Estimation of Lamellipod Protrusion Forces (Richard Chadwick)
Sep 5, 2023
Richard Chadwick
8/3/15
BioNanotechnology Summer Institute '15
Well, thanks for inviting me.
Yeah, so
First, I'm not going to be talking about hearing today. I'm talking about cell mechanics, but
just
some other words of introduction. I
really encourage you to stop me in the middle and ask questions. I like that and
So what I want to begin with is just explain a little about the NIH and
what kind of a place it is. It's in Bethesda, Maryland. It's
basically a campus
without students, although there are some students now. You can do it. You can get your PhD
by working in a lab there, but you have to be in a university as well that gives a degree program. So
you can
do both. I mean you can be at Johns Hopkins and work at
NIH or
University of Maryland and work at NIH or anywhere.
Now
NIH is it's it's the United States' premier biomedical
research
they fund givers. I mean
they give the budget of NIH is like 35 billion dollars.
That's like a state's budget in some states.
It's
90% of that money goes to universities
in terms of grants, and I'm sure there are a lot of NIH grants here, and they're very hard to get.
Particularly if you're
coming from the engineering side.
How many of you are biologists?
Okay, and how many are engineers?
And how many are physicists? Oh good. That makes it good. So
the message I want to
give to you, the take-home message, is if you want to work on biological problems,
work with a biologist. And if you think you want to know about
force, and in biology, it's there's never been a time where people care about
forces so much as right now. And
everybody wants to measure forces and
all the interesting questions come from biologists. So I encourage
everyone to form a good collaboration with
with a biologist. And I know it's very difficult
because I myself, I'm not a biologist. I'm a
I'm actually an aeronautical engineer, my PhD, but over the years I've
become kind of a
quasi cell biologist. But to keep me straight, I need to work with a cell biologist. And that person is Claire Waterman,
in my case. She's a well-known cell biologist who happens to care about forces.
And the other
people involved on this project
actually, Claire and I and
Daphne, she was a postdoc of mine. She has a position in Crete,
Greece.
We wrote a grant and got
normally, people in NIH can't write a grant if you're in intramural program, but they're trying to encourage people in different
institutes
to cross boundaries and
with
they give like seed money for that. And we got such an award, a Director's Challenge
Innovation Award, and that brought
Daphne from Greece to work on the project. And the other person is the microscope
expert in
Claire Waterman's lab, who put together
the
instrumentation. And I
I'm a modeler. I like equations, so you're going to see some equations. I hope you don't mind.
Okay, so the important thing is to have a question. And I have to say,
honestly, we didn't have a good question.
This, the good question, the question came later, but and this is the question.
What, so when cells move on a substrate and the
the
it depends if they're moving on in 2D, like
in this panel, or if they're embedded in tissue and moving. They seem to move in different ways.
So the classical thing is that actin, this green
rod here, is growing and pushes the cell forward and
then
molecules in the back are
usually thought to be myosin. It's pulling the back
down. And in the meantime, there are adhesions that hold it. So it's kind of inching along in this way. And
on the other hand, when they're in tissue,
like a
metastasizing cancer cell,
they don't seem to move that way at all. I mean, they don't move in a straight. So they seem to bleb.
They seem, they form little bubbles and these
blebbing things push through the tissue. And these are pressure driven. So there's, there's, the question is,
and here it's known that it
whoops, I didn't want to do that. Here it's known that
actin is not involved.
It's pressure that's involved. Whereas here, it's thought that pressure wasn't involved and it's actin is involved.
Now the problem with actin being involved is that
actin can only, a monomer can only get in on the end. And the end is touching a membrane.
So there's right away a problem about
how does the
actin monomer get onto the end of F-actin and at the same time push forward?
This
always bothered me as a, you know, from the physical sciences. And the way around that is that there were some very
sophisticated
Brownian motion type theories that were developed
called the Brownian ratchet. And basically what's supposed to happen is that this membrane is
fluctuating. And
it just so happens the timing of the fluctuation is that
something gets in there and pushes. I mean, it's kind of hard to believe, but
in physics where, if you ever read Feynman's book, he talks about a
Brownian ratchet and the impossibility of it. And yet
here, that's what's making this, that's what the biology field says is making it move. And
of course, there's energy involved. So maybe it can work.
