 So as the last thing we do, I want to show you something that's really state-of-the-art today. Most of the previous proteins design I've showed you have the fact in common that they kind of piggybacked on existing folds, and I just tried to stabilize a new sequence in that fold. That is smart, and it's in general what you want to do. Don't reinvent the wheel. But what if we could create completely new folds? Remember this landscape that I spoke about in the bioinformatics lecture that we know most of the existing fold space. But instead of this space, what if I could design a protein out here? It's a completely new fold, a fold that we've never seen in nature before. In this case, there is no way we could rely on plain bioinformatics. It's not going to work. We don't have anything like it. This is something that I need to stabilize based on the laws of physics, basically. There is one group in the world that I would argue significantly further ahead than anybody else here, and that's David Baker in Washington, the groups that are behind this Rosetta program. And what they do in most of this case is roughly the same thing as for Rosetta, but you can piggyback as much of known structures. You're assembling small fragments. In this case, we try to assemble small fragments into a new fold that we've either guessed or had divine inspiration or tried to optimize what it should look like. And then we're iteratively placing side chains, see if we can improve the stability, moving the side chains around, maybe even running a short simulation to see if we can stabilize things. And then we need to go in the lab, test this, take back the successful cases, try to modify the successful cases a bit. There is an insane amount of work here, but there is also an exceptionally impressive stream of new papers coming out where they show that they can do this. They can design vaccines, they can design proteins, and they can design completely new folds for proteins. This is David. One thing that they did just two years ago, I think, was that they, instead of building traditional proteins, they figured out what if we could design some sort of large building blocks, almost like the fibrous proteins we talked about earlier in this class. Now we can't have very long genes, so that their idea is that what if we start by creating some small bricks. And what they did here is that they usually use small helix and sheet combinations, two, three secondary structure elements, then they would use several disulfide bridges. So sustenance, that would mean that I'm going to lock them up with disulfides, create very nice small rigid components. And for a few of these, they also used chemical treatment afterwards to try to get the N and the C terminal to stick together. Again, the reason for doing this is creating small building blocks where each building block will be very rigid and stick together. Once you've done that, you have your small bricks. And if these bricks now stick together, we can start to change the amino acids on the surface of the brick. So if I have two bricks here, what if I create a periodic pattern here so that the bricks will like to pack? Then the bricks might spontaneously self aggregate into larger structures. And do they? This is a shock and awe paper too that I have for you in Canvas. They had a whole range of structure where they can create almost crystals formed out of proteins with repeating patterns, because again, each building block here is small, but the patterns they create can not quite reach macroscopic scales, but almost. The use cases for this is unlimited. You can imagine using this in bacteria, we might create those more miniature structures and everything. The reason why I don't have any super specific examples for you yet is that it's still research, of course, but we're literally creating a type of protein structure engineering that is like nanotechnology. The only difference is that traditional mechanics-based nanotechnology, that has structures in the ballpark of 900 nanometers. These proteins, they have structures in the ballpark of 10 nanometers, so that it's almost on the board of picotechnology, not quite. But far smaller structures that can do far more advanced things, and I bet we're going to start to see miniature machines here in the near future too. When I introduced the membrane protein class to you, I figured that there's only one class of proteins that is not so common, the beta-barrel proteins in membranes, so maybe I could let go of those. Chance has a way of playing funny tricks at you, because because I did that just two weeks later after that lecture, of course, David Baker published an amazing paper in Nature, where they show that they could design beta barrels, so now I have to share beta barrels with you anyway. So beta barrels are not common membrane proteins, but they do exist, and when they form some sort of semi-beta heat-like structure on the outside, and then they gradually slide into the membrane by themselves. We know very little about their folding, we also know that there's delicate stability and everything, so this is probably some of the hardest proteins imaginable to fold, but David managed to do that. This is also a paper I've shared with you, because it's a remarkable example how to use protein design in practice and designing completely new beta-barrel proteins. So what did they do? Well, first they used all these rules that we've gone through, having anchors on the top of the membrane proteins to anchor, say, tyrosines to the head group regions here. You need a hydrophobic exterior of the pore, and then a moderately hydrophilic interior of the pore, so literally have a pore functionality to go through it, and then carefully select the number of beta strands we have to form a large beta sheet. But it turns out that's not enough. Because if you do this and create something that is a very stable beta sheet, it's not going to work at all. The reason for that is that if something is too stable as a beta sheet, they will just form a beta sheet out in the water. And if it's stable as a beta sheet out in the water, it's no particular reason for it to go into the membrane. So the challenge here is not just design, but for proteins you also need to do anti-design. So the greatest challenge for them here was actually not to design a beta sheet. Anybody, me included, could do that. The hard part is that they had to reduce the property of this protein to form a beta sheet out in water. And when they reduced this property sufficiently enough and made it relatively hydrophobic, then you got it to a state where it would not form beta sheets out in water, but it will form beta sheets while it is inserting into the membrane. And then they managed to show that a dozen or so of these actually do form stable new folds of beta-barrel membrane proteins.