There's a really deep and interesting thing about planaria, but let's talk about head and tail decisions. In most textbooks, what you'll see is they show the worm being cut into thirds. So there's like two cuts and then they draw this like gradient and they say, 'Well, this is how you know where the head and tail is.' That's all well and good, but what they're neglecting to show you is the simpler but much more challenging version of one cut. Because if you make one cut in the middle, the cells on the left side of the cut are going to make a tail. Cells on the right side of that cut are going to make a head. Their positional information is exactly the same. They were neighbors until you came with your scalpel and cut them apart. And just purely logically, you can't tell locally whether you should be a head or a tail. You have to communicate with the rest of the tissue.
I wanted to have one giant eyeball. I wanted the whole thing to be one giant eyeball. So, what I would do is I would go in and inject like an eight-cell embryo. I would inject every single cell like, 'Okay, this thing's going to be a giant eye.' No, you can get multiple eyes, but you will never get a bigger eye. There's something that we haven't cracked yet where they know what the size is. The first two-headed worms were seen around 1903, made by a completely different process. Nobody had thought to recut them for over a hundred years. Why? Because everyone thought it was obvious what would happen.
This week, Michael and Shingu spoke with Dr. Michael Levin about how bioelectric signals guide development, regeneration, and collective cellular decision-making. Levin explains how voltage patterns help tissues determine body structure, why cells communicate like networks, and how manipulating bioelectricity can induce regeneration, alter anatomy, and even create two-headed planarian worms. We also discuss morphogenesis, memory in living systems, and the possibility that biology operates through high-level information processing far beyond genetics alone.
Cool. Please welcome Mike Levin.
It started actually really early. I was interested from an extremely early age in engineering and biology. I spent a lot of time looking at bugs and insects, you know, when I was a kid, and thinking about the difference between the things I was building with electronics and later computers and the living things that put themselves together and have preferences about what happens and all of that. And so for the longest time, I was kind of interested in this big question of how minds exist in the physical universe, how they scale, how they change, what happens and all of that. And what I was always looking for is what I now call cognitive glue, which is the policies and mechanisms that take parts and come together to be a whole that knows things and has goals and preferences that the individual parts don't have. And I've always wondered what that was. And we kind of knew what it was in neuroscience, right? So it's the electrophysiology of the neural networks and so on that are critical for those kinds of things. But we didn't really know what it was in development.
And in 1986, I was at the Vancouver World's Fair with my dad and we went into a used bookstore as we did all the time and we found Robert Becker's 'The Body Electric.' It's a book and it's a book that is kind of well known in the field. Now the most interesting thing to me in that book was a bibliography going back decades showing that people were thinking about this already and doing experiments and there's a whole history of developmental bioelectricity and it occurred to me that this was kind of a perfect merger of the things I was interested in because not only does it answer an evolutionary question, which is where did the brain learn its amazing tricks, right? So clearly it was much older than that and was here long before neurons and brains appeared. But also it enabled us to have a kind of entry point to understand how do morphogenetic systems make decisions.
In other words, beyond just the chemistry of the mechanisms, the question that we always ask when we look at embryos, like how does it know? How does it know how many fingers? So we had a standard story that basically all of the parts work together according to the laws of chemistry. Everything is a dumb machine. Nothing actually knows anything, and the resulting outcome is a sort of emergent consequence of what happens. But I and many others were very suspicious of that story and it seemed to me that bioelectricity, by taking inspiration from what we learn in neuroscience about top-down causation and multiscale kinds of decision-making and so on, it seemed like that was a perfect way to address it. And so that's pretty much when my plan really crystallized around what I wanted to do. But the interest goes back much earlier than that.
So this was pre-undergraduate studies. Did you already go into your undergraduate thinking that you wanted to, in high school, that you wanted to combine electronics that you were playing around with with biological questions, or even more so, questions of consciousness potentially?
Yeah, even long before that, as a teenager, I was doing behavioral experiments with worms and tadpoles and things like that, trying to understand how they see their world and things like that. And that became my dream, to sort of advance the science in that direction. I didn't really think it was realistic or possible, but I figured I would do as much as I could and see how far I could push it.
And what was then the first experiment that you designed and did yourself that was in this direction?
