 So next speaker is Ayari Fuentes, who is going to talk about multi-scale experimental evolution of antibiotic resistance. So thanks Ayari for being with us, even if not in person. Yeah, thank you very much, Akopo, for the invitation, and I'm really sorry that I'm not there. But well, it's a great opportunity that I can speak with you either way. So what I'm going to talk about is about a couple of different work that I do in my lab. It has to do with experimental evolution in antibiotic resistance and particularly not just evolution, but actually adaptation about antibiotic resistance in different scales. So of course, we all know that antibiotic resistance is a multi-scale problem, which means that it's a very difficult problem to tackle. So in general, we heard a lot about antibiotic resistance in terms of populations of the big population of epidemiology in the settings of hospitals and those kind of things. But antibiotic resistance has his base in the molecules, in the genes, in the cells. So all these scales actually are important to understand why it's a big problem now. So it happens not just in spatial scales, but also in temporal scales. So what we were interested in is what happened actually when like the first pulse of antibiotic resistance that a bacteria can sense. And in the other hand, what happens in evolutionary times, but evolutionary times which means like weeks or maybe months. And the first part of the talk is going to be about rapid adaptation. And by rapid adaptation, I mean we are going to use single cell multi-microfluidics and imagine bioinformatics. So what we are going to see is a bacterial population that is subjected to a pulse of antibiotics and in a very, very small populations, which actually are going to do a difference between what we're going to do a difference. And the second part of the talk is going to be about evolutionary adaptation. And I mean evolutionary adaptation because we are going, the timescale is going to be much longer. And the timescale is going to be enough for mutations to actually fix in the populations, which is not the case in the first part of the talk. It's about 100 generations and the population density is about millions or hundreds of millions. So just to start, this is the people that is going to, that was involved in this, in this work is Charlie Bruno, which is there, Alvaro Sanmigliano, which is in Spain, and Rafael Pena Miller also from the Center for Genomic Scientists here in Mexico. So the system that we are using is Escherichia coli MG655, which is the most common Escherichia coli actually, the lab Escherichia coli. And what we did is we put a multi-copy plasmid, a plasmid is just a strength of DNA that's replicated itself within, it doesn't need the bacteria machinery to replicate, it's auto-replicating. So this multi-copy plasmid has an average of between 19 and 20 copies per cell. And it's not a conjugative plasmid, which means that they stay in the same cell. And what we did is we put Betalactam resistant genes, Ablatem1 gene, and in the same, in the same promoter we introduce a GFP fluorescent marker. So what we, what you see here is a clonal population, a clonal population means that actually those are cells from the same cell. And the green is going to see what, when we see green, we are going to see the amount of plasmid that is within a cell. So the other thing that we are going, the device that we are measuring these cells is a microchemostat. A microchemostat is a microfluidic chip, which is a really small piece of a biopolymer. And we are going to control it dynamically, which means that we can actually control the environment that the cells are seeing. So in here we are going to have a very good control about the, when we put, when we put the antibiotic, about the amount of antibiotic that we are putting and the time of the antibiotic that we are putting in. We are using, all of these is actually put inside a fluorescent microscope in that, in that sense we can see it and we can measure everything and we are going to use quantitative analysis. The first video that I wanted to show is, this is one chamber, one chamber of the microchemostat, and this is a clonal population, this is a clonation from one cell. The first thing that you can notice is that you can see different amount of green, different degrees of green. It's not all green, which means that even though the population is in average 19 or 20 copters per cell, actually we have lots of heterogeneity in this clonal population. So when we are going to start seeing red, when you start seeing red, this is the pulse of antibiotic, and this antibiotic we are going to put some dye in it. So when they enter into the cell, it turns purple or red, and this means that this is dead. If you notice it, not all the cells die at the same time actually. There were some of them that lasted a little bit more. So what are the kind of things that we can measure here? So in here, the first square, the VAC one, is how we see the cells from the microscope. And then we can measure a lot of things using these bioinformatics. So we can measure the GFP, which is how green it is. We can measure the DS red, which is exactly the amount of red that there is in the environment and within the cells. We can track every cell, which means that we can track the mother, and then we can track the daughters. And we can see if they are the same color, if they are not the same color. And then we also measure the length, which is going to be very important in here. I don't know if you noticed, but we can see it actually in the DIC image that the cells are getting bigger, are getting longer and longer when the antibiotic is entered into the cell. And then we can measure division, which means that we can measure growth rates, which is important, of course. And then in this sense, what we did is we put, we use this system, we put a pulse of antibiotic of 80 minutes. And