15:07:09 Okay, Thank you.
15:07:11 Thanks everyone for sticking around still at 3pm.
15:07:17 And so it's a real pleasure to be here to get all the in person, wipe from all of you. After a year long Nirvana in zoom where I lost my sense of time.
15:07:29 This is a great experience of kind of waking up.
15:07:34 So thanks to the organizers SEPA Pankaj Daniel and auto Yeah, for, for putting this together I really enjoyed a lot. Fortunately I can I can be here in person just for a couple of days.
15:07:50 But I'll stay connected. That in zoom again.
15:07:53 Alright so
15:07:57 to start off my talk about work done by a very talented PhD student, mainly your character. I wanted to say one uncontroversial thing, which is that, you know, ecology and evolution, depends on the environment.
15:08:18 It varies with environment and so by that.
15:08:21 If that's true, then that implies if we want to do experiments on what we're interested in ecology and evolution.
15:08:26 We have to try hard to control the environment.
15:08:31 So not only do we have to think about what we put into our experiments in terms of different species, but we also have to think about the properties of the environment.
15:08:42 And of course one important project property that often it's very well controlled as the kind of chemicals, be put in an agile environment the temperature and so on.
15:08:51 But another one which will be the focus of my talk is physical structure.
15:08:56 There's a very natural one, which is not only popular but I would say beautiful, it's the well make structure. You just tear in a chemist, very well or you take your test tube and shake it really really well, both cases, you can ensure that limit quite
15:09:15 well. And it's, it's beautiful from an experimental standpoint because we can realize it quite confidently and reliably.
15:09:25 And it's also beautiful in a way from theory standpoint, because it allows us to make strong simplifications because our models become infield models.
15:09:35 As far as space is concerned.
15:09:38 Of course, in papers written at that level. It's usually said as important caveat. Well, there's also space.
15:09:47 And indeed, as we've seen in the previous Park, of course, there are you know situations where microbes grow in their intricate spatial structures biofilms in colonies.
15:10:03 Here you have I've pulled out an example from the breezy lab where they image the microbes that grow on our on.
15:10:12 But even, you know at lot much larger scale where you might think that perhaps the situation is well mix, it's still not clear whether it's really well mixed because in the ocean you have complicated currents that generate.
15:10:29 You know long distance transport, that has some spatial structure, even at that level.
15:10:35 We don't understand yet well what that implies for microbial dynamic so ecological.
15:10:43 So, um, right. So, so spatial structures important and also I think in our discussion in the courtyard discussion which I could only participate the last 10 minutes space was mentioned, multiple times is kind of an important frontier, which I totally
15:10:58 would subscribe to.
15:11:00 But the difficulty of one of the main difficulties I would say is that there's only one well make structure, this one here, but kind of infinitely many specialty.
15:11:11 So which one should we choose or stated differently. Do we have a way of actually systematically sampling the spatial structures.
15:11:24 And definitely a challenge. That's essentially the goal of what I want to show you today is, I want to study a certain class of spatial structures.
15:11:33 And then identify a way how I can systematically change that spatial structure and try to understand how that spatial structure then influences the self organization microbes that grow, and divide and that's patient structure.
15:11:51 It's going to be at the level of the micro is going to be relatively simple just one species.
15:11:56 But because I want to focus on that question of physical structure.
15:12:01 And to put that into a broader context I pulled out a quote from last week. I think he said this.
15:12:11 We spend a lot of time studying model organisms but relatively little time creating model environments.
15:12:16 So, this is an aspect where I want to tune one.
15:12:21 I think important not also have the environment, what's the physical structure the scale of that environment.
15:12:28 And also there's an important
15:12:33 essay in the literature I think cited almost 9000 times by Simon 11, which is emphasizing the problem of scale and ecology.
15:12:44 It's a bit of a diffuse concept but what I took away from this is your many phenomena in that was written about macro ecology many phenomena depend on the scale you're looking at the system.
15:12:57 And also on the scale at which these systems are operating. So if we mix together observations done on different scales, we get confusing answers that we shouldn't do that should be can do experiments that are aware of the scale, looking at which.
15:13:16 Alright so to to now make this a little bit more concrete and allude to the kind of physical structures.
15:13:22 I want to study.
15:13:25 Let me point out that bacteria are really good at colonizing spatially secluded spaces of poor like structures.
15:13:35 For instance, in the soil, you have course it's full of pores, there's a wide pore size distribution, and you do find microbes in them.
15:13:44 Also on animals on most larger living things use, you have poor like structures which are colonized by microbes. For instance, on us on our skin. We have the, we have skin follicles sweat glands, the patients plans, and they are often filled with microbes
15:14:02 see acne here.
15:14:05 And to a very heavier high density.
15:14:07 And I already want to point out a paper that is, is a strong relationship to what I'm presenting here which is from Lieberman's lab calm.
15:14:19 Okay.
15:14:22 And also, for instance, in the gut, you have intestinal crypt, which are also filled by microbes, for instance, but since fragile is.
15:14:34 And such situations are observed many different species including the fly I will later use a species that's taken out of the fly guard by my collaborator we're letting them.
15:14:47 And also they are up to your growth identities and the spaces.
15:14:54 And my whiteboard okay what are the properties then of these populations and, at least in the case of intestinal crypts it has been really measured quite well in nicely that these, these, these habitats are are colonized very stable instability here was
15:15:11 was measured.
15:15:14 And suddenly the following way you took, they took a mice, a mouse and colonize them with with, but soon as fragile as like a wild type unlabeled wild type.
15:15:26 Maybe it was labeled with one person marker. And then, after this pre colonization they try to bring in another strain, which was labeled differently.
15:15:38 And that strain wasn't able to invade unless it was given kind of an advantage right but it had also an antibiotic marker so you selected for the one that came.
15:15:49 But, at least in the absence of antibiotics, the resident was very very stable, and that's called colonization resistance is kind of a measure of stability which is quite significant and you might wonder.
