[Audio] RadioBio interviews Dr. Dan Weinreich on antibiotic resistance and microbial evolution


The winery Club at Brown University Uses antibiotic resistance genes to study how a gene can evolve in bacteria to give rise to increase Fitness? Today we are going to learn about the exciting world of gene evolution and discuss science philosophy with our guests dr. Reiner himself Hello and welcome to radio bio, my name is Craig anise and Today we’re joining with dr. Daniel Heinrich. He is an associate professor of Brown University in the Department of ecology and evolutionary biology Thank you for joining us today. Thanks for having me So to start off who are you and what do you study? Well I am professor of biology, I’m interested in evolution evolution and natural selection I’m particularly interested in the kind of simplest mathematical models of how evolution works how natural selection and random processes and mutation interact to change the genetic composition of a population over very short periods of time we take that as a Proxy for what’s gone on in populations on earth since the beginning of life three point something billion years ago But we don’t make any attempt or claims to study Super long term processes, so we work on a small number of mutations over the timescales that we can realize in a laboratory population So do you work with like a model organism is there specific one that you I would describe myself as a theoretician okay? We work a lot, but we do have a lab and we work a lot with microbes with bacteria e.coli We’ve done a lot of work with e.coli. We have two big projects right now in the Brewers East We’ve also worked with a kind of virus that that is a threat to bacteria though Not to humans, but what these organisms all have in common are some technical Facilities some technical opportunities that let us construct experiments that are in alignment with our radical Framework so we write down some algebra Necessarily making a ton of simplifying assumptions, and then we want to ask okay We assume certain values for parameters in our theory what actually goes on in biology, so These organisms we can grow in population sizes from ten individuals to a billion Easily we can manipulate their DNA and introduce mutations that might be interesting we can sequence their genomes very easily We can control the environment in which they live which is a really important element for us because natural populations like fruit flies or mice or humans live in very complicated environments that impinge on their evolution in ways that Are essentially unknowable? And lastly we can freeze these organisms and then follow them out and compare evolved Strains to their ancestors and see how in what way they’ve changed When you say we you just clarify Yeah, so I run a small research lab at Brown University I have a small amount of money from the federal government to employ a couple grad students and a couple postdocs and lots of undergrads but really when I say we I mean the intellectual community that I operate in so I have colleagues around the world not like a Dense number of them, but I have a few friends that around the world who are interested in similar questions And we all kind of take advantage of these same opportunities It’s really a great time to be in this field because the theory is really Developing rapidly and the experiments and the technology would genome sequencing and and related kind of technological breakthroughs in the last 20 years It’s really a boom time for the study of evolutionary genetics. It really is So you’ve done a lot of work studying the beta lactamase gene Can you tell us what that is how does it work and what does that mean? Yeah where everyone so beta lactamase is an enzyme And its job is to digest Drugs antibiotics, so Drugs that it digests are all called beta lactams. That’s why this enzyme has a name That sounds so similar just beta lactams penicillin is an example of a beta lactam that everyone’s heard of so is amoxicillin which my kids got when they were young and There are more Recently developed drugs that are in the same chemical family these enzymes Render the drug Biologically inert so these drugs kill bacteria they interfere with Biological processes the bacteria need to survive, and if you expose bacteria to the drug they die But if the bacteria has the gene for this enzyme the beta lactamase Then they can protect themselves because the beta lactamase that’s just you know floating through Their world when it comes into contact with the enzyme, it’s digested, and it’s rendered inert okay, so how do we use the beta lactamase to gene to study evolution yeah, so Actually, let me since we’re at Merced. Let me give a shout-out to my colleague Miriam Barlow. Who’s here and whom I met a Few years ago more than 10 who turned me on to the system. I had this I Told you guys earlier that I’m interested in kind of linking theory with experiments are these theoretical theoretically framed questions about What we call evolutionary or mutational trajectories, that’s to say? sequences of mutations that over time Allow an organism to get better at doing whatever it does so I had some very simple questions in mind that to which there were not good empirical answers and In particular what I was interested in is the following Suppose you know that some combination of say five mutations together give The organism some big boost and just to cut to the chase in the case of beta lactamase There are five mutations that allow bacteria to survive in a hundred thousand fold higher concentration of a particular beta lactam and biotic Relative to the concentration that kills them without those five mutations okay, everybody’s got the gene, but the gene is Evolving and these five mutations together allow the bacteria to survive at hundred thousand for higher concentration the drug so that’s a huge signal and My question was how well do those five mutations play together, so? At one extreme you could imagine that I have five mutations I put in any one mutation and I get 20% of the