Optovolution: Teaching Proteins to Think Like Computers

Intro — dynamic proteins and why this matters

0:00Hello internet. This is your captain speaking Lester Narre joined as always by my co-host and our resident PhD Krishna Chowdery. Today we are going to be touching on a protein evolution story in the lab directed by humans. So historically if we've wanted to make proteins do things that we wanted them to do, we've used you know the process of evolution uh in order to do so. But there's a lot of limitations with that approach. This next story is taking that method and putting it on steroids, giving us the ability to now be able to

0:41turn these on and off and sort of create almost like a computer, these simple logic gates that give us much more control. And it's fascinating how this is going to have implications in a variety of arenas. And the way in which they did so is very interesting. So we are going to learn about the science from the ground up today because this is from first principles.

Directed evolution 101 — how we normally engineer proteins

1:19>> So we want to make new proteins. >> Yes. >> All the time. Right. That's a lot of biomedical research going on right now is to make new enzymes, new types of proteins to do all sorts of things. And traditionally, a lot of times what we can do is use inlab evolution to make novel proteins. Now, those techniques for reasons that we're going to get into only make certain types of proteins, specifically ones that always remain on. But if we want proteins with actual more natural dynamics like proteins that turn on for certain things and off for certain things or if we want to have like mini logic gates like the the ones that we have in our computer for the and or the or or the exclusive or things

1:59like that it's been really difficult to use the traditional technique to build those types. This is taking that step forward. Okay, it's a breakthrough called optoe evolution from EPFL in Switzerland. It's a cell journal publication and the mechanism is really really cool. What it's doing is it's using continuous evolution with light to steer the evolution of dynamic proteins. Not just proteins that are going to be in one state all the time, but proteins that can transition and be in dynamic states. It couples like effectively like an oscillator, like a optical oscillator

2:39with internal cell cycle oscillators. There's like incredible like deep physics underlying this and it's it's actually really really cool. Um, and to get started, let's just go through some of the history of directed evolution. Okay, >> so evolution is really kind of an optimization, right? We all know it as nature's way of engineering biological systems. There's the idea of survival of the fittest, which is you've got some population inside of the population, inside of that DNA, there's a lot of variation when it goes from one generation to the next. So you have variations in the genetic code, which leads to variations in phenotype, which means that, you know, like I'm going to

3:21have black hair or blonde hair or whatever. And through natural selection, the organisms that function most effectively are going to reproduce more and they're going to proliferate more. And that's basically the central tenant for evolution. Now, humans have harnessed this process for a very long time to do artificial selection, artificial evolution. Early farmers influenced evolution by choosing which crops or which livestock to actually move on to the next generation. And if you look at, you know, the thousands of years of selective breeding have given us all of the crops that we know and love, but they started out not great.

From crops to enzymes — artificial selection at the molecular level

3:58This particular photo, the one on the upper left, that's watermelon. That does not look like watermelon. >> Does not look appetizing. >> That does not No. Right. Um the one in the middle, that's corn. On the right, that's a carrot. That looks like a stick. >> That looks like a twig. >> That looks like a twig. That's not what we see in carrots. Right. Um, so all of the food that we eat today is a result of artificially selecting the ones that we prefer and then having them proliferate to the next generation. So this is not a new thing for humans to sort of take over the evolutionary reigns and steer organisms to their will. Okay? Here what we're doing is at

4:38a biomolecular level we're doing targeted directed evolution. Okay. This is the idea where you induce mutations in the DNA. So you make thousands of versions of an enzyme let's say from that DNA and then you select the ones the versions of that enzyme that do the best for the next round. So you know let's say I have a thousand different versions. I'm now going to create thousands of different enzymes out of those the top 100. I'm going to select that DNA and then now do it again. Do a bunch of mutations. create a thousand different versions again test it I do this natural selection at a genetic level now right at a molecular

5:20level and what I'm doing is through that process I can make catalysts replace toxic chemicals this is this has been

Frances Arnold, directed evolution, and the 2018 Nobel Prize

5:29huge for the biomolelecular industry we've created biofuels out of this we've created agricultural chemicals we've even created a better way to produce a drug for treating type 2 diabetes So this has been around for a very long time. Actually in the 1990s there was um a professor Francis H. Arnold at Caltech. She started doing this. She's really been a pioneer of this type of directed evolution. She won the Nobel Prize in 2018 for her work. Um interesting tidbit about her undergrad at Princeton. >> Woo. Tigers. Yeah. It was an undergrad in mechanical and aerospace engineering which which I thought was interesting because from there she went and did a PhD at Berkeley in chemical engineering

6:10and then a faculty at Caltech. And I think this is just a really cool way to think about like what you do in undergrad is not like going to totally sequester you for the rest of your life. >> I think they say what past is not prologue or something. >> Yeah. I mean, she did aerospace and mechanical and now she invented directed evolution and won the Nobel Prize in 2018 in chemistry, right? So, it just shows that like in undergrad really what you're trying to do is learning how to think, >> right? And that's what's important. And then you can move on and use that skill later on to do whatever whatever you want. It's not over yet. You still have a chance. >> Exactly. Um,

The core limitation — why classic directed evolution only makes “always-on” proteins

6:50so she pioneered directed evolution. Here's the problem. Directed evolution is all about static selection pressure. For example, if I want to make an enzyme that breaks down a certain toxin, what I'm going to do is create a bacteria with these with these with this DNA that is constantly inside this toxin bath. >> Okay? So now only those bacteria that have the enzyme to break down the toxin are going to live. But naturally, what that means is that the enzyme is going to be constantly on. It's going to be constantly trying to break down that toxin. It's never going to turn off. >> So, it's going to this whole process of

7:32directed evolution is going to select >> for proteins that have continuous function, not dynamic function. >> That's very different from proteins that we see in life. >> Proteins are dynamic, >> right? >> They move from one configuration to the other. Here what we're seeing is a particular lian that green dot is moving in into the protein and the alpha helix in blue is changing shape. The protein is literally changing shape and toggling between one state and another state. And this is very natural for proteins to do. They want to toggle between an active and inactive conf confirmation in response to some kind of stimuli.

