15:09:06 Unfortunately she couldn't be here in person, but I'm sure this talk is going to be great regardless. 15:09:10 So take it away. 15:09:15 Great. Yes. 15:09:18 I was hoping to make a quick visit up to Santa Barbara but yeah as I as I just mentioned coven had other plans. 15:09:29 Anyway, happy and healthy here in San Diego, and excited to get to share my work and the work of my lab, and actually some of the work that we started in 2017 when I was an instructor for this course at KTP. 15:09:47 So, what, what is the question that really drives the work that we do in my lab. 15:09:59 The question that I really 15:10:03 started to, to think about when I was finishing up my, my graduate work was this idea that we have probably missed out on a lot of super interesting biology, because we usually say organisms in isolation, and I was guilty of that my PhD training was in 15:10:23 in equal light genetics. 15:10:27 And, but as more and more studies came out about the diversity of the microbial world. And I think one common theme was that these microbes, live in the context of communities they don't grow in isolation. 15:10:41 And so I became very fascinated with this idea that that there's probably a lot of super interesting biology out there, if we can find ways to approach it. 15:10:52 Now, what are the ways that biologists often approach biological problems well that is through the use of model systems model systems have been incredibly powerful through, you know, over for over a century of biological research, and the model systems 15:11:12 can range from, you know, a number of different types of organisms depending on the questions that somebody might be interested in. 15:11:20 So things from model systems based on faith. 15:11:26 Bacteria yeast. 15:11:28 C. elegans and so on. But at the time when I was starting to think about this, there weren't very many model systems that were being used to study the biology within microbial communities or microbiome. 15:11:44 And so I started to think about how I might approach this problem, as somebody who's trained in the use of model systems to study biology and what are the types of things that I might that what are the types of characteristics that I might look for in 15:11:58 potentially setting up system to study microbiome. 15:12:04 And one of the concepts that I kept sort of coming back to was that there's this trade off between her potential trade off between how experimentally tractable, a microbiome is and the potential you might have for insight into some of these novel aspects 15:12:22 of biology found within communities. And so this is just a sketch I made a couple of years ago, where if you have the number of species on the y axis here. 15:12:33 So, you know, a typical microbial community would could have, you know 10s hundreds thousands of species living together. 15:12:42 However, as you as you increase the number of species, the experimental track stability of that system declines quite rapidly at the moment I think you know there's many ways we can push that curve out further. 15:12:58 But this was sort of my estimate for how that curve looked, and then the opposing curve, I think it's this potential for insight. So, we want you know at least a handful of different organisms so that we could start to learn some of this interesting biology, 15:13:12 but I think probably at some point this this curve also maybe maxes out. I don't know where that actually is. 15:13:20 And so what I was thinking about is, you know, is there somewhere. That can hit at these curves in the right place, something that that is experimentally tractable but still has some significant potential for insight into new biology. 15:13:35 And so I thought that maybe there's this Goldilocks zone for a model community so similar to how planetary scientists think of, you know, can we find planets that are not too hot, not too cold for life can we find microbial communities that are not too 15:13:49 simple but not too complex communities that we can still work with but have enough interesting biology within them. 15:13:55 So which communities fit that mold. Well, the one that the ones that I landed on where the microbial communities that are found in fermented foods. 15:14:08 And so these communities. As you may know, are highly reproducible humans have been cultivating this communities for thousands of years. They're relatively simple. 15:14:21 And we, we think we understand, you know, what they grow in and therefore, we might have some ability to culture, the organisms that that grow in these communities. 15:14:31 And so I had proposed using one of these systems in particular, as a new model for understanding communities. 15:14:41 And the fermented food that I chose to focus on was actually the communities that form on the surface of cheese, as its aging and so this is called the cheese rind biofilm. 15:14:52 Now, when a cheese maker makes a cheese they take fresh milk, they ferment it, they. 15:14:58 After the fermentation that the acid from that fermentation causes the protein to coagulate, and traps the fat within it. And the water leaves, so the Kurds are formed into a whale of cheese, and that cheese is placed into an aging room, so a cave, which 15:15:14 is cool and humid. 15:15:19 And during that aging period which can last from weeks to months or even years. A microbial community forums on the surface of the cheese. So you can see here these top three panels. 15:15:30 These are three different types of cheese that have three different types of microbial communities that have formed on the surface, while in the cave. 15:15:38 Now these communities, we found range from low to medium complexity so anything from three species upwards of 20 or so different species living together. 15:15:51 Despite being not super complex in terms of the number of species they're actually phylogenetic Lee quite diverse. So they have representatives from across several different bacterial phyla, as well as filaments fungi, and yeast and as we're learning 15:16:09 more and more. 15:16:11 Tons of faith in them as well. 15:16:15 We've shown that these communities are completely cultural and and amenable to in vitro reconstruction. And so what I mean by that is that we culture all the isolates from a given cheese, and we can culture them in isolation and then we can rebuild communities 15:16:32 in the lab in our in vitro cheese system, and that showed on the right, this is one of our 96 cheese plates, where we've inoculated the surface of cheese based medium to recreate biofilms that grow on the surfaces of these walls. 15:16:48 So, when I first set out to work on the system. 15:16:55 I had no background in food science. 15:16:58 I had no background in studying microbial communities except for having taken the MPL microbial diversity course of what's whole as a grad student, but I managed to convince Andrew Murray, and the Bower Fellows Program at Harvard, to give me the opportunity 15:17:18 to set up set up the system from scratch. And so over my, my term of five years there. I, along with two very talented postdocs and several other undergrads work together to basically set this system up from the ground. 