boosts one-fifth and I put in the second mutation and I get a second boost of one-fifth a Purely additive model it doesn’t even matter which mutation They’re all interchangeable so that would be that’s sort of at one limit and in that kind of a world If I can put it this way life for natural selection is very easy there aren’t a lot of traps there isn’t an opportunity To evolve to say a dead end mmm-hmm at the other extreme You could imagine a world where you don’t get any payoff until you have all five mutations together and if that were the case It would be very difficult for natural selection to evolve these five mutations to discover and to discover and keep these five mutations because no Subset of them does anything so natural section can’t tell what it’s looking at until all five come together and the probability of any particular Specific set of five mutations appearing all at once is really really low It’s it’s analogous to I think of it as being analogous to the difference between building a stone wall and a stone arch so if you want to build A stone wall you add one more stone you’ve got one more unit of wall It doesn’t matter which stone you pick up you just keep adding the wall gets higher That’s good. You want to build a stone arch you put up some kind of wooden scaffold and you pile up your stone, so you don’t get a stone arch until the Keystone goes in and Four-fifths of a stone arch, there’s no stone arch at all, and we did not know in any system. Yeah, it works We did not know in any system What the challenges were for natural selection no one had ever Thought to ask that question And I met Miriam and I said I have this question and she was finishing her PhD at University of Rochester and her professor Barry Hall had a had been working in beta-lactamase He’d been asking related questions, but hadn’t really gone this far with it And she said you should look at beta-lactamase and that’s what led to that 2006 paper and and the punchline is The truth in that system is somewhere between those two extremes, so there are you can’t just throw those mutations together? Anywhere do you want, but in some orders you can get and? in some orders the mutations will each one improve drug resistance and The point is that’s the property the natural selection can see so if your mutations are not improving drug resistance or Fertility or ability to find food or whatever the natural section doesn’t know what could they are and in this system There’s enough of a signal The natural selection can make progress, but we didn’t know What the answer that question was so did you limit your? Your analysis to only these five residues that are known to be Like how fitness fitness effects. Where did you look across the whole? Genes yeah, so it’s a really good question. We only looked at those five and in every case The effect you get from adding a mutation Depends on what the background is what other mutations already there, so I think I said One way of thinking what the study was to ask how well do these mutations play together? And the answer is how well they play depends completely on who they’re being asked to play with so the total genome the complement so the rest of the genome which calling is the genetic background that has a set of genes that may or may not work together well and Every particular mutation in a beta lactamase gene has to work well with all these other genes No, actually we held everything else in the genome constant so when I say background. I mean the other four mutations, so I’m going to put in so one of them changes glycine to G 238 a Allanon s Syrian policing becomes Syrian opposition to 38 and by and that improves drug resistance But the amount by which it improves drug resistance depends on the presence or absence of the other four mutations in the gene And that’s an example of epistasis right so thank you that When I said the effect that a mutation has on Whatever the trade is of interest if that effect depends on genetic background that is epistasis. That’s the definition of epistasis Well so you mentioned earlier evolutionary trajectory yeah, so can you explain to our listeners What that is yeah, and how does that play into this research? Yeah, so we have five mutations and You could imagine that you can get from the starting point to the ending point by adding let’s just number them left to right one to five so you could add mutation one and then mutation two and then mutation three and The mutation for another mutation five that would be one trajectory that would carry you from the beginning to the end but you could also imagine adding mutation to and the mutation 3 and the mutation 4 in the mutation 5 and then mutation 1 So the end point is the same, but the sequence of events differs? And so that’s what we describe as a trajectory is some particular sequence of events ok So there are any limits to the secret of a sequence of events that could occur if you have five mutations then in principle You can have up there there are 120 Trajectories 120 orders in which you could add those mutations the way to think about that is starting from the beginning You’ve got five options once you pick one of those you’ve got four remaining options Once you’ve picked that one you’ve got three remaining options once you pick that one. You’ve got two remaining options There’s only one so that’s five times four times three times two times one. That’s 120 okay, so there hundred and twenty trajectories if you just think about the mutational process and We did not have any data in any system that told us, what selection thought and the punchline of the paper was of those 120 only 18 of The orderings have the property that drug resistance goes up at every step and That’s as I said a few minutes ago. That’s the only thing natural selection can see natural selection is in trend Profoundly short-sighted, it’s like Wall Street. What have you done for me lately so if this mutation doesn’t? To use the metaphor pay dividends now Natural selection will take a pass okay? So how did you decide whether or