 Hi everyone, I'm pleased to be talking to you today about my work on ESBL plasmids in Klebsiella in a hospital setting. So first a quick bit of background about Klebsiella. So Klebsiella is a gram-negative bacteria and it's frequently an opportunistic pathogen in humans. Now there are many different Klebsiella species as we can see from the phylogenetic tree on the right. However, species that fall within the Klebsiella pneumoniae species complex, which are the seven species shown in the box, are the most common in humans. And species within the Klebsiella pneumoniae species complex are often multi-drug resistant when we find them in hospitals. And third-generation kephalosporin resistance is becoming a real problem here in Australia and that's usually mediated by the carriage of extended spectrum beta-lactamases or ESBLs in Klebsiella, which are frequently found on plasmids. So we've already done quite a bit of work looking at Klebsiella diversity and strain transmission in our hospital. So in 2013 to 2014, we had a one-year study called CASPA. And in this study, all patients in the ICU were screened for carriage of Klebsiella and we also collected all Klebsiella infection isolates hospital-wide. Overall, we had 440 genomes that came out of the study and these genomes were all sequenced using the Illumina platform to generate short reads. And at the time, one of the PhD students in our group, Claire Gory, worked on this data and she spent quite a lot of time understanding the strain transmission dynamics, which you can find in these three papers here. And so one of the main findings that Claire had was that isolates that we were pulling from the hospital called Klebsiella pneumoniae actually belong to three different species, all of which are in the Klebsiella pneumoniae species complex. And these isolates were incredibly diverse. We did, however, see some instances of strain transmission and these are the lineages or sequence types highlighted in blue. But as I said, Claire mostly focused on what was happening at a strain level in this population and she didn't look too deeply into the plasmids. And we know that when it comes to the movement of antibiotic resistance that strain transmission isn't the whole story. So what we really wanted to understand was what were the dynamics of ESBL transmission in our hospital? So firstly, what is the actual burden of ESBLs during this study period? What are the plasmids that are carrying these ESBL genes? Was there any plasma transmission occurring? And what was the impact of any plasma transmission on ESBL burden? And so we used a combination of Illumina and nanopore sequencing to answer this question. So we took our 440 genomes and the first thing we did is we de-replicated them so that we only had one genome per episode. So that made sure we had a unique genome per patient and per body site to remove any duplicates. And so here is a bar chart showing you the total number of episodes per month of the study broken down by whether the isolate was a carriage isolate or an infection isolate. And when we screened the Illumina data for ESBL genes, we found that 18% of our genomes had an acquired ESBL gene. And by far the most common ESBL gene was CTXM-15. So then to get at what plasmids were carrying these ESBL genes, we then went and did an additional round of sequencing where we selected genomes to do nanopore sequencing on. And the way we selected genomes is we made sure that we were going to have at least one completed genome per species, sequence type and ESBL gene combination. So once we had our completed genomes, then what we wanted to do was to find some more plasmids. So the first thing we did is we've separated our genomes out into what specific ESBL gene they carried. And then we took the plasmids that carried the same ESBL gene and we compared them all in a pairwise fashion. And so here I have an example of two CTXM-15 plasmids. And the first comparison that we would do for each pair was to compare the nucleotide sequences. And we would then get a nucleotide similarity score, which fell between zero and one, one being identical and zero being not the same at all. And so in this toy example here, we can see our nucleotide sequences are pretty similar. We have a score of 0.98. The next thing we did is we wanted to check how similar the gene content was. So to do this, for each pair of plasmids, we worked out which genes were homologous and homologous genes got the same gene symbol, which is what I'm showing here. And then we calculated out of all the possible gene symbols across these two plasmids, how many did they have in common? And that gave us our gene similarity score. And for this example here, the gene similarity score for these two plasmids is 0.8. And then the final thing that we did is we had a look at the order of genes. So if we took these two plasmids and we had the sequence start in the same location, how many of the genes were not just the same, but also in the same order? And so this is what I'm sort of demonstrating here. We have our two plasmids and the gene symbols are pretty much all in the same order, except for the middle-ish genes, gene 4 and gene 10. And so that was a gene alignment score. And in this particular example, the gene alignment score is 0.88. So back to our question, how many different ESBL plasmids are there? This is what we did. We compared every plasmid pairwise and on this graph here, I'm showing just the nucleotide similarity scores and the gene content similarity scores. Each point is a pair of plasmids coloured by whether they come from the same strain or sequence type or not. So grey if they don't and black if they do. And you can see that there's a pretty clear break in the distribution. So in the top right-hand corner, we have pairs of plasmids that are sufficiently similar at both the nucleotide and a gene content level as to be called the same. So when we applied these cut-offs, we found that we had 12 distinct Clebsiella plasmids across our ESBL collection. 