 As a chair, she was one of the first people to notice or to have or to raise concerns about the general need for more exercise and better nutrition to improve life, to improve health, right? That is something that was there for decades and we are still talking about it. So here is a person who was here, talked about this so many many years ago and we are still recognizing it. Unfortunately upon her death in 1969, she bequeathed her estate to the Department of Biochemistry, which has been used to support the lectureship in this department for many many years. And as you will tell you, this lectureship features the former associates of the department, including former graduate students, prospective fellows and faculty. So for now today it is a really great pleasure for me to introduce a former faculty member of this department who is going to deliver this lecture, Professor Anert Melon. So Professor Melon got his PhD from Cornell and then went on to do his postdoctoral training with George Cross at Rockefeller University in New York. And actually that's where I had about Anert for the first time because I was in a laboratory at Hopkins that was working on the GPI anchored protein biosynthesis in the ER. And it was a very tough competition at that time, but Anert handled it very well. So he went on to be a faculty member at Rockefeller and then in 1993 he joined the department of biochemistry as assistant professor. So he went through all the ranks and then until 2005 and he continued during that time he was here to work on the GPI anchor, anchored proteins biosynthesis, GPI by the way is Glucosil Phosphotidinoceto anchor proteins, very important and he made major contributions in that area. When he got here he started new projects to look at the transfer of phospholipids in membranes and between membranes. You may talk about the phospholipid scrambleses and intracellular transport of sterols. Now for reasons that I'm not going to tell you, Anert decided in 2005 to return to New York. So he has been to New York as a professor of biochemistry at the Cornell Medical School for up to now as a model fact. But then we thought he would be the appropriate person to come and deliver this lecture in biochemistry. So I don't want to take too much of your time, you are really welcome back to Madison and we really missed you. So thank you very much James, I'm a little confused to be back actually, a little daze to be back because I sort of walk around a few streets, a few alleyways and everything is different. There used to be passages going from A to B but there's now a glass steel box that interrupts that passage and there are many of these that have been built and it's not been that long it seems to me but it's obviously lots of progress, at least in terms of putting these impediments into small passageways that one but so much more lab space but it's also good to see so many friends. So lots of people here that I obviously know for and have known for years and haven't seen for years so it's a lot of fun to be here. I thought I would add one slide to James' introduction about Gladys Eversen. I was looking trying to find out, to see what I could find out about her and I discovered this obituary that was published 10 years after her death in the Journal of Nutrition and it says something quite interesting which appears to my sense of humor and I think the students who had lunch with me were probably disturbed at that sense of humor because they probably got a little more loose talk than they were used to at that kind of format. So here is a person who in quest of information would occasionally relieve the library of a long-awaited new book without the formality of signing it out and when she would elude the campus police in her rush to the laboratory to rescue samples left in the drying oven she was merely achieving her goal of advancing science and the comment here is that the books were always returned and the campus police always had another chance. So this is someone I would really like to have as a colleague. And then she did, there's this other little bit of description here which appeals to the iPhone age I suppose which is fond of describing unproductive activity as a poor use of one's time and then this business of leaving notes endlessly, apparently she left notes for a whole range of people her night-hour working habits which cause great concern to the janitorial and security staff and nocturnal communicates left for the students and faculty apparently more and more illegible as she grew older but still kept on leaving these notes disturbing the daylights out of everyone I think. So this is someone, this is a colleague that the department and one should aspire to perhaps. Okay, so I want to start with a very broad description of the problem that I'm going to talk about some of it will be on elementary terms just to make sure that everyone knows what it is I'm saying. I work here, this is spread of institutions here going from Rockefeller University to Sloan Kettering Cancer Center to Cornell University Medical College so I was in buildings over here and in between I was in Madison. Okay, so just to start with self, this is a book that I particularly like and if students have any sense of any appeal for quantitative things for length scales, time scales just to be able to do back of the envelope calculations there are phenomenal amount of information in this book, Cell Biology by the Numbers so they are very precise on a useful scale. So here is a mammalian cell grown stuck on plastic so it's spread out 50 microns or so in diameter in dimension, a yeast cell with 5 microns across. What you see in both these cartoons is that the cells are extremely compartmentalized this is something that we all know. Very many compartments you have the green of the endoplasmic reticulum the Golgi apparatus, various ways in which the cells have boxes within them that do particular functions. So I focus on yeast because a lot of my lab works in the yeast system although we do other things as well so here is the yeast cell now blown up and you see here the cell wall in yellow and the endoplasmic reticulum in green but in reality there's not just compartmentation but there's a whole lot of architecture that is important in the cell so if you look at this model of a yeast cell that was done in my lab using this focus iron beam scanning electron micrograph you see a lot of the cell has not been depicted here but what's important is that the plasma membrane is coated with endoplasmic reticulum on about 50% of its surface and this is an ER network that is attached to the plasma membrane through tethering molecules so there are protein molecules anchored to the ER that reach out and contact anionic lipids on the plasma membrane stitching the ER along the length of the plasma membrane and you can imagine what impact this might have on various communications this is the sort of thing that is important in store operated calcium entry where you have the stim 1 ER sensor that talks to a calcium channel in the plasma membrane and it's also been implicated in lipid traffic so we'll see something about that in a few moments but the endoplasmic reticulum for all practical purposes is where lipids are made so it's a biogenic membrane, it's able to make its own components integrate those components into its own fabric and also transport those components to other parts of the cell that's an oversimplification, lipids are made elsewhere as well but the bulk of anything one talks about in terms of lipid synthesis occurs in the endoplasmic reticulum and because of that you need some transport so you need to take what you make in the ER to the plasma membrane for example the bounding membrane of the cell that has a particular lipid composition, thickness, fluidity and so on that defines its barrier properties as opposed to the ER which is biogenic and a little more promiscuous in terms of what it'll allow for permeation and also here you can imagine an organelle such as the mitochondrian where you have to take lipids that are made in the ER to populate mitochondrial membranes so one example and I will highlight this briefly is the problem of how sterols move around inside cells so sterols are made in the endoplasmic reticulum the bulk of the biosynthetic apparatus famously HMG CoA reductis the enzyme that is targeted by statins is located in the ER and the 20 odd steps of cholesterol biosynthesis occur in the ER membrane similar state of affairs in yeast where agostrol is the predominant sterol instead of cholesterol but even though