Well, as an undergrad, I didn't have access to any serious equipment and things like that. But what I did do was to look at the effects of exogenous fields. So in that case, electromagnetic fields on sea urchin development. So there are two limitations to that method. The first limitation is that this is exogenous stimulation. So whatever you find, you don't know that this has anything to do with how the system normally controls itself. What you found is perhaps a cool engineering trick and maybe you've learned something, no doubt, but it wasn't the same as asking how does the system control its own parts? How do parts come together? How is information integrated in the system? How are decisions made? So external stimulation is not sufficient for that. Also, external stimulation is a blunt tool.
In other words, for sure you can do some amazing things and people have. The kind of my heroes in this field have done incredible work on showing all kinds of phenotypes that made it clear that something important was going on. But the limitation of it is that after when you've applied a field, you don't know exactly what you've done mechanistically. In other words, biological tissue, the impedance is very complex. And so where do the currents go? What happens to the cells? How are they affected and all of that? It's very hard, especially with magnetic fields, because the ELFs go everywhere and so you have no real spatial control over it. Yeah, I wanted to — this was part of my strategy afterwards as a postdoc to develop the first molecular tools to really address endogenous native biological signaling, no external stimulation, no magnets, no magnetic component, just to understand how the system controls itself.
And I suppose it was quite obvious then from the beginning that you would focus on gap junctions and membrane potentials or ion channels. So how to manipulate them, or how did you create your toolbox?
Yeah. Well, it wasn't obvious at all. These were very hard choices because until then most people were studying electric fields and some people were studying ion fluxes. So Lionel Jaffe and Rich Nuccitelli designed this thing called the vibrating probe at Purdue. They designed this thing called the vibrating probe which allowed you to characterize ion fluxes and those were very important, but I really thought that for a number of reasons I thought that what we should be looking at is resting potential voltage gradients, which is quite different from either of those. And so I focused on two things. I focused on ways to characterize the voltage potential non-invasively. Until then you could do it with electrodes, so standard electrophysiology, you poke the cell with little glass needles, but the problem is it's invasive, you only get one cell at a time, if the cell moves or divides you can forget it, and you can't get the whole embryo, right? So what I wanted to do was to look at voltage-sensitive dyes, fluorescent dyes that allow you to non-invasively soak the whole thing and get images and videos of what's going on. So that's the side of reading the information.
And then in terms of writing the information, I wanted to manipulate the way that the cells endogenously create and change that potential, which meant ion channels and so ion channels, ion pumps. And so as a postdoc I began to assemble this toolkit and what I would do is I would contact various neuroscientists in the field and I would ask them for plasmids for their ion channels. So at first I told people why I wanted it. In other words I said I was going to try to misexpress these in other places in the embryo to have control over the resting potential and thus study bioelectricity. I stopped doing that because my boss, Mark Mercola, my postdoc mentor, came in one day and he was laughing and he had a letter in his hand and he said one of the people that I had emailed wrote him a letter and warned him that one of his postdocs was basically dangerously insane and that this was such a crazy idea. I only say that because nowadays it seems pretty obvious, but at the time this was considered completely nuts. Like to look at the kind of neuroscience components outside of the nervous system seemed crazy to most people. So he was warning my boss to be careful that I was...
And this was not in jest. He actually...
No, no, he was 100% serious. He was warning. I mean, I get it as a PI now. I would appreciate the warning. It could be a waste of time or...
Or worse. Sometimes people have nervous breakdowns, you know.
So yeah, he was giving him a heads up and so I stopped telling people why I want these things and I would just ask for them and mostly everybody was nice and sent them out. And so that was the toolkit. So the idea was to learn to manipulate the native interface, the way that the cells control each other, the way that the cells are hacking their own neighbors and the way that the system was collecting information. That is where I wanted to pipe into that system.
And then in practice, once you received these dyes, you would immerse the entire embryo in it and then it would find the locations where it would accumulate and you would read it out through a microscope, or what would be the experiment design?
So there are two, right. So the dyes is the reading component, that's reading the bioelectric code. And actually the first person to do that was my collaborator Ken Robinson, who was also a real hero of mine in the field. He did tons of work in the 70s and 80s and 90s on this stuff. And what we were working together on was bioelectricity of left-right asymmetry in the chick embryo at the time. And so he and his postdoc Thorleif Thorland used the dyes to basically see the first asymmetric voltage gradient in the chicken at the exact stage I told them it would be at because of my predictions about left-right asymmetry and how it would work. And then we used it on frog embryos and everything else. And so yeah, it's microscopy. So it's epifluorescence microscopy. Nowadays the tools are much fancier. There are better dyes, FRET dyes and FLIM and confocal microscopy and super-resolution, all that stuff. But at the time it was just epifluorescence. So that's the reading part.
