 So, good morning everybody, my name is Theo Javel, I'm working in Basel in the laboratory of X-Fan Invasion and today I will talk to you about my PhD work that consisted in studying the response of E. coli to carbon starvation at the senior cell level. So what happens when we put bacteria in the flask of fresh nutrients, well it's been well established and we have seen a couple of examples of that last week, that bacteria will start growing exponentially consuming the nutrients and once those nutrients are exhausted they will reach what is called the stationary phase where the number is stagnating and if we leave them long enough in this phase they will start dying. And as we've seen last week most of what we know quantitatively about bacterial gene expression has actually been established on exponentially growing cells and arguably one of the reasons for that is that it's very easy in exponential growth to put the cells in a reproducible steady state. But if you think of it and we have had a glimpse of that in the talk right now, in most environments nutrients are limited so we actually expect bacteria to spend at least a substantial part of their lifetime in a state of starvation for some nutrients. We also expect that the selective pressure associated with starvation have therefore shaped the evolutionary story of bacteria and it's fundamental for understanding it to understand what's happening during starvation. Stationary phase is in a lab proxy for this state of starvation because the cells stop growing due to the exhaustion of the nutrients as I was explaining. Given the fact that the bacteria seems not to be active and seems to not change their numbers we can ask okay can bacteria do anything at all in starvation or is it simply dormant and waiting for new nutrients to come in order to regrow. In 2014 it's been shown by Geffen and colleagues that bacteria are actually able to sustain gene expression during starvation. To do so they use genetically modified E. coli carrying a transcriptional reporter where a promoter of interest is driving the expression of a fluorescent protein. By following the fluorescent signal emitted by the cell they were able to follow the ability of the cells to express genes. So they put those bacteria in a flask of fresh nutrients as I explained before let them grow exponentially until they reach stationary phase. They ask what is the production rate of the fluorescent protein a long time and they could see that as expected as bacteria enters stationary phase this production rate is dropping massively due to the exhaustion of energy. But the surprising result is that instead of dropping to zero this production rate was dropping to about 10% of this exponential phase value and remain relatively stable for an extended period of time. The authors call this constant activity in stationary phase and this really signals the fact that bacteria are not dormant in stationary phase but are able to keep on expressing genes even though they are not growing anymore. Now it's been reported many times that the concentration of various gene products in star cells is fundamentally different from exponentially growing ones and this results in very different phenotypes of the exponentially growing cells compared to the star cells. Star cells are notoriously associated with for example relevant traits such as increased tolerance to a broad range of stresses going from osmotic stresses to antibiotic passing by oxidative stresses and phages and is therefore fundamental to try to understand what sets the concentration of the gene product to understand this phenotype inside the star cells. One fundamental difference between those cells and their exponentially growing counterpart is the fact that the turnover of protein is now greatly limited by the fact that the dilution stopped entirely because the cells are gross arrested. So we expect that star cells are not able to change the protein very efficiently. So the question we want to answer is actually what determines quantitatively the protein concentration inside the cells during starvation. Starvation is notoriously associated with heterogeneity and so we want to answer this question at the single cell level in order to understand how this variability can be characterized. To do that we use a microfluid device called the mother machine. We've seen a couple of examples of that last week where individual bacteria that you can see as those black rods are growing into dead end gross channels that are those white lines and we can feed them whatever nutrient we want through the main channel. In our experiment we used about like in my work I was studying about 20 promoters of E. Coli and this was made possible by the fact that our microfluid chip is multiplexed so we can study up to eight strains in parallel and all of those strains are carrying a different transcriptional reporter for a different promoter. Now we use time-lapse fluorescence microscopy to obtain a long time, a time-time resolution images about the phase contrast that will inform us about the morphology of the cell, their length typically and images about the fluorescent that is an information about their gene expression. I don't have the time to enter into much details but we use a sophisticated pre-analysis pipeline that allows us to recover quantitative measurements out of these images. By that I mean that we recover information about lengths and fluorescence inside individual cells where the measurement noise has been removed so that we have a very precise