The future of cell-free biotechnology

0:00Michael Jewett:This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. Since we started this show eight years ago, it's become an archive of amazing and impactful work by my Stanford colleagues. Research is not something that just happens in the lab, and as you'll hear on this show, the research at Stanford can impact areas like health, technology, law, and business, and many other topics that can affect everyday life. We hope you'll tune in to learn more about how research has the potential to help your life and to help the lives of people you care about in your family and your community.

0:32You can kind of think of DNA like a cookbook. It stores the instructions. We transcribe that information into a temporary copy, like your grandma's recipe card, and then you translate that into proteins, which, as you said, they're the doers of the cell. What's so special about these cell-free systems is they have all of the processes of biology going on. In fact, like energy metabolism to support how we make proteins, but we can produce those proteins outside of the living organism. So that can allow us to make things like medicines when and where they're needed.

1:07Michael Jewett:This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. If you're enjoying the show, please share it with people you like, people you love. We'd love to have them on board. We'd love to share with them the future of everything. Personal recommendations are the best way to go. Today, Mike Jewett will tell us that you can get better medicines, better vaccines, better biofuels by using biological systems where we strip away the cells and we just use the guts of the cells to do useful things. It's the future of cell-free biotechnology. Today, we're continuing our feature, The Future in a Minute.

1:43Michael Jewett:At the end of my interview with Mike, I will ask him some rapid-fire questions, and he'll give us his rapid-fire answers in the future in a minute. And also, before we get started, remember to tell your friends and family about the show, especially if you're enjoying it. We'd love to have them on board.

2:05Michael Jewett:So biotechnology is one of the great things that has developed over the last 50 years. We can use living tissue, yet living cells, to create medicines, to create vaccines, to create biofuels to run cars, engines, and many other applications, including things that will help for sustainability. There are even bacteria that can eat toxic materials and turn them into non-toxic materials. Well, the problem with that technology is it relies very heavily on cultivating and growing large numbers of cells in huge vats the size of a swimming pool in order to get enough materials to then do all this useful stuff.

2:43Michael Jewett:Well, it's possible that we can break open the cells and take the pieces inside the cells that do that work, isolate them, and make much more effective and efficient little machines, biological machines, to create these very same medicines, vaccines, and biofuels without having to keep those cells growing. Mike Jewett is a professor of bioengineering and chemical engineering at Stanford University, and he's an expert on these so-called cell-free biotechnologies. He'll tell us how they work and how they're starting to change how the biotechnology and bioeconomy might look like in the future. Mike, you study cell-free biotechnology.

3:26Michael Jewett:What's cell-free biotechnology? Hey, Roz, that's a great question. So, you know, biotechnology is all around us. You know, We've used this to make medicines for decades. And generally speaking, people use microorganisms, living cells, to make those types of products, like an anti-cancer drug. My lab does something different, which is called cell-free biotechnology. So rather than using living organisms or cells to make stuff, we actually use the components of cells to make those types of products. So what we do is we take cells, we rip off their cell walls, and we collect the insides. and then we use the insides as a molecular factory to produce everything from medicines to chemicals to diagnostics.

4:11I guess you can kind of think of it like taking a car and lifting the hood up and pulling the engine out. It's like using the engine but without the constraints of the car chassis.

4:20Michael Jewett:Gotcha. Okay, so far free be it for me to like question the wisdom of this but let me just ask a couple of questions. It seems to me that cells, the good part of the cells is that they're living and they do things like repair stuff that's broken. So like you said, cars, and that's a fair analogy, but cars don't fix themselves. Whereas to some degree, cells fix themselves. Like if they have an injury or a problem going on, they have some sensors. So are you throwing out the so-called baby with the bathwater when you give up the intact cell and all of its kind of amazing biological capabilities that we actually don't really fully understand yet?

4:57Yeah, so that's actually the totally surprising part. We're not actually throwing the baby out with the bathwater. In this case, we're actually just getting rid of the evolutionary baggage of the cell. So I'm an engineer, Russ. And so what I try to do is, you know, take biology and use that to, you know, let's say, create better access to medicines or convert carbon oxides into the atmosphere into, you know, sustainable chemicals. And one of the challenges as an engineer is that those cells that repair themselves and grow and have all these kind of great features. That's terrific if you're a cell living in the environment and you're trying to adapt and live and survive.

