Scientist Stories: Cuomo Christina, Genomics and Fungal Pathogens

Hey, good afternoon, everyone in the, in the room and online. And it's a pleasure to be back here in Lipsett for the NHGRI Seminar Series and welcome any of you to come in the future. So I'm Julie Segre, and I'm here to introduce Christina Cuomo. It's a real pleasure to welcome Christina to the NIH. Christina is a longtime colleague, although it has taken me a while to get close enough to be able— in fields to be able to work with Christina. So as a bit of a background, Christina has her PhD in genetics from Harvard and then did a postdoc at UCSF and then joined the Whitehead Institute and started to focus on the human genome and joined the Human Genome Closure team, so, uh, was part of closing chromosomes 8, 11, 15, 17. Remember when we used to do chromosomes one at a time? That was in 2005, 2006. But really, at the same time, Christina was developing an independent research project, becoming a leader in fungal genomics, and in 2007 published one of the first fungal genomes for Fusarium Graminarium. This is always my problem with microbiologists, I can't pronounce any of these things. But really went on to sort of be the pioneer and be the leader in the genomic studies about fungi and sort of how would genomic information reveal really cool biology about— so for example, in 2009, a study on the evolution of the pathogenesis and the sexual reproduction in Candida genomes. And also, so while Christina's research has focused mostly on human pathogens, she's taken also a larger view of human health and thought about environmental health and was one of the first people who talked to me about the chytrid fungi that were affecting frogs and is really just very well-versed in the different fungal genomes and works with so many people and is seen as really the person who is leading fungal genomic studies. So today, Christina's talk is going to focus on population genomics and the evolution of virulence traits in Cryptococcus neoformans. So welcome, Christina, and good to have you here.

