Scientist Stories: Richard Young, Co-Founder of Dewpoint Therapeutics

It's delightful to be here today. I'm going to try and do the impossible, which is to teach you all about what transcription is. And to do that, I'm going to cover a broad spectrum of issues. Some of them may be confusing to you. Stick with me. And I promise you, you will have some new insights into life. But let me start by telling you that among my many passions, aviation and its history counts as one. And I am sure that from the earliest days we watched birds, we were interested in doing what they do, which is to fly. And if you've read a little bit about the Wright brothers' discovery of how to do this, you understand that they weren't just bicycle mechanics who accidentally stumbled into this. They were passionate scientists who did experimental science, developed principles, and tested each one, often to their personal suffering. They learned how to control what's called pitch, yaw, and roll. So pitch is up and down. Yaw is the odd feeling you get when you're in the back of the airplane and the pilot pushes the rudder a little too hard and the back end of the plane slides around. That often happens when you're landing in a crosswind. And roll is that business of the lateral movement of the plane with the wings going back and forth. They learned how to do that. They learned how to do it first. They flew their full-scale airplanes as kites, crashing them routinely, fixing them, altering things. All experimental stuff, and ultimately they succeeded before anyone else in flight. And only 10 years later, from the time of that discovery to 10 years later, there was commercial aviation. And today there are a million human beings in the air at this moment. That happened because of the convergence of experimental science technologies and most importantly concepts that gave us predictive capabilities. In aviation, you can sit down in front of a computer, create an airplane on the computer, and it will fly exactly, exactly as it does on the computer. Absolutely no variation from what the computer creates, what you simulate. Your pilot gets in the air in the brand-new plane, never flown before, and it will fly precisely as it flew in the simulator. In biology, we don't have that predictive capability. Just don't have it most anywhere. It's too complicated, too many layers of complexity. But in transcription, It's emerging. So I'm going to talk about transcription in normal cells. I'm going to talk about how it influences human traits and disease. I'll tell you about some new insights that have only been gained in the last few years. And then I'll introduce Ciros, which is exploiting some of these insights. Here's transcription. Your DNA blueprint is transcribed into a working version we call RNA. And that RNA can be used either as a template for protein synthesis to make the proteins, the machines of the cell, or it can be regulatory, as Phil mentioned with some of the regulatory RNAs he described. Here's a crystal structure of what it looks like if you have a transcription factor protein. Transcription factors are the proteins that are the molecular switches for genes. Bound to DNA. And this is a core enhancesome, first described by some of our colleagues over at Harvard. There's 7 transcription factors bound to this segment of 50 base pairs in DNA. This is at the center of what we call an enhancer. It's protein bound to specific sequences in the DNA. The way the process works is you form an enhancer with these transcription factors bound to DNA. These little round barrels represent what are called nucleosomes, part of the packaging material of your genome. That enhancer will loop around with cofactors, mediator, cohesin, other various proteins to contact an enzyme called RNA polymerase II, which is the apparatus that makes the working copy of RNA. And each of these steps is separated because there are many independent controls that occur here. In you and me, where we have 200 or so different cell types, what's important to know is that of the protein coding genes, and there are probably a similar number of non-coding RNAs that are produced in human cells, so about 24,000 of each, A subset are selectively transcribed. What I mean by that is the transcription factor of the cell make a decision what 30% of the genome to transcribe in any one cell type. And that creates cell identity. And an important thing we know about this is that only a few transcription factors are necessary to specify that identity. We call them master transcription factors. And I've made a list of these for factors where we can reprogram a particular cell type. So here's a skin fibroblast into a cell of interest that might be useful for, say, regenerative medicine in the future. So we— the question is, could we figure out all of the master transcription factors for all of your cells? And I think at a first step we've done that. I'll show you that. But I want to summarize this point. Transcription is DNA to RNA. These switches called transcription factors bind to specific DNA elements in each cell. Those enhancers actually physically loop and regulate their target genes. And in each cell, we have cell-type-specific master transcription factors that determine what that cell is, and in fact, determine the whole process of development from a single cell to your 30 trillion cells. All right, so what does this have to do with traits and disease? It has everything to do with traits and disease. So take a look at these two individuals, Willie Shoemaker and Will Chamberlain. They have what I consider the same genome, but they have variations. They vary about 1 every 1,000 base pairs. So do you. Your mom's dad genome— your mom's genome and your dad's genome in you varies the same. So the 2 genomes in you are as variant as you are to me. That variation leads to the probability of developing genetic disease of about 70%. 