Michael Levin The electrical blueprints that orchestrate life TED
Chris Anderson: Mike, welcome.
It’s good to see you.
I’m excited for this conversation.
Michael Levin: Thank you so much.
I’m so happy to be here.
CA: So, most of us have
this mental model in biology
that DNA is a property
of every living thing,
that it is kind of the software
that builds the hardware of our body.
That’s how a lot of us think about this.
That model leaves too many deep mysteries.
Can you share with us
some of those mysteries
and also what tadpoles have to do with it?
ML: Sure. Yeah.
I’d like to give you another
perspective on this problem.
One of the things that DNA does
is specify the hardware of each cell.
So the DNA tells every cell
what proteins it’s supposed to have.
And so when you have
tadpoles, for example,
you see the kind of thing
that most people think is sort of
a progressive unrolling of the genome.
Specific genes turn on and off,
and a tadpole, as it becomes a frog,
has to rearrange its face.
So the eyes, the nostrils, the jaws –
everything has to move.
And one way to think about it
used to be that, well,
you have a sort of hardwired
set of movements
where all of these things move around
and then you get your frog.
But actually, a few years ago,
we found a pretty amazing phenomenon,
which is that if you make
so-called “Picasso frogs” –
these are tadpoles where the jaws
might be off to the side,
the eyes are up here,
the nostrils are moved,
so everything is shifted –
these tadpoles make
largely normal frog faces.
Now, this is amazing,
because all of the organs
start off in abnormal positions,
and yet they still end up making
a pretty good frog face.
And so what it turns out
is that this system,
like many living systems,
is not a hardwired set of movements,
but actually works to reduce the error
between what’s going on now
and what it knows is a correct
frog face configuration.
This kind of decision-making
that involves flexible responses
to new circumstances,
in other contexts,
we would call this intelligence.
And so what we need to understand now
is not only the mechanisms
by which these cells
execute their movements
and gene expression and so on,
but we really have to understand
the information flow:
How do these cells
cooperate with each other
to build something large
and to stop building
when that specific structure is created?
And these kinds of computations,
not just the mechanisms,
but the computations
of anatomical control,
are the future of biology.
CA: And so I guess the traditional model
is that somehow cells are sending
biochemical signals to each other
that allow that development
to happen the smart way.
But you think there is
something else at work.
What is that?
ML: Well, cells certainly do communicate
biochemically and via physical forces,
but there’s something else going on
that’s extremely interesting,
and it’s basically called bioelectricity,
non-neural bioelectricity.
So it turns out that all cells –
not just nerves,
but all cells in your body –
communicate with each other
using electrical signals.
And what you’re seeing here
is a time-lapse video.
For the first time,
we are now able to eavesdrop
on all of the electrical conversations
that the cells are having with each other.
So think about this.
We’re now watching –
This is an early frog embryo.
This is about eight hours
to 10 hours of development.
And the colors are showing you
actual electrical states
that allow you to see
all of the electrical software
that’s running on the genome-defined
cellular hardware.
And so these cells are basically
communicating with each other
who is going to be head,
who is going to be tail,
who is going to be left and right
and make eyes and brain and so on.
And so it is this software
that allows these living systems
to achieve specific goals,
goals such as building an embryo
or regenerating a limb
for animals that do this,
and the ability to see
these electrical conversations
gives us some really
remarkable opportunities
to target or to rewrite
the goals towards which
these living systems are operating.
CA: OK, so this is pretty radical.
Let me see if I understand this.
What you’re saying is that
when an organism starts to develop,
as soon as a cell divides,
electrical signals are shared
between them.
But as you get to, what,
a hundred, a few hundred cells,
that somehow these signals end up forming
essentially like a computer program,
a program that somehow includes
all the information needed
to tell that organism
what its destiny is?
Is that the right way to think about it?
ML: Yes, quite.
Basically, what happens
is that these cells,
by forming electrical networks
much like networks in the brain,
they form electrical networks,
and these networks process information
including pattern memories.
