest average mRNA expression of LANCL2, while the average mRNA expression of EGFR was highest in GBM, head and neck cancer, clear cell renal cell carcinoma (ccRCC) and LGG (Fig.3A-B). mRNA overexpression of LANCL2 and EGFR was found in 35.63% (57 of 160 cases) and 48.13% (77 of 160 cases) of GBM samples, respectively (Fig3.C). However, the mRNA overexpression frequencies of LANCL2 and EGFR in LGG samples were only around 10%, and little difference was shown in astrocytoma, oligoastrocytoma and oligodendroglioma (Fig.3D). The correlation between mRNA expression and CNV of LANCL2 and EGFR was then analyzed. Results showed that mRNA expression of LANCL2 was signi cantly elevated in GBM samples with LANCL2 ampli cation, compared with GBM samples with diploid or gain of LANCL2 (Fig.3E). Likewise, the correlation was the same in EGFR (Fig.3F). However, mRNA overexpression of LANCL2 or EGFR was not signi cantly associated with OS and PFS of GBM patients (Fig.3G). Moreover, concurrent mRNA overexpression of LANCL2 and EGFR was found in 26.25% (42 of 160 cases) of total GBM samples and 54.55% (42 of 77 cases) of EGFR-overexpressed GBM samples (Fig3.H). In addition, linear regression analysis demonstrated that mRNA expression of LANCL2 and EGFR was positively correlated (p<0.001) (Fig.3I). Nevertheless, concurrent overexpression of LANCL2 and EGFR was also not signi cantly correlated with OS and PFS of GBM patients (Fig.3J-K). Interestingly, mRNA expression levels of EGFR were signi cantly elevated in IDH1/2-wild-type GBM samples, while no obvious change of LANCL2 mRNA expression was found, suggesting a signi cant association between EGFR mRNA expression and IDH1/2 status (Additional le: Figure S1A-B). However, mRNA overexpression of LANCL2 or EGFR was also not signi cantly associated with OS and PFS of IDH1/2-wild-type GBM patients (Additional le: Figure S1C-D).

CNVs, mRNA expression and the prognostic values ofLANCL2 and EGFR in low-grade gliomas
To evaluate the prognostic values of LANCL2 and EGFR CNVs and mRNA expression in LGG, the study of LGG (n=514) was analyzed. The dominant type of LANCL2 and EGFR CNVs in LGG was diploid, and chi-square test revealed that the CNVs of LANCL2 and EGFR were signi cantly correlated (p <0.001) (Fig.4A). The mRNA expression pro les of LANCL2 and EGFR were then investigated, showing that LANCL2 and EGFR mRNA expression was signi cantly increased when these genes ampli ed. However, no difference among LGG samples with shallow deletion, diploid and gain of LANCL2 or EGFR was observed ( Fig.4B-C). In addition, mRNA expression level of LANCL2 were also positively correlated with that of EGFR in LGG samples (Fig.4D). Similar as the GBM samples, CNVs and mRNA expression of LANCL2 or EGFR were signi cantly correlated with IDH1/2 mutation status ( Fig.4E-H). All patients with LANCL2 or EGFR ampli cation contained wild-type IDH1/2. These indicated that LGG patients' survival may also be affected by the IDH1/2 status. Therefore, we evaluated the prognostic values of LANCL2 and EGFR CNVs and mRNA expression in IDH1/2-wild-type LGG patients. Unfortunately, results also showed that ampli cation or mRNA overexpression of LANCL2 and EGFR was not signi cantly correlated with OS and PFS of IDH1/2-wild-type LGG patients (Fig.4I).
Ampli cation and co-ampli cation LANCL2 and EGFR were also frequent in glioblastoma from the tumor banks, and were associated with poor overall survival of glioblastoma patients.