Also, the other mechanism is that, that this, the actin rod itself is vibrating and getting in out of the way and
allows space to
to get in.
By the way, if you're interested in this talk, it's published now, and it's in Scientific Reports, so you can look it up.
So the question is what, in this case, when you have a lamellipod, this long thin
front of the cell, which is moving that way,
is it,
is pressure also pushing? If it can push here, why not
here? And how would you decide that? So the tools that we use
is the atomic force microscope. And
so we had to put together a microscope there where we could see the cell. And
this, so we, this is the standard
bioscope AFM.
I'm not good at this. And we have a spinning disc
hooked up to it. I showed this last time I was here.
Anyway, this system is, right now, it's,
we're reconfiguring it, but it worked pretty well. And
this shows the kind of image we got from,
it's a PTK1.
That's a kangaroo rat
kidney cell. It's, we chose that because it's very thin. And you can see that, you can see
actin staining.
These are the
actin rods that are supposed to be pushing. And you can also see the adhesions in red. And you can see the
microscope, atomic force microscope cantilever. And we use the sharp tip, I want to say.
And a lot of what I'm talking about is what you can learn when you indent with a
sharp tip. And this shows some of the initial data that we got.
We were interested,
and well, the standard thing with atomic force microscope
is you try to get Young's modulus, which is a measure of the stiffness of the cell. The larger the modulus,
the stiffer it is. And
because we could see where we were on the cell, we could poke around
on adhesions, on fibers, in different places.
We can also measure the
height using this. And we can plot out
different domains of Young's modulus. And it's varying here up to around,
I guess it's 14 kilopascals. But generally there, it's in a few kilopascal range.
I have to say that just blindly using the atomic force microscope and using the standard
Hertz theory, which is what everyone does, will get you into trouble. And you, because the lamellipod is so
thin,
that you'll feel the bottom. And so you have to make a correction for that.
And the corrections are in the literature.
And I worked quite hard on that correction myself. So there's no questions so far.
Yeah, that's a good, I would say that this is about,
let me, I don't want to advance the slide. This is about
20 nanometers,
maybe 10 nanometers.
No, I'm sorry. This is about 2 microns here. You see that? 2 microns.
Maybe the whole thing is 10 microns.
Okay.
What happened?
Okay. Now,
when we started to analyze the data, this is a typical,
I should explain a little about the atomic force microscope. So there's,
what you have is a piezo and a cantilever attached to it. And the piezo goes down.
And the tip of the cantilever indents the cell. And it gets bent up.
So there's two signals. You learn how much you're going down.
That's on the x-axis.
And then how much the cantilever got bent up is on the y-axis.
So those are the two things you learn from making this measurement.
It does measure a force. So I will explain how. You have to calibrate the cantilever.
So if you know, so it's just, if you know the spring constant for the cantilever,
and you know, for example,
this is 5 nanometers. And if the spring constant is so many nanometers,
so many nanonewtons per, so many newtons per meter, you can calculate the force.
So we know this is also force. But if you multiply this by the spring constant,
you have the force. But I have to say we're, we're pushing down.
So it's not obvious what the force is.
But I have to say we're, we're pushing down. So it's not obvious what is pushing forward.
Okay, that's in a different direction. But what's unusual about this curve is there's this bump.
Most of the time you don't see that. I mean, we did see it often.
But most AFM curves just follow this, this path.
And you get the Young's modulus by fitting a theory to this, that red,
the red line fitting this part of the curve gives you the Young's modulus.
This whole talk is about what's that bump due to.
Okay, so now I'm, this is a model of, of the lamellipod.
It's, I have to say, controversial.
There is a paper that's out that from Chao-Wei Zhang from Harvard.
It's in Nature Methods. She has a dual storm high resolution imaging technique
where she imaged lamellipods and got, and showed that there's really two layers.
