 Okay, so welcome back to our second seminar in our inaugural International Steenbach Lectureship. So for those of you who weren't here yesterday, I'll just quickly reintroduce Professor Thomas Langer from the University of Cologne in Germany. And so Thomas is a world leader in mitochondrial biology and especially as they relate to proteases. So yesterday we heard a fascinating talk about a pair of proteases in mitochondria that regulate fusion and fission in response to stress and response to changing metabolic conditions that had some pretty profound implications on whole organismal physiology. And today showing the eclectic nature of his work will stay rooted in proteases I believe but really pivoted to a very different topic and about membrane organization and in lipid trafficking. So how do you actually move hydrophobic molecules from different membranes throughout the cell? Something that's I think largely untapped and through mechanistic biochemistry that Thomas is really helping us understand. So with that please welcome again our inaugural Steenbach, International Steenbach Lecture, Professor Thomas Langer. And thank you for coming again. As Dave pointed out today actually I will talk a little bit about something a different topic that is of interest in our lab but which of course still concerns our pet organelle and these are the mitochondria. Yeah and today as you will I will introduce you a bit into our work on this lipid trafficking within this organelle. Now actually I would like to start out with a very general question and that is how do cell membranes maintain their lipid composition and this is actually something that is really poorly understood I think broadly poorly understood but it's also very important to understand because a defined lipid composition is of course important for the function of any cellular membrane and the proteins that reside in this membrane and that is of course also true for mitochondrial membrane just as an example and this is a lipidomics analysis of mitochondrial membranes in yeast and they have a very defined lipid composition like any other cellular membrane and that is very important for the functional integrity of mitochondria. If you disturb it mitochondrial functional integrity is also disturbed. I mean what I would like to highlight here is that the lipid composition is common to other cellular membrane in that PE phosphatidyl ethanolamine and PC are very abundant phospholipids in mitochondrial membranes but mitochondrial membranes are distinct from our other eukaryotic membranes in that they have a phosphodiagyl glycerol phospholipid cardiolipin you see here the structure that is actually synthesized within mitochondria and that is they share with bacteria so this is a typically bacterial derived lipid and it's characteristic for mitochondrial membranes in particular the mitochondrial inner membrane but it also seems to be present in the mitochondrial outer membrane. Now cardiolipin is a specific lipid because so-called non-biolayer lipids so it forms hexagonal phases in vitro so if you reconstitute liposomes in vitro and have cardiolipin in them there's not a it doesn't arrange in lip in in bilayers but forms these so-called hexagonal phases and in vivo there are many functions associated with cardiolipid that all have to do with membrane fluidity kind of membrane fusion and fission because this is the specific function of these non-biolayer lipids. Now in other cellular membrane that do not contain cardiolipin another phospholipid is fulfilling this function this is phosphatidyl ethanolamine so we have here phosphatidyl ethanolamine and cardiolipin both known bilayer lipids and they basically are sought to help membrane reshaping processes in a broad sense. Now cardiolipin is synthesized within mitochondria there's an enzymatic cascade that resides in the mitochondrial inner membrane I don't want to go into any detail here these are just the names of the yeast enzymes and these results then in the formation of a mature cardiolipin molecule and then there's some so-called cardiolipin remodeling occurring which base as a consequence of this remodeling the cardiolipin molecules all have the same acyl chains or symmetric molecules and this all occurs in the mitochondrial inner membrane before redistribution between the membrane membranes occurs. Now the synthesis of this cardiolipin starts from a precursor phospholipid which is phosphatidic acid and this precursor lipid is imported from the endoplasmatic reticulum and in fact the endoplasmatic reticulum as you may be aware of is the major site of form I should say the site of phospholipid synthesis of lipid synthesis in the cell and this complex diagram should just highlight that you know many lipids phospholipids are synthesized in the ER and then redistributed to different cellular organelles including the mitochondria. Now for instance phosphatidic acid which is somewhere here is then convert is imported from the ER and converted into cardiolipin but there are also other phospholipids must be imported into the into the mitochondria to preserve the mitochondrial lipid composition but of course there's also a lot of lipid exchange with other cellular organelles that basically receive lipid synthesized in the ER. Now the way the field is thinking of lipid trafficking is basically summarized in this cartoon which says there are different ways and we do not understand much about it. Basically what you see what is just what I just would like to point out here is that the spontaneous transport of phospholipid is not really occurring easily simply for biophysical reasons or bio energetic reasons so it's very difficult to remove a hydrophobic lipid out of a bilayer so there is no evidence that this is of any relevance in vivo even at sites where two membranes are in close contact. However these sites are sorts to be important for lipid trafficking that is mediated by specialized proteins so-called lipid transfer proteins that pick up lipids from one bilayer and transport them then to another bilayer and this is actually appears often to occur at membrane contact sites. Of course a very dominant way of transporting lipids in the cell is via vesicular transport and we have the whole secretory pathways and this actually allows reshuffling of lipids between different cellular compartments. However the connection so the mitochondria are not connected to the secretory pathway. There are some recent findings from mainly from Heidi McBride's lab at McGill that there are mitochondria derived vesicles but they have not been linked to lipid trafficking so far at least. So it was really unclear after a few years ago how mitochondria receive their lipids and how lipids are redistributed within mitochondria. Now how did we get into this? Maybe you may remember if you have been here yesterday my lab is