what you can see here, and in the, right, in the left part, you can see the track cells and we are there measuring which one are bigger and which one are normal and which one are dead. So as you can see, the result is very heterogeneous. There are some cells that are stressed, which means that the yellow ones are the ones that are going to get bigger. There are some that are dead and there are some that are actually normal. So what we wanted here now is to compare to a system when there is no this, there is no this heterogeneity. So what we did now is we took the bacteria, we removed the plasmid, and we introduced this gene, this beta-lactam gene, both in the chromosome to see what is the, now what is the response of this small population. But now it has exactly the same, the same resistant gene, but now in the chromosome. What we put here, the video is not very good, but you can see that, oh, I'm going to repeat it. You can see that the antibiotic enters and the response is very, very synchronous. Almost all the population turns yellow and then it turns blue again very quick. This is just all, like all the experiment put in the top of it, we have the population that has the, that has the beta-lactam gene, the mutant gene in the chromosome, and in the bottom, we have the population that has the plasmid. So what we can see here is that the straight population, when we have a, when we have a plasmid is a small amount, and when we have it in the chromosome is a big amount of the population. So this means that the population, as I already told you, is going to respond like really synchronous when they have the chromosome mutation. But this heterogeneity that the plasmid is giving does make the population respond as a heterogeneity population to the other population. So now what is the consequence of this effect in a bigger population, like a population level? So what we did next is to do a population-level experiment with a pulse of antibiotic, but now in plates, in 96-width plates. This has instead of, well, it has like really big populations, we have 200 microliters of bacteria in each of the, of that wealth. And what we see is exactly the same, that the survival probability is going to increase a lot when the, when the blood, well, when the mutation is going to be buried in the plasmid. So this multi-copy plasmid is going to increase survival to fluctuating environments. By fluctuating environments, we mean, I mean, pulse of antibiotics. So what we did here is this experiment is just, we have the populations in 96-width plates, then we do a transfer to, in different times, from half an hour to six hours. And then we remove the antibiotic and we let the population grow again. So that is the experiment on how we measure survival. And now the next step is what happened if we let one of these plasmid mutate, which is something that happens in nature. So in here, we are going to have the cell, the cell is going to have this in average 19 or 20 plasmids, and what if one of these plasmid actually mutates? And when it mutates, it's going to generate a therozygous cell. A therozygous cell, I mean, they are going to have two different mutants in the same cell, two different, well, yes, two different mutants in the same cell. And then for this, for the system that we are using, this beta-lactam system, what is going to happen is going to have another beta-lactam in the cell. So we wanted to see what's happening here. So what we see, what we did is to build another bacteria that have the same plasmid, but now there is going to have a beta-lactam-12, which is just one mutation for the beta-lactam-1, and we are going to put it in red. And now we have two different systems, one that the green one, the one that have Latin-1 and the one that have Latin-12. And by the other hand, we are going to have a third one that is going to have exactly the two of them in the same cell. So it's going to be a therozygous cell. So when we see red is a population that have Latin-12, when we see green is a population that have Latin-1, but when we see yellow, it's going to be a therozygous population. And we wanted to see why actually this therozygous population exists in nature. And here now we are going to use what we call a mother machine, which is another microfluidic device. And in here, the mother machine, what we do is we put the cells in there, the mother cell is going to stay on the top. And so we can follow actually, we can track all the daughters. So we see there this is a drug-free environment. And the first thing that we wanted to measure is if there is like a tendency of going to one of, we start with a therozygous population. So we wanted to see if at the end of the experiment in a drug-free environment, we have one more of the Latin-1 or more of the Latin-12. And we realize that we expected that this, this platelet dynamics is actually just a noise-driven process. So some of them are getting redder, are getting more red, some of them are getting greener, but it's completely random. So what happened now if we introduce pulse of antibiotics? So in this video, in the top of it, we are seeing a therozygous population, but now with a pulse of ampicillin. Ampicillin is an antibiotic that Latin-1 is resistant for. And then in the bottom one is going to be a pulse of septicidim of the same population, but septicidim is an antibiotic by which Latin-12 is resistant for. So as you can see, when we introduce the antibiotic with the ramp of antibiotic, the amount of antibiotic that we are using is very, very high. So we wanted to see if the one that we put ampicillin in shifted into greener and the other one into red. But by the end of the experiment, everything is almost dead. So we see elongation, we see stress bacteria, and then we measure actually the shift. We can see it, we can see it. If we introduce ampicillin, there is a shift towards the green and when we introduce septicidim, there is a shift