| Objectives:Gliomas are the most common primary central nervous system (CNS) tumors in adults (-60%), with glioblastoma (GBM) being the most devastating ones. Because GBM infiltrates deeply into the surrounding brain parenchyma, therefore complete surgical resection is almost impossible. Despite maximal safe surgical resection, followed by standardized radio-chemotherapy, tumor recurrence from the primary foci or even distal location is inevitable and patients with GBM still have a very dismal prognosis. Discovery of biomarkers indicative of glioma malignancy is believed to be valuable in terms of early and accurate diagnosis of glioma, monitoring of treatment responses, alerts for recurrence, and even development of individualized medicine for glioma patients. Given its special location and biological activities, human cerebrospinal fluid (CSF) has long been deemed a rich reservoir for the discovery of various neurological disease biomarkers. In this study, by analyzing the CSF form patients with different pathological grades of gliomas, we aim to profiling the compositional changes of the CSF proteome, both reflecting macroscopically as thermal stability and microscopically as changes in the abundance of high- and low-abundance proteins, which accompany the malignant transformation of gliomas. We believe these efforts would be valuable for the ongoing endeavor toward mining the CSF proteome for glioma biomarkers.Methods:In Part One, after appropriate sample preparations (ultracentrifugation, buffer exchanges, determination of total protein concentration), we analyzed 39 glioma samples, including 36 CSF and 3 GBM tumor cyst fluid (CF), with highly sensitive differential scanning calorimeter (DSC) for changes in thermal stability and by ELISA for concentrations of the 10 most abundant proteins in the proteome; In Part Two, we first systematically reviewed the research articles reporting proteomic analysis of glioma CSF by search the Pubmed website, retrieved the protein lists (biomarkers) validated as differential expressed between glioma CSF and controls, analyzed them by bioinformatics tools like PANTHER to ascertain the underneath relationship between these proteins biomarkers, and suggested novel candidate focus protein from the network generated by IPA software. Finally, we validated these novel candidate biomarkers in our 29 samples (including 26 CSF and 3 CF) by ELISA; In Part Three, we first prepared 44 glioma patient samples (including 42 CSF and 2 CF) by ProteoMiner kit for depletion of high abundance proteins and enrichment of low abundance proteins, desalting, total protein determination, tryptic digestion, iTRAQ labeling and finally the CSF proteome was quantitatively analyzed by 2D nano reverse phase liquid chromatography (LC)-electrospray ionization (ESI)-mass spectrometry (MS)/MS. After protein identification by Mascot software, differentially expressed protons lists were extracted and, among them, candidate protein biomarker were sleeted after reviewing the literatures. The resultant differential expressed protein "panel" was further validated in the CSF by ELISA and re-examined for reliability in terms of discriminating GBM from other gliomas of low malignancies in a smaller set of 6 prospectively collected CSF samples. Moreover, we also collected culture superannuates from 5 cases of primary GBM tissues, as well as from 9 tumor cell lines. The latter included 4 human GBM cell lines (A172, U87, U251 & SHG66),2 glioma cell lines (H4 & SW1783) and 3 non-glioma cell lines (K-562, AGS & N87). In order to trace the origin of these potential glioma biomarkers, their concentrations in these cell supernatants were also determined by ELISAResults:In Part One, the DSC thermograms of the CSF from non-neoplastic group manifested a major transition temperature or melting temerpaure (Tm) of 73.2℃, which increased to 73.8℃ and 74.1℃ in the benign brain tumor group and low grade glioma (LGG) group. The most prominent shift in Tm was observed in the GBM group, which elevated to 75.6℃ and was statistically higher than other groups (P<0.001). The total protein concentration of the GBM CF was significantly higher than CSF groups (P<0.001). And the totoal protein concentration was highest in GBM CSF when compared with other CSF groups (P<0.05 or P<0.01). Five proteins, i.e. ALB, IgG, AGP, AAT and FIB, accounted for more than 99% and 98% of the total mass of the 10 proteins examined in CSF and CF, respectively. Except for AGP, the GBM CF was featured by significantly higher protein concentrations in the rest 9 proteins than CSF groups (P<0.05 or P<0.001). Among the CSF groups, GBM CSF had significantly higher ALB, IgG, AAT, FIB, APOA1 and TRF concentrations than other groups. Moreover, in terms of relative abundance, CF had higher AAT, FIB, Cys-C and C3 abundance while GBM CSF had higher IgG, AAT and TRF abundance; In Part Two, we retrieved a total of 8 peer reviewed articles reporting 19 differentially expressed proteins as validated by different methods. According to Gene Ontology (GO) annotation and PANTHER analysis, the majority of these 19 proteins were denoted with the molecular function of "binding", cellular location of "extracellular space" and participated in biological processes like "inflammatory responses", "stress/hypoxia responses" and "apoptotic process" et al. Therefore, these glioma CSF proteins