15:45:22 Okay. My name is three.
15:45:30 I'm a postdoctoral fellow at MIT in the lab of might allow, which some of you may have met Slashdot off, but I, at least not in the subject matter that have been has been talking about here so I'm gonna maybe one upfront that is going to be a sharp left
15:45:39 turn, sort of weirdo land, relative to where everyone has been comfortable spot, but I'm happy to hear any feedback about particularly interesting things that you might be interested in that we can also try and do with some of the things I'm going to
15:45:52 tell you.
15:45:53 So what I'm fundamentally interested in is how bacteriophage evolve. And in particular, what are the constraints on the revolution.
15:46:01 And what I'm going to show you today is sort of a fairly straightforward experiment we did in the last year or so. That gave us very surprising results and maybe teaches us something about the bound on this evolutionary process.
15:46:15 So the first thing I want to start with because no one really has talked about painting and explicit sense here is just like a picture of what we think of a stage in the world today.
15:46:25 Most of you have probably seen stage of a schematic type like this, or a transmission electron micro graph like this, where essentially there's little Lycosa he will head as a tail and some other rq, that is the stock, and then some tail fibers that everyone
15:46:42 sort of has a general understanding that it uses to stick to its host and I'm injects his DNA replicate and then produce way more face particles.
15:46:51 But as we sort of started studying fades more seriously what we've come to appreciate is pager actually probably the most diverse, almost certainly the most abundant genomes on the planet.
15:47:04 Right. And the key point here I want to make sure that they're not the most abundant biomass clearly that that's definitely not the most abundant genomes, on the plan.
15:47:13 And I don't know, a lot of these fields so I'm happy to explore it if you notice something weird and funny that you want to look at, but I just want to highlight everyone that almost all of them have a common feature, which is they encapsulate a nucleic
15:47:25 acid genome DNA, RNA, it can be sort of either version in often a protein a capsule. But the protein issues capsule can also have limits. So, for example, in eukaryotic viruses, a lot of them are what are called envelope viruses which is they have a lipid
15:47:43 membrane. That's not totally new carrier, there are plenty of bacteria feeds that to the same thing. And in fact they bought off bacteria.
15:47:57 So macrophage sort of run the gamut, so to speak. Okay, so that's the first thing I want to find out. The second thing I want to point out that we are now slowly starting to appreciate going back to what I said earlier about the numbers is as we start
15:48:05 doing sort of an inventory of the world in terms of sampling, the genomes of bacteriophage around us. The families of fairly distributed across some geographic right like we don't enrich for particular types of activities in particular place in the early
15:48:19 days when we sampled them, we actually thought there was such a correlation that in the ocean, there's a particular type of back to Faith soil has a different type, God has a different type so on and so forth that turns out not to be so true.
15:48:30 It's just that we go very bad at isolating bacteriophage and arguably the Stila. And so as the sequence and identify more we expand the repertoire and we find that basically there's no strong correlation, like you can find any phase anywhere you want.
15:48:44 And I also want to highlight that the diversity extends also to genomes size. So if, for people who can't read it very clearly on the x axis here is just size of the genome, each dot represents a fully assembled genome that is present in NCB is reference
15:48:58 genome database, and the y axis represents the different families that are introduced in the previous slide, by their families based on the way the particle looks and now more modern way as we have of categorizing by Gene families, what you can see is
15:49:11 that they span a huge diversity and you know, even within the same family right so this is not similar to other organisms where a family on average has a human genome composition that then changes over long periods of time.
15:49:25 There's a lot more flexibility in fades, you know, then pretty much anything else we see
15:49:32 and concomitant with this sort of fluctuation in a genome size is actually the size of the particle itself. So this is the, from what I understand, still holds the record for the largest page that can be isolated and propagated on its own.
15:49:50 Right. So it's a faith that infects a particular vacuum called listening bacillus. It's about 500 kilobytes per as long, the genome has about 675 genes, 18 t RNAs right so it's basically approaching the point where you'd start thinking of this like as
15:50:07 many cell.
15:50:09 Right.
15:50:09 On the other hand, among the smallest there are smaller ones but among the smallest genomes of a face is actually something that a lot of people in work with equal I dread to have in the lab.
15:50:22 It has four jeans. Right. It has the replication gene, the gene that allows it allows us to sell, and then two genes that are organized the code.
15:50:25 But Ms to.
15:50:29 That's it.
15:50:31 And it is also a fade right so both of these go through replication cycles and obviously broadly similar ways but the details sort of change a lot. Yeah.
15:50:41 So that's a very good question. Both of these are little page. So neither of them can actually go Isagenix so I want to point out something here is, in general, the CFO very day which is the blue one on top here and also the largest part of this pie.
15:50:58 They are taught to contain most Isagenix age. Um, I'm not saying that we have a good genetic predictor of which phases Isagenix or not. That's not true.
15:51:04 We actually don't have that predictor. But observational, we can find that most Isagenix page cluster in here through various methods that we found.
15:51:19 So, so what's interesting is I was going to say is that this is the one that we've isolated. We are now starting to find fades, that are off order like one megabit.
15:51:28 And there are eukaryotic viruses that are essentially functional pseudo cells where some of them are about 1.1 mega base but i think is the largest ones but the Mimivirus i think is the current record holder.
15:51:39 And the other thing I would say that slightly different is unlike bacterial genomes, the genome type is also dramatically different right because there's RNA double stranded on a single turn on a positive negative sense all sorts of craziness that they
15:51:52 do. So that's the one other caveat, I will.
15:51:56 Okay. Um, so, so basically just this was just an introduction to give everyone a flavor of what we think of age today. Yeah.
15:52:09 So that's a very good question. In fact, we'll go into what we're going to talk about now so the question just for people in zoom is what host resources does do fades rely on broadly, I'd say the most consistent thing that may be true across all fades
15:52:23 is almost none of them encoder full ribosome.