15:17:36 So we went out and we sampled cheeses from around the world, we profiled the microbial diversity that was present in the system. 15:17:44 We showed that we could actually cultivate representative members of, of all the general that we found by sequencing, so we could actually culture, the members of this community which is often a big challenge in setting microbiome. 15:17:58 And then we show it as I mentioned that we could actually rebuild communities in the lab that mimic what we saw in the cheese caves. 15:18:05 So super grateful to have had that opportunity to just set up a new system be given the space and, and creative freedom and resources to do that which is not an easy feat. 15:18:22 So after that five years, and since I've been reading a lot at UC San Diego. 15:18:29 We've basically been now. Now that we have the system now we want to figure out how it works. 15:18:35 So we've been working on methods for analyzing these communities and understanding the basis of microbial interactions within the system. 15:18:47 And so our approach, sort of spans many different types of methods we, I think one of the nice things about working with this type of system is that we can move between the study of the Institute communities in the caves in the cheese making facilities, 15:19:02 and in vitro communities in the lab, which are reduced complexity, by necessity. 15:19:08 So, we can use sort of top down approaches that move from Institute to in vitro using things like comparative to know make sense metagenomics which I'll tell you a bit at the end of the talk today. 15:19:19 And then, a lot of the time we've spent on focusing in the lab is on bottom up approaches where we start with the in vitro communities. We have developed approaches for doing genetic screens, which I'll tell you about today, as well as in vitro community 15:19:32 minute manipulation, that give us insight into the types of processes that are happening between species in the system with a goal of eventually sort of going back to the NC to Jesus, and showing that that is relevant. 15:19:50 Okay. 15:19:53 Any questions so far I'm happy to take questions as, as they go along. 15:19:58 I think we're good for now. Okay. Great. 15:20:03 Okay, so the first part of the talk today I wanted to tell you about the work that we've been doing on high throughput genetic screens in cheese, and this is work that's been led by to really fantastic people in my lab. 15:20:15 My Nolan moron Who is she actually was a TA for the KTP course, a couple of years ago when I came up and Emily Pierce, who was a PhD student who took the course, and she's now recently graduated by first he sitting graduated in January, she's now a postdoc 15:20:32 in the lab, and they have, they basically they basically invested the first several years of our time here at UCSD just getting this type of system setup for looking at my computer interactions. 15:20:45 Actually Rachel I do have a quick question before you genetic screen. Yeah. 15:20:50 How do you think about can you educate me a little bit again sorry about what these communities are doing on this, on the surface of the cheese. That is when they create the rind is there some sort of relatively reproducible metabolic process that's happening 15:21:05 there. 15:21:05 Yeah, variable from you know Greer to come on there or whatever, right. So, the, the metabolic processes that happen on the right, it varies from cheese cheese and as you saw the types of communities are different, depending on the type of cheese and. 15:21:23 And in a way, cheese makers have basically are basically enriching for different types of communities because they possess different types of metabolic activities. 15:21:31 And so, in this type of cheese that I have a picture of right now like a camera bear style cheese this, we call them blue me rain cheese's. 15:21:39 The, the bacteria and fungi are a lot of that sort of goofiness that you get in the cheese is actually from D acidification that happens as well as pretty all Asus. 15:21:52 So basically the, the organisms that are growing on the surface are creating enzymes in this case they should creating a lot of pretty aces into the cheese that helped break down the protein and destabilize the structure of the protein network that's 15:22:06 present inside the cheese. 15:22:10 In the process of, of that their metabolic metabolize those peptides and amino acids further and further along as the longer you age the cheese and those create a ton of volatile compounds, as well as other small molecules that we we taste when we eat 15:22:27 these cheeses. So, depending on the microbe the microbes that you have on the cheese you end up getting a different combination of activities the metabolites and volatiles and so that the texture and the taste and the aroma of each she's could be different. 15:22:46 Does that answer the question. 15:22:49 Okay. 15:22:50 Thank you for the thumbs up. 15:22:54 Okay. 15:22:56 Yeah, I have like a whole other like our lecture on, you know, cheesemaking. 15:23:03 We can we can force can ask me to come later in the course i'm going to do like a cheese tasting and and lecture, coordinated. 15:23:12 Okay. 15:23:14 So, how did we set up, how do we identify interactions between species in this community. Now, one of the problems with working in a model system that is made of non model organisms. 15:23:31 Right, so the species that we have in our cheeses are not things that have been developed for that have like standard genetic tools developed for them or anything else like that. 15:23:43 And so we had to come up with a sort of way around getting around the the fact that we don't have access to a lot of genetic tools for some of these organisms, and the approach that we came up with was this idea that maybe we could use equalized libraries. 15:24:00 Equally mute and libraries as a way to sense the types of interactions that are happening within a given community. 15:24:08 And so we sort of caught. Think of this as it equally as an interaction sensor in this case. 15:24:14 So, what we do is we take large pools of equal I mean library so these are random barcode transpose on sequencing libraries, and we introduced these libraries into our communities. 15:24:30 And depending on what the species in the communities are producing or using it changes the environment that Nikolai is finds itself in. And so, you know, in the case of maybe these, so these little oval cells are equal light with mutations in different 15:24:48 parts of its genome, and maybe these cells that have died had mutations in parts of the genomes that were specifically important in the context of this community. 15:24:58 So I'll walk you through this sort of step by step of how we actually find those jeans. 15:25:04 So you get a better sense for how this works. 15:25:09 Okay, so we have our, our starting library Nikolai in this case the whole I library has over 150,000 unique insertions and it's genome and each insertion is tagged with a unique barcode. 