Proteins in Organic Solvents: Engineering Enzymes for Sustainable Chemicals Manufacturing with Samuel Thompson

Join us as we explore the innovative world of protein engineering with Samuel Thompson. Samuel's work focuses on engineering proteins to function in organic solvents, environments that would be hostile to traditional cell-based life. This approach has significant implications for bridging the gap between the enzymes market and the trillion-dollar specialty chemicals market, potentially leading to decentralized chemical production with a much lower environmental footprint.

In this episode, Samuel shares their personal journey, from growing up in West Texas to their current role as a postdoc at Stanford and the University of Washington. They discuss how their queerness informs their science and the long-term vision they have for their work — a commitment to solving complex problems often overlooked by mainstream science. With support from the Homeworld Collective, Samuel is pushing the boundaries of what is possible in protein engineering, aiming to create sustainable solutions for chemical production that could transform industries in the decades to come.

Chapters

• 0:00 Frontier Protein Engineering in Climate Biotech
• 3:41 Exploring Queer Protein Engineering
• 7:47 Long-Term Vision in Protein Engineering
• 12:52 Protein Engineering for Non-Biological Applications
• 23:11 Protein Engineering in Organic Solvents
• 29:47 Protein Engineering in Organic Solvents
• 37:59 Innovations in Protein Engineering
• 45:15 Biotechnologists Embrace Collaboration and Simplicity
• 50:51 Gratitude and Vision in Biotech

Transcript

[00:00:00] Samuel Thompson: And when it comes to the structure of chemistry and the structure of biology and how we see molecular structures, biologists have been eating the crumbs of physicists for years, right? Electron microscopes, synchrotrons, those are all things that physicists developed for themselves.

I think there's a space for biologists to step up and say, Hey, this is what we need the physicists to make for us.

[00:00:22] Dan Goodwin: Welcome to the climate biotech podcast, where we explore the most important problems at the intersection of climate and biology, and most importantly, how we can solve them. I'm Dan Goodwin, a technologist who spent years transitioning from software and neuroscience to a career in climate biotechnology.

As your host, I will interview our sector's most creative voices from scientists and entrepreneurs to policymakers and investors. Our guest is Samuel Thompson, a postdoc at Stanford University, working on one of the coolest corners of protein engineering.

Normally, when we talk about protein engineering, we don't realize it, but we're talking about folding proteins in water. Our bodies are 70 percent water molecules are bouncing around making proteins fold. So we never really think about what is those molecules around the protein. But if you want to start thinking about proteins as a nanoscale, infinitely programmable machine, you have to start thinking about making proteins work in really weird environments, so it instantly dissolve a cell if you drop it in there.

But what would it take if the protein could actually function in something like an octanol or something like an organic solvent necessary to work in a chemical manufacturing process. What's really beautiful about Samuel's work is he's always done things his own way.

And so what you're going to learn in this podcast is about the philosophy of how you start building a field where you don't have the same amount of tools. Like very mature areas, like working with E. coli or other directed evolution techniques, or use something with AlphaFold. The stuff that Samuel is doing is so far into the unknown that he's had to build a lot of these tools himself.

Homeworld is really proud to be one of Samuel's supporters through our garden grants program. And through that, we've got to know Samuel both in and out of the lab. Their work is far on the frontier of cutting edge protein engineering, which has major implications for long term sustainability.

Today in this conversation, we'll explore the content and the motivation of Samuel's work to engineer proteins in weird environments. How weird? Think about environments that would instantly dissolve any cell based life. And why is this important? Works like Samuel's can bridge this world between, the quote unquote enzymes market and the specialty chemicals market.

Enzymes today are a billion dollar market, but chemicals are a trillion dollar market. So this thousand fold difference could mean massive implications for decentralized chemical production, much lower environmental footprint, and a difference of kind efficiency, which lets us think about, chemical plants that fit in our hand.

 Samuel Thompson is a postdoc with a joint position between Polly Fordyce's lab at Stanford and David Baker at the University of Washington, where they develop microfluidic and computational methods to engineer proteins that fold and function exclusively in organic solvents. 

Samuel, we're so excited to have you here today. So thank you. We'd love to just introduce you to the audience. And so who are you, where did you grow up?

[00:03:19] Samuel Thompson: Oh man, who am I? I am nobody, let me tell you that. If you're listening to this podcast and you haven't heard of me at all and you have no idea who I am, that is the correct answer. Especially looking at the lineup of some other people who have been on the podcast. So yes, I'm very proud to be working in this very, I'm going to use this word intentionally, this very queer space of protein engineering and protein science.

And before I get into who I am, I was reflecting a little bit on this question as we were preparing for the podcast. I was thinking that there's a couple of themes, in both my work and my life that I think will come across as I tell my story and the scientific themes are.

Our self assembly. I love machines that build themselves. And I think that's an essential part of life and biotechnology and what makes biotechnology different from so many other human created technologies. And then in particular, I'm really interested in separation and templated synthesis and how templated synthesis allows us to build things and allows biology to build things that are not possible with a lot of the synthesis techniques that we have.

And then in my personal life, the things that have led me to this work are my own aversion to doing the norm A lot of things that I would call slow, but maybe hopefully are more kind of early exponential phases. I tend to get into things at that period, where it seems slow, but hopefully it's building to something new.

