 First day of Oregon, he's going to show us some really cool eye popping, eye popping candies with substance. Okay, is this on and decent? I see a thumbs up in the back, so I'm going to go ahead. Great. So yeah, I'd like to thank the organizers for the invitation and also for organizing this really fantastic workshop. It's been a lot of fun, both in terms of the variety of topics, but also the really fascinating people that you've managed to assemble here, so thanks a lot. So I'm going to tell you about some things my lab has been doing to explore gut microbial communities by directly looking at them. And I'm going to start off with something that everybody in this room, I think, already knows, which is that each of us is an ecosystem composed of one human and an awful lot of microbes. So we have more microbial cells in our body than human cells. And these microbes, mostly resident in the gut, play an enormous role in processes associated with things like digestion, immune response, and even early development of organs and other sorts of systems. We also know that aberrations in microbial communities, so differences in microbial community composition, are correlated with a large and growing list of diseases and disorders. We heard just yesterday about Crohn's disease. In addition, asthma, colitis, inflammatory bowel disease, asthma, Parkinson's disease, all of these are associated with differences in gut microbial community. It's not clear in most of those whether the differences are correlative or whether the communities actually cause disease, but in certain cases, links seem actually to be clear. So we've learned nearly everything we know about these topics through sequencing-based techniques, so taking typically fecal samples and sequencing DNA or RNA to get lists of what species are present or what genes are present and being expressed. And this is immensely powerful. As mentioned, it has opened up this horizon and frontier of looking at host-associated microbiotics. It's immensely powerful, but it has some severe limitations. So as several people have noted already, we have almost no information about the spatial structure and temporal dynamics of these host-associated microbial communities. And we know, just thinking about ecosystems in general, that spatial structure is important. That's come up, in fact, several times at this workshop. So even though this has come up several times before, I think it's worth dwelling on for just a moment anyway. So if we imagine, for a moment, that we were thinking about a tropical forest, and imagine that you knew that that forest contained trees and elephants and monkeys and leopards, but imagine that you had no idea at all that trees couldn't run around, that monkeys could climb trees, the leopards could run after monkeys, and the elephants couldn't climb trees. You wouldn't be able to construct at all a good picture of how that ecosystem functions and what it meant for a forest to be a forest. That's the situation we were in with respect to the gut microbiota. So quite basic questions, like what is the spatial organization of gut communities? Are there spatial niches that help diversity exist? Are unknown? What are timescales for fluctuations, both intrinsic fluctuations and responses to perturbations, like, for example, antibiotics? How do colonies nucleate and grow and establish themselves? How do species compete and interact, and how are space and time relevant for that? These are largely open. So these are, I'm not gonna answer all these questions, it would be wonderful if I could, but I'll point out an approach that we've been taking that I think is an interesting one to tackle these sorts of issues, which is to adopt a model system that allows controlled and controllable experiments on gut microbial communities, and to look at it. And by look, I mean, in a very literal sense, look at what the microbial communities ended up doing through imaging. So as a model system, we're going to make use of zebrafish, specifically larval zebrafish, for a bunch of well-known reasons. So at larval stages, they're quite transparent. So here's a larval zebrafish here. Here we put a red dye into the gut. The gut isn't normally red, just so you can see where the intestine is. They're transparent, they're genetically quite tractable. They're vertebrates like humans, so they have a lot of physiological similarities with us. In addition, they, like other vertebrates, have large and diverse gut microbial communities. And best of all, it's possible to raise them initially germ-free, so devoid of any gut microbes. And then one can introduce into the water that they're living in, particular microbial communities. And the things I'll first show you, these communities are composed of normal native commensal gut bacteria, often that have been engineered to express fluorescent proteins. So what we can do then is construct few species, model gut microbiota in this model organism, and learn what kind of relationships can occur, what mechanisms can take place inside a host. I should point out that most of this stuff has been done in collaboration with a colleague of mine, Karen Gillerman, also at the University of Oregon, who's pioneered a lot of these