 works and the idea is that fireworks need to like finish with a crescendo of the best ones need to be at the end and so basically this workshop is like this and so the best ones are at the end I think and so here is one of the two at the end uh no then there is another talk but this was not in the original schedule but here we have Jana and Olivia in this session who are presenting some of the most I think innovative work on Plasmid that I'm aware of so Jana up to you wow thank you uh with such an introduction I I hope not to disappoint and to go out with a bang um just to confirm you can see my screen the the proper version not the uh presenter view okay okay and let's see if we can do this different thing and again same problem okay and now it should be okay perfect okay oh perfect yeah so today I'll present some work we've done to estimate Plasmid conjugation rates and really my original plan was to um present this paper that we recently published in the journal Plasmid on methods to estimate Plasmid conjugation rates from in vitro liquid culture meetings in this work we compared several existing methods both on simulations and experimental data showed that there were some issues with some of these existing methods proposed improved methods and also tools that people can use to to estimate these rates easily I will still present this work but based on some of the discussions over the past few days I thought I would have a slightly more extended introduction on why to care about conjugation rates in the first place I think Fernando de la Cruz said that it's primarily the modelers that care about conjugation rates yes we do I think for good reasons but I think we should all care about conjugation rates also and then in the discussion more what do these conjugation rates in vitro mean for other systems or how can we translate them to in vivo or environmental scales how do they relate also to what we get from uh genomic studies and this is more a few ideas for discussion and it would be nice if we still have time at the end of the session to go into that so I probably don't need to convince any of you that plasmids are interesting and one of the big reasons why they're interesting is because at least the conjugative ones can spread both horizontally and vertically and by spreading horizontally they can spread genes genetic material into strained backgrounds in different environments where these genes weren't present before and this is primarily interesting for antibiotic resistance here in the context of this workshop but also a whole range of other virulence genes and things like that and this is often seen as being in a sort of trade-off with their vertical transmission so horizontal transmission tends to be costly because the bacteria have to build the the pilus they are yeah they open them up themselves up to predation by phage and things like that and so you see different sort of lifestyle modes developed for certain plasmids that co-evolve more with their hosts that develop a lower cost and perhaps also lower transfer or those that are really promiscuous they spread far and wide in many different hosts but perhaps are less stably maintained and yeah I think we've seen a bit of both talks this week people that are really interested in sort of the growth part of things so the interactions between plasmids and their hosts perhaps other mobile genetic elements within the same host what this means for the cost that these plasmids confer to the host and also aspects relating to selection for the genes born by the plasmids themselves as well as some talks more focusing on the infectious transfer and I think what I would like to argue here is that to really get a good idea of plasmid biology on the one hand but but also to have tools to predict sort of the epidemiology so the spread of plasmids in different environments we need to be able to distinguish the spread of a plasmid that is due to this horizontal component from aspects that are more vertical so transmission of the plasmids together with its hosts into different environments essentially so that's why I think we need really good estimates of these plasmid conjugation rates and here are just a few questions that one could be interested in answering in an experimental system where I think it's important to again distinguish effects of environmental factors on growth versus on conjugation so one is for instance how yeah sub-lethal doses of antibiotics might affect plasmid transfer this is a question that's been studied before and you can imagine that the antibiotics of course inhibit growth of some of the strains involved in transfer but they might also actively act on the conjugation machinery essentially and the transcription thereof and things like this then there's other questions about what makes certain plasmids very successful is it their low cost or their high transfer what is high transfer in the first place how do we compare transfer across different plasmids across different environments and with that can we also predict which plasmids are the most problematic for spread so with that how do people normally estimate plasmid transfer rates there's certain parts that are quite conserved let's say the main idea behind sort of the standard protocol is you take your donor and recipient bacteria you throw them together in whatever media you're interested in using you incubate them for some time during which all the bacteria grow and donors recipients meet each other they conjugate the trans-conjugants grow they also conjugate with recipients and so on and so on and then after a certain amount of time you stop this process and you somehow count the final population sizes whether that is by plating on different selective media or if they're fluorescence labeled or so you can attract them in a different way and this is kind of the conserved but this is also where the mess starts because now you have these final population sizes and what do you do with that how does this tell you anything about the underlying dynamics of plasmid transfer and so as I said the first thing we did in this paper is to sort of catalog the different methods that are out there that people have used to try to estimate a plasmid transfer from such mating cultures and this is just the first half of the table where I show some of these methods and you can already see sort of the simplest methods in a way are simple enumerations of the amount of trans-conjugants you find at the end of the experiment against other bacteria involved in transfer so for instance trans-conjugants over the donors or you could say out of the recipient population which fraction carries the plasmid and then there's other methods that are based more actively on understanding of the population dynamic models either in a heuristic way or a more direct way as we'll show in just a few slides so what we did here was to simulate certain population dynamics that we know and then compare the performance of several of these