sterols are made here they're only about 5% of all lipids so 1 in 20 lipid molecules in the ER membrane is a sterol molecule whereas in the plasma membrane sterols are highly concentrated so 1 out of every 2 or 3 molecules of lipid is a sterol in the plasma membrane so you need to traffic the biosynthetic bolus of sterol here to the plasma membrane and other compartments in the cell you also need to traffic sterols to mitochondria where steroid hormones are made and this is an image of what a yeast cell looks like you can take a molecule called dehydroagostrol which is only one double bond different from agostrol it's a conjugated double bond system makes the molecule fluorescent it's a nasty fluorescent but you can it's UV excited but still you can image these and use it to study various properties at the level of fluorescence microscopy ok so what do we know about how this transport occurs so one is transport to organelles is going to occur by some pathway that doesn't involve secretive vesicles there are no vesicles that are known to take anything from the endoplasmic reticulum to mitochondria so some non vesicular pathway but that's also true of transport of molecules to the plasma membrane you can shut off vesicle, mediated, budding, fusion etc all of those pathways can be shut off pharmacologically in a mammalian cell through the use of a sec mutant for example in yeast and sterile transport occurs just fine so in general and sterile is the example I've used here but in general lipids can go around by non vesicular means which doesn't mean that vesicles don't carry lipids vesicles of course made of membranes they're trafficking lipids of course but in general there is a bulk pathway with more than enough capacity to take lipids around from their site of synthesis to the plasma membrane and to organelles and the usual way to think of that is that you have a molecule that looks like albumin sort of some protein in the cytoplasm with a hydrophobic binding pocket a little more sophisticated in this case because the binding pocket is literally a pocket but then there's a flap that would close it off so some molecule like this flying around in the cytoplasm in a collision encounter picking up lipid from one membrane and in a second encounter dropping it off somewhere else so that's how you can imagine how that works so some molecules that usually allow for bidirectional transport because they are simply catalysts they are like cyclodextrin in the cytoplasm just bouncing around between membranes and allowing for a homostatic regulation of lipid composition in their different compartments and sometimes they act at these contact sites where the ER is in close proximity to the plasma membrane as I showed you in the model in an earlier slide so we know lots about these things about these types of molecules there are very many of them that have been identified years ago through assays in vitro assays where lipid molecules were seen to go from microsomes to mitochondria and this is the work of Don Zilversmith at Cornell in the 1970s and many molecules now have been identified that would play such roles in the test tube there's no shortage of them Tom Martin has some of these here that are involved in phosphoenositide metabolism so they all work in moving lipids around between vesicles in the test tube what they do in cells is less clear because there are molecules in hand you can figure out many things you can crystallize them you can do molecular dynamic simulations you can study what their requirements are for transfers you can get an idea of how the machinery works but their physiological importance is not entirely clear so this is a crystal structure of one of these sterol carriers you can see the cholesterol molecule here with its hydroxylen buried here and then this little flap over here is this the omega loop so that's a structure we did along with several others not so long ago okay, so that sets the system for making things in the ER and shipping molecules from the ER but what about the ER membrane itself and this is basically the subject of my talk today so the problem is set up as follows all phospholipids are made on the cytoplasmic side of the ER membrane this pale blue green slab is the membrane bilayer so if you want to make a phosphotydylcholine molecule and use the Kennedy pathway to make it this is the reaction you start with diacylglysrol and take CDP choline which is a cytoplasmically generated molecule doesn't cross the ER membrane phosphocholine is transferred and you make your phospholipid here and release CMP so that reaction occurs on the cytoplasmic phase of the ER and that's true for every major phospholipid phosphotydylinositol ethnolamine serine and all these lipids are made on the cytoplasmic side and really the problem is how do you get across because some of what you make has been transferred to the other side to grow the bilayer uniformly otherwise because of the bilayer couple hypothesis which is more than a hypothesis at this stage you start getting membrane curvature because one leaflet is expanding at the expense of the other leaflet you need that kind of expansion in particular cases for example you want to sculpt a vesicle you might need something like that but in general you want to have the ER uniformly except for places of where it's curved, tightly curved where you have tubule formation and so on so you need some method to transfer the lipid across the membrane and here I've already given you an indication that that method involves a protein so not only do you need the lipid movement across membranes for just membrane growth you need it also for a couple of other many many places in sales where you need this but I highlight a couple of them one is this problem of lipid asymmetry at the plasma membrane I'll show you that in a slide the plasma membrane is very strongly asymmetric and you need under particular physiological prompts to dissipate that asymmetry and the other is for glycoprotein synthesis which is a preoccupation of mine, James mentioned my interest in GPI anchor biosynthesis but the same problem applies to end glycosylation and other forms of glycosylation in the endoplasmic reticulum so the example of the plasma membrane I've simplified the plasma membrane to one phospholipid which is phosphatidylserine which is localized to the cytoplasmic phase of the plasma membrane exclusively there's no PS displayed on the outside and yet when needed you need to display that PS and that happens under particular conditions blood coagulation will not happen without the display of PS on the surface of activated platelets so if there is a defect in the process that exposes PS on blood platelets this order called Scott syndrome and there are several other versions of that PS is displayed in apoptotic cells and this is part of the program for cell death and it then becomes a marker for picking up those cells macrophages use that as one of the recognition markers for phagocytic clearance of the cells and similarly you need every morning when you wake up and if you haven't been looking at your iPhone all night and disturbed your entire circadian rhythm PS is exposed on the tips of your photoreceptor cells in the eyes or the rod outer segments which are bland in terms of their PS exposure they look exactly like a plasma membrane with no PS on the outside suddenly turn on PS in the top 10% and PS becomes exposed to the outside recognized by the retinal pigment epithelium and the top 10% is eaten and new discs photoreceptor discs come up from the bottom so every morning this ferocious phagocyte in the retina takes off the top 10% in response to a PS signal again requires exposure of PS specifically as a signal on the outside okay so you need lipid flipping to occur in some sort of physiological time frame for these things to happen and then this impossible problem here which I have to highlight I tend not to talk about this but I really have to talk about it because it's a preoccupation for years I lose sleep over this so this is a textbook picture and this is a part that many will be familiar with message of protein comes through the translocon and if that protein exposes a glycosylation sequon which is a triplet of amino acid starting with an asparagine residue that will eventually get modified by the oligosaccharide followed by any amino acid and a hydroxy amino acid that little triplet of amino acid is recognized by a multi subunit enzyme called oligosaccharide trans rays and this collection, this oligosaccharide is slapped onto the asparagine residue that