Dynamic proteins as molecular transistors

8:10>> It's effectively acting like a molecular transistor in some sense, right? There's an onoff switch. There's a zero and one. And we don't want a protein that's either always zero or always one. >> We don't want to be always asleep or always awake. >> Yeah. Yeah. We want we want things that actually wake up and do things based on their particular environment. And that's where this paper comes in. Okay. This paper is called light directed evolution of dynamic multi-state and computational protein functionalities. That's what they're doing. They're using light to do this directed evolution and from that they can actually figure out ways to create dynamic proteins. That's the optoe evolution um solution that we have. And

8:53so before we get into the mechanics of all that, >> yes, >> let's do a little bit of housekeeping. >> Yes. So, I think I want people to understand how incredible it is to be able to have the opportunity to have this show and to talk to you about Breaking Science news stories every week. As some of you who

Housekeeping — new format, support, donate, APS plug

9:14watched our last episode may know, we are trying out a new format here where we're going to have multiple episode drops per week rather than one big monolithic episode. So, you can expect us to have about two to three episodes out per week as we try out the format. Big story episodes like we're talking about today are focused on really getting deep in the weeds on all of the mechanics from first principles on a particular research paper. Occasionally, we'll do a deep dive in a general topic. We have some of those coming up that are really exciting. And we'll always end the week with our SNL weekly weekend update style rundown where we'll cover several stories, a little bit less

9:56in-depth, but more topics because there's so much happening in the world of science every day right now and we want to make sure we cover as much as possible. If that sounds incredible to you, that sounds like something that you want to see in the podcast universe, if you are not particularly motivated to listen to another talking head share the same opinion that the other talking head talked about, one of the best ways you can support the show is a like, a share, a comment, a follow. Bring it into the water cooler at the office, at the lab, at school. making sure that we are battling that billionaire algorithm is the best way for us to get this show out to as many people as possible. We are on

10:37our journey to build an ESPN for science in all ways and all of the support that you, our audience and community give us is the way in which we can continue to to produce the show. Speaking of producing the show, it's just two of us, myself and Christian. All this, the camera switching, the production value, the quality of the content is centered around two folks. And so, one of the other ways you can support us is we do have a donation portal if you would like to be a patron. FFPod.com/donate helps us monetarily support this show and remain independent. And if if you are interested in representing

11:19FFP Pod in the public, let us know that you'd like to see our merch merch store go live. We do have some interesting stuff ready to go and we just want to make sure that there's interest in having FFP related clothing items, maybe a couple special items that we will not talk about just yet. But with housekeeping over, we can jump back into the story. Except for >> and one more because this should be coming out on Wednesday. So if you if anyone is at APS at the APS global summit, we used to call it March meeting but now apparently it's called global summit. The American Physical Society conference is the largest conference of physicists. I will be giving a talk on

12:01Thursday. If you're going I think there's an app or you know you've got the program look me up Krishna Chowy. I'll be giving a talk on Thursday. Come check it out. I am uh you know actively still doing stuff and trying to talk about it to other physicists. >> It's very very very important and very very very important that you go there. Okay. I'm just he's so embarrassed. I'll stop. I'll stop. No, I thought you were about to >> go look at the previous episode if you want to see what we >> uh and with that again we are so appreciative of those of you who tune in to the show and for and a lot of you new uh folks who are coming in from our Netflix appearance on Dave Chang, welcome. Uh we're going to have a great

12:43time here. Uh you might learn a little bit, you may not learn a little bit, but you still might find it interesting and entertaining. Yep. So let's get back to the story. We want to talk about opto

Optovolution begins — coupling proteins to the cell cycle oscillator

12:55evolution which is this new way that this paper has figured out how to use directed evolution. This idea of inlab evolution but now to create dynamic proteins not just proteins that are remaining on but can now toggle between an onoff switch like normal proteins. Okay. The physics that underlies the core principle is from coupled oscillators. Okay. This is something that we learn about in undergrad physics. You've got two oscillators that are connected by some kind of mechanism. >> Mhm. >> Over here we see just two masses on a spring. And you can see in this case the energy from one is getting transferred to the energy of the other. And so one of them oscillates and then the other

13:36one remains stationary. And then when the other one oscillates the first one remains stationary. >> The idea >> is that if I were to apply this to my protein >> challenge, right? I want to create a protein that can oscillate between an on and an off state. So what if I can tie its performance to a fundamental oscillator of life? >> Okay, >> which is the cell cycle. >> Okay, >> the reproduction cycle of the cell, right? Because the cell has to grow, then it has to split and then it has to grow and then it has to split and it has to grow and it has to split. So if I could somehow couple the evolution of

14:16the protein in interest with that oscillator >> Mhm. >> such that it needs to oscillate in order to really talk to this fundamental oscillator of the cell cycle then perhaps I can optimize for proteins that can be dynamic rather than only proteins that can be on and off. I want to ask a quick question just to help me understand when you say oscillate >> uh can you opine on that a little bit in terms of what what exactly you mean by oscillate inext. >> So in in in physics when we talk about oscillate we mean that it moves from two states in some type of way. It doesn't

14:56have to be continuous. It doesn't have to be discreet. For example simplest oscillator would be a pendulum. It's going to the left, then it's going to the right, then it's going to the left, then it's going to the right. That's an oscillator. Um, a mass on a spring, it's going up, and then it's going down, it's going up, and then it's going down. Um, a ball on a hill, >> a ball in a valley, it's going to the left, then it's going to the right, then it's going to the left, then it's going to the right. In physics, actually, it turns out everything is a simple harmonic oscillator in that sense.

