The Company Trying to Reprogram Aging

To make a medicine for you, I don't want to turn you into a bag of stem cells. That would actually be really bad. What I want to do is take your old skin cells, your old liver cells, your old immune cells, and turn them into young versions of themselves. So can I decouple cell age and type in the other direction? Can I reprogram cell age but preserve cell type? And so we're searching for those groups of factors similar to the factors Yamanaka found that might be able to reprogram age independently.

Welcome one and all. This is Ashlee Vance. I am in our Still in Progress podcast studio, and we have a slightly different podcast for you today. We're going to be interviewing Jacob Kimmel, the co-founder and CEO of New Limit. I got to tell you a little bit more about the show because it's different. But before that, I would urge you all, please Go to Core Memory, read the story we put up on Friday about Brad Smith and Tiffany Smith and the Smith family. Brad has ALS. He's the third person to receive the Neuralink implant. Me and our team spent the last few months hanging out with Brad. I was reporting the story. We've filmed, uh, a mini documentary which you can see on Core Memory or our YouTube channel. About them. You can read the story, see the film. The film's going viral on YouTube right now, which is great. Come join everyone else and subscribe. We could use the help, and it would be appreciated. More importantly, it's just— it's a really moving story. I think it's— Tiffany, the wife of this family, is incredible. The kids are incredible. Brad's incredible. I think you'll all enjoy it. And now for this episode, it's a little different than we usually do. We've, we've been filming with NewLimit as well. It's a biotech company. They are in this field of iPSC technology-induced pluripotent stem cells. I talk to you guys about them all the time. It's the most exciting area of biotech for me. We have an episode coming on NewLimit in a few weeks, but just today the company has announced that it raised $130 million from Nat Friedman, Daniel Gross, Kleiner Perkins, Human Capital, and Khosla Ventures. And so obviously they're going to be pushing on with their efforts. Our interview with— oh, also, they raised more money from Patrick and John Collison, Joe Lonsdale, Fred Ehrsam, other rich people, Josh Kushner. That I— it's kind of a who's who of tech investors. So company's doing well. Jacob is one of the most eloquent and I think, um, just gives really great explanations of how this technology works, what NewLimit's trying to do. So I think you'll really enjoy this episode. And then in a few weeks we'll have a Core Memory show for you guys on the site and, and on YouTube. Um, with that, We will go to Jacob in a minute. Please, I know it's cheesy, but like and subscribe to all our stuff. It really does help. And this episode is sponsored again by E1 Ventures. They're a venture capital firm based in the Bay Area. They fund all kinds of deep tech stuff. Just wonderful, kind people that I've gotten to know a bit from doing the show. And, and thank you to E1 Ventures and Here is Jacob.

My name's Jacob Kimmel. I'm one of the co-founders and the president here at NewLamette. So I run the firm day to day and lead our research programs.

I've been here once before. It was probably like, I don't know, I want to say like 18 months ago. For people who haven't heard of you guys or learned about this technology, can you give us the overview of what you're trying to do?

Of course, this is what I love to do more than anything. So we work on a new technology called epigenetic reprogramming, and we're trying to make new types of medicines based on this tool that will allow us to treat age-related diseases and eventually the aging that occurs in every single person. So epigenetic reprogramming is a very weighty set of syllables, but it describes a pretty simple concept. So all of the cells in your body have the exact same DNA code, and yet they do very different things, which might seem simple on the surface, but is actually kind of profound when you realize it. Your tongue and your eyeball have the same code, and yet they have very different jobs. Turns out that's controlled by a set of chemical modifications on top of the DNA, which basically tell each cell which genes to use and when. As you age, unfortunately, those chemical marks get messed up, for lack of a better term. And so the tools we're using actually try and rewrite that code back to the way it was when you were young. So your old cells use the right genes at the right time and ideally behave a bit more like their younger selves, allowing you to treat diseases that would otherwise be intractable and to eventually address the challenges that emerge as each of us get older.

So we have, we've got neurons, which are brain cells. We've got kidney cells, all the cells, obviously, when we're, call it 18 to 25, in this optimal state. And these cells degrade over time and just can't do whatever they used to do as efficiently.

Yeah. And we think one of the main reasons for this is that it's— the genes are actually still all there, which is kind of profound to think about. Again, it's like all of the instructions the cell needed to perform its job when you were young, when it was working well, they're still preserved, they're still in that cell, modulo a few details. And yet they're not using the instructions the right way anymore. And so we think that really by re-imprinting these epigenetic marks, allowing them to use the right genes the way they did when you were 18 to 25, right in your prime, then even in your latter decades, you could actually restore some of that youthful function you've lost.

And this is based off what— what is it, roughly 12 to 15-year-old science is the foundation of this. There was a Japanese researcher named Yamanaka. He discovered these for proteins that turn genes on and off and seem to be able to reboot cells, right?

Yeah, exactly. So there's actually even science going back even earlier. We knew that we could reprogram an old cell back to a young state and give rise to an animal with a normal lifespan ahead of it. So this actually goes back to some early technology around cloning. So a lot of people have heard of Dolly the sheep. Cloning actually was much broader than just Dolly. Even just Dolly had a sister. No one knows the name. I thought her name is Polly. People forgot about— yeah, poor Polly. But so even with that technology, which you can you can think of as like a really complicated way to do this science of reprogramming, to take the epigenome and change it back to a youthful state. We knew you could take a cell from an old animal, reprogram it back to a youthful state, and then give rise to a whole other animal with a normal life ahead of it. And there's even a group in Japan that used that cloning technology to clone the same animal effectively 13 times in a row. And so you could take an old cell, reprogram it back, give rise to a young animal, wait for that animal to get old, reprogram it back, and do that 13 times. Now mice live 2 years, so that experiment takes about 20. And if aging were somehow immutable, if there was some sort of damage occurring that was totally irreversible, you'd expect that last mouse to be pretty messed up and not live a normal life. But instead, what we were able to demonstrate as a community, I should say, not we here at NewLimit, was that just through reprogramming, that animal could live a totally normal lifespan.

It's rebooting.

It's rebooting. It's restarting at sort of the same epigenetic state, the same use of its genetic instructions that you did when you were young. So all of that technology is actually pretty old, but it's not something you can imagine turning into medicines. You know, it's probably too complicated to do something like cloning as a therapy for people. So just as you referenced, there's some science about 15 years ago from a remarkable scientist named Shinya Yamanaka where he discovered that whole complex process could be achieved just by turning on a set of 4 genes at the same time. And so these genes are now known eponymously as the Yamanaka factors. And each one of them is what's called— And he won the Nobel Prize in 2012. You know, one of the quickest Nobel Prizes, I might add, because it's so profound, such an advance in the field. Field. I think the prior belief was that this would just be way more complicated than that, that, you know, there are tens of thousands of genes in the genome, millions of molecular interactions. There's no way we could reset a cell from its developed adult age state back to this youthful embryonic one in a really simple activation of just 4 genes. And so his result, I think, really broke down a lot of the cognitive biases and prior beliefs the field had. And so those genes are really special. They're called transcription factors. And I like to think of them as sort of like orchestra conductors in the genome. They don't really do much themselves, kind of like an orchestra conductor, maybe not the person you want playing first chair violin, but they can point at other genes effectively and tell them when to play, when to turn on, when to turn off. And so just by activating a few of these genes, you can actually control much broader programs in the genome and rewrite those instructions that determine which genes are being used by which cells at which times.

And it's wild that even in the mouse experiment you were talking about, It's not like it's like we found some magic cell that's being at the heart of doing these different generations. You can take like a skin cell of—

Skin cell, blood cell, almost any cell. And create an animal. You can create a whole animal with a normal life.

Like reboot, take that skin cell back to its kind of stem cell state and then reboot a whole animal. It's wild to me.

Yeah, it's like truly remarkable that all of this biology works as well as it does. I think what makes it somewhat more intuitive is to remember that this is also just how reproduction works. You know, regardless of the age of two parents when they have a child, the child's age starts at zero. And so if you have a kid at 20 and a kid at 40, both of those kids start as infants. One of them is not plus 20 years. Maybe even a little silly to talk about because it's so intuitive to all of us.