7 out of 10 of us will develop a genetic disease in our lifetime. And much of this is due to transcriptional variation. So let me just show you a little evidence for this. This just came out this year. It's a survey of variation in all of our transcription factors.. And Martha Bullock, a professor over at Harvard, Manolis Kellis here at MIT, got together and studied human transcription factor variation and came to the conclusion that differences in binding behavior of these transcription factors from one individual to another could be responsible for many traits. When I say trait, what do I mean? Your athletic ability, your— the rate at which you age, anything in biology. Is a trait. Most of the variation that we know of that causes disease or is associated with disease is in enhancers. Most of the diseases that are seen over at Children's Hospital in Boston are diseases where there's a variant that's affected a protein coding sequence, a protein itself. But the variants for the vast majority of adult diseases are occurring in these enhancers where the transcription factors bind. Cancer, to which about a quarter of us will succumb, is a disease that involves changes in gene expression. That's one of the hallmark effects of cancer. And the hypothesis, if we could understand all the changes that lead to gene expression, we might act on it. But to me, the common feature of cancer is this misregulation. What's remarkable in all of the diagrams in the textbooks is there's a tremendous amount of detail out here in the cytoplasm, out here outside the nucleus, that just disappears as we go into this little spot in the center, which is transcription. Our textbooks just don't have good pathways there. And this helps you understand why, if transcription is so fundamental to all these traits, to what diseases we acquire, to our rate of aging, and possible therapies of those that variation might be beneficial to us, why hasn't it been more aggressively pursued? Sued in the industry? Well, there are two key challenges. To me, one is the Orville and Wilbur Wright issue of having a lack of predictive models, not knowing the principles by which you could solve this problem in a computer. I'm going to make the argument that's close to change. And the other is historically, The transcription factors themselves have been considered undruggable. These two things are the reasons why we haven't gone beyond a small set of what are drugs for what are called nuclear receptors. So let me tell you why this is changing. First of all, it's changing because I'm going to show you I think we know what the master transcription factors are for most cells. I think we figured out some key dependencies. I'll tell you what that's about. I think we understand some of the— what are called— what's called the core regulatory circuitry, the pathways that are operative. And I think it's increasingly straightforward to drug this space. So we published this only a few months ago. We predicted for all human cell types that we can get our hands on the master transcription factors that, if they are masters, would, if we took a human fibroblast, turn that cell into another cell type. We tested this in two ways. One was we asked whether or not we'd predicted all the known reprogramming factors, and we had. But more importantly, we started with human fibroblasts and asked whether we could turn them into retinal pigment epithelial cells. These are cells that feed the rods and cones in your eye and are critical for preventing macular degeneration. And in fact, this may be the next regenerative therapy. There are trials in Kyoto to take human skin fibroblasts, turn them into what are called induced pluripotent stem cells, differentiate them into RPVs, put them back into the eye of individuals with macular degeneration, and initial results with one patient look quite good. So our test was, could we take a human fibroblast and turn it into a retinal pigment epithelial cell with the predicted master transcription factor? We were able to do that. Those cells had function in a transplant into to rat retina. What about enhancers? Well, a huge fraction of them have been mapped, maybe all of them. And in fact, what they do is they encompass about 70% of our genome. So the protein coding sequences are 2% to 3%. The vast majority of the genome that is not repeat, that is not the consequence of invasive viruses in our evolutionary history is enhancers that are used in different cell types. They've been mapped. Can we figure out pathways? I think so. The master transcription factors have this unusual property that they bind their own enhancers along with the other masters in this coordinated fashion that creates what we call an interconnected autoregulatory loop. In every cell type that we've investigated, these master factors, that's what they do. This is kind of the control room then that allows for reprogramming, that allows for control of these cells. And remarkably, and what's important for taking advantage of this therapeutically, there's some unexpected dependencies. So many of the factors that we've been studying for many years that are involved in this regulatory activity of bringing enhancers to promoters, that are involved in various kinds of modifications of the apparatus that packages the DNA, that are involved in modifying the DNA itself, I would have thought had we inhibited these activities that we'd simply kill the cell. But instead, It turns out over and over again, we find that there's certain genes that have built special dependencies on these factors so that we can get what appears to be a therapeutic index, a concentration of an inhibitor or an agonist that doesn't seem to harm healthy cells but can have a really striking, almost digital effect on a gene that's responsible for the pathology of a particular disease. I'm going to illustrate this by telling you a little about super enhancers, a discovery we made only about 3 years ago that went into CIRROS. And these are unusual kinds of enhancers. They're clustered enhancers that occupy a tremendous amount of DNA in the vicinity of each gene that plays a prominent role in that cell's