techniques for generating notobiotics zebrafish. So there's our model system, the zebrafish, and now we want to look at the gut microbes in it. Now this is rather challenging in itself, because there's a lot of things one has to be able to do to get to this to work. One needs quite large fields of use. So the gut is hundreds of microns long by a couple hundred microns wide, a couple hundred microns high. We want to image that entire volume with single bacterium resolution. We want to do that imaging in 3D because the structures are three-dimensional, we expect, and we need to do that three-dimensional imaging quite quickly, because every minute or so we get this happening. This is the peristaltic motion of the gut, which we actually already heard a beautiful talk on from Jonas Kramer. So every minute or so we have these peristaltic contractions, so we need to be able to take a 3D image at timescales faster than roughly once per minute, otherwise our images would be horrifically blurred. And then also, because we want to look at the temporal dynamics of this, we want to do this imaging for many, many hours, and so we need a technique that has low photodamage or phototoxicity, okay? So those are the constraints. So how can we do that? So normally when one thinks about three-dimensional imaging or three-dimensional fluorescence imaging with like labeled things, you can think of confocal microscopy. That's the first thing that often comes to mind. But this actually won't work. So confocal microscopy, as probably most of you know, you illuminate your sample and excite all the fluorescent probes in this sort of green cone here. And then you stick a pinhole in conjugate to your focal point that blocks all of the light that isn't coming from your focal point A. So in other words, everything that you're actually detecting is from that one point, and you can construct a three-dimensional image by moving that point all through three-dimensional space. And that works great, it's wonderful, but it has a couple of inherent limitations. That it's slow, to get a three-dimensional image, you have to scan this point in all three dimensions. There's tricks you can play with multiple pinholes and stuff, but essentially you need to scan in three dimensions. And it's inefficient in its use of light. We're getting information about this one point, but we're shining light on this entire cone. So it's inefficient and causes lots of photobleaching and phototoxicity. So about a decade ago, or actually a bit less, a new approach, or depending on how you look at a new or very old approach to three-dimensional imaging came about known as light sheet fluorescence microscopy. And here the idea is that we're going to illuminate our sample, so we're gonna excite our fluorescent probes with a thin sheet of light. And then using just regular wide field detection, we're going to detect whatever is emitted at that plane. So at any instant we're constructing a two-dimensional image from that slice. So now to construct a three-dimensional image, all we need to do is move in the one remaining dimension, so it's fast, and everything we're exciting, this plane here, we're also imaging, so it's very efficient in its use of light. So this is fast three-dimensional imaging with low phototoxicity. So schematically what the setup looks like is something like this. We have a scan laser beam, rapidly scan back and forth to create a sheet. That gets magnified down by this lens and intersects our sample here, just indicated by this golden orb sample right here. That sample is a larval zebrafish, and it's held in a weak agar gel that's sticking out of the end of this glass capillary here. And then with this second perpendicular lens, we're detecting the emission from that excited plane. What it actually looks like is this. So here's our set, we built one of these, actually now we've built a couple, but this was our workhorse one. We built that several years ago, and there's a little zebrafish at the end of each of these glass capillaries, and here's those two lenses that was mentioned. So notably we can look at multiple specimens, one after the other, which ends up being very important to get any sort of statistics on what's going on. Okay, so we have then a model system and a way to image it, so now what does it look like? So what I'm gonna show you now is a scan. So each frame of this movie is one slice. We're gonna scan through in depth. And this is a fish that was initially germ-free that was inoculated, or the water around it was inoculated with one particular commensal species. It's an aeromonas species, and we're imaging it about 24 hours after that inoculation. So as we scan through, we can see, and those bacteria are expressing a fluorescent protein. So we scan through, we can see all these bacteria. We see a dense cluster of bacteria there. We see the outlines of the gut, nicely kind of outlined due to the fluorescence, autofluorescence of the mucus in the gut. Can you through like that? So just for kicks, we can make a sort of 3D rendering of that. And again, you see individual microbes as well as these dense clusters that this species forms. So already, like with the very first images we took of this, you can