Jul 9, 2021
Conducting a Cell Orchestra
A conceptual bioelectricity review
You’re the conductor of a prestigious symphony orchestra. With a stroke of your baton, a perfectly con fuoco harmony rings out. You raise your arms, lifting the volume of the orchestra with you, but wait — something is off. The trumpet player at the back sounds like he doesn’t know what he’s doing.
In this metaphorical scenario, each musician is a cell in your body and the conductor is our current healthcare research.
Any problems that occur in an orchestra are caused by the musicians not sounding like what the conductor wants them to sound like. Likewise, any health problems that occur are caused by our cells not doing what we want them to do.
You silence the ensemble to tell the trumpet player what he’s doing wrong and how he’s supposed to sound like. Then, picking up the baton, you start over again, everything sounding just as it’s supposed to.
However, where our current research is going, fixing a cell is not a simple as fixing a musician. Current technology target the level of the biological “machine code” — proteins, genes, cells. We have to first understand every part of the cell (the genome and its products) and then micromanage it by making very specific changes to the genome.
Here’s the thing, the human genome is made up of 3.2 billion base pairs. Only 2% of it actually codes for something (proteins). But we don’t fully understand any of it.
You can imagine why understanding DNA is so complex. We have to make sense of a language so abstract it’s only made up of 4 letters, calculate for nth order effects, temporal-spatial coordination and the many other variables that come into play.
This bottom-up approach is especially challenging when trying to control large-scale 3D structure, like regrowing an arm or repairing vision. Scientists face the “inverse problem”: the difficulty of knowing what low-level intervention to make, in order to achieve a specific change in emergent outcomes.
In this article, we’ll explore how an alternative approach can be used to answer the main question of:
How do systems detect when their shape has been altered by injury or disease, compute what steps to take to restore their correct form, and decide when to stop growing (after their anatomical goal has been reached)?
The structure of an organism isn’t a static state. It’s an ongoing process of individual cells working together to maintain that state. The cells in your body have to work hard every day to make sure that you stay looking like yourself instead of a blob of human stuff.
Specific situations where this occurs is during ageing (as individual cells die but the overall organism continues), regeneration (for example in species that can replace whole limbs, eyes, and hearts), and remodelling (somehow, a tail transplanted onto the torso of a salamander will transform into a limb, a structure more appropriate to its new location).
This is made possible by the communication networks between cells. Cells send each other signals so that even though they can’t see what’s happening on the tissue or organ level, they do the right thing individually to make up an operating cell collective. Communication is 🔑. It’s how 100-people orchestras play together in harmony and how humanity has reached its peak of existence.
While we exchange words, cells exchange ions. They speak the bioelectric language.
Bioelectricity: a dynamic distribution of electrical properties in somatic cell networks which mediates large-scale coordinated information processing in pattern homeostasis, orchestrating cell activity toward large-scale anatomical state.
Pattern homeostasis (or anatomical homeostasis): the ability of the system to activate the necessary sequences of cell behaviours to progressively reduce the error between the current state and the species-specific target morphology.
A target morphology is what a normal [organism] should look like. A human’s target morphology = 1 torso, 2 arms, 2 legs, a face, etc.
Those were a lot of big words so the visual below may be a helpful summary.
Moving on to more big words.
Ions are charged particles. Resting membrane potential (or Vmem) is the difference in charge between the inside and the outside of the cell. If a bunch of positive ions (cations) flow out of the cell, the overall Vmem of the cell will be more negative. This is called repolarization. The opposite of repolarization is depolarization when a bunch of positive ions flow into the cell and the overall Vmem of the cell becomes more positive.
If those words sound familiar to you, you probably heard them when learning about the nervous system. The way that neurons send propagate signals is similar to how all the other cells in our body send signals to each other as well! Our non-neural cells are also yelling at each other all the time. Sometimes these messages can be so loud that even cells halfway across the body can hear them. And as you may have guessed, they don’t communicate through sound, but through ions.
You could say that every cell in your body is a brain, just not capable of consciousness (hopefully).
The really cool aspect of Vmem, is that it controls progression through the cell cycle.
When the cell is hyperpolarized (-), it transitions from G1 to S in the cell cycle and its DNA is replicated. When the cell is depolarized (+), it transitions from G2 to M in the cell cycle, and mitosis, or cell division, occurs.
So what happens when a group of cells is overly depolarized? → increased cell division. And what does increased cell division result in? Cancer. In fact, with voltage-sensitive dye, you can physically see which parts of an organism are more negatively/positively charged.