15:16:02 You know what gives.
15:16:05 What about these structures gives colonization resistance.
15:16:09 Do you always get it, independent on, let's say the scale of these cavities.
15:16:16 And what's going on. And this is what I'm trying to investigate in virtual system.
15:16:22 Later on, but just to give you an idea of what what might generate this colonization resistance.
15:16:28 I have this kind of violent analogy to, to the Battle of the term of Bueller Bueller happened a couple of thousand years ago, where
15:16:43 the Spartans. Um, you know, we're able to fend off supreme ways to period invader the Persians for quite a long period of time, simply by taking advantage of the topology, the surrounding, they found a narrow mountain pass position themselves there and
15:16:58 then providing hard they still lost, ultimately, but they, they were able to use you know the structure of the surroundings to their advantage. So maybe that's also going on the level of microbes, we'll see.
15:17:12 But what these paintings also show is that we would like to see not only kind of the actors, the fighters. We also would like to see the environment here.
15:17:24 And, and, and that's where the hero of the story comes in, which is New Year Kavita who developed a microfluidic device to allowed us to study this problem scale dependence here at systematically, and he built on a lot of work in microfluidic devices.
15:17:43 But he had one crucial innovation so let me try to explain this so what you see here is essentially just the structure of a microfluidic channel, we see the scale 50 microns of these.
15:17:54 So as we bring in, mostly flow in media, but as we bring in bacteria these bacteria see these imaginations and they can and principal enter them and colonize them.
15:18:07 And what's critical about this device is first of all that this scale is not too small right so one device that many people might know is the mother machine which is just the width of a single bacterium, so you get a sausage factory of bacteria, where
15:18:27 nobody can overtake.
15:18:29 Another bacterium so this is relatively wide. The height is about 10 microns.
15:18:35 And the small but crucial innovation compared to these previous works, is to systematically change the length of these cavities, just to explore to ask the question, know if I change the scale of my, my cavity.
15:18:48 How does it affect the self organization, yeah.
15:18:54 This is not really drawn to scale here you see the diameter 50 microns and you see the height of the tallest one is 10 microbes or 10 in the in the in the direction, out of the screen.
15:19:12 Out of the screen and how about the height, the death. What is this, oh this is drawn to scale or Okay, got it.
15:19:18 And well because it looks like kind of a pan flute, call it Mike flick pants.
15:19:26 And what we studied in there. At first, we also look at other organisms now but what the body study first this this bacterium as you go back to Indonesians is, which is derived from the fly gut by our collaborative letting them, who studies that in situ,
15:19:46 and perhaps can see some of the nominally be will explore here also, and fly.
15:19:55 Alright so when we first I want to show you kind of still picture of what happens at steady state at the low end of that microfluidic device.
15:20:09 A bit dirty but you see that here in these smaller cavities.
15:20:10 There's practically nothing.
15:20:12 Few sales find their way in here.
15:20:15 But then when these cavities. Show some substantial density.
15:20:21 It's.
15:20:23 They are characterized by some, I would say characteristic links here. I'll see, which is the first question that we, we asked right so can we understand why there's kind of a transition from nothing to something.
15:20:36 And when it appears, there's a group of the things.
15:20:40 And maybe in your head you already forming a model for this. So what's going on here is to these microbes they are not multiple.
15:20:50 So they diffuse passively in the system.
15:20:54 So they are, they are brought along by the flow, and then, occasionally they lucky they diffuse in here.
15:21:00 And then they have the chance to to divide. And many of this game of division in this cavity, but also you have outflow due to the future. Right. And so if the cavity is too small, what's going on then is the diffuse of outflow is too large compared to
15:21:22 proliferation, so that this is not happening and this is very, you know, this is the physics here that's going on. The biophysics, and a simple model for this is shown here, where we just model the, The dynamics here in one dimension along the vertical
15:21:40 axis, we define, we say that the rate of change of the density of this population as position as a function of position excellent time t is controlled by two terms one is the fusion.
15:21:53 The other one is growth with the growth rate are.
15:21:57 And that needs to be supplemented by boundary conditions.
15:22:01 So first I define.
15:22:12 And since I define it that way, I have that boundary condition.
15:22:16 I define, essentially, the x equal to zero the origin at the place where the density goes to zero.
15:22:16 And then, well at the floor of the of that cavity, nothing leaks, so we have to ask cells to be reflected at that boundary which corresponds to this patient.
15:22:28 And if you analyze that then you get a very simple answer. So you find that their characteristic scale pops out, which is can only be proportional to the only links that you have in this model, which is the square root of the feasibility of a growth rate,
15:22:46 dressed with some pre factor, and predefined is that if this you know if this Ellis, smaller than this critical scale, then you can only find extinction, all the modes the normal modes in the system go to zero.
15:23:04 And the flip side of this is that when l is larger than MC then you get growth.
15:23:09 In fact, in this model because it's linear, you get endless growth, right, which is of course has to break at some point and this will be kind of the more interesting part of the story is how it breaks.
15:23:20 But this is already something we can test with a kind of this establishment transition is well described by this simple model, because the facility is something we can measure by tracking the self carefully.
15:23:34 And we find that this is essentially equivalent to the passive activity of a ball of radios pointed microns.
15:23:41 And the growth rate we can also measure relatively carefully and that gives us this critical length scale, which agrees very well with what we kind of can extrapolate but by going across those pipes is a question.
15:23:59 It was the choice of species important like it seems like maybe these, these bugs don't stick to each other, they don't make any extra probably sack right or any Yeah, they do not, these ones they maybe there's a very mild sticking to one another.
15:24:13 That.
15:24:15 But what we can see occasionally but there's no obvious biofilm production there. But, yeah, so, so if you look at different systems you have a strong biofilm former this, this cannot.
15:24:29 Yeah.