Thanks. Thank you, Julie, for that introduction. And thank you also to Diana Proctor for the invitation to come. I'm going to focus today just on one area of what we work on is talking about Cryptococcus. And the priority that we've given to Cryptococcus has been highlighted by this recent WHO report This came out just within the last few months where the WHO prioritized for the first time some of the most health-threatening fungi. And they did this because of the increasing threat of fungal infections and the rise of antifungal resistance. And as piloted in these quotes, that because of that, they're becoming resistant to the treatments that we have. And the top of the list of the most important threats, they highlighted this critical priority group that includes 4 species that we work on in my group. And one of these is Cryptococcus neoformans. Cryptococcus is a global disease causing meningitis, and it's been estimated to cause over 180,000 deaths per year. But it is of particular burden in Sub-Saharan Africa, because of the large population of HIV-positive individuals there. And in that Cryptococcus is causing disease in immunocompromised individuals. But because of that also, it affects people worldwide in the United States, causing disease in people who are immunocompromised or living with cancers or who are undergoing things like organ transplants.. And this picture just highlights the annual incidence of Cryptococcus around the world. Wanted to also highlight that, you know, while I highlighted a lot of the, some of the other fungi that are human pathogens, including Candida and Aspergillus, that Cryptococcus is very distantly related from them. Here in this phylogeny produced by Antonis Roukas in a paper last year. This is highlighting in red some of the fungal pathogens within the human— within the fungal kingdom. Cryptococcus is in a phylum called Basidiomycota that contains a genus of fungi that are commensals and cause dandruff called Malassezia. But otherwise, this is not where you find human fungal pathogens. You mostly find them more around 4 or 5 o'clock in this phylogeny where you find Candida, and then dimorphic fungi down around 6 o'clock. So it's not very closely genetically related to other human pathogens. So some of our knowledge is really going to be— need to be newly discovered about what makes it a pathogen. And within Cryptococcus, there are multiple species. How many species has been a topic that has been pretty widely debated and wanted to just recognize that there are these two large subdivisions called Cryptococcus gattii and Cryptococcus neoformans. I won't be talking about Cryptococcus gattii, but just wanted to mention that this is really different in that it affects immunocompetent hosts and has been found within the United States, particularly causing an outbreak in the Pacific Northwest. And has been subdivided into these lineages or species that are on the top of this tree. But the group I'll be talking about is Cryptococcus neoformans that I've mentioned affects immunocompromised hosts. It has environmental reservoirs in different types of trees, in urban environments, in pigeon guano, and also in soil. And as I noted, has a global distribution.. And I'll discuss how it has been broken down further into these distinct genetic subgroups, here denoted VN1, VNB, and VN2, that have distinct global distributions. And I'll mention further that there's a related species called Cryptococcus deniiformans, and then these two can be distinguished serologically. So while Cryptococcus is an environmental pathogen, it's been demonstrated that isolates from the environment and isolates from the clinic are very different in virulence. So first, just to review its mode of infection, as I've mentioned, there are these environmental reservoirs that, from which spores can be released and infection then can start as being inhaled and begin as a pulmonary— inhaled into the lungs and begin as a pulmonary infection. It can then disseminate to the central nervous system and cause meningitis. And those infections can be diagnosed by a positive culture. And if you compare these clinical isolates that are recovered from patients from those recovered from environmental sources side by side in a mouse model, in this paper from Anilipp and Seva, they're very different in how they affect the mice, where the environmental isolates as a group in this study did not cause the mice to die, but the clonal isolates did. So although this is an environmentally acquired pathogen, this suggests that there are isolates that are very predisposed to causing clinical infection, which seems unusual for an opportunistic pathogen. But perhaps there could be things in the environment that are predisposing it to cause disease. It's been hypothesized that this could include interactions with other eukaryotes such as amoeba. But we'd like to understand what are the factors that explains these kinds of differences. So to do this, we launched a large genomic study to understand how natural variation contributes to virulence. And we began this by trying to better understand the population structure of both clinical and environmental isolates and taking this global perspective. We also looked at recombination and hybridization in these populations to lay the foundation for doing genome-wide association studies. That then we tried to use to look at genetic traits associated with clinical isolates. And then we married the whole genome sequencing to phenotypic analysis that I'll describe that was either a range of things from in vitro phenotyping to using patient metadata to try and associate traits with the genome variants. So to start with the picture of the population structure, we did this using phylogenetic analysis from the whole genome sequence. And this revealed that while initially there were described these 3 population subdivisions, our analysis showed very clear support for 4 major subdivisions. And that's shown in this phylogeny in that we recovered this new split at the top of the tree of the B&B group separating into 2 lineages. And that the colors around the external circle, the blue and the red at the innermost circle are showing clinical and environmental origin. And what I hope you can observe even at a distance, that there's a good interspersed pattern of these two colors. So there aren't any big pockets of being a clinical isolate or an environmental isolate. They're fairly interspersed. This suggests that there's not any 1 or 2 or 3 major clinical lineages, but that these patterns are interspersed, suggesting again that association approach seems like a valid way to proceed. Just to