Purpose: Despite significant advances in the understanding of the biology, the prognosis of glioblastoma (GBM) remains dismal. The objective was to carry out whole-exome sequencing (WES) of Indian glioma and integrate with that of TCGA to find clinically relevant mutated pathways.

Experimental Design: WES of different astrocytoma samples (n = 42; Indian cohort) was carried out and compared with that of TCGA cohort. An integrated analysis of mutated genes from Indian and TCGA cohorts was carried out to identify survival association of pathways with genetic alterations. Patient-derived glioma stem-like cells, glioma cell lines, and mouse xenograft models were used for functional characterization of calcitonin receptor (CALCR) and establish it as a therapeutic target.

Results: A similar mutation spectrum between the Indian cohort and TCGA cohort was demonstrated. An integrated analysis identified GBMs with defective “neuroactive ligand–receptor interaction” pathway (n = 23; 9.54%) that have significantly poor prognosis (P < 0.0001). Furthermore, GBMs with mutated calcitonin receptor (CALCR) or reduced transcript levels predicted poor prognosis. Exogenously added calcitonin (CT) inhibited various properties of glioma cells and pro-oncogenic signaling pathways in a CALCR-dependent manner. Patient-derived mutations in CALCR abolished these functions with the degree of loss of function negatively correlating with patient survival. WT CALCR, but not the mutant versions, inhibited Ras-mediated transformation of immortalized astrocytes in vitro. Furthermore, calcitonin inhibited patient-derived neurosphere growth and in vivo glioma tumor growth in a mouse model.

Conclusions: We demonstrate CT–CALCR signaling axis is an important tumor suppressor pathway in glioma and establish CALCR as a novel therapeutic target for GBM. Clin Cancer Res; 24(6); 1448–58. ©2017 AACR.

This article is featured in Highlights of This Issue, p. 1241

Translational Relevance

The advancement in effective therapeutics for glioblastoma (GBM) has been minimal, and the median survival still remains at 15 to 17 months only. Hence, there is a need to elucidate novel altered molecules for effective therapeutic possibilities. In this study, we explore the mutation spectrum of patients with GBM to unearth novel pathways altered by mutation that predict survival in patients, which revealed neuroactive ligand–receptor interaction pathway to be the most significant. Calcitonin receptor (CALCR), the most mutated gene in this pathway, was studied further, which revealed this receptor to be tumor suppressive in nature, and activation of it by its ligand, calcitonin, led to a decrease in tumorigenic properties of glioma cells. Moreover, CALCR inhibited transformation of astrocytes in vitro, glioma stem-like cell growth, and in vivo glioma tumor growth in mice. Hence, GBMs with wild-type CALCR, calcitonin, which is prescribed for postmenopausal osteoporosis, could be considered as a treatment option.

Glioblastoma (GBM) is the most common and highly aggressive adult primary brain tumor. GBMs show a significant amount of proliferation, invasion, angiogenesis, necrosis, and are also treatment refractory. Each GBM tumor carries an amalgamation of genetic alterations that determine cancer prognosis and response to therapy. Intensive studies on candidate genes show various genetic alterations typical to GBM, for example, TP53 mutation and loss, EGFR amplification and mutation, PTEN mutation and loss, etc. (1). In recent times, two independent groups have carried out whole-exome (WES) and RNA sequencing analysis of GBM tissue samples from The Cancer Genome Atlas (TCGA) and have found out various novel genetic alterations that may play an important role in GBM pathogenesis (2, 3). From these studies, it is evident that three pathways, receptor tyrosine kinase (RTK), TP53, and RB, are significantly altered in GBM tumor by mutations or copy number alterations. However, even with the increase in our understanding of the tumor, advancement in therapeutics is minimal, and the median survival still remains at 15 to 17 months only (4). Hence, we need to elucidate novel altered molecules and pathways in GBM such that more effective therapeutic possibilities can be explored.

With this objective, we propose to understand the genetic spectrum of patients with astrocytoma through WES to find novel targetable pathways in GBM. In this study, we performed integrated analysis of mutated genes from our Indian patient cohort as well as TCGA cohort to find out mutated pathways that predict survival in patients with GMB. This analysis revealed neuroactive ligand–receptor interaction pathway to be the most significant pathway that predicts poor survival. We characterized calcitonin receptor (CALCR), a member of this pathway, and demonstrated the tumor suppressor role of this gene in GBM. Furthermore, we found that mutational inactivation of CALCR abrogated this tumor-suppressive function of the gene, making glioma cells more aggressive.

Collection of patient tumor sample and blood

Tumor and matched blood samples from patients were collected from All India Institute of Medical Sciences (AIIMS, Delhi, India). Tumor samples were resected in the neurosurgical room, and a portion was snap-frozen in liquid nitrogen, and stored at −80°C. In addition, blood samples collected from each patient were snap-frozen and stored at −80°C. The remaining portion of the tissue was fixed in 10% buffered neutral formalin, processed for paraffin sections, and was used for histopathology and IHC. For RNA isolation, fresh tissue was snap frozen in RNAlater. Normal brain tissue was collected within 3 to 4 hours after death from patients who died from nonhead injury/non-CNS disorders. The study has been ratified by the ethics committee of the AIIMS and Indian Institute of Science (IISc), and patient consent was obtained as per the Institute Ethical Committee guidelines.

Cell lines, plasmid constructs, and siRNA

The GBM cell lines, LN229, LN18, T98G, U251, U87, U343, and U373, were obtained from Sigma Aldrich. The immortalized human astrocyte cell line IHA (NHA-hTERT-E6/E7) was obtained from Dr. Russell Pieper's laboratory, University of California, San Francisco (San Francisco, CA). The immortalized human astrocyte cell line SVG was obtained from Dr. Abhijit Guha's laboratory, University of Toronto (Toronto, Canada). The immortalized mouse astrocytic cell line, IMA2.1, was a kind gift from Dr. Stefan Schildknecht, University of Konstanz, (Konstanz, Germany). All the cells were cultured in DMEM supplemented with 10% FBS. The cells were grown at 37°C in 5% CO2. Dr. Hiroaki Wakimoto, Brain Tumor Research Center, Massachusetts General Hospital (Boston, MA), kindly provided us with the glioma stem-like cell (GSC) lines MGG4, MGG6, MGG8, and MGG23. The GSC, 1035 (or SK1035), was a kind gift from Dr. Santosh Kesari, Pacific Neuroscience Institute (Santa Monica, CA). The GSCs were cultured in Neurobasal medium supplemented with 0.5 mmol/L l-glutamine, 20 ng/μL EGF, 20 ng/μL FGF, 40 μg/mL heparin, and B-27 and N-2 supplements. The cells were grown at 37°C in 5% CO2. The plasmid pPM-C-HA-CALCR was obtained from Abmgood (catalog No. PV007283). The construct used contained transcript variant 2 of CALCR (transcript ID: ENST000009994441), and one of the eight mutations detected (chr7:93091387) was not present, and hence, its effect was not tested. Mutated CALCR was generated by site-directed mutagenesis (SDM) using QuikChange Multi Site-Directed Mutagenesis Kit (catalog No. 200515). The vector control (VC) plasmid was created by expelling out the CALCR open reading frame using restriction enzymes NheI and XhoI. The overhangs were end filled and ligated to generate the VC plasmid. The shRNA plasmid constructs against RAMP1 (TRCN0000273872, TRCN0000273874, and TRCN0000273814) were obtained as a kind gift from Dr. Subba Rao and Dr. Saini from MISSION shRNA Library (Sigma Aldrich). RasV12 (KRas) construct was a kind gift from Dr. Annapoorni Rangarajan (IISc). Cells were transfected with Lipofectamine 2000. For selection of stable clones of CALCR and shRNA plasmid constructs, G418 (500–1,000 μg/mL) or puromycin (1–2 μg/mL), respectively, were added in the complete medium and selected for 1 to 2 weeks. ON-TARGETplus Human CALCR (799) siRNA – SMARTpool (5 nmol/L) was obtained from Dharmacon (catalog No. L-003635-00-005), and for each experiment, 100 nmol/L siRNA was transfected using DharmaFECT transfection reagent.

