 Hello everyone. Is this working fine? Yes. Yeah. So thanks for the organizers for inviting me and also for all the workshop which has been amazing so far. So my presentation is in two parts. So the first part is published. So what's published in PNIS last year and was about the evolution of conjugative classmates with high transfer rates. And the second part is actually saying that a lot of what we said in the first part was not really false, but was missing a whole part of the story which I'll present here. And this also has effects on the spread of AMR. So this experimental evolution study we did was to evolve classmates with variable opportunity for horizontal transfer. And we did this by so passive and plasmid carrying cells like in standard co-evolution studies, but with treatments that added every day a variable proportion of immigrant, what we call immigrant cells, so which don't have plasmids, so which are potential hosts. And so in the absence of transfer the plasmid carrying cells would be diluted and plasmids need to transfer to subsisting populations. And the expectation we had was that with high immigration plasmids would be selected to increase the rates of transfer. So we had a no immigration treatment which is a standard treatment and then treatments with increasing amounts of immigrant cells every day up to 99% of plasmid free cells added every day. And there was no antibiotic selection during evolutions, so plasmid carrying cells could potentially go extinct. And I did this with so with E. coli and with one plasmid, so a one plasmid which is a plasmid with a very simplified map here. So it's a conjugative plasmid with different AMR genes, including the ampicillin resistant gene which we used to identify plasmid carrying cells. So this is the population dynamics that we observed in which so for 10 days roughly we had a plasmid carrying cells subsisting at high levels in the populations until after which time there was a decrease in plasmid carrying population density as measured by ampicillin resistance, especially with high immigration treatments. So I took one clone per lineage at an intermediate time point at which we could collect plasmids from each population before they were extinct and I measured their conjugative transfer rates. And so this is the results of conjugation assays. So you can see on the Y axis the transfer rate of the plasmid and the clones are ordered by immigration treatment with high immigration on the left and the black line is the transfer rate of the ancestor. So as we had predicted we saw significantly increased conjugation rates for all of the high immigration treatments. So plasmid evolved higher transfer rate. And in the next step so to characterize the mechanism for this I sent all of those evolved clones for sequencing. So Illumina sequencing and we mapped some mutations to the sequence of the ancestor. And this is the results from the sequencing analysis. So here so each circle is a plasmid genome with no immigration treatments at the center. So as you can see the was very little evolution mutations detected result in migrations which treats with the phenotypic data and most a lot of mutations with immigration treatments which are on the outside of the plot. So we saw quite a few different things here. So in gray what you see is large deletions. So for a few clones we saw large deletions which are deletions of resistance, antibiotic resistance determinants. So because plasmid evolved in the absence of antibiotics they were free to lead to to those antibiotic resistance genes. Except that we still selected clones on ampicillin to identify platelet carrying clones in the first place. So all of the clones retained the ampicillin resistance gene which is in the middle of these deletions. And so this does suggest that there is actually more deletions going on in the population and we are biasing strongly for plasmid carry before non-deleted mutants at the step before yeah to get the clones. And then we saw various plant mutations and I'll focus on two of them. So the ones on the left here in FinO are the mutations that we basically expected to get in the first place. So FinO is the major repressor of the transfer operand expression in incaved plasmids and as Fernando explained yesterday those plasmids are usually repressed and in the lab we usually we often use the repressed mutants which are have very high transfer rates and are mutated usually in FinO. But we didn't get this many of those clones so we only had five clones with this and weirdly they also didn't have very strong strongly increased transfer rate. So they didn't explain much of our data. Instead most of the mutations we observed were all clustered in a small region of the copper gene which isn't involving transfer but instead is involved in the regulation of application. And to understand what this we're doing we didn't have to do much about with quite old literature because those mutations have been very well characterized in the 90s. So what copper does is that it's a small RNA which so which hybridizes to the mRNA of frappe protein and frappe is the main is the initiator of replication and is limiting for plasmid replication and plasmid copy number and the mutations we observed were exactly the same as mutations that were observed in an important second well in one loop of the copper RNA and they prevent its finding an inhibition of frappe production. So those mutations are textbook copy number mutations and we looked at if they explained transfer rate evolution and in here I plotted across clones the clone transfer rate on the y-axis and the cop A mutation status on the x-axis and now because we now have copy number mutants we have to look at the frequency of this mutation because not all clones have fixed cop A mutation in their genome and we saw that over all there was a statistical significant correlation between them with cop A mutants