The writing part is, so when you have a plasmid that encodes an ion channel, the sequence, what we would do is in vitro transcribe it. So you have your RNA and then you inject that RNA. You microinject it with a little pulled glass needle. You microinject it into specific precursor cells of the frog embryo and then so that's one technique and then they would express and you put in a little tracer so you can see which cells exactly got it. That's one way to do it. You can do gain of function by putting in active channels. You can do loss of function by putting in dominant negative channels that will inhibit native channels. You can put in completely heterologous channels. That's the beauty of it. The cells are reading voltage. They don't care how it got there. So we used channels from yeast, you know, pumps from yeast and channels from other species, all in all kinds of heterologous systems. And then of course there are drugs, right? So the pharmacology of ion channel openers, blockers, all that stuff.
What led you to formulate your hypothesis about the left-right asymmetry formation?
So as a grad student with Cliff Tabin, we discovered and characterized the first set of asymmetrically expressed genes. So these were the first genes that are expressed on the left or the right side and they actually, there's a whole cascade where they turn each other on and off and then downstream they allow the different organs to know which side is left and which side is right. And as a physicist, you probably know this is a fascinating problem because at the macroscopic scale the universe doesn't distinguish left from right.
Chirality, like, right, fascinating, right.
And so the question is how do embryos reliably know which side is left and which side is right? You can't get it from genetics because genes don't tell you which side is left and right, right? You need to orient to the outside world. And so...
You need to introduce a symmetry breaking. What is doing that in biology?
Exactly. You need symmetry breaking and then you need amplification because the symmetry breaking will tell each cell that way is left. But that doesn't tell you, but I'm on the right side of a midline, right? So it has to be again, it's this, I was really interested in this because not only is it a cool physics problem, it's also again this like change of scale because it's one thing to know which direction is left and right but it's something else to know where am I relative to the rest of the embryo and that's actually extremely important because all of morphogenesis is this kind of collective intelligence that integrates relative to a large-scale target that individual cells don't know but the collective does.
And so I was very interested. So we found these set of genes but then of course whichever earliest gene you have that is on one side or the other you have to ask well how did that one get expressed on, so what's upstream of that? And so that's what I decided to tackle as a postdoc and I started with a simple but kind of interesting experiment. What I did was I took the early frog embryo and I cut it in half and I cultured the two halves separately and we had by then, because of my earlier work, we had markers that were left-sided markers or right-sided markers. So you could tell what does the tissue think it is? Does it think it's on the left or the right? And what we found out is that if you actually segregate the two halves, they get confused.
In other words, 50% of the time each one would do left side or right side. Actually, there were two interesting things. One is that the tissue itself would get confused, but the cells would never get confused. In other words, you never saw speckles. You never saw, well, I think I'm left and the next cell over thinks I'm right. The whole tissue would be in agreement. And this came up again and again in my work later on, but the tissue was in agreement, it just wasn't correct, you know, half the time it wasn't correct.
The other half as the reference point, so to speak, that you took away.
Yeah, yeah. So what that told us is that you can't, which actually contradicts a mainstream model in the field, which is that individual cells can figure this out because of the chirality of cilia and things like that. I actually think it's got nothing to do with that. It means that small bits of tissue can't uniquely determine left and right. They need to coordinate, right? You need to have communication across, actually across considerable distances, millimeters in that case.
And so I started to think, okay, how would this be? And then that's when the whole gap junction thing came up because I started to think that, again borrowing some ideas from neuroscience and just drawing a very simple, you know, my first model was a very simple electric circuit where I thought, okay, if you have a battery at the ventral end of the embryo and the rest of the gap junctions kind of go into this. So the embryo at that point was a flat, sorry, like a frisbee, like a flat. But there was a primitive streak, like the special region down at the bottom, and then all this stuff around. If what was happening, the battery was driving either current or electrophoresing something else around. That's why you needed the whole thing because if you interrupt it anywhere along, and in fact I showed that if you make single cuts anywhere along the edge you would confuse it. So it had to communicate in that way. And so that was the first model of a battery in the frog, it's the ventral side, and in the chick it's the primitive streak.