The eukaryotic cell cycle as biology’s built-in oscillator

15:23Harmonic in that like there's I think I think it just comes from like harmonies and the fact that it's you can fit it with a sine wave. Mhm. >> And an oscillator, I think it just means it's oscillating from one state to another state, >> right? And so the cell cycle is an innate biological oscillator because it's oscillating from growth to division to growth to division. Right. So it's a cycle. >> There's a cycle that has some kind of period. >> Yes. >> Um and a starting position and an ending position >> and then it goes back to >> and it go and and it repeat there's a repetition of the movement between those two. >> Exactly. Yeah. Yeah. And so, and so what we want to do here is in order to create

16:03an oscilly protein, something that can toggle between an on and an off, I'm going to couple it, connect it somehow to the original oscillator of biology, which is the cell cycle. And would the potential benefit of doing something like that be the biology? It's it's sort of like the same reason we use um um was it quazars for clocks >> uh because it oscillates rhythm on a on a cycle that is so consistent and so repet like it's a dependable system to base off of. >> Yeah. With a quazar, it's like so far away that it doesn't depend on my position the the earth's position around

16:45the sun and so on. So, it's always in the same spot in the sky, right? And so, it doesn't depend if I'm in July or March, the quazar is always there. And so, I can use that as my clock to figure out how long my day is. In in this sense here, we're using the cell cycle because it's the cell cycle is just going to go, right? Otherwise, the cell dies, >> right? So if I can make a protein that I need to be dynamic tied to the cell cycle, then if the protein arrests into either an only on or an only off state, it's going to kill the cell. >> Mhm. >> And then that protein is not going to proliferate in evolution. That's the sort of high picture idea of how we're

17:25going to use evolution here in this case, right? Because now it's survival of the fittest. But in order to be fit, I need to be able to toggle the internal dynamics. >> The protein needs to be able to go from one state to the other. >> Mhm. >> Right. That's what's key. >> So, let's talk about this biological oscillator that I've been going on about. This is the ukareotic cell cycle. Okay. Mitosis, >> which is something that I hope everybody who did high school biology knows about the idea of binary fishision, right? The cell doubles its DNA and then it splits in two. and then it doubles its DNA and then it splits in two. Well, the doubling of that DNA happens during interphase, specifically the Sphase. That's the synthesis phase. The

18:07splitting in two is the mitosis part. But after splitting in two, now each of these daughter cells have only one >> copy of DNA. In order to replicate again, they need to go through the Sphase again. They need to synthesize the DNA and double their DNA and then split up again. Now this entire cell cycle is driven by something called cycl dependent kinases CDKs. Okay, these cycl

CLB5 and the lethal “always-on” problem

18:31dependent kinases are formed and degraded at very specific times. And if we go back to to photo 11, what we'll see is what's happening is a certain compound comes up during let's say the G1 phase, which is the very beginning of interphase. So this is right after mitosis, right after the split, a certain cycl E shows up in the cell. Okay? And that's going to trigger the next phase. Okay? Then during the Sphase, during synthesis, another cycl comes up. That's going to trigger the next phase. Now, how is that triggering happening? The triggering is happening because the certain cycl shows up and it immediately gets degraded. So it's only

19:13there for a short amount of time. So it's like this boop signal, you know, it doesn't stick around. It's it's maybe like um is an analogy. It's it it's like the ignition when you're starting your grill and you click the button on the grill to get the ignition going >> and it's only there at the beginning but then the fire's going and it's not there. >> Yeah. Yeah. Exactly. And we don't want the you know the the whole time, right? You want it to just sort of start and it starts the phase and then it keeps going. And how does the cell regulate that? Well, the cyclones show up as molecules and then they get degraded. So they're only transient, right? >> Okay. The key here, and it'll make sense later, but the key for them was to find

19:53a cycl, a particular sort of um factor here that is like starting the cell cycle that is essential for one part of the cell, but if it sticks around for too long, then it's going to completely destroy the cell cycle. >> Interesting. >> Right. With the with the starter that you have on the grill, it's going to be annoying. The sound is going to be annoying, but the grill is still going to work. Yes, >> we need to find a type of starter molecule that if it sticks around is going to be deadly >> for later. And what they found is a cycl CLB5. What this thing does is promote DNA replication. So essential, right? This is the thing that starts DNA replication for the cell. But it

20:34naturally gets degraded and if it's still present, it's going to lead to cell death. >> Okay? >> Because of later on, it's just going to mess with whatever whatever is going on. It's not important like how it does it, but we just know if it sticks around, it's bad. CLB5. Okay.

