a16z Podcast - a16z Podcast: All About Synthetic Biology

Episode Date: January 11, 2019

with James J. Collins, Vijay Pande (@vijaypande), and Hanne Tidnam (@omnivorousread) The idea of 'designing biology' -- once science fiction -- has over the last 20 years become just... science. In this episode, a16z bio general partner Vijay Pande with Hanne Tidnam talk all about the field of synthetic biology with James J. Collins, professor of bioengineering at MIT. Collins, whose work in synthetic biology and systems biology pioneered the field, has also launched a number of companies and received numerous awards and honors (including a MacArthur "Genius" Award, an NIH Director's Pioneer Award, and Sanofi-Institut Pasteur Award). This wide-ranging conversation about the birth of synthetic biology covers everything from the founding story of the discipline to what "engineering and designing" biology really looks like in action -- when instead of engineering electrons, you are engineering toggle switches for genes -- to the disciplinary differences (and synergies) between how biologists and engineers see the world. What are the engineering and design principles, techniques, approaches that work best when applied to science? How does building a company in this new space look different, in terms of platforms and products? And how is this new field changing education in science, all the way down to kits that allow you to play with the machinery of a cell... at home... and even in middle school?

Hi, and welcome to the A16Z podcast. I'm Hannah, and in this episode, General Partner on the Biofund, Vijay and I talk all about the field of synthetic biology with Jim Collins, Professor of Bioengineering at MIT and one of the pioneers of the discipline of synthetic biology. We talk about what engineering and designing biology really looks like when instead of engineering electrons, you're engineering toggle switches for genes, including the scientific story behind the creation of that very first gene toggle on and off switch. We also talk about the disciplinary differences and synergies between how biologists and engineers see the world, what the engineering and design principles, techniques, or approaches that work best when applied to science, how that looks different when building a company in the space and thinking about, for example, platforms versus products, and even how it's changing education in the field all the way down to middle school. So let's start at the very beginning. How did the field first begin to emerge?

What's the sort of founding story behind SynBio? Yeah, you know, the very beginning really was the island of Misfit Toys. So if you go back to the 1990s, the dominant story was the genome effort. And what's interesting is the biomedical engineers did not play a major role in the genome effort. In biomedical engineering departments, the curriculum, the research interests more or less stopped at the tissue level and was only beginning to get interested in the cellular level. In the late 90s, however, as the genome effort was really picking a lot. up speed and beginning to produce these parts lists for different organisms. The leaders of the genome

began turning to the engineers and increasingly to physicists to figure out, help them figure out how these parts put together. Can you explain what the missing information specifically was? So the sequencing would lead you to the ability to annotate the genome, to identify coding regions, identify the genes, identify promoters, but couldn't really give you information about underlying networks that were making up these living cells. So in what way were the genes, the proteins, the RNA elements interconnected and leading to the interesting biology and disease biology. The notion even of network was not dominant or even prominent within molecular biology.

You said one thing that really struggling is you said it was not necessarily biologists, computer science and physicists. You then had this effort to really try to pull the physicists and the engineers into molecular biology. And why them? I think because we deal with complexity well. to get after what became systems biology was the recognition that you needed to embrace complexity. Before the network, what was the predominant kind of idea?

You had really the fixation on the actions of single protein, single genes, and at the level of kind of integrating those, you had pathways. And you began to see efforts primarily out of Stanford, beginning to use mathematical modeling that opened up the doors for folks like me to do basically systems biology. Could you reverse engineer these networks? works. And let's actually explain what you mean by that, reverse engineering a large scale

because that's not a concept that's native to traditional. It's an engineering concept. So it's a strong electrical engineering concept. So the notion is that you have a system that's interconnected or wired together. And so the example that could relate to many of the listeners would be the wiring system or the wiring setup of your house. Say you buy a new house and you've got a circuit breaker box down in the basement. And let's say it's not labeled by the previous owners. Just like my house. Just like my house. So what do you do?