having higher copy number and more over we could say that this was causal because we had a few plasmids which had only one cop A mutation and no other mutation on the plasmid and those had higher transfer rate so they do increase plasmid copy number and plasmid copy transfer rate. And so we did a few more experiments to understand this and what the way we think this works is that it is a gene dosage effect the tax in two ways so with higher copy number you have both a higher number of DNA to transfer so a higher number or it is to be recognized by the transfer machinery and then there is a gene dosage effect on the transfer machine itself so more proteins to transfer plasmids. So at this point we thought that we might also have an effect on other phenotypes because the AMR genes themselves might also have a gene dosage effect and so this has been mentioned several times in the workshop as well so it's not new so we confirmed that indeed the clones with high copy number also have higher levels of antibiotic resistance and to understand how this impacted phenotypes at the population level I went back to populations that evolved for nine days in with or without immigration and what we saw and I played with those populations on normal levels and high levels of antibiotics and I saw that clones evolved so no populations evolved without immigration so with no selection for transfer have a very low proportion of highly resistant clones whereas populations evolved with immigration so with selection for transfer have a very so the majority of the populations have a high proportion of highly resistant clones so this is what we published and so at this point we understood to this as a copy of selection because of the mutations we get are plasmid copy number mutations the couple effectively selection for transfer and for high level antibiotic resistance so in conditions that promote increased transfer rates you get high copy number and this also leads to high levels of AMR and it also works in the opposite way so we were quite happy with this however there was quite a few phenotypic data that didn't totally fit with the story so I don't have time to mention them in detail but some of the phenotypic data were a bit rare and also the sequencing data we had also told us that there was an additional level an additional story going on and so this is some this additional info from the sequencing data so what you see here is the coverage of plasmid rates mapped on the plasmid sequence and so it's relative coverage compared to the chromosome so a coverage of one means that there are as much plasmid as the chromosome per cell so yeah and here we have ignored the two blocks so it's different it's what happened with data cells which were two different treatments we had and this is just to declutter the graph a bit but to say basically the same here so what you can see is that in black is the what is a wild type and this one plasmid which has a constant plasmid copy number which is around three or four and most of the evolved clones have an increased copy number but the same qc directly we're looking at this graph is that this isn't uniform over the sequence so there is a massive drop in coverage for a lot of the evolved clones in this region and this region is again the antibiotic resistance region and the drop corresponds to the insertion sequences which flank the resistance region so and so I still can't believe that when I when I looked at these graphs initially I couldn't understand what was going on for a very long time and and now it's in retrospect it seems quite obvious so what is happening here is that we have within each sequence clone there is a cohabiting a wild type plasmid which full length sequence and deletion mutants which have lost the antibiotic resistance region so within a lot of the evolved clones one deletion mutants are present and we have this scenario with a full resistance plasmid and deletion mutants which are which have increased copy number whereas in most of the cases the wild type plasmid uh full length plasmid doesn't have increased copy number and to characterize this uh a bit in more detail uh so I had one clone which had only um the deleted uh plasmid so which didn't have the high the full length plasmid so it had a clean deletion uh and so I just uh did a piece here uh using the flanking regions and I sent this for sequencing and from this we from analyzing the sequence what we saw is that this is this deletion is a clean deletion of the full resistance region uh and from the two identical insertion sequences which are on the sides there is just one left um so this explained quite a lot of the work phenotypes we saw uh including the fact that there is um that the phenome mutations that we saw uh had very little effect on uh plasmid transfer rate and this is because most of those phenome mutations uh are on these deletion mutants so they mostly transfer themselves and like can detect this in conjugation assays which rely on them on plating uh on ampicillin plates uh but another thing it explains is the population dynamics so this is uh the population dynamics data I showed initially and I always found them quite weird because what you see here is that for 10 days the plasmids are transferring quite well actually despite the high level of immigration of so they don't mind the dilution of the transfer very efficiently and only after 10 days do they begin to decrease so this doesn't really fit um with a simple a simple story and now knowing that there is evolution of deletion mutants what I did is that I took clones from uh thousand populations at the end of the evolution experiment uh but that I plated this time just on lba and not on ampicillin and I did PCR to detect the deletion junction so to detect deletion variants and what I saw is that uh most of the populations are full of uh the deletion variant so the majority of clones uh have this