 the hearty quarter of the population who's held out for the last day of this meeting, it's really great to get to be here and talk about our stuff. And so I'm gonna, I also like how I think the organizers purposely, maybe it wasn't purposeful, but I think they kept a really nice percolated meeting between kind of theory talk, empirical talk, theory talk, empirical, so perhaps, you know, you can't get just like all of your excitement all at one moment, you have to kind of keep switching your brain, and I think it's actually really good to have this kind of back and forth. I'll have a little bit of math in mind, but not so much. And so I'm gonna talk, also I think about something that's a little bit at the borders of some of the other talks. We've had a fair number of talks about metabolism, and that metabolism is exciting, which I love. It's often been a rare thing to be the one in the crowd excited about metabolism. We've also heard about antibiotics and things that kill you. And now I'm gonna kind of bring those two things together and show kind of some commonalities between thinking about things like antibiotics and thinking about metabolism, because sometimes you even do things that hurt yourself. Maybe you did last night in downtown Trieste have a little too much metabolism in that sort of way. So first off, before I forget, I wanna just think, got some wonderful folks in the group. A lot of what I'm gonna talk about today was done by a postdoc, Jessica Lee, and some modeling by Siobha Shriyazi, a graduate student I share with Chris Rameen, and then little bits of things were from Tomaslav and Janelle, two other postdocs, as well as some former members of the lab. And all this, rather than talking about kind of old things, I've decided to also talk entirely about new stuff. And there are three parts to this. I'm gonna try to keep them really simple and just kind of convey a main message for a few of the parts and really get into the third one in a bit more detail. And so kind of, you know, the primo, segundo, you know, the go through a series of plates. So first, I'm gonna talk a little bit about communities, since this is a meeting about communities, and talk about toxins in the context of a community and how they can be made as some substrates are used and hurt the organism that's making them as well as go out there, could hurt other organisms, but some of them might actually be able to help the system from kind of a practical standpoint. I'm gonna zoom in and talk about a molecule that can actually sense these toxins and talks to the rest of cell physiology in a really unusual way. And then we're gonna really zoom in to the main thing I wanna try to talk about is heterogeneity and phenotypic heterogeneity and are kind of surprising, finding for at least surprising for us that that plays a critical role in terms of which of the cells are growing or not within this sort of scenario. And so, first of the organism in the compounds, so the organism is methylobacterium, and so this is a, it's in the bradyveraizobium group. It's basically a plant-associated microbe, but it happens to have kind of an unusual metabolism that can eat single-carbon compounds, and it's been a model system for about the last 50 years for this kind of lifestyle. And so things like methanol can get eaten and they get turned into formaldehyde. It's already a hint for what the bad compound decom will be. That can get oxidized through a pathway that depends upon a nice simple cofactor named tetrahydromethanopterine, try to say that. Oxidized through the formate, and that can either get burned off to CO2 or assimilated through kind of a modified amino acid cycle into biomass. And so all organisms have to do some of these reactions to make intermediates, to make amino acids, to make DNA, et cetera. This is their assimilatory main pathway, main dissimilatory pathways for growth if they're eating something like methanol, methane, et cetera. And there's lots of ways, both in terms of kind of applied aspects as well as environmental aspects, that C1 metabolism plays a critical role, both in terms of like micro-plan interactions. A lot of these are growth-promoting microbes in terms of things like oxidizing the methane that would otherwise be released into the atmosphere. And there's a whole bunch of interest in biotechnology based upon using methane or methanol as a feedstock. But I'm not gonna hit those topics, but I have to talk about it. And then of course the compound here is formaldehyde. Every single carbon when you're doing this goes through formaldehyde. This would be like if your diet was neat ethanol, right? Where then every single carbon that you ever took in went through acetaldehyde as an intermediate. This is a pretty hardcore way to go. And so this formaldehyde is produced at two millimolar a second. This would be the concentration that it would increase per second if you kept making formaldehyde but didn't use it. The briefest period of time you'd become a chip-chip experiment, right? Everything would be cross-linked to everything else. How does this not happen? And we're talking seconds, not minutes, not hours that it would take for this to occur. And so interestingly if you look at their growth and I'll come back to these aspects later, if you just throw in kind of outside formaldehyde they can take it up to about two millimolar no problem. Three millimolar begins