What's the deal with bioelectricity?

What is bioelectricity all about, and is it actually useful for human applications?

This would not be a biology blog if I did not address the internet’s favorite topic: bioelectricity. My guess is if you’re online at all you’ve heard the word at some point. Very briefly, the field built around studying bioelectricity aims to determine whether there is an underlying “language” driving the genetics-centered view of cell behavior, namely embryogenesis. Put more plainly, bioelectricity is looking to see if messing with voltage gradients across cells can help artificially control tissue regeneration (think organs, limbs etc, like in those weird salamanders).

The axolotl salamander.

I picked this topic because bioelectricity is, by far, my readership’s most requested topic. I have gotten emails about it, Twitter DMs, and even friends in biology have asked me to share my views. As far as I can tell this popularity is owed to Michael Levin at Tufts University, whose work has become rather mainstream in the online world I inhabit. I’ve heard from a number of freshly graduated software engineers that Levin’s work specifically drove them to work on biology, since it is so exciting. But understanding that excitement around bioelectricity requires looking into things far upstream of the Levin lab.

I started writing this essay over a year ago initially because I wanted to see what this whole thing was about. This effort was further incentivized by the modest growth of my blog, and with it a drive to give the People what they want, which, as I said, is to see what bioelectricity means. However, my initial efforts made my head hurt, and I put the whole thing away. That is until recently, when I learned more about attempts at treating brain cancers with electric fields. Though bioelectricity is less about that and more about other things, I revisited the field once again, and I think I have something this time.

What follows is my mental model of bioelectricity and what it means in context of applications to human health and engineering.

Revisiting the textbook

Whether you were aware of it or not, you actually likely learned about bioelectricity in some form when you were in elementary biology. For whatever reason high school or principle level biology classes love to pay some lip service to the idea that ions exist and have some function in a cell. But this is typically, like, 2 pages of a biology textbook, while the other 800 pages will cover every transcription factor you can name. Though I doubt experts in this field would frame it this way, I personally view bioelectricity as principally a challenge to the view that genetics is the thing that matters. We’ll come back to this.

Going back to your high school biology, there is an important concept which is that in most cell types, cells exist in a resting state where the inside of the cell membrane is slightly negative relative to the outside, typically around -70 millivolts. This voltage difference, called the resting membrane potential, is maintained by ion pumps and channels that carefully regulate the flow of sodium, potassium, calcium, and chloride across the membrane.

The most famous example of this machinery in action is the neuron, which I’m sure you will remember from your schooling days. When a neuron fires, voltage-gated ion channels open in rapid succession, allowing sodium to rush in and potassium to rush out, creating the action potential that propagates down the axon in milliseconds. This electrical signaling we’ve understood reasonably well since Hodgkin and Huxley’s Nobel Prize-winning work in the 1950s.

This is what I call “fast bioelectricity,” because it happens in milliseconds. A signal travels down a gradient, and the result is ions release at the end and you feel pain, or contract a muscle.

Fast bioelectricity exhibited in action potentials driving down an axon. (Source)

But here’s where things get confusing, and where the field starts to challenge our gene-centric worldview. Beyond neurons, every cell in your body maintains a membrane potential, and these voltages aren’t just sitting there doing nothing. Non-excitable cells (skin cells, gut cells, cancer cells, and most importantly embryonic cells) use their bioelectric states to communicate, make decisions about growth and form, and coordinate collective behavior over hours, days, and weeks. This is “slow bioelectricity,“ which I term such because the effects take a lot longer to become visible, generally happening over minutes to hours. The majority of this blog post is dedicated to this kind of bioelectricity, which I will simply refer to as “bioelectricity” going forward.

Current as a driver of cell states

Bioelectricity, as it’s studied today can be best understood by looking at a 2011 paper by Tufts biologist Laura Vandenberg. The question the paper is trying to answer is pretty simple: how does a clump of cells know how to build a face?

