Ep 120: Shifting mutational landscapes (with Deepa Agashe)
What is mutation bias and how can scientists study it? How does changing a population’s mutation bias influence its evolutionary trajectory?
In this episode, we talk with Deepa Agashe, an Associate Professor at the National Centre for Biological Sciences in Bangalore, India. We first talk with Deepa about mutation bias and how she uses E. coli to understand it. We then focus on a 2023 PNAS paper about the fitness effects of experimentally changing the mutation bias in E. coli. In this research, Deepa and her team used a strain (MutY) of bacteria containing a mutation that knocks out an important DNA repair enzyme. They then isolated subsequent single mutations produced within both MutY and wildtype lines and studied the fitness effects of those mutations. Surprisingly, more than a third of mutations in the mutant lines were beneficial, and often across several different environments. Zooming out, the big picture is that shifts in mutation bias seem to generate new kinds of mutations that weren’t previously accessible to lineages, and a greater fraction of those may be beneficial in some circumstances.
Cover photo: Keating Shahmehri
-
Art Woods 0:00
Hey, Big Biology listeners, just a reminder that we're moving into a more intensive period of fundraising over the next few months. So you'll notice a couple of changes to our show.
Cameron Ghalambor 0:09
First, behind the scenes, we're applying to foundations and other funding sources that will help keep Big Biology on the air.
Art Woods 0:16
So if you know of any funding sources that support scicomm podcasts, please share them with us, you can reach us at info at big biology.org.
Cameron Ghalambor 0:25
Second, please consider making a donation, either through our website or through Patreon.
Art Woods 0:30
If you donate through the website, go to www dot big biology.org and click on the "About" link. It's super easy to make a safe and secure donation using a credit card.
Cameron Ghalambor 0:41
While you're at the website, you should also click on the shop link where you'll find all the super cool big biology art on T-shirts, coffee mugs, and other stuff.
Art Woods 0:50
Or you can go to patreon.com/big bio and become a patron for just a few dollars per month.
Cameron Ghalambor 0:56
Patrons get cool insider stuff like access to behind the scenes audio and extras from our guests about their lives, their hobbies and their careers. Now, onto the intro for today's show.
Art Woods 1:06
So Cam, you've been co-hosting Big Biology for a while now. How are you liking it?
Cameron Ghalambor 1:17
Well, it's an amazing opportunity to talk to other biologists. I also like to hear your dad jokes. But as you know, it's also a lot of work. I've been really lucky that I could lean on you and Marty for pointers.
Art Woods 1:31
Thanks, but it hasn't always been easy. We went through a steep learning curve, which you'll realize if you listen back to our early episodes, and during that time, we learned a lot about what works and what doesn't. One example, so early on, we thought that listeners would like much shorter episodes. So we went through a phase of producing both the full conversations and some small 15 minute mini versions. Turned out, almost no one downloaded the short ones, and so we stopped doing them.
Cameron Ghalambor 1:56
So sampling from what works and what doesn't work is actually a nice transition into a major theme of today's show, which is what mutations get sampled during the evolutionary process. In other words, what kinds of mutations happen and how do biases in those mutations shape evolution?
Art Woods 2:15
We all know that "genetic variation" is necessary for selection because evolution. We know much less though about how that variation arises in the first place, how it persists in populations, and how the nature of that variation can influence evolutionary trajectories. In fact, this analogy is often replaced by simple ideas.
Cameron Ghalambor 2:36
For example, a simple idea is that mutations occur randomly throughout the genome.
Art Woods 2:41
Another important idea has been that most mutations are effectively neutral and that the occasional deleterious mutation will be rapidly removed by selection. This led to Motoo Kimora is neutral theory of molecular evolution.
Cameron Ghalambor 2:54
Another similar idea is that the occasional beneficial mutation will be swept to fixation by natural selection.
Art Woods 3:00
But as we discussed with Arlin Stoltzfus in season five, we now know that mutations aren't completely random, that certain types are more common than others, and that mutation types and amounts can vary a lot depending on where you are in the genome.
Cameron Ghalambor 3:15
Collectively, these kinds of patterns are known as a mutation bias. For example, the well known transition-transversion bias, where in an A is replaced with a G more commonly than it's replaced with a C. But we know a lot less about how important or predictable this type of mutation bias is for fueling adaptive evolution.
Art Woods 3:37
One important player here is the DNA repair enzymes that find and repair mutations.
Cameron Ghalambor 3:43
The interplay between processes that cause mutations and these DNA repair mechanisms ultimately creates the range of genetic variation that can be sampled during the evolutionary process.
Art Woods 3:55
Our guest today Deepa Agashe, an associate professor at the National Center for Biological Sciences in India has been thinking a lot about the causes and consequences of mutation bias.
Cameron Ghalambor 4:06
We talk with Deepa about her recent research using mutant strains of E. coli. One mutant, called MutY, no longer has an important DNA repair enzyme, which leads to both higher mutation rates overall and different distribution of mutation types.
Art Woods 4:23
Most of our conversation with Deepa is focused on a 2023 PNAS paper. In it, she and her group did a mutation accumulation experiment in which they isolated the fitness consequences of a single new mutation in both wild and mutant lines.
Cameron Ghalambor 4:39
Surprisingly, they found more than a third of the mutations in the mutant lines are beneficial not only in one environment, but across several different environments.
Art Woods 4:48
These results are thought provoking on many levels, because conventional wisdom says that beneficial mutations are rare and that tradeoffs are common. In other words, what works well in one environment tends not to work well and others.
Cameron Ghalambor 5:01
So what's going on? It appears that mutation bias actually plays an important role in generating a new range of mutations that can be sampled.
Art Woods 5:10
In particular, what Deepa's group found is that the mutant bacteria create relatively more transversions. Compared to transitions, which is a shift in the mutation bias.
Cameron Ghalambor 5:20
This bias opens up a whole new spectrum of mutations that can be sampled by natural selection.
Art Woods 5:26
We explore deeper how this evolutionary dance between mutation bias repair and natural selection can challenge commonly held assumptions about the process of evolution.
Cameron Ghalambor 5:36
I'm Cameron Ghalambor
Art Woods 5:38
And I'm Art Woods.
Cameron Ghalambor 5:39
And you're listening to Big Biology.
Cameron Ghalambor 5:55
Deepa Agashe, thanks so much for joining us today on Big Biology.
Deepa Agashe 5:59
Thank you for having me. I'm excited.
