a16z Podcast - Undruggable Drugs

Episode Date: May 6, 2020

In this episode of the a16z Podcast, we take a deep dive into the world of drug development—specifically "undruggable drugs": a category of protein, protein family or even piece of RNA that’s so difficult to target that many researchers don’t even want to touch it. Jay Bradner, President of the Novartis Institute for BioMedical Research, shares with a16z General Partners Jorge Conde and Vijay Pande, and a16z's Hanne Tidnam, all the new tools, technologies and breakthroughs which are causing the science of therapeutics to explode in some of these areas where it's been incredibly difficult (even impossible) in the past. From molecular glues to cell and gene therapies, Bradner shares the behind-the-scenes science stories of what it really takes to make a new drug that shatters the category of an "undruggable" target. 

Hi and welcome to the A16Z podcast. I'm Hannah. In this episode, we take a deep dive into the world of drug development, specifically undruggable drugs, which is a category of protein or protein family or even a piece of RNA that's so difficult to target many researchers don't even want to touch it. Jay Bradner, president of the Novartis Institute for Biomedical Research, shares with A16Z general partners Jorge Condé and Vijay and myself, all the new tools, technologies, and breakthroughs, which are causing the science of therapeutics to explode in some of these areas where it's been incredibly difficult or even impossible in the past. From molecular glues to cell and gene therapies, Bradner shares the behind-the-scenes science stories of what it really takes to make a new drug that shatters the category of an undruggable target. So I thought maybe we could start about just talking about what the category of undruggable

really means to the industry. What does that traditionally mean? This is a favorite subject, but also for me a sore subject. object. The term undruggable refers to as yet the inability to drug a protein or a protein family or a piece of RNA. It's an unfulfilled promise. Imagine drug hunting with small molecules where I've worked and trained as sculpting a drug molecule that fits into the pocket of a protein. Well, what if there's no pocket? That protein may be regarded in our discipline as a priori undrugable. So is it always a shifting kind of category or was there a particular group that

always was understood to be that kind of undruggable? It's very much both. You know, Mars is unwalkable until we arrive there. Serious human diseases of the non-infectious nature are often caused when pathways go awry. And these cellular pathways are driven by little machines called proteins that are globular and they have invaginations where biology occurs, enzymes that metabolize food and such. When these pathways go awry, we try to identify a critical note in that pathway, typically a protein, and work to understand functionally if it's too active, in which case we try to inhibit it, or not active enough, in which case we try to activate it. In the discipline of drug discovery, this biological knowledge is very powerful. But sometimes, we're

we regrettably find out that it's a type of protein or protein fold that has never been drugged before. And this creates real challenges. So this is the undruggable when we have no idea how to get that protein there. These are the undruggable proteins. And there are whole families of very tantalizing protein targets creating a conceptual risk that often keeps many scientists away from pursuing coordinated efforts in drug discovery. In my time, as a professor, I studied the way genes were turned on and off. And in cancer, as a cancer doctor, I was interested in the proteins that would cause the growth program to be activated,

to turn one cancer cell into two and so on and so on. These proteins, called transcription factors that bind DNA and turn genes on, are considered to be beyond the reaches of drug discovery or undruggable as a class, which is regrettable because the perception that they may be hard a drug, has kept many scientists away from even trying. So people don't, they literally don't touch it because it seems like such a challenge? There are a couple of important exceptions.

The estrogen receptor binds estrogen. It is therefore drugable by the sex hormone, estrogen or estradiol. But the most commonly activated gene in all of cancer called Mick, the protein that sits around the human genome orchestrating, the growth symphony, has never been successfully drugged, even though it is one of the best validated targets in over the last 30 years in cancer science. It's so interesting because I sort of assumed that it had to do with a lack of biological knowledge. But you're saying the biology is very well understood, but we just haven't understood how to approach it.

So what is changing now? Where are we in the landscape of these undruggable categories of drugs? I mean, one way to think about that is that in a sense when we mean undruggable, it's undruggable by the way we normally do things. Correct. And only when you start to develop these new methods, you realize even in the old targets, there's other things you might want to hit and other ways to hit it. That's right. One of the things that's really interesting here is, you know, when we think about targets, we add adjectives to the targets. We have undruggable targets.