5:37That's actually not necessarily what I want as an engineer. And so oftentimes there's a tug of war that exists between what cells want to do, let's say live and survive, and what I might want them to do as an engineer, which is to produce a medicine. And so the advantage of our approach is that we kind of get the best of all worlds, we get the capacities of biological processes, like how do we convert inexpensive substrates into things like medicines and chemicals, but without the constraints of that living evolutionary baggage that oftentimes is totally fighting what I want to do as an engineer.

6:12Michael Jewett:So that actually is, that's a great answer. I knew you would have a great answer. I didn't know what it was. So let me ask you, tell me about this evolutionary baggage. Like that's very interesting idea. Tell me about the things that like cells tend to do that can like mess up your engineering goals. Yeah. So so cells are like finely tuned in terms of their ability to, let's say, keep the processes that they need going alive. So and one of the major ones is how a cell uses, let's say, energy to support its growth. So we oftentimes use the components of microbial cells like bacteria or yeast. And those cells, as they're growing, oftentimes like 50 to 70 percent of the energy of the cell is going towards a process called protein synthesis.

7:03And it's to support that process of protein synthesis so that those cells can replicate and divide and be happy. And so one of the challenges is when I, as a biotechnologist, might give those cells an engineering objective encoded at the level of DNA. And I realize we just jumped through a lot of topics, but maybe we can come back to those. The cell might not want to take its energy and resources to support my objective. It would rather actually keep itself growing and make more of the components it needs for, let's say, protein synthesis. And so that's kind of this tension that we have oftentimes as engineers for those that use kind of cells as factories.

7:41Michael Jewett:Gotcha. Okay. Okay, good. So we've established that the cells have amazing properties, but they also have some features that make them not so perfect for your engineering. And as you said, you bust them open. Forgive my like coarse language. And then you take the pieces like the engine you pointed out and you can use that then to do things. So what are the big wins that accrue when you take this approach? I'm guessing that, for example, we've all seen these pictures of these huge industrial vats where they're growing huge amounts of bacteria, trying to coax the bacteria to do all these things.

8:15Michael Jewett:Maybe it occurs to me that maybe one of the advantages is scale, that maybe you can do this on a larger scale. So tell me what the advantages are. Yeah. So good question. So some of the advantages are actually that instead of maybe having to do things at these massive scales, we might open up opportunities and let's say distributed manufacturing or decentralized manufacturing. So I can give you an example. If you're a large pharmaceutical company and you want to produce a medicine, we do. We grow these organisms in, let's say, 50 ,000 liters or 100 ,000 liter tanks. That's like an Olympic swimming pool, right?

8:51And so when we build the manufacturing plants for those products, it's expensive. It can cost hundreds of millions of dollars to a billion dollars or more. And that creates risk. And so the consequence is that there are oftentimes many types of products from biology and biotechnology that can never reach the customer because the risk is too great that maybe something changes and maybe the customer doesn't need that process. The cell-free protein synthesis systems or biotechnology systems that we use kind of flip that on its head. And we open this chance to use biology maybe at lower scales. So I'll give you an example with vaccines because you asked like, what's a big win, right?

9:33Yeah, yeah. So one of the things that terrifies me is something known as antibiotic resistant bacteria. So in fact, it's now projected by like 2050. the possibility is that there will be more lives threatened by antibiotic-resistant bacteria than by cancer today. And that's scary because, you know, it can lead to many people, you know, potentially getting sick or even worse. And so myself and many others are trying to think about ways to develop new approaches to create medicines, maybe more quickly, at scales that could be distributed and certainly also reach kind of global locations. So in fact, like, you know, 30 % of the world's population lacks access to essential medicines.

10:20And so what we can do with these cell-free systems is we can grow up our cells and our cell-free systems. We can lyse these cells and we've got like this engine. We feed this a piece of DNA and now it can produce a medicine. What's neat about our system is that it can be freeze dried. So you kind of think of it like astronaut ice cream or freeze dried fruit. And so you have this dried powder. And what's neat is we can just add water to kind of make this medicine. So a big win is we've created a just add water biotechnology. In the case of these medicines for antibiotic risk of bacteria or pathogenic bacteria, we can produce about 150 ,000 doses of vaccine to prevent infection by bacteria in about a liter.