15:52:27 So, the ribosome in particular I think almost all of them rely on part of the horse machinery. Many of them can encode RNA polymerase of their own, and DNA polymerase of their own, or more generally the genome replication machinery because depending on
15:52:49 what your genome is it sort of is a little more sketchy.
15:52:49 I think I said this at the beginning but the way we classify them into families, is that is by how they look.
15:52:56 Yeah. Is that a good like or yeah so I think back to the way like linear started biology right that's basically how phase classification worked until I'd say about the 70s to 80s, so most of modern molecular biology that was invented in phase was invented
15:53:14 when we had no idea right because in fact pages how we figured out DNA is the genetic material.
15:53:20 So, a lot of this stuff is since then. What I will say is the problem with even using genomic data to try and sort of do Plato grams of fiber grams is unlike almost any other organism.
15:53:31 There is no globally University concert page protein or vital protein. So there is no equivalent of 16 hours, so whether 16 S is useful in fact we are not sort of is sort of a little bit of a debate but it is undoubtedly present that there is a university
15:53:45 consulting. So in fades you do gene clustering, but you ask what content does your genome have and you try and rely on that as the closest way to, to sort of compare phylogenetic.
15:53:59 Okay.
15:54:01 This one
15:54:04 this one. Yep, who follow up to Michael's question which you answered very, you know, very thoroughly but I wonder, even if there is no universal property and maybe you can do a little bit of a hopscotch like the one family and other one overlap over
15:54:20 a part of the genome it's not the same part as overlap between family B and C and one can try to do something, this way. Okay, so I'm actually going to introduce something in the slide that maybe get a little bit of that but I'll also address what he
15:54:31 just asked okay so actually following very nicely to suggest question is one of the other things that's a little different and fed genomes, is they undergo vast changes in their genetic material, relatively rapidly.
15:54:45 So for example, so this is the work from Graham had first lab in Pittsburgh. And what they did is they've been isolating phase of Mycobacterium for a while now.
15:54:53 So I think that current collection stands at about 1000, maybe a little less than 1000 Mycobacterium page. And when the genome sequence all of them what they actually find these colored bars so each line here represents a genome of one isolate the colored
15:55:05 bars represent regions of homology right shared between two fish. And what you can see is, on average, yes, there are these blocks that looked like there share, but you add and remove jeans, kind of as you want.
15:55:18 And sometimes you jump entirely to new part of sequence space where it seems like it's an independent horizon.
15:55:24 Right. But at the same time, particularly shared genetic genes might still be as concerned as they are in a truly concerned, page. So it's not like it's a continuous scan across the genome.
15:55:37 So rather to maybe answer your question, more specifically, so the lab of Eugene Kuhnen who's at the NIH, they do a lot of this work, and what they figured out is that if you instead of doing it by actual genome order or synchrony.
15:55:51 If you construct a genome map as a bag of gene, families, and then you ask in a network topology, how you get from node to node or which nodes are most closely associated because they share multiple edges.
15:56:05 you can then sort of reduce the graph, what looks like a clustering of page families. It works, sort of, but the problem is computationally from what I understand scaling problem is very poor.
15:56:16 So the more fades you get the worse and worse it starts performing.
15:56:23 Yes.
15:56:28 He recognizes the micro bacterial genes or are many most of them unique to the phases because there's only sharing genes in common if they're all effective on the same host will be very surprised.
15:56:38 Yeah, so that that's a very good question. So, maybe the best way to put it, so I don't have something explicitly about this but maybe to address your question.
15:56:47 So if you think about equal life age, right. So, all of whom can in fact equal life, there are a few families that are dramatically different from each other so the classic example is t 74.
15:56:56 They have almost nothing in common, right like I think
15:57:00 just the T. That's right.
15:57:02 It just actually wants a funny story about the T fed 31 through seven isolated at the same time. It just so happens that even numbers are related to each other and the odd numbers are not 32 to 46 are accidentally, they're all the same family of age,
15:57:17 and in fact the yeah that's why they call the event page.
15:57:21 But to answer your question more directly.
15:57:24 The problem is we actually don't even know how to think of faith proteins as belonging to a host or not. And part of the reason for that is I will get into comes a little bit to your follow up part of the question which is most pages don't have an annotation
15:57:37 we actually don't know what these genes do, if at all we even know for sure that they are gene right so often it's just automatic annotation that predicts there should be a gene in this open reading fair, definitely say ages have genes in which are recognized
15:57:49 them please no bacterial absolutely so metabolic G. So for example, a number of faith to carry tyrannize right the tyranny is our from various host that they've grabbed at various times, and even gain and lose them.
15:58:03 So related faith can change the DNA composition they carry with them over time. The other example of things that you grab from your host is actually analogous to me I'll talk about today, which is you can you take machinery that might help you disrupt
15:58:16 post defense architecture. So suppose your host has a specific way to disrupt your replication you steal something that the host users to stop it from killing the host itself.
15:58:28 And because all hosts carrier self protection system. So you steal that so for example the metal as of restrictions systems, sometimes gets stolen. So things like that.
15:58:37 Okay. So just as a quick follow up I also want to save, maybe as a follow up to my answer to Sergei, this is a similar diagram, but not using Symphony, where the x axis and the y axis represent the pages that are in this collection, as of 2015 I think
15:58:52 as I said there was about 700 of them here.
15:58:55 And the color represent the percentage of shared genes in the genomes. And you can see that that clearly clusters. Right. So there are some genomes are clearly more like each other than others, but on average this quite a bit of variance.
15:59:10 And basically, between clusters, there's very few if any shared.
15:59:15 And remember these are just micro bacterial page. So there's no reason to expect and actually Daniels question is a good one. There's no reason to expect sino bacteriophage are some hope more close to any one of these clusters than anything else.
15:59:27 If anything they do their own special thing. And, as David pointed out one of the things that's actually quite fascinating is in Sinopec two in particular, there are some states that have stolen photo system architecture, in order to supplement the metabolic
15:59:38 energy needed for a stage of production, so basically they supercharged the bacteria because they know that there's going to die because they're killing bacteria.