towards the red. But then why is this optimal? Or why we see this in the population? Like in that sense, there are cells that have less resistant genes of one kind. So what we realize is if we shift, if we do like a fluctuation environment between the two of them, the two antibiotics with septicidim and ampicillin, this is optimal. And actually we can save the population. So in this video, what we are doing is we are putting in some time septicidim and sometimes ampicillin, but in the same amount that the previous one. So this is a lot of antibiotic, but actually shifting this antibiotic can actually stabilize the population. And we can have the population for several times. So maybe this is why it's optimal exactly. So the conclusion for the first part is that this multi-coupu plasmid acts as a platform for wrapping gene application. And this accelerates the rate of adaptation, which is the first part. And that means that the terrygenius response of the antibiotic can make that the population actually can survive longer time to this pulse of antibiotics, a really high pulse of antibiotics. And a terroplasmid is unstable in environments, which constant selection, but actually is very stable and stabilize the population in fluctuation selection. So this was interesting for us. Then the second part of the talk is going to be about evolutionary adaptation, which as I already told you is going to be like proper evolutionary experiments like with enough time, within days, weeks or months, with enough time to mutations to fit in the populations. In the other, in the previous part, mutations were already there, and we were just seeing how they act in rapid adaptation. So these are the persons who were involved in this part of the project, which is Sandra Cisneros, Lucia Graña and Dejanira Perez, all of them were from Mexico. So, to introduce this, I'm going to introduce a very simple and naive population dynamics model. So what we are going to do is first just to see how a kernel population actually adapt to antibiotics. So I'm going to, the way that we are going to model adaptation to antibiotics is going to be through a mutation. So we are going to have a wild type susceptible population, which is this BWT, and then this susceptible population are going to have two avenues in order to make it resistant. It's going to be, it could be very strong resistant at a very high cost or very mild resistant at a low cost. So this is how it's going to, they are going to see in the terms of the model. So if we increase the antibiotic concentration, of course, the susceptible one is going to die quickly. And then the very resistant one is going to die at the end of the X axis. But then in terms of bacterial density, in absence of antibiotic, of course, the susceptible one is going to grow better, but then the resistant one is the one that's going to grow last. So, well, this is just the kind of the general definition of how we are going to model these populations. And the change between those populations is going to be just a probability of a point mutation with an epsilon bigger than zero. So that's a, well, that's a usual ODE problem. Now, if we introduce a ramp of antibiotic, which means that we are going to try to model an evolutionary experiment. In this evolutionary experiment, every day we are going to change the media to French media. Every day is going to have like, again, the complete amount of resources, but we are going to start introducing antibiotic in an increasing way. We are going to call it an adaptive ramp. So if we do this in the model, what we see is that if the relative frequency between those three populations shifted like kind of quickly between all the population to be susceptible to be a little bit resistant. If the amount of antibiotic is not very high, we are going to have like in this graph that why I am showing you, we are going to have the population that is my resistant actually. And as a small part of the population is going to be very, very resistant. Now we are going to define what we are going to call the rate of adaptation, which is how fast actually they add up to the antibiotic. And it's just going to be the delta MIC. MIC is a minimal inhibitory concentration, which is the concentration at which we don't see growth anymore. And this is going to be divided by two times the time of adaptation of this MIC that we are getting. And then, as I already told you, this is the case that we are not, the increase of the antibiotic is not very high, the antibiotic which we are subjectivity. And then we are going, what happened now if we increase the antibiotic of the ramp, of this adaptive ramp. So what we expected is that exactly the amount, the population get to the same MIC bought in a shorter time. And actually, which is start to be interesting is that the population is different. No, the shape of the population is different. And if we measure actually the final population within these two cases, if we just measure resistant, this is not going to be distinguishable. The phenotype is going to be exactly the same. But if you see the composition of the population is very, very different. When we realize this, we were puzzled about what happened now, or how these different populations respond to, if we remove the antibiotic, and then if we put the antibiotic again. So, well, in this sense, the rate of adaptation as we expected is correlated with the strength of selection. If we put more antibiotic, this means that we have the strength selection is higher and the rate of adaptation is higher. But then we put the drug-free environment, we put the same amount of the drug-free environment, and what we see is that the population actually responds very different. The population in the mild selection is not going to be naive. Again, like the removal of the resistance is going to decrease in a slower rate. And in contrast with the strong population that actually