provide a little more context of the type of initial work we did, we did find some additional substructure within one of the lineages called VN1 where it split very cleanly into 3 separate clades that we gave additional names to. And we found that these were also separated by geography where one of them that we called VN1C that is in the principal components analysis that is on the right part of the slide was entirely composed of isolates from Botswana, which is in— were collected from— all collected from the southern— represented the most southern part of Africa that we collected isolates from. So this, whereas in contrast, the other two groups in this tree were, and in the PCA, were globally collected. So while VN1 is thought of as a global lineage, this shows that the finer scale picture is a little bit more refined where one of the groups is actually specific to Sub-Saharan Africa. So then to pivot now, once now that we have a clear understanding of the population structure into carrying phenotype association, wanted to just give a picture of what kind of data we were thinking about for some of the phenotypes. We wanted to be able to test phenotypes that were going to be as relevant as possible for thinking about virulence traits. So for Cryptococcus, two of the big ones were production of melanin, which is a virulence property, as well as these other ones that are listed here to do with temperature growth, resistance to oxidative stress, and drug, which I won't have time to talk about. And then I will talk about as well some of the clinical metadata as I go through the talk. To carry out the genome-wide association, because we saw such strong population division, we decided initially to separately analyze the lineages and carry out separate analyses for these like VN1 or VN2 separately. And we then subset the variants for their functional impact. But we also, 'cause we know that we are looking for probably rare variants to be able to look for things associated for virulence, we decided to collapse signals by gene to be able to find rare variants. And then the test that we run was called GEMMA, which is a linear mixed model. So the first test that we ran was for loss of melanization, which is an assay that you can just run on a plate and measure for each isolate the level of melanin. And the top hit that we got from the GWAS was a loss of function mutation in a transcription factor called BZIP4. And we had 4 different mutations that were associated with this transcription factor in 4 different clinical isolates. And then what they actually look like is shown on this plate. The 3 at the bottom and then the one, the top row on the right is a little more melanized. And then there's some controls of a wild type isolate that is melanized. And a mutant that is not. And then this is just some reproducibility of the assay. This was a gene that in a transcription factor screen by another group was shown to affect melanization. So this was already validated, not by us. So this was a nice kind of positive control that we were able to pull out a real signal. And then we had the other hits that were coming out in this, analysis where all, many of them were in or next to a serine-threonine phosphatase that we haven't looked into as much because they were very common variants. In this first analysis of GWAS, we also then took a very, I guess, aggressive swing at trying to look for variants associated with just being a clinical isolate. So asking, are there SNPs associated with clinical origin compared to environmental. And when we did that, we found that yes, you can find SNPs, and they are— if you look at the top of the list, what you see is a lot of genes that are linked to virulence factors and genes involved in the oxidative stress response. But many of these mutations are either common or in genes and some of them are kind of in just linked to the genes in intergenic regions. So nothing directly yet to follow up on, but I think we took this as a positive sign that this analysis looked like it was going to be able to find some signal. But we definitely recognized that just comparing a clinical to an environmental isolate may not reflect the level of virulence. So what we wanted to move to next was to be able to look a little closer to reflect the level of a strain's virulence by moving to metadata of the, how the patients that the isolates came from fared during the course of their infection. And we did this by partnering with the clinicians who were involved in this clinical trial called ACTA. This was a trial of patients from hospitals in Malawi and it was a trial for antifungal regimen. And the patients were all HIV positive and had developed meningitis and the trial was comparing two different antifungal regimens. So we sequenced about 300 Cryptococcus isolates. These now are from a different lineage than the previous analysis. These are mostly VN1. And with this clinical— with this study, working with the clinicians, we selected this metadata that the clinicians thought might be useful for us to think about comparing, including the fungal burden, how fast the infection was cleared, and then things to do with the, the how the patient fared during the infection, the mortality, how long they survived, and things to do with their mental status reflecting the meningitis. And again, one of our preliminary assessments was plotting some of these features on the phylogeny to ensure to evaluate if there was any jackpotting again of phenotypes of genetically related isolates, which again, we did not observed suggesting that there's no, again, single genetic lineages representing strong phenotypes. The other initial look we took at the metadata was just comparing it to determine what had the strongest correlation just with itself. In doing this, we found a very strong correlation between the fungal burden or the colony-forming units coming from the CSF and the patient mortality. And when we looked at the fungal burden in the CSF, there's a very wide range that was observed in the counts that came out. So this plot is showing for isolates from these different VN groups and a few from VNB, that there's a very wide dynamic range here plotted on the log scale of these CFU counts. So from this, we thought this was going to be a good place to start of a GWAS with the level of fungal burden because it was such a— it looked like it was a good measure of death and it had a wide dynamic range. So we ran the GWAS with the fungal burden and we were able to find that there was a very strong association for variants with genes involved in virulence. And this provided some interesting hits, including one of the