15:25:22 So we can measure the starting population of barcodes within these libraries. 15:25:29 And then we can take the library and grow it by itself in our, our environment of choice which in our cases are in vitro cheese medium. 15:25:38 And then we can look and see what the barcode abundance is after the scrip and cheese, and so this tells us something about what are the genes that are specifically required for growth in a cheese like environment. 15:25:49 When you call it is growing by itself. And so you'll see there's changes in the frequency of some of these barcodes that after we measured them. 15:25:58 And then the real experiment is to now grow this library of barcode mutants on our cheese environment in the presence of a community. 15:26:09 And then again we sequence the barcodes, at the end of this process, and look for changes in the, the frequency of barcodes, compared to the growth alone condition. 15:26:22 And so by going through this process of comparing the barcode abundance under different conditions we can actually detect a couple of different interesting patterns. 15:26:33 So, in this case, we see a barcode that was present in the starting library, but it's absent in both the cheese and cheese plus community condition. So this suggests that this is a gene that is normally would just normally be required to grow and achieve 15:26:49 like environment, it doesn't matter if the community is there or not. 15:26:55 In this case, we have a bar code that's present in the initial library, its present when the equal a library scoring on cheese, but now it's no longer present in the community condition. 15:27:06 So this suggests that there was something changed in the presence of the community that made this no longer able to grow. So, this we call this an interaction induced requirement, Nikolai. 15:27:18 And then there's others that have interesting mutants, where we find them in the initial library, they're not in the cheese alone library so they can't grow the by themselves on cheese. 15:27:30 But in the presence of the community. Now, the barcode frequency has recovered. And so, we think of this as an interaction alleviated genetic requirement for the cola. 15:27:42 So this is the sort of general setup for how we 15:27:47 do this type of experiment. 15:27:50 And the first set of of the first community that we decided to apply this to was this in feature Cameron Bure model. And so Cameron Bure one of these blue me Ryan cheese's it falls into these relatively low diversity, communities, and we have a model 15:28:09 in the lab of this community that's made of only three, three different species so half the alpha which is a bacteria to gamma per do bacterium jam candied um it's actually used to put it grows in a filament as form and Penicillium Camembert t which is 15:28:24 a filament as fungus. 15:28:26 And so what we do is we take this, these three species we grow them up separately and then we mix them back together and inoculate them onto the surface of a cheese curd auger plate. 15:28:42 And in addition to that community will add the sensor to that. 15:28:46 The sensor library to that community, as well as for the sensor library on its own, and cheese. And so then we can measure the fitness of each of the mutations in the whole I library. 15:28:58 And so these plots are looking at the the mutant fitness and we're only looking at the negative fitness here. 15:29:04 So this is a scale from going from zero to negative four. So the more negative the fitness means the worst that meeting is doing, compared to wild type or a No. 15:29:16 Sorry, wild type or insertion and energy that that doesn't have an effect in that in this condition, and the y axis is the T score. 15:29:27 Look at the significance of that estimate. 15:29:31 And so in the sensor alone condition so this is just the whole life growing by itself on cheese. We have 160 genes that have negative fitness, in this environment. 15:29:43 And in the sensor plus community we have about 126 genes that are required, or that have a negative fitness. Rachel. Yes, um, how, like I Kurt Can you like how reproducible Can you make these measurements like the if I put Aero bars on this over like 15:30:01 if you replicates. 15:30:02 How big would you imagine maybe. 15:30:06 Yeah. So, for each. So there's a couple of different ways to think about that each gene in this library has about 15 independent insertions and so the way we do these fitness calculations is averaging across all of those insertions and doing, you know, 15:30:24 statistics on the reproducibility of of the measurements within a single gene. And so in this case all of these measurements are very high confidence. 15:30:36 Because of that, because of that high density of of insertions. Does that make sense. 15:30:41 And then, among the, let's say you have 15 different insertions, 15:30:47 are they so you're kind of averaging them out, are they more, let's say the same insertion in a different replicant would one be consistently more fit than the other, or is it indeed kind of noise. 15:31:01 Right, yeah I don't know if we've looked at that specifically we so we do this across triplet kids now I actually think in this first paper we didn't implement a triplicate statistics into it but we found for for the next piece that I'm going to tell 15:31:18 you about we actually needed more confidence in our estimates. 15:31:27 So I'm not sure. 15:31:29 From replicate to replicate. So we're kind of just in this particular one, we're just treating the independent insertions as kind of independent replicates. 15:31:39 Thanks. Yeah. 15:31:42 But in the end, the next set of experiments that I talked about we do actually, we had to develop a whole new statistical pipeline for for accounting for the potential variance between replicates and within a go. 15:31:57 Rachel. 15:31:59 I have a question. Um, so when you culture this equal a library with the community the cheese community. 15:32:06 Does this co culture affect the cheese community composition. 15:32:11 Yeah, so we we looked at that. 15:32:14 So, equal the presence of a cola doesn't have a significant impact on any of the community members growth, but the community members all make equal ly grow less than it does when it grows by itself. 15:32:43 With the fact that we are using equally library effects. The these patterns. So is this species dependent. These results. 15:32:48 So, sorry. Can you say that again. 15:32:51 So, like, I know that there's only the library or we're local for equally. But if you assume that we are using the library mutant library of other species. 15:33:05 And if you use that in that in this experiment with that change the results that we're seeing, so do the other species respond to the presence of Nikolai is that, I mean, it does the fact that we are using the equal i is that is that affecting the results 15:33:24 of the genetic signals. 15:33:28 I see what you're saying, Okay, I think, so I think what you're saying is if we were to use some more native member of the community and use the library and would that change the results. 