 Good morning Today we are going to entertain ourselves with membrane proteins This is largely a continuation. Well, this is the third class of proteins that I told you about on Wednesday and It's also the big law of love both of my group and the most groups at DBB biophysics and biochemistry at Stockholm University They're fun The problem with a book on this topic is that 15 10 or 15 years 15 years ago when Alexi wrote this book and I would even argue most textbooks. We knew I wouldn't say that we didn't know anything about membrane proteins But our knowledge of this field is expanding so insanely rapidly that there are virtually no textbooks whatsoever that keep up with it So I'm not really going to talk about research here depending on how much time we have later on in the course I might spend one lecture on research topics in the department and everything and what people are working on But there are going to be quite a few relatively new results here But before we do that, I figured let's go back to our discussion items and spend some time with the globular and fibrous proteins 12 questions shorter today You lead the discussion take the questions and try to answer them and then we'll see if I have any comments So where do you find it? Right, so that's true and and so so what's common for all these proteins at a there Would you mess up all the three things you mentioned there are items that are so large that you can actually see them Not just with a microscope but without a microscope even and that's what makes them kind of unique This is not really something that is used in biotechnology today But I bet one could imagine starting to use biological building blocks even to form slightly larger structures So we talked a little bit about things like potentially using artificial proteins for organs or something, right and Imprinciple you could probably create a complete heart or something that well You couldn't create a heart You could create something with the shape of a heart that then a protein material on it and hopefully these would then be Proteins or something that didn't throw off the immune defense Certainly not possible today another problem is that if you ever get to the point where we would like to create organs or something You would also need them to do something. So just creating something like a blood vessel is probably easier But potentially a pretty cool way of forming building blocks long-term Yes, so it's and it's a bit stupid in a way, right because the name mostly has to do with the shape but When we say globular, it's not that first all globular proteins aren't really spherical But this name came up before we knew so many protein structures And it's very hard to change but but the point of them is really that they're water soluble and that has Some sort of relation to the shape, but it's not super strong So what's the difference between those protein types and membrane proteins? You can forget about membrane proteins as we're going to talk about that today Yes, so they're larger they're hierarchical, but there is also something they don't have the globular proteins typically have Well, I would say a well-defined specific functions, right? Remember when we spoke about globular proteins yet all this beautiful and that's why I can that's why I said that Fibrous proteins are kind of boring. It's just bricks, right? Well globular proteins that all these beautiful things that they combine groups. They do things they transport oxygen globular proteins are very biological in the sense that they do things in the body The fibrous proteins that's just building material We need both but No, there's there certainly enzymes that are fixed to membrane pro to membranes although In general, they don't know as we're going to see today when we talk about membrane proteins What is the reason for having a membrane protein in a membrane? Right usually that it's as you're going to see these are slightly more complicated proteins You need to transport something you need to do something and there is usually no specific reason why you absolutely need an enzyme To be attached to a membrane wall, right? There are a couple of examples like that and that's usually because they're involved in some process that occurs Very close to the membrane, but in general most enzymes are small and globular proteins and that's a very good question By the way, so how does an enzyme work? I know we haven't talked about it any guesses Based on what you know about proteins and thermodynamics and barriers this far Yes, so it has to do it's gonna have to lower an energy barrier, right because an enzyme is what a catalyst so in general an enzyme You know getting I need to be a little bit careful with an omicrater But the catalyst can't be consumed itself as part of the process, right? So catalyst after the entire process is done. We would still have the catalyst Occasionally the body still uses ATP Not for the process itself, but to somehow prepare the catalyst but So how could you imagine this reaction works if you have a couple of Source compounds or that you want to do something with a process you want to catalyze a reaction you want to catalyze Right so so it usually the reason why you have an activation barrier or something in a reaction is usually that Before you have for instance again, if you have two compounds and you want to fuse them to create a product or something You need to form a chemical bond for instance, right? Once that has bond has formed it's going to be beautiful because it's stable and nice But while forming that bond you're gonna need to say for instance to push push to charges very close to each other or For whatever reason put these two molecules in a conformation. That's pretty unfavorable Otherwise it would just be downhill. So what an enzyme usually does is that it binds the two Molecules and stabilizes them so that it is possible to put them very close together because you have lots of other groups around them This will then make it easier for these two residues i.e. Lower the energy barrier So they will form the bond and then they will be released again, but we'll come back to how enzymes work later Very cool produce So we got to one and two this far and three and three yes, and How is it that? So that it's a secondary structure, sorry It's a super secondary structure where you alternate between strands and sheets and how are they then organized three-dimension? Well, it's not so much circle form right, but that they go back and forth so that all the beta sheets for all these beta strands for one continuous beta sheet So it's not that you it's not that you literally mix individual helices and and strands because then those strands would not be Stabilized by anything but really that you have one layer of strands and then one layer of helices and If you take a fold like that and turn it and all the way around so that you have a small barrel of beta sheets And then helices around it. That's to be called a Tim barrel which is actually another that's not just a super narrow structure That's an entire fold So while we're at it what it's what is a Greek key? Yes, so this is a similar super secondary structure for beta sheets by far the most common super secondary structure for beta sheets So why do we introduce this concept of super super secondary structure? And that's really the same way why we introduce secondary structure right because remember these plots I showed you with tons of atoms all over the place proteins are complicated that it's too difficult to make sense of proteins if you're gonna try to Define and discuss proteins in terms of the location of every single atom or residue so the reason for having these terms that it makes it possible for us to talk about Common concepts without going into all the gory detail about the amino acids and everything So for instance when somebody talks about a Greek key, you already know roughly what type of beta sheets We're talking about and what this type of structure might be able to do While the Rossman fold as I mentioned that there's frequently small binding sites at the edge of the sheet or between the sheets and the Helices or something so if you're a skilled structural biologist or working with drug design The second you identify these things you can usually start to almost guess where the binding sites might be Before you have any functional data whatsoever It saves time for you Six how do beta sheets influence dimerization? Yeah, well, so the idea is that you frequently have a molecule where you have one small beta sheet that needs molecule Right, but at some point you have an end of a beta sheet And they're gonna be a bunch of potential hydrogen bond partners that when you have the isolated protein a solution These form hydrogen bonds with water if you now take two of these sheets together They can let go of their hydrogen bonds with water and just form one long sheets that continues from one monomer to the other monomer That is by far the most common way proteins interact if they're gonna interact fairly strongly It's much imagine two alpha helices, right? There aren't really any free hydrogen bonds in alpha helices There might be some side chains, but an interaction between two alpha helices will in general be pretty weak That's not bad. Occasionally you want weak interactions say hemoglobin the four subunits But if you want to read a strong interaction, it's awesome to have these strong Hydrogen bonds like five ten hydrogen bonds and then be the sheet So all this relates to seven. What is a fold and what perhaps just as important? What is a fold not? Right you could or a three dimensional Organizational what your mathematics typically call it topology of something so topology would be the way for instance If you take a knot and if you make a knot with a piece of string, right? There are actually mathematics there are actually mathematical ways of describing for instance about I know a ball line not or something You might do this with several different types of string You might do it in several different ways one not can be a mirror of another not but mathematically. It's the same concept So that's an even higher description rather than now of course in biology We don't we don't use the mathematical descriptors, but this could rather be for this is something that has six alpha helices in a particular confirmation or something But what is it not based on what you know about bioinformatics that you might suspect? If two proteas share the same fold does that mean anything? It might mean what? They might be homologous. Yes, but there is absolutely no guarantee what's there for it And of course if two proteas share the same fold They might have similar functions But it does not imply that they are evolutionary related So this is frequently an example of what you call convergent evolution. Did you talk about that in bioinformatics? That's some very in particular small simple folds as I mentioned there are only ballpark of a thousand folds So nature has to reuse folds even if things are not evolutionary related because they're a small stable And this is interesting I don't think I'm not sure I talked about that because this starts to be interesting You remember at the beginning of this course. We started to talk about does any protein? Sorry, does any sequence fold into a protein and here we started to see that there are possibly only a thousand folds in nature Maybe two thousand So that's kind of reduces the problem right the question is if you pick any sequence either It's gonna have to be stable in one of those 1500 folds or so or it's not gonna fold Now this actually turns out to be an astronomically small fraction of proteins that are stable in one of these 1500 folds So if you just go into the lab and make general sequences You're gonna form lots of beautiful molten globular that is just collapsed sequences, but they're not going to be stable proteins So nature has evolved our genomes to contain sequences that do form stable proteins So that events a matter of what you mean what you're learning a simulator So of course when we talk about protein design and protein engineering for instance if you have a choice Would you try to create a new fold or we try to reuse an existing fold if you if you not need to design a new protein That does whatever process and then sign Something that an enzyme does does not exist in nature to catalyze some important biotechnology process That will be the foundation of a new ten billion dollar company Would you try to design a new fold from scratch or would you try to reuse an existing fold and get an existing fold to catalyze your process Right because unless you reuse you're gonna be dead. There's no way on earth You will ever be able to invent a new fold that stable. Well, I wouldn't say that that people have done it, but it's It's an exceptionally steep uphill battle to try to create new folds and since nature again pick any of these folds You will likely find at least 50 if not a hundred different functions for a particular fold So nature the fold is really the Organization of the protein in which the amino acids are stable depending on the specific residues one fold can have a ton of different interactions So it's certainly possible to remember those four helix bundles. I showed you technically a four helix bundle is a small fold But you saw at least three or four different ways that that could work And then we realized you might be able to design a protein carrying hemoglobin by using a four helix bundle So pick something small and simple. Why shouldn't you pick something large and complicated? It's hard to do right and in particular the the actual folding kinetics the process by which this fold is going to be so much More complicated if you just have four helices either they are stable together or they aren't stable together And if you put lots of residues there that makes them form really super stable helices They are going to be helical if you have a 1500 residue protein that would have four different subunits and therefore sorry four different domains You won't have any idea how it's going to fall in the first place or how these different domains are going to interact or anything And that's kind of the same way nature does it What is more complicated your proteins or a bacterial protein? Bacterium is much smarter Because you keep carrying all these extra cargo to do complicated interactions such as the nervous system or anything Bacterial proteins are small because they're optimized to be efficient not to be smart Learn from bacteria whenever you can much better We'll get to consciousness maybe today actually. We'll see if I have time for it. Yes So that's a good question Well when we talk about full classification first full classification kind of gets important when you start at many folds, right? So when you talk about the smallest fold I Would I would argue that this depends on who you are if you're a biologist you would say that, you know Anything that's a smaller than 100 residues or something. That's just a polypeptide. It's not a real protein ignore it Now of course physicists on the other hand They love this small simple toy systems because well that's kind of the whole point of physics, right? Reduce it until the point that it's you should always reduce as much as possible, but never more so that This eventually comes down to what you mean by a protein. When is something a protein? When is something a protein versus just a polypeptide? Yes, and I would say so that something is a protein when they can has a fold and that says that it has one stable Configuration of it always reaches right on the other hand you could argue that if you have