And hopefully you'll see a little bit of a teaser of that exponentiality. And then queerness. I do a lot of the things that I do both historically and in terms of my interests because I am queer. And I was thinking about that in part because when I was in grad school, I had the opportunity to interview Jay Keasling on a panel, and this is not to call out Jay Keasling, he's a wonderful person.

But I asked the question of, like, how does his identity inform his science? And he was like, oh, not at all, those things are totally separate. And I was really disappointed not because, it's a fault of his, but that's because what, that's what I wanted, because my queerness informs my science so much.

I grew up in West Texas. I grew up in a really conservative family, so conservative that I wasn't allowed to listen to music with words in it. And, my compromise for that was that I started listening to Japanese pop music and that will come up. But one of the other things was, it really inspired me to leave.

And when I was in fifth grade, my sister got these promotional materials for early college programs. These programs let you do the last two years of high school and your first two years of college at the same time. So when I was in fifth grade, I was like, I'm going to do that. That was one of my first early exponentials of planning that out.

And so by the time I got to my sophomore year in high school, I had applied to one of these positions. I only applied to it because my parents were convinced that I wasn't going to get accepted. So I got into This wonderful math and science magnet program called the Texas Academy of Mathematics and Science.

And that's where I started doing research because it was all about tracking people into science careers and into, big name institutions. And so they had programs to get people involved in research. They had clubs where they had RAU like summer fellowships. And I started working on synthesizing polymer micro gels.

I had no idea what I was doing. I just, went to the first lab that accepted me and they accepted me because someone else did. But that was where I fell in love with building molecular machines was this idea of Oh, polymer hydrogels are super important. You can do all kinds of, materials with them.

But the downside I realized in all of the processes that we were doing, because we were trying to make these swimming particles, things that could go along a vector path and do something. And it never really worked, but it really did make me fall in love with it. The problem on the material side, and I realized that the best material for doing those kinds of functional things was proteins because you have the ability to code information into the sequence in the way that you don't with a lot of our random polymer synthesis where you end up with, very Poly dispersed, heterogeneous compounds and materials 

you have this very reliable, very reproducible technology, and that's what evolution builds on because you have that reliability and reproducibility and that encoding of information into the templated synthesis process, both for gene and proteins. When I started undergraduate, I knew that was the direction that I wanted to go into in protein engineering.

I started working for Alice Ting, where we were engineering these ligases that would attach fluorophores onto cells and protein so you could track where they were going. But what I learned from that was that I ultimately wanted to go into a computational field just because we were doing rational immunogenesis and that space, that engineering space is so large.

So you either need to be able to do it in high throughput or you need to be able to do it computationally so that you can have strong predictions. I took a brief stint between that research and in grad school. I worked in two structural biology labs. One, Kathy Drennan's lab where I was working on carbon dioxide fixing enzymes, and then when I took a gap year in Japan I decided that I was going to do this program for sending people abroad to do research and I had been minoring in Japanese at the time and it seemed really useful.

And so I went on mental health leave because that was the only way to get out of my registration at MIT without deregistering as a student. So I took a year of mental health leave to go work in a lab in Japan. And maybe I am crazy. But while I was there, one of the things that came up, it was a wonderful experience in so many ways, formative in so many ways, in terms of my understanding of the world, but also my understanding of science, because one of the things that we did while we were discussing science, professors were talking about the kinds of protein work that people were doing that was really famous for a time and then went away.

for, the oldest professors, it was crystallography and they were talking about how everyone was going to the butcher shops and trying to get, Cow's blood and pig's blood to purify some protease so they could crystallize it. They just needed enough protein to crystallize.

And so that's what everyone was doing when that professor was in grad school. And then one of the other professors who did grad school in the eighties and the nineties was saying, oh, when I was in grad school, everyone was trying to put enzymes in organic solvents. 

They just stopped doing it. And I was like, Ugh. Oh my gosh, that's the problem that I'm going to solve and literally like in that conversation that was, this is back in, in 2011, I decided that was what I was going to do. So literally everything from my grad school applications to my NSF GRP application, like all on through postdoctoral, etc.

I've always said, my long term goal is to understand the biophysics. of engineering proteins for organic solvent. So even though I haven't been working on this is something I've been thinking about for a long time. This is my, my early exponential phase of this project. 

[00:09:17] Dan Goodwin: And to just to jump in there, Samuel, to play this back, you picked your life direction and research in 2011 in Japan

at a random talk he just happened to go to.

[00:09:27] Samuel Thompson: No, it wasn't even random talk. It was a conversation. We were just sitting around the lab shooting the wind. And I think there's a value. We live in a very fast paced world, especially in Silicon Valley. I'm here at Stanford.

everything moves very fast. I think everyone wants things very fast, but I think that there's a space for people digging into a problem and really meditating on it and thinking about it. And I say that selfishly because this is what I'm trying to do. it takes a lot of humility, Believe that you're investing in something that is very long term and that maybe no one really cares about or that people have given up on and to decide that you're going to commit a really significant period of your life to doing this. 

ultimately what I wanted to do is I wanted to commit to something big enough and grand enough that I thought would be impactful, 20, 50 years later, if not right now. I think that I'm still working on getting to that fast exponential part of this project.

But it is why I'm grateful to be in this position where the Homeworld Collective and people like David and Polly are supporting me on this.