Cameron Ghalambor 6:01
Yeah, we're really looking forward to talking to you about your research in evolutionary biology, and in particular, about the causes and consequences of genetic variation for phenotypic diversity. And I guess to start, I'd like to say that I first became familiar with your research, maybe ten or so years ago, when you published a series of really nice papers on the consequences of genetic variation on flour beetle population ecology and evolution. And I've followed your work ever since, and I see that you've now added E. coli as another sort of model system for setting experimental evolution. And, you know, I guess I was kind of curious, I see the obvious advantages of these kinds of systems for studying evolution in experimental contexts. But I guess I was just curious on your path to choose this type of system for studying evolution, as opposed to, you know, working on charismatic organisms in nature, like butterflies or birds or mammals. So how was it that you came to focus on these types of experimental evolution systems?
Deepa Agashe 7:15
So when I started off in my PhD, trying to figure out an organism with which I could do experimental evolution, in a PhD timescale, so four or five years, I decided that I didn't want to work with stickleback fish, which were what my PhD advisor worked on at that point, because I would still be a PhD student maybe. So I needed something that was short generation time, that was, you know, something I could work with easily. And also something for which I would not require ridiculous amounts of logistical support, let's say, right? So I remember making a list, I read a bunch, I knew already that I wanted to ask how genetic diversity influences population dynamics, that was one of my core questions for my PhD. And so clearly, I needed something where the population dynamics would be interesting. And then it'd be nice to use a system where previous work had already shown that, you know, populations would fluctuate in their size, interesting things could happen. And then I could ask how those things differ if I changed genetic diversity. And insects is something that I had worked with a little bit earlier, I knew I was happy with working with insects, no problems, and so on. And I could find in the literature, three systems that seem to suit a bunch of criteria: Drosophila flies and flour beetles, on which there was beautiful work from the early ecologists, and you know, three entire volumes of work on flour beetles encapsulated, across a few decades of research, and then blow flies, which you know, feed on rotting flesh. And so I remember going to my advisor, Dan, and saying: "Hey, so here's my shortlist, what do you think?" And he asked me a bit about all of these systems, and he says: "Let's not do blowflies, we don't have working stuff in the lab, and it's going to stink everything up."
Art Woods 8:49
We don't want everybody smelling that in the lab .
Deepa Agashe 8:51
Yeah, so blowflies were out. And then I thought between Drosophila and Tribolium, Tribolium was a little bit more interesting to me in many ways, because the ecology was much more, a bit better known in some sense at that point. And the ecology could also be retained a little bit in the laboratory, unlike what I could understand about Drosophila at that point. Because these beetles have evolved as pests, you know, they used to hanging out with, they're human pests, you know. So they're okay with just being given tubs of flour, that's kind of their natural ecosystem. And so it's easy in the laboratory to simulate something that closely resembles what their natural habitat is, which would be difficult to do for other systems. So that's how I chose beetles.
Deepa Agashe 9:34
And I think the advantages of continuing to work with beetles now is that, of course, it's the pest species, so anything we find out about it is potentially useful, for control and management. It is a big problem in India, as well as really globally because of huge amounts of grain losses. The other useful thing is that it's a system that I could get, you know, from anywhere. So one of the first things we did when I set up my lab here is that we went out to different cities and collected from grain warehouses all over, flour beetles. And one of these inbred lines that was set up in my lab, actually, was a lonely fly beetle that I found in my own house, which had probably escaped from, you know, some box of wheat that I had left lying around for a while. So it's got its advantages, you know, otherwise, it's like a good model system, like many others are short generation span, easy to use, and so on. With bacteria, I again chose to work with E. Coli., because just the genetic resources are so huge and readily available. And for the questions I wanted to ask, which were focused on very broad principles of how mutations influenced fitness, how selection works. You know, one doesn't need anything peculiar about that organism that one wants to focus on. One was something where it's very easy to manipulate the things that one wants to. So that's how I chose E Coli, and then it just works.
Art Woods 10:49
This is maybe a diversion from our main thread, but wanted to ask about the flour beetles that you collected from different grain warehouses in different cities. Were they really different one from another? Or was there a lot of homogeneity across the collection?
Deepa Agashe 11:02
Yeah, so that's an interesting question. We haven't published a lot this work yet. But we did a lot of phenotypic assays, when we first brought them into lab, after some few generations. And we found enormous phenotypic diversity across these flour beetle populations. In terms of morphology, they all look the same. I can't tell any of them apart one from the other. But if you look at how well they survive in different resources, how many eggs they lay, how fast they develop in different kinds of cereal flours, there's a lot of diversity, right, the immune function, a lot of diversity. And yet, we recently were just wrapping up some work by a new PhD student who joined my lab later on the genomic aspects of the diversity here. And across all these flour beetles, the diversity is very low, very, very low, right? And so, I mean, I'm not saying that the genome is 100% identical. It's not there aren't differences. And there are clear population level differences, but it's nowhere close to what I expected, at least looking at the phenotypic diversity that we observe.
Art Woods 11:58
So what explains the discrepancy between those two sets of observations? Like how do you get all the phenotypic diversity without also underlying genetic diversity?
Deepa Agashe 12:07
Yeah, so one way to think about it is that perhaps a lot of the phenotypic diversity reflects plasticity. I mean, even though we try and maintain them in controlled environments, and so on, so forth, I think they have a lot of responses in their phenotypic traits that are related with how crowded they are. And so if there are small differences in basic growth rates of these populations, and that changes the degree of crowding, I feel that alone could explain some of the phenotypic variability that we observed in our assays, even though we tried to control for some of these effects. The other possibility is that whatever little bit of genetic diversity we do have in these, that segregate these populations, that has some large impacts, which affects or explain some of the phenotypes that we observe. I don't have the answer exactly to what is going on. But those are my guesses. And we'll have to think a little bit more.
Art Woods 12:58
Yeah, great. So last year, we talked with Arlin Stotzfus about mutation bias, and in quite a theoretical sense, based on the book that he recently published, and you've been studying mutation bias intensively in a quite a nice experimental context. Last year, you had this fantastic PNAS paper, we'd like to focus on the details of that paper. But before we go there, maybe can you just say, what is mutation bias and why does it matter?
Deepa Agashe 13:24
So mutation bias is the observation that some kinds of mutations occur much more frequently than others in pretty much all the genomes that we have surveyed so far as a body of scientists. So as an example, if you can split your mutations that occur into point mutations versus sort of big chunky movements of DNA across, so what's called structural variation, most organisms will have many more point mutations. But there are differences in the frequency of each of these types of mutations, and you can split mutations, many different ways you can think of them as those that occur in coding regions versus non coding regions, those that lead from an A or a T base to become a G or a C, or vice versa, so that's called ATGC bias. Ones that are transition mutations, those that lead from pyrimidine to pyrimidine or purine to purine, versus transversions, which cross across these two boundaries, so purines to pyrimidines and pyrimidines to purines. And so you can split basically different types of mutations, different ways. And it's almost every organism for which we have good data, across bacteria to all kinds of eukaryotes, it's clear that they all have a bias in which kinds of mutations are more frequent, and which kinds are less frequent.