within the same amount of drug-free environment, all the population is going to be naive again. And then if this population is subjected again to adaptive ramp, of course it's going to respond in a different way. So we have this very simple model, and we decided that there were some things like interesting in there. So we decided to do the actual experiment to see if this happened. Like if the population actually responded different just because they have a different composition. So we decided to do the experimental evolution adaptation and this adaptive ramp. And again, we use the same system that we were using, the Skericacoli MG65. We remove, of course, the plasmid. And here we are not using plasmid, it's just the Skericacoli, like the normal Skericacoli. And we are going to use serial transfers and ambicillin. So we are measuring every day the MICs for 22 drug concentrations, and we freeze everything. And we are going to have like two different treatments, the strong selection and the mild selection. For the strong selection, what we are going to pass every day is what we are going to call the IC90, which is the inhibitor concentration at 90%. And what we did is we measured the MIC and we transferred the population, like the next population, the last population that we see actually grow, that we detect grow. And by the mild selection, what we did is we measured the MIC and we transferred the IC50, which is the population at 50% of the inhibition. So they were two very different treatments. And we did this for every day. So that is why we call adaptive ramp, because every day we have a different MIC, which means that we have a different IC90 and a different IC50. And we did this until we reached the 10 MIC of the parental strain. So this is what we call the phase one. So what we see here in this graph is in the XY axis, we put generations and in the Y axis, we put normalized MIC. In average, both of the populations are 10 MIC, but as we can see here, the strong selection is going to be, it's going to be, to get there really fast. And actually the mild selection is going to be very noisy. It's going to be slower and it's going to be noisy. So it takes longer time. So what we see here is exactly what we saw in the model that the rate of adaptation is again correlated with the strain of selection and with the intensity of selective pressure. But again, when we, if we measure the populations, if we just measure the resistant population at the end of this experiment, they're exactly the same. Like they, we cannot distinguish within one or the other, but the evolutionary history is start to be different, of course. So we did the whole gene sequences of this part. And for the strong selection, what we see is that just three mutations were at 100% of the populations. ACR, RPOD, and CLPXXLON. So these were the only mutations that we saw. So the first thing that actually surprised us is the very few mutations that they have. And they were like completely in the whole population. These mutations actually drive this fast response to the antibiotic. And in contrast for the mild selection, there are parallelism between the replicas and we found just two mutations at 100% of the population. Like we have more mutations, which is not surprising because the amount of the length of the experiment for the mild selection were longer. So of course they were more mutations accumulated. But in there, we found more mutations within the replicas. So now that we have these two different populations, we remove the antibiotic. And we remove the antibiotic for eight days for the both treatments. And what we see here is if we measure the MIC, we measure how the, well, the first result is that both populations actually decrease their resistance in a drug-free environment. But the rate at which the resistance is decreased was very different. For the strong selection, if you can see just in few generations, they've removed it lowers a lot. And for the mild selection, actually it's quite stable and then it's removed. The resistance is lowering, but in a small rate. So also for mild selection, the resistance actually is not, well, the population is not completely naive as we kind of see in the strong selection. So this is again very similar that we have in the model. And what we see here is, I kind of already told you, is that the reduction of resistance is also negatively correlated with the level of resistance. And if they become very, resistance very fast, they decrease the resistance. And this could mean that the mutations that acquire in the strong selection were actually very, very costly. So they've removed it very fast, which is the contrary to what happened in the mild selection. So now that we have here, that we are here, we put again another ramp of antibiotic to this population and we wanted to see what was the effect of this evolutionary history in the second pulse of antibiotic. And now this is the result. As you can see, they respond exactly kind of the same. It's a bit noisy, but they respond very, very fast. The rate of adaptation is very fast. And now they follow very similar evolutionary trajectories. And now if we compare the two treatments and the two different phases of antibiotic, what we can see is that the only phase that we can actually distinguish between them is the mild selection, the first phase of the mild selection. All the orders are completely distinguishable. And then the surprise of a lot in the sense that the mild selection, the third phase of the mild selection, actually, the rate of adaptation is very fast, and they actually became even more resistant than the other one. So of course, this population is better, like it's very prepared to tackle to antibiotic, but this population never saw like very high amount of antibiotic. And this is very, well, this is, this is, this was something that surprises. So as you can see here, when we measure the rate of adaptation, the