[00:10:26] Dan Goodwin: It's so good. And I want to just underscore your phrase, identity driven science. behind what you do, because there's so much soul that we have observed in what you do and why and how you do it. And so I've never heard that phrase before. I absolutely it definitely hits a chord. Andit also just hits a personal note too with me, because I think what took me to do my PhD was this idea of engineering a heroic archetype.

And, what would you do in 10 years where you have a big story to tell and that's also feels very spiritually aligned with the way I'm hearing you set up your work that then ultimately took you to Stanford where you are today.

[00:11:00] Samuel Thompson: It did. So that's a great segue. First came UCSF where I didn't think anyone was going to let me work on this immediately. And so I felt like I was having to pull together all of the skills that I needed to convince someone that I was positioned to do this. 

[00:11:15] Dan Goodwin: Because there's probably very little money for this, right? People would say, oh, this was done decades ago. It's kind of niche, right? You're doing something that is not immediately mainstream. Was that part of the challenge?

[00:11:26] Samuel Thompson: I think it's both. I'm sure there's probably some money for synthesis out there, but also it's more a matter of credibility, right? There's a couple of ways of getting credibility. Either you've been successful elsewhere and you're transferring that credibility and whatever money you've made there to the people that you're To another field.

But as a graduate student, especially a first year graduate student, you have no credibility, right? If you come up with a crazy idea, no one's going to believe you. And so it's a matter of amassing both the scientific capital and the political capital in order to do the work that you want to do.

And so for me I got very lucky in that. I interviewed at a lot of labs doing computational protein design, had great conversations with Tanya when I was starting my PhD. I was like, okay, that's definitely the kind of lab that I'm looking for.

And so I worked with her pairing computational protein design, with high throughput screening in this case a cell based high throughput screening. The publication that you'll see from that may not reveal all of the work that I did in computational protein design, just because the cell based screening ended up being more complicated than we thought.

And so a lot of my PhD was some of the lessons that we learned about how to actually do that high throughput screening and some of the complications of doing high throughput screening in cells. But one of the things that did come out of it was connection with the Rosetta community and David Baker.

So at the end of my PhD in 2019, I was presenting my work to Rosetta Con, and I knew David from having interviewed with him and having been at several conferences, we were talking about the field of computational protein design and where it should go.

And I mentioned that one of the things I thought we needed to push was to push it more towards non biological applications. He said, that's absolutely what we should do. Would you like to work in my lab? And I just said. Yes. And the type of project one would work on.

And I need these skills. I need to be able to do this in vitro. And I need someone who works in microfluidics, because I think that's the technology that's going to scale it up. So when I was applying for my postdoc positions, I was coming to it with a proposal saying, Hey, this is going to be a joint project between you and David Baker.

So there were several different labs that I interviewed with. And ultimately, I chose polyfluoridase at Stanford, largely, I think, on the end. The connection that we had, and I think her enthusiasm for the project, but also her use of very simple microfluidic systems. She has a very complicated microfluidic systems, but the systems that I wanted to use that had really been pioneered in her lab and had been building off of work by Denny Taufik and other people were these fax sortable droplets.

So how do you take a very simple microfluidic system and generate a droplet that you can plug into the system? A commercial fax order. And so the kind of the first there were two go no goes for the project. And one of them was, can we design a protein proof of concept that actually goes into organic solvent and do those low throughput experiments?

And then the other go no go for the project was, can we? Make these triple emulsions where we have a fax sortable droplet. So it's got an aqueous outer layer that's compatible with the fax machine. It's got an oil shell, which is very similar to a cell membrane that fax machines sort all the time.

Then an aqueous core All the reactions take place and inside that aqueous core, I wanted a droplet of octanol so that I could do all of these experiments of getting proteins to actually go into the octanol and do chemistry. So that's recently in advanced materials interfaces where we're actually seeing can we actually make these things and can we do chemistry in them?

[00:14:36] Dan Goodwin: I want to thank you. I absolutely love this and I want to just underscore your idea of the cold start problem of when you want to work on something that's really ambitious, but you're a nobody. It can be really hard. And so it takes that moment of finding a P. I. To believe in you or telling a story that people find really compelling.

And that's also what we tried to solve for in the home world grants, right? When we did the garden grants, we read your problem and your solution. We got really excited. And we try to judge on the content of the idea and the solution, Not caring about seniority, but that's not, that's the edge case right now.

So I really love that you have been cold starting this problem for over a decade now.

[00:15:14] Samuel Thompson: Oh, and 

[00:15:14] Dan Goodwin: 

[00:15:14] Samuel Thompson: I would say I'm also really grateful because, in the project that I've pitched, something that I think is very different about my approach versus some of the other ones that came out of Francis Arnold's lab and other places is I'm saying, Hey, I'm stupid.

Let's start with a toy system. Let's like make the simplest thing possible. Let's not even worry about the chemistry as much right now. Let's just see if we can Make things that go into organic solvents and fold and then we can add on the chemistry later and I think that maybe reflects me coming at this from a materials and a biophysics standpoint as opposed to a chemistry standpoint Where to me like getting the structure and understanding the structure is in many ways enough and interesting And if you're focusing on it from the chemistry application, you want to go straight to the reaction And I think that's really what separates what I'm trying to do versus what's been done In the past.