five residues in an alpha helix Is that a protein fold? Doubtful right? That's a piece of secondary structure So what the definition that we tend to use is as once you have a sequence that's large sorry a Sequence or fold if we call it that a structure that's large enough that there is going to be one or more residues that are completely buried Not accessible to water then you have you have an inside of the protein Even if it's just a single residue so for I think for the last decade or so people have argued that the smallest protein We know is this TRP cage, which is a tryptophan cage There's one tryptophan residue in the middle of a sequence and when this folds this tryptophan is not accessible to water So you have lots of other residues around it and from a biology point of view This is an exceptionally unremarkable and uninteresting protein, but from biophysics or folding concepts in it You can learn a lot from it how it folds Which of these classes of proteins has the largest diversity so apart from memory proteins? Yes yes, and That's in particular because building blocks have to be fairly simple right because the whole point of building blocks is that simple structure And use them hierarchically and there are way more functions and globular proteins membrane proteins. It's a good question What do you think are they? More or are they simpler or more complicated than globular proteins both in terms of function and structure or anything? Yes, you think that they're gonna have to be slightly simpler That's a very good idea. I'm not gonna tell you the answer think more about it. We'll cover that later today. I Would have agreed with you 10 years ago That's it's worse than that. I did I've actually written that in a bunch of research applications 15 years ago We'll come back to membrane proteins 9 10 11 and 12 are the ones we have remaining I'll let you pick Right and if you think about three-dimensionally that would mean what? Yes, but it's still really mixed. It's typically mixed in the same you say If you have you have a small domain that contains just alpha both alpha helices and beta sheets Although they're mixed in the sequence, but they're also very close to each other in space But this type of protein would frequently be what? Completely different domains right so you might have a membrane protein that is Alpha helix in the membrane and then outside the membrane you have a domain That's full of beta sheets. You're actually going to see one of those today Super-secondary structure already talked about so that's just really a concept to identify these very very common fact common patterns that are reused 11 and 12 Since protein isn't expensive as all you can buy it a ton of it probably a couple of dollars or something now If you put it in a shampoo for some reason the shampoo becomes very expensive And that's pure marketing 12 what's a permanent way? Yes And so that you you first reduce the disulfide bridges that makes the hair very floppy and And soft and then you oxidate them again And that means that you have now if you keep the hair in a hair or whatever either in the curls or just comb it out When you then reoxidize the disulfide bridges, that's actually going to fixate the hair in that form for a while at least And that's pretty much exactly the same thing as you do in biochemical studies when you call that you cross-link protein So if you want to find out if two parts of a protein and are two parts of a protein If you want to find out if they're close together in an experiment What you typically do is that you? Introduce cysteine and when you have these cysteines you can create a small disulfide bridge by oxidizing And if you created a small disulfide bridge with say between two parts of a protein that should move suddenly They won't move anymore So this is a fairly nice way of doing some engineering to understand the structure of a protein even if you don't really know what the structure is Because for this we only need a sequence, right? You can guess that two parts should be close in structure But you don't need to know the structure Good. Let's get on to membrane proteins Membrane proteins are almost entirely alpha helical Actually, that's not true Well, it is true kind of Most membrane proteins are alpha helical. There are some beta sheets membrane proteins in a barrel shape I'm not sure if I have any pictures of that. Let me get back to that later on today In the ballpark of 25 to 30 percent of all the proteins in a cell are Either inside the membrane or they're associated to the membrane in the sense that they're bound on the membrane surface You would not believe this and you look in the protein data bank for a long time When I was your age, I think there was one membrane protein structure in the protein data bank now There are probably 200 different structures. That doesn't mean 200 different folds Because the second people manage to determine one fold. There is shortly 15 or 20 structures of it, right? So I there's probably a dozen or maybe two dozen different membrane proteins false It's a much much more restricted landscape and that's simply because they're very very hard to determine structures for Why is that? Yes, so these exist in oil, right and it's pretty hard to crystallize oil Or fatty acids There might be 2% of structures, but this is the reason why they're important. It's 50% of all drug targets If you think that is high I've heard numbers and I should add that this slide that it's 75% of the revenue on the drug market Are drugs targeting membrane proteins? Because some of the really old drugs that is aspirin or so are fairly on specific functions Modern cool drugs against cancer or something they frequently hit membrane proteins So if you're in a pharmaceutical company It's more likely that you're working on membrane proteins than that you're not And that's because they're typically so important for signaling or so they do something very specific For a very long time both I And many other people used to argue that membrane proteins Might conceptually be simpler than Globular proteins and the reason for that is that we have the membrane as some sort of external Boundary condition here, right if you have an alpha helix an alpha helix really needs to go from one side of the membrane to the Other side of the membrane Why is that can't it stop halfway? Why would it be bad for an alpha helix to stop halfway in a membrane, right? Remember that you have all these peptide bonds here, right and an unpaired peptide bond That would then you would have a partial charge Right in the middle of a hydrophobic environment. That's astronomically bad nature hates that And that's why you end up having these extremely regular patterns of 20 amino acids Which is initially if any of you try to do predictions above membrane protein Topology in the bioinformatics course This is pretty much why it works right because you always you need 20 residues to go through the other side all the Time so even if you start being very uncertain whether this might or might not be an alpha helix If you are going through the membrane you will go through to the other side So then friend of order could say that that's great They're just alpha helices at least most of them and if you have these alpha helices Virtually all of them are going to be pretty much straight So you don't have these large crossing angles and then it's just going to be a matter of deciding how these Whether one two three four five six seven. It's a gpcr How these alpha helices packed together much easier than a general protein folding problem? It's a really good argument I've used it in real research applications that actually were funded the only problem is that is wrong We're gonna talk a little bit about structure prediction and third there is I talked about structure prediction That's pretty much the reason it works I likely won't talk about gpcrs, but I'm gonna talk about drug design and then my talk about gpcrs next week What is this you were almost right This is a fatty acid and this is fatty acid This is a lipid. These are the coolest molecules. You can imagine They are so Undervalued in life From many points of view these molecules are cooler than proteins Now I have to confess that I'm slightly biased here because I spent five years doing my PhD studying lipids But there are remarkably cool molecules So what type of protein is this? It's not a protein at all Sorry, sorry. Yes So why are these important compared to produce well from one point of view and they're not they're not really We typically will be clue them in biochemist your occasional at least organic chemistry But it's not really a bio molecule in that sense. It's not a molecule. It's not built in your ribosome It's not built from DNA Where do these come from? You eat them right and it's not just that you you of course you eat proteins to you eat food Protein digest the amino acid these amino acids are used to build other amino acids But these fatty acids you get pretty much directly from a hamburger or something careful what you eat You are what you eat literally The cool thing with it as an individual molecule, it's kind of boring It doesn't really have much structure at room temperature because these tails are exceptionally floppy But the cool thing is that they have average order if you put them together say in a micelle or vesicle or membrane This is a really boring way to show it And this is how well, I guess we would have thought this 20 years ago But you know better things than that so we can actually let's look at some simulations of lipids so we can see how these molecules behave So this is a this is actually not the single lipid I'm only showing a simple lipid But this is a part a lipid that is part of a large membrane so you have lots of neighbors around it But I'm only showing one lipid and This is also fairly short time I'm only showing one rotation this entire tail is very disordered and it moves So this was like two picoseconds or something in a nanoseconds details are going to be flying all over the place You have a large head group here, but there isn't really in a specific structure in the head group either So these molecules they have average ordering when they are together that here the chains are Relatively ordered and very far down the chains are very disordered and this is a dipole. You have a minus sign the phosphate and the plus sign Nitrogen there that goes that always almost always points up just so slightly But it's so the cool thing with this as individual molecules. They're completely dead and boring But as a collective of molecules they start to form all these shapes cells compartments and everything So I think this is a movie that I at some point borrowed from a colleague or internet Let's see this movie is bad and I'm gonna see if you can find it's bad So this is a cell it's just a computer animation of course and then you're zooming in on the membrane And then you're gonna see the lipids in a bit and then there is some corny speaker voice that I've removed But the whole idea is that it's it's exceptionally flexible, right? And I think it's gonna hit it harder. Yes, you can even hit it so hard that it breaks But the cool thing even at that even if it breaks lipids will actually heal themselves and reform I'll get back to that in a second and a real membrane contains not only lipids But lots of things like cholesterol Sugars and many of these things on the surface of the membranes actually what you recognize that antigens or binding sites on the surface of The cell there are a couple of things. So first, why do membranes heal each other heal themselves so well? Why do they form in the first place? Let's free energy for what? Well, if from one point of view, it's a really good answer that There was apparently lower free energy to go in that direction. So of course they went there But why is it lower free energy? What what is the reason why they form and why do they form this way? Right, do you remember that we talked a little bit how much it cost to solve it one? CH2 group for instance in water Like you have 15 CH2 groups in each chain They said a room. It's an exceptionally large hydrophobic surface here So trying to expose just one of these to water would be very unfavorable and here you have hundreds of thousands of them But if you turn the hydrophobic parts against each other and then just expose the charged parts to water Because that's the other part and say there are full They're effectively one full minus charge and one full plus charge here Putting one of those charges in the middle of the membrane would be just as bad So this creates a very nice stable Membrane or lipid bilayer here that if disrupted or something it will literally heal itself It's flexible What type of molecules can you strap transport through this one? Could you get water through? Well, you can you can always get water through the question is how high the free energy is So if you if you pick a if you pick a small patch of a membrane or something Something that you could simulate could you see a water going through it in the simulation? Well, if you simulate for like one second of real time, but you can pretty much forget about it It's eat my left shoe territory water does not go through this because water is so polar The problem with this picture and most textbooks is that they have drawn the lipids completely in the wrong way all Beautiful parallel tails. They're perfectly ordered like an army or something that is not actually that is how a set of lipids Look if it's a crystal or liquid crystal not in your body So in your body Things will look rather like this and this is actually this is taken from a simulation So you see this is complete chaos. So this is like a liquid hydrocarbon on the inside and This is something we've largely learned from simulations that it's exceptionally chaotic a real Membrane in a in a cell would also have tons of different lipids and everything you would have cholesterol bound Cholesterol actually makes the membrane slightly more rigid. I can probably show a picture of that on Monday Sorry, I should have included that today And then you have these head groups here that are full of charges and this acts as a very efficient Boundary here against the water. So here you have water, but you have virtually no water whatsoever here And this stuff like ten on a second or something So the problem here that there is no structure here. There's literally no structure whatsoever You can't well you can't crystallize it, but if you crystallize that you would get a completely different Confirmation so this structure you can't crystallize But on the other hand the order isn't random either right we know that there is water here There are head groups here and there are is hydrocarbon in the inside So how do we know this in this case? It's a simulation, but how do we know my simulation is right? well You can still do experiments on it So the reason why x-ray crystallography works is that you're scattering things from a molecule and hopefully get a beautiful three-dimensional Structure based on how much electrons you scatter You can try to do similar experiments here by scattering either x-rays or neutrons or something against these systems The only problem is that since they're so flexible and floppy and everything you only get a very small number of structure factors So you can get some information about the average order like this And this is based on really an electron density So you can show that you have lots of water out here. This is the center of the membrane So you have lots of water out here The green part would be the positive group in this case the choline the phosphates are slightly further in The carbonals are these COO groups that connect the head groups to the to each tail really and in the middle You have the tails and it turns out you can actually separate CH2 from CH3 groups by marking them with you mark the hydrogens with deuterium or something So we have a fairly good idea about the average order and here to you see there is no water whatsoever on the inside. They are They are very hydrophobic Based already on this plot Could you guess what residues you're gonna see here and what residues you're gonna see here in the membrane protein? Because this is still part of the membrane Yes, but what goes where? Sorry pole. Yes, the polar residues here, right and hydrophobic residues there. It's a pretty good guess What if you were to try to put something in here? That's let's be extreme. Let's put an eye on here Like plus 20 k cal per mole Based on that you can calculate how likely relatively it is to it is to see that one there How would you do that? Compared to have it the same water. Yes How many kT is 20? 