To validate the analysis results of TCGA database, we analyzed the copy numbers of 100 GBM patients' samples from our tumor banks by Taqman Copy Number Assay using uorescent probes targeting LANCL2 and EGFR. The log 2 copy number values larger than 2 was regarded as ampli cation. Results showed that compared with the copy numbers in normal brain tissues and grade I gliomas, the copy numbers of EGFR were signi cantly elevated in GBM, while the copy numbers of LANCL2 had no obvious changes (Fig.5A-B). Interestingly, when the GBM samples were subdivided into newly diagnosed and relapsing tumors, the copy numbers of LANCL2 and EGFR were signi cantly increased only in newly diagnosed GBM (Fig.5E-F). The ampli cation frequencies of LANCL2 and EGFR were 62.00% and 55.00% in 100 GBM patients, respectively (Fig.5C, Table.1). LANCL2 and EGFR co-ampli cation was found in 47.00% of the total GBM samples and 85.45% of GBM samples containing EGFR ampli cation (Fig.5D, Table.1). Pearson's correlation analysis also showed that the copy numbers of LANCL2 and EGFR were positively correlated with each other (Fig.5G). Chi-square tests showed that ampli cation of LANCL2 was not correlated with IDH1 and TERT mutations, and MGMT methylation, whereas ampli cation of EGFR was signi cantly associated with IDH1 and TERT mutations. On the other side, co-ampli cation of LANCL2 and EGFR was not related with TERT mutation and MGMT methylation, but was correlated with IDH1 mutation (Fig.5H). Without regard to the in uence of IDH1 status, LANCL2 or EGFR ampli cation, and their co-ampli cation were signi cantly associated with decreased OS of GBM patients (n=81) (Fig.5I). However, LANCL2 or EGFR ampli cation, and their co-ampli cation were not correlated with OS of IDH1-wild-type GBM patients (n=20) (Fig. 5J).

Protein expression and localization of LanCL2 was independent to EGFR in gliomas
To investigate the protein expression pro les of LanCL2 and EGFR, 72 GBM samples (including newly diagnosed and relapsing tumor samples) and 4 low-grade (grade I) glioma samples from our tumor banks were used. Compared with the grade I glioma control, the log 2 relative protein expression values larger than 2 was regarded as overexpression. We found that overexpression of LanCL2 and EGFR was found in 38.89% and 58.33% of the total GBM samples (Fig.6A, Table.2). The protein expression of EGFR was markedly increased in GBM samples, whereas the expression levels of LanCL2 had no signi cant change (Fig.6B-C). Interestingly, overexpression of LanCL2 was observed in relapsing GBM compared with newly diagnosed GBM (Fig.6D, F). On the other hand, although both the newly diagnosed and relapsing GBM samples displayed elevated EGFR expression compared with the grade I glioma samples, no signi cant change was found between the newly diagnosed and relapsing GBM samples (Fig.6E-F). Pearson's correlation analysis showed that the expression levels of LanCL2 and EGFR were not correlated (Fig.6G). Chi-square tests showed that overexpression of LanCL2 or EGFR was not signi cantly associated with IDH1 or TERT mutations, and MGMT methylation (Fig.6H). No signi cant association was also found between the expression of LanCL2/EGFR and OS of GBM patients (Fig.6I-J). Subsequently, we used tissue microarray to investigate the expression pattern of LanCL2 and EGFR in GBM cells. Results also showed that the expression scores of both LanCL2 and EGFR were markedly increased in GBM tissues, compared with normal brain tissues (Fig.7B, D). LanCL2 was expressed in both the normal brain tissues and gliomas. The protein expression level and intracellular localization of LanCL2 were correlated with the grade of gliomas. The higher the glioma grade, the higher the expression intensity of LanCL2. LanCL2 was mainly found in the nucleus and cytoplasm of high-grade glioma cells (grade III-IV), whereas it was expressed on the nuclear membrane of low-grade (grade I-II) glioma cells (Fig.7A). On the other hand, EGFR was barely expressed in the normal brain tissues and low-grade gliomas, but was overexpressed in the grade III-IV gliomas. It was mainly located in the plasma membrane and cytoplasm of both low-grade and high-grade glioma cells (Fig.7C).