tell this isn't a homogenous, well-mixed flask. There's structure in here. There's individuals, there's clusters, there's clusters in particular locations, all of that. So there's structure in this gut microbiota. Now, as I mentioned, we can look at this over time. So in this movie, we're again looking at this same species, but now each frame is a projection from a full three-dimensional data set. So I've squashed the full 3D image down, and I'm gonna show you that over time over about 14 hours. And here we started very soon after that inoculation. So at the very beginning there's something like eight bacteria in the gut. And then we watch, and those eight divide and divide and divide giving us tens of thousands or hundreds of thousands of bacteria. So my colleague likes to refer to this as the Garden of Eden movie. This is the initial population of this world. So because at each of those time points we have this full three-dimensional data set, we can actually identify the bacteria in it and quantify those populations over time. This also is challenging because we have a very strong background from that mucus in the gut. And just to make our lives difficult, that background is very variable both across specimens and over time. It's highly variable. There's also autofluorescent host cells, which are probably the secretory cells making the mucus the complicated things. And the data sets are gigantic, being hundreds of gigabytes per fish. So we spend actually a lot of effort writing computational tools to do, to recognize both individual bacteria and large clusters and figure out like where are the bacteria, how many are there and so on. I'm not gonna go into how we do that. You can ask me about that later if you want. The upshot though is that we can quantify number of bacteria and do simple things for example, like just plot the total number over time. Those are the points here. And get those numbers quite precisely over many orders of magnitude. Precisely enough that we can very nicely fit particular growth models like here a logistic growth curve and determine things like growth rates for the actual in situ gut microbial population. As an added bonus, because we have that full three dimensional structure, we can already ask questions like, do those two things that we just saw, individual bacteria and clusters have different growth kinetics? And it turns out that they do, these clusters that we, so if we separately plot the population for clusters and population for individuals, this is just within one fish, we very robustly find a growth rate that's about twice as high for the cluster populations than individuals. And there's more to that story, but it's an old one so I'm not really going into it. But we can look at, we can determine spatial structure and look at how that correlates with things like growth kinetics. Okay, all of that is actually just with one species and even within one species that are interesting things like that to look at. But especially here what we really want to think about is things that are more than one species. So I'm going to tell you a little bit about two species interactions. So kind of reiterating some of the points at the beginning, we don't have a great idea of what determines the assembly, the stability and the fluctuations of actual gut microbial communities, for the reasons I illustrated earlier. And I thought I would elaborate on that by pointing out some quotations from a illustrious native son of Trieste on this slide. So one can ask, for example, can we think of microbial communities as a quote, supervenzymes, okay? In other words, could I predict the outcomes of microbial competitions by ignoring the actual gut, but rather by thinking about what sort of interactions would I get in a flask? Or what sort of interactions would I get if I just knew the metabolic networks of these communities? So can I think of microbial communities as a supervenzymes? Can networks of metabolic interactions between microbes be understood and predicted based on networks within? Can we engineer environments to modify microbial consortium? So these are really fascinating questions. And I looked all over town for a statue, but I didn't, so these are quotes from Daniel over here. I colored this a bit copperish to look kind of statue-like and that was the best I could do. Anyway, so a lot of open questions about how we can try to understand competitions or interactions within the gut. So I'm gonna tell you about two species, and I thought it was also worth elaborating on that a little bit. So you might be thinking to yourself, especially after yesterday, just two species, what am I, some kind of Cartesian fanatic who's very fond of reductionism? And the answer is partially yes, that I am one of those who does think that a bottom-up perspective of trying to build up complex communities from few member constituents can actually be powerful. But there's another point that I thought was worth making, which is that by looking at few species systems that we can directly observe and quantify and make sense of, what I'd like to convince you by the end of this is that that will help us figure out what processes might be relevant in shaping communities. In other