oligosaccharide comes from a lipid so you first build it all 14 sugars on diphosphate isoprenoid lipid and that little snake here is 100 carbons long in mamadian cells made of 20 repeating 5 carbon units so this part is okay except that the synthesis of this lipid starts on the other side of the membrane with dolicol that is phosphorylated to make dolicyle phosphate and in the first reaction sugar phosphate is added and that's the tunicamycin sensitive reaction for those of you who use tunicamycin for let's say experiments with the unfolded protein response this is what upsets the pathway preventing proteins from being glycosylated then a second sugar is added these are both anacetyl glucosamines followed by 5 manos residues in green and then this whole thing is flipped over the over the membrane to be continued with 4 more manos residues and 3 glucose residues so this flipping has to happen in real time we cannot do n-glycosylation will not grow so this whole pathway is essential in yeast and so it has to happen on the time scale of a yeast cell doubling okay so those are the 3 examples the growth of an ER membrane the exposure of PS on the outside of cells in n-glycosylation so just a little bit of history so spontaneous flipping I've already alluded to this this doesn't happen in real time if you take a synthetic membrane it will flip over once a day roughly so this is not physiologically relevant it is a background level of flipping for which compensation needs to occur so you don't want PS exposed even at that slow rate but it's not relevant for any of the processes that I've described and the first time anyone knew about that was in 1971 in this paper by Kohnberg and McConnell this is Roger Kohnberg of transcription factor fame but this is his PhD work when he had more sense he worked on lipids instead and the other thing to know is that flipping occurs really quickly in biological membrane so if you take a bacillus membrane bacillus megaterium was used in this particular example then you get flipping that is commensurate with cell growth you have lipids exchanging between leaflets of the bilare in under a minute and that was the first time anyone measured anything like that was this paper here by Rothman and Kennedy this is Gene Kennedy who discovered lipid biosynthetic pathways and that's Jim Rothman who's gone on again to do other things should have stayed with lipids as well okay so these transporters that are needed to promote fast flipping come in two flavors one is the ATP dependent flavor so there are proteins that use ATP hydrolysis and couple that hydrolysis to the movement of lipids and because they do that they're necessarily slow they're no better than about 100 lipids per protein per second because that's the rate at which ATP is hydrolyzed and those proteins are called flipasies if they move a lipid from the outside to the inside exo to endo or they're flopasies if they do the other thing but these are both ATP driven transporters that move lipids against their concentration gradient so they're uphill transporters there's the other variety which are called scramblases and these are the ones needed for end-like oscillation and for growing the endoplasmic reticulum bilare and these come again in two flavors but they don't require metabolic energy their flavors are because they are regulated so there are those in the plasma membrane that are turned on by calcium micromolar calcium and then there are those in the endoplasmic reticulum that are constitutively on so the glycosylation scramblases are constitutively on the ones that allow the ER to reset its phospholipid imbalance in response to synthesis is that transporter is constitutively on as well and these are super fast if you take them out of the membrane or at least take the activity out of the membrane by detergent solubilization and reconstitution they transfer lipids faster than 100,000 per second so this is, that's because that's the best number one can get if it's quite possibly 10 to the 6 per second which becomes the rate at which an ion channel would transport an ion so that's one frame, one thing to think about so I'm going to show you the sort of setting in which I'm going to talk about this problem of lipids scrambling and it begins with a picture drawn here by this gentleman in the front row Adam Steinberg who does this it's the same picture that's on the poster and it's one way to think about how a lipid might move across a membrane so it's basically short circuiting the bilayer so you want to connect the two leaflets of the bilayer you can view the lipid as a giant ion except it has the problem of having a pair of legs attached to it so what you do with the legs you can roll everything up into a ball and push it through some sort of cavity one of these lipid transport protein type cavities and then drop it off on the other side it's possible to do something like that or have something like this which is a concept that we've developed here actually I think I was talking about this idea with Adam and Ivan Raymond walked up and said oh it looks like a credit card and then Adam drew a cartoon that looked like a credit card being swiped through a card reader which is more or less what this picture tells you here that are walking through a protein cavity and that protein cavity is hydrophilic and there's an energy penalty to pay to expose that hydrophilicity to the membrane and so one way to compensate is to have an apfipathic adapter and so the lipid would be the adapter so it's head which is vitronic or charged in some way talks to the hydrophilic part and the legs talk to the membrane and you can imagine a path like this where the two leaflets are short circuited so this is what would happen if you have a protein the barrier that has been reduced and so in doing something like this is about 80 kilojoules per mole or 20kals per mole for those of you who think in kals so this is not going to happen at a appreciable spontaneous rate and it's downhill movement basically it's allowing the short circuit allows basically diffusion to go in whichever direction that you need in order to dissipate a concentration gradient okay so I began this working on this because of a conference I attended I was attracted to this idea I heard about the glycosylation pathways first and then the general problem of lipids crambling at the ER and then started that in New York and then came to Madison where a number of people in the lab started so the project began with Thor Regberg who helped make particular lipid substrates that we were interested in very short chain lipids he's an organic chemist from the University of Stockholm the process of making this lipid was enormously difficult for me and totally trivial for him so that's how the lipid got made initially and then the project was picked up by a couple of graduate students the first of whom was Sigrun here who joined the lab I think in 93 or 4ish as a rotation student along with this person here who worked on a different project but is now sitting in the front row in an orange set if anyone wants to spot him and I moved to New York with Yolanta here who was also in the front row and works in Madison so these were the people in the lab when I started and Bill Watkins came a year or two later we also had a couple of German visitors from Humboldt University because I had a grant with Andreas Hermann at the Humboldt and so they would come for a few weeks at a time and then leave all very productive for developing assays by which one could measure this transport process it's a non-trivial assay to set up and these were the people who set it up also set up the reconstitution activity as well which is solubilize everything in detergent and put it back together in a vesicle therefore allowing anything to move forward any identification to happen so that's all ancient history very long time ago really very long time ago deep in the last century okay so the problem is that most scram blazes are not known so when these projects were started there were no scram blazes that had been identified there was not a single name there were proposed activities that I talked to you about and of course if you don't know the scram blazes you don't know how they work and so you can imagine all kinds of things like this credit card model that I've alluded to so you know so unconstrained by fact we just a few principles you can make up nice drawings and stories like that so that's the that's the problem so the situation has changed a little bit the endoplasmic reticulum that I'm interested in still is an