15:59:58 Okay, so the last thing I want to introduce is on the other side of the equation, why potentially have faith diversify. So it turns out, just as much as spades have diversified because of a billions of years ago evolution back the outcome of really smart
16:00:01 overcome it. So, yeah, so that that's sort of like what what they do.
16:00:14 ways to kill faith, because they don't want to be susceptible to parasites. So the one that is the historically the most recognized as the restriction system the restriction modification system.
16:00:24 More recently, a lot of you have probably already heard of CRISPR which. Now a lot of people think of as just a DNA editing system but actually the reason CRISPR seems to have been physiologically relevant is a famous defense mechanism, right, because
16:00:37 it is the equivalent of our adaptive immunity it steals a portion of the face DNA and therefore has a targeted nucleus against fades it in fact, but those are just two of the many types.
16:00:47 So some step to ISIS for example actually make DNA into talent that specifically into Caillat into fades DNA and it's actually not known how they protect their own DNA.
16:00:57 Right, so that's something that's interesting. There's a particular class of systems, which we'll talk about a lot more today called the abortive infection systems, which is basically the equivalent of a switch sizes, a suicide system, where the infected
16:01:09 host commit suicide to prevent a burst of page to protect its neighbors.
16:01:15 And then there are systems that we are only now starting to even address as fate system so one of them which is actually quite interesting, is actually it is thought so, in eukaryotes, one of the things that happens in our vertebrate immune system is
16:01:27 there's a response to viral infection, which is to produce cyclic die nucleotides. And this is a very famous immune party called the sting, or suggesting pathway.
16:01:37 And for a long time it was thought to be an innovation of vertebrates, it turns out, prokaryotes invented it way back in the day, we just hadn't found it till now.
16:01:45 And in fact they've even expanded on the repertoire, the only ug MP MP.
16:01:50 It turns out they use any number of cyclic di nucleotides. And what's interesting is cyclic give the nucleotides are in fact signaling molecules that battery use for growth rate control.
16:01:59 So a lot of you who have studied stress for example might recognize that in other contexts. And finally, there are now ways in which people labs in particular the Sierra Club is using computational tools to try and enrich for genes that should be fades
16:02:13 defense systems, but we have not yet validated them, and a number of these contain gene groups that we don't recognize what they probably do so. Suffice to say a lot of bands defense, we don't understand how defense is enforced.
16:02:28 How big is the set of genes in a typical bacterial genome that are involved in page defense is that can you give me an a number.
16:02:36 I have, so I don't think the calculation in that way has been done in any reasonable sense so what I'll say. So one of the things that soda club now has been doing for a while, is they argue similar to antibiotic resistance Island, they should be islands
16:02:50 of fades resistance right where you want to pass on to your friends and progeny these islands that are enriched for fades defense, presumably have locally enriched face.
16:03:00 And so there are arguments that there are these sort of recombination islands that contain face defense but it's not totally clear but I can give you maybe the inverse correlation.
16:03:09 So, on average, I think most bacterial genomes contain between two to 10 restriction system across types. They contain between 10 and 100 PA system. So about of infection systems.
16:03:20 Many of them have a crisper system, sometimes multiple CRISPR systems because CRISPR doesn't target just DNA, it can target RNA as well. And so yeah it's like sort of a ever growing list, as we find out more and more order hundreds of genes.
16:03:36 Yeah,
16:03:37 Oh easily. Yeah, okay.
16:03:40 look something like this.
16:03:42 If you find something new in the environment that conflicts like this How do you know what to face, as opposed to a man environment or virus or some other type of organism.
16:03:51 Yeah, so don't know the genes, you don't know all the proteins.
16:03:54 Yeah so true definition of.
16:03:56 Yeah, so I think the way it has been done thus far, so I can maybe the walking definition, is the better way to say it. I don't think there is a clear definition it's an operational definition.
16:04:07 So the most simple way is if you can propagate it on a bacterial host that you will have available.
16:04:12 I because then there's a clear sort of life, knowledge, back.
16:04:16 From what I understand, one of the ways people have been evaluating number of virus like particle is agnostic of what the impact. So for example in the Ocean Sampling often they pool or viruses viruses of bacteria plankton.
16:04:31 I'll gay, all together.
16:04:32 The other thing that I think sometimes they do is if you sequence, there are some genes that are not as lost across state so for example the code protein is actually a good example of this.
16:04:44 The, the thing that forms the caption that are not that many mentions of them there's only like a handful of them. And those if you can find something that's close to a known back to your page you might argue that it's enriched for a battery, but that
16:04:56 that's a totally reasonable question I don't think there's a clear answer.
16:05:03 Someone. Are there any Sage inject not only bacteria but other organisms.
16:05:10 So there's this crazy thing so maybe I can say it this way.
16:05:14 We have never observed that. However, if you look at the genome sequences of viruses. It turns out so for example the herpes virus or protein is homologous to a lot of bacteriophage.
16:05:26 We have no idea how that happened right because you can argue that evolved from the same parent. But there seems to be a lot more ad mixing subsequent to the original divergent, which suggests that they share.
16:05:41 Somehow they inherit DNA from each other, my hypothesis would be that three DNA is probably taken up by the host of one of them, and then used by the other one, rather than they both in fact, in fact, like the same post, but that's a guest, because one
16:05:52 of the things I will say maybe to your point of macrophage is unique in the sense that it can actually traverse the multiple membranes in the cell wall of bacteria, but it's a lot of other viruses don't have that right so plants are again different because
16:06:03 they have to traverse the cellulose barrier. So in that sense, there are distinction, what they can and can't do, physically.
16:06:11 Okay.
16:06:14 So, I'm just going to jump into sort of what I work on. So these are sort of two very broadly phrase questions that I'm personally quite fascinated by the first is, given that we see all this diversity of fades in the wild.
16:06:26 My interest is presumably they must do cool things during the revolution. It's just that we haven't had the ability to see it till now. So in the lab.