Other methods are provided in Supplementary Methods.

Integrated analysis of Indian and TCGA GBM exome identifies pathways with prognostic value

We have carried out WES of 42 astrocytoma tissues of Indian origin (10 grade 2, 13 grade 3, and 19 grade 4/GBM) and matched peripheral blood samples (Supplementary Table S1). An average of 40 ± 17× coverage was obtained across all samples (for details, please see Supplementary Information). The matched tumor blood sequences were analyzed using MuTect tool (5) and Indelocator (6) to identify tumor-specific nonsynonymous single-nucleotide variants and insertions/deletions (indels), respectively (Supplementary Tables S2–S4). The genetic spectrum of our patient cohort (Indian cohort) was explored through the analysis of top mutated genes, chromatin modifiers, and DNA repair–related genes that are known to be mutated in GBM as per previous reports (Supplementary Fig. S1A–S1C; ref. 3). As observed before, TP53 was found to be highly mutated across all grades of astrocytoma with higher percentage of mutation (65%) in lower grade samples/LGGs (grade 2 and 3) compared with GBM (32%; ref. 7). Other top mutated genes in GBM as per TCGA study, such as PTEN, PIK3CA, and NF1, were also found to be mutated in GBM samples in an Indian cohort (3). We also found IDH1 and ATRX to be mutated typically in the LGGs as shown before (8, 9). To compare the mutation spectrum of the Indian cohort with that of TCGA, the three signaling pathways that were found to be significantly altered in patients with GBM, the RTK pathway, the TP53 pathway, and the RB pathway (3), were considered. The analysis revealed that the Indian patient cohort behaves largely similar to TCGA cohort (Supplementary Fig. S1D).

To identify the defective pathways with genetic alterations in GBM that predict survival, an integrated analysis involving GBM-specific mutated genes derived from Indian and TCGA exome data was carried out (for details, please see Supplementary Fig. S2A; Supplementary information). Of the several pathways carrying genetic alterations found in GBM, Cox regression analysis revealed that 62 pathways predict survival in GBM significantly (Supplementary Fig. S2A; Supplementary Table S5). As expected, this list contained many pathways that were previously implicated in GBM survival like PI3K–Akt signaling, Ras signaling, mTOR signaling, insulin signaling, focal adhesion, and regulation of actin cytoskeleton (Supplementary Table S5). However, the top five pathways that predicted survival with very high significance were not reported previously for their association with GBM survival (Fig. 1A). The “neuroactive ligand–receptor interaction pathway” is particularly notable as GBMs with defect in this pathway (mutation in at least one of the genes) have a poor median survival (3.13 months; P < 0.0001; Fig. 1B) and it is highly mutated in both TCGA and Indian cohorts (Supplementary Tables S6 and S7, respectively). Multivariate Cox regression analysis with age, G-CIMP methylator phenotype, MGMT promoter methylation, and IDH1 mutation status revealed that the neuroactive ligand–receptor interaction pathway is an independent predictor of survival in GBM (Fig. 1C).

Figure 1.

Clinical significance of neuroactive ligand–receptor interaction pathway and CALCR in GBM. A, Top significant mutated pathways (with ≥5% genes mutated and ≥5% samples mutated) that predict poor survival in GBM as per univariate and Kaplan–Meier survival analyses. The log (base 10) of the P value is plotted on the x-axis. The number on the right of each bar represents the percentage of genes mutated in the corresponding pathway. B, Kaplan–Meier survival analysis of GBMs stratified by the genetic status of neuroactive ligand–receptor interaction pathway. “Altered” refers to patients having mutation in one or more of the genes in the pathway. C, Multivariate Cox regression analysis of neuroactive ligand–receptor interaction pathway (denoted by $) with age, G-CIMP methylation, MGMT promoter methylation, and IDH1 mutation status. D, Kaplan–Meier survival analysis of GBMs stratified by CALCR mutation status. Note: Although there were 8 patients with mutation found in the analysis, only 7 patients had survival information. E, RNA levels of CALCR from microarray data in TCGA Agilent, TCGA Affymetrix, and GSE7696 datasets. F, Real-time qPCR analysis of RNA levels of CALCR in the Indian patient cohort (control = 5; GBM = 20). Kaplan–Meier survival analysis of GBM samples expressing high versus low CALCR RNA levels in –TCGA Agilent data (G), TCGA Affymetrix data (H), GSE7696 data (I), and Indian patient cohort (J). K, RNA levels of CALCR in glioma-derived cell lines compared with immortalized astrocytes. L, Protein levels of CALCR in glioma-derived cell lines versus immortalized astrocytes. The P values for panels A–K are represented by *, ** and ***, which denote P values of <0.05, 0.01, and 0.001, respectively.

Figure 1.

Clinical significance of neuroactive ligand–receptor interaction pathway and CALCR in GBM. A, Top significant mutated pathways (with ≥5% genes mutated and ≥5% samples mutated) that predict poor survival in GBM as per univariate and Kaplan–Meier survival analyses. The log (base 10) of the P value is plotted on the x-axis. The number on the right of each bar represents the percentage of genes mutated in the corresponding pathway. B, Kaplan–Meier survival analysis of GBMs stratified by the genetic status of neuroactive ligand–receptor interaction pathway. “Altered” refers to patients having mutation in one or more of the genes in the pathway. C, Multivariate Cox regression analysis of neuroactive ligand–receptor interaction pathway (denoted by $) with age, G-CIMP methylation, MGMT promoter methylation, and IDH1 mutation status. D, Kaplan–Meier survival analysis of GBMs stratified by CALCR mutation status. Note: Although there were 8 patients with mutation found in the analysis, only 7 patients had survival information. E, RNA levels of CALCR from microarray data in TCGA Agilent, TCGA Affymetrix, and GSE7696 datasets. F, Real-time qPCR analysis of RNA levels of CALCR in the Indian patient cohort (control = 5; GBM = 20). Kaplan–Meier survival analysis of GBM samples expressing high versus low CALCR RNA levels in –TCGA Agilent data (G), TCGA Affymetrix data (H), GSE7696 data (I), and Indian patient cohort (J). K, RNA levels of CALCR in glioma-derived cell lines compared with immortalized astrocytes. L, Protein levels of CALCR in glioma-derived cell lines versus immortalized astrocytes. The P values for panels A–K are represented by *, ** and ***, which denote P values of <0.05, 0.01, and 0.001, respectively.

Close modal

Mutation or reduced transcript levels of CALCR predicts poor survival in GBM

To gain biological insight into the prognostic association of neuroactive ligand–receptor pathway, we chose CALCR for further studies, as it was found to be the top mutated gene (Supplementary Table S8). The location and nature of the eight tumor-derived mutations in CALCR is shown (Supplementary Fig. S2B). Univariate Cox regression analysis revealed that mutation in CALCR predicted poor survival in patients with GBM (Supplementary Fig. S2C). Furthermore, Kaplan–Meier survival analysis revealed that GBMs with mutation in CALCR survived lesser (Fig. 1D; median survival = 4.83 months). Multivariate Cox regression analysis with other markers, such as age, IDH1 mutation, and G-CIMP status, revealed CALCR mutation to be an independent prognosticator. On comparison with MGMT promoter methylation status, CALCR mutation showed a trend toward predicting poor survival with near significant P value (P = 0.075). However, CALCR mutation status lost its significance in multivariate analysis involving all prognostic markers (Supplementary Fig. S2C), which could be due to the fact that there were fewer number of patients with CALCR mutations. We evaluated the distribution of CALCR-mutated GBM samples with respect to highly mutated genes in GBM (TP53, PTEN, EGFR, PDGFRA, NF1, and IDH1) and GBM subtypes (G-CIMP, MGMT, classical, mesenchymal, neural, and proneural). Although there was enrichment of CALCR mutation in patients with wild-type (WT) NF1, PDGFRA, EGFR, and IDH1, odds ratio testing for mutual exclusivity showed no significant correlation/mutual exclusivity between CALCR mutation and any of the highly mutated genes or the different GBM subtypes (Supplementary Table S9).