not a sequence of amino acid substitutions is? Good enough for natural selection or not do you look at you just a selection experiment you? had microbes growing in a certain concentration And you up the concentration because you added a mutation you tried to see if they were still surviving That that’s what people had done That’s how we identified these five mutations and now when I say we I mean the community I did not do that work, but another group actually Barry Miriam’s PhD adviser did a key experiment But people had identified these five and he’s contributed this to these five mutations as jointly giving this big effect there’s a hundred thousand fold increase in resistance, but no one had asked about the orderings so I Haven’t had a chance to tell you what my training is about my undergraduate degree is in computer science And I used to give this talk Is you to know? When I did the work and I was on the road giving this talk I would I always said without any I did what any good computer scientist would do there are five mutations That’s only two to the fifth combinations thirty-two. I just built them all I just built them all I Told you that one of the reasons we like microbes is that there are these technical opportunities, and one of them is genetic engineering yes even easier Nowadays, there’s a lot of excitement in the community you mayority of interviewed somebody uses CRISPR technology Which allows you to engineer in specific mutations into eukaryotic genomes, which are profoundly more complicated than bacteria yes But for many years, there’s been technology that allows you to put any mutation you want into a bacterial genome And so I this was this is Technically very simple worked at an undergraduate who spent the summer building these I mean I built one or two to make sure I know how to do it and then I supervise them, but he did all the work, it’s just a you know PCRs and sequencing Have a student in a lab right now. Who’s building some of these and undergraduate? It’s the technology is is really accessible, so you have his library of mutations, and yeah Well, it wasn’t even a library we had we had pure culture so you Engineer in the first mutation, this g2 30s that I mentioned earlier, but in the freezer. I have 32 tubes One for each of these variants, I just in pure culture measure How well it grows in the presence of the bacteria? Another in the presence of the antibiotic another really cut-and-dried technology that Miriam turned me on to I mean this is stuff that medical Microbiologists are very concerned with a patient presents the doctor they presents of a hospital or to a doctor with an infection That’s not responding to an antibiotic so they’ll culture the bacteria and measure its Resistances this is well-established protocol for doing that and we just adopted that protocol we take a man like M I see we’re talking exactly about mi. Sees yeah, it means minimum inhibitory concentration. That’s right, so it’s the lowest Concentration that blocks bacterial growth and the way you do it for any one of these 32 Constructs is you make a series of? Media that have a rain each of which has a different Concentration of antibiotic and you inoculate them with your strain, and you simply ask What’s the lowest? Concentration of antibiotic at which the bacteria fails to grow it’s done visually if you could not be undergraduates do that routinely in the lab Tool, so what kinds of questions are still unanswered in your system of study Beta-lactamase is called beta-lactamase because it hydrolyzes beta lactams yes, and it was purified using technologies that you start with a total cell you grind it up you get a bunch of Proteins or enzymes and you find the fraction that has the beta lactamase activity and then their biochemical techniques to purify The fraction and you just follow the fraction that has the beta lactamase activity until you have a fairly pure sample and you know in 1964 you do Edman degradation or some protein sequencing technologies or don’t you got in your hands, but that Doesn’t mean that beta lactamase is the only thing that hydrolyzing beta lactams does the only thing that a beta lactamase can do and I in my view, that’s Generically true for all enzymes that have a name because that was the activity that was followed by the chemist who purified the thing First place and then cloned the gene blah blah blah blah blah, but from the cells point of view. That’s just the label and it’s reasonably well established I think that many enzymes will bind a different substrate and maybe do the same chemistry but to a different substrate and a rate That’s one millionth the rate that the biochemist used to purify the thing but from the cells point of view maybe that’s important So that’s called enzyme promiscuity And I have an idea that I don’t I don’t have the technology running in the lab So I can’t even imagine writing a grant to do it, but I have an idea for how to get after the question What is natural selection think about these promiscuous activities like you could imagine the natural section doesn’t care? So in the way, that would work is you’ve got of there You’ve got a bunch of variance of beta lactamase out in nature, and you would ask well You’ve got a single. Gene. It can fold in multiple confirmations and each of those confirmations can do something a little bit different now What’s natural selection going to do with that well now? I do need right natural selection primarily operates on genetic variability, so let’s imagine we got a second allele Which has a slightly different? Conformational Ensemble it’s it mostly folds into the same states as the first variant in movement and thus mostly has the same kind of fingerprint of Enzyme enzyme reactions that can catalyze, but it’s slightly different what natural selection might have an opinion I’ve got two competing variants and the only difference between them. They both are great beta lactam ASIS. They both at the same m.i.c Okay, but one of them has one millionth fold activity in some other On some other substrate it