 All right, is this on? Yeah, it seems like we're OK. Hey, John, thanks for having me here and thanks for the kind introduction. Really excited actually to tell this group about some of our work that's sort of connecting ENCODE-like assays to really clinically entrenched problems, and in particular in the area of cancer stem cells and glioblastoma. So glioblastoma is the most frequent human brain tumor. And despite resection of the tumor and radiation and chemotherapy, the median survival is very short and is almost uniformly fatal. So there's this lot of evidence that not all cells in glioblastoma are created equal, but rather there is this subpopulation of very aggressive stem-like cells in glioblastoma that propagate the tumor. So by the time you've resected this lesion here, there's actually stem cells, tumor stem cells, that have sort of infiltrated throughout the brain, right? And these seem to be resistant refractory to therapy. This is sort of a model of cancer stem cells. I don't mean to say that all cancers have stem cells, and it's a really controversial area, but in glioblastoma, there is evidence that not all cells are alike and that there are these purple cells that can both self-renew and propagate tumors. And for the purpose of this talk, I'll call them cancer stem cells or gliomastem cells, one can model these cells, and this is what we do. We actually take human tumors resected at MGH in surgery, and we can either expand them in stem-like conditions, in serum-free conditions, as gliomospheres. So these grow up spheres. They look a lot like neural stem cells that you might be isolating from stem cells or something. But these gliomospheres, they express CD133, which is a canonical marker of the stem-like state, and they're very aggressive. So you can put as few as 50 of these cells into a mouse brain and orthotopic xenotransplantation, and they'll cause a tumor. So these are tumor-initiating and they have tumor-propagating cells. You can take the same human tumor and you can expand cells in serum. And these grow a cell line, kind of the conventional type of cell line that we all might work with. These cells, though, they don't express CD133. And you can put hundreds of thousands or millions of these cells into a mouse, and they won't do anything. The mouse is just fine. These cells won't propagate tumors, and they lack stem-like characteristics. You can differentiate these cells. So you can put these cells into here, into the serum conditions, they'll differentiate, and they're no longer tumor-ogenic. So there's this directional thing that these are sort of a primitive state, and they will differentiate. But you can't go back. You take these cells in serum, these non-stem cells, they won't go back to this initial state. They basically won't proliferate in the gliomospheres. So this was very interesting to us, and led us to this question of whether we could reprogram these differentiated cells into tumor-propagating gliomospheres. So we basically set up this strategy where we were going to look at the enhancer landscapes and the transcription factors, and then we would sort of test the ability of these TFs to reprogram these cells into the gliomospheres. So whether we could identify something like a TF code for glioblastoma. So this is just very familiar now to this group, but this is basically, I think, a pretty striking picture that says, well, if you take three different patients' tumors and you expand them in the stem-like conditions, you can see that the SOX2 locus is very active, and there are these sort of H3K27 acetylated peaks in the chip-seq profiles that are enhancers in the SOX2 locus. But if you take these same tumors and you expand them in the differentiated cell line conditions, this is a very quiet locus. SOX2 is off, the enhancers are off. Conversely, if you look at some locus like BMP4, which is associated with differentiation, you see precisely the opposite. So in these cases, it's not which patients' tumor you're looking at. It's whether you're isolating and expanding a stem-like population or a differentiated population that is underlying at least this landscape. But I think you can see it's more striking when you now, each little row here is a different candidate enhancer or K27 acetyl peak in the genome. You can see there's a large group of shared enhancer-like elements shared between the stem-like cells and the non-stem cells. These have motifs for some cell cycle regulators, right? These guys are all proliferating. But what we're interested in here are these stem-like cell specific peaks. You can see these guys are all active in the stem cells and not in the differentiated cells. Again, motif analysis identifies HLH motifs and SOX motifs suggestive neurodevelopmental program. So then what we can do is we can look at RNA-seq data and we basically identified a set of about 20 different transcription factors that are predicted to bind to these motifs and that we posited might be explaining this different signature in the CSEs, in which are expressed specifically in the CSEs. So we went back and took the differentiated cells and we over-expressed each of these transcription factors one by one to ask whether they could reprogram, right? Kind of like the Yamanaka experiment. But we're looking for cancer stem cells. So basically if you take the differentiated cells and you try to expand them in sort of the spherogenic conditions, basically nothing happens and you can see that here. You get zero sphere forming cells. But if you now start adding these different transcription factors, there are