They include representation
of large-scale anatomical structures
where various organs will go,
what the different axes of the animal –
front and back, head and tail –
are going to be,
and these are literally
held in the electrical circuits
across large tissues
in the same way that brains
hold other kinds of memories and learning.
CA: So is this the right way
to think about it?
Because this seems to be such a big shift.
I mean, when I first got a computer,
I was in awe of the people
who could do so-called “machine code,”
like the direct programming
of individual bits in the computer.
That was impossible for most mortals.
To have a chance
of controlling that computer,
you’d have to program in a language,
which was a vastly simpler way
of making big-picture things happen.
And if I understand you right,
what you’re saying is that most of biology
today has sort of taken place
trying to do the equivalent
of machine code programming,
of understanding the biochemical signals
between individual cells,
when, wait a sec, holy crap,
there’s this language going on,
this electrical language,
which, if you could understand that,
that would give us a completely
different set of insights
into how organisms are developing.
Is that metaphor basically right?
ML: Yeah, this is exactly right.
So if you think about the way
programming was done in the ’40s,
in order to get your computer
to do something different,
you would physically
have to shift the wires around.
So you’d have to go in there
and rewire the hardware.
You’d have to interact
with the hardware directly,
and all of your strategies
for manipulating that machine
would be at the level of the hardware.
And the reason we have
this now amazing technology revolution,
information sciences and so on,
is because computer science moved
from a focus on the hardware
on to understanding that if
your hardware is good enough –
and I’m going to tell you that biological
hardware is absolutely good enough –
then you can interact with your system
not by tweaking or rewiring the hardware,
but actually, you can take a step back
and give it stimuli or inputs
the way that you would give
to a reprogrammable computer
and cause the cellular network
to do something completely different
than it would otherwise have done.
So the ability to see
these bioelectrical signals
is giving us an entry point
directly into the software
that guides large-scale anatomy,
which is a very different
approach to medicine
than to rewiring specific pathways
inside of every cell.
CA: And so in many ways,
this is the amazingness of your work
is that you’re starting to crack the code
of these electrical signals,
and you’ve got an amazing
demonstration of this
in these flatworms.
Tell us what’s going on here.
ML: So this is a creature
known as a planarian.
They’re flatworms.
They’re actually quite a complex creature.
They have a true brain,
lots of different organs and so on.
And the amazing thing about these planaria
is that they are highly,
highly regenerative.
So if you cut it into pieces –
in fact, over 200 pieces –
every piece will rebuild
exactly what’s needed
to make a perfect little worm.
So think about that.
This is a system where every single piece
knows exactly what
a correct planarian looks like
and builds the right organs
in the right places and then stops.
And that’s one of the most
amazing things about regeneration.
So what we discovered is that
if you cut it into three pieces
and amputate the head and the tail
and you just take this middle fragment,
which is what you see here,
amazingly, there is an electrical
gradient, head to tail, that’s generated
that tells the piece
where the heads and the tails go
and in fact, how many heads or tails
you’re supposed to have.
So what we learned to do
is to manipulate this electrical gradient,
and the important thing
is that we don’t apply electricity.
What we do instead was we turned
on and off the little transistors –
they’re actual ion channel proteins –
that every cell natively uses
to set up this electrical state.
So now we have ways
to turn them on and off,
and when you do this,
one of the things you can do
is you can shift that circuit
to a state that says no, build two heads,
or in fact, build no heads.
And what you’re seeing here are real worms
that have either two or no heads
that result from this,
because that electrical map
is what the cells are using
to decide what to do.
And so what you’re seeing here
are live two-headed worms.
And, having generated these,
we did a completely crazy experiment.
You take one of these two-headed worms,
and you chop off both heads,
and you leave just
the normal middle fragment.
Now keep in mind, these animals
have not been genomically edited.
There’s absolutely nothing different
about their genomes.
Their genome sequence
is completely wild type.
So you amputate the heads,
you’ve got a nice normal fragment,
and then you ask: In plain water,
what is it going to do?
And, of course, the standard
paradigm would say,
well, if you’ve gotten rid
of this ectopic extra tissue,
the genome is not edited so it should
make a perfectly normal worm.