words, that's kind of an abstract statement, but in other words, not just saying, what's an effective numerical parameter that we can put into a lack of a Volterra model or something like that, but what physically does that interaction parameter mean? What are the physical processes that might govern interactions? And I think those are easier to figure out in model systems that one can actually control. You can see if you believe that by the end. Okay, so what I'm gonna do is tell you a little bit about a story about two commensal species. I'll be a little bit quick because we published that last year and you can read about it more if you want. And then what I really wanna do is tell you a couple of new things that are unpublished, one about engineering invasion of a commensal species and a very short bit about perturbing one species to end up looking actually like another, and I'll comment a little bit on the future. Okay, so first let's look at a two species model system where we're looking at two commensal species. One is an Aramona species, I'll color code that in pink and the other is a Vibrio species, I'll color code that in blue. They're both native gut commensals, they're isolated from zebrafish and they're actually very abundant in conventional zebrafish guts. They're both very good colonizers, so if you just mono-associate, if you just expose germ-free zebrafish to just one of these species, each one will colonize to high levels, like 10 to the fourth or 10 to the fifth per gut. And they're both like all of these aerobic bacteria. Also, if you grow these in like rich liquid media, like an LB broth, they'll both happily coexist with one another. So what happens, yeah, so let's look at that in the fish. So first what I'm going to show you is the result of a very standard sort of dissection and plating experiment. So here what we've done is exposed the fish to the pink bacteria armonis in the water population, sorry, in the water at day four, then at day six we dissect out the guts, there's a gut being dissected, in this case by me actually, we've played out the contents and count bacteria, each dot is from a different fish and there's, like I said, about 10 to the fourth bacteria per fish. Just a good colonizer. Here in contrast is what happens when again you add armonis at day four, but now at day five you add the blue species, this Vibrio species to the water and then wait 24 hours in play. What you find is lots of the Vibrio species and an annihilation of that armonis population. It's about two orders of magnitude lower in abundance than it is on its own and it's also highly variable. So challenged by this Vibrio population we have this small and variable armonis population. So the question then is how can we interpret this? Now again the standard, I don't mean this as a criticism, but the standard thing to do would be actually just to stop there and say okay they're interacting, we can write down the ratio between them and characterize that by some numerical parameter, but let's go beyond that and try to figure out what exactly is going on here. So what do these look like in the gut? So this movie again takes place over 12 hours or so and what we're gonna see, thought I could turn the lights off. Oh, hold on, I'm gonna try that. Okay, I think that's the exact opposite of what I should be doing, but whatever. Anyway, so we have the blue and pink species, these are both in the same gut. I'll play this, there we go. I'll play that one more time. So what you see, I guess I'll just keep looping. We have that nice big pink population and it's there and it's there and it's there and all of a sudden it gets expelled from the gut. Meanwhile, the blue population just keeps steadily growing and growing and growing. Here's another example and then I'll turn the lights back on. So we have these dramatic collapses. Here's another example. Here I've split blue and pink into put them above and below each other but they're both again from the same fish. Here we're gonna lose a big chunk of the Armonis population but what's left over is happily there and it's in fact growing. I'll show you that in the next slide. And then in just a minute, yeah, we're gonna lose that one as well. What we find is these very dramatic population collapses of that pink population, the Armonis population. So quantifying this and this, we have 15 examples or something but a very stable drop by two orders of magnitude and then regrowth in fact but not quite to the original level. Here we have one collapse, regrowth, another collapse occurring like that. So what we see is not that the growth rate of the Armonis population. So what we see is not a change in the growth rate of Armonis but rather these dramatic collapses are what drives its population dynamics. So what's going on with this? So to look further, we thought more about what these two species are doing and in fact went back to just looking at each of them on their own. So here what we're looking at is the blue species, the Vibrio species, the one that wins in this invasion and we're just looking in real time at just one plane within the gut and this I think is actually my favorite movie of all. We see these very individualistic, very