top hits was a gene that is the target of azole drugs, ERG11. This included multiple genes involved in capsule, which is the polysaccharide coat that is on the outer part of Cryptococcus and also considered a virulence factor. There's a gene called SGF29 that I'll mention, that I'll come back to, that is involved in histone acetylation and also melanization and virulence. And other genes involved in thermotolerance or iron. So this is a pretty encouraging list. You can see there are definitely kind of single hits that come up on this Manhattan plot, which I'm showing here, just the 14 chromosomes of Cryptococcus and the p-values of the individual SNPs. So in terms of follow-up, we selected the most highly significant loss-of-function mutation and working with John Purvis' lab at Duke, they tested this gene in a rabbit model. This was a loss of function mutation in a phosphofructokinase that is involved in glycolysis. When this gene was deleted in the lab strain H99, in a mouse model, it did not affect the survival of mice in an infection model. But when they deleted— when they tested this deletion in a rabbit model that has some advantages for recapitulating what infection looks like in a human, it significantly reduced the fungal load that you measure from CSF in the plot on the right. And that is the phenotype that we were associating on. So this was was recapitulated. I also want to highlight importantly that while we find many interesting hits from the GWAS, a lot of the top hit— a lot of the top genes that are highlighted now with this version of the slide and noted with HPs in the in the cloud above the— within the red plots are hypothetical proteins. And in fact, as noted here, nearly a third of the most highly significant variants are these genes with very little or no predicted functional annotation. So there's a lot of— a lot that we have yet to learn about Cryptococcus gene function, kind of as I alluded to in the slide of phylogeny. Cryptococcus is again just a lot of phylogenetic distance away from other known fungi that from which we can infer function. So to start to do that for one gene, again, we took what was the most highly significant GWAS association, a frameshift in an unknown protein, and we took two approaches. We looked at the difference in virulence of two natural isolates and tested them side by side in a mouse model on the left. And in that model, we were able to show that there was a significant difference in the survival of mice just under a significant P-value. However, when we go and delete this gene in the lab strain H99, we do not see a significant difference in, the probability of the mice survival. So to us, this suggests that there might be other factors going on in these natural isolates that we have not yet uncovered that either might interplay with these genes or contribute to their virulence. We also are taking advantage of all these sequenced genomes to look at what larger other forces may be at play and looking at selection across all of these lineages that I've mentioned. And when we do this using a test for selection across the entire genome, what we're able to highlight is genes that have undergone selective sweeps in each of the lineages. And one major signal that this has highlighted is the importance of sugar transporters, carbohydrate transporters, and inositol glucose transporters in particular in VN1 and VNB lineages. And here I'm just showing a more detailed phylogeny of those transporters where we've done a more detailed analysis of selection. And highlighted in red the individual genes that are under selection in different lineages. So inositol and xylose are compounds that are abundant in plants, which might then explain why they've been selected, is that they're important in the environment. But why it might be important for the pathogen is that inositol is really abundant in the human central nervous system. So that might be a real benefit for Cryptococcus as a pathogen. And it's been shown that it's actually required both for virulence and mating in different studies. So this is helping to highlight how this, you know, really strong selection in the environment probably is an example of coincidental adaptation for the human central nervous system. Lastly, we are— we'd really like to be able to, of course, be able to zero back to looking at how an isolate from the environment then moves into a human and adapts. But there— it's extraordinarily rare. There's only been one case published of tracing the exact origin of a clinical case back to its environmental source. For the other, you know, many, many, many cases, we don't have that exact mapping. We also, for the populations that we've looked at for VN.1, have only had very few environmental isolates collected. But we have now gone back and tried to, for VNB, match geographically around these hospital locations where we've been sequencing large clinical sets of isolates, go back and collect environmental isolates in the same location and see if we can come more close to matching isolates. So that was done. And then we repeated the study with the collecting was done by Daniel Louredo and Poppy Sefton-Clark in my group then repeated the GWAS. And the hit that I wanted to talk about from this analysis was the gene I mentioned, one of the genes I mentioned earlier called SGF29. That is very strongly associated with clinical isolates when we do this test. So this is in this now collection we have of, of 650 VNB isolates. And we see this mutation really exclusively coming up in clinical isolates that we see loss of function mutation occurring. Of this gene called SJF21. This is, as I've mentioned, is involved in chromatin modification. So might have a lot of pleiotropic effects, but looks very, very specific for these clinical isolates. And what's interesting about it is that a recent paper has also described it occurring within this interesting series of isolates that are linked to the lab strain called H99 as it's been passaged by different groups. And in one of those passages moving on the right-hand side of the slide from this H99.0 as it's becoming hypervirulent, it has— this paper noted there's a loss of function mutation of this gene that occurred. And so as I've noted here, this gene is, is involved in histone acetylation. And so we expect that there might be a consequence of a lot of different epigenetic changes as a consequence of— so a bit complicated to look into, but definitely intriguing that this might be a common change that's occurring in clinical isolates. One other common mechanism we've uncovered that's occurring in clinical isolates is aneuploidy, which is very