Hello I’m Julie Theriot at Stanford University and I’m delighted to be taking part in this iBiology talk series on great problems at the interface between physics and biology. The particular problem that I would like to explore with you today is a question of how to go about discovering design principles for cells and organisms. This I think is a very fundamental question in biology that reveals itself just when you look at any living organism. And I’ve showed you a few examples here at the cellular level, where on the left we have one of Santiago Ramón y Cajal’s beautiful drawings of neurons in the brain. In the middle we have a light micrograph of an algae, Spyrogyra where you can see the beautiful regular helical patterns of its chloroplasts. And over here on the right we have the little glass exoskeleton of a diatom, which is a unicellular organism that’s able to build itself this beautiful little glass house in which to live. Now if we were to see these kinds of very precise, robust and delicate structures in something it was built by human, we would assume that there was some sort of underlying design principle that the human had chosen to achieve some function, chosen materials and chosen the structural organization that would achieve that function. But at the level of cells, in the absence of an architect, there have to be some other kind of underlying physical principles that enable this kind of organization to emerge. Now we can think about this problem of a cellular self-organization at many different scales. And arguably one of the first interesting places where self organization happens in biological systems is just at the level of protein folding. And I would actually argue that at this point we basically understand pretty well how proteins fold up. The interactions between the atoms in the protein as it’s folding are very familiar from fundamental physical and chemical principles and using just very basic equilibrium theory, it’s possible to predict the folding of a protein with an astonishing degree of precision. Surely there are details left to understand but the physical principles I would say we do understand. But as soon as we start to getting to larger scales then things all get to be a great big mess. So if you take a bunch of these proteins you put them together you make a cell. If you take a bunch of cells you put them together you can make a multi-cellular organisms such as this mouse embryo shown here. And if you take a bunch of organisms and put them together you can actually have very elaborate larger scale kinds of biological organization. One example I’m very fond of is this termite mountain shown here, which is several meters tall. And this enormous and elaborate structure with little nests and storage rooms down at the bottom and a big system of air conditioning built up top is put together by little termites that are only a few millimeters long. And I would postulate that there’s not a termite architect is telling all the rest of them where to go. Instead they’re able to use just local information and local behavioral rules to put mud together to build this big structure. I think that’s a very good analogy for what must be happening at the level of the organism and at the level of the cell. So fundamentally as we try to understand this level from physical principles our equilibrium theories that worked pretty well for protein folding just failed dramatically at all higher scales of organization. So one way maybe for asking this question about how we’re going to discover design principles, is to say in the context of protein folding, for example, we have a privilege state – the equilibrium state. But what equilibrium would that be for something that is like a cell or like an organism with no obvious global privilege state and yet is able to generate robust, reproducible and precise forms. Now that’s a difficult question. And I think it becomes even more difficult and even more puzzling when you realize that of course almost all biological structures are not only precisely defined but also are dynamic. There’s one case where you can see this very clearly which is cell motility. Now there many different kinds of cells, unicellular organisms or cells within multicellular organisms such as humans, that are able to crawl around in a directed and purposeful way. And one of the most motile cells we work on a great deal in my group is shown here. These are cells from the skin of a fish. And their structural elements can be highlighted. So for example in this image by my students Sunny Lou. She has used a fluorescently labeled mushroom toxin to highlight the structures of all of the actin filaments within this particular cell. So particular over here in this movie that was made by a former student Patricia Yam. She is using essentially the same technique to build up a little bit of texture in the podium inside this living cell. So as a cell starts to crawl, what I hope you can appreciate is that the net movement of the cell arises because the mesh work is assembled at the front or at the top, and that is disassembled at the back. But remarkably those two rates are almost exactly identical to one another. Because the cell moves forward without changing its top to bottom length. Similarly the rate of mesh work assembly over on the right side of the cell is got to be the same as the rate of mesh work assembly over on the left side so it is able to move forward without turning. So how is it that the cell is able to put together all these very large scale structural elements that are constantly turning over and yet do so with such precisely balanced rates? How does it know how to make the shape? Well we don’t know the answer to that. But I hope I can at least take a step back and focus on a slightly simpler question or what should be a slightly simpler question to challenge you with this idea of trying to come up with physical principles that will enable cells to do things like measure length, and count things and measure spacings as they are building structures. And in particular I’d like to focus on a relatively simple case, which is when cells build structures with filaments wrapped by membrane that protrude outward from their surface. There are many examples in biology where cells are able to build these