Voltage-sensitive dye could potentially replace invasive cancer diagnosis methods such as biopsies. It’s already being used to image the brain, so why not use it for the rest of the body?
On the therapeutics side, not only can we detect the charge of a cell, but we can modify it as well! But before we get into that, seeing life through a bioelectric lens requires a change in perspective.
Let’s start with a hypothesis.
Since all babies develop in the same way, to get from embryo → baby our cells just need to follow some a set of instructions aka DNA.
As an embryo, if we messed with its facial features, what would happen? Would it develop into a baby with an abnormal face?
It would be unethical to test this on actual humans so let’s ask our frog friends for a generous donation.
🧑🔬: “Hey, can we use your embryos for a science experiment?”
🐸: “…”
🧑🔬: “Alright, I’ll take that as a yes.”
🌟 Frog embryos acquired!
Let’s move the mouth is up here, make the eyes sideways, displace the jaws, kinda just move everything around… We’ll call the messed-up frog a “Picasso” frog.
We watch a normal frog embryo and a Picasso frog embryo grow and…
They turn out the same?!?!
The Picasso frog corrected its facial features as it was developing. This shows that growth is not a set of specific movements that reliably turn standard tadpoles into standard frogs. That information regarding anatomy is not completely hard-coded in the DNA.
So how do these frog embryos know what a correct frog face looks like? How do they find new ways to grow to reach the correct structure?
Answer: attractor states.
Attractor States
Think about the last time you had some really good food. 🍽
:)
I don’t even know what you’re thinking of right now (though I wish I did). I just told you to think of a memory involving food and your brain filled in the rest.
Incomplete cues or stimuli can retrieve memories because the connections in a brain (or network dynamics) evolve from an initial state towards a more stable attractor state. This attractor state encodes the memory. The concept is visualized in the diagram below.
Imagine rolling a ball down a hill. You’re responsible for the initial push, but you don’t have to guide it all the way. The ball will roll itself the rest of the way down until it reaches a stable state. It will then stay there. Like a potato.
Our bodies are just like that hill and ball. Millions of years of evolution have resulted in our cells being able to build a non-cancerous cell, grow normal limbs, etc. They know exactly how to do it. All we need to do is provide the right push to trigger the correct signalling pathways and reach our goal.
Our cell’s attractor state, or end goal, is its target morphology. Using their built-in computational systems, cells try to minimize the difference between their current state and their stable attractor state (end goal). They execute something like the following steps:
if damaged (current pattern doesn’t resemble encoded target pattern):
issue commands to individual cells to bring overall pattern one step closer to the target morphologyif current pattern = target pattern:
stopelse:
repeat
In the Picasso frog example, instead of starting from its normal place in the attractor basin, its facial features started from all different places. However, since it was still in the same basin, it minimized the errors and got itself back to its normal stable attractor state. Cells don’t just follow a set of instructions to create-an-organism, it’s a very responsive and adaptable process.
With bioelectricity, we can give specific inputs to cells, not control everything they do. Let’s say, we wanted to grow an extra eye. To do that, there are many many steps that’ll need to be taken but we don’t need to worry about them because our cells already know how to complete the steps. They’ve grown not one, but two eyes already! All we need to do is communicate the right inputs to push the cells in the right direction, and they will do the rest of the work by themselves. It’s like calling a pre-existing function buildAnEye() instead of coding everything from scratch.
Going back to the orchestra analogy, the conductor doesn’t know how to play trumpet better than the trumpet player, they just have better pattern recognition and an outside perspective to identify when things don’t sound quite right. Likewise, we don’t know how to do cell stuff better than our cells do, but we can recognize when they’re not acting quite like they’re supposed to.
Here are some more examples to illustrate the differences between modifying bioelectricity and gene editing:
- Gene editing: trying to change the shape of an anthill by changing each individual ant’s behaviour.
Modifying bioelectricity: having the power of Dr. Dolittle and being able to speak to ants to build a different anthill. - Gene editing: regulating the velocity of each molecule in a gas
Modifying bioelectricity: changing the temperature and pressure - Gene editing: changing your phone wallpaper by taking out the motherboard and changing each transistor.
Modifying bioelectricity: tapping some buttons on your screen.
DNA provides the instructions, bioelectricity decides which instructions to follow.