You go down and you flip the different switches on or off, and you discover what they connect. So you're beginning to reverse engineer the circuitry of your house. You then, through additional experiments that you're running the microwave, same time with your toaster, that in fact, oh, those two circuits are actually interconnect even further that's not represented by that. And you begin to infer the underlying structure of the wiring diagram of your house. But our first response was to run away from that problem as fast as we could, because

there weren't sufficient data. to do it. In the late 90s, we just had microarrays, a peer technology that is no longer in vogue, but would allow you to survey thousands of genes in their expression states simultaneously within a given cell. These are incredibly expensive, and when I got pulled in, there were only seven publicly available data sets. So the idea you could reverse engineer a large-scale network was ludicrous. And what were those types of data that were lacking? So there was largely expression data set. And in those days, it was really microarrays. So even though the technology just appeared, we didn't have those data sets. We didn't have a wet lab.

the capability. And so we began to think about, could you put together molecular parts in a network? Could we instead take a tinkerer's approach to molecular biology? So the expression is the output. What's the input? Like, how were you perturbing it then? Well, so on that time, we had no perturbation, right? Perubation would be that you're going to kick the system, where you're going to stimulate the system in some way. Flip the light switch on. Let's say you have your radio on in your kitchen upstairs with the volume on. You're going to run downstairs. You're going to flip the switches one by one until you hit the volume go off. That's a perturbation to your circuit that allows you to map out the circuit. We introduced techniques to do that, but even before we get

there, we stumbled into what became synthetic biology. We said, okay, could we instead of figuring out how these large-scale networks are wired together, could we put them together ourselves? Could we do it with intent? So could we design circuits with a desired architecture and a desired function. And that's the key difference, right, in this entire approach, that you're designing exactly. We are really doing biology by design. In some cases, we're taking natural parts and putting them in new and different ways. In other cases, we're building new parts and then putting those together in new or even established architectures. We spent about nine months thinking about what we wanted to build, and we arrived at building a genetic toggle switch. And

this was motivated by electrical engineering, where toggle switches, which are also called flip-flops, R.S. latches are very simple memory elements, memory switches that can be flipped between 0-1 and the binary state or on-off by just a transient electrical signal. Toggle switches are at the heart of digital memory. And we cycle through many different schematics and circuit diagrams and arrived at one that was basically a mutually inhibitory network, where we have two genes that are set up so they always want to be on, but they're arranged so they're trying to shut each other off. And we showed mathematically, and computationally, that you should be able to have it as a bi-stable system.

It either wants to exist in state A, gene one is on and gene two is off, or state B, gene two is on and gene one is off. And in principle, you should be able to flip it between those states by transient delivering a chemical that would shut off your currently active gene, allowing the gene that had been kept off by that active gene, allowed to come on that now would produce enough protein that would shut off the gene that had been on, and you can remove your stimulus, and you flip from state to state-a to state-by

your state-beat of data. It's like electrical engineering except not electrons. It's now a wet circuit, but in this case, we actually did mathematical modeling computation. And now the next segue

was I reached out to Eric Eisenstadt, who was a program officer at the Office of Neighbor Research, who was running a gene circuit modeling program. And I told him what we're doing, and I more or less called him once a month for over about six months, said, could you give us money

to try to build this in a lab? And he kept putting me off, putting off. And I basically warmed down. I said, you know, I think this could be really big. but we need to be able to build this, and I'd like this to be part of your program. And was the reason why it wasn't clearly, I mean, clearly having an on-off switch

like that seems incredibly powerful, why was it not immediately just because it seemed so hard to do? We had talked to about six or seven very serious molecular biologists who told us this couldn't happen, this couldn't work. They said, for it to work,

you're going to have to have a very tight off-state, two very tight-off states, that's going to be impossible. You're going to have to introduce a very large plasmid. It's going to put a burden on the cell. The cell's not going to like this foreign element. The joke was they used to pat us on our heads and say, boy, stick to engineering because biology is really complicated.

