Scientist Stories: Michael Elowitz, Life at the Single Cell Level

Thank you very much. It's really fun to be here today. So, like many of the— or I guess all of the previous speakers, my background really is in physics, but I find myself now actually a professor of biology even though I never really took any biology. So, I don't know.

Okay.

I don't know if that's a good thing or a bad thing, but what I want to talk about today really centers on, I think, one of the most fundamental ideas in all of biology, and that idea is that all living things, from bacteria to humans, are made up of cells. Okay, this is not a new idea. But what's happened over the last, I would say, I don't know, 10, maybe 20 years, is that suddenly we actually have a very different way of looking at cells, of analyzing cells, of seeing what they do and how they do it. Than we used to have for many years before in biology. And it's given us a very different kind of view of how cells work. And so I'm gonna try and tell you a little bit about some of our work and a little bit of other people's work as well that is helping us understand how cells do the amazing things they do. Because if you think about it, what do cells have to do? They have to regulate their own growth and division. They have to communicate with other cells. They sometimes have to kill other cells when necessary, in case of the immune system. They have to keep track of time Cells can have clocks inside— in them that keep track of like a 24-hour day, for example. They can do all of these amazing things. And it really presents, especially from the point of view of a physicist, but I think from anyone's point of view, this basic question is, how can these little guys do all of these amazing things? And I think the change in perspective we have is that normally when we think of cells, we rarely think of them individually. So if we think about bacteria, you may have seen images like this. This is an image of a bunch of bacteria. Each of these little rod-shaped objects is one E. coli cell, okay? Similarly, when we think of human cells, we probably think about organs. We think of hearts and brains and so forth. And very rarely do we actually think from the point of view of one individual cell in that organ. How is that cell operating? What kind of information is it receiving from its environment? How does it make decisions about what to do and so forth, okay? So, how do you think about E. coli? So, these guys are about a micron in size. And so it's very hard to get into their world and think the way a bacterial cell does. But one thing you can do is try and find a kind of good model for a cell. And so yesterday I was having breakfast and I was thinking about bacterial cells, which are these little rod-shaped things, right? And what's so amazing about them is that they're very small, they're very simple kind of minimalist elegant shapes, but if you were able somehow to look inside of one of these things, what would you actually see? You'd see a very high density of proteins of nucleic acids like DNA and RNA, all in a very, very high-density environment where even the ability of one protein molecule to move over to another location could be much slower than it would be if that same protein molecule were just in a drop of water, let's say. So, the physics inside of a cell is very different. Now, even more important than actually the physical movement of molecules in the cell is the way in which cells do what they do, the way they think, in a way. And I think this has been alluded to before, but I'll just go through it very briefly again, a kind of very schematic idea of how we think about how cells do things. So we think about genes. So I could represent a gene this way. This line here could represent a segment of DNA sequence, right? A particular string of letters, of A, G, Cs, and Ts, different bases. And this part here, which I've labeled in yellow, codes for a given protein. So when this gene is expressed, it means that I'll make a protein which I can also call Y. So this is the gene Y and the protein Y. Now, that's great. That's one gene. Cells have thousands of these genes. So, in many cases, this gene is not just constantly making the protein, it's regulated. There's a protein that binds to the DNA near this gene and turns it on or turns it off. So, for example, there might be another protein X X, and this X may bind to the DNA near Y and cause it to turn on expression of Y. And so, we could draw a link in a way in an abstract circuit diagram and we could say the protein X activates expression of the gene Y, okay? So, this is just one little link. Now, things get more complicated. Maybe the protein X binds to, it sticks specifically to protein Y, and when it does that, it changes the activity of X and maybe makes it more active, okay? So we can think maybe Y in that sense has a positive impact on X. We can draw another link in this kind of circuit diagram, okay? This is all kind of a hypothetical schematic example but it gives you a flavor for what we're doing all the time in biology across all fields. And you could— things could get more complicated. Maybe the protein Y activates its own gene and you have a feedback loop where Y activates itself. Or maybe X then goes and activates some other gene, Z. Okay? And then you would maybe draw another arrow. This is what we're doing all the time across all fields of biology. We're finding out which proteins and genes interact with each other in all kinds of systems ranging from human cells to bacterial cells, and we're drawing these diagrams. And these diagrams in some sense represent the regulatory logic of the cell. They're like circuit diagrams but for chemical circuits that are operating inside cells. And what we really want to understand are what are the principles of circuit design for these kinds of circuits? Just like we have principles of circuit design for electronic circuits that have been great for all the devices we're familiar with, what are the principles of circuit design that apply to these kinds of genetic chemical circuits that operate inside cells?

Okay?

And you might think that these kind— this kind of information would give us everything we need to know about these circuits. It would tell us the logic, how they're— the wiring diagram of how they're wired up. And in some sense it does, but in practice it turns out that these kinds of diagrams are very hard to actually use in many cases to make predictions about how a cell will respond to a given stimulus. It's just very difficult in practice to take this and say exactly what is it going to do, okay? Why is that? Well, I think there's a few basic reasons, okay? First of all, these circuits are dynamic. They don't sit there in a constant state. The concentrations of all of these proteins are changing continually over time, okay? You may have a lot of X one minute, X protein, and then later on very little of it. Y may be coming up going down, etc. It's changing all the time. You never have a cell almost that's really in a stable state where it just sits there doing nothing. The second thing that I'm going to talk a lot about today is in addition to be dynamic, these circuits are inherently noisy. So by noisy, what I mean is that they're subject to stochastic fluctuations. Their behavior may be inherently non-deterministic, okay? And I'm— we're going to talk about this in more detail. It's a very fundamental issue. And finally, these circuits don't look like the yellow one, they look a lot more like this white one. They're very, very complicated. We find many interactions, and the more we look for these interactions, the more we find. Okay, and so sometimes some of that information, which may in a way be correct in the sense that these interactions may exist, these arrows may represent real chemical interactions between particular molecular species, but not all of those interactions may be relevant all of the time. And so a major challenge that we face is trying to get past all of this complexity and try and figure out is there some core circuit that we can understand, some basic simpler module lurking inside of this really complicated-looking web of interactions. Okay. So these are the challenges. And recently, my lab and other labs have kind of come up with various approaches. So the approaches I'll talk about today involve, first of all, this idea of synthetic biology. And that idea is very simple. It says instead of trying to just deconstruct or perturb these very complicated circuits that we find in nature, what if we could simply put together our own very simple circuits and put them into cells and see what they do? Maybe that would give us a kind of bottom-up way to understand and elucidate principles of circuit design by designing, constructing, and analyzing our own circuits from the bottom up. Another approach is it's very important to ask what are these circuits doing in individual cells? So traditionally in biology, you might grow up a large population in a little test tube, you might grow up 10 to the 9, a billion bacterial cells in a test tube. And then you kill them, lyse them, release all their contents, and then analyze biochemically what was happening on average among those cells in a tube before you killed them. Well, what if instead of doing that, we could actually watch these circuits operating dynamically, watch what's going on in each individual cell? Would we learn anything different? And the examples I'll show you, we do.