common and has been known about in Cryptococcus. Work that we've published previously using genomics has found there's a lot of changes you can detect from using genomics by looking at read read-depth analysis. And of course, landmark analysis from June Kwon Chung's group demonstrating that aneuploidy of chromosome 1 will allow resistance to azole drugs to emerge. We had also identified a case where you could get targeted duplication of genes that can confer drug resistance. And this is an extreme example. Where in this case you could get 13x amplification just of the drug target in this one clinical isolate. So we went back and looked at first just at the active clinical isolates that I described for ones that display aneuploidy. And I wanna focus on first the ones And these plots in dark blue that have an entire chromosome, that's the aneuploid. So in the phylogeny, they're noted with the dark blue circles around the tree. So they're of diverse origin. In the chromosome plot, they are, they're most commonly chromosome 12 and 9, but they are multiple chromosomes. The light blue is partial chromosome aneuploidies, which can be of more different chromosomes. And then on the rightmost part of the slide demonstrates growth data that's showing that the aneuploids grow more slowly. So this is doubling time and both at 30 and 37, there's a significantly slower growth that you observe for the aneuploid cells. There has also been work from Neil Stone and Tehana Bekanech demonstrating that in a series of isolates from the same patient that kind of echoing what the Quan Chung lab has shown that you could pick up an extra copy of chromosome 1. But this is unstable and confers heteroresistance to azole drugs over the course of recurrent infections. We do not see that there is very much chromosome 1 aneuploidy in these ACTA isolates, but most of them were taken at the beginning of the trial. So the patients have not yet undergone their full AZL therapy yet. So that might not be surprising. So most of the isolates were capable of growth on fluconazole. Only one had this chromosome 1 aneuploidy. And there's some others that were not aneuploid that could, that could grow on, on fluconazole. But what we see much bigger picture is in larger cohorts is just a huge bias for clinical isolates to be aneuploid compared to environmental isolates. So here looking at VNB as a whole, this group of 650 isolates that we've sequenced, we see that 13% of isolates that are clinical are aneuploid compared to 3% of environmental. And again, we see this kind of the same signal of maybe chromosome 4 and 12 being the most frequent clinically. Environmental, it looks like also chromosome 4 being— this was very common, but it's much smaller numbers. And we also looked back at a larger set of VN1 isolates, looking back not just at our data, but other published data. So 1,400 isolates. And here also we see a very big difference, 12% of clinical compared to only 3% of environmental. Again, chromosome 12 comes up as very frequent, but now different other chromosomes. And I just wanted to highlight one other piece of data, which is that, I mean, Cryptococcus will tolerate all the other kinds of unusual karyotype behavior in that it can form hybrids with its related sister species, D. neoformans. When it does so, it undergoes widespread aneuploidy and loss of heterozygosity. This is the example of— this is a single genome now of two different hybrid of both A and D. Looking across the y-axis, both these genomes are in a single cell, but in some isolates, there's, there are extra copies of one or the other chromosomes. And in the bottom one, one or the other chromosome might have been lost or gained. And yet, when we analyze these at a genetic level, all of these isolates that are shown at the top in terms of SNPs are nearly identical. So, and if you look at a finer scale at some of these AD hybrids, this is one that has an example of one that has resolved itself to a close to haploid level, but has undergone crossing over between these A and D haplotypes. So to wrap up this, just this part of the, some of these unusual features, I just wanted to highlight the level of genome instability that Cryptococcus can undergo that appears very higher, much higher in clinical isolates, and question if it could be due to stress or provide some kind of advantage to which we don't yet know the answer. And while we know there's an advantage to chromosome 1 because of the role in drug resistance, it does seem that other chromosomes are more frequent. And whether or not that's just because they're tolerated or if they might provide an advantage, I think is yet to be understood. We know that in Canada, there are particular triploidies that provide an advantage for living in certain niches. So it seems possible there could be those things to find in Cryptococcus. And then I think it's an open question of how when it's— how can Cryptococcus can tolerate this level of aneuploidy when it's clearly such a growth disadvantage, for example? And then to wrap up the, the Strainome-wide scans that I talked to you about, we've begun to show that clinical isolates can be used to find variation linked to melanization and clinical origin. And I didn't show data on drug resistance, but it's in some of our papers. And the direction we're moving for this work is to increase our power. We're definitely underpowered for the types of variants we're trying to find. And we also would really like to just measure virulence directly by pooling isolates to be able to kind of side by side be able to associate on virulence. So we're sequencing additional isolates and working on those additional phenotypes. So some of the time that remains, I wanted to talk about newer work that's ongoing that I thought might be of interest to a genomic audience. And it's a comparative genomic project stepping back, thinking about how Cryptococcus has evolved and kind of an unusual finding that came out of that. So again, I kind of, I started out some of the talk talking about how Cryptococcus is not very closely related to other pathogens. But now I'm, I'm going to kind of zoom in around it and look at what's kind of immediately outside of it. And it's within this group of species listed here. The Cryptococcus neoformans and gattii are related to These other species, some other Cryptococcus and some other Quaniella, some of the names have changed, I'll show on the next slide, that are mostly saprobic taxa that have been isolated from different types of arthropods, from other types of plants. There's some that are mycoparasites or soil. And so we, working with Joe Heidman's group at Duke, thought