[00:15:59] Dan Goodwin: and I totally agree with the intuitions there. Because it seems like what you're doing is you're doing one part cutting edge using the latest techniques, but then you're also looking at the lens of what are the highest throughput already working machines that we can combine. And so I think we're going to get there.

I do think it's going to be helpful to the audience just to be rooted a little bit more in the specific problem you're solving. And when I was listening to a recent talk you gave, I think you did a really good job of setting up the We just assume water is everywhere when we talk about proteins. And it immediately made me think of the famous David Foster Wallace speech where he tells a story about two fish coming up and one fish says, how's the water?

And the other fish says, what the hell is water? It makes me think of that's what we've been doing with proteins forever. Right? When, especially in the post alpha fold period, where we got a lot of computational people straight into bio, we're really looking at crystal structures, we think that's what matters, but there's this implicit assumption that, hey, there's a lot of water molecules bouncing around, creating that shape that we trust.

And why is it important to consider water? And can you help us understand a little bit of why we might want to consider proteins? Without water 

[00:17:01] Samuel Thompson: great question. Thank you for directing me in that way. From a biophysics standpoint, if you take stat mech, if you do any kind of folding, from the very beginning that solvent is the most important thing. Solvent is what drives folding At the same time, we don't really consider protein folding in other contexts.

We know that water is super important, and we know the way that it contributes to the energy of protein folding. But we don't have a great comparison. So what I basically, argue is that going back to this idea of self assembly, is that, Proteins are these chains, they're these polymers that self assemble, and they collapse by the hydrophobic effect.

So they're excluding the polar solvent, the polar water from their core. And so if you want that folding to occur in an organic solvent, you have to be careful about the organic solvent that you choose. And so we chose octanol, which is very hydrophobic and allows us to invert that phase separation, such that we have a polar core and a hydrophobic exterior.

We're trying to prove that proof of principle. We have, I think, a good start of showing that we can design these things find proteins that actually leave water and go into octanol. They form secondary structure. They form some kind of defined 3D structure, although we're trying to verify if that, Okay.

3D structure matches between the simulation and what we're designing. But yeah, really, then the question is, okay, why would you even do that? And I think there's a lot of industrial applications on the catalysis side. Work has been done in trying to make high value products. And that's probably the first Thing that one would want to do if one was, putting on their business hat and starting a company is going for these expensive things.

But then there's also a bulk applications. Using biofuels. For example, taking all of the waste cooking oil from restaurants and hotels and converting that into biofuels would essentially replace 000 7 percent of U. S. Diesel consumption You can imagine using enzymes for doing recycling for taking mixed species of plastics and doing recycling And if you could do that in organic solvents, you can perhaps do it in a way that actually pulls apart the polymers and makes the polymers more accessible to what you're using to recycle them. And that makes the process easier and less energetic,

 we want to think about the ways that proteins that do self assemble can be used in high technology to increase the longevity and enable some of these energy transformation technologies and make them more, more long term, more scalable, more robust over time.

[00:19:23] Dan Goodwin: It's the motivation that I think gets a lot of people excited about working with proteins that it's, the infinitely programmable nanomachine that biology is invented. But I think there's this implicit thing that we always have to say, okay, and it's going to work in E. coli, but to say it's going to work in a lithium ion battery that would explode if you put water in it, I think makes it a fun, very important topic.

And so I think it's one of those things where, you know. Trillion dollar market cap of chemical production. So obviously there's probably like, you know, there's two things going to happen. One, like first people are going to doubt the problem. And then once people see that, Oh, wait, this is, potentially a huge unlock for chemical synthesis.

Then you say, okay, people surely must've tried this before. And when I was going through the background of your work and the background of the field, it really is remarkable to me that people have been trying this all the way back to the eighties. then I would love to just do a little bit of a, like a story about how the field developed and, specifically what the role of AI is

[00:20:19] Samuel Thompson: it's going to be impossible to really adequately cover the field. I think in this. Time period, just because, as you mentioned, so many people have done so much. I think there are a couple of names that stand out and I was really pleased in what you sent me that that open AI chatbots can actually pull out some of the biggest names.

And I think that two of those big names are. Klebanoff and Arnold. So Alexander Klebanoff at MIT and Francis Arnold at Caltech and they've had different contributions and many people, as I said, have contributed this work. Klebanoff He was really considered the founder of enzymes in organic solvents.

But a lot of what he was doing was crystallizing proteins subtilisin for example and putting them into organic solvents. So you exchange the solvent and that transfer, that exchange, I think is really key to what he was doing. But he was actually putting enzymes into neat organic solvents so anhydrous solvents.

And he was the one who was really looking at the water shell that, that you're It is bound to the protein in the organic solvent. And then Frances Arnold really started her lab basically saying that the protein folding problem is going to be fixed very soon. And we're going to be able to engineer whatever protein that we want. And so as an organic chemist, I'm going to take that and I'm going to. Enzymes and proteins and whatever we want, and we're going to put it into organic solvent, and we're going to replace all of organic synthesis with enzymes, and there's huge benefits from that.

Organic synthesis pathways have multiple intermediates, they have multiple protecting groups, multiple purification steps. And so oftentimes if you can substitute an enzyme in, you can replace many of these steps in one, because in part you don't have to add on other steps. all of these protecting groups.