In the fullness of time, there may be no such thing as an undruggable target when you take in sort of the full armamentum of different modalities that we might go after a specific target. Can we take the other side of that coin for a second? Is there such thing as a novel target? A novel target in the language of drug discovery is maybe the first recognition that a protein is really involved in a disease process and that biological experiments have validated that protein or gene in that process. Novel targets may be fully drugable, like the proteins that sit on the surface of a cell that because of successful prior campaigns to drug kinases are now, as a group, considered

easily drugable. But sometimes novel targets are in these undruggable protein families, and this gives us pause. I believe that some of the best validated targets in disease biology would have a clarified path to helping patients if only we could get out of our own way and really commit to approaching these proteins as drugable, to challenge the dogma. To let go of the old concept. That's right. I'd love to hear what some of those successes that really sort of forged a whole new path forward for people were and then also break down the tech behind what made those possible. I think a very fine example where drug discovery has taken down an undruggable protein

target is our work to develop the first inhibitor of what's called a phosphatase. In this case, a protein called ship two. Phosphatases are some of the most interesting proteins in disease biology. There are phosphatase is very important for diabetes and a couple extremely important for cancer. You might know what a kinase is. This is a protein that drops what are called phosphate groups onto proteins. And there are a great many important kinase inhibitor drugs that followed once Novartis developed, the first, if not one of the first, called Glevec for chronic myeloid leukemia.

As there are interesting kinases that drop phosphates onto proteins, there are counterfeiting are counteracting phosphatases that pull them off. Oh, interesting. And it's for no particular reason that kinases are so commonly drugged and phosphatases are not, except that for 20 years people tried to make phosphatase inhibitor drugs, and they just couldn't do it. It's one of the most famous protein families in the undruggable class, and there's something really peculiar about it.

Phosphatase drug discovery campaigns almost always produce. a very potent and sometimes very selective inhibitor of a pure enzyme, studied, say, outside of a cell. Okay. But these molecules don't work when the enzyme is inside of the cell. The pocket that's drugged in the phosphatase is very positively charged, and you know how opposites attract. The molecules, then, that are discovered are very negatively charged, and they can't get

into cells. Scientists bang their heads against the wall for decades trying to make phosphatase drugs for cancer and diabetes and other disease states and were unsuccessful. Well, some very creative scientists at Novartis did a really interesting experiment. They imagined that maybe a way to inhibit the phosphatase isn't to go for the most active site, but to try to inhibit the enzyme through what we call an allosteric site to sort of sucker punch the phosphatase at a different part of the protein. And so we performed two high throughput screens, one with the full-length phosphatase that has two or three globular domains, like three beads on a string, and a second full high-throughput screening campaign where we just looked at the active enzyme pocket itself.

We found a thousand hits in this assay, and we threw all of them out except two. We only kept the molecules that would work in the full-length protein, but wouldn't work in the small-format protein. Basically, that you'd find the molecules that would hit the pocket that's only present when the whole protein is there. Exactly. Drug discovery is like trying to find a needle in a haystack. We perform thousands, sometimes millions of experiments.

with chemicals to try to find the one chemical that does what we want. We threw out all the molecules that would inhibit the active site and kept only molecules that worked when these other sites were present, called allosteric sites. After many years of very careful science, we produced the very first inhibitor of a phosphatase. And the way this molecule works is it glues the ship two protein together. We call it an intra within the same molecule, an intra-molecular glue.

What a cool concept. It's extremely cool. For years, we've been just looking under the street lamp, trying to find different ways of accessing the active site, not considering that there was a whole different way to approach this protein and protein family. So was it fundamentally different in the discovery process in using that high throughput?

or what was it that really led to that breakthrough, you think? It was a concept shift around what to try, but then was there also different tools involved that changed how you approached it? Proteins, when we visualize them using what we call crystallography, they look like static structures with little pockets, and we, like molecular locksmiths, fashion a molecule to bind into that active site.