Okay.

And again, our goal is to kind of get a quantitative understanding of basic principles of circuit design for biological circuits. Okay. So, let me start with one of, I think, the most amazing, at least to me, amazing properties that— but very simple properties that cells have. And that is the ability to keep track of time. Okay. How do you make a clock? Now, we have our own circadian clock. Anybody who flew from another time zone is probably aware of that. Flies have clocks, circadian clocks, and even single-celled bacteria also have clocks. So the clock is not a sophisticated property that only arises in the human lineage. It's something very primitive and very basic and fundamental to life. And so people have analyzed for a long time clocks in human cells, in flies, in cyanobacteria, and other systems. And here's an example from flies. What they realized was that there seemed to be They want to understand how does this clock work, and they realize is that there's a kind of feedback loop where there's proteins that get produced, and then it— they go out of the nucleus of the cell, and it takes them a while to come back into the nucleus, but once they come back in, they're able to turn off their own expression, and that causes the concentration of those proteins to gradually decline over the course of the remaining part of the day until they get to a point where they can no longer turn themselves off, and then they start to get produced again, and you get a new cycle. Okay, so it's a— the basic principle seemed to be a kind of time-delayed negative feedback loop. Something turning itself off, but only after a time delay.

Okay.

So that's how the natural clock works. But it's very hard to understand. It turns out to be quite complicated. And I don't think we totally understand it. And so a number of years ago, when I was a grad student with Stan Leibler, we asked a very simple question, which is, instead of trying to analyze this natural clock in flies, could we actually try and build the simplest possible clock circuit we could imagine and put it into cells. Okay. How would we go about doing that? How would you design a genetic circuit that would keep— that would create a kind of oscillation inside of a cell? That was the problem. And we thought about this game of rock, scissors, paper. I don't know if— I'm sure everyone's played it. What? Spock. Spock? Yeah, Dr. Spock. That's Spock, scissors, paper, right, exactly. Okay, good. And so, the idea of this game, as many of you know, is that scissors can kill paper, paper can kill rock, and rock rock can kill scissors. Now, imagine if you had sort of a population of rocks and scissors and papers, okay? And so if you had a lot of rocks, they would very effectively kill all of the scissors and then all the scissors would go away. And because there's no scissors to kill the papers, the papers might start to proliferate and then they'd start to kill the rocks and the rocks would go away and that would allow the scissors to start building up, okay? You could imagine that this thing might give you a kind of oscillation. Now, it turns out you can make this kind of a system actually out of genes. So, we took 3 genes that were at the time very well understood as genes go. So, this gene is called TetR and it's a repressor. All 3 of these are repressors actually. So, a repressor is a protein that turns off the expression of another gene. So, this one called TetR is found in a genetic cassette that makes some bacteria resistant to the antibiotic tetracycline. So, you know there's a— big issue with antibiotic resistance. So they have little— bacteria have little cassettes that enable them to detect and then respond to antibiotics and allows them to survive in the presence of antibiotics. So that's where this Tet protein came from. And we hooked it up in such a way that it turned off the expression of this repressor, which is called lambda C1. This one comes from a virus, a bacteriophage that I think Rob probably told you about this morning, that infects E. coli. And this is a repressor that normally allow— the bacteriophage, the virus, uses this repressor to keep itself dormant so that it won't come out of the cell. But we're gonna take it out of that context and use it for our own purposes. And so this, for us, is just another repressor. It happens to come from a virus. This one is called LacI. This is a repressor protein that's in a very famous and well-studied system that allows cells to— that allows E. coli to metabolize lactose, the sugar, the milk sugar. And the idea here was to just take these 3 genes and hook them up in such a way that one represses the other, represses the next, and represses the first one. These blunt arrows, by the way, I should say, this is biology language to mean represses. So if I draw one of these sort of square arrows, it means that, in this case, TetR turns off the expression of lambda C1. Okay, so what would you expect this to do? Well, actually, it's very hard to reason about this. I sort of gave you a rationale for thinking that maybe there would be oscillations, but you might also think that maybe the system all these proteins would just go to some constant level of expression and they'd all make a little bit and nothing would change with time. And so you can do a mathematical model of this system, and when you solve the model, you can see these nice oscillations where you get first the level of one protein building up and then the level of the next and so on, okay? And it turns out that this is pretty robust as long as a few conditions are met, which we can meet experimentally, okay? So theoretically, we actually expect this system to give us pretty good oscillations inside cells. So the next step was to build it. Here's what it looks like. Here's a kind of map of the genetic system that we built, okay? So what we did is we took a plasmid. A plasmid is a circular piece of DNA that has sequences on it that allow it to replicate itself inside of a cell. So the idea is that if you take this circular piece of DNA that's a plasmid and you put it in a cell and you grow up the cell, the cell keeps growing and dividing, growing and dividing, and the plasmid will grow and divide along with it. And the plasmid will not get lost. It'll keep— It'll stay within the cell just like the chromosome would. And so we made a plasmid and the plasmid has each of these circuit components. So this orange one is the TetR, the red one is the LacI, and the blue one is the lambda C1. And next to each one is a corresponding sequence which binds specifically to the protein that's regulating that repressor. So for example, next to this orange one is a red sequence that binds to the LacI repressor. So the idea is the lacI repressor comes over here, turns off the expression of TetR, the TetR protein comes over here, turns off the expression of lambda C1, and so on, okay? And this plasmid also has other sequences on it that allow us to select for it in cells and allow it to maintain itself, like I said, in the population. And we put on one other plasmid into these cells, and that one is what's gonna allow us to see what's going on in these cells, because even if the system was working, How would we even know? Because we need some way to look at it. And in 1994, I guess, a bunch of fluorescent protein genes, or one fluorescent, a green fluorescent protein gene was cloned. So people took the gene that creates this green fluorescent protein in jellyfish, and they took that gene and they stuck it in bacteria and they stuck it in worms, and they showed that it was sufficient to give you a green fluorescent protein that you could see in any organism. Okay? And that set off a really— a huge revolution in biology because now we don't just— now we don't have to kill cells to see what's going on and open them up. We actually can just watch the cell in a microscope and by looking at how much of this green protein is there, we can see what's going on. And so, the second component of this circuit is to take this green fluorescent protein gene and put it under the control of one of these elements, in this case, TetR. By the way, interrupt me if anything is confusing. Just feel free to ask a question.