And the amazing thing is
that it is not what happens.
These worms, when cut again and again,
in the future, in plain water,
they continue to regenerate as two-headed.
Think about this.
The pattern memory to which these animals
will regenerate after damage
has been permanently rewritten.
And in fact, we can now write it back
and send them back to being one-headed
without any genomic editing.
So this right here is telling you
that the information structure
that tells these worms how many heads
they’re supposed to have
is not directly in the genome.
It is in this additional
bioelectric layer.
Probably many other things are as well.
And we now have the ability to rewrite it.
And that, of course,
is the key definition of memory.
It has to be stable, long-term stable,
and it has to be rewritable.
And we are now beginning to crack
this morphogenetic code
to ask how is it that these tissues
store a map of what to do
and how we can go in
and rewrite that map to new outcomes.
CA: I mean, that seems
incredibly compelling evidence
that DNA is just not
controlling the actual final shape
of these organisms,
that there’s this
whole other thing going on,
and, boy, if you could crack that code,
what else could that lead to.
By the way, just looking at these ones.
What is life like
for a two-headed flatworm?
I mean, it seems like
it’s kind of a trade-off.
The good news is you have this amazing
three-dimensional view of the world,
but the bad news is you have
to poop through both of your mouths?
ML: So, the worms have
these little tubes called pharynxes,
and the tubes are sort of
in the middle of the body,
and they excrete through that.
These animals are perfectly viable.
They’re completely happy, I think.
The problem, however,
is that the two heads
don’t cooperate all that well,
and so they don’t really eat very well.
But if you manage to feed them by hand,
they will go on forever,
and in fact, you should know
these worms are basically immortal.
So these worms, because
they are so highly regenerative,
they have no age limit,
and they’re telling us that
if we crack this secret of regeneration,
which is not only growing new cells
but knowing when to stop –
you see, this is absolutely crucial –
if you can continue to exert
this really profound control
over the three-dimensional structures
that the cells are working towards,
you could defeat aging
as well as traumatic injury
and things like this.
So one thing to keep in mind
is that this ability to rewrite
the large-scale anatomical
structure of the body
is not just a weird planarian trick.
It’s not just something
that works in flatworms.
What you’re seeing here is a tadpole
with an eye and a gut,
and what we’ve done is turned on
a very specific ion channel.
So we basically just manipulated
these little electrical transistors
that are inside of cells,
and we’ve imposed a state
on some of these gut cells
that’s normally associated
with building an eye.
And as a result, what the cells do
is they build an eye.
These eyes are complete.
They have optic nerve, lens, retina,
all the same stuff that an eye
is supposed to have.
They can see, by the way,
out of these eyes.
And what you’re seeing here
is that by triggering
eye-building subroutines
in the physiological software of the body,
you can very easily tell it
to build a complex organ.
And this is important for our biomedicine,
because we don’t know how to micromanage
the construction of an eye.
I think it’s going to be
a really long time
before we can really bottom-up build
things like eyes or hands and so on.
But we don’t need to, because the body
already knows how to do it,
and there are these subroutines
that can be triggered
by specific electrical patterns
that we can find.
And this is what we call
“cracking the bioelectric code.”
We can make eyes. We can make extra limbs.
Here’s one of our five-legged tadpoles.
We can make extra hearts.
We’re starting to crack the code
to understand where are the subroutines
in this software
that we can trigger
and build these complex organs
long before we actually know
how to micromanage the process
at the cellular level.
CA: So as you’ve started to get
to learn this electrical layer
and what it can do,
you’ve been able to create –
is it fair to say it’s almost
like a new, a novel life-form,
called a xenobot?
Talk to me about xenobots.
ML: Right.
So if you think about this,
this leads to a really strange prediction.
If the cells are really willing to build
towards a specific map,
we could take genetically unaltered cells,
and what you’re seeing here
is cells taken out of a frog body.
They’ve coalesced in a way that asks them
to re-envision their multicellularity.
And what you see here
is that when liberated from the rest
of the body of the animal,
they make these tiny little novel bodies
that are, in terms of behavior,
you can see they can move,
they can run a maze.
