Scientist Stories: Dave Schaffer, Pioneer in Bioengineering and Gene Therapies

Now, speaking of award-winning faculty, it's my great pleasure to introduce today's homecoming speaker, Professor David Schaefer. Dave received his bachelor's degree in chemical engineering from Stanford. We all make mistakes. But then he began to atone by receiving his PhD in chemical engineering from MIT with a minor in cell and molecular biology. He then went on to postdoc at the Salk Institute for Biological Studies in La Jolla, where his work focused on stem neural stem cells, and viral gene delivery vehicles. In 1999, Dave joined the Berkeley campus with a joint appointment in chemical engineering and the Helen Wills Neuroscience Institute. The Wills Institute, as you may know, is a collaborative campus research unit that investigates fundamental questions about how the brain functions. Very important interdisciplinary research area that is growing rapidly, not only on the Berkeley campus, but throughout the world. Dave is currently a full professor and now holds joint faculty appointments in chemical and biomolecular engineering, the Helen Wills Neuroscience Institute, bioengineering, and molecular and cell biology. In addition, since 2007, he has been affiliated with the Berkeley Stem Cell Center and is currently the director of the Stem Cell Center. Now Dave occupies a unique position within chemical engineering really as a pioneering leading, a leader in both stem cell biotechnology and virus-based gene delivery vehicles. And his work includes both fundamental and clinically oriented elements. His stem cell research is dedicated to understanding the biology of stem cells and expanding their therapeutic potential, whereas his work on gene delivery targets effective ways to control how stem cells develop into specialized cells that perform all the necessary functions of life. And Dave's program combines these two avenues with beautiful synergy because virus-based gene delivery is one of the most promising approaches to introducing genetic materials for modulating stem cell behavior. Furthermore, Dave has very effectively bridged basic research and commercial applications and technology through his involvement in successful startup companies. He's been recognized with numerous awards and accolades, including the Glenn Award for Research in Biological Mechanisms of Aging, the AICHE Excellence in Teaching Award, and the department citation for excellence in teaching, among many others. He's also elected— or has been elected a fellow of the American Institute of Medical and Biological Engineers. Finally, Dave is a superb mentor. Which I have seen many times firsthand as one of his many collaborators, and is evidenced by the excellent graduate students and postdocs from his lab who are already well underway in their own careers in the biotechnology industry, as well as in academia. And finally, Dave happens to be an exceptionally nice guy, and we should congratulate him today because it happens to be his anniversary. So thanks for taking the time on this very important day, Dave. His presentation is titled Evolving New Synthetic Viruses, Sparking the Gene Therapy Revolution. So please join me in welcoming living proof that nice guys can and do finish first, Professor Dave Schaeffer.

Thanks so much, everyone. It's terrific to see all of you here on a bright and early Saturday morning. So I wanted to describe some of the work that's been taking place in our lab here at Cal over the past 15, 16 years or so, but I also wanted this to be kind of a two-way discussion rather than a one-way lecture. So if anybody has any questions throughout the next 40 minutes or so, please don't hesitate to raise your hands at all. So I wanted to start by talking about healthcare and how the healthcare field has been evolving in our country really over the past several decades. And as we all know, you know, the cost of healthcare has been progressively increasing. So if we take a look at the fraction of the GDP, the total value of goods and services produced in our country every year, around 5% of that just 50 years ago was spent on healthcare. And that number has been creeping up ever since so that we're now at about 18, 19%. And if you take a look at a few decades into the future, this is going to reach around 50% of the GDP by the year 2050 or so. So this is obviously not sustainable. It's going to break the bank at some point, and we really need to be able to come up with better and better healthcare solutions to bend this curve and hopefully begin to take it down. So if we take a look at some of the underlying reasons for the shape of this curve, some of them are the fact that the nature of healthcare problems in this country has been progressively evolving. We've moved away from acute healthcare conditions like bacterial infections and viral infections, things that can be solved with a course of antibiotics, and all of a sudden now we're facing these longer-term chronic conditions, such as diabetes or cardiovascular disease or neurodegenerative disease. And these are chronic long-term conditions for which we currently don't have cures. We have treatments, we have long and somewhat expensive treatments, but we don't have cures. So if we take a look at the molecular therapeutics that are available to be able to treat these kind of conditions, they fall into a few generations of, of molecules. So small molecules, things like Advil and aspirin, have been around for almost a century. Protein therapies, things like insulin or monoclonal antibodies, have also been around for a number of decades. And what our lab and our group is very excited about is what we view as the next two generations of healthcare products. Products based upon DNA or genes and products based upon cells or genomes. So before I describe to you what gene therapy is, this is gonna be the focus of my talk today, I wanna provide a little bit of background about biology. So this is the central dogma of biology, and if you take nothing else away from this talk, this is a really important set of discoveries in the field of biology over the past 50 years, and this represents about a half dozen Nobel Prizes on this slide. So first of all, we know that the material that stores information in our bodies is DNA. This is the permanent information storage medium, and our DNA within every single cell contains about, turns out, about 2 gigabytes worth of information. So considerably less than my cell phone stores, right? So this DNA ends up getting transcribed into messenger RNA that gets sent out into our cells and ends up docking with molecules called ribosomes, which turn this into proteins. And proteins and messenger RNA are kind of temporary residents of our bodies. They degrade rather quickly. DNA is a permanent part of the body. Now there are many situations in which we can identify a protein that could be able to treat a disease. Unfortunately, if you try to use a protein as a therapy, these stick around in the body for a matter of hours or perhaps days if we're lucky. So we need to continuously redose every single day or every single several days if we want to rely upon a protein as a therapy. But if we can deliver the DNA, this becomes a permanent part of an organ or