to impart a little bit of a lag. But beyond that, they just will never grow. They don't grow. A little bit of interesting things that the board will talk about in the last part. And above about five, you get rapid death. And so this is a log scale here in terms of CFU per mill, just over a few hours in a concentration-dependent way, they die in a curve that look a lot like Bruce's work on antibiotics, for example, just they get killed. So it's definitely quite a challenge even for an organism that has this as its lifestyle. And so let me first mention the bit about communities. And so the community context, so this as well as some of the other later work was done by Jessica Lee. And a surprising place where you find C1 compounds is actually in plant biomass. There are a lot of methoxy groups in plant biomass because lignin is kind of a really complex polymer, you know, branched and non-repeating structure. And many of these different aromatic groups, they're all these OME, these ME are all methyl groups. And it's about 10% of the carbon in lignin and lignin is about say a third of a plant. And so it's actually a fairly sizable part of what's going on there. And critically for organisms that are degrading lignin and are trying to eat these different methoxylated aromatics, when they do so, take something like vanilla acid and convert it into proto-catechemic acid, this comes off as formaldehyde. And so they hurt themselves and it gets released into the media. And so it's a real challenge. The organisms universally will eat these compounds faster than those because they're making this nasty compound on the inside of themselves. And so on the inside, that's a private bad, right? But once it's outside, now it's a public bad. And any organisms in that community, whether it's a natural community or in these case kind of model practical synthetic ecologies would have to deal with that formaldehyde. And so in a project spearheaded by Daniel Segre and also involving Trent Northern, we have been looking at using a few different organisms. I won't get into all the details. We'd use organisms that degrade either the cellulose and hemicellulose or degrade the lignins, deal with this sort of aspect, feed organic acids to something like uroia to get nice and fat with a lot of lipids, make biodiesel, make DOE happy, we get funding. That's basically the cycle of things. And so the role for methylbacterium is here in terms of either being the one to take off these methoxy groups or to deal with this formaldehyde. And we've done some work in terms of this first bit but I'm not gonna talk about that today. But just I wanna briefly mention about this idea of being able to kind of purposely engineer into a synthetic ecology, a scavenger for this type of toxic molecule, the removal of a public bad. And so all I'm gonna do basically is show the kind of result that we can get from this, that if you look at growth, for example, of pseudomonaspeuta on vanilla acid alone versus in co-culture with methylbacterium, and this is formaldehyde buildup, having the methylbacterium around, not surprisingly, can dramatically prevent the formaldehyde from building up. It can speed up growth in the community. And when we go to then these higher order pairs and trios and quartets of interactions, having all the species present, including methylbacterium to do this detoxification, allows the entire community to grow much faster into higher yield. It turns out there's actually even some of the other organisms that I had on the last slide are even more sensitive to formaldehyde than either putida or exturcans are themselves. And so just wanted to make the simple points, when we think about metabolism, we often think about these excreted metabolites as being good things. pH as well as things like acetate were brought up as kind of these dual acetate can be a food, but it can also be a problem. Formaldehyde is really a problem, very strongly. And so it's a relevant challenge in this kind of system. And here's an example of having the removal of a public bad be rationale for inclusion in this kind of a system. So that's all I'm gonna talk about in terms of the community bits. So for a moment then to zoom in on how do these cells deal with formaldehyde. And so a really amazing former graduate student of mine, Dipty Nyak, she and I, we were puzzled by this fact that these guys grow at this amazing rate, two millimolar a second. And yet if you give them anything more than about two millimolar in the media, they really struggle and why is that? Could they get better? And so we decided to try to evolve them to grow and to grow on formaldehyde alone. And so just in a very simple kind of pilot experiment, just took a few populations evolved for a fairly short period of time on formaldehyde. We started on methanol and then put in some formaldehyde and kind of like Roy's morbidostat, we kept marching the formaldehyde up, up, up, up, making it harder on them every single batch. And now we were able to evolve these populations in a very short period of time to grow on 20 millimolar formaldehyde. That's the same concentration of formaldehyde that killed 10 to the seventh of the cells in two hours, now they're growing on it, right? So yes, this is possible. Amazing, what is it to, and they can actually grow on formaldehyde faster than methanol, these strains. So what did they do? We know so much about