Additional investigation revealed CALCR transcript levels are downregulated in GBMs derived from TCGA, GSE7696, and Indian cohorts (Fig. 1E and F). Kaplan–Meier survival analysis revealed that GBMs with low levels of CALCR transcript have a poor survival in TCGA, GSE7696, and Indian cohort (Fig. 1G–J). Furthermore, we found that glioma-derived cell lines have reduced CALCR transcript and protein levels compared with immortalized astrocytes (Fig. 1K and L). From these results, we conclude that CALCR may function as a tumor suppressor gene in GBM and either mutational inactivation or reduced transcript levels lead to a highly aggressive GBM with poor survival.

CALCR is a tumor suppressor in GBM

CALCR is a G-protein–coupled receptor (GPCR) with an N-terminal ligand-binding domain, a seven-pass transmembrane domain, and a C-terminal domain (10). When its ligand, calcitonin (CT), binds to the N-terminal domain, the receptor undergoes conformational change that leads to activation of G-protein α present at the C-terminal cytosolic side, which leads to the regulation of various downstream signaling pathways, thus affecting a variety of functions (11). To test the function of CT–CALCR signaling in glioma, we used various glioma-derived established cell lines, wherein the WT status of CALCR was confirmed (Supplementary Fig. S2D; refs. 12, 13). We first used LN229 glioma cells, which express relatively high levels of CALCR (Fig. 1K and L). The effects of exogenously added calcitonin on various cancer cell–associated properties of empty vector (VC)-stable and CALCR-stable LN229 cells were tested (Fig. 2A). Exogenous addition of calcitonin inhibited colony formation, proliferation, migration, invasion, and anchorage-independent growth of LN229/VC stable cells very efficiently (Fig. 2B–F). LN229/CALCR stable cells also showed significant reduction in these properties compared with LN229/VC stable cells (Fig. 2B–F). Moreover, exogenously added calcitonin inhibited all the above properties of LN229/CALCR stable cells even more efficiently than LN229/VC stable cells (Fig. 2B–F). Similar results were obtained in CALCR-expressing cell lines T98G and U87 (Supplementary Fig. S3A–S3C and S3D–S3F, respectively). These results suggest that calcitonin inhibits tumorigenic properties of glioma cells in a CALCR-dependent manner. Furthermore, to test the importance of CALCR in calcitonin-mediated functions in glioma cells, the effect of calcitonin in CALCR-silenced LN229, T98G, and U87 cells was evaluated. The inhibition of proliferation and migration by calcitonin seen in nontargeting siRNA transfected cells (siNT) was significantly abrogated in CALCR siRNA transfected cells (siCALCR; Fig. 2G–I; Supplementary Fig. S4A–S4D).

Figure 2.

Effect of calcitonin on various properties of CALCR-high LN229 glioma cells. A, Overexpression of CALCR in LN229 cell line verified by checking RNA and protein levels. B–F, Colony suppression assay (B), proliferation assay (C), migration assay (D), invasion assay (E), and soft agar assay (F) for CALCR in vector control (VC)/CALCR (wild-type) conditions in presence or absence of calcitonin. All assays are quantified and the bar plots are provided at the right side of the representative images. For each experiment, the quantification at the end of the assay is used for the bar plot. The value of the control condition (VC + BSA) was normalized to 100%, and the values for the other conditions (VC + CT, CALCR + BSA, and CALCR + CT) were calculated with respect to normalized VC + BSA. G, Silencing of CALCR in LN229 cell line verified by checking RNA and protein levels. H and I, Proliferation assay (H) and migration assay (I) in CALCR silenced conditions in the presence or absence of calcitonin. All assays are quantified and the bar plots are provided to the right side of the representative images. For each experiment, the quantification at the end of the assay is used for the bar plot. The value of the control condition (siNT + BSA) was normalized to 100%, and the values for the other conditions (siNT + CT, siCALCR + BSA, and siCALCR + CT) were calculated with respect to normalized siNT + BSA. The P values for all panels are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively, and NS refers to nonsignificant P value (≥0.05).

Figure 2.

Effect of calcitonin on various properties of CALCR-high LN229 glioma cells. A, Overexpression of CALCR in LN229 cell line verified by checking RNA and protein levels. B–F, Colony suppression assay (B), proliferation assay (C), migration assay (D), invasion assay (E), and soft agar assay (F) for CALCR in vector control (VC)/CALCR (wild-type) conditions in presence or absence of calcitonin. All assays are quantified and the bar plots are provided at the right side of the representative images. For each experiment, the quantification at the end of the assay is used for the bar plot. The value of the control condition (VC + BSA) was normalized to 100%, and the values for the other conditions (VC + CT, CALCR + BSA, and CALCR + CT) were calculated with respect to normalized VC + BSA. G, Silencing of CALCR in LN229 cell line verified by checking RNA and protein levels. H and I, Proliferation assay (H) and migration assay (I) in CALCR silenced conditions in the presence or absence of calcitonin. All assays are quantified and the bar plots are provided to the right side of the representative images. For each experiment, the quantification at the end of the assay is used for the bar plot. The value of the control condition (siNT + BSA) was normalized to 100%, and the values for the other conditions (siNT + CT, siCALCR + BSA, and siCALCR + CT) were calculated with respect to normalized siNT + BSA. The P values for all panels are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively, and NS refers to nonsignificant P value (≥0.05).

Close modal

To study the importance of CALCR for calcitonin functions, the effect of calcitonin on CALCR-low cell lines (U343 and U373; Fig. 1K and L) was also tested. Calcitonin showed either no or very minimal effect on colony formation, proliferation, migration, invasion, and anchorage-independent growth of U343/VC stable cells (Supplementary Fig. S5A–S5F), which could be due to low expression of CALCR in these cells. U343/CALCR stable cells showed significant reduction in the above properties compared with U343/VC stable cells (Supplementary Fig. S5B–S5F), suggesting the fact that mere overexpression of the receptor is able to inhibit these functions. Furthermore, exogenously added calcitonin inhibited the above properties of U343/CALCR stable cells very efficiently compared with that of U343/VC cells (Supplementary Fig. S5B–S5F). Similar results were obtained in U373 cell line, which also expresses low levels of CALCR (Supplementary Fig. S6A–S6C). These results demonstrate that calcitonin inhibits various properties of cancer cells in a CALCR-dependent manner.

To address the signaling downstream of CT–CALCR cascade, we investigated the status of AKT, ERK, and JNK signaling, which are known to regulate various properties of cancer cells (14). The exogenously added calcitonin efficiently reduced the pAKT, pERK, and pJNK levels in U343/CALCR stable cells but not in U343/VC cells. This reduction in pAKT, pERK, and pJNK levels was seen in both LN229/VC stable and LN229/CALCR stable cells (Fig. 3A). Furthermore, silencing CALCR in LN229 cells efficiently reversed the inhibition by calcitonin on AKT, ERK, and JNK signaling pathways (Fig. 3A). Receptor activity–modifying proteins (RAMP1, RAMP2, and RAMP3) are single-transmembrane proteins that induce trafficking of CALCR, thereby regulating CT–CALCR signaling cascade (15). We found that RAMP1 transcript levels, but not RAMP2 and RAMP3, are higher in LN229 cells where mere addition of calcitonin inhibits various cancer cell properties (Fig. 3B). To know the role of RAMP1 in CT–CALCR signaling cascade, we tested the ability of calcitonin to inhibit colony formation and proliferation of glioma cells in RAMP1-silenced condition (shRAMP1) in the absence or presence of exogenously expressed CALCR (Fig. 3C). Calcitonin inhibited both functions equally efficiently in shNT and shRAMP1 conditions of LN229/VC, and this inhibition was enhanced in LN229/CALCR cells, although similar in shNT versus shRAMP1 condition (Fig. 3D and E; Supplementary Fig. S6D). Collectively from these results, we conclude that CT–CALCR signaling acts as a tumor suppressor pathway as it inhibits various properties of cancer cells by inhibiting AKT, ERK, and JNK pathways and RAMPs are not required for CT–CALCR signaling cascade in glioma.