planktonic, very fast Vibrio just zipping around in the gut. This is the picture you should perhaps have of your own guts, just this dense swarm of bacteria zooming around. Now these Vibrio are kind of like a swarm of bees. You could change the kind of container that they're kind of like a liquid. You can squeeze on the bag that it's in but they'll always just re-adapt that shape. So when a peristaltic event happens, as is kind of happening right here, they don't really care. They just deform and respond to that. They're unaffected by these peristaltic events. In contrast, as actually we saw back at the beginning, this armonis, the pink species, largely exists in clusters and these clusters tend to be in the narrow mid-gut of the fish and they're essentially non-motile. This is in fact a movie. If you look really carefully, you'll find like a couple of bacteria swimming around somewhere up there. Anyway, we have these dense non-motile clusters and they're very strongly affected by peristaltic contractions. Whoa, that's a bit much. Very strongly affected by peristaltic contractions. They're pushed back and forth whenever these waves go through. So they're affected by these contractions even in the absence of the blue species. So if we look at the dynamics of the pink one alone, we do see these large population collapses but two things are notable about them. First of all, that typically even after a collapse, the armonis can grow back to its origin, back to the carrying capacity, back to its original population size but also that the rate, the probability per unit time or the rate of these collapses occurring is actually about twice as high when the blue species is there than when it's absent. So the picture that emerges then is that what's going on in this two species interaction is that the pink species isn't having some suppressed growth rate or something like that. It's exhibiting quite normal growth except that it's punctuated by these stochastic collapses that are driven by the peristaltic motility of the gut. So if that picture is true, we should be able to do a couple of things. So first of all, we should be able to come up with some meaningful quantitative model of this. So how could you do that? Well, as I mentioned, things seem to be described well by logistic growth. So you can imagine sort of constructing a model in which you have just logistic growth, just characterized by a growth rate and some carrying capacity. You could also imagine some variance in that carrying capacity that ends up not affecting anything. Yeah, so growth rate and carrying capacity. Then you could say, again, that you have collapses built into this. And those collapses are characterized by some fraction, like what fraction of the population disappears and by what the rate of these collapses are. So what is their probability pre-init time? So you have this model, very simple one with these four parameters here. Now, this model is trivial to simulate. It's just logistic growth with these random collapses. And so you could ask, as a function of these parameters, what is the mean and variance of the population that I should expect after, let's say, 24 hours? So you see me like that and you get some mean and variance of these. So those are functions of these parameters. If, for example, I keep everything fixed but vary that collapse magnitude, the standard deviation and mean of that final population trace out some particular curve. If I keep everything else fixed and vary the probability pre-init time of the collapses, the trace out almost exactly the same curve. And it turns out that I can collapse those two parameters characterizing the collapses onto one effective parameter that turns out to be the probability pre-init time times the log of the collapse fraction. Now, we stumbled on that just numerically, but now we actually have analytic solutions to this whole population dynamic model, which ends up being super interesting. So we actually even understand at a deeper level why that collapse occurs. Now, the upshot then is that we had this model with four parameters but two of them collapsed onto this one parameter. And one of those parameters was the growth rate, which we measure extremely well from the experiments anyway. So we have just two parameters in this model that we can, that are sort of free to fit to the data. So we can do that. We can go back to that plating data I showed you before and say what are the best fit model parameters characterizing the carrying capacity and this collapse parameter, given the mean and variance of the armonis populations that we saw before, and you fit it and you get some number. Not particularly interesting what the number is, but the z is 0.13 plus minus 0.5. But that z remember is the product of the probability of collapses pre-init time and the magnitude of collapses. And we know both of those because we're watching the fish. We get those from these time series. So from the imaging data we can measure p and f and what we find is exactly the same collapse parameter, the z, as we got from the static data. The fact that they're exactly the same is just kind of a cosmic joke or something, but the fact that they're approximately similar was like extremely