a couple that kind of lit up and you could sort of use your, if you were very optimistic, you might guess that POW3F2 gives you a little bit of something growing in the plate. No real CD133, but we thought we might be moving in the right direction. So we started trying combinations of factors. Sox2, you get a little more sphere formation and so on. Ultimately by sort of iterating between additions of factors and then looking at the enhancer maps that we were getting, we were able to sort of come up with this cocktail and ultimately define four TS, POW3F2, Sox2, Sal2, and Olig2, that when you added them to the cell lines, they would reprogram into spherogenic stem-like cells that could propagate tumors. And this is just showing you a picture that you can take these reprogram cells, you put them in a mouse and they'll cause a tumor and kill the mouse and the tumor looks like, to some degree like a glioblastoma. Okay, so we'll call these induced GSCs. As you might hope, you can see that these CSC specific enhancer elements, after you've reprogrammed the cells with the core TS, they get K27 acetylation. So you're really reprogramming also the chromatin landscapes. The chromatin landscape in this case is a pretty good surrogate for the functional state of the cells. Another, I think, pretty critical data point here is that you only require transient expression of the core TS. So in some reprogramming or directed differentiation experiments, the factors are required continuously to maintain a new state. So that raises the question of whether you've really reprogrammed, whether you've really achieved an epigenetic self-sustaining transition. But here, what happens is we can show that the exogenous expression plasmids that we've expressed actually get shut off. We can tell this by looking at the three-prime UTRs. And the native loci for these four core TS, actually come on, you can tell this by looking at the RNA expression patterns. And you can also tell this by looking at the loci and you can see that the enhancers in the PAL-3F2 locus, the enhancers in the SOX2 locus, right? And all of them are coming on, right? So you sort of reprogram these guys and they're now self-sustaining. And part of how you know this now too is that you can now switch these IGSEs back into serum and they'll readily differentiate and they'll totally shut off the core TFs. So it's telling you you have an endogenous regulatory program now that you've got going. So how might this be useful to understand glibostoma? Well, for one thing, we could now go back and map the direct targets of these core TFs, which we, well, PAL-3F2, SOX2, SOX2, OLG2 by chip-seq. There's a great deal of overlap. They're sort of binding to each other's loci and we think sustaining their expression. I can tell you that all four of these guys are essential for tumor propagation of the GSEs. But more than that, when you now infer the downstream targets of these core TFs, here are the core TFs and you can use the binding patterns and expression patterns to infer their downstream targets. I can tell you that essentially everything we've looked at and then tried to knock down in the GSEs has turned out to be an essential factor. So this is sort of like, you know, there's various proposals out there, whether you use super-enhancers or specific TF signatures or what, here's an example where by knowing the core TFs that are driving the circuitry and examining their direct downstream targets, you now have a set of regulators that seem also to be essential for the tumor-propagating state. We were particularly interested in ASCAL-1, which is a transcription factor that it's a neurodevelopmental TF that's activating wind signaling in the GBMs. As well as the R-Core II histone dimethylase complex, which actually three or four different members of this complex are direct targets of the core circuitry. So in a sense, knowing this core circuitry, now, like knowing the core circuitry of IPS cells, told you a lot about the pluripotent regulatory circuitry, I think we've learned a lot about the GSE circuitry by using sort of familiar ENCODE type approaches and combining them with reprogramming approaches to define the circuitry. So, but I kind of told you this was at some level, it's a very fundamental project, I think, but it's clinically driven. We're doing this because we want to understand the nature of these cells that are propagating tumors in humans. So the question then is, what's the status of these core TFs in primary glioblastomas? So here we wanted to, we actually did a series that we did flow cytometry as well as immunofluorescence analyses. We labeled antibodies that recognize each of the core factors. We also used CD133 antibody. And we stained slides of primary glioblastomas from patients. And we can see there's a great deal of heterogeneity, as you might expect, glioblastomas are notoriously heterogeneous tumor, variable expression of the core TFs, but there clearly are a subpopulation of cells, the pink ones here, that we know are co-ordinately expressing all four of the core TFs in the tumors. And when we do flow cytometry, we look at all the markers. These guys, I can tell you are also CD133 positive, which is one of the markers that enriches for tumor propagation. So it's very likely that there are surrogates within primary tumors that look very much like the stem cells that I just sort of functionally described in our model. And such as shown here. We've also looked at single cell data in another way, and that is by single cell RNA-seq. So by single cell RNA-seq, again, of primary