16:06:33 If you evolve page against particular need to try and overcome defense systems which is presumably one of the strong selective pressures against page.
16:06:42 What are the molecular features that fade sort of employed, like how do they actually go about this process. And second, can you actually taste the dynamics as they start infecting new horse was that they didn't have the ability to in fact before and
16:06:55 gain mutations in order to now invade a new environment.
16:07:00 That's sort of generally my question.
16:07:01 So, when I joined the lab last year, I sort of like started working together with a graduate student Chantal was working on one of these different systems that protect against people.
16:07:15 And what I'm going to show you today is the evolution experiment we designed using her understanding of that system. And so the simple question was, aside from all this complicated business of multiple different systems and all that, can we even observe
16:07:27 a phase evolved to overcome a specific selective, but just one selective pressure being the genetic difference between our assistant and understands to pose.
16:07:37 Yeah.
16:07:41 I'm just I'm just wondering, we've, we've been working with Paige for quite a long time. Do we have any idea of how the same phase that has been studied for a while has you know changed over time, we've domesticated would, would that tell us a bit about
16:07:55 this, this question.
16:07:57 Yeah. So, that's a good question. So, part of the problem is I don't know how reliably we've held stocks have paid from historical record for us to sequence now and be confident that we can actually trace domestication right because you'd need that right
16:08:13 we have all present aliases but presumably they've all been passage to some greater or lesser extent, and it is very hard to go back to the wild and find it for right like we can find it for like faith but then there is no guarantee that that was the
16:08:29 ancestor of what we now see in the lab.
16:08:32 I think is maybe the background so, and hopefully as I go through this maybe we will get back to why that may not be as big a problem as might see.
16:08:40 Okay, so just a little bit of background of what the actual specific selective pressure to you. So I told you that there are a number of different systems that the lab lab in particular studies, a bunch of them.
16:08:51 So Sean was interested in working on trying to understand this particular defense system so it's called the type three toxin antitoxin system and in particular, it's called talks I am because it's in an opera, and the way it functions is docs and is the
16:09:05 the protein that is encoded, and it is a toxic protein because it's an ordinary right it's a relatively prolific ordinance which you can imagine is very toxic to pretty much anything that needs to make army.
16:09:16 But it also produces this or any antitoxin talk side, which you can think of is like a fairly straightforward substrate inhibitor right it's a substrate mimic that inhibits the function of the toxic.
16:09:26 And you can imagine how the system is stable, because the RNA is toxicity is not exposed to the cell, until the antitoxin is removed.
16:09:34 But what she observed is that upon infection with before.
16:09:40 It seems like the presence of this genetic system can arrest the replication of default in cells that happen.
16:09:47 And just to give you a little bit of a lack of sort of experimental detail and how we can measure this or and of course the question she wanted to understand is how does this work.
16:09:55 And so just to give you a little bit of experiment in detail because this sort of is relevant to the rest of my talk, the way we test this is in the background here in sort of light gray is a lawn of bacteria, and what you play it on it are serial dilutions
16:10:07 of the feedstock that you have from left to right. So more concentrated less concentrated page. And here you can see that in a, in a wild type lawn, basically the fans produce these blocks quite robustly.
16:10:21 On the other hand of the lawn of cells is expressing this defense system, the same that seed and dilution of fates produced no blocks at all right so it's almost a perfect protection against the page that is trying to replicate on the small.
16:10:35 Additionally, she showed that you can measure the amount of faith that come out of a single cycle of infection. So that's what the blue line traces here so and you don't have the defense system.
16:10:45 After about 35 minutes, you get about 100 phase particles from a single round of infection.
16:10:50 When you have the defense system, essentially get know burst at all.
16:10:56 I'm going to simplify all the other thing I should point out maybe to answer the question here is the for the one we work with in the lab is about 170 KB Gino has about 208 nine proteins at our names and about two small army.
16:11:10 So that's the stable version of this genome that we work in the law.
16:11:15 So to cut her story short, and this is the reference to her work if anyone's interested. This is what she found happens as a model for how to for protection is in engaged.
16:11:25 So when T for in fact a host. This system is present, free floating in the bacterial cell. The inhibited toxin detox the antitoxin is degraded.
16:11:36 And then the toxin is released, and it chews up all RNA, including the army that the feds producers in order to produce the boss, necessarily, this is a suicide system.
16:11:48 Right. And I should point out that the Suicide is actually not because of the toxin killing the RNA of the host, it's because the for one of the first things that does when it infects the host is it chops up the host genome.
16:12:01 We recently lost all power over your autonomy at that point as a bacteria that before has taken over.
16:12:06 He recently lost all power over your autonomy at that point as a bacteria that before is taken over. So after that point whatever happens, you're only doing to protect your neighbors there's no way back for you is this mechanism the Norton planktonic.
16:12:14 Sorry. Is this not happen in planktonic bacteria in Atlanta, as in growing bacteria growing back here.
16:12:22 Yeah, this is exactly what happens on which mechanism you ask, which the default or the antitoxin is this defense mechanism.
16:12:30 Right.
16:12:38 See, committing suicide doesn't pay well doesn't obviously pay. If you're all getting mixed around and things. Yeah, so, protecting your siblings, right.
16:12:41 So, so I will tend to talk, I don't know why the TS systems of all. To me the way they are, but it is definitely true that you can find them across horse, and many of them now we know they protect against change.
16:12:53 So, presumably, a large part of a selective pressure maybe in regions where you can at least locally create a small enough colony that your neighbors are you're confident, is your is your siblings.
16:13:06 That's the guess there's no
16:13:12 secretly fH defense then you don't necessarily need to be so related anymore. Right.
16:13:19 Yeah.
16:13:20 Oh, you still eat. Okay, basically you can cheat right so suppose if I lose this system, and I only rely on my friends to protect me against page.
16:13:29 Why would this be stable as a static accept like him clash like eventually you'll crash so maybe the argument is because there are so many more fades relative to bacteria in general.