methylotrophic in this strain, had all these guesses. When we did resequencing, as is often the case, nothing that we knew about came up with any mutations and instead two loci in all three populations had different independent mutations. And of course there were genes of unknown function. Had no guess whatsoever what they were. I'm gonna ignore one of them for today. But the other, so we called these EFG, for Evolved Enhanced Formaldehyde Growth, this one really caught our eye because it is only found in methylotrophs, yet methylotrophs all over the phylogenetic tree, from alpha, beta, gamma, delta, et cetera, but even NC10, like one of these deep divisions, et cetera, only methylotrophs have this gene. And so that correlation between lifestyle and this made us feel like even though this formaldehyde selection was kind of a crazy ridiculous, just sadistic thing to do to our populations, clearly we found a part that has some relevance. We just have no idea what it is or why it would be relevant. And so in terms of very simple genetics, we could of course swap the alleles, the evolved allele in with the wild type one, or put the evolved allele into the original strain, and it causes you to either lose or gain from aldehyde growth. And so these alleles were necessary and sufficient for that to occur. And then we also made a deletion strain while we were at it, and that also allowed you to grow on formaldehyde. And so it means that these beneficial mutations were actually loss of function. And thus there's a gene, a protein there, EFGA, that actively is preventing you from growing. Why? Why would you have a protein that kind of does this sort of self restraint? What kind of scenario could that be? And one thing we saw is that if you don't have this protein, if we really hit them and with some very, very high level of formaldehyde, cells without EFGA have decreased survival. So there is some kind of ability to grow versus ability to survive sort of trade off, but I'll come back more to kind of what its purpose might be. So what does it do? In the past, this would be my entire 50 minute seminar, so I'm gonna put it here in four little panels. With a use of shimu down at rice, determine the crystal structure of this protein, because we had no idea what it was. And the structure made us realize it looks a lot like an oxidative stress sensor from some gram positives. But even though the amino acid similarity is very different, so maybe it's a sensor. Very remarkably, if you put just EFGA into E. coli, you can move the phenotypes with it. It will cause E. coli to stop growing on sublethal formaldehyde concentrations, but will increase survival of E. coli on lethal concentrations of formaldehyde all by itself, which also suggests it doesn't have kind of any particular methyl-by-caram specific partners, and must interact with something in a more general way. And from further evolution experiments, if it was fun the first time, why not do it again? We did a larger number of these. We found a whole bunch more EFGA alleles, but we also found mutations at other loci, including mutations in peptide deformalase. What the hell does the first step, in terms of cleaving off the formal group from the formal methionine of nascent peptides have anything to do with this, right? I haven't talked about the ribosome, and certainly this was not at all in our ballet week, in terms of what we were thinking about. And as it turns out, EFGA binds to peptide deformalase in a formaldehyde dependent manner. The beneficial mutations all abrogate that interaction, and this causes translation to halt instantly, faster than our positive controls. We used some various antibiotics. They took a while to make the cells stop translating. With EFGA, that was active, you add formaldehyde, translation stops. And yet you don't die. Whereas if you add something like canomycin, it takes a while for it to stop, and death happens rather quickly. So it was really fascinating. And so why would you have this, right? What's the advantage of stopping translation that doesn't seem like that's a good idea, right? It's like, please stop paying me my salary. Like, why would I ever want to say that, right? What could possibly be a rationale for this? And our first guess is it might have something to do of a growth survival trade-off. Maybe something like antibiotic persistence. And which is the idea that some cells phenotypically are actually tolerant to antibiotics, and they can switch back and forth between growing cells that are sensitive and non-growing cells that are resistant. And you see these sorts of kind of classical dynamics. Almost all of those mechanisms of persistence involve the ribosome. And so we thought perhaps EFGA is doing this sort of thing that maybe it creates some sort of persistent-like state. And so if you add formaldehyde and look at the kinetics of death, sorry, this is a little small, this is three hours, and this is a log scale again, going from like 10 to the eighth to non-detectable, you see killing that's kind of tri-phasic. A little bit of a lag, faster death, slower death. So it looks like persistence. There's this second tail. However, EFGA had nothing to do with it. You get this in either wild type or the strain lacking it. So it's neat, something else is causing this, but that's not what EFGA is doing. So unfortunately Roye's gone today, so I don't get to again give him credit for