different methods on those simulated dynamics and see if we can recover sort of the truth that we put into it so here I'll show you a few results on a scenario where all the bacterial populations have a plasmid that conjugates at exactly the same rate but in my different treatments I get an effect on the growth rates of different strains that are involved so green in a way is the control donors recipients and trans-conjugants all have the same growth rate orange is now a situation in which the plasmid confers a cost to the bacterium that carries it so the donors and the trans-conjugant strains and the purple is a scenario in which the recipient grows faster than the donor so both recipients and trans-conjugants grow faster for instance because it's a different species and if you now look after either four or eight hours with this trans-conjugant per donor method you immediately see that we get qualitatively different results so first of all we see that the time when we stop the experiment matters so after eight hours we get higher conjugation proficiency than after four simply because there's more time for the trans-conjugants to grow which maybe within a single experiment is not so much of a problem but becomes a problem if you want to compare results across the literature and obtained by different labs and for different incubation periods and things like that without the growth rates mentioned of the strains involved you cannot really compare these things and measure them back the second thing that we see and I think that's even more problematic here is that we would now infer we would draw the conclusion that our plasma is transferring at a higher rate to the faster recipient and this is caused by the fact that clonal growth and horizontal transfer are confounded in this particular scenario because the recipient or the trans-conjugants are growing faster after the initial transfer into these bacteria we just get more resulting trans-conjugants at the end of the line and this is being seen as some higher level of conjugation proficiency. Luckily this was already seen as a problem back in the 90s by Lorna Simonsen at all who developed a method based on considerations of the population dynamics so they effectively modeled what's happening in such a mating culture both these conjugation processes and growth processes going on and they devised what they call this endpoint formula with which you can actually get the conjugation rate given information on the population growth rate of all of the strains together as well as measurements of the final population sizes at the end of the experiment. So this is already really good because now there's no more dependence on when you exactly terminate the experiment but we and there's also some other advantages like it doesn't depend on the initial densities of the strains as strongly anymore but what we still see and it's a bit hard to see on this depiction is that there's a slight difference depending on our growth rates because effectively this method is taking into account one combined growth rate for all of the strains involved so that's one assumption that we relaxed simply by allowing all of the different strains involved in this process to have their own individual growth rates and that means that you need to measure those individual growth rates that's additional work for an experimentalist but with that you also get more precise estimates and we finally recover here that in each of these treatments we actually had a plasmid conjugating at exactly the same rate and very similar to the Simonson method we basically derived what we call an endpoint formula so you only need to measure sort of at the beginning and at the end of the experiment and you don't need measurements all through time. So growth is one thing that can happen in these experiments but another thing that might be confounding are measurements of the conjugation rate from these donors to the recipients is that at some point in the experiment trans conjugants are there in such appreciable numbers that they also start contributing to these conjugation events and they might be conjugating at a different rate than the original donors in fact in other work together with Fabian Benz and Erik Bakeren we found that the donor identity strongly affects the rate of plasmid transfer and in this case we had a first generation conjugation experiment as we call it from clinical donors into a recipient strain and then we took the trans conjugants resulting from that experiment and mated them together again with the same recipient strain in what we call a second generation conjugation experiment and we found that for this particular strain that difference was two orders of magnitude and now this is not a difference to transitority repression but really due to differences in strain identity and yeah you can think for instance of different restriction modification systems and so on that are suddenly the same now between the new donor and the new recipient. So just to get back to our method to estimate plasmid conjugation rates we don't actually remove this problem we cannot say okay this is the donor conjugation rate estimate and this is the conjugation rate estimate from trans conjugants but what we did manage to do is derive within which time window you should measure the first experiment so that it's not troubled yet by the contribution of trans conjugants to this conjugation process because initially they are at very low concentration as they are arriving in the mating experiment and it takes a while until they start dominating what is visible in these cultures. So we developed an R package and a shiny app so that's a web application that allows people to estimate conjugation rates on their experimental data. The idea behind it is that it should be as easy as possible to use I really hope that's the case but please please try to use this if these are experiments that you carry out and let me know what you think of yeah the way this is usable and so on just send me an email or add an issue to this github page it would be very nice to know how it's being experienced by those that that are actually using it. So all of that was to say something about plasmid conjugation rates in vitro but what does this actually mean for dynamics in more complex settings because as we've discussed also for the past few days in vitro you're always looking at just one very specific situation and in the environment maybe the plasmids are experiencing a very different surroundings and dynamics and it's I think important to look at what the rates of plasmid transfer might be at these environmental levels because this is exactly where we would want to intervene and do something about surveillance first of all like in the hospital which plasmids could be problematic which plasmids are spreading our antibiotic resistance from one type of bacteria to the other but also which environments might be introducing resistance or actually sort of absorbing resistance