No, I feel like we take it for granted. Until I started getting into this field, I'd never actually even thought about it. You're just like, oh, two old people make this total young kid.

Yeah, of course that works, right? Like we all, we all understand that. And there was a belief in science for a while that maybe there was something about reproduction that was really special. Maybe like the cells involved in reproduction just like never aged and they were protected in some way. There's a fancy name for this called the disposable soma theory, that only reproduction matters and the rest of your body can get old. And I think what Yamanaka really showed is that that's not true, that you can actually execute the same reprogramming event, the same reinvigoration of that youthful cell state using just a few genes at a given time. And so that's really what's been inspiring here to us, is to ask, can we now take that biology and try and make medicines with it? Maybe not directly those 4 genes, but can we figure out ways to use small numbers of factors like that, small enough numbers we can even imagine making medicines for them, and achieve really dramatic changes in cell function the same way Yamanaka was able to achieve with those 4 factors?

And just to underscore that one point you make, I mean, I think people hear about longevity. I mean, you know, all this stuff is hard. It takes many years, but like there is this fundamental thing that we know is that somewhere in nature this is happening because life is rebooting. I mean, there must be— there is— I don't know if we can guarantee we can replicate this recipe, but we know something is possible because it happens all the time.

Yeah, I think it's a really important point is like I think a lot of people have the intuition that aging has to happen or is in some way necessary for biology. It's like physics where like all organisms must get older. And as you highlighted, humans haven't like aged into extinction yet. So we know somewhere in this process there is a rejuvenation event occurring. And so whether we can make a medicine from it is a totally different question. And we hope to one day answer that in the affirmative, but we at least know it's possible. You know, in physics you'd call this, or mathematics, you'd call it an existence proof. We know what we're trying to do is not entirely crazy, maybe just crazy enough.

And there's, there's a few things I want to dig into. When Yamanaka factors arrived, it was, it was first used to take these older cells, whether it's a skin cell, a liver cell, a heart cell, back to the stem cell state, which is this blank slate where you're building from there. And so for the first time, we had these, these— instead of harvesting stem cells in these controversial ways, you had this lab-grown stem cell that then you could do all kinds of interesting things with. But there's a whole another track to this, which is you take that liver cell or heart cell and instead of rebooting all the way to the blank, the blank slate, the blank state, you want to keep it a liver or heart cell, keep its identity, but you want it young. You want the young platonic version of that cell. And that's really where you guys are focused.

Yeah, absolutely. That's really our goal is to be able to say Yamanaka showed you can reprogram a cell's age while changing its cell type at the same time. So I take an old cell from an old mouse on death's door turn it into a young stem cell. So it's both young, that's one big difference, the age changed, and it's now a stem cell, so the type changed at the same time. So there's also some pretty amazing science where other folks have shown you can take an old skin cell and turn it directly into a neuron, which just kind of blows my mind every time. You can take a skin cell and turn it into a neuron with a big axon firing action potentials, like this crazy dramatic change in what a cell does. And remarkably, that neuron actually looks old still. And so you can change the cell's type without changing its age. So No Limits is basically trying to ask that question in the other way, which is, you know, so somebody's shown you can break apart cell age and type reprogramming. Can we reprogram age without reprogramming type? To make a medicine for you, I don't want to turn you into a bag of stem cells. That would actually be really bad. What I want to do is take your old skin cells, your old liver cells, your old immune cells, and turn them into young versions of themselves. So can I decouple cell age and type in the other direction? Can I reprogram cell age but preserve cell type? And so we're searching for those groups of factors similar to the factors Yamanaka found that might be able to reprogram age independently.

And it's weird to me, as an outsider covering this. So Yamanaka finds these 4 transcription factors, this very precise combination of 4 things, basically proteins, and it has this effect. And then most, for whatever it is, 12, 15 years, it's people experimenting with those same 4 things. You guys are looking to see if there's other transcription factors, other recipes we can use. And I was just curious. I mean, why did people get stuck on these four? I would've thought like you immediately start exploring to find all these other things that may have these magical properties or seemingly magical properties.

Yeah, yeah, you're entirely right. It's kind of shocking. I think of epireprogramming, the ability to turn on these combinations of transcription factors and see changes in cell state, be that cell age or cell type. It's really the most powerful change in cell function we've ever been able to evoke in biology. I used to work on small molecule and antibody types of medicines and Honestly, it makes those seem a little bit trivial when you think about the complexity of what you're achieving. And so other groups have looked for new combinations of factors. That's how we got amazing results like, oh, I can turn a skin cell directly into a neuron. Marius Wernig, not far down the road at Stanford, discovered that. Others have found converting between other cell types. But for reprogramming cell age, I think for a long time it was difficult to even think about how you would find another set of factors. The tools didn't really exist. When Yamanaka found his first 4 factors, one of the reasons he was able to discover those was because he was changing the cell type at the same time. And so it was really easy to know when you had achieved the goal. You could just look at the cells under a microscope and you could see visually these are a very different cell type than they started out as. Go ahead.

Well, so, but what was he even doing? Because how many— there's a very long list of transcription factors, right? Yeah. That are possible. How many are there?

Yeah, there's about 2,000. This is debate. If you get a bunch of transcription factor biologists into a room, they'll get really heated about what are the 1,200 or 2,000, but somewhere in between.

Roughly. And then, and then he hits on these 4 and he hit on these 4 relatively quickly. Like, were there— were they somewhat obvious that these would be the 4 you might want to try for this kind of thing?

I don't think it was obvious at all. I mean, at the time, Shinya Yamanaka didn't even have a laboratory. He was running a core facility. He was doing this work in a way where I don't think he had the levels of institutional support one would expect. I mean, this was really like the motivation of him and his group pushing out on a limb, trying something very very few others were. But the way he was able to actually get that to work was he had a pretty simple hypothesis. He looked at embryonic stem cells, so the stem cells we know that start in embryos and eventually form whole organisms, so they can become any cell type in the body. And he measured which transcription factors are really highly active in those embryonic cells. I see. Then he measured some non-embryonic cells, some adult somatic cells, and said, which of those transcription factors are gone? Maybe if I turn those light switches back on, these cells will turn back into the embryonic state. Maybe these are the control mechanisms. And so then what he did was he just tried random combinations of them. So he actually selected a group of, I believe it's just 24 factors, and then tried as many combos as he could and lucked into this combination of 4. And it turns out it's not unique. Actually, other combos work too. So you can make these stem cell-like reprogramming events using a number of different sets. That's sort of how he stumbled upon it.

Okay.

And I think one key insight there is the reason he was able to test so many things was that his assay, his way to measure if it worked, was really, really simple. Look under a microscope, are these cells, did they go from these fibroblast-like sort of skin cell-like state, which you can see visually, to a stem cell-like state where they grow really fast and they're really small and they clump up together. And so you don't need fancy equipment to measure this. You can just look under a microscope and you know. And I think that's why finding these factors that reprogram age alone has been so tricky. Until recently, the difference between an old and a young liver cell, an old and a young skin cell, is not as apparent with really simple tools. You can't just look under a microscope and know. And so we needed technologies like the ones we employ here at NewLimit where we can measure every gene that's being used in an old and a young cell at the level of each individual cell. This is called single-cell genomics. In order to be able to even ask if the experiment worked, if I test a set of factors, did I even make the old ones look younger? We needed tools that were high enough resolution that had the complexity to actually enable us to answer it.

So if it's a liver cell, is it clearing toxins as efficiently in a younger liver cell? Is it clearing toxins more efficiently than an older one? Is it processing something like alcohol faster, like as it did in your college days kind of thing, right? I mean, you have to find these clues of what's getting turned on and what makes a younger cell have these qualities that are beneficial and useful.