Dajbog yeast strain — deleting endogenous cyclins to force control

20:48So, they created a strain of yeast. The the model organism here is yeast. They they created a strain of yeast called Dajb. This is named after the Slavic sun deity, the god of the sun. And he is in charge of the day and night cycle. Okay. We're going to learn about why it's called day and night cycle because what we're trying to do is use light >> in this entire process. So, that's why they called it dashbog. Okay. Okay. >> Here's the trap that they're setting. When they made dagebog, which is this particular yeast strain, it has end indogenous cyclines, right? That that are there because in it wants to replicate. They deleted them all.

21:30>> Okay. >> Okay. So now this yeast strain is completely dependent on what the researchers give it. >> I see. >> Okay. We've held it hostage and now the starter thing, it doesn't have its own starter. >> We are going to provide the starter. So instead of the grill having the button on it where you can click it and it has the igniter that starts the fire, we remove the battery from that on the grill so the it can't do it. And we're bringing in one of those long lighters ourselves and pointing it but we control when it's going in there. >> Exly. Exactly. And now that lighter, that critical cycl CLB5 that we have, we're going to place it under the control of the protein of interest, the

22:12one that we want to evolve. Yes. Yes. >> Okay. And that protein of interest is going to act as a transcription factor for that cycl. Why is this important? I told you that that cycl is extremely dangerous if it's on all the time. >> Previously, the problem with directed

Light as the external controller — survival only if the protein toggles correctly

22:27evolution is that we were only getting proteins that were on all the time. >> Now, that's going to cause cell death. >> That is no longer the fittest, >> right? M >> in the evolutionary perspective if this protein is on all the time, >> right? >> The cell's going to die. >> We we've basically created it's it's it's it's like in the movies where there's a scene, we've put a a a collar around the protein. Yeah. >> And if it continues doing the thing we don't want it to do, the collar self-implodes. Yeah. Um, but we basically now have a a means by which to prevent a type of protein we don't want to persist. Yes. From persisting. >> Exactly. Isn't that I mean it's such a

23:09such a neat >> thingy thing >> and it's inserting itself because there's a natural process that's already happening. Yeah. >> And we're basically trying to make a portion of the candidate proteins in the entire population not procreate. >> Yeah. Anytime there's a mutation >> Yes. where this thing turns on and stays on, the cell's going to die. Anytime there's a mutation and this thing turns off and stays off, it's going to die. The only way the cell is going to survive is if my protein of interest, which is the thing that I'm trying to evolve, turns on, creates the cycl, and then turns off to stop creating the cycling. Right? And >> now this is the magic of the light.

23:51>> Okay? >> The survival condition is the following. That protein of interest is has a little chromophore. So, it's sensitive to light. >> Okay. >> I'm going to turn it on with light. It's going to turn on. And then when I turn off the light, it's going to turn off. >> Okay. And if it ever goes against that signal that I'm giving, it's it's done. >> It's done. So, so we have this external we're basically trying to uh have the ability to say to say to control the state >> through the light as an external source of of of influence. >> Yes. >> The the internals are such that we've

24:34constructed it that it will always it should when it responds to our light to our external perturbment it will survive. Yeah. If >> if it remains on when we've turned it off, >> then it'll die. And so so now we've basically created a mechanism by which we externally can control the state of the protein >> via light as the mechanism to do so. Because if it no longer responds to our external probing, >> uh it will no longer survive. Yeah. Meaning through the natural evolutionary process, we will kill off the ones that are not what we want to be there. >> Exactly. Yeah. So now we're only keeping

25:16proteins that are toggling on and off. >> On and off >> and and as long as the light is in sync with the cell cycle, which means, you know, it's like over like so it's going to be like on for an hour, then off for an hour, on for an hour, then off for an hour because that's about it's about takes like two hours for yeast to go through the cell cycle and reproduce. And so that's what we're doing. We can drive this oscillator now >> with light. >> It's like you're, you know, pushing on a swing >> and as long as the swing keeps going, you're good. But if the swing stays up, >> that yeast cell is going to die. >> Can I ask a quick clarifying question?

Why this matters for therapeutic and research applications

25:49Um, you know, >> we we've talked a lot about proteins on the pod before and why they're so important. >> Yeah. Can we just briefly touch on for folks who may not be familiar with why we would even this is even relevant at like having control over the the protein's ability to exist or not exist is a powerful tool. >> Yeah. >> But to what end maybe like or like >> Yeah. Sure. >> Yeah. That kind does that make sense? >> Yeah. So proteins are the workh horses of life. >> Okay. >> We when we think about life we think about like DNA, right? A lot of times we think about genetics. The reason why genetics is important is genetics is the

26:30blueprint for creating the machinery. >> Okay. >> Okay. You can imagine genetics, the DNA is the IP, >> but the protein is the car and the vacuum cleaner and the lights >> and the camera and the computer. >> The DNA is the the uh KFC 13 herbs and spices secret recipe. >> Yeah. The protein is the actual >> is the chicken. >> Is the chicken. Right. And so what we want to do when we're when we're doing this directed evolution here, >> in the KFC analogy, we're changing the recipes to create the best chicken, >> right? We want wings, we want tenders, we want, you know, da da da. >> And so far with the directed evolution,

27:10the way that we've done it is, you know, the toxin's always been there. So this particular machine, the protein, has just always remained on. It'll always eat the toxin. It'll never not eat the toxin, and it's just always going gang busters. Mhm. >> That's not how a lot of proteins in life work. A lot of proteins will toggle between being active and being inactive. And they only get active when there's a certain stimulus. For example, the lactate enzyme. The lactase enzyme which digests lactose only gets on or only really is expressed when there is lactose in the environment. Now there the difference is the protein isn't actually there in the first place until