Discussion
Ampli cation and overexpression of EGFR are frequently occurred and extensively studied in GBM. Eley et al found that approximate 34% (40 of 118 cases) of GBM samples contained EGFR ampli cation, and 50% (20 of 40 cases) of EGFR-ampli ed GBM samples displayed LANCL2 co-ampli cation (19). Concomitant ampli cation or copy number gain of two genes is a common phenomenon in cancers, such as MYCN and DDX1 in neuroblastoma, ERBB2/HER2 and TOPOIIα in prostate cancer (27)(28)(29). Similar as LANCL2 and EGFR, these genes are located in the same ampli cation region, which is the driving factor of their co-ampli cation. Since the co-ampli cation of LANCL2 and EGFR was found in GBM in 2002, studies of LANCL2 are barely reported in glioma till now. Only one study using integrative radiogenomic analysis found that the copy number and gene expression of LANCL2 were signi cantly increased in multicentric GBM (27). In this study, we rstly analyzed the TCGA database and found that the ampli cation frequencies of LANCL2 and EGFR in GBM were the highest among 32 different types of tumors, indicating the high speci city of LANCL2 and EGFR ampli cation in GBM. The frequencies of LANCL2 and EGFR ampli cation in 575 GBM patients were approximate 28% and 44% respectively, and 62% of GBM samples with EGFR ampli cation contained LANCL2 co-ampli cation, which were higher than the frequencies reported by Eley's study. In addition, the ampli cation frequencies of LANCL2 and EGFR in GBM were six to nine times higher than those in grade II-III gliomas. These ndings were veri ed in 100 GBM samples of our tumor banks, which showed higher ampli cation and co-ampli cation frequencies of LANCL2 and EGFR than the results analyzed in TCGA database. Moreover, EGFR is the top gene with the highest ampli cation frequency in the TCGA database of Glioblastoma Multiforme (PanCancer Atlas) (data not shown). These suggest that ampli cation of EGFR or coampli cation of LANCL2 and EGFR were potential diagnostic markers for GBM patients.
Survival analysis of TCGA database and our tumor banks showed that ampli cation of LANCL2/EGFR and their co-ampli cation were signi cantly correlated with poor OS, but not PFS of GBM patients. It seems that they were of value for the judgment of the prognosis of histological GBM patients. Subsequently, we found that IDH1/2 mutation, but not MGMT methylation status was correlated with CNVs of LANCL2 and EGFR. However, ampli cation of LANCL2/EGFR and their co-ampli cation were not associated with the OS and PFS of IDH1/2-wild-type GBM patients. Similar studies also nd that EGFR ampli cation is closely associated with wild-type IDH1/2 (28). CDKN2A/B deletion, but not TERT mutation or EGFR ampli cation, was associated with worse OS and PFS of IDH-wild-type GBM patients (29). Our ndings indicated that CNVs of LANCL2 and EGFR were not the independent prognostic biomarkers for IDH1/2-wild-type GBM patients.
Numerous studies showed that both the mRNA and protein overexpression of EGFR, which were highly correlated with EGFR ampli cation, were the signatures and prognostic predictors for GBM patients (30)(31)(32)(33). However, paradox was delineated that the mRNA expression of EGFR, not protein expression, showed a close correlation with EGFR ampli cation (34). Currently, no study has yet reported the expression pro les and prognostic values of LanCL2 in GBM. In this study, we found that the mRNA expression levels of LANCL2 and EGFR were positively correlated in GBM of TCGA database. LANCL2 mRNA expression was signi cantly increased in LANCL2-ampli ed samples, so was EGFR. In our tumor banks, we found that the protein expression of EGFR was elevated in GBM samples, whereas LanCL2 expression did not signi cantly change. mRNA and protein overexpression of LanCL2 or EGFR were not associated with poor OS and PFS of GBM patients, let alone in IDH1/2-wild-type GBM patients. Although the frequencies of LANCL2 and EGFR ampli cation and mRNA overexpression in LGG were low, the relationships between CNVs and mRNA expression of LANCL2 and EGFR in LGG were similar with those in GBM. Since the CNVs and mRNA expression of LANCL2 and EGFR were also associated with IDH1/2 status, the prognostic values of LANCL2 and EGFR in IDH1/2-wild-type LGG patients were evaluated. Similar with GBM patients, CNVs and mRNA expression of LANCL2 and EGFR were not the independent prognostic biomarkers for IDH1/2-wild-type LGG patients.