25 kT or something so e to the power of 35 minus 35 Which is like I think I've written that three times 10 to the power of minus 15 And this is again what I consider eat my left shoe territory. It's not gonna happen. Sorry If you see that in the structure if you see something like that in the structure, what do you say about that structure? It's wrong something went really wrong there, but you're smart enough to realize that there's no way that can happen but This needs another problem, how do you what you do if you were at some point you we well I'm not sure about your cells, but I certainly have ions and charged proteins on the inside of my cells And so at some point we're gonna need my all the amino acids that I get from food at some point I'm gonna need to get them into the cell So to the body this does pose a bit of a problem It's good in a way because if anything could go through your membranes, we would be dead, right? The membranes provides some sort of compartmentalization But we still need to be able to get things selectively through the membrane or nothing will work and For a long time this was a bit of a problem because well obviously there are things that help us do that proteins in particular But we couldn't really determine any structures of them So we knew that membrane proteins were super important, but we couldn't really determine We couldn't learn anything from them. We could just get very low resolution indirect data And this is of course why they're cool and why lots of researchers has spent huge amounts of time on it When do you think we got the first membrane protein structure? Bitter in 1980s or so I would say and this is secret So one of the first really important structures that came in the 1990s was this small protein This is a very special protein in Archea called Bacteriorhodopsin And it's part is only called a purple membrane So these membrane these bacteria if you drive them out to have a membrane That's actually about a picture of that but you can search for purple membrane It literally looks purple when you have it in a dish The reason for that is not that the membrane is purple, but this membrane is so full of The gray stuff here is lipids all the red stuff is the protein It's so full of protein that it's more is this a membrane with protein in it or is it protein with a bit of lipid in it? I Guess it depends on your definition, right? Like 50% of the mass or so in this membrane is really protein So it's the protein that makes it purple not the lipids The reason for that is that you probably think of membranes as these large bilayers, right? And then with a protein here and there this is a much better approximation of a real for real Membrane there is way more protein than you're thinking that at least 30% but sometimes more What this protein appears to do is that it converts light energy photons to Electrical and chemical energy So it somehow pumps protons across the membrane that creates a load And the way it does that is that you have a photon hitting so this is the protein And you can kind of imagine that the membrane is here, right? There is this purple chain, which is called a retinal group and when a photon hits that one you have an isomerization So you have a double bond in the middle of this group that suddenly changes between from trans to cis So the photon uses the photon energy to kick this group up to a higher relatively unfavorable state And then when this falls down this protein is going to change this confirmation a bit And that helps us to move a proton across the membrane Which is really a charge protons are typically moved as part of water rather than isolation Or you might move it as part of a sequence of different titratable residues, which you do in this case You see all these residues that are titratable So one residue here can donate a protein to another and that way You can effectively it's not really going to be the same proton perhaps, but effectively you end up with one more proton on the Side here and one less protein on the other side. I'll get back to that in a second The cool thing with bacteria we wrote ops in is that by now we have a bunch of different structures At first sight they're all the same there are one two three four five six seven helices But can you start to spot some differences here? Right, there's a kink and there is maybe a helix that moves slightly closer to another one There are some conformational differences in this residue There are some residues that point in different directions and this is really how you understand any type of biology You need structural information is awesome, but even when you see the structures I'm not sure about you for to me. It's not entirely obvious what this does just by seeing the structures There are a number of ways you can get data from this you can even use say you can even use x-ray data And then use choppers you have super short pulses of x-rays Pemptosecond or so even and then you can have a flow You then can have a cell where the membrane Protein flows through here so that the laser hits the membrane protein And then we somehow measure the scattering or something a very very small time later And that means that you actually get a time-resolved x-ray experiment Because if this protein say moves with a certain distance per millisecond if you hit it with a laser and then you take a measurement a small piece later when this protein has moved just so much that you Know that you're measuring the protein one millisecond after the laser struck it Destructure you're gonna get is a structure that corresponds to the state one millisecond after being hit by the laser, right? So You can't really you can't directly an x-ray experiment is an average So you can't directly have an x-ray just take a picture in one millisecond But this way you can actually get time-dependent structural information There are lots of this is a fairly advanced such experiment But there this just stopped flow or continuous flow methods is a very common way to getting things time-dependent In general the problem with many of these experiments is that it's hard to get super high resolution data Because you're not gonna have a crystal or anything But by looking at the scattering as a function of wavelength where you can you can start to study features This is gonna be one that because this is because this is Wave numbers is one over the wavelength So very low numbers here corresponds to large features and as you go up here You're gonna study smaller and smaller and smaller and smaller features And then you can start to identify if something happened on that scale and that's what people have done over decades so it turns out that there are a bunch of different confirmations here and The numbers you see here. What are those? They're rate constants, right? And that relates to the kinetics we spoke about earlier this week So it's possible to use experiments. There is no way I can measure the actual transition But by measuring how long it takes for these processes to happen I can indirectly get the rate well rate constants or this actually one over this would be a rate constant But this is characteristic times for how long it takes for each step to do So based on this, can you say something about where the highest energy barrier is? 16 millisecond, right? So this is going to be a higher energy barrier. That is a very low energy barrier And I'm not this we're not going to talk about bacteria rotops in this entire lecture But let's see if I can show you a small movie there. Yes So first you see that they get this isomerization that pushes part of the structure very quickly to another state You're going to get the slight twist of these entire two helices here And then you're going to get the entire structure relaxing back again And then it moves back in and what has then happened is that because you were exposing or hiding these different residues here This caused the proton to move over the protein to the very small structural changes Today you can see so this is not a simulation actually. This is based on just interpolating between the structures But today you can simulate these features For deep protein couple the receptors people have done a ton of simulations to show what happens To produce like this So why would you be interested in a protein like this? Hey, our K are mighty cool, but They might not there's not a lot of health care that is focused on our K or obvious reason So this type of receptor could you imagine any other place where you have a protein that should convert light To a signal a human eye, so you have proteins called options and you're on your retina That small molecule is actually called a retinol that I had told you that you would have guessed what it was right And that's why I have that's where your retina is called retina. It's based on the name of that molecule The cool thing with the options is that you have different options with slightly different sequences And this sequences makes the opposite Make these options sensitive to photos of different wavelength So they have optimization maximum for different wavelengths, and that's how you get the sensation of different color in your eyes And that's a great example of something a human dust, but that would be very efficient for a bacterium or something There is no way a bacterium needs a retina. It doesn't need to perceive the world So an archaea is a much simpler organism and just uses this again Entirely similar type of protein, but just converts light directly to energy the bacterium needs This is part of a much larger column That's right These are also very similar to a large class of proteins called G protein coupled receptors But I don't think I'm gonna have time to talk about them today. So I'll talk about them next week But they are super important pharmaceutical This is part of a large concept that you're gonna see again and again and again a membrane proteins Something sitting in a membrane and transporting something from one side to the other That is pretty much the main thing membrane proteins do not the only thing there are lots of other things But it's by far the most common membrane proteins. I think There are a ton of different ones. We'll go through some of these So my first question is Why do we need these special proteins or something to transport things here? I've written ions, but this could equally well be water. There are proteins transporting water, too Why would you need a protein to transport water? Right, but what's the big deal? Can't we have our water on the outside and some water on the inside and be happy with it? Like make up your mind if you got sick if you have a certain amount of water on the inside Let that stay on the inside and let the water on the outside stay on the outside Well, you don't need so much water for that But for instance what happens with osmotic effects, right? If you change the salt concentration or the temperature or something At some point the pressure on the inside of this cell will start might start to build up If you can't release some water then you're gonna explode And that's kind of bad I've heard So at that point the cell somehow needs to find a way to let water out and later on you need to be able to let water in That type of So there are even mechanosensitive channels that sense how much the membrane is pulling and potentially open up and let things through There is a wide class of water channels called aquaporins that Peter Agier got the Nobel Prize for in 1997 if I recall correctly. No, not 1997, 1998 Eight or so I think I should know I don't remember anymore No 2001 2001 ha ha So what do you need to do with an iron channel and what what what do these proteins really do apart from opening up a hole? Well, so remember what I said what happens if you put it an iron in the middle of a hydrocarbon environment It's so expensive that it won't happen So Andrew Parsey Jen showed in 1968 and that actually is a year that I'm sure about that What all these channels do and this is mighty cool because this would like decades before anybody found out anything about the structures What he argued is that to stabilize an iron going through a membrane, which is hydrophobic You will need to shield the electrostatic and you will need to shield the electrostatic so that you get the epsilon here up to at Least well 20 or 40 or 20 or so so that relative to the electrostatic here That might be two or so very hydrocarbon environment So what this protein literally has to do is going to be to shield the electrostatic environment and So that you can't take these iron so that doesn't have to interact with the lipids And by then you can start to say something what what what what type of rest used gonna Are we gonna need to have on the outside of an iron channel outside facing the membrane? Hydrophobic right because otherwise on the outside of this if it's not hydrophobic It's going to be just as costly to expose a polar residue to the lipids I'm sorry. Sorry. No, I'm sorry. I didn't mean the surface but literally facing the biler and on the inside of this So when you call this a poor what type of rest user we're gonna need to have here Charged or polar at least if you compare this with globular proteins. What does this mean? What do we say about folding of globular proteins? Right, so membrane proteins should be the opposite of globular proteins when it comes to structure That's a really good statement. They it's wrong, but it's a really good statement We'll come back to why it's wrong. I would have guessed oh, too But then we have something else here too a pump The reason why I have this plot is that this is a huge part of what your cells are doing all the time This is an engine and that's kind of a clutch. So the way your cells Use a large part of their energies for neuronal signals. All of you have a nervous system That's not obvious a bacterium doesn't an archaea doesn't so then that's because you are vertebrates, right? You have a spine and a brain That's occasionally pretty useful, but there are some problems with it. They consume truckloads of energy Really lots of energy for it. That's why bacteria. I don't have it. It wastes too much energy This energy is primarily stored in the form of ions potassium ions and Sodium ions and these ion channels are so cool that they're super fast to react that this ion channel only let's do potassium This ion channel only lets through sodium. They pretty much never ever make mistakes way better than any computer or anything when it comes to accuracies and because there are too many there is an Excess concentration of potassium on the inside and sodium on the outside The