remarkable. I even put an exclamation point on it. So we have then, yeah, so it's also telling us something kind of interesting which is that doing an ensemble measurement of a bunch of fish at one time point agrees with what we measure from doing measurements of individual fish over time. So there's sort of a ergodicity of fish at play here. Okay, so we have then a quantitative model that seems to actually describe what's going on here. The other thing you would want if this picture I've been saying is true, is a way to alter this gut motility. That's the sort of fundamental driver of this apparent interaction between the two microbes. And what you would expect is that if I could snap my fingers and turn off peristalsis, I would get rid of this apparent competition between these two bacteria. So I can't quite do that, but I can come actually pretty close because one of the wonders of zebrafish is that there are a lot of very interesting mutants that exist. So here what you're looking at are the enteric neurons of the larval zebrafish. So you all have a whole nervous system sort of surrounding your gut, you have these enteric neurons that help mediate processes like peristalsis. And there are zebrafish with mutations in a particular gene called RET that have no enteric nervous system. Interestingly, the same mutation occurs in humans where it gives rise to something called Hirschsprung's disease. It's also characterized by a loss, although it may not be a total loss of enteric neurons and people that have severe transport issues in their gut and also actually dysbiosis of the microbiota in them. Anyway, there are issues with this fish. They have other kind of health issues as well, so it's not exactly a clean experiment, but we'll take what we can get. So these redfish, it turns out, they do still have peristalta contractions, but they're weaker than normal wild-type fish. So what happens in those? So here in a wild type where the heterozygous fish, we again, like we saw before, have blue-dominating red when invaded and left to sit for 24 hours. Doing this in the RET mutants, the pink and blue happily coexist. So we can get rid of this apparent competition between the Vibrio and the armonis in a different host genetic background. Vibrio and armonis can happily coexist. And you can look at them too. Yes? Change with the V-pedium? That's a great question. You'll get a better answer later, and the answer is not much, but you'll get a better answer later. Yeah. So to first order now, yeah. Right, and here there are red and blue, happily even co-localizing in these redfish. Okay, right, so host genotype and in fact the enteric nervous system can influence apparent competition between gut microbes. So with this part, what hopefully I've convinced you is that we can actually get a quantitative understanding of competitive or apparently competitive population dynamics like in situ. And I think the really key thing is that the host environment matters. Or here's a concrete example of the host environment mattering. And it matters because of the physical nature of it. So things like gut motility and peristalsis and the kind of flow that that creates, also echoing in this previous talk, are really crucial to bacterial behaviors. And bacteria need to respond to these forces either as individuals or as aggregates, and that can determine a lot of things about them. So to give an answer, at least in this context, to Daniel's question, can we engineer environments? Yes, we could reshape for example the host type to engineer the type of environment that the gut microbes are seeing. Okay, so probably some of you are wondering, what is it about the Vibrio that makes the armonis more susceptible to these collapses when it's present? What is it that makes the collapse rate twice as high? And sadly, I have to point out, we don't actually know the answer to this question. It's a very good question. Our guess is that maybe it has, or a guess is that it maybe would have to do with disruption of adhesion of those clusters. One can expand that question actually more generally and ask, can we in general understand various specific molecular factors and how they may play a role in guiding interbacterial interactions? And can we use such factors to intentionally reshape gut communities? So I'm gonna tell you a story about that. So this involves Vibrio cholera. So not a native fish commensal, but actually Vibrio cholera, the agent that causes cholera. And this is a collaboration with Shado Xavier at Stone Catering and especially Brian Hammer, microbiologist at Georgia Tech. So Vibrio cholera, yeah, causes cholera. And also, like many other bacteria, as probably people in this room know much about it, my, has what's known as the type six secretion system. So this is this molecular sort of syringe-like apparatus or sort of protein-based syringe-like apparatus that punctures other bacteria and delivers various toxins to its victims. And there have been a lot of beautiful in vitro experiments exploring the type six system in recent years and also from like metagenomic studies and sequencing-based things, indications that the type six secretion system affects microbiota composition in human and other