I mean, those do sound like reasonable concerns. I had an incredibly talented graduate student, Tim Gardner, who was a mechanical engineering trained student out of Princeton. Tim presented our modeling, and this was a famous O&R meeting, where he was more or less shouted off the stage by some very serious molecular biologists, including a future Nobel Prize winner, who said it was impossible. This could not work. And in part, Tim, who was green in molecular biology, had been, as a mechanic. engineer who had designed autonomous helicopters to fly around Princeton, didn't know any molecular biology, presented this engineering concept. Nonetheless, in nine months, he had a functioning

bi-stable toggle switch. At the same time we were doing this, Michael Illowitz and Stan Liebler, who were two physicists sitting in both the physics department at Princeton and the molecular biology department at Princeton, were taking a remarkably similar approach to synthetic gene circuits, completely independent of us, and en bonones to us, and that we worked in E. coli with three different repressor proteins using dynamical modeling. They worked in E. coli with the same three repressor proteins using dynamic

modeling. And instead of creating a toggle switch, they consisted of two genes trying to shut each other off. They created what they called a repressulator, a ring oscillator, similarly motivated by electronics, electrical engineering. They consisted of three genes in series. A tries to show off, B, B tries to show off, C, C tries to show up A.

Same parts, just different circuits. Exactly same parts, different circuit. They submitted to nature. We submit ours to nature. end up being published at the launch, the turn of the century in January, late January, in nature back to back. The next week, the head editorial said, oh, you know, physicists are beginning to move into molecular biology.

They said you should look back at those two papers because there's something interesting going on. You know, you raise an interesting point about how everyone on the biology side thought this wasn't possible, almost like the biologist, some of them have an immune response against engineering. That's a great way to put it. And that seems to continue even to today. It is.

I have many, many close friends in college. who biologists, but I think within different academic disciplines, there's often too much tribalism. So the very common critique that we get is, oh, they're not biologists, or they're not chemists, or they don't know basic biology, they don't know basic chemistry. And I, well, yes, yes, yes,

yes, and yes. I don't think it's fair, though, and I think that these problems really can benefit from a disciplinary approach and different perspectives, that we don't know the detail at the level that the biologists do, and one of the more common critiques we'll get, and any time we submit papers to molecular biology journals is that the columns of, they don't have sufficient controls that were done.

But I maintain that in almost every biology experiment paper I've read, they've never had the controls that I'd like to see, which is the idea that you can knock out a gene and that what's lacking now can be attributed to the gene is absurd from a system. It's interesting because, you know, if you think about an electroengineer to a physicist, an electric engineer might now understand like a P&P junction and a transistor the way a physicist might, but can still use them and make parts out of them and still make things. And so I think there's just a role for each one of them,

But there's a fundamental difference between sort of doing science and doing engineering. Yes. And that culture clash is pretty distinct. Yeah, I was just going to say it's not entirely just that academic tribalism, right? It's a discomfort with a different kind of mode of knowledge. I would love to hear if you can describe a little bit more what beyond the tribalism is valid, what you try to think about or what you reject entirely. Yeah, no, it's fair.

I think the differences in science and engineering are healthy and valuable. I think in many cases, engineering, benefits from science in that most of the work that we do from an applied standpoint is really looking to see in what way can we utilize or exploit principles, discoveries that have been made in science and move toward an application. In the same way, science benefits tremendously from engineering in that as a new technology gets developed, it can just open up the ability to address an answer new and different questions. And so there really is a great synergy and it becomes interesting on the controversy. Well, yeah, there's a synergy.