It really streamlines the process because the protein is large, it's bulky, it binds the substrate in a specific orientation, and it has an incredible regiospecificity and stereospecificity that's very hard to accomplish when you're using small molecule catalysts or just free reagents and solutions. So there's a real big advantage to synthesis if you can use these large protein catalysts.

Klebanoff was crystallizing things and exchanging the solvent. And Francis Arnold largely was doing this kind of step based approach, which is very much in line with her views of directed evolution where you gradually shift the solvent or gradually shift the amount of solvent.

And so often times with Francis Arnold, even when she's talking about working with enzymes inorganic solvents, it's Actually, organic solvent mixtures. So it's water and the mixture of something else and from these two studies I think or these two different directions and the holistic sum of what the field did I think that the takeaways are that you get either a highly stable enzyme in Organic solvents that are nonpolar, and it's highly stable, but it's not very functional because it has all of these bound water molecules to it.

So it loses its dynamics, and the dynamics are so important for function that it oftentimes can't undergo the motions that it needs to perform its catalysis. And then for a lot of the challenges that are taken on by trying to do the step wise approach, where you're going from water into a polar organic solvent and gradually shifting the solvent, you find that you very quickly end up in these conditions where you are unfolding the protein.

Because again, protein folding phase separation. And so if you have something that is an amphiphilic solvent, it's able to stabilize both the folded and the unfolded state of the protein. And I think that's where my kind of conceptual innovation or conceptual difference at the very least is where I'm basically Thanks.

If you want to find a protein that folds in organic solvents, you have to make a jump. You have to make a big jump into these non polar solvents, these very hydrophobic solvents. And because one, that's the most similar to the membrane where we see protein folding. You have to really reinvent and invert the protein architecture in order to get it to fold in those environments. And, To do that, the question is AI enough of that? I don't think AI is sufficient at this point. And that's mostly because AI needs good training data. And so there's nothing really in the PDB that's anything like the proteins that we're building or anything like I think that the Proteins that would fold in these solvents should look like if my hypothesis is correct.

So I think there is definitely going to be a place for AI in the future where you're wanting to diversify the structures you get or build new things, but you really have to start from actually generating the data to train those AI models. If you want to enable the next generation of AI assisted protein design in organic solvents.

You need something that gives the AI that base foundation of what are the rules of this new environment.

[00:25:23] Dan Goodwin: I get really excited by the intuition that you're driving on here. And I mean, a great little personal backstory here is that in my PhD, every experiment we did took six months for basically, one to two samples.

[00:25:34] Samuel Thompson: were doing mouse experiments?

[00:25:36] Dan Goodwin: I was doing mouse and we were doing expansion microscopy and in situ sequencing techniques.

And it was a lot of microscope time. And so I just became really attracted to anyone who would say, Oh yeah, I just, I did a million cells this morning, just, I just faxed it, and so it feels like there's a lot of kind of that intuition about what you're doing and the paper that you just published, I think is really cool.

And so the title is the fax. I think everyone will know what fax is, but it's the fluorescent sorting cell sorting technique. So fax sortable triple emulsion peak of reactors for screening reactions and biphasic environments. And we'll put a link to it. Yeah. And I think just this phrase, triple emulsion Pico reactors is worth rooting everybody in and why this is a step towards exactly what you were just saying, which is it's not just alpha folding it and be done, but really creating a whole new stream of data.

So what is a triple emulsion Pico reactor and what were you doing in this paper?

[00:26:33] Samuel Thompson: So a triple emulsion pico reactor is this technology for miniaturizing reactions. We have this goal of taking kind of the CPU like benchtop experiments ifying them, where you can do a ton of really simple experiments all at once. So we encapsulate very small reactions, picoliter scale reactions, which is why they're pico reactors, inside of an oil shell, and that oil shell has water on the outside of it, so it's compatible with these cell sorters.

So we're mimicking the architecture of a cell, but turning it into a test tube. And so it's a matter of what can you put into that droplet? A lot of times when I'm talking about producing proteins for organic solvents. Everyone asks me, Oh, do you have to make a ribosome that works in organic solvents?

And maybe that's far down in the future, but that's really not what we're doing right now and not really something that we're capable of. So we're still limited by the molecular biology that we need. We need to do DNA, Transcription translation to make the protein. So we can order a DNA library.

We can order these massive libraries. They're very cheap. We can make them very cheaply because we're using, very small amounts of reagents to peak leader reagents. And but we need the protein to go somewhere. And so we need optional in there. And that's where the triple emulsion comes from.

So it's oil in water. in fluorocarbon oil in water. That's the triple emulsion. But it gives us the ability inside these very small reactors to both do the molecular biology to create the protein and then also have the oil to assay if it's going where we want it to go and doing the chemistry that we wanted to do.

That paper doesn't really take it all the way because we're not fully going to the point of putting non aqueous proteins in there, but we get about halfway there where we're starting to build up some of the assays that we would use to screen for these proteins. In particular, looking at this class of substrates that actually partition into the octanol and stay in the octanol.

And so we found. In the course of doing this work, these substrates that kind of model the reaction that we want to do for producing biofuels, but they never leave that oil droplet in the core of the pico reactor. They never go into the water phase. So we can make proteins in the water part of the droplet, see what goes into the oil droplet based on whether or not it activates this fluorescent substrate in there.

And that's what we're pushing on doing right now.