tumors, one can learn a little bit about the regulatory programs in the primary tumors. And this is a short talk, and I'm trying to focus on regulatory elements a little bit, so I'm not gonna give you much on this single cell RNA-seq. Except I wanna tell you one interesting property. We looked at a few hundred cells from a few different glioblastomas. We saw a lot of heterogeneity within each tumors. There's a variety of transcriptional programs. These tumors at the RNA level, as you might expect from that last picture, are very heterogeneous. But what, we wanted to take this opportunity to ask about the stem-like transcriptional programs and its presence in primary GBMs. So we defined a stem-like signature based on, which is composed of genes that are expressed at low levels in these differentiated cell lines, okay? But are highly expressed in the GSC, the stem-like model that I talked about earlier in the talk. And then we asked, what is the expression pattern of these genes in single cells from a person's tumor? So this is just one tumor from a patient, and we looked at about 70 different cells. And I think you can see that there is a population of cells at the right side of the slide here that are highly expressing a lot of the genes in the signature, okay? So together with the last picture, where I said, well, we've got the core TFs, and we've got a protein level analysis. So we've got stem-like cells that emulate our model in tumors. Now I'm telling you that we've also identified cells with a transcriptional signature. It looks a lot like what our model looks like, okay? I'm telling you all this, but now I'm gonna tell you about a little caveat, okay? The problem was that if you take these cells, and you now ask about their cell cycle signature, turns out they're not cycling. They're totally dormant and quiescent, okay? So we have this model of CSEs with the right neurodevelopmental programs that seems to be emulating something in a tumor, yet our CSEs are proliferating like mad, like probably most of the cultures that people work on in this room, because if a culture's not proliferating, it's hard to work with. But the real in vivo GSEs with this developmental program seem to be out of cycle and pretty much quiescent and dormant, okay? So this presents a bit of a challenge to us as we think about modeling and targeting, okay? I'm gonna tell you now how we're trying to get at this population. I'm gonna come back to the population in a second, but I need to just sort of talk for a minute about therapeutics, okay, before I get back there. Okay, so receptor tyrosine kinases. So RTK inhibitors, of course, have been a dramatic advance in cancer research and Brian Drucker and Charles Sawyer's and others have shown that this miraculous inhibitor of BCR-Able practically cures CML. This RTK inhibitor basically ablates this whole population and this whole tumor. So this led to a lot of excitement and now RTKs are being used in many settings. GBMs, gliomas seem like a perfect setting to use RTK inhibitors, right? Look at the genetics of gliblastoma. You've got EGFR amplification, a huge proportion, you've got PDGFR amplification and others. Altogether, about 67% of brain tumors, gliblastoma brain tumors, harbor a amplification of an RTK. So of course this led a lot of excitement to then go in the clinic and try RTK inhibitors that could target these genetic lesions and hopefully have an impact on gliblastoma. They haven't done anything in the clinic, though, right? They haven't changed the course of this disease whatsoever. So we wanted to understand this better and you'll understand where I'm going in just a minute. But we took one of our stem cell models, one of these proliferating GSC models and we took in particular one with a PDGFR amplification. And we treated it with the satinib, which is a really good PDGFR inhibitor. You can see that the satinib, this particular population is very sensitive to satinib, pretty low concentrations of this inhibitor kills the cells. This is in contrast to a couple other glioma lines. So if you take a glioma stem cells from an EGFR amplified one or a MIC amplified line, they don't care about PDGFR, so they're insensitive to satinib, okay? But look at this. Even in these cells that are supposedly totally dependent on PDGFR, there is this sort of window here where about five to 10% of the cells just won't die and they remain viable. If you culture them out for a while, they actually start proliferating very slowly, okay? And they're quite insensitive. We call them persisters. I'll call them GSC8 persisters or whatever persister stem cells. They proliferate slowly, but they're pretty much impervious to very high concentrations of this satinib. So this cell within this genetically amplified PDGFR dependent population, these cells can acquire a different state that is independent of the RTK. So this is just showing you that we take the naive cells and we can culture them in the presence of this satinib and over time we kind of acquire this slow-cycling population, okay? This is a little bit analogous to persisters that have been defined in lung, melanoma, breasts in other settings. One interesting thing is it seems to be a reversible because we can remove the pressure of the drug and these cells will go back to their initial state where they're proliferating very rapidly and PDGFR dependent, okay? It's kind of a reversible persister state. Okay, this I think was pretty interesting so we can look at the regulatory elements of the different cell states, right? So this is all from the same GSEs. Here