 human gut microbiome. Okay, well thanks Daniel and the organizers for inviting me here and I realize that I'm the man standing between you and lunch and it's getting late so I will try to be short and if I'm getting overtime please signal me and I will try to even be shorter. So I'm going to talk about this drug-bacteria interaction landscape of human gut microbiome. Just to give some background, I have background in modeling but today I'm not going to present much of the modeling work. Instead I'm going to try to play or try to be analogous to Tycopra who was a Danish astrologist who has been looking into planetary motions. He was one of the first ones who made very accurate maps of positions of the planets and their dynamics which eventually enabled basically the physical laws of motion that could be that could be eventually realized and I think the reason why my lab moved from mostly computational work to now actually mostly experimental work is because at one point I realized that we're hitting the wall when we try to understand microbial communities. We have reached the limit of how much we understand metabolism. In fact actually I don't have any longer illusion that the central metabolism is very well understood. I mean we hit so many points even with saccharomyces cerevisiae we don't realize what's going on. So we decided that we need more experimental systems, we need more data and this is part of the work that my lab at the moment is doing. So the question that I'm going to talk about among the various different when addressing at the moment is looking at the drug-pactra interactions in the human gut microbiome and this is really, really largely unknown. We can look into bacterial. People have been doing a lot of metagenomics studies making correlative analysis between for example how the obesity diabetes many different types of diseases are related to gut microbial composition, how diet can influence it, also some studies coming up how drugs can influence it, but these are mostly correlative studies and not causative. Of course there are some causative studies and sometimes I like to make a joke sometimes in private, sometimes in public, there's almost race now to relate everything to gut microbiome you know so eventually I will leave it on that. So the before I go to experimental results and since I may run out of the time I would like to first hand acknowledge the people that have been working on this people in my lab of course and the collaborator especially Nasus Tipas and Pierre Bork who are group leaders also at EMBL, Hadelberg and everything on the gut microbiome I'm going to talk about has been done in collaboration with them. So this is going to be my only slide that's based on published work and everything else that will come after would be unpublished work. This is just a motivation example so they have in two nature papers looking into the effect of type 2 diabetes on the gut microbial composition you find a very big signatures until it turns out that the signature that we have been observing is not the signature of related to anything to type 2 diabetes but it has because of the metformin which is the first line of the drug that people take as a treatment for type 2 diabetes and since most of the people take it there hasn't been enough statistical power to or people even didn't think about it so most of the changes in the microbiome that people associated with type 2 diabetes actually turned out that it was metformin right. This is just one example and there are only a couple of other examples that exist but how far this goes nobody knows so we decided that we should start some steps in direction so what I'm going to talk about is first of all this is going to be mostly all in vitro studies so we need to grow this gut bacteria so we need to grow them in reliable reproducible way and ideally in a fashion that in a defined media so we can actually go back in to see what these interactions are do molecular analysis and so on I will then talk about drug bug interactions and then bug drug interaction if the time permits I will also talk about bug bug interactions okay on media and the on the bacterial resources we have now approximately 100 representative gut bacteria that can grow in a defined medium I'm not going to the details but basically this bacteria have been chosen so as to represent there's a healthy gut abundance into the healthy gut we also added some known probiotics pathogens this is associated bugs also the bugs that have been known to associate with the drug in effects and so on we have been growing them into 15 defined out of this donor novel media that has not been described before and to four different rich media and basically on the phylogenetic side we get quite high coverage of the known diversity approximately 75 percent of the known abundance on the metabolic diversity or at least the metabolic diversity as we know of today we cover almost all of it okay so this is the let's say the heat smaps summarizing actually almost four years of work so what we see here is the different bacterial species slash trains over here different growth media rich media semi defined defined some even minimal media and basically the conclusion is that almost 85 percent of the species they grow in minimal or defined media and most of the species where at least actually two different defined media that we can grow and of course this has been actually in the beginning quite surprising for us and we have been doing a lot of follow-ups you know and then making sure that we don't have contamination we don't have media follow kind of a media contamination etc etc and this actually has been quite clean we have reproduced these results many many times now we also find just to give some highlights because i don't have time to go through all the results we find many novel musin degraders so actually we actually just double the number of species that have been known to colonize or to utilize musin and this has large implications for many diseases associated with the intestines because the colonization of the musin and weakening of the intestinal barrier is one of the first