 All right. Can you see my screen? Perfectly well. Great. So yes, thank you very much to the organizers for putting together this super interesting program. So I'll be talking about this question of why some bacterial genes, including antibiotic resistance genes, seem to be found on plasmids rather than the chromosome. So this is all theoretical results, just one you beforehand. So the work here was motivated by this observation that there seems to be a trend in the types of genes that are present on plasmids versus the chromosome. So here, for example, I'm showing you data on the proportion of chromosomes versus plasmids. So the proportion of DNA on chromosomes versus plasmids, that code for antibiotic resistance genes. And it seems that there is indeed this effect where antibiotic resistance is overrepresented on plasmids. Of course, there's caveats to this sort of data, notably sort of biases that might exist in the databases that you use for this type of analysis. But nevertheless, there seems to be a consensus that typically certain types of genes, including antibiotic resistance, virulence factors, heavy metal tolerance, and bacterium bactericine systems are overrepresented on plasmids. And there is a considerable body of work that has considered this question. And much of the theoretical work around this idea, mainly if we could focus on this tradeoff between horizontal transfer and segregation loss. So the idea is that the advantage for a gene of being on a plasmid is that it can be horizontally transcribed. But then the disadvantage is that its plasmids are less stably inherited than the chromosome. So you get this segregation loss effect where sometimes you lose the plasmid and therefore the gene. More recently, there's been some really nice work that has looked at the effect of having multiple copies of the plasmids. And what that means for the types of genes that end up on plasmids versus the chromosome because of these genetic dominance effects that come into play when you have this type of polypody. But the work that I'll be presenting today is very firmly in this kind of first type of tradition where we're thinking of low copy number plasmids and really kind of in this framework of thinking about horizontal transfer versus segregation loss. And to sort of summarize the key ideas from this body of work, as was discussed in great and interesting detail in the previous talk, there is this idea that essential genes tend to be on the chromosome, although I was interested to hear that not always. And this is because if a gene is essential and plasmids are subject to segregation loss, then we'd expect to see very beneficial genes located on the chromosome. There is also this idea that public goods end up on plasmids because this is potentially a way to enforce cooperation. So if you have the trait that goes for the public good and it can be horizontally transferred, then you prevent the emergence of cheaters, although some of these ideas are still up to debate. And then most relevantly for antibiotic resistance, there is this idea that genes that code for locally beneficial traits end up on plasmids. This idea was proposed a while ago and then it was developed in a really lovely modeling paper by Bergstrom et al. So the premise of this work was this idea that plasmids are too costly and not transmissible enough to persist as pure parasites. So that reflected what was known about plasmid biology at the time. So under this assumption, looking at a closed population, the author's sign that any trait that is beneficial will end up on the chromosome because of this segregation loss effect. And then that leads to the loss of the plasmid. So in essence, in a closed population, beneficial traits are always on the chromosome and the plasmid is lost. And this led to this idea of the plasmid paradox. So plasmids can only persist if they carry beneficial traits, but if the traits are beneficial, they go on the chromosome. So how do plasmids happen? And the solution that these authors propose is that if we consider a locally beneficial trait and an infallux of cells that firstly don't have that trait and secondly don't have that plasmid, then the advantage of being able to pass the trait onto the incoming cells via horizontal gene transfer allows the trait to be maintained on the plasmid and then that allows the plasmid to be maintained itself in the population. And hence this idea that plasmids encode for locally beneficial traits. However, this work really was premised on the assumption that plasmids have too high a cost and too low a transfer rate to persist as parasites. And more recently there's lots of emerging evidence that makes this assumption quite questionable. So firstly, transfer rates appear to be higher than what was measured at the time, in particular transfer rates in biofilms are higher than in liquid cultures. The fitness cost of plasmids is known to decrease through compensatory evolution and there is experimental evidence suggesting that parasitic plasmids can indeed persist. And finally, there is also gymonomic evidence showing that the same plasmid backbone can exist with variable gene content. So taken together all this evidence suggests that there is a need to revisit the question of why certain genes are on plasmids taking into account this the presence of this gene free version of the plasmid. And so this is the work that I'm presenting here. And now I'm shifting and I'll be talking specifically about antibiotic resistance. But note that the results also hold for other genes with similar properties. So I'm developing here a model where I look at competition between six different cell types. So the large circle represents the chromosome. So this chromosome can either carry the resistance gene marked with R or not in which case the chromosome is considered sensitive. And then the cells can either have a resistant version of the plasmid here shown in small circles and no plasmid at all or a sensitive version of the plasmid. And so I model multiple processes here. I'm not going to go into the details of these processes because the results that I present are robust to how I'm modeling them. But essentially I'm looking at a logistic growth model. So there is replication and cell death and some sort of carrying capacity. The plasmids can transmit between cells and can be loss three segregation loss. And then if there is if the cell carries no resistance, it is subject to an additional death rate from antibiotic exposure. And we also assume that both the resistance gene and the plasmid have a fitness cost. So if you have either version