Gene Therapies That Work? - EP 86 Adrian Veres

All right, we're talking gene therapies today. It has long been this, this much-hyped, much heralded technology from biotech land where you, you send a therapy into your body. It is much smarter than, than your typical therapies. It's going in, fixing your DNA for what ails you. The massive problem with gene therapy tech is that it has been Very expensive, hard to produce. It's been hard to send into the right places in the body. Often these gene therapies get all sucked up in the liver and don't go where we need them to go. Our guest today is Adrian Veresh. He is the co-founder and chief science officer at Dyno Therapeutics. They have been working on making A better gene therapy. For a long time, they were kind of doing this AI bio thing way before it was cool. They've been using AI machine learning technology to reshape viruses. Viruses are often how these gene therapies are kind of built and put into the body. So they've been trying to create a modified virus, call it a platform, a new way to deliver these therapies that is better, smarter. And faster than what's come before. They are a little different than many biotech companies. They're not trying to make just one specific cure for, for a number of diseases. They are— they're trying to be what people call like a platform play. They want to sell their technology to the likes of Roche and Novartis and other companies, some of whom are already their partners. Adrian is going to give us He's gonna give us the basics on how gene therapy tech works. He's gonna make you smarter on all this stuff. It's just fascinating technology. It is a tremendous listen, which is what I always say, but this is true. I am Ashlee Vance. This is Core Memory. Here we go. Adrian, thank you so much for coming.

Excited to be here.

I'm excited too. The, you know, I always try to do my own research reading things, but then it is the age of AI, so I always like type the person's name in and then I'm like, what would you ask them? And, and at least OpenAI is a big fan of yours. And it tells me I have a coup. Um, it said you're clearly one of the more accomplished young scientists in the gene therapy world and have stayed almost entirely entirely out of the press and that I, I'm breaking new ground here by getting you on. Let's do it. This is— I feel like we're going to be talking about your company, Dyno, and gene therapy. I always feel like this is a difficult subject. I think people hear about gene therapies in the news. Even me, as someone who covers it off and on, I always have trouble understanding Where we're at with this technology, sort of how it works. I mean, you always, you know, I think for, for the average person, you just, you, you think you're getting some injection and it's rewiring all of your DNA. And, and then, you know, we hear about these things, but then they don't seem to be out in the wild a lot. And, and so, yeah, I mean, the, I always like to just explain the basics of some of what's going on here. Does that seem like a decent place to start?

Absolutely. And I think the other thing to realize is I think we're going to hear a lot more and more about gene therapy. I think we've been in the background phase. And if what we and everyone else is working on, what's going to happen is we're going to be hearing more and more about it because it's just going to be a bigger and bigger part of medicine. So it's a great time to get the basics in.

I feel like that's the thing, is it's just Stuff you hear about in the news for so many years, and it obviously would come with all this, this promise if we have a brand, you know, a whole new field of incredible, quite personalized therapies for people. And then it's just, it always feels like a little bit of a letdown because they're incredibly expensive. They're rare. They seem to have bad consequences sometimes. And so, yeah, anyway, that's where— so, so let's, let's start here as one One item, which I think kind of gets at what you're saying. The federal government recently seemed to kind of like open its mind towards gene therapies, especially in the case of rare diseases. And so, so for, for a very long time, these have been people who are this, this difficult population where it's just not usually cost effective for a big pharma company to develop a drug for someone with this rare genetic disease. And so they get kind of neglected. It turns out for some of these diseases, we call them rare. It's actually quite a decent number of people. And if you add up all the rare diseases, it's a very large population of people. And so, you know, it's a difficult circumstance that is— I read about it really quickly, but it seems like this particular administration is saying we've kind of come far enough with gene therapy technology where if you're in quite dire circumstances, you know, we're— and there's enough evidence that something might work, we're going to be more lenient in letting you try this therapy without having to go through all the usual trials. Is that sort of where we're at? And is that indicative of something new?

That's part of it. I think that's tying into— we're seeing a lot more new ideas on the regulatory side, as well as on how people are framing what is a gene therapy. Like, is it more of a medicine? Like, is it more similar to a pill that you just take once, or is it more like a surgery where, you know, it's like a bigger intervention that you do once with a very specific intent that's actually different from person to person? And a lot of these ideas are pretty new. The reason we're also talking about them is the moment feels right.

Okay.

Like, the convergence of technologies that's enabling this is just getting better and better and it's helping. What's going to happen now is a lot of the challenges that you mentioned, like cost, safety, how widely applicable it can be, all of these are really technological challenges at the core root. There's really nothing that says— laws of nature don't say we can't do this. We just have to find a way to do it. And I feel that the progress we've been making as a field in the past few years has just been really incredible, and it is going to continue. And that's why I think we're going to hear— you'll be hearing more and more about gene therapies.

Okay, well, yeah, so let's bounce off that. I mean, why don't we— let's define a gene therapy. Can you help walk us through the different forms that are available and then some of what you're seeing that's starting to change?

Yeah, I think at the simplest level, what a gene therapy is, is acting at the genetic level to improve health. That's really what a gene therapy is. The point is you're going to intervene at the genetic level to have the impact that you want to have. The contrast with this is if you take a small molecule drug, you can choose to act at many different levels of the biology to You could act at the symptom. You could act at the very last thing that's giving you a certain feeling of pain, for example.

Yeah.

You can try to suppress that. Part of gene therapy is, as a framing, it's restricted to where you're doing this intervention. And what's helpful to know is we've been wanting to do this for a very long time. The reason for it is, in understanding disease, if you go back about 50 years, we quickly realized that there's a lot of diseases that we could tie back where that disease is at the genetic level. Meaning once we started, we understood that there's this thing called DNA. It's this stretch of letters. And we started asking, can we find— there's this terrible disease that exists, can we understand its cause tied back to the letters that have changed? And once we started doing that, the answer was, yes, we can. It used to be really difficult. Now it's gotten dramatically easier because of technology. But very quickly, we were able to point to, hey, there's this gene that causes this disease. In fact, there's many genes that are named in that way, like the cystic fibrosis implicated gene is really called CFTR. The CF stands for cystic fibrosis. As soon as we knew that, the next step was like, okay, can we fix it?

Can we just turn it off?

Yeah, there's a mistake. Can you just go in and fix it? That journey, It's been going on for decades now because it turns out that it wasn't so easy to just will it into action. There's a lot of components that we needed to bring into place, even actually just the changing of the letters of the DNA. That's like the promise of CRISPR, which everybody's heard about CRISPR.

Yeah.

Why was it such a big promise, such a big thing in our field and continues to be is because it brought in the ability to go make those changes. But then another problem was, say you're actually able to poke at the DNA, like change the actual letters with some tool, some technology like CRISPR. Well, you have to get the CRISPR in the cell at the right place for it to act. And that's another big challenge. And actually, probably the bigger challenge that's held back gene therapy is the ability to do delivery effectively, go in and actually, there's a bunch of different ways you can intervene on DNA and other components that are close to DNA. But actually getting the machines in, in the right cells safely, cheaply, that's been a big challenge. And that's actually the challenge that we at Dyno, we founded the company just to focus exclusively on that challenge. That's what we focused in.

To work on that. And so when we talk about small molecules, like a pill that we take, I mean, is it fair to kind of talk about that more as kind of mitigating a problem, but it's not fully turned— something's not turned off or turned on like you want it to be? You're always still dealing with this condition in some way.

I think that's a bit unfair compared to amazing medicine.

Yeah.

I think the reality is the small molecules give you more degree of freedom in terms of where you can go, and you can use that to have a causal intervention. If there's a way to do the intervention, small molecules can do it on certain contexts. The problem is that it's just a very hard hacking problem. It's like if maybe a programming analogy might help, you can't fix the soft— actually change the source code. And sometimes there's just an actual problem in the source code, or even if the problem isn't in the source code in the first place, if you could go in and change the source code, that'd just be a very direct way to fix it. And instead, with small molecules, sometimes you have to find some other way of like, oh, this thing is happening here. So if I Poke it here, you can have a direct impact. And genetic medicines complement other traditional medicines, but they offer a different angle of attack, a different point of leverage that's very, very high. Because once you change the source code, the premise is you should be able to make a change that lasts forever.

Yeah.

For a very long time. And that's the potential of— that's what's such a big potential and why the field has continued. you know, in the face of many, many challenges, just kept at it.

Yeah.

