Dr. Sloan, thank you. Thanks very much. First of all, I'd like to thank the organizers for inviting us to present our work today. Here are my disclosures. I think everyone in this room knows that gliomas are very malignant and very difficult to treat and that there's been very little progress in the last three decades or so. There are a number of reasons for this. One of the most commonly cited reasons is treatment resistance, and there are a number of mechanisms cited for this, but two of the most commonly cited mechanisms are the phenomena of glioma stem cells and intratumoral heterogeneity, which we see both radiologically and histopathologically. So briefly to review the phenomena of glioma stem cells, these are typically cells that comprise somewhere between 2% and 10% of a tumor. They are notable for persistent self-renewal, sustained proliferation, and tumor propagation. So if you take a small number of these cells, implant them in an immunocompromised animal, you can recapitulate a tumor that looks very much like the original tumor. There are no definitive markers for this, but the most commonly used markers, at least in human gliomas, are CD133 and CD15. And this is a bit of a cartoon, but the idea here is if you consider glioma stem cells as the pink cells here, commonly used cytogenic therapies, such as chemotherapy and radiation, tend to kill the non-stem or bystander cells. It looks like the tumor gets smaller, but the stem cells remain and eventually begin to proliferate and the tumor regrows. Conversely, if we could theoretically figure out how to target the stem cells, we could theoretically get sustained remission. At least that's the theory, although no one's really proved that this works in humans, at least in the glioma literature. So when we think about radiological heterogeneity, as has already been cited here before, when we look at a typical glioma, what we see is an enhancing margin, which is hypercellular, typically has neoangiogenesis and cellular atypia and nuclear atypia. On the other hand, when we see the necrotic core, which is typically found in an RGBM, we see pseudopalisading necrosis as the hallmark. So our hypothesis was that radiologic and histologic intratumoral heterogeneity would be reflected by distinct subpopulations of glioma stem cells, and these subpopulations would be driven by distinct molecular repertoires that could be targeted alone or in combination. Oops, sorry. So what we did was we took patients who were undergoing surgical resection for gliomas, and as part of a resection, again with IRB approval, sampled these distinct regions. And in particular, today I'm going to talk about the enhancing region here that we see with the red arrow and the necrotic region, or region three here, that we see in the blue arrow. And I'm going to focus on those two. And then we characterized glioma stem cells from these subpopulations and attempted to inhibit distinct targets both alone and in combination, and both in vitro and in vivo. And when we did single-cell sequencing from these tumors, what we found is in the enhancing margin we had high levels of Olig-2 as well as VEGF and VEGF receptors and CD31, which, as we know, is representative of the pronal phenotype and angiogenesis, which is more or less what we expected in this region. On the other hand, when we looked at the necrotic region, we found high levels of CD44, CHI3L1, as well as upregulation of carbonic anhydrase 9, GLUT1 and GLUT3, which are high-vitidity glucose receptors, MCT1 and 4, which get rid of cellular waste. And, again, these are characteristics of the mesenchymal phenotype. When we tried to corroborate these using immunofluorescence, again, we found high levels of SOX2, Olig-2, and von Willebrand factor, suggestive of stemness and angiogenesis in the enhancing region. And, conversely, in the necrotic core, we found CD44, CKL40, and carbonic anhydrase 9 in the necrotic core, again, characteristic of the mesenchymal cell phenotype. We then looked at epigenetic profiling of these tumors. And we looked at two polycomb repressor complexes, repressor complex number two, which is driven by EZH2, and notable for the marker H3K27M3. And what we found is that that co-registered with CD33, I'm sorry, CD31, a marker of vascularity, and CD15 in the proneural phenotype, but we didn't see that marker at all in the mesenchymal cells in the necrotic region. Conversely, when we looked at the H2K119UB, which is a marker of the BMI polycomb repressor 1 complex, we found that that co-registered with the carbonic anhydrase 9 in the necrotic region, as well as the CD15 stem cell