estimate of what is the actual length of the cell and the fluorescent content. On the top plot here you can see an example of individual cell traces in terms of fluorescence and we also get a reliable estimate of instantaneous rate. Instantaneous rates are the instantaneous gross rate at the time of the measurement and fluorescent production rate so the volumic production rate that is the amount of protein that is produced per unit time per unit volume and this last rate is a proxy for the activity of the promoter we are interested in and you have an example of these individual cell traces of this rate on the lower plot here. In the stationary phase the population is constant in that mono curve. Now that could be because every cell is not growing or because cells are growing but also dying. Yes. The numbers, right? So your traces show cells to be growing. So no. What about death? So this is not in stationary phase. This is in exponential phase. Sorry. Oh this is in exponential phase. Yeah, because basically typically what we do is we switch the cells from exponential gross to stationary phase and I will come to that. Sorry. Yeah, this is just an example of the way we treat our data but this is exponentially growing cells. Yes. But so my first question is that in the stationary phase is it that cells are growing and also dying? So no. That's why the number is constant? Yes. Or is it that cells are not growing? So cells are entering gross arrest. This is one of the results. Within our time frame we observe no gross and no substantial death of the cells. Just wanted to add one thing. If you call it generally once glucose runs out, for example, it takes one or two days before you start seeing viability going down. So when you, it takes like a long time before cells start actually dying. So within this time scales, yeah, I think it's just like cells not growing. All right. So as I said, we want to study starvation using this device and what we typically do is we connect to the microfluidic chip, a media containing glucose, M9 plus 0, 2% glucose, and a media containing no carbon sources that is simply M9 not supplemented with any carbon source. Using a pressure controller and a feature of our microfluidic device, we are able to switch very rapidly from M9 glucose where the cells are growing exponentially to M9 0 where the cells will be starved for carbon. And what we observe is that the cells enter very quickly a gross arrest. So on this plot you can see the gross rate and the average gross rate across the population is the black line and it shows that the cells in general at time zero where it's the moment of the switch are entering gross arrest. And you can see that individual cell traces that are showed as these thin blue lines and the ribbon corresponding to the standard deviation across the mean shows that individual cells are also entering gross arrest. So there is an homogeneous gross arrest that is very abrupt at the moment of the switch. So now that I've showed you that we are able to follow individual cell and their gene expression a long time as we switch them from exponential gross to carbon starvation, I can come back to the question that we want to answer and show you what determines the present concentration inside the cell during starvation. So we primarily follow the volumic production rate that is a proxy for the activity of the promoter we're interested in. And out of the 20 promoters that I've studied today I will show you only three of them that are covering the spectrum of behavior we observe for all the promoters we studied in our work. On this plot you can see the first example for RPLN that is ribosomal promoter and on those plots the thick colorful line corresponds to the average across the population while once again the thin colorful line corresponds to individual cell traces. Here I show the volumetric production rate normalized by its value in exponential phase and I'm showing you 20 hours of starvation out of the typically 60 hours that we perform simply because most of the dynamic is happening at the entry into starvation. In the case of promoters such as RPLN the volumic production rate is sharply decreasing at the time zero and then remain relatively stable around zero for the rest of the stationary phase. On the other end of the spectrum we have promoters such as BOLA that is controlled by RPOS. RPOS is an alternative sigma factor that is implicated in the response to starvation and stress. And as expected for those promoters we see an increased activity but remarkably this peak of activity at time zero is then followed by an exponential decrease of the production rate, so the promoter activity, towards zero. Finally we have promoters such as HSLV that is a heat shock promoter in this case where we don't see an up regulation or down regulation at the entry into starvation but we simply see that the production rate is bettering out. Now since most of the dynamic seems to happen at the entry into starvation we were wondering if we consider only simplified dynamic at the entry into starvation are we able to predict what would be the concentration resulting from this production activity. And concentration is the interesting parameter here because that's what will drive the phenotype of the cell ultimately. So we are wondering how those difference of behavior are translating into differences of concentration. So here I will show you a very simple model where we consider the interplay between production the simplified production dynamic, dilution and