You know, decade after decade.

And over these couple decades, I mean, I hear all the time about monogenetic conditions, the rare diseases that we're talking about. Often they fall into this group where it's a single gene that is causing a lot of mayhem and it's disrupting a very important process. As time has gone on, well, obviously there's other conditions that have many genes at work. And when we're talking gene therapies broadly, are we in this moment in time just very, very focused still on these single gene problems? Or is there a thought— I could never tell if the— once we get to the multi-gene level, it's just too complicated to even begin to address that.

I think it's a matter of the order of events. It's very natural that where there is a single gene, the other thing that's very helpful in a monogenic disease is you understand the causal biology. You really know that if you have all the evidence that if you go in and make this change, you're going to have the impact you want to have. And that's actually just really valuable in biology because often in biology, the trick is in biology is you can have a story that's like, really beautiful, it makes a lot of sense. There's a very clear argument. And then we've seen time and time again that when we actually test the argument, you're missing some component in the story.

Right.

And so it doesn't work out. For monogenic diseases, that's so crystal clear, so that's why the focus is there. But nobody in the gene therapy field is saying, cool, we're going to do single gene diseases, and then we'll have to find something else. I think we're actually all excited about the potential to intervene at the circuit level, like to have to intervene on a combination of things to really shape cellular behavior. And that's absolutely coming in pretty soon, actually, given where the technology is, the pace of technological development right now.

And so Dyno is focused on viruses, viral vector delivery. What are the methods at our disposal in 2026 if I want to deliver a gene therapy? What are my options?

There's a couple. Yeah, there's a couple options. Um, the challenge, ultimately the challenge, it's helpful to understand the challenge is you have a piece of DNA or a piece of RNA, you have to get it inside of a cell. And there's really like 2 ways to do this. Um, one is you have to put it in some sort of box that's going to get into the cells. And That box can be chemically formulated. This is something like a lipid nanoparticle where you just mix a few components around, and depending on the conditions, you can now achieve that goal. Then there's another set of things, which is what we normally do in biology, is rather than invent new things, we look at what nature's made and we borrow and steal it.

Yeah.

Viruses do this by default. There's many different viruses that do it. take them to do this delivery for you. Better yet is you can actually, which is what we do, is we engineer the system to actually make it do not just what it was naturally evolved to do, but we design it towards a specific purpose and really turn it into something we picked up from nature into a technology that's going to be able to— give you the characteristics you want for any disease that you might want to be treating.

This episode is brought to you by Brex. I spend all of my time inside of the world's most ambitious companies making rockets, semiconductors, robots. Very often the business is struggling, and then all of a sudden things click and take off. Then the company's expenses spin out of control. They either have to clamp down, start counting pennies, We're cross their fingers, hope for the best that the quarter or the year comes out okay. Brex was built to handle exactly these situations. It helps you move fast, spend smarter at the same time. They've got credit cards, banking, expense management all in one place. They've got AI technology looking over everything, taking care of tasks for you. No more chasing down receipts. We use Brex at Core Memory. It has been a godsend and given me back so much time. Giants like Anthropic and Zoom use Brex as well, and so do tens of thousands of startups. You should use Brex, and you can find out exactly how to do that by going to brex.com/corememory. Okay, and then tell me how badly I butcher this, but, but so viruses have this, this kind of outer shell called a capsid, I think. Um, I'll go and then you—

Some viruses do, but yes, they all do in some way.

Okay, okay.

So far so good.

And, and, and so, so nature is very good at this, like you said. I mean, we look to nature, obviously we all know once you get a virus that gets in there, it goes to work pretty quickly sometimes, um, doing, doing what it's trying to do. And, and so you want to create a virus that's not just doing what it normally wants to do. You want to reprogram it to do what we want out of it, but to kind of use its mechanism. And so there's this, this outer layer that it seems like we can manipulate and, and try to get our, our the genetic code that we're— the tweaks that we're trying to make into this outer layer and then have this system feed into your body?

Yeah, so let's talk about that. There's a couple interesting things. I think it's actually fair to— these aren't viruses anymore. Once you start using them as, you know, you take out just the shell.

Okay.

Like, we've borrowed a component from a virus, but it's helpful to know, like, fundamentally what makes a virus a virus is it can go in, It's a machine, it's a self-replicating machine. It's kind of one of the minimal, it's not the most minimal, that's probably prions, but it's one of the smallest self-replicating bits of biology out there. And its goal is to self-replicate. Well, we don't have that goal, right? We actually want to, in its process of self-replication, it has to deliver itself somewhere where it can take over the cell for its own purposes of replication. We're borrowing just the delivery component, so the capsid. So that's actually the component that we're engineering. We're engineering the capsid. And you're also hinting at something interesting, which is there's quite a bit of viruses. We know many of them, but we actually know a tiny fraction of what's out there in people, also in the animal kingdom overall. But what the field has done is actually there's a virus that Has piqued a lot of curiosity because it just seems to—it's just kind of not a very good virus fundamentally. It's in fact, it can't even replicate on its own. Like it actually needs another virus around to be able to replicate, and it seemed to be that it's very simple. So it's protein only, and it doesn't cause much of an of a reaction to the body, and also it doesn't cause disease. So if you have, if you are infected with that virus, you actually never really know. And so that virus is called adeno-associated virus, it's AAV. And because of these set of properties, it was just very intriguing for the field for decades being like, wow, if we had to take over the components and use this to deliver gene therapies, this is a great starting point. And that's the story. Those are the giants whose shoulders we're on, is this progress of finding this one, testing this hypothesis, building it and building it. What's happened in the last few years is a convergence of technologies, including AI and this ability to explore the sequence space in new ways than before that have converged both in the lab and in AI, and enabling us to really just customise these capsid shells to do exactly what it is we want them. That's been our focus.

Do we know why this particular virus— what was its purpose then in the world?

It's hard to speculate on these things. By default, if I had to speculate, I'd just say like most viruses, they want to replicate. That's it. Their purpose is to self-replicate and continue.

It seems like it's bad at replicating or—

I mean, it's making do. It doesn't have to be the best. Like, if it weren't able to do it at all, it wouldn't exist.

Okay.

So clearly it's good enough. And, you know, that's good enough for me. The interesting thing is the, you know, it's got this funny name, adeno-associated, because people were studying adenovirus. And then when they were looking like in electron microscopes, they saw like, wow, there's this other thing that sometimes pops up. It looks like a virus. Could it be? And it took a while to figure out like, yes, it actually is. And it's interesting in its own right.

Okay.

And yeah.

And so when you guys are, when you're talking about changing the shell, I mean, what, what are you doing? What are, what is actually happening? And then, and then what, what do those changes allow you to do?

Maybe it's helpful to describe what this is because it's actually, it sounds complicated, but it isn't. And you can picture something very, very simply. So this capsid is, it's helpful to picture like a, I said a box, That undersells it a little. Picture a really nice origami box.

Okay.

What you want to picture is this icosahedron thing, and there's one piece that is repeated 60 times. That piece is actually very small. That's a protein, and that piece is defined by about 700 amino acids, which are the letters that define proteins. So it's just a string. It's really quite literally just a string of these 700 amino acids. Now, those amino acids are incredible, right? Because those 700 amino acids are able to together create this icosahedral origami box. And then that box is able to— you can put DNA in that box. So when we do the engineering, What we're doing simply is really just changing the letters. Of those 700 letters, if we find the right change, well, we can directly— that can change it if we want it to go to the brain, because we want to make gene therapies for a neurological disease. If you find the right set of changes, you can direct it there. And now there's a lot of very interesting science for how we do that, but that's the core concept of the problem we're out to solve.

And nature has had millions and millions of years to work on this. But the thought is that we can make these adjustments and still be good at creating a change, like even one-upping nature here.