hydrolyzes something else at one millionth of the rate that it hydrolyzes the beta lactam and the other one Has that same second activity, but five times stronger? right now There are these two things so let’s imagine these ooh things competing the population their equivalent with respect to their named Activity they protect the bacteria just as well against the drug, but one of them is a slightly higher activity Albeit extremely though what natural selection might not like that that might be a meddlesome activity I in the actual second the organism says I didn’t want that other thing hydrolyzed I’m not able to grow as effectively so natural selection would suppress the variant that has this slightly higher secondary activity Naturally might say gee that’s awfully handy like that actually is something that the organism can take advantage of and enrich For the variant that has this these are really in my imagination These are really low-level activities, but my intuition is natural selection might have opinions on the value of those additional activities my intuition is Those additional activities are just part of the process of being an organism It’s not like an engineered thing where there’s a pipeline where we do this here, and we do that there and it’s just a bag of mush and and so I put some ideas for how to get after that experimentally but So imagine this is your last chance to teach the world a single concept in biology. What would that? What would you say to them? Yeah, that’s the one question that is the tough one that I was like whoa and I was that girlfriend about it this morning That’s pretty philosophical and honestly. That’s where I go with it Because my intellectual passion for evolution is Includes is motivated by a strong sense of aesthetics like I don’t pretend that there’s a purely scientific motivation that and I just am Well, let me start. Let me put this in a little more rigorous context. I told you I was an undergraduate computer science major and as an undergraduate I had the great fortune of walking into a graduate seminar at my undergraduate institution taught by a guy Who had invented something called genetic algorithms ok and so these are computer programs that evolved? mm-hm and the original application was to Use them to find computer programs to perform tasks That are so complex that we don’t really understand how to do them, so if you want the next digit of pi I can write down an algorithm that will just spit them out one after the other but there are lots of challenges in the world like Well playing Games like chess and checkers where at least when I was in undergraduate, which is the the 80s? They hadn’t been solved by in a mathematical sense so there were only heuristics and Problems like that are really hard to program and so the idea here was let’s get a Population of computer programs that have a slightly different each of which has a slightly different set of heuristics to solve the problem We’ll just play them against each other and find the set of heuristics That’s most successful, and and there would be of so as a population just like in in my field in population genetics and there’s variation and there might be some kind of Variational operator that introduces new variation maybe by mixing The left half of one heuristic with the right half of another which is inspired by genetic recombination You’re talking about your genetic algorithm is also a selection. Yes, that’s right So you have to have some kind of a they call an objective function that allows you to tell Which of all of the members in these populations are doing better and what you’re doing not as well like this Like analogous to a fitness function that’s right, and I was a computer scientist. I was taking classes in data structures and and algorithmics and stuff and this Possibility that this is essentially I mean this is Darwin’s paradigm, but I saw it as an algorithm like there’s mutation and recombination and reproduction and survival and mutation and recombination and Reproduction as well. It’s just a loop and so I got very interested in what that algorithm could do and That’s still what I’m really interested in and and so that’s kind of like an intellectual motivation, but the thing that really Makes me so excited is to couple that kind of intellectual traction with the extraordinary diversity and complexity of life on Earth so Somehow in my last lesson to humanity that would be the a message I would try to communicate But that’s why I was saying to you my answer is kind of philosophical, and I don’t really know how to put that into a 50 minute lecture Or a 50 minute podcast, but that didn’t look back to me that That that there can be a material explanation For both the world that we experience and the fact of our experiencing it mm-hmm blows my mind So that would be what I would want to communicate very cool very cool That’s like is that like the idea that if you if you knew all the algorithms for everything, right? You know you could calculate the perfect statistic like that sum that one event would happen But your time out like but how do we explain what we observed now, right? This is because we if we could do it. We could use these prossies simulations, maybe and we could find out maybe How it happened like at least series yeah, I mean I’m I’m not that the piece that I’m Resisting is the idea that you could have perfect knowledge like these processes are so dependent on on stochastic components randomness like which mutation happens next who knows Yet my beta lactamase experiment tells you that if this mutation happens then the next four mutations like there are Forks in the road and If you just by chance take the left fork the next Period of evolution will be different, then it would have been if you had taken the right fork in the road that That’s not actually that clear in my data, but other data sets that are like it Even in beta lactamase totally make that point There is also a lot of randomness in populations with respect to who actually Successfully reproduces, so evolution is