his really great paper that made us think about this direction. He showed that translational inhibition can actually speed up growth. Now you know, who of you amongst, you don't have to raise your hands, but thinks that that sounds like a crazy thought, right? How could you grow faster if I slowed down my own translation? So they studied antibiotic suppression. And antibiotic suppression is the idea that one antibiotic can actually help you grow on another. And it turns out this only really happens, and Bruce can correct me if I've gotten some of the details wrong. Only really happens between one class and another class, and it's only one direction. Only adding a translational inhibitor, like Spearomycin will save you, your MIC increases on a replication inhibitor, like Trimethoprim. And neat thing about it is it's generic. You can use essentially any replication inhibitor and any translational inhibitor, and you will see this sort of effect. And you can see this, like with growth rate in the Z-axis, you can go above the MIC on Trimethoprim, and as you march across levels of Spearomycin, you go from being not growing to faster, faster, faster, to the best you can do and back down. And this is, like I said, is stereotypical. Now, why is this? It's because the cells, even if they're not replicating, E. coli is actually pretty dumb when it comes to their ribosomes. They will continue to translate and they become longer and longer and longer. And so, with any replication inhibitor, you get these elongated cells, and here's a distribution of cell length, it doesn't matter which one you use, whereas if you use a translation inhibitor, of course, they're not translating, they don't get bigger, but they don't always keep dividing, right? The opposite is true. They're smart enough to know not to just divide every 30 minutes no matter what's happened, but they're not smart enough to stop translating if they're not replicating, and thus they elongate. And you can see this massive elevation in protein per cell. And the idea is that this unchecked translation itself is bad, right? Imagine you can translate a rate 10 and you can replicate a rate 10 and it's fine. You're in a steady state, you're happy. If your replication is now inhibited to five, but your translation won't slow down, they're both positive values, but you're not at steady state. You violated FBA assumptions and a whole lot of other things and you're out of whack. It's actually better for you to slow that translation down to match. So at least you can grow even though it's at a slower rate. Now, obviously, if you slow it down more, then you're not gonna grow at all either. And that's exactly what they see with these sort of dynamics. And so the hypothesis was, even though E. coli never evolved to kind of, these at least particular drugs that we're using that we hit at them in this sort of way, that EFGA may be functioning like an internal antibiotic, so as to say, that does exactly this, that tunes the cell's translation to the level of damage being created by formaldehyde. If formaldehyde behaves relatively like one of these replication inhibiting drugs. And so I'm not gonna get into all of the various aspects about that. We've seen sensitivities to these various sorts of drugs, swap between the different genotypes. We've seen that EFGA, this protein, can suppress trimethyprim. So we get the exact same sort of chart as over here. So our protein can work just like something like canomycin. And then probably the neatest evidence that we have, this was worked with Andreas Vasdekis, a physicist at University of Idaho, looked at, did some image cytometry to look at what happened to the cells. And we wanted to see if you could stop this dysregulation of cell size. Does EFGA actually stop the cells from elongating if formaldehyde ever caused them to elongate the first case? And sure enough, it's a little bit hard to see, but formaldehyde does create these elongated cells that look just like we added trimethyprim or something. And you see this large increase in kind of the median cell size and a huge variation cell length, et cetera. If they have EFGA, this is even in E. coli, it stops. They can make themselves, they have restraint, they don't translate in the presence of this formaldehyde damage. And so these early experiments are consistent with the idea that EFGA functions kind of analogous to a translation inhibiting antibiotic. And so the thought then is, we know, this was the world that I thought was fun and this is in a textbook and I know it's going on in the cells, but that's housekeeping. And other people, of course, study translation and they know metabolism happens because it makes the amino acids that you use to translate, but you kind of ignore that. And so we knew that there was a connection from the point of view of damage, right? Our metabolite is gonna damage all sorts of things, DNA, proteins, RNA, all sorts of things are gonna go on. But what we didn't realize is that there was this sort of linkage might be possible in terms of something that would actually directly sense this and directly modulate translation. And to our knowledge, this is the only case in a bacteria where a metabolite is sensed and it directly modulates translational activity. If anyone knows a counter example, please tell me, actually at the coffee break because I'm gonna be gone by the time Bruce is done talking. So