steps that gives to leads to many diseases and also immune responses etc quite interesting actually we don't not only see what are the metabolic requirements of different bacteria we see actually quite substantially inhibitory effects of amino acids and short chain fatty acids in particular for many different bacteria and this actually has been a very good lesson at least for us because when we think about growing difficult to grow bacteria we always think about what's missing and people tend to add and we also tended to add you know what's missing let add extra amino acids let's add this and this and this but it turns out that actually inhibitory effects can be actually quite so important so i think we should start not thinking not only thinking about what's missing but also watch too much and i think many times we don't get spaces growing because we just simply have too much of amino acids for example or short chain fatty acids these inhibitory effects can be simply too strong another message that we learned was the phylogenetic distance is not always a good predictor so we observed two different strains of the same bacteria that can have very very different metabolic properties they can prefer very different media on here you can see a genus level map actually this is clustering based on their growth preferences on different media and the color of different genuses if this orange genus for example just simply spread all over right so you can get actually this kind of differences at the level of species sometimes also at the the level of strains and simply because of this and also that this kind of resource has not been available before our ability to model this metabolism also has been limited so if you basically we took around 50 different or at least all available gut bacterial metabolic models we had in hand and try to see if they can just simply recapitulate this growth in defined media only eight percent of them could do this so it means that actually just from gene annotations we cannot accurately predict what the metabolic capabilities of different species are so I hope you can appreciate the value of such a resource and I think this is one of the first steps I think we need to do for many different community systems before we can go and model larger communities some others of course one of the things that reason we started doing such a resource because we want to move to to individual studies in a larger drug screens or even the community studies so another few interesting observation they had that there are some species actually they prefer defined media right so you give them everything they don't like it so much and there are some species that grow only in defined media and show very very poor growth in in the rich media right again is the question of the of the inhibitory effects and the bacteria that prefer defined media actually we find that they're at least significantly more prevalent right seeing that they have this kind of strategy that they can be in many different places they just require so they're basically living off into into niches where the resources are scarce what's also interesting is that relative growth in certain media correlate very well into relative abundances across different people this is the chart or here of course I'm not trying to say that this media are media are anyway by any means represent you of the good environment but this means that this media we can actually use for in vitro screenings at least with the growth relative growth rate differences we can to some extent or the effects that come because of that can be to some extent tackled and one of this media actually there are many defined media they do this and the MGM is one of the the rich media actually they also does it not shows good correlation okay so moving on from the resource to looking into the drug bug interaction landscape so what we have done is that we have screened 1200 different drugs basically these are from the press week library these are all ft approved drug compounds and we are screened them against the 40 different representative bugs again from the coming from the these are the subset of the hundred bugs that we have actually looked into and of course this study will continue and we will hopefully reach the whole 101 point so all in all we have tested 48,000 drug shrimp pairs and this has been done on all in biological triplicates and under anaerobic conditions basically what you do is that you throw different different drugs into different bacteria and then basically you record the growth curves okay again yeah many years of work summarizing to one slide what you see here I mean it's I don't have time to go into all the details just want like to present some highlights so the first thing actually that was quite I don't have proper word quite astonishing at least for me was that most of these who many of the human targeted drugs in this case almost 25 percent of them they're extensive impact on the gut bacteria so the inhibit gut bacteria of course you can see that antibiotics do the job as well but this is not surprising they're made to kill bacteria human drugs actually are made to target humans they have quite a large impact on gut bacteria and this is actually a conservative estimate because we are done our screen on one concentration of the for across all board which is 20 micro molar and if you look into estimates of the what's expected concentration of the different drugs inside our gut for example made from in the example I talked over here this is way above what we have tested right so if we for many of number of drugs actually we are now increased concentration and see what happens of course as you increase concentration you see starting effects of many many other drugs hitting gut bacteria uh another interesting observation is that the abundant commensals are more affected by the human targeted drugs that meaning that the effects that are going to have on the human microbial community and the effects thereafter side effects for example or the effects secondary effects on the other bacterial communities and eventually up to horse are going to be quite