kinds of surface protrusions that clearly have defined and measured length. So for example over on the left this is an electron micrographs, a cross section of the surface of an intestinal epithelial cell. We see it’s covered with these little fingers that are sticking up. They’re all wrapped in membrane and that enable an increase of surface area for the absorption of nutrients. Now in the middle we have a scanning electron micrograph of a ciliate. And this is a single cell organism that looks all hairy. Each of these individual little hairs is a beating cilium that the cell is using to swim around. And you can see both on the intestinal epithelium and on the cilia that all of these many hundreds of different surface protrusions are of really pretty much exactly the same length so it has to have some way to measure that. Now maybe even more interesting case that I’ll come back to is the stereocilia on the surface of the hair cell in the inner ear. And this little cluster of surface protrusions is what enables us to perceive motions in the air as sound within our brains. And you can see in this case that not only are these stereocilia in perfectly well defined length but in fact different stereocillia on the surface of the same cell all have different length. And they organize themselves in this incredibly precise staircase step pattern like a set of organ pipes. So we want to be able to come up with some proposition of general organizational principles that can allow these kinds of structures to be built by cells. And in order to do that we need to know a little bit more about their structure. In particular all of these surface protrusions are made by structural filaments that are bundled together and then wrapped around by membrane. These filaments themselves are also assembled from smaller subunits. And in the cases that we are going to be talking about, these filaments are actually self assembled by multiple different copies of identical subunits. And your first approximation, the same sort of simple equilibrium theories that enable us to describe protein folding, can also enable us to describe these very simple sorts of protein-protein interactions, that build in this case helical filaments which as many copies of the same subunit interacting over and over again. In the simple systems we can also precisely measure aspects of their rate, of their turnover. And so for actin filaments that we are focused on today there been a couple of really very beautiful experiments measuring precisely all of the rates that govern their self assembly and behavior. So for example in this one very famous study from Tom Pollard done in 1986. He was able to measure the on-rate and the off-rate of individual subunits landing on the ends of the filaments, both on the two structurally distinct ends and also as a function of whether the subunits contain ATP or ADP. So all of those numbers are known. Okay so if you have a system like this where you have a bunch of protein subunits that are able to self-associate, is that something that just by itself can give you a well defined length? Well the answer actually is no. It’s very clearly no. It’s been shown experimentally and also described theoretically that if you have this kind of self assembling subunits then that the default length distribution is going to be exponential. The fundamental reason for that is that each little subunit doesn’t know what the rest of the film is doing. It’s behavior is governed just by these local rates, these local interactions but because each one of those additional lots of reactions is essentially independent, the exponential distribution is the default. So we can get some aspects of filament structure in particular its diameter and also its dynamics defined by just the structure of the protein subunits that make it up. But for length, if we want the length to be some precisely defined number, there has to be some other mechanism for determining that. Now of course inside of living cells, filamentous proteins never operate on their own. There are always lots and lots of other proteins, in fact hundreds or possibly thousands of other proteins that interact with these cytoskeletal filament structures in order to determine their behavior. And this is a textbook illustration of just a few of the different kinds of proteins that affect cytoskeletal filaments, and the details here aren’t important but the basic idea is just for every aspect of dynamics or structures of these filamentous proteins there are some other protein that is able to interact with those subunits and regulate their behavior. So there are other proteins that can regulate the on-rate and the off-rate of the subunits at the ends that can bind to the filaments and regulate their stability, they can force the elements to bundle together etc, etc. Okay so now if we know that there are other proteins that will interact with these filament forming proteins. We go back to our problem of how do you set a defined length. You might imagine that a really easy way to do it would be just to have a protein whose job it is to determine that length. And in fact there are some cases in biology where we know that that happens. One very well characterized case is in bacteriophage lambda. So this is a kind of virus that infects bacteria. And it makes these spectacular little sort of landing pods structures. And there is a tail that is part of the structure where if you look at that tail in many different individual virus particles, as you can see in the micrograph, those tails are all the same length. And it’s been shown that there is a protein called H protein, a big long protein whose length determines the length of that tail. Internal inframe deletions that make the protein shorter also make the tail shorter and duplications that make the protein longer also make the tail longer. So that’s great but it doesn’t have to be a protein that acts as the template it could be a different sort of macromolecule. And in fact there are other viruses such as tobacco mosaic virus shown here that instead of using a protein to determine their length, they use the RNA genome to determine the length. But the idea is still the same. If you have a template that’s a great way to solve this problem. But of course we’re not done. Because the