And then a gap junction path through which something is going under electrophoretic force.
So bioelectricity basically revealed itself as the answer to a question that you already had, which is the left-right asymmetry.
Yeah, I mean, on the one hand you might think this is tremendously lucky. You know, this idea, I mean, bioelectricity was something I was interested in for a long time. My PhD project was not unusual or weird, you know. I mean it was unique in many ways, but back in those days I didn't talk about any of these ideas, you know. I was sort of taking the time to just learn the mainstream thinking, that I need to learn how to do things at the state-of-the-art level. I could maybe break them later if I stayed in the field long enough. But you know, I was just doing like good, you know, standard molecular biology kinds of things. And it's kind of, it was, you know, kind of tremendous good fortune that that particular project lent itself immediately to applications in bioelectricity. It didn't have to be that way. But of course, the flip side of that is that, you know, many people were studying early development. Most people didn't look for bioelectric kinds of things because they weren't interested, right? So it's, you know, I think it's partially my emphasis and partially luck.
You said it led to applications. What were these applications that the first experiment led to?
Well, the first experiment basically allowed us to design the tools that we then used again and again. So the dyes, the strategies of putting in different ion channels and pumps, the strategies of changing the gap junctional paths because with molecular biology you could make gap junctions that were either opened or closed or had different permeabilities. To use the drugs, I developed drug screens, something called an inverse drug screen to identify in new contexts which channels and pumps might be important, which was very important because natively if you want the mainstream molecular biology, cell biology community to pay attention, they need to know where does the signal come from, what are the genes that are important for it. And I don't think the genes drive this process at all, but they do provide the hardware. So you need to know what genes are important.
And then of course on the other end, having — and that's something else that I developed — is ways to chase it down. So that having said, okay, here is this voltage pattern that's important for this and that. How does it control cell behavior? So what are the cell behaviors that are changed? What are the gene expressions that are changed? Like what else happens, right? So being able to track that. So that was kind of the idea. I thought that the only way that any of this would be successful is to merge it with the explosion of work in cell and molecular developmental biology. And that meant two things. That meant first of all addressing questions that they were interested in. So not just obscure, even if fascinating, phenomena that they could sort of marginalize, but actually like if you want to know why the gene Sonic hedgehog is on this side and not on that side, the molecular explanation is not sufficient. You actually have to go upstream of that and understand the biophysics that allows the whole thing to orient in space.
And the second thing is to tie it to the pathways that we're interested in. So do all of the characterization. What genes are downstream? What genes are upstream? You know, what is the interplay between the physics and the genetics.
Was this controversial at all when you first suggested that there's something upstream of gene control?
Oh, absolutely. This was incredibly, incredibly controversial. It's still in many corners. It is very controversial. Amazingly enough, I mean, it's kind of remarkable. A few months ago, we published a paper in Trends in Genetics where we said things that were basically, you couldn't say those things even recently, even a few years ago. I mean, I'm amazed that we could say them now. But yeah, this was considered very weird.
I would feel bad if I didn't get out of this interview with an understanding of this. So, you have the planarian worm. Okay, you cut off the tail or the head and then you...
Or you cut it into thirds.
Yeah, you cut it into thirds. You cut the middle part or whatever and then you, how do you actually adjust the potentials on the end? Is it just soaking it in the drug or like what's the experimental design that leads to two heads for example?
Right. So there's a couple ways to do it. So planaria, for a number of reasons, a lot of techniques that work, for example, in our frog model don't work in planaria for technical reasons. Their cells are tiny, you can't inject them, there is no embryogenesis really available so you can't inject blastomeres, like there's no transgenics. It's a very difficult system to work with. So what you have left in planaria are drugs.
And the earliest way I generated the two heads was the following. There's a really deep and interesting thing about planaria. So in most textbooks, well, there's many amazing things about planaria, but let's talk about head and tail decisions. In most textbooks, what you'll see is they show the worm being cut into thirds. Okay. So there's like two cuts and then they draw this like gradient and they say, well, this is how you know where the head and tail is. It's because there's a gradient and it's different. Okay.
No, no. They're talking about a diffusion gradient of Wnt protein or something like that, right?