[00:28:56] Dan Goodwin: Wow. And what is this fluorescence signal that you're creating?

[00:29:00] Samuel Thompson: So it's very simple. They're very standard fluorophores. These are just esterified substrates. So anything that would work with an esterase or a serine hydrolase. But the real difference is we're choosing fluorophores that are hydrophobic enough when they're esterified.

So there's no, polar charge groups on them. 4MU is one that has worked well. Resorufin is the one. That we use in the paper, but you take the charge group on them and you esterify it. And this is something that people have been doing for a long time, back when I was working in Alice Ting's lab, we were doing it to deliver these fluorophores across cell membranes.

It makes them hydrophobic enough that they can cross the membrane, but then that ester gets cleaved by an esterase and then it's charged and accumulates in the aqueous phase where it's fluorescent. What I'm doing is I'm taking that to the very extreme where I'm saying, can we put a super hydrophobic chain.

onto that fluorophore that esters a really long carbon chain such that the partitioning coefficient is so high that once it goes into the optional when we dissolve it, it's actually not even soluble in water. We dissolve it in the optional and it just never comes out. We can never detect it. If we put Lipase like catalytic activity around it.

We have to have something that actually goes into the octanol to convert it. And then the fluorophore is charged and it goes into the aqueous phase and is fluorescent there. And we have some early evidence showing that we can actually do that. That we have catalysts that go into the octanol and turn over 

that substrate. Whereas if you have an enzyme in solution, you see no fluorescent activity.

[00:30:23] Dan Goodwin: I love that you prefaced a question. You, I love that you had brought up a question you get a lot, which is people saying, Oh, do you have a ribosome that works in these weird environments to do that? And that's fine. It makes sense. That's a common question you get. I feel as if another common question you're probably going to get is, could you do a GFP?

That folds correctly in these environments. And would that be an essay?

[00:30:44] Samuel Thompson: Perhaps at some point in time, you will be able to do it, do a GFP. But we've been relying on chemical reporters. So Roger Tien who really his lab found GFP and developed that into the technology, it is also developed these reporters flash and reash that use Quenched fluorophores that react to a peptide tag.

It's an arsinylated fluorophore and you kick off these chelating agents, and it binds to the peptide tag. It puts the fluorophore in a slightly different conformation that's fluorescent. We've been adapting those for our use, and we've been very lucky that, not only do they seem to work in our assays and allow us to build ways of detecting the proteins that are in organic solvents, but there are also phase separated versions of those fluorophores as well.

it's been a really interesting way of combining Old approaches of chemistry that I think have somewhat fallen out of favor into this new environment where we really don't have a lot of the nice genome integrated tools that we would want to.

We do have this genome integrated tag or genetically integrated tag that we use to activate the fluorophore, but we do need that kind of small molecule technology to make it work.

[00:31:54] Dan Goodwin: Yeah you're living your flat side of the exponential curve, for sure. Because you're in an environment where there's very few tools, and it sounds like you have to solve a lot of the stuff yourself. It's very funny to listen to this now, because there's so many implicit solutions that you're just saying oh, we need reporters, so we found this molecule, we needed, this light pace activity, and this, and it's very funny because we could accidentally step over that thing like, oh, this is all really easy, but there's hundreds inside this, from biologist to biologist, like I'm hearing so many.

Implicit solutions that you hit your shins on, you did, they had to figure it out and some, I'm curious, looking back at this and the result is, of course, this extreme throughput of screening these variants which were some of the engineering challenges that got you stuck for a long and maybe if you can, if it's possible, say this one was expected to be hard and this one was not expected to be hard, but took us a while.

[00:32:44] Samuel Thompson: I think that we benefit and we are a little bit hobbled by those technical challenges in part because there's many ways in which the things that I'm doing would not be special if it were not happening in organic solvents, right? Sometimes I look at the very basic things that I'm trying, the basic techniques that I'm trying to recapitulate.

And I think this is, very important. Simple things. This is the kind of thing you would teach a rotation student. But, in my context, it's blows people's mind, right? So the bar is very low. 

[00:33:16] Dan Goodwin: I folded a protein.

[00:33:19] Samuel Thompson: even that came on. I made a protein. Right? Very simple things.

Goodness. one of the challenges that I thought was going to be harder than it was is I thought it was going to be hard to find. a proof of concept of a protein that will actually fold and partition into organic solvents. And that's part of why I thought that we were going to need high throughput screening to begin with.

And that turned out not to be the case at all. We've just found these things all over the place. I think the base hypothesis, really has a lot of legs just because we've been able to in low throughput screens, a lot of hits for positive controls just to allow us to iteratively advance our assays, the more confidence that we have a positive control that allowed us to test more and higher throughput methods and have greater confidence that our screens were actually working.

Another thing that I wasn't entirely certain about, or that we got really lucky, is the fact that we are making these proteins in these biphasic reactions. We're making the proteins using in vitro transcription translation and allowing them to partition into octanol, which means that both of those proteins.

Solvents, the aqueous and the non aqueous are present at the same time, and I wasn't sure that it was all going to be compatible, and again, that really just worked out. The thing that has been hard, that we knew was going to be hard, was how were we going to determine the structure.

We knew that probably crystallography was going to be hard, even though we have tried that. And that because we're working with small things, NMR was probably our best bet for getting structural information. And that has worked marvelously well in the few tests that we have done. But the real challenge there that I think I hadn't quite anticipated was how challenging it was going to be to see.