interested in mitochondrial proteases and that's actually not directly linked to it and for a long time although we were working on membrane bound proteases we didn't think much about lipids I have to admit. However we got interested in a protein that we now consider as a membrane scaffold in the mitochondria inner membrane and these are prohibitins because we identified these proteins in a proteomic approach as a binding partner of one of our pet proteases which is the M triple A protease. It's an ATP depredated protease in the inner membrane. It's not important for the talk today. Now this protease basically forms large assembly of about 200, 2 megadolts in the mitochondria inner membrane with these prohibitins and these are highly conserved proteins you find them in all eukaryotic cells highly conserved. There are always two members of this family PHB1 and PHB2 in the cells and they form large complexes in the mitochondrial inner membrane and we purified them some years ago now from the yeast and some single particle M analysis combined with cross linking studies suggested that these prohibitins subunits form large rings in the mitochondrial inner membrane that have a diameter of about 20 to 25 nanometer and this suggested to us already that this may be a kind of a scaffold membrane organizing protein in the mitochondrial inner membranes and multiple copies of PHB1 and PHB2 form this ring-like structure. I should say that prohibitins themselves are highly conserved but they're also part of a even more distributed protein family that is called the SPFH family they all share domain common domain here in this part of the same topology in the inner membrane are now sought actually to form similar structures in other cellular membranes so we may look here at a conserved family of membrane organizing proteins presence in different cellular membranes but prohibitins we believe are mainly exclusively present in the mitochondrial inner membrane. Now we what do you do to really define the function of such a protein we took in my lab actually two approaches and one was more similar to what I said yesterday we made a knockout generated knockout mice for one of these prohibitins subunits prohibitin 2 and deleted them in the mouse found that this embryonic lethal so it's really important function in mitochondria but when we then looked at tissue specific knockouts in the various tissues over the year turns out that in all tissues this has a very devastating effect so that by itself we could talk about it so we looked in the brain or in the cardiomyocytes or in beta cells in all cases we have a very severe phenotype and as a kind of a recap to my talk yesterday it turns out that an impaired mitochondrial fusion due to this stress-induced processing of this long OPA1 from spy OMA1 is actually the cause of cell deaths in these in these models so these are all examples by the loss of prohibitins we activate OMA1 and this results in stress-induced processing now this actually fostered us to look at this in more detail and I told you the story yesterday but and it also told us a bit what the physiological consequences or of a loss of the scaffold protein is but it really didn't tell us much what the molecular function of this protein is and that's not so easy to answer if you think of a scaffold protein that doesn't have a enzymatic activity so therefore we also took a in parallel a different approach and used yeast and yeast genetics to basically look for G and were puzzled from the very beginning that in contrast to the mammalian system where the loss of prohibitins was essential the loss of prohibitins and yeast had no obvious phenotype so that was really puzzling to us and therefore we did a so-called synthetic genetic array where we looked basically in G in genes that genetically interact with prohibitins and that are essential in the air in prohibitin deficient cells so if they are redundant functions of these two genes maybe we learn something about the function of the prohibitins now in this screen that Christoph Osman a former PhD student in the lab performed he came up with 35 genes that he named CHAP genes for genetic interactions with prohibitins whatever and we have to say we stared at this list for at least a year and tried to make sense out of it was not so it was didn't seem render it didn't seem random so they were functional classes but we didn't really understand what this all means until he did another screen on top of that I don't want to go into that that actually led us to note that all of the mutate all of the genes that basically are essential in prohibitin deficient yeast cells affect in one way or the other the accumulation of phospholipids in particular the phospholipid phosphatidyl ethanolamine and cardiolipin these are these non hexagonal type non bilayer type lipids in the mitochondrial membrane so all of the gene or the most of the genes that we identified in the screens affected in one way or the other the accumulation of PE and cardiolipin and Miriam Greenberg labs had shown before that also the common loss of both phospholipids is leisling yeast so we have here intimate cross talk between prohibitin genes with phosphatidyl ethanolacoramine phosphatidyl ethanolamine accumulation and cardiolipin accumulation and that was the moment when my labs got interested in phospholipids I still blame Christoph for that till now because it's a difficult topic anyway this this work basically let let us to propose a model how prohibitins work as membrane scaffolds so we think they are membrane organizers that form these large rings in the membrane and recruit specific proteins to specific sites so they are functional domains in the membrane that are defined by let's say defined protein complement and we did in the meantime some proteomic studies in in mammalian cells and I think a rather fair understanding which kind of proteins are there at this scaffold but based on our yeast studies we also think that they may affect the asymmetric distribution of specific phospholipids in the inner membrane and may basically enrich domains of in the inner membrane of lipids like PE or cardiolipin so in that sense they define functional domains in the inner membrane that have a defined protein and lipid composition I should say this is just a model despite all the work and an alternative model that is not completely different but a bit different is shown here on the right side keep in mind that the mitochondrial inner membrane is highly protein rich people estimate up to 80% of this membrane are proteins so maybe it's very important for certain functions like fusion for instance or yeah to have protein free lipid patches in the inner membrane so it could be that these scaffold proteins actually do the reverse they have a kind of fence like function and keep the proteins from specific spots