tissue in the body. So we can begin to think about single administration, very potent, long-term therapies to be able to, for example, treat these chronic conditions I described to you earlier. So, you know, sounds great, you know, sign me up. You know, why can't I go to my general practitioner and get a prescription? And there are, of course, challenges within the field, and there's this old joke in the field of gene therapy that dates back decades that there are only 3 problems in gene therapy: delivery, delivery, and delivery. So DNA is really hard to get into cells at a really high efficiency, and the reason for that is that most time when foreign DNA is coming into our bodies, it's associated with things like bacteria and viruses, and our our bodies, our immune systems, have really mounted a series of barriers or challenges to prevent that from happening. So it's kind of hard to convince our body all of a sudden that we want to stick a piece of medicinal DNA into enough cells to have a therapeutic benefit. So this delivery problem has been really the Achilles heel of the entire field over the past 30 years. So investigators came up with an idea several decades ago of why not use a virus to actually carry pieces of DNA inside of the cell? Viruses have evolved to do this over millions of years. It's how they make a living. It's part of their life cycle. So maybe we can actually trick a virus into carrying a piece of DNA medicine into a cell rather than carrying its own DNA. So over the past couple of decades, people have really converged on some very, very good viruses. And this is a particularly successful bug for reasons I'll describe to you. And interestingly, probably most people haven't heard of it, and the reason for that is that It doesn't cause disease. The majority of you have already been infected with it, typically by the time you started going to daycare, age 5 or so, but it's never caused human disease, and as a result, it's a very harmless virus, a really good starting material to turn into a therapy. So it's a very, very small virus. It only has 2 genes inside of it. This gene encodes some proteins that replicate the viral DNA, and this gene is really gonna be the focus of my talk. This encodes the gene delivery vehicle. So specifically, it encodes some proteins, 60 copies of which self-assemble to form this protein shell. This shell then becomes loaded with a copy of the viral DNA, and then when introduced into a host, it travels through the tissue and then delivers the DNA to the nucleus of the cells. So you can convert this into a Trojan horse, trick it into delivering a piece of therapeutic DNA, by quite simply taking the two viral genes, cutting them out of the viral genome, which liberates space for the insertion of any gene of interest, a piece of therapeutic DNA, and then this protein shell, this delivery vehicle, becomes loaded with this gene of interest, and now you have potentially a therapeutic gene delivery vehicle. So this has a number of advantageous properties. 90% of you have already been infected with the parental virus, the original virus, and never noticed it. It doesn't cause disease. In addition, it's been used in a variety of clinical trials, over 120 to date, It's somewhat efficient in these trials. The natural versions of these viruses are somewhat good at carrying DNA inside of cells to the point where it's beginning to succeed, and I'll show you a couple examples. In addition, once it ends up releasing its DNA in the nucleus of cells inside of our tissues, that DNA is stable. It's maintained for years or potentially decades, so we can really begin to think about this idea of one administration, therapeutic benefits for years or potentially the rest of our lives. And, like I mentioned, it's beginning to enjoy some efficacy in trials. So a couple examples of this in the New England Journal of Medicine, which is kind of the premier magazine or the premier journal in the medical field, has published several publications showing really strong successes in this field. So first of all, hemophilia B, which probably most of you have heard of. It's a bleeding disease. It's caused by mutations in a factor, a protein, that is responsible for blood clotting. So as a result, hemophilia people can't clot blood, and you're familiar probably with bleeding episodes that result. So this is caused by mutations in a gene. In addition, this is another disease. These two, in fact, are blinding diseases where these kids have been unfortunate enough to have inherited from their parents defective copies of genes that make sight or vision possible. So again, we have a broken protein. Can we actually deliver a correct copy of the DNA to repair that broken protein or replace that broken protein. And all of these successes to date, and I'll show you a couple examples of these, have really culminated recently with the very, very first regulatory approval by the European equivalent of the FDA of a gene therapy drug. And this occurred in 2012. And in fact, the very first commercial sale of a gene therapy product occurred about 2 weeks ago or so. So all of these are based upon this particular virus, adeno-associated virus, which for the rest of my talk I'm gonna abbreviate as AAV. OK, so the way that this may work, for example, in the case of hemophilia, is you take your therapeutic gene delivery vehicle— it's this shell of proteins— load in the correct gene encoding, for example, the factor that's responsible for blood clotting, and then inject it into circulation, into the bloodstream. It finds its way through circulation. At some point, these particles end up exiting the bloodstream, find the surface of a cell, enter into the cell, find the nucleus, and then deposit this DNA therapy in the nucleus of a target cell. So that's the idea. That's what we would like to happen. So here's one example of this. This is a clinical trial result just published last year. And I'm just going to cut to the chase. Each one of these lines right here represents the concentration of that blood clotting factor. It's at around 5% of normal. It's not all the way up to the normal level. But these used to be patients who had severe hemophilia. And all of a sudden they've been converted into a very, very mild form that can be readily controlled. It can infect non-dividing cells, and in fact, in this case, the target cells are liver cells, hepatocyte cells, and they're non-dividing. I'll show you examples of neurons right here. So in the second trial, this is the disease, one of those blinding disorders called Leber's congenital amaurosis, and these patients, unfortunately, you know, as the name congenital implies, they're kids. And as a result of these mutations within these genes that are important for vision, these kids progressively go blind. So you can see this kid is making it through this rather simple maze, needs lots of help from the handlers, and bumps into the wall 7 times. Takes him over a minute to walk 30 feet. So these investigators back in 2008 took the correct copy of that gene, loaded it into the virus, delivered it into the patient's retina, and the results speak for themselves. So this has been done now for over 30 patients. There's a Phase 3 clinical trial taking place in Philadelphia