hosts or in other words, or it's correlated at least with differences in composition. Okay, so we've got Vibrio cholera. It's got the type six secretion system and wild type cholera has its type six apparatus and it also has factors that give it immunity to type six. So it's not killing itself and its siblings. So it has type six and immunity. We've got a mutant that I'll, or a strain that I'll refer to as killer that's been engineered to be always expressing the type six system. So it's always there, ready to stab its neighbor. We have the victim, which lacks the type six secretion system and also lacks immunity to type six. So it can be killed. And we have the defective killer which has immunity but doesn't have a functional type six system. So we've got killer or victim, these combinations. So these are from humans. We wanna see, can we determine the activity of this type six secretion system in a gut? Will they even colonize the zebrafish? Thankfully, the answer is yes. So all of these different engineered strains will colonize. Not as well as the wild type Vibrio that I've been showing you, sorry, the commensal Vibrio I've been showing you before about an order of magnitude lower, but still, they colonize quite robustly. So can Vibrio cholera use the type six secretion system to defeat another species? So let's do the same experiment as before. We're gonna have armonis as our sacrificial lamb and see can it be invaded by this Vibrio cholera. So the victim, okay, so let me show you what we see. So what I'm gonna show you is a scan through the gut and we're gonna just look at the armonis fluorescence and here's what it looks like after we add, 24 hours after we add the victim Vibrio strain. You see lots and lots of stuff, lots and lots of armonis. It looks, you see big clusters, you see individuals, you see a nice big happy armonis population. Here's armonis after we've added the killer. I'll pause right there. It is empty. Here we see the outline of the gut. Here we see nothing. The armonis has been annihilated by the killer. I think there's one bacteria, oh yeah, there was one bacteria just a moment ago. Oh, maybe it's three. Anyway, very few bacteria after the killer comes in. Quantifying that, so here's the, yeah, good. The armonis abundance, if it's invaded by the victim, by the defective killer, by wild type, we get, you know, 10 to the three-ish armonis, invaded by the killer. It's a couple of, it's like two orders of magnitude lower with lots of extinction events down here. Time series are pretty dramatic too. Here's armonis invaded by the victim. And again, we see these expulsions, but armonis happily grows back. Whereas with the constitutively active killer, there we go. Expulsion, expulsion, nothing much left. So, good. So invaded by this type six secretion, the constitutively active type six secretion system strain, armonis collapses, and in fact the collapse rate is even, it's twice as high as it was with that commensal Vibrio. So the type six strain has a big effect on the invasion abilities of this Vibrio strain. So what is that type six secretion system doing? So the obvious answer to that is, like I've said before, they use, the bacteria use this to stab other bacteria, so it's killing armonis. So we thought, okay, yeah, that's what it's doing. Basically as a control, we thought, kind of as you asked, let's see if there's any changes to the gut motility itself. So yeah, we analyzed gut motility, looking both at its frequency, and as I mentioned before, the amplitude of the contractions and how much stuff is displaced during any of these contractile pulses. So here is that amplitude normalized by the germ-free fish amplitude. For germ-free fish, so no bacteria, fish associated with the defective killer, so basically identical, gut motility amplitude, and here are zebrafish associated with the killer, the constitutively type six active. So we see a more than 100% increase in the strength of the peristaltic motions of the gut induced by these type six positive bacteria. So we're really quite surprised by this. So yeah, it looks like these bacteria are making the host peristaltic contractions much bigger. Coming back to your question a bit now, I have a graph of this and I can show it later, but the commensal vibrio, its mean is actually over here, but it has actually a large variant, so we might think there would be some of that effect with commensal vibrio as well, but nothing nearly this strong. So yeah, so this is affecting the host gut motility. Now, that raised the question, how could it be doing this? So this brought up something that we perhaps should have been anticipating before. So this type six, you're all sleepy after lunch, but I'll ask you a question anyway. Where was the, how was the type six secretion system discovered? Ha ha ha ha ha ha, that might make it too much like an exam or a, anybody know? Type six, type six. How was the type six? What do you mean by host cells? Can you be more specific? Close, so yeah, the type six was discovered actually by bacterial abilities to kill eukaryotic cells, actually dictyostelium, so what we saw yesterday, actually, so dictyostelium. After that, people looked at like tissue, like cultured cells, but