millisecond these channel opens it boom it will instantly let ions through There's only one small problem with that and what is that? Yes, and what would happen when you have the same number of potassium on the inside and the outside Absolutely, nothing happens when you open them work. So these channels are Very efficient, but something is going to need to transport the potassium so that you have excess potassium here and excess sodium there But that's easy, right? We can just have another channel that achieves exactly that or How do you achieve that? There is a problem there You're gonna need to pay right because you can't spontaneously transport potassium to from a place where you have low Concentration to a place where it's hide concentration. It goes against the gradient You can say that from a completely different. So why why can't that happen? No forget about energy bears Nothing to do with energy bears Has to do with an important concept. We picked up early on in this course This thing blob in the middle what I'm saying that it does is that it takes for is a sodium from a place We have low sodium concentration or Actually, you can start assuming that we have the same sodium concentration on both sides And then this place would take sodium here to have more sodium on one side and it would take potassium here So that you had more potassium on the other side Sure, but that's how the day that's our nature and everything does it. So what I this has to do with entropy, right? So you're taking something that's relatively disordered you have the same concentration on both sides and creating a state that is more ordered That's reducing the entropy And if you're reducing the entropy you would move to low as you would move to higher free energy, right? So forget about everything we know about nature and how nature might do it You can say that's physically impossible that can never happen spontaneously You will have to you there has to be something else that pays that energy and that other thing is exactly what you spoke about This is a molecule called ATP aces It's a wide class of Proteins Discovered primarily by James Christian school, you know or who's in Denmark and got Nobel Prize for it So what ATP aces does? Not entirely shocking that ATP aces surprisingly uses ATP which is the way Your body's stores a lot part of its energy. So this molecule binds Ions and ATP and then it undergoes some very complicated rearrangements. Let's see if I can no, sorry This is just a still movie. I'm just rotating it here. It's not going to move and Then it uses this a one at once it has bound this ATP This will force the entire iron channel to move over to a slightly different confirmation because when you're bound ATP It is apparently more favorable for this channel to change its shape slightly as this channel It's changes its shape. This ATP is losing one phosphate. So it's moving over to ADP. That's how we pay the energy and While doing so it's also moving Ions in opposite directions So it's pumping ions against their gradient. So it's and then eventually you're releasing the ADP and This channel will be back to its initial state How frequently do you think this happens in your body? How much a just for fun to get it? How much ATP do you think you use per day? That's like your body's weight There's like 30% or so of the energy turnover in your body is used on the system And that's of course by bacteria has gotten rid of it because it's it's it's a nice thing to have a brain and a nervous system but you do pay and Of those 20 or 30 percent roughly two-thirds of it is spent on your neurons in your nervous system We know quite a lot about all these states. There's particular a big group in August led by Paul Nissen that have determined I don't sure it's more than a dozen channels I should know exactly how many and we've collaborated with some of them But the cool thing is that we have Not just individual states, but we pretty much have all the skills of the entire movie We know from every single state here. Do you have two ions bound? They have three ions bound How has it moved between the different confirmations and then it's a remarkable puzzle to Identify all these things in the right order and show how it might be possible to move between them a Minus under sign up in our team has worked a lot with Paul on this doing simulations, too And what you typically get is this so this is say an electron density is from an x-ray structure And I know you're smart you you instantly realize how this protein works, right? The reason we're showing that that it's it's hard. It's really hard even in principle. We know where all the atoms are here But to get from where the atoms are to understand how a protein works, it's not trivial So what you end up doing here is that you need to sit down and Assign different helices and everything where they are sitting in this particular structure And that's going to be the same for say cryo-em Then you can start to place side chains how the side chains must be placed You will eventually be able to say what our potential binding sites for ions in this case You're actually lucky because it's a fairly high resolution structure so we even see the ions in the structure this green electron density and Then you have to start comparing different structures see what happens with these cavities as we are moving this Between different confirmations. This is a this is just a short simulation from Magnus I don't know we're not going to see the undergo any state changes here But there you see the three ions bound on the inside This is also an example of you do work like 20 years ago people have been perfectly happy with the structure But today when a group determines the structure you want to simulate it You want to the place where you think you might have had nine is the iron really stable there Is the energy for instance if you're forcing the iron out by pulling it the energy you see is that a reasonable energy? Can we predict that this is a reasonable transition to undergo so in many cases we use simulations Not necessarily to predict things but to understand that the model we had from an even coarser x-ray experiment seems to be right and That's why I'm going to say it's so dangerous We you just see these with all the atoms in the PDB and think that's the structure and it's it of course is the structure But you should be aware of that the experimental structure is that this is the experimental result everything else is an interpretation It's usually a good interpretation by some very smart scientists But just as the way simulations have lots of shortcomings What you really should do is of course take your simulation, right? Calculate the electron density from your simulation and see if that agrees with this density Saying that an atom has deviated by one angstrom could as well be an error how you place this from the experimental data here So that's half of the process and I'm going to get back to the iron channels in a little bit But there is something I've cheated a little bit in here. How does this end up in the memory? Well, yes, there's going to be a membrane roughly here There's like it's not like you can cut open a big hole in the membrane push this in Because if you cut that hole open the entire contents of your cell would escape Again, you're going to need to produce new proteins all the time, right? And if you did this every millisecond in your protein every time you did this You would equilibrate the distribution of ions and everything your cell would die once every millisecond So we're gonna need some complicated mechanism in this asserting this and for a very long time We didn't know There are a couple of different models here that are worth knowing about one of them. I haven't told you yet You remember these floppy fluid models. I showed you of the membrane that is a Model originally proposed by Singer and Nicholson. You don't need to remember those names But this model is called the fluid mosaic model And the idea is that you have a membrane that really consists of Individual lipids can move around and an individual membrane protein even diffuse around inside the membrane just like the diffusion lab you did a Human membrane protein under normal conditions will diffuse around but in 2d instead of 1d and the probabilities of two proteins Interacting colliding something participating in reaction. You can actually calculate that based on their diffusion rates there is a fairly large lab down on floor 3 in this building led by Yelma Brismar and They are using membrane proteins and ATPases and then they're trying to mark these with fluorescence probes When you mark these with fluorescence probes, you can see where they are The only problem is that this protein might be 10 nanometers And what is the light the limit of resolution of light in a microscope? Roughly the wavelength of light, right? So you can't really identify where they are This is why this group is one of the largest as we did using these new type of super resolution microscopy that got the Nobel Prize two years ago So that you can actually identify in particular where ATPases are with a resolution that is in the ballpark of the size of the protein So you can identify not just in what cell but in exactly what part of the cell we can Mark these so you see when they open up So you can actually in live cells see what is the relative activity of the ATPases when do they open where do they open? How do they open and now we're not talking about computer simulations This would real microscopy on a real live cell and there are even techniques that can occasionally do this in a live animal Say a rat brain or something This is still a biophysical technique But what I'm trying to do that we we're we're learning a whole lot of what we've learned about the energy turnover and everything Of course comes from experiments like that very simple biophysical. Well, no They're not simple biophysical experiments very complicated biophysical experiments But then we can actually also measure that how fast are these proteins diffuse in the membrane and that's how we've learned that It's not really a rigid crystal and that's why I want to emphasize this classical picture with all straight tails It is so wrong. This is like a two-dimensional liquid. It's water It's like water, but it's not water of course, but it can only move in two dimensions But inside the membrane anything is free to diffuse So that's the first model the fluid mosaic model the other model is a model proposed by Rick Popo and Don Engelman and that's called the two-stage insertion model or the Popo Engelman model and What they're arguing is that membrane protein insertion or folding happens in two stages You have individual helices inserting in membranes And this is a crossover simplification I'm not saying that they insert directly, but the idea is that each helix should insert independently of the other helices in the membrane protein and Then magic happens and somehow these helices aggregate to form a protein Now for a real protein, there would of course be some sort of loop between the helices or something Let's just ignore that for a second, but they're part of the same protein Is this right? So could you imagine an experiment where you could test it? So what experiments would you then do? So the one problem is of course is that if you start removing part of your protein You might very well have destroyed the structure of your protein, right? So that might be this model might still work really well. It's just that you screwed up the structure So remember there was not a coincidence that I showed you Rolopsen So what they did is that they took one of these seven transmembrane helical proteins And that is of course expressed by a gene But instead of having this is one gene Let's cut this gene into two pieces So you have one gene with three helices and one gene with four helices and the cool thing And then you knock out the original gene so that the single gene with seven proteins do not exist anymore And this means that you now have one gene where you express three helices and another protein with four helices And the cool thing is that they showed that you still have activity in the cell So somehow these two different parts they diffuse and find each other in the membrane You certainly lost like 50% of the activity because it and that's reasonable if you think of diffusion or something You there there will be a limit that if they are free, they're not gonna meet each other that frequently But this is pretty much the first proof that of this concept that we're first inserting and then aggregating proteins. I Would argue that this might very well be a future Nobel Prize It's starting to get a bit old and There is this thing that this covers when discoverers become too old They're no really obvious They become so obvious that they're part of the general truth and then they can be hard to argue that they're great discoveries But don't angle mine in particular has done an amazing amount of stuff for membrane proteins And that's I realize that's we've been I've been reasonably good in this course at predicting over prices I've done lots of mis predictions But another problem if you have this is the same protein But does it insert so you have two loops here and one loop there or will you have two loops there and one loop there? What decides the order in which the protein is inserted? Well, Popo and Engelman didn't really say right in principle. Could you imagine both? Well, that would be a bit bad. What if that ATPase was inserted both ways So you would have you would now spend 70 kilos ATP per day and on average the molecules would just cancel the contributions Right, that would be somewhat bad. So we need in general. It can't be random We have to control the nature has to control the way membrane proteins are inserted Otherwise you couldn't uphold any sort of difference between the two sides So good our von Heine and our department actually discovered this some decades ago Which was one of the first examples of bioinformatics and they were just studying what is the occurrence of charge residues? And it turns out you actually have charge residues in membrane proteins It's just that they don't occur here, but they certainly occur in the loops and Gunnar noted something fairly fun that There are more charges on the loops on the inside of the cell Sorry, there are positive charges on the inside of the cell While the charges on the outside are usually negative. So this is what you call the positive inside rule How do you think you prove that? So it's very easy, right? Just do forceful mutations swap that with that that with that that with that and that with that They're in the loops, so they should not affect the structure and suddenly you can show that you get proteins that insert the opposite way very cool experiment I Would argue that this too could be a Nobel Prize in the future The reason gonna won't get the Nobel Prize is that he's the chair of the Nobel committees Sorry, they can the secretary in the chemistry Nobel committee now against and I which kind of disqualifies him Otherwise, I would I think that Engelmann and von Heiner had would be in a very good chemistry price The other sorry the other part though with this process was that at some point you're gonna need to get these ceilings is together And from one point of view we still subscribe to this The beautiful simple picture that these proteins are so much easier than globular proteins We still believe that one this far this looks simple this far, right? We don't know we don't know exactly how they assert, but we do know that they assert It's beautiful every single helix is