There's like a, a ventral layer, which is dorsal and which is ventral.
I get mixed up. The, yeah, this is dorsal and this is ventral.
So there's two layers of actin cortex, which is, which is a mesh work of actin and myosin.
And it's an interacting network. I'll show a model of it on the next slide.
And there's a gap in here. It's a little unclear how far that gap goes.
But in this model, it's going all the way to the end.
And she definitely saw two layers in the lamellipod.
And the things that I'm trying to learn by doing this experiment are different parameters.
And that will help me estimate the force pushing forward.
But there are basically mechanical properties or physical properties.
There's the gamma. It's the tension in the cortex.
You can think of it as the membrane plus this thin layer of actin cortex.
And it has a certain thickness, H.
And there's also the elastic modulus of the cortex, bending modulus of the cortex.
There's the bending, a bulk rigidity of the cell.
And there's the height of the cell.
And then there are these bunch of different gaps that I want to talk about.
How much the actin actually gets compressed back by pressure acting on it.
And then there's the distance to the focal adhesion.
So this is the model that we use to analyze the data.
And actually we could get all of these things from one AFM curve.
So that's kind of amazing.
So this is a model of the cortex.
I took this from, I think, a paper from Ava Palook's lab, who studies blebbing.
She has a lab in University College London.
And so this is just the bilayer.
And then under there, there's about a 100 nanometer layer of actin filaments with myosin.
And they're creating kind of a, the myosin is pulling in there.
And that's what creates the tension.
So the membrane cortex is under tension.
And we can, and the thickness of this thing is around 50 nanometers, I think.
I don't think it's one micron, like it says.
So this is, now comes the AFM part.
So the whole theory for AFM was invented by Hertz.
The same Hertz that, you know, megahertz Hertz guy.
And why he was interested in it is he was interested in building telescopes at the time.
And his PhD thesis in the 1880s was he was interested in two glass spherical lenses resting on each other.
And the weight of them compressed the area between them.
And he was interested in determining the contact area and everything.
So this theory of his has since been, it's what we use in the atomic force microscope.
It gives a relation between Young's modulus and the two things that you can measure with the motion of the piezo and the motion of the cantilever.
However, it's designed for either two spheres or one sphere in elastic half space.
It's not designed for a thin layer.
So that's why you have to correct when you have a uniform material in a thin layer or just a uniform material.
In this case, there's just cytoplasm under this.
So this shows and every sharp tip has some finite radius that you try to fit in there.
And so you usually use a paraboloid with a radius of curvature that we don't really know here, but we can fit it.
And one of the hard things about contact mechanics is it's like a mixed value, mixed boundary value problem.
Out here, there's no stress after when it's not touching.
But here you have a known displacement.
So it's a displacement is known over part of the boundary and stress is known over another part.
And you don't know where it leaves.
So that was the genius of Hertz.
He could figure out how to calculate where it leaves.
But for this problem, we needed a whole new theory because Hertz doesn't apply.
And one of the things that we want to remember, we want to get the membrane tension, the bending modulus, all those things.
So this is the equation that actually we solve.
It has bending in it.
Bending has a fourth derivative in it.
And this is the bending modulus.
This is like just this is the tension part.
This is like a drum and this is like a beam.
And then there's pressure.
And this equation was also known in the literature of structural mechanics.
It's the Floppel-Von Karman equation.
Anyway, this is how you express the...
For those of you that don't want to look at this, you don't have to, but it's not that hard.
So the thing is, on the left hand side of this equation, this is the shape of the paraboloid, which we know.
This is the maximum flexion downward.
This is the radius of curvature.
This is just the radial coordinate.
And here's the membrane tension.
And this is the pressure, which we don't know.
That's the pressure acting in the contact region.
And this is the singularity solution of this.
This is like the point force solution.
So we're integrating over a bunch of point forces with an unknown pressure to get a known deflection.
This is an integral equation.
This is the kind of equation Hertz solved, but he didn't solve it with this log in it.
He solved it with this.
He solved it for this.