Cameron Ghalambor 14:40
So maybe let's jump into the details of how we actually study mutation bias. So in your work, you've compared wild type strains of E. coli to this really interesting sort of mutator strain where the repair mechanism, so this is a DNA repair enzyme, has essentially been knocked out. And so can you describe a little bit like how you're able to kind of take advantage of these mutator lines to study mutation bias?
Deepa Agashe 15:13
Yeah. So for many decades, a lot of people working on natural genetics have sort of worked hard on understanding how do mutations occur in the genomes of organisms. And how are these mutations fixed or repaired? And through all of this research, we know that there are a bunch of specific genes that typically encode a set of enzymes, some of them are organized into different pathways. And these enzymes effectively go and scan for where there's breaks in the DNA, right, due to, for example, UV damage, C bases can become T bases. And so that leads to a sort of a chink in your DNA, right? And that can be structurally recognized by these enzymes. Our polymerases, which are the big enzymes that replicate our DNA, also recognize these problems in the DNA structure. And so they will get stalled if the breaks are not repaired, if there's a break in the DNA. So all these surveillance and repair systems basically go in and they identify when there are mismatches, where an A is pairing with a G, for example, and they will fix it, then. They will remove the offending base, and then put in the right base, and then things can go on smoothly. So this is effectively the process by which mutations occur in our genome. And these mechanisms to repair mutations are, of course, not 100% accurate. And they also don't manage to catch every single mutation that does occur, or fix it in an error free way. So some of the enzymes that do the repair work are also a bit shoddy at their job. And then the different enzymes also have different preferences or tendencies to recognize certain kinds of mutations more easily. So this MutY gene that you refer to, which we used in this paper, the enzyme encoded by the MutY gene, that's the enzyme that goes and recognizes A and G mismatches in the DNA. And it also recognizes mismatches caused by oxidation of the G base, so it's called 8-Oxo-G. And then that weird oxidized G can pair with things that it's not supposed to pair with, and that causes problems. And so it goes and fixes those kinds of mutations. So one easy way to change the mutation biases in an organism is to delete or, you know, cause the loss of function of some of these repair enzymes, because they have a tendency to repair specific kinds of problems or mutations, if you get rid of them, then those problems are going to persist, right? And so what happens then is you typically will have an increase in mutation rate if you delete these guys, and make hyper-mutators, but you also will have a shift in the mutation bias. And what we were able to do is capitalize on this change in the mutation bias, focusing not only on the rate, but the bias change. And I think one of the reasons why we found something that people had not found before is because we looked at the mutation bias whereas a lot of previous work has focused on the mutation rate aspect of the mutaters.
Cameron Ghalambor 18:10
Yeah, so just to kind of follow up on that. So if I kind of follow the, you know, the way these kinds of experiments are historically done, you know, I'm familiar with these kind of mutation accumulation experiments. But in your PNAS paper, if I understand correctly, what you did was you set up these strains, and then you kind of like, captured just a single mutation, while controlling for the rest of the genetic background. And so I was as kind of curious, how do you do that exactly? Because, especially in these MutY strains, you've knocked out the repair mechanism. So it would seem like they would just be accumulating mutations very quickly. And so can you explain, just like, logistically how, how you're able to get a single mutation like, isolated in these different lines?
Deepa Agashe 19:05
Yeah, sure. So some of it is just brute force, and maybe I'll explain a bit. So what we do and you're right, absolutely, when we delete the MutY gene, we have a tenfold increase in mutation rate. This is known from previous work from the bacterial geneticists, like I said. And so what we do is, each of these lines, you propagate every 24 hours, we take a random colony, and then we choose it from the agar plate, and we allow it to propagate, right? And so what happens is, if there's been a mutation at the beginning of that colony, whatever cells we pick up, and whatever cells seed our next randomly picked colony, that mutation is going to get propagated, and it gets to accumulate. So in the wild type E, Coli, we have to do this process of daily bottlenecking, single cell bottlenecking, for at least two months or so to get a single mutation to show up on average in a lineage. It's a Poisson process, and so we can't predict it exactly, right? But that's the we have some ballpark estimate of if I now sample lineages at a particular time point after certain number of transfers, I expect to see one mutation in at least some reasonable number of them. And then that time point just occurs much earlier for the MutY strains, right, because the mutation rate is much higher. But the mutation rates still for the MutY strain is not so high that we cannot do it. And so the MutY, I think, took a few days time to get to the point of one mutation. And then the brute force part comes in. So then my PhD student, Mrudula, she's one who did all of this crazy work. So she picked up all these colonies, and we sequenced about 300 of these lineages, just whole genome sequencing. And then identified mutants that had a single change. There were, of course, many others that had two, and some three, some had none, by chance, those were all put away in the freezer to be used for something else. And then only the ones which had a single mutation were used for the subsequent experiments. And we got about a hundred such mutants from sequencing three hundred.
Art Woods 20:59
And then how do you know, during the subsequent experiments, that they're not accumulating mutations during the course of the experiment? Right, so like, it seems like the ones with the MutY genotype, they would be accumulating mutations faster during the course of any sort of physiological measurements of things like fitness or growth?
Deepa Agashe 21:23
Yeah, that's a good question. We were worried about that for sure. But like I said, MutY the mutation rate is still not high enough to expect that early on during our roughly 12 hour assays that we do growth rate assays for, we don't expect to see a mutation, so I'm sure there are mutations in some fraction of the population by the end of the experiment. But because they didn't occur early on, they would not influence the growth rate measurements that are the main thing we're measuring
Art Woods 21:49
From the population altogether. Yeah, I see.
Deepa Agashe 21:51
Correct? Yeah, yeah. This would ben a problem for other mutators, which would have a super high mutation rate, like we know, some that have 1000 fold higher mutation rate compared to E Coli. Those would be hard to measure.
Cameron Ghalambor 22:03
Yeah. And just another kind of follow up question is, so you knock out this seemingly very important gene that, you know, is involved in DNA repair. And you mentioned that it sort of focuses on a specific type of mutation. But are there other repair mechanisms that are sort of acting compensatory to sort of like, take up a little bit of the slack when MutY gets knocked out?