Yeah, absolutely. So in addition to just measuring what we call internally like a looks-like assay, you know, we're not trying to ask right away whether it functions like a young liver cell, like you're highlighting, processes alcohol, caffeine, regenerates, doesn't die when you drink, that sort of thing. Those have to be sort of designed for each cell type individually. And building those assays is really challenging as well. Before we even get there, We're just asking at the level of the genes a cell is using, does it kind of look like a younger version of itself? Does it use the genes a young cell would? And for us, we hope that's a clue that maybe it'll act like a young cell too. And even those acts-like assays, you might expect naively, as I did, that a lot of the stuff is just known, that maybe you can just go and go to your favorite search engine or talk to your favorite set of weights, and they'll tell you, ah, these are the 4 things you go in the lab and you measure, and you know if the cell is old or young. It turns out a lot of that foundational science just hasn't been done at a very rigorous level. And so we also had to build a lot of those assays assays, these measurement systems to know, is this old liver cell acting like a young one? Does it process alcohol the way a young one would? Does it resist some of those damages from things like a high-fat diet the way a young cell would? And so standing up each of those assays is likewise really challenging. And I think in doing those, even if we were able to build all those assays without the looks-like assay, you'd never be able to do enough of them to find new factors. So to sort of bring it back to your question of like, why did it take so long? It's been 15 years. I think that's really the answer. It's like those sort of acts-like assays, you know, Does it process alcohol? Does it regenerate? We could build those 10 years ago, but we can't really do that many of them. You need some middle step where you go from this crazy number of factors to that smaller number. And for us, that was sort of this notion of, I can measure all the genes the cells are using, and that gives me a decent idea of which ones are even worth testing with an alcohol diet or something like that, which is much harder to scale up to very large numbers.

Okay, let's dig into that. I mean, let's assume this 2,000 transcription pool is roughly correct. And so you're hunting for this combination of these many things. I mean, you have to find maybe 2, 3, 4, or 5 of those that when are together are doing something. So that's already like mathematically complex. And then you've got all these downstream effects that you're checking for. And so this is mathematically a large computational field.

It turns out it's about 10 to the 16 possible combinations.

Very large.

Yeah, it's very, very large. That number doesn't mean anything to human beings, it's just like, oh, it's like some scientific notation thing. So to give you a sense, there's about 10^11, 10^12 stars in the Milky Way. So we're talking about like 10,000 galaxies worth of stars of experiments. Like you just can't do all those experiments. So it's like, it doesn't matter how smart you are, you're never gonna be able to test all those things.

And this is in many ways the heart of the company as well was then. So we have this wet lab type science, we have this biological stuff that's going on, and then you're bringing this computation to bear, not just, new tools like single-cell, being able to look at single-cell genomics. But then we've had an AI revolution. We've had just raw sort of computing horsepower being added to all this. But I mean, that was kind of the thesis of the company was let's build that model that goes and finds these things both computationally and biologically.

Absolutely. So just as you alluded to, I'm really proud of how many experiments we can do. I think we can do way more than anyone else really in this field. The number of experiments we can do is still something like 11 or 12 orders of magnitude short of what we would need to do to just search everything exhaustively. And so we've also got to be really clever about picking which questions do we even ask before we start trying to get an answer in the lab. And I think, you know, 10 years ago, this wouldn't have been a very tractable problem. But because of this AI revolution that's come to pass, now if we're able to gather a sample of that data, we can run a large number of experiments, but not nearly enough. We can train AI models that are able to predict what might happen for a new combo of transcription factors that we've never tested before in the lab. And that model can make a guess. Will this make the old cells look younger? And for us, those models don't even need to be perfect. We're not trying to get the model to answer the question necessarily. We're trying to get it to tell us if the question is even interesting to be asked. Yeah, limit the scope of what we're gonna do. And I really think this is going to apply not just to our work, but to broad areas of the biological sciences. This is a pretty common problem. There's just way more things to test than you're ever going to be able to do in the real world. And so these sorts of in silico models, I think, are going to more and more come to dominate how we discover drugs in the future. And this is really what I think is one of the most exciting and high-impact applications of those tools because these are such potent pieces of biology. If we're able to build models, which I'd like to brag that I think we have today, that are able to make those predictions in a reasonable way dramatically increases our chances of actually finding those magic combinations that have the effects we're looking for because we spend more time testing hypotheses that are reasonable and less time testing hypotheses that are very unlikely to work.

So, and I don't know if this is apples to apples, but if I'm thinking of a large language model, it's sucked in trillions of words from the internet and then you're applying transformers onto these things and getting this token predictor of what the next word would be. Like in this case, are you, you're feeding the model all the 2,000-something transcription factors, you're showing it what has happened data tied to what's happened with Yamanaka factors, data that's tied to your experiments? I mean, I don't even know what you actually feed the model.

Yeah, yeah, it's simpler than you'd think. So I'll start in the simplest version and then I'll sort of back out and get a little more complicated. But I think your analogy to large language models is actually a pretty apt one because those play a role in how we solve this problem too. So the main idea is if we can show a model some mathematical representation of transcription factors, so some set of numbers, and let's just go on the journey for a minute where we pretend those numbers represent the biology of the transcription factor. Can I go from sets of those transcription factor representations and then some representation of what an old cell looks like? You kind of got to know where you're starting to know what reprogramming would do, where it's going to take you. If I don't know the starting place—

So this is a math— You're mathematically coding—

You're mathematically representing both of these things. So I want to take some notion of what's the transcription factor, represent that as a set of numbers a machine learning model can see, not so dissimilar from how a large language model looks at text. Text, and then it does what's called tokenization. It turns that into a set of numbers the model can understand. We sort of need to do the same thing for transcription factors and old cells. And then maybe by way of analogy to a large language model, you can show the model these representations of the TFs, the representations of the old cell, and the same way a large language model completes a sentence with missing words, we can ask, what would the outcome of this experiment be? Complete the missing word here being the outcome of what would occur. And so the, the way we're training these models is that we've got those representations, we've got the many sets of transcription factors we've tested where we sort of the answer. We know what that missing word looks like. Did this make the old cell look young? Did it make it look older? Did it do nothing at all? And we train the model to complete that and say, ah, okay, these made the cell look younger, these made it look older. Now, your analogy again to LLMs is not too far off because where we actually get those numbers that represent transcription factors is really using a lot of that same technology. So we can use things like language models trained on proteins and language models trained on DNA And so we can show those models the protein sequence of each transcription factor and show it the DNA sequences the transcription factor binds to, and then extract from that what's called an embedding. So this is the same way that in a large language model, you might take two documents, you turn those into a set of numbers, and then in that embedding space, you can ask, are these documents about similar things or different things? Well, we can do that for transcription factors and then say, are these two transcription factors similar or different? So maybe if our model has only seen some subset of the combinations, it hasn't seen this unknown combination, but it's seen one that has pretty similar transcription factors in it, it can make an educated guess about what might happen. And so because we're starting from those foundational models of protein language and DNA language, our models aren't like learning all of biology from scratch. They start from a pretty smart place. The same way that people would take a large language model and maybe fine-tune it for a task they're working on, we're taking these protein and DNA language models and using that knowledge to really jumpstart our own work. Okay. Okay.

And then you guys gave an update recently. I'm going to mess up some of the numbers, but you'll help fix them. I mean, as far as I remember, you've been working on this technology for a couple of years. In the last 6 months to a year, you felt like the model's humming along in a better state. And you'd found 100— was it 100 transcription factors you thought were promising?

100 sets. 100 combos.

100 combos. And so you've been doing this work. You're searching this huge space and now you've found 100 combinations that are promising to go look at. Is that the right?

Yeah, absolutely. So we found right now a little over 100, the number's a little out of date, combinations of transcription factors where when we put them in old cells, and these are real old cells from real old people, so it's in some ways simpler than a lot of the science you might hear about. These aren't like engineered cancer cell lines with gosh knows how many chromosomes. They're not mouse cells. They're real cells from old humans.

It's just like old blood, right? It's old blood cells, old liver cells.