and how can we stop or like target those environments where where it's needed the most. So I don't have any answers to how to do this the best but I thought I would at least mention a few different ideas in this direction that we can then discuss further. So one thing is the translation of in vitro rates estimated for a very particular plasmid maybe a very particular donor and recipient pair as well from in vitro to in vivo models and again in this study with Fabian Benz and Erik Bakeren we did exactly that we took two plasmids that we had previously conjugated in vitro and put them into a mouse and we found that qualitatively we saw the same plasmid performing well in the mouse as we saw also in vitro and performing at a higher rate basically than the plasmid that was doing worse also in vitro but we did not actually map the rates onto each other because one problem for instance in doing this is that we don't know whether the mouse guts represents a well mixed liquid environment it probably does not so there might be contributions of plasmid transfer on surfaces which we know for some plasmids happens at a different rates than it does in liquid for instance and also a lot of the meeting dynamics so that the kinetics basically of which bacteria find each other and have the opportunity to mate is likely different in the guts than it is in flasks or test tubes so this is something where I think modeling can help in the future to try to describe those settings more realistically additionally there's evidence from other experimental systems that this translation from in vitro to in vivo on a qualitative level doesn't always work so this is from Lofty Eaton et al where they studied transmission in zebrafish and they actually found different conjugation partners playing a role in vivo than in vitro simply because of different abundances so I think in their in vitro experiments they didn't see one bacteria at all that was there in sort of low prevalence in vivo and actually played a large role in the onward transmission or maintenance of this plasmid also because it was the host where the plasmid could persist better over a longer time and I think that's another key difference in translating in vitro results where often conjugation experiments take 24 hours or maybe if you have slow growing strains two to four days or so versus in vivo we're really talking more on the scale of weeks to months of this plasmid being present in a particular environment and also being maintained in a certain strain then secondly there's this evidence from meta analysis by shepherd at all of just the sheer range of rates conjugation rates that plasmids occupy and here this is all as measured with the Simmons and Endpoint method so in principle it should be comparable methodologically and they find a plasmid transfer rates in vitro again that's sprung from 10 to the minus 16 to 10 to the minus 8 and more specifically for a single plasmid that's been used in many studies or one they find data points that span again from 10 to the minus 18 to 10 to the minus 6 which is crazy it's unheard of 10 to the minus 12 or 12 orders of magnitude difference for a single biological entity across different experiments and I think we really need to find out as a community what is causing this is it just differences in experimental setup like mixing and again interaction possibilities between conjugating partners or is this all caused by environmental factors and temperature and things like this biotic factors and knowing better when a plasmid is at the bottom or the top of its possible range is going to be really important for predicting spread also again in more complex environments yeah and lastly this is just because we have a lot of people interested in genomics here if we're thinking about genomics and the presence of a particular plasmid in a particular strain and what that is telling us about the plasmid transfer rate I think we're actually talking about something very different than when we're approaching it sort of bottom up through in vitro measurements of plasmid transfer and I would argue in a way that what we're observing through sequencing is maybe bringing us back to this very beginning point of both growth and selection in particular environments as well as plasmid transfer of course we need the transfer to initially establish into a strain but then the fact that it can actually be observed there and be present in high enough numbers to be picked up by some random sequencing study is probably because it's at least partially selected for and I think this is also something that we could potentially even target with specific sequencing studies and so on whether we can try to disentangle some of these effects more maybe over shorter time scales or where also the phylogenetic tree of the chromosome of the bacteria that they're in has a lot of temporal resolution and the plasmid is just sort of jumping in and out and then you could start saying more about plasmid transfer rates so with that I'm at the end I really hope I didn't go too far over time I lost track of where we were I showed you this paper that we wrote to estimate plasmid transfer rates in vitro I hope I could also trigger some thoughts and discussion about what we need to do to translate these rates into more complex settings and that would be something that I would love to to discuss with you more didn't do this all alone I would like to thank my supervisors Tanya and Sebastian I recently finished my phd with them I'm also looking for a post-op position so if anyone knows anything please do let me know and I would be happy to answer questions wow that was amazing so any questions for Yana I think there are questions now so what about we let Olivia talk because I don't have like questions on the talk but I map for the discussion but maybe it's better if we let Olivia talk and then we discuss I don't know Olivia are you here I also see a question from Cornelia and is there a question for yeah okay well it's also a quick question so your transfer experiments they're always done in broth or were they done also in service meetings because it's a question of mixing because in my lab we typically do surface meetings because they are more relevant for soil systems or plant systems but you use liquid yeah yes exactly so this is all for liquid media and there's this nice paper that shows that under very very particular conditions I think if you have really high density liquid and you put it on a filter for a very short amount of time that it can still be roughly approximated but there's there's no real good measures that I currently know of to do this on filters and that's exactly part of this this discussion that I think we should have is how we can create measures that do take into account some of this spatial structure for instance this different interaction indeed and maybe relate some of the transfer rates we measure on filters to those that you would get yeah in in liquids for instance thanks