Hello, everyone. My name is Leo Kim, and I'm a sixth year MSTP student at Case Western Reserve University. I would like to thank the organizers of this meeting for this opportunity to share some of my work that I've done during my PhD training in Dr. Jeremy Rich's lab at the Cleveland Clinic and UC San Diego. In our lab, we study the Kendo stem cell populations in glioblastoma, or GSCs. And as in other stem populations, they're defined both functionally and molecularly. We characterize them functionally through their ability to self-renew and their ability to form spheres in in vitro culture conditions, as well as propagate tumors and intracranial xenograft models in immunocompromised mice. GSCs also express stem cell markers, such as CD133 or CD15, on their surface, as well as the stem state-defining transcription factors, such as Olig2 and SOX2. They also have the ability to differentiate along multiple lineages, such as the oligodendrocytic or astrocytic lineages. So then in this study, we wanted to know how GSCs feel their epigenetic plasticity as they differentiate along multiple lineages. To do that, then we performed heavy isotope metabolite tracing experiments using carbon-13-labeled glucose or glutamine tracers. We first took primary tumors from patients after dissociating them. We enriched for the Kendo stem cell populations using the marker CD133, which enriches the GSCs from the differentiated glioma cells, or DGCs. After we expanded both populations to about a million cells, then we exposed them to near-stem cell media for 48 hours to equilibrate the two cells, and then performed the tracing experiment over 24 hours using the carbon-13-labeled glucose. Then we prepared the cells for mass spectrometry. Using the metabolic profiles from those experiments, we could easily distinguish GSCs from DGCs based on differentially enriched metabolite levels, as well as carbon-13-labeling patterns. And in the carbon-13-labeling patterns, we found out that the GSCs undergo rapid glucose uptake, which undergo glycolysis to generate pyruvate, and through the Vorburg effect. Unlike other kinds of cells, the pyruvate from glycolysis enters the T-C cycle in the reductive manner, or the reverse pathway, which allows for rapid generation of aspartate, which is an important precursor for nucleotide biosynthesis. Given the metabolic profiles from those experiments, then we integrated them with the transcriptional profiles we have of those same cells, and we found that the largest metabolic network supported by the genetic program enriched in GSCs was the malate-aspartate shuttle, which takes the T-C cycle intermediate malate into generating aspartate, and which in turn gets fed into nucleotide biosynthetic pathways. The malate-aspartate shuttle transfers cytosolic electrons into the mitochondria in exchange for T-C metabolites, allowing for flexible utilization of cellular metabolites. Analyzing the TCGAA dataset, I found that the malate-aspartate shuttle component expression patterns suggested directionality, where TCA cycle intermediates are fluxed into the cytosol in exchange for electrons. And this pattern can be confirmed at the single cell level using single-cell RNA-seq profiles from bulk tumors, where we found the specific enzymes, such as the malate dehydrogenase-2 in the mitochondria, or the aspartate glutamate transporter, GLAST, are strongly enriched in tumor cell populations, while the other components were not. And within the GSCs hierarchy, we again saw a directionality inferred from the gene expression patterns of the malate-aspartate shuttle components, and these gene expression patterns are epigenetically maintained at the enhancer level as seen by a 3K27 acetylation ChIP-seq. And when we knock out MbH2 in GSCs, we observed a significant impaired in vitro proliferation, as well as impaired sphere formation in vitro. And MbH2 knockout also reduced Olic-2 expression in GSCs. There's a small molecule inhibitor against MbH2 reported, and using that against GSCs, we again saw a strong impairment of the in vitro proliferation, as well as sphere formation. To further investigate the metabolic consequences of MbH2 knockout, we performed the glucose and glutamine tracing experiments again, and we found out that MbH2 knockout significantly decreases alpha-ketoglutarate levels in the cells. And when we look at the RNA-seq profiles of these cells, we again interestingly saw that RNA methylation was the most affected programs in these cells. And indeed, when we look at M6A RNA methylation states of GSCs with MbH2 knockout, there's a significant reduction in these RNA methylation moieties. So based on these data, we hypothesized that since alpha-ketoglutarate is an important code factor for RNA and DNA demethylases, that the increased alpha-ketoglutarate levels after MbH2 knockout in GSCs could be affecting these pathways. Then we tested this hypothesis by supplementing GSCs with exogenous alpha-ketoglutarate. And indeed, we observed a concentration-dependent reduction of 5MC and M6A RNA methylation states with increasing alpha-ketoglutarate levels. To better understand the genetic consequences or gene expression consequences of this phenotype, we performed RNA immunoprecipitation followed by next-unit sequencing, or RIP-seq, of GSCs compared to normal stem cells, neural stem cells, and found that both these stem populations maintain important transcription factors using RNA methylation, where GSCs have high methylation of important transcription factors like olig-1, which correlates to high expression of olig-1, and similarly in olig-2. And when we knock out MbH2 in GSCs, then there's a strong reduction of M6A RNA methylation. And this reduction in methylation then reduces olig-2 expression at the protein level by destabilizing this RNA molecule, which can be evidenced by the reduction in half-life. To better investigate the clinical relevance of MbH2 expression patterns in GSCs, I examined available data sets that link gene expression patterns with predictive therapy responses and found that high MbH2 expression across multiple cancer cell lines predicted poor response to satinib, which is a Src inhibitor involved in many RTK signaling. And work by Brad Bernstein's group has already investigated how GSCs remodel epigenetic landscapes in response to satinib. And further mining the data, we examined the MbH2 locus also undergoes significant epigenetic remodeling in response to satinib, and in exposure to satinib, there's increased MbH2 expression in GSCs. To examine whether MbH2 inhibition synergizes with the satinib treatment, we knocked out MbH2 or inhibited with small molecule inhibitor, and indeed found that GSCs in in vitro conditions exhibit a therapeutic synergy between the satinib and MbH2 inhibition. And for in vivo phenotype, we observed an MbH2 knockout extended survival of tumor-bearing mice. To examine whether the in vitro synergy between MbH2 inhibition and the satinib can be achieved in vivo, we treated mice combinatorially with the satinib and the MbH2 inhibitor or monotherapy with each agent alone. And indeed, we observed that the combinatorial approach achieved the best tumor control compared to monotherapy or vehicle control. So in summary, we came to this model where GSCs maintain high levels of MbH2 expression and malate aspartate shuttle system to promote not only the availability of aspartate, which leads to increased macromolecule biosynthesis, but also reduces the alpha-ketoglutarate levels low so that RNA can be methylated, promoting the stability and expression of important stem transcription factors. But when we impair the system with a molecule inhibitor or genetic tool, then we can reduce the availability of biosynthetic precursors for cellular proliferation, as well as increase alpha-ketoglutarate levels, which promotes demethylation and instability of important stem factors. So with that, I would like to thank all the lab members, collaborators, and the funding sources. And thank you very much for your attention.

Leo Kim

glioblastoma stem cells

epigenetic plasticity

metabolic profile

malate dehydrogenase-2