but there is still a sort of of a nominal difference in how they view the world and how they... It is very different, right? And that the engineer often is very purposeful, whereas the scientist

is really driven more by curiosity. Can we discover? Can we learn more? And it's not that the engineer doesn't want to learn more. An engineer can do remarkably well in many instances

with little, if any, understanding of mechanism and that we'll get after a black box so you can represent a model that gives you some sense of relation to be the input and output of, say, going back to our perturbation. And yet that model

may have little, if anything to do with the underlying biology, but it nonetheless has predictive value that's useful to get after a given application of purpose. That drives my biologist friends nutty. But that doesn't mean that the engineer is uninterested in mechanism. That the more we know about mechanism, the better we can do, the more facet. The more predictive of the model. The more predictive of the model, absolutely. Let's talk quickly about the bacteria. So why E. coli? You know, on E. coli, the E. coli was easy to pick. It was and remains really the workhorse. Broadly molecular biology. It's been very well

studied. It's still not fully understood, but we had a number of well-characterized parts, established protocols, and so a couple engineers could relatively quickly get up to speed to do experiments on it. Having said all that, I'll give two points. One is that when I first moved into space, I was told not to work on bacteria because everything's been worked out. We understand. It turned out to be far from true. And second, it's linked to that is that it's fascinating to me that this still is a high percentage of the genes are not functioning annotated in the E. coli, 30 to 40 percent. Not unless it was enough there. that we could begin to design and build with some level of predictability and some level of

comfort that you could insert these circuits and get them to function the way you like. Can you tell us about the latest in bacteria that you guys are now working on? It was then a very nice piece by Louis Serrano and Attila Becksky that came out also in E. coli about four months later in nature. And then it was about no and a half years of nothing coming out because the field didn't exist. You had a number of modeling papers, but nobody really had experiments going. We and others then got excited about engineering bacteria, and particularly E. coli as living technologies, as programmable cells that could sense something in the environment, make a decision, and act on it. And in 2004, began creating these programmable

cells to be wholesale biosensors. But the notion of putting it in somebody's body was crazy. The idea of the microbiome was not prominent. Nobody was talking about it. And so we talked about how you could engineer these, place them on chips, computer chips, electronic chips, and placed them in a quad at a university to see, was there an anthrax release or was there... Detectors. Detectors of lead paint in a home or on lead on imports and showed that you could engineer these quite readily. And then by the late 2000s, early this decade,

a lot of excitement arose. Could you engineer bacteria to serve as living diagnostics and living therapeutics? Could we engineer bacteria to detect and treat cholera? And we decided to engineer lactococcus lactose lactose, which is found in dairy products, fermented milks, and showed you could engineer it to eavesdrop on cholera by taking the quorum sensing system out of cholera and repurposing it into l-lactics.

And that you could then harness the natural lactic acid-producing capability of lactis to treat or inhibit cholera infections. It's incredible that you were tackling from the very beginning both the diagnostic and the therapeutic aspects. Did that not seem like biting off a lot? No, no, I mean engineers, we probably take on two. much. We, engineers are famous for being able to focus, but do it with lack of focus in the sense that they can take on big problems, and thus lack of focus, but take on those big problems

and try to accomplish it. We thought for sure it would be fine, even though it's really viewed as two distinct fields. And we launched what became Synlogic. And in a matter of small number of years, they now have multiple clinical trials. They're a public company, and they're using engineer E. coli to treat different rare genetic metabolic disorders to treat inflammatory bowel disease and now to treat cancer. What are the engineering principles that you tend to think are the most useful

in this crossover space that are kind of your north stars? Part of is that, like, you know, when one is trained as an engineer, you learn certain principles. Do you think bioengineers can take some of those existing engineering principles and bring them over?

Or are bioengineers inventing new engineering principles? You know, it's an interesting question. I think that in many cases it's more techniques and approaches than it would be principles. I tend to think of principles really being technically grounded

notions that have established a relationship between certain variables of parameters that you come in. So what's interesting is that a major challenge in engineering biology and more broadly, specifically synthetic biology, is that we don't really know the design principles. That's right.