that's going to report report out their results later in October this month, and this may be the very first FDA-approved gene therapy in the United States. So he's asking, does AAV cause an immune response? So I'm not gonna talk about this problem in depth, but it's a major challenge within the field. All of you have been infected with AAV. I certainly have. I've been working with it for 15 years, and no big deal, it's harmless. It doesn't cause any diseases, but I've been accidentally vaccinated against it as a result. So I probably have high concentrations of antibodies against the virus floating around my bloodstream, and as a result, if I try to use a natural version of that virus, it's gonna get taken out by the immune system. It's gonna get neutralized. So we've been working on creating new versions of AAV that overcome this and other problems, and I'll talk about how that works. So there's now a series of successes within the field. Another clinical success was announced about 2 weeks ago at a European gene therapy conference, 2 weeks ago today., and it was a success in treating this horrible debilitating disease which causes degeneration within the brain and mental incapacitation for kids before the age of 5. So all of these are successes. However, there's some challenges or problems within the field. These successes are due to the fact that there were some very savvy academics, professors, who identified the low-hanging fruit of human disease. Disease targets which for one reason or another, the natural versions of this virus were just good enough to be able to treat, because these natural versions are not great as biomedical gene delivery vehicles. And even so, they had to brute force the route of administration, the delivery, in many cases. So in this case, for Leber's congenital amaurosis, that movie I just showed you, it involved an injection deep into the eye behind the retina to be able to get the virus to the right place. Hemophilia B, this is a disease where nanograms per mL, very small concentrations of the protein, are naturally present. And 5% of that is good enough. That's not representative of the majority of diseases. Lipoprotein lipase deficiency, this clinically approved product, 70 injections into muscle. So the field recognizes that even these quote unquote easy targets, the natural versions of the viruses are not quite good enough, and certainly to build upon these easy targets and to begin to treat things like congestive heart failure or Alzheimer's disease or other diseases such as type 2 diabetes, We need better viruses. We need better gene delivery vehicles. So the question was, why not do 100 instead of 70 injections? One challenge, it turns out, is manufacturing. And this is a very chemical engineering idea. It's bioprocess engineering. These things are hard to produce at scale. And this right here, for hemophilia B, they applied the maximal manufacturable dose. And that hit the 5% level. So if we can get a protein or get a virus that's 100-fold more efficient at delivery, for the same dosage, then all of a sudden we can hit 100% instead of having to settle for 5. So that's, that's the idea. Can we make better viruses, better carriers, which will mean that maybe even we have to manufacture less and you can help out on the bioprocess side as well. Okay, so this is the motivation really for, for a lot of the work that we do. This is starting to work, which is terrific, but most disease targets are beyond the reach of the current delivery technology. We need better viruses. So why aren't the current viruses good enough? The first of these is a problem I already described to you. We've all been vaccinated against these viruses, and as a result, it turns out the most common exclusion criteria and the most common reason for turning somebody away from a clinical trial and telling them that they can't get this therapy is that they have antibodies against the virus, against that protein shell. In addition, if the virus is lucky enough to make it past the antibodies, it biodistributes, it goes to the wrong tissues inside the body, It's not very good at penetrating deep into those tissues to reach the target cells. If it's lucky enough to reach the target cells, these viruses, the natural versions, have zero ability to target the right cell and stay away or leave alone the off-target cell. And in addition, even if it can be able to start infecting or reaching that target cell, the uptake and the internalization and the delivery within those cells is often quite poor. So this is a series of challenges or problems that really the field has been facing., and all of them have this really underlying no-brainer reason, which is that nature never evolved viruses for our convenience to use as medicines, right? And so we can illustrate that idea by saying that on this axis right here, this is some kind of success or fitness of the virus, the ability to succeed within some setting, and let's say every single point on this axis is a different sequence or a different composition of that protein shell. These viruses have been evolving for millions of years within a natural setting, And they've likely, you know, reached some kind of optimum or maximum in their ability to succeed within this natural setting. All of a sudden we're plucking them out of the natural setting and we're trying to use them for medical application, getting them to infect cells that they really never evolved to be able to deliver genes to. So we're changing the rules of success and this sequence right here, which may have been optimal, may be completely suboptimal right here. However, we came up with this idea about 15 years ago or so that You know, evolution is a very powerful engine to create novel and useful biological function. Why not simply continue the process of evolution of these viruses but change the direction so that we end up directing that evolution towards new synthetic versions of these viruses that now have exactly the properties that we want them to have for therapeutic application? So in other words, by applying this concept of evolution as a design principle, as an engineering method, we can re-evolve or re-optimize these viruses and create variants that are now optimal for our therapeutic application, rather than being at the mercy of whatever nature has thrown our way. So the way that this works is that we take some input DNA sequences, and these are the sequences from a number of natural versions of AAV that encode that protein shell, because that's the delivery vehicle that we want to engineer. We then play Darwin. We genetically diversify these by using any number of methods, and we create enormous libraries We might call them a series or a collection of drug candidates, each of which may be a potential solution to a delivery problem we'd like to solve. So we generate on the order of 100 million versions of these viruses. We then produce or package these into collections of viral particles, each of which is composed now of a variant protein shell that may have the properties that I want, but it contains inside of it now, initially, during this method for optimization, the DNA that encodes that protein shell. So it's barcoded with its own genome. We then screen or select these by any