With this perspective, DNA is not the software of the cell anymore, it’s the hardware. The 4 nucleotides adenine (A), cytosine ©, guanine (G), and thymine (T) are used to build cellular machinery, but the bioelectricity controls what machinery is made. It’s like how the code in your phone controls which transistors are on or off.
When bugs are detected or improvements are to be made, your phones undergo a software update. The basic underlying machinery doesn’t change, but the code that controls the machinery does. Similarly, by modifying cells’ bioelectricity, we don’t have to go through the convoluted and risky process of changing the genome, we’re just changing the code that controls the genome.
This makes bioelectric changes easily reversible.
But how are attractor states determined in the first place?
Pattern Formation
In the beginning, you were an embryo, a cluster of undifferentiated stem cells. Each cell had the power to become anything, a knee cell, a brain cell, a pinky cell, any cell at all. Still, all our cells today have virtually the same DNA.
So how is the decision made on how to use this power? They meet with a morphogen, or molecular career counsellor. Morphogens are signalling molecules that control what the future of a cell will be. An important kind of morphogen is transcription factors. They determine which genes will be transcribed and when. These transcribed genes can then regulate the expression of other genes in a cascade of gene regulatory networks. At the end of the cascade are molecules that control cell behaviours such as cell migration and cell adhesion.
A higher concentration of a certain transcription factor in an area could make the cells in that area transcribe the genes to make knee proteins. Another transcription factor could transcribe heart proteins, and so on. Some more well-known mammalian morphogens are retinoic acid, sonic hedgehog (SHH), transforming growth factor beta (TGF-β), and Wnt.
Yes, there’s actually a morphogen called the sonic hedgehog morphogen. If you’re curious, it controls the development of the embryo and the organization of the central nervous system, limbs, digits and many other parts of the body.
Once different parts of the embryo choose a career, the embryo starts to take on pattern. Pattern formation is when cells take on different identities and assemble themselves in the correct place and orientation (to form tissues and organs). Each organism has a normal pattern which is its stable attractor state.
Can we change a cell’s stable attractor state? Spoiler: yes we can. Read on to learn how.
Multistability
Memories can be untrustworthy sometimes.
Me: “My birthday party was in grade 3 so I must’ve been turning 9.”
Mom: “No, you were turning 8.”
Me: >:C
Personally, I’ve had many debates with friends and family about the time and place of the same event. That’s because memories are rewritable. Cell memories (attractor states) are also rewritable.
Meet the planarian, a flatworm that has cracked the code to immortality. When cut in half, sixths, or even two-hundredths, each part of the planarian will rebuild what’s missing for a complete worm. Every single piece knows what a correct planarian looks like, builds the right organs in the right places and stops when all is well.
Normally, when cut in half, the planarian will regenerate into two planarians each with one head and one tail. But, by submerging them in Vmem-changing chemicals, we can trick them into believing that they’re actually supposed to regenerate into a two-headed worm or a two-tailed worm.
Now, if we remove these two-headed worms from the chemicals, place them in water and cut them in half again (poor planarians), what happens? We’d expect that they regenerate back into one-headed and one-tailed worms because we didn’t alter their DNA in any way, but no, they generate back into two-headed worms!
The pattern memory of these planaria has been permanently rewritten (like how we can trick ourselves into believing unfactual memories). We can also write their memories back into being normal worms.
Since there was no genomic editing involved, this shows that the information structure that tells these worms how many heads they’re supposed to have is not in the genome, it’s in an additional layer, the bioelectric layer.
Yet, there’s something even wackier.
Not only can you change attractor states, but you can also create 2 (or 3, or 1000) equally stable states that the planaria can choose from.
In one experiment, scientists found that by putting a bunch of planaria in a octanol (ion channel blocker) solution, 70% of them turned out one-headed and 30% turned out two-headed. However, the 70% weren’t normal. Their bioelectricity was also altered, it just wasn’t shown. Yet, when cut again in plain water, instead of 100% being one-headed, again, 70% were one-headed and 30% were two-headed.
This is because the scientists created 2 bistable attractor states that are equally stable for the planaria.
This phenomenon is similar to the two diagrams below. Which face of the cube is the front face? Is the animal on the bottom a duck or a rabbit? They’re both at the same time but you can only see it as one at a time.
There can even be attractor states that we don’t know about.
A frog embryo normally turns into a green, webbed, fly-gobbling animal (a frog). But, by altering the embryo’s bioelectricity, we can move its current state so far away from its default basin of attraction (embryo → frog), that it instead falls into another basin of attraction entirely.