And that we'll put things together and they'll be motivated by electrical engineering but recognizing that obviously we're not making it in electrical or electronic circuit. But in many cases, the circuits don't work. at the level as well as we like or as we predicted. And I think it reveals that we don't have those design principles in that establishing the relationships, for example,

that really got electrical engineers, we're still, I wouldn't say haphazedly put in them together, but it's still very much trial and error, and it leaves great opportunities for getting after design principles. At present, I think, for really the first two decades of synthetic biology, it's really been more engineering approach, and that is this idea that you can design a schematic for your circuit, find the right parts, put them together, and begin now characterizing

both the parts and they're functioning in a circuit, and iteratively improve upon the behavior. The engineering will go after it in a highly parallel fashion. So as he was doing the toggle switch, he would explore two dozen possibilities simultaneously and do it relatively quickly and identify the one or the couple that gave him the closest to what he was going after, and then he would expand around that after a couple weeks. weeks and do this iteratively and get through what could be basically many dozens of hundreds of possibilities. Was that because of just a different approach or was the technology actually enabling him

to do that for the first time? It was really, I think, more of the approach. And the distinction is that the young biologists in my lab will instead pick one design, one embodiment of that design, and they're going to drill down and grate the tail over, instead of two weeks, over six weeks, realize that it didn't work, and then they're going to go and create the next design on the basis of that, drill down one by one. More serial. Serial, and they will have a much greater in-depth understanding of that one case,

but it's going to take them 20 times longer to get to a functioning prototype. And so there is where there still is really a cultural divide on the approach. Would they argue there's something lost when you do it simultaneously, the engineering-led approach? I haven't heard them say that it's lost. outside that they would say that... Like, why not? You know, the area would be interesting is that I do think they feel that

they'd like to get an in-depth understanding. And again, it's now coming up against the culture clashes. The scientist is curious, wants to understand, is looking for discovery. The engineer wants to get it to work. And doesn't necessarily care do they understand how it works, but that it worked. You know, the scientist really wants to understand

and making it work is proof of understanding, but understanding is the real ends and making it work is the means. Whereas from the engineer, making it work is the end. ends. Yes. And understanding is the means. Yes.

And because both care about the ends, the means actually they care less about. Yeah. And so they're optimizing for different things, even though from the surface they may look like they're doing the same thing. And I'll draw an interesting further distinction on level of comfort is that I've been in the CINBio world now for about two decades, and I have yet to get significant buy-in from the engineers in my lab for doing directed evolution. So you would think after you get your prototype that's working, okay.

but not at the level that you like that you would be open to using one of the great techniques Nobel Prize winning techniques. Noble pried by Francis Arnold, who's considered part of our community and she's marvelous and a great pioneer. You would use directed evolution techniques to optimize

yours as taking advantage of this beautiful feature of biologic systems. It is completely foreign and anathema to the engineers in that they view it as cheating, that they want to be able to themselves retweak this system by design to get it to function the way like.

The chemists in the biologists in my lab much more readily are embracing. That's kind of funny. And I think there's now going to be three prongs because, you know, the success of direct evolution over most rational methods for, let's say, protein design, enzyme design, something like that. It's just really hard to do that rationally.

There's a third arm, you have rational design where maybe trying to use physics and quantum mechanics, the directed evolution where you just sort of optimize for some function. Machine learning is an interesting way to sort of have it both ways. and that from all the data you get from directed evolution, could you use that to actually come up with a predictor or readout? I do think machine learning within synthetic biology is one of the next frontiers.

And the challenge there is a couple levels. One is can we have enough data to feed the models? And by there the data is can we generate enough diversity and can we functionally characterize that diversity to feed the algorithm? I think the marriage of direct evolution with synthetic biology really is also the future. to machine learning. Synthetic biology allows you to cheat in that it can move you in that genotype phenotype space into an area that you can't get to readily through evolution. Do you break that down? If I'm going to tweak a system, whether it's naturally evolving

or I'm going to use directed evolution to evolve it, you're really going to move incrementally. Now, at fairnish, maybe you get a five-fold or even a hundredfold or two-hundredfold after many, many rounds, as Francis showed, for enzymatic activity. But in cases still that you're going to be kind of starting in some place, say if you've got on your X-axis genotype and your Y-axis phenotype, you're stuck over in this corner. Directed evolution will allow you to move from that corner, but you're going to stay in the general space. You're going to kind of randomly move around. Synthetic biology allows you to take your starting point far away from that corner. Or to hop.