Figure 3.

Regulation of ERK, JNK, and AKT signaling by calcitonin–CALCR signaling axis and the role of RAMP coreceptors in CALCR-mediated tumor suppressor functions in glioma cells. A, Western blot analysis of phosphorylation of downstream signaling molecules ERK, JNK, and AKT in CALCR overexpression conditions (in U343 and LN229 cells) and CALCR-silenced condition (in LN229 cells) in the presence or absence of calcitonin. The quantification for each phospho-protein is provided at the bottom of the corresponding blot. The total protein for each lane was normalized to the actin levels, and subsequently, the phospho-protein was normalized to the normalized total. For CALCR overexpression experiments, the value for VC + BSA for each phospho-protein was normalized to 1, and the other conditions (VC + CT, CALCR + BSA, and CALCR + CT) were calculated with respect to the normalized VC + BSA. For the CALCR silencing experiment, the value for siNT + BSA for each phospho-protein was normalized to 1, and the other conditions (siNT + CT, siCALCR + BSA, and siCALCR + CT) were calculated with respect to the normalized siNT + BSA. B, Transcript levels of RAMP1, 2 and 3 in LN229 cells compared with control brain tissue. C, Knockdown of RAMP1 in LN229 cells shown by real-time qPCR and Western blotting. D, Colony suppression assay in LN229/VC and LN229/CALCR stable cells transfected with shRAMP1. E, Proliferation assay in LN229/VC and LN229/CALCR stable cells transfected with shRAMP1. The quantification for the 6th day of proliferation as given in the Supplementary Fig. S6D is given in the bar plot, wherein the value for LN229/VC/shNT + BSA condition was normalized to 100%, and the rest of the conditions were plotted accordingly. NS, nonsignificant P value (≥0.05).

Figure 3.

Regulation of ERK, JNK, and AKT signaling by calcitonin–CALCR signaling axis and the role of RAMP coreceptors in CALCR-mediated tumor suppressor functions in glioma cells. A, Western blot analysis of phosphorylation of downstream signaling molecules ERK, JNK, and AKT in CALCR overexpression conditions (in U343 and LN229 cells) and CALCR-silenced condition (in LN229 cells) in the presence or absence of calcitonin. The quantification for each phospho-protein is provided at the bottom of the corresponding blot. The total protein for each lane was normalized to the actin levels, and subsequently, the phospho-protein was normalized to the normalized total. For CALCR overexpression experiments, the value for VC + BSA for each phospho-protein was normalized to 1, and the other conditions (VC + CT, CALCR + BSA, and CALCR + CT) were calculated with respect to the normalized VC + BSA. For the CALCR silencing experiment, the value for siNT + BSA for each phospho-protein was normalized to 1, and the other conditions (siNT + CT, siCALCR + BSA, and siCALCR + CT) were calculated with respect to the normalized siNT + BSA. B, Transcript levels of RAMP1, 2 and 3 in LN229 cells compared with control brain tissue. C, Knockdown of RAMP1 in LN229 cells shown by real-time qPCR and Western blotting. D, Colony suppression assay in LN229/VC and LN229/CALCR stable cells transfected with shRAMP1. E, Proliferation assay in LN229/VC and LN229/CALCR stable cells transfected with shRAMP1. The quantification for the 6th day of proliferation as given in the Supplementary Fig. S6D is given in the bar plot, wherein the value for LN229/VC/shNT + BSA condition was normalized to 100%, and the rest of the conditions were plotted accordingly. NS, nonsignificant P value (≥0.05).

Close modal

Patient-derived mutations abolish tumor-suppressive functions of CALCR

Because our study finds GBMs with CALCR mutations to be more aggressive with lesser median survival, we hypothesized that mutations might have abolished the tumor-suppressive functions of CALCR. To test this possibility, each of the seven mutations were introduced into CALCR through site-directed mutagenesis (Fig. 4A) and the ability of exogenously expressed CALCR mutants by themselves as well as in conjunction with calcitonin on various cancer cell properties in U343 and LN229 glioma cells was evaluated. We introduced empty vector (VC), WT, and each of the mutant CALCR constructs into U343 and LN229 cells and assessed the effect on glioma cell proliferation, colony formation, migration, invasion, and anchorage-independent growth in the presence and absence of calcitonin. Although the WT CALCR overexpression (please see “BSA” condition) inhibited glioma cell proliferation, colony formation, migration, invasion, and anchorage-independent growth efficiently, CALCR mutants failed to inhibit these functions significantly, although to varying extents in both U343 and LN229 glioma cells (Fig. 4B–F; Supplementary Figs. S7A–S7E and S8A–S8H). Similarly, exogenously added calcitonin by itself (in LN229 cells) and in the presence of WT CALCR (in LN229 and U343 cells) inhibited all five properties very efficiently. However, the inhibition by calcitonin was significantly abrogated in LN229 and U343 CALCR-mutant stable clones although to varying extents (Fig. 4B–F; Supplementary Figs. S7A–S7E and S8A–S8H). The loss of function (LOF) was profound in CALCR mutants, A51T, V250M, A307V, and R420C (“severe mutants”), whereas it was minimal in CALCR mutants, R45Q, P100L, and R404C (“mild mutants”). To assess the effect of varying LOF by different CALCR mutants on tumor aggressiveness, we generated LOF score by combining the level of LOF for five different properties in each cell line, and this was finally correlated with GBM patient survival (for details, please see Supplementary Information). The analysis revealed that the severe mutants had higher LOF score, whereas the mild mutants had a lower LOF score (Supplementary Table S10). In comparison with patient survival, there was significant negative correlation between LOF score and survival (Fig. 4G; Supplementary Fig. S7F).

Figure 4.

Effect of mutation of CALCR on U343 glioma cell properties. A, Representation of the structure of CALCR. Star denotes the approximate positions of the mutations in the protein domains. B–F, Quantification of proliferation assay (B), colony suppression assay (C), migration assay (D), invasion assay (E), and soft agar assay (F) for CALCR overexpression in U343 cells transfected with WT CALCR and the seven mutated CALCRs [the eighth mutation Y186C was not included in this analysis as this position is absent in the transcript variant 2 of CALCR (transcript ID: ENST000009994441) used in this study)]. The value of the control condition (VC + BSA) was normalized to 100%, and the values for the other conditions were calculated with respect to normalized VC + BSA. Comparison of mutant CALCR phenotypes with that of WT CALCR was done using one-way ANOVA. The P values are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively. G, Correlation between LOF score from the results in U343 cells versus patient survival. This analysis did not include R45Q, as the patient harboring this mutation did not have survival information. Please see Supplementary Information for LOF score calculation.

Figure 4.