unsolved membrane as far as the scram blazes are concerned none of the glycosylation scram blazes were identified and I can tell you what we're doing about that but basically everything in the ER remains we identified which is quite shocking considering that genomes have been sequenced one knows the names of everything in the ER more or less not everything surprisingly there are still proteins in yeast open reading frames without assigned function but I have a suspicion that many of these proteins will end up being dual function proteins they are known for one thing and are doing this on the side or doing this on the real as the real job in doing the other thing on the side whatever it is but it's not it's there are ways around this and I can talk about that in a moment two of the plasma membrane scram blazes have been identified through heroic sort of fact screening efforts by the Nagata lab in Kyoto one is the TMEM16 family this is a family of calcium activated ion channels many of which are also lipid scram blazes so they are able to pass ions and transport lipids at the same time and this is the class of protein that is involved in scott syndrome the bleeding disorder so TMEM16F there are ten of them in mammalian cells if 16F is messed up then blood platelets don't expose PS in your bleeding disorder and then you have the XK related family XKR8 was the first of those and these are the scram blazes that are turned on by caspases so they have a caspase mediated cleavage that activates them and this is what exposes PS in apoptotic cells so unregulated ones have been identified and then quite surprisingly we have one unregulated constitutively active scram blaze which is rhodopsin and it's the prototype for other class AG protein couple receptors so the other GPCRs that we've looked at have also got scram blaze activity if you purify one of these proteins stick them in a vesicle you get scrambling rates that are equivalent to what you would get with any of these proteins as well so this comes of course from photoreceptor discs and this is what I mentioned the top 10% get eaten every morning by these neighboring pigment epithelium cells there are about 40 of these rhods that dock into one pigment epithelium which uses the top 10% of the rhods for breakfast basically so I got interested in this because of course I was interested in the endoplasmic reticulum and not getting anywhere fast when I got invited to write grant proposal I got a phone call from someone who's accent I couldn't follow and but it turned out that what he wanted was a proposal to do something about macular degeneration and they were shopping around asking people who had nothing to do with the eye to write proposals to try and bring people in from the outside who might weigh in on this problem I'd never heard of macular degeneration I didn't know anything about the eye it turns out of course that the eye has a huge amount of lipid traffic going on you don't recycle retinal unless you traffic the retinal outside the rhod discs out of the rhod cell into the neighboring cell all kinds of traffic has to happen in real time but as a result of this grant which I got I got it here actually I discovered these papers and these are papers from 93 and 2000 which basically say that the photoreceptor discs in the eye in the rhod outer segments have an endoplasmic like reticulum like phospholipid scrambling activity which is if you put in a spin labeled phospholipid into one side it equilibrate super fast with the other side and this was the work done by Wayne Hubble originally and then repeated by Klaus Peter Hoffman's group so the photoreceptor discs have scrambling activity and I thought well these discs have really only rhodopsin Hubble thought that it might not be rhodopsin they have hardly any other proteins it's a very small proteome although every time you look around there are a few more proteins stacked onto it so this would be a good place to start you take, you find cow eyes there are plenty of them you prepare the retina, you get the discs you do what we know how to do with the ER remove rhodopsin and see what's left and so on and progress that way so the end result of all of that was that it was rhodopsin so we took it away and lost the activity we purified it, it came right back we forgot about the cow eyes and the retina and made the rhodopsin in a cho cell a cost cell, a hex cell so there was no mysterious disc component that was coming with it a purified protein made somewhere else and that was also true of the adrenergic receptor and the adenosine A2A receptor other class A GPCRs these were all scrambling lipids when brought into the context of a liposome and so I sat on that result for quite a long time because it's a famous protein and one doesn't want to throw mud at it in terms of assigning a new activity it's what it is and so we published that quite a bit after the initial discovery and then this was followed up very quickly by trying to see if the process had anything to do with the light sensing function the signaling function of the protein and so you can make flavors of the protein that first of all don't have retinal these are views of the seven helical bundle from the cytoplasmic side so there are two opsins which don't have retinal and two rhodopsins that have either 11 cis retinal here or the old trans and particular mutations and you can play with the protein in different ways to get representatives of each of these pools and all of these as far as we could tell was scrambling lipids so this is the work of Mike Gorin who was a PhD student in this department with Brian Fox and also did some of his work with Jay Bangs in microbiology and then came to me as a post log so he's now at Regeneron having set up this whole system to study lipid scrambling okay so it is opsin rhodopsin so I'm going to say opsin because we don't care about the retinal anymore and it's a headache to work in the dark room with this protein and you can do just as well with the retinal free version of it so there are several immediate questions that come into play but first let me tell you just quickly that we you phone up people who do crystallography and this was before crystallography of these of GPCRs quite so prominent as it is now so you phone up people who have done the crystallography who have samples of crystallography grade proteins they send you the proteins, you reconstitute them and you get activities so some are in the presence of antagonists some inverse agonists and so on so all of these and several other class A rhodopsin family GPCRs that we've assayed always scrambled lipids this fast so what does that all mean I mean no one, I mean this is a weird thing for G protein couple receptors to do in the plasma membrane you don't want PS exposed on the plasma membrane as I've told you because they are perfectly good reason that you don't want a macro page to come and swallow up the cell so what's the physiological relevance how is it regulated if at all and then how does it work so those are questions to consider so here we are back to the photoreceptor cell this is a nice electron micrograph from a colleague showing you the disks lined up here and I'm going to highlight next in the next slide and what is surprising is that all the three classes of transporter that I showed you talked to you about earlier on which is the flip pages, flop pages and scramble pages, all of these three classes which are typically present in different membranes in say an epithelio cell are all found within the disk membrane so you have rhodopsin or opsin that is scrambling lipids so moving lipids bi-directionally you have a p-type ATPase which is a flip page that moves lipids to the cytoplasmic side and you have an ABC transporter which is doing an unlikely thing because most ABC transporters transport the other way but this one is acting as a flip page as well and the interesting thing is that this one is specific for aminophospholipids favoring PS and this one moves retinal phosphatidylethnolamine and this is a connection with macular degeneration so I can explain that afterwards at the end of the talk if someone has a question but this is an interesting lipid that if it's not transported gives rise to a fluorescent lipofucin aggregate in the eye which is the starting point of the age-related macular degeneration so the end result of having these things go on is that lipids are being moved from the luminal phase of the disk to the cytoplasmic side so minus one inside plus one outside you keep having these ATP driven things moving a lipid from the