16:13:39 Anyone who loses and attempts to cheat is knocked out because they're killed a new phase of produced, I guess, I don't know.
16:13:49 I'll be my best guess.
16:13:50 Yeah.
16:13:51 So I yeah I wouldn't say like the evolutionary sort of pressure to hold such a system like this, I don't actually know.
16:14:05 Yeah, so that's the other question so I sort of skipped over that, the way the system is induced is actually quite clever. So what it does is the way the, the antitoxin is degraded.
16:14:15 Is that the antitoxin is extremely unstable. So you need fresh transcription to always maintain the complex. So as soon as the whole, the fades degrades the whole genome, there's no more transcription.
16:14:25 So basically it's a surveillance mechanism to check is your genome intact, like integrity maintain.
16:14:30 And that's the argument that how this system is meant.
16:14:34 But that's not a universal strategy for activation that's just this particular example.
16:14:39 Okay, so, So that's what we knew when I got there. So, the question I had was can we now take the food right and basically run a simple experiment and evolve it to overcome the state's defense mechanism.
16:14:53 And the protocol is actually a fairly simplistic one it was actually developed in various forms, but in the beginning, I'd say in the 1920s back when fates therapy, especially during war was particularly relevant.
16:15:04 And it was a way to generate fades cocktails against pathogenic strains of bacteria wounds. So it's called the opponent's protocol after the person, I think at least was one of the people who was doing a lot in Georgia, I believe, are the country Georgia.
16:15:21 And it has more recently been sort of like attempted in various ways to sort of reinvigorate face therapy, but I thought it's a useful protocol for what I'm going to do so.
16:15:29 The way the protocol works is fairly straightforward you take the ancestral population of age. You inoculate that population into both hosts that are sensitive and resistant, using the particular defense selective pressure that you want.
16:15:43 After a certain amount of time, the sensitive strains get cleared and you produce a new phase population and the resistant strain in the beginning, doesn't clear at all.
16:15:52 So there's no faith produced. But what you do is you pull both nonetheless, clear out all the hosts that you only take the license.
16:16:01 And you presumably generated genomic variants by passing through the sensitive population. And so now you restart this entire experiment. And you basically a sampling for the variants you generate in the sense to host to ask any of them now invade their
16:16:14 assistant was fairly straightforward experiment system.
16:16:18 Just to give everyone a little bit of a sense of the numbers, the way I set up the experiment and you can definitely change these parameters, you go through, essentially, the phase go through from 100 to about 10 to the six stage number amplification.
16:16:33 But if you think about it that's only two cycles of infection, because each burst is 100. Right. So, in exponential growth you're, you're essentially in two cycles you generate your new population entirely.
16:16:43 But to infection cycles, around.
16:16:46 What do you need to mix them.
16:16:48 So the reason you need to mix them I suppose a rare animal can actually in fact resistant loan, in order for it to increase in sort of like increasing frequency in the population, you need selection on some way right because otherwise it just stays at
16:17:02 the low level.
16:17:04 Because I'm not testing into two o'clock at each round, you're just having the population as a whole, yeah yeah but it seems like you could just take the ones that are you could continually propagate the ones in the, in the talks i n minus condition and
16:17:16 then just on the side take the light side and dip it into the other one in a, in a fresh culture each time. Absolutely, absolutely a spurious I absolutely you can do that is just at the nice thing about this is you don't need to do it separately when
16:17:27 you're doing it together you have a nice visual output of is the population as a whole evolving. And you can also I say it in neutral tones Yes, you can do it without mixing.
16:17:35 Okay.
16:17:41 I understood that's all I was FDA, you can totally do so. I guess that maybe the way I will say it is
16:17:45 without mixing, and if you just test the population every round on the plus talk science cells, and you see clearing at one point. You don't know if there was one clone that like to all that clear.
16:17:55 Or if there are multiple lineages in the population that are leading to it and then you have to isolate them separately. What this allows you to do is it allows you to increase the frequency appreciably, and then you can ask Was there an initial strategy
16:18:05 that was bad that then had to get better and things like that. But yeah, absolutely. You can come up with various ways of doing.
16:18:13 And just to remind everyone this is the phenotype were tracking parallel on long. So similar to the, the, the suggestion I mentioned in the previous the finger in the previous slide, the yellowish color in the background is the battle on the clearing
16:18:28 and dark color you see is the face blocks that are growing, and you can spot a serial dilution and you see that in the beginning, the four spots on on the empty vector, long but does not on the, on the defense system containing.
16:18:53 Yeah, so this actually goes back to what I said about that phase when it in fact, it kills that host. So if there are way more fade than any host Western she will clear all the hosts, and there's no more host to recover right so if the multiplicity of
16:18:55 fades relative to initial hosts is higher than one on the host immediately die.
16:19:07 And there's no more recovery for you that that's why you see that small question that the toxin is under an irreducible promoter or what's its native promoter that was found in the sequel isolate so we just basically pulled up like Final base pairs upstream
16:19:23 it, and it's on this plasma TV or 322 which is sort of this classic low copy studying natural isolated DNA. Right, I'm going to cut off questions for the next five minutes From now on, it sounds like more just quick bullets of what we found.
16:19:32 Okay, so this is what I did, I evolved six individual replicates at the same time, replicate populations. The top replicate is a control evolution which is never exposed to the toxin the toxin containing bacterium, it's only exports to sense to back him
16:19:57 And what was really cool. Very quickly, they started evolving the ability to overcome, not science. Right. And by the time we finished our experiment which is about 25 transfers, all our populations had fixed what looked like a clone, that could overcome
16:20:11 toxin.
16:20:12 So this was great because I was very curious what he found. So everything I'm going to show you from now on, is from plating these populations out to single clock's ticking o'clock and testing everything that, so I know have a colonial strain, if you
16:20:27 will have before that can overcome this differences. Okay, so, because page genomes are tiny, you can sequence them super efficiently. So we sequence, the, the vault loan to see what happened, and not really much happened in terms of mutations, until
16:20:41 you take a look at the coverage map.