15:33:38 Yes, Yes. Yeah, so we did that, we tried to do. 15:33:47 Pseudomonas species that we isolated from this type of cheese and added it did the same types of experiments as we did with the whole I, and there are there are some genetic signals that are similar, but 15:34:08 it actually became really challenging to figure out how similar they were because the Pseudomonas that we use didn't have a well characterized genome like we sequenced the genome but there's so many genes that we're on characterize it was really hard 15:34:30 know whether we were getting the same types of signals in Pseudomonas as we were in Nikolai. And, and we tried to do some direct comparisons of Okay Are they like autologous genes that are being, you know, responding between the two but that was also 15:34:39 really challenging to do so. 15:34:43 It wasn't as helpful as we had hoped it would be. And, and that's kind of why we've keep falling back on the whole life just because it's genome is so well characterized we can leverage that pretty well. 15:34:57 Thank you so much. Yeah. 15:35:00 Yeah, but all that data is in is in this paper that was published and then the next study as well we, we also did it with two bonus. 15:35:07 And we're continuing to try and make libraries and some of our cheese species. 15:35:13 And, which seems fine and gamma party a victory we can do it but once you get outside of gamma party bacteria gets really tricky. 15:35:21 Okay. 15:35:24 So, What do you, how did these two conditions compare sec have kind of similar numbers of genes that are have a fitness effect in these conditions. 15:35:35 So when we look at the overlap. There's about 89 of these genes that have a negative fitness in both that alone and the community condition so they see these core requirements that I referred to. 15:35:49 Then in orange, we see genes that are don't have a negative fitness in the alone condition but now do have a negative fitness in the presence of a community. 15:35:59 so these, we would call community induced requirements. 15:36:05 And then this, I think super interesting category of genes that do have a fitness effect in the alone condition so if you basically get rid of one of these jeans equally does not grow well when it's by itself, but now it grows just fine when it's in the 15:36:23 presence of a community so these are these community alleviated genes. And when we look at this gene it's quite the set of genes is quite a large number and so we went back and looked at the functional categories that were present in this community alleviated 15:36:35 set. 15:36:37 And it turns out that most of these genes were found in amino acid biosynthesis pathways. 15:36:44 You can see here. 15:36:53 Genes, and Rick required for the biosynthesis Valium leucine leucine for example. So these genes if you got rid of them equal like can't grow by itself, but in the presence of the community now equal like doesn't need these amino acid biosynthesis jeans 15:37:03 anymore. 15:37:04 And so what this suggested to us is that when he call he is growing by itself, even though it's surrounded by protein I mean cheese is like mostly protein, it actually needs to make its own amino acids, so we think that it doesn't have the capacity to 15:37:20 break down the protein around it so that it can actually access those amino acids around it. 15:37:27 And so it actually needs to make its own amino acids, even though it's surrounded by protein. However, in the presence of the community, it doesn't need to make its own amino acids anymore suggesting that there's some type of prospecting that's happening 15:37:41 in the community. 15:37:44 Now, we didn't know who this cross feeding was coming from although we could have made some educated guesses based on that, but I think I wanted to highlight here. 15:37:53 One of the nice things about having one of these in vitro community, is that you can grow them in whatever combination of species you want. And so what we did next is we took the equal I library, and we grew it with each individual strain, each, each 15:38:10 individual species from the community. We grew them with the, all the pairs of species and the complete community. And so we can actually try and parse out, who, who any of these given effects might be coming from. 15:38:25 Now in the case of this potential amino acid crossbreeding what we saw is that it seemed to be the filament is fungi and the yeast geometry that we're providing this alleviation of amino acid synthesis requirement. 15:38:42 And so what we think is happening is that these fungi are so creating produces, which is breaking down the large full length case in protein into smaller bits that equal I can actually use. 15:38:55 Now, the other thing that we saw when we go through and do the sort of comparisons of gene fitness across different scales of complexity. So whether it's single partner to partners or three partners. 15:39:09 What we see is that the effects on gene fitness in about half the cases are not linear. And so we think about half of the interactions are subject to higher order effects, which means that we wouldn't necessarily be able to predict the interactions of 15:39:22 this three member community from just looking at the single or double interactions. And so I'll show you just one example of this, this is higher order impacts on a multi drug effects system and Nikolai. 15:39:35 And so this is the ACR AB multi drug effects pump. 15:39:41 Beautiful crystal structure of it here. 15:39:43 Now these two jeans when you call it is growing by itself, don't have a negative fitness, which, as you can imagine, would be the case. However, when you grow equally with geometric come. 15:39:55 Now, these genes become important, so there's a negative fitness if you get rid of either one of these, suggesting that you have to aim is producing some sort of antimicrobial compound that causes equal it need these this multi drug effects pump. 15:40:12 However, the presence of the community. Now these genes are no longer important suggesting that there's some higher order effect going on here so one of the community members maybe is doing something to prevent the production of the antimicrobial or detoxify 15:40:28 it in some other way. And so again we went through this process of kind of narrowing down who might be the causative player in this higher order effect. 15:40:38 And what we actually found. So, is, is this is the bacterium half via seems to be the the member that we can attribute this at activity to, we still don't know exactly what the mechanism is we think it might be creating some sort of anti microbial resistance 15:40:56 factor that equally. 15:41:00 So basically detoxifying the environment so equal I no longer needs its multi drug effects pump. 15:41:06 So, one of the points that I think is super interesting just by looking at this relatively simple community is that we see these interesting, you know, different types of interactions and different and relatively complex interactions. 15:41:22 So, in this case this fungus theatrical is both providing resources for Nikolai but also providing producing antimicrobials that would potentially inhibit its growth. 