might be involved in tumor progression and tumor related immune responses. According to Human Protein Atlas (HPA) database, most of these protein had increased expression in glioma tissue compared with normal CNS tissue. However, only one protein (attractin) had been identified by more than one article and this relatively low "reproductivity" might be related with the fact that there was no standard experiment protocols for CSF proteomic studies. IPA analysis showed these 19 proteins were endowed with important functions like "inflammatory responses" and "cell death and survival", and participated in pathways like "acute phase response pathway", "VEGF signaling", and "p38 signaling". Core proteins in the first network generated by IPA were largely located at the nucleus (e.g. MAPK3/MAPK1, HNRNPA2B1 and CEBPA) and they had functions like DNA binding, catalytic activities and regulatory activities. In comparison, IL-6/STAT-3 pathway was found central to the second IPA network. Accordingly, focus proteins like GAL, HSPA5, WNT and IL-6 were further selected and validated in our CSF and CF samples. According to ELISA resutls, GAL, HSPA5 and WNT levels were significantly lower in CF than CSF, while no statistically significant difference was observed between different CSF groups. In comparsion, IL-6 level was not only significantly higher in GBM CF when compared with CSF (P<0.001), but also significantly higher in GBM CSF as compared with CSF from LGG (P<0.001); In Part Three, we identified a total of 445 proteins form the CSF, and these proteins were mainly located at the extracellular space (24%), cell membrane (13%) and cytoplasm (11%), participated in processes like regulation (22%), cellular process (18%) and metabolic process (9%), and had molecular functions like binding (48%) and catalytic activity (22%). Among these 445 proteins,144 were related with the malignancy of astrocytoma (up-regulation 84, down-regulation 60),14 were related with glioma malignancy (up-regulation 5, down-regulation 9),17 were differentially expressed between high grade glioma (HGG) and LGG (up-regulation 3, down-regulation 14, P<0.05),179 had more than two-fold changes between recurrent GBM and primary GBM (up-regulation 89, down-regulation 90), and 135 had more than two-fold changes between high and low grade oligdendroglioma (up-regulation 53, down-regulation 82). IPA analysis revealed the top functions of the networks as interconnected by these differentially expressed proteins were protein synthesis, lipid metabolism and small molecule biochemistry, intercellular signaling, cellular immune responses, humoral immune response, inflammatory responses, DNA replication, recombination and repair, cancer, cell motility, molecule transport, cell death and survival, cell cycle et al. Six proteins, namely CNDP1, PDIA4, QSOX1, PF4, SFRP3 and VCL, were validated by ELISA to be up-regulated in CSF from GBM compared with that from gliomas of lower malignancy. SFRP3 and VCL, especially the latter, had higher concentration in primary GBM cell culture supernatants compared with CSF, indicating their potential direct tumor origin. Moreover, QSOX1, SFRP3 and VCL levels were higher in supernatants form glioma cell lines when compared with non-glioma tumor cell lines, indicating their secretion had glioma specificity. Besides, the level of QSOX1 in the supernatants increased with the malignancy of the cell lines (from H4, SW1783 to U87) from which the supernatants were collected. Therefore, QSOX1 might be a potential biomarker whose level showing correlation with the glioma malignancy.Conclusions:we presented the evidences for the first time that, by analyzing the thermal stability of the CSF proteome, DSC has the potential to be a diagnostic tool for the discretization of GBM patients from those with gliomas of lower grade or benign intracranial lesions. The changes in the levels of the major abundance proteins might be one possible explanation for the observed shifts in Tm. Because DSC only reflected the thermodynamic properties of the whole proteome, especially those high-abundance proteins, it therefore was necessary to apply high through-put proteomic platforms for detection of the changes in the low abundance region of the CSF proteome during gliomagenesis. Accordingly, we then systematically reviewed, for the first time, the available literatures on the proteomic profiling of glioma CSF, retrieved potential biomarkers from these reports and validated a novel biomarker (i.e. IL-6) discovered through bioinformatics analysis. We were also the first to apply iTRAQ-LC-MS/MS to quantify the glioma CSF proteome and identified 6 proteins (e.g. CNDP1, PDIA4, QSOX1, PF4, SFRP3 and VCL) as differentially expressed between GBM CSF and CSF from gliomas of lower malignancy. These low abundance proteins, either secreted by tumor and its microenvironment, or as host responses against tumor, might be valuable biomarker for GBM patients and future research is indispensable to reveal their specific biological activities in glioma pathogenesis. Nevertheless, it would be more preferable to combine expensive and labor intensive proteomics techniques with convenient DSC scans in the pursuit of glioma malignancy biomaers. Our present work provided evidences for further larger scale clinical sample validation and even clinical translation. |
PDF Full Text Request