bottom line is this really simple elegant idea of having a template simply doesn’t work in any of those cases that I showed you, in the intestinal microvilli, in the cilia on the surface of the swimming organism or in the sterocilia on the surface of the hair cell. So what determines length in those kinds of cases? It’s got to be something that’s dynamic. So let’s focus in particular on these hair cell stereocilia because I think this is just an exquisitely beautiful piece of biology that’s actually filled with really really interesting question and physics. Now the organization of the stereocilia as I showed you before is in bundles where each individual protrusion is a defined length and they’re organized in such a way that the protrusions lined up so that they’re each next to a neighbor that’s slightly longer. And the reason why that’s important is because every pair of neighboring filaments of different length in the structure is attached by a little structure called a tip link. And when these hair cells in the inner ear have the tips of their those bundles bent by the movements of membranes that react essentially to fluid flow that comes downstream of the bouncing around the eardrum. Then the movement of those stereocilia relative to one another causes the slightly longer stereocillia to slide, with respect to slightly shorter neighbor. And so tugging on that tip link then opens ion channels that enable changes in membrane polarization, which can eventually be perceived as an electrical signal and interpreted as a sound by the brain. Okay so it’s obvious that it’s very important that these stereocilia be built a certain length in a certain pattern on the surface of the cell. And also actually it turns out that they need be tuned to different frequencies so that you can hear both high pitches and low pitches. And I want to share with you just an absolutely astonishing study that was done by Lewis Tilney and his colleagues back in the 1980s, that I think shows the magnitude of this problem and how difficult it really is going to be to try to understand how it works. What they did here was simple but extremely painstaking series of experiments where they dissected out the cochlea from newborn checks. And then use scanning electron microscopy to examine the hair cells in different parts of the cochlea. So one end there are hair cells that are tuned to hear high frequencies at the other end hair cells tuned to hear low frequencies and then graded frequency determination in between. What they found is if they look at the end that can hear high frequencies they found these short, little, stiff, stubby stereocilia. But if they look at the end of the cochlea that was to hear long frequencies they found these long, elegant sort of floppy stereocilia. Okay that’s pretty cool but the really amazing part is then they went ahead and measure the total number of actin filaments, the total length of those filaments, the total surface area membrane that wrapped around all these things for all these different cells throughout the cochlea. It turned out that those numbers were absolutely conserved. So there were 9,000 microns of actin filaments making up each of these very different kinds of structures. There’s 180 square microns of membrane that’s wrapped around each of these very different kinds of structures. And yet each individual cell tunes the overall length number and spacing of these stereocilia to be able to hear the pitch that it’s intended to hear. So this means is that you know every single one of the cells basically start with the same Lego kit. It had the same number of structural proteins and the same amount of membrane And yet each cell was to arrange them in slightly different ways depending on where it was in the cochlea and what pitch it wants to hear. Now we don’t understand how that works. Not only do we not understand in detail how the length of these structures can be set but we also don’t understand how those length can be so finely tuned in order to achieve this kind of biological specificity. Now that’s just one example, but I think it opens up this huge can of worms about how we go about understanding cell structure determination. And I think this is a subset of the very general question raised by Schrödinger of what is life. Certainly life has to obey physical and chemical principles but there are additional principles that I believe we have not yet discovered that are necessary to describe these kinds of processes that we see inside of living cells and organisms. And cells are very complicated. The inside of the cell was extremely crowded as you can see in this election micrograph reconstruction. It’s very inhomogeneous. Everything is constantly far away from equilibrium, and its material properties are very weird. It’s not viscous, it’s not elastic, it’s not plastic, it’s some combination of all those things and actually can change its material properties over time. So with all this kind of weirdness and complexity in strange material properties how are the cells nevertheless able to build these exquisitely regular beautiful structures that we see all the time. Well I think it’s very clear that equilibrium thermodynamics and statistical mechanics are not gonna get us there. It’s a good first step, but it’s not going to really be able to describe the physical principles for cell organization in a living, in a real living system. And I’d like to propose that one way to really try to get at this and to start building a general theory of the cell is to start with questions that seem like they might be relatively simple. How do cells measure things? How do cell count things? The hope is that once we understand the principles underlying those kinds of relatively simple processes, we should be able to put them together to understand robust and dynamic cellular organization also in more complex contexts. In complex context for the life of the individual cell and also in order to build a whole multicellular organisms such as ourselves. And at some point I hope that if we understand these physical principles and these organizational principles well enough, then we should be able to also engineer them. And be able to get the cells to build structures that we choose, rather than just building the structures that they choose.