15:41:34 However, one of the other bacteria in the system can actually modify that activity. 15:41:40 Rachel. Yes. Have you tried the experiment of getting a co white to secrete a protease, to see if it will now grow in the absence of the other guys. If it has amino acid knockouts yeah we haven't done that. 15:41:59 No Andrew we haven't. 15:42:02 But that would be cool to do, we have Yeah. 15:42:07 I wonder, yeah, it would be cool to maybe like put one of the fungal pretty faces into, into Colin haven't expressed that. 15:42:17 Okay, any other questions on this part I'm going to move on to the next section. 15:42:31 Okay. 15:42:31 So, a couple of hours. 15:42:35 Before we move on, sorry it took me a while to find the microphone. 15:42:39 If you had to sort of rank, the number of genes that are essential as a function of the number of species in the community. Would you say that that number sort of went down in all of your combinations or kind of very context specific interesting so did 15:42:58 the total, so did the total number of jeans with the negative fitness go down in the presence of the community is that the question. Yeah, essentially, yeah. 15:43:07 I haven't ever thought of it that way. I mean, we definitely have some genes that become more important in the context of a community, and some that become less important so I guess I'm just wondering how things would change as we scale up if you had 15:43:24 a 10 species community would you expect a lot of these sort of biosynthesis kinds of genes to be supported because somebody is probably chopping something up somewhere right right. 15:43:35 Yeah, super interesting question. 15:43:38 I don't know 15:43:41 anything else. 15:43:43 I can actually ask something Hi there, I'm just wondering, so it seems to me that this is looking at genes that are specific to equal i right i mean all the knockouts it's all equally jeans, how relevant are representative, do you think that is what might 15:44:00 actually be going on in the cheese microbiome. 15:44:04 Yeah. So it's a great question and I think related to one of the previous questions in the sense that we've also tried doing this with some of our native isolates from cheese and making these libraries and trying to sort of use that to tell us what's 15:44:21 important with, I guess, not as much success as the COLA and so the way I think of it is really that Nikolai is sensing, what is what's happening in the environment so. 15:44:36 So, this these experiments have told us that, you know, there's crossbreeding going on, you know the the fungi are chopping up proteins that other community members would be able to access so I don't think that that necessarily is going to be equal I 15:45:03 Again we think that it's sort of like monitoring the environment a bit. So I think there are some trade offs obviously with using equal live versus versus non, I guess, versus more CO evolved our native members of the community. 15:45:05 specific whether you know other species would be able to benefit from those types of interactions, and I'll show you in this next part as well. 15:45:24 But I think a lot of what a lot of what we end up doing in our lab is we're sort of kind of straddling the line between track stability and, like, you know, more natural. 15:45:43 And, you know, sometimes we have to push things more into the stability side because we want to really get specifically down to these mechanisms that, and, you know, equally provides us a way to really pretty quickly see some of the processes that are 15:45:59 happening. 15:46:00 But I'll mentioned this at the end, I mean, 15:46:05 even a Nicole I like about half the genes that we pick up our hypothetical genes we don't know what they're doing. 15:46:11 It's much more in in model organisms when we put them in here. And, and another major caveat with any of these experiments, is that really we can only easily interpret findings that are related to genes that we already know the function of. 15:46:28 And so we focused here on this last part on amino acid biosynthesis because we know very clearly what those jeans do and so we can very quickly make a hypothesis and test it in the lab. 15:46:41 So I think, you know, with any of these systems that's I think a big roadblock to our interpretation of what's really happening in these communities, is that we just, there's so little we know about the biology within communities that, that, you know, 15:46:57 it's going to take a lot more years of effort to, to figure out what's going on. 15:47:04 Yeah, sorry that was kind of a long, but you definitely hit, hit on something there that I think about a lot. 15:47:12 Yeah. 15:47:13 Yeah. Thanks so much. Yeah. 15:47:17 Okay. 15:47:22 Um, so, the, this next part I wanted to tell you about a slightly different version of these experiments that was done by my grad student, Emily Paris. 15:47:32 And she was really interested in what specifically different fungi might be doing in these communities we have this diverse kind of collection of fungi that we've isolated from cheese and fungi are very understudied in microbiome is in general, most people 15:47:47 who study microbiome is focused on the bacterial components of the systems. So we were super curious to know what what these fungi might be doing. And so she set up a similar type of experiment where she grew each of these different fungi in the presence 15:48:02 of this equalized sensor library. She also did it with the student library but I'll just show you the equal results today. 15:48:12 And I'm going to summarize, her data in in basically this one slide, which is what took us to the next set of experiments. So, there were two. So this is looking at the the fitness alone on the y axis so each of these dots is a gene and its fitness score 15:48:34 is alone is on the x axis and the average fitness with the fungal partners on the y axis, and she's color coded these based on their functional category that they're involved in, and the size of the.is based on the number of fungal species that, that 15:48:53 she saw a fitness affected, so large circles mean that's fairly conserved affect that many fungi are having the same effect on this gene 15:49:04 and. 15:49:06 And so basically, she's looking at genes here that are either significantly different that are all significantly different. in the presence of fungi versus alone. 15:49:17 Now, one of the things that she saw were detrimental effects on cell wall, integrity, which I'm not highlighting here, but the biggest effect that she saw was actually on iron metabolism. 15:49:31 So, these are these pink jeans fairly large so these were things that were conserved across almost all of the fungi, actually all of the fungi that we tested. 15:49:42 And what you're seeing here is that the, the fitness alone for these genes is around, minus five. However, in the presence of any of the fungal partners, this average score but it's similar across all of them. 15:49:55 The fitness now is boosted to about, minus two, or so. So there's a significant increase in the fitness in the presence of fungi, and any of these genes that are related to iron metabolism. 