Old people and young people, and we can put the transcription factor combos in those old cells, and more than 100 of them make the old cells express genes that are much more similar to a younger state. And so the way we phrase this internally to keep it simple is they make an old cell look younger based on its gene expression. So we don't know for every one of those sets yet whether it makes the old cell act younger, whether it can actually function better in a way that would help you help treat a disease, help you just feel better when you're generally older. We don't know that for every set yet, but for a number of them we do. And so we actually have 4 now, 4 sets where we can put those sets into old liver cells in old animals and even old human liver cells and see them function better. So they can regenerate better or they can resist injury more effectively. So example, in liver cells, one of the things that changes with age is when you drink alcohol, old liver cells get way more damage than young cells do. Like it's so dramatic, you don't even need statistics. You can just like look at the chart and know there's a difference. Many of us probably believe that because we've run the experiment in our own lives. So, you know, some personal distributed replication, I should say, has been performed in this case. And we can put some of these sets of transcription factors in and dramatically reduce the amount of damage these cells experience when they actually consume alcohol.

So these are like actual mice getting—

These are actual mice that get drunk.

Yeah.

Yeah.

What do you give them?

Just— It's surprisingly hard to get a mouse drunk. As one of our SAB members has told me, they evolved to eat garbage. And so they actually, they drink an alcohol-based diet. So it's kind of like drinking like beer as their main calorie source for a couple of weeks. And what that leads to is some damage in their livers. And they're actually remarkably resistant to being like what you and I would recognize as inebriated. We can look at a few later.

And they're— yes, I'm excited. They get a shot or—

No, so they actually, they just— oh, so to get the transcription factors? Yeah, yeah. So we just give them a shot intravenously. So it's the equivalent of like an IV for a human you would get in the hospital. And that's how these medicines would eventually make their way to humans as well, is this would be something you got as an IV when you went for an infusion. So very similar to some existing medicines that patients already get. The medicines we we use are called lipid nanoparticles, CARB-mRNA. So basically, lipid nanoparticles, really fancy phrase for a fat bubble. If you've ever taken olive oil with some water in a tube and you like shake it up, you'll notice there's little bubbles of fat with little dots of water inside. We do the same thing for these RNAs, and RNAs sit between DNA and proteins. So they're sort of the intermediary between actually going from code in your genome to having proteins that do a job in your cell. And so we can deliver those RNAs, which encode the transcription factor, and we just wrap them up in a fat bubble, and then your liver cells suck come up. It's actually what the liver cells evolved to do. So when you eat a cheeseburger, you get fat in your blood, your liver pulls the fat out of your blood. So it's kind of like a Trojan horse, like we're tricking the liver cell into thinking this is just a normal fat bubble. And actually it's a normal fat bubble carrying a couple of mRNAs that might make that liver cell younger.

And how long does it take for the mouse to start rebooting a bit?

Yeah, like until we see differences?

Yeah, yeah. Hours?

Hours. Yeah, we're talking, I think the shortest timeline we've ever measured is like 12 hours, but you start to see benefits quite rapidly.

And the liver is like actually The cells in the liver are rebooting on their own, or it's growing, it's just growing a batch of newer, better liver cells?

It's definitely the former. So what we're seeing is that the cells that are already there, the cells you already have in your body, they have all the right code, right? The DNA is still there, they're just using the wrong genes. And so those cells start to have their epigenomes changed. So rather than using the wrong genes like an old cell does, they have the epigenetic marks placed on it, the little bookmarks on your genome that say, use these genes, don't use that one. And so we don't have to make all new cells. It's like you still have all the code, you've got all the cells, they're just kind of like misbehaving.

Okay.

And so we can make them behave much better. And so they're able to use the right genes at the right time. You don't have to grow a whole new liver.

That's wild.

I mean, I know we talked about it, but this is helping wrap my head. I mean, so the cell's just sitting there and over time it's just, it's programming, it just starts to break down and you're just telling it, reminding it to go flip some other switches back on.

You're just kind of reminding it the way you might have to remind yourself every once in a while with your to-do list. What's a high priority? What's not. And so you don't have to introduce anything crazy. You're not taking new cells and putting them in the organism. You're just taking the cells that are there that aren't doing their job anymore and trying to remind them, so to speak, in an anthropomorphic way, which genes to use.

And that's really what those transcription factors are doing, is going in and placing little reminders on your DNA and saying, "Use these genes, don't use those ones." I know there's a lot of theories about this, but I don't know, maybe what I'm after, what your theory is, Are those genes getting turned down over time because somewhere in evolution we found it was like calorically expensive or something to have those on? Or is this just this like slow degradation process and just the coding just gets off kilter as time goes on? Or what, like what's your guess as to why this happens in the first place?

Yeah, it's basically unknowable, I think.

Yeah.

Whenever biologists get together and start talking about evolution, it's like time to go to the bar because like we can't answer any of these questions. My own hypothesis is it's probably the latter. I think if you imagine like, how did we end up with the genomes we have today? You know, sort of taking the selfish gene theory. Basically all of the code in your DNA was evolved so that that DNA would be passed on more aggressively. And so anything that occurs in your life after the period of max fertility, it basically doesn't have any selective pressure. Like there's no feedback tuning your genome to be more effective after you've had kids, right? It's like after you've had all your kids and you know, primordial human beings are running around on the savanna, after you've done your child rearing, basically nothing that happens to you is getting feedback into the genome. There are hypotheses against this. There are things called like kin selection, like, ah, well maybe having the older people in the community age faster had some benefit. But I think you have to start twisting yourself into kind of funny arguments to believe it. And so I sort of like the simple hypothesis, which is just like your genome optimized for you to have a lot of kids really young, and everything that happens after that is just unoptimized. And so one framing I like to take on it is when you're a biologist and you're trying to improve a system, trying to improve a system that evolution already had a couple billion years to iterate on is really, really hard. You usually got pretty close to an optimal. Trying to improve a system that evolution didn't bother to optimize at all is a much easier thing to spend your time on.

Yeah.

Because you might not even have to do anything too sophisticated. Even simple changes that just didn't have any selective pressure could have improve it. I think this is where some of the results in the field are really striking. You know, one result a lot of people point to in the field of aging came from an old colleague of mine, a woman named Cynthia Kenyon, where she demonstrated you can break one gene in a worm and make it live 2 to 3 times longer. If evolution had optimized for lifespan, like, that shouldn't work. It sort of tells you right off the bat that if something so simple can dramatically expand lifespan, that lifespan really wasn't optimized to begin with.

And why don't we all just break this gene?

It turns out, for like various complicated reasons, you know, that gene does something different in worms and it in humans, it's, you know, in humans has some other negative side effects, like making you a little bit more rotund and a little bit less muscular and other things you wouldn't want to intervene on.

Makes you live twice as long?

It also doesn't extend lifespan as much in mammals, like mice or humans. So again, sort of does a different thing in worms than maybe it does in humans. But that's sort of just one example of the fact that if evolution had like optimized to make organisms live for a long time, it shouldn't be so easy. You shouldn't be able to find these like quick wins.

Yeah.

So it seems like there's low-hanging fruit. I know with the transcription factors, I mean, one issue has been as you're trying to go to reboot to that younger state, it's like, can you get the cell to that younger place without going too far and going back to the blank slate? And then there's worries that obviously this is happening inside your body. And so you don't want something going wrong. Like you might go younger and then this process somehow you get a runaway process that's making tumors or or something horrible. Like, is that still— have you got even on the liver? Like, do you guys feel like now you have actually figured some of those bits and pieces out or there's just tons more work to do?