Make these proteins at a scale for NMR. I was hoping that we would be able to synthesize these peptides chemically and that has worked in some cases, but it's been highly variable, right? So I was thinking that if the ribosome can make it, then surely, the chemists will have no problem with making a really simple peptide by coupling, since they're fairly short, but that has been a challenge and that is really currently what we're grappling with is how do we get proteins that are contaminant free and at high enough yield for NMR?

And that, that is one of the things that I'm really actively working on right now.

[00:35:48] Dan Goodwin: Fascinating. Oh, I just, I love this work so much. And so what I would love to do is, before we move to rapid fire questions at the end, just to hear what you're excited about moving forward. My understanding of your work is that you've been building, you've invested heavily in these projects.

foundational tools and figured out all these methods. And I'm sure you're going to figure out how to produce the proteins at sufficient scale to get the NMR working. And then what applications are you most excited about or what frontier tools do you want to show next?

Yeah. What are you excited about Samuel?

[00:36:19] Samuel Thompson: I'll say the first thing I'm excited about is I'm excited about getting this project to a position where I can start recruiting people. At the beginning, I was giving people a little bit of warning saying that this could fail massively. Rotation students may not want to work with me.

And I'm very interested to hear, the new ideas that come from other people. I think for me, I'm really interested in. This question of what kind of odd topologies can we get where we're exploring whether or not there are structures that are possible or structures that are beneficial in different environments.

And I'll give you one example of something that I've tried that I'm not entirely certain will work. We know that exposed backbone is not good for these proteins and so loops are particularly hard. I think ultimately we're going to have to find some way of adding unnatural backbones in order to actually allow us to do interesting loops.

But in the meantime, one of the things that we've been playing around with is the use of proline, because proline can cap the backbone in certain places. And one of the things that we found is we can't make beta sheets, but we can make beta hairpins. that alternate back and forth between proline and another amino acid.

And so we end up with these like really crazy looking beta barrels that have super high proline content. So I'm really interested to see what creativity this very harsh environment brings out in terms of the types of structures that we can make that aren't seen in proteins currently.

And then once we can do that, or even before we can do that, looking at what are the very simple catalytic activities. Again, we're starting as simple as possible. We want to understand things like what does it mean to buffer a reaction in organic solvents? And what does it really take to get good catalytic turnover in these environments?

I think there's, again, going to be the same process of going through, just the base building up of how to run a good enzyme assay in organic solvents. That's reproducible and controlled, etc. We're gonna learn a lot about that process, and it's gonna be a lot of reinvention. So I'm excited about that new phase of troubleshooting and adapting,

[00:38:28] Dan Goodwin: It's going from that cold start problem to the cool kids party. I can totally imagine a future world where. there's quote unquote, normal protein engineering, and then there's the organic solvents people. And they used to be the fringe weirdos.

And it turns out they're actually doing some of the best and biggest deployments of proteins.

[00:38:45] Samuel Thompson: and I think we'll eventually even go beyond proteins. Like I mentioned at the beginning, I'm really coming at this from this idea of a self assembling polymer. I'm open to working with any kind of polymer material as it becomes available, and as we have the technology to do that templated synthesis.

I can't help, but just bring up, there's this parallel, which is the big extra topic and bio, which is the mirror world. But it's like the organic solvents world, which is a little less kill everybody vibes, but also wild new frontiers of templated synthesis.

Yeah, I think that I find myself. Thinking about the protomolecule and the expanse, like there's a lot of cues to a very organic component of that, which also does not have the happiest outcomes, but also is very much a double sided technology where it has, great potential, but also a lot of destructive power.

And yes, as scientists, we always have to keep those things in mind.

[00:39:36] Dan Goodwin: Oh, it's so cool. Samuel, I want to wrap up with our rapid fire questions but before I do, I just love getting a chance to geek out. I learn a lot from this. I get excited by the work you do, and I feel like the time is right.

I totally buy your thesis that we're going to be emerging from the flat part of the exponential and up the elbow to the rapid growth phase. And I'd love to go back from the technology back into the person. I love your identity driven science view. And I think this is where, we get to share some more of that with the audience.

And I remember the first time we talked, we geeked out probably for 25 of our 30 minutes just on sci fi, and I always cherish that. the first question is, what's a single book, paper, art piece, or idea that blew your mind and shaped your development as a scientist?

[00:40:21] Samuel Thompson: I think I talked about that in terms of the idea when I was in Japan, that conversation that people had wanted to put proteins in organic solvents and couldn't do that. That really changed me. I think a lot of Kim Stanley Robinson's work also really shaped me the Red Mars series, this idea of taking a long term project of how you terraform a planet and all of the different kind of political aspects of thinking about the different stakeholders and the different approaches to doing that.

That also has been a big part of how I think about my project in terms of thinking about these technologies in terms of where do we need robust technology to deploy? And I think I always go back to Mars because If we're on Mars, we're going to be so limited in our resources that you really need something like biotechnology, where all you have to do is just streak out a plate and grow it on your waste products and you have what you need.