So there's a couple challenges, right? One is nature's never Nature was never optimizing for our own goal. It wasn't trying to make an amazing delivery vehicle for brain gene therapies or muscle gene therapies. I mean, I guess we are now, so nature is finding a way in a complicated path, but the actual evolutionary process at the viral level wasn't aiming to create that. So that's why it's obvious that once we set a new objective, we're going to be able to do the engineering. Now, that engineering though is pretty hard because what's happened is if you look at nature and it's a very, very finely crafted machine, that's that little protein shell. We've actually run this experiment of what happens if you take those 700 letters and make for every one amino acid, there's 19 others. You can make the change to any other amino acid. We've done this experiment. This is work done in the Church Lab at Harvard prior to the creation of the company where we made every change. And quickly, that explains to you why the problem is hard. When you do that process, about 70% of the changes just break the shell. So it tells you that if you're going to be changing it, you really have to understand, okay, if I'm making a change here, got to make another change somewhere else to make up for it and keep exploring the space. That's one piece. To your question of also how come there's room to do the engineering, one thing that's really cool is if you look at existing AAVs in nature, there's a few that are very different from each other. You can pick out 2 things that are clearly AAVs, like they make the same type of protein origami shell, but they differ in half the positions. That tells you that actually there's room to engineer. There's not one solution, there's many possible solutions. If we explore that space effectively, we can get there. And then the other big piece of good news that we found was once you get that data, so for example, once you make every single change and find the good ones, you start doing combinations of changes. One thing we noticed, and this was before the AI moment that's happening now, this was around 2017, 2018, we noticed that actually machine learning techniques are exquisitely good at picking out patterns and allowing you to get traction on this engineering problem. Much better than people, much more aggressive, but also just able to pick out amazing patterns. And that was the core bet of our company, was let's apply these technologies of let's measure many changes all at once, which was the traditional Church Lab type of technology. And then let's combine this by applying machine learning and AI to take the information from all of those changes and put it together towards a very specific practical objective and then put it together and see what happens. And now we're a few years into that journey, and the results are really exciting to us in terms of the implications of the impact of this output.

Yeah. And I wanted to kind of come back to that a little later in some ways, because you guys did start— the company was founded in 2018, I think. And so that's like Transformers are just being written about in a paper. I mean, that was way before this Big AI revolution. It's way before 8 million, it feels like, uh, AI biotech startups were, were peppering me with, with, we're using AI to redesign proteins and, and all this stuff. Like, you guys clearly, you know, were kind of on the, the cutting edge of all this. And, and maybe we could describe what I was reading about, um, Dino. I mean, the way it's— It's presented in the literature is that you're, for this capsid, this outer shell, that you're using machine learning to try millions and millions of different combinations of these, what, of these amino acids, of these proteins, and then trying to get to a point where you have the most promising ones. And then I don't know if you test on animals or not. I presume you probably test on mice and things like that along the way. And then, so you're—

Yeah.

You're starting with these millions of things. Machine learning is kind of helping winnow it down, and then you get to sort of animal testing, and then you move forward with the most promising ones. Is that the general idea?

You have the right pieces.

Okay.

But let me put them in the right order. One thing that the field of gene therapy learned pretty early on was, especially for delivery, it's really valuable to get to the ground truth as much as possible. And this means working in animals from the beginning, especially in the most relevant animal species, for example, like primates.

Okay.

And the focus there is also actually learn— these are challenging experiments. These are ethically weighty experiments, making sure we learn the most from each one of these that we have to do. And what we're able to do is we have the ability to Make a collection of these different capsids, call it about a million. It's not that large of a number, but I'll get into why that's actually better in many ways. It's actually, we chose that number because then we can get measurements on each one, and we can get that in vivo. So in one experiment, you can measure for this half a million to a million capsids, how many went to the brain? How many went to the liver? How many went to the muscle? How many went to all the different organs? And allow you to see how are you changing from the natural propensity of the capsid, which—

Oh, okay. So you put many in at the same time?

Yes.

Okay.

Yes.

And then there's markers on there so you can track them.

Yeah. So the way to think about this is since you have that protein shell and inside of it is DNA, every time we make a shape on the— a change on the outside of the protein, which is what we're trying to do, we record the nature of that change on the piece of DNA that goes inside the shell. It's using a barcode. We're essentially encrypting it inside the box. But then what that allows you to do is when you actually— you can mix several of these together. They're labeled by the piece of DNA that's inside the shell. And when you put that in an animal and it goes to the brain, if you then in the brain take out the DNA and say, all right, what genes have been delivered here by AAV. And then you read the barcodes, and then you can say, okay, so which capsid did that? That's a new— the natural one didn't do this or did it very poorly. What about which of the new ones did it? And that gives you exactly the type of data that you want to understand. Now, the problem is the first time you do it, it just doesn't work very well.

Okay.

Like, you barely see any signal, right? Like, you'll see a wisp of like, wow, maybe it went a little away from the liver and a little into the brain. But as you iterate, as you take that data, feed it into machine learning algorithms and ask it, okay, learn from this, find the combinations which are going to be most promising in silico. When you're doing this design, you also get to specify exactly your wishlist. You can say, go here, go here, don't go here, don't have these features, be easy to make so that this doesn't cost a ton of money at the end. You can actually specify all these features and have the machine learning compute stack just run through billions and billions of possibilities, and then come down to, all right, the next million you should test is this one. Then we do the experiment again, and again, and again. That's the approach that we've been taking and we found to be very fruitful.

Are the markers, do they fluoresce, or how do you find that barcode?

So the barcode is a piece of DNA, and because it's a piece of DNA, it's— the 2 things are going to happen. And we actually measure both because we care a lot about this data. Is, you know, DNA gets expressed into RNA. That's, that's how, you know, genes turn from their DNA information into a protein that can do stuff. And we can— and both DNA and RNA are information-carrying molecules. Like, their literal purpose in the cell is to just ferry information around, which means they're actually kind of easy to read and write. So we just read them off with sequencing.

You just sequence it? Yeah. Okay.

And we do this in a pooled fashion. And because everything is barcoded, when we see a certain sequence, we know that that came from a certain virus we designed in a certain way, and then can benefit from that information.

So you give an animal an injection?

Yes.

And then it acts for a while, and then you just take a blood sample or a tissue sample?

So we take out tissue samples of every tissue. So that we can learn as much as possible.

See what happens, what went where.

Exactly. Sequence each of those tissues. And then ultimately, this just becomes this giant table.

I see.

Literal, you could open it in Excel. That's every tissue, every virus you tried. Did it go— is it more than the baseline or less than the baseline? And then from that measurement, you can look at it. Actually, sometimes you see like, wow, that's really promising. We should just Validate that one because it's, you know, it maybe it does what we need it to do, or, or at least it looks like a step in the right direction. We can confirm that, but then you can then also repeat this process.

All right, pod friends, I am here to tell you about SendCutSend. They are a manufacturing phenomenon, whether you're working on your own car or you're a giant automotive company, a rocket maker, If you need metal cut, machined, or bent, you're going to SendCutSend. They'll do it for you right, and they will do it for you fast. Go to sendcutsend.com/corememory for a 15% discount on whatever you're trying to make. We love SendCutSend. Go visit them today. And you mentioned the liver pretty early on, which is an issue here, right? Because we've seen that gene therapies tend to— the liver likes to absorb things and So it's picking these things up and this has been a problem, right? Is you want to create something to help a condition in the central nervous system, in the eye, in the brain, and the liver is gobbling everything up and then it's creating toxicity issues. And I mean, that's part of this quest, right, is to get this out of the liver and to the spot where you want it to go.

Nailed it.

Okay. And this is— I see on your website, I mean, you guys really emphasize this a lot. So this is because this has been the central knock on delivery is just That we come up with all these great ideas and the liver sucks everything up and kind of ruins the party.

It's more than sucks everything up. In some cases, the viruses are homing there because that's where they're going to do their replication. And so if you take the delivery component from that virus, it's not surprising it's going to keep going to the liver. That's one piece of it. The other piece of it is it's helpful to understand, you know, why are we seeing the toxicities that we're seeing? The reason for it is actually, if you take just the capsid from a natural virus, it almost looks like it should work. And then when you try it, you're like, wow, this thing's actually quite safe. You can put in a lot, like a huge number compared to almost any other virus before the immune system has even almost any reaction. And then so what we do is we just turn this crank because, for example, if you're making a muscle gene therapy with an existing capsid, it goes there a little, So if you increase the dose enough, you're going to get enough delivery. But you're faced with this trade-off of you can't go as high of a dose as you would need, but you want to go high enough to have a shot at having a strong impact. But then you're kind of— we call this the therapeutic index. It's just very narrow. And that's why we're so focused on, well, if we can just massively detarget the liver. It's going to widen completely the therapeutic index. But that's also a place where we could do just that. That would be great. But actually, we're more ambitious than that because we can detarget the liver to help with safety and at the same time, just make it go more to the place it needs to go.