number of methods. I'll show you a couple of examples. And then isolate the fittest. We can use these either as the starting point for additional rounds of evolution so we can progressively improve or climb in the function of the virus. At the end of the day, pop open the particle, sequence the genome, and recover the information that was responsible for this improvement in function., and then we can take that particle, load it with a piece of medicinal DNA, and use it for downstream clinical application. So the idea is we can evolve new viruses that solve all of these problems, and we have 30+ publications in this field and multiple examples for each one of these problems in which we've solved it in one context or another. It's that outer protein shell that is— that's the vehicle. So if you administer it into the bloodstream, for example, it's that shell that gets the virus to sneak past the immune system. To biodistribute and go to the right tissue, whether that's the liver or the muscle or the lung or the heart, for example. And then it's able to penetrate deep into that tissue, find the right cell to deliver the DNA to. So all that information, the delivery information, is encoded in that protein shell, and that's what we have to evolve, that's what we have to improve. And then once we've done that engineering, we can load in a piece of medicinal DNA to treat a particular disease target. Influenza, yes, or rhinovirus or coronavirus. Yeah, the rhinovirus or coronavirus are the ones that are used for the cold. There's a series of properties that we need in order for a virus to be a really good gene delivery vehicle. It has to be a DNA virus, because DNA is, you know, that's the permanent genetic material inside of our cells. It turns out that one of the cold viruses is actually an RNA virus, and so it doesn't maintain itself for a very long period of time. That's one of the challenges, and there are a series of other properties that we need. We want it to be able to establish a latent infection, so it can kind of hang around in the nucleus for years and keep that medicinal piece of DNA around. So if you go through the checklist of properties that you want, this virus is almost perfect. Can we take advantage of the fact that viruses often replicate as part of their life cycles and use that as a way to spread medicinal DNA? They're really— so within the gene therapy field, For reasons of safety, we typically want to have a really efficient gene delivery vehicle that's replication incompetent, that's disabled, so it can't divide. Because if it can divide, then I can't control dosage, I can't quite control where it goes. However, there's one exception, which is that there's another field called oncolytic viruses. So oncolytic, you know, onco is cancer, lytic means lice or kill. You can engineer viruses that will selectively find tumors replicate in those cancer cells, kill off the cells, and then, you know, as a result, serve as a therapeutic for cancer. So the very first oncolytic virus received regulatory approval by Amgen just several months ago. So that was for treating melanoma. That virus is not very good at treating a lot of other disease targets, so we're actually— we're initiating an effort to apply the same approach of directed evolution to an oncolytic virus. Which is gonna be vaccinia, it turns out, you know, the vaccine strain for smallpox. So he was asking about retroviruses. The two most successful viruses, if I went back, I probably won't go back 15 slides or so, but I had a series of clinical successes. The last one was for treatment of lysosomal storage diseases using lentivirus. Lentivirus is a euphemism for HIV. So people have taken HIV and converted it into a gene delivery vehicle. Typically, in that case, it turns out those viruses don't work very well if you inject them into the body. You have to harvest cells, put them into a dish, add the virus, and then put the cells back. That works really well for the blood system because it's easy to harvest blood, and blood transfusions and bone marrow transplants are part of our healthcare system. Lentivirus is great when you can bring the cells to the virus. AAV is really the premier vehicle to use when you can bring the virus to the cells or deliver directly in the body. And HIV lentivirus is a retrovirus. So for gene therapy, we want a virus that can establish latency, that can go into the nucleus and just shut down and just produce the therapeutic protein from that DNA. For an oncolytic virus, we want something that lyses cells. So we want something that's actively replicating and shredding a tumor, right? Selectively. Yeah, the question was about risk management, what potential risks, either known or unknown, could arise from a brand new generation of therapeutics It's an absolutely great question. And that type of thinking goes into really choosing what diseases you would like to treat. So one of the criteria for us is that we want to treat a disease that has serious unmet medical need and no existing therapies. So lysosomal storage diseases, for example, kill kids by the age of 5. So in that case, you probably— I don't need to tell you. You could probably tell me this much better. You're willing to take a lot more risk in that case because there's less to lose. Or cancer, end-stage cancer patients, in that case you're willing to take lots of risk. So that's where I think the field should start. You know, we should be choosing or identifying the disease targets for which there is no current therapy, no standard of care. Yeah, as you move on to other disease targets where perhaps there's an existing therapy, it's not as good, but there is an existing therapy, there is a standard of care, then we really have to think about the risk management side of things, but I think the good part then would be that, you know, thinking optimistically, if this really begins to succeed in a number of clinical trials, we're gonna learn a lot more about what the true risks are. So the question was about autoimmune diseases, where our immune system makes a mistake and starts attacking one of our own proteins. So this includes multiple sclerosis, lupus, rheumatoid arthritis. And you can deliver genes to components of the immune system, cells in the immune system, or to the tissue that's being attacked that are anti-inflammatory, that are immunosuppressive. And there's some really exciting developments in the field with this new population of T cells that's been discovered called regulatory T cells. They're suppressive of immune responses. If you could genetically modify those, you know, you could potentially have a really good therapy. So for each one of these targets, tissue targets, I'm gonna give you examples of ways in which we've engineered brand new viruses that are orders of magnitude better than what nature has given us. And the idea then is that we're interested in initially treating disease targets that are, quote unquote, lower-hanging fruit, so that we can begin to have a broad impact in these diseases for which there currently are no treatments. And then as we learn more about how to do this gene delivery, and as the field becomes de-risked, we can start thinking about attacking or going after larger disease targets that afflict hundreds