vertical here, and now we're just gonna need to find a way to pack them The beautiful thing is that don found that So they found that there are some small proteins in particular protein called glycophorin A that is a single helix but they dimerize and In these protein they love to have glycine in the middle Actually, not just glycine in the middle, but you have a glycine and then you have one two three other Residues, and then you have a glycine again Do you see why this is so remarkably good? It's not might not be obvious. Remember the pictures I showed you was it yes on Wednesday when you show the surfaces of the helices And I claim that there are edges and valleys Those edges are ridges correspond to the side chains, right? And that's kind of what limits how close you can push two proteins together to each other These glycine residues when they're spaced three apart It's not entirely easy to see two dimensionally But this one is actually a bit out in from the screen here while these two glycines are the ones that are closest to each other So you're now gonna have Two residues that are close in space and there really aren't any side chains there If you have two helices like that That's gonna be like having two helices and both of them have a small depression on their surface because you Don't have any side chain Those two depressions are gonna love to just bind together and this will form a dimer in the membrane And they showed this with lots of experiments But you know else what else was was cool that what would happen if you start to insert if you use a charge Residues they will never ever insert in the membrane But what would happen if you started to say put a slide an asparagene or something? That's well something that's polar, but maybe not charged here Or here What's gonna happen to insertion it's less hydrophobic There's not gonna insert quite as well right You can actually show that it does insert a bit because it's not a free charge. It's just it's bad It's not astronomically bad But what happens when these two are inserted and you now have two side chains that can form hydrogen bonds But you just put them in a place where they can't form hydrogen bonds because there is no water So they can form hydrogen bonds with each other right there is no water Remember the thing that I was told about hydrogen bonds that it's mostly entropy and water because it's a matter where you have the partner Here there is a difference between not having any hydrogen bond whatsoever versus having a hydrogen bond So this becomes a gigantic driving force for dimerization There's a beauty so this there's an entire example of these motifs It's called the gx3g motif and then there are probably dozens of papers where people have tried different mutations inside these x's When this was found I think I would argue that not the general consensus, but a lot of people partly included Thought that we had solved the memory protein folding problem So we found this is called a motif and a motif is just really a small pattern in a in a protein or something That expresses something important to binding site or something. So this is good This was the first motif we found and the other the other question is how many other motifs that yes There are and when we know enough of these motifs we can actually start predicting membrane protein structure because we will know how they interact So that's a good question. So of course things have evolved a bit. So how many motifs like this? Do you think we have nowadays? Yes, it's one This is a really good idea, but it didn't work And where people have certain and they're pretty much this is like a freak of nature probably there are no motifs like this This is the only one So this is the problem is that membrane proteins are not like we thought they were it's not It's not simple. It's not At all simpler than the global approach is I would argue But in that case, so why but why don't we see any complicated membrane proteins all the membrane proteins? I showed you this far. I've been relatively simple, right? Well forget about the ATPase for now because that was a later product How does it come that the membrane proteins that I showed you are so simple the early ones? Yes Do you see any correlation there? What is it easier to get a structure for a complicated or a simple membrane protein? So the problem is right all the ones we got the early structures for us were the simple and low hanging ones While the difficult ones we couldn't get any structures off So the more we've learned the more we realize how little we know So this is really the problem which is failure, but it's actually this is a I talk about this as failure But I think this is a remarkable success of science when I was your age and Significantly later than that we thought we had solved this Bacterioridopsin 1990 that was the year I got my undergraduate degree. No, actually was 91 No, 91. I started my undergraduate degree. Sorry. I'm not that old But I'm apparently a bit suffering from Alzheimer or something Simple beautiful protein seven transmembrane helices it fulfills every single rule I spoke about and then in 1997 This is the reason why though what I kept saying 1990 97 was the year Peter Eger published aquaporin Do you see something here? Yes, but on the other hand, I've also said that there is no rule in biology without an exception this is the exception right because Well, it doesn't quite go through but it's this helix ends and then there's a similar helix from the other side taking over So how bad is that going to be? Not really right because they can form peptide bonds those peptide bonds will form peptide bond with that helix It's just that it's going to be a different part of the sequence But every single peptide group here can form a hydrogen bond with the next helix So it's kind of reasonable to argue ha fun freak of nature, but that's can probably still be stable The last few years though, we start to see proteins like this. Do you see that? It's a helix oriented horizontally in the membrane some really complicated binding side Disordered residues things that go in and then turn out again This is just an example. I've stopped updating this slide that there are dozens of proteins like this now Membrane proteins are not simple. Sorry. It's just that it seemed like a good idea at the time This is how we thought in the 1990s It was reasonable to think that membrane proteins look that way because by 1990 every single membrane protein We knew of look that way Which is of course because we only knew one membrane protein, but 100% of those proteins look that way Do you see any beta sheet membrane proteins here? They do exist, but they're very rare and there is this pretty much a small class of proteins in the Outdoor membrane poor proteins you can if you take a beta sheet and curl it up and rotate it So you let the end of the beta sheet join the beginning of it. You can form a small barrel We know even less about those because we only have a handful of examples So I might not even talk about them but if we look about membrane proteins most of you are taking the bioinformatics class and if you look at the hydrofobicity For instance for all residues most residues are Hydrophobic the exposed residues are hydrophobic and the buried residues are hydrophobic So everything that goes into the central membrane part is hydrophobic while things out in this head group can be Slightly polar There is a problem with that. I'll let you think about that. You can look at the secondary structure and again This is statistics over all known Mem I forgot whether it's over all known membrane protein structures or sequences, but it doesn't really matter that If you have to take a guess for a membrane protein that you don't know anything about it's going to be alpha helical There are some very few examples of strands, but they're pretty you can pretty much forget about it But there is a problem here last slide You had a pretty good idea that I told you was wrong What is the problem with this slide? So exposed here means exposed to the membrane surface Buried means that sorry that towards the lipids and buried means that you're turning something to the inside Of the membrane protein Membrane proteins are not the opposite of globular proteins. They're hydrophobic everywhere And if you think a little bit about it, that's kind of makes sense in this insertion model because if the if the if the Helices insert one at a time They're gonna have to be stable in the membrane one at a time where they would not insert, right? So then you can't have a helix that's half hydrophilic Once you had folded a membrane protein that would be nice, but it's never gonna insert So this means that they're not the opposite And in fact that they're gonna be way harder than globular proteins because globular proteins You can at least predict if it's something large and charged it's gonna be on the surface if it's something hydrophobic It's gonna be on the inside Membrane proteins all bets are off and that's why they it's not just easier. They're significantly harder than globular proteins Which is kind of fun because it does provide job security if you're working with membrane proteins But it's not easy to predict their structure They do and that's how each helix of course is a Bible But remember that modern membrane proteins are not just plain simple helices, right? That they can be almost over the place There are some things you can do to identify structure in bioinformatics that if you look at the sequence variation and Information content here really means that How much let's see did I screw that up? Is that in the wrong order? Well, I forget about this by those but the point is that residues that are exposed to the membrane here They're kind of passive. They should be hydrophobic and Things that are related to packing between residues. They seem to be slightly more conserved And what that means that the things that are packing between two helices or something that it appears that it matters Slightly more what residue would put there Because if you replace an alanine with a tryptophan, they're not no longer going to be able to pack While things that face the lipid they should just be hydrophobic so that the lipids that we're happy in the lipids So there are some ways of being able to at least try and predict membrane proteins, but In contrast to globular proteins membrane protein structure internally is almost entirely Lena Jones packing We need to pack the helices in some way and these are much much much weaker forces than electrostatics And that's what makes them really hard to predict The only advantage is that we get some guidance that the helices really need to go from usually Need to go all the way from one side to another but even that isn't always true This might be a good place to take a break after the break I'm going to talk about how membrane protein folding happens in reality because there is one big Elephant in the room that I haven't talked about how do the helices insert in the membrane in the first place It's 10.25 now. Should we meet here at 10 to 11? Alright, welcome back. So that must have been the world's longest 20-25 minute break What happened is that I had a slight problem with recording equipment after the break probably because either the battery ran out Or because I had to reconnect the computer So normally I would just tell you to read the chapters in the book But given that this particular lecture on membrane proteins is much more focused Well a on our research, b on our colleagues research and c on the slides It might have been a little bit difficult for you to read this up entirely on your own So rather than saying that those of you couldn't be at the lecture It's your own problem. I decided to re-record the second part of the lecture at least I might do this just so slightly faster than I normally do at the lectures but Let's get started. So before the break we spoke about sequence variation and All the stuff we did up into this I kind of cheated because I just said assume that there is some sort of way that the membrane proteins can make it into the Membrane, but I didn't really tell you how that happened in reality So the short story is that it doesn't happen quite that easy in real life in real life We have lots of other Mechanisms molecules in particular the ribosome and the translocum So what you see on this slide is that you have the messenger RNA string in green there Which carries the genetic information into this gigantic molecule in blue the ribosome and what the ribosome does is that it connects amino acids with peptides bond to each other just as we described in the very first lecture in the course and Then this ribosome is typically attached to a membrane So this is not any membrane at all This is a particular part of the cell called the endoplasmic reticulum that you're probably aware of which is really the protein factory of the cell So what the ribosome does is that it gradually pushes out these New protein the nascent chain or the polypeptide that it says here through the exit tunnel in the ribosome and Then the ribosome in turn is connected to this translocum and a translocum is really a channel But it's not an ion channel. It's a protein channel at the time when CY I think made this illustration on the left there well, the same the illustration is probably older than the 2005 paper, but Eventually we managed to get structures of these translocum by Birgit Al And that's what you see on the right So that translocum is really think of this like a small sort of barrel with two big arms around it And what you see on the lower right Sorry on the top right is where you see the translocum from the side. So this is sitting right in the membrane plane and Then on your lower right, you see the translocum right from the top and In particular, you see a small green helix and a small blue helix right in the middle on the right there So what happens is that these two since these are not connected the blue and the green helix the translocum can Really open up its arms and let something in particular helix out into the membrane So what happens is that? Depending on what sequences you have in this chain They go through the translocum and if it's very hydrophilic it will just well It might stay here a bit But it's not really going to garden the membrane because that would be too costly and eventually This will be a globular protein that ends up on the inside of the cell Remember, it's not the plasma membrane, but the ER membrane But for some other sequence We might have very hydrophobic residues and these residues are going to stay here a bit because they don't want to continue Because if we push them out in the cell, it would be very unfavorable And then this translocum likely breeds a bit all the time so that it occasionally opens For a hydrophilic residue that wouldn't matter. It would be bad to go out But the hydrophobic residues they would happily now diffuse out into the membrane and then this would be a membrane protein Not too surprisingly perhaps So having said that you probably all agreed that membrane proteins contain hydrophobic residues While globular proteins contains hydrophilic residues and the translocum kind of helps us select which is which But that's the model that we have to test this in practice to and this is actually be done by a colleague of mine Gunnar von Heine in our department and one of his Students a few years ago Tara Hesse And they came up with this smart experiment that if we really want to measure The extent to which a certain segment say helix inserts in a membrane We're gonna need to find a way to compare how much is inserted versus how much is