by any questions or thoughts, I'd love to hear it, but it has to be over coffee, no, don't have lunch. You'll have to take off, I'm afraid. And so we're certainly pushing to understand much more in terms of exactly what's going on here, how exactly does this cause translation to stop? What's the broad cellular response, et cetera. One thing I wanna mention, there are EFGA homologs in almost every single bacterium. So if you study a microbe, which most of you do here, then chances are you might have a homologue of this. Some of these we've found to be encoded in operons that make aldehydes. And there's actually a lot of things that when they get used, make an aldehyde as an intermediate. So this actually may be a relatively more generic kind of interaction than previously appreciated. And the last thing I'll say too is we've seen because it behaves kind of like canamycin, you get metabolism drug interactions just like drug-drug interactions. So giving a little pre-dose of formaldehyde massively sensitizes the cells to some antibiotics and saves their butt against other antibiotics. So what metabolism is going on in our gut may greatly affect the sensitivity of antibiotics of the same bugs that have, if they have these sorts of systems going on. All right, so the third part is about the populations and finding some heterogeneity. And so I think it's also a lesson about, you only see what you look for and the kind of method that you use. And so we had been doing these growth experiments and we felt really good about ourselves because we were doing really nice fine kind of titration of formaldehyde and we got nice pretty data like these, et cetera. We saw these really long lags. You go to like four millimolar formaldehyde right at kind of the board of what they can take and not take, we thought, wow, look, there's this really long lag and it may be hard to see but this is like two days before they start growing. It's not a lag. It's not really a lag. This is CFU per mill, exponential death and then exponential increase. So they're not just all sitting there. There's massive death happening and then massive regrowth happening. And so we wanna understand what is causing these cells to change really rapidly from this death phase to this growth phase. And so the first easiest hypothesis, especially given EFGA mutations where a loss of function, these are just mutants, right? And it takes you about a day or two. Perhaps, you know, if you extrapolate this down, we're just getting a high number of EFGA knockout mutants or something and up they go. Short answer is no. There are no mutations and we can take these cells, grow them up in some other media, re-test them, you get exactly the same thing again. All right, so it's not genetics. Well, so then what? We thought maybe they're changing their environment. This has also been a theme of the meeting, right? Organisms change their environment. Maybe during this death phase formaldehyde's above some threshold that they can take and we've also heard about threshold tipping points, things like that a lot. If they can get below some threshold between net death and net growth, before everybody's dead, your population survives. If you can't get there before everyone's dead, well then the whole population's dead. And so this seemed like a very reasonable scenario. And in fact, we went through the exercise of trying to really kind of model this out. Don't look at any of the equations, even if you want to, because I'm gonna tell you this is wrong, but I will mention that this was work of CIVOSH and like I said, co-advised by Chris Vermean and another faculty at Idaho. And so we made a model of what was going on and from this model, we could even felt really good about being able to make some quasi steady state assumptions, look at kind of phase planes and null clients and all sorts of funds and groovy things. And our model fit the data really nicely. Now mind you, what's the data? An exponential down and exponential up. It's not that hard to fit an exponential down and exponential up, but this all looked great. Until you do an experiment to ask about this sort of question, what is the basis for why cells would go from death to growth? Is that formaldehyde was going down in the media. And when you measure formaldehyde, it has not yet gone down. This whole transition happens long before when formaldehyde begins to go down. So they are not making their world better on a time scale that's causing this thing. So it looked great, the math was great, wrong reason entirely. So okay, the world's not gotten better and they haven't evolved. They have no genetic change. Then what about phenotypic differentiation? So back to this idea of something like persistence. So persistence like I said is this idea that some cells in the population randomly switch into growing essentially at no speed at all. And it's exactly because they're not growing that they are resistant to things like ampicillin. And then at some rate, they can switch back on to growth. And so that phenotypically, some number can survive a brief enough insult and come back and restart. Now the thing though about persistence is growth is reasonably by stable. Like typically you have the growers and a small proportion of non growers that rare and they're exactly resistant because they're not growing. This clearly can't be that because if it's involved, I need to explain