of thousands to millions of patients. So the first example is the retina. So if you take a look at the back of your eye, you know, this is the lens, this is the vitreous humor, the liquid inside the eye, and at the back of the eye right here we have this layer right here of matrix protein, and then a few layers of neurons, and at the back of the eye are the light-sensing cells within the retina, the photoreceptors that are responsible for vision. In addition, we have this region right here, which is actually a very important region that's a really high concentration of neurons called the macula or the fovea. And this is why when you read, your eyes scan back and forth across a page. It's so you can keep this really important region focused on where you need high visual acuity. So unfortunately, across the retina and especially within this fovea, there are a couple hundred ways in which this can happen. So mutations in over 200 genes in diseases called retinitis pigmentosa, Leber's congenital amaurosis, retinoschisis, other diseases, cause these photoreceptors to degenerate. The retina thins out, the photoreceptors go away, and vision is lost. People go blind. This also occurs for some larger diseases, such as macular degeneration, age-related macular degeneration, which also kills off these photoreceptors, unfortunately, robbing people of their vision. So what you'd like to be able to do is deliver a drug into this vitreous humor of the eye. You know, it sounds a little bit— it causes me to cringe, you know, thinking about a needle going through the skin, flare of the eye, but it's actually a very simple procedure, 15-minute outpatient local anesthesia. It's done, it turns out, hundreds of thousands of times in the US every year to deliver the protein drugs that treat age-related macular degeneration, like Lucentis or Avastin, which Genentech makes. So it'd be very simple to deliver a DNA medicine in here. Unfortunately, however, in order for this DNA medicine to have an effect, it has to reach the photoreceptors at the back of the eye. There's a transport problem where these viruses can't sneak their way through a couple hundred microns of really dense tissue. The natural versions of the virus can't do this. So in that movie I showed you earlier, the way in which the investigators did this was to jam a needle through the eye behind the retina and then deposit liquid behind those photoreceptors. And this causes a transient retinal detachment. Putting fluid or forcing fluid underneath the photoreceptors. And this is problematic for a couple of reasons. It damages the tissue, it's pretty traumatic, especially if somebody already has a retinal disease. And furthermore, this is a fundus image, this is what an optometrist sees when they're looking at the back of your retina using one of those scopes they peek through. So this is the retinal surface and you can see this marker right here is the region where the virus went. If you're injecting liquid behind the retina and you're not protecting the entire surface area of the retina, you're only protecting a small a little circle. But that was good enough for that clinical trial I showed you earlier. But can we do better? So the idea is, can we evolve a version of a virus that is capable of transporting all the way to the back of the eye? And I'm going to cut to the chase. This took about 9 months to do, but we evolved the virus through multiple rounds of evolution for the ability to reach this population of cells, the photoreceptors, which we could separate away from all the other cells in the eye. And we finally— our 100 million variant library converged down to just a few variants, just a handful, about a half dozen. And this was the dominant variant right here. This is a cartoon of that protein shell. You can see there's some modifications here in green. And it raises the question, how are we doing? Okay, so we loaded a gene, a marker gene, a green protein into this virus. Just to orient you, this is the parent virus, the original one, the only one that's been in clinical trials in the eye to date. This is just the vitreous humor down here. The photoreceptors are back here, and you can see that when you put this virus into the vitreous, it doesn't really do very well. This is the variant that we evolved. So it's hitting every population of cells all the way to the back of the eye, including the photoreceptors. And furthermore, this is that fundus image again, and we're seeing that we're protecting the entire— or delivering genes to the entire surface of the retina, as opposed to just some little circle. Okay, so can we do some therapy with this? So this is a disease called X-linked retinoschisis. It affects about 1 out of 5,000 boys. This is what you would call a well-defined, low-hanging target for gene therapy. We know the broken gene, and if we can supply enough of this therapeutic DNA to enough cells within the retina, you could potentially treat this disease. So this protein, it's believed, is responsible for cell adhesion, kind of gluing glue to stick the cells together. And when you lose the protein, the retina quite literally falls apart. So this is a cross-sectional image from the eye of an 11-year-old boy. And he's forming these cavities or schisis in the retina because the retina is delaminating. It's falling apart. And as a result, this 11-year-old is going blind. He's losing his vision. So we wanted to try to initially within models of this disease see if we can treat it and then move towards human clinical trials. Okay, all right, so this is a mouse model of this disease. This is the normal mouse. You can see that it's fine. This is the diseased mouse, the one that has this X-linked retinoschisis disease. You can see the eye begins to fall apart. And if we load an inert protein into our evolved variant, which we call 7M8, doesn't do anything to the eye. If we load the correct, the therapeutic version of this virus into any number of natural versions of AAV, doesn't do anything.. But if we load the therapeutic protein into our engineered virus, it's able to rescue the tissue. The tissue is thicker, indicating that we've saved photoreceptors. And then when we analyze the electroretinogram, which is essentially the current, the electrical current in the eye that tells us that the neurons are still functioning, that the animal has vision. This is the wild type, the natural normal mouse shown here in the dotted line. In green is the mouse that's gone blind. Its electroretinogram, its neuronal circuitry is dead, it's flatlined. However, if we end up delivering the therapeutic DNA, we're able to rescue this almost all the way back to normal. And this effect, unlike a number of other controls, this effect is sustained for months, essentially for the life of that model. So we're currently taking this therapy and moving it towards clinical trials, which I'll describe towards the end. Okay, another example is lung. And the most famous, perhaps, monogenic, single gene disease within the lung is cystic fibrosis. And in fact, the gene that was responsible, whose mutation leads to cystic fibrosis, was originally developed, or identified rather, by the current director of the National Institutes of Health, Francis Collins. So however, if we got