11:00Michael Jewett:So you can think about like a carton of milk. Okay, so this is great. And so actually, I asked a good question, but I had the wrong answer in mind i was thinking you would tell me you can scale up and do huge vats of things but actually you're scaling in the opposite direction to create almost a a more manageable deliverable technology add water get your antibiotic and um this this bypasses amazing things like we don't need to have a cold chain of refrigerators everywhere from you know wherever it's manufactured to wherever it's being taken and that that refrigeration can be a huge issue for as you know very well for living cells and all kinds of tissues.

11:39Michael Jewett:So that's a great example. So I've read about recently the work that you've done. And one of the things that you guys focus on is creating proteins. And proteins, as many people will remember from their high school biology, proteins are the things that do stuff in living systems. The DNA that we always hear about encodes for proteins, which are these beautiful three-dimensional structures. They contract in your muscles to make force. They are in your eyes and they bend light to make a focused vision. They digest your food. You're able to make these proteins outside of the cell. That's right. And to kind of understand that process, you're totally right.

12:24You kind of need to understand a little bit about how the biology works, right? At the heart of biology, you know, we're programming instructions at the level of DNA, you can kind of think of DNA like a cookbook, it stores the instructions, right? We transcribe that information into a temporary copy, like a grandma's recipe card. And then you translate that into proteins, which as you said, they're the doers of the cell. What's so special about kind of these cell-free systems is they have all of the processes of biology going on. In fact, like energy metabolism to support how we make proteins, but we can produce those proteins outside of the living organism.

12:57So that can allow us to make things like medicines when and where they're needed, like we just talked about. It also can allow us to explore the landscape of protein sequence and protein function generally. So actually, there's been a complete wave of innovation in biotechnology and medicine over the last even just handful of years with something known as alpha fold, which allows us to take kind of this protein sequence encoded by DNA and define what the structure is going to be ahead of time, It led to the Nobel Prize in 2024. And like the big interesting opportunity for the future and a real frontier of science now is, right, how do we get to protein function?

13:38And these cell-free systems allow us to express proteins very rapidly and kind of explore that functional protein space even more than before.

13:49Michael Jewett:Yes. And one of the things that I think you're doing is in the area of biosensors. And I believe that some of these are at least are proteins. And can you tell me about what a biosensor is and how that's looking in the context of cell-free technologies? Yeah, we're using biosensors. You can think of it like diagnostic or even like a light switch. And so what the biosensor is going to do is it's going to tell me the presence of a molecule that I might care about. And one of these key advantages of cell-free systems in the context of diagnostics is I can create something like a COVID test, but for something else that we might care about.

14:26And you're right, our lab has been looking at these particular sensors involved in kind of water and environmental sampling. So, you know, you're probably aware more than two billion people on the planet lack access to clean water. And so being able to understand kind of when and where water is safe to drink is an increasingly challenging and important problem. Biology, of course, senses things all the time. Right. If you put your hand on a hot plate, you know, you realize it's a hot temperature. Microorganisms are sensing their environment to say, aha, there's food. I'm going to swim that direction because I'm hungry.

15:01Right. And, you know, in this case, there are lots of biosensors that even detect like contaminants like lead and drinking water. And so our lab has been super interested in trying to develop, you can kind of think of it like a pregnancy test for water or COVID test for water to tell us when and where our water is safe to drink. That's tricky because actually contaminants in water, oftentimes you can't see them, you can't taste them, you can't smell them. But these biosensors that we use are proteins that, like a light switch, tell us whether or not the contaminant is in the water. How that process works is leveraging the information flow of biology where we use something known as a transcription factor.

15:43So that's the biosensor that we're using. For lead, it's something called PBRR. this biosensor when it's in the presence of lead will allow for that information encoding of dna to be transcribed and translated into a visual protein that we can see like a green fluorescent protein from a jellyfish or you could imagine making a small enzyme that can take a substrate and turn it into something that's color metric that you can visually see with your eye so that's

16:10Michael Jewett:really cool so if i'm so if there's lead in the water the water is going to turn green basically Yeah, totally. And the hard part, of course, is like getting it to like work so that I can see it in a timescale that I care about. And, you know, like I want to see it in 15 minutes. We all got used to that. You're thirsty. That's right. Totally. And the other thing, of course, is it turns out that biology doesn't need to sense lead at the legal limit approved by the FDA, which is five parts per billion. Right. So biology, of course, it needs to know that it has lead around or other metals because it needs metals for its enzymes or other proteins.

16:50But it doesn't need that the human specified performance characteristic. And so in our instance, what we had to do is we had to teach the transcription factor, this biosensor protein, to read lead at the legal limit, which was a really fun problem.