the growth. Cause they're still growing while the antibiotics here. This switching back with persistence happens after you've taken the antibiotic away. And so the thought though is that maybe phenotypically, even though we see what looks like a lag, a lot of the cells may be just sitting there in terms of an optical density standpoint. And there's a rare set of phenotypes that they're able to grow they're different in some phenotypic way and they begin to outnumber and it's not until they reach a high enough number that you see the things take off again. If it was by stable, so let's imagine that you could have some like a persister but like a growing persister. And let's imagine it's one in a thousand and instead of only being able to take up to two millimolar you could take up to five. If we were to plot tolerance to formaldehyde phenotypic tolerance, you know, nearly all of our cells would be tolerant up to two then we'd have this kind of discrete class that could take it up to five and then that's it, right? This is what this kind of data would look like if this was discreet to, you know, bimodal distribution sort of scenario. What we see instead is a continuous distribution that fewer and fewer and fewer and fewer cells can make it to higher and higher and higher and higher and higher levels. So rather than just having kind of like some really discreet class and a lot of phenotypic heterogeneity that's been studied, I think because of ascertainment bias has been really discreet differences. Persisters are really different thing growers. Spoilation is really different thing growth. You know, competence is really different thing growth, you know, et cetera. And this more like kind of subtle quantitative variation phenotypically is often, I think, a little bit less followed out. And so we see instead, like I said, this kind of continuous distribution of fewer and fewer cells that can actually grow under the exact same conditions that are killing most of their neighbors. How is that possible? Yeah, Carol. Very easily. You have a series of plates with more and more and more formaldehyde and so you take them and you just plate. And you know, then how many do you get at, you know, on zero? How many do you get on one? How many do you get on? Yeah. Yes, you do. It seems to be totally, or massively fast genetic. So I mean, I will say, with a star, right? It could be some unknown, massively fast genetic thing that behaves on the same time scale as phenotypic. Does that sound good? Yeah. Yeah, I think, I would too for now. But frankly, they're gonna behave very similarly to each other. It would be easier if it was genetic because then we could sequence. We have something we can do to get right at it. All right, so back to the model. And so we wanted to ask, you know, if this was the case, what would kind of then the consequences be? Could we understand this from this perspective? And so we had a model that looks at kind of the use of the methanol and the formaldehyde that are present in the media across all the types of cells. And then the reason why this looks all nasty with these integrals and stuff like that is we're now using a PDE to explain the cells where we're following the cells through time and as well as in terms of X where X is the tolerance state. And so we have kind of this continuous rainbow of different tolerances that are around and that tolerance plays into three different parts. The growth rate seems to be relatively independent of the tolerance, so it doesn't show up there. Death, we're having, we can do different sorts of functions. This is kind of just a simple one. If your tolerance is above the formaldehyde in the world, then you don't die. And if it's below, you die proportional to what the concentration is. And then it also shows up in terms of the tolerance itself being able to change. And so this is basically a diffusion and an infection types of terms in terms of this tolerance. Whatever gene expression and such, we can get to kind of what might be going on here in these populations. So this sort of heterogeneity model, again, it can certainly reproduce fairly well the sorts of dynamics that we see including these rapid death followed by regrowth. But the critical thing is it gives us the insight what should we be looking for? And so at the beginning, we're saying that we've got this almost like exponential distribution of the phenotypic states. That I had plotted on a log scale before, so it was a straight line. This is not on a log scale. By the time the populations at Nadeer, then there should be very few, this is in total CFU, left of the sensitive types. But the rare tail of phenotypic types have been enriched as well as you got kind of continued diffusion off of that point. So we should see here kind of a mixture. And by the time this population is growing back up, they should all be in this kind of, this higher phenotypic state. But if you go a long period of time after where formaldehyde's not around, it should relax back to the original standing distribution. And indeed, this is what we see. So if we now do a dynamic experiment of exactly what Carol was asking, here we're just showing the zeroes, twos, and fours. So plate counts at zero, two, and four. Initially, you have this distribution, those that could take four were one in 50,000, something like that. And within a short period of time, kind of at the dip of