Effect of mutation of CALCR on U343 glioma cell properties. A, Representation of the structure of CALCR. Star denotes the approximate positions of the mutations in the protein domains. B–F, Quantification of proliferation assay (B), colony suppression assay (C), migration assay (D), invasion assay (E), and soft agar assay (F) for CALCR overexpression in U343 cells transfected with WT CALCR and the seven mutated CALCRs [the eighth mutation Y186C was not included in this analysis as this position is absent in the transcript variant 2 of CALCR (transcript ID: ENST000009994441) used in this study)]. The value of the control condition (VC + BSA) was normalized to 100%, and the values for the other conditions were calculated with respect to normalized VC + BSA. Comparison of mutant CALCR phenotypes with that of WT CALCR was done using one-way ANOVA. The P values are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively. G, Correlation between LOF score from the results in U343 cells versus patient survival. This analysis did not include R45Q, as the patient harboring this mutation did not have survival information. Please see Supplementary Information for LOF score calculation.

Close modal

To address whether the LOF shown by CALCR mutants is due to their inability to inhibit downstream signaling pathways unlike WT CALCR, we assessed the effect of CALCR mutations on downstream AKT, ERK, and JNK signaling pathways in U343 cell line (Supplementary Fig. S9). Addition of calcitonin to U343/VC cells led to a minimal reduction of pAKT, pERK, and pJNK levels. However, overexpression of WT CALCR led to a significant reduction in phosphorylation of the above signaling molecules, which was further reduced when calcitonin was added to U343/CALCR cells (Supplementary Fig. S9, compare lanes 3 and 4 with 1 and 2). However, the mutant CALCRs both in the absence and presence of exogenously added calcitonin inhibited these three signaling pathways much less efficiently with the inhibition being negligible in the severe mutants (Supplementary Fig. S9, compare lanes 5 to 18 with 3 and 4).

From the above results, it is clear that the mutant forms of CALCR exhibit varied LOF phenotypes. This could be explained by the fact that different mutations may have a varying impact in altering the structure of CALCR. The crystal structure of the N-terminal domain, along with calcitonin ligand bound to it has been crystalized, and the structure has been determined (PDB ID: 5II0; ref. 16). We used three servers, mCSM, SDM, and DUET, to quantify the influence of R45Q, A51T, and P100L mutations (located in the N-terminal domain) in disruption of the protein stability of CT–CALCR complex (measured by the change in Gibbs free energy ΔΔG between the wild-type and mutant structures). All three tools predicted A51T mutation, one of the severe mutants, to be destabilizing. However, the above analysis predicted R45Q and P100L (both mild mutants) to be neutral in nature (Supplementary Fig. S10). Collectively from the above results, we can conclude that mutation in CALCR abrogates CT–CALCR tumor suppressor signaling axis, thus contributing significantly to the more aggressive phenotype seen in GBMs.

Effect of CT–CALCR tumor suppressor axis on GSCs

Next, we evaluated the function of CT–CALCR axis in patient-derived GSCs as well as glioma cell line–derived GSCs. According to the RNA and protein levels, we divided the GSCs into CALCR-high (1035, MGG6, MGG8, MGG23, and LN229) and CALCR-low (MGG4, T98G, U87, U343, and U373) cell lines (Figs. 5A and 1L; Supplementary Fig. S11A). The WT nature of CALCR in these lines was confirmed except MGG6, MGG23, and 1035 (Supplementary Fig. S2D). Calcitonin was exogenously added to each of the GSCs, and the effect on neurosphere growth was observed by sphere formation and limiting dilution assays. Although calcitonin inhibited the sphere growth in most of the GSCs, the percentage inhibition of sphere growth was found to be significantly higher in CALCR-high GSCs compared with the CALCR-low GSCs (Fig. 5B–E). The glioma reprogramming factors, SOX2, OLIG2, SALL2, and POU3F2, as identified by Suva and colleagues, were found to be downregulated in calcitonin condition, and this was more prominent in CALCR-high GSCs (Fig. 5F; ref. 17). Furthermore, addition of calcitonin to CALCR-high GSCs (MGG8, 1035, and LN229) led to a significant inhibition of pERK and pAKT levels, which were less pronounced in the CALCR-low GSCs (U343 and MGG4; Fig. 5G). Next, we evaluated the effect of CT–CALCR axis on the growth of xenograft-derived neurospheres (xGSC). Addition of calcitonin to xGSCs significantly reduced the number as well as size of the neurospheres (Supplementary Fig. S11B and S11C). In addition, the effect of CT–CALCR pathway was tested on reprogramming of established glioma cell lines to form GSCs. We observed that addition of calcitonin during reprogramming inhibited neurosphere formation significantly in CALCR-high cell line (LN229) unlike CALCR-low cell lines (T98G, U87, U343, and U373; Supplementary Fig. S12A–S12E). These results demonstrate that CT–CALCR axis potently inhibits the growth of GSCs as well as reprogramming of glioma cell lines in a CALCR-dependent manner.

Figure 5.

Effect of CT–CALCR tumor suppressor axis on GSCs. A, Transcript levels of CALCR in patient-derived GSCs and cell line–derived GSCs as detected by real-time qPCR. The transcript levels divide the GSCs into CALCR-low (MGG4, T98G, U87, U343, and U373) and CALCR-high (1035, MGG4, MGG6, MGG8, and MGG23). B and C, GSC growth as neurospheres was assessed in the presence of BSA or calcitonin. The percentage inhibition of neurosphere growth in calcitonin condition as compared with BSA when all spheres (B) and spheres with size >30 μm in diameter (C) were considered. For each GSC cell line, the number of spheres in the BSA condition was normalized to 100%, and then the calcitonin condition was calculated. The difference in neurosphere growth percentage is plotted. Student's t test was performed to evaluate the statistical significance between the two groups. D, Representative images of neurosphere assay for CALCR-low (U373 and U343) and CALCR-high (MGG8 and MGG23) GSCs. L.M., low magnification (2.5×) and H.M., high magnification (10×). E, Limiting dilution assay for GSCs. For BSA and calcitonin conditions, 1, 5, 10, 20, 50, 100, and 200 cells were plated (n = 12). At the endpoint, number of wells not having any sphere was calculated, and the graph was plotted using extreme limiting dilution analysis (ELDA) software. Total number of cells (dose) is plotted on the x-axis, and the log fraction of nonresponding/empty well is plotted on the y-axis. The dotted lines represent the confidence interval (0.95). F, Transcript levels of glioma reprogramming factors (SOX2, OLIG2, SALL2, and POU3F2) in calcitonin condition compared with BSA in various GSC cell lines tested. The log2 values are plotted for calcitonin calculated with respect to the BSA condition (normalized to 0). G, pERK and pAKT levels in GSCs in BSA versus calcitonin conditions in CALCR-low GSCs (U343 and MGG4) and CALCR-high GSCs (MGG8, 1035, and LN229). The quantification for each phospho-protein is provided at the bottom of the corresponding blot. The total protein for each lane was normalized to the actin levels, and subsequently, the phospho-protein was normalized to the normalized total. The value for BSA for each phospho-protein per cell line was normalized to 1, and the value for calcitonin was calculated in comparison with the normalized BSA. The P values for all the panels are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively.

Figure 5.