15:24:38 So flow what flow does essentially it creates if you think about the flow lines, what happens is the flow lines they penetrate this cavity.
15:24:48 And there's kind of a, in a way, like simplistically speaking there's kind of a lowest flow line, which is washing outsells. And so if the cells are diffusing away from that flow line, then they get into this into this area where everything is dominated
15:25:03 by the fusion, rather than flow and actually flow is a very handy way of, in addition to effectively tune the size of this
15:25:16 of the cavity where you only have to fusion and growth and into control how large this wash layers.
15:25:23 I see because I was wondering why that doesn't factor into your MC or in your model, right, we are essentially only modeling under the ceiling of the washout flow, that's what we were modeling and the rest is taken care of by this boundary condition Yes,
15:25:38 a related question What about the, wouldn't the growth rate be dependent on the ups in the layer as nutrients are depleted. Right, that's what you would think but it turns out and everything that I'm showing you actually growth is not limited because
15:25:53 turns out, but there are different ways of getting it that one is simply estimating how fast molecular diffusion is how much is being consumed. Down here, and at these scales at least growth rate is is effectively a slightly reduced at the highest densities,
15:26:12 but it will not be able to explain the phenomena that I'm going to show you.
15:26:16 In fact, even at the highest density is we have a direct measure of growth. And it's hardly different from the.
15:26:27 In fact, that's a, but let me point out one one other important kind of test of this.
15:26:34 Even though this is just a linear theory it describes what the density profile should be here of the cells, as a function of the distance to the floor of the pipe which simply a cosine profile.
15:26:48 And so this is something you can also measure quite carefully and it agrees quite well.
15:26:55 To kind of the smallest pipes, up to a certain point, when the density becomes very small that's really where the flow then makes a difference. Of course this a symmetric shape that has something to do with the flow and it's not captured by the simple
15:27:07 theory.
15:27:10 All right, but we know the next question is what happens in the super critical case we have growth, where so to speak here.
15:27:18 Yeah, there you have growth.
15:27:20 How do you then find a steady state. In this is really a steady state, okay so so let's look at the time lapse movie of the same experiment.
15:27:31 And what you see is shown here at steady state.
15:27:37 So as you go to larger and larger cavities. You see kind of the same picture of jostling cells in these cavities, the density gets higher, but there's a rather sudden transition where you kind of change the size of this cavity by a little bit.
15:27:54 And suddenly kind of this very dark fraction of cells appears.
15:28:01 And in this doc friction continues to be there.
15:28:04 So, what you can, what you can show is by kinfolk imaging, that the cells you are actually touching one another.
15:28:11 So, they are literally jumped on this transition happening here, and it's quite sudden it's, it's a discontinuous transition.
15:28:31 Oh, it's the last one I just showed you kind of the microwave tantrums very long, and I just showed you that first part where you have the onset of colonization.
15:28:41 Yeah, this is now for larger than the critical size or so is getting washed out of one and into the next.
15:28:50 Say it again, or so was getting washed out of one and then but yeah so so they they are, they are also moving around here. That's true.
15:28:59 That's true is that religious the oscillation on the writers that might imagine you see an oscillation, and we may be in the boundary be looping. Yeah, I don't know, I think, I think there's, there's a lot of fluctuations especially close to, to the transition
15:29:16 here.
15:29:19 Are there bubbles bubbles. Well, this.
15:29:27 That's at least what we think what it is.
15:29:28 These, this is kind of debris.
15:29:29 It's definitely things that we can actually track as this population is growing up.
15:29:35 So, so one way of measuring the growth rate is by tracking by taking these things and tracking them from the floor to the top and asked how fast is that advancing if everything is growing exponentially at the same rate.
15:30:04 speed of these things moving upward is proportional to the distance of the floor and that's exactly what you find and the growth rate is the same as what we have in the Duluth case
15:30:05 also wanted to show you how this happens then amicably because quite interesting that you see the hours.
15:30:11 So at 10 hours the three ones are filled, but actually at the cavities that are close to the transition takes really long to actually realize that yeah we should be in a different state, and develop that transition.
15:30:29 So, yeah, the next question will be about what could be going on there.
15:30:36 And the natural thing to do is to start with the model that we had before about the onset of that growth and to ask, could be modified perhaps to describe this phenomena.
15:30:50 And we already discussed the growth rate. So we do not think that. So what we need is a non linearity to actually find reach steady state, and natural non linearity that often us of course, talking about is one that affects growth at high density, but
15:31:06 here we really have very clear evidence that that growth, it most by 5% changes by 5% at most.
15:31:15 And that's not enough to explain the phenomena that we see here. So, maybe the non linearity has to be in the other part, which is about diffusion.
15:31:29 Oh, I forgot to say like if even if you simulate this.
15:31:34 Well, these are these are dry simulations I think the fact that of water around the cells is very very important, but you see this in dry simulations and, but we also have done read simulations so this is David lemma and he's our collaborate on this project
15:31:47 to do simulations of sales and in water and without water to general these phenomenal.
15:32:02 But yeah, to, to upgrade the model. You know, we could, we could imagine and that's actually quite natural. To say that the facility the past if the facility of these cells is a function of density.
15:32:14 And indeed, from colloidal physics, actually, it's known that the defense of it in general is density dependence.
15:32:22 And you can actually decompose this quantity into two parts.
15:32:28 One is a term that's an equilibrium quantity. So that's the gradient of the automatic pressure is a function of density.
15:32:39 And so, it tells you kind of how much pressure you generate.
15:32:46 As you increase the density and so this is kind of the driving force for pushing out itself.
15:32:51 And then you have a second factor here, which has to deal with mobility so that's a kinetic term, it's very hard to understand our priority. In fact, both of these terms go in opposite directions so there was nothing miraculous is increasing with density.
15:33:09 But the mobility goes down with density simply because you kind of get traffic jams between the cells as the density gets higher and higher.