And the way we go from here to here is I just differentiated that with respect to r.
And we get an equivalent kind of Hertz kind of problem that we can solve.
Because this integral equation is known how to solve that.
So in math, you know, in math, you shouldn't start all over again.
You should try to reduce things to what you already know.
So in this case, the tricky part, after I differentiate this, I added and subtracted a constant.
So I don't change the equation, but it shows a superposition of two different known problems.
This happens to be a cone.
And this is a flat-end cylinder.
And you can see that here, that the problem I want to solve on the left can be the superposition.
This is a half space.
This is nothing under here.
This is just tension and bending.
I'll get to the bending in a minute.
A cone plus a flat-end cylinder.
And both of those are problems of the period in the literature.
So because of this, we're able to get a simple formula relating the characteristic things about the bump,
this delta star and d star.
There's a closed-form formula to get the tension.
If you measure d star, it should be a delta star.
And you know the spring constant, you can actually get the tension.
And when we did that, we found the tension was in this range, hundreds of piconewtons per micron.
This is larger than the bilayer.
So it shows that there's some kind of active structure in the cortex due to the myosin pulling on the actin.
Yeah, okay.
In the previous slide, you had the superposition of a cone and a flat-end.
Why didn't you make it as a circle?
No, it is a circle.
These are axi-symmetric.
They're both axi-symmetric.
Good question.
Now what about bending?
How do I get...
So I can get the tension.
How do I get bending?
So it turns out that this part...
So this is where the cantilever tip leaves the cell membrane, where the arrow is.
So this is a parabola.
And then this is a curve that has to come back to where it started from.
And this part of the curve is determined by the bending modulus.
And so what you get...
So the shape of that curve is given by Bessel functions.
And the parameter in there is the ratio of the tension to the bending modulus.
And if we know the tension...
It turns out there's only one value of this kappa that works to make it come back to the right place.
It's an eigenvalue problem.
And that you can get numerically.
And once you know that, you can get the bending modulus.
So we find that the bending modulus is in this range.
It's expressed in terms of the Boltzmann constant and temperature.
Somehow that's the unit that's used for bending modulus.
And we also get...
If you know the bending modulus and the tension, and you know the strain...
And you can get the strain because it's the difference in the length of that curve compared to just the flat curve.
Then you can also get the thickness of the cortex, which we found to be in this range,
which agreed with those storm images that I showed in the beginning.
And then from simple bending theory,
knowing the tension and the thickness and the strain, you can get the elastic modulus.
And that's much larger, as it turns out, than what's just measured from the bulk modulus.
So the myosin-actin interaction is stiffer than just the regular bulk part of the cell.
So you get a lot of information from one bump.
Now, I'm interested in getting the pressure.
And the first question is, how much pressure can you get from acting, pushing?
And that's a different experiment.
And I used what's in the literature.
So people have used an optical...
There's something called a philopod, which is just like a...
Instead of a sheet, it's just a cylinder going out.
Cells do that too.
And by putting the spheres to represent an optical trap...
You can measure the force on the optical trap when you calibrate the trap.
So what you measure is the actin force pushing against it,
minus the tension in the membrane pulling back.
So you don't really...
Can't get the actin force directly from that.
Somebody else...
But what you know is that this total force was 3 piconewtons.
Somebody else recently measured the resistance of the membrane part of that
when this was retracting.
So that was 15 piconewtons.
This is the PNAS paper.
I forget where this one is right now.
Sorry.
So based on that, we know the total is 15...
The philopod actin pressure will be this 15 plus 3 piconewtons.
And then I need to divide by the area of the philopod to get the pressure.
It's 2.3 kilopascals.
So the question is, is that a significant part of...
How much is the just cytoplasmic pressure?
And how do I get that?
So basically we use a force balance.
This is not exactly the correct one,
but the idea is that pressure shown here is pushing on an area,
and it's resisted by tension acting on a perimeter.
But what's wrong with this slide is there's also actin pushing.
So this is a combination of actin and pressure.
And I show that on the next slide better.