Effect of CT–CALCR tumor suppressor axis on GSCs. A, Transcript levels of CALCR in patient-derived GSCs and cell line–derived GSCs as detected by real-time qPCR. The transcript levels divide the GSCs into CALCR-low (MGG4, T98G, U87, U343, and U373) and CALCR-high (1035, MGG4, MGG6, MGG8, and MGG23). B and C, GSC growth as neurospheres was assessed in the presence of BSA or calcitonin. The percentage inhibition of neurosphere growth in calcitonin condition as compared with BSA when all spheres (B) and spheres with size >30 μm in diameter (C) were considered. For each GSC cell line, the number of spheres in the BSA condition was normalized to 100%, and then the calcitonin condition was calculated. The difference in neurosphere growth percentage is plotted. Student's t test was performed to evaluate the statistical significance between the two groups. D, Representative images of neurosphere assay for CALCR-low (U373 and U343) and CALCR-high (MGG8 and MGG23) GSCs. L.M., low magnification (2.5×) and H.M., high magnification (10×). E, Limiting dilution assay for GSCs. For BSA and calcitonin conditions, 1, 5, 10, 20, 50, 100, and 200 cells were plated (n = 12). At the endpoint, number of wells not having any sphere was calculated, and the graph was plotted using extreme limiting dilution analysis (ELDA) software. Total number of cells (dose) is plotted on the x-axis, and the log fraction of nonresponding/empty well is plotted on the y-axis. The dotted lines represent the confidence interval (0.95). F, Transcript levels of glioma reprogramming factors (SOX2, OLIG2, SALL2, and POU3F2) in calcitonin condition compared with BSA in various GSC cell lines tested. The log2 values are plotted for calcitonin calculated with respect to the BSA condition (normalized to 0). G, pERK and pAKT levels in GSCs in BSA versus calcitonin conditions in CALCR-low GSCs (U343 and MGG4) and CALCR-high GSCs (MGG8, 1035, and LN229). The quantification for each phospho-protein is provided at the bottom of the corresponding blot. The total protein for each lane was normalized to the actin levels, and subsequently, the phospho-protein was normalized to the normalized total. The value for BSA for each phospho-protein per cell line was normalized to 1, and the value for calcitonin was calculated in comparison with the normalized BSA. The P values for all the panels are represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively.

Close modal

CT–CALCR tumor suppressor axis is a potential therapeutic target

To confirm the tumor-suppressive function of CT–CALCR axis, we tested the effect of CALCR co-transfection on the ability of Ras (KRasV12) to transform SV40-immortalized mouse astrocytes (IMA2.1 cells) in vitro. We observed that Ras transformed IMA2.1 cells efficiently as seen by the increased number of colonies in soft agar (Fig. 6A and B, bar 1). Furthermore, we tested the effects of WT and mutated CALCR on astrocyte transformation. Although WT CALCR inhibited Ras-mediated transformation very efficiently (Fig. 6A and B, compare bar 2 with 1), CALCR mutants failed to inhibit Ras-mediated transformation efficiently, although to varying levels (Fig. 6A and B, compare bars 3 to 9 with 2). In particular, the severe mutants with the highest LOF scores (A51T and V250M) were found to have completely lost the ability to inhibit Ras-mediated transformation. To ascertain pathways that could be involved in CALCR-mediated inhibition of transformation by RasV12, we evaluated the effect of CALCR on the phosphorylation status of ERK, AKT, and JNK molecules in RasV12-transformed cells. Ras-transformed IMA2.1 cells showed an increase in pERK and pJNK (Supplementary Fig. S13; compare lane 3 with 1), which was abrogated significantly when CALCR was exogenously introduced (Supplementary Fig. S13; compare lane 4 with 3). These results demonstrate that CALCR targets ERK, JNK, and AKT signaling molecules independent of Ras and perhaps involving other signaling pathways. To evaluate the importance of CT–CALCR axis as a therapeutic target, we tested the effect of calcitonin intraperitoneal injections on the xenograft tumor growth of LN229-luc cells in NIH female nu/nu mice. We found that treatment with calcitonin inhibited LN229 xenograft tumor growth significantly (Fig. 6C–E). In summary, these results confirm the tumor-suppressive nature of CT–CALCR axis and highlight the importance of the axis as a novel therapeutic target in glioma (Fig. 6F). When glioma cells harbor WT CALCR (left), binding of calcitonin leads to inhibition of JNK, ERK, and AKT phosphorylation resulting in reduced oncogenic properties of the cells such as proliferation, migration, invasion, and anchorage-independent growth capacity. This ultimately contributes to better survival of patients with GBM with less aggressive tumors. Also, treatment with calcitonin could be an option for therapy for these patients. However, when CALCR gets mutated (right), binding of calcitonin to CALCR, change in conformation of the receptor, or activation of G-protein by CALCR may be abrogated, and this will damage the inhibitory signals downstream. Therefore, increased levels of pERK, pAKT, and pJNK are observed with a resultant increase in oncogenic properties of the glioma cells mentioned before. Hence, this will result in poor survival in patients with GBM with very aggressive tumors.

Figure 6.

Effect of CALCR on Ras-mediated transformation of immortalized astrocytes in vitro and in vivo xenograft tumor growth. A, Transformation of mouse immortalized astrocytes IMA2.1 by RasV12 oncogene tested by colony formation in soft agar in presence of WT and the seven mutated forms of CALCR. B, The total number of colonies in RasV12/CALCR condition was normalized to 0%, and the percentage of colonies in RasV12/VC and RasV12/CALCR mutants was calculated from A and plotted. C, The pictorial representation of the experimental design of in vivo xenograft mouse model for testing the therapeutic efficacy of calcitonin. Day 0 begins with the subcutaneous injection of LN229-luc cells in the NIH nu/nu mice (n = 4). Luciferase reading was taken every 5 days till day 30. The calcitonin injection regime followed was days 10–17 (1 I.U./mouse/24 hrs) and days 18–25 (1 I.U./mouse/48 hours). D, The tumor growth (days 10–30) for BSA and calcitonin injected mice are shown by bioluminescence imaging. E, The total flux of luminescence from the tumors of BSA and calcitonin treated mice are plotted. The scale has been adjusted to – minimum flux = 600 radiance counts (photon/s) and maximum flux = 2,000 radiance counts (photon/s). The P values for panels B and E represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively. F, Graphical representation of the clinical efficacy of the key findings from this study. The panel on the left describes therapeutic utilization of CT-CALCR axis in CALCR WT glioma. The panel on the right describes how CALCR mutation abolishes CT-CALCR tumor suppressor pathway leading to an aggressive GBM tumor.

Figure 6.

Effect of CALCR on Ras-mediated transformation of immortalized astrocytes in vitro and in vivo xenograft tumor growth. A, Transformation of mouse immortalized astrocytes IMA2.1 by RasV12 oncogene tested by colony formation in soft agar in presence of WT and the seven mutated forms of CALCR. B, The total number of colonies in RasV12/CALCR condition was normalized to 0%, and the percentage of colonies in RasV12/VC and RasV12/CALCR mutants was calculated from A and plotted. C, The pictorial representation of the experimental design of in vivo xenograft mouse model for testing the therapeutic efficacy of calcitonin. Day 0 begins with the subcutaneous injection of LN229-luc cells in the NIH nu/nu mice (n = 4). Luciferase reading was taken every 5 days till day 30. The calcitonin injection regime followed was days 10–17 (1 I.U./mouse/24 hrs) and days 18–25 (1 I.U./mouse/48 hours). D, The tumor growth (days 10–30) for BSA and calcitonin injected mice are shown by bioluminescence imaging. E, The total flux of luminescence from the tumors of BSA and calcitonin treated mice are plotted. The scale has been adjusted to – minimum flux = 600 radiance counts (photon/s) and maximum flux = 2,000 radiance counts (photon/s). The P values for panels B and E represented by *, **, and ***, which denote P values of <0.05, 0.01, and 0.001, respectively. F, Graphical representation of the clinical efficacy of the key findings from this study. The panel on the left describes therapeutic utilization of CT-CALCR axis in CALCR WT glioma. The panel on the right describes how CALCR mutation abolishes CT-CALCR tumor suppressor pathway leading to an aggressive GBM tumor.

Close modal

In this study, we analyzed somatic mutations in an Indian cohort of 42 gliomas compared with matched blood samples using WES. The mutation spectrum was found to be largely similar to that of TCGA (3). Although TP53, IDH1, and ATRX were mutated typically in lower grades, mutations in PTEN, PIK3CA, and NF1 were largely seen in GBMs. Integrative analysis of the exome data of Indian and TCGA cohorts identified several pathways to be associated with GBM survival. GBMs with mutations in one or more genes that belong to “neuroactive ligand–receptor interaction pathway” predicted worst prognosis. We also show that CALCR, the top mutated gene belonging to the above pathway, along with its ligand functions as a tumor suppressor axis and inactivating mutations/reduced transcript levels in CALCR defines a highly aggressive subset of GBM with very poor survival.

