Model Systems: Disease Diagnostics

The Hood group has identified promising protein biomarker candidates and the corresponding peptides from each protein being synthesized for further targeted proteomic validation, in both primary glioblastoma multiforme (GBM) tissues and blood samples. Novel GBM subpopulations have been identified through a preliminary survey of single cell transcriptomes. Over the past year the Hood group has focused on:

Stratifying heterogeneous GBM cell populations through single cell transcriptomic profiling
A microfluidics platform is being supported by other funding to conduct single cell transcriptome analysis of glioblastoma cells. The cells are digitally stratified with heterogeneous cell populations to identify the tumorigenic cancer stem cell. Twenty four genes were selected from glioblastoma cell line U87 which exhibit changing transcript levels in glioblastoma (source: public databases), and qPCR primer pairs for these genes were obtained. Reliable amplifications of seven genes were obtained from thirty two single cells. By using gene expression levels of only three of the seven genes, the thirty two U87 tumor cells were able to be separated into at least five subpopulations. Since the three genes encode cell surface proteins with effective antibodies available, these subpopulations can be separated by subsequent functional assays. The goal is to analyze 1000 single cells.

Single cell blood biomarkers using proteomic technologies
A high throughput glycopeptide capture technique to enrich and identify cell surface proteins, and provide molecular information on the pathophysiology of glycobiology, has been developed. This glycoproteomics technique has been applied to mouse E14 messenger-embryonic stem cells, and 1230 N-glycosites among 3864 putative N-type glycomotifs in 544 glycoproteins were identified. Compared to other membrane targeted proteomics approaches and non-membrane restricted total proteome and transcriptome analyses, this approach enriched low abundance plasma membrane proteins more precisely.

A computational algorithm for mining public brain tumor databases
A computational approach has been developed to build a gene association network, without the assumption of linear relationship between genes. This network is specifically amenable for biomarker discovery. The approach has been demonstrated, not only for validating known cancer biomarkers, e.g., EGFR for brain tumor glioblastoma, but also for unveiling new classifiers separating cancer from non-tumor tissues at three levels: network, gene pair, and single gene.

The proteomics of the cell surface and secreted proteins to identify novel glioblastoma multiforme markers (resulted in an Invention Disclosure) In the last year the Hood group developed an approach that uses Single Reaction Monitoring (SRM) proteomic based assays for tumor characterization at both tissue and serum level. The approach has three essential components: generation of membrane peptide libraries from the brain cancer cell lines, selection of potential targets from the transcriptome database and the development of SRM assays for the detection, and quantification of target proteins from tissues and serum. With the aid of labeled synthetic peptides the expression of several cell surface proteins can be measured across GBM tissues in parallel.

Stem cell markers using proteomics technologies
Based on a previous quantitative proteomics survey of pluripotent stem cells, e.g. the embryonic stem cell (ESC) and the embryonic carcinoma (EC) cell, a set of proteins highly enriched in the undifferentiated pluripotent state has been expostulated for a possible role in the core gene regulatory networks governing ESC pluripotency. Among these is a transcription factor whose expression level decreases as the ESCs undergo differentiation. To place these findings in the broad context of cancer, better dissection of the pluripotent network in ESC is needed to better understand the stem cell origin of cancer.