27:52lactose is there. Lactose then triggers the transcription of that DNA to create the lactase enzyme that then digests lactose. But here we want to create proteins that >> can toggle while being present, right? We don't we don't want to initiate their production and then there's a time delay because the DNA has to be read into RNA which has to be read into this and then become a protein and then the protein goes right. What if we want instantaneous >> reactions from a protein? Well, we want it to be off and then we want it to turn on. That's a good question. Yeah. Okay. But does that make sense? >> Yeah. No. No, it does. And and so the the the the justification for trying to research this area is if we can control

28:36the state of the protein. There are variety of ways in which with that capability we can try to pursue any number of let's say therapeutic outcomes. I mean like for example even with just the old directed evolution right the reason why um the 2018 Nobel Prize was given to this is because there's been tons of research and tons of pharmaceutical uses from just classic directed evolution and that directed evolution was only giving us proteins that always remain on. Now imagine if we can have a dynamic >> protein that comes out of this and later

29:16on we'll get into something that's very very cool >> that's helpful. Thank you for I just wanted to regground really quick. Please continue. >> No, that makes sense. So that's the that's the idea. That's the whole idea is that we're trying to um use light >> in synchrony with the cell cycle. The light is going to drive this protein and as long as this protein can turn on and then off and then on and then off, we're good to go. >> Based on the input of the light being present or not present. >> Exactly. Yeah. So now let's get into some proof of concepts. >> Does this work? Okay.

Proof of concept 1 — shifting EL222 from blue to green light

29:47The first thing we're going to do is spectrally modify the light oxygen voltage domains of a certain protein called EL222. This is a protein that is ubiquitous in research. Okay. Um it's a transcription factor. So what that means is this protein I shine light on it that's going to attach to DNA whatever part of DNA that you want. And then that part of DNA is going to be transcribed into mRNA. and it's going to cause the gene to be expressed. So, we can turn on and off genes using EL222. >> I shine blue light of 450 nanometers

30:30>> and whatever gene of interest is going to turn on. >> Interesting. >> That's an incredible technology already. >> Right. >> Okay. >> Right. >> But it's only one color. >> Okay. What if I wanted to turn on two genes? >> Mhm. >> Right. >> Mhm. Well, you could say, well, I just get an EL222 on gene A and a EL222 on gene B. But then when I turn on blue light, both are going to get expressed. >> What if I want to turn on one and then maybe sometime later I want to turn on the on the other and then I want to turn on both later. I'd need two colors of light, >> right? >> So, so currently we sort of have a very naive way to start the factory. Yeah. >> We can basically say start the whole factory. >> Yeah. >> What we're trying to say is we want to

31:11start the chip department only. >> Yeah. And then maybe a time later >> later we'll get the phone department going because the chips are ready or whatever it might be. >> Exactly. So let's get a little bit into this um this protein here EL222. Okay. A lot of these proteins that are light sensitive use something called a chromophore. Okay. This particular one uses a flavin monucleotide chromoplur. That's the FMN. If you look at the protein here, the bulk is the protein and then that little gray sorry no the the bulk of the protein is in gray. the black little compound there that looks like a nucleotide. Nucleotide is the building block of DNA. In this particular case, that little black molecule needs to be there. That black

31:54molecule is going to absorb the blue light photon, go into a higher energy state. Like in quantum mechanics, just classic quantum mechanics, it's going to go into a higher energy state. The protein is then going to sense that higher energy state and change its shape. >> I see. >> Okay. M and for the longest time that blue photon which is at 450 nanome blue remember is high energy >> which means that that chromophore which is the little triggering molecule requires a high energy photon to bump up to that high quantum state and then that causes the shift in the protein structure. Some random alpha helix goes from one state to another state and then that activates the protein. The

32:35challenge is that chromophore is pretty happy with blue light. >> It doesn't want to accept other nanometer wavelengths, right? So shifting the absorption to let's say a lower energy green light cuz green light there's a bunch of lasers we have with green light. It would be really nice if we could make a chromophore that shifts to green light. >> Okay. >> Okay. >> Okay. >> How do we do that? >> Yeah. >> Right. Because we talked about purple lasers recently. >> Yeah. And >> but that's hard, >> right? Green light is easier because it's lower energy, so it's it's fine, right? Right. Um, so I'd like to be able to create an EL222 that doesn't just respond to blue light. But there's another one. There's another version

33:16that responds to green light. That way, if I want to start the chip chip department, I shine the blue light. Then if I want to ch start the phone department, I shine the green light and I have independent control of which gene I'm going to turn on. We we we want to have more granular ability to turn the factory on to generate the proteins we want to under the circumstances we want to and then be able to also stop that process >> when we want to and then have this control of saying if the factory starts producing the thing we don't want it to do the factory will automatically shut down. >> Yeah. Yeah. And so I want like I want I want multiplexing ability. So here's what here's what they did. They they selected under green orange and red

33:57LEDs. And again, you do the same thing. You do directed evolution. You shine it on green. Those that survived make it through. Those that don't. Oops. >> Mhm. >> What they found was key mutations. They actually were able to create this almost impossible light oxygen voltage transcription factor. The chromophore didn't change. Okay. the little black molecule didn't change. >> The original triggering one we talked about >> didn't change. It turns out there's some intermediate state that that thing has that does respond to green light. It goes into an intermediate quantum state, but the original protein was not sensitive to that change

34:39>> through two changes in amino acids. There was at the 83 at the 83 location T went to A. So the theonine went to alanine. One amino acid was switched for another. And at the 80 location, the glycine was moved to an arginine. Both of those 83 and 80, as you can see,