large and if you look into the drug side effects right so if you look into the drug side effect databases look into side effects that human drugs or horse targeted drugs have on the horse or in this case humans is quite interesting to see that if you have the drugs that show a hit in our screen meaning that these drugs are affecting commensal bacteria then they tend to show actually side effects that are very similar that you see with antibiotics right so this is kind of an let's say indirect in vivo relevance of our screen that we would actually expect that these drugs are affecting gut bacteria this will also show side effects that are similar to antibiotics which would also kill this bacteria which we also hope so now this resistance very interesting question is about need to rethink about antibiotic resistance and when we think about the antibiotic resistance of course we think about antibiotics but i think now we need to also start thinking about non-antibiotic drugs the human targeted drugs that can also lead to resistance and here is in just one of the few examples that we're following up so on this axis you see the number of antibacterial drugs with anti-common selectivity and then here you see the the number of human targeted drugs with anti-common selectivity here's the E coli wild type over here now if you make the the mutant of the tall C which is one of the the efflux pumps of the E coli so it's one of the common mechanism through which bacteria basically become resistance they just pump out the drugs they become resistance and now if you take make this E coli mutant it moves diagonally upright so it's not moving or like this over here so if it becomes resistant to antibiotics it's also becoming resistant to human targeted drugs right so this we are done also with not only E coli some of other bacteria over here and then we see that there are many times the mechanisms that through which the the back check can be resistant to host to antibiotics also apply to the to the mechanisms through which they can also be resistant to host targeted drugs which will mean that the wise first also will be true that if we are eating let's say a lot of certain drug that is affecting the commensal bacteria which we don't know about it can also lead to development of antibiotic resistance okay so moving on i would from moving from drug bug interaction i will move to drug interactions meaning that so far we have looked into how drugs affect the growth of bacteria what we're also doing of course on the smaller scale because of the analytical limitations looking into how bacteria affect the drugs itself right how they're for example metabolizing drugs so we are looking into 19 drugs that have been selected on based on different criteria of course the main criteria has been that we can analytically measure them easily into lab this has to be the drugs that are orally taken and sufficiently sufficient quantity so that there is some chance that they will end up into our gut i just mentioned that it's not necessary that you need to eat the drug so that it ends up into the gut because of the the hepatocirculatory system because of the circulatory system a lot of the drugs even if you inject them would end up into the gut as well okay and so we are in total around 19 drugs that cover many different classes i mean we have not put any chemistry filter on it right basically selected try to diversify whenever we can so i have many different chemistries many different therapy targets etc okay the screen again is conceptually very simple you grow the bacteria with the drug you measure the drug concentration in the beginning and you see at the end whether the drug concentration has decreased or not so one of the main conclusion again it was quite surprising at least for me that many of the drugs are affected or affected by bacteria use the word affected at the moment and it will clarify in a few slides why i don't say they're metabolized by the bacteria but many of the drugs are affected by bacteria i don't actually now recall the numbers i think 12 out of the 19 drugs actually we look into their concentration is reduced in the medium spectrum this also has been quite surprising for me because if you look into literature the number of known interactions that have been described are quite few and ours is a quite actually relatively small screen so just to extrapolate which i should not do but if i would increase the screen i would expect similar heat rate across the board if you look at the types of bacteria that are doing it these are mean we again have been screening only with the 20 bacteria so far but they have been selected for their diversity in terms of types of functions that they play in the gut ecology or their phylogenetic diversity we see all types of bacteria all the from commensal to potential pathogens to probiotic bacteria they are interacting with these drugs in this fashion so the reason i've been talking about affection and not metabolism so far is because at one point we realize that this what we are seeing partially is not only biotransformation which is going to be in zymatic modification or less a chemical modification of the drug but also to some extent actually bioaccumulation so what we have done basically is that now we do two different type of extraction protocols one is let's say looking only into supernatant other is the hard extraction where you break the cells and again to a very hard solvent extraction to get everything of the drug that we can and this way we can now distinguish between two different mechanisms one is the biotransformation or zeno metabolism other is the bioaccumulation and this is the summary of the results over here so what we are looking at is the different bacteria that we found some of at least showing one hit the depletion of the drugs that were here and of course in the bonus readout that we have is also how the drugs affect the bacterial growth as well in the same screen everything in the gray scale this gray scale thing you see these are our positive controls these are the drugs that have been known to be metabolized by the gut bacteria from the before