are the naive sort of proliferative GSEs and here are the persisters, right? That after they're going, they're cycling very slowly, they're PDGFR independent. You can see enhancers that are common in various different patterns and then there's this big group here that seems to be quite a bit stronger in the persisters. I think what's kind of striking about this is if you go back and look at the tumors, the tumors actually look much more like these persisters. So in a sense, these RTK-inhibited persister cells that are very slow cycling actually look a lot more like the primary tumors of the GSE, the primary glioma tumors, okay? So that we might be doing a better job of emulating them. We want to go back and understand a little bit their circuitry and their regulation. On many levels, these persister cells look much more like tumor cells and they look in particular like those tumor cells that are quiescent or non-cycling. And I'll just give you one example that we were kind of excited about that's I think relevant to this conference. This is kind of a complicated graph, but basically we're comparing the primary tumors, different cells in the primary tumors and at the top here are genes that are associated with the quiescent tumors. They're anti-correlated with cell cycle. Here we're sort of comparing the naive GSEs to the dormant GSEs, and here are genes that are high in the quiescent drug tolerant ones that we selected. So things up here are high in the dormant non-cycling tumor cells and they're also high in the drug tolerant non-cycling GSE model. What are the genes? Well, we were pretty struck that this is just packed full of KDMs, lysine dimethylases, genes that encode these enzymes that remove the methyl marks off of histones. You might notice that in contrast, the methyl transferases actually are down-regulated in these drug tolerant cells. We thought this was pretty interesting here that these KDMs are being up-regulated. In fact, KDMs have been, particularly at KDM5 here, has been previously implicated in drug tolerant cancer cells. Let's see. I'll tell you just, I don't have a slide here, but we spent a while looking at KDM5 actually and it turns out that KDM5 does not seem to be the critical player here. We can inhibit KDM5, we can knock it down, we can actually pharmacologically inhibit it, but the cells don't care. The persister cells don't care. So they don't seem to be dependent on KDM5. This is the one that removes lysine form ethylation. Okay. But there's another set and there are these KDM6 and KDM3 families that remove repressive marks that actually attract, seem to be a little more interesting. I'll tell you why. First of all, if you look at K27ME3, repressive sort of polycomb peaks in the naive cells, you see pretty good signal. And then as these cells transition into this drug tolerant persister state, there's a global erasure of K27ME3. You can actually see that, let's see, if I'm gonna highlight it, you can see that here in the SOX2 locus where you see pretty good signal for K27ME3 in the naive and then it kind of goes away a little bit in the persisters, but it's a very global effect. So you're erasing this repressive mark. You're doing the same thing, maybe equally dramatically to K9ME3, another repressive mark, and I think you can see a region here where you've got some pretty strong K9ME3 and the persisters sort of have erased that and it's very flat. There's a KDM6, which is the one that ruins K27ME3, there's an inhibitor, GSKJ4, it's not a perfect inhibitor, but it actually seems to work pretty well because if you look at the naive cells compared to the persister cells and then the persister cells treated with the drug, the K27ME3 comes back up. So the drug sort of blocks this demethylation effect. And actually what was really striking is that these persisters are really in red here. You can see that these long-term persisters are very sensitive to GSKJ4. So they're very sensitive to an inhibition of the KDM6 at very low concentrations, suggesting that this erasure of these repressive marks may be essential for these cells to transition into this drug-tolerant quiescent state. So why do you need to do this? What's going on? Well, the persister cells have an altered developmental program, as you might suspect. So this is just looking at comparing naive to persister for K4 methylation and for K27 acetylation and for K27ME3. And we see a number of loci where there is K27ME3 sort of silencing genes in the naive cells. And then this gets erased in the persisters and up comes K27 acetyl, up comes K4ME3. We looked in general at the pathways. There are a couple of things that come on. One is very primitive neurodevelopmental genes that are associated with primitive neural stem cells. The second is notch target genes. And in fact, there's some very interesting literature in the neural development field that suggests that in radial glial development and very primitive embryonic development of neural stem cells in their transition to transient amplifying cells, notch and RTKs are sort of antagonistic in having a complex interaction. So the most primitive neural stem cells that are quiescent have notch. The transient amplifying cells get the signal from EGFR that causes them to rev up and proliferate a little bit. This is happening in a very narrow window of development but we think this may be emulated in glialostoma. And in fact, if you now go back and think about notch signaling, you can see that notch signaling is actually activated in the persisters. You can just do Western blots to stain for active form