Okay.

Yeah. Actually, the other way. The more TetR— TetR is repressing GFP, so the more TetR, the less. Exactly. Okay. And then what did we want to do? We want to make a movie and see what's going on. And so here's a movie of E. coli growing. It's really fun to watch these things if you've never seen them because it just shows you, you know, over a time scale, each division— cell division here. Is maybe about 45 minutes. And you can really see how these cells just grow and divide, grow and divide, grow and divide exponentially, and how you could very quickly go from one cell to a major life-threatening infection or something, if you imagine this happening in your body. Okay. So— what?

Yeah.

So this is what cells do. So we put them on a nice flat surface to make it easy to make these movies, but you can do it in various ways.

Place.

Now, if we look at that same movie, but now I'm going to show you what it looks like in the GFP channel, the green fluorescent protein. So it's exactly the same colony I just showed you, but we're looking at the green part now. It's a black and white movie, but everything you're seeing is the green. Can you see that? Yeah. So what are the cells doing? You see they're kind of pulsing on? Okay. So we were so excited because at least for myself, I didn't think it was really gonna work. So I was quite surprised. Anyway, so we see this kind of pulsing. These cells turn on the GFP and then the GFP goes away. You get this beautiful pulsing. Now, like I said, I was really excited, but then I showed it to a lot of people. I was like, look at how cool this is. And the trouble is you should never do that, because then people were like, well, look, you said you were gonna make a clock. Look at those. Those cells are totally out of sync. One cell is going really fast, another one's going much slower. If this was a good clock, the whole colony would be oscillating together. So, they were like, you know, "The clock doesn't seem very good." So, now, it turned out— it turns out that people have kept on working on these clocks. And believe it or not, they've actually made some that look much better. So, here's an example from Jeff Hastie's lab at UC San Diego where they made a slightly different design. And here you'll see these guys turning green, you see these bursts of green, and you see much better synchronized now. This is in part because they really optimized the circuit much better than I did in my original one. Oh, then it gets washed away, but— and in part because they used a different circuit architecture, a slightly different design of the circuit, okay? So, what did we learn from this exercise? Well, I'd say a couple of things. It is possible to program behaviors in cells. We can actually design a circuit and put it into a cell and make cells do something fairly complex, in this case oscillate, that they wouldn't be able to do otherwise. But like I said, it led to this troubling question, which is there was nothing in the circuit I designed that said that the cells should all go out of phase so quickly, right? So this led actually to a question that's a fundamental issue for the circuit,, but I think even more so is potentially a fundamental issue for cells more generally, which is why is it that two cells, even two sister cells, that are genetically identical, they're like genetic twins, right? And they're, why is it that they would behave differently? Especially they have exactly the same circuit, right? Why do two cells do different things? Okay, this is a very fundamental problem, you know, biology at many levels. And one idea, is that the reason different cells might do different things is because the components of cells are subject to stochastic fluctuations. So by stochastic, I mean non-deterministic, okay? I mean that things inherently random, okay? Now, coming from physics, inherently random or stochastic processes are actually quite familiar in physics. I think that probably the best example is in quantum mechanics where certain— the outcomes of certain experiments at the level of an individual electron or particle may be inherently non-deterministic. But I was also kind of bothered by that because I found myself explaining the variability in the repressilator and I would say, "Well, it's just because of noise. It's just because of fluctuations." But I didn't totally know how to make that precise. I didn't understand how is it that we can really tell whether this variation is coming from some inherent stochastic fluctuation in the cell, or does it just reflect the fact that different cells, each different cell is different environment for the circuit to operate? And so, if the circuit is operating differently in one cell than another, maybe it's just because those cells are different from each other for other reasons, different microenvironments and so on. How can you really tell, in other words, if the behavior of a genetic system is whether it's stochastic or whether it's deterministic, right? So if you think about what is a deterministic system, you would say, well, you know, it's like a billiard table, an idealized billiard table, right? If I set up two exact copies of that billiard table, put the balls in exactly the same positions, and I hit the cue ball in exactly the same— with exactly the same force from exactly the same angle on the two tables, all the balls should do exactly the same thing on both tables, right? That would be That's the definition of a deterministic system. If you start two copies of the system in the same initial condition, then they should all develop in exactly the same way. So, wouldn't it be great if you could do that with cells? Wouldn't it be great if you could take two cells and set them in exactly the same initial state with every little molecule in exactly the same place? Now, like I said, we can't do that because even if you look in the microscope at two sister cells that have just divided, you can just You can't see that they're different from each other. You can't prepare two identical cells. So how can we tell if a process in the cell is really deterministic or not? Okay? So this is what we would like to do. We can't do it. Okay. But here's what we can do. We can take two identical genes and put them in the same cell. Okay? So it's subtly different, but here the idea is imagine that I have two copies of the same gene. From the point of view of the cell, they both look the same. If the cell is going to turn on one, it should turn on the other because they're this very they're the same sequence. The cell can't tell the difference. Okay, so one way to tell if gene— at least if the process of gene expression is stochastic would be to take two exactly identical genes, except we'll cheat, and we'll make a little point mutation. We'll change one letter of one gene to change its color from green to red, okay? So we'll take a green fluorescent gene and a red fluorescent gene, but to the cell they look exactly indistinguishable. We'll put them under the same kind of regulatory control, and then we'll ask, if you see a cell that's brighter in one color, is it also brighter in the other color? If the gene expression was deterministic, it would be. If it's stochastic, then you might have a cell that's bright in one color but dim in the other color. So we can imagine two extreme regimes. In a deterministic regime, all cells would have the same levels of green and red, so they— in a computer screen, they would all look yellow because they'd have equal levels of green and red. Some might be bright yellow and some might be dim yellow, but they'd all look yellow. If things were very stochastic, we might see some cells that are very green and others are very red. Does that make sense? Okay. So we did this experiment. We put this green and the red gene in E. coli. We made sure— we did all kinds of little technical things to make sure that they were really responding exactly the same way and that they were indistinguishable from the point of view of the cell. And then we just took pictures of them. So this is a picture of just in a sort of normal microscope just to show where the cells are. And here's a picture if you look in one channel, let's say the green. And here's a picture if you look in the other channel. And if I flip back and forth— whoop, whoop, sorry. My flipping's not working. OK. Let's try it again. If I flip back and forth, I think you can see that some cells— oh, what did I do? Sorry, sorry. Some cells are brighter in one channel and some are brighter in the other channel, okay? And actually, one of the nice things with these images is now I can put them into the two channels of the computer, the green and the red channel, and you can really see with this snapshot, you can see how non-deterministic gene expression is. Because the fact that you see all these red and green cells in addition to the yellow cells is really visually just telling you that the process of expressing these genes is noisy. And we can make this more quantitative because we can measure exactly the amount of the— how much green and how much red protein there is in each cell, and we can see how well correlated they are, and we can analyze that more carefully. Okay. And what's interesting is if you do this over all kinds of conditions, you can see that actually under some conditions, the cell can be very deterministic. You can actually make gene expression quite accurate. To do that, you have to crank up the expression very high.