Or to hop over into another spot and then fluctuate around using direct evolution. And so it's the marriage of the two that allows you to explore spaces that would, take hundreds of years, if not millions or billions, or what really has happened on the planet. Well, the interesting, you've seen... It sounds like wormholes to another side. You've seen evolutionary approaches apply to electronic circuits, right? Where they make a radio and it's just like some bizarre circuit that works. But it's hard for human beings to understand.

It's funny with biologists and engineers are two different cultures. The machine learning is like a third one still, possibly. Do you think that science as a whole will be moving in this direction going forward more with this blend of engineering approaches? There's been a blending between science engineering, especially in any of these interdisciplinary activities. I have Stanford colleagues that were trained as engineers but are writing science papers and doing science.

And I'm trained as a physicist, but I've moved into engineering. And I think when you're at the interdisciplinary sort of interface, a lot of people do whatever it takes. And so that spirit is there. Now, there will be people who stay in their domains that just love biology or just doing biology, and that's all they want to do. And that's also fine. But I think the rise of the interface

has really sort of led to this new area. I do see more and more of the interface between quantitative approaches and science, not per se engineering, but quantitative advances of really getting after data analytics. And embracing this notion that machine learning and more broadly,

computer-based modeling and data analysis, is becoming dominant as people are trying to figure out how to embrace our increased ability to collect a lot of data and analyze data. Yeah, so one thing that I was talking internally and almost as an investing theme is, you know, where is synthetic biology on the hype cycle? You know, my read on is it's gone past peak hype and is sort of in the, now getting

into production, getting too useful. Synthetic biology has been hampered by basically too much hype from its very beginning. And I think there's multiple reasons for that. One is that it's a very romantic, exciting field, this idea that you can engineer life to address some of the world's biggest problems is exciting. And second is that I think we've had an unchecked field. that is not yet an established discipline. And so you have fringe elements that will grab their 15 minutes of fame to put out stuff that is frankly not realistic.

Sounds right, likely will work, but could take centuries to accomplish. And we've been hurt by that. Now, if you go back to the mid-2000s, our field took a strong left turn toward bioenergy. And there were many factors driving it, starting around 2004, where you had gas prices going north of Fort dollars

at the first time. You had then President Bush calling for alternatives to petroleum. You had large oil companies looking for alternatives. You had venture capitalists seeing, boy, there was a real opportunity here. And they heard about this exciting new space inthetic biology where people were engineering bacteria and fungi to invert different food stuffs and sunlight into fuel. Sounds like a panacea.

Sounds brilliant. And they poured hundreds of millions of dollars into it. You had Department of Energy and other national finances, similarly pouring huge monies in. And it was too early. too early in that you had brilliant academics showing that you could do this in small bioreactors in their lab at university. But to think

you could scale this up to the million-liter bioreactor was absurd and that you had loss of efficiencies you just couldn't compete economically. And so many of these early companies got to be pretty good at producing $1 worth of gasoline, but it cost them $8 to do it, which

was a horrible business model. Unless gas went up 10x. Unless gas went up 10x. Yeah. And so, you Now, that part of the field collapsed on itself on the hype. What happened then in the late 2000s when the bioenergy focus hit Earth came back down is that the field got grounded and really began building out more broadly tools, constructs, platforms, and turned toward biomedicine. I think the field got really re-grounded and saw that there was great potential.

And alongside that, you had other developments in biotech that were complementary to synthetic biology, broadly engineered cell therapy outside of Sinba, like Carty therapy, and you then had the biotech discovery or development, I should say, of the century, which is on CRISPR, enabling very rapid, easy, inexpensive genome editing

that together now opened up possibilities that just weren't going to be readily apparent to us 10 years ago. I mean, other directions that go for molecules that are just worth more than oil. I mean, a barrel of perfume is a lot more expensive than a barrel of oil. Yeah, and so then the question is, can you gain a lot by engineering a living cell to produce it versus using a chemist

in a regular petri dish? And that's where I think the jury's still out on many of these companies. It seems like they can make it. Now the question is, is this our business model there? Yeah. I'd love to know about company building in this space, what that looks like. How is it actually different building a company in this space versus traditional bio? Yeah, and there's like two models.