inside to the outside and that can't go on for ever a disk last 10 days and this can't keep going on so what this protein is doing and that's a speculation is that it's resetting the lipid number on the two sides of the bilayer so you keep allowing the lipids to sample both sides of the bilayer while moving particular lipids to the cytoplasmic side so this allows the bilayer to stay normal so the imbalance is corrected by this which basically acts as a release valve for all the pumping done by the other proteins it sounds like a short circuit but these are particular lipids that need to be moved out whereas this one just resets the total number it doesn't care which lipid so long as the packing is arranged so that's a possible function for rhodopsin but not yet for the other GPCRs which we can come to later the other is regulation so I alluded to this already which is every mavalian cell are not all exposing PS except on demand except when triggered so here is a cost cell expressing opsin you can see the protein label from the outside with red and the nucleus in blue and it's been treated with calcium and anionophore or it doesn't matter if you don't treat it and then you stain with a fluorescently labeled anaxin which binds to phosphatidylserine molecules and there's no labeling at all whereas if you take TMEM16F this is the Scott syndrome the bleeding disorder with ion channel that's also doing scrambling and you express that in cells nothing happens unless you add calcium and anionophore but when you do then you get fluorescent staining on the outside so this one works at the plasma membrane exposing PS on calcium trigger but this one doesn't do anything at all and that's not a surprise because hex cells any other cells that you want have plenty of different types of GPCRs none of them are exposing PS on the outside so this whole family of GPCRs is silenced at the plasma membrane and that may have something to do with the composition of the plasma membrane its thickness, its saturated lipids its cholesterol content so that's one way for the membrane perhaps to regulate the activity so mechanisms so that's what I want to spend the rest of the talk on so how do we measure these things I've made several comments about setting up assays which were done here this particular assay and the easiest assay we've done several different ones is to use a fluorescently labeled lipid you can buy these, we used to make them at one point and this is a phospholipid with a normal head group a phosphodiester a glycerol entity and then here on a short carbon chain you have a fluorophore and this fluorophore is nitrobenz dioxazole the nitro is relevant because when you chemically reduce it with dithionite which is this molecule over here the nitro becomes amino and the fluorescence is killed it's irreversible, it's gone so what we do is we reconstitute large unilamella vesicles like this we can have proteins in them or not the fluorescent lipid is included at the time of reconstitution so it's on both sides of the membrane and then you treat with dithionite which is this reagent here and it doesn't cross the membrane, one does controls for that and it can be expected because it got two minus charges it's not going to cross easily and when you treat with dithionite all the fluorescent lipid on the outside is killed so the nitros become aminos there's no fluorescent left and the inside is protected because it doesn't go out in an empty vesicle or in a vesicle with an irrelevant protein in it but if you have ops in here or another scramble is here then the lipids can exchange and all will become exposed to dithionite at some point so the assay goes from 50% in a dead system where you only modify the outside to 100% when you're able to exchange both leaflets and the reconstitutions are done like this I'm explaining this because it's important for what I say later we purify the protein, it's in dodecal multicide or other detergent, it's in a micelle so it's got a nice belt of detergent around it we start with premade vesicles to which we add a little bit of detergent to make them permissive so they are now able to take in protein and then when you combine the two you start removing the dodecal multicide with bio-beach it's a polystyrene adsorbent and then the proteins jump in some vesicles say empty some have one, some have two proteins and so on and the vesicles look very nice they have a nice size range that's a cryoem image done at the New York structural biology center and then you get assays like this so you add dithionite at this point and if you have no proteins in the vesicles you get a result like this 45 ish percent that's what passes for 50 in biology and then if you reconstitute increasing amounts of protein you get increasing extent of reduction and we top out here we never do better than this and that's because the reconstitution system is such and not just for scramblesies for any transporter there seem to be a population of vesicle that simply will not take protein they just refuse to take protein possibly because their detergent has been wicked away by the beads before the protein had a chance to get in whatever but there's a fraction of those that don't take protein so the assay goes from about 45 to about 80 85 percent instead of 50 to 100 the other thing to notice here is that the kinetics are all the same so even though these graphs look nice and you see nice curves and so on the rate of reduction that you see for liposomes is the same as you see for a proteoliposome which is we are not getting any kinetic information here these traces are governed by the rate at which dithionite bleaches the nvd fluorophore so all we measure are endpoints this is the amount of reconstituted protein you get increasing endpoints but you can still use these kinetics to make an estimate you know how many what the size of the vesicles is how many lipids there are per vesicle and that's where this number of more than 10 to the 5 lipids per second comes from so one point I made earlier I'm going to repeat here is that this this is not seen when you reconstitute any old protein you can take TMEM16 proteins that have amongst family members for example some will do it some will not you can dial up people and they'll send you proteins and they won't do it so this is a particular reconstitution of a particular protein that can first scrambling activity so because of the endpoint assay you can do something interesting you can play with the amount of protein you reconstitute so here is the protein to phospholipid ratio and because it's a yes no thing a vesicle that has a protein has the activity one that doesn't have it you can ask what is the probability that a vesicle has at least one scrambles you increase the probability like this when you have no protein of course no scrambles lots of protein you have lots of scrambles and it goes up very nicely according to personal statistics you can use that to deduce the size of the functional entity that was reconstituted because this is done in curious units it's the grams of protein per mol of phospholipid so you know what the from the fit constant what the masses of the protein that was inserted and it turns out in the case of our opsin molecule to be 90 plus 100 kd which is a bit which is a bit upsetting because an opsin monomer is 41 kd so the only way to deal with this result is to assume that the opsin is going as a dimer or a multimer and the way we think of that is the following which is you start with a monomer you have your permissive vesicles and as you remove detergent the first thing that happens is something like this where the protein spares detergent by making a face like that before jumping into the vesicle so you multimerize before insertion and just to emphasize one point the insertion it's a single insertion that confers activity it's not cooperative you don't need one insertion followed by a second partner and you know that because these graphs are all going straight up there's no sigmoid here so you never see that so a single insertion of a dimer or a multimer confers the activity to a vesicle and we know that we start with monomers there's nothing funny going on there because you can do experiments like this which is you can prepare a snap tag opsin you can immobilize it on a glass slide you label the snap tab of course with a fluorophore image all these fluorescent dots on your slide and then up the laser intensity so you can watch them bleach and they all bleach in a single step if these were multimers or dimers they would bleach in two discrete steps