Believe it or not, we don’t really have good answers. You may have heard of some things that point directionally at answers, namely the developmentally-critical Hox genes that are important for planning (Hox-mutants make some funky creatures), but this doesn’t really explain things fully. Developmental biologists had long assumed this was essentially a genetic story, whereby transcription factors flip on and off and eventually things align roughly in the shape of organisms that are functional. Vandenberg and her colleagues thought something else might be going on a little upstream that drove these genes to turn on and off.

Hox-variants, whereby basic bodymap planning is disrupted, leading to morphologic mutants.

The experiment was pretty simple in concept. They took frog embryos, bathed them in voltage-sensitive fluorescent dyes, and watched the initial stages of embryo development. They filmed what happened as the embryos developed, making time-lapse movies of bioelectrical activity across the surface. What they found was hours before any face structures actually form, a pattern of electrical activity appears on the embryo that looks like a face. Bright spots of hyperpolarized cells (meaning more negatively charged inside) show up exactly where the eyes, nose, and mouth will eventually be. They called it the “electric face” (great band name potential). The whole thing unfolds in three waves, each one lining up with different stages of development.

When they disrupted this bioelectrical pattern by blocking a proton pump called V-ATPase the embryos developed with horrifying craniofacial defects. Some of them grew two brains (!!). Others had no jaws, or malformed eyes, or messed up nasal development. And critically, blocking the ion flow also screwed up the expression of a bunch of important patterning genes (Sox9, Pax8, Otx2, and others). Clearly the bioelectric pattern wasn’t just correlated with face development, but rather was driving it way upstream. Really, the voltage pattern was telling the genes what to do.

It is important to really sit with this idea, because it is foundational to the field of bioelectricity. We are trained to think of electricity as transmission (A calls B), but in this field, you have to think of electricity as memory (A and B are holding a state).

To understand the “language” of slow bioelectricity, you have to stop thinking about a “telephone wire” and start thinking about a Lite-Brite (or a pixelated screen).

Let us take this Lite-Brite analogy and map it on top of the frog embryo.

Imagine a grid of 1,000 cells in a developing frog embryo. Most of these cells are sitting at around -60mV, a state that we will call “blue.” Suddenly, a patch of cells in the middle depolarizes to -20mV, a state which we will call “red.” You now have a red circle on a blue background. To the embryo, that red circle isn’t just a voltage difference but instead it is a coordinate. The cells inside the red circle “know” they are different from their blue neighbors. The bioelectric pattern is spatial. It draws a map on the tissue that essentially says “build an eye here.”

But how do cells know what their neighbors are doing? The answer are the physical tunnels that connect adjacent cells and allow ions and small molecules to flow freely between them, called gap junctions. If Cell A is at -20mV and Cell B is at -60mV, there’s a voltage gradient between them. This gradient physically pushes charged signaling molecules (like serotonin or calcium) from one cell to another through those tunnels, a process called electrophoresis. In other words, voltage directs traffic. It controls where the signals go.

Nano banana representation of the gradient concept. It isn’t perfect, but you get the idea.

The next question is the one that puzzled me for a while: how does a voltage change actually turn on a gene? There are two main mechanisms. The first is the calcium trigger. When a cell’s membrane depolarizes, voltage-gated calcium channels on the surface feel this change and open up. Calcium rushes into the cell, and because calcium is a “second messenger,” it binds to proteins that travel into the nucleus and physically flip the switch on specific genes. In the Vandenberg paper, an example of this is Pax6, the eye-building gene. Voltage change leads to calcium influx leads to gene expression.

The second mechanism works between cells rather than within them. Certain transporters, like the serotonin transporter SERT, are powered by voltage. A voltage gradient across a tissue can force the charged neurotransmitter serotonin to accumulate in specific locations, say, that red circle from earlier. High concentrations of serotonin in that spot then trigger proliferation or differentiation. The voltage pattern becomes a chemical pattern, which eventually becomes a tissue.