16:20:44 So this is just one region in the genome.
16:20:56 And this is about like between 10 and 20 fold access coverage that piles up right there on top of what looks like an opera. And from this point we're pretty confident is an opera right containing these two GPMD and 61 point.
16:21:02 I'll show us where something I need to say as well comes back into view, 34 is arguably a place we did a lot of molecular biology of the 300 ish genes.
16:21:10 62 genes are absolutely essential to make a new virus.
16:21:15 Right. Any one of those genes last year gone, there's no way to make any virus.
16:21:18 So the majority of genes are completely neutral in page replicating and I'll be in equalize that sensitive at 37 degrees.
16:21:26 Many of them we have no idea what they do.
16:21:29 So maybe going back to your question, even the first genome is almost entirely unknown what its capabilities are right.
16:21:37 So, what we think happened is that there was an amplification segment of this opera on in the Fed genome. Now the reason for this is not just the sequencing, we did various other tests like PCR if you do you see lathering which sort of this classic way
16:21:53 of seeing these ladder show up, and then we did the actual experimental test but what we did was we reasoned. Well, the simplest hypothesis is that they amplify the region because they want more copies of the gene.
16:22:05 So if we put these genes on an inducement plasmid and over express them. You might be able to rescue, even the ancestral default that people wouldn't need to do this amplification itself.
16:22:16 And exactly as you might have. When you don't induce, you actually see very little, except in the case where the gene we think is leaky, and therefore rescues partly, but when you do induce overexpression ancestral T for basically recovers this phenotype
16:22:30 we thought we want, which is our the database, which is the ability to overcome the resistance have provided by this.
16:22:38 Okay, so what's particularly interesting and maybe this is more interesting as a biochemist is.
16:22:49 If you believe the stock code on in this plasma protection goes away.
16:22:49 If you record them RNA.
16:22:52 The protection is still present, so which means it is the peptide product not the RNA, that provides protection. But what I told you the native antitoxin is an RNA.
16:23:00 So between the Fed has basically be think invented its own antitoxin.
16:23:06 In order to overcome the selective pressure.
16:23:08 Right. And it has been living in de force genome, and we just didn't know about it.
16:23:14 Right.
16:23:15 So the other thing I will say, like many of the unknown genes of e4. It's a pretty short gene so it's only 85 minutes it's, it's a very positive. So the pi is actually pretty high, and from what we can tell it's fairly unstructured it doesn't really look
16:23:30 like anything. Right. The reason I point out the PI in particular, is you might imagine, the antitoxin mimics and RNA structure, so it should be negative.
16:23:39 It's actually the opposite. It's a very positive protein. So Shonda who was my collaborator she is very interested in the mechanism of this sort of inhibitory system because it seems to be an independent way of innovating, the toxin compared to what the
16:23:52 hostess right so that's the first sort of like results that I thought I'd share with the second thing which I think is of particular interest to some of the questions that were raised earlier, is you can ask well, is this something that just shows up
16:24:03 and before it turns out, I totally there are many fans that RT for like, so these are four of them that are very easy to find in just repositories of page, 3042 36 and Rb 69.
16:24:17 They have the locusts that looks slightly different, and in particular, obviously, All of them have the gene.
16:24:23 And you notice that Rb 69 natively is better at overcoming the defense system, even without the segmental amplification.
16:24:30 And the reason we think for that and now we've shown is that because RB six nine seems our better adopted dream.
16:24:37 As an anti dogs.
16:24:39 Alright so you can imagine similar to what I told you where the states are evolving against different horse. If you have whole alarms of the toxin system in different Nikolai each page is presumably trying to find the best adaptation to its own host,
16:24:51 rather than some global minimum that there is constant coalition on both sides of the song.
16:24:58 That's the second thing that I want to tell you. And the last thing which I think is particularly interesting to me is, what does this segmental amplification do to the genome of a page.
16:25:09 So something that was Sava interesting side tangent of my reading is the way of fades God and faith genomes of packaged into capsules is actually quite fascinating.
16:25:20 So it is actually the strongest at PS motor known.
16:25:24 So the page, in particular the T for motor that drives genome packing into the capsule is the best at PSV know it drives back into what's called the headful mechanism which basically is exactly what it sounds like it packs it until it's full.
16:25:39 The motor stalls, you cut the genome, and that's your one particle, and then you move on.
16:25:44 So you could imagine when you amplify a part of your genome, that's really bad because now you are confident your head with everything you needed.
16:25:51 And that's exactly what we saw.
16:25:53 So, when you see this amplification in the face arise.
16:25:57 Subsequent in the evolution, the actual stable species that shows up our species with genomic deletion. But they have lost part of the default genome, because they are now maintaining this larger fragment because of presumably the selection we've even
16:26:11 post.
16:26:13 And what's interesting is we are pretty confident the segment or deletions are compensatory not beneficial intrinsically, because in the control population there was no amplification or deletion.
16:26:24 So just passing does nothing but the different replicant populations delete entirely different parts of their genome.
16:26:32 Right. So this at least strongly argues that the particular part you delete doesn't really matter it more matters that you need to delete something, and particularly as maybe I think all of you can inform what you did it can be essential.
16:26:46 So we think what this is showing is a heat map of non essential parts of the page you know of the T for genome.
16:26:53 Right.
16:26:53 And I can actually show that to you in a very nice way where one of the things I was working on as I told you is trying to understand how phase you in fact evolved in fact different hosts, who had many different hosts in the lab with me.
16:27:04 So if you take the stage isolates that have these various genomic deletion. So, this is the exact same as as previously so left to right, a serial dilutions, the different roles represent these different clones with different genomic deletions, the top
16:27:16 row is the ancestor.
16:27:18 Right.
16:27:20 When you have the toxin so the defense system, exactly as we predict all the clones that evolved can overcome that to the ancestor and the control evolution can't.