Although CNVs, mRNA and protein expression of LanCL2 and EGFR were not the independent prognostic biomarkers for IDH1/2wild-type GBM patients, the roles of LanCL2 and EGFR in GBM cells are of importance and can't be ignored. EGFR is extensively studied in GBM. Plenty of studies showed that EGFR and its active mutant EGFRvIII played critical roles in tumorigenesis, proliferation, angiogenesis, and invasion of GBM (35)(36)(37)(38). However, the role of LanCL2 in GBM remains elusive. In this study, LanCL2 protein overexpression was only found in relapsing GBM compared with newly diagnosed GBM, indicating that LanCL2 overexpression may be correlated with GBM recurrence. In contrast, no signi cant difference of EGFR expression was found between newly diagnosed and relapsing GBM. Moreover, no signi cant correlation between LanCL2 and EGFR protein expression was showed. These ndings suggest the expression pattern and role of LanCL2 in GBM are independent to EGFR. A study demonstrated that the N-terminus of LanCL2 protein could be myristoylated and LanCL2 was located in the plasma membrane, juxta-nuclear vesicles, and the nucleus (39). Here, our immunohistochemical result found that the expression and localization of LanCL2 was correlated with the grade of gliomas. The major localization of LanCL2 in GBM cells was in the nucleus and cytoplasm, whereas it was mainly expressed on the nuclear membrane of LGG cells. Another study showed that LanCL2 is a non-transmembrane G protein-coupled receptor, and its nuclear enrichment was induced by ABA or its unmyristoylation to activate ABA signaling (40). Therefore, we speculated that LanCL2 maintains inactive in the cytoplasm of LGG cells, while LanCL2 transforms to the active form in GBM cells and then translocates into the nucleus.

Conclusion
In summary, this study showed that ampli cation and mRNA overexpression of LANCL2 and EGFR, and their co-ampli cation and coexpression frequently occurred in GBM patients, compared with patients with LGG. LANCL2 or EGFR ampli cation, and their coampli cation were signi cantly associated with reduced OS of GBM patients, rather than IDH1/2-wild-type GBM patients. However, mRNA and protein expression of LanCL2 and EGFR were not correlated with the prognosis of GBM patients. LanCL2 protein was overexpressed in relapsing GBM samples rather than newly diagnosed GBM, whereas overexpression of EGFR was found in both newly diagnosed and relapsing GBM samples. Moreover, the protein expression and intracellular localization of LanCL2 were correlated with the grade of gliomas. LanCL2 was mainly found in the nucleus and cytoplasm of high-grade gliomas (grade III-IV), whereas it was expressed on the nuclear membrane of low-grade (grade I-II) gliomas. Taken together, CNVs of LANCL2 and EGFR were the independent diagnostic and prognostic biomarkers for histological GBM patients, but not for IDH1/2-wild-type GBM patients. The protein expression pattern and role of LanCL2 in GBM were independent to EGFR. LanCL2 expression was correlated with GBM recurrence, and its activation may trigger its translocation into the nucleus.

Declarations
Ethics approval and consent to participate This study was approved by the Research Ethics Committee of Shenzhen Second People's Hospital and Sun Yat-sen University Cancer Center. All patients were given written informed consent.

Consent for publication
This study has been approved by all authors for publication.
Availability of data and material Tables   Tables   Table 1. Amplification and co-amplification of LANCL2 and EGFR in GBM samples of tumor banks