degradation that are like coming together to lead to a certain concentration. In our experiments the dilution is dropping to zero instantaneously so we model it like that. The degradation we can measure experimentally for fluorescent protein. I don't have the time to explain it here but it's not going to be different for different promoters so we really expect that actually the difference will be led by the difference in terms of production activity. In the simplest case that corresponds to our ribosomal promoter case the production rate drops to zero at the same time as the dilution so the promoter simply stops being active and in this case we expect the relative concentration to actually go down under the effect of the degradation. Now simply introducing a gradual decrease similar to what we've seen for the promoters at Petor's app this heat rock promoter is enough to produce an expected increase into the relative concentration and this is due to the fact that even though the dilution has stopped the production keeps lingering on for a while and therefore the concentration is expected to increase. Now if we add a peak of production at the entry into starvation such as we've seen for about an hour deep we've modulated further the expected concentration. And so here it's really fundamental to realize that it's the delay between the arrest of growth and of the production that shapes the concentration in the cells. And so of course yes dilution stops so that means the turnover of protein will be diminished but the fact that the dilution stop also offers the cell the opportunity to change dramatically the concentration of certain gene product even though there is no upregulation of the promoter. Sorry just to be sure so but in the previous plot so I should expect that for longer time scales of course the concentration should drop. So given that the degradation rate in our that we measure is exponentially decreasing we actually expect that it sort of stabilize and also like yes in this model since the production drops to zero you would still expect that it sort of decreases but like at very long time scale. Because the degradation goes to zero. Yeah I like sort of yes it exponentially decreases. So we basically use inducible promoter so we keep the production of the fluorescent protein then we let the cell enter starvation and stop the induction of the protein so they start with a certain pool and then we measure very at very long time dynamic the fluorescence and see it decreasing. So that we neglect the photo bleaching effect because we measure very rarely and we are able to measure the level of the protein. Now when we look at our to come back to our experimental data when we integrate the production along time along with the concentration that we measure the degradation that we measure exponentially we indeed observe that the concentration of the for the three different promoter is behaving as we expected for a more very simplistic model. So that really means that the dynamic at the entry into starvation is really important in setting this concentration and what we were surprised by is the fact that it seems that individual cells as can be seen by the standard deviation like the ribbon around the mean seems to behave quite similarly and more similarly than to in the other cases of nutrient switches or something like that that we observe in the lab. So we wanted to quantify how homogeneous the cell are responding to starvation and to do that we can look at a metric that quantifies the amount of variance that is explained by the fluctuation of the average behavior. So I will explain that with this cartoon on the top of the cartoon you can see that we have individual cell traces that are those seen lines and all cells are considered acting independently. In this case looking at the variance across all the observations or across cell and time we can see that it's much bigger than the variance of the average behavior across the population. On the contrary if all cells are behaving exactly the same we expect that the average behavior across the population recapitulates the average the behavior of individual cells and the fraction of the variance of the total observation that will be explained by the variance of the population average will be big. Now when we look at individual cell traces of concentration for different promoters in the case of RPLN we see that before times zero it looks like indeed the cells are fluctuating a lot crossing each other and after we switch them to starvation it looks more coherent. This is even more true for HSLV and BOLA where we look at individual cell traces it really looks like after we switch them to starvation they respond very homogeneously both in terms of timing and in terms of amplitude to starvation. And it seems that most of the variability or most of the dynamic is most of the action is happening early on in starvation so we want to ask how homogenous is the response of the cells or how homogenous is the behavior of the cell at different time points in our experiment. So we can divide the total span of our experiment into three different time windows what's happening during the exponential growth, what's happening during the early starvation that we fix at 10 hours because this is an arbitrary threshold but that is based on the fact that most of the changes we observe are happening early on and what's happening during late starvation that are the remaining 50 hours of starvation. And when we look at this fraction of explained variance we see that