Abstract

Background: Only 30-100 nm in diameter, exosomes are membranous vesicles of endocytic origin that are released into the extracellular space by a variety of cell types including cancerous cells such as glioblastoma multiforme (GBM). Several recent studies are indicative of the fact that exosomes are involved in a mode of targeted intercellular transfer of molecules and induction of different functions in the recipient cells. Conveniently, it has been described that GBM tumor cells shed exosomes into the extracellular space which are retrievable from the cerebral spinal fluid and the blood plasma. This property of exosomes can be utilized to better study GBM. Objective: We took advantage of the exosome blood born mode of communication in GBM. The objective was to compare and contrast the mRNA expression profile of the exosomes extracted from the blood plasma of patients with GBM and normal control subjects using microarray in search of non-invasive novel diagnostic strategy for GBM. Method: Blood samples were collected from 9 patients diagnosed with de-novo primary GBM immediately prior to cytoreduction surgery. Furthermore, blood samples were collected from 7 normal healthy controls from volunteers recruited at Massachusetts General Hospital. Blood was next subjected to a filtration step through a 0.8μm syringe filter to separate the serum. Aliquots were made and kept at -80°C. Exosomes were separated from the serum by ultracentrifugation. Briefly, 1ml of serum was diluted with PBS and centrifuged at 110,000g for 70 min at 4°C. The resultant pallet was treated with Dnase (Ambion, Austin, TX). Next, 700 μL miRNeasy lysis buffer was added to the tube to lyse the exosomes and release the RNA. The RNA was isolated following the manufacturer’s recommendations. Next, the RNA’s quality and concentration was assessed with Agilent Bioanalyzer and sent to Miltenyi to carry out the linear amplification and hybridization to Agilent microarrays. Agilent 4x44k Human Microarray chip was used for hybridization. The generated raw date was processed, normalized and analyzed using R/Bioconductor and GeneSpring Software. Results/Discussion: We were interested in the ability to categorize the samples based on only the expression array data for the two groups, without any prior knowledge of the identity of each sample. To this end, we filtered the data to exclude probes with constant intensity in all samples and included probes with high signal intensity (intensity >6 in minimum 4 out of 16 samples) and probes for which variation across all samples were high (Std. dev >0.8). The remaining 206 probes were analyzed with an unsupervised clustering which showed the samples were perfectly separated between control and GBM. Furthermore a principle component analysis (PCA) revealed a similar separation. This observation suggests the existence of significant differences in the expression profile of GBM exoRNA as compared to that of control exoRNA. We next sought to identify the genes that best separated the two groups by conducting gene-by-gene student t-test between the groups on all the genes in the full 44 thousands data set and corrected the resulting p-values for False Discovery Rate (FDR). While this method revealed the most statistically significant genes, we also looked for genes that in addition, had an intensity fold change of greater than 2.5. In doing so, both the statistically significant and biologically significant genes were identified. This analysis revealed 210 down-regulated genes and 24 up-regulated genes. Furthermore, a PCA was preformed using the selected upregulated and downregulated genes which once again displayed significant differences between the two groups with clear separation. Gene ontological anaylsis of the most down regulated genes using online tool DAVID revealed that the majority of the resulting GO terms are related to ribosome functions. Unfortunately, in a high background level of normal RNA from plasma exosomes, indentifying uniquely eleveated levels of tumor derived RNAs deemed challenging. GO analysis for the 24 upregulated genes were related to transcription which could be biologically explained by the high transcriptional need of growing tumor cells. However, we did not see a strong enrichment score. It should be noted that lack of GO association does not rule out the possibility that only a few individual genes might be strongly dysregulated Conclusion: This work is the first to report the ability to differentiate GBM patients from normal controls based on a gene expression blood test, and the first to report differential expression analysis using RNA extracted from exosomes/microvesicles isolated from clinical serum samples. We demonstrated that unsupervised cluster analysis of exoRNA from GBM and control can separate the two without any prior knowledge of the samples. Furthermore, using FDR corrected gene-by-gene t-tests we have identified the most significantly up and down regulated genes in GBM patients. The next immediate step would be validating the upregulated and downregulated selected genes using qRT-PCR techniques. While currently brain biopsy is the ultimate diagnostic procedure for GBM, these tumor exosomes and their constituent RNAs present unique genetic information about the tumor and can perhaps replace the need for biopsy in the future.