Yeah, I don't want to claim we've got it entirely solved, but I do think everything you said is totally valid. You know, someone might, might listen to us chat and say, well, if the Aminaka factors can make an old cell young, can't you just kind of like dial it in, right? Can't you get like the right dose? And, you know, there are folks who believe that. We tend to think think that it's going to be very challenging to use some of those existing factors and find a way to reprogram your cells just enough without abolishing their cell type. And exactly as you said, basically sending it to cell too far back and then creating a tumor in the body. And so what we're really focused on from even the beginning of our discovery process is trying to find sets of factors that have much lower risk of changing a cell's type, which would lead to something dangerous like you're describing, like forming a a tumor. So when we perform our early screens, we test what these TFs do in cells, we measure all the genes they're using. We look at not only does it make this cell look younger, we also measure, does this cell still look like the right type of cell? If I start with a liver cell, ideally when my younger cell comes out of the other end, it's still a younger liver cell. It doesn't look like something else. And so we can see that even early on, and we filter stuff out. As an example, the Yamanaka factors make cells look not at all like the cells they started with. We filter stuff out really early based on that sort of risk. And then we can go downstream in animal models and test these things directly. So you can ask, you know, does this make cells proliferate uncontrollably like a tumor would? And, you know, hopefully check those TFs off. You can then also do things like turn the TFs on for much longer than you probably would in a medicinal context and just make sure nothing bad happens. To give you an example, you know, in medicine we often talk about therapeutic index, the difference between a dose that helps you and a dose that might be harmful. So even medicines that are over-the-counter have a therapeutic index. You take too many Tylenol and you destroy your liver, right? So, you know, we understand that you can find safe doses and then also you can also have too much of a good thing be a bad thing. And so for the Yamanaka factors, the difference between the safe dose and the beneficial dose is less than twofold. So it's like, take one Tylenol, your headache goes away, take 1.4 Tylenol and you die. So it's probably not gonna be the right direction to go for every cell type. What we're hoping for is that the therapeutic indexes we can find are really, really big so that the dose we're giving is so safe, the chance anything negative occurs downstream is really, really low. And thankfully, you know, we're sort of get to stand on the shoulders of giants The FDA has been thinking about this problem for a long time. It's not a risk just for reprogramming. It's a risk for lots of different types of drugs you put in your body. Maybe they could cause cancers, do something bad that you didn't understand. And so there are tools you can use to ask, you know, for example, if I put this in animals and I basically look at them for a long time, can I measure if they don't get tumors? And we now have a lot of data on sort of what sorts of models are predictive of what will happen in a person. And we can leverage a lot of that existing safety knowledge to make sure that our medicines are safe. Safe too.

Some of the work you guys do, the first time I came here, it's kind of cool, the starting point. I mean, you just have— we talked about it a little bit, but you have these— you've got bags of old blood, you've got bags of young blood, right? And then you're trying to turn one thing into the other. And so you use those cells to run the experiments on the old blood and then see if it starts to resemble what's in the the good stuff.

Yeah, yeah, exactly. I mean, so we've tried to keep this pretty simple from day one. I think in biology, often doing the right experiments is hard, and so we often allow ourselves some approximations. As an example, cells from real people, they don't grow crazy well in plastic dishes. They evolve to grow inside a human body. And so everything's easier if you're willing to just do your experiments in cancer cells. Those guys will grow on anything. And so folks will often make those sorts of approximations early on, But we worry that ultimately what we're trying to do is make medicines for patients. We're not just trying to write a paper, like that's great too, but we're really trying to make medicines. And so we want to start from the earliest days doing all of our experiments in real human cells, hoping that they'll translate to human cells inside of a patient. We've even started now doing some of our earliest screens in vivo. So we can make, for example, mice that have mostly human livers. And so it's not a perfect representation of what'll happen in a person, person, but it's a human cell inside a working organism, it's probably about as close as you can get before you get to the clinic. And so we're willing to bite the bullet on doing a bunch of hard stuff like getting hundreds of bags of blood and isolating all the cells. It's certainly arduous. I've been on calls at 2 AM trying to get hepatocytes sent to NewLimit so that we can get the right— hepatocytes are Greek for liver cell— so we can get the right liver cells at the right time, dealing with a bunch of these logistical and operational issues. We're willing to bite the bullet on that because we think at the end of the day, it means the results of our experiments are much more likely to actually be meaningful. To actually translate into people.

I heard you talking about this mouse with the human liver on this presentation. I mean, how do you, how do you put human liver cells inside of a mouse and actually—

It's so crazy this works. It's so crazy this works. So we can't take credit for the idea that this works. I can tell you a little about what we've innovated on, but these mice were actually invented by a member of our scientific advisory board called Marcus Grompe. And so basically the trick is that there are these diseases which cause the hepatocytes, the liver cells, to be sick. And so they don't work very well, and they actually start to die if they don't get a certain medicine. And so these are diseases that happen in real people, but you can also engineer that disease into a mouse. And so now we've got a mouse which doesn't have an immune system, so you can transplant in human cells and they don't get rejected the way that, you know, if you transplant in the wrong blood type into a human, it gets rejected. So we just got rid of the immune system in these mice mostly, so we don't have to worry about that. And then the mouse liver cells are sick. And so all you need to do really is inject them with a dose of human liver cells that don't have that genetic disease, and then you remove a drug you were giving the mice that helped their sick liver cells stay healthy. So now the sick liver cells start to die off. They start to get sick. The human liver cells are not— don't have that genetic disease. And so they start to outcompete the mouse. And over time, you're really pushing each of these human liver cells, and you're asking the question, how well can you regenerate in this sick environment? Kind of resembles like a disease in a human. The rest of the liver is sick. The cells are dying off. Maybe these cells are seeing damage, and they might want to die themselves. Themselves, and you're asking not only can you not die, but can you repair the damage that's occurring elsewhere? Can you make more liver when you need it? And so over the course of a few months, what you end up with is a mouse that has very few mouse liver cells left and mostly human liver cells. That's kind of crazy to think about, but you're holding a mouse that has basically a human tissue in it. So it's not a perfect model. Again, these mice aren't that big. They weigh about 30 grams. So I'm not going to claim it predicts exactly what will happen in a human in the clinic, but a lot closer probably than what happens in a normal mouse and probably about as close as you can get before you get into human beings.

That is weird. And the liver, sorry, it's probably— No. Really stupid question. The liver doesn't try to grow, like—

No, it's crazy that this works, right? It's not a stupid question at all. So the liver has all sorts of internal signaling that basically tell it how big to get. It's kind of a crazy biology problem when you think about it. Like, why don't my organs grow too big? Like, how do they know? How does my heart know what size to be? How does my liver know what size to be? Turns out there's all these signaling mechanisms, and a lot of them are simpler than you'd think. They're like mechanical. When the liver starts to press up against other tissues, it says, ah, okay, I'm the right size, and that tells the cells to stop dividing. And all of that circuitry, all of those genetic signals, are actually conserved across a lot of species. So the human cells inside a mouse, when they fill up the right amount of liver space, the mouse body says, "Okay, liver's big enough, don't grow anymore." And so these mice can actually live reasonably long periods of time with human liver cells in them, and they function just fine. So what we've been working on, and this is sort of where we started to introduce innovations on top of what our SAB member Marcus invented, we've started working on actually building those livers with mixes of young and old cells rather than one donor liver at a time. We mix a bunch of them together and then we can use genomics to actually ask what happened in the liver cells of each individual person. So rather than testing our medicines effectively in one human being who might be weird and idiosyncratic in their own way, we can mix cells from a bunch of people together and say, okay, which interventions that we try actually work in everybody? Because what we're trying to do is make medicines for everybody, not like donor number 6, which is what a lot of firms end up doing. Up doing. And so we've also found ways to take that system and put combos of transcription factors into the hepatocytes. So now we end up with this mouse whose liver is not only human, but all of its hepatocytes have different combos of transcription factors, and we can turn those on and off, mimicking what happens with a drug, and ask at the end of the day, did we make those cells look younger? Do they express the right genes? And more importantly, did they regenerate and actually outcompete some of their brethren and build a bigger part of this new human liver inside the mouse?

Yeah.