not inserted ie pushed through the translocum Which you also called translocated And if you look in the upper left on this slide They took a small protein called leader peptidase which is really just three small helices that were hydrophobic But then they took in particular the third helix of this small protein and started changing that helix That's the one in red In addition to changing that helix they also made sure that there were two glycosylation sites before and after the helix and This is not a course on biochemistry So I'm not going to go into details with this glycosylation dust, but it's essentially adds small sugar fragments on the Peptide and by using these sugar fragments and fluorescence We will get a signal when this is Translocated But sorry We would get a signal with this up in the air lumen, but when it says in the cytoplasm the signal is cancelled so what that means is that If the entire segment is translocated we're going to get both the glycosylation one and the glycosylation two signals But if it's inserted is that we're only going to see one of these signals So if you look on the lower left that's exactly what they've done There are a few different sequences there. They've tested and depending on What signals you see here in this case on the gel we can decide how much of this protein was inserted versus how much was not inserted translocated And based on what you know in the course before you know how to use that measurement and turn that into a free energy Right because the second you know what is the probability of being state a versus what is the probability of being a state b? Well, that's determined by a Boltzmann distribution, right? And when you take that boltzmann distribution and really assault for the free energy instead of the probability You're going to get the result that the difference in free energy is minus rtln and then the quotient between the probabilities And so in chemistry we say rt a physicist might say kt here, but that's just a matter of units And that's what you see down there, right? It says an app for a parent. I'll get back to that in a moment But the cool thing is that the second you can measure two things in relation to each other. You can usually convert that to a free energy For going from one state to the other And that is beautiful. Then you can create this nice wonderful red plot you have on the right that for each amino acid Which has how costly is it to insert this in a membrane? And somewhere there I would expect to hear some protest from some of you because what's the problem with this? Well on the one hand, it's beautiful, right? You have that we gain some energy free energy from inserting the hydrophobic ones like isoleucine, leucine, and phenylalanine While the most expensive ones to insert are the charged ones, or EKD But look at the Y scale The Y scale appears to be extremely compressed. We are we only gain half a kcal per mole for inserting isoleucine And we only pay like three three and a half kcal per mole for inserting a Charged residue. Do you remember the first part of the lecture? Taking an ion and forcing that to be in a hydrocarbon cost how much? Well These residues are really ions. They're amino acids, but they're charged and therefore they're ions And you can even take some of these side chains say the arginine side chain and Use that as an isolated molecule and pro-arginine. That would be called a guinedinium ion And the cost of inserting that in a hydrocarbon is roughly 17 kcal per mole. We know that from simple partition coefficients So there is something fundamentally wrong that it should cost 17 and in practice it costs roughly three There is something here. We do not understand So can you think of any solutions to that? Well, one of them could of course be that it's magic. We have a translocon Well, not necessarily magic, right? But it's obvious that if you have a biological molecule that can drive this process It's going to be completely different than from just inserting it directly, right? Or, so what did we say about free energies? Last week or a couple of days ago at least Well, the free energy is a state Property so the free energy of being in a particular state only depends on that state It can't depend on the way you took there So if you're there the free energy from going from A to C That depends on the free energy in A and the free energy in C. It cannot depend on what happens in B It can't be cheaper for go to go A, B, C, then just to go A, C If it's under equilibrium If that was not been the case, we end up in the argument I had a few days ago about perpetuomobilist, right? Because if it was cheaper to go one way You could pay the cheap way and gain the energy of being in a particular state The cheap way and gain the energy back The way where the difference was much larger and then we just keep gaining energy So that doesn't work In principle you could argue that the A state here versus the translocated state At the bottom where I don't have any letter are slightly different But in practice they're not they're virtually identical in ion concentrations and everything So you could of course still imagine that the body does something amazing in B But no matter how amazing that is if we believe that there's equilibrium between A and C It doesn't matter what happens in B B might speed up the process That might be a catalyst It might certainly be good for kinetics But it cannot change the equilibrium distribution So there is a famous quote from a old episode of Simpsons that I like to use here That young lady in this household we obey the laws of thermodynamics I think it's saying it's the daughter And that's the beautiful thing with thermodynamics It doesn't matter what type of magic happens in B That cannot change the equilibrium distribution between A and C So are there any other things we can think of? Well both we and others have looked at this with simulations over the last 5-10 years And I would say that there are two potential explanations The first one is that I've reminded you about a couple of times is That a membrane is not a lipid bilayer And what we're showing here is really a protein sitting right in the middle of a lipid bilayer Right? But remember that membranes contains in the ballpark of 30% proteins And those proteins while they are not super hydrophilic They are certainly not as hydrophobic as a pure hydrocarbon that you would have in C So if we now imagine that you had lots of protein or something in C A real membrane might actually not be as hydrophobic as a lipid bilayer And that's actually confirmed if you look at MD simulations that we are We had a really talented student and I do also want to do this a few years ago So that can kind of explain that it's somewhat cheaper to insert a charged residue like arginine or lysine in particular in a membrane than in pure hydrocarbon Because a membrane simply isn't a pure hydrocarbon But can we think of that some other ways? Well when it comes to arginine You could also imagine having the arginine helix in the middle of the layer But then because there are so many charges the arginine is not stable there So what the arginine would do in practice is somehow slide halfway out or something So that these residues could snorkel or interact with the charged head groups If you only have one arginine residue you could certainly imagine that's the case The only problem is that if you repeat this experiment But then use two arginines and place them symmetrically and just slide them further and further out Well the problem is you can't slide to both directions at the same time, right? So in that case we would expect those type of sequences to be extremely costly And they aren't really that costly So when it comes to arginine though the simulations I just told you about that Pretty much explains it within a factor two or so On the other hand, lysine? Lysine most certainly won't slide halfway out But what's the problem with lysine? Well remember that a scale I showed you is not just a matter that we are not paying as much as we should The entire scale appears to be compressed So the lysine residue itself appears To not we don't gain as much from inserting lysine as we would expect This too can probably explain be explained by the fact that membranes are not quite as hydrophobic as we would expect So that the scale the difference compared to water is simply not quite as large as we would have expected The problem though is that when we and others have done simulations of this We don't really see much of that effect for lysine So lysine we haven't really been able to explain I think that we have we can with a bit of hand waving But I'm a bit more uncertain when it comes to losing than arginine But the other thing is what we say at the bottom of the slide here that there is actually Had you asked me 10 years ago whether membranes are stable membrane proteins are thermodynamically stable in a membrane I would have said of course But if you think biologically that's not absolutely necessary It's enough membrane proteins need to be stable in a membrane But they only need to be stable for biologically relevant times say months or something until the protein is degraded by the body So what you might imagine is that some proteins might actually not be happy in a membrane So why do they stay in that case? Well, if you look at these helices the second you start to slide them out You would need to drag either the n or the c terminus of the helix Through the most hydrocarbon part of the bilayer right you would expose three peptide bonds And it's actually worse than the peptide bonds because you also have these effective helix dipoles So it will be extremely expensive for his helix to slide out Second from the body's point of view and the cell it would be bad if membrane It really helps that membrane proteins are really nice and stable They don't really move up and down if membrane protein helices were diffusing a bit randomly up and down I think it would be very hard to have stable membrane proteins So what might be the case is that The translocon uses Well, the ribosome uses energy to push things in the translocon and the translocon then helps us to insert things So that we get the particular helix inside these head groups that we have on each side And once it's here, even though it might not be super happy there It would be a very high energy barrier for the helix just to slide out directly And that might very well be the case so i'm Big caveatemptuers here. This is certainly not something that has been proven experimentally But it might it just might be possible that Membrane proteins at least some of them might only be kinetically stable in membranes rather than thermodynamically stable And that relates very much back to what we've talked before in earlier lectures about valleys for thermodynamic stability And barriers for kinetic stability and the free energy landscape So having said that you probably know this quiz by now If I show you two small helices here one on the right That's polyalanine and one on the left that contains lots of arginine and to make life a little easier I can even add lots of charges on the one on the left I have occasionally run this quiz without doing the further two or three Slides prior to this one and it's surprisingly how many the senior professors Don't realize how important the one on the left then Because it turns out that the helix on the left is not just a one that occurs in a membrane protein But it's possibly one of the most membrane protein helices we have So this is a segment in a voltage gated ion channel If you look at the protein here, you see a big central pore right in the middle of the pore You have this small channel that lets through an ion And on the outside you have four small domains in the corners that are these so-called Voltage sensor domains and they contain some four helices and the fourth helix in each of these domain contains these arginines So why do we have an arginine helix there? Well, these proteins are super important for say conducting nerve signals in nerve cells They're important for your heart beats Converting those signals into muscle activity They're important when the sperm fertilizes an egg to close the egg It's a super important part of your nervous system And these are really the proteins that consume all those ions that the sodium potassium ATPase was pumping And if you think a little bit about it if you want something to change Function or change form shape confirmation when you change the voltage Well having a charge is the obvious way to do that If you have a charge and there's a voltage change or voltage difference There's going to be a force on the charge And that force is ultimately what's going to move these helix up and down in this protein And that's the reason why we get these nerve signals So a nerve signal this so-called action potential, which you can also see in an EKG That is really just a chain reaction of these Voltage gated channels one channel leading the opening of the next leading to opening of the next leading to opening of the next In reality, it's slightly more complex because we also have sodium channels But I'm I'm going to leave that out for the topic of this lecture So the way these voltage sensors work if you look on the Structure to the left there is a blue helix in the rear. That's the one with the arginines And what happens is that somehow this helix is going to need to move up or down So it actually turns out that in the very first structures people got Rob McKinnon in particular They saw this helix together with the helix before it Lying like a small paddle right in the interface between the membrane and the water So this was simply an artifact of the crystallization conditions that the while this was happy in the membrane for whatever reason the The micelles or the detergents that were used today to overexpress the protein the protein wasn't stable in that confirmation Now to roge credit He was I think the first group to realize that this must be wrong to and a few years later They published the correct structure, which is the one you see here But it's not entirely sure right because if the blue helix is going to move up and down You're going to have at least four arginines there that are exposed to the membrane and that can't happen We just said that's eat my let chew territory So the way this happens is that first they don't exist in isolation as you see in the figure in the middle the lower middle There is also the central poor domain and the central poor domain is really going to be the neighbor of most of these arginies So they're at least the one in the middle They rather face another helix than the lipids not quite as good as water, but certainly not as bad as pure hydrocarbon The other part is that a lipid bilayer is not just oil You remember that you remember what the lipids look like you had these carbonyl groups You have the phosphate groups. There are lots of negative charges in each lipid So what likely happens is that in particular the upper of these arginines, which is also the one facing most outward It's not really located in the hydrophobic center of the membrane But it's located so far up that it will rather form salt bridges That is an electrostatic interaction between a positively charged iron and a negatively charged iron in the lipid And rod actually showed that's the case because if you put this protein in a different type of lipid environment One that lacks negative charges in the head groups. It doesn't work So that's pretty cool The other thing that you can think of here if you look in the right hand side of this You see that there is almost a spiral stair shape of these