Deepa Agashe 22:30
Yeah, so I don't know that I should say compensatory, necessarily. But there is another one, another enzyme, that works on same pathway, it's called MutT. And so it kind of goes and mops up 8-Oxo-G that's floating around. So it does something the same pathway, but it's not able to, like really compensate for the lack of Y. And then on the other side, there are other systems, the most famous of these is the mismatch repair system, which has its own proclivities for fixing other kinds of mutations. And so if you knock out each one of these different enzymes, you're going to end up with different sorts of mutation biases, which is very convenient for us, but, but they don't all compensate for each other.
Art Woods 23:23
So turning to some of the results, I think the maybe the first thing that struck me and Cam was how many of these mutations that you captured turned out to be beneficial. So in the mutator line, it was something like 35, or 36% of the mutations were beneficial. And in the wild type, it was still up around 28 or 30%. And by beneficial, I think you mean like their growth rates are a little bit higher than the wild type, or their fitness is slightly higher in a measurable way. This just seems surprisingly high , I think because I have this image of you know, mutations mostly being deleterious. So why are there so many beneficial mutations happening?
Deepa Agashe 24:04
I don't have a great answer for this. I can only speculate a little bit. So initially, we were very worried about this and like yourself, you know, several other people that I showed these results, to they're like, "Wait, what, why do you have these insane number of beneficial mutations?" So the problem I was facing is that very few people have actually measured fitness the same way we've measured to construct these distributions of fitness effects. The best comparisons are available with viruses where, you know, people were able to make tumorigenesis and through fitness assays with hosts infectivity in plants. So there they had these sort of similar kinds of measurements. But then viral genomes are also strange in their own ways, and so it's a bit hard to compare. Basically, there was no other comparison that I could really work with.
Deepa Agashe 24:49
So the way that we have largely estimated what is the distribution of fitness effects, the statistical distribution, and from that infer what fraction of mutations are deleterious or beneficial is many times, or I would say most of the times nowadays, it's something that's estimated from data on polymorphisms in populations, right? So you go out and ask, what is the allele frequency distribution, for example, of different kinds of alleles in a population? And then you use that distribution, given that mutation has happened, selection has happened, recombination has happened. And then you use that spectrum, the site frequency spectrum, to estimate something about the DFE, the distribution of fitness effects. And experimental data of the sort that we have, you have a set of mutations, you've measured single mutations, that is effects very, very rare. In many other cases, previously, when people have measured DFEs, they've measured them on, for example, the mutation accumulation lines that have gone on for some time, but accumulated an unknown number of mutations, right? And so then epistasis is kicks in across different mutations, and you don't quite know what's the fitness effect of a single mutation. So we struggled with this a little bit, because there were very few comparison points that made sense. We did also control for biases that might arise during the mutation accumulation experiment, by a correction factor. And this is something that I worked with a collaborator, Lindi Wahl, was really important. She's a mathematician, and, and she basically said, "Well, during colony growth, when we do the transfer from one day to the other, there's still about 20 generations that occur, you know, when the colony has to grow up large enough for us to see it, and then work with it. Is it possible there's some selection that's going on in that timeframe?" And so she did some simulations, did some modeling, and we came to the conclusion that, indeed, there is some selection. And so that whole distribution should be a little bit left shifted, compared to what we are seeing it. And so we corrected for that, but we still have the numbers that you mentioned earlier, right? We still have way too many.
Art Woods 26:51
And just so I understand this sort of process of selection you just mentioned. So you mean that you're selecting for slightly positive mutants, because single cells that contain those are likely to grow slightly faster, and then be slightly more likely to be chosen when you're choosing a random colony? Is that the selection process you're referring to?
Deepa Agashe 27:09
Exactly, exactly.
Art Woods 27:11
Got it
Deepa Agashe 27:11
Precisely, yeah. So then we correct for that, not in terms of the selection coefficients that we measure using growth rate, but in terms of the entire distribution, right? So we can say, if we can bin these mutations into highly beneficial, you know, and as opposed to the relative frequencies in the DFE, we can now say, okay, the ones that were really high frequency, you know, high beneficial mutations, probably the actual frequencies lower. And so we remove some weight from that bin, and we add some weight to the deleterious lot that we measure, which we are almost certainly, you know, less likely to pick because of the selective process, that you mentioned, in colony growth. The one last thing I want to say about this, why are there so many beneficial mutations? Well, two last things.
Deepa Agashe 27:51
One is that we're actually not the only ones finding this. So a bunch of other people studies with yeast, in particular, earlier have found fairly high proportions of beneficial mutations in mutation accumulation studies like what we've done. And again, I think it's not clear why. But I was happy to hear that it wasn't just us having weird results. And those were, again, were kind of unexplained. In arabidopsis mutation accumulation lines also, as early as the late 90s, Ruth Shaw and others had found results that were consistent with a much higher fraction of beneficial mutations than you'd expect, based on what we sort of commonly think of.
Deepa Agashe 28:30
And then, finally, this is something that a reviewer suggested for this paper, maybe what's going on is the way we're measuring fitness through measures of exponential growth rate, maybe that is a much more spread out distribution, right? So we're getting not just more beneficial, but also more deleterious mutations in general. And if we measured fitness using competitive growth, or some other metric, perhaps that would compress the DFE a little bit more. Or another way to think about it is perhaps that the measure of fitness we have is very short term, it's rather immediate, like, you know, how fast can you grow right now? Compounded over some several generations, maybe that fitness metric is not as large, the values are not as large. So we haven't been able to test this comprehensively yet. So in the end, I don't have a good answer for you. But we've replicated this.
Art Woods 29:22
Yeah yeah. Sounds very compelling. I guess that's one of the great advantages of using E. coli, right. If you want to go back and measure fitness other ways on the same strain, the same clones, then you take them out of the freezer and do it right. That seems like a tremendous advantage.
Deepa Agashe 29:37
Yeah. Yeah. It is a wonderful to be able to do that. Yes.
Cameron Ghalambor 29:41
So, Deepa. I'm curious, when you look at these beneficial mutations, I'm assuming you know where they are in the genome. Do they tend to be distributed across many different genes? Or are they do they tend to be for example, more common in like particular pathways because you are measuring them on these kind of growth medium. And so do you see any kind of pattern in the places where they tend to end up? Or is it just kind of random?
Deepa Agashe 30:16
I think it's fairly random. But I would also say that the number of mutations we get in our studies are very small compared to the total genome size, right? So there's 10 to the six base pairs in E. coli four times, that actually. And so at most, the number of mutations we're getting is, in any one genome, is a few dozens, no more. And so it's a bit hard to say whether that distribution is truly random. What we don't see for sure is whether they're all clustered in certain types of genes, types of pathways. Other datasets by others with E. coli, with which they did mutation accumulation lines for longer, and they did sequencing at the end have reported this wave like pattern. This is also seen in many other bacterial genomes, where mutations seem to have peaks and troughs, but they don't seem to be connected with anything functional that one can see, obviously,
Art Woods 31:05
You mean spatially along the genome, they occur in peaks and troughs?