Which we hope is associated with their ability to regenerate and function in a human being who might have a disease or just normal aging. I give you—

I give you guys credit because I feel like you always explain where you're at in fairly reserved, clear terms. You're not promising, overpromising to me. I think a lot of times when I hear about some of the science in this field, it's always with mice. It's always in the liver because the liver seems to regenerate. It seems to like absorb all this stuff. A bit easier, won't there be some critics who— and I know it's early days and you have to start somewhere, but yeah, wouldn't there be like, okay, all your exciting— not all your exciting stuff, but some of your most exciting stuff, it's in a liver, it's in mice. This is like where everybody goes. How excited should we actually be about the results you're seeing?

Yeah, well, you know, for obvious reasons, like, I'm stoked on what we're doing, but I think it's a totally fair critique. You know, people hear we're trying to make you live longer and then they get like 10 minutes into us talking and it's like, oh, it's all about livers, it's all about these tiny little 30-gram animals, how much does this actually matter? The way we thought about this at NewLimit is that you need to start somewhere. I really admire firms that have the ability to stage their ambitions over time. And so we think internally about what's like the Tesla Roadster, the first thing you can build that really works before you can scale out to eventually have the Model Y be the most popular car in the world. And so here for us, we think that for a bunch of reasons, like you highlighted, you can get these mRNAs into the liver really easily, and the fact that we know the liver is important, We know if you make it younger that it'll have benefits for people. We started there, but that's not the limit of our ambition. We eventually want to extend out into other cell types. So we also work on T cells. We talked about isolating cells from the blood earlier. We're hoping to fix your immune systems when you get an infection. When you're old, you can actually fight it off the way you do when you're young, so that when you're 75 and get sick from your grandkids, it's kind of annoying rather than like sending you to the hospital, which is unfortunately what will happen today. And eventually we want to be working on many other cell types in the body, but you have to do this in a stage-wise process. We think you really are going to need these specific reprogramming medicines for different cell types in your body. There's not going to be just a single pill probably that you can take. We'd be very excited if that were the case. You could get like one single set of factors that works everywhere, but we think you're going to need to do it in a specific way. And so the stage-wise process I think is the right way to roll this out. We're not going to go to Mars tomorrow. You have to first reach orbit, sort of metaphor that I like to use internally. We think where we're at today is that the results we have in animals using reprogramming are as strong as what some companies have used to go into the clinic and actually run human trials. We're not going to do that next year, and I don't want to overstate or overpromise there. But just to give you a sense of scale, these are the sorts of data that people think are about as good as you need to be able to run an experiment in human beings.

And on the liver and the T cells?

On the liver right now, is sort of where we're most advanced, is on the liver. We've got results that, you know, really are the same sorts of results people have triggered clinical trials on. And so we think it's still very early days, but for a drug development program, this is pretty promising. You know, it's as fast as I've seen a program move. We started working on the liver 9, 10 months ago now, and already we're at that position because of some of the technology we built in T cells enabling us to move really quickly across cell types. So I think this is still a really exciting time. You know, yes, drug discovery is a long road, but I think we're in a place where anyone listening to us chat right now could anticipate that in some number of years you're going to see these reprogramming medicines in people and some number of years after that, you're going to see them actually deploying out not just into patients who are very sick, but eventually people who are healthier and healthier, but who want to stay healthy, who don't want to be degrading as they age. And so it'll take a while. This is unfortunately not quite like software, but I don't think it's pure science fiction anymore. I do think we're going to start seeing clinical results in a number of years down the line.

So I mean, in some sense you answer the question completely. I just, I feel like sometimes I here, it's like, ah, and it's working in the liver. The second we go to, if you wanted to deal with something like dementia or Alzheimer's, the second we try to reboot this in the neuron, all this stuff that we were so excited about that worked in the liver is just not applicable. So you're on this journey, but it's like, can this stuff even translate to the other bag of things?

Yeah.

I mean, we don't know today, and I don't want to overpromise there. We don't know if the types of approaches we found can allow us to discover sets of transcription factors that work in every cell type, but we have the ambition to go out and find out. And I think we know from the Yamanaka factor reprogramming case that in principle, it can work everywhere. Whether we can turn that into a medicine is a whole other question, and we don't know the answer today, but in principle, the existence proof is there, which is not something you can say about every therapeutic program, that you know there's a possible hit in your dataset somewhere. So I think even if we just fix the liver, and even if we just fix the immune system, No one knows for sure, but my own personal bet is that still buys you multiple years of healthy, active human life. And I can't really imagine a product more valuable than that, even if it just works in the liver and the immune system. You know, what's more valuable than more time to spend with your loved ones, more time to spend on the things you love, honing your craft, deciding which hiking trail to go on rather than which nursing home to end up in?

Yeah, yeah, nursing home.

You know, so we're really excited even if that's the first place. It works. But we do have the ambition that eventually we do want to address almost every cell type in the body as well. But it's still early days. I don't want to overpromise and say like, oh yeah, for sure it's going to work in Alzheimer's. Don't think anyone knows today, but we want to be the company to try.

And because this technology is so exciting, a ton of money poured into this. There's you guys, there's Retro, there's Altos Labs, there's others. But even just among those three, I'm probably talking about, you know, billions of dollars in funding has been poured in this, so many of the world's leading scientists who work on this stuff. So all that is great. It's like kind of moved from academia into industry. Things are going to go faster. It is also— there is an argument to be made that maybe you guys are publishing less, you're sharing information less. Altos Labs in particular is very secretive. This is such a prized technology. Wouldn't it be more wonderful if, like, all of the this was out in the open. I mean, this strikes me as like one of the things I'm most interested and excited about because it could be just— it's like antibiotics level of—

Yeah, that's entirely the ambition. It's like antibiotics level improvement to human health.

Right. So yeah. Is there a downside that all you guys have come into this and sucked up all the brains into your research caves?

Yeah, well, I hope we're not a cave anymore. Ashlee Vance is referring to our our first office, which looked like a cave. I don't think there's really a downside. I think there's still a lot of attention in academia. And if anything, I think this particular problem, this particular science still gets less attention than it deserves in terms of the potential impact.

It does. It's crazy to me.

Yeah, it's like— I looked on the New York Times. I don't want to totally libel myself or whatever, but I think when I looked, it was like Yamanaka won the Nobel Prize, I think since then there was one other story that mentioned iPSCs and Yamanaka factors over years. And then this is bizarre to me because this seems like the— yeah, potentially the most exciting thing.

It's one of the most exciting things happening in like the natural sciences, not just biology to my mind. I mean, talk in my own book, but like, I really believe that, you know, New York Times can be no core memory. We can all be— we can all be at the same level of scientific—

we have got in our early days, we've got heavy iPSC. I can stand behind that. Yeah, it's not.

I appreciate it. Yeah, I don't think this science gets enough attention already. And so I think the fact that we've started to see some level of investment in this science is, on the industrial side, is actually really a net boon. I think what you're going to end up with is because of this industrial investment, you'll see more academics working on this problem on net, not less.

Yeah.

I think the ecosystem needs both pieces. You know, there are questions that you can answer in an academic setting on very long time horizons, really low chances of success, but we can't really do in a company. It doesn't make sense as a business. And so that ecosystem has lots of interaction and cross-pollination. And I think seeing that there are places where talented people who have worked on the science can go after their academic trajectory, seeing that there's a hope that if you discover something really exciting in the academic world, there's a pathway for it to actually reach humans through some of these great organizations, not only ourselves, but some of the other companies that we admire as well, like Retro and Altos. Then I think that leads to more attention, not less. And I think really, I think about growing sort of the size of the pie of people paying pay attention to this and way less about dividing up the slices because right now the pie is just microscopic. And I would rather not talk about how many microns your slice is. I'd rather be arguing about how do we get our pie to the size of meters in diameter. So that's sort of how I think about it. And ultimately the types of problems we're trying to address making medicines, they can really only be done in industry. And I think the history of this plays out across therapeutic modalities. You look at the history of biotech, recombinant DNA was invented at UCSF Stanford, but it was actually pioneered and turned into medicines at Genentech. That's not an accident. And that's been repeated with cell therapy, repeated with monoclonal antibodies. You really need the focus and the concentration of an industrial effort with, I think, a for-profit motive is really important in able to actually translate this and turn it into a product that people can have in their own lives. That's ultimately what matters. Like, I see business as the vehicle by which you turn ideas into benefit for real people. And I don't think we've discovered as a human civilization really a better mechanism mechanism of that than a company where you get a lot of smart people who wouldn't otherwise organize together to do more together than they could alone.