They are completely different
from frogs or tadpoles.
Frog cells, when asked to re-envision
what kind of body they want to make,
do something incredibly interesting.
They use the hardware
that their genetics gives them,
for example, these
little hairs, these little cilia
that are normally used to redistribute
mucus on the outside of a frog,
those are genetically specified.
But what these creatures did,
because the cells are able
to form novel kinds of bodies,
they have figured out
how to use these little cilia
to instead row against the water,
and now have locomotion.
So not only can they move around,
but they can, and here what you’re seeing,
is that these cells
are coalescing together.
Now they’re starting to have conversations
about what they are going to do.
You can see here the flashes
are these exchanges of information.
Keep in mind, this is just skin.
There is no nervous system.
There is no brain. This is just skin.
This is skin that has learned
to make a new body
and to explore its environment
and move around.
And they have spontaneous behaviors.
You can see here where
it’s swimming down this maze.
At this point, it decides to turn around
and go back where it came from.
So it has its own behavior,
and this is a remarkable model system
for several reasons.
First of all, it shows us
the amazing plasticity of cells
that are genetically wild type.
There is no genetic editing here.
These are cells that are really prone
to making some sort of functional body.
The second thing,
and this was done in collaboration
with Josh Bongard’s lab at UVM,
they modeled the structure of these things
and evolved it in a virtual world.
So this is literally – on a computer,
they modeled it on a computer.
So this is literally the only organism
that I know of on the face of this planet
whose evolution took place
not in the biosphere of the earth
but inside a computer.
So the individual cells
have an evolutionary history,
but this organism
has never existed before.
It was evolved in this virtual world,
and then we went ahead
and made it in the lab,
and you can see this amazing plasticity.
This is not only
for making useful machines.
You can imagine now programming these
to go out into the environment
and collect toxins and cleanup,
or you could imagine ones
made out of human cells
that would go through your body
and collect cancer cells
or reshape arthritic joints,
deliver pro-regenerative compounds,
all kinds of things.
But not only these useful applications –
this is an amazing sandbox
for learning to communicate
morphogenetic signals to cell collectives.
So once we crack this, once we understand
how these cells decide what to do,
and then we’re going to, of course,
learn to rewrite that information,
the next steps are great improvements
in regenerative medicine,
because we will then be able
to tell cells to build healthy organs.
And so this is now
a really critical opportunity
to learn to communicate with cell groups,
not to micromanage them,
not to force the hardware,
to communicate and rewrite the goals
that these cells are trying to accomplish.
CA: Well, it’s mind-boggling stuff.
Finally, Mike, give us
just one other story
about medicine that might be to come
as you develop this understanding
of how this bioelectric layer works.
ML: Yeah, this is incredibly exciting
because, if you think about it,
most of the problems of biomedicine –
birth defects, degenerative disease,
aging, traumatic injury, even cancer –
all boil down to one thing:
cells are not building
what you would like them to build.
And so if we understood
how to communicate with these collectives
and really rewrite
their target morphologies,
we would be able to normalize tumors,
we would be able to repair birth defects,
induce regeneration of limbs
and other organs,
and these are things
we have already done in frog models.
And so now the next really exciting step
is to take this into mammalian cells
and to really turn this into the next
generation of regenerative medicine
where we learn to address
all of these biomedical needs
by communicating with the cell collectives
and rewriting their bioelectric
pattern memories.
And the final thing I’d like to say
is that the importance of this field
is not only for biomedicine.
You see, this, as I started out by saying,
this ability of cells
in novel environments
to build all kinds of things
besides what their genome tells them
is an example of intelligence,
and biology has been
intelligently solving problems
long before brains came on the scene.
And so this is also the beginnings
of a new inspiration for machine learning
that mimics the artificial intelligence
of body cells, not just brains,
for applications in computer intelligence.
CA: Mike Levin, thank you
for your extraordinary work
and for sharing it
so compellingly with us.
Thank you.
ML: Thank you so much. Thank you, Chris.