Deepa Agashe 31:09
Yeah, along the genome.
Art Woods 31:10
Huh
Deepa Agashe 31:12
Anyway, but so yeah, so what we did do is ask whether, you know, the beneficial mutations were all clustering in certain types of genes or regions. And we don't see any obvious signature like that. We also asked the beneficial mutations were more likely to be in gene bodies, and so in the coding regions, rather than non coding, or vice versa, again, no obvious pattern. Where deleterious or beneficial mutations are more likely to be synonymous, no pattern. So we tried all of these different, you know, ways to understand what was going on, and really nothing else seems to matter, except for the transition transversion bias.
Art Woods 31:47
Let's shift over to another, I think, really interesting result that you guys got. So you grew these strains across each of 16 environments, and then used that design to ask a really interesting question, which is, how correlated are the fitness effects across the different environments? And I would just say, you know, for me, and I think maybe you guys had this expectation, too, that there might be a lot of what they could characterize as antagonistic pleiotropy. Meaning that if you see increases in fitness in one environment, you see offsetting decreases in another environment. And in fact, you didn't see that, right, you saw mostly when there were positive effects in one environment, they were also tended to be positive in the other environment. So I guess, a two part question, why isn't there more antagonistic pleiotropy? And then the converse, why are there so many positive effects across environment?
Deepa Agashe 32:42
Unfortunately, again, I don't have a very good answer for you. I can speculate. So one way to address this is to look at what the mutations are. For example, one simple hypothesis might be all these mutations that are beneficial in many different environments, are hitting some global regulator of some sort. Or they're sort of stopping the production of something really costly for growth. And so if you don't stop stuff, which is not really relevant for growth in all these different substrates, maybe that's beneficial, and that's what's going on. So we didn't see anything obvious like that. But I should say, we haven't examined this in terrible amounts of detail yet. I don't know whether that's the right hypothesis to explain what's going on yet or not. Another way to think about it is to say, well maybe many of these mutations are generally sort of functionally stopping things from happening, right? So stopping expression of stuff which is not useful, in general, when this media that's provided is not a very good medium for growth, when fast growth is not possible. And therefore, it's beneficial in general, regardless of what kind of carbon source you're growing on, to have all this useless stuff shut down, and not have to pay the cost of expressing all these other genes that are kind of useless. And maybe one thing I would say to help understand this a bit better, is a lot of research on the E. coli genome and transcription, translation kinds of processes have shown has shown over the years that E. coli is one of these organisms that's really, really geared to grow fast, you know, it's got its translation machinery is such that it's got gobs of, you know, copies of the genes that are important for making ribosomes, for making tRNAs that will bring amino acids during translation. And my lab has done some of this work as well recently experimentally showing this. And so it really looks like everything E. coli is doing, it's been optimized for really rapid growth. And as a result, it's pretty wasteful in many ways, you know, because the benefit of having everything ready to go is so high, that that's what it's supposed to do. And so maybe the results we see are peculiar to E. coli, potentially, and may not be the case in something else that is a slow grower, that's how it's been, you know, that's the selection it's experienced over its evolutionary history. And there maybe doesn't have as much sort of excessive wasteful expression of things. And so they're, you know, random mutations may not have as much beneficial pleiotropic effects as we observe. I don't know what's the case.
Cameron Ghalambor 35:12
Yeah, that's super interesting. I hadn't really thought about, you know, the sort of life history of E. Coli in that context. And that's a kind of an interesting thing to think about. I was also really surprised by these results, and also had this expectation of seeing more evidence for tradeoffs and antagonistic pleiotropy. And I guess one thought that I had was your different environments are different kinds of substrates. And so you know, it's possible that, you know, the one reason why maybe you see beneficial effects across environments is because these are all different substrates, you know, you're still dealing with maybe the same kinds of pathways that are involved. But like, I was thinking in my mind, like, well, what if you had crossed malate, with like, high temperature or low temperature and the environments became a little bit more complex? Would you then be more likely to see more evidence for kind of the mutations being more locally adapted, and seeing more trade offs across environments? I don't know, what do you think about that?
Deepa Agashe 36:19
Yeah, it's entirely possible. We didn't manage to test that particular idea, but the one thing I would say is, I hesitate a little bit to go very far out of some kind of space of what we're dealing with when we're thinking of tradeoffs, because I feel like tradeoffs, some perhaps, meaningful for organisms when they're moving along some axis, which is environmentally relevant. So for example, a mutation that's beneficial under, I don't know, galactose growth, right? Whether it is beneficial or deleterious as a function of temperature, may not be as relevant or as important to understand as whether it's important for amino acid metabolism or some other sugar and so on. Because I don't know if it's clear what I'm trying to say, but basically, I think there are several different axes of the environmental input that would be relevant. And I feel like across different axes, while you might see more pleiotropic, antagonistic pleiotropic effects, I don't know how to interpret them. Whereas I think along this a similar-ish axis of carbon use or carbon resources, it's a little bit relevant or easy to think about saying, okay, as the organism experiences different environments, it still always has to have a way to metabolize carbon, take it in, use it. And so I feel like that gives me a little bit more surety that I'm understanding something about what these tradeoffs may actually mean, during evolution. Yeah, I mean, I think your question is fair, but I don't have an answer for it.
Cameron Ghalambor 37:51
Yeah, I guess I've been thinking about this in my own research in the context of kind of like, benign versus like stressful environments. And it seems that there often are tradeoffs between being able to thrive in a benign environment. And that usually means that you're a good competitor, and you can exclude, you know, sort of poor competitors. But then, in more stressful environments, having tolerance to those kind of stressful conditions seems to come at the expense of being a good competitor. And so that's one way that you can also promote, like, niche diversification and have like, local adaptation across these different environments. So I was looking at these results, and like thinking like, oh, okay, you know, so how might these fit in with, you know, some of the systems that I'm familiar with?