of notch. You can see old gene expression signature associated with notch signaling. And the persister cells but not the naive cells are very sensitive to gamma secretase inhibitors which target notch signaling. So this kind of leaves me here as I'll wrap up with this picture here. We think that the persister cells are sort of emulating a quiescent neural stem cell state whereby we're actually starting with these naive GSEs that are proliferating an RTK dependent but through demethylation and perhaps a resetting of their epigenetic state, we get these persister cells that are very slow cycling. They upregulate very primitive stemness genes and they're notch dependent. As I mentioned, this parallels an event or sort of inversely parallels an event in normal development whereby neural stem cells that are quiescent have notch signaling and switch to an amplified state with RTK signaling. We're very interested in how KDMs are working, we think, to sort of help allow the transition between states or back to this primitive state and allow particularly a ratio of repressive marks or repressive state can allow the right enhancers to come on and sustain the state. We're also very interested in whether now, given the observation that the primary tumor cells look a lot more like this persister state and that they're very slow cycling, suggest that perhaps notch inhibitors or better yet inhibitors of some of the chromatin enzymes might synergize with RTKs and give you a better chance of targeting some of these glioma cells. So my final slide, I've shown you that enhancer landscapes distinguish these tumor propagating GSEs from conventional cell lines that don't propagate tumors and don't have this functionality. I've shown you that we can identify core TFs that are sufficient to reprogram GSEs and these sort of define, their targets define essential drivers of the stem-like state. I've also told you about that there's a large-scale erasure of repressive marks seems to allow GSEs to adopt a more quiescent drug-tolerant state that we think is emulating the primary tumors and how this may have implications for glioma therapy that targets this other population, which we think is completely missed by existing therapies that all of them, because they all target proliferative cell states. Knowledge, Maria Suva is now a faculty at MGH formally, a postdoc in the lab, led the reprogramming work, Brian Lau, Jim Severs have been working on some of this, the persister cell story. The single cell work was a collaboration with the Thai Taroosh and Aviv Reghev at the Broad Institute in the Claremann Observatory and with MGH neurosurgery and Chekepstein and Noam Shores at the Broad have been critical for a lot of the chromatin mapping work that we do. Glad to take questions. Dependent pathways for these GBM cells, are you saying that now there is an epigenetic understanding or basis for oncogene addiction that you're able to target with? Some of these drugs? Let's see, so I think there's a point to this, that when cells become dependent on oncogene for a particular state, they will respond favorably. And I think that model holds very well for that 8% of cells that are proliferating rapidly in the tumor. The model doesn't hold so well for the other cells in this tumor that are out of the proliferative state. CML, CML oncogene, it works perfectly. In CML, 99.9% of this chronic leukemia. 99.9% of the cells are in this proliferative state, are driven by BCR-Able, and so if you give them Gleevec, you practically cure them. There's about 0.1% of the cells still in the bone marrow. You probably need to give them Gleevec for a while until those sort of peter out, right? But it's a cure almost. Gleevec is almost the opposite, right? That fraction of oncogene-addicted cells is that 8% that's proliferating, where there's 92% sitting in this dormant reservoir or something like that. So the model's great, it just depends on the tumor, and a lot of tumors have a very high fraction of dormant cells in this other state, other epigenetic state that can transition, and that poses a huge problem to using RTK inhibitors on their own in the clinic. You have to combine it. Great talk, quick question in regards to the state of your persistent cells. Do you think they're under epigenetic instability? So the global reduction in K27ME3, for instance, do you think that that's true of all cells? Or if you do IHC, you see that there is still a variability in some having high levels, some having low levels, and by the chip, it all looks down because the signal from all over is negated. Could it be that you have a massive amount of heterogeneity? I don't think that, I mean, it's certainly possible, technically, and in the tumors, I can't even say what they look like in the tumors. But in the actual persistent population that we're studying, where I showed those things being erased, I think if there was a lot of epigenetic variability from cell to cell, there'd be some regions with middle levels and things like that, but in regions that are really repressed in deserts where K27ME3 is sitting, it shouldn't be coming on and off and it's gone. So the fact that those deserts also get totally erased and there's no rationale that you think some population of cells is expressing in the desert, tells me it's a global effect hitting all the cells in our model. Again, I don't know what's going on in the tumor, it's very hard to work with single cells from the tumor in terms of understanding their chromatin, but it's a good question, I mean, we need to try to understand something. Great, thanks.