Okay. And then on you guys' website, I noticed you were calling out promising work around muscles, around the eye. around the brain and central nervous system. They kind of get lumped together sometimes. Like, I would like to— I'd love to break down for people still how this works when it's doing what it's supposed to be doing. I don't know if we want to go through all 3, or should we pick one of these?

Pick one. Let's pick the brain.

Okay, so let's do the brain, and let's kind of describe. So you've done all this work to make this capsid the way you want, and you're getting past the liver. And then, so what are we doing in the brain?

So you've got this protein shell with a piece of DNA in it. It's going in the blood. You inject it in the blood, and it's just moving around in the blood. The first challenge you face is the brain actually— nature's designed it quite well. One of the things it has is this thing called the blood-brain barrier. Think of it as just a bouncer.

Yeah.

That's literally deciding, oh, you want to get in the brain? You have to have very specific things to let in.

It's the ultimate VIP.

Exactly. But, you know, we can learn about that bouncer, and we can allow our protein shell to make it through that first barrier. That's actually one of the big challenges in the brain, the first one.

Okay.

Now you're through the blood-brain barrier. There's a neuron sitting there. So the virus just touches the cell, And this is part— I'm going to have to use somewhat vague words because we barely understand how this actually works in practice at the molecular level, but we have hypotheses. And it's going to start making it inside the cell. And then the virus is— actually, the cell is going to start detecting, like, this is kind of weird, I should just put it in the recycling bin. And when that process starts to happen, the capsid knows, okay, it's actually time for me to if I don't do anything, I'm going to end up in a recycling bin, so it's time for me to get out. It sort of wriggles out of the state that it's in and then has and finds a way to make it to the nucleus where the DNA of the cell resides. Once it's there, the job of the delivery platform is accomplished. You've now delivered a new piece of DNA inside of a neuron. And I've described one neuron, one protein, but one capsid. Obviously, you're doing this with billions of it all over the brain. That's the goal. And now what can happen? Now it's up to you. If you were making a drug, you can now do any number of things. For example, that piece of DNA could express a CRISPR system that goes in and fixes a specific genetic disease. Maybe you don't even need to do that. Maybe the gene is just broken and missing. So all you need to do is you can just express the gene directly. And that's it. That gives you a path to curing a specific disease. And this also highlights, actually, I'm not describing to you a drug. I'm not actually telling you I'm going to go in and express that one gene to cure that one disease. That's actually because that's not what we're doing. We are not building drugs, which is different for our field. We're building this delivery technology.

The mechanism, yeah.

And then working with many partners who say, I need delivery technology, and I have a very, really good thesis for how I'm going to put in a payload inside of this capsid and go and intervene causally on a disease. And we're doing that together. And that type of partnership has also been just really, really fun in the story of our company.

Yeah. And just going back to the neuron for one second, I mean, so once you've made these changes, the neuron is doing what? It's either producing more or less of something, essentially, or—

So it has to start by producing It has that piece of DNA. It has to start by turning that into RNA. And this is where things get kind of cool.

Okay.

And let's find the right altitude. But that piece of RNA can do a number of things. The piece of RNA itself could go in and make changes to other pieces of RNA. So the simplest is it could go and shut off another piece of RNA. So that enables you to turn a gene off. which for many different diseases, that's a great starting point.

Yeah.

Let's do that. Actually, there's pieces of RNA which now are more complicated and can do interesting things. They can actually edit the RNA directly, and that's allowing you to now change the behavior of another gene in a way that's going to help improve health. That's very direct. So that's one way, and this is even just RNA. It's actually, these are maybe the more sophisticated for the connoisseur approaches. The simplest one to think about is You go from DNA, you go from RNA, then you make a protein. And now that protein, maybe the natural protein is missing in some way, so you just have to replace it. And there's many gene therapies that have that approach. In some cases, that protein itself is then going to go do something more complicated. So for example, if that protein is a gene editor itself, then that protein's going to go and find the genome, make that change. And you can now kind of see, what the potential of genetic medicine is, right? Like, you have all of these different ways to intervene on this, like, central repository of information that is the DNA and RNA of the cell. And, uh, ultimately, it's such a point of high leverage to go after disease.

And like you mentioned, I mean, so if we're talking, um, muscle, eye, brain, you guys— normally I hop on some biotech company's website and they have their pipeline of their drugs and which stage of a trial they're at. When I was looking at you guys, Like you mentioned, you have this list of pharma partners that you're with. So they are licensing or creating some deal with you around this delivery mechanism, and then they're testing whatever they want on the eye, the brain, their particular— whatever they've discovered, their gene edit that they want to push in there.

Exactly right. And it's not just pharma. It's not just the largest companies, although They're very interested in genetic medicines, and that's an amazing moment there. It's also biotechs at every stage. It's also increasingly nonprofits or families who are trying to make a drug in an N-of-1 setting. And that's really speaking to the moment that the field is having through this convergence of technologies of if you can bring in, solve one component really well, it's going to unlock everything else. I also want to tag that ultimately, if, you know, if in your intro you were hinting at, man, progress in genetic medicine has been kind of slow. Yeah. If you had to blame it all on one piece, it's probably delivery.

Well, okay. I was just about to ask you. Okay. I feel like we've set a framework, you know, to have everybody up to speed on the basics here, what you guys do. And so, yeah, this is where I wanted to open up is I just have a bunch of questions. I mean, I had Jake Beecraft from Strand Therapeutics on my podcast. I had Jennifer Doudna. on, you know, both of whom were talking about this delivery problem. And so, okay, like 20-something years ago, 20— what? Yeah, 26-ish years ago, we sequenced genome. Almost immediately everyone's like, oh, we're going to cure everything. This is going to be pretty easy. It turns out to be really hard. Some years pass, we get CRISPR, and it's like, okay, well, now we have this great tool to make all these edits, and everybody gets all excited again. And then it turns out, okay, this is hard. Sometimes when we make these edits, we're Really ripping up the DNA in aggressive ways. We can't do this on as big chunks of DNA as we want, and we need to be more precise with how we go about this. And so, I mean, this has been— what happened? Like, was this overenthusiasm, or like, were people just not paying enough attention to this delivery issue? Or this is just the nature of how technology is developing. You got to do all these basic steps and then focus on this, this end, endgame? I think it's like, do we, do we fuck up, or—

I don't know. No, not, not at all. I think all of these things, it's very hard sometimes to know, especially in biology where these systems are so complicated. It's sometimes hard to know what is like really hard versus actually this is easy, you just need the right solution. Like once you actually put the right solution in place, like bam. You just solve it. So it's sometimes hard to know which of these combinations of things and where do you want to go. For a very long time, actually changing DNA in a programmable fashion was this impossibly complicated problem. It felt like, how could we do this? There's actually very few ways to copy from nature. That's our first trick is, can we copy from nature if it's doing this somewhere else. And actually programmable changes to DNA, not something if you look in most of nature, you don't see that. And it took some incredible scientists to find and be actually, wait, this actually has the features, like some very strange system that actually is doing this for a reason to protect itself from viruses, funny enough. We can borrow this. And then suddenly that problem went from top of the list of, wow, we really need to sort this out, to we have traction now, we need to engineer it. Delivery, maybe part of it is we felt like we had more of a handle on it from the get-go, because for many years, actually, we'd picked up these viruses from nature, we were going somewhere, it was going pretty well, it was pretty safe. And then we also upped the ante of where do we actually need to go? Part of the challenge is once you realize, oh no, to have impact, It's not sufficient. If all disease that we needed to treat was in the liver, I mean, we'd be almost done, right? But that's the challenge is actually we want to go in many different places. We're going to go there really effectively. So we've just got to put these pieces together.

Okay. And then when you guys are calling out muscle, central nervous system, eye, is there a reason it's in those buckets? Like, if you figure this out, why can't we be delivering this to all the parts of the body?

But we want to get everywhere.

Okay.

We absolutely want to get everywhere. The reason we started there is it felt strategically to us and really to the field that this is a place where you have a ton of unmet need and probably no other path to get there. For example, on our list, we used to have liver. Now LNPs just got pretty good. But actually to get into CNS or to very effectively get into every muscle, We think using AV capsids is, you know, ultimately the— by far the front runner. And with the engineering approaches that we have, it's, it's, you know, starting to look like we have the right solutions in hand and we have momentum to make them even better. So it's an exciting time for that reason.