Deepa Agashe 38:45
Yeah, can I say something responding to that? So it bothered me a lot, by the way, not finding so much antagonistic pleiotropy. So this is, you know, this is something we also reported in a different paper in 2018, from the same PhD student, Mrudula. And it bothered me then as well, you know, what gives? Because tradeoffs are everywhere, you know, whenever people have measured tradeoffs, they're there. And so I was reading more about this, and I found that actually my postdoc advisor, Chris Marx, and a PhD student in his lab that I briefly overlapped with when I was there, they had published a paper which showed nice evidence with E. coli with the Lenski long term evolved lines, showing that actually mutation accumulation, as the alternative explanation for how trade offs occur, seems rather more supportive than antagonistic pleiotropy. So maybe I'll explain that slightly. And I suspect that's what's going on and in many other systems. So what they found was that if you looked at some of the hypermutaters that arose in the Lenski populations, they tended to have bigger fitness declines, over time, compared to populations that never experienced any hypermutators. And the idea then is that as you accumulate more and more and more mutations, you're more likely to hit a mutation that's going to be deleterious in some other environment, but was either neutral, or maybe mildly beneficial in this current environment, right? And so it's not that single mutations have these two opposite effects, but that the accumulation of mutations over time is what's relevant and important. And that leads to tradeoffs, right, as a byproduct, not because of selection directly, as a byproduct. And I find that really attractive. And so we plan to test this using the, you know, all these waste mutants that we had with multiple different mutations accumulated in our main lines, hoping to use those to directly test this hypothesis.
Deepa Agashe 40:40
But what was really interesting for me was in the phenotypic assays with the flour beetles that we had done, I told you about, where these are, presumably just very recently diverged populations, they show phenotypic diversity, very little genetic diversity. But we asked whether their performance on different flour resources showed any tradeoffs, right? And we don't find any tradeoffs. If there are beetle populations that are great at eating wheat and have high fitness on wheat, the same population has higher fitness on all the other resources, we tested them on, right? Everything from corn, which is terrible for them to other millets, which are used in India, which they're okay with, right. And we see this kind of lack of trade off across 20 different populations that we tested in our lab. And this was again, something which I never published yet. By the way, I really should do it. But I sat on it for a while thinking, what do I do with these results exactly? I expected to find tradeoffs, there are no tradeoffs. And so my suspicion is, if you look at populations that have diverged very recently, you will not find tradeoffs as often because the number of mutations is not very high. If you allow populations to diverge a lot more, I suspect you will find many more tradeoffs.
Art Woods 41:52
Yeah, that's super interesting.
Cameron Ghalambor 41:53
Yeah so it's very parallel to the kind of arguments that have been made for the evolution of senescence. And very much, you know, the debate about whether the evolution of senescence arises because of mutation accumulation or antagonistic pleiotropy. And it can be very difficult to tease those two hypotheses apart, selection always becomes weaker, you know, for any of these mutations that have effects late in life. And selection's obviously weaker in its ability to remove them. And so, yeah, you're kind of making a good case for maybe mutation accumulation over antagonistic plateau. So that's cool, yeah.
Art Woods 42:48
I wanted to push
Art Woods 42:50
back to another aspect of your experiment that I sort of said in passing maybe a half an hour ago, but I want to dig in a little bit more to this particular effect that came out of your experiments. And that is that the MutY line, so the mutation bias, or the higher mutation rate lines, and those mutations that arose in those lines, were statistically more likely to be beneficial mutations, right, than the mutations that were in the control line. And so something about the shift in mutation bias also produced a greater fraction of beneficial mutations. So why is that that seems really key to, for us to understand that.
Deepa Agashe 43:38
Yeah, we struggled with this a lot initially, because it didn't make sense. So and this is where, again, Lindi Wahl collaborator came in. So I think she had an insight of the sort of theory people seem to get once in a while, because they can zoom out of the details of the experiment beautifully. And then she did simulations to sort of test her hunch, and turns out, that's what's happening. So what Lindi showed is that, through these simulations, anytime a population has evolved for a long time, imagine the same fitness landscape, right, same environment. It has, as time goes by, we know that it will accumulate some beneficial mutations, these will get fixed. And the deleterious mutations will, with some success, be removed from the population over time. So the deleterious mutations can come back again, right, because they've been removed, they can be sampled again. But the beneficial mutations, once they're sampled and fixed, are kind of no longer available, they're taken out of the available pool of beneficial mutations that existed at the beginning of adaptation. And so, for example, as predicted by Fisher's geometric model and other other ways to think about adaptation, as you go along an adaptive walk, populations generally have depletion of available beneficial mutations, so the fraction of beneficial mutations decreases as you go closer and closer to your fitness peak. And so what Lindi realized and showed through the simulations was that if now, you have been sampling a particular type of mutation much more frequently over this adaptive walk, that depletion is irreversible in the fraction of beneficial mutations that you're sampling. So imagine a population that's been sampling largely transition mutations, right? Just type A, it doesn't matter what you call it. And then over time, the mutations of type A that are available, that are beneficial, keeps reducing, keeps reducing. Now, if this population suddenly switches to sampling many more mutations of type B, that has a much bigger pool of available beneficial mutations that it simply hasn't sampled before. And so that is key, I think, to understanding our results that when MutY strain, the MutY strain that we made, started sampling, about 95% transversion mutations compared to E. Coli's bias, which allows it to sample only about 40% transversion mutations, 45% transmission mutations, there was a huge mutation space opened up to these populations, and all the beneficial mutations that were always available, simply now become accessible to the population. And that is what explains why we see this big increase in the beneficial fraction when we change the mutation bias to be opposite of what it was before. And that opposition of the old mutation bias is key to understanding these results, and to the predictions that we make in this paper.
Cameron Ghalambor 46:34
Yeah, so if I could just kind of like, make sure that I understand that correctly. So it isn't necessarily that there's anything special about those mutations, it's simply just the case that when you shift from the prevailing bias towards an opposite bias, you now open up this new pool of mutations, and can sample them where you previously weren't able to. And so they're not more enriched necessarily for being beneficial, it's just that you have a new pool that you can sample from, is that correct?
Deepa Agashe 47:12
Absolutely. Perfect. Yes.
Cameron Ghalambor 47:14
Okay. That's really interesting. And I think it really changes the way that, at least for me, personally, the way that I think about how mutations arise and how they get picked up and sampled during the process of evolution, it's a kind of a game changing way for me to think about things. Yeah,
Deepa Agashe 47:37
I can't tell you how excited I am about this discovery.
Art Woods 47:41
Yeah, I had a question too, about the sort of generalizability of this. So you know, this has to do with particular shifts in biases among transitions and transversions. Do you think is it possible to generalize this to sort of other kinds of mutations? So anytime a lineage evolves a new bias, even if it involves, like bigger chunks of DNA that opens up a different set of possible beneficial and deleterious mutations?