15:50:11 Okay, so what is going on here. 15:50:14 So, it turns out, these are all genes involved in the dare for uptake. 15:50:19 They're required alone, so the alone condition fitness for each of these genes is in this empty circle at the bottom, so strong negative fitness alone. 15:50:30 And then this is their fitness score and the presence of each of these fungi so all of the fungi are boosting the fitness of these mutations. 15:50:41 Now what is going on here. 15:50:43 So, what we think is happening is actually, this is showing that maybe bacteria can actually steal fungal severe force. And so, in Cydia for our production of bacteria the bacteria secrete these iron key later molecules called Sudhir force that helped 15:50:58 them create iron from the environment, they export them, and they take them up through their own specific system for the city or four. 15:51:08 So what we were seeing is that when you get rid of the uptake system for equalized city or for inter Bacton, you actually decrease the fitness of equally significantly so it can't take up its own city Air Force, it means that basically starves for iron. 15:51:23 However, in the presence of any of the fungi, what we saw is that the fitness was no longer as dramatically decreased. So what we think might be happening is the fungi might be secreting their own city Air Force, and maybe Cole I can actually take up 15:51:40 the fungus of Air Force. And if we look when we look back in the literature from about 20 or 30 years ago. It was initially reported that e coli has uptake systems for fungal severe force, and one of these systems is the few system, FHU. 15:51:58 Now, as far as I can tell, nobody's ever really thought about why equal I might need fungal so therefore, uptake systems, or, or what, why, or when it might need them. 15:52:08 But it has them and they've actually been fairly well characterized genetically and biochemically. 15:52:14 So we knew exactly which genes to potentially test to see whether these are actually fungal serious we're actually moving through this few system and that was leading to the fitness benefit. 15:52:28 Okay. 15:52:30 And we also tested this from fungi, from different environments, including skin and soil, fungi. 15:52:37 And so, 15:52:40 sorry. 15:52:42 Yeah. And so basically what we found is that when we knocked out the the few system the equal I could no longer benefit from the fumbles of Air Force, suggesting that it what they were actually going through the system. 15:52:55 And as I said, we, we tested several other fungi so we think this is actually telling us that this is maybe a fairly common phenomenon in, in nature where fungi are creating their own severe fours and bacteria like e coli but these few systems. 15:53:11 When we look by informatics Lee they're broadly distributed across bacteria that bacteria generally may be benefiting from the presence of fungus of air force in the environment hi Rachel. 15:53:25 Yes, that was loud, I had a question. So, these fungal. So the city or for uptake genes were sort of rescued by the presence of fungi, but there is there any similar pattern for the production of the city Air Force. 15:53:43 No, I might that be I mean his production. 15:53:46 Yeah, so getting rid of it is like bad you lose the iron but good and that you produced some major investment there and well it's actually less interesting that than that in this particular case because we actually can't see the effect of secretion, because 15:54:01 of the pool nature of these libraries. And so, so let's say you made a mutant of the export system, but you're growing it in the context of this pool of mutants. 15:54:12 So all of its neighbors are actually producing so therefore still, even if that one mutant cell is not. And so you're getting actually crossbreeding within the library that would complement the equalizer of indigenous system that doesn't work with the 15:54:28 the uptake so we can actually see effects on the uptake but not the biosynthesis. 15:54:34 One more quick question. Yeah, it was iron intentionally limiting in this system and, like, what, what are the limiting nutrients in a cheese like environment. 15:54:44 Yeah so iron is limiting in cheese and any dairy product. 15:54:49 It's limiting in many different environments. And so, iron limitation and sort of fighting over iron is a very common theme in microbiome. 15:54:59 And so, I mean based on our analyses, I week. 15:55:04 There's several other prop stories that we have that all basically come back to iron, as a major limiting nutrient in the system. Yeah. Thanks, that we didn't intentionally do it it's just dairy is sort of rip normally low in iron concentrations. 15:55:29 Additionally have another question. 15:55:32 I'm sure, presumably, we can tell that there's a bunch of, you know, iron being secreted and a bunch of amino acids being secreted but is there any sense in which you can figure out by 20, equal lion which species in your system might be using these amino 15:55:48 acids natively for instance, maybe by throwing the equalizer abundances have some species Go, go down. 15:55:57 And maybe those. 15:56:00 What court in the natural Jesus be eating these cider fours. 15:56:07 Yeah, that would actually be really interesting to do I don't think the effect is very strong with the pool approach because it's like one and, you know, several million cells that have that particular pathway but if we were to just take the. 15:56:23 Let's say the mutants or the biosynthesis means alone and put them in the community that would be Yeah, that's a that's a super interesting idea. 15:56:33 Thanks. 15:56:42 Okay. 15:56:45 All right, I'm going to move on. 15:56:51 Okay, so I think what these genetic screens have have given us is is a window into species interactions it's not telling us everything but it's, it's one way to sort of probe, what's going on and what it's told us is that these communities are succeeding. 15:57:13 A lot of things into the environment like enzymes and amino acids and Cydia fours antimicrobials. 15:57:22 And, you know, we, one of the things that I think is super interesting is that, in the case of fungi which appear to be doing many of these, these things, I guess. 15:57:36 One of the things that's interesting is that the fungi are having such a big impact at all, and especially given that they're relatively understudied in the context of microbiome. 15:57:48 And so I think the sort of at least to me is that we should be paying more attention to fungi and maybe other microbial eukaryotes in the context of microbiome because they have this ability to potentially alter the growth of the bacterial members of 15:58:05 the community. 15:58:07 And then the other thing that I think is really interesting from this is that, you know, a given fungus, for example, can both be doing stimulatory and inhibitory things at the same time. 15:58:20 And I'm really interested in thinking about like how do those different types of interactions sort of add up or or not or you know like how does how did those get processed by the microbes. 