of the plasmid, you have a cost that is just associated with the plasmid backbone. And if you have the resistance, you, chromosomal resistance, you are subject to a fitness cost. If you have plasmid-borne resistance, you're subject to the fitness cost, which we assume is the same as the fitness cost of the chromosomal version. And then importantly, if you have both chromosomal and plasmid versions of the resistance gene, you are paying both fitness costs. And then the question that we are interested in answering is, when is plasmid-borne versus chromosomal resistance evolutionary stable? So what I mean by evolutionary stable here is that can one form of resistance be invaded by the other? So here in the top case, the green version cannot be invaded by the orange. So it is stable. And in the bottom version, introducing the orange into the green population allows the orange to take over. So the green is evolutionary and stable. So here I am looking at evolutionary stability of the two forms of the resistance as a function of antibiotic pressure. So this is the additional depth rate for the sensitive cells and cost of resistance. So in this white zone, I don't see any resistance. So here the population is sensitive. And then there is this very small blue zone where only plasmid-borne resistance is stable. This area is very sensitive to parameterization and also sensitive to model structure. So I won't be talking about that in detail. So the two main outcomes in this model are this orange zone where only chromosomal resistance is evolutionary stable. So in this zone, if plasmid-borne resistance is the only form of resistance available, the population will have plasmid-borne resistance. But as soon as I introduce chromosomal resistance, that will take over the population. And then once chromosomal resistance is established, it cannot be invaded by plasmid-borne resistance. So that's the orange zone. And then in this pink zone, I observe by stability. So what that means is that if chromosomal resistance has been established, it cannot be invaded by plasmid-borne resistance. But similarly, if plasmid-borne resistance has been established, it cannot be invaded by chromosomal resistance. So in this parameter space, resistance genes can either be on the chromosome or the plasmid, dependent on initial conditions. And we'll talk about that in more detail in a little bit. So to understand this question of chromosomal resistance parameter space versus the bistable parameter space in more detail, we can look at how changing different model parameters affects which space we're in. So I won't go into these results in detail, but to summarize, we see chromosomal resistance as the only possible outcome when the benefit from resistance is high. So when the antibiotic-associated mortality is high or the cost of resistance is low, we see it when plasmid fitness is low. So when the transmission rate is low or the cost of the plasmid is high or the segregation loss is high. And we see it when cell density is low, which essentially means there's less plasmid transfer. And then conversely, we see despistability. So either form of resistance being possible when the benefit from resistance genes is low and when the plasmid fitness is high. So I said that in this bistable zone, which form you eventually see will depend on the initial conditions. So here I am looking at exactly that. So these are the results of a simulation where I start off the population with a mix of sensitive and resistant cells. So and on the x-axis, I'm varying the frequency of the sensitive plasmid in the sensitive population. And on the y-axis, I'm varying whether the resistant population is mostly made up of a chromosomal resistance or sensitive resistance. And so there's two effects that I want to talk about here. So there's this effect that high initial frequencies of the sensitive plasmid benefit chromosomal resistance and this frequency dependence where high initial values of plasmid-borne resistance benefit plasmid-borne resistance. So to start with, let's look at the effect of the sensitive plasmid. And this is interesting because right at the beginning, we talked about this local adaptation idea being dependent on the absence of the sensitive plasmid. So now that we see this effect where the sensitive plasmid being present is advantageous for chromosomal resistance, it's worth revisiting that local adaptation hypothesis. So at the top, we're looking at the scenario that was modeled in the Bergstrom paper, where the influx of cells into this local population don't have a version of the plasmid. And now I am looking at what happens if we actually assume that these incoming cells that don't have the resistance gene do have a sensitive plasmid. So here I am showing you exactly this variation on the x-axis. So I'm modeling a local population where resistance is beneficial and then cells coming from the outside that don't have resistance. And again, here the blue corresponds to plasmid-borne resistance and the orange to chromosomal resistance as the outcome of the model. And what we see here is that indeed under the assumption where the incoming cells don't have the sensitive plasmid, then having this influx is beneficial for plasmid-borne resistance and allows the gene to be on the plasmids. But as soon as I start having plasmids in this influx, then it's actually an advantage for chromosomal resistance. So the takeaway from this is that this local adaptation explanation for why certain genes are on plasmids only works under the assumption that there isn't a gene-free version of the plasmid. So we've sort of questioned one explanation for why genes are on plasmids. So can we come up with an alternative hypothesis? And here it's interesting to look at this second effect where the eventual outcome in the model depends on the initial frequency of chromosomal versus plasmid-borne resistance. So the higher the initial frequency of one form of resistance, the higher the fitness of that form of resistance. So this is something known as positive frequency dependent selection. And the intuition for why this occurs has to do with this low fitness of cells that have the dual both forms of resistance. So to gain an intuition for this, consider a cell with chromosomal resistance that is attempting to invade a population with plasmid-borne resistance. What's going to happen to this cell is that it's going to be susceptible to invasion by a plasmid carrying the resistance gene and then it will have this dual form of resistance meaning that it will have low fitness because it has to pay this dual cost and therefore