That's all well and good, but what they're neglecting to show you is the simpler but much more challenging version of one cut. Because if you make one cut in the middle, just cut them down the middle. The cells on the left side of the cut are going to make a tail. The cells on the right side of that cut are going to make a head. Their positional information is exactly the same. They were neighbors until you came with your scalpel and cut them apart. What gradient at that point, right? That doesn't do the trick. So that means that purely logically you can't tell locally whether you should be a head or a tail. You have to, because that information will be the same, much like we talked about the left-right asymmetry. You have to communicate with the rest of the tissue. So you have to say, is there already a head here? Am I on the other end? Where am I located, right? And so I started to think, and so we're talking this was like 2003 or so. I was thinking, how is that coordinated? Well, my favorite thing, the gap junction. And so I said, okay, what if we just block gap junctions? And so pharmacologically you can block gap junctions. And the first experiment was just to block the gap junction. And sure enough, and this is the work of Takashi, one of my first postdocs at the time, yeah, that's how we got the two-headed worms.
And something else just to point out there, having worked with them for a few years, 2007-8, we recut them and that's it. You take the two-headed worms and you just cut them into pieces in plain water. No more drug, no more.
The first two-headed worms were seen around 1903, made by a completely different process. Nobody had thought to recut them for over a hundred years. Why? Because everyone thought it was obvious what would happen. Their genetics are unchanged if you just get rid of that ectopic secondary head. Of course, they're going to regenerate normally. Like, why would you even do this? So this goes back to our earlier discussion of thinking about these things in a different way makes you do different experiments, right? And so we said, well, yeah, the genetics are the same, but I didn't really believe that the genetics were storing all this stuff anyway. And so I said, why don't we recut them and see what happens? And it turns out that two-headed worms, if you recut them, they make more two-headed worms. And also when they reproduce naturally, which is to tear themselves in half and regenerate, they also make two-headed worms. So that goes back, like that's the one example I have where it actually does go cross-generational. You know, we know it does that.
So that's the first way we did it. More recently, we were able to do it with specific voltage changes using ionophores. So these are compounds that basically allow specific ions, protons or potassium or whatever, to cross in and out of cells. So you can make changes that... Okay, so in planaria it's pharmacology basically.
So the idea is you block the gap junctions. It doesn't know whether it's supposed to form a head or a tail. And I guess, how does it know in the end whether to make two tails or two heads?
The standard, so there's two things that are going on here. First thing is that by default, if you're a region and you are broadcasting, hey, is there a head here? And nobody is the head, then somebody has to make a head. So you do that, that's the first thing. So in the absence of a head, the default is make a head.
That's the first thing. The second thing is that there's a particular biological pattern and it's not local but it's a large-scale pattern that says how many heads are you supposed to have. And what you can actually see, and the coolest part to me, the coolest part is that you can take a perfectly standard worm, one head, one tail, and you can use the ionophore to change that biological pattern and you can confirm on the voltage dye that it now says two heads, but that worm doesn't have two heads. So that's the most interesting part is that this is not a pattern of what the tissue is now. It's not a pattern that reflects the current anatomy.
It's a counterfactual memory of what I will do if I get injured at a certain time, right? What I think it is, and I think this is quite significant, it's a primitive form of the kind of mental time travel that we're able to do as cognitive beings. So if you're an extremely primitive system, all you know is what's happening right now and you can react to that. But if you are a more sophisticated cognitive system, you can have thoughts about things that happened before. So memories, you can have anticipations about what will happen later. It's information that is not true. You can think about things that are not true right now. It's extremely powerful and they call it mental time travel. So I think this is a very simple version of that because that worm, if you look at it anatomically, one head. If you look at molecular markers, perfectly normal, anterior markers in the head, posterior in the tail, no, you wouldn't know. You wouldn't know anything was wrong until you looked at voltage. And it's kind of like the neural decoding that neuroscientists try to do. You look at the voltage and you go, wait a minute, this thing has a false memory. It thinks that it should have two heads. But you won't know that it's a latent memory until you injure it. As soon as you cut it, now you get two heads.
Yeah. I think that's pretty neat.
And can you impose that instruction after you've cut? Like when you do the thirds and you can just take the middle part and then encode that and it will still...
Absolutely. So you can either do that before or...
I mean there's a time limit because if you wait too long, if you wait past 12 hours, it's already decided what it's going to be and then it's too late. But to address your other point about the heads or tails, we eventually learned, by 2011 we had learned how to go to the tail fate. And so then we could make no-head worms or in fact we could fix two-headed worms and send them back to being one-headed by changing the pattern back.