Okay. And I don't want to put, uh, words in his mouth because I don't understand this field well enough, but I feel like when I talk to Jake Beecroft, you know, and other people, they always tell me The virus stuff is just really expensive. It's just, it's never going to be the path because it's, it, when you see how much like these, these one-off gene therapies cost, it's millions and millions and millions of dollars. And they say it doesn't get like that much better, um, even for broader conditions. And, and I guess Strand has committed to lipid nanoparticles like you're talking about, mRNA being delivered, you know, as, as a much cheaper mechanism. Is there something, first of all, like I don't know, I guess address that, but is there something you guys have done that makes yours somehow cheaper than what other people have done with viruses?

Absolutely. There's 2 challenges really. One is, well, first of all, I just want to say on that perspective, I think it's always a tough bet to bet against engineering technology. I think there's a lot of things that a lot of people say, it's like, that's not going to happen. I'd say let's try it out and let's see what plays out. I think part of it is just developing the right technology. There's some incredible companies that are saying, for example, manufacturing even natural AVs before we engineer them to make them easy to manufacture. Manufacturing natural AV is a pretty expensive process. If you break down the process of why it's hard, there's a bunch of different reasons. There's a very creative startup in San Diego that's saying, what if we just made this in plants? Because it turns out it's just expensive to grow human cells in giant containers that you have to stir and prevent from being infected. But you know, one thing we're amazing at is just growing. Have you ever seen a tobacco field? That's a lot of biomass. And can we move the engineering there?

It's making that AAV sequence in a plant?

Yeah, that's exactly what they're working on. And it sounds as Crazy as that. Who is that? Circium.

Circium. Okay.

And that type of approach is exactly the type of technology unlocks that we can bring. It also comes down to the community aspect of there's a few different pieces of technology, and in each one, including on lowering cost, folks are working on doing this. Maybe you're saying actually plant, that sounds making avian plant, Wow, that sounds too sci-fi. Can you just automate better, make better bioreactors, and just use AI to guide you through the manufacturing process so it becomes cheaper? So I think we're already seeing huge unlocks on the cost in the manufacturing side of things. Let's actually talk about the capsid for a second. One thing is imminently engineerable is if you make the capsid 10 times better, you can lower the dose 10 times. So that's it. You've got a direct cost unlock there by just making the capsid itself more efficient. If you make it go to the brain more than the liver, the ones that were previously going to the liver were wasted. So you had to make up for it by adding more. Now you can just make cost savings right there.

Okay.

And then the next step is the actual process of manufacturing. The way this nanomachine assembles is also engineerable. And we can focus on that too to drive down costs. So I think ultimately for cost, the perspective here is it's kind of like solar panels. used to be super expensive, and people were saying it's not going to scale. And we started doing more and more, and the cost went down and down. I think that same trajectory is going to happen to genetic medicines, including delivered by AAV.

So do you think— are we— you know, that solar process was a multi-decade journey. Like, and I— maybe it's impossible for you to speak on behalf of your partners in all this, but, you know, if I was getting your therapy, a therapy based on your underlying mechanism, your platform today, and what— I don't know, let's pick the brain again. You know, one of your partners has come up with this great drug for the brain, and it's going to be delivered with your mechanism. Would this be today like a multimillion-dollar type thing?

No, far from it. Okay. I think we're looking at costs that are closer to the thousands or low tens of thousands, depending on the dose and the approach. And Even that's going down. So already, we've dialed in, in the past, less than the past few years, dialed down a 10x drop, and it's going to continue. So I think the reason for it is obvious, is people looked at it and saw it's going to be a huge problem.

Yeah.

If it costs $1 million to make this drug, the impact of this drug is going to be tiny. And we've put our technological and engineering capabilities to lower that cost. And it's worked. So it's advancing.

All right. This will be my nastiest question. You know, you guys, in some ways, you were so far ahead of the game and where we are now. I just wonder if you were almost too specific, because if I talk to a company like, say, Chai today, they'll tell me we're not just doing proteins and amino acids designs around viruses, we can do anything, you know? And you look at, I don't know, just some of the work Google DeepMind puts out. I mean, it seems quite broad in what they're attacking. I, if I open my inbox today, I guarantee you I would be being pitched by a biotech company that tells me they're using AI to design all kinds of stuff. And so, you know, George Church, um, who was the professor that you worked under the most?

So I worked with Doug Melton.

Doug, right.

So making beta cells in vitro, you know, different, different thing.

But I mean, but I mean, you know, both pioneers in their field, at the cutting edge, and then it feels like the world has caught up in some ways to where you guys were thinking. Yeah. So is this a curse or a blessing? Were you too narrow, maybe?

You're observing something great. And actually, there's 2 important things here. One is, I saw this elsewhere, so I'll borrow their phrase but say it, Show it in vivo or it didn't happen in biology is really important. I think actual proof—

In the body, not just in the—

In the body, not just in simulation, is really key. So that's one piece where I'd say once you do that test, you actually see much more clearly what works and what doesn't. The other piece is we noticed this too, meaning we noticed that as we got better and better at engineering capsids, we realized that everything we learned on the machine learning side, well, could be applied to but proteins more generally. And it's also a place where we're starting to share that technology. So we're presenting our foundation-scale models for broad protein engineering that have come from our experience of building capsids. The other thing that's cool with capsids is these are really hard to make. So in a way, it's like if you're going to the gym and doing the hardest possible workout, that's going to help you for the general problem.

I see. Sure.

And that's the experience we've brought in. We're also excited to share our technology both on the models that do the protein engineering, as well as the approaches to use those models effectively, not just have a thing that only a machine learning wizard scientist can use, but actually make it very easy for any scientist to deploy this technology at scale confidently. We've benefited from many, many reps of doing this ourselves where we Don't just ship the model, but ship the model, use it, put the sequences, test it in vivo, see what works, be disappointed by all the stuff that didn't work, learn from that lesson, and repeat that. I think being early is never a curse, but you got to keep your mind open to how can you have impact from what you've learned.

How will this play out when we see the work that Google DeepMind has done in science? I mean, they've won a Nobel Prize, and then we see OpenAI, Anthropic, these giant companies with billions of dollars behind them and not infinite compute, 'cause it's being dragged in all sorts of places, but the world's biggest computers. And they've gotten interested in biotech, obviously, sometimes through a partnership and then sometimes trying to do their own internal work. If I think about fields like brain research and things like that, I'll talk to brain researchers who are doing connectome stuff, trying to build these maps. And, you know, they have to, like, to do the work that they want to do, they have to fight for time on these Google supercomputers. And it's like, have you Have you buttered up the person at Google enough to sort of be put at the, at the front of the line for a startup? I mean, you know, I obviously, I can see a world where a focused startup is gonna make more progress than some broad, um, some broad company that this is not their sole business and, and area of attention. On the other hand, it just seems like, man, we get to some point where these the foundational model companies have so much resource in their model and their compute that if they— yeah, I mean, do they become competition here?

Maybe. The way I'd frame it is there's 2 things to say here. One is access to compute is key.

Yeah.

And, you know, we've mentioned our pharma collaborations. We've also had the privilege and the honor to be partnered with NVIDIA as we've been doing this work. And that's been, you know, a fantastic collaboration to like push us and work with, collaborate at the very leading edge. So we absolutely want to stay there because it's critical. I think there's 3 things. To make AI work, you really need a combination of 3 things. You need compute, you need algorithms, you need data. Data is a place where if you don't have the right data, it's just really hard to intuit the shape of the problem, especially when you're working with completely new problems or new questions. delivery in humans or in primates is a problem where you just don't have the data. You have to go make the data to apply it. And it's also not data that's publicly available. So we've created that own dataset. The other part that to me is exciting about all this progress and the scale of people wanting to contribute to this is we're doing this for our mission, right? We're doing this to make genetic medicines happen. We're calling this concept genetic agency. We want to be in a place where you can make decisions. There's not one drug available for some diseases. There's several drugs for every disease, and you can make choices about which one is right for you and when. That's a really meaningful problem. It's going to take a lot of effort, and applying the combined effort of all of these individuals, organisations, now increasingly AIs, We're going to need all of that. It's just not a problem that we're going to look around and be like, wow, that was easy to solve. The more we increase the scale of the resources that are deployed to this, I think that's all good news.

A bit on that point, I asked you on the cost elements, but again, maybe it's hard because you can't talk about what your partners are doing or you don't want to speak on their behalf, but this initial batch of work that's being done with your technology, is that going to be Will the therapies that result from that in the near term be for limited, more on the line of the rare disease type populations, or are some of your partners going after quite broad diseases in the general population?