There's how to build a traditional bio company and how to build like an engineering-driven tech company. And this seems distinct from both. You have to make a decision. And that is, are you a platform company or are you a product company? And what happens is that the companies in this space typically are being launched by the academic. As the academic comes in thinking they can do everything. And so they want it to be a platform company that's feeding a product company.

And when you then talk to the real company builders, they'll say, no, we've got to make commitment. And I think they're hybrids. I mean, you can have these multi-product platform companies. Let's break it down what you mean. specifically mean by that? So by that I mean that on a platform company would be that I'm going to develop internal capabilities that will allow me, for

example, to engineer bacteria to sense different things and produce molecules that would have an impact. To make many products. To make many, many products. And it might be that I'll even focus on therapeutics, but that I could have the capability to engineer E. coli that would allow me to address 20 different diseases.

And that my model then might be I'm going to team up and partner with farmer and large biotech. The product company would be, no, know what I'm going to do. I'm going to engineer bacteria to go after your recycled disorders. And I'm going to launch a clinical trial, and that's my product. I'm going to move it toward clinic. I could build and maybe have one or two other internal programs. But it's a struggle. In what way are you going to play it in this space? And you see companies that are playing both ways. Synlogic is really a product. It's a therapeutic

company that is focusing on developing therapeutic products that will go into patients and that they themselves maybe through partnerships to market them but develop them. Ginko, Biowworks for which I'm not involved, is more of a platform company

developing capabilities to engineer organisms to optimize enzymes and through a partnership model could have a platform. The platform companies in this place are playing to the technology investors

and the product companies are playing more toward the traditional biotech investors. I think that's starting to change in that I think there's been a cycle so even the biotech investors are curious about platforms. I'm encouraged to hear that,

It goes booms and bust. So when the NASDAQ collapsed 17 years ago in 2001, the local biotechs ran from platforms. Prior to that, from 98 to 2001, it was platform happy. And then they became fixated on products to a fault in that many of the companies would rush a molecule into trials to say they had a product. And most of those molecules were horrible and they ended up failing.

The other thing that happens is that it's very tempting once you've built a platform and you actually have an asset that is, let's say, past phase one or phase two, that becomes worth so much. It's a hard financial decision to take money away from that and put it to other things. And then you become a platform company

becomes a single asset company. And I think that the model is that if you can move your assets, develop assets, move them at trials, you then also have that struggle, do you now flesh out your platform to, and I see that in companies I advise and I'm involved with,

and I'm a big believer that you should also have the long view, that it's worth building that internal capability, whether it's around design or whether it's around manufacturing. So, for example, having the ability to manufacture your product in-house significantly increases your value. What is the go-to-market like in this space when you're building these synthetic bio products or platforms?

What is the way you see this kind of entering the system? How is that process happening? I think the challenge is really making a commitment as to what is your go-to-market. Because in many cases, these early companies, particularly coming out academic, are aspirational in that you want to take on all these problems. And yet, if you're really then going to get down,

it's very hard to go to market and what's your choice. And then it really becomes specific. So in one area I'm involved with is diagnostics, is it through a partnership model, or are you really going to develop a product, get it to approval, and now introduce in the market? Well, that poses all fascinating new challenges of who's going to manufacture, how you're going to distribute,

what are your distribution? And in many cases, I think in synthetic biology, it's still at a stage where either they're following traditional paths and the traditional paths in therapeutic space and even the diagnostics is through partnerships. And so it's at what stage. But then it's the challenge is at what point you partner. You partner too early, you're not going to get as much value for your investors.