deletion variant so what we thought was plus well not kasmid loss but what we thought was kasmids not being maintained over time was actually an invasion a progressive invasion of deletion variants uh and the ampicillin resistant variant being displaced by those deletion variants um so then what I've done since then is to try to understand how those deletion variants affect population dynamics um and they might have effects on both vertical and horizontal transmission of the resistant uh variant so first I uh so I I did an experiment where I compared just vertical transmission of different uh setups so in the first one you just have the wild type one kasmid so just been diluted over and over um again in the second one I put both one deletion variant which has low copy number so wild type copy number and the wild that's full length wild type kasmid and then in a third setup I had the wild type kasmid coexisted with a deletion variant which also has high copy number and this is uh what the dynamics of the wild type copy number uh carrying cells um looked like so what I saw is that the proportion of cells that with ampicillin resistance decreases over time strongly but only for the variant which has when the deletion variant has high copy number so there is some competition within cells and displacement of the wild type kasmid uh but this is not uh this doesn't have a significant effect uh in this experiment if the deletion mutant has a low copy number uh so this suggests also that in the evolution experiment what we saw is first evolution of high copy number and then the deletion mutants with high copy number displays the wild type kasmid uh then I did um a different experiment to look at how um deletion mutants in uh the recipient cell affect uh the spread of the resistant kasmid which is in the donor cell so I had one percent of the one wild type kasmid donor that I mixed with 99 percent of a recipient that either has no kasmid or has different variants of deletion mutants um and so here you can see so I yeah I mixed them and then I just diluted them over a hundredfold every day for three days and here you see again the cell density of ampicillin resistant cells uh in um over time so in black you have what happens in the absence of deletion variants in the recipient and the wild type kasmid takes takes three days to begin to significantly invade the population so it's not really fixed at three days but almost um and for most of the uh when we add most of the deletion variants uh which is that this spread is totally stopped so this is likely due to entry and exclusion uh from the um deletion variants in the recipients so we had one of the plasmids uh deletion variants where this wasn't working but after looking at the chromosomal background this is not that the plasmid is transferring to the recipients it's just that the um this deletion variant is an extremely costly one so what we see here is just a clonal spread of the donor cell or the donor background and finally I did another experiment where I wanted to look at what happens if we if the two variants so the wild type and the and the deletion variant compete for transfer so in this treatment I had one percent of cells which contains the wild type ampicillin resistant plasmid one percent which have either no plasmid as a control or different deletion variants and a majority of recipient plasmid free cells so here both plasmids types begin at similar frequencies and compete for access to recipients and this worked a lot better than I thought it would be it would so I expected some effect but not a massive effect because the first the wild type plasmid also can transfer quite fast but what I saw is that so in uh in control treatment the wild type plasmid invaded uh so began to significantly invade at three days and then was mostly fixed by four and five days uh but all of the deletion variants uh were very good at totally stopping this part of the wild type and this included one of the deletion variants which has lower pin number in in orange which shouldn't have higher transfer rate so I didn't expect it to but actually when I then looked at I characterized colony individual colonies at three days and I saw that in the in the control treatment with just the wild type plasmid most of the colonies are still plasmid free so the wild type plasmid has just begun to invade the population whereas in most of the deletion variant colonies I could detect the deletion variant so it seems to have a strong advantage even when it doesn't have a high copy number it seems to have a strong advantage for horizontal transmission so it's not necessarily transfer rate because after three days you have lost transfer and selection but it has an advantage so to conclude uh yeah this is a bit messy because I did this five minutes ago but so the first thing is that copy number is really an evolvable trait and it acts not only on vertical transmission as we've seen in many talks this week but it also acts on horizontal transmission at least in this plasmid so it's the question is open on how general this is for all the conjugative plasmids and it also acts on evolution as we've seen in other talks and also so I can't prove it but I think that here what's happening is that because we have high copy numbers this promotes the evolution of deletion variants in the first place because they can become binary with each other the second thing is as has been discussed a lot in the workshop already we see the importance of non AMR related plasmids for the dynamics of AMR and they will act both within cells and across cells to prevent transfer and this is a paper I found where uh people looked at uh the they were trying to explain why some types of peopleized strains didn't have antibiotic resistance plasmids and what the the main factor they found wasn't CRISPR or restriction multiplication but it was a plasmid of the same type without resistance