you can actually easily see now they were why they were known for even decades because every bacteria you throw on them they do the same biotransformation on them so it has been one of the easy things to catch up everything that you see either black and green is some a novel finding that has been not been described before the black are the biotransformation hits and the green are the bioaccumulation hits bioaccumulation meaning that these drugs are being accumulated by the bacteria they're taken up but not metabolized surprisingly we find very little overlap on the two sides meaning that the bacteria depleting the drugs are either bioaccumulating or biotransforming actually it's very rare we found I think only four cases for interaction cases where the growth of bacteria is affected so in the most cases the growth of bacteria at least under our acid condition is completely unchanged since the bioaccumulation turned out to be so prevalent in our screen and at least actually quite an underappreciated mechanism of how drugs can be influenced by how bacteria can influence the drugs we did some mechanistic follow-up for some drugs and one of them is geloxetine which is an antidepressant unfortunately a lot of people take it in quite large quantities if you look at the over time period of the year or the time period of that people take this drug so what we have done managed to do is to put a click chemistry tag on the drug over here meaning that we can now pull it down with the biotin or steptavidin over here and here is the pull down as you can actually get quite clean pull down if you do mass spec then you can look at the enrichment of the which proteins are basically binding shrugs so what we also done I won't show the direct data here is the metabolomics also treat of the of the bacteria after treating them with with the geloxetine here is the summary map so every line here in this cake style map you see is the enzyme that we pulled down and the circles of the metabolites that actually are differentially expressed in the bacteria after treating with geloxetine and what I hope you can appreciate is that they actually lay mostly in the proximity with each other meaning that a lot of central carbometabolic enzymes actually binding geloxetine are leading to metabolic perturbations inside the cell that brings me to kind of conclusions of at least this part of the talk so basically what I would like to convey the message that we see really profound implications for for the drug the drug discovery and the usage of our findings for the first part we have very good resource for representative gut bacteria and growth media for in vitro studies we see a lot of collateral damage a substantial fraction of the human targeted drugs impact commensurate bacteria this of course has a reliance for drug side effects emergence of antibiotic resistance drug repurposing and also inciting to drug mode of action for example especially for antipsychotic antidepressant and it so many of these antipsychotics antidepressant which also hit in our screen on the both bug drug and drug bug side it actually doesn't if when people start taking it doesn't work immediately sometimes it takes weeks and sometimes in months before you see the mode of action and it's suspected the mode of action might be through the gut and which is has recently been actually shown for metformin case so the metformin if you inject it doesn't work you need to take it orally and the gut microbiota plays an important role in the mode of action for this drug okay and also the bioaccumulation seems to be a very widespread yet underappreciative phenomena and of course it has a direct relevance for the for the drug efficacious in pharmacokinetics and overall of course one can imagine that you should be quite relevant for drug design in general so for personalized medicine and if either I could stop here or I could have five minutes more so to the oh I have 15 more minutes okay I hope people are not too hungry then by then but okay I will try to be okay I will try to feed you some drugs so moving from the single species to community right so what I've shown you so far this is all single species work looking into you know one drug one bacteria interaction of course these effects can be quite complicated as soon as you can get emergent effects higher order effects as soon as you move from single bacteria to communities but it's important to keep in mind that if you don't understand how single bacterial metabolism works or how a drug affects single bacteria there is no way we are going to understand if the higher order level so we need to map these interactions just to give them kind of a flavor for what we might expect to see so even if you take a simple community with five bacteria that's actually the follow-up from our the drug bug screen again the duloxetting case the community effects can be very counter intuitive so this is the case where erectile which is the the rate bar over here so this is a five bacteria make sure you grow them as a community in a single shot either or actually you can also get them into stable formation you grow them without duloxetting or with duloxetting so this red guy over here is actually if you grow them alone with duloxetting it's very sensitive to duloxetting so duloxetting basically inhibits this growth very very strongly for this bacteria yet in a community context this bacterium actually is rescued by duloxetting and this basically happens because the same mechanism I showed you before so duloxetting the bacteria that bioaccumulate duloxetting their metabolism is shifted they start secreting nucleotides and many other metabolites which are actually beneficial for this year actually over here and that benefits that comes from metabolic cross-feeding is it basically masks the duloxetting effect so this means that the community effects actually can be counter intuitive and you can actually understand them based on the the single bacteria study we have done before this is of course a small community and not representative of the gut environment so one of the other resources that we're building now like we have the