Thesis focuses on the characterization of the mechanisms involved in regulation of the CHI3L1 gene expression, which is overexpressed in glioblastoma, the elucidation of the peculiarities of its functions, as well as glioblastoma typing based on the gene expression profiles. Among all human cancers the tumors of the central nervous system comprise from 2 to 5%, and about half of them are glial brain tumors. They develop from astrocytic glia or oligodendroglia, 50% of them are diagnosed as glioblastoma, tumor of fourth, the highest malignancy grade according to the World Health Organization grading system. Medium survival for patients with glioblastoma is 12-14 months, and despite prominent success of radio- and chemotherapy during the last decades the survival did not increase significantly. The CHI3L1 gene that encodes chitinase-3-like protein 1 is one of the most overexpressed genes in glioblastoma as compared to adult normal brain. The CHI3L1 protein belongs to glycosylhydrolase family 18 and possesses growth-stimulating and proangiogenic properties. In human organism CHI3L1 is a marker of differentiated macrophages, where it is expressed at the last stage of their maturation. In tumors CHI3L1 is expressed by both macrophages recruited to the tumor and tumor cells themselves. To investigate the CHI3L1 protein level in different brain tumors we analyzed 54 tumor samples of different origin. Among analyzed brain tumor samples we found the CHI3L1 level to be absent in diffuse astrocytoma and other tumors of non-astrocytic origin. Alternatively, CHI3L1 was produced at high level in anaplastic astrocytoma and glioblastoma. Next, we found that in the patients with primary glioblastomas, G/G-homozygous by SNP -131C->G in the CHI3L1 promoter, there is a tendency to decreased CHI3L1 expression. We investigated the influence of the transcription factor p53 on the CHI3L1 production in the U87 cells derived from glioblastoma and showed that the CHI3L1 expression may be reduced due to the ectopic expression of the TP53, p53 protein activation or by blocking p53 degradation. We demonstrated that the specific inhibition of histone deacetylase SIRT1 reduced the CHI3L1 expression in U87 cells. We purified CHI3L1 from MG63 conditioned medium using heparin affinity chromatography and showed that purified CHI3L1 similarly to known growth factor IGF1, increased the viability of glioblastoma U251 cells after 24 hours incubation. Comparing the structures of two homologous proteins that have different effects on cell proliferation: CHI3L2 model, obtained by homology modeling, and resolved CHI3L1 structure we showed that CHI3L1 has two different functional units - the reorganized ligand-binding site and the surface positively charged motif. Using site-directed mutagenesis we demonstrated that CHI3L1 binds heparin through a cluster of positively charged amino acids on the protein surface and the presence of heparin-binding site might be responsible for substrate-independent growth of 293 cells with the ectopic production of CHI3L1. 293 cells, stably transfected with CHI3L1 that carried mutated heparin-binding site, formed lower number of colonies in semisoft agar as compared to the cells stably transfected with wild-type CHI3L1. We developed an approach for glioblastoma subtyping, which consists of the following steps: merge of the gene expression data from different works, determination of the number of groups in the resulting data set, cluster analysis and verification of results using the independent method - neural network analysis. Gene expression data could be divided into two large groups, and expression of 401 genes is significantly different in these two groups. To confirm our findings, we created Kohonen map, built on the basis of expression of 15 genes with increased expression in the first subtype, and 60 of 401 genes with increased expression in the second subtype, and showed that two glioblastoma subtypes separated on the map. According to the expression profile of 401 genes selected in this study we identified two subtypes of glioblastoma, "proliferative" and "proneural" that could display peculiarities of the development of these tumors. Glioblastoma subtyping based on the gene expression profiles may help in the future to develop individualized tumor treatment strategies.