so acting as a barrier and the final thing that I would like to say is that I think this is this type of data that I found is quite easy to miss both experimentally and in genomic data so experimentally the standard way of looking at conjugative transfer is just to play to the diabetic resistance so we missed most of the story for a long time and on genomic data so I'm not an expert at all and I have no idea how easy and how standard it is for people to look at a copy number and also even more on the difference in coverage across a plasmid sequence but I know that even in my case where we had I took yeah I could take a lot of time to figure out what was happening even in a very simple case so yeah I should stop here right yeah I I was planning to say exactly the same thing how many things we miss like if you don't can you hear me yes I'm talking yeah if you don't look properly or if you just have a hey there is Fernando hello Fernando you are peered hello yeah I was from the beginning but I am in the shadows do you have anything to say I have but I haven't risen my hand yet so let's other let other people talk let me ask questions but nobody has a hand up no okay so then I like very much your talk very much because you know these things happen not to you but to everybody but you were lucky that you found the answer you know because many people have things they cannot explain and you know they are there so what happens with F or with R1 is very interesting this kind of evolution and the loss of antibiotic resistance because it is known for a I think it's R1 that they did also a transcript analysis as the one I analyzed in R388 that I spoke yesterday and they found that the only again that the only genes which are expressed to high a to high level are the antibiotic resistances and it is known from from a very long time that a R1 can lose what was called then the R determinant yeah so there are there are variants of R1 without resistances known from from a very long time because that happens even when you don't look at it but what is interesting is you say that for evolution this rising copy number is important but in fact all like plasmids have very low copy number so there must be something else which impedes these variants to take over real populations do you know what this could be it it it can be the cost but if the cost is mostly because of the R determinant it is lost and then they won't need a copy number I think one one problem in not problem one a one consequence of the design of your experiment is that if you have dominant and mutants in your variants they will displace the wild type much more than you expect because the low copy number won't be able to replicate at all in a strain with with the other one because the way plasmids replicate or disrepay plasmids replicate is that they have to reach a concentration of the initiator to start replication so if you have one plasmid that needs less or that is more that is more the sides are more if it's the copaging I don't remember very well but maybe you're just shutting down replication of the other one so you're excluding it officially from the cell and from replication that's one thing but there must be other things that happen as well and you could think why these high copy number mutants never occur in nature as far as we know so I think one of it some of it is the cost and but the fact that I have in this experiment I have strong selections for for increased transfer because so I have one so all of these deletion plasmids I found that in this experiment they all have high copy number but I got one that has no copy number that I used in some of the experiments okay so this one I was in so I had a plasmid free control population well I had 12 of them and in one of them I got contamination by plasmid weeds so which got contaminated from the evolving populations and this one has low copy number and so in this population after the plasmid has invaded the population there was no more selection for transfer so I think this one came from a high copy number one and then reverted to low copy number as soon as it had no need to invade your explanation for the absence of high copy number plasmids in in nature of the R1 that overall in nature you wouldn't have this strong constant selection for transfer sure yes that's maybe yeah okay we have thank you very much thank you we have Olivia who wants to ask a question I think hi yeah my question's really quick really great talk Tatiana I really love your work so this is great um so you did actually do the test for your the effect of your deletion variants on AMR invasion with the deleted variant of low copy number and you saw that AMR didn't invade is I'm just trying to understand so is your understanding similar to mine where it's actually just that the deleted variant has less cost and so it has a higher fitness and so it's actually just clonal expansion that's making it so the deleted variant can invade faster than the plasmid that has the AMR cassette no so I mean I did a few experiments so and but in none of them is very it's not easy to know what the factor is a lot of factors are happening in all of them but I think what is different is that it has a it seems to have higher horizontal transmission and once it is in a cell then you get entry exclusion so if it arrives in a background before the wild type does then the wild type cannot uh enter and this seems to be the effect seems to be strong enough even with low copy number and um so entry exclusion is just working very well in any case okay then thanks Tatiana it was a really amazing talk and I have a piece of news on the Rohan Mehta thing so he was not here he said an email saying that he was double booked and he couldn't appear and we are trying to have him speak at five or at six thirty uh I will let you know we said the money man um and it's okay I feel less bad um now we have Sarah that's for you