Assuming all of the above is true, the logical application of these findings is towards cell regeneration. If it is true that voltage is driving embryogenesis, it is likely also true that the same process can be induced to return “adult” tissues to some embryonic form, whereby they can be directed. That is the basis of the work since, and likely where many people have learned about bioelectricity.

Hacking the Worm

Our embryonic stem cells run out by the time it matters for regenerative purposes in old age, so studying the theory behind bioelectricity in a living organism has required finding a suitable model. To this point, the planarian flatworm has become that model.

Planaria, Masters of Regeneration

If the body of the worm becomes damaged and detached, the segment regenerates to become an entirely new worm, head and all. This is not a one time thing either, you can actually cut up the segments dozens of times, with each tiny piece growing its own head and becoming its own being. Remarkable in its own right, to me the most interesting question here is how pieces of the worm know to grow a head and not, say, another tail. Genetics, at least to me, cannot really explain how some component of memory is stored to retain the mapping of “this end is the head end.”

To explore this, researchers in Michael Levin’s lab at Tufts used pharmacological tools to manipulate the bioelectric state of regenerating worm fragments (Beane et al., 2011). Normally, after you cut a planarian, the wound site at the head end is more depolarized than the tail end, implying that this voltage difference appears to be what tells the tissue which end is which.

First, they asked: what happens if we prevent that depolarization? By blocking an ion pump called H,K-ATPase (another high school biology favorite!!!), they hyperpolarized the anterior blastema, and the worms failed to regenerate heads entirely. The anterior wound became a kind of null state, growing neither head nor tail, just nothing. Then they asked the opposite question: what happens if we force the tail-facing wound to become depolarized like a head wound? Using ivermectin to open chloride channels and depolarize the posterior blastema, they got worms with two heads, one at each end. The mechanism appears to work through calcium signaling: the depolarized membrane opens voltage-gated calcium channels, calcium floods into the cell, and this triggers the downstream gene expression that builds a head. Block those calcium channels, and head regeneration fails even if the voltage is right.

If you’re interested, you can hear Levin talk more about this here.

From their paper: “Wildtype fragments have a membrane potential gradient with the head blastema most depolarized and the tail blastema least depolarized. SCH-mediated H,K-ATPase inhibition results in relative hyperpolarization and ivermectin treatment in relative depolarization. Hyperpolarized blastemas become either tail or headless, while depolarized blastemas always result in head formation. IVM=ivermectin, hpa=hours post amputation, dpa=days post amputation.”

I find the “software” versus “hardware” analogy quite tiresome, but I will use it here. Essentially the software represented by the bioelectric pattern in the worm’s cells can be rewritten without touching the underlying hardware (DNA in this case), with visible changes to the organism’s outcome.

Putting these findings in context, what I think this means is that bioelectricity is less so about building structures during development but rather it’s some mechanism for maintaining a kind of memory of what the organism is supposed to look like. The voltage pattern stores information about the body plan in a way that genes do not, at least not alone, anyway. Part of why I think bioelectricity is so confusing is because the phrase itself implies some type of electron movement, like we think of with computing, but it’s actually more like chemicals that have a charge moving from one cell to another. The term “voltage gradientology” seems more fitting, actually. I think that would clear up a lot of confusion.

We’ve now established that voltage gradientology is important for cell behavior, namely as the means through which genes are activated, but we also know that we can’t take ivermectin to regrow lost limbs. What exactly is missing?

The Gap

You don’t need me to tell you this, but humans are a few orders of magnitude more complicated than the lowly worm. Though it would be nice, we can’t flatten human biology to apply principles of bioelectricity more easily, namely because the specifics matter quite a bit.