Proof of concept 2 — red-light control and the “trash collector” insight

34:58they're very close to that black >> molecule. So, they're right next door. >> And just those two changes made the entire protein sensitive to that intermediate quantum >> level that green light triggers. And now on the left, you can see now the the yeast is growing. >> Yeah. Yeah. Yeah. >> Using those >> Yeah. Yeah. >> changes. >> Oh, God. >> Right. And this this is what's kind of interesting here is again you you you come up with the plan, you try it and then you find out oh actually there's this thing that we didn't understand which is this intermediate stage that exists >> and if your fundamentals are good then you can then continue the process

35:40because then you're like okay well now if we manipulate these intermediate stages >> we should still be able to get the results we're looking for. >> Exactly. And now now we have a whole new >> EL222 star I guess >> that is allowing for orthogonal color multiplexing. Right. Now I can have green light. Now I can have blue light. Blue light on green light off or vice versa. >> Right. >> Cuz because the the combination of the two it is is it its own. >> Yeah. Now I can now I can >> control a bunch of genes. Imagine if I could do this over and over again, right? If I keep doing the directed evolution such that the bandwidth of the response on that protein is really narrow. So green light of this

36:21particular frequency is only going to trigger that protein, right? >> Green light of this other, you know, teal versus aquamarine is going to trigger some other thing, right? >> Mhm. >> Cuz with lasers, we can get really really narrow bandwidth. We can be like it is at exactly 450. It's not 451. It's not 452. It is 450 nanometers. Right? So again, like for biomedical research, this is huge. It's allowing multipplexing of genomes. >> Yeah. >> Right. Of gene turning on and off. >> Yeah. Yeah. Yeah. Yeah. And and and to maybe distill it down to kind of a basic concept level, we're creating a controller

37:02for researchers to better drive the process, the research processes they want to around protein synthesis. Is that the right Yeah. And gene expression. >> And gene expression. >> That's the big one. Yeah. Okay. >> Yeah. And protein synthesis is gene expression. >> What is what the drive it's the thing that drives the gene expression. >> Yeah. Exactly. >> Okay. >> Okay. So, breakthrough number two. >> Okay. >> Has to do with trying to use it with red light. Now, with red light and infrared light, you can't use the EL22 because red light is way far. You know, red light is at what, like 800, 700 nanometers, whereas this thing was at 450. So, it's like almost half the energy, >> right? Um, so you don't have any hope in using the same protein. You're going to

37:43use something else called the 5B PIFF3 system. Okay? This is something that uses 660 nanometers to turn on and 740 nanometers to turn off. Okay. >> Why would you want to use red light? Well, red light is very good for deep tissue penetration. >> Yes, >> it's the same reason why sunsets are red. >> Okay? the sunset is red and the sky is blue because blue light scatters more because of rally scattering. And so that's why from every direction in the sky you're getting blue light because the sun is illuminating it with all the colors. But really the shorter frequencies are the ones that are sorry the the shorter wavelengths the higher frequencies are the ones that are coming to us from all different angles. But

38:23when the sun is setting >> it has to go through a lot of atmosphere. So for the same reason all the shorter wavelengths higher frequencies are getting scattered away from us and the red light is penetrating through. Same thing happens in tissues in biological tissue. If I want to do this kind of you know light directed turning on and turning off of genetic factors and things like that in my biomed research but deeper in the tissue the blue light is only going to get through so much. If I use red light or infrared light that's going to get deeper into the tissue. And so it's very much to my advantage. >> Have you uh seen the uh speaking of red light? Yeah. And getting deeper into the

39:04tissue, the the new masks that have been created in sort of the >> I guess uh beauty and skinare and they're blaring red light at your face >> because it gets deeper into >> deeper into the tissue. Yeah. >> I'm I'm just same thing. >> Same thing. Exactly. Now these these these proteins just like the EL222 that we were discussing with the blue light these all rely on a chromophore which is some kind of molecule that is going to absorb the light that's going to get triggered into a high energy state and then the protein around it is going to be sensitive to that high energy state and then change its configuration and become active. The particular

39:45chromophore here is fico cyanobilin PCB. This thing is hella expensive. Okay. Um a,000 it's $1,400 per two mg. It's also highly unstable and it's not natively produced in yeast and mammals, right? So in order to actually express it in yeast, I have to give it this >> chromophore and I have to give it the red light. It's just it's just really expensive and it's got to be like timely right? >> Because it degrades really fast. So the campaign that we want to do with this opto evolution is can we make this thing respond to red light without that chromophor >> without the like additional secondary piece that is required to actually make

40:26the process >> exactly without without PCB. >> Can we get the lobster tail without the caviar? >> Yes, that's exactly right. And so that's exactly what they did. They subjected these dashbog cells to red and far red pulsing and no external PCB. Mhm. >> There were mutants that had loss of function mutations that survived and created and were actually responsive to PCB, >> but >> they had deletions of this particular enzyme, this particular gene called Y1. >> Now, why Y1? Because that Y1 is basically a trash collector enzyme. >> Okay, >> that's weird. Why would a trash collector mezzy >> cause me to be now sensitive to the red