allowing basically the formation or maintaining this lipid free protein free lipid patches in the mitochondrial inner membrane preserving specific functions so this is the way how we think of that and we try to basically clarify that the mode of action in more detail now another outcome of this genetic screen or this work in yeast that we did is that we now all the sudden had a set of genes in hand that were linked to the accumulation of PE and cardiolipin and I mean we looked then in the literature what could these genes do and I have to say I was really surprised to under to see how little is understood so many open reading frames that we really didn't know what they do for some of them we have learned that in the meantime so we decided to look at these genes in a bit more detail because they apparently affect the accumulation of PE in and cardiolipin in more in mitochondria in some one way or the other now there's always an important decision how to pick a gene to look at it in more detail in this case it was however very easy because I had actually two students in the lab one did this genetic screen that I just described you and then I had another one who basically did a quantitative mass spec approach looking at potential substrate of the I triple A protease why me one that you may remember from yesterday's talk which is an ATP dependent protease that resides in the mitochondria in a membrane this was Tanya Engman at the time another PhD student and she did this quantity of quantitative proteomics and actually also identified two proteins that the corresponding genes were identified in Christoph C's gene that was in some more obvious that this attracted our attention and though we looked at those in more detail now it turns out that these proteins are part of a highly conserved protein family all of these proteins in the meantime have been localized to the mitochondria in this intermembrane space in yeast there are three members UPS one UPS three two UPS three and in mammals there are also three members really one slow-mo one and slow-mo two so that's a highly conserved family of proteins residing in the intermembrane space and Christoph and then another Christoph was another PhD student in the lab Christoph potting actually looked at this more in detail using some biochemistry to characterize these proteins and it turns out that these proteins are highly unstable UPS one and UPS two and are constantly degraded by proteases that reside in the mitochondria in a membrane this is this why we want protease here we look in yeast but constantly degrade these proteins so the steady state level of these proteins is very very low and they are only stabilized upon a formation of heterodymary complexes with a common binding partner that is termed MDM 35 this is a CX9C protein which basically and we think that this is important for the import of these two proteins into the membrane and then forming this heterodymary complexes so we have this situation that we have two proteins part of heterodymary complexes the only upon heterodymaryization these proteins are stable and protect against are protected against degradation the steady state level is very low we hardly detect them by western blood depending on the antibody that we have but when we delete these genes there seem to be a lipid specific effect UPS one deletion affects the accumulation of cardiolipin in mitochondrial membranes whereas UPS two deletion affected the accumulation of PE in the mitochondrial membranes now we of course wanted to understand this in more detail and started to look in more detail at UPS one so Matthias Harkin at the time did the lipidomics analysis by mass spectrometry and basically looked at phospholipids in yeast strains lacking lacking UPS one and what he observed was that we saw the reduced reduction of cardiolipin but what may was maybe more informative to us was the fact that he saw the accumulation of phosphatidic acid in these mitochondrial membranes isolated for mitochondrial lacking UPS one so we have the accumulation of phosphatidic acid and reduced cardiolipin levels now when you look at the enzymatic cascade that I just showed you before that resides in in a membrane phosphatidic acid is the precursor form that accumulates cardiolipin is of course the product so UPS one MDM 35 is an intermembrane space protein soluble intermembrane space protein so it was obvious to speculate that maybe for UPS one has anything to do with the trafficking of you phosphatidic acid to the inner membrane to basically allow its conversion into cardiolipin I think it was an obvious guess simply from the topology of these proteins together combined with this lipidomic data and we did some genetic yeah we supported this model actually with some genetic data I don't want to go into into it now to to basically look at this in more directly we basically are Takashi Tatsuta and this I have to say we're at that time all new essays in the lab set up a lipid transfer essay in vitro and as I'm not sure whether you are all aware with the type of these essays that are used in the lipid trafficking field quite extensively I just wanted to introduce it to you very briefly so what you do here you basically generate donor and acceptor liposomes that differ in the lipid composition so the donor liposomes have the lip contain the phospholipids that you are suspecting to be transported the acceptor liposomes are free of them free with them you label them differently so that you can also distinguish them by their fluorescence and most importantly you load one type of the lipid in this case with the donor liposomes with sucrose to make them heavier and this allows them after doing this reaction when you then incubate these two lip types of liposomes with your protein let them incubate if trying to be to allow transfer then you can later in a flotation gradient separate this liposomes again and so we basically separated these liposomes again and then determined the lipid composition of the acceptor liposomes and any of these lipids that we find that in the acceptor liposomes must have been transported by this lipid transfer protein now that's the basic principle of this there of course variations of this essay now doing that when Takashi did this the result was really striking because when he looked at the different types of phospholipids only one phospholipid was efficiently transferred and this is phosphatidic acid and this showed us with together with other experiments that UPS1 MDM35 are are able to transfer phospholipids in vitro from one liposome phosphatidic acid from one liposome to another liposome so UPS1 MDM35 seem to be lipid transfer proteins for phosphatidic acid and we looked at this in more detail and this is the scheme we came up with how this UPS1 MDM35 proteins actually transfer phosphatidic acid now UPS1 picks up PA molecule