And unlock or lock? Correct. To super glue, perhaps it closed. but in fact proteins are very dynamic they live and breathe they have waves of change through them in the cellular context and so we've learned that by binding to these other sites these allosteric sites we can confer rippling effects through the protein that can activate it or shut it off there is a paradigm for drug discovery that could be vastly oversimplified that you get the crystal structure, you look at the pocket, you find something that will bind to them pocket as a competitive

inhibitor. But, you know, I think really the beauty of what you're talking about is that, you know, you're thinking about just a different paradigm. When we say undruggable, it's undruggled by the original paradigm. And so allosteric inhibitors is an example of a new paradigm. It doesn't show up in the crystal structure. I mean, by construction. So again, it has to be a new paradigm. You have to think of a different experiment. So it can't be crystallography to an inhibitor. It has to be something else. And here used htia high throughput screening uh but uh you know it i think the higher level thing is that if you're starting to just think out of the box think of a different approach you'll come up with other experiments or really frankly other technologies which might bring in data science or

computation as well um or other types of experiments to be able to sort of just do it differently yeah and i think the the set of weapons that we have you know to fight disease right we make medicines, right? The armamentum of tools that we have has expanded dramatically. So if we go way back in time, our medicines were chemicals, small molecules, and then we had proteins, and then we had pick your flavor. We had gene therapies and cell therapies. And now we have, you know, if you just look at the different modalities that exist today, it's a pretty long list, right? I mean, you have bispecifics, you have, you know, antibodies, you have, you know, so on and so on and so on. So if you're a maker of medicines, and you're looking across a pretty broad range of diseases, how do you think about

what is the right weapon for that fight? There's never been a better time to be a drug hunter. We have never before had such a clarified understanding of disease processes, but so much more to learn. And we've never before had so many therapeutic modalities or weapons to apply to take out a critical protein in a disease pathway. We try to pair up the best possible biology to validate a target, to convince ourselves that the juice is worth the squeeze, because drug hunting can be a decades-long endeavor. And having that rock-solid biology in the beginning is just essential. And then secondarily, we try to choose the sequence, truthfully, of approaches that we'll try to

try to identify the first prototype drug. We have molecular glues that take undruggable proteins and glue them to another protein so as to leverage the large binding capacity of a second protein to compensate for the weak binding capacity of a small little molecule. We have induced fit molecules that show us that an active site exists beyond what we thought was available to us through, say, x-ray crystallography. We have gene therapies where we use viruses to deliver disease genes to compensate for the lack of gene, inherited disorder, or to augment a biological pathway with an invented gene, as we have in carty cells.

We have RNAs that can take out a protein, coding RNA, or RNAs that can deliver a protein where it's needed. We have so many weapons at our disposal. We try as drug hunters to marry up the correct technology to the target and how we intend to disable it. So is this category, this new kind of paradigm shift of molecular glues, how does that lead the way for other kinds of opportunities? I'm very taken with this new class of medicines that we call molecular glues. The story of the phosphatase inhibitor is one of a molecule that glues a protein into a confirmation where it can't function. Even more compelling are molecular glues that short-circuit disease biology pathways by bringing two proteins that would normally never engage one another into very close proximity.

That's fascinating. Literally short-circuiting cellular biology. And what happens? We have conventionally thought of medicines for cancer and other serious diseases to bind and inhibit a protein target, but sometimes we drug the wrong part of the protein. I'm right-handed. If you cut my right hand off, I will eventually learn to write left-handed. And this often happens in cancer drug discovery. We make an inhibitor for one part of a protein, and then the protein mutates or evolves, and another part of the protein starts to compensate. What if we could, in binding,

this protein, destroy it altogether, take out all functions of this protein. It could be a very powerful way to disarm a disease protein, but moreover, we could start to drug proteins that we haven't been able to approach before, proteins involved in scaffolding, as opposed to, say, enzymes. So is it like a protein delete almost? It's like a protein delete, yes, when it works. This idea was floated by Kenton and Roberts almost 20 years ago, the idea that you could make molecules with two arms, one that would bind the protein of interest, and another arm that would bind the cellular garbage disposal system called the ubiquitin proteosome system. And these sort of two-headed monsters would bring proteins to the trash to be destroyed.