But can you make eight heads? Can you take like sides of the worm and keep like growing?
You can. There's a — we did a — I think it's a Developmental Biology cover for Junji Morokuma's paper that has a four-headed worm that's the shape of a cross. It has like that. Yeah, you can.
So, okay. So, I have the two-headed worm. I cut it in half and it grows two heads again. How does the cut in the middle know to make a head?
So what happens here, and we're soon going to reach the limit of what we know because there's a lot that we don't know. What happens is that the electric pattern in the tissue rescales. So one of the most interesting things about this is that when you have a particular pattern that is worm-sized, it's sized as the whole worm. You cut the thing into pieces. Each piece then, the pattern rescales in each piece. It's like cutting a magnet basically, but it's much more complex because it's not just two poles. It's not just head and tail. For example, we showed that you can make these planaria develop the heads of other species. So it's not just a dipole, but it's a similar thing in that you can cut it, but the pattern rescales and within hours you'll see that now whatever pattern you had, you now have here.
So, the pattern that was in the head. So, I cut this thing in half. There are two heads. The pattern in the head ends up in the cut. Like...
It's not the pattern in the head. It's a pattern that's...
Like the relative change ends up, it used to be from head to head, then it's now, it cuts, it rescales, cuts ahead. That's crazy.
Yeah. Super, super interesting. And we've been trying to computationally model that.
We have models that do that. So you can make models of electric circuits that work that way, right? Because that's what's happening here. The pattern is, the whole thing is like RAM basically, right? It's a self-sort of maintaining. Once you put in the pattern, it holds. As long as there's energy, the thing holds and it maintains and it rescales and it has some properties. We have computational models that work like that. But testing them specifically in the worm is hard because of the limitations of the planarian model, and those are actually, it's really interesting why they're limited like that.
If you wanted to make a lot more heads, sorry, I'm stuck on that. How many heads would you like? Do you reach a saturation point at some point where one head says, 'Oh, there's a head already next to us, so we're not going to turn it into...'
You do. And you do, and it's very interesting. The size control is very interesting here. We can make ectopic eyes. We've never succeeded in making an eye that's bigger than a normal eye. They know exactly what size they should be. We have not cracked the size control. I tried at first, I wanted to have one giant eyeball. I wanted the whole thing to be one giant eyeball. And if you do that, and this is the fun...
Yeah. So what I would do is I would go in and inject like an eight-cell embryo. I would inject every single cell like, 'Okay, this thing's going to be a giant eye.' No, you can get multiple eyes, but you will never get a bigger eye. There's something that we haven't cracked yet, that's where they know what the size is. And it's very important too because in planaria, for example, this again shows you the difference between the molecular approaches and the bioelectric approaches. There are standard ways of manipulating the Wnt pathway where you get ectopic heads in planaria. When you do that, the head is not the right size compared to the rest of the body. Eventually they might rescale, but it's not the right size because you have to specify the size yourself and the identity, and the Wnt protein or the Wnt signaling apparently only specifies the identity, it doesn't control for the size. The bioelectrics is not like that. When you make a...
When you make a second head via bioelectric control, everything is perfectly matched. The size, you know, everything is perfect because you're at a higher level, right? Because you're not down here where you have to micromanage these other things. You're up at a higher level of control. So the size control is really interesting. Heads absolutely will sort of demarcate themselves and there's a minimal size, and there's something else here which is the standard model in the field says that heads suppress heads. So that once you do have a head, it sends out some sort of a suppressive signal that then suppresses heads somewhere else. But if you cut the head like this, right, if you cut it, you know, this way, this side of the head will regenerate. Well, it'll make two heads if you keep recutting. You have to, it's like the high school experiment. You get a Y. You have to keep recutting. You have to prevent them from joining. But let's say you cut and you get rid of one side. This side will just regenerate the other side. There's no suppression there. It immediately regenerates right next to that head. There's more, you know, head tissue. So it's much more sophisticated than that.