15:33:18 And so it's very hard to say how the behavior of this quantity is, in general, but we can try to analyze this equation here at steady state and ask what properties, what features this this this quantity.
15:33:35 If you see it as a function of density must have, in order to generate the transition that we saw and we can exploit in fact that kind of an exact mechanical analogy here.
15:33:47 see you define that as a, we define that as a pressure because it really act like a pressure in the system or effective pressure in the system which means the the gradient of that quantity the spatial gradient of that quantity is represents a particle
15:34:08 current current due to the future.
15:34:11 And if you write the same equation in terms of this pressure, what you find is the equation of that time so the second derivative of that pressure is a function of x is the rate of change.
15:34:26 When it's given by this quantity be here, this is slightly more general than what, what we have in the system more general. So be here is the rate density of birth, and generally can make it also a function of the density.
15:34:42 As long as the density is just a function of the pressure.
15:34:48 We can redefine we introduce your function you.
15:34:49 And that, you know, can be.
15:34:52 So we define that right hand side simply as a gradient of a function you and then this equation is just Newton's equation, where
15:35:03 you have the second derivative of the position, like the acceleration is the gradient of some potential with respect to the position, right, and that we know how to solve for any kind of potential you.
15:35:17 And so we can ask what properties, do we need to have need to ask about this function you.
15:35:26 And then about this function, which, which implies properties of the facility in order to see that transition
15:35:34 City in.
15:35:46 In, not in this in in our system, but in the, in the, in the, that includes that part so you can you can include that and ask this question even more generally, within you.
15:35:57 Anyway, what, what we do find by doing this analysis is what we need to have is that this density, this effective disability density pendant effusive it needs to have a negative slope, meaning that if you serve it decreases with density, they are fundamental
15:36:08 reasons why it low density is it has to increase.
15:36:13 You know that. Actually, exact laws, you can write down for the low density limit.
15:36:18 But what we need to have here is that if you see it goes down at high densities.
15:36:24 And when that happens intuitively you can understand this transition as driven by a positive feedback.
15:36:31 Imagine you have in your population can have a density fluctuation up.
15:36:37 Then, locally your the facility goes down, which means you have less flow out of that region,
15:36:45 which means your density increases further because growth growth continues at the same rate. So you have this positive feedback, which can only stop when sales actually touch.
15:36:55 So when you get that Oscar there's a question from somebody on zoom, go ahead still on us.
15:37:03 Oh, thank you. I was, I was trying to wait for the q amp A at the end but my question was just.
15:37:07 Do we know what the nutrient gradients are in this device, and in particular could get this spatial pattern that you've shown, somehow emerge as a us through an interplay with nutrient gradients.
15:37:23 Based on for example where you put the nutrients in or where they accumulate or something like that. We haven't really talked about the nutrient gradient so far right, I haven't we haven't done any measurements of nutrients in the system we have only
15:37:35 done measurements of growth which is kind of the critical feature.
15:37:39 At least within the model and the growth rates, simply do not change they are the same, at the, at the highest densities are tiny, tiny changes, way too small to explain what you see here.
15:37:53 So, even if they are gradients, and they will be gradients, and they do not seem to affect growth, and therefore on. Not likely. I would say to be the reason underlying this phenomena.
15:38:06 Right. Thank you.
15:38:10 Yeah.
15:38:13 Just as a follow up to this since you have this general theory which allows you to put any dependence of the birth rate on density if you just put a standard, the density increases until it reaches a carrying capacity, what do you get.
15:38:28 Well, if you only have that you do not find this phenomenon. If you only have that you do not find that you can ask, you can't play the game how strong, how strong does that need to be in the presence of a negative slope in this DLC to still see this
15:38:42 phenomenon, those kinds of games you can play. So if you have this plus gentleman, plus some density dependent diffusion, you get a depends on how strong the fact
15:39:03 So, since this doesn't seem to depend on are the differences in growth rates of the cells don't seem to matter much suppose during your flow. You also include small beads that are the same size as your cells.
15:39:04 is another question it.
15:39:14 Can you like artificially increase the concentration of the effective particles such as fewer cells, you start seeing the timing transition in a way that the cells don't even need to rise.
15:39:24 Right. I mean those.
15:39:26 The problem is to actually bring a lot of cells, as far along of beats in there. I think in before you even bring in the bacteria. Yeah, so I guess like I was thinking more, suppose you include the bacteria, and then in your media you just have a low
15:39:39 concentration of beets.
15:39:56 the growth that's happening in these do expect them to also get trapped in the jam region, I guess that's the part I'm trying to understand like would you not expect diffusion not be enough to remove them in the same way it's not enough to the fusion
15:40:06 not but they get pushed out by growth that they grow I see, okay, but the cells themselves who the cells themselves.
15:40:18 Okay, so So indeed, so then, then we asked you know, I think, one of the questions.
15:40:25 When I see the theorists is okay, this might be going on here. But, I'm like, do we expect this for maybe the even simpler system you can imagine the simplest one here is one where we say okay let's say the bacteria or just spheres.
15:40:40 They proliferate, and they just repel that just push against one another right it's a model of proliferating art spheres. Do you expect this to see.
15:40:53 And indeed, that's the case because this is actually something we can measure and we did a very careful simulation of odd spheres in liquid measure this density pendant The facility is proliferating heights fears and you get this hump like shape, which
15:41:08 actually can be also related to a fundamental illogical laws that are known in the literature.
15:41:19 And if you then analyze this equation with this density pen, the facility, you do find a fault bifurcation so I'm plotting here the density at the floor.
15:41:32 This is a packing for action actually of cells there at position elsewhere the density is highest as a function of the ratio of this length over the critical length right so this is larger than one, we have growth.
15:41:46 And for this extreme system you do do find it, that roughly you know as you increase this parameter at around 10%.