15:58:32 How does that end up, you know, resulting in a final composition of a community based on maybe who's stimulated and who's inhibited. 15:58:42 And so I think these are just sort of things that have come out of these genetic screens that are interesting avenues for future research. 15:58:52 And the other thing that I mentioned before is that over 50% of the genes we identify even an equal is important interactions have unknown functions. So, a huge amount of space and opportunity for discovering new biology here. 15:59:09 Okay, so I wanted to 15:59:13 fit in one more story because it ties back to something that I started at KTP a couple of years ago. 15:59:20 And so if I have the time, which I think I do. 15:59:26 Although somebody can stop me if I'm running over Okay good. 15:59:29 So I wanted to talk about horizontal gene transfer. 15:59:33 So this is a different type of interaction than the sort of interactions that I've been talking about so far but when I was first starting this system I started to think that maybe cheese could be an interesting environment to study horizontal gene transfer, 15:59:47 and there were a couple of reasons for that. One is because, you know, in some cases, these organisms are coming from different types of environments and they all end up growing together on the cheese so they could be coming from the air of the caves 15:59:59 they could be coming from the skin of the animals from the skin of the people who are making the cheese at different places and ending up growing together so they might be meeting each other for the first time. 16:00:11 And so that might present novel biotic pressures, so you know maybe there's new antimicrobials that they've never seen before or other pressures that that they need to figure out how to deal with. 16:00:23 And there's also a lot of interesting, a biotic pressures in cheese that maybe some of these organisms have never seen before, so it's low pH, salty. 16:00:34 There's, you know, differences in moisture. So, as well as things like iron limitation, and other resource limitation that that could, you know, lead to microbes having to rapidly adapt to these changes. 16:00:51 And so, you know, how do microbes often do that will often that's through horizontal gene transfer, and we often think of this in the context of antimicrobial resistance where you know gene can be transferred through different mechanisms like fade or 16:01:05 conjugation or free DNA. 16:01:07 But this is happening all the time in microbial communities and so I thought you know maybe we could use cheese, as, as a system for studying some of these processes in the lab. 16:01:18 And the other side of that is I thought, well, it might be interesting to actually look at what genes are being selected on and and transferred and maintained in these genomes, as a way to actually tell us what is important in this environment and in 16:01:32 these communities so you know if we find genes that are related to specific functions that have been horizontally transferred in the species maybe it's because those genes are specifically important in the context of this system. 16:01:46 And so this project was started by a postdoc in my lab, when I was back at Harvard Kevin bottom. 16:01:52 He took a computational approach, which was based on an approach developed by Eric arms lab at MIT. And what he did was to look for regions of high identity between pairs of genomes between different species. 16:02:09 And so if we look for pairs of species from sequence genomes isolated from a species isolated from cheese, where the genome wide identity was less than 89%, but then there were regions of high identity of 99% or more over at least 500 base pairs. 16:02:27 And so we mapped out the this pattern of of potential HTTP and cheese, and we see extensive evidence for att across these different bacterial species and cheese so over 4000 genes were found within these. 16:02:45 These HGTV regions. 16:02:49 Now, when we look at what genes are present in these regions, it's not just a random collection of teens of the things that are, that we were able to annotate and functionally characterize actually not on this graph art like transpose on, and other mobile 16:03:04 element related genes which are where the highest category, but the second highest category of things that we knew what they were. Are these this category of metallic hadn't iron so therefore and vitamin B told transport systems. 16:03:17 So a huge number of genes present in these HTTP regions that are probably related to iron metabolism based on what we know about cheese. 16:03:26 Yeah. 16:03:29 This graph that you were showing of the GT and Jesus. 16:03:33 How do you know that that a GT occurred while these microbes are in the cheese, and not, you know, a long long time ago somewhere else. 16:03:41 We have no idea. 16:03:44 Yeah. So, I mean, they. 16:03:48 I guess the only evidence we have to suggest that is that these organisms were often independently. 16:03:55 Like isolated from cheese itself so these are all cheese related organisms and we see evidence of HTTP or shared regions between their genomes, but other than that we have no idea when these exchanges could have happened I mean they could have been cohabiting 16:04:11 on a calendar, and it happened there. 16:04:17 Yeah, but also presumably while your cheeses are growing you could see you know genes that you didn't see on this graph before but do you see because of growth in your communities. 16:04:31 Is that is that something you're able to see or. 16:04:33 So if we can see it, like, while the cheese is being made. 16:04:37 Yeah, basically like you sample, all of these microbial genomes before your cheese is growing and then after a while your community is assembled and run yeah yeah so one of my postdocs, Christina sock is is analyzing data set like that right now or went 16:04:53 went in and sampled cheeses as they were aging over time, and applying a ton of different sequencing based methods to actually try and see if we can detect het while it's happening during the development of one of these communities. 16:05:08 Cool. Thanks. Yeah. 16:05:12 Okay. 16:05:13 Oh, can I ask a quick question from zoom. 16:05:17 Um, but relative like relatively fractionally how many of the bacterial genomes have evidence of like what percentage of their genomes is occupied by hgt type. 16:05:30 I don't know if we estimated that. I mean, there's going to be a lot more HTTP that they've experienced from other sources. These are just the ones. 16:05:42 When we look at these particular set of genomes. 16:05:45 So it would be, I think, a pretty dramatic underestimate put I'm not sure if we actually ever tried to calculate that. 16:05:54 Yeah, because I feel like that could be almost like a proxy for the question, like the fractional like the the shared HTTP among cohabitate and choose dwelling bacteria versus all of the HGTV might be a proxy for the previous. 16:06:10 Oh, interesting. 