16:27:30 If you played them on lambda license you can have equal life just take lambda license denies equal ly.
16:27:37 One of them can no longer in fact equally with the lambda licensing and what's beautiful is we actually know the reason.
16:27:43 So turns out in this particular genome, the gene that is deleted in this region is called are two. It's actually one of the ways we figured out the structure of genetic material wave of jeans back in the day.
16:27:55 Right.
16:27:55 It is actually a known defense system against a profane element that protects its host against other fields. So, something to keep in mind is profit actually fight for the other team, once you become a licensed right because until you are a free living
16:28:10 Phaedra Phaedra against other other horse, but once you become part of the host you want to protect your host against other page.
16:28:17 And we know the system exists and exactly as you might have predicted last in that gene loses the ability to now protect against the vandalized.
16:28:26 But what was fascinating to us is differently deletions show sensitivity in different hosts that we have no idea why they're like that. Right. We think this actually is also a good way for us to unearth defense systems that were not known, because the
16:28:42 feds had the ability to overcome them. So you could never identify them, and vice versa, we can hopefully identify what the feds used in the first place, to overcome it.
16:28:52 So that's another part of sort of my project not necessarily to do with the genome evolution but specific arms race evolution in terms of the.
16:29:01 So before.
16:29:02 So, I should point out, equal or here is I have a collection of 50 odd environmental Iceland of Nikolai.
16:29:09 So these are all equally, but in general to your question.
16:29:13 We have actually not been very good at isolating broad host range page for a number of reasons which we can talk about later if you want, almost always the range is limited to neighboring species, it's very hard to really cross anything major right so
16:29:26 for example, I would say I'm pretty confident you can't find a phase in fact both gram positive and Gram negative, because the actual details of how you get into the host probably are just too complicated for that to co exist, but there are probably some
16:29:39 things that you can find friends for.
16:29:54 So that's really all I have, so maybe I'll just summarize and maybe add a little sentence for something that might be relevant to people's here. So as I pointed out today.
16:30:02 Page diversity suggests that there's actually a fairly dramatic history evolutionarily both as an arms race between bacterial host genomes and the genomes themselves.
16:30:12 And for everyone sure something to keep in mind is that might also be one of the things that affects how communities are stable right because these communities, other than the resources they themselves us also trying to fight off these invaders of page
16:30:24 age, that are specific to some hosts and not others and trading sort of ammunition, so to speak. In this ongoing war,
16:30:33 specific to what I showed you today we came up with an evolution platform that's quite successful at getting people to vault overcome a different strategy, and in particular the way it evolved the defense strategy, gave us a hint that actually it has
16:30:47 seen this different strategy before and it already knew how to overcome it, it just didn't overcome it in the lab setting that we had imposed.
16:30:54 And so that's again something to keep in mind is page, evolved relatively rapidly to in fact new horse potentially, and they can do this in many ways.
16:31:09 Either mutating existing gene which other people's evolutions the lab see or potentially even gaining genetic material, both either by copying existing ones, or by gaining from their environment, either from the host or from other states even.
16:31:18 So in a specific sort of conclusion is that the segmental amplification of the genome Franklin fragment in before also leads to compensatory deletions.
16:31:31 And this is something maybe that's a more broad statement which is page are constrained right so that genomes are not like in an envelope that's a little free in terms of size that they can move.
16:31:39 They're constrained in both directions because often that package to a full hand.
16:31:41 So deletions are also costly and amplification that also cost.
16:31:48 The way you evolve your genome.
16:31:49 Right. But having said that their ability to replicate that DNA seems to be fairly error prone such that they can actually do all of these strategies and sample.
16:31:58 A lot of genotypes relatively rapidly, which is something that's kind of cool to see happen in front of.
16:32:06 And the last part I wanted to point out is that the genome deletions are neutral in the selection regime that we imposed, but very clearly were costly in hosts that we had not expose them to.
16:32:16 And what this actually suggests to me is that phase in the natural world are evolving against many hosts parallel right like it's not that you have one favorite constantly evolving against one bacterium, and it's this sort of paired arms race forever.
16:32:29 That's not what's happening, but what's probably happening is everyone is sharing everyone and everyone's trying to find everyone at the same time.
16:32:36 That's really all I have. So I just wanted to say thanks to the lab are pretty big lab, most of all, studying biology and biochemistry but.
16:32:56 have you tried to put the two different types of phases one was deletion. One was antitoxin. So then you would have the best of both worlds right that one can recover it's a missing DNA from the other one or the other one can yeah so that's a very good
16:33:09 question and experiment so that's actually something I'm trying so there's actually a couple of variants of that question right so you can ask, even before the genome deletion spec.
16:33:19 Suppose I introduced a horse techniques one of the divisions costly so I introduce one of these holes into the selection system.
16:33:22 So now I enforce replication into that host as well. What do you see, do you see it actually break down into subpopulation where each subpopulation fights its own host, or do you actually see what you're asking do you generate a generalist.
16:33:35 And is that a mixing of that various strategies. Absolutely. And also, can you gain something you've already lost. So I'd say, in many cases that seems to be true because like in other cases people have shown that before recombined with other people genomes
16:33:55 that exceeds. So it is probably very likely that it can recombine back.
16:33:58 So, back to the question of host strange. And you may have explained that when you were telling us about the various defense systems, and I just missed them forgot.
16:34:09 So, even the within, let's say, environmental isolate so we call it from some great lakes or something. So do you find the great deal of diversity in defense system so are these things that are just routinely horizontally transfer transfer.
16:34:27 So how would that transport is a little bit of an open question. As I said, the sonic lab things, these lie on what are called islands of recombination so presumably there's some mode of transfer that's not inheritance, a vertical inheritance.
16:34:43 Also, defense system seems to jump in and out of this island, so it's not a constant Island everyone shares.
16:34:50 So, exactly what I was saying in the last slide,
16:34:55 is equal rights that we have.