15:41:55 You get a jump, like, if, if you would let slowly increase the parameter you jump to the highest possible density is jamming density. And you can also then look at the density profile it's kind of continuous below here which we call gas like phase, and
15:42:11 then we jump to this partially jam phase.
15:42:16 And what's also implication of that theory is if you if you take that parameter then back down again.
15:42:23 So you lower that control parameter, then you should jump at a different place back down. So you you expect history says this is also called the tipping point in the specially in the biological literature.
15:42:38 And so this is something we also did test.
15:42:43 By studying and our pen from the device here in various ways, a growth rate up shift by actually changing the temperature.
15:42:52 We also did it by modifying the flow because when we, when we modify the flow we can effectively change the parameter, l, because right because this will show layer manipulated by by flow.
15:43:04 Here I'm showing you modulating this control parameter by growth rate because that affects this critical things, LLC. So, as we increase the growth rate, we go, let me go back one time so.
15:43:18 So we start with this, and we end up a jamming one more of these cavities. If you go back down again.
15:43:26 This, this cavity which was jammed here does not engine, which is an indication that there's history.
15:43:35 That's a nice consistent with this.
15:43:41 Where are the cells coming from, I mean these experiments are 30 hours are you apply, are you supplying cells at a fixed density at a constant rate, no no it's a very very tiny initial Benson yourself, okay.
15:43:53 They get invaded and then all the rest is just growth inside these cavities and being washed out, and you know what nutrient is limiting their growth.
15:44:04 It's not I mean it's a really high nutrient right they grow, essentially at maximum rate.
15:44:11 It's not, you know, they, they don't feel stressed as far as growth is concerned, we believe.
15:44:28 I mean I cannot exclude that there's a tiny bit of sticking to one another, but we don't see large clump.
15:44:39 No, no, that the forces are way too small, I mean these are walk, walk, walk, bacteria, and so right they aren't on automatic pressure and the kind of minute forces that are experiencing here.
15:44:51 Do not the form that I think we think mechanically they are kind of as a static package, they actually quite marginal least stable population so I think the mechanics could also be interesting for some parts that I'm discussing later
15:45:11 more of what.
15:45:14 Right. Yeah, yeah, yeah so that will be interesting definitely to also see how this depends on shape of the bacteria.
15:45:23 I want to come back now to this question of colonization resistance right so after describing kind of this self organization that's happening here is a function of scale and how sensitive it is to the scale.
15:45:34 We then asked, this, this question like, it's how stable is this population are these populations to invasion by kind of an outside invader. So we have another strain, which is derived from the same stream which is unlabeled here that strain is green.
15:45:54 And it's also resistant to tetracycline so tetracycline so those don't go.
15:46:02 If we flow in that invader in the absence of tetracycline, we don't see any invasion for five days. That's the longest time we run these experiments.
15:46:13 So, we have to bring in antibiotics to reduce the growth of the resident here. And so I show you now in the movie what happens the first 2021 hours.
15:46:29 Interesting things happen you get to kind of conversion from jams back to Gaslight state, but you see know where any green guy proliferating.
15:46:39 So let's. Yeah.
15:46:45 It's the same as the as the wild type in the absence of the drug.
15:46:54 So, what happens in the next 24 hours is shown here.
15:46:59 So you see, um, but I think it's a movie it's a movie.
15:47:03 So, some invasions happens to some green guys, enter in these populations and they start to proliferate so just have wait long enough invasion happens.
15:47:19 But I think intriguingly after five days, the kind of invasions we see so sometimes we see many places where invasions happens sometimes just build a few, but invasions only happen in the state where the population is relatively little crowded so it's
15:47:34 it's in this Gaslight state and not in this jam state so these jam populations were completely protected we never saw any case, where that occurred.
15:47:45 And so you can ask why. Why does that happen. Well, it has to do now again with the fusion, but it's kind of a different quantity it's it's self diffusion that matters now it's the diffusion of a label particle within a sea of unlabeled ones.
15:48:00 Because, an invader and what it needs to, to, to do is to kind of innovate this crowded population and start to proliferate there and take over the population.
15:48:12 And we can measure the self diffusion in the system, both in the guests like state and the state in the Gaslight state it's it's it's hardly, but it's it's it's on the scale of the one that I introduce passive diffusion of a ball of radios point I'd microns.
15:48:29 In, in the jam case, it's, it's very small, we cannot track individual cells.
15:48:35 But sometimes when we see a kind of little bit of diversity in the, in the population where we have here a dark stripe, and have a population of dark and a population of green in the same population, we can look at the fluctuations of the line and infer
15:48:51 the self diffusion there, and it's it's drastically different and for orders different in that facility.
15:49:00 And, and that has a couple of implications.
15:49:04 And that has a couple of implications. The first one is for a cell to invade this population, what it needs to do in the Gaslight state is kind of to enter, and diffuse against this outflow, and get somehow into this region.
15:49:18 Once you're in that region, you essentially in a remix population and you're competing with everybody else in here. So, if you're fitter than the population there.
15:49:29 You have a good chance of taking over.
15:49:31 So Oscar really depends on just how much tetracycline you're using and how much you're reducing the proliferation of the residents, right.
15:49:42 So are they still proliferating at some lower rate at the so proliferating at some of the stories, there's only a few set that rate to zero, they would wash out, and you get right.
15:49:51 That's right.
15:49:52 And so there's some critical. There's some.
15:49:56 There's some parameter which is the difference in proliferation rates between the invite and the resident but yeah so it's an interesting question. Yeah, it's an interesting question how the kind of the rate of invasions, and on their growth rate.
15:50:12 As a microphone.