that if we sequenced these species related to Cryptococcus, that might give us a view into how the human pathogens differ if we could just zoom into their genomes. So we initially had started this with Illumina sequencing, but We have since moved to completing all of these genomes with long read sequencing and have now this kind of complete view of this Cryptococcus clade at the top and the sister genus Quaniella at the bottom. And the top group in red encompasses the pathogens. And then there's several Cryptococcus below it that are not pathogenic. And then all of the Quaniella are just related saprobic species. So one way the pathogens differ is that their genomes are a little bit smaller. That's the next, that's the first column. They're about 18 up to 19 meg. The genomes of the other species are a little bit bigger. And so they have slightly fewer genes as well. But in evaluation of the conserved gene content, all of these look pretty complete in terms of our efforts at predicting genes. So we looked at gene conservation by building orthologs across all of these species and looking at conservation patterns. And this is work from Marco Quello at Duke and Sage McGinley-Smith, who's an undergrad in my group. And we looked initially at patterns to see what was the predominant patterns, but then have focused in initially on the groups that were unique to some of the pathogens, such as this group those just found in the pathogenic species. And what's striking, just for starters, is how few there are of them. There's only 53 orthologs. And they include some genes that are of potentially interesting function, glycosyl hydrolase, some transcription factors, oxidoreductase, a whole lot of genes of unknown function. But what they do not include is all of the many of the, and all of the canonical virulence factor pathways, which are found in all of these species. So when you think about all of the genes required to produce capsule or melanin, which we think about as Cryptococcus virulence factors, those are found in all species. What we focused on next that was the surprise that came out of this project had to do with karyotype when we got to the point of complete genomes. And that while most of these, many of these genomes and all of the pathogens had 14 chromosomes, chromosome number varied from 3 chromosomes to 14. And that you could infer based on some of the ones with 14 in 2 of these species that it appears that we would estimate the ancestral chromosomes were— that there were 14 chromosomes that we could align. We could paint over the 14 chromosomes between these 2 species, between these 2 groups, and they look fairly syntactic. But if we look now across the Quaniella where the chromosome number varies so widely, and we start painting from one of the ones that has 14 now across the other species, you can— what I hope you can appreciate is that there's been a really unusual pattern of chromosome fusion that has occurred in these species. And in particular, with these, the species that have 3 or 5 or 6 chromosomes. There's mainly been one chromosome that's been the object of all the chromosome fusion. And this is just a little more detail on what has taken place. So starting from the, the two species in the top two rows that have multiple chromosomes, What has occurred is a series of fusions where there's been an inversion that has occurred next to the centromere, which is then lost. The centromere is lost in then the fusion chromosome that is then present at the bottom. So this is not— I'll show a model on the next slide. But in the process of all of these fusions, it's not just simply chromosomes joining together. It's a series of inversions linked to centromeres. So all of this then process probably inactivated all of the centromeres. So you end up just with the one centromere that is the ancestral one. One other thing that we have followed up on is that the centromere length is pretty different in all of these species. We can predict this computationally by looking at this class of transposons. I'm just showing an example on the right-hand slide of what centromeres look like in Cryptococcus neoformans. It's not exactly the same It's not the same elements in all species, but we can identify similar elements at all of the centromeres in these species. Further, Marcia Parra in the Heitmann lab has proven these are centromeres using chromatin IP methods. But what is striking is that the centromere size has changed in these species. But it's not that the species with 3 chromosomes have the tiniest centromeres, but definitely the Quaniella as a group have very small centromeres compared to the Cryptococcus. This is a bit of a whiteboard slide to end this part of the talk. We're trying to think about how this has occurred because we need to join things together while making inversions and inactivating centromeres. So this could occur by fusing centromeres and by fusing telomeres and then inverting or the other way around. I think some things to add layers into the ideas we think about is that we don't see any loss of telomeric genes. So there haven't been telomeres getting chewed away that are kind of instigating the process. And neither really do we see kind of centromere genes kind of adjacent to centromeres being lost. So it's really a fine-scale process that kind of initiates this kind of fusion events and definitely unusual that they're all being joined to a single chromosome. So just to wrap this part of the talk, and this is the end of, this will be the end, is that we are thinking about kind of the evolution of Cryptococcus as thinking about, you know, that a lot of the canonical virulence traits for how, what we think about as a pathogen are already present in a lot of the saprobes. A lot of the capacity to be a pathogen, if we just think about gene content, might already be there. And then we have to understand specialization at a finer scale to think about how the pathogens have adapted to become pathogens. And that the last story is what I just talked about, is thinking about karyotype evolution and how this unusual mechanism occurred. So I wanted to acknowledge a large number of people who contributed to the work I talked about. In particular, I'll call out Poppy Septian-Clark, who did a lot of the Cryptococcus population genomics, and Sage, who's worked on this comparative project in my group. At Duke, we've worked very closely both with John Perfect and Joe Heitman's group, Jenny and Marco and Marcia in their labs. Then the ACT-A clinical trial was a close collaboration with Tahana Bacanek and Tom Harrison's groups. I would be happy to pause here and take any questions.