identity, including the genes that make the master transcription factors. So it's occupied— these are so-called gene tracks that tell us where the material's occupying the genome. And in this case, it's occupying the genome around this microRNA precursor that's essential for embryonic stem cell state. I wondered why we evolved to have this. This is in every cell. And it turns out one of the reasons that we've evolved to have this is because it concentrates the response to signaling pathways. So each of our cells lives in a milieu of literally 100 or so signals from its environment. They can be local signals from the cell next door. They can be signals from far away. But it depends on that signaling environment in order to be who it is, in order to respond to stimuli and do its normal job. In every single signaling pathway we've investigated, the pathway is hardwired into the super-enhancer. And if you want to understand the mechanism for you aficionados, that there are transcription factors at the end of these signaling pathways that have very poor specificity. They depend on the master transcription factors to define their sites in the genome. And so the masters, which are specific to each cell type, are telling the signaling pathways to what genes to go, the genes that have prominent roles in cell identity. Importantly, what we're finding in cancer and in other diseases is that the cells that are pathological are developing or acquiring super enhancers over their lifetime. Genes that are responsible for pathology. This is especially true in every cancer we've looked at, at oncogenes that are drivers for those tumor cells. Cells evolve some means of developing a super-enhancer at that oncogene and optimizing not only its expression but its response to oncogenic signaling pathways. So it's a double whammy, and importantly, These super enhancers are special. They're special because those oncogenes often are producing a short-lived message on which those tumor cells are exquisitely dependent. And a small change in the activity of a super enhancer component, transcription factor, a cofactor, leads to the collapse of the expression of that oncogene. Which leads to the collapse of the tumor itself. This is not just a fantasy. This we've demonstrated with colleagues at Dana-Farber for a whole variety of lethal cancers with, with just one inhibitor developed in Nathaniel Gray's lab over at Harvard against a cyclin-dependent kinase called CDK7. Which seems to have this property of developing a special dependency in very, very aggressive cancers. Transcription, I think, struggle more. This is just a whole series of chemical targets for the variety of enzymes that are doing the chemistry of transcription. This is not just molecular Velcro. It's a series of chemical reactions that enable the transcription apparatus to function. And for every one of these modifications I show you in red, there is a blue enzyme that creates that modification, and there's a red enzyme that will remove it. And the catalysis then that is controlling gene expression expression, and is controlling super enhancers, engages many if not all of these various enzymes in a balance, all of which is druggable. So this is just by means of introducing siROS. Some of the concepts that I've described to you I developed together with Jay Bradner, an oncologist over at Dana-Farber. Who as of this year has become the president of Novartis Institutes for Biomedical Research. Nathaniel Gray, a really extraordinary medicinal chemist, developed drugs for Novartis and moved to Harvard. So we've got a guy in academia moving to Novartis and a guy from Novartis moving into academia. I think it's a good balance here in, here in Boston. Cambridge. We thought that what we were doing was going to lead to a convergence of predictive ability of attacking traits and disease that had never been done before. And we could not do this in our laboratories despite actually very generous support from Koch and Dana-Farber. And so we got together with Nancy Simonian. And an extraordinarily talented management group with exceptional experience. Eric Olson, largely responsible for Vertex's cystic fibrosis drug, and an incredible team of directors that came both from the investment community and from academia. And what we've done is to develop what we think is the first platform to exploit this convergence. So it has a variety of technologies, uses a variety of principles, and what we're interested in doing is solve patient problems. And in fact, I think one of the biggest challenges that the company has is it's such a broad opportunistic space that every once in a while it's not clear where we should put our focus. But the managers, of course, it's their job to— so I just thought I'd list here a few of the programs that CIRoS currently has. It has a couple of programs that have led to small molecule therapeutics that have a filed IND. That was done last month, and one that's planned for later this year. These are the first-in-class programs that, that Ciros has planned. And again, I just want to point out to you, um, that I think now there is a predictive biology in transcription. It's imperfect. It is not, I think, one where I could sit down in front of a computer like I can and design an airplane. And design the solution to any and all diseases. But it's remarkably close. I can't— I was very surprised that we could predict the master regulators for retinal pigment epithelial cells and to do that in a reasonably short period of time and then to travel to Kyoto to talk to Shinya Yamanaka, who won the Nobel Prize for reprogramming, and have them begin to integrate into their clinical program the idea that that set of regulators that we predicted would now be used to shortcut the path from individual patient cells, skin fibroblasts, to their own RPE cells, and that all that would happen in a 12-month period. That's pretty remarkable from my point of view. I think that's just the first line of evidence that we're on a tear. So thank you very much.