41:08light that I was no longer sensitive for? I don't have the chromophore that I need for the red light. But if the trash collector is not there, all of a sudden I can be responsive to red light. What What does that mean? That means the trash collector was collecting some other molecule that I was using. >> Yeah. Yeah. Yeah. Yeah. >> That I'm now using for the chromophore to be sensitive to red light. >> Right. Right. >> And now that the trash collector is gone, there's some random other molecule that I can now grab and use to be sensitive to red light. Yes. >> Isn't that cool? Yeah, that is very good. The the the trash collector uh was taking away the active ingredient that allowed the response to the red light. When you get rid of the trash collector, the active ingredients then available

41:48>> then available. And that's exactly what they found. They found this active ingredient bivverin >> BV and when you give the yeast cell artificial BV, >> it'll respond to the red light. If you take out the trash collector Y1 gene, it'll be responsive. >> It will also be responsive. >> It'll also be responsive. So now what does that mean? That means that I've the endogenous bilivin >> can actually trigger my red light receptor cells. All I have to do is delete the trash collector. >> Right. Right. >> Enzyme. >> And this this >> and now it's way cheaper. >> It's way cheaper. We This goes back to what this thing you brought up before and continue to bring up which is >> if we can use the body's natural system

42:29to drive the outcome we're looking for versus artificially inserting something, it is >> always better. I don't want to say always, but just in case. >> Yeah, just Yeah, you know what? Yeah, you're you're very correct. In biology, there is never an always. >> Generally speaking, Generally speaking, >> you're right. Yeah, I should. >> It's better. >> Most of the time. >> Most of the time. Yes. Yeah. But biology is a notorious.

Proof of concept 3 — single-protein AND logic gates

42:55>> Well, actually, >> yeah, biology is not like physics that way. >> But that's a really interesting point. Again, this and the tracing of the process to the root to the first principle and the why then allows you to make more calculated choices about how do you want to intervene or or interact with the system >> to to to decrease the external as much as much external disruption as possible. >> Exactly. Exactly. >> This is that's that's and the journey to get there. >> Very clever. >> Yeah. Yeah. And finally, Breakthrough. I think this is going to be your favorite. >> Oh my god. No, there's another one. Yes. >> Okay. So, just a really good So, the break breakthrough number one. So, >> breakthrough number one was we made the

43:35blue light sensitive protein >> sensitive to >> multiple um >> green light >> green light >> which was good because now and also in later we can we can be sensitive to to multiple >> right >> here >> then now we figured out the the the red light trash collector piece >> uh with the the rale scattering. >> Yes. >> And now there's the last one is single protein computers. Okay. >> Okay. >> Right. >> We are now trying to create an andgate. >> Okay. All the students of logic know about the and statement. Right. I need both of my inputs to be >> No, no, no. I need either of my input zero or one to be one and I'll get um

44:18>> am I showing No, I'm showing uh this is this is a orgate. Okay guys, whoever's whoever is looking at the uh um thing, this is like a not and gate, right? Because the A and B if they're one, this the output is zero. >> So this is a NAND gate. >> Yes. Yes. >> Sorry that that's my fault. Okay. But in any case, an ANDgate is when both the inputs are one and that's when the output is one. Otherwise, it's always zero. >> Right. Right. Right. Okay. >> Which the idea is like Yeah. >> This and that are true. That means that means the total thing is true and prosper. Yes. Okay. So historically biological logic circuits we can construct them but usually what they require is a multi-gene cascading

44:59process meaning gene A is on and gene B is on which means that gene C will be on okay so it requires this and that to be on and both of them go and become that I'd like all of this to be in a single protein right okay because that way it's not a slow I don't have to wait for one gene to turn on another gene and another gene >> it's not a sequential process it's becomes just an instantaneous Boom. It's this protein, right? And um it's not also resource intensive because to transcribe one gene, it's a lot of ATP. To transcribe another gene, again, a lot of energy. And so what I want is a single protein advantage. The computation is instantaneous. >> Yeah. >> Right. >> And there's negligible

45:39>> metabolic load. Yeah. >> So what do I do? I use my TT on and T off system. This is something that we have discussed in the um it was the time capsule. >> Oh, one of my favorite. That one was so Yeah, that was that was a crazy episode. >> That was a great If you guys have not watched that episode, I think I put that in my favorites list for last year. >> Very fresh. >> No, that that was this year. Wait, the Oh, I'm thinking about I'm I'm confusing time uh time crystals and the and the >> and this is the time capsule, the honeycomb structure. >> Yeah, that one was also really good. But the idea is with with um RTA, this is this TET on Tet off system. Basically, whenever there's doxycyc

46:19>> Mhm. this um particular >> that's right >> protein is going to get transcribed the gene is going to get transcribed and so what they did was they combined this teton system the RTA with something called a pest deg

46:41rapid destruction okay so here's what I've done I've created um a protein where if I at doxycycan it's going to come on >> yes >> but as soon as it gets transcribed part of it has this degon system to degrade it's like a degrade flag the proteosome is going to come in and degrade it >> okay so it's only going to be transiently active for a very short amount of time okay >> now instead of being a continuous on and off like a continuous thing that is either in an on or off state now you're kind of gating it to a a a window of time that has an explicit start and explicit end. >> Exactly. And so here is why the logic

47:23works because it requires input A which is transcription on from the light that I'm getting. >> Yes. >> And input B. >> Doxycycling needs to be present. >> Okay. Both need to be present and they need to outpace the degradation. >> Okay. If I turn off input >> Mhm. >> then the whatever protein is still there is going to get degraded immediately. So now it's a super time sensitive >> and switch. >> Yes. Yes. Yes. Does that make sense? Yeah, it does. It does. It does. >> It needs the and of the of the light and the doxy if both of And the light I can trigger immediately, right? The doxy can sort of just hang out, >> right? >> But the light I can now trigger