in all case the blue bar the corresponding to the transition is substantially higher than what's happening during growth and during late starvation and this is particularly the case for HSLV and BOLA. So this really says that the cells individual cells are acting much more homogeneously during this transition than they would during growth or during late starvation in terms of this promoter activity. So this really sort of like shows that the response to this abrupt growth arrest is maybe a programmed response. Now I've shown you that like most of the dynamic is happening early on in starvation but if you remember one of the point I started with is that in the literature there was some report of constant activity during starvation. So this means that we have to decipher okay what fraction of the phenotype of the cell is dominated by sort of basically what is the importance of this early production dynamic. And can we decipher whether this plays a big role in the fact in the phenotype of the cell or whether the residual promoter activity is actually dominating because over 50 hours of residual activity maybe it doesn't really matter that the cells have been very active at the entry into starvation. So once again what we can do is divide our time window into three different time periods what's happening during growth what's happening during early starvation and what's happening during late starvation in terms of production. And at every time point we can ask out of the pool the total pool of protein present in the cell for reporter during starvation what fraction come from which production phase. So we expect that at time zero the 100% of the protein will come from what has been coming from growth because we didn't start considering production that production that is happening in starvation. Until time 10 hours we don't expect any green to be visible because we consider early starvation to span from zero to 10 hours and after 10 hours we will start considering the protein produced in the late starvation so the remaining 50 hours. So in the case of LPLN you can see that across time the red sector is largely dominating and this is not surprising because the production drops almost instantaneously to zero. So the state of the cell at any time point in starvation is dominated by what the cells have been inheriting from their exponential growth phase. Now if we go from to our promoter that is petering out you can see that the blue sector is taking over very quickly at the entry into starvation. So this means that the phenotype of the cell like that they had in exponential growth is quickly forgotten and quickly replaced by a new phenotype that is set by the early production events. And after 60 hours of starvation you see that the blue sector still dominates the fraction of the protein so it means that after 60 hours of starvation the phenotype is still dominated by the early production events. And this is even more true when we look at promoters such as Bollard at Displayopik where here the blue sector at the end of the starvation is largely dominating the pool of protein. So this really tells us that the phenotypes of the cells are set early during starvation and are kept for a very long time. And so a way to interpret that is when cells are running out of food they have limited resources and limited time to act and change their phenotype. And they will do so very quickly and after that they will be set in the state they managed to reach for a very long time. But of course the question that we can ask is okay does that matter at all for anything? And so like if what we think is correct if the phenotypes are indeed really set early and that relevant traits are set early we could ask, okay what is the consequence of perturbing this gene expression on some relevant traits? And one of the relevant traits that I was mentioning in the introduction is the fact that star cells are globally more tolerant to a various range of stresses. So we decided to expose the cell to an oxidative stress, yes? Hey Drew, is this the plot that you just showed? How much production contributes to the protein pools? Can you say something about how much do you think an active remodeling would contribute to the protein pools, you know active degradation and then making proteins out of? So what we would expect is for example if the cell were able to regrow now if I would switch them back to exponential phase and ask the same question, I would expect that due to dilution the turnover would be very quick and that like the phenotype will be reset very quickly, right? But the thing is protein degradation has no chance to reach this amount of turnover because it consumes energy. Most of the proteases in E. coli are ATP dependent and in the absence of carbon source the cells are just unable to do it. So in the literature there are some reports of active degradation at first and we also observe that actually on our fluorescent protein. The degradation is faster at the entry into starvation and then starts decreasing. So it would increase the turnover and we would expect to have a faster replacement but practically we don't think it's really happening somehow. So in your experiment in order to induce the starvation you just switch the media from glucose to nothing basically to minimal medium. So what's your intuition about where do the cells take the resources in order to express the other protein? I mean through degrading the protein that are already there through some medium that maybe it's still around or? So we don't