Okay?

But if you're making a lot of protein from a gene, it's actually pretty uniform. These are all yellow. Here though, if you go down by about 30-fold in expression, you can see it's much less deterministic. Now, what's interesting is that the cell is not usually operating in this regime. It's usually operating in the noisy regime, okay? The cell makes, for example, 10 copies of the lac repressor normally, 10 proteins on average. So, there's big fluctuations. From cell to cell in how much lac repressor there is. So, normally the cell seems quite comfortable to operate in a highly variable and noisy way, okay? I should also mention because the previous talk was all about evolution, noise seems to be an evolvable trait. You can make mutations which affect not just how much a gene is expressed, but how variably it's expressed, how noisy the gene is. So, noise is also evolvable. Okay. And this then raised a fundamental question, which is, what is noise doing for the cell? Is it a nuisance? I think that's how most of us usually like to think about it before. We think about it as something that just messes things up because the cell would like to operate very precisely, but instead, it can't quite control the amount of expression of any of its proteins, so it just ends up just being a not very efficient machine, okay? That's one possibility. Another possibility is that maybe this variability is actually useful for something that the cell wants to do. Maybe it's adaptive in the language of biology. So to think about this, we turn to another organism, another bacteria. This is called Bacillus subtilis. These guys come from the soil originally, but we study them in the lab. And in our lab, we love Bacillus. We love it because it's basically the Rob Phillips of bacteria. Okay? It's individualistic and unpredictable. So, unlike E. coli, for example. E. coli is nice. We like E. coli. We live with E. coli. But E. coli, if you take those cells and you put them in various conditions, you usually find that all of the cells respond respond in more or less the same way with some variability like I showed you, but not qualitatively differently. But with Bacillus, you have individual cells doing very different things. That's what this movie is showing. So these are Bacillus cells, and you can see most of the cells are growing and dividing, and they're labeled with this green fluorescent protein, okay? Sometimes they'll turn into these white objects, okay? Those are spores. The spores are these dormant forms of the bacteria that can sit around in the soil. These are very much structurally similar to anthrax spores. They can sit there for 100 years and then later when conditions are good, they can re-germinate and turn back into cells. So that's one thing they're doing here. The other thing that really struck us in these movies is that once in a while you see this. You see a cell suddenly switch into a different state that here is labeled with a red fluorescent protein.. And that other state is what's called the competent state. That's biology lingo meaning that those cells can take up DNA from other cells. So this is a kind of mechanism for bacterial sex, is that DNA can be released by some cells and taken up by other cells and integrated into the genome of those other cells. So it's a mechanism for exchange of DNA. And what's striking to us about this is that, look, it happens very rarely. In this movie, only one cell the cell does it, and it's transient. The cell goes into this state and then it goes out again, okay? So it seems to be, we would say, a probabilistically controlled decision, and it seems to be a transient decision at the same time. So this raises a lot of questions. How can a cell control not how it's gonna respond to something, but the probability with which it's going to respond to something? Might there be a role for noise in allowing this to happen? Those were the kinds of questions we were wondering about. Now, we did a lot of work on this system, and I'll just tell you the end result, which is after making lots and lots of movies where we labeled these cells, we put green fluorescent proteins, red fluorescent proteins, blue fluorescent proteins, all these different colored fluorescent proteins in different parts of the very complicated circuit, and at the end we realized that that very complicated circuit could really be explained by a much simpler simpler circuit, which is shown here, okay? And the idea that we saw in that simpler circuit is that there was one sort of master gene, okay, one master to what we call transcription factor, which is a protein like a repressor that controls the expression of other proteins. It's called comK, com for competence and K for competence in German maybe? I don't know. Okay. And comK, like that first gene I showed you, Y, it has the property that It's autoregulatory. It turns on its own gene. So if there's more ComK protein, then that gene will make even more ComK protein. It's like an autocatalytic positive feedback system, okay? There's a positive feedback loop. But we also saw that there's a negative feedback loop where ComK can actually turn off another gene which is called ComS, okay? Now S is interesting because the ComS protein, which is shown here, basically stabilizes COMK. So the COMK protein, the master gene, is normally being degraded all the time by the cell. So the cell's producing it and then destroying it at the same time. It's a kind of futile cycle. Why is it doing that? And the thing that's destroying it is this MecA complex here, which is a protease. That's a protein that destroys other proteins. And COMS blocks that. It prevents the protease from killing the COMK. And this turns out to be an indirect negative feedback because if you have more ComK protein, that leads to less ComS protein, which reduces the inhibition of this protease, so it reduces the rate at which ComK is being destroyed, and so it has a net— well, it's supposed to be net negative on the loop. Wait a second, I got myself confused. More ComK, less ComS, more degradation. So more ComK, more degradation. —so it's negative feedback. Okay? So you have a fast positive feedback and a slow negative feedback in this circuit. Okay? Now, what you can actually do—I won't go through this in detail—but you can write down a system of equations which tell you about how you would expect this kind of dual feedback system to behave. So these are just simple differential equations where K, for example, is a variable that refers to the amount of COMK protein in the cell, and S is a variable that refers to the amount of COM-S protein in the cell. And these terms represent these kinds of feedbacks. And when you analyze those equations, you see something very striking, which is they can give you a situation where the cell can sit happily— so let me tell you what this graph is. This graph is the COM-K concentration of COM-K protein on one axis and COM-S protein on the other axis. And don't worry too much about what all these lines are. The main point is this: there is a state that is a stable state. The cell can sit there, and if you perturb the cell away from that state, if you perturb the concentration of the proteins, it just goes right back. Okay? That corresponds to that growing vegetative green state we saw in the movies. But something amazing can happen in this system, which is there's a kind of threshold or a line. And if you were to somehow push the system across that line, if you were to give it a little bit of extra ComK, it wouldn't be able to cross the line going the other way. It can't go back. There's no way backward. You can only go forward after you cross that line, and you go along this black trajectory where you start making more and more ComK. That's that ComK positive feedback loop kicking in, and then that kicks in the negative