This is what some scientists found out to be possible with the African clawed flog xenopus laevis. They were able to get xenopus laevis stem cells to go down a non-default development pathway and turn into a brown, small, blobby thing that’s programmable! To pay respect to the frogs these blobby things came from, they were named xenobots.
Xenobots can wander around, do mazes, self-heal and cooperate in groups. They work together to build things out of other loose cells. Check out this amazing article by Nisha Lerdsuwanrut to learn more about how xenobots work.
Although the two-headed planaria and xenobot look completely different from what they were originally supposed to look like, remember that NO changes have been made to their DNA! They are still 100% planaria and 100% frog! All that was changed was the bioelectricity and therefore what decisions the cells made. Makes you wonder if humans also have another attractor state that’s waiting to be discovered. 🤔
Feedback Loops
We’ve established that control of cell functions isn’t linear, but we haven’t fully answered how cells know when to stop (how they determine when the correct morphology is achieved).
Cells are really really good at communicating with each other. In fact, their very existence depends on what messages they receive (they can be told to kill themselves through apoptosis). It is through the taking in of messages and responding with more messages that feedback loops are created.
We experience feedback loops every day. When you’re hungry, you eat food. When you ingest enough food, your body tells you to stop eating. That is an example of a feedback loop.
At the cellular level, an example of a positive feedback loop is the NFκB pathway. It’s turned on by K+ loss, which in turn downregulates transcription of the potassium importer HK-ATPase139 (aka K+ loss prevents more K+ from being imported). An example of a negative feedback loop is the process of depolarization activating the V-ATPase hyperpolarizing pump, which pumps protons out of the cell, making the cell more negative.
If that made no sense to you, essentially when cells reach the correct morphology, it will send signals to other cells to stop growing.
Thus, time is a very important variable that must be taken into account. Bioelectric states are constantly changing. And we can keep track of these temporal dynamics with quantitative mathematical modelling.
Methods
There are a lot of ions in your body. They’re everywhere. It’s like electricity in your computer. What matters more is where ions go. Ultimately, controlling bioelectricity boils down to which ion channels should be turned on and which should be turned off (like transistors in a computer).
Ion channels can be blocked or activated chemically using drugs or genetically using CRISPR. For example, using K+ channel-inhibiting drugs → prevents K+ ions from flowing out of the cell → making the cell more depolarized → initiates wound healing in mono-layers of cells.
Here is a list of well-characterized modulations of membrane potential or ion channel activity, and the effect on cell proliferation:
The challenge with pharmacological inhibitors/activators is that for many ion channels, there are no specific blockers available. In addition, for the ion channels that are broadly expressed throughout the entire body, it’s difficult to target specific locations (to grow an eye in an eye socket and not somewhere else). Advances in CRISPR and other gene-editing technology have allowed for a lot more precise ion channel modulations.
It’s further important to note that we’re focusing on the bioelectric state. Since voltage is the sum of many ion types, the same voltage can be induced by different ion channels as long as the Vmem pattern is correctly established. The specific channels that are modulated don’t matter as much as what the final result is.
Problems cracking the morphogenetic code can solve
Real world applications!
Cancer
When cells turn cancerous, they become like unicellular organisms, treating the rest of the body like its environment so it can do whatever it wants. It’s almost like a terrorist group. However, instead of capital punishment (chemo, radiation, etc.), we can use rehab and therapy (provide patterning cues to the surrounding environment) and cause them to behave like normal cells again.
In the example below, even though the damaged DNA is still there, if we can force the cancerous cells to electrically keep connected to their neighbours and keep receiving signals about what they should be doing in a larger structure, they will not make a tumour.
Regeneration
Regeneration is the most near-term application of bioelectricity. After a traumatic injury, let’s say, a shark bites your arm off. Instead of going through the scarring process, your arm could go through the growing process and become a fully functional arm again!
This was proven possible with frogs. By placing a cuff (bioreactor containing progresterone) on the amputated hind leg of a frog for just 24 hours, it permentantly changed the memory of the frog’s leg cells. In the span of 9 months, instead of becoming a stump, it became a webbed leg again!
Birth defects
Ready for some more frog experiments :)
By injecting a developing frog embryo’s brain with nicotine (a carcinogen), it did not develop to be as advanced as normal frog brains. However, activating the HCN2 ion channel, activated the electric signalling pathway towards normal brain development and learning ability.
By altering the butt bioelectricity of, a tadpole this time, scientists were able to stimulate the eye-growing signalling pathway on the butt of the tadpole. So now the tadpole has an eye on its butt.