So I remember originally watching talks by you maybe in like 2017 or even earlier. The famous experiment is like the planaria, cutting up the planaria, right? That's the kind of like the aha moment that's easy to communicate. What confused me immediately was like, okay, you have this mass of cells, whether it's a worm or anything else, how can you change the voltages in the cells in a specific region? Or like basically if you're soaking the worm in a drug, I mean, how do you control where the voltages are changing? And the second question is, if you're sensing this with voltage-sensitive dyes, how do you deal with all the artifacts from the dyes or like bleaching or these kinds of things? Like to what degree, how lossy is the whole process of actually trying to make these changes? Because it seems like a very complicated system and readout.
Yeah. Great questions. Okay, let's talk about the dyes first. Yeah, there are about a million controls that you need to do. So for sure, there's bleaching. For sure there are issues with dye availability, so differential dye penetration in different regions. So the way you handle that is to use ratiometric dyes or now FLIM, fluorescence lifetime imaging, which are basically ratios that weigh away some of those issues. There are still plenty, the whole technology of reading voltage non-invasively, especially in deep tissues where the light path isn't, you know, there's a lot to be done in the future. There's a lot of room for development, but even then the tools were good enough and the current tools are good enough to make a lot of progress, but there's plenty of room for development out there. And there are lots of controls that you need to do for sure with the bioelectric changes. So there's a couple of things...
I guess wait, on the dyes, just if we take like a toy example or whatever, if you have the worm, you cut it in half and you want to know, so you soak it in the dyes and they just diffuse into the worm, and then how do you get the ratio? It's like two different dyes and you have a...
You can do it, there's a couple of ways you can do it. And by the way, now we also have genetically encoded voltage reporters. So there are proteins that report voltage that you can do with transgenics, you know.
So it's like from the embryo level, you just put it in and it grows with those reporters?
Or you can knock it in later, you know, you can transfect or something like that, right? So there's a wide range of tools now that you have. And so, yeah, you can collect at two different wavelengths or you can have two dyes. One of my favorite methods that we used to use has two dyes, and one dye is sensitive to depolarization. The other dye either is nonresponsive or goes the other way. And so then you can computationally...
They diffuse together or they're even on the same molecule?
Correct. They can be, or they diffuse together. But again, oftentimes some of these reagents have a wavelength at which they are voltage sensitive and another wavelength at which they're not voltage sensitive. And so then you can ratio away things like penetration artifacts.
And did you then develop these dyes yourself or took whatever the neuroscientists were already working on?
Yeah, we don't develop the dyes ourselves. We're not chemists. We work with people like Evan Miller, who is an amazing chemist, makes these dyes, or the genetically encoded reporters. The thing is, neuroscience sometimes they make dyes, but the thing with neuroscience is that they're mostly interested in very rapid spiking. So what they want are dyes that have very rapid on-off kinetics. We want the exact opposite. We want something that is going to catch all the slow stuff because the things that we're interested in change at the scale of minutes, hours maybe, right? Not milliseconds. So a lot of the dyes that we're interested in are not, in fact, the dyes that most people are using because they're trying to catch the kilohertz kind of spike, right? Yeah.
And is the resolution purely determined by the microscope resolution or by...
It is. And in fact, now we're looking at the nuclear envelope potential. So the nuclear envelope also has a voltage gradient. We're going to have a cool paper on that soon. Oh wow. Yeah, we didn't discover that, it's been known for a long time through electrophysiological evidence, but yeah, the nucleus has a voltage. So yeah, the resolution is just your microscopy. I will however say that the vast majority of what we do with this is not at high magnification. Voltage, as now we know, which we didn't know at the beginning, the bioelectric code is not a cell-level code. So yes, cells read their own voltage, but the real magic of all of this is in large-scale patterns. So most of the time our problem is the opposite. We're not trying to zoom in, and although that's cool too. You can zoom in, like within a single cell, you can see different domains of voltage in the same cell. Like there are lots of, cells don't have one voltage. There's lots of little, it's like a soccer ball of different, you know, like different...
I don't know if it's rafts or what it is, but for us, the biggest thing is to pull back and see the large scale because if you're dealing with a frog leg or a whole planarian or an eye or a frog embryo or something like that, you're not trying to zoom in. You're trying to pull back and say what is the large-scale pattern here. Yeah.
Is this already visible on the single-cell level? Like when you have the single-cell stage of an embryo?
And I guess how, is that just a dipole? Do you just have plus minus on the single-cell level or what?
Yeah, the single cell egg, there's sort of two things going on. There's inside versus outside, which is how you measure the voltage, right? So all around...
Yeah, internal versus external path.