expect that there is medium still around because the switch is very, very quick and actually the fact that the cells and our growth are so quickly is showing us that we replace the media very quickly. So we would rather bet on something like a bit of recycling of internal components although this is limited by the protein degradation. And also potentially some reserves in the literature there are some reports of glycerol stocks that are very short lasting and stuff like that. So I would expect the internal pool of the cell to play this role but we actually don't know. So in your experiments you shift very quickly right to the carbon free medium? Yes. So how much of this very quick response determines this initial condition dependency in some sense? So if you were to shift it more slowly I guess you would see less of this effect. So we also conducted these experiments by using batch cultures where we flow batch cultures through the chip and the cells are sort of then growing in this limited nutrients and enter progressively stationary phase. And so the cell in the chip are experiencing the same sort of progressive change of media and we see a remarkably similar pattern in terms of gene expression. So it's actually I think it would be interesting to push it in even slower growth and so on. But for a standard like starvation stationary phase entry we actually observe the same thing. Okay thank you. What is the mechanism of very rapid activation of RPOS? Is there an anti-sigma factor for this sigma factor? So how can it you know again I know that the sigma factors are usually sequestered with anti-sigma factors and the purpose of this is to do everything much faster than you would do if you just relied on transcriptional regulation. So what is what is going on here? The main up regulation is linked to the fact that RPOS is during exponential phase constantly degraded and this degradation stops when the ATP gets depleted. So the pool of RPOS is actually piling up. On top of that you have a tiny bit of transcriptional up regulation and you have the activity of a protein called CRL that will actually sort of help RPOS to win the competition for the RNA polymerase, RNA polymerase versus sigma 70. So you have like a couple of different mechanisms that are actually pushing RPOS activity to shoot up very quickly. But the main one is really like that you stop degrading it and so suddenly sort of the constant expression is getting bigger. Okay and a related question again or maybe not so related. So here maybe not but I kind of got confused so here in the left panel you see virtually no degradation so everything stays roughly constant. And in the previous few slides you were showing that you measured rapid degradation. So what what is the difference between? So what I've seen what I've shown in the previous slide is sort of basically is that like the production activity is dropping right? And so that's why here you don't replace the pool of protein whereas here you sort of like your protein production keeps on lingering on for a while and that's why you start reshuffling the while you start replacing basically adding up some protein and sort of diluting the other ones out basically. So you replace what you produce but you degrade what you produce and you don't degrade what you don't produce. But in a couple of slides ago you had something which I interpreted the overall proton white degradation but maybe I was wrong. Yeah something like well there was some experimental data you said that you measured it. What we measure is the degradation of the of our fluorescent protein and this is what drives RPLN concentration here to go down on the left. And this is an effect that also touches on HSLV and BOLA but because those are producing more than the degrade you see the concentration increasing. But in the case of RPLN because the production drops very sharply the degradation is actually contributing to reduce its concentration. Yes so does that matter at all and what we can do is test that by submitting the cells to an oxidative stress. And asking okay now if we start perturbing the gene expression in various ways how does this affect the relevant trait that we're interested in. Here tolerance to oxidative stress. So the first condition we use is not perturbing anything at all. We simply switch the cells from M9 glucose where they grow exponentially to starvation just as I did up till now. And then submit the cells after 20 hours of starvation to 5 hours of oxidative stress. And switch them back to fresh media in order to ask what fraction of the cell is able to regrow after those 5 hours of oxidative stress. And this is a measure of viability how many cells are able to regrow within 20 hours in M9 glucose. When we do that and we follow the BOLA reporter as previously shown we observe the concentration is ramping up very quickly and is relatively stable until the point where we hit them with the stress. And when we ask okay what fraction of the cell are able to regrow after the oxidative stress we see the fraction of surviving is almost 100% in our cases. So like we use a rather mild oxidative stress when the cells are let to express gene as normally. Now the second condition we can use now is okay if we kill our POS we expect that actually the starvation and stress response will be killed as well. Because this is the main sigma factor implicated in this so it's kind of like the extreme where we prevent the cells from having this stress and starvation response. And we can submit them to the same