48:03immediately and turn off and as soon as I turn off it gets degraded. >> The the light is our controller. The the doxycyc is a part of the substrate that's necessary to facilitate the environment where we have control. But we don't necessarily as long as the doxy's present. >> Yes. >> That's the only that's the thing that matters. >> Exactly. And and they were able to actually do this. They were able to actually create this and switch in a single protein. Now just imagine right like computers are based on logic on computational logic. And here we're able to do computational logic with a single protein at like the second time scale resolution, right? It

48:44just it just opens up a lot of really cool avenues for biomedical research, for fundamental science research because now if we want to figure out what this particular gene does, we can use this andgate to be like, well, >> what is the logic of this particular gene in relation to the entire logic table of the rest of the >> genome? I is. So would you say there are a lot of ways you could utilize it is one way you could utilize it as basically uh being able to isolate in a very complex biological environment certain things you want to look at and control it so that you can watch the process or the outcomes knowing what the

49:25in like knowing that you're controlling like the inputs into the system. >> Exactly. Yeah. Exactly. And from that we can as I said like create the logic table and that really lets us understand how genes relate to one another how they relate to external stimuli >> right >> like all sorts of stuff right and the idea is with this opto evolution we can now create proteins that have this dynamic behavior >> that need and that are not always one >> this is a logic this is this is one that requires inputs to turn on >> which which means then you could then

Why programmable proteins matter — biology as computation

49:58you know have proteins that only turn on based on an external stimuli that we define. >> Mh. >> And so, right, you know, if you're putting something in the body, you don't want it to be always be on. >> You want it to be on only when you want it to be on. >> And now we kind of have the early makings of a mechanism that we can repeat that is discrete, controllable. Y >> and has also the ability to terminate at the end which is like different you know the we we talked about a story for example where it's you have the halflife of a drug in the body and drugs can have a really long halflife which then goes really and but

50:39>> you can't really >> you can't just decide today we want to turn it off. You kind of just need to let it >> the system dissipate naturally. >> In this case for this particular use case we can say like nope we're done. >> Yeah. Yeah. And and I just I the the possibility of all the stuff that we can do with this is endless, right? So So it's really it's really like like when I say that like like you know right now I don't have exact ideas about how this is going to be used. But that's kind of the beauty of it. It's so it's such a new paradigm to take this oscillator tie it to the cell cycle and make it such that the protein toggles on and off. Right now it's a whole new avenue of directed

51:20evolution that we can now use to create a myriad of different proteins. A space that we didn't even have access to before. >> Right. >> It would it would have would a a an analogy that makes sense be like it's like when we first figured out how to use binary in computing systems. >> Yeah. Like even before the transistor, right, the the vacuum tube was big was huge. >> Right. >> Yeah. >> Right. And we didn't know that we were going to get Uber or or Facebook when we were starting to just be like, "Oh, we can like have an on andoff game." >> Yeah. Uh that that's >> Yeah. Imagine all the logic that we can now put in a cell in a cell. >> In a cell, we we can have the same level

52:00of sort of software style engineering that we have in computers. Again, I know there's a lot of other details in >> Yeah. Yeah. Yeah. This will be more machine language, >> right? But as a as a rough kind of reference point for people to try to understand who might not be who might not grasp the concept as organically uh we are building our own ability to create programs at a biological level. >> Yes. Exactly. >> Like the equivalent of software program. >> It's the it's it's the beginning of that. >> It's the very very early early early early stages. >> Yeah. It's exciting. >> This is very very good. And this was this came out of um I want to make sure

52:41>> that's right echol something something >> echoly technique federal de la there we go >> university deirut beirut not beirut b a y r e u t h >> not the uh anyway uh >> we got to be careful of that so no but but this is and again you know um >> it's so it's very this was very dense. >> Yeah. >> Um but I I think it's again the the level of depth of previous knowledge and work. >> Mhm. That created the base by which

53:21these researchers are now building on top of to try to move away from this previous directed evolutionary approach and trying to find a way to be able to create these control mechanisms within it and the three different breakthroughs. >> It was just it was just proof of concept. They were just doing proof of concept. Imagine what other people are going to come up with. Right. >> Right. Right. And that's another interesting point is like not all people are sometimes like well not all not all science is has a consumer product at the end of it. >> It's just more science >> right >> but imagine the amount of more science we can do with this new tool >> which will ultimately end up in real world yeah >> uh impacts this is very different than

54:02>> uh folks who well I was going to I was going to talk about dark matter which we'll never figure out and we'll never know in our lifetime. So that's one area where we don't have the ability to have real world impacts. Um, fascinating, fascinating story. Uh, we've had a lot of protein stories. I mean, we we talked about the the Yale uh, uh, Lego block story last year. >> Yeah. I mean, at the end of the day, when you get to the molecular level, everything is proteins, right? Right. >> And we have to we have to be we want to

Wrap-up — what this could unlock next

54:32be able to do more with them. Again, we are in our new format on the show multiple times a week. You'll see us pop up in your feed on the podcast feeds if you're on Spotify or Apple on the YouTubes and you'll catch some of our clips on the socials. This gives us more times to get deeper into each story instead of just 30 minutes and rushing through. We can really really dive in and get deeper. We still have the rundown. The rundown will be a standalone. Catch that later this week. And I am your host Lester Nar joined as always by my co-host and our resident protein synthesizer and PhD Krishna Chowdery. We will see you guys for the

55:15rundown.