So this is the correct force balance.
This is the cytoplasmic pressure.
This is the filamental pressure due to the actin.
And you have to know the height at the edge of the lamellopod.
And I'll get to that.
That was a big issue getting this paper accepted,
because we don't...
I'll come back to the problem.
And this is balanced by the tension.
So that's just a simple force balance.
And because of the model, the actin is not filling up the whole system.
We can rescale the filopod pressure.
Actually, it's only this part that's pushing, the bottom part in the model.
So we're able to...
And since we know the tension, we're able...
And I'll show you how we get the height at the edge.
We're able to calculate the pressures in the cytoplasm.
So this is comparable to the pressures due to actin polymerization.
So it seems like they're both doing the job.
But now to...
One of the problems was that if there's so much pressure in there,
why doesn't it bulge up, like in a bleb?
Why does it remain flat?
So I realized I had to overcome that difficulty.
Sometimes the reviewers have raised very good issues,
and this was a very good point.
And so I had to go back to the drawing board.
So I guess that will come on the next slide.
So let me first explain how we compute all these gaps.
I mentioned Brownian ratchets.
So there's a formula in the literature that gives how much a membrane can vibrate.
This is the amplitude of the vibration.
This is the Boltzmann factor.
And this is the pressure in the cytoplasm.
And this is the bending modulus.
Since we know everything here now, for each cell...
By the way, we only had five cells that we got measurements off of.
So for each cell, we could determine the pressure and the bending modulus.
So we could calculate the gap due to the membrane
and see how many nanometers it is.
I'll show you a table pretty soon that will show these values.
And at the same time, the way we calibrate a cantilever
is that we watch it vibrate thermally,
and you can measure the root mean square of the vibration.
So there's a known formula, and it's using the fact that
there's a balance between thermal energy and elastic energy.
And that balance is given here.
It's also related to kT.
Instead of the stiffness of the cantilever,
this is the stiffness of the actin rod.
And using, again, elasticity theory,
we can relate that to Young's modulus
and the thickness and the length of the rod.
So we can calculate the amplitude of vibration of this cantilever.
And I realize now that there's a slight mistake.
That's how much it moves.
It doesn't explain the gap.
There's another formula for the gap.
And at the same time, there's a pressure in there that's pushing up,
pushing forward, pushing back.
It's isotropic.
And this delta is the amount that's shown here
that is getting pushed back.
And this allows the actin monomers to get in there to polymerize.
And this shows some measurements that were done previously
using qualitative speckle microscopy,
where you can see where there's a lot of actin turnover in a gap near the edge.
And this compares favorably with these measurements.
Okay, now I want to get back to the question about why is it bulging.
And it is bulging.
I mean, actually the model should be tapered
because this shows the height of the lamellipod.
It is thinner at the very edge than in rearward.
However, when it's under pressure,
there's tension and bending in this equation
that allows us to calculate a new shape.
And it's this edge that we need, which is around 300 nanometers.
It bulges from about 100 to about 300 nanometers.
And that's an important part of the theory.
So it turns out you can calculate the probability
that an actin will get onto the end of the rod.
And the theory for that, it's by physicists in the 20s.
I found it in a Chandrasekhar paper,
but I think it's from Uhlenbeck originally for the physicists here.
And it's the theory of Brownian motion of an oscillator, of a damped oscillator.
So that's the theory I use.
And after a long time, the probability of the gap exceeding the monomer size
is given by the integral of a Gaussian.
So this is the integral of the Gaussian.
And what the formula says, the reason the 7 is there,
that's 7 nanometers, that's about the size of an actin monomer.
And I ran into a problem with that too,
because if you read the Brownian motion literature,
they want that number to be as small as possible
because they want the gaps that they calculate to be larger than the monomer.
And so they use 2.5 nanometers.
But that's not, I mean that is based on the assembled repeat, half repeat.
But the actual monomer, the crystal structure is known,
and it's between 6 and 7 nanometers.
So it makes it harder for a given, you know,
the bigger the monomer, the harder it is to get in there.