Well, thank you, Christina. People can go to the microphones in the aisles and also post questions online, and we will be monitoring those.

And I see people already going to the mics, but I'm going to ask the question anyway.

I've got a couple. But so I'll start with one about the transmissions and your thoughts that they are coming from the environment into patients. And I wonder, I mean, that was originally the model with Pseudomonas aeruginosa in cystic fibrosis kids. And then it was shown that on top of that, there actually was a fair amount of transmission occurring in the hospitals. And so I guess I wonder if there are any strains like that have, that you're seeing repeatedly that maybe would suggest that there is some patient-to-patient transmission within the facilities and some strains that maybe have optimized a fitness landscape for colonizing human hosts?

We are working with a dataset now that— so the data that we worked with for the ACT-A trial was from these just 3 hospitals from Malawi. But we are working now with a dataset that's from a 4th location. And in that dataset, we definitely have strains that look like they are suspiciously very closely related. And so we have a call next week to talk to the people about this because yes, it does look suspicious that it could be, you would think it could be transmission, but I wouldn't, we haven't seen anything like this before, but yes, it does look suspicious. I think that there, you know, we kind of went down this road when there was in the news few, you know, pre-pandemic, a few years before the pandemic, there was a supposed Cryptococcus outbreak in this hospital in the UK where they thought they had contamination in like a garden inside the hospital. And so we worked with Andy Borman and a couple other people there and sequenced the isolates from patients and isolates they collected from They tried to collect isolates from the garden and they didn't even get any Cryptococcus. And so they just collect, they collected a couple other isolates from nearby. The long story short, nothing was related in that case. So a lot of these cases, there's just a lot of diversity. We do have this one case of this fourth, this other site in Africa where it looks like there could be some genetic relationships.