The first problem is cellular. Planaria are roughly 30% composed of stem cells ready to become whatever tissue is needed. When you cut a planarian, those neoblasts are sitting right there at the wound site, waiting for instructions. Humans are not like this. We used up most of our pluripotent cells during development. By adulthood, our stem cell populations are limited, specialized, and tucked away in specific niches. [EDIT, credit to Metacelsus who corrected me in comments: Unless something went terribly wrong, there are no pluripotent stem cells whatsoever in the adult body. (If there are, you’re likely to get a teratoma.) Gastrulation (about week 2-3 post fertilization in humans) is the last point at which significant quantities of pluripotent stem cells are present.

What you’re thinking of are multipotent stem cells which can differentiate into several, but not all, lineages.] Even if we knew the exact bioelectric pattern that means “rebuild a hand,” we might not have the cellular raw material capable of responding to that instruction. The signal is only half the equation. You also need receivers. Now, it is a separate question of whether we can make more stem cells, but that is beyond the scope of this piece. Sorry.

The second problem is complexity. A planarian is, at the end of the day, a relatively simple animal with one major axis: head to tail. The bioelectric instruction set for regeneration is correspondingly simple, where depolarization at the anterior wound means “make a head.” A human limb is a different beast entirely. It has three axes: proximal-distal (shoulder to fingertip), anterior-posterior (thumb side to pinky side), and dorsal-ventral (back of hand to palm), which are important during embryo development as we come into being. As things progress in utero, we become bones, muscles, tendons, ligaments, nerves, blood vessels, skin, and fat, all arranged in precise spatial relationships. The bioelectric “map” required to specify all of this is presumably orders of magnitude more complex than anything we’ve decoded so far. I’m sure there’s some way for me to use the language analogy here, but again I find it tiresome, so I won’t. Something about “we don’t know the grammar.”

There is some work in humans here, so I don’t want to discount things entirely. As an example, it has been demonstrated that the skin edge is voltaically dead. When you cut your finger, you short-circuit the battery. This creates an immediate, measurable electric field at the wound edge (40-200 mV/mm). This electric field is the signal that tells keratinocytes (skin cells) to migrate. If you apply a drug that kills the electric field, the wound healing slows down or stops, even if all the chemical growth factors are present. All of this is to say, even in adult cells, the principles still apply, but this is just another specific example. Again, a common principle is missing.

The third problem is that we simply don’t know the code yet. The field has cracked specific cases, like frog faces. These are genuine breakthroughs. But as of now these seem more targeted towards an individual example.

What voltage pattern means “index finger”? What combination of membrane potentials specifies “cartilage” versus “bone”? How about a glial cell versus a functional neuron? We don’t know.

And that’s really it. We don’t know. There are people working on this, which is wonderful and I actually do expect something to come out of it eventually, but as best I can tell, it does seem like the field has generated a good corpus of knowledge around frog or worm-specific bioelectric patterns, but nothing that seems strictly universal. Making the jump from those models and into humans will require finding a way to dig into human-relevant patterns, which my guess requires working on embryos. Doing this in an uncontroversial way will likely mean finding some way to non-invasively monitor developing human embryos. If that can happen, I bet the field accelerates very quickly and I imagine regrowing limbs doesn’t sound so crazy. If this field is important to you, maybe consider working on that problem specifically.

So after a while thinking about this topic, I hope you can agree with me that bioelectricity actually isn’t that complicated. What is clear to me now is that the movement of ions across and within cells matters as it pertains to activation of molecular processes. It is also clear that there are specific voltage patterns and ion exchanges that happen that act as the on / off switch for cell states and behavior, and that feels like the thing to anchor on and learn more about.

Hopefully I have convinced enough of you that the field of bioelectricity should be renamed “voltage gradientology.” It would have saved me quite the headache.

I have other things I could say here, like elaborating on methods for monitoring in utero development, but I ultimately wrote this to understand the field and I think I’ve sufficiently laid that out there. Maybe I’ll write more about it in a future post, but I think I’ll call things there.

Thank you for reading, and happy new year.