Even among my longevity chums, which I have many and I think are pretty—

Why am I not on the invite list? No, no, you're in it. Get me on the text thread.

You're in it. I just meet all this group of people I've always tried to learn from and understand this world. I would characterize this as pretty quite rational, science-forward people, not the kind of fringe end of longevity, like people I trust. And, you know, half the people are people like you and they're super into this. And then the other half are just like, this is never— with transcription factor, this is never going to work. And they're really down on the technology. Yeah. Yeah. I mean, so, so you might— I'm sure you face that and obviously you've decided excited. We're going to see if this works or not. But what is the— there is a case to be made among some people that there's like, we'll never get this recipe quite right.

Yeah, yeah. If I could do my best, like, you know, Berkeley rationalist steelman argument, like, let me like indict like why what we're doing might not be perfect. I think a lot of the arguments I hear are something like biology is really complicated. There are lots of molecules running around in in the cell, lots of challenges that can emerge with age. Things like mutations in the DNA aren't going to be fixed just by changing the epigenome. That really tells you which part of the code to use. If the code's messed up, it doesn't fix that. And so, you know, we're not going to get to living forever just with epigenetic reprogramming alone. And I basically agree with that. I don't think we're going to be able to live forever with epigenetic reprogramming alone. But taking the median lifespan from, you know, 76 in the US to 95, eventually to 110— we know thousands of people live to 110, that's not impossible— is a very audacious goal that I would love to be part of actually realizing in the world. And I don't think saying that epigenetic reprogramming is not the perfect answer to all of aging means that it is not going to be one of the highest impact levers that we can actually access. And I would argue probably the highest impact mechanism that we could work on at this time. So I definitely hear some of those arguments and I think they're valid. But again, I think this comes to the sort of the stepwise staging argument. If you spend your entire time on whiteboards trying to find what the perfect what the perfect plan is that takes you to immortality from today, and you don't do anything until you find that perfect plan. Well, people have been whiteboarding around for a few hundred years and haven't found the perfect plan yet. I don't think it's coming. I really believe in sort of a bias for action that produces new information in the world. And I think the path we're on has the potential to add decades of healthy life to every person. And for us at NuLimit, we're wonderfully excited by that mission. Would this be one of the most important moments in the history of humanity, on par with the invention of antibiotics. You know, we went from an average lifespan of around 45 a couple hundred years ago to an average lifespan of 76. We like forget this is like the miracle of the 20th century. I'd argue as important as any other technology of that era. And I think we've got the chance here with epireprogramming to do that again. And so while yes, it may not be the solution to all problems, it's a big enough solution that it's worth working on.

And I'm sure there's different idealized— I guess the idealized form is maybe you pop a pill and you pop that pill once and you're rebooted across your whole body. I know, you know, that's far off. Yeah. And so in a semi-idealized form, near-term, say it's like 5, 10, 15 years out and things are working really well. People would— they would what? They would show up for like an IV and have this a little bit of a cocktail that would reboot multiple kinds of cells. And then they would have to— this, this This isn't a one-time thing like a pure vaccine, right? At least in the early days, like you would show up every few months, every couple of years to get this reboot through this, this, this cell cocktail.

Yeah. Yeah. You're asking like exactly the right question, which in some ways is like, okay, let's imagine we even find the factors that make an old cell look young. That's awesome. But it's not sufficient to make a medicine. We can imagine a version where it makes an old cell look and act young for 10 minutes, in which case, if you need an IV, it's actually not that useful of a medicine. And so you need it to actually be a durable effect. So I'll start first with like, what do we know is possible and what do we have today and how do we get to maybe spanning the gap between those? We know epigenetic reprogramming can last decades. The most extreme form of that is that you were once a bunch of embryonic stem cells that became other types of cells and your eyeballs don't spontaneously forget that they're eyeballs and become a pancreas. And so we know these epigenetic marks can last over, in principle, your entire life. Lifespan. We also know that from other groups, you can go in and make targeted epigenetic reprogramming events, you know, change the epigenetics around a single locus in the genome, and that can last multiple years in a non-human primate. So those experiments have been done. That's an existence proof exists. We also know in the laboratory, you can go and make some of those specific edits and then make cells divide hundreds of times, kind of simulating like multiple lifespans worth of divisions where they have to copy this mark over and over again. And it still durable. So I think the upper bound on what's possible is basically you could have, in principle, a one-time dose that resets your epigenome back to a youthful state.

Okay.

We don't have that today. That's just like, in principle, it's possible. So what we've shown today is that these effects can last in animals and human cells growing in, in, uh, inside mice and humanized livers. These effects can last at least a few weeks. And already, I think, for certain types of patients, that's actually enough. There are patients who every 3 weeks go in and get an IV for an lncRNA medicine, and that's been going on for about a decade now. So we think even that duration is tractable for a medicine. But certainly for a broad population, what you'd want to do is extend that durability first to months. Ideally, you'd have like a 6-month dosing first. And there are patients who get a medicine like that called Inclusiran as well. Every 6 months you get an IV, keeps your cholesterol lower for that particular medicine. Ideally, you then move out to years. Maybe that number of years gets quite large and this becomes something that you do every once in a while as part of your regular checkup. Ideally, it's not something you're worried about all the time, but once every couple of years you go in for your physical and you get your IV of your reprogramming medicine and you don't worry about it again for for quite a while.

You're so eloquent and well-versed on both the biology and the computation. What's your background to get to this place and what drew you into this?

Sure. So I did an undergrad degree in effectively molecular biology, and then I did a PhD at UCSF right up the road in San Francisco. So technically my PhD is in developmental biology, but I'll jokingly say, don't ask me to look at an embryo because I know less than most would. I really did my PhD at sort of the intersection of computer science and molecular biology. And so I like to tell people, I try and look for problems where by being a dilettante in both of those fields, you can make progress as opposed to just being an expert in one. And so I really have been working on aging my entire career. My sort of story of why I work on biology is I originally wanted to be a physicist when I was a kid. I really thought Richard Feynman was the coolest guy Guy alive. And then I realized that when I read all these physics books, all the dates of the important discoveries were like 100 years ago, and then they just kind of stopped. And I was like, wait a second, it seems like maybe progress in this field isn't going quite as quickly. And by contrast, I was reading a biology textbook one day, and the beginning of the book starts like you'd expect. It's like a chapter about how life starts, like here's how a cell becomes an organism. And I was like 15, so I did not understand all of the details, but I understood there were a details there. And I kind of naturally expected I'd get to the end and I was like, okay, this is where the answer is about why everyone gets old and dies. And there's just no chapter on how aging works. And I thought originally, like, this textbook must be bad. Let me go find another textbook. I realized there's never a chapter on how aging works. And I just became fascinated with this originally as sort of just a really profound, important problem in the natural sciences, sort of on par with like, what is matter composed of? How does energy emerge? Is why do biological systems inevitably break down? And then as I had matured and grew as a scientist, realized if you could answer this question, it's not just a curiosity. If you could make even a marginal impact on the rate of aging in people or reverse it to some degree, it would really be the largest impact we could have in all of medicine in terms of adding happy, healthy years to individuals. And if you think about like what are the most valuable products one could produce, I really do think it's time. It's like more years to do the things you want to do, more years with the people you care about. Nothing, you would trade nothing for that. At the end of your life. And so I've always been fascinated with this. And so, you know, as a graduate student, I worked on trying to study how muscle stem cells aged. Why, no matter how many bicep curls do I do when I'm 75, can I not get jacked? That was sort of the question in my mind.

Unless you're Bryan Johnson.