at the inner leaflet of the outer membrane and upon assembly with MDM35 it is stabilized and it is basically we think stabilized in a transport competent form it this triggers the dissociation from the membrane it allows then and transport it allows then docking to the outer leaflet of the inner membrane where this complex again dissociates allowing the release of the bound phospholipids from from this lipid transfer protein and this phospholipid can then be used for cardiolipid synthesis now this is just a cartoon very simple we think for instance that this PA transport does not occur via free diffusion of this lipid transfer protein we favor much more the idea like many other people in the field that this lipid transfer proteins are actually closely located to membrane contact sites so that which basically reduce the distance between that needs to be where lipid needs to be transported from the outer membrane to the inner membrane making the process simply more efficient in vivo than in vitro okay one point was very interesting to us that we observed when doing this in vitro lipid transfer asses and that was that when we added cardiolipin to the acceptor membrane so the two receiving membrane this cardiolipin seemingly inhibited the transport of phosphatidic acid and that we found interesting because these are concentrations that you really find in vivo under physiological conditions the cardiolipin concentration is around 15% depending on that so we looked at this in more detail and it turns out that at this physiological cardiolipin concentrations this UPS protein cannot dissociate anymore from the acceptor liposomes and basically the back transport is inhibited and this we think is a very interesting observation if you think about the stability of these proteins because maybe this actually serves as a kind of a sensor of the lipid content or cardiolipin concentration in the acceptor membrane so we can at least envision that under low cardiolipin concentration you have this transport bi-directional transport in both directions so UPS1 picks up PA transports it to the inner membrane it's converted into cardiolipin and then dissociates again picks up the next PA and basically boosts cardiolipin biosynthesis until the cardiolipin level in the inner membrane reaches a concentration that prevents the dissociation of UPS1 from the inner membrane now under these conditions because there's so much cardiolipin now in the inner membrane this protein cannot dissociate anymore and therefore it's degraded by the proteases and limiting the PA transport this is I should really highlight that this is a model that we have at the moment there are clearly more experiments needed to support it but at least it gives us an hypothesis how actually the sense may utilize that's a lipid transfer protein to sense the concentration of a defined phospholipid in the my the hundred in a membrane now this work was all done in yeast but I just would like to emphasize that this process is highly conserved and occurs also in mammalian cells so we looked in human cells and this was work by Christoph potting in the lab human mitochondria contain homologs of all of these components we have a UPS1 homologous term pre-le-1 MDM35 homolog tribe one and he did biochemical experiments showing that we have exactly the same reg alone here the pre-le protein is degraded is a substrate of this why me one protease constantly degraded under these conditions and you may remember I showed you yesterday as a news this is an example for illustrating the knockout of why me one because it always accumulates in all tissues this so this pre-le protein is constantly degraded only upon assembly with tribe one it stabilized the dense against degradation and this heterodymeric protein complex serves as a lipid transfer protein for a particular for phosphatidic acid so we have exactly the same situation and the same regular on in in yeast and in mammalian cells and I should actually I forgot to mention this all of these components were identified in our screen the yeast screen for prohibitants they're all synthetic leasel for prohibitants so this is all seems to be one functional regular on there okay now in mammalian cells I think this we think this is interesting because one of these protein tribe one are actually is p53 inducible and we think that this is a very important response to preserve cardiolipin concentration in the mitochondrial inner membrane because as you may know cardiolipin one function of cardiolipin is to be an acceptor for for cytochrome C and it's therefore very important for a cell to keep cardiolipin concentration high to prevent the release of cytochrome C from the inner membrane and then finally from the mitochondria in case of the apoptotic signaling is increased so we think that this lipid transfer protein by preserving the cardiolipin concentration protects the cells against apoptosis and indeed when we take away these lipid transfer proteins in knockout knockouts or something then these cells are much more susceptible to apoptosis and we can rescue it by providing simply more of the phospholipid to the cells so this just highlights the physiological relevance of a defined lipid concentration in this case it's cardiolipin to protect against apoptosis okay now one point to switch gears a little bit here that was difficult for us to made it more difficult for us to recognize that these proteins are a novel class of lipid transfer proteins is simply the fact that they did not show any sequence similarity with any known lipid transfer proteins that was a bit puzzling and therefore we were always interested to solve the crystal structure of these lipid transfer proteins and we were really fortunate that we were able to collaborate with Steve Matthews at Imperial College in London who actually worked on these proteins for completely different reasons he was interested in p53 regulated genes and we actually teamed up and they and I should really say this is their structure they solved it they solved the crystal structure by NMR and x-ray crystal crystallography of a hetero dynamic dimeric complex of tripe on a slow-mo one which is a homolog of European one UPS one in a million cells and this is the structure of this hetero dimeric lipid transfer protein you see here this slow-mo one which is UPS one or really one and maybe I should UPS one looks identical I should say because the group of Toshi ended this in parallel in yeast so I could also call it UPS one you cannot distinguish it forms this anti-parallel seven stranded anti-parallel better sheet which forms this kind of concave barrel that is then closed by this diagonal long helix there and together with in green shown here the triable MDM 35 protein which forms I come to that which basically together close this cavity I mean the striking thing about this structure