Like you guessed, I can't comment on any specific, but I can tell you that the answer is yes. It applies to each one. It's also part of it is we have several partners that have different strategies. Some partners for If you're a foundation focused on one disease, they're going to be focused on that one disease, and it's going to be as rare or common as they want it to be because that's their mission. Then there's organizations that are working across several diseases, and we're happy to support that. That comes down to why we made the company the way we wanted to do it. As we created these capsids that unlocked possibilities, We wanted to make sure that the impact wasn't limited by the drugs we could make using that one capsid, but instead the impact could be maximised by everyone being able to— who wanted the technology could access it and use it and build their programmes. And so it's kind of this full-court press on disease using genetic medicines with highly engineered delivery capabilities.

Okay. All right, we're going to broaden out for a second because I have, I have like more general questions for you. I mean, one we've been dancing around, or we've addressed it somewhat, but, um, just to put a fine point on it. For 20 years, I have been told that biology, we've, we've gotten to the point where biology is like engineering. And we've just gone through many reasons why it does not seem exactly like computer programming. I mean, is this ultimately like a— False promise. I felt like 20 years ago, synthetic biology became this term that started to enter the world. And for the reasons I outlined, we sequenced the whole genome. It's like, okay, this is just, this is code, this is math, this is all stuff we know well. I think your background, you're a physicist and a chemistry undergrad at Harvard. I mean, so you're kind of approaching this from the— a brain of somebody who wants to engineer the world and deal with these concrete things that we can manipulate. I just, and I don't feel totally let down. I love this stuff. I cover it all the time. I know things take a long time to develop, but like, have we been sold a false bill of goods? Will biology always be so much more complicated than code in a software program?

It's funny because I want to say yes, but I worried you'd be disappointed. To me, that's a good thing.

That's a good thing. Okay.

But let me explain that. I think unpacking part of it is, I think the challenge with biology is this multiscale computational process. And part of it is we have this desire to be like, cool, okay, we understand one of the layers. That's it. We're good.

Yeah.

Because if you write code or if you actually engineer a system, if you're going to have multiple scales, you actually want to have very elegant abstractions that separate them so that you can operate on one independent from the other. Biology just doesn't play with those rules. It's just squishy and it's allowed to just make shortcuts between the scales and how things impact. That's the main challenge is it's actually very hard for us to keep track of a lot of that. That said, every time we make improvements on the ability to measure the biology, the ability to make sense of what we measure, the ability to interrogate it in new ways, we actually unlock more of that bigger picture. There's probably always going to be surprises. Even in the far future, we're going to look at something, we're going to find some bug somewhere that does something where we're like, wow, We'd never thought of this. But I think that's what is framing the problem that's hard. That's also why it's beautiful. There's this complexity that's just such a puzzle, and it is so satisfying as a scientist. For a while, when we were starting Dyno as a company, it was just so fun pitching machine learning people on this problem, being like, hey, 700 letters goes everywhere in the body.

Yeah.

That's it. Now there's a question mark in between. help us puzzle the question mark. And you can see people be like, oh, wow. Yeah. So, you know, it's coming. Um, will it ever be completely computable? I, I don't think so. Um, but I think that's great.

Yeah, I just, you know, I just get, I get hopeful and then I get frustrated. It's like if I cover iPSCs, I cover mRNA, what you guys are doing, I get so excited. And then if I talk to Each person in these various fields, you know, the the IPSC people will tell me no, you know, like ten years we got this thing working for sure. Gonna reprogram our organs back to their state, and then I go talk to my other very smart rational friend, and they're like no, whole field's bullshit. You know, it's not gonna work. I always feel like there's there's a rug that gets pulled out from under us on on each one of these areas, and that I mean I got so synthetic biology when I first started to hear about it just I was like this makes sense. This is blowing my mind. How amazing! And then we have done amazing things, but yet, yeah, it just, it feels just like a brutal, brutal slog. And I think everyone's— I think the way it gets presented to the public sometimes by people like me is like, you know, somebody is going to unlock this universal key that cures all illness. And it just, it just seems like, um, because it is, it does get presented as DNA gets presented as software that we can just fix. And it just seems so difficult.

I think it's I think what's helpful there is I understand the disappointment, and I think it kind of comes down to what's different between biology and other forms of technology. And the difference in my mind is biology is underpinning medicine, and with medicine comes just very real stuff, disease of yourself, disease of loved ones, and with that comes a desire for, you have this iPhone in your pocket that's this incredible piece of technology, and then you might have a family member with a disease that we've been studying for 50 years, and there's nothing you can do, and it's hard to believe what's going on there. And I think that's one of the elements where if that makes people feel impatient, I think that's actually that makes sense. That's normal. We feel that. That's a totally reasonable feeling to have in that position. So I think part of it is we're holding biology to a different standard than other technologies in that sense. And I think maybe we should.

Yeah.

But I also think that if someone's saying, oh, all of this stuff is just like vaporware, wet vaporware, that's also wrong. That's also just an overly skeptical view. And the nice thing there is we're seeing the proof in the progress in the clinical trials. Those are an extremely high test. You can't fake it. You actually do the trial and you see, does it work or does it not? And the picture is going to be really clear. But we're seeing the work— I used to work on making beta cells in vitro to cure type 1. It took years to get that into people. But then once it was in a person and there's just people going around who've received these cells that have all been made in the lab, They just don't have to inject insulin every day anymore. It's not perfect yet. They have to suppress their immune system for the privilege of not injecting insulin. But for a bunch of people, that trade-off makes perfect sense. That's incredible. And that progress is absolutely real.

Okay. You grew up in Romania?

I grew up in Canada.

In Canada?

I was born in Romania. I grew up in Montreal. Great city.

So your parents are Romanian?

Yes.

Are they academics?

Mix. Dad's a physicist. Mom's a Programmer. Okay. Okay. Yeah.

I mean, your academic credentials are quite extraordinary. The, um, all these honors and everything at Harvard. I mean, you did undergrad and grad at Harvard. And, and so what you were, did you assume you did quite well in your teenage years at school?

I did quite, did well in school. I think the thing that was really special for me before, you know, college was, and I got very lucky about this, is I I was— he was a, you know, very serious nerd participating in science fair competitions. And that was like really fun. That was just such a permission to play. And I think—

Science fair kids always do well, man.

That— yeah, you know, you just get that bug at that age and, uh, it was hard to let go.

No, but like through my career reporting, it's one of the— it's a, it's a high signal, you know, somebody who does well in these international, um, well, even national science fairs tend— they tend to, tend to do quite well in life.

I mean, for me, the, the— it's almost stronger than that is like for me, going back to the first science fair I participated in, that's where I like bumped into biology.

Okay.

And that's where like it started.

That's what I was gonna ask you. Yeah. And so like, if you were starting over again, given everything you know about AI, many of the things we talked about, you know, if you had to pick physics, chemistry, biology, energy. So I mean, would you still pick biology now? Is there any element to this where you see what the AIs are doing and you're like, oh, They're going to do a lot of the work from here on out, and maybe this will be less exciting for me.

I think my feeling about what's going to happen there is as the power of AIs grows and they can help us do more and more things, it's going to become like purpose is going to become increasingly more important. Like if you can just apply this thing to solve anything, applying it on the things that matter most. And that's a bit of a personal decision of what do you feel matters? Is it something beautiful? Is it creating beauty, which is really what art is about? Is it understanding how the world works, which is what science is about? Doing this with more and more AIs just sounds better. And I think for everyone in the field, we're going to get to continue doing it with increasingly better collaborators, increasingly focused on the things that actually only we can do or we have the strongest leverage. That sounds fun in the years to come.

So if we sit around, when you're going to sleep at night, and I mean, this is the most optimistic case scenario, are you— because you have this window, Into into all this technology and where it's heading. I mean, do you do you sit around and you're like, fuck, in ten years we just like cured every disease? I can see that very much on the horizon. Or are you still more like, oh, this is this is this is a forty, fifty, sixty, seventy, hundred year journey we're still on. And I mean, every disease is a lot. You don't have to like take that verbatim. But I mean, you know, like do you sit back and and you're like, man, the world. In 10 years, people have no idea what's coming for them. It's going to be a huge change in health.