regulation of ComS, and so ComS levels start to go down, and then ComK levels start to decline, and you return back to that initial steady state. Now, this dynamic has a name in throughout science, which is called excitability, okay? Excitability means you have a system that normally just sits there doing nothing, but if kicked hard enough, and if you kick it a little bit, nothing happens, it still does nothing, but if you kick it over a threshold, then it responds in a big way, much, much bigger than the, in a way, the way out of proportion to the size of the kick. And in particular, the kind of response you get is really independent of the way you kick it as long as you kick it hard enough. Okay? That's excitable system. Now, excitability is very well known in biology actually in at least two contexts. So, I think the classic is neurons. So neurons, as many of you may know, they transmit signals from one neuron to another through what are called action potentials. These are spikes of voltage across the membrane of the neuron that travel like traveling waves along the axon to signal to other neurons.. And the mathematical equations that we use to describe these action potentials have very similar structure to what I just showed you, okay? They're also excitable. The membrane of a neuron is an excitable medium just like this competence system is. Now, there's a very big difference in the kinds of components that are being used here. So, in bacteria, those spikes are spikes of differentiation, and although they went by very fast in that movie, they actually in real life last about 12 hours. Seconds, okay, so they're very, very slow, whereas an action potential in a neuron might take, you know, on the order of milliseconds, okay? So here we have something where you have a very common mathematical structure across two very different kinds of biological systems. Now the other biological system I think that's even more familiar that I think can be described as an excitable system is the toilet. So I think for— this I think is actually a perfect analogy because if you think about a toilet, you have a handle, to do the flush, okay? And if you push the handle very hard or you just push it a little bit, you know, as long as you push it far enough regardless of how hard you hit it, you'll get exactly the same flush if everything is functioning properly, right? You'll get a very stereotyped cycle of activities of new water coming in and old water going out, okay? Now, the other thing you can imagine with the toilet is you have the option of jiggling the handle on a toilet, okay? What would that correspond to? Imagine you have a system that's sort of inherently noisy. So instead of having very precise control over the handle of the toilet, you have your sort of shaking and you have very— you jiggle it, okay? If you're jiggling this thing constantly, what you might find is once in a while, you're going to hit it over the threshold and you'll trigger a flush. Okay. What does that got to do with anything? Okay. Well, possibly nothing. No. We think that that actually is a very good analogy for what's happening in Bacillus. So let's go back to Bacillus. So this plot that I showed you, I have this stable state, right? Now, mathematically, that state is very stable, but in a real cell, there's fluctuations because of these stochastic interactions among genes and stochastic production of proteins all the time. And so the cell can't really ever stay at one fixed point. It's jiggling around here all the time. It's like it's jiggling the handle. And once in a while, it will actually cross the threshold and go around the loop and come back. So in this sense, our hypothesis was that maybe cells are using noise as a way to occasionally trigger a differentiation event. And we wanted to test that. We wanted to test the idea that noise was really required, that it was actually something that was necessary for differentiation. And this was troubling because how do you— to do that, what would you like to do? You'd really like to have a way you could turn off the noise and leave everything else the same? How can you take a cell and keep it exactly the same but just get rid of the noise? I told you noise is sort of an inherent part of the way the cell's behaving. Reduce the temperature? OK, reduce the temperature is a great physics way to do it. OK? Because in physics, that's exactly what you would do. That's actually perfect. The trouble is cells are not often happy at near absolute zero. And in the case of the cell— it just so happens with these guys, if you put them on ice, they already freak out and lyse. Okay, so that's— but it's a good— actually, it's a very good instinct. Anybody else? Yeah? Can you hold them? Hold them? Like squeeze them with a tweezer? Oh, just somehow— well, I guess you could try and pinch them. I mean, I don't think though that would affect fluctuations in the production rates of genes and so forth because it would— you might put a little pinch on them, but it wouldn't give you a reason to make it less fluctuating. It might just change the average behavior a little bit. Reduce the density of cells, you mean? You mean grow them in isolation from each other or you mean— Reduce concentration of everything. Actually, I think what you'd want to do is increase the concentration of everything, right? Because then you'd have higher numbers of molecules and that's actually very close to what we're going to do, okay? So the only problem with that is if you change the concentrations of everything, then you're really changing a fundamental aspect of the system. Because the concentrations really matter. A protein maybe have no effect at one concentration, but a higher concentration may have a big effect. What we'd really like to do, but I think you're on the right track, what we'd really like to do is just change the size of the cell, but keep the concentrations constant. So, this is like, in physics, you might study, if you studied a tiny little droplet of water, you might see big fluctuations, but if you studied a big tub of water, then the relative importance of the fluctuations would be much lower, okay? Okay. So, can you actually do that? Can you actually take the cell, make it bigger, but keep all the concentrates? Now remarkably, this is one of the things I love about bacteria. This is an experiment you can actually do in bacteria because normally bacteria grow and divide like this. You have one cell, it forms a membrane, grows and elongates, becomes two cells, okay? But what we can do is we can take and make a specific mutation in the bacteria that prevents it from growing this septum. It prevents it from disconnecting between cells. But the cell seems to be completely unaware that we've made this mutation. And so locally, it all looks totally normal. It grows into long filaments, but the density of chromosomes, the concentrations of proteins, all of that stuff is exactly the same as it would be in normal cells. So in other words, we can make the cells longer. And now if you think about it, The concentration of everything is the same, but the absolute numbers of molecules have become— are larger. And therefore, if you think about fluctuations in the central limit theorem as generally going like the square root of the size of the system, square root of n effects, for example, this is like if you flip a coin, right? If you flip a coin 10 times, then you might expect, you know, if you counted the number of heads and number of tails, you might expect between 3 and 7 most of the time, something like that. But if you flip the— flip the same coin 1,000 times, you would expect the relative deviation from 50/50 to be much smaller, so you'd be closer to 50%. So exactly the same idea. Now, so here's what we would expect, that the small cells, because they're— oh, sorry, go ahead.

Yeah.