treatment. When we follow the concentration of our reporter we see that as expected it remains extremely low during the entirety of the starvation. And now if we ask what fraction of the cell is able to regrow we see that almost none is able to survive this oxidative stress. So this really highlights the fact that our POS as was reported before is fundamental in this process and we believe that part of this is due to the fact that it allows cells to kick up some genes during starvation. But if you pay attention you'll see that during exponential growth there is already a much lower level of the concentration or like a lower level of the concentration of the reporter. So this means that the RPS knocked out strains are not able to upregulate upon the entry into starvation but they also start with a lower level. So probably this response that we observed is a combination of not being able to increase the level but also starting with a lower level. And what we're really interested in is what would happen if we simply prevent the cells from expressing during starvation. So we use chlorophenical in order to inhibit gene expression only during starvation. So we let the cells express normally during their exponential growth and as we switch them to starvation we also put them into chlorophenical that is expected to reduce their ability to express genes. Following our fluorescent reporter here we see that we don't get a complete inhibition but we still decrease substantially the level of expression of these reporters. So this shows that the chlorophenical condition is not killing gene expression entirely but is dramatically reducing it. Now if we ask what fraction of the cells are able to regrow we see that we get about 20% so it means that compared to the situation without perturbation preventing the cells from expressing their gene normally during starvation is decreasing 80% their ability to survive an oxidative stress that is concurrent with starvation. So this shows that like expressing gene is important during starvation for a relevant train. But my main point was to say that actually what matters really to set the phenotype is the early production events. So now we can ask okay now if we let the cells express normally during the first five hours of starvation would this change anything we would expect from what I explained earlier that it would actually allow the cells to recover quite a bit of their viability because they would be able to set their phenotype early on and then expressing or not expressing wouldn't make such a big difference. What we see is that the concentration of our fluorescent reporter is indistinguishable from the concentration without perturbation which was expected because we've seen that most of the dynamic in terms of gene expression was happening early on. And now looking at the fraction of the cell that is able to survive we see that we recover an extra 50% of viability compared to inhibiting gene expression during the entirety of starvation. So this really shows that having early gene expression being allowed to express only during five hours is really allowing to recover like 70% compared to the RPS knockout. So this really tells us that this early gene expression as we were hypothesizing before is fundamental to set the state of the cell and then the cells will keep this state for a very long time during starvation. And so it's fundamentally different from the exponentially growing case where we can look at the cell, check how much it expressed or what is the concentration of the protein at the time t and because of the fast turnover we get an information about the gene expression status and like it's not really required to know the full history of the cell instead of exponential growth to understand its current state. In stationary phase we think that it's really fundamental and everything is played very early on, all the action is happening early on and then the cells are frozen in this state and they have to deal with whatever comes with what they were able to do early on. So I will summarize with this. I showed that the delay between production arrest and growth arrest is resulting in a rapid reorganization of the protein across a single E. coli cell. I've then showed that the phenotype is then set early on in starvation and is kept for a very long time and that this early expression is critical for relevant phenotypes such as the survival to concurrent stress. With this I would like to thank the Sanimvegan group and especially the people that have been directly implicated on this project and I would also like to thank you for your attention and I'm happy to take any questions. Thanks a lot. We have time for a couple of questions. So maybe I missed it, but so you measure degradation rates. Do you observe any specific time course of degradation from exponential phase to the shift until long term observation? So we are not able to measure it with our method in exponential phase. But it also matters much less because the dilution rate will set the rate of the turnover of the protein. But because we are relying on the fact of being able to sort of shine light once and then 10 hours later on the cell and measure the level of protein, we wouldn't be able to do that on exponentially growing cells because obviously this is not the same cell that we can observe across time. So in the way I'm doing it, I'm just assuming that the starting level in starvation is the level of degradation in exponential phase, but it's anyway a negligible effect compared to the