17:06Michael Jewett:So I was just going to go there, and you've already brought us there, which is you're taking these natural proteins, which might be very good at something. But you might need them to be better or different at that same thing. So in this case, I guess you're trying to make it even more sensitive. Is that the right – was that the direction you were going? Like it was too happy with high lead levels and you needed to tell it, hey, I need you to detect lower lead levels? That's right. So we want this sensor to turn on at like the lowest level of lead possible. We want the light switch to turn on, right?

17:35We use a genetic circuit. But you can kind of think about like the circuit of your light switch. And we need it to turn on at super low levels. And so that was one thing we had to solve for with this particular lead sensor. But the other thing we had to solve for is we wanted the sensor to only turn on in the presence of lead, but let's say not other atoms or other elements that might be in your water like zinc. And so it's something called selectivity as well. We needed to address both the sensitivity problem, but also the selectivity problem. Right. I want the light in my room to turn on only when I'm doing it, not when my child runs around the corner and flips it on and I don't see them.

18:10Right. And so anyways, that's kind of one of the challenges.

18:14Michael Jewett:So in addition to taking these natural molecules, a key part of your research is basically engineering them, evolving them, if you will, to have slightly different properties that are better fit to your use case. So that gets you into the – you're creating entirely novel proteins in many cases. Yeah, so we're creating both novel proteins, new to nature proteins, or just tuning a natural protein to make it into something that we want. In the context of like this lead sensor, if we just stick with this story for a second, the way we did that is a protein is made up of many amino acids. There are 20 amino acids in biological systems.

18:53Those amino acids are arranged in different orders, and it's that sequence of amino acids that leads to the protein structure and the protein function. We don't know how to necessarily change the sequence of amino acids a priori to predict the human specified function. And so that's what we iterate on. So what my lab would do is we make mutations or changes in that sequence of amino acids, probably around where like that protein would bind lead or where the protein in this case might bind to DNA because it's modifying essentially how the process is transcribed. And then we would just iterate quickly through many different sequence changes till we find something that worked.

19:33And that can be an arduous process in some contexts. What's neat is in today's world, we're trying to use machine learning algorithms to accelerate that process.

19:43Michael Jewett:You're listening to The Future of Everything. I'm Russ Altman, and we'll have more with Mike Jewett next.

20:01Michael Jewett:Welcome back to the Future of Everything. I'm Russ Altman, and I'm speaking with Mike Jewett from Stanford University. In the first segment, Mike described what he means by cell-free biotechnologies and why they're so useful and why they can lead to biotechnology in powder form. Just add water. In this segment, I'm going to ask him about sustainability. I'm going to ask him about how he uses AI and a little bit about education. Don't forget, at the end of our conversation, I'm going to ask Mike a few quick questions as part of our segment called The Future in a Minute. Mike will then give me some quick answers.

20:33Michael Jewett:At the very end of the last segment, Mike, you mentioned that you had used some machine learning to kind of supercharge your design of new proteins that had these special lead binding capacity that you needed for your sensors. Is AI playing a big role in your work? Definitely, AI is playing a big role in our work. And I think everyone's – and actually, one of the key things and the only way we solve this problem with the lead biosensor, so we didn't get really to the end of it. But actually, we were able to engineer this transcription factor, PBRR, to read out lead at the legal limit and in the absence of zinc.

21:09So actually, to get a little more specific, one of the challenges with trying to identify lead and zinc is they're both like plus two cations. So they have different orbitals and different geometries. So if we think about the chemistry, but they're kind of similar. They're positively charged metals. And so as we look at how to discriminate between those two, we had to teach this transcription factor and the shape of the active site of this transcription factor to recognize one but not the other. And so in order to do that, we generated a lot of data and we said, hey, let's use machine learning based models to try to guide iterations of this design.

21:47You might have, you know, just kind of back to the previous segment, we were changing the amino acid sequence to get better designs. And so we actually use the machine learning model to predict for us what sequence changes we should make so that we could read out at the level of lead, but not at zinc. Because actually, the first time we tried to engineer the lead sensor, it was terrible. Because basically, we made a better sensitive lead sensor. We could actually report out on lead at the legal limit. But it turned on in the presence of the zinc. So it's actually a totally useless sensor. And actually, the machine learning model allowed us to predict and identify sequence changes that could do both.