is that it is strikingly similar to known lipid transfer proteins although there is no sequence for more or more logic I think it's really a striking example for convergent evolution here and what this is basically an overlay of these two structures and they are literally highly identical you see here this is the helix that is here replaced in this case in red here by the MDM 35 so the C-terminal helix in this phosphatidyl inocytol transfer proteins that contain a so-called start domain is basically the C-terminal helix is replaced by MDM 35 tri-protein in the case of the hetero dimeric lipid transfer protein class so we and we think that this and have some mutational and I did some mutational analysis that then phospholipids in this case for the for UPS 1 phosphatidic acid basically bind in this cavity with the head group here there are some charge interactions and with the as a chain having some hydrophobic interactions with this beta sheet so it's a there's a classical for this fold of a classical lipid transfer protein there's no sequence similarity but this is completely consistent what we see actually in our biochemical experiments so we were really happy about that I should say this MDM 35 has also quite interesting structure because it's expected between helical bundle and this is typical for this so-called CX9C proteins these are proteins that are oxidized in the intermembrane space of mitochondria by another protein that is basically termed Mia 1 and this structure of these two proteins Mia 1 and tri-up is really almost identical with the exception of an N-terminal extension which is important for the function of Mia 14 in this redox regulation so this was all fine and we have now can conclude from that now we have identified a new class of lipid transfer proteins that transfer lipid phosphatidic acid from the outer membrane to the inner membrane to allow cardiolipid synthesis this is just to wrap up what I've told you so far so these are heterodimeric complexes UPS 1 MDM 35 really one tribe one transfer of phosphatidic acids we think this transfer occurs at contact sites which basically facilitate the lipid exchange between both mitochondria and membrane I showed you the reaction cycle association dissociation of this complex this structure similarity and this is a highly conserved protein family okay now for the rest of the time I thought I introduce you to some more recent work that we did looking actually at the other member of this protein family and this is this UPS 2 MDM 35 complex remember we have we started out identifying these genes here with UPS 1 deletion affecting cardiolipin whereas PR UPS 2 deletion affecting phosphatidyl acenolamine so we of course wanted now to know are also these heterodimeric complexes lipid or do they serve function as lipid transfer proteins in mitochondria and of course if you all the work that we have done already ah sorry I should introduce you to that so now how can how can UPS 2 affect PE accumulation in membrane what I didn't tell you so far is that besides cardiolipin mitochondria also synthesized phosphatidyl acenolamine in their membrane and that I think is a function of mitochondria that is often forgotten because PE is in fact a major constituent of all cellular membrane and PE synthesized from the mitochondria is actually exported from mitochondria and distributed in all cellular membranes now the relative contribute their different PE synthesis pathways in the cell and the relative contribution of the mitochondrial PE synthesis varies a lot between different tissues and different organisms in yeast the mitochondrial PE is a major source of PE in the for the all cellular membrane in membranes it differs a lot between different tissues now and how do they synthesize PE now mitochondria do synthesize PE from precursor lipids that are again imported into mitochondria from the YAR in this case this is phosphatidyl serine that is synthesized in the YAR transported from the YAR across the outer membrane to the inner membrane and in the inner membrane there's a phosphatidyl serine decarboxylase that decarboxylates phosphatidyl serine converts it into PE which is then redistributed across mitochondrial membranes but can also be released from mitochondria and then by methylation basically converted into phosphatidylcholine another major constituent of cellular membranes so this is really an important function of mitochondria depending on the organism you're looking at because it is important for the synthesis of two major phospholipids membrane constituents of cellular membranes PE and PZ but as I said there are different PE synthesis passes in a cell and it depends really on which cell you are looking at good by the way you see again here highlighted how many lipid trafficking event must occur between these membranes and actually we hardly understand any of them actually in this case we don't understand a single one so it's really ugly okay so how can UPS2 and MDM35 affect the accumulation of PE in mitochondria now it's an obvious idea that it somehow has something to do with this trafficking event so we started out and actually purified heterodimeric complexes in vitro this turned out to be a bit more complicated because these proteins contain cysteines which makes it a bit more difficult to express but we generated a variant that basically could be expressed and we could purify heterodimeric protein complexes and much to our satisfaction it turns out that also these heterodimeric protein complexes serve as lipid transfer protein however they have a strikingly different substrate specificity in this case UPS2 and MDM35 are highly specific for phosphatidazerine whereas other other phospholipids are not too hardly transported in vitro so we have now UPS1 specific for phosphatidic acid UPS2 specific for phosphatidazerine now we also wanted to know what is the human orthologue of that and for that we used a yeast screen where we basically complemented by expressing the human orthologs of UPS2 in yeast cells lacking yeast UPS2 and basically look for the suppression of the phenotype and as you can easily see expression of this of this protein slomo2 restores growth as well as the accumulation of PE in UPS2 deficient yeast cells identifying clearly slomo2 as the orthologue of yeast UPS2 and slomo2 also forms a complex with tribe in a heterodimeric complex which acts like the yeast protein in phosphatidazerine transport and this is basically shown here this is a different assay now but doesn't matter it's for a sense sequencing assay where we look at the transfer of nvdps and you see this protein although less active and the yeast protein is clearly able to specifically transfer phosphatidazerine but not phosphatidic acid in this assay so we have another type of lipid transfer protein UPS2 MDM35 slomo2 tribe that is in this case specific