15:50:14 Can it be that it's hard for antibiotic also to kind of get into the jam state, that seems to be the case because we actually use quite high concentrations of tetracycline, and
15:50:28 you have to use them in order to even appreciate, no sufficiently suppressed the, the resident knows that one of the basic antibiotic resistance mechanism is out any genes is just to form a biofilm or this crowded state, right, and then the antibiotic
15:50:47 you'll have a hard time to get to the bottom of this population which is still growing normal. Right, right. So can be that in your experiment, what you see is that like there is virtually no fitness difference between your antibiotic resistant strain
15:51:00 and your residence drain. So that's why you don't see this invasion. Well, we can we actually do see it and not, I mean definitely that there's a substantial difference here when we do this antibiotic act because you see, this is the state prior to that
15:51:16 well to get texts we see all of these are being jam, that's a function of the growth rate, and you know when we started that and to be able to get this is all going away.
15:51:26 They are here they get diluted out.
15:51:29 And so there's even here there are pipes where the invader can establish but the resident not anymore as a result of that suppression so it's it's definitely substantial but, actually.
15:51:41 How confident are you that the only antibiotic resistant ones are green. Say it again. How confident are you that the only ones that are antibiotic resistant or green, like you have a bunch of cells growing exponentially for five days, maybe they evolved
15:51:56 antibiotic resistance and, oh that might be. So you're suggesting that maybe the evolution is happening. Yeah, it's.
15:52:05 I'm not. I can't exclude it because this is.
15:52:10 Yeah.
15:52:14 Thousands Yeah.
15:52:17 Right, but I think up to 10,000 So, actually, I think one interesting, you know, type of experiment we want to do is to to study kind of these devices at the sweet spot where you have effectively remix state escapes your state.
15:52:34 And then, you know, in the limit where you have around 10,000 cells and maybe it's doable, and then you have many, many replicas of effectively well mixed.
15:52:45 Well, to be honest, where you can maybe study evolution as a relatively small scale but non negligible I think evolution might still be happening because it's an potentially interesting regime.
15:53:05 As you see there's not very much evolution at this point it's really just ecology and invasion of a foreign strain.
15:53:10 That's, that's future plans.
15:53:13 And, yeah, I cannot fully exclude that the resident has picked up an occasional.
15:53:22 No drop resistance gene.
15:53:25 I would just say that in this jam population it's super hard to actually do this I mean, just theoretically it's super hard because
15:53:36 the reason is if I go back now.
15:53:47 I'll go back to that later why I think that competition in the gym state.
15:53:57 Like difficult. Um, but anyway the point here was that invasions we only saw in the dilute state so if there seems to be a matter of time simply until invasion happens but be virtually never saw it here.
15:54:09 And the reason for the difficulty of invasion is,
15:54:15 is that the invader has to kind of infiltrate the population, we're going to beta has to do is to say two things.
15:54:23 It has to invade the population from the outside, and then it has to take over.
15:54:29 And they really two challenges the first of all to to invade to defuse against this.
15:54:36 In this densely packed populations is very slow. That's what we're measuring here, it's very unlikely to by chance to use down here.
15:54:44 And even then, when you hear even if you haven't increased growth rate. This is a tiny effect of population size, so to speak, where the competition is very inefficient.
15:54:55 So, that's the reason why we expect this. But why this is kind of consistent with our theoretic picture of what's going on the flip side of that is that these populations are probably not very good at evolving resistance, because a mutation from mutation
15:55:11 to take over. It has to arise in this tiny population, even though the population is actually an absolute number much bigger than the one here has to arise in here and take over in the tiny population so that's also hard there's a trade off.
15:55:30 Yeah, despite this trade off, there's actually also interesting things that go on when you turn on selection so here, it's hard to see but we kind of rather than doing the wild type first and then the invader we mix them both together, and occasionally
15:55:47 you find cavities, that are mixed right so they are jammed. And initially mixed.
15:55:55 And then we asked what happens when you turn on now selection so now we bring in tetracycline.
15:56:00 And what happens is happens in two steps I tried to illustrate this so first of show you the whole movie, we see when you turn on selection and the tier, the green population takes over, but kind of in two steps.
15:56:13 The first one is shown here. So at first.
15:56:17 So you wrote, reduce the growth rate of the dark ones, but until you lose a part of the dark fraction, but you also lose a part of the green fraction because also the green ones that kind of rely also on the presence of the dark strain to actually be
15:56:33 so abundant as it is here. In fact, the frequency in this first phase, the frequency of green is not practically not changing.
15:56:42 But then, this population turns into this effectively well mixed geishas state.
15:56:50 And then competition becomes quite efficient, and the green guys take off.
15:56:56 So this is just, you know, initial initial observations that we want to follow up and future work, this is not in the pre print, but there seems to be an interesting interplay between total population size and selection strength, the system.
15:57:16 Yeah.
15:57:16 Suppose the channel is large enough that even when you add antibiotic it never leaves the jam state. Yeah. Does that mean you just never fix the green guys like basically you have to exit the state before selection can act the way it needs to well, possibly
15:57:32 I haven't done this experiment but but it could be like that, because you kind of lose both types. Even if you lose just one of them.
15:57:45 Yeah, but I guess maybe one other way of asking it is if you reduce the amount of antibiotic you add in this context, right so it's that the D amount which you you crush the state is lower in channels, yeah that's that's a deeper chance to maintain them.
15:58:01 them. Do you ever noticed that at the deepest channels you never fix, I don't know, we haven't, we haven't yet done these experiments.
15:58:13 Yeah. gem state is similar to if you have grown them or, you know, on top of petri dish.
15:58:19 Or is it the two special structures that should give you different you know results when you mean with this is this is microfluidic chamber right gems state.
15:58:26 So if I want to say if you do that, similar experiment where if you grow the two types of cells or like, you know, lager containing in antibiotics. Would you see something like, similar to the gym the state or are they kind of.
15:58:41 I'm just trying to cuz you mentioned that if a number of special configurations right and they're there but I'm just curious about these for these two different special configurations, what do you see something similar or something different I Are you
15:58:52 thinking of a colony growing over the mix.
15:58:55 Yeah, I mean, then you have kind of a moving front.