A bunch of my SF friends keep selling me on, like, this is how you feel, and it's just kind of not how I view it. I think— I mean, I do hope that in 10 years, we are ripping through diseases in terms of places where we can have impact, and we can just put together pieces of technology to achieve the impact that we want to have.

Well, you're saying hope, but you think that— I think we would. Yeah. Like, is there— what's, what's the low-hanging fruit list of the diseases that fall into that 10 years?

I think if we start listing monogenic diseases, yeah, you know, things like Huntington's disease, ALS, you know, SMA is already on the list. Like, we've already— what's spinal muscular atrophy? Okay. So this was one of the first gene therapy with— one of the first gene therapies, Olgensma, was approved for for SMA. And some of the kids that received that therapy are now entering their teenage years soon. And there's maybe a— actually, we're going to get to learn, like, is the function of that gene even necessary once you're done growing? And we will learn that information soon. So that list is going to grow. But also diseases like Parkinson's, which are not anywhere near a rare disease.

Right.

It's one of the most common diseases. But it has single gene handles that we can pull on. And there's quite a few approaches that are being— the trials are starting soon is how I would describe it. So in 10 years, we'll know the answers, which of those work.

Okay.

And at that point, we'll be in a deployment phase for this low-hanging fruit. We're in a scale and deploy. also looking to bring in more pieces, bring more diseases, more areas of opportunity to the application of these technologies. I think the other part here is as we grow, there's aging broadly is a place where a combination of different genetic interventions should help, will help. Question is just let's find the right ones. And we're absolutely going to be in that direction.

How into the aging stuff are you? Is there anything in particular that excites you?

I think I'm a bit more myself as a pessimist there, just honestly, partially because of the time I trained as a physician. I never practiced, but that was a very grounding experience on what can and can't happen. So it's maybe bringing a certain pessimism to me in terms of— what are the— if you look at the max age or how healthy you are until you die, understanding these perspectives for me is very grounded in the reality of just meeting many patients through medical training. I am quite hopeful that increasing healthspan by quite a bit is absolutely possible as we understand the drivers of this healthspan. And I think it's amazing the the focus we're bringing and also the will we have to test bold hypotheses. I think to me that's amazing, and I'm really glad we're doing this investment. In terms of what it's going to take to increase the maximal age of a human, that's going to be tough. Part of me feels like I just don't see as much the proof of concept unlock that I'm more in that camp of I don't see the reason to know that we have the pieces that creates this unlock. But at the same time, if there's many, many different diseases we can stack and fix, that's going to have a direct impact on healthspan as well.

I've got Bryan Johnson's olive oil up there. We can have a shot.

Amazing.

Well, you were making fun of your San Francisco friends. I go back and forth between Cambridge and San Francisco, and I Obviously, see what you're talking about. I mean, there are two distinct flavors to these communities. You know, one, when I go to Boston and Cambridge, just absurd. Everyone I meet is an MD PhD or a PhD PhD, and and but you know, and just at the cutting edge of of incredible breakthroughs. I mean, the Church Lab, Boyden Lab, all these you know, I feel like they're just knocking out all these people that end up making fantastic startups and and everything. But you know, there is this more conservative tradition. slower-moving sort of vibe there. Out here, it's a bit reckless at times. I just keep running into 20-year-olds who have a bunch of mice in their lab, and they'll just come out and be like, I've cured insomnia. And it's like, have you? But so, yeah, I mean, do you take pleasure in both communities, or do you find San Francisco just over the top?

I was robbing my friends there, but I think the trick is find the best of both worlds.

Yeah.

And that's really fun as an approach. I think even in the work that we do at a practical level, combining bio and AI, you want to try to play to the strengths of both rather than be constrained by the relative weaknesses of both. That's essential. Yeah, it's fun to see the different vibes in the different communities.

But you don't yearn to move the company to the Bay Area?

No, we're not. No near-term plans. We're very happy where we are. We have an amazing space. And look, to us also, the focus has been we're just so blessed by this incredible team of really creative, really committed scientists. the same way that Silicon Valley is this for tech in the world. This is what Boston is for biotech. And so we want to continue to benefit from that and continue tapping that. But we do have some remote folks across the country, including NSF, and playing to win everywhere.

Yeah. Aaron Marogi, who writes for Core Memory, has even hosted a Core Memory podcast, is studying to be a doctor like you.

Was he a He was one of our early employees.

Early employees. What do we— should we shame or praise Erne?

All praise. All praise for Erne.

I think we're going to start like an Ask the Doctor Q&A series where people can like send in questions to Erne to be answered. And yeah, do you have a favorite? And I'm trying to humiliate him here.

Trying to think, what can I do? Can I share right here? No, but Aaron, I mean, I think part of what actually probably makes him effective for Core Memory is I think he sees what things are, sees them as they are, and also is willing to speak up.

Yeah.

Which when you're making a company is sometimes it's annoying, right? You're like, oh, this person speaks up in this way.

It's annoying.

Yeah. But then you actually sit down and you're like, oh yeah, there's something here. And then I think actually getting good at that is key. So Aaroné having taught us this loop and having set us this loop of, you should have a culture where if something's not how it should be, you should speak up so we can change it, make it better. That's been a great impact beyond the scientific contributions More directly.

He was a coward. He had to go get his medical degree. He was—

I, I had a lot of fun, uh, myself being like, hey, just finished my medical degree. Like, here's some thoughts I have. Yeah, you know, someone who, who, you know, realized I, I'd have a bigger impact, um, you know, not practicing myself.

Okay. He did send me a couple questions to ask you, one of which is, is low-hanging fruit, but let's do it anyway. Peptides, all the rage. Should people be playing with their own peptide stack?

Let's put it this way. I'm not playing with my own peptide stack. I think what people should do is up to— there is a framing where as long as it's understood that it's safe, let people do how they please. But I think the question is for depending on the peptide, like, is it safe or not? It is maybe a question.

Yeah, the question is more like we seem to be tilting to quite an ever-growing number of people experimenting with things they're buying off the internet, which I guess like the weightlifting community has been doing for decades. And this is just kind of becoming mainstream in some ways.

Yeah, I think there are some, you know, what's giving me pause for myself anyways is the, There are some places where there's interesting surprises when you study something for a long time that you don't understand.

Yeah.

And I think the trick is when you're making a drug, the trade-off is often very immediate and worth it. So you are willing to tolerate some side effect because there's a very clear benefit so that you're willing to make that trade. I think the risk here is that there might be times when we realize over time, like, oh, this had a side effect. And we might also realise, actually, we don't understand the benefit or the benefit's not nearly as clear. And then people might end up kind of disappointed because they're going to have certainty on the side effect, but not on did it do anything. And they might— would have done things differently if they knew that information. At the same time, I think if we're looking at the peptides, like the GLP-1s, The fact that they're blurring this line of medicine into consumer product.

Yeah.

That's great. I think that's amazing from the perspective of not tying into this thing of like medicine is this, you know, thing where a small cast of appointed few can tell you what you should and shouldn't do in some conditions. But, you know, you as a person can look at information and understand what something is and decide if it's right for you. And that's that. that transition. But it's helpful to point out GLP-1s have been studied for a very long time. And again, the impact is so clear, so it makes the trade-offs easy.

Okay. I ask this next question. I always go to China at some point in the podcast. China. And I hope people understand that I love China. I love going there. I am not like some chingoistic, crazy American You know, anyway, these questions do not come from that place. I just generally think people are not paying enough attention to the rate at which China is making progress in science and all sorts of fields. I just, I feel like most Americans are kind of sitting around expecting we're going to dominate all these things for a long time, and I'm not sure that that's that's the case. What's, what's like the— so China can run trials faster, cheaper than the United States. It is catching up, if not, um, on par with us in many areas of cutting-edge science at this point. Um, you know, I, I see a world where these technologies are getting powerful, and, and if we are creating— I don't want to call it like superhumans, but if humans are starting to become healthier, um, gaining functions we didn't expect. I, I kind of fully expect most of these things to maybe turn up in China first. What's, what's like, what's the pro case for American science? What is, what's like the best thing we have going for us? Or, or, you know, or answer that question in any U.S. or China, um, sort of frame you prefer.

Yeah, I think to me the— I agree, the rise of like Chinese biomedical science has been amazing, and it's really interesting to see.

I think I have friends, you know, who have very good jobs, like professorships at the University of Michigan, um, who are not unlike tech in some ways, are not like being actively recruited home. They're just choosing to just like go move back. back to China.