arginines If you're now going to take the helix and Somehow change it This is going to need to both move up and it's going to need to rotate to maintain the same interactions between the different arginines That's an extremely complicated conformational transition Anything can happen in practice, but if I would have to guess that would likely take a minute or so Can you think of something else that you could do with this helix you remember that very early on in the course I brought up these different types of helices and then I said that there might be some special locations when you need One type of helix instead of the other Well in hindsight, I could say that was not purely for academic reasons The reason why I brought up in particular those three ten helices is that just now and then they do occur in nature And this is one of those cases So if you remember the three ten helix compared to an alpha helix and alpha helix had three point six residues per turn And that is if you have these residues and they have three neighbors between them all the time They're not they're going to end up being slightly Shifted for each turn and that's why you get the spirals there But if three ten helix had exactly three residues per turn So if you now take this helix and then wind it slightly tighter You're suddenly going to have all these charged residues line up exactly over each other That would likely lead to a helix that can move straight up or straight down much quicker We have seen that in simulations. There are some other groups that have seen that in simulations, too But a problem is of course here. We're looking at an experimental structure where we don't see it Well, actually you do see it in the very lowest part of the helix, but not in the upper part And I think this is a bit of an open question. It could certainly be that This has been overexpressed in a crystal relaxed for weeks It might very well be that the helix relaxes or What we have come to think I believe more is that You might actually have a small part of sliding three ten helix so that the very short part In the center when all these residues need to jump from one partner to another if that part is three ten helical That's actually going to be sufficient and then you can let the helix relax and be alpha helical Above and below that because it's slightly better in terms of free energy for helix to be alpha helical than three ten But the beautiful part is how much we can learn from the combination of X-ray structures functional studies in particularly mutating these residues and models and simulations It actually turns out that since this Voltage sensor has two states It's going to be in the down state relax state when you have a potential across the membrane and it's going to be in the open state when you have well when you remove the potential when you've depolarized it So what state do you think it's going to have in an x-ray structure? such as the one you're seeing well, you can't really apply a membrane potential in an x-ray structure partly because well part of our technical reasons but more important than that You typically don't have a membrane around your protein anymore, but you have some sort of detergent And of course if you now grow this for this would relax in a much faster than a millisecond But if you now grow this for weeks in a crystal, it's going to be in an extremely depolarized conformation and fully open and That's a bit of a bummer because we could probably learn a lot from the transitions that we also knew the down state The beautiful thing is that there have been a number of groups including us that have been able to simulate this And we actually know what the down state is here, but that's primarily based on simulation data Had this been one simulation you could doubt it too. Well, that's a nice confirmation But it's still a bit uncertain But by the time you have five or ten independent simulations and they all end up with the states that are roughly identical This is now so good that I would argue even even experimentalists have come to accept the computational models of these down states And I think that's a gigantic shift in how we see structural biology. So computational studies are no longer a Post experiment analysis technique, but it's really something that we can use for At least determining specific states in addition to trying to crystallize it I'm going to show you one of the coolest simulations This remember the simulations I showed you earlier this week. They had timescales in the ballpark of 50 or 100 nanoseconds Have a look at the counter where it says 0.0 mu s those that is a micro second counter So a thousand times longer timescales than in the nanoseconds What you're seeing here is the protein and these blue parts is really going to be the part where we Have the gate and where we close the channel right now It's open the green parts are the voltage sensors and the red helix there That's this s4 segment that's going to be moving. So let's go That's 10 microseconds already 20 30 40 You're gradually going to see that the select that the pore the gate in the central actually closes Do you see there that it's virtually impossible for nine to get through this appears to be a very asymmetric state But that's the way physics work. We don't have to force a symmetry. You can still be symmetric on average We're now at 200 microseconds is an insanely long simulation and these are now timescales Where we could actually do electrophysiology experiments and everything and see very easily And you will probably also see well now we looped But before we looped you probably also saw that the voltage sensors are moving quite a lot So this protein is not quite that rigid Let's have a look from the side instead One two three we go So I think here we're going to get to 215 microseconds or so and then it's going to slow down and when it slows down You're going to see this helix here. Do you see that it moves down the blue arginis there? They move one step down at the time and I think we're going to get one more step there. Yes And then I think we're going to speed up again in a second So now we have this entire red helix that moved vertically down And I think now we're going in the other direction here instead These simulations are not possible to do in a normal computer But what you see up on the left is a special purpose built machine called Anton designed by David Shaw Who used to be a professor of computer science, but has spent a lot of years working in the financial industry and stock market And the reason why it's possible to do these things with a dedicated Computer it's kind of the same reason that if you try to save an image If you try to save a jpeg image in Photoshop on a computer That might take 0.1 seconds So you could save 10 second 10 images per second But on the other hand you could also save images on your iphone and your iphone can probably save like 100 still images per second So that's kind of strange. How does it come that a phone that just draws one or two watts can be more powerful than a computer? Well, the reason for that is that a computer in general is programmable While an iphone has a special part of its chip Designed specifically to just do image compression and that's why all these digital cameras are so good at it Now that works. Well, if you had chips that you're going to sell a billion copies of there are lots of people who want a camera which they pick compression But these computers when you're only might sell 10 or 100 of something that becomes Exceptely expensive because this is literally building a new microprocessor that cannot run a program But it can only run a molecular dynamic simulation But again, this is another one of these gigantic shifts in it because suddenly it's possible to see in Almost on-the-fly processes as they're happening on time scales that are at least a thousand if not 10,000 times longer than what we've been used to before however As beautiful as that is there is a problem with that can't can you remember what that problem is Based on what I said about simulations earlier in the week What is a simulation? Well, the one thing I told you that's a simulation is not right. It's a Simulation exactly of our particle or collection of particle moves And the reason for that is that we don't know the initial conditions exactly and the second We don't know the initial conditions exactly. They're going to deviate exponentially So a simulation is just a matter of sampling from a Boltzmann distribution So even in this particular case David Shaw and his team they were ran a bunch of different simulations I think was five or so of them to show that this was statistically reproducible and they got it several times The challenge here is that many This type of simulation suddenly makes experimentalist love simulations There are so many things you can simulate But I think it's very important that you and others need to be the same voice there and say well as beautiful as that simulation is What we need to think about is how accurate the statistical sampling is and can we really use this to predict Which state is the more stable one rather than believing that this is a microscope? It is conceptually a microscope in the sense that it can help you understand things But it is not a microscope in the way that it's showing you exactly how atoms move So again caveat emptor And you are the person who is responsible for being the same voice in these cases All right, that was the voltage gated channel part As I told you voltage gated channels are important in nerve signaling And that's what you see up here from the top left a nerve pulse is coming in a nerve cell But nerve cells are not infinitely long and at some point one nerve cell has to convey the signal to another nerve cell Or well, you might need to change the signal or you might need two nerve cells combine a signal There are lots of reasons why your body needs to have nerve cells talking to each other The way the body does this is that when you get the signal at the end of the first cell You release small bubbles with Chemical molecules neurotransmitters, but these neurotransmitters are actually frequently just small amino acids single amino acid molecules like glycine or glutamate And then they diffuse over this very thin synactive cleft here ballpark of 0.1 millimeter or so And then the green molecule the neurotransmitters will bind to the pink molecules here, which are ion channels But these are not ion channels that are controlled by voltage But ion channels that are controlled by the binding of a small ligand Or that is ligand gated ion channels There is a huge variety of these channels and it's really these channels when they open and now they're going to let through a specific ion Say a chloride ion. This is now going to need to a reduced potential on the inside of the next cell And in that case that's going to lead to a new electrical nerve signal in that cell And then we have transmitted it across the synapse or the synaptic cleft that you see there in the middle These channels are mighty cool And i'm biased here too because we've worked on these channels too for a long time They the neurotransmitter can be a large number of small things like gamma Aminibuturic acid glycine acetylcholine serine So they can bind not quite anything but a very large number of different drugs And then they can conduct either anions or cations. So depending on exactly what channel this is and The difference between these channels can be as little as two or three residues They can either hyperpolarize or depolarize the membrane, which means they can either Inhibit or create a nerve impulse Complicated we'll get back to that There is something else that it says here anesthetics and ethanol I'm going to jump to another slide and then we'll talk more about that This is a real photo taken at massachusetts general hospital right next to harvard university In a room called the morton auditorium. You can actually still visit. It's a pretty cool place And this was october 16 18 46 This patient is Edward Abbott and I think he's having a tumor on his neck removed or something And this was the second example In the world where people were able to sedate a patient perform surgery on them and the patient actually lived Now we might have used other types of Sedative or anesthetics such as ethanol or something before but the key thing here This was really controlled. He was in no pain and he survived The first patient actually survived too. That was a few days earlier that somebody had a tooth extraction But for whatever reason I haven't found any photos of that How does anesthetics work? Well The scary thing is that anesthesia has a history of 150 more than 150 years and it's mostly trial and error Very good trial mind you but still trial and error So there is a famous result in anesthetics called the mire overtone hypothesis or the mire overtone correlation for anesthetics So what you see in this plot is that On the x-axis you have the olive oil to gas partition coefficient at body temperature So on the left you're not very soluble in oil and on the right you much prefer oil to water And on the y-axis the lower down you are on the y-axis the better this molecule acts as an anesthetic Mac or minimum alveolar concentration to get an anesthetic effect If you look at this it's kind of amazing how perfect this correlation, right? So that it appears anesthetic effect is purely based on hydrophobicity of the molecule So in that case, what would you imagine that an anesthetic how an anesthetic acts in your body? Well, I would guess that you're going to predict that it goes into the membrane and you are an exceptionally good company there, right? Jesus if it's this if it's entirely correlated with how hydrophobic it is This must be that the molecules dissolves in the membrane And then they must they must do something strange to the membrane Things molecules like xenon there in the middle is this xenon is never going to bind to anything So it must somehow be altering the membrane properties now That's a very good idea But today we have a huge number of different inhalation anesthetics If you go out and have surgery today here at carolinska They're likely going to administer a silver fluorine or a desk fluorine that you have there on the right They're not going to give you ether as they did in 1846 But the funny thing is that these channels that we have sorry This is a picture from the top of the channel. I'll get back to that in a second We have actually found specific Residues in the sequence of the ligand gated iron channels And if you mutate away that residue and change this is something else You can actually show that rats are suddenly Not susceptible you can't sedate them to some of the anesthetics. I showed you on the previous slide And that's strange. That's quite a paradox, right? So on the one hand We know that anesthetics are purely hydrophobic. This should be a pure hydrophobicity effect that they go into the membrane but on the other There might be if there is a specific residue there almost has to be a specific binding site in these channels for anesthetics And people have thought about this for decades well two decades at least including me What I show in this slide is a fairly remarkable thing that in the early 2000s Nigel Unwin finally Was able to determine a structure of one of the channels the nicotinic acetylcholine receptor Which is super important in humans too This was based on cryo electron microscopy Again 15 years ago the the center the detectors were not as good as they are today But when you have a couple of thousand engineers like this one We were able to build a model not just