But what’s even more insane is that all on its own, the eye started growing nerves toward the brain. So now the tadpole can see out of its butt!
If this is possible then fixing any sort of defect should also be possible.
Ageing
Even though everyone ages and we’ve all just accepted our life trajectory towards impending doom, ageing is a disease. There are a bunch of factors that cause ageing but the overall theme is that our cells start to get tired and not behave like they normally would.
Michael Levin is collaborating with Wyss Institute’s Biostasis Project, to understand and control biological time.
The future
Once a disruptive new technology is discovered, people think that it will be able to solve every problem known to humanity. This is called the forest of resemblances. However, in the same way that life is so amazingly complex, life also has an amazingly lot of things that can go wrong. Cracking the morphological code won’t be able to solve every single problem there is but will bring us a great deal closer to a panacea (if any).
Here is what the future of bioelectricity may look like:
- Diagnoses. Remember the image near the beginning of this article where you could physically see the different resting membrane potential of a tumour vs the area around it? Bioelectricity could provide a non-invasive and precise way of diagnosing illnesses.
- Microdomains. For the entirety of this article, we’ve been referring to bioelectricity as the average of all charges in an area. But, different areas of a cell membrane have different charges. The organelles inside of cells also have multiple charges. Exploring these microdomains could provide us with an even more high-quality understanding of how cells work.
- I’ll let Michael Levin explain the last point:
I’ll jump way forward because it’s more fun that way. Regeneration and limb regeneration is critical, but we can fantasize further than that. In the sort of asymptote of all of this, I see two things that I think should be possible. On the one hand, I think this is progressing towards a total control of growth and form. At some point, when we really know what we’re doing — when we actually know how morphology is handled — you will be able to sit down in front of a kind of a Computer Aided Design (CAD) system, and you will be able to draw whatever living creature you want, in whatever functional anatomy you want. It might be for something with an application here on Earth. It might be an organ for transplantation. It might be a creature that you’re going to use in colonizing some far off world. Whatever it’s going to be, you are going to be able to sit down and specify at the level of anatomy, the structure and function of a living creature at the high level, and then this will sort of compile down and let you build the thing in real life.
How you can also be a cell-whisperer
You can also play around with bioelectricity on your computer!
BETSE, or the BioElectric Tissue Simulation Engine, is an in silico tissue simulator. It can visualize the effects of ion channel changes and voltage gradients.
EDEn, or the Electroceutical Design Environment, is a program to identify drugs that can target user-specified ion flows. Learn more about it here.
Ludobots is a free online course by Josh Bongard, co-creator of the xenobot, to train the basics of evolutionary robotics. You only need to know basic python to do this course! Check out my progress here. After completing ludobots, you can move on to voxcraft, which is the online simulator used by Dr. Bongard and team to design xenobots!
TL;DR
- Bioelectricity is the movement of ions across a cell membrane — changing the resting membrane potential (Vmem)
- A hyperpolarized Vmem transitions cells from G1 to S in the cell cycle and its DNA is replicated. A depolarized Vmem transitions cells from G2 to M in the cell cycle, and mitosis occurs (increased cell growth).
- Anatomy is not hardcoded in the DNA. Cells have an attractor state (ex. anatomical homeostasis) that represents what they’re supposed to look like. They then compute the necessary actions to reach the attractor state.
- Cells can have multiple attractors states and their states can also be changed (ex. double-headed planaria & xenobots).
- DNA is the hardware of the cell, bioelectricity is the software. To change cell behaviour, we don’t necessarily need to edit the DNA, we can simply input some ions, change their attractor state, and the cell will figure out how to reach that state by themselves. It’s similar to calling predetermined functions like buildAnEye() instead of coding everything with 1s and 0s.
- Pattern formation is when cells take on different identities and assemble themselves in the correct place and orientation (to form tissues and organs). Morphogens facilitate pattern formation.
- Cells determine when to stop growing with positive and negative feedback loops.
- Bioelectricity can be changed with ion channel inhibiting/activating drugs or through gene editing ion channels.
- Problems that bioelectricity can help solve: cancer, regeneration, birth defects, ageing.
Thanks for making it to the end of this article! I’m Amy, a 15 y/o synbio researcher who loves all things bio. If you’d like to stay in touch, feel free to subscribe to my monthly newsletter, connect with me on LinkedIn, or follow me on Twitter :) Have an awesome, froggy day!