for UPS2 rather than for UPS2 for for for the specific for phosphatidazerine rather than UPS1 that is specific for phosphatidic acid okay now of course these were all in vitro experiments and we wanted to really verify these results by in vivo experiments thus do these proteins really act like ps transfer proteins in vivo now this is a bit more difficult than you may think and this is a scheme in yeast about on the different pathways for the synth that lead to the synthesis of PE and it's really a nightmare but this is why yeast is so beautiful to work with sometimes because you can really manipulate this genome now like crazy even more like like Chris Parker with Chris Parker's human cells so we have a kind of a competition in the lab at the moment who deletes more genes in one cell type but the yeast people are clearly ahead in this case we basically deleted all non-minor hundred pathways for PE synthesis and I don't want to go in too much into detail is there's a psd2 protein in the Golgi and there is a pathway that basically utilizes the head groups from single lipids for the synthesis of PE we can basically delete these corresponding genes we can also prevent the synthesis de novo synthesis from from ethanolamine of PE from ethanolamine or choline by just not providing this to the growth medium so we end up with cells where we deleted these proteins don't add this that can only synthesize PE along in a psd1 dependent manner that's actually the only thing you should understand here at that point so we really look at these cells that only have these pathways nothing else so any PE that we see is generated by psd1 but when you delete them psd1 it's lethal these cells die now then Mari Alton an PhD student in the lab actually did a labeling experiment now with in the hope that if you delete your ps2 now the PE synthesis should really break down now when she did this this is the result she labeled the cells with serradial labeled serine and the result was rather disappointing because when she looked at the accumulation of PE and also pc which is actually just generated then from pc so it's all pc that is there newly is also synthesized by psd1 it was completely normal and that was really a bit disappointing this is a quantification there was really absolutely no difference in the accumulation of pc in these cells and only a minor decrease in the synthesis of PE in these cells now that was really now a problem for us and startups make thinking how can this be we have here lipid transfer protein that clearly synthesizes ps in the transverse ps in vitro ps is required for psd1 to convert it into PE so how can this be that this is another lipid transfer protein how can this be now one idea we came up with is actually outlined here so maybe the following happens in these cells now psd1 is a enzyme that actually resides in the mitochondrial inner membrane it has a membrane anchor it's anchored to the inner membrane by one transmembrane region and exposes the catalytic domain to the intermembrane space maybe it's a bit of a strange idea but maybe psd1 actually is able to act in trans and decarboxylate phosphatidyl asanolamine in or when it is still in the outer membrane and therefore you would not need to have this ups2 mdm35 dependent transfer of ps to the inner membrane so this model would predict that we have basically two pathways how pe can be synthesized within mitochondria by psd1 one pathway that is ups2 mdm35 dependent and depends on lipid transfer between both membranes and one pathway that depends on the transaction of an inner membrane and of psd1 localized in the inner membrane at the outer membrane now can this be the case now to to test this possibility takashi tatsutsa in the lab actually set up an in vitro assay to prove or to test whether psd1 is in in vitro able to decarboxylate phosphatidyl serine 2pe in a bilayer that is just in juxtaposition now what he did is the following so he basically synthesized psd1 in a cell-free system we use an E. coli system for that in the presence of liposomes and this resulted in the insertion of psd1 into the liposomes and there are some ways to test the activity of psd1 in these liposomes there is an autocatalytic cleavage don't want to go into detail so we were very very confident that this psd1 is really catalytically active now then he purified these proteolyposomes by a flotation gradient and then incubated with these proteolyposomes containing reconstituted psd1 with liposomes that contain ps and i should forgot to mention but these liposomes of course did not contain any peo ps so the only ps that was provided to these reconstituted psd1 molecule was provided in trans by these liposomes that were mixed with these proteolyposomes and then we purified these and analyzed this by mass spec and what he found is actually summarized here in DTCs a time-dependent accumulation of PE by these psd1 molecule in trans and we did many controls to really convince ourselves one is for instance that you just inactivate psd1 so it's really depending on the catalytic activity of psd1 we can exclude the fusion of the liposomes during this reaction so we are very confident from this assay which is a purely in vitro assay that psd1 can actually decarboxylate phosphatidylsirin to phosphatidylsannolamine in a lipid bilayer that is actually provided in trans to this man so this would be at least consistent with our model now another prediction of our model is so this seems to be not completely off so another prediction of this model is that if this molecule acts in trans it must be critical that it can reach the outer membrane so the spatial arrangement between the inner and outer membrane should be very critical if this is the case then it would be really interesting to see because it could be a very general principle that inner membrane proteins can actually act in trans as long as the inner membrane is really close juxtaposition to the outer membrane now how can we test this possibility now this actually made us getting interested in a protein complex that was actually in recently identified in three labs independently of each other in walter neupers judinaries and nicolas faunas lab uh in germany there that uh they identified a hetero oligomeric protein complex that they're called the membrane of membrane contact site and crystal organizing system micos which basically sits in the inner membrane and is basically important for the ultra structure of mitochondria and for the alignment of the inner and outer membrane that's a rather complex molecule and there's intense research going on on this molecule and this complexes are sought to sit on this crystal junctions so these are these invaginations where the crystal are formed and there are also interactions with the outer membrane that maybe preserve the arrangement of these two membranes with each other