An integrated survival analysis involving GBM-specific mutated genes derived from Indian and TCGA cohorts identified several pathways being associated with GBM survival. Of these, genetic alterations in “neuroactive ligand–receptor interaction pathway” genes were found to be associated with poor survival with high significance. This pathway consists of a large number of GPCRs, which upon activation by their respective ligands have been shown to regulate neuronal signaling in specific ways, thus influencing a variety of animal behavior (18). Epigenetic alterations in neuroactive ligand–receptor interaction pathway have been shown to be associated with the risk of developing pancreatic cancer (19, 20), small-cell lung cancer (21), renal cell carcinoma (22), hepatocellular carcinoma (23), etc. In TCGA pan-cancer data study, neuroactive ligand–receptor interaction pathway came up to be the fifth most highly mutated pathways in cancer (24). However, there are no studies that show the functional significance of genetic alterations in this pathway with respect to cancer development. This study demonstrates that mutational alteration of neuroactive–ligand interaction pathway promotes GBM aggressiveness. Our finding has translational relevance as agonists and antagonists for many GPCRs are available, and hence, one could utilize a drug repositioning approach wherein the known GPCR inhibitors may have potential anticancer therapeutic implications (25).

Calcitonin, the ligand of CALCR, is a 32 amino acid polypeptide hormone synthesized primarily by thyroid (26). Binding of calcitonin to CALCR regulates a variety of signaling downstream resulting in the regulation of bone metabolism, calcium flux, and cancer cell proliferation (26–29). Although mutations the CALCR gene have been reported in lung adenocarcinoma (30), the functional importance of CALCR mutations with respect to cancer development is not known. Our study demonstrates that CT–CALCR signaling inhibits cell proliferation, migration, invasion, and anchorage-independent growth of established glioma cell lines with a concomitant inhibition of downstream AKT, MEK, and JNK signaling. Furthermore, we show that somatic mutations in CALCR led to LOF in glioma cells, and patients with mutated CALCR exhibited poor prognosis. In fact, the LOF score, calculated from the experiments (depicts the severity of the mutants), correlated significantly with patient survival, although the small number of samples preclude the ability to make strong conclusions.

Our study reveals the potential of the CT–CALCR axis as a novel therapeutic target as observed by its effectiveness in the inhibition of GSCs that led to a significant decrease in glioma reprogramming factors. The role of CALCR as a tumor suppressor was further validated by its potency to inhibit the first event in tumor initiation, that is, transformation. Indeed, CALCR inhibited RasV12-mediated transformation of immortalized astrocytes in vitro. Moreover, CALCR was found to be capable of inhibiting pERK and pJNK even in the presence of constitutively active oncogenic Ras, which suggests that CALCR is capable of directly regulating signaling molecules present downstream of Ras. Although Ras is known to activate MEK kinase, which in turn can activate ERK and other pathways such as JNK (31), it is also known that downstream to GPCR, other upstream molecules, such as PKA (32) and Rac (33), can regulate ERK and CDC42 (34), Gαi (35) can regulate JNK. Thus, we demonstrate that CT–CALCR axis is a direct inhibitor of oncogenic signaling downstream to Ras, which further underscores the importance of CT–CALCR axis as a compelling tumor suppressor pathway in glioma. This was substantiated by the result that intraperitoneal injection of calcitonin in NIH nu/nu mice, inoculated with LN229-luc cells to form subcutaneous flank tumor, reduced the tumor burden significantly.

Thus, our study identifies CT–CALCR axis acts as a tumor suppressor pathway in GBM. Mutational inactivation of CALCR or reduced CALCR transcript levels leads to an aggressive GBM with poor survival. Our findings have multiple translational implications. Most importantly, for GBMs with wild-type CALCR, calcitonin could be considered as a treatment option (36). In fact, salmon calcitonin is prescribed for postmenopausal osteoporosis (37) and also in therapy of giant cell granuloma (38). For GBMs with mutated CALCR receptor, a combination of inhibitors of AKT, MEK, and JNK pathways could be tried.

No potential conflicts of interest were disclosed.

Conception and design: J. Pal, V. Patil, C. Sarkar, K. Somasundaram

Development of methodology: J. Pal, V. Patil, C. Sarkar, K. Somasundaram

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Pal, V. Patil, K. Kaur, C. Sarkar, K. Somasundaram

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Pal, V. Patil, A. Kumar, C. Sarkar, K. Somasundaram

Writing, review, and/or revision of the manuscript: J. Pal, V. Patil, K. Somasundaram

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Kumar, K. Somasundaram

Study supervision: K. Somasundaram

The results published here are in whole or part based upon data generated by The Cancer Genome Atlas pilot project established by the NCI and NHGRI. Information about TCGA and the investigators and institutions that constitute the TCGA research network can be found at http://cancergenome.nih.gov/. We also acknowledge the use of GSE7696 in this study. We thank Drs. Hiroaki Wakimoto, Samuel Rabkin, and Santosh Kesari for providing us with GSCs. We thank Dr. Stefan Schildknecht for providing mouse immortalized astrocyte cells. J. Pal acknowledges Indian Institute of Science for the research fellowship. V. Patil and J. Pal thank DBT, Government of India for financial support. The NGS facility, Indian Institute of Science is acknowledged for exome sequencing. The Centre for Animal Facility (CAF), Indian Institute of Science, and Dr. Krishnaveni (CAF) are acknowledged. K. Somasundaram acknowledges CSIR and DBT, Government of India for research grant. Infrastructure support by funding from DST-FIST, DBT grant-in-aid and UGC (Centre for Advanced Studies in Molecular Microbiology) to MCB is acknowledged. Prof. Partha Majumder (NIBMG) and Arjun Arkal Rao are acknowledged for their invaluable inputs.

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.