Yeah, I can see why. And I feel previously, if you looked at 20 years ago, it was obvious that if you wanted to be cutting-edge research, you should be in America for biotech. Now there's parity, or depending on the field, there's just key differences. And I think you're right to point out that there's a real risk that we've been at the forefront for so long that we sort of assume that this is just our place to be rather than a place where we need to keep innovating. The hope for American science here is if you look at other historic times, like in rocket science, of times when there was great power competition where the point was the technological race, we know we can do it. I think we're finding the will to do it, which is exciting. Ultimately, I think if competition makes us accelerate, that's awesome. Again, these are such hard problems with just a clear value. If we hit them, we should do it. I think there is a risk, which is that different countries all do have somewhat different diseases. What's most common can be related to genetics, can be related to the food you eat, the society you have. And so there's a real risk that if we Right now as Americans, we more or less get to prioritize the list of diseases based on where we want to apply this focus, or we do this jointly with Europe. We have a responsibility to do this for the whole world, but there's a risk that we want to continue being able to allocate and know that we're applying the best technology in the right places. I do think the changes that are coming, for example, from a regulatory perspective are key. It's like a wake-up call to be, this is just done faster and simpler here, so why are we doing it in this cumbersome way? But at the same time, if you look at, for example, the story of Baby KJ, we still have the goods, we can run the race. And the wake-up call from that perspective, I think, is good news. Let's find the will to speed up.

Yeah, I'm long American ingenuity, creativity, work ethic. I'm short on— I see some will. I don't see massive— I don't think people are paying enough attention to the scale of what we're competing against.

I think in biotech, it's probably a place where Like, people see it, feel it, get it.

Like, most people do not really— I mean, and I, again, I— God, I hope this doesn't come off as like really crass and like, um, nationalistic or something. Like, like, most of the— at least 50%, if not more, of the people who are like postdocs working in a lab are, are Chinese immigrants, right, working at U.S. labs. Is that fair?

I don't know if it's up to 50%. It's a lot. It's a big number. But I think more generally, right, like like myself as a Canadian.

Yeah.

I, you know, I loved— I love Montreal, amazing city. Would have loved to stay in Montreal if I could work at the absolute cutting edge of biology, which is what I wanted to do. But it was clear that coming to America is the right thing to do. I think that's just been the case for a long time, and I think the fact that that is shifting is a bit new and, uh, we'll want to to respond to that and continue to, to benefit from this massive influx of, of talent, of the most committed and, and, and talented people.

That was sort of my point, is like, at this particular moment in time, maybe we don't want to scare off like a bunch of brilliant Chinese researchers and make it really hard for them to, uh, to come here and, and work in these labs, you know?

Yeah, I mean, at the same time, there's, uh, there's, you know, America's got so many, so many good things going for it. Yeah. Down to its principles, you know, got to play to those advantages. That's going to remain true.

All right. You're making me feel a little better, maybe. I don't know. The cool thing about biology is usually whoever's solving these puzzles, this will be dispersed throughout the world. So that part seems that good. I asked you all of my biggest questions, and I've taken up quite a bit of your time. Is there anything, like, in particular you want to talk about that I didn't hit?

To me, it's this— we've talked about it, but maybe it's helpful to, like, put the pieces together. Yeah. Of this, like, moment. Yeah, yeah, yeah. That's coming in genetic medicine. To us, we've— when a moment is happening, we found it's helpful to label a thing because then you can notice it. Yeah. To us, we've We found the best label is this concept of genetic agency. And to me, I think that's tying together all of this story. So let me just define what is genetic agency. To us, it's the ability of an individual to act at a genetic level to live a healthier life. Right at the beginning, we talked about genetic medicines as being the operant that does the acting. Thinking about genetic agency as this now Someone's ability to do this. And I think part of it is this is really a movement that's building in the field between the ask for people who want to have this, who look at the technology and say, seems like we can put the pieces together and can you do it faster? And creating the pressure for that and the various systems also saying, wait, this is true. We actually need to rise to the occasion.

You're talking about somebody who's ill?

Yes. Or a family member.

Okay.

Yes.

And you're saying, you're saying, okay, well, okay, going back to where we started, I mean, obviously the US government is addressing this some way for these very rare cases, but you, I mean, what I feel like you're maybe saying is something even broader, which is, is like, um, the tools are getting quite good and, and, and, and this applies to a broader group of people who, who, if they're willing to kind of, you know, be on the cutting edge, could be helped. Is that—

Yes. And the proof of the principle is there too. Like, people are looking at the baby KJ story. Yeah. And saying, can we do it again? Yeah. And, and of course, the, the people who, who did that are, but more people are looking at it and saying, you know, what, what needs to be true for it to happen to me? How can I drive that? And it's creating this community that previously was like a drug maker. That's something like making one drug. Now it's this distributed community of technology builders. AI has a role to play here.

Wow.

It's probably one of the purest applications of obviously unalloyed good that AI can play is furthering this future. So the folks who are building AI want to enable that, want to apply it in that way. So I think that moment is happening and it's really exciting to to see, for me personally, to being a part of, but also to invite others who want to be a part of this. There's just many angles by which folks can contribute to this mission and movement. So we're excited for that.

Okay, I'm gonna get one final thought from you on that. If anyone, the story of Baby KJ is fascinating. This is a child who had a rare genetic disease, and we, in very short order, were able to create a therapy which needed to be very fast because his development as a kid was gonna be stunted lifelong. And then it was also a debilitating illness. And, and it was miraculous. A bunch of scientists all got together. It kind of proved out how far we've come with these gene therapy technologies. If you want to learn more, Aaron A. did write a story for us on Core Memory. He also did a podcast, and we also talked about it with Jennifer Doudna, if you want to check it out. But so like what you're saying, I mean, I'm just trying to wrap my head around it fully is, is because those are such isolated, specific cases. And you seem to be saying just that, what, that that's just exciting that we've been able to do that and we've shown that this stuff works and there's speed and now the government's adjusting and that maybe even though we're starting with rare disease, just that people should be— they can contribute to this, but also be cognizant of technology that's available.

Yeah, I'm saying it's a herald of a time to come. So notice, pay attention to the movement, get curious. That's part of the beauty of the movement is it's also just really interesting for many different reasons. So approaching it with curiosity is probably the— everybody can do that. And then also in some cases, approaching it with with more than curiosity, with agency in terms of if I want to get involved in this mission, there's many different ways to do it. And again, just the problem is so hard that anybody, someone raises their hand and says, yeah, I want to participate.

Yeah.

How can I? There's probably a way to find. Now it's just a matter of finding that point of leverage and coming into the genetic agency family. Okay. Okay.

Last question, and then I'll let you go. And it's along those lines. If you were 18 years old today, would you become, given what you just said, would you become an MD? Would you become a computational biologist? Would you become a biologist? Would you become a computer scientist? I guess we need all of them, but what's, what would be, if you know, knowing what you know now?

Biologist. Yeah. Remove the computa— like the computational part is is probably going to be more direct. You'll be able to collaborate with amazing computational AI biologists as well, but I would focus on the biology just because it's so interesting and so powerful once you understand the interesting bits. I'm confident that for the rest of our lives, it's going to be a thing we talk about, a thing we marvel at, and a thing we just get a ton of value from.

Okay. That was specifically for my son, uh, my oldest son who I want to become a biologist. It may or may, he may or may not listen to this podcast.

He should come to Dino. Happy, happy to give him a tour.

I'm trying, man. I'm trying. He, he told me he wanted to be a computational biologist when he was like 12, and I haven't been able to let this go, even though I think his interest may have shifted. Um, but that's, that's what I would do if I was his age. So, um, Adrian, thank you so much for coming on. Um, really appreciate it.

Everyone can go to Dino. It's dynotx.com.

Okay. And, and, and you can— there's all kinds of cool information and get a better sense of some of the things we were talking about. And thank you for trekking all the way out here.

Thank you.

The Core Memory Podcast is hosted by me, Ashlee Vance, and/or Kylee Robison, or both of us together. It is produced by me and David Nicholson. Our theme song is by James Mercer and John Sortland, and the show is edited always by the John Sortland. Thank you so much to Brex and Sen Katsen for all your support, and thank you most of all to everybody for listening or watching. We love you. Please leave us a like, a review, a subscribe—all those tremendous things. Thank you, and we'll see you again.