of the transmembrane domain But also the entire extracellular domain And I so remember then this came because both we and a number of other groups We were ecstatic that this really confirmed our predictions that it should be a pentamer And inside each monomer we should have four helices in the transmembrane domain And which should predominantly have beta sheets in the extracellular domain. This was beautiful Now The caveat is that what you're going to see in the PDB is of course not an image like that But rather something like this well the hydrogenous is added in the molecular model here And as good and beautiful as that experiment is you probably realize too that we don't quite have the detail here that But there's been quite a bit of modeling involved And in particular one of the hardest part is the modeling is to decide exactly how do we align the sequence That is each amino acid to the structure the rough shape that you see in the blue cryo unit here If you start making just one minor error So if you place amino acids one unit off in a helix, that's going to be an RMSD of roughly three angstrom So make a very large error the second you place one residue wrong What's worse is that some of these Residues serine 267 in particular that we had found that were extremely important for anesthetic activity Well depending exactly how you make that alignment, they're going to be located either inside each subunit So just the red one here or between say the red and the orange subunit So we have no idea where those binding sites were Oh, sorry. I'm going to spend a little bit more time on that I might talk more about that later in the course, but I think it's an example that Simulations has helped quite a bit here too What we have done since then that I don't have any slides on Is that we've been able to use simulations to prove the existence of both these binding sites And then you can go back and take the models that you have in the simulations predict what residues you should knock out to get a specific effect And then you can test that and you can test that with a technique called electrophysiology That I will get back to a little bit later on in the course, but we kind of we kind of like talking about our own research So that's worth its own lecture There is one more class of proteins that I'm going to bring up if of course and that's called tyrosine kinase receptors or RTK receptor of tyrosine kinases Most of the membrane proteins I showed you here They were large ones in the sense that they had a very large transmembrane domain But there are also examples of receptors where you have a fairly big blob outside the cell You just have a single helix that anchors it in the membrane and then you have some other domain on the inside And that's the case for this type of receptors What then happens is that In a normally working cell you would get a ligand a chemical molecule or something that binds to these extra cellular domains When this molecule binds this induces dimerization because it brings the two parts together And then these two green helices are going to stable each other exactly in these forms that I showed you before the break right? And when these are then held together Some sort of signal will be released on the inside that for instance leads to cell growth differentiation or something Now in a normally working cell The ligand eventually leaves and when the ligand leaves Since they were just borderline stable together these helices will eventually diffuse away from each other again And when they diffuse away from each other again, they no longer signal, which is good because otherwise bad things would happen Now Unfortunately in some cases there might be mutations in the helical sequence sequences So remember that I said that it's possible to insert other arrested use That might affect how much two helices attract each other And that happens in some cases for these so the blue helices here They have a mutation that causes them to bind much stronger together The sad thing is that they bind so much stronger together that even when this ligand and the extracellular domain leaves They stay bound together and when they stay bound together, they keep signaling on the inside for the cell to divide and grow And the cell is going to go completely berserk and that's what leads to tumor growth but In principle, we're not talking about a gigantic protein now. We're talking about a single helix interacting with another single helix. So what if You could Design a locking key. So the red helix here would be some super smart designed helix that The red helix would bind even stronger to a blue helix Then a blue helix would bind to another blue helix So each blue helix would much rather be paired with a red helix than pairing up blue blue Now if we could do this, this would likely break all those bad blue blue interactions And in that case we would stop the cell growth So there have been a number of groups attacking this problem in particular build the grotto almost a decade ago together with Joanna Slusky who worked the DBB for a couple of years actually So what they did is that they realized this this problem is so simple. This is just a matter of two helices crossing Remember that I said that in principle there are only two potential angles helices will cross each other with roughly plus 20 or minus 50 degrees Based on the packing of the ridges and valleys in the helix Well, just two confirmations might be a little little But let's bring up every single helical crossing we have in the literature or at least in the protein data bank And then we can look what sequences you typically have in those crossings So if we now in the A part of this figure here if we have a particular helix that we would like to stabilize the one in the front What residues should we build in the helix in the red helix in the back? So now it's we don't have that much freedom. You only have 20 amino acids to place We have to make sure that the helix stay a helix. You can't pick any amino acids you want So this is just going to be a matter of finding amino acids that stabilize these two helices at a pair Now that is an expensive computational problem, but it's a tractable computational problem compared to folding an entire protein So what Joanna and bill and co-workers did is that really they created they wrote computer programs to do this And the nickname for this was champs computed helical anti membrane proteins So the idea is that you just select backbone geometry you want you find the helices in proteins to use as a template And then you design these complementarity so that these two helices One of the helices we can't change because that's going to be the bad blue helix We have in nature But we can swap almost every single residue in the other helix to make sure that this is a helix that binds to the first helix And this actually works So the you could of course are you well, how are you sure that this method is going to produce good results and everything? Well, the way you show this is that you can show an experiment that these helices actually have affinity to each other that they do bind each other This is a few years ago The reason why you haven't seen a lot of results like this on the market is not that They are rare I would argue that there are hundreds if not thousands of studies that are now able to very specifically design completely new folds Proteins targeting well small artificial proteins targeting other proteins But the way pharmaceutical Production works is that from the time you discover something until you start going through clinical tests and eventually put it on the market Is frequently going to be at least 10 years and frequently 15 years So I think if you wait another decade or so, I expect we're going to see much much more of this type of designer drugs And that is related to what I told you last lecture Do you remember that I talked about biologicals? This is a biological too. It's a simple biological It's just a single helix rather than a complete antibody, but it follows the same principles We design something with amino acids to get a specific shape get specific interactions And hopefully these interactions are going to be so specific that it works better than the traditional small drug Now if this works is going to be much more potent because they're it's more specific It is also the specificity also means that there are much fewer side effects At least when things work, I'll get back to that The disadvantage is that it's a protein so if you were to eat a pill with this it would be degraded in your stomach by the enzymes So you would need to re inject keep injecting this which is bad both because of quality in life and because companies are Well say what you want about it But companies are in business to make money and the companies realize the patient has to Inject something they are likely going to prefer some other drug That they don't have to inject even if the quality might not be quite as high It can also be very expensive to synthesize these drugs Because you well it's no longer a matter of simple chemistry production, but you know how to overexpress and purify these proteins for instance from bacteria The other reason that can make this expensive to synthesize is that one way to get the way from these injection criterion Is to use for instance artificial amino acids or something that makes sure that these proteins are not degraded But just as we don't have artificial amino acids bacteria normally don't have them either So then it becomes an even more complicated production process And then well whether this is an advanced knot they clear fast from the body Which means that you keep needing to take new injections of them Normally this works really well, but as I told you earlier on in the course during a break There are some examples where really bad things have happened And one of this was a few years ago when the german company called to genero They were developing a an antibody reading which called tgn 40 and 12 was their internal name This worked by activating a special receptor called cd 28 I'm not going to go into details of t-cells. You've taken a course on that before But the whole idea is we can we can use this to activate and then try to kick start the normal immune defense Rather than doing all ourselves We basically kick start the body's own defense and then we have the body do the actual workforce Really neat idea in theory And it worked beautifully and all the computational tests It worked beautifully in all the simple chemical tests And then when you got to phase one when you start studying this Testing this on patients. There was a disaster complete disaster So I think there were seven young people that said they had to amputate fingers their heads swell up Swell up and I even think that one of them developed cancer. So the question is what happened Well, most of this is not public yet So I've only this is partly gays based on educated guesses from me and others what might have happened So take this with a fairly large grain of salt like half a kilo salt But what might have happened is that this is really a super agonistic antibody So we want to use this to kick start your immune defense in a way that's stronger than what the immune defense would do by itself Now obviously the reason why we're designing this is that we want to administer this in humans eventually Mice don't really have a whole lot of money to pay for drugs. So that is much better to target humans and That of course to use that in testing and everything we need the human and mouse to share most of the protein structure function and sequence And they kind of do that. It's 93 percent. It's not awesome, but it's probably okay to start designing things The only problem is that In the extracellular part of the cd28 protein, there's only two-thirds identity And then you're starting to get into the territory where it's really dangerous So there might be some substantial difference not necessarily in structure, but in specific function and interactions So what might happen there is that if you now go through all these tests And we do the animal tests your rats and everything and then you measure what dose you need in a rat Well What if you now get the rose and realize you want dose x here and then you take dose x and give that to a human The problem is that that molecule that kind of didn't work that well in a rat that we needed quite a lot of it Suddenly when we administer this in a human it fits like it's not that it just fits. It fits perfectly It starts kicking in on every single cylinder and it's not just super but hyper activates your entire immune defense Your immune defense and your entire body will go crazy Mind you that's speculation So why are people that stupid? Wouldn't they have understood this? Well, it turns out we did There are lots of procedures like when it comes to clinical tests This is not something like a professor decides in a Tuesday afternoon that hmm. It would be fun to get some drugs tested tomorrow I'll see if I have any volunteer students But you Well, you license that you hire a company and they start following all the specific procedures in the country where they're going to Do the tests There are lots of rules and regulations that based on the activation you saw in the mouse model And the dose you had there what is the dose you're going to start with in a human? And typically that first dose is going to be like one tenth or one hundredth or something that you had in the first animal model So when you first start this the doses are going to be so small that we don't expect any activation at all The only problem is that I think that most of those protocols were historically developed based on traditional small drugs Uh, remember these small hydrophobic molecules such as aspirin or something And this is a completely different type of drug The activation and the barrier stability Well, the specificity is much higher here for once. The barriers is probably going to be completely different. The kinetics is completely different So these traditional safety margins with a factor of 10 or something simply don't hold here In principle, I think that you could have done this here too But you should probably have started with a dose that was a million or a billion times lower than you would normally use in the animal model so It's one of these cases where hindsight is 2020 when this has happened I think it's very easy to realize that we're going to need completely different tests for biologicals Not just that what you also realize that since with biologicals, you're starting to play with the body's natural immune defense There are some pretty amazing things we can do But the same things can also be very scary and that's simply not something we used to see that extreme in drug development But that's going to be your job in part like when you go out and work at tessero zenica or pfeitzer or merker any of these companies You should be aware not just of what we've told you because again hindsight is 2020 But there will likely be other classes of problems like this when we use biologicals that we have no idea how to handle in the future And sadly because I don't know what they are. We won't be able to help you with them But remember that past performance is not an indication of future ones I think that's all I had for you today both before the break and then after the break when I re-recorded things here As always, there are a bunch of study questions here that I'm not going to go through today But on monday morning, I suggest we have our normal chat session and work through these 16 ones And then I'm going to talk more about kinetics of protein folding