15:58:59 But in terms of, like, in theory land, these two situations have some similarities, but also differences in terms of the kind of fluctuations, you expect to see, this is, this is a pushed wave, on, on the place these colonies of quote unquote waves.
15:59:22 Yeah.
15:59:34 President of Sales that would occasionally turn on fluorescence in a subpopulation of your resident cells you could possibly see all sorts of interesting phenomena.
15:59:42 Say that again if you occasionally turn on have some recombination dependence scheme to turn on fluorescence in your residence cells so you were labeled little clones.
15:59:52 Yeah, you might be able to see all sorts of interest I mean we are working on strains, maybe off the top you're thinking of where we create a locked.
16:00:02 Related system where we essentially we have a fluorescence initially and then we cut out a certain cassette from the genome and after that recombination event.
16:00:13 You have a different fluorescence and you turn on different things like an antibiotic resistance gene, or maybe expression of ups, so this is something we would like to start because then you can study evolution in this at least one step evolution in
16:00:30 these devices, this what you have in mind, or what I was also just thinking that turning on small colored clones without any special advantage might be a formative.
16:00:41 Right.
16:00:58 Facebook particles. Oh, but then they shouldn't divide, but if they distill divide, they said, so it's a different color but you'll have little colored rice or clones Yeah, great suggestion, that's those and video we have those, and looking at them, we
16:01:01 don't have them yet with a marker.
16:01:04 Yeah.
16:01:06 What do you know about the metabolism of this particular bug it seems to me like its capacity to enter the jamming treatment transition is somewhat dependent on its ability to grow pretty fast and the absence of say oxygen or something.
16:01:20 Why do you think oxygen. I just imagine that like in most other situations when you have a very dense population of cells of this link scale that they consume the oxygen at the top end of the at the top of the oxygen the fuses and, and also in the flying
16:01:37 God, that's another. I mean, we don't have the. At this point, the ability to grow anaerobic strains in our lab.
16:01:44 But this is from the fly got and this is an hour of X ray.
16:01:51 Yeah, I'm just I'm sort of imagining right like the part I didn't quite understand is, it seems like you're to have this feedback loop, you need to have sufficient carbon source at the bottom of the at the bottom of the channel at the very beginning,
16:02:03 so that you have enough growth in order to write in order to cross the transition point.
16:02:09 And I'm just, it seems to me like then then staying there also as a function of the antibiotic also requires that there's like enough growth in the system.
16:02:17 Other words at the growth is independent of whatever the, whatever the spatial structure is, and that strikes me as maybe particular to this bug.
16:02:25 Yeah, or particular to our device that provides you know nutrients that efficient escapism might be still so small that simply oxygen and nutrients are able to diffuse down, so.
16:02:45 Oh yeah, that's, that's another another possibility just knock out the respiratory pathway. That's what do you want me to do. I mean, the strain.
16:03:04 I've taken as a quote and give it to my students. See what happens, or we come back to you.
16:03:04 Um, yeah so let me conclude.
16:03:08 So this is obviously.
16:03:10 I mean, it's not providing evidence that this is going on in nature in any real system it's essentially showing what can happen
16:03:21 as a function of physical structure it points out that that self organization of microbes can be very sensitive to small changes of physical structure, and to the point that you know you can imagine that it acts as an ecological filter right depending
16:03:40 on the size certain microbes can grow, others cannot.
16:03:45 So I find that interesting right so there's potential for adaptation and in two ways if you're a host you can try to to shape the environment. So maybe enrich for the kind of backs you're interested in.
16:03:59 Of course microbes can adapt also to the structures that they find another point I want to make is that there seems to be a trade we see a trade off between stable colonization and how strong competition within species competition has and maybe how efficient
16:04:19 evolution is even though we really didn't track evolution here.
16:04:24 Maybe that has some, you know, maybe that can be seen other systems as well. and maybe has implications for the response to antibiotics, but of course this is Twist, twist that I mentioned, even though, if you know the selection is not very efficient
16:04:40 initially in these very crowded systems, as they uncrowded. They might slip into a state where, where it's very efficient.
16:04:51 And maybe the last point is yeah there's an extreme scale sensitivity of population dynamics and maybe it's a good idea to study evolution on multiple scales.
16:05:03 Um, I didn't want to leave you with our showing also that it can look very different. So for instance, if you look at a call I
16:05:11 can see a situation like that, where in, you kind of in the smallest devices, everything looks kind of normal as in the system as well. But then you get funky modifications and equal I becomes seems to become very elongated is kind of cleaning out these
16:05:32 cavities, as it's growing, is this something I'm not understand, it's not in our model at all, is that multilingual or not.
16:05:41 This is not multi or he's also played with multiple cases.
16:05:46 And sometimes you enrich in these devices for ones there are none model, because only those can grow.
16:05:58 Right, right. I think something like that might be going on here too. Yeah.
16:06:06 What happens if you have two speeches, with different diffusion coefficient and you mix them together would there be a one out compete in the other in the crypt, because I think right that depends on the state again in a very crowded state, I think it's
16:06:20 very difficult to form a community.
16:06:22 Because then, by chance, one is pushed out.
16:06:26 The other survives, if you, if they have a very strong mutualism.
16:06:31 This might not happen and so maybe they, they, they kind of dancing around this transition point to stay relatively dilute.
16:06:43 In this geisha state, I think, effectively like a small wellness population so communities, one should be able to form and this is something we want to look at, obviously just one population, because I kind of I'm thinking about applying this sort of
16:06:58 ideas to the gut creates right and that was partially your inspiration in the human gut, and maybe there are different depth, one is totally customizable by one species as there is only colonized by by another one dependent on their depth course that's
16:07:16 also an important feature that's not in here, that's in the scripts you often have also you have mucus Joker's makes everything very viscous that sells something we're looking into whether we can modify this to bring in some form of mucus like substance.