so we thought if this is uh if we basically look in cells that lack micos complexes we disrupt the ultra structure of mitochondria and that should somehow have affect the ability of psd1 to act in trans so this we did and in jody's lab actually they generated a strain that tronazine freedmen in her lab who basically lack all these core micos subunits so all six genes so we have now a yeast strain with six genes and we then looked at these strains and deleted two more genes in these strains uh just to again get rid of the non-mideahondrial pathways for pe synthesis and then marie actually again labeled phospholipids in these cells and she was really happy when she found that in these cells now the synthesis of pe was impaired and in particular the accumulation of pc was impaired and this is i think again a quantification so you see here the pc accumulation was completely impaired in a micos dependent manner whereas she saw in fact relative when she plotted it relatively an accumulation of pc in these cells now this told us that these contact sites between the inner and outer membrane are somehow important for the efflux of pe and its conversion in pc in the endoplasmatic reticulo so the ultra structure of mitochondria seemed to be indeed important for psd1 to allow psd1 to synthesize to decarboxylate pe and ps2pe and convert it into pc so this is the model we are actually proposing on these and other experiments and that is that they are basic or the key of this model is that there are two independent pathways for the synthesis of pe in mitochondria there's one pathway that depends on ps specific lipid transfer proteins up is to mdm35 or slow mo2 tribe one in human mitochondria this allows psd1 in the inner membrane to convert ps into pe however there is a second route that does not depend on these lipid transfer proteins but it depends on micos complexes and on the structural arrangement of mitochondria and this allows the decarboxylation of ps to pe in the outer directly in the outer membrane and its conversion into pc now maybe if you give me five more minutes i would actually elude to another aspect of this works that we were actually prompted to look at when we said because this work then for the first time linked this contact site complexes to the mitochondrial phospholipid metabolism so apparently the arrangement of this membrane is very important to or the conformation of this contact sites is very important for allowing lipid synthesis to occur in these cells now we therefore thought maybe what is actually the yeah i should start differently so can we basically substantiate this role of micos in this phospholipid in more by a few more experiments and this is just one piece of evidence where we which is basic genetic evidence when you basically delete this non-mide pe synthesis parcel that's indicated here by this double delta in micos deficient cells then they basically show a strong synthetic growth defect which is completely consistent with the function of pe of micos in this pe synthesis pathway and we can resuppress this phenotype by simply providing ethanolamine as an oxotroph in the to these yeast cells another experiment that we did to substantiate this role of micos in the in this for this phospholipid trend is using artificial tether protein now this is a bit tricky to do but what we did here is we basically design an artificial protein that contained on inner membrane transmembrane region a spacer region of 12 amino acid and then a transmembrane region in the outer membrane in the gfp fusion protein and we could express it in these cells and we were very happy to see that the expression of this tether restored the growth or deficiency of micos of micos deficient cells that were basically grown at 37 degrees i should highlight that this tether it is looks very clear for this phenotype but there are many phenotypes associated with the delta micos in or with the loss of micos complexes in the cells and one is for instance that these cells lose respiratory activity and this is not suppressed at all by this tether so we think that this artificial tether is a partial partially suppresses functions of micos complexes but it seemed to suppress the functions that are linked to the pe metabolism this also tells us that it's not all about pe metabolism in these micos deficient cells which made the micos feel a bit more happy i think so okay so i think these experiments somehow substantiate this model that we would actually like to propose but they're also brought us made us more thinking which of the functions of micos are actually really associated with this pe metabolism and here we looked actually at the cells lacking not only a micos subunit but also the lipid transfer protein so in this case now we basically interfered with lipid transfer and did this in micos deficient cells and we were really surprised to see that in this case the loss of ups2 proteins the lipid transfer protein completely restored the respiratory growth of these cells and maybe more impressive when we look at the ultra structure of mitochondria we also have a dramatic suppressive effect and you may not see this so clearly in the wild type situation but when we look at cells that lack central micos subunits like humic 10 or humic 10 or humic 60 they show a completely aberrant crystal structure very typical and by simply by deleting this lipid transfer protein that is a lipid transfer protein for ps we could completely suppress this phenotype in these years cells so apparently this subtle changes of phospholipids in the mitochondria in a membrane can have very profound effects on the ultra structure of mitochondria this is actually the last bottom or the last conclusion i would actually like you to pick up and this is really this fact that the lipid composition of a membrane has a really profound effect on the ultra structure and the structural arrangement of mitochondrial membranes and this will be confirmed by lipidomics we do see only subtle changes in these micos deficient cells but we these but these can be implemented also by deletion of ups2 and therefore we think that the decrease of pe levels by deleting of ups2 in delta mico cells basically restores the ultra structure in these cells and this is actually also supported by some previous studies that we did where we looked at cardiolipin synthesis mutant doesn't really matter what it is or actually maybe it matters this it's a PG synthase which is a central step in the synthesis of cardiolipin also these mitochondria are completely distorted and show aberrant crystal morphology and also in this case by simply deleting in this case the lipid transfer protein for the precursor protein a precursor lipid phosphatidic acid we could completely suppress this phenotype and in this case we were could actually show that basically by deleting this lipid