[Speaker:AUDIENCE] I just have a question. There's no question of local concentration of batteries as opposed to— Yeah, it's actually a great question. So might you expect there are little local fluctuations in one part of the cell that somehow don't affect other parts? So what we know about these cells is actually we've measured diffusion of proteins in the cells, and it's very fast compared to the time scales of all of the other things that were happening. So there's kind of, those kinds of effects could happen, but they would be on a faster time scale than things we're looking at. That's a good question. So what we'd expect is that small cells would be really variable. So if we looked at the amount of green fluorescent protein, very different from cell to cell, but long cells should be much more similar to each other. And we should have just less noise, but the same average behavior. Okay, so the question we're gonna ask is something like this. Does being bigger and therefore quieter make you less competent? Okay. And so here's a movie where we tried to do that. So these cells have this mutation and we've just sort of activated it. And so now they're gonna grow into these long filaments. The zoom level on this movie, you'll notice, is much broader wider because they're going to grow very long. And you can see the red is going to label these competent cells. Ooh. Yeah. And you see you get some of the cells turning competent, but much fewer, it turns out, than you would normally. And what's really amazing is if you painstakingly go and measure how frequently a cell becomes competent as a function of how long the cell is at that moment, you see something very striking. Which is the longer the cells get, the less they're able to differentiate. Okay? So we take this as pretty good evidence that noise is not just a nuisance. It's something that cells need in order to differentiate. And it's something that they're using in order to be able to control the probability of doing something rather than just operating in a deterministic way. So it is a mechanism for them to create diversity in what would otherwise be homogeneous population. Okay. So, okay. So what else can noise do for us? The last— let's see, I wonder if I should— 5 minutes? Okay. Let me do— I'll tell you— let me tell you about another thing we've done recently which I think also involves some of the same principles but in a little bit more subtle way, but I think is sort of fun. So, more recently we started looking at a special kind of circuit, genetic circuit. And that is the genetic circuits that allow cells to communicate with each other that are called signaling pathways. And what's really amazing, I think that's becoming more and more clear over the last decades, is that from worms to flies to humans, we see the same signaling pathways, the same kinds of these molecular circuits being used across all of these different species. That are multicellular. And we see the same signaling pathways being used in very different contexts in our body. So the same signaling pathways operate in the brain and operate in the skin and operate in the lungs and so on. And so the question we've been interested in is what's so great about these signaling pathways that makes them so capable of operating in all of these different contexts? And so these signaling pathways have great names. There's the efferent, the hedgehogs, the EGFs, the TGF betas, the Wnt, Notch, calcium, integrin, okay, all of these different things. So I'm gonna tell you about one in particular that we started to look at, and that's the calcium signaling. So in many different systems, calcium ions are used as a signaling molecule. They're released by some cells or released in the cell, and they activate other responses. And the way in which it's thought to work looks on the face of it to be very straightforward. Calcium ions bind to a protein called calcineurin, and that protein is an enzyme that's called a phosphatase, and it has the job of removing phosphate groups from another protein, which here is called Crazy1. That stands for calcineurin-responsive zinc finger, but we just say crazy. And so calcineurin dephosphorylates Crazy1, and it drives it into the nucleus. And so the idea of this pathway is very simple. If you have more calcium, you have more calcineurin activity, and that gives you more Crazy1 in the nucleus. And since Crazy1 is one of these transcription factors, it's a protein that turns on other genes, and it turns on over 100 other genes, that would give you more expression of all of the downstream genes. So it looks very linear and straightforward until you actually look at it in single cells. And so if you try to— what you can do is take Crazy1 and stick GFP to it, so make a new gene which has which is chimeric. So it's Crazy1, and then at the end of Crazy1, it starts to be GFP. And so it makes a protein that has a GFP stuck to Crazy1, okay? And if you look at that, you can imagine what's gonna happen. If there's no calcium, all of that Crazy1-GFP would be in the cytoplasm of the cell, so it would look a little bit dimmer and it would be spread out. But if you have calcium, it would all be driven into the nucleus and you'd have a really bright spot of GFP. Localized nuclear crazy one. So all we did is we took cells that contain this protein and we made movies of them. We put them on a coverslip. It's very simple. And we just watched. And I hope people can see— each of these is a cell, these spots. I think if you look at it, you'll see a little twinkling. Twink, twink, twink, twinkling, twinkling. So follow any individual cell. Tunk, tunk. So this movie that I'm showing in real life goes about 6 hours, so it's sped up a lot. But each of those little twinkles of localization lasts about a minute and a half in real time. Okay? So we were stunned because we thought, you know, just you put in some calcium and you'll get some level of nuclear localization. It'll all be very simple and boring. And here's what it looks like if you actually kind of try and make a plot of it. On the y-axis is the amount of localization, on the x-axis is time, and these are these little spikes of nuclear localization. These are the twinkles that we saw, chun, chun, chun. And you see they're very erratic, they're unsynchronized between neighboring cells, they look very stochastic, okay? And there's one crucial property of this that was really striking to us, and that is as we changed the input, the amount of calcium, we saw no change at all in the duration of these little twinkles of activation, okay? But we saw a big change in how often they occur. So more calcium, a higher frequency of these events. Less calcium, a lower frequency. Okay? So in other words, the system is being controlled by frequency modulation. A higher input gives you a higher frequency, okay? It's a frequency-modulated signaling system. So what's going on? Why in the world would the cell want to control anything in this crazy way, right? Why FM? And so I think there's a few— FM is very familiar in engineering in our daily lives. We have FM radio invented in 1933. And I think the idea there is that the FM, by modulating the frequency of signals rather than the amplitude, you can make your radio— like your FM radio is much less susceptible to noise. You've probably noticed that if you've driven through a tunnel. In neurobiology, neurons change the frequency of action potentials. But here's the main— here's the context which I think is most relevant to the eukaryotic cell, okay? And that is rocket thrusters on spacecraft. So if you're familiar with how people navigate, how people set up these spacecraft to control them, what they don't do is they don't give them a gas pedal and they don't say, "If you want more thrust, give it more gas." That would be the obvious thing to do. What they do instead is they create these engines that basically can be all the way on or all the way off. It's called bang-bang. And what they do is they want more thrust, they leave the engine on more of the time. And I'll show you, if it works, an example. If anybody's played this video game— this is called Lunar Lander. So all I can do— it's a very boring game, actually. I basically can— I need to get— like, if I don't want this thing to crash, I control the fraction of time I better look at it. OK. The fraction of time that I've got the engine on, but I can't control how strongly, how much gas I'm giving the engine, right? And so if I screw up, then that's what happens. OK. OK. So that's what we think the cell is doing. The cell is controlling this engine. Now, why in the world would it do that? So this is the last idea. So here's the idea. This is a transcription factor. It regulates 100 different genes. But just in your head, think of two, okay, A and B. And in general, all of these different genes that are being regulated respond differently to the concentration of Crazy1 in the nucleus. Some may turn on at lower values but not go as high. Others may turn on at higher values but go higher, okay? And if you imagine that this system worked the way you would— like a gas pedal, like a simple AM system, you would say, "Well, at low input, I would be over here, and I'd make a lot of these blue proteins and fewer of the green ones." But if I went to higher values of calcium, I'd be over here, and I'd make more of both proteins, but perhaps I'd make even— now I have more green than blue. So in other words, at low values of the input, I make a lot of the blue but less green, and at high values of the input, I make a different ratio between red and green. So the point here is that the ratio of the expression of the different target genes depends on the constant— the amount of the input. So we say it's uncoordinated because if the input is trying to control all of these genes in concert together in fixed ratios, it's not going to work, okay? But the FM system does work because at low values of the input, I give a few bursts but each burst always gives me the same amount of these blue and green proteins, okay? So now I go to high values of the input, high calcium, then I just get more of the bursts but I don't change the ratio because the ratio of expression of the two genes is really a property of the burst itself, okay? And it doesn't matter how many bursts I do, I'm not going to affect that ratio. And so therefore, this system coordinates genes, okay? So what we think is that cells are using frequency modulation of nuclear localization bursts, pshew, pshew, pshew, as a way to actually coordinately control the expression level of over 100 different target genes to keep them all in the same proportions. And that's something you can actually measure. Each of these graphs is a different gene in yeast, and what we're showing here is that each of them responds in exactly the same way to changes in the input. So they all respond in a very coordinated fashion because of frequency modulation. Okay, so what I told you is something I think with these examples that I hope is sort of a general statement about life at the single-cell level, and that is that it's noisy, dynamic and complex, but it's not inscrutable. If we look at these things and we analyze them quantitatively and with the— try and think about them through mathematical models, we can actually get fundamental insights into what are the principles of their design. It's not something that's just a big mess that we'll never make any sense of. It's something that we really can and are figuring out slowly, but we are figuring it out, and these things can be understood, I think. The other thing I told you at the beginning is that we have another way to go about this entirely, which is in addition to just taking the examples that we found in nature, we can make our own genetic circuits. We can program our own cellular behaviors. And you may have seen this. This is from the New York Times. You may have heard about this synthetic cell. I don't know how long ago, maybe a year ago, that was announced by Craig Venter's team got a lot of news coverage. And what they actually did was they showed that they could synthesize chemically an entire genome of a bacterial cell, and they could put it into a cell and remove that cell's original genome. Now, what does that mean? What it means is that not only can we make a little circuit, genetic circuit like a repressor, we can start to think about making very large and complex circuits that might involve many, many genes or even the entire genome. And so this is a very exciting area. It's called synthetic biology. And the final, final, final thing I'll say about synthetic biology that to me in a way kind of really changes your way of thinking about biology is this. What is biology? Normally I think most of us think of biology as the study of living things that are out there, whether it's plants, bacteria, humans, all of the ecosystems. Right? But for the first time, we can actually start thinking about biology in a new way. So, if you think about it, those systems, as many— even though there's this incredible diversity of life on Earth, that's really a subset of the potential diversity of biological systems, right? You can start to think about all kinds of other biological systems that work with the same kinds of molecules and using the same principles, but perhaps look differently. So, you know, here's a natural organism that we see, but maybe we can start to think about whether other kinds of— what other kinds of organisms are actually possible. I think it's a useful way to think about biology, not as just a set of particular species, but as rules for generating different kinds of species. So I'll stop with that, and I should just thank— there's a lot of people involved in the work that I talked about, and happy to take questions. Thanks. Thank you, Michael. Not only can he do great science, I can testify he can also land the lunar— It's more fun to crash it though. I have a question. Do bacteria age like multicellular? It's a great question. So, you might think that they wouldn't. Why? Because it looks like they're dividing symmetrically. The question, by the way, if anybody hears, do bacteria age? And you might think they wouldn't because they seem to divide symmetrically and it would seem like the two sisters are the same and how would they age? But it was recently shown a few years ago that by tracking individual cells in these little microcolonies as they're growing, that there seems to be an aging effect where the cell that gets— one cell systematically gets, in a way, the old cap, the old pole of the cell. And if you keep tracking that member of the division over time, it actually slows down its growth and eventually stops. So the answer seems to be yes, that they do age.