16:34:57 There's a huge number of them, and all of them are equally. So if you look at the code genome right like the four ish kb.
16:35:05 Pretty much conserve the remaining is like this random assortment of presumably defense systems that we don't know about yet that set this level of variation.
16:35:13 So in fact, something that might be interesting to keep in mind is in the lab when you evolve horse to evade about age, they generally do it by surface modification right they escaped by changing the receptor.
16:35:24 That's doesn't seem to be what like the wild generally seems to prefer it does this crazy defense architecture swapping, so it's more, I think I came to think of this as immune profiles of people, right, we each have a unique immune profile based on our
16:35:37 history of infection. And that's sort of the way these, these bacteria are haven't Ecole I discovered that the benefits of a drug cocktail.
16:35:50 So I think this is so easy to beat Why, why doesn't take, you know, three, three genes to beat them. Yeah. So there's a couple of things here so there's maybe two different arguments people make.
16:36:01 So the first argument is there is some intensive cost to maintaining all of these systems, and there are fade that beat each one of them so by chance if you lose one because you don't see the face right now.
16:36:13 Your future is in question because the feds move in and out, in some sort of stochastic manner. That's one, two and this might be a potentially questionably interesting or not is back you're also gain benefit by gaining new genetic material like classmates.
16:36:28 These work against plasma, so defense systems can't discriminate what the source DNA is often, so things, for example crisper is a good example of crisper is very good at blocking plasmid gain by course so naturally competent bacteria now fighting this
16:36:50 But that's not how these different systems are set up in general, it seems like,
16:36:57 Oh, please go ahead. Oh sorry yeah I just had a quick question it seems like it seems really interesting that you've got your adaptation was through this like massive duplication.
16:37:11 And I wonder if you've looked in other fields genomes to see if there's any signatures of, you know, yeah, highly repeated regions within these very tightly constrained genomes okay so there's two answers, both of which are a little unsatisfying.
16:37:27 I'll say that off the bat.
16:37:28 So first thing is, most phase when you assemble them, you don't assemble with the intention of having these duplication because they're most assemblies to shorter in sequence.
16:37:37 So it's almost impossible to know what the two genomes structure looks like. So for example, this, it's because I have a reference, I can tell you the coverage is increased right, there's no real otherwise way for me to align it in such a way that I can
16:37:49 see the amplification in some structural manner. So one of the things we're doing is long read sequencing and only now people started doing longer sequencing for viruses, we don't know the answer to the second answer maybe as a something I'll point out
16:38:02 is also a very quick at shrinking things that they don't need and potentially gaining back, other things so it's not clear that there should be a stable thing ad infinitum probably move back and forth, and in fact in a similar evolution experiment for
16:38:16 if I'm not mistaken pox virus in eukaryotic cell line. The lab of heartbeat Malik in the heart, saw similar thing where they call it the accordion were essentially you see an expansion of our region.
16:38:36 that improves efficiency and accordion snaps back. Um, I would argue sure it's a little harder because box size is in the envelope is so the genome can actually increase and decrease in size without being as costly to the envelope.
16:38:45 But once you lose a large part of your genome, especially if you fix. There's no more sampling to get back in the genome. And so that's why I think the next experiment of like putting back white it for now asking, Can you get better, and then bring back
16:38:58 all genetic material would be interested in each, each successful infection. You said you have a birth size of 100 or 200. Isn't it true that each of those hundred would have a different random deletion subset.
16:39:12 Yes, absolutely. So the one thing I will say is, that's not strictly speaking through the senses, to where the deletions happen is at one point of the head full backing.
16:39:23 Right. And basically the way headphone packing happens is you have a concatenation of DNA genome, all connected to each other through the combination.
16:39:33 And basically the way you back is you back into the contract, and then you move on to the next head.
16:39:37 And so presumably in this ensemble, sometimes you delete an essential gene that page particle is just a dead particle, you know, so presumably what's fixing is the thing that appreciably rises above some floor before it gets lost a draft that doesn't
16:39:52 delete anything that is essentially. That's because you're growing in the context of one bacterial thing, but just in the natural environment it would seem like each birth size sort of covers all kinds of strategies.
16:40:03 And so for example you could imagine if I have these bacteria in the environment, generate sub populations that are each specialist to both the defense system, and a horse that needs another defense system similar to this question about why bacteria have
16:40:15 have the shuffled gotcha sets of defense, you'd imagine phase would also start diversifying to have shuffled cassettes of everything they need.
16:40:24 I think, okay,
16:40:35 while he's waiting. Yeah, well let's figure it out. So I want to follow up on bonus question. Absolutely cola actually has a doesn't also have this toxin anti-toxin Justin's.
16:40:47 The yy the other ones are not working. Yeah, so that's a very good question. So it turns out so there's a number of people in the lab lab who actually studied this, it turns out some of the toxin antitoxin systems don't have anything to do with page defense
16:40:59 in which they protect. So you could similar principle, the monitoring and the stoppage of a transcription.
16:41:07 Not all of them seem to the other large system for example yeah so some of them can.
16:41:14 Not all of them seem to, none of them but the others, March system for example yeah so some of them can. I think one of the things to.
16:41:17 I guess I don't know the answer for white doors ordinances are in a, unable to like actually crush the page system as well. observational, they aren't.
16:41:27 One thing I will say is there's a number of things that also influenced the protease of equalizer, where it might start changing what gets degraded. In terms of the horse protein right because the horse proteins all start crashing as well.
16:41:37 So there might be specific ways in which it has targeted particular proteins progression degradation, but that's a guess, I have no idea. It gets back to the question of the host range and then so.
16:41:49 So how high throughput is it possible to do you know very large number of, you know, close relatives of eco lie, and a large number of pages and measure the yeah so that's actually basically one of the things I thought I could do for the postdoc, so right
16:42:05 now, what I've got is a geometry and a 96 well play because one of the advantages is Paige are extremely population dense. You don't need much volume to get your populations quite high.