really good at delivering medicinal DNA to the lung, we could think about treating things like emphysema, or chronic obstructive pulmonary disease, or hypertension, pulmonary hypertension. There are a number of disease targets we can think about. So unfortunately, natural versions of AAV are really bad at the lung. AAV2 was attempted or was utilized in a couple of human clinical trials to treat cystic fibrosis. It was perfectly safe, but those trials fell flat. No efficacy whatsoever. And this right here, we evolved this virus to be able to better mediate delivery on lung tissue from postmortem cystic fibrosis patients. So these are little pieces of lung tissue inside of these wells of a tissue culture plate. You can see AAV2 is quite poor, and we evolved this version right here, which we called AAV2.5T, and if we quantified how good it was, it's about 99 or about 100-fold better than the variant that failed in clinical trials. So it's not surprising that this variant was so bad. That's shown here with the blue circles, and it's essentially down to the level of noise. So if we take a look at this variant, it— I'm not gonna go through this in detail, but it experienced a couple of modifications, a couple of changes to make changes in the structure of that surface protein, we could have actually stared at the structure of that protein for 100 years and not guessed which modifications to make, but evolution provided us a solution to that problem. So evolution involves two steps: genetic diversification to create a gene pool, and then selection for more fit variants within that gene pool. So on the genetic diversification side, we use, it turns out, about 20 different techniques from molecular biology to create 20 different libraries where diversity, changes in amino acid sequence, are introduced into that shell of proteins in a variety of different ways. So our library is actually a pool of 20 different libraries, each of which has diversity dialed in a different way. And we've been collecting those over, or engineering and creating those over the past 15 years. So how do we identify the ones that are fit within the gene pool and reward them? 'Cause the way natural evolution works is if there's a fit gene, it has the opportunity to increase its frequency in the gene pool. You know, that's the essence of Darwin, right? So the way that we reward the fit viruses is that in the case of the photoreceptors, we took the entire library, all 100 million variants, injected it into the vitreous of the eye, waited a couple of weeks, recovered or isolated the photoreceptor cells, and then used a molecular biology technique called polymerase chain reaction, which is invented in the Bay Area, it turns out, over in Emeryville. Nobel Prize-winning discovery. The person who got the Nobel Prize did his graduate work here at Berkeley. But polymerase chain reaction, or PCR, is used then to amplify the successful gene copies or copies that the viral— the gene encoding that viral capsid or that shell protein. So rather than having the virus replicate, we use a molecular biology technique to amplify the gene pool and reward the fittest. Alright, so can we use this to treat cystic fibrosis, or at least begin to treat models of CF? So CF is the most common single gene disease within the United States and Europe. It's due to mutations within this particular gene, which I mentioned was discovered a couple of decades ago by the current director of the NIH. This is a chloride ion channel, and in the absence of this chloride ion channel, unfortunately, for reasons that are not fully understood, patients suffer from chronic bacterial infections within the lung. And this leads to a median lifespan of around 35 years or so. Gene therapy has been tried, but there's been no efficacy, and it's because it's widely regarded or widely accepted that the delivery efficiency of the natural versions of this virus are very poor. We need better viruses. So this is postmortem human tissue from a CF patient that we delivered the therapeutic DNA to using this virus. You should see in green here a stain that lights up or marks this protein, the CFTR, the chloride ion transporter. It's completely missing within these CF lungs. If we end up loading the CFTR into our engineered virus, we're able to completely restore production of this therapeutic inside of cells. Okay, so we see the protein. Does it actually affect this chloride ion transport? Do we get better chloride current, electrical current within the cells? So you should see here when you add this drug, there should be a spike in chloride ion transport, but this is a cystic fibrosis lung. That's lacking that transporter, and as a result, you don't see it. When we deliver the correct copy, the CFTR, to these lung tissues ex vivo, or outside the body, you can see all of a sudden we're able to restore completely the chloride ion transporter. All right, so can we then take this into a model on the way to human clinical trials? This is brand new data that we haven't published yet, but we have collaborators who generated a pig model of cystic fibrosis. And the pig is important because its lung is really similar to the human lung. So if it works here, that's an important step towards human trials. So this is a normal pig. Perhaps you can't quite see, but in red here— I switched colors on you— this is the stain for that CFTR protein in the normal pig. If you move into a CF pig, where our collaborators have knocked out or deleted the CF gene within these animals, there's no red here. Take my word for it.— but if we end up loading the CFTR into an evolved virus, we're able to restore CFTR gene expression within the pig lungs. Okay, so does that affect chloride ion transport within the pig lung? So this is a normal pig, and this is the current of chloride ion within that pig, within this layer of cells. You can see it's right here. Within the CF lung, it's completely dead. If we end up delivering our virus, we're able to restore it close to normal. This actually increases the ability of the lungs to kill off bacteria. So these animals, once treated, have lower bacterial load within the lungs. So these are now, we believe, really important preclinical data that we're going to work with the FDA as a— to be able to enable an IND and head into human clinical trials. All right, so the final example is gonna be in the brain and spinal cord, where actually the majority of my own work has been over the years. And there are a couple of devastating diseases, so these lysosomal storage diseases, there are actually 35 of them, which really kill off neurons or compromise the function of neurons in kids before the age of 5. So horrible debilitating disease. In addition, you're probably familiar with ALS by a different name. It's known as Lou Gehrig's disease. So this is a paralyzing disease. And of course, if we get good at this, we can start thinking about larger disease targets. That will benefit more patients. Okay, so this is the brain and the spinal cord right here, and there's a problem, which is how do you get DNA to these tissues? They're protected by the blood-brain barrier, so we can't just stick the DNA into circulation. In addition, these are very, very large tissues. If I wanna treat these lysosomal storage