10 of these had ink F replicons. They were generally pretty large, at least 100 kilobase pairs in size. And the vast majority of them carried additional AMR genes on top of the ESBL gene. So then we wanted to understand if there was any evidence of plasmid transmission. So to do this, we took our scores and we used them to make a network. And what we found when we made this network is that most plasmids were in a single sequence type. So here I have an example of a CTXM-15 plasmid B and each of the dots here represents a genome coloured by its sequence type and the genomes are connected by a line if they share the same plasmid. And you can see that all three of these genomes have the same plasmid, but they're also all the same colour so they all belong to the same sequence type. And generally this is what we saw for many of our plasmids. They either belong to the same sequence type or they were just a single strain that just carried that plasmid. However, we did find one plasmid that was found in multiple Klebsiella sequence types and even multiple Klebsiella species. So this plasmid we called plasmid A. It carried a CTXM-15 gene. It was pretty big at about 250 kilobase pairs. It had ink FIB and ink F2 replicons and it carried five to eight additional AMR genes on top of the CTXM-15. And you can see by the different colours here in this little network is that it's present in multiple sequence types and in fact the light purple dot actually represents a different Klebsiella species called Klebsiella varicola. So we wanted to have a look at what was going on with all of these different plasmid A positive genomes and what was happening with their sequence types more generally across the study period. So what we found was that plasmid A was present throughout the entire study period. And the first thing that we noticed was that Klebsiella ST323 which is the top row here in this graph in blue was carrying the plasmid throughout the entire study period. So each of these points is coloured in if the genome is positive for plasmid A. And we can see there's ST323 all the way along. We see single instances of ST221 and 5822 both of which carry plasmid A. And then what we see is about halfway through the study we see the emergence of ST29 that's now carrying plasmid A. So we hypothesised that ST323 has donated its plasmid to ST29 and this is our first instance of plasmid transmission into a new strain. And then we also see a couple of instances of our Klebsiella barocola ST347 emerging now carrying plasmid A. And again, we hypothesised that ST323 has donated its plasmid and this is now our first instance of species transmission. So what was the impact of plasmid A transmission across the study? So here I have a bar chart showing the total number of ESBL episodes that have occurred per month during the study. So we had 57 ESBL episodes across the whole study period. 23% of those were due to ST323. However, if we then add in all of the additional ESBL episodes that were caused by strains acquiring plasmid A we now have 53% of all of the episodes throughout the study were due to plasmid A. So a small number of transmission events has led to over half of all of our ESBL episodes during this study period. So this is our hypothesis. ST323 persists throughout the study period. We see occasional transfer of the plasmid into new strains. Two of those strains then go on to cause further outbreaks in the hospital and infect other patients. And we wanted to know if any of these newly resistant strains persisted beyond the end of our study. So we had a collection of all Klebsiella infection isolates from the same hospital that spanned from 2017 until 2020. So we had a look in these isolates to see if we could find plasmid A. And this is what we found. So here again, I'm showing each bar is showing the total number of ESBL episodes per month from 2017 through to 2020. And the bar is colored if that particular episode is carrying plasmid A. And so you can see many of the same colors that we've seen before. So the Klebsiella varicola ST347 is now very prevalent, especially in 2017 and early 2018. But we also see the emergence of a new strain now carrying plasmid A ST2856, which then goes on to cause an outbreak, as well as a single ST5823 that now carries plasmid A. And if we look at the impact of plasmid A in this early time point from 2017 to the first portion of 2018, 64% of all of the ESBL episodes of this time were due to strains that carried plasmid A. So we also just wanted to double check that when we looked at the plasmids from this later time point that they were indeed the same plasmid. And so here I just have a little depiction of what's on the plasmid. So this particular plasmid carries metal resistance as well as plenty of AMR genes shown in red. And here is a alignment of three of the plasmids, one from each of the