often enough so they bleach in a single step they're all monomers when they start sorry so they're monomers when they start but they enter the vesicles as dimers or multimers because of this graph okay so how does this thing work so we have two options one is a opsin protomer would scramble lipids by some magical means some short circuiting of the bilayer like this or it could be that the the interface between these protomers is somehow privileged and that's where the action is and we got around that by looking at some disease mutations there's a disease called retinitis pigmentosa where the rod cells are basically destroyed over time and there are several mutations lots of patients come through mutations are mapped so one knows about several mutations in this disease many are assigned to rodopsin and it's clear why they are disease mutations because the protein is not made or it's not traffic properly or doesn't bind retinal all kinds of things but there were several of them that were enigmatic there was no obvious reason perfectly nice protein and we got interested in several sites that face the membrane so this is this F45 on helix 1 pointed towards the bilayer and a couple here on helix 5 that also pointed the bilayer and we thought why then this might be a scrambling issue because they are facing the membrane and could be active that way and it turned out not to be the case all of these mutations that we tested I'll show you what they are on the next slide all of them scrambled lipids but they did something interesting so they are F220CV209M F2F45L all of them did something interesting it was obvious to the postdoc who was doing these experiments that there was something interesting here because she could use less protein to get a maximum signal than she needed to use with wild type even without making this graph she knew that there was something up because less protein gave her the same activity and the graph tells you exactly why because you populate you get up to a maximum probability by at a smaller point on this titration scale and if you do the fits you end up with this whereas the wild type comes in at a close to 100 all of these mutations have the molecular mass of a monomer so just individual mutations are allowing this to happen they prevent the pre-dimerization and they go in straight as a monomer so we know that a monomer of opsin will scramble lipids and that's what's depicted here so you have a large unilimela vesicle with 200,000 or so phospholipids you have a single opsin molecule and the leaflets here are being exchanged at faster than 100,000 per second so that's the setup we also had another quadruple mutation which was nastier to make and less physiologically important but this also went in as a monomer so we know this from multiple angles that an opsin monomer will do this so the opsin and these other GPCRs are famous proteins and there are so many structures available and there's not a hint it just looked like a little Teflon blob there's nothing in the protein that suggested how it might work so you know we're sitting around wondering whether we'd have to mutagenize the daylights out of the protein you know just do brute force stuff walk around the outside of the protein shell in the membrane and see whether we got any hits but then of course for years I've been talking to colleagues in the physiology department at Cornell and these are simulation people they're molecular dynamic simulation people there's now much more access for MD simulation just because of the computation power and the way the analyses are done so it was what could not have been done perhaps 10 years ago became extremely easy to do now easy for them to do and for me to watch so what they did was what they did was park an opsin monomer in a slab of membrane and the monomer now is important because it's a simpler computational system when you have one protein instead of the dimer for those interested and the interesting part also is that in total they were able to drum up 50 microseconds of time on this computation which is not literally not something you would think of not that very long ago but now it's possible the computers have names they're called Anton 1, 2, 3 etc but that's where the computations are done but the result was absolutely striking so but the guy who was doing this George Khelashvili was sitting there bored you couldn't talk to him he's saying well nothing's happening and he'd go off and drink coffee and then after a while it got to a point where you couldn't get him on the phone you couldn't talk to him at all because he'd seen something like this so this is an aggregate picture from the full 50 microseconds of simulation showing water molecules going between trans membrane helix 6 and 7 of opsin so here's a whole chain of blue water molecules populating this entire path if you switch out the water and look at where phospholipids are you find that phospholipid phosphates which is basically the linkage between the head group and the glycerol have populated the same pathway all the way to the top so there's a large entryway on the cytoplasmic side and this whole caveat like arrangement shows you where phospholipids have once been along this pathway so this is again between helix 6 and 7 and so the way this pathway developed and I'll walk you through this very quickly is that if you look at the bottom end the cytoplasmic end of helix 7 is a ionic linkage here between two residues and there's a tyrosine here that flaps around stochastically some of the time it talks to helix 6 some of the time it looks inward pointing to helix 2 the details are not important it's just sort of flapping back and forth and when it flaps in and this thing breaks then there's an opening there's a widening of this gap here and water starts coming in so you have to have the tyrosine disappear inward and this linkage to break and that happened with sufficient frequency and once the water comes in then lipids follow and so there are three lipids here it's a little confusing the color combination so there are lipids in red, green and blue so the red one is the one that has advanced the most after this after this opening has been created and the red's gone up here through some part of the simulation and as it comes up here about midpoint where this where this structure has a waist the waist expands just a little bit so the constriction is dilated and then the lipid goes up further and when it goes up further a couple of other lipids follow and eventually the red lipid manages to get past the waistline and the other two fall back in this particular arrangement and so if you take a snapshot and it's an extremely suggestive snapshot at one point of this pathway and it's a side view now to show you the helix 7 in orange and 6 in orange and 7 in red you have three lipids with their head tucked in very similar to the cartoon that I showed you earlier where you short-circuit the bilayer and this is a case where all three lipids were at this point the red one went further and the other two fell back but it looks exactly like this picture that was I think drawn in 2008 and appeared in a review article in 2009 that Adam drew and that Ivan I think called the credit card model for obvious reasons or at least he said that and that's how the drawing appeared but it looks extremely similar to that and this is now a possible way in which this in which lipid scrambling would occur according to the simulations okay so that was published sometime last year and there's a point that I want to make before I sort of wind down which is this has again nothing to do with the lipid sign with the signaling activity of rhodopsin or of other G-protein couple receptors so this is again a view from the cytoplasmic side there's a fair amount concerning what about what's known in terms of helix movements when you signal when you add a ligand in this case a switch of the 11s to all trans retinal in the case of the adrenergic receptor you'd add a ligand so there are particular movements of these helices the most dramatic one is helix 6 coming out like this so the easiest way to look at it and this is the you can conceptualize all gpcr papers now to 80% precision by imagining that you have a helical bundle like this ligand binds to the top and helix 6 moves out that's all that's happening for most of them and then the details are of course interesting, important and so on but that's what's happening and the G-protein binds that movement is what is depicted here and is not related at all to the movements that the simulation produced in terms