diseases or Alzheimer's, this affects a very large volume of tissue in the brain. In addition, if I would like to protect the spinal cord, you know, what am I gonna do, treat it like a pincushion? And make multiple injections. That's risky. It induces risk of infection. It causes damage to the tissue. So for this recent clinical trial I mentioned to you that was successful a couple of weeks ago for treating this lysosomal storage disease within kids who are 4 years or younger, they actually had to drill 8 holes into the cranium, into the skull, and do 16 injections into the inside of the brain. You know, it sounds awful, but this is one of those cases where you're willing to take on risk because the consequences of not doing anything are so severe. And as a result of doing this, they were able to restore about 10% of normal levels of this missing protein. So again, we need much better gene delivery vehicles. We can do better. So a few years ago, we discovered this property that, as we all know, neurons have a cell body. Pretend that's my elbow. And then they extend wiring, and they have these axons or these projections. That touch the next neuron, and then that touches the next neuron. And conduction of electricity along that series of cells that are all touching each other is what's responsible for the function of the brain, for, you know, thought, sense, and motion. So we discovered this property that it turns out that a lot of these cell bodies are distributed over large volumes of tissue, but a lot of the wiring, a lot of these projections, go to the same place for certain parts of the brain. And we discovered that AAV has this interesting property that if you deposit it near the end of that wiring, it'll actually take up the virus, traffic it back to the cell body, and you could use that as a way for getting delivery or better spread of virus within the brain or in the spinal cord. So, for example, this region of the brain called the cortex is very large, but all the cortical neurons send their wiring to one single location. Or the spinal cord is obviously very large, But motor neurons in the spinal cord send their projections, send their wiring to muscles throughout the body. So maybe if I did an injection in muscle, I could actually get it taken back up into the spinal cord. So these investigators— I was on this paper, I wasn't on this one— but these investigators did exactly that. And they did injections of the virus into muscle, used that as a way to get the therapeutic DNA into the spinal cord, and used that to treat this mouse model of Lou Gehrig's disease. Okay, sounds great. You know, what's the problem? Well, a company tried to develop this technology, and this transport property of the virus going from the ends back to the cell body is way too inefficient for human clinical use. It was just not good enough. We're talking about 1 in 10,000 particles actually transporting back to the cell body. We need to do better. So we evolved a virus that specializes at doing this, and these are unpublished data as well. As well. So I mentioned the entire cortical region of the brain sends neuronal projection, sends their wiring to this single location. So if I could put a virus here, maybe I can actually deliver it across this very, very large volume. So that could be a way of treating large volumes without having to do 8 drill holes and 16 injections. So we took the entire library of virus, put it into this location, harvested the cortex, recovered the virus that made it there, did polymerase chain reaction to reward it for being good at getting there, to increase its frequency in the gene pool, repeated 5 times, and then came up with a series of really good viruses, potentially really good viruses. We needed to test these. To do so, we loaded in a red fluorescent protein just to be able to get a readout, and this is the result. So this is that single injection right here, and you can see that the virus is trafficked back to and lights up the entire cortex of the brain. So in addition, this is a parting or last image I'll leave you with. This large volume of the brain sends their wiring to just a couple of small locations. So we did a couple of injections, one with a green protein, one with a blue protein, one with a red protein, and you can see that by doing a couple of just very simple local injections, we're spreading this gene product, spreading the therapeutic DNA across the entire volume of the brain. So this is a better version to be able to reach difficult-to-reach surgical locations without having to brute force the delivery. And furthermore, we believe it's much more efficient than the current viruses. Now that we've discovered that, you know, my time at Berkeley and Berkeley's incredible brain trust and capability enabled us to try to make some progress within this field, we now are interested in therapeutically translating this, taking it into human clinical trials. So as Doug mentioned, I've been coming increasingly involved in taking the next step step to take the technology and work with Berkeley to translate this into the private sector, to put it into a company which can now advance it towards clinical trials. So we started a company a couple of years ago called 4D Molecular Therapeutics, which is a spin-out of Berkeley and QB3, for those of you who are familiar with QB3. And this company has been very rapidly forming a number of partnerships with other companies. So we have a partnership with Roche, we have a partnership with Unicure, which is the only company to have a clinically approved gene therapy product, the one in Europe right now. And we have a couple of other partnerships. We believe we'll be able to announce another partnership with a large pharma company over the next couple of weeks. And in partnership with these companies, as well as with our own resources, we're taking a lot of these viruses and pushing them into human clinical trials so that we can build upon, I think, some of the success that we and others have had at Berkeley to be able to create new therapies to treat human disease. So I'd like to close by leaving you a couple of points. Gene therapy is actually working. There's been a lot of effort within the field, especially to overcome this delivery problem, and there's a joke that it took 20 years to become an overnight success, but it's really now working. And what's made this possible is that we're one of these strange groups that views viruses as gifts of nature because they've evolved a bunch, a lot of ingenious strategies to carry DNA inside of cells, and we can use them to do that, to carry a piece of therapeutic DNA rather than their own DNA. But quite obviously, viruses evolve for their own purposes and their own ends, and their properties as a result are not compatible. They're not what we need them to be to serve as biomedical therapeutics. However, we've developed a powerful engine called directed evolution, which we applied to this virus to create new versions of the virus designer viruses, synthetic viruses, that are now precisely designed to have exactly the properties that we need them to have for a variety of different tissue targets. So we've fundamentally reprogrammed the virus to bind to different receptors on the cell surface, to