Unless you're Bryan Johnson, in which case you might still get jacked when you're 75. And so worked on basically trying to address that question by building new tools to measure how these cells behaved. At first, I was really trying to apply modern deep learning and computer vision to microscopy.

That's sort of how I got into the machine learning field.

Build. This is now like 10 years ago, right when things were kicking off. This is like right after AlexNet. And then I went on to a company called Calico, which is owned by Alphabet, run by a really legendary CEO named Art Levinson, one of just like the most intelligent people I've ever had the pleasure to work for. Yeah, he was CEO at Genentech. He was one of the early employees. He's chairman at Apple. Again, the guy's like truly a genius, somebody that I really admire and look up to. Had the pleasure to work there for a number of years across some of the both the drug development side and then later ran a group in the aging biology division and so worked on epigenetic reprogramming there as well. So again, have sort of traveled this path of always trying to understand how does aging work, how can we ideally intervene upon it? And then because I'm not an expert in either molecular biology or computer science, trying to find those problems at the intersection where being good enough at both is sufficient to really make progress.

Yeah, okay, so, but you're not, you mentioned before, it's only in the last 10 years that you thought some of this was possible, maybe even more recently if we're honing in on the AI part. It's just, it's interesting 'cause aging's gone through mega transformations, longevity field, depending how far back you wanna go. If you wanna go all the way to like Fountain of Youth and, you know.

Longevity Field Foundation, 1505, Ponce de León.

But you know, we do go through these waves, right, of between magic and science and this is real. And then now it's a cult and now, you know, like, these oscillations, but you've always been coming at this from this— this is not like— you've always been coming to this from the science side. You're not— you didn't grow up in the whole sort of longevity cults of the Bay Area.

No, no. If anything, I don't think I'm still part of the longevity cult in the Bay Area as far as I know. No, I come at this very much from the hard science angle of ultimately what I find rewarding is creating real value in the world and making people people's lives better. And the difference between science and magic is that science actually works. And if the science isn't solid, you're just not going to create value for people at the end of the day. It doesn't matter how exciting it sounds. And so I would rather be a little bit more conservative in some of my statements. I would rather be working on something that is a little bit, you know, nearer term or closer at hand if I think it's actually going to work than to maybe play, participate in some of the more outlandish sort of let's live forever activities in the field. Yeah, I don't necessarily disparage folks who are interested in that domain, but I personally am really interested by delivering real value for real people. And I think sticking to the most rigorous aspects of the science is the way to do that.

And there's 4 founders of the company?

So there's the 3 co-founders, myself, Brian Armstrong, Blake Byers.

Okay.

And they're co-founders of the company, but they obviously have lots going on in their lives as well. And what was the— you guys were all just sitting around one day and we're like, transcription factors are so exciting, we should Were you going to do something here or what was the genesis?

Not that far off, honestly. So I was working on epigenetic reprogramming previously. I was running a lab at Calico. I had published a couple pieces of science around this and I think my co-founders Blake and Brian independently had been spending some time together thinking about what are the really important problems we can work on. They'd both been very successful in their other endeavors and thought, what's one of the core quests we can go on that would really improve humanity. And I think if you sit down and really think about it, like, what are the most important problems in the world? I can list off a few. It's like creating maybe artificial intelligences, massive potential to make the world better, maybe making humans multi-planetary. You need a backup system for the human race off of Earth, creating infinite energy, making fusion, um, not dying. Like, these are kind of all in the same order of importance. And I think in that latter category, my co-founders felt, you know, there, as we were just talking about. More attention could be going here. Like, yes, there are a few envoys. It's not to disparage anyone who's already in the field, but just relative to the potential impact, there's not a lot of attention. We ended up getting connected through some mutual friends talking about the work we were doing around epigenetic reprogramming. And I think all of us were excited about the idea to actually make medicines using these tools, not just to do the basic science. That's great too, but to try and translate it, to try and make medicines this decade, not just this century. And if you look at the history of the biotechnology ecosystem, I believe it to be true that an incumbent has never introduced a new therapy therapeutic modality. Never once. It's always a small startup. It's always a scrappy group of people like a Genentech, a bunch of guys in the South San Francisco warehouse next to an old steel mill that introduced the new therapeutics. Can dive into some reasons why, but that's just empirically what happens. And so we felt if we want to see these medicines in the world, we probably need to build a company. That's just historically how this has happened. And so we got to talking about what that company would look like, and that eventually became Newlimic.

Okay. Okay, so I know you guys just did this update where you brought people up to speed on some of your liver work, some of your T-cell work. You talked about how the model, you're going to keep scaling it up, getting data better and better, hopefully to find more of these promising combinations. But just to give people a sense of what you're working on very focused over the next year, 2 years, and if it is going into any other types of cells?

Yeah, yeah, absolutely. So I'll maybe break this down into like 3 bullet points. So one of the things we're trying to do is actually advance products. So we have a few what we call leads now. So these are like real prototype medicines you can hold in your hand. There's a tube of them. It's not science fiction. And we want to start taking those toward the clinic. And so ideally by the end of this year, we'll have what we call a preclinical candidate, a medicine that we think this really is worth betting on. This really could be—

For the liver or it could be—

This would be in the liver right now, but almost certainly if we found it in T cells, I'm not going to be upset. Probably in the liver. And then the other sort of piece of this is on the technology and what we call our discovery engine side. So we want to scale up the amount of data we're generating. We want to build models that are even more performant. And sort of the high-order bit there would be not only can we test roughly an order of magnitude more sets of TFs, but of those sets we test, can we use those models so that the number of them that actually yield interesting results goes up several fold? Right now it's roughly 1 in 100 of the things we test is actually interesting. Which is way better than it was when we started. But still, you know, you could go from 1 in 100 to 2 in 100 to 5 in 100. And really what that reflects is that for every dollar you're investing, you're getting back X more discoveries. So we want that to improve over the next couple years. And then finally, we want to work on more than just the liver and T cells, to your earlier question. So we're going to be onboarding our third cell type over the coming year. It'll likely be endothelial cells. So these are the cells of your vasculature, your veins and your capillaries, your arteries. In some ways, this is the tissue that's inside of every other tissue. And I think you'd be surprised surprised to learn how many of the challenges your other organs experience can really come from challenges in getting access to sort of the information superhighway of your body, which is your bloodstream. So those are sort of the big 3 headlines over the coming years. And I hope if we have a chance to connect again a few years down the line after that, we'll be talking about those milestones being things like actually entering human trials.

And then today, I don't know if you got the full scope of what we get to see, but I know— so we are going to get to probably see this mouse with a—

You can see a mouse.

Human liver.

He just looks like a normal mouse, but you can see him. Yeah. He's got a human liver. And kind of the remarkable thing is he doesn't look weird. Yeah.

Yeah. And then what else might we see in that building?

Yeah, so people can show you some of the data from that. Folks will probably show you isolations of cells from blood. Folks can show you the automated microscopes we use to measure how well T cells are able to perform their job of killing other cells. Folks can show you big factories of viruses that we cook up that we we use to deliver reprogramming factors. Folks can show you areas where we take transcription factors from an idea into actually building molecular constructs we use to start experiments. People can show you DNA sequencing and how we profile cells, a couple different types of that. So there will be all sorts of fun stuff. You get to look at robots moving liquid around, always everybody's favorite.

Love a robot, a pipetting robot, always the best. Well, thank you so much.

Thanks for your time.

We're, we're looking forward to exploring.

Yeah, thank you, Ashlee. Great to have you here, and thank everyone for your attention.

That was great, man.

Awesome.

Thank you.

Thank you.

I hope everyone enjoyed this episode. As always, Core Memory is hosted by me, Ashlee Vance. Our guest today was Jacob Kimmel. The show is Produced by me and David Nicholson.

Our theme song is by James Mercer and John Sortland, and the show is edited by the not egregious John Sortland.

And it was sponsored by E1 Ventures. Turn to E1 Ventures for all your venture capital needs. Thank you so much, E1, and we'll see you on the next episode.

Good.