of what's happening for for scrambling so helix 6 moves out in a different way parts coming from 7 so there's a big 10 angstromish movement at the cytoplasmic end of those two helices and various other movements occur but they're not related to the signaling movements that would bind the G-protein so this constitutes some form of constitutive twitching of the protein and this is something that is turned on by agonist or other binding so Adam possibly be giving this talk himself turn the cartoon into something more detailed so this is based on the simulation so you actually got the PDV structure of the protein in one of the states that was captured during the simulation and then you have have it parked in a realistic lipid bilayer with lipids that are moving between helix 6 this is the one here and 7 that is over here I think the blur is his the rest is from the simulation okay so let me summarize a part of this story so the first point is monomericopsin and that's true of these other GPCRs as well scrambles lipids that rate greater than 10 to the 5 per protein per second after reconstitution into LUVs so these are simple LUVs it doesn't matter if they are compositionally it's slightly more complex but we haven't solved the problem of whether cholesterol inhibits this activity or not because when you reconstitute with cholesterol you get a heterogeneous vesicle population which you cannot deal with if you have an endpoint assay so we are solving that problem in different ways single vesicle imaging all kinds of things but at the moment this is basically a PC or a PCPE vesicle and the scrambling rate is huge and that explains the result it puts a molecule into the observation of Wayne Hubble in 1993 and Hoffman a bit later about why disks or how disks are scrambling phospholipids in the assays that they reported so the MD simulation provide indication of hydrophilic path that's dynamically exposed between transmembrane helix 6 and 7 6 is also interesting for the signaling pathway but it's movements here are distinct from those that are needed for G protein binding and the pathway looks exactly like this credit card model I don't want to say that actually there's a little story which is about Conrad Block in a lipo centric topic here so Conrad Block as a PhD student rushed up to his advisor saying I've done the experiment it's exactly what I thought and the advisor said how boring so you know we don't want to be quite in this situation here of saying that we have a result that looks like something that was predicted okay and then this one comment here about the obsin mediated lipid scrambling may be required for maintaining the bilayer the disk bilayer keeping the number density fixed in the presence of these other ATP driven transporters so coming back to the cell that I started with so the E cell we have the possibility that there are proteins like this now obsin like proteins is this type that sit in the endoplasmic reticulum and arrange for bilayer exchange of lipids between the two leaflets of the bilayer and then you would have a similar protein sitting at the plasma membrane except that it would have to be turned on by some trigger either calcium or caspase cleavage the TMEM16 protein 16F that's involved in the in the exposing PS in platelets has in its the crystal structure that was published maybe five years ago and there are several more structures of that since that looks very much as if it's got a groove it's a dimer and each of the monomers has facing away from the dimer interface literally has a cut out it looks like a T-cozy handle the kind of thing that you would grab onto and the inside of that is hydrophilic it looks exactly like the original cartoon that I showed you so in that case the groove exists it's opened up because a helix moves out allowing a continuous passage in the case of a rhodopsin-like molecule it seems to be dynamically exposed because of these fluctuations in the membrane but there are many there are questions left so we have molecules at least some of them and clues to mechanism but then there have to be tests of the mechanism so there are those who would say the MD is fine where it is on a computer but one has to actually verify it in some way and there are obvious ways to verify it there are 16 residues one can try staple the helices together modify its polarity those tests have already been done quite deeply for the TMEM 16 family of proteins and the groove pathways you mess with it and you upset the scrambling activity so this general concept of having a transpile hydrophilic groove and allowing lipids to sort of swipe their way through at least is verified for the TMEM 16 family regulation so why are these proteins silent in the plasma membrane one is you can ask that question specifically about the mechanism but it is also interesting to know what role the membrane plays in dictating the activities of membrane proteins is the bilayer thinner there is it more rigid do lipids not access the bilayer there is some speculation that even though the lipid may move from one side to the other the release time can vary as quickly as they are transported across so there is a pile up so there are all kinds of questions to be dealt with there and then there is this perpetual problem of identifying the ER glycolipids scramblesis and if anyone wants to have their ear bent about what we are doing about this I can tell so I want to end with an acknowledgement slide there are several people here that's my wife up there who did the ops in work and actually she did the ops in work I am looking at you Judith because you said biochemistry student to my lab when we put him on this project and he didn't suit him at all and she looked at him sort of not doing so well he then went off I think to Rick Amasino to do more pleasant things grinding cow eyes and things like that so she said she can do all of this so she is the first author on the ops in discovery paper Mike Goran is the student here from who got his PhD with Brian Fox and he is the one who did the constitutive activation of the constitutive activity of these proteins big employer is the one who discovered the retinitis pigmentosa mutations that make monomers of ops in on their way to reconstitution George Khilashvili is the molecular dynamics simulation person Jeremy Ditman is a faculty colleague who does math when he is depressed there is nothing else to do when he is in seminars and is bored so that is what he does he has endless pages of scribbles so he is the one who helped to model that person distribution properly allowing for basical size distribution amongst other things it was two pages of fiddling during a seminar and that is what was used in the paper so I have various other collaborators Oliver Ernst is the structural biologist doing ops in work Josh Levitz is doing single particle analysis and Vadim Ashavski I did not talk much about his work but he is the one who follows up with these RP retinitis pigmentosa mutations in mice looking to see whether that the dimerization deficiency is what would provide an explanation for these funny mutations that no one has understood the molecular basis for okay so I will end with a movie from George and I am happy to take questions that is coming so that is the ops in simulation I see we can't do anything about unless you know unless you pinpoint some you have to make a mutant perhaps that is an obvious way to do it there is a membrane there is a membrane issue so I have alluded to the fact that plasma membrane has got more saturated lipids higher cholesterol and so on which is in the disk membrane which is where it signals from that is where it also scrambles and that is the original Hubble experiment but there is rhodopsin in the plasma membrane and there it doesn't signal and that plasma membrane is a high cholesterol membrane so there is a possibility that the two are coupled that the membrane controls what the protein is doing because it is not clear because other GPCRs for example would signal at the plasma membrane saturated lipids and so on but it is not so easily answered unless you have some control over the scrambles pathway so to do the experiment in reverse so you were you were surprised at the auxiliary function for the rhodopsin the opsin molecule so how many scrambleses are there going to be for your average mammalian cell so well how many different activities are there so there are like those I counted out for you which is there are at least three for the glycosylation pathways there should be at least one for phospholipid scrambling in ER possibly