16:06:13 Yeah. 16:06:15 I hadn't thought about that. 16:06:19 I could Yeah, I think there could be ways that we can estimate like the total HTTP versus the ones I mean it's a little bit flawed just because in LA, at least in this particular study, we kind of picked, you know, whatever set of genomes had already 16:06:33 been sequenced we didn't go in and specifically, like isolate organisms from one she's although we could do that now with this new data set that we're working with. 16:06:43 Yeah, that's it. That's an interesting idea. 16:06:46 Thank you. 16:06:51 So many great questions. Thank you all. 16:06:54 Okay. 16:06:56 So iron again. And in this case that actually goes back to that. The other question of what about. 16:07:04 Sorry, it goes back to the the idea of the city or for uptake is being important again. And so, in this case, we found this is just one example but we found multiple examples of specifically severe for uptake systems that were being transferred between 16:07:21 species. In this case, this is an integrative conjugated element that we found on several genomes inserted next to try a, it has all of the city or four uptick regions as well as a bunch of other cargo on it, but I'm not showing here, as well as the machinery 16:07:39 for excising itself replicating itself and conjugating itself into another bacterial genome into another bacterial cells. And so, again, what we think might be happening here is that within these microbial communities, in particular, and this has been 16:07:57 seen in other microbial communities as well, that if you possess the ability to take up, iron, containing Cydia force from your neighbors. You might. 16:08:08 This might provide some growth benefit because you don't have to produce the air force yourself. 16:08:16 Okay, last bit, which will be short, but I wanted to talk about this because this is something that we worked on that started at KTP when we were here in 2017, and it was the first time we've ever done and of course sequencing in my lab, and we're super 16:08:46 about it and so we ordered some cheese from cheese makers and threw them in an airport and generated a bunch of data, which one of my grad students has been working on for the past three and a half years. 16:08:48 So, the question that we've sort of decided to focus on with this data set. This long read data set is, what is the diversity of plasmids and faith in these cheese rights and this is something you know related to the horizontal gene transfer so plasmids 16:09:03 they're often things that are mediating horizontal gene transfer as our age, but also just think interesting in terms of increasing our understanding of the diversity of biology within the systems. 16:09:16 And so we perform Long, long read sequencing and this is all work that is being done by MD, PhD student my lab, counting, who has been using audio to map out the taxonomic diversity from the cheese's so this is a natural rain cheese from Vermont, called 16:09:35 Bailey his in blue that we've studied for a long time. 16:09:38 And you can see the, the different bacterial and fungal species here, but in. 16:09:46 In this particular panel here these are all the, all the lines are the viral contacts that are associated with this assembly, and then above that are the plasmid associated sequences. 16:09:59 And so you can already see by I there's quite a few different plasmids and viruses present in the single cheese rind. 16:10:10 And we did this across three different types of cheese a natural line of blue Rhino wash rain so those are present those three types that I showed you before, as well as to kefir samples as sort of an alternate form of fermented dairy. 16:10:26 And I think the take home message here is that most of the plasmids that we identify in these systems are actually not seen previously and any database. 16:10:35 So, a lot of potentially novel biological diversity, found in the system. 16:10:42 And so that would be represented sort of roughly in this space here. Were looking at blast values in alignment with plasma databases, so poor alignment and poor, many that are poor. 16:10:59 Evaluate over, or at large value, or height significant value over a short portion of the plasmid. 16:11:07 And the really nice thing about nanopore sequencing and the long pieces you can reconstruct circular plasma genomes really nicely so we have a lot of complete plasmids from this is from just the one cheese. 16:11:24 And then we can start looking at the genes that are present on these classes so what what are the plasmids bringing along with them within these communities. 16:11:32 And so in this particular plasmid we have genes for beta lactamase resistance we also have beta glucose side and carbohydrate utilization genes. 16:11:44 We see this really interesting. 16:11:48 Metal resistance and uptake is very common across these plasmids especially cadmium resistance which we don't fully understand why that is but we see a ton of cadmium resistance in this in these cheese associated plasmids. 16:12:08 different restriction modification systems, as well as a crisper cast system, and the crisper cast system has spaces that targets. Other plasmid replication proteins, so it targets the rep proteins from a different category of classmates so it's actually, 16:12:09 And then this one super interesting. It looks like a sort of a defense plasmid It has 16:12:25 it looks like that this plasma is protecting against both page and other plasmids and its host. 16:12:34 And so then we started to look at the page, and there's a ton of age in these cheeses. So, this is looking at the the similarity of these pages two pages in databases. 16:12:49 And you can see here, these are sort of looking across the different types of communities. 16:12:56 And there's similar. This is actually a genome wide similarity score that they have so 16:13:03 very few of them have a high similarity to known pages so a lot of novel pages present in these cheese and configure samples. 16:13:13 And you can see here just the sort of general numbers. So in the natural rain cheese we have about 40 different page presence, which is sort of mind blowing to me that that there's that many users. 16:13:27 It's predicted to be lyric pages, and these are like Isagenix pro viruses in the sample so pretty large diversity of features in the system. 16:13:37 And of course I'm getting my toddler knocking on the door now sorry. 16:13:42 And I just wanted to highlight this one which I think is super cool. We found this one happening of age appropriate age, that looks like an encode some cheese flavor jeans and addition to other interesting. 16:13:53 It has a text execution system, but it has blue Tammany semifinal game Alliance, which are known to be important for the development of cheese flavor. 16:14:03 And as far as I know people have never found them on page four so sort of interesting. 16:14:10 Okay, so just summarize, we set up a new system, where we can take apart communities and put them back together. We've developed approaches to probe the biology within the system. 16:14:26 And what that's taught us is that there's still a lot to be discovered.