Transcript
Deniz Kent: So that's really been the unlock, is creating a language that both the machines understand and the biology understands. And if you have this common language, then you can start teaching machines how to control biology.
Daniel Darling: Welcome to the 5 Year Frontier podcast, a preview of the future through the eyes of the innovators shaping our, world. Through short inside packed discussions, I seek to bring you a glimpse of what a key industry could look like five years out. I'm your host, Daniel Darling, a venture capitalist at Focal, where I spend my days with founders at the very start of their journey to transform an industry. The best have a distinct vision of what's to come, a guiding North Star they're building towards, and that's what I'm here to share with you.
Today's episode is about the future of biology. In it, we cover machines controlling cells, manipulating biology with light algorithms for drug development, and how biotech is becoming a low cost mass production industry. Guiding us will be Deniz Kent, CEO of Prolific Machines. Prolific is the first biotech company to harness light as a more efficient way to produce lab grown food, life saving drugs and novel bio solutions. Offering a full stack toolkit from bioreactor to AI software, Prolific's groundbreaking technology provides dynamic control over virtually any cell function in any cell type. Until now, biomanufacturing has been limited to indirect cell control by expensive, inefficient and imprecise tools like chemicals. Prolific's platform enables direct control using light to produce new and superior biosolutions, faster, cheaper and at greater scale.
Based in Silicon Valley, the four year old company has raised over $80 million in venture capital from the likes of Mayfield, Breakthrough Energy Ventures, In-Q-Tel and Fonterra. Prolific's co founder and CEO is Dr. Deniz Kent, an expert in biological systems who earned his Ph.D. in the center for Stem Cells and Regenerative Medicine at AH King's College, London. During his studies, he discovered a new human liver stem cell, worked on cures for asthma at GlaxoSmithKline, and developed research in the field of cancer immunotherapy. Deniz nice to see you. Thanks for coming on the show.
Deniz Kent: Thanks for having me.
Daniel Darling: So we're here to chat about what I view as one of the defining topics of our generation, which is how humans are gaining control over cells and our biology, or essentially biotech innovations that are enabling us to play with the building blocks of life. You're on the inside of this industry. How fast is this future approaching and how quickly is it accelerating?
Deniz Kent: Very fast and very quickly. So it's been incredible to watch even just in the last five years of doing prolific, how far the field has come. I think we're really on the cusp of biology being able to do a lot more things than it has historically been able to do.
Daniel Darling: And that's exciting, super exciting. And we're talking about manipulating cells. So how has that been done to date? And tell us about this quest of yours to transform our capabilities using light.
Deniz Kent: Traditionally, cells have been controlled with molecules. So you have these cocktails of proteins and chemicals, and you mix them up and you add them to the cells. And different proteins and chemicals do different things. And this has worked reasonably well. But there are many downsides of using molecules to control cells. They can be very expensive. They're very hard to control in terms of where they go. They're very hard to control in terms of when they go where they go. They can introduce sterility issues. There can be batch to batch variation, a whole plethora of issues. And, at, Prolific, we're building a new toolkit to control cells with light, because light can systematically address all of those issues. So it's the cheapest possible input into biology. It has very tight control on the time axis, very tight control on the space axis. It's extremely reproducible. It's inherently sterile. So there's a bunch of reasons why I believe, and a growing number of people believe, that light is the best way to control biology. And there's many different things that you can do with that. So you can make pharmaceuticals, you can make functional food proteins, you can make meat and fish. Eventually you can make tissues for transplantation.
Daniel Darling: Yeah, and I'd love to unpack some of those use cases, but tell us a little bit about how you're using this light.
Deniz Kent: So what we build is genetic tools that can do various different things. at the heart of it is always, light sensitive proteins. And these light sensitive proteins change shape when they get hit with light. And we leverage this shape change to do a variety of different things. And you can use this shape change to activate receptors on the surface of cells. You can use this shape change to turn one type of cell into another type of cell. You can use the shape change to control transcription and, produce something. And the way you use the shape change to control these things is by tethering these light sensitive proteins to targets. Everything from UV to far red light can control pretty much every function inside of the cell.
Daniel Darling: And so basically, science has discovered what are the right colors and what are the light spectrums that cells respond to in very specific ways. And you're enabling them to issue that out to the cells and issue those instructions, if you will, to influence that change?
Deniz Kent: I wouldn't even say science has discovered, it's like evolution discovered this stuff. So microbes evolved these light sensitive proteins because they were useful for them. You imagine you're an algae and you're in the ocean and you need to be able to regulate your buoyancy because if you go too far up in the water you get fried. And if you go too far down in the water, you don't have the energy that you need. And so you need a way of being able to gauge where you are. And so, microbes have been using these licensitive proteins for way longer than humans have existed. And they evolved as basically like sensory mechanisms for microbes.