transfer protein we prevent the accumulation of PA and CDP DAC in the inner membrane and this are really is a very low percentile so we're talking about below one percent still it has this profound effect on on the lipid composition so this made us believe now that we have to think much more about lipid composition when we think about mitochondrial structure and ultra structure now with this actually I would just like to conclude with this slide which tries to wrap up all what I have told you today at least in the second part so I've introduced you to this new class of lipid transfer proteins UPS1 MDM35 is a lipid transfer protein for phosphatidic acid affects cardiolipin synthesis and in the second part UPS2 MDM35 is a lipid transfer protein for phosphatidic serine now there are two pathways leading to the formation of phosphatidic acid as an olamine in the inner membrane UPS2 MDM35 dependent pathway and amicose dependent pathway and this is central for the conversion of PE it's export of PE and it's conversion into PC in the endoplasmatic reticulum and at the end I've just highlighted that this lipid composition is a very important determinant for the ultra structure of mitochondria and subtle changes in the lipid composition can have profound effect so for instance in mecos deficient cells we see that a subtle decrease of the PE accumulation as we think in the mitochondrial inner membrane by deleting UPS2 completely restores the normal crystal or morphology in these mecos deficient cells now with this I'm at my end thank you but most importantly also thank co-workers involved in this and for the second part this was really the work of Mari Alton and Takashi Tatsuta along your co-worker in the lab they built on a lot of work of previous members on the lab that are basically now all over the place who basically Melanie and Matthias and Takashi were really instrumental in identifying UPS1 as lipid transfer proteins and Christoph actually started the whole endeavor in the lipid metabolism and the crystal structure we did together with Steven Matthews I should say he did the crystal structure and Jody provided us lab provided us with the delta meco strains and from Paul and Benedict Tetsami M for us thank you very much well you can only correlate it but it correlates with respiratory deficiency in this mecos cells so if you basically restore the normal crystal morphology also suppress respiration so it completely goes with the function but I wouldn't go so far that I really see it's called yeah but it definitely correlates with that so that's a clear correlation between that yeah it is interestingly in us and so we made different lengths and actually we see almost exclusively suppression with a 12 amino acid not with a 14 and not with a 10 so this is really striking I mean we only see that I don't know what I mean you can now make you know length estimation what what basically maybe what do you need to what is the size of the domain of PST1 very difficult but that's what we observe so there is definitely a length dependence it works best with a 12 amino acid long spacer yeah so under normal conditions I think there are different ways because there's a PE synthesized outside mitochondria and then it can be imported from the ER depending on the cell type you're looking at it's more or less under the conditions where we looked where we basically it's can only come from PST1 because in these cells it can only be generated from PST1 and you may have noticed that in this delta meco cell we see we see a complete block of PC synthesis whereas PE accumulates and we think this is because PE is transported by UPS2 into the inner membrane and accumulates there you cannot get out of it basically like kind of a sink of that so this the second one we haven't looked at but we surely will do the first one I can say a clear yes so there's clear requirement of defined phospholipids for fusion that's work of several labs michael frohman's lab showed that you need phosphatidic acid for fusion there's some data from Hirumi Suzuki's lab that you need also phosphatidic acid for fission so there is a clear requirement you need cardiolipin at many stages for the oligomerization of OPPA1 for instance MGM1 in yeast so these phospholipids play a crucial role for fusion and fission and this is what you would expect because of the biophysical properties and this is clearly the case but whether they activate the pathway I was talking about yesterday we haven't looked because in yeast it doesn't exist and this was mainly done in yeast in the membranes we are about to look at it we tried to look into that but didn't find them but I think yes because the phenotype is a bit different than the than the phenotype of the other protein there should be definitely other binding partners there are some yeast specific proteins there and we looked in yeast that I think are candidates for that but whether they are lipid transfer proteins I would no I don't know we don't know I mean again there are no sequence similarity with anything as this is an example it doesn't mean too much but yeah but it could well be that they are more binding partners actually I would expect so there have been proteomic studies where people looked at the distribution of in a membrane proteins in a on a sucrose gradient just where you have different densities of the different membranes and from these studies it looked from these biochemical studies the pst1 was localized close to the membrane contact sites that were later shown to be formed by or dependent on m-cost complexes in that sense I would expect that pst1 is close to contact sites which would also make sense if you think about that this is likely the site where the phospholipid transfer occurs but really hard data confirming that have not been done now in yeast and people including us also look now in mammalian cells where these proteins are really localized to to see yeah whether it's really there and whether this is basically a hub where you have the transfer process occurring and at the same time as these data suggest allowing the transactivity of this pst1 in the outer membrane but it is actually a prediction we would make based on the available data and according to this this data well I think this is a well this is a very good question and I mean there's a lot of discussion about it why has it involved that you have pst1 in the inner membrane why do you have this strange pathway and I mean one can speculate in one way or the other but the answer is simply we don't know whether this is really why this has evolved like this may have evolutionary simply evolutionary reasons but why this has been maintained as the only enzyme sitting there is still not clear let's put it like this really yeah but it's really the only enzyme there good okay i thought that's the thing Thomas for two great lectures