bacterial resource for the single bacteria yeah just a question in this case I don't know the top of my head we have measured them but total biomass I don't expect to change because the duloxetting is not reducing growth so much for the other bacteria mostly for for the year actually yeah and this is not the most dominant one again I know what I'm showing is the relative abundance so this is not the total biomass and also measuring total biomass is not so easy with gut bacteria many of them you know the flocculate cluster together so odys I mean is a barrier of approximation I think what we need to do either look into the total CO2 produce or basically just measure the the dry cell weight so like we have the single bacteria resource we also want to build a community resource that we can use studies like I've just shown in the slide before the first so this actually I forgot the name Alvaro Sanchez that did a very good job in introducing type of experiments we are doing basically start with the community you basically get them into particular media and then you basically keep the transplant every 48 hours exactly some experiment the difference being this is of course the gut bacterial community and run under an Arabic condition and this is the initial mixture is known right so we put this bacteria together so we know exactly where we start we know exactly who is there and so we can really reliably follow so we don't have the problems with the phylogenetic assignment over here and we know how each of these bacteria grows in this in this come in this particular media very well actually it was quite interesting to see that we get a very rapid assembly so this is basically the number of transfers each transfer is up to 48 hours you get exactly the same same result if you do transfer after 24 hours we do it 48 hours because it's well let's say experimentally more convenient to do so basically get establishment a very diverse subcommittee in both media as you can appreciate there is no competitive exclusion meaning that there's no one winner of the system you get actually mixture of the bacteria so we have indicating that there is either a strong niche partitioning and or there is also metabolic or other kind of niche creation going on in the system we have of course done this now on a much larger scale so we are starting with three different bacterial mixtures so they are either we exclude for example probiotics or pathogens or mix all of them together we are 15 different growth media defined growth media which are exactly the same that I showed you in the resource slide earlier and two different pH just rest to test how the pH would affect the system so 7 and 5.5 so 5.5 is closer to what you would expect into right after the after the the stuff comes out from the from the stomach we do nine transfer we know that four three transfer also sufficient just do nine transfer just to be sure and we do the barcode sequencing at the end so we get a range of defined communities as snapshot of it at very very different diversity you get as few as four or five bacteria you get up to 20 bacteria into the system it's just a pca plot I mean I don't want you to read through this I mean doesn't mean anything unless you look closely into it but this is basically just means that these are assemblies are media dependent different media different assemblies as one would expect what's interesting here is that you get lots of winners and losers so I mentioned that we also characterize in depth all these bacteria and the same growth media before what you can now do is that to comparisons like this for particular media look into monoculture growth rank and then we see the abundance sign into community how they do right so you can and then the winners are the one basically which grow very poorly in the same media when they're growing alone but in the community context they grow very very well right and the losers are the one for example equal I often turn the south to be the loser when you grow equal I alone is one of the best cross and we just you know buy for the easiest to grow and the best crossing across almost all the media that we look into it basically quite strongly loses out and even the many different media conditions so you can now if you make statistics of who is winning often for example like to Berserker say our cluster in Ramoza are winning under many of different conditions I got a lenta or this guys over here actually are losing out in many different conditions you look into so you can start now looking into who is winning from being in the community who is losing out into the community how to understand this so I don't turn again time to go through the details but what we are doing now is again the condition media type of experiments that you grow spaces one then you add take the condition media and the grow other spaces you get very interesting results just a snapshot these are different spaces these are commensals this is an opportunistic path version and this media it doesn't grow alone at all very barely detectable any growth use a condition media of a commensal bacteria and then it starts growing over here so the goal basically is going to be the reconstruct the metabolic interaction map the human gut bacteria this is computer simulation purely this is for the fun of it because we can do it we can simulate almost a hundred bacteria together take the genome scale models and and simulate every possible metabolic interaction that can grow between them basically you keep your supercomputer busy for as long as your institute would allow it you can start mapping into possible interaction landscape like this and the goal basically is going to be see how many of this turns out to be true in our experimental setup and with this actually this is the things my all everything about gut microbiota I realize that a lot of I mean I came only today which actually is a pity but had many family constraints so I could not attend the talks before but I realized that the niche generation and the cooperation mutualism seems to be quite strong interest in this in this audience so either I can stop here or I have