increase the efficiency of delivery, to target delivery towards particular cells, to enable the virus to become stealthy and evade the immune response. And in work I'm finally— I'm really not talking about at all today, by taking these evolved products and actually reverse engineering them, we've been able to gain some kind of fundamental mechanistic insights into the ways in which these viruses interact with ourselves. So I'd like to close by really acknowledging the people who actually did the work. The work in the retina was done by Leah Byrne, Tim Day, Denise Alkhara, and Ryan Klimschak in collaboration with John Flannery here at UC Berkeley, and we have a couple of other collaborators for current work as well at other universities. The work in the lung was done by JT Kerber in collaboration with Joe Zabner. This is the fellow that has the CF pig model at the University of Iowa. And then furthermore, the work on this transportability within the central nervous system, within the brain, was done by Jorge Santiago Ortiz, Bamiol Wang, in collaboration with Oleg Karpova at Genelia Pharm. So thanks so much, and I'd be happy to address any other questions. Sure, the question is why are they better? You know, if I can rephrase it, what better ways do they interact with cells and tissues mechanistically in order to enable this improvement? So we feel that one of the powerful features of this approach is that we don't need to know precisely what the mechanistic problem was in the first place. We can evolve a solution to the problem, and then we're in the position of being able to, you know, in a luxurious position of having the solution to the problem, and then we can do the reverse engineering to understand what the mechanism of improvement was. So we've really found that, you know, it makes sense that in order for a piece of DNA to travel all the way through a tissue and reach the nucleus of a cell, there are a series of barriers that could potentially be rate limiting, and we have found in various tissues that different barriers and sometimes multiple barriers are the problem for that tissue. So in the case of the retina, for example, that I mentioned, there is this matrix layer, this dense layer of proteins. It turns out that a lot of the natural versions of the virus are too sticky. They bind to that and can't penetrate any further. So the evolved version of the virus actually reduced its affinity, reduced its binding for that matrix, so it could slip through, but it also acquired the ability to bind to a completely new target protein on the surface of the neurons, and that enabled it to get access to those neurons in the way that the original virus could not. So that's typically what we find. It's multiple modifications acting in combination which knock down multiple barriers that are rate-limiting, or were rate-limiting. Yes, so the question was, there's age-related macular degeneration. There's some drugs for that, right now, well, there are two forms of AMD. There's wet and dry. There are no drugs for dry AMD. For wet AMD, there's this product Lucentis from Genentech, another called Eylea from Regeneron. So I would think that right now, since there's an existing therapy for that, we probably wouldn't go after that as the initial target. You know, once we're positive that we've really solved all the problems for gene therapy in the eye, then we would want to go after that because those are protein therapies, require 10 injections a year, into the eye. But for the inherited forms of these macular degenerations, retinitis pigmentosa is a really good example. You know, this is an awful disease that really robs people of their vision. And Leber's congenital amaurosis does the same thing, but does it to single-digit age kids, you know, kids who are 5, 6 years old. So those are the ones we're really going after. So our initial target is gonna be a disease called chorderemia, so that's what we wanted to head into clinical development with. And that's a single gene disorder that causes retinal degeneration, or macular degeneration. Yeah, the question is what happens if we're successful in getting a piece of DNA to a cell and then the cell dies? So in that case, most of the diseases that I'm talking about involve delivery to a stable, non-dividing cell. So neurons, hepatocytes, skeletal muscle, heart tissue. So these are the tissue targets that we're going after initially. And those cells don't turn over very much, with a couple of exceptions. So the blood system is a more challenging one, because those cells are constantly turning over. And that's why delivery to T cells is a bit more challenging. So we're going after ones where the cells are really stable, and as a result, the DNA will be really stable as well. Yeah, the question is, how do we choose what diseases we should be going after? Unfortunately, mutations in about 7,000 genes cause these rare monogenic inherited diseases, and we want to— our goal, the guiding principle of the company is to try to get our delivery vehicles into as many clinical trials, as many patients, and as many products as possible. So that's why we're working with partners so that we can put our vehicles into their clinical trials, but we also want to develop clinical trials on our own, and then your question is, you know, which ones do we think that we should target first? We have a series of criteria, you know, the biology has to be really well known. The broken gene has to be identified. There should be models where we know if we deliver that broken gene that we're getting efficacy within the model before moving on to human trials. We'd love to be able to treat diseases that have maybe 10 people who are afflicted, but we should probably start with the ones that have 10,000 people who are afflicted. So there's a series of criteria that we've really had a lot of discussion with amongst ourselves and with key opinion leaders, KOLs, in the field to really identify what our initial targets are. Yeah, his question was asking what's the wiring or what are the pathways in the brain that we could potentially use to deliver a therapy, including this cortical pons projection. The brain has all this circuitry and all this wiring, all this interconnectivity that's really fundamental and central to its function, and that's good news for us because it means that if we're treating the entire brain, you know, we can pick a couple of strategic injection sites and it can spread spread everywhere. If you're treating Huntington's disease or Parkinson's disease, which are much more focal or much more local, then we can choose strategically a local injection site which will result in delivery just to those affected regions. So there's a lot of opportunity. Yes, so from a product, from the point where we have a therapeutic piece of DNA inside our virus for the first time to the point of having the first human clinical trial testing that is about a year and a half. And we're reaching that stage now for several disease targets. Then it has to go through clinical trials, which is probably another— for these rare monogenic disease, which the FDA would probably fast-track because there are no other therapeutics for it, you're talking about 3 to 5 years to get a clinically approved product.