16:07:41 These gentlemen transitions that you were showing up, because any of this depend on the fact that your height increases.
16:07:51 You know, in an order way, if you, in other words if you sort of mess up the ordering of all of these I presume that the diffusion coefficient the coefficients of migration would sort of be different, right, would, would that affect the German transition
16:08:07 like oh between neighboring months. Yeah. No, that was an important control we scrambled up the device we have one device with just randomly select sizes and see the same.
16:08:19 Okay, so it just depends on the length of the. Yeah. because, you know, once they are colonized that kind of becoming independent of the small migration events that are happening.
16:08:35 And so, um, yeah so this is a question I'm interested in, in the future, like how does colonization community assembly, and all of that depend on scale.
16:08:47 And so, maybe a strategy to study this systematic size or shape variations.
16:08:54 And so with that, let me thank again, UUER, who was really leading this work.
16:09:04 Very talented by a physics grad student, and he's standing on the shoulder of many students and postdocs from the lab, and I'm looking for postdocs, so anybody in here are out there, interested in this.
16:09:19 Thank you.
16:09:41 Okay.
16:09:42 So given that you told us about the historicity of the way the jamming transition goes up and down. I guess I'm curious if you add the antibiotic and then drop the flow rate I would guess like as you drop the flow rate it takes a while for the jamming
16:09:54 for the jammed regions to become Anjem because you're going down the gradient versus going up right because the way that his services seem to work is, as you increase Lu start jamming faster rather than when you're decreasing Lu stage am longer than you
16:10:14 technically should. So, if you drop the flow rate while Aunty, adding the antibiotic, does it actually take longer, or does it take a different time scale over which you fix the green.
16:10:25 I mean, these go in opposite direction right so as you lower the flow, you effectively increase the size of calories that goes in the direction of becoming more crowded, bringing that into the articles in the opposite direction and then depends on what's
16:10:42 So, so, so the before you put the product in. So then what you've got is some fairly effective the one that gets there first, right, colonizes and even other ones that are identical keeps them out because of growing yeah so if you start putting in numbers
16:10:43 wrong.
16:11:04 Do does it seem like a reasonable enough so so i think i think Tommy.
16:11:10 She's not on your bed Tommy's recent work, which looked at just the genetic structure of car in our port.
16:11:22 Seems to be consistent with very strong priority effects. What is it, but if I'm asking. Okay, now you know you know how fast they grow you know the size you know, if you just put in numbers and say, Are you When do you expect that to be in a regime where
16:11:34 you get a party effect from this mechanism.
16:11:38 That's a good question. I, I haven't, I haven't done these estimates so you, what would you need, you would need growth rate, the futures of immigration right so you need immigration rights that's very hard to know.
16:11:51 But immigration is not even just immigration of course is huge than this fact, the immigration is relatively small.
16:12:00 This gives you.
16:12:02 So if you if you look at the sizes and viscosity of yeah yeah whether male the woman in nectar. I don't know whether I mean I don't know Ben if you've talked to, Ted for company ever enough but the the nectar supposed to be Lauren's party effects but
16:12:16 it is I don't know whether it's really the things that are essentially phenotypically identical, or whether the subtle differences and this one changes the environment.
16:12:25 The first one changes the environment in some way which is more compatible with itself than with a slightly different.
16:12:33 Presumably this, that certainly do experiments on ones where there's something which is labeled otherwise by saying, and it's a rich it's a very rich medium that is going into, I think, is it, is it plausible or there that we think we really would put
16:12:49 in some numbers and even back level of level to see whether it's a good suggestion or is there a question on zoom still on us go ahead.
16:12:59 You. First off, thanks for accommodating as remote participants really appreciate it.
16:13:05 I had a, I had a sort of a double question the first one. It seems that you have a symmetry in your setup.
16:13:13 Because of the direction that the flow goes, and the direction in which your wells increase in size and it wasn't. Perhaps you mentioned this in your talk about and I missed it.
16:13:25 But did you check that your pattern remains the same if you reverse that a symmetry so if the well is increased in the other direction relative to the flow and and the other question was regarding at the section slash turbulence at the at the openings
16:13:44 of the wells, it seems that you're working with the diffusion model but because of your because of your flow.
16:13:50 I can imagine that smaller wells feel this turbulence at the entrance in a different way than the than the bigger wells and, and that could also be driving the patterns.
16:14:01 Right.
16:14:02 Thanks for the, for both questions so the first one. Yes, this is important control to check whether you know when we change the direction of flow with ease patterns change they don't.
16:14:14 Again it's indicating that you know that once these, these whales are colonized further migration is not really critical for these phenomena.
16:14:26 The second one about the flow really the model was super simplistic we just modeled the bacteria, kind of underneath a ceiling of flow washing away cells right so the dynamics is here okay if you're down here you diffusing and growing but if you diffuse
16:14:43 into this layer of death when you get washed away.
16:14:48 And we don't yet.
16:14:51 Try to simulate or try to model, both of these situations because we thought, okay, a simple absorbing boundary condition on the top is taking taking care of that.
16:15:16 First proclamation you got one motor there, and it's just, just a great deal of the pressure driver. There's no turbulence of course.
16:15:19 No, it really is just given by the great depression for the for everything it's.
16:15:27 Everything is simple but higher dimensional.
16:15:34 Oh,
16:15:34 yeah, you have to you have to make a two dimensional model.
16:15:38 I agree you know you don't need the three dimension.
16:15:41 That's right, that's right.
16:15:44 It's yeah so so I think this is in reach such a model that would take into account in two dimensions.
16:15:52 Both the flow, as well as.
16:15:54 Yeah.
16:15:56 The interesting instabilities that happens if you make that larger because then you get convection roles and.
16:16:06 into the minus five typically in this.
16:16:10 Can your question was what is the Reynolds number and then Oscar said 10 to the minus five. it's very low