one for moving sphingolipids in the Golgi because the glycoshingolipids are made on one side of the early Golgi is made on one side as we move to the other side and so on these are activities so how many molecules are going to do this so GPCRs there are 800 out of them 900 of them in the mammalian human genome in principle they are all doing this or at least all the class A GPCRs are doing this they are not doing it in a real sense because they may be I mean they are in the plasma membrane and they are perhaps silenced by the membrane but they could contribute to the phospholipid scrambling requirement of the endoplasmic reticulum when they are first integrated into that membrane before they are trafficked out of that membrane so it could be a general fold all of them have the same characteristic 7 trans membrane helix same arrangement of residues and so on quite a reasonable amount of conservation so the question is whether there is a different class of proteins altogether that would do this I don't you know there may be others I mean there are plenty that don't I'll give you an example of one way to test for this which is you take endoplasmic reticulum so we just narrow it down to that organelle and dissolve it in a detergent and fractionate it by some mechanism we've been using velocity gradients forever because it's a convenient way to fractionate you get a peak of activity which means that lots and lots of proteins on that gradient have nothing to do with scrambling so if there is a large family of them they are coincidentally co-fractionating around 4s on that gradient so that's not a good enough answer but it's an approximation to how many different there are lots of things that fractionate there but it's not really conceivable that all of them would belong to one protein family there will be many more than we think at the moment just because of the way the activity the requirements for the activity thank you so you mentioned you were investigating what the cholesterol might block this scramble activity any clues from the modeling as to how a lipid with a particular head group or a particular saturation pattern might lock this scramble activity to a particular step so one is these are all the lipids that have been used are these fluorescent lipids of different stripes we've also used fluorescent groups on the heads of the lipids so you can modify amino phospholipids with fluorophores on the head all of those things are indistinguishably fast asile chains don't seem to matter to the extent that our ability to resolve kinetics are not optimal so we only have an endpoint data so what we did recently that was also published last year was to try to see whether there was a limit to the size of the head group and to the size of the head group so we made a whole series of pagulated lipids so different to a nested set of pegs 500, 2000, 3000, 5000 and this was in the context of the TMEM16 protein where there's a defined groove of about a nanometer or so in size so the expectation was that big heads wouldn't go through they all did so the question is how much the groove is physically transferring lipids in this fashion or whether it's doing some of this but also doing something to perturb the membrane in its immediate vicinity so the shell of lipids around this groove that is somehow defective and also allowing things to pass so the problem with the peg experiments is even though they're sort of spectacular and unnerving because large things are going through pegs can unravel so we're not precisely sure whether they stay as globular entities in any way before passing through so those experiments are being redone with dendrimers now with defined structures so we'll see whether that happens so head groups don't seem to matter for the moment the asile chains don't matter the fluorophore can be anywhere very promiscuous what this protein will not do is transfer one of the dolicol intermediate so the MAN5 gluknak II diphosphate dolicol needed for anglycosylation is not scrambled by this and there the chain might matter or it could be the diphosphate linkage which is not found in any phospholipid no more questions thank you that was beautiful so I understood that retinal is not required for the transport but is it inhibitory does it actually change the endpoint or I guess the kinetics is hard but do you get a less efficient transfer so the question was whether retinal blocks transport so the work that Mike Gorin did and that we published a few years ago we went out of our way to check this but I'm not entirely convinced that we had it right so the thing to do is to make rhodopsin so dark adapted rhodopsin with 11 cis retinal in it you sit in the dark room when you express the protein you add the retinal to cultured cells all in the dark, all dim red light I can't do this, I can't see anything but Mike was able to do stuff reconstitute in the dark with the fluorescent lipid add dithionite in the dark 10 min dissolve the preparation to get rid of the dithionite and then take it out to do the endpoint measurement in the fluorimeter so he did all of that and there was still activity the only problem was that he needed a rather large amount of rhodopsin to reconstitute for that activity and it looked as if it was a sixmer or an eightmer which is what we reported what it could have been is that most of it was not doing anything and that Mike in a moment of weakness light on or did some so a small fraction was converted to anopsin and so what we were measuring was we had to load in a lot of protein in order to capture the small fraction so we are revisiting it, I doubt that this was a case I mean he's not turning on the light but anyway that's a possibility because it's unusual to have such a big oligoma going into the vesicle so I think that's possible so in that case you'd be blocking the path and it's close to that path so it's sitting there close to it like this and so one would imagine that you'd block the path so we are doing that again but okay I think Anat will be here for a few more minutes if you have more questions so yeah so we have there's a whole family of opsins I mean apart from the mammalian type GPCRs there's a whole family of opsins there are bacterial ones, there are kaol ones some don't have retinal some don't even have the lysine on which the retinal would be would be attached so we did this same type of experiment with bacterial adopsin from the purple membrane so you grow the cells there are complications there because you can strip the protein in different ways to get rid of the shell of annular lipid that comes with it but under conditions where bacterial adopsin was able to pump protons in a light dependent way it was scrambling lipids in a light independent way so same kind of thing and it was doing it as a trimer so again using these same reconstitution plots that I showed you the fits there yielded precisely a trimeric molecular weight which is what bacterial adopsin's core structure is it makes itself into a trimer before in the purple membrane it expands into these big lattice so this one the last video oh I see so a couple of comments first of all bacterial adopsin doesn't lose the retinal it's a two photon cycle it regenerates on this one secondly it looks like mammalian adopsin but it isn't it's got seven transmembranes bands but the arrangements are different the rigidity of the bundle is quite different so in fact the hope was so the hope was that it wouldn't do it and then one would start before this MD simulation was done the idea was to try some chimera type thing to see if you could convert a bacterial adopsin into a mammalian adopsin except that it was doing this we again did some simulation work with bacterial adopsin and there the pathway seems to there seems to be a polarity generated initially by the interface between two of the protomers before it goes along so it would not look exactly like this we haven't followed that up you should do so if you take an ABC so it depends where you do the measurement if you do this in a synthetic vesicle you get almost no transfer you get very small amplitudes and that's because you're principally moving lipids from one leaflet to the other and you get strain immediately by populating one leaflet at the expense of the other in a mammalian cell in a live cell where you do such measurements the type that I mentioned at the outside there's rapid exchange between the biosynthetic compartment the plasma membrane there's a lot of vesicular exchange and so on so that gradient the potential is lost more or less immediately thank you