1.
The Cancer Genome Atlas Research Network
. 
Comprehensive genomic characterization defines human glioblastoma genes and core pathways
.
Nature
2008
;
455
:
1061
8
.
2.
Frattini
V
,
Trifonov
V
,
Chan
JM
,
Castano
A
,
Lia
M
,
Abate
F
, et al
The integrated landscape of driver genomic alterations in glioblastoma
.
Nat Genet
2013
;
45
:
1141
9
.
3.
Brennan
CW
,
Verhaak
RG
,
McKenna
A
,
Campos
B
,
Noushmehr
H
,
Salama
SR
, et al
The somatic genomic landscape of glioblastoma
.
Cell
2013
;
155
:
462
77
.
4.
Stupp
R
,
Hegi
ME
,
Mason
WP
,
van den Bent
MJ
,
Taphoorn
MJ
,
Janzer
RC
, et al
Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial
.
Lancet Oncol
2009
;
10
:
459
66
.
5.
Cibulskis
K
,
Lawrence
MS
,
Carter
SL
,
Sivachenko
A
,
Jaffe
D
,
Sougnez
C
, et al
Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples
.
Nat Biotechnol
2013
;
31
:
213
9
.
6.
Cancer Genome Analysis
. 
Indelocator
.
Cambridge, MA
:
Broad Institute
.
Available from
: http://archive.broadinstitute.org/cancer/cga/indelocator.
7.
Ohgaki
H
,
Kleihues
P
. 
Genetic pathways to primary and secondary glioblastoma
.
Am J Pathol
2007
;
170
:
1445
53
.
8.
Vigneswaran
K
,
Neill
S
,
Hadjipanayis
CG
. 
Beyond the World Health Organization grading of infiltrating gliomas: advances in the molecular genetics of glioma classification
.
Ann Transl Med
2015
;
3
:
95
.
9.
Kannan
K
,
Inagaki
A
,
Silber
J
,
Gorovets
D
,
Zhang
J
,
Kastenhuber
ER
, et al
Whole-exome sequencing identifies ATRX mutation as a key molecular determinant in lower-grade glioma
.
Oncotarget
2012
;
3
:
1194
203
.
10.
Ji
TH
,
Grossmann
M
,
Ji
I
. 
G protein-coupled receptors. I. Diversity of receptor-ligand interactions
.
J Biol Chem
1998
;
273
:
17299
302
.
11.
Kobilka
BK
. 
G protein coupled receptor structure and activation
.
Biochim Biophys Acta
2007
;
1768
:
794
807
.
12.
Patil
V
,
Pal
J
,
Somasundaram
K
. 
Elucidating the cancer-specific genetic alteration spectrum of glioblastoma derived cell lines from whole exome and RNA sequencing
.
Oncotarget
2015
;
6
:
43452
71
.
13.
Degrauwe
N
,
Schlumpf
TB
,
Janiszewska
M
,
Martin
P
,
Cauderay
A
,
Provero
P
, et al
The RNA binding protein IMP2 preserves glioblastoma stem cells by preventing let-7 target gene silencing
.
Cell Rep
2016
;
15
:
1634
47
.
14.
New
DC
,
Wong
YH
. 
Molecular mechanisms mediating the G protein-coupled receptor regulation of cell cycle progression
.
J Mol Signal
2007
;
2
:
2
.
15.
Morfis
M
,
Tilakaratne
N
,
Furness
SG
,
Christopoulos
G
,
Werry
TD
,
Christopoulos
A
, et al
Receptor activity-modifying proteins differentially modulate the G protein-coupling efficiency of amylin receptors
.
Endocrinology
2008
;
149
:
5423
31
.
16.
Johansson
E
,
Hansen
JL
,
Hansen
AM
,
Shaw
AC
,
Becker
P
,
Schaffer
L
, et al
Type II turn of receptor-bound salmon calcitonin revealed by X-ray crystallography
.
J Biol Chem
2016
;
291
:
13689
98
.
17.
Suva
ML
,
Rheinbay
E
,
Gillespie
SM
,
Patel
AP
,
Wakimoto
H
,
Rabkin
SD
, et al
Reconstructing and reprogramming the tumor-propagating potential of glioblastoma stem-like cells
.
Cell
2014
;
157
:
580
94
.
18.
Engel
SR
,
Grant
KA
. 
Neurosteroids and behavior
.
Int Rev Neurobiol
2001
;
46
:
321
48
.
19.
Wei
P
,
Tang
H
,
Li
D
. 
Insights into pancreatic cancer etiology from pathway analysis of genome-wide association study data
.
PLoS One
2012
;
7
:
e46887
.
20.
Li
L
,
Hanahan
D
. 
Hijacking the neuronal NMDAR signaling circuit to promote tumor growth and invasion
.
Cell
2013
;
153
:
86
100
.
21.
Jahchan
NS
,
Dudley
JT
,
Mazur
PK
,
Flores
N
,
Yang
D
,
Palmerton
A
, et al
A drug repositioning approach identifies tricyclic antidepressants as inhibitors of small cell lung cancer and other neuroendocrine tumors
.
Cancer Discov
2013
;
3
:
1364
77
.
22.
Liu
X
,
Wang
J
,
Sun
G
. 
Identification of key genes and pathways in renal cell carcinoma through expression profiling data
.
Kidney Blood Press Res
2015
;
40
:
288
97
.
23.
Zhang
MH
,
Shen
QH
,
Qin
ZM
,
Wang
QL
,
Chen
X
. 
Systematic tracking of disrupted modules identifies significant genes and pathways in hepatocellular carcinoma
.
Oncol Lett
2016
;
12
:
3285
95
.
24.
Dees
ND
,
Zhang
Q
,
Kandoth
C
,
Wendl
MC
,
Schierding
W
,
Koboldt
DC
, et al
MuSiC: identifying mutational significance in cancer genomes
.
Genome Res
2012
;
22
:
1589
98
.
25.
Liu
Y
,
An
S
,
Ward
R
,
Yang
Y
,
Guo
XX
,
Li
W
, et al
G protein-coupled receptors as promising cancer targets
.
Cancer Lett
2016
;
376
:
226
39
.
26.
Masi
L
,
Brandi
ML
. 
Calcitonin and calcitonin receptors
.
Clin Cases Miner Bone Metab
2007
;
4
:
117
22
.
27.
Purdue
BW
,
Tilakaratne
N
,
Sexton
PM
. 
Molecular pharmacology of the calcitonin receptor
.
Receptors Channels
2002
;
8
:
243
55
.
28.
Robbins
EW
,
Travanty
EA
,
Yang
K
,
Iczkowski
KA
. 
MAP kinase pathways and calcitonin influence CD44 alternate isoform expression in prostate cancer cells
.
BMC Cancer
2008
;
8
:
260
.
29.
Terra
SR
,
Cardoso
JC
,
Felix
RC
,
Martins
LA
,
Souza
DO
,
Guma
FC
, et al
STC1 interference on calcitonin family of receptors signaling during osteoblastogenesis via adenylate cyclase inhibition
.
Mol Cell Endocrinol
2015
;
403
:
78
87
.
30.
George
J
,
Lim
JS
,
Jang
SJ
,
Cun
Y
,
Ozretic
L
,
Kong
G
, et al
Comprehensive genomic profiles of small cell lung cancer
.
Nature
2015
;
524
:
47
53
.
31.
Xia
Y
,
Makris
C
,
Su
B
,
Li
E
,
Yang
J
,
Nemerow
GR
, et al
MEK kinase 1 is critically required for c-Jun N-terminal kinase activation by proinflammatory stimuli and growth factor-induced cell migration
.
Proc Natl Acad Sci U S A
2000
;
97
:
5243
8
.
32.
Dumaz
N
,
Marais
R
. 
Integrating signals between cAMP and the RAS/RAF/MEK/ERK signalling pathways. Based on the anniversary prize of the Gesellschaft fur Biochemie und Molekularbiologie Lecture delivered on 5 July 2003 at the Special FEBS Meeting in Brussels
.
FEBS J
2005
;
272
:
3491
504
.
33.
Eblen
ST
,
Slack
JK
,
Weber
MJ
,
Catling
AD
. 
Rac-PAK signaling stimulates extracellular signal-regulated kinase (ERK) activation by regulating formation of MEK1-ERK complexes
.
Mol Cell Biol
2002
;
22
:
6023
33
.
34.
Wang
L
,
Yang
LD
,
Burns
K
,
Kuan
CY
,
Zheng
Y
. 
Cdc42GAP regulates c-Jun N-terminal kinase (JNK)-mediated apoptosis and cell number during mammalian perinatal growth
.
Proc Natl Acad Sci U S A
2005
;
102
:
13484
9
.
35.
Bastin
G
,
Yang
JY
,
Heximer
SP
. 
Galphai3-dependent inhibition of JNK activity on intracellular membranes
.
Front Bioeng Biotechnol
2015
;
3
:
128
.
36.
Wookey
PJ
,
McLean
CA
,
Hwang
P
,
Furness
SG
,
Nguyen
S
,
Kourakis
A
, et al
The expression of calcitonin receptor detected in malignant cells of the brain tumour glioblastoma multiforme and functional properties in the cell line A172
.
Histopathology
2012
;
60
:
895
910
.
37.
Leggate
J
,
Farish
E
,
Fletcher
CD
,
McIntosh
W
,
Hart
DM
,
Sommerville
JM
. 
Calcitonin and postmenopausal osteoporosis
.
Clin Endocrinol
1984
;
20
:
85
92
.
38.
Pogrel
MA
. 
Calcitonin therapy for central giant cell granuloma
.
J Oral Maxillofac Surg
2003
;
61
:
649
53
.