Hi, great, great talk. So, the aneuploidy sort of lower fitness yet more common in the isolates sort of locked in a question. Have you ever tried competing a virulent and nonvirulent in an immunocompromised mouse or rabbit? Because I'm wondering if it's not just colonizing— get better at colonizing people, but the immunocompromised environment is a specific niche and that these mutations are only useful in that context and actually reduce vigor in even a wild type or a normal immunonormal person, I guess.

That's definitely a direction we wanna go in that we are putting barcodes into strains so we can start computing them. And I, we had some of our work not on Cryptococcus but on Candida, we had seen some nice differences in strains in the gut but versus the blood. That, yeah, maybe we would start to see things in certain types of patients. I mean, we don't— I don't think for— yeah, we haven't worked in a— I have to think of— I'll have to talk to John Perfect about if there's a good— how we're accounting for the immunocompromised part in the animals.

Yeah, that's not— I was wondering about this massive fusion event that you're seeing across chromosomes. Could that somehow be caused by some sort of spore competition in meiosis where the larger chromosome spores inheriting larger chromosomes are somehow getting advantage over the other spores? Or if there might be like a selfish element, like a spore killer gene that somehow all the chromosomes want to be on that on the chromosome that has the sporcular gene or something like that.

It's an interesting idea. We would take any ideas like that because we've been trying to— I think we've been stuck figuring like why it seems like it would just be more of a disadvantage to be dragging around that big chromosome during mitosis. So yes, thank you for the idea.

I had an idea and then we'll go to the online. And, or I had a question, sorry, I didn't have an idea. Do you think that these could be used as a model for chromosome fusion or karyotype evolution? I mean, you've got it now, but are you seeing this happening, or could you use these, you know, species to try to force karyotype evolution and learn more about the process?

So, well, Joe has done an experiment where he has used CRISPR to induce a break and then sees what happens in Neoformins. So I guess you could think about doing those types of things, right? If your hypothesis is something was initiated by a break, there is— it seems that in Neoformins that centromere— centromeres are also favored sites of rearrangements. I think that's just part of the story is that, I don't know if it's because of proximity, if centromere clustering is part of what's going on and that's initiating some, you're just starting initiating rearrangements and then something goes awry. We have a question from Andrew Hazley. Sorry, I don't know how to pronounce your last name. He asks, could there be an underlying relationship between the aneuploidy tolerance and the unique karyotypic evolution in the environmental strains? Tie between aneuploidy tolerance and the unique karyotypic evolution in the environmental strains? Well, there seems to be less aneuploidy in the environmental strains. I think that— I mean, we've always more worried about like, when are we just losing aneuploidy when we're culturing things? Because is obviously, since it's a growth disadvantage, we've had some concerns we're biased just by culturing, we might be losing things. But yeah, it seems to be more of a factor in clinical strains. I think that the— yeah, this— the— we haven't— I haven't seen as much of, uh, I think I'm beginning to come convinced that there's a— the both the frequency is higher in clinical strains and that there, there's a specific chromosome. The biggest signal is chromosome 12, um, in clinical strains. Okay. And then I actually had a question, which was, did you look at the bird poop isolates separately from the environmental isolates? I mean, I wonder whether passage through a host actually— I mean, we know for Candida albicans that passage through a mammalian host increases the mutation rate. And I wonder if it's something specific about the mammalian environment or— I, so they tend to be more of the VN.1 isolates that are, that are more of the VN.1 isolates that came from birds as well as VN.2. We did also, we had this, this, the one case that's been linked to the environment came from a bird. It's, it came from a cockatoo. Someone had a pet cockatoo and they got Cryptococcus from their cockatoo. But there wasn't a series of isolates, unfortunately, from the cockatoo. It was just a single isolate. Great.

We've got our last question.

Great, Dr. Snyder. Quick question about the selective sweep in your slide. So do you have in your isolate, are they all haplotype or do you have diploid also? My question is, do you have a selective swipe or could you have also clonal interference in your dataset?

There are some diploid Cryptococcus that we, we, we, we did look at that separately in the genetics paper that we published, but there, I didn't talk about any of those here. Yeah.

Okay. Thanks so much, Christina, for a wonderful talk and for visiting us at NIH.