dilution rate. Sure, sure. So you wouldn't be able to tell whether cells increase their degradation rates from exponential phase to whether they activated. No, that we are not able to tell. Yes. So the transition into starvation phase, the processes that are happening to the protein, do you think it's a program that once it's triggered it will just run until the end? Or do you think it would be stopped and adapted when in that very moment nutrients become available? So I think that the RPOS activity is a program response, but I think part of it is also coming from the dilution rate that stops, right? And if you would put them back into fresh nutrients, I expect like in the middle of this transition phase, I actually expect that it would start to regrow quickly and that all the sort of processes that are responsible of like... Well, basically you start returning over your protein very quickly and your phenotype is able to recover very fast. On the other hand, if we let the cell long enough in starvation, then you see a very long lag phase before they start regrowing and the duration of the lag phase has been reported to be a function of the duration of the starvation. But if you let the cell in one hour during carbon starvation and you switch them back to glucose, they are able to regrow without a lag basically. So maybe there could be some kind of feed forward loop trigger, like making this asymmetric process, like that switching on and off are different, yeah? Yes, yeah. So my question is exactly about this lag which you mentioned in answer in the previous question. So do you have any intuition of what sets the duration of it? Because here's where I'm confused. So the main message of your talk is that first five hours or maybe 10 hours are critical and then nothing changes after this. And that means that there should be no duration or no influence on duration on the lag if you stay 20 hours or 30 hours or whatever. Yet you just said that the duration of the stationary phase is very variable. So just tell me what is how do you understand this lag on reentry to the exponential growth? So it's a fair point. What I'm mostly focusing on here is sort of the phenotype that is relevant for starvation. But of course, as you mentioned, the longer we let them installation, the more they will struggle to get the right phenotype when they have to regrow. I think this could be due to several things. Like you have this ribosome, a hibernation system that starts coming in. And also, we also know that if we let the cells long enough in starvation, they will eventually start dying. They cannot sustain this state for a long time. So they accumulate some damages and that they have an accumulation of misfolded protein and components that are toxic for the cell. And then they get sort of, they have to get rid of that and put things on their feet again before being able to regrow. So this is sort of, I would say, this is sort of an aimed phenotype in the sense like it feels like the cells start kicking in. So lots of responses, osmotic stress and so on, that they might not primarily need in starvation, but where they wouldn't have the time to adapt in case these kind of stresses would come later on. So some kind of preventive response, if you wish. And another aspect is the thing that are going to degrade anyway because in the absence of energy, the cells are not able to sustain for a very long time. Okay, so we have one last question. So I was wondering, you showed for three different promoters kind of the dynamics. And now then in the end, if I understood correctly, I mean you showed the knockout, the behavior for the knockout of RPOS, which I guess is not surprising. And then the inhibition of most, how would you say, of most processes. So I was wondering, did you take any more, do you can comment anything on take home messages from the different promoters that you studied? So it looks like if we see which promoters are upregulated and which are downregulated, typically our ribosomal promoters are following the trend that I showed for the ribosomal promoters. They sharply go down and sort of this is also looked like an active process, an active sort of downregulation. This is in general true for the promoters that we would associate with sort of exponential growth, whatever it means, but like fast growth, such as ribosomes and so on. It feels like this is an active process of saying we don't want that. Also what I find interesting in the case of the ribosomal promoter is that they usually consume a lot of resources. It's like very costly to express ribosome. And so downregulating those very quickly is actually also allowing the sales to use the pool of resources that remains to express a lot of other things. And because the dilution stops, with this limited pool of resources, they're still able to do a lot somehow. That is my vision of the way why these promoters are getting downregulated. Then I think the very important promoters such as Bola and the one that our RPS regulated, they get kicked up very high so that they get a very high expression. And you also have some promoters like this HSLV that is sort of simply pattering out but even this decreasing activity is enough to increase the concentration and to understand really what sets exactly sort of the level of the peak and sort of the time dynamic of this decrease of production activity is not clear yet. But yeah, that would be definitely an interesting question to answer. Okay, let's thank the speaker again. Okay, so we move upstairs for coffee.