Yeah. [Speaker:ERIC_BULLOCK] Hello. My name's Eric Bullock. I teach chemistry at Santa Barbara City College. If we go back to the first— you don't have to I'm sorry to change the slide, but if you go back to the first circuit you put into the E. coli, I think, I'm thinking, you know, they didn't all show the green at the same time, wasn't regular. But— and so you said something like, "Is it the microenvironment or is it noise?" What I'm thinking is, isn't that what the noise is, the microenvironment? So you have slightly different access to nutrients and things like that. Is that what noise is?

It's actually— it's a great question. What do we mean by noise? And I think one of the problems is we mean many things that are subtly different from each other. So, for example, that experiment with the two colors that I showed you was really there to discriminate between what we call intrinsic noise, which is noise just in that process of gene expression, from things that might cause variation, or from variations that are upstream of that that involve other proteins that could affect both of them. For example, suppose one cell has more ribosomes than another, then it might give you more expression of both genes, okay? Now, you could ask, well, why did that cell have more ribosomes anyway? Well, maybe that's because of noise in the production and expression of the ribosome genes. So, there's a kind of, A little bit difficult to disentangle, but what we try and do is we sort of draw a kind of conceptual bubble around one part of the system and we ask how much noise is being generated within that part of the system and how much is being— is sort of noise in other components that are outside of that part of the system or is things outside of the cell entirely. And so, it's hard to give a general answer to it, but it's something that you can kind of go and analyze in any particular system.

I'm thinking maybe if you have a system where you have the nutrients flowing, where you can have very accurate microenvironments or, you know, identical microenvironments for all the cells that might do something.

Yeah. So the question is how much of this variability would you still get if you control all the microenvironment as well as you possibly can. And people have now made microfluidic devices where they can keep cells in really a constant flow where they're, you know, they have exactly— they're all exposed to exactly the same, you know, environment and they still do vary in their expression. So it seems to be an inherent part of the cell. All right. Thank you.

Hi. I'm Mark Zagajewski from Lexington High School, Massachusetts. You said that, like, what else can you do with noise? And I'm wondering if you're talking about a system that has one of those little sort of activation points and you're jiggling it around, doesn't that make it more sensitive? Doesn't that like increase the sensitivity to whatever the input is? Isn't that like another effect that noise has? I mean, rather— you sort of said, "Well, occasionally it just kind of go off on its own." But if it's a system that's responding, doesn't, you know, a no-noise system versus a noise system, doesn't it make the system more sensitive? Like, so it kind of contrary to what you'd think noise would do, like make it worse at detecting a signal, it actually could improve the ability to detect a stimulus.