And so what Prolific does is it like takes this sensory equipment out of the microbes and puts it into mammalian cells which don't normally have these proteins. And then there's a component which is, you have software, so you build these closed loop control systems where you can have an AI agent that will basically control this process for you. Because you don't know what the right light patterns are a priori. So you have to go through a phase of trying to figure out, okay, you've engineered the cell and you have the hardware to illuminate it. But how do you do that? Do you just leave the light on the whole time? Do you do a little burst of light? If you do a burst of light, how big is that burst? Like how intense is the light? How long do you leave it on for? Cells can tell the difference and different patterns do different things. But the ultimate goal is to train machines to be able to control these systems and be able to experiment with the light patterns and dynamically adjust the light patterns in real time. So we don't really have technology like that right now.
We have these proteins and chemicals, we mix them in, we add them into the bioreactor and then we kind of just hope for the best. But this would be a paradigm shift by process control because it would allow you to read and react dynamically to what's happening inside of your reactor. And ah, so you can say, you can pre program it to say, okay, I want you to optimize for getting this tighter or I want you to optimize for this particular glycosylation pattern or whatever it is that you want to optimize for. You can have sensors in the reactor that will measure those things. And then you can build an algorithm that will start experimenting with the light patterns in order to see how best to get to that thing. And if it sees it going in a different direction, it can also auto adjust to go back towards the thing that you want. We haven't really had anything like this before. And I think that it will have some profound implications on what biology can do.
Daniel Darling: That's a really fascinating train of thought. So essentially you have within your bioreactor all of these sensors that are monitoring your light emission and the impact onto the cell and then essentially informing the discovery process around what light needs to have what impact at what dose, et cetera. Like that. And you're not doing that in a manual process, you're passing that into the software process from there. Is that a real big unlock that enables this to happen? And does that then produce the different types of repeatable processes and essentially instructions to do it again and again?
Deniz Kent: Yeah, yeah. I mean, one of the challenges with biology is that it's inherently variable. You have a single cell line and you could split it into two reactors and do the exact same process and you don't always get the same result, which is, which is crazy to non biologists, but if you speak to biologists, this stuff happens quite frequently. And so that's all the more reason to be able to have systems that can adapt dynamically because of this inherent variability in biology. We want to be able to both understand that variability and measure it, but then also be able to react to it. So regardless of what sort of intrinsic variability exists, we want to have the product be consistent because otherwise the consumers don't want it. Quite rightly so.
It's like you want to make sure that the burger that you're buying today, you want it to be the same burger, that you're buying from the supermarket the week after. You don't want it to be a different burger, you expect it to be the same burger. Even more so for an antibody that's meant to save your life, or a vaccine that is meant, to prevent.
Daniel Darling: Something, what has been the technology on the software side or on the AI side that has suddenly enabled you to do this?
Deniz Kent: The technology that's enabled this is really more on the genetics side because we haven't had a universal language between machines and biology before. That's really what's been lacking. Because molecules are not something that machines can really understand or produce. So machines that can control biology today can add, liquids, they remove liquids, they can move cells from A to B. Some of them can even image cells. But we don't have any machines that can directly interact with cells today.
The reason why we haven't is because we haven't had an input that machines can control and understand and produce. And light is that input because light is just electrons going through a circuit board into an led. And so machines can do that. Machines can't just produce like FGF2 or TGF beta for example. You have a, we have a whole supply chain to make these things and they normally get produced in microbes and then you have to be purified and cold chain transported and et cetera, et cetera. This is the whole thing. But with light you can just be hooked up to the grid and use electricity to make the light. So machines can understand that. But importantly, if you engineer the cells to be light sensitive, then cells can also respond to that. Now you have an input that both machines can understand and the cells can understand because of these light sensitive proteins. And so that's really been the unlock, is creating a language that both the machines understand and the biology understands. And if you have this common language, then you can start teaching machines how to control biology.
Daniel Darling: That is an incredible, incredible thought, essentially enabling machines to communicate and interface and control biology. So fast forward that what does that start to look like in your eyes? What does biology start to look like in terms of what's happening inside of the labs?
Deniz Kent: You think about like what biology is doing well today it's really restricted to the very expensive things. So like antibodies have been a, big success story. Vaccines have been a big success story. Anytime biology has tried to make something that is cheap, AKA the commodities, we've run into trouble. Like biofuels didn't grow super well as they cultured meat hasn't gone super well. Precision fermentation hasn't gone super well. In my opinion, the reason for that is because the tools that we're using were never designed to make commodities. They were designed to make super high value products like antibodies. And so I think what you're going to see is with a better toolkit, the product categories that can be unlocked with biology will expand. It will no longer be just the super expensive therapeutic proteins, it will be fuels and food and commodities.