Kenton and Roberts didn't get it to work, but it was a powerful idea. but it would be many years until we would identify a chemical solution to targeted protein degradation. We made small molecules that have the power to bind a protein of interest and a group of enzymes called E3 ligases that are professional garbage disposal enzymes. Over the last five years since this first discovery was reported in Science Magazine, now pursuing this research as therapeutic discovery at Novartis, we've degraded more than 50 or 60 proteins with this technology. Wow.

And what's interesting is it pretty general. I mean, because I think because it's just binding X to Etheligase, you can sometimes have a lot of things in X. That's right. It doesn't work every time. But with a little bit of luck and a lot of chemistry, we have been able to degrade almost everything that we've intended to approach with this chemistry. So let's actually talk about what the luck piece versus. the other stuff going on. How much of it is, you know, in these in the molecular glue, or I'm just going to unclassily call it the protein delete, but how much of that was a sort

of creative shift of thinking? You know, if you had to say what the single biggest maybe cause or driving factor that led to these major breakthroughs of these undruggable areas, what would you credit those to? I think there are creative moments for scientists where perhaps inspired by nature or the finding of a colleague or a quiet and isolated moment of personal introspection and inspiration where you come upon a consideration that you've just never taken before that maybe nobody had ever taken before it's uncommon but at these moments you kind of open yourself up to a new paradigm a new mode of say pharmacology in this case and then the wheels just start turning.

And you imagine if that is possible, then perhaps this is possible. If this is possible, then perhaps what is possible? Today, we're now because of this science able to delete, as you say, proteins at will. But I'm very taken with the idea that molecular glues, perhaps even smaller than these big and bulky, two-headed monsters that we presently make, will allow us to short-circuit many other cellular pathways. Once you can engineer chimeras at will, then you can start to have so many different mix and matches. Then it's about sort of being creative about the potentials of the disease and the disease biology of what you want to bring together. That's right.

All right. So, yes, the eureka moment. Is it that the eureka moment happens and then we have new tools to drive them forward faster? Or are we fundamentally, are the tools, the new tools fundamentally changing how we think about discovery? Targeted protein degradation is the ensemble. of some very advanced technology that ultimately allows us to test and deliver on the premise. These are advances in chemistry, the ability to make different kinds of molecules with different types of chemical reactions. It's the use of protein structure in computers to continue to

sculpt and optimize the structure of the molecule. It's advanced cellular systems. We now have many brains that we can make from stem cells derived from patients with severe neurologic conditions that we don't even understand, but that allow us for the very first time to ask questions that were previously unaddressable. And increasingly, we have access to so much data that we use certain data scientific technologies or digital technologies to lead us to new hypotheses and to confirm or refute old ones. It's been a lot of time talking about basic research which is like, for us, that's like the fun stuff. But in the end, we have to, the reason why we're all here is to bring this technology

to patients and make drugs. Within the Novartis infrastructure, like, how do things go from the institute to other parts of Novartis? How does it become a drug? Or doesn't, I have to say. Which darlings do you have to kill and why? Regrettably, pharmaceutical companies have abdicated basic research and even translational research, often treating innovation like a commodity that can be bought and sold. But there are some

challenges that will only ever be successfully approached if we commit a great degree of basic research in disease biology. We're comfortable taking the time to serially deploy new technological approaches to take down undruggable targets. We have sometimes a 10 or 15 year horizon to take down a protein so pivotal for disease, our basic research in disease biology is unapologetically translational. We exist to make medicines. But our research doesn't end in an animal model, a mouse model of, say, cancer. Our remit is the concept of a drug and its invention through the proof of concept in the only relevant model system of disease, the patient. Sometimes the idea that we had in mind for the medicine when we invented it is ultimately not the best use of that medicine

for patients. And so we perform early clinical trials as signal finding to identify the right patient for this new and first-in-class medicine. For sure, other pharmaceutical companies have drug discovery capabilities. For sure, the biomedical, biotech ecosystem is serious about disease biology. What is unique at Nibber is just the gravitational energy of 5,600 drug hunters working in a framework of intense scholarship to dissect disease biology with therapeutic intent. You know, we've seen cellular therapies. We've seen gene delivery, you know, gene therapies, what's next? Either as a technology or as a means towards addressing, you know, new disease areas. The creativity in our science right now is,