be outcompeted. So the intuition is the same for the other way around for plasmid-borne resistance trying to invade a population with chromosomal resistance. The key idea is that these dual forms of resistance will always be generated and for the rarer form of resistance, these dual resistance cells will be a larger proportion of the overall frequency of that form of resistance. So in this context, being rare is a fitness disadvantage, giving rise to positive frequency dependence. It should be ending now. You've spoken 22 minutes already. Okay, I will wrap up. So these results are very robust to specifics of model structure. They're also robust to having movement of the gene between the plasmid and the chromosome and they're robust to having fluctuating antibiotic selection instead of constant selection. The thing that these results are sensitive to is the assumption that dual resistance, so having both forms of resistance is disadvantageous. So far I've presented, I've talked about having to pay the cost of resistance twice. We can relax that a little bit so it doesn't need to be exactly twice the cost of resistance but as soon as dual resistance becomes beneficial, then the population just, those dual resistance cells outcompete the rest of the population. So this gives rise to this hypothesis that maybe resistance genes are on plasmids because they arrive on plasmids. Because once one form of resistance has had the time to be established in the population, it cannot be outcompeted by the other form of resistance. And so because transfer rates of plasmids are higher than the transfer rates of chromosome genes, one estimate is 10 to the seven times higher. We suggest that perhaps the gene location of resistance genes is explained by these higher acquisition rates. And so we've done some modeling that supports this, additional modeling that supports this but I don't have the time to go into that. So just to summarize, why are some genes on plasmids? We've seen that this local adaptation explanation works but only if we assume there is no gene but free version of the plasmids. And for moderately beneficial genes, positive frequency dependence selection means that an established form of resistance is difficult to displace. Under these circumstances gene location is determined by the acquisition rates. And so this rises, gives rise to the hypothesis that plasmid-born genes are moderately beneficial, rarely acquired than other and shared between multiple species because under those circumstances the effects of horizontal transfer are particularly important. And so our suggestion is that perhaps resistance genes are on plasmids because they fulfill these criteria. And with that I'd like to say thank you to the collaborators on this back, Iyana Huisman and Sebastian Bonhoeffer. This was published in Evolution Letters last year and then more recently we expand on some of these ideas around frequency dependence in this philosophical transactions fee, a special issue. Thanks very much for your attention. Thank you, Sonia. Are there questions for Sonia? Tal Dagan has one, right? No, she's just applauding. Okay, I can do that as well. You want to say something? Yes, I just wanted to say that this is de novo acquisition. I do not really agree with this assumption because we published 2020 a paper on ink P1 plasmids that be captured from the rhizosphere, so from the root zone of tomato and lettuce plants, and they have this exactly the same backbone as the R751, but they are free of any resistant genes. So this indicates that so they have like an integron but this out gene cassettes and they have IS 1071, but we have even some plasmids that have no acquired genes, but just IS 1071. So this indicates that the plasmids are around there and they do not necessarily carry the resistant genes, they are just captured when they are needed. Yes, I mean exactly like that's kind of why we wanted to do this work because it was the sort of previous theory that had started with this assumption that the gene free reasons of the plasmids don't exist. So we really wanted to look at because I agree with you like there is evidence that these gene free versions of plasmids are around. So we wanted to do modeling that included the possibility of having those plasmids exactly. Somebody else? I have a question or a comment I don't know and is that in my recollection of things genes are on plasmids, adaptive genes, because they are required just from time to time. So in a population only sometimes there is selection for the plasmid and then in your models it's always kind of just one environmental condition that you're modeling. But this is not real life. So I think antibiotic resistances are on plasmids because most of the time the bacteria are not challenged by the antibiotic so that they don't require and the cost of that is is unacceptable. But when there is resistance the plasmid comes. So you can have plasmids going down to say one in eight, 10 to the eight cells in a population if not required and suddenly comes up to one if there is selection. So it's how you say see that because there's no need for all this modeling you're doing. So it is certainly a good point and it is true. So if you look at the effects of fluctuating selection under some conditions that does select for genes that are present on plasmids, however under some other condition it selects for genes that are present on the chromosome. So I think that intuition is one that I share but I think it's a more complicated story. So we've been doing some work on this with a master's student but so I would say yes I agree to some extent but I think the story is more complicated and the effects are relatively minor but it's something that we're actively investigating. Okay thank you. Sam do you wanted to ask something? Yeah. Last question has to be the last question because we are late okay I'll hurry up. I've been fluctuating my hand up and down because I can't quite work out if my question makes any sense. Well first thing is what happens if you invert your thing and you say actually it's advantageous to have both resistances because there's a dosage effect and so two copies is worth more. What happens to the system? So then that is what gets selected for so yeah that's just something that we explored and if it becomes advantageous to have both resistances then the cells that with both resistances outcompete the others. So there's no bi-stability anymore? No so then it's just the ones that have both versions are stable. And okay thank you Sonia we have to go ahead and we have the last speaker before the break who is