main sequence types from the first part of the study in 2013. And we can see by the order and color of the blocks that these plasmids are remarkably similar. There's been a small deletion here and we've seen an inversion of this green block in these two representatives here. And when we expand this to have a look at representatives from all of our previous STs plus the new ones, we can see that the plasmid is remarkably similar. So there's almost no point mutations different between any of these plasmids. And the changes that we do see are mostly around insertions and deletions. So we have some small deletions here and this one here and this new sequence type seems to have incorporated a larger new portion of DNA. So I think one of the main takeaways from this study was that plasmid transmission was pretty rare in our hospital, but when it did happen, it had a really big impact on ESBL club cell burden. And I think the other thing that I took away from doing this analysis was it really made me think about how we could go about performing these types of plasmid transmission analyses going forward. So I set specific thresholds that made sense for my data at the time, but I don't necessarily think that thresholds that I've used here could be replicated and make sense across all studies in the future. And I think what it really comes down to is how we wanna define the same versus different when it comes to plasmids. I think one of the really key tricky things that plasmids cause when we're trying to understand this is trying to have a better understanding of how frequently mutational events occur in plasmids. So when we look at chromosomes and we look at point mutations, for many bacterial species now, we have a pretty good understanding of how frequently we expect those point mutations to arise. Whether I don't think we necessarily have that good of an understanding of how often we expect to see that in plasmids. How many point mutations between a pair of plasmids is enough to say this was a recent transmission event versus this happened a long time ago. And on top of that in plasmids, we don't just have to worry about point mutations, we also have to worry about larger scale insertion and deletion events, especially in these AMR regions. How often do these larger scale insertion, deletion, inversion events actually occur? And I have a sneaking suspicion that this is probably gonna vary a little bit by plasma type and also potentially by host background. And I think we really need to get a good understanding of how frequent these different events are if we really wanna be able to use these kinds of genomic techniques to understand when we've got plasma transmission happening in a clinical setting. And finally, I'd just like to thank everyone that was involved in this work. So I'd especially like to thank Kat Holt, who was my supervisor at the time that I did this and everyone in the Holt group. I'd also like to thank everyone in the hospital microbiology lab, especially Adam Jenny. They collected all of the isolates and made them available for me to study. You can read some more about this work in the preprint that I'm linking below. And lastly, I'd just like to say that I'm really keen to hear any of your thoughts or if you've got any questions about what I've presented here today. Unfortunately, it's the middle of the night here in Melbourne. So this talk is prerecorded, but I'm hoping to be awake and available for questions during the discussion portion of this session. But otherwise, you can contact me on Twitter or you can send me an email. Thank you very much. Thanks very much, Jane. That was a fantastic talk. And I'm glad you're still awake. So we have a couple of questions. I'm gonna give you Yana's one. I'm gonna just give you one because of time, but we'll come back to the other stuff from Yana Hussman. Who is Yana? I'm gonna have to ask you how to spell your pronounce your name because I keep relaying your questions. But anyway, Jane, you mentioned two episodes of Plasmid A transmission into a new ST. In one case, in green, you first find the ST without the Plasmid and later with. But for Klebsiella varicola, you don't have another status of the strain without the Plasmid. What made you think this was a new transmission event during the time you were studying? Yeah, this is a really good question. Look, it's just a hypothesis. It's really hard for us to pin down exactly when the transmission events happened. It's not published, unfortunately, this information, but we do know that ST323 had been circulating in Melbourne for a while, not just in our hospital, but also in other hospitals. So we do still think that ST323 is the donor strain, whether that Plasmid transmission event happened in our hospital or in another hospital in Melbourne, we don't know the answer to that, but it's our best guess. Great, thank you. Yana, any follow-up on that? I'll let you type. Jane, that was brilliant. I have questions, which I'll say for later. And I don't know how much of the rest of the session you intend on staying up for. Yeah, I'm hoping to stay for the discussion section after the following talk and then go to bed. Thank you very much. Okay, so let's move on to the next talk from Rob Moran. This is high-resolution sequence annotation and comparative analysis as the foundations for Plasmid surveillance. Rob, do we have you online? Hey, yeah, here I am. Let's see if I can screen share.