Yeah, so actually that's a great point and it's something that— there's something that's called stochastic resonance in physics which is more or less along the lines that you're saying, that you can actually improve the performance of the system very counterintuitively in some context by adding noise. Now here what we think, I think is also kind of in the spirit of what you're saying, is that it's not just that occasionally it'll do that. There's signals that the cell responds to, and in response to those signals, it might control the average level of expression of one of these genes. So it can modulate the frequency of these events. So it is responding to— Exactly. So the cell can respond, but it can respond in a probabilistic way by controlling its probability of doing something rather than having all the cells do the same thing.

You mentioned that signaling is the same across a wide range of organisms, and I've heard people talk about how one of the biggest events in life was the transition from prokaryotic to eukaryotic. And you— do you get the sense that eukaryotic is almost like a sort of an institutionalized large scale noise, or can you see when it becomes advantageous to revert to a multicellular system, and which physics would be the same across that, which would be modified?

Okay, so I guess your question is sort of what fundamentally changes when you go from prokaryote to eukaryote in terms of your ability— let's see, in terms of the noisiness of the systems, is that what you mean?

Yeah, and in fact, there's a certain degree of differentiation from one cell to another in the single cell. Ultimately morph into sort of an institutionalized differentiation?

Well, so let's see. So one of the fundamental differences— let's see. So eukaryotic cells in some cases could be much larger than bacteria on average, so that's one difference. And you might think that that would make them less noisy, but what you have to remember is that even in a eukaryotic cell, you still have, at the end of the day, just two copies of every chromosome, okay? And so what that means is that a lot of the noise is actually generated because of this very small number of copies of the DNA, and that aspect is preserved in both prokaryotic and eukaryotic organisms. So, from that point of view, I think they're both subject to noise.

I don't know if that totally answers your question. [Speaker:MALE_SPEAKER] Yeah, that almost makes it sound as if by needing a cell to be larger, and less noisy, that almost necessitates some kind of specialization.

You wonder if that carries into that. Yeah, I mean, one example that comes to mind is like, you know, these muscle— there are multinuclear muscle fibers that have many nuclei in them and that are in some sense kind of like these long bacterial cells that we talked about and might be expected to have reduced noise.

Excuse me. Here's a really bizarre question, okay? Several years back when Jurassic Jurassic Park was out, there was a documentary I think on the Discovery Channel or something like that. They're talking about the feasibility of doing this, you know, bringing dinosaurs back. And Bob Bakker, the guy with the long beard, you know him? I think in the University of Hawaii. He's a paleontologist. He says— he was contending that like ostriches, for example, which are descendants of dinosaurs supposedly, have legacy legacy genes which could be turned on possibly and to create a dinosaur from that. You know, these legacy genes exist in all of these descendants from all sorts of animals but, you know, he was specifically talking about dinosaurs. Is that anything that's even remotely feasible and even in the future?

The Jurassic Park problem. And it's a fun project. I think we need more funding. But it's— no.

You gotta get that rich guy with the helicopter.

Right. Right. It's a great— it's like one of those classic questions to me is— to everybody, right? That's why I think that film was so exciting to people is that it just makes you wonder that. In terms of realism, I don't think I really know the answer to that.

It's certainly not obvious to me how to do it. Are there such things as these dormant sort of legacy genes which lie within various organisms that can be turned on in some way?

I guess, as you say that, I guess what I'm wondering is how would you go about doing doing it now, I mean, if you could obtain sequence data from, you know, if you could sequence— there's more and more efforts to sequence ancient DNA now. I don't know if that's well beyond anything anybody could sequence. Like Neanderthals, for example? Well, Neanderthals, but that's a much more recent, much more recent. But, you know, I think if DNA— I guess the question is like if DNA synthesis became really good and cheap and if you actually had information information about what the sequence you wanted is, then would you be able to— I think in the book they talk about this actually, or in the movie, about trying to put it into some other organism in order to— I don't know. I don't think I have anything intelligent to say about it. Yeah. Maybe just one last question. Yeah. Patrick Elbaz, Herzliya High School, Montreal. Just a quick question, more of a technical one. If we wanted to maybe grow some fluorescent bacteria with our students, is there a simple way that we can do it in a high school environment? Oh, yeah. That's actually very easy. So, if you want to just take— you can take dish, you know, like plastic plates and you can make agar, which is what we use to grow the cells on. That's— those are all very cheap, inexpensive components. And you could just pour those plates and then you can get you can get fluorescent bacteria in different colors and you can streak them out onto these plates so that you can see individual colonies. And yeah, you can do all that kind of stuff. What I really meant was if you take regular bacteria, okay, if they culture regular bacteria, can they actually do the technique where they make them fluorescent or do you have to buy them already fluorescent? Oh, no, I think you can buy them You could do that, yeah. You basically would— it's called transformation. And the idea is you'd want to take the cells and you'd want to take the DNA and mix them together, but you'd have to prepare the cells the right way. And then you would want to streak them out and see if you could get, like, make the cells turn green. Yeah, we can talk about it maybe afterwards about how—

what's the easiest way to do that. [INAUDIBLE] —do this every year. It's called Bio-Rad is the company. They also, when I was at UT as well, UTD, University of Texas Dallas, we used to get a lot from them as well in the biology department. It's about $75 for the kit. It's pGLO. You're going to take plasma and GFP. You put it into an E. coli. Arabinose turns it on. It glows. It's a transformation lab. It's very inexpensive. It's failproof. You can put the kit in your summer storeroom, regular temperature. You don't have to put anything in the freezer. And it works every year.

It's wonderful. Thank you. I should pick some up. Yeah. Let's, uh, thank Michael again. Okay.

Thank you very much. So we have one more presentation on Los Cumbres, but before that, I think we should give a big hand to Boris, to— let me name them all so we can— to Boris and to Rob and to Richard and to Michael for some incredible talks. And I think you sense why we're doing so much biology at the KITP and what is different. I'm sure you got the sense of what is different when physicists approach questions in biology and also when they talk about it. I'm sure you'll have noticed the difference between a set of 4 talks on the excitement of biology as given by this group of exceptional scientists and a set of similar talks, say, given by biologists. And it is that special point of view that the people who come to our programs and participate in this bring that we think is special and important. And the excitement that I'm sure was conveyed is why we think it It's crucial. So let's thank them again.