few slides on the niche generation that are mapped through metabolic analysis so I think stop or continue continue okay so okay technically I still have 10 minutes so so this is the work entirely or turned by Olga Pomorova who has a PhD student with me and is being continued by Natalia her postdoc in the lab so we're interested into metabolic cross-fitting especially the question of niche generation and mutualism I mean I don't need to convince you these are interesting questions to look into and the reason so this whole project started with modeling right so we started doing modeling you know we actually see quite interesting patterns we started when we started measuring metabolites and even if I look into literature so there are lots of lots of implications of metabolic cross-fitting very few examples of metabolites have actually been measured and shown going from one space to other spaces these are hard experiments to do I mean you can look at the DNA RNA and you can say okay this DNA came from this space is this RNA came from this space is fine you cannot do that with metabolite the pyruvate when it leaves the cell bacterial cell fungal cell once into the medium this could belong to anyone right there is no way unless you do clever labeling experiments which are also very difficult to do in a community context you can't do that so there's a reason why it's so difficult to do and the Olga has been actually quite interested in in getting at least one very good well mapped community and what she has been looking into is metabolic dependencies between yeast and lactic acid bacteria why yeast and lactic acid bacteria these are one of the let's say microbial communities of actually they co-occur in many many different environments especially with food associated environments if you look into wine fermentations kimchi sourdough kefir cocoa fermentations bread fermentations everywhere almost everywhere you would naturally find saccharomyces cerevisia some other yeast almost always you'll also find lactic acid bacteria so all guys been puzzled by this there must be something going on so again this is summary of a lot of different work but basically we managed to make models model predicted that there must be some amino acid exchange between this community and we have done together with uesauer and marcos ralser uesauer from itia Zurich and marcos ralser from Cambridge a lot of extensive metabolomics experiment the idea have been again doing a condition media experiment just to give you a flavor so we have a media or actually many different media where lactic acid bacteria do not grow alone right so put them into won't they grow only in presence of each you can also do in condition media grow each remove each add lactic acid bacteria then they grow suggesting there is a flow of metabolites or some diffusible factor from each to lactic acid bacteria and if you do take now basically the samples during the each culture and then lactic acid bacteria culture which is after you remove the yeast basically what we are hunting for in the mass spec is the the ions that show this bell shaped profiles that this would imply that basically the metabolites that are secreted by yeast and taken up by lactic acid bacteria these are the results of untargeted mass spec profiling you see very very interesting profiles but the red ones actually the the ones that we're looking for there are lots of molecules and many of them actually assigned to the amino acid that we predict by the model are basically exchange between yeast and lactic acid bacteria and just to to drag your attention that these are not only so we have done come complementation assays and so on so forth to basically prove we also done quantitative metabolomics and supplementation assay in the same media and we basically can reliably show that this interaction are indeed taking place hey just again we are done also mutant screen and here's some example of the the four different mutants that we find interesting these two mutants are that each mutant that increase the growth of lactic acid bacteria two mutants that decrease the growth of lactic acid bacteria or at least one of the two bacteria we taste and here is the quantity amino acid secretion profiles and you can see this perfectly mirrors the growth of lactic acid bacteria meaning that you can also genetically manipulate your community and show show what's happening over here all these mutants are related to nitrogen catabolite repression which I will come in a second and just to convince ourselves that this is an active secretion not cell death basically this was a month of sleepless night sport for organi because we got troubled by the the thought that there is no active secretion yeast cells are simply lysing if you take different mutants there lies more and then basically there is no active secretion but simply not cooperation but you know cell death now we have done many different experiments that show that this is not the case many lines of evidence one is that if you would now make condition media of the different growth growth along the growth curve of the yeast they all show proportionate secretion and there's not much cell death going on over here you can actually also take this different mutant I showed before that showed different degree of the support to lactic acid bacteria we have done the live dead staining and basically first of all if you just calculate the number of days we observe they are by I mean they are far far away from explaining this lactic acid bacteria support that they agree and plus that if you try to make correlations between how much support different mutant use and the death rate there is simply no correlation this is indeed an active secretion and not cell death yes because I will come to it I mean again I don't have the data here because we also done gene expression analysis and the German gap one there's general amino acid permeates one expression is proportional to the amino acid that you secret into the medium so this is really dependent on it and you can see in the mutant that are not secreting much gap one expression is also going going down as I said I