yeah, the type six secretion system was even though people predominantly think of it now as a bacterial, bacterial-bacterial stabbing mechanism, it was actually discovered as a tool for killing eukaryotes. And in fact, people have figured out that it's a particular actin crosslinking domain that's part of the type six apparatus that's responsible for this. So, almost done. What happens, so our collaborators created a type six active strain that lacks the actin crosslinking domain. And what happens there is the gut motility amplitude goes back to normal. So it is the actin crosslinking ability of the type six that stimulates this gut-parastaltic activity. And then, if you use those things to invade the aeromonas, the aeromonas can happily withstand the invasion. So, can ribbericolor use type six to defeat another species? The answer is yes. And yeah, so expression of this particular tool, this type six secretion system, can strongly affect an existing microbial population. And it can do so by actually altering the environment that these bacteria live in. Reflecting actually a theme that's come up several times at this workshop. And so we have a specific mechanism for influencing community structure and for bacteria to shape this environment, but not the one we expected. And as far as we know, actually it's the first instance of the type six secretion system influencing an actual kind of physiological function of an animal. So I'll end just by showing you two interesting movies that point out that there's a lot to do with this overall system. So we've been talking about perturbations in a way. So speaking a bit more generally, we've been talking about perturbations like invasion, how they can reshape the structure of a community. We can think of other perturbations as well, like for example, antibiotics, and ask what they might do. So this theme of physical structure being important ends up being reflected actually in these as well. So this is worked by a really phenomenal graduate student, Brandon Schloman. I'll skip to the movie actually. Here, what you're looking at is again, Vibrio, the nice commensal Vibrio species, the one that I used to label in blue, nice commensal Vibrio species. And as we see before, there's nice planktonic fast individuals, but you see also a large non-motile aggregate. What we've done here is expose the fish to a very low dose of antibiotics, about 10% of the MIC. And what we find is that that induces, as is known actually from in vitro stuff, filamentation and the loss of motility, then in the gut end up being funneled into these aggregates that then go on to be, for example, expelled from the gut in the same way as the armonis ones do. So we can induce changes in the physical structure of microbes with low dose antibiotics, which we think is really important both for the persistence within the gut, but also their transmission and dispersal to new environments. So that's something we're starting to look at now and I think I won't tell you a bit more about that, but I'll show you one other movie. Blah, blah, blah, this works all kinds of antibiotics. So everything I've shown you is for two species, but we can also think about whether we can do what I claimed at the beginning, build upon this to construct more complex communities. So we haven't yet actually looked at more than two species in the gut, but we do now have several, about six or so, that are engineered with extremely good chromosome-integrated fluorescent proteins. And just to illustrate with one slide, why this I think will end up being extremely interesting, is we already see a nice variety of structures. The bottom movie is the one I showed before, a Vibrio zipping around. The top one is a commensal plasium monospecies that's also motile. These are both playing at the same speed. It's also motile, it's not as fast, and it forms these odd sparse aggregates that are probably largely full of cross-linked mucin or something like that in the gut. So one can imagine a wide variety of landscapes that these guys can create. So I'll skip to the conclusions, where are the conclusions? Yeah, those are my kids. So the things to keep in mind, I think, are that imaging is actually a neat approach to determine microbial population dynamics actually in vivo or in situ. And it tells us that populations are spatially heterogeneous and dynamic. And that we can get from that insights into interactions between species, and they tell us that things like spatial structure, the physical properties of the environment, like peristalsis and so on, and particular mechanisms like the type excretion system can mediate these sorts of competitions. So this has been work from my group, collaboration with Brian Hammer and Shao Xavier for the Vibrio cholera stuff. Nearly everything we do is a, especially the beginning and ending stuff, is a close collaboration with Karen Gilliman, a wonderful microbiologist at the University of Oregon. And a lot of the commensal to species stuff was done by a student of mine, Matt Jimmelito, and a postdoc of hers, Travis Wiles. All the type six stuff is Savannah Logan, there she is, and an excellent undergrad, Drew Shields, who I forgot to put in bold. Yeah, so thank you very much.