Daniel Darling: Basically fast forwarding five years and you're on this trajectory here and this starts to become more mainstream. What are the kind of intractable challenges in biotech that this could start to help address?
Deniz Kent: I think I discussed a couple of them. I think one is making cultured meat viable economically. I think another is taking these difficult to express or difficult to produce proteins and making them producible. I think another is democratizing access to biologics. For example, we just recently announced a grant that we received from the Gates foundation and they are very interested in antibodies for Africa for malaria and other diseases. And like the African nations just can't afford these antibodies. And they can't afford them because the production costs are too high. And so there's a lot of opportunity and in having better production systems that can be cheaper, that can then dramatically expand the number of people that can access these.
Daniel Darling: You know, I want to go back to a really fascinating part of your business, which is this machine biology interface that you're creating. And if you think about the amount of discovery acceleration that artificial intelligence is doing with the likes of AlphaFold, et cetera, really advancing the industry forward, how do you foresee the discovery process being impacted by artificial intelligence in the years ahead? Do you think more will be done in terms of this software during the discovery process versus the scientists, given that it now has an ability to interface and learn from the biology in almost real time using your technology?
Deniz Kent: The system that we have where you know, 90% of drugs fail, this is clearly fundamentally flawed, right? It's going to be addressed because there's just too much money on the table to not address it. You'll really see is that the failures can happen a lot sooner. Right? So currently a lot of the failures only happen once you go into trials. And that is a function of the system that we have to test these drugs before they get to clinical trials is not super great. We either use monolayers of cells or we use animals like rats or mice. People will build things like what Prolific is doing that will allow you to far more quickly make the decision that this is not going to work or is going to work. And I think there will be a large AI component of making those things. Like for example making a higher fidelity representation of a, disease. You don't need to fully understand the cells in order to do interesting things with them M and useful things with them.
So this is the real power of AI in my opinion, is like systems that you don't fully understand. A cell can still be optimized to get you what you want so you can have the system you don't understand, but you can start testing these different light pat and let's say you want to be able to produce this drug or you want to be able to produce this food and you can't normally produce it. Whether you understand the system or not, it's kind of irrelevant because you may be experimenting with these light patterns. And, some light patterns could be working really well and others not very well. And you could have absolutely no idea why that is the case. But that doesn't actually matter. What matters is can you produce the thing that you want it to produce? I think we'll be able to utilize biology to do a lot of useful things before we can truly understand it. We can understand it enough to do some useful things, but maybe not enough to build a full simulation of it.
Daniel Darling: Yeah, that makes a ton of sense. Essentially what you're talking about is now we have a machine and algorithm that can test a multitude of, different types of options against a cell and then monitor that and optimize for the end outcome of that cell. But we don't actually have to really understand what is happening within that cell to drive that outcome. But humor me for a moment because we're talking about the future. What would happen if we did have a full understanding and kind of digital representation of a cell? How fast could scientific discovery start to occur? If you fast forward to that future.
Deniz Kent: The costs of doing science would plummet. Right, because Prolific, one of Prolific's biggest costs is that we actually we run a lot of experiments in the real world. We have a ton of reactors ranging from, you know, very small 15 milliliter amber systems to big stainless steel tanks, all of which are expensive to run. And so we're learning in the real world. But if we could just like not have to run any of those, and we just run like a billion virtual bioreactors to figure out what the best light patterns are, obviously the impact of that is profound. I'm pro-digitizing the cells. I think another area that you'll see a lot of work and it's already happening is in like synthetic cells. So you can actually build cells from scratch. And that is another thing that's going to take a really long time, but that may completely change biology.
Daniel Darling: Amazing. Well, look, there's so much innovation coming our way. And thank you for introducing, one that sounds incredibly impactful to biotech and into multiple different types of industries. So appreciate you coming on and sharing it with us today.
Deniz Kent: Thanks for having me.
Daniel Darling: I really enjoyed learning about a frontier area of bioscience from Deniz It was an aha moment for me when he spoke of establishing a machine biology interface, a common language for machines to control biology. This is a huge concept that could accelerate scientific discovery and biological production dramatically, enabling the sharp curves of innovation occurring in both biotech and AI to converge. It's clear that this could democratize previously high cost drugs as well as power biotech to develop more everyday goods from food to fuel using biology to follow Deniz and the remarkable work being done by his team. Head over to their account on X, @ProlificMachine. I hope you enjoyed today's episode and please subscribe to the podcast to listen to more coming down the pipe. Until next time, thanks for listening and have a great rest of your day.