viability, all the cells are of type four. You get the exact same counts on four, two, and zero. Every single cell there that could ever make a colony is able to make it even with four millimolar. And then that persists up. And then this goes away quickly. And so the sorts of questions that Bruce asked, not gonna show all the data for it right now, but if you take them immediately after recovery and you do the experiment again, they just grow right off. If you take them, you put them in succinate for a small period of time and grow them up, retest goes back to the original experiment. But, and if you do it at like short, soon after succinate, they're like intermediate. And so there's a time scale of a few generations that they have that keep some of this inheritance, but it will relax back to this initial state. This fully enriched population, because there's so many of them, you actually see some that can resist six and resist seven. So there is like an extended tail even further that would have just been too rare to have seen in the first population. And like I said, they rapidly lose this tolerance, but not instantly, if you take, put them in media now without formaldehyde around. And so, I'm making this case that it's phenotypic heterogeneity and it's rare cells that are actually doing this. There is an alternative hypothesis. And maybe some of you watch this show even all around the world. Anybody recognize this show? Right? So with zombies and the living dead and whatnot. So what if these cells are just partly dying? And then here they're undying, unpartly dying, right? And it may sound silly, like what do I mean by partly dead? Microbial ecologists have a word for this, right? Viable but not culturable, right? That they're there, they may be metabolically active, they just won't make a colony. Like asking them to make a colony is really, like that's a hard task. And we often forget that not every cell that's alive will necessarily make a colony, right? The plating efficiencies may change. And so if they basically have lost their plating efficiency, most of it, and then regain their plating efficiency, that could also explain this sort of dynamic. And thus this recovery phase is just the zombies going back to the regular humans. And so how do I really know that they're rare? And until yesterday I could only show one type of data for to be able to get at this. So there's an excellent assay that I learned about here three years ago. Matthias Heinemann gave an excellent talk about transitions and used exactly this assay. And thankfully I actually remembered that much later and we could give it a shot. So he added a membrane dye that interpolates into membranes but won't go cell to cell. Like you dye and the D-Y-E dye the cells and then you take away that and then you start growing them. Each time they grow, you have twice as many cells that are half as bright. And then you have four times as many cells that are as quarter as bright, et cetera. And so if it's just all the cells start growing at the same time, you'll just have this nice unimodal distribution fluorescence measured by the flow cytometry, nothing surprising. However, imagine the majority of the cells taking the dye and they never grow at all. So they're just gonna stay as fluorescent as they are. And instead there's a minority of cells that are somehow phenotypically different. They're gonna start to lose their membrane dye long before you can ever see them in the population if they were initially rare enough. So by the time you see them, they're gonna be a little bump in the distribution far out to the side and the original distribution will just get outnumbered and those new cells will rise up in number and also I continue to kind of finish diluting out the membrane dye that they have. So it's a really lovely assay for being able to, in a simple way, look at phenotypic heterogeneity. And so this is what it looks like for just a plain experiment. Just grow cells on methanol from time zero to 53 hours, they were all bright. They became less, less, less, less, less. Nothing very exciting at all, right? But for this experiment, we see exactly what I was mentioning. The original time point, this fluorescence stays basically exactly the same and just gets lost and instead you see this little bump that rises, like you can see this blue trace, right? And rises and rises and as it's rising, it's actually continuing to dilute the dye out but you can see it popped up all on its own. It was a rare small number of cells that were actually able to make this sort of transition. And we're saying, you know, our numbers seem to be something on the one in 10,000. If it's harder than that, even one in a million cells that actually allow the population to survive and yet it's just a one in a million phenotypic. And then just yesterday, I finally got images from Andreas Fasteckis. We've been doing this with using automated microscopy and this just shows one particular field. And so you put cells down just on methanol and essentially every cell will grow up and make a nice little colony and have a wonderful little life for themselves. However, if you put this two and a half million molar formaldehyde, only a very rare number and it's exactly the same numbers. It was about 2% of the cells here. And we also had an estimate from the plating of 2%. And so direct observation in this. And the other nice thing about