Deepa Agashe 48:08
Yes, I think so. I think so we've been thinking a bit about this. In the paper, we did show with simulations that the same effect will be valid for GC to AT biases. So you know, if you've been sampling largely mutations that lead to an increase in the GC content of the genome, then a switch to a mutation bias that leads to an increase in the AT content should have a similar beneficial effect and vice versa. So yes, we think it should be a general effect. But I do think it's important and interesting to think about what are meaningful axes in which bias can be split, right? And for me, biologically, I think it's much more meaningful to think of biases of the sort that are GC, AT and transition-transversions, because there are clear molecular mechanisms, which can mediate those shifts, right? So you know, DNA repair enzymes, for example, also alter the GC AT bias. And we know that these enzymes are often gained or lost in the phylogeny. This was shown by our work as well as work of many other people before for specific enzymes. We know that these are also horizontally transferred in bacteria very frequently. And so there are clear mechanisms in which you can have a quick and rapid change in the mutation bias along these axes. For other axes, such as synonymous or nonsynonymous, or, you know, indels versus SNPs, I'm not so sure that this is a meaningful way to think about it, because there isn't a clear molecular mechanism that distinguishes these kinds of- maybe for SNPs and structural variation, actually, there is a mechanism one can think about because structural variants are typically, actually structural variants are also a huge class of sets of mutations driven by many, many different molecular mechanisms. And so it's possible that those are meaningful to think about. I think we have so little data though on the, you know, on the DFEs of the structural variants, I think there are just a few papers coming out now that show these kinds of effects. I think it's fascinating to look at whether shifts in these will also have these roles, or these impacts. But, in principle, I think, yes, it should be generalizable to any bias.
Cameron Ghalambor 50:17
Yeah. And so you kind of tested this idea with a sort of phylogenetic analysis. And, and you found that when you compare across sort of a diversity of different strains of E. coli, these types of shifts seem to be very common. And so it made me think whether what the time course of this is and how dynamic is this process? Because it would imply that shifting your mutation bias has clear advantages, because it opens up this pool of mutations that you couldn't sample before. But do you think that this is kind of like something that goes back and forth like you, you have one mutation bias, you deplete the mutations that are available, and then it creates an opportunity for a huge advantage for a mutation that allows for a shift in the bias in the opposite direction? And so I can imagine this just kind of like back and forth sort of process kind of happening. Is that sort of a result that maybe comes out? Or is that extending it maybe too far?
Deepa Agashe 51:27
No, I actually, I think you're right. That's how we've been thinking about it. And then I will maybe add on some layers of complexity to it, which is why I think it won't actually work as nicely and as beautifully as the intuition suggests.
Art Woods 51:39
Damn.
Deepa Agashe 51:41
Just one quick correction. So the phylogenetic work that we did was actually not in E. coli strains, but across many different species of bacteria. And so it actually spans millions of years of divergence. And we cannot put exact timescales along which these processes happen, because unfortunately, as I'm sure many of the people listening to this will know, you don't know where a longer branch, a particular gain or loss event happened, right. So we don't have a way to date that. All we can say is at the nodes, what differences might be expected and what's going on. But yeah, so I think the reason why these kinds of swinging back and forth across alternate mutation biases would be complicated, or would not probably occur in real life is that the environment keeps changing as well, right? So the minute the environment changes, your fitness landscape is now different. And so you know, your existing mutation bias also has a bunch of beneficial mutations that become newly available. So what is the rate at which these back and forth can happen? Would be interesting. I think that's something that we should explore from your from a theory or simulation perspective. Experimentally, I think it's very difficult to do. But I think these are all very interesting questions for a lot of future work.
Art Woods 52:55
You're saying, in effect, you know, that we don't really know, either theoretically or experimentally, I hear you saying that. But if you had to say, just intuitively, is this something that happens on the timescales of, you know, tens to hundreds or thousands of generations? Or is it something that plays out over, you know, millions of generations over millions of years? Like, very long term, or very high frequency? Is it possible to distinguish those things right now? And the answer may well be just no.
Deepa Agashe 53:23
Yeah, I guess the answer would have to be nuanced like I don't think it's, I can give a meaningful answer simply because it depends. So this whole effect that we're talking about really depends on depletion of beneficial mutations, right. And so the timescale at which that happens, is itself dependent on the population size, environmental fluctuations, you know, yada, yada. So it's hard to give one answer. And with bacteria, it's further complicated by horizontal gene transfer. And so at sort of tiny, relatively tiny, timescales, you know, million years give or take across strains, people have shown horizontal transfer across bacterial lineages of these DNA repair genes. Across millions of years timescales, we have been able to say, okay, there are gains and losses clearly that have happened across bacteria lineages. In the scale of experimental evolution, so in the Lenski lines for example, people have observed mutaters arising, you know, several times. Whether they have been back and forth mutations that have led to different biases, nobody has examined that directly. What is clear is even in the timescale of 60,0000, 70,000 generations of the Lenski lines evolving. Rohan Maddamsetti and others have shown recently that there were a lot of mutations that arose in these lineages, which could have led to changes in mutation biases. Whether they did, how long did they stay, we haven't done a systematic analysis of that yet. But it would be really fun to do.
Cameron Ghalambor 54:52
Yeah, so I'd like to follow up on that because my impression from the Lenski experiments is that the frequency at which a beneficial mutation arises seems to be kind of rare. At least, that's my impression. It's, you know, thousands or tens of thousands of generations before there's a change . But then your results would suggest that maybe, you know, these beneficial mutations should arise more frequently. And so how do we reconcile those two kinds of perspectives? Are they in opposition? Or are they complementary?
Deepa Agashe 55:28
That's a great question. So I have two ways to think about this. One is that what we are measuring as fitness, like beneficial mutations, are basically fast growth, right? That's what we're measuring. But in the Lenski populations, as well as, I guess, in nature, it's not only important how fast you grow, but also how you survive in the lean periods, right? So in the lens, key lines, transfer happens every 24 hours, for example. And they definitely spend a big chunk of that time in stationary phase, which is where nutrients are kind of depleted, things are not looking as great. So it's not as important necessarily, you know, whether you grow fast, but also, can you make it through that time? And so that's one perspective, that the way we're measuring beneficial mutations is one way to measure it, it's not the only thing that matters in terms of fitness in the real world.
Deepa Agashe 56:14
The other perspective is that what we measure in our DFEs are the instantaneous availability of what's beneficial, right? But the fixation probability of that beneficial mutation depends on the population size, other beneficial mutations that arose at the same time. Were there other hitchhikers that were deleterious, etcetera, etcetera, epistasis, right? So a lot of other things are happening that determine the fate of any particular mutation over time. And so even in the Lenski populations, my guess is what we are measuring, as our DFEs that say, how many beneficial mutations are available at this point of time, that's probably very large, but what actually fixes is a tiny proportion of those beneficial mutations. Because there's a lot of reasons why a great mutation doesn't actually make it, right? So a great mutation has to first arise. Okay, fine. We say it could arise as a pretty high probability. But then it has to make it. And that's very hard. And so I think that's the dichotomy there.