a process that occurs not to individuals But the whole populations right I’m born with the genes. I have and for better or for worse that’s these are my cards, and I’m going to play them out, but the population over time changes in his genetic composition and A part of that this key part of that is who gets to reproduce. That’s where the next generation comes from and You can have the greatest allele in the world Let’s say for resistance antibiotics But you fell into an open manhole and died and that good gene didn’t save you so there There’s a lot of randomness at a lot of levels so the best we can hope for I think are sort of statistical Averages and things of that nature, so that was the part to your question that I was wrestling with those I don’t want to imagine that we can have certain knowledge But we can I think we can learn a lot Yeah, we learn a lot, so I mean relatedly I have to admit that I’m kind of more interested in framing good questions and Less interested in answering them There are a lot of things that are interesting that that that are curious about the world That I don’t know how to begin to answer It’ll be great if we could sprout wings we could fly around And that would be pretty awesome, and yet that hasn’t happened and doesn’t seem like it’s gonna happen, but why I don’t know People have theories, but that’s not a place where I feel like there are experiments that one could do So a large part of what gives me joy is Identifying places where there’s theory that can guide experiment? almost that’s more fun than what the answer actually turns out to be so that there are 18 paths instead of 38 or 120 like whatever answer there was mm-hmm would have been neat because the question was well formed and The answer will be compelling no matter. What but do you have any recognitions for how to start asking the good questions? So you recommended looking to the literature, but you have any other? advice for how to formulate a good question No I Mean that is really You know the $64,000 question. Yeah, speaking of answering a question with a question Yeah as a new someone that’s new to a graduate program Learning to ask the good questions is something that I’ve been like trying to work on So I guess more literature, yeah The literature is important. I think another Good idea is to talk to people and most importantly and this is Possibly the most important lesson I learned in graduate school to ask plenty of questions of other people and this also connects with what I was saying earlier about how? Some of the students, I’ve enjoyed working with the most really Have a confidence. That’s ego free, so they’re perfectly happy to say I don’t understand and And you know the best collaborations that I’ve experienced are with people that have You know a shared interest in a problem and everybody just kind of drops their guard and starts chiming in with ideas um So that’s not really A thing you can bottle a technique you can deploy on on demand or on demand but I would say You know my students Come back to me with questions, and we sort of brainstorm together, and that’s the way it works with my collaborators as well and so that’s yeah, I don’t know this idea that that drove the 2006 study I Don’t exactly know where that came from that was mine mm-hmm But I don’t know what I was thinking about the minute before I was thinking about that mm-hmm But it’s like now, it’s a cottage industry like there are lots of people doing those studies And so we’re getting these lovely datasets that like mine so maybe in another system there were six mutations And so if there are six mutations then there are 6 times 5 times 4 times 3 times 2 times 1 which is 720 possible paths and maybe it’s a different fraction of paths that are accessible and We can kind of take averages and look for kind of trends and patterns across these data sets and try to learn Kind of deeper things about biology like for example The the data aren’t very strong in this dimension yet, but the little bit that I’ve seen suggests that these patterns of epistasis the constraining consequences of epistasis are much more common among mutations within a single gene and less so in mutations between genes so that sort of reinforces an idea that’s old Which is that genes code for proteins and proteins kind of are these atomic objects that do their thing? It’s like a subroutine in computer science where there’s a well-defined interface, and what goes on inside of it You don’t have to know and so if you change one component in a big software engineering package it shouldn’t break other things But if you change one line it might break that one module that modules inputs and outputs that Okay, function that does it remains in there right so it doesn’t really matter inside Let me just point out that I’m bringing a lot of Suppositions about the way biology ought to work Which aren’t that well established yet in my view so as I say there aren’t great data sets that compare the kind of epistatic patterns within a Gene among mutations within a gene to epistatic patterns among mutations between genes but the data since we have are consistent with that idea of modules I Think there’s a great temptation to look at Molecular biology the way an engineer would and that that’s good like those are good tools but it’s not always sufficient to understand what’s going on because Evolution is not an engineering process oh Yeah, so if our listeners want to get in contact with you. What’s the best way female email? Yeah, so it’s just my last name Weinreich at Brown edu well Yeah, be very happy to talk to anybody about any of this stuff. I mean You know this is the curse of being a professor’s you can talk a lot longer than anybody wants to listen About your favorite subjects that just comes with the territory Well, thank you so much for joining us today sure my pleasure Radio bio is supported by the quantitative and systems biology graduate group at the University of California Merced

Add a Comment

Your email address will not be published. Required fields are marked *