rate of the, the higher rate of adaptation is the phase three of the mild selection, even higher than any of the strong selections. And if we, if we see the relative fitness and compensation adaptation, as we can see here, the only, just the phase two of the strong selection is actually over one in the relative fitness. This means that the fitness cost of this whole population is going to remain a constant through the experiment, no. And these resistant mutations of the strong selection, they are going to be rapidly compensated in this cost of selection. And this is how the ramps, the actual ramp C, like the trip phases of the ramp. And the only, what the difference between them actually is just the phase one, in terms of the, of the length of the experiment. And now we see more accumulate mutations in the mild selection. And if we do the whole game sequencing now for the, if you compare it in the first phase in the second phase and in the third phase, what we see here is that the, for the strong selection, we see like mutations in the whole population at 100% of the population, but then they decrease. And then they, they go up again, but they are not the same mutations necessarily. So for the third phase, actually, the rate of adaptation is equally fast, but they are another, they are other mutations are not the same mutations at the first phase. In contrast, for the mind selection, the thing that was very interesting is that there were more mutations, there were more mutations that were in the, in the, in the whole population, and even though if they were not in the whole population, for the phase two actually for the, for the part of the environment, they were covered now in the 100% population and the, and the mutations actually were stable, they, they don't, they didn't decrease for phase two. And then for phase three, we have more mutations for the mutations actually is stable were maintained. It's like this, my selection of, of anti biotic kafas is like having a memory that it doesn't have this strong selection. So, just to conclude, and now this part is that the rate of adaptation is going to be proportional to the strength of selective pressure. This strength of selection is going to shape the mutation of the specter. The mutations are going to depend on the evolutionary history on how many, like the time of which the expert, well, the bacteria were subjected to antibiotic and how much antibiotic they see. And this, my selection favors the stability of these resistant mutations and evolutionary history and external selection determine the rate of resistant adaptation. And I think we think this is relevant because a lot of time when we are, when we are talking about antibiotic resistant, we are neglecting all these small amount of antibiotics that are, that are in the environment and actually we can see here that they are shaping a lot of the of the population and these, these resistant mutations are stable and in contrast with the other ones. So, we'll all together, we can see that drugs persistent adaptation of course in multiple time of scales. This might selection increase genetic diversity in the population and this accelerate resistant adaptation in fluctuation environments. And we also saw that this Plasmid copy number variability is going to increase also genetic diversity and it's going to result in the resistant population. And this is going to be important in survival with inflation environmental also. This multi copy blackmail replication is going to increase intracellular genetic diversity. So then, what we are seeing here is that if we have more diversity, more diversity, actually populations are going to be able to respond better to the antibiotics. And we also think that this is going to be relevant. If we now explore this in polymicrobial communities, which is the thing that a little bit more natural in polymicrobial communities, we are going to have very diverse population with very diverse MICs. And this is going to happen along well it's going to happen a lot in there. So, I think this is it. Thank you very much for listening to me and if you have any questions, I'm happy to take it. Thank you very much. So we have time for a few questions. So I have actually a question. So when you were showing the model and that there were the two, the week selection and the strong selection regimes. In both cases, it seemed that the weaker resistant and the strong resistant strain were coexisting. Yes, I didn't understand what is the mechanism that determines their coexistence. In phase one, just at the end of phase one. Yes, yes. Well, here is just that mutation is just a random parameter, isn't it? And just if we have a higher amount of the antibiotic, the strong mutation is going to stay in the environment longer time. So if we don't, it is just more costly. So if you don't have enough antibiotics, like in the case of my selection, you don't need this strong selection. This is the like very strong selective bacteria, but they are coexisting all the time. So we are also, these experiments are parametricized with the experimental data. So this is not in a stationary phase. No, this is just every time I stop the simulation. I don't know if you understand. Yes, in part, I'm not sure. If I see, I mean, I understand that there is a cost for the strong resistant. What I don't understand is why, I mean, there is a balance of the cost and the benefit and either the weak resistant or the strong resistance displays the other. Or it's like just because there is a balance, I mean, there is like sort of selection mutation balance where yes, it's because of that. Yes, yes, yes. But the thing that I wanted to say, this is not in a steady state. Okay, yet. So this is, if you continue the simulation, this is going to shift again. Okay, okay. Thanks. We have time for, for other questions. Okay, if not, thanks. Thank you again. Thank you very much. Thank you very much.