Cameron Ghalambor 57:18
Okay. Yeah. No, that's super helpful
Art Woods 57:18
I wanted to ask, we're starting to get toward the end, but I have one more sort of broadly theoretical question, which is about, you know, what your results and other people's results, say about sources of genetic variation that underlie evolutionary change. And it feels like your results as a whole suggests that the origin of new mutations is really important. And, of course, the sort of converse of that is that many populations contain standing genetic variation that exists, and then over some long period of time it gets drawn on under the right circumstances, the right selective environments. And it feels to me like there might be a difference between, say, unicellular populations that have very fast growth rates and potentially very large population sizes, versus more complex multicellular organisms, plants and animals that have maybe more complex phenotypes, and that are perhaps underlain by more complex forms of genetic variation that can accumulate as a kind of standing genetic variation. So I guess this is a long winded way of saying, What's your sense of how important are, you know, new genetic variants versus standing genetic variants and how does that depend on the context in which evolution is happening?
Deepa Agashe 58:36
Great question. I seem to have no answers for most of your questions.
Art Woods 58:39
Well this great, you know, let's discuss.
Deepa Agashe 58:43
Yeah, so I think you're right, you know, the kind of context in which these standing variants are segregating versus new mutations and how important each of those are, but we don't have enough data to be able to compare honestly, right? And so we sort of struggle with this a bit as we were writing this perspective, or introductory article for American Naturalist last year, trying to ask, what do we know about the rule of standing genetic variation versus new mutations? You know, which one is important? And really, that comparison is so hard to make, because in the systems where, like wild, natural populations of animals or plants, or where sending genetic variation is typically measured, and people work with those to ask what's happening with adaptation. Typically sending genetic variation is found to be relevant, much more relevant than new mutations. And the mutation rates are not as high, population sizes are not as high, and so on, so forth. But the problem is, there is no comparison for de novo mutations, right? Because the mutation rates and the sizes of the samples that we can do with these kinds of organisms is just nowhere going to match up. And so I know of almost no studies where people have tested this or been able to test it. Maybe something like yeast is where it's possible to test it, because the generation time is large enough mutation rates are not insanely low that you can sample new mutations, and you can have standing genetic variation that you allow population to have. I think a recent paper a couple of years ago has tried to do this a little bit in MBE. But I don't remember right now what results they found. But there are almost no papers or no studies that have compared this. And so I think the data are a bit skewed, in what kind of variation is observed or easy to observe in which kind of system. And so making cross system comparisons is very hard.
Cameron Ghalambor 1:00:32
Yeah. So, Deepa, one big question I have for you that we've talked a little bit about this interplay between selection and mutation bias. And as I was reading your work, I was thinking about this paper by Grey Monroe and colleagues in Nature from a few years ago, where they found that, at least in arabidopsis, mutations were much less common in coding regions and sort of functionally important genes. And that seems to make a lot of intuitive sense that, you know, these are important genes that have important functions, and you don't want mutations landing inside of them and messing things up. But then, you know, then I'm looking at your E. coli results, and I'm seeing wow, you know, a lot of these beneficial mutations are getting picked up and leading to these positive fitness effects. And so how should I think about the this tension between, on one hand sort of like protecting, you know, important genes for mutations disrupting function, versus the potential, you know, for benefits of new mutations coming, and being part of this adaptive walk, you know, to higher fitness? What would you tell me about how to think about the two sides of that coin?
Deepa Agashe 1:01:49
Yeah, so there's certainly a genetic load associated with sampling lots of mutations, because sure, a bunch may be beneficial, but a bunch are also deleterious. And that's something that's problematic for thinking of essential genes, for example. One way, I've seen some examples of this problem being solved, in some senses, different domains of the same protein encoded by the same gene sometimes evolve at different rates, right? So some domains appear to have many more polymorphisms across strains, or species or lineages, whereas other domains are super conserved, and they will not show any polymorphism. The problem here is it's always difficult to know how much of that difference in polymorphism that we observe is mutation driven, or selection driven, or some combination of both? And I think that's part of the issue in the Monroe paper is that I think they've tried to understand how much of this is due to mutation versus selection, and they've suggested that it's largely mutation that's driving these patterns. And that's consistent, by the way, with a lot of data that's observed in bacteria, yest, a bunch of other organisms before as well, that it does seem like mutation rates are suppressed in certain parts of the genome. Is that something that has evolved for its evolutionary advantages? Or is that a byproduct of the fact that, for example, several highly expressed genes tend to be, you know, opened up for transcription, very often transcription is mutagenic? So there's some correlation there. Or other kinds of factors in eukaryotes like histones are the modifiers of the genome that protect the genome, in some cases, from mutations or allow mutations to occur? I think it's very complicated. And I have colleagues at my institute, for example, who've been asking or looking into cancer genomes, just because the datasets are so incredibly rich, to ask these kinds of questions, you know, which regions are protected from mutations? Which reasons are not protected from mutations? And why, what's the molecular basis? And what I gather from all this is it's a nightmare, like there are so many of these proteins that hang out near DNA, sometimes all the time, sometimes occasionally. Anyway, that's a long winded answer to say that basically, I think we don't understand everything yet.
Cameron Ghalambor 1:04:06
It does definitely seem complicated. And I don't know, maybe potentially, to some degree context dependent on the environment that you're looking at. But these are like really big questions and important ones for sure. Yeah.
Art Woods 1:04:22
Well, Deepa Agashe, thanks so much for coming on to Big Biology. It's been a fantastic conversation.
Cameron Ghalambor 1:04:28
Thank you so much, yeah
Deepa Agashe 1:04:28
Thank you very much. I enjoyed it a lot.
Cameron Ghalambor 1:04:51
Thanks for listening to this episode. If you like what you hear, let us know via Twitter/X, Facebook, Instagram, or leave a review wherever you get your podcasts. And if you don't, well, we'd love to know that too. All feedback is good feedback.
Art Woods 1:05:05
Thanks to Steve Lane who manages the website and Molly Magid for producing the episode.
Cameron Ghalambor 1:05:10
Thanks also to Dayna De La Cruz for her social media outreach and Keating Shahmehri for producing our awesome cover art.
Art Woods 1:05:17
Thanks also to the College of Public Health at the University of South Florida and the National Science Foundation for support.
Cameron Ghalambor 1:05:23
Music on the episode is from Podington Bear and Tieren Costello