So this is the probability that an amplitude A of a vibration,
of the membrane or the gap left by this bending rod,
will be greater than 7.
Otherwise it won't get in.
So what you see is like if a monomer were like 10 nanometers,
I mean if the vibration were 10 nanometers,
there would be like a 20% probability it could get in.
And if it were, this is 10, if it's, you know,
if it's 7 or something or less than 7,
there's still a finite probability, but it drops off very fast.
And it becomes unlikely that a monomer will get in.
And if it can't get in, it can't exert this force.
So this shows that this is also useful in calculating
or explaining the difference.
So the bigger the gap, the more,
the easier it is for a monomer to intercalate.
And that's why we want the pressure to be high,
so the gap will be larger.
Okay, this shows a lot of data.
So first of all, this shows all the different cells,
cell 1 through 5.
And these are the AFM parameters varying between,
this is how much the cantilever bent from the laser light detector,
between 2.4 and 6 nanometers.
This is the indentation of the membrane cortex,
and it's varying between 300 and, you know, around 100 nanometers.
We could use one of the formulas I showed
to actually calculate the effective contact radius,
which came out to be slightly different on each cell.
And these are the tensions in piconewton per micron in the cortex.
This shows how this method agrees with previous results,
pulling a tether or a micropipette aspiration.
These are much more harsh measurements to make,
but it shows the method works.
These are the bending moduli, also agreeing with other techniques.
No, that's the tension.
The bending modulus hasn't been measured on cells like this with a cortex.
Another way of expressing the bending modulus,
this is called the persistence length.
This has been used in vitro by looking at actin polymers
and looking at how wavy they are.
The shorter the wavelength, the less the bending modulus.
So this shows how, in the cells,
the persistence length seems to be smaller than in a test tube,
which has been measured in the literature.
These are the thicknesses of the cortex that we got.
It's comparable to super-resolution light microscopy.
These are the elastic modulus, moduli.
These are the pressures due to actin polymerization,
and these are the pressures due to the cytoplasmic pressure.
These came out to be a little bigger, but they're comparable, I would say.
This shows the different gaps.
This shows the lower gap due to the pressure pushing the cortex back.
All these are larger than seven, which is important,
which means it helps polymerization.
The upper one is even way, way, way larger.
The upper one is larger because I assume that there's a series spring in there,
which is softer.
These show the membrane fluctuations.
This is the fluctuation of just the membrane without the cortex,
and it barely...
That was another thing.
The reviewer said, well, the membrane...
You should take the...
It should be free of any molecules attached.
This estimates...
I took the measured bending moduli, measured stiffnesses,
and I reduced it by a factor of 10 just to get an estimate of the membrane itself.
These still don't give you very much...
It's not much different than seven.
These are slightly larger, so it seems to be a little more efficient
for the thermal fluctuations of the actin.
It's a little bigger fluctuation than the fluctuation of the membrane.
Basically, to summarize, and I understand you...
I should talk about some other things after this,
about getting grants and how to go to NIH, which I'll be glad to do.
I just wanted to say that I think the first thing is that it's so important
for a biologist and physicist or engineer to work together
and to struggle through each other's thinking
because we're thinking in different ways.
Maybe this is not the most important biological problem,
what's pushing a cell forward, but it's the most important...
It was a problem that bothered me, and I don't think it's...
I think it is relevant to biology,
but I wish I had come up with a better question than I did,
but that would be the next paper.
In any case, using the AFM, you can measure lots of stuff,
but you can't do it without a modeler.
That's another thing.
You just can't go into the literature and pull out formulas
because no one talked about this bump before.
When Daphne got many curves with bumps on them,
instead of throwing them out, we thought about it,
and it resulted in a new way of thinking about how it works.
The conclusion is that, first, this was really hard to get it to this point.
People just want to believe it's actin pushing,
and they don't want to believe it's myosin squeezing in the cortex to push it.
But this shows that it's both.
The other point is that the pressure helps the polymerization
because it creates a bigger gap for the monomers to get in.
That's basically what I wanted to say.