direct observation that you never get with plating is what happened to the cells that didn't make a colony? Did they like have so much formaldehyde damage, they managed like a couple of doublings and then just kind of ground to a halt? Did they at least try to extend their cells and maybe get all messed up in their morphology? Turns out they do absolutely nothing. Those that do not make colonies because their translation is probably halted, they never change the way they look at all. And so we get that kind of information from this sort of assay that we never would have gotten from the dye assay, of course. And then our last approach for quantifying is we're gonna use a barcoding technique, put in a population of barcodes and see how many barcodes survive various sorts of bottlenecks. And I think that'll be even the most quantitative for being able to hit really high stress levels. And so what I think is really fun about this is I think it's really shows that kind of a minority of cells can be incredibly important for what a population is doing. And in this case, it was really just these very, very, very few, these ones that can make a colony that end up making this sort of bump that allow these cells to be able to navigate these sort of transitions. Pankaj alerted me too, there was a paper about cancer cells and anti-cancer drugs where this sort of phenomenon has even been worked out mechanistically, where there's a minority of cancer cells that have an expression pattern that allows them to resist a higher level of drug than the rest. That's the kind of direction that we wanna go in terms of, of course, understanding what the hell's going on here, what are presumably either the regulatory circuits or DNA methylation patterns or something, like what is the basis, of course, of being able to create this heterogeneity. Yeah, Wenying. Oh, that's a high, this is a four. The experiment we did was at only two and a half percent. It's a challenge, unfortunately, to directly observe them. You want it to be rare, but not too rare. So we can't hit them as hard as we want to because we're mean. And so like, I'd love to see them at four, but I can't ask Andreas to look at 100,000 cells to find the five that are growing. So we had to back down for that. EFGA, yeah, we're making a construct for that. So like, yeah, the ideal is to, so the nice thing is we can enrich phenotypically for a population that we can even just use bulk transcriptomics or proteomics and look, what's different? How are they different, how are they similar or different to being genetically resistant, the Delta EFGA strain, for example? And we can try to understand, like, what's our marker for this? Once we have a marker, then we make a fluorescent reporter go and look in single cells. Because I want to know, two minutes before we hit them with formaldehyde, what made you, what can I predict if you're going to survive or not, right? That's where we want to go with this. But right now we don't, so EFGA is guilty just, all right, like just by association with these sorts of things. And I think it's possible, or it may be more complicated, it may be the tails of the dystopic cells that have abnormally low EFGA and abnormally high peptide deformalase. Because in the cell, you know, it's like, you know, it's like a predator prey sort of thing. Like they have to, EFGA has to stop all the peptide deformalases from working, probably. And so like that ratio might be, who knows? Yeah, great question. And I think last kind of soapbox point is, you know, in terms of thinking about how common this heterogeneity is, so this first paper from Matthias Heinemann's group was using the extraordinarily exotic organism, Escherichia coli, and the transition from glucose to acetate, right? And saw in fact, there's a huge long lag and like 100,000 cells will make the transition from glucose to acetate. This doesn't sound like the hardest thing in the world to do. And yet it's very, very rare. Showed up as a lag, but it's not really a lag. It was some cells were doing it and some cells never will and they have excellent work in terms of understanding why that was true. And the thing that's really kind of exciting or demoralizing, however you want to think about it, is myself and others have done a lot of work studying this transition, for example, from succinate to methanol and looking at fluxes and transcripts and proteins and metabolites and blah, blah, blah, blah. It turns out only about a quarter of the cells go through that transition. So during those, the eight hours of transition, what have we really been studying? We're studying three quarters of the cells that are going nowhere and one quarter of the cells, they're not in lag, they're growing. And so it's really, it's the linear interpolation then in terms of one quarter growing cells and three quarters of whatever. And we've been, we see various things spike early on and like, oh, these are the genes important for the transition. No, that's actually what like, you're never gonna grow again, looks like. And so I think the way we interpret and think about things is massively different, even in relatively more mundane sorts of transitions. And so, again, just a number of excellent collaborators both at Idaho, Yusuf Shremu in particular on this project, Daniel on other projects, et cetera, thankfully some people who give us money and happy to take any questions.