population in that region, but we did not find it, that marker, in the enhancing region. We looked, again, on westerns, and, again, we corroborated our preliminary findings, high levels of Olig2 and SOX2, as well as EZH2 in the proneural cells, which were in the enhancing region. And, again, in the necrotic core, we found BMI1 complex CD44 and YKL40 in the mesenchymal glioma stem cells that originated from the necrotic core. And we think this is clinically relevant because when we looked at our 96-sample tissue microarray, we found that the cells that had low levels of both these epigenetic repressors had very long survivors. So these are newly diagnosed GBMs treated with standard STOOP protocol, and the median survival, if they had low levels of these transcripture factors, was 28 months. On the other hand, the ones that had high levels of both did very poorly, with a median survival of about five months. And the ones that were upregulated in one or the other were intermediate at 18 or 19 months, respectively. So then we wanted to look at sensitivity to repression of these transcriptional regulators in vitro. So when we grow the proneural stem cells, and these are two different representatives of proneural stem cells from the enhancing margin, when we inhibited BMI1, we had some impairment of proliferation and moderate amount of cell death. When we inhibited EGH2, on the other hand, we had marked decreased proliferation and significantly increased cell death. Conversely, when we looked at the mesenchymal glioma stem cell, again derived from the necrotic core, we found the exact opposite. BMI1 was much more effective at slowing proliferation and killing cells than the EGH2 inhibitor. Similarly, when we looked at limited dilution, which is an assay of stem cell self-renewal, we found that looking at the proneural stem cells, inhibiting BMI1 modestly or moderately impaired self-renewal, but EGH2 inhibition markedly impaired stem cell self-renewal in the proneural stem cells, and conversely in the mesenchymal stem cells, the BMI1 inhibition was much more effective. We wanted to try to understand this in a somewhat physiological context in terms of cellular stress, so we looked at normal stem cells and the two different phenotypes of glioma stem cells, again from the same tumor, in conditions of normal glucose and oxygen, and then with stress of impairing glucose, impairing oxygen, or combining stresses. We found that in the normal condition, all these stem cells, normal stem cells and the two glioma stem cells, all fared quite well. Inhibiting glucose or inducing hypoxia favored the glioma stem cells, but there was not a significant difference between the two. However, when we inhibited glucose and induced hypoxia, clearly the mesenchymal stem cells had a survival advantage. Finally, we wanted to test this in vitro, so we took mesenchymal stem cells, transfected them with GFP, and proneural stem cells dyed with tomato red, and planted them in animals and then treated them every three to five days with inhibitors of either EZH2, BMI1, alone or in combination, and found that the control animals reproducibly died between 45 and 50 days. Inhibiting either one of these epigenetic repressors improved survival, but this effect was additive and possibly synergistic in combination. I'm going to skip this in the interest of time. But again, survival curves show that the proneural cells had a modest improvement in survival, inhibiting BMI1 only. They had an increased survival advantage if you inhibited EZH2, but when you inhibited both, survival was additive and possibly synergistic. And this is also true when you combine these two cellular phenotypes in the same animal. Inhibiting both transcriptional regulators was additive and more than doubled survival. So in conclusion, intratumoral heterogeneity correlates with differential glioma stem cell niche, micro-environmental cues, cellular, epigenetic, and genetic cues. At least two to three distinct glioma stem cell clones representing different tumor cell niches can be derived from a single patient, and targeting these distinct glioma stem cell niches in combination appears to be more efficacious than targeting any single one of them. I want to acknowledge my collaborators, Amber, Christa, or Fogel, who did a lot of this work, as well as my medical students, Ellen Krita, Raghavan, and Jim Robottom, as well as Ellie Barr, and Jeremy Rich, and Jill Barnholtz-Sloan, as well as my funding. Thank you very much for your attention. Thank you.

gliomas

malignant brain tumor

treatment resistance

glioma stem cells

intratumoral heterogeneity