mean I don't have to go through all the experiments we have done a lot of also looking into mutants gene expression metabolomic etc but this is a summary side what we think is happening what basically we are seeing is a niche generation in action right so metabolic decisions of the each that are enabling lactic acid bacteria survival and I use the word decisions because they said this in each to actually make so here is the the summary model from the work so you have extracellular nitrogen sources amino acids ammonium etc so the the uptake mechanisms are you are the same for different I mean uptake mechanisms are not specific to amino acid they are general permeates the nitrogen basically comes in now if at the same time if you start making synthesizing nitrogen inside or the amino acid inside that can happen either because you're mutant or you're the torsive torsive signaling pathways I've activated for example adding rapamycin and forgot to mention that if you would add even slightest amount of rapamycin to your media the growth of lactic acid bacteria boosted many many folds right in the in the wild rubbish culture so you can manipulate by rapamycin then basically what you get is a nitrogen overflow because you have excess nitrogen coming in first of all east has all it needs but it starts to think either you have bad nitrogen source or some kind of other signaling if it's things that it's under nitrogen starvation it starts also synthesizing amino acid inside and this basically leads to a nitrogen overflow phenomena and part of that can also be toxic because you can't simply just accumulate nitrogen inside and this basically leads to amino acid secretion into the medium and until for a while actually people have been thinking the saccharomycesis is a bad amino acid secretor it's not bad amino secretor you just need to get under right conditions its secret is actually fairly good amount of amount of amino acids and basically this creates a niche for lactic acid bacteria now this interaction you can manipulate either by making mutant for example that these two mutant I showed or add rapamycin you can get a lot more lactic acid bacteria or you can make another mutant or till it change decrease it or you can reduce the nitrogen sources into the medium and this is actually quite proportional so if you would look at the how much amino acids are secreted at different nitrogen load you can basically make lines linear curve to it just a few thoughts on this and again another evidence that this is active decision of the east because if you look at the which amino acids are being secreted so these are the red ones over here amino acids that are secreted and these are the biosynthetic costs of the of different amino acids only the cheap amino acids are secreted right so each is not secreting something that costs a lot to make it secretes the cheap amino acids over here is this relevant into ecological context we are looked into different east isolates that come either from the grapes or from for example from the kefir and then you see whether they do the similar growth promotion for lactic acid bacteria in that they do this is our let's elaborate is s90 lactic acid bacteria also east and this is the growth of lactic acid bacteria so this has in the scale we have been looking up to now we can see for example candida californica kefir isolate is lot lot more growth support so they're actually secreting the same effect of the the growth support you also see in grape juice experiments so it doesn't need to be in the laboratory medium that we have designed and we also done metabolomics and the mutant analysis in the grape just showing that this is again the same mechanism through which you see this impact on the on the grape juice and those who know about wine fermentations or the beer fermentations lactic acid bacteria actually quite notorious to contaminate this fermentation sometimes we want to have them for mallactic fermentation but they can also contaminate quite heavily and once you get them actually it's quite hard quite difficult to get rid of them and you can see the reason for that each actually are creating niche for lactic acid bacteria over here and of course it has relevance for many other communities some of them were experimentally following up like kefir for example but I won't talk about that and that brings me to my last slide as a teaser now we have been looking into one way interaction right and this is of course is much harder to think ecologically why would each secret things and we actually quite convincingly shown that each is secreting things because it's profitable for each right it doesn't need a nitrogen it has plenty of nitrogen accumulating more nitrogen is toxic since it secretes it other bacteria can take up into it now how that situation actually can be precursor for mutualism or cooperation depending on how we use the terminology and in a very very simple experiment you can see this happens is simply replace glucose by lactose and then you choose lactic acid bacteria strain that is lactose positive and this is what we exactly get so these are the lactic acid bacteria isolated from kefir which is a milk drink and this is the wild type is so these are all each at the same od and the dilution is on this axis you can see now the each grows only closer to lactic acid bacteria and vice versa right and you see exactly the same into the liquid medium here's the monoculture versus co-culture monoculture you don't see almost no growth co-culture you see both lactic acid bacteria each grow this is the glucose control if you do the second mix when glucose each doesn't care about lactic acid bacteria each does its stuff like it lactic acid bacteria grow closer to the east as we saw in the previous experiment if you look into again we are done metabolomics you just show the the lactose consumption lactose is consumed only even to the co-culture okay and with that I would stop here and again once I can acknowledge I mean to all the people who have been into the lab who have been worked into this and the the corp and the collaborator labs as well and thank you for your attention