And so can you hold it off? Can you maybe partner off maybe less essential, less central? Well, and part of it is that, you know, synthetic biology has the hope to do so many different things. And almost now you have the challenge of figuring out of all the things you could do. What should you do? It is. that's a hard choice to make because it's really opportunity costs on the negative that will

drive many of these companies bad, and it's really what makes or breaks companies. And separate from the technology, it really comes down to the team that can make those decisions and that has the flexibility to change up depending on markets that change or opportunities that come or go. I wanted to think about how this field is now sort of percolating down into education, and I was really surprised to read that it's percolating quite far down into these, you call them bio bits, these kits, these classroom kits. Can you tell us a little bit about those and what kinds of kids are using them, what they're learning?

Yeah, so I'm 53 years old. The best Christmas present I ever got was when I was in third grade from my dad's mom, and it was one of the old classic chemistry kits, big yellow tin that had these powdered chemicals and I did all these different experiments and they had fantastic combustibles

and explosive. I don't think you can buy those anymore. You cannot buy them. I have hands on it. You can't blow yourself up, basically. You can't blow yourself up or poison things. I said, things on fire. But you have these marvelous hands-on activities for chemistry. You have marvelous hands-on activity for electronics, robotics, mechanical-idental erective kit.

But you don't have anything for biotech. And this motivated us could we develop synthetic biology tools. So these BioBits kits are new synthetic biology educational kits that my lab have co-developed with Mike Jewett from Northwestern. So with

Ali Huang, Peter Nigh Gunn, and Jess Stark, grad students and postdocs in our respective labs, we developed these kits that are basically motivated by chemistry kits, but for synthetic biology. Mom's not going to be happy if Sally's growing her brother's stem cells in the kitchen. And what we used were

cell-free extracts, so you can open up a living cell, remove the machinery of the cell. And typically, you'd play with these in a test tube or petri-deers, but we showed you could freeze-dry them in pellet form. And so now you have shelf stable, room-temperature stable, pellets that you then add

to water and in now your test tube at your home. You have the machinery of a cell that's outside of living cell. Oh my gosh, it's like those owl pellets that you can bring home from preschools, but for cells. But it's for cells. They now allow you to play with how proteins are made,

how circuits are made, and we created these biobit kits that already go down to the middle school level. Really with the idea to introduce kids at an early age to the ability to engineer biology, given that I think in many ways this century is going to be the century of biology

and really put a point on it, I think it's going to be the century of synthetic biology. And we need to get the use, youth excited about the potential of this field to really change the world. So what do you think changes when you start introducing that framework and that mindset so early? What do you think in kind of biology, design thinking fundamentally changes? Well, I think you expose the young people, so they begin to see the world differently.

So if you're a designer, you're going to see a problem and think about how you can design it. Just the way your chemistry set did. Well, you think about some of the greatest challenges that we face right now, world hunger, global warming, these types of things are inherently interlaced between chemistry and biology? Yes. And if you want to have an impact on these things, you'll have to be able to engineer biology to some extent.

Yes. I think that synthetic biology will provide solutions that are going to contribute to solving broadly environmental challenges. You're increasingly seeing companies turn towards synthetic circuits to alter traits in a programmable fashion of plants and to enable them to be more robust to pathogens. I think climate is really just beginning to excite the community and broadly environmental issues, on two counts.

One is, in what way could we use algae and other organisms to pull more carbon out of the atmosphere? Can we help reverse the trend on using biology? And second, can we address issues such as the breakdown of coral reefs? So could we create synthetic ecosystems, synthetic microbiomes for coral reefs to make them heat-tolerant and to prevent the bleaching of schemes?

So I think those two areas are ones that I see in the next 10 years, we're going to see a good amount of activity. And we need more talent in these spaces. We've got to reach out and make them comfortable with these technologies and introduce from the concepts, the eventual design principles. And that's just the next 10 years. I mean, we talk about the biosensurate. The funny thing is, I think, people tend to overestimate

what people can do in a few years. That's what generally feel like and underestimate what you can do in 10 years. I don't even think we can conceptualize what 50 years is going to look like. Thank you so much for joining us on the A16Z podcast. Thanks for having me. Thank you.