[00:41:22] Dan Goodwin: You don't want to be relying on these very fragile supply chains and ecosystems. You want something that makes itself. I really love how much of a stealth there's like a stealth astro What is

it? Yeah. Yeah. I just, I love that. Like a lot of people had like big biotech ideas are also really big into traveling the cosmos and colonizing Mars and, or whatever our word is for visiting Mars and thriving there.

So I love it., I have not read the red Mars series. I'm going to check it out. So next question what's the best advice line that a mentor gave you? 

[00:41:57] Samuel Thompson: I think that one of the best advices. That I got was that really transformative biotech is simple. And so you have sequencing, which is reading. You have Restriction enzymes, which is cutting. You have CRISPR, which is cutting in a more specific way.

And then you have, I think, in my case these droplets, this phase separation these microfluidic compartments, which separate. And I found so many ways in which I collaborate with people who are doing crazy science that I. Would have no idea could be applicable to this because I'm working in this area in Polly's lab So I think that if you want to have a big impact on science find some kind of simple Modular technology and that really scales

[00:42:48] Dan Goodwin: That's fantastic. one of the lines that changed the way I think is there's a, sorry to be a cliche, but there's a Steve Jobs quote on this, which is that when people start doing something, they do it because they have a simple idea. when you start working on a simple idea, it quickly blows up into a bunch of complexities and this and that and he has this point that where a lot of people fail is that they try to go to market or they try to publish a paper that is a simple idea turned really complex, but the great people are the ones who are able to go back to a really simple idea.

And so simplicity, I think in his view is the idea is encapsulating a lot of complexity, not ignoring it. And I honestly, I wish somebody had helped translate that to me in biology sooner. So I love, I love that answer. So similar now, it's kind of, that's the best line, advice line a mentor gave you.

I'd love to swap it and get some advice you'd give to other people. And one way to prompt that is to ask what is a skill or sub area that In and around biotech that you'd suggest that developing biotechnologists should look into, and there's two ways to answer it. There's, what is an overlooked area or, what is a great foundation that you think empowers people to do great work?

[00:43:51] Samuel Thompson: Okay, so here's where I'm going to be contrarian and queer and say, I'm not going to take your binary and I'm going to answer this question in my own way. Think, for me and this maybe also connects with mentorship advice, so Nergis Mvalvala, who I believe is the Dean of Science at MIT right now, but she worked on the LIGO project for detecting gravitational waves.

She came to MIT. UCSF. I was a grad student there and gave a talk. And basically what she was saying was, physicists, we all get together and we look at these big multi billion dollar projects, multi decade projects. We decide what are the things to go after that's going to enable The most science for everyone that creates the biggest, widest tent for everyone.

And she says, I don't think biologists are doing that. And I really think that biologists, and especially young biologists, should pay attention to politics. And I don't mean party politics, even though that's important too. I mean the power of a group of people collectively working together to come up with a problem, to 

come up with a goal to come up with a challenge. We in biology are very separated. We do our individual labs and our individual small scales. And that's great. And miniaturization is great. But there's a real need to come together and say, this is the thing that we need for all of biology to be advanced.

And I'm going to make my pitch for what that is. That is nanoscale resolved, angstrom level, wholesale imaging. If you could. Watch a whole cell, with that depth of field at angstrom scale resolution and nanosecond resolution. You would change the way that we do biology and chemistry because you would turn it from this struggle to get observables into something more like ecology where you're just taking pictures.

There's looking what happens, analyzing the data and reporting back. And when it comes to the structure of chemistry and the structure of biology and how we see molecular structures, biologists have been eating the crumbs of physicists for years, right? Electron microscopes, synchrotrons, those are all things that physicists developed for themselves.

I think there's a space for biologists to step up and say, Hey, this is what we need the physicists to make for us.

[00:46:02] Dan Goodwin: You answered the next question there, which is if you had a magic wand, where would you send it? I think you've answered that too. People listening to this obviously won't be able to see this, but I'm just smiling ear to ear because I love the push. The line biologists have been eating the crumbs of physicists for years is a really great salty line.

I hope that gets some people to shake their fists. And I think it's a really wonderful motivating line to end on which is that the power of a few. Really motivated people can never be underestimated. and both to inspire more people, but also, just even a small tight group can accomplish a ton.

So I want to end just by saying, I absolutely love the work you do and who you are and how you do it in your very own unique way. And it's just wonderful. Before we wrap, is there any way that you want to share with how people listening to this can find you or your work?

[00:46:48] Samuel Thompson: I love that question. You can find me in person. I have never been one for a big internet profile or digital strategy, but I love meeting with people. And I think especially now it's so important. I'm on the Fordyce and BakerLab websites, reach out to me, email me, find time to get coffee with me, chat with me.

That is the best way. Find me in person.

[00:47:11] Dan Goodwin: Best answer. Samuel, thank you so much for taking time. I really enjoyed this conversation.

[00:47:16] Samuel Thompson: I've loved it. And I've really appreciated you and your roadmapping and your vision.

[00:47:19] Daniel Goodwin: Thank you so much for tuning into this episode of the climate biotech podcast. We hope this has been educational, inspirational, and fun for you as you navigate your own journey and bring the best of biotech into planetary scale solutions, we'll be back with another one soon.

And in the meantime, stay in touch with homeworld on LinkedIn, Twitter, or blue sky. Links are all in the show notes. Huge thanks to our producer, Dave Clark, and operations lead Paul Himmelstein for making these episodes happen.

Catch you on the next one.