at an unprecedented high. Still, though, we can expect, I think, new waves of innovation. The first generation of gene therapy is truly remarkable, but we must get beyond simple gene replacement. And imagine hybrid gene therapy vehicles that use chemical cues or respond to small molecule drugs to allow more precise dosing of these genes. Our ability to interact with the human genome

in a therapeutic context, is brand new and so exciting. With CRISPR, we can selectively damage the genome. With base editing, we can subtly replace a single base in the human genome. And now with prime editing, we can insert new bits of genome at will, bringing new functions to cells in the body. These are powerful potential technologies, but they're often looking for their killer application and disease biology.

So some of those new functionalities like molecular, lose or deletes. How do those translate into medicines for patients with diseases? Do those end up looking like pills that we can still just, you know, take and solve the problem? Or is there some fundamentally different way these medicines have to make it out into the world to patients? Some of these next generation medicines will look just like classic medicines to the patient. There'll be small white tablets that they take once a day. They won't know that something totally different is actually going on in your body. Correct. But many of these medicines will be indeed very different. Definitive therapy for spinal

muscular atrophy is today the intravenous injection of a virus that delivers new bits of genome to the neurons in the brain. Cellular therapies for sickle cell disease may require a bone marrow transplant to get them on board. These will not be picked up at the pharmacy. We need to be very cognizant of this because we intend to make medicines that will reach patients all around the world. And so often in drug discovery, we work backwards from the patient, not just their biology, but the way in which doctors will be able to treat the patient. And sometimes this means that certain therapeutic modalities will just not be appropriate or acceptable. For example?

Well, sickle cell disease is a great example. Now with genome editing, we can remove switches in the blood cell. that cause a normal form of hemoglobin to be expressed, compensating for the mutated hemoglobin that causes sickle cell disease. But to bring these cells to patients, we can't just hang a bag and infuse it in the vein. We need to ablate or remove with chemotherapy all of the normal bone marrow to get those cells on board. This will be really challenging in Ghana, where the burden of sickle cell disease is extraordinary and where the medical and tertiary care nursing support to safely perform bone neurotransplantation as yet just doesn't exist.

One of the big questions when you're thinking about designing the drug is drug delivery. The idea that when you give the patient that medicine, how do you ensure that that medicine gets to the target to act? And what you're describing is almost like an expanded definition of drug delivery. It's not just what happens when the medicine gets inside the patient. It's how do you get the medicine to the patient in the first place? And this used to be a trivial concern for a drug manufacturer, right? Because there's an entire infrastructure to ensure that. But it's not trivial when you're talking about some of the kinds of therapies that you've just described. The idea now that you're going to have to figure out how to manufacture something and get it

through the entire system so that that patient can receive it is a relatively new consideration. It makes these medicines as much medical procedures as they are therapies. Like where does the role for Novartis and for you specifically end in this sort of more complicated journey? If we are to reimagine medicine, we have to be agnostic about the learnings of the last generation and imagine that new therapeutic concepts can heroically be delivered to patients around the world. We innovated the first CAR-T cell therapy. This is a white blood cell taken from the patient, infected with a virus, expressing a novel gene, expanding and growing inside of patients as a living therapy as it kills cancer cells. The ultimate in personalized medicine,

it is a medicine that is manufactured each time for one patient. This is very expensive. It is very complicated. The outcome of the manufacturing process can even be uncertain. But in the desperate situation of advanced B-cell leukemia, this medicine is often the patient's last hope. To bring this medicine to patients all over the world, we need to reimagine the process of its manufacture. We have to reimagine the site of its manufacture. We have to reimagine the dose strength so that perhaps we could manufacture less cells and have them grow inside the patient, in effect collaborating with the patient to manufacture the medicine. And we don't always deliver the ideal medicine the first time it comes forward. We learn so much through clinical investigation

and ultimately the process of drug access that we in research follow this process very closely. Our responsibility doesn't end when the medicine goes to market. We take the learnings from the patients and the clinicians to try to imagine the next generation of the medicine. That was beautiful. Thank you so much for joining us on the A16C podcast. Thank you. Cool, molecular glue, man. I'm all about the glues.