Podcast: Curiosity Unbounded, Episode 3 — Decoding the tree of life

Sally Kornbluth: Hello, I'm Sally Kornbluth, president of MIT, and I'm thrilled to welcome you to this MIT community podcast, Curiosity Unbounded. In my first few months at MIT, I've been particularly inspired by talking with members of our faculty who recently earned tenure. Like their colleagues in every field here, they are pushing the boundaries of knowledge. Their passion and brilliance, their boundless curiosity, offer a wonderful glimpse of the future of MIT.

Today, my guest is Greg Fournier, associate professor of geobiology. Greg's work centers on the microbial world. He and his team study microbial evolution to further our understanding of the evolution of life on Earth. Greg, I've been looking forward to this. Thanks so much for being here.

Greg Fournier: Thank you for inviting me.

Sally Kornbluth: I heard that as a child growing up in rural Connecticut, you liked to wander in the woods with your friends and would come upon signs of life — human life from years earlier, barbed wire embedded deep within a tree, crumbling stone foundations. I understand that it was these early discoveries that helped fuel your interest in the evolution of life itself. How did that interest lead you to become a geobiologist? 

Greg Fournier: I think when you have the opportunity to really explore in the natural world, especially a place that doesn't have trails or a guide, that when you are 10 or 12 years old, as far as you know, you're the first person to ever discover it. That leads you to genuine exploration experiences. In a way, you always want to get back to that feeling. It's a bit different than the adventures you can have when there's already a trail, where there's already a mountain peak, where there's already a list of things you're supposed to see. I think I've always been drawn to that kind of exploration. The work we do — looking at the evolution of microbes and genes and genomes over billions of years — is not only new as a discipline, but it's also still a place where you can really explore. You can reconstruct the histories of these large data sets and you can discover something that no one else has really put together before. In a field that's old as evolutionary biology, it's exciting to be able to bring a new perspective and still be part of that kind of tradition.

Sally Kornbluth: Yes, I can imagine that. As a kid you don't have a sense of the long, long history of time. When you come upon things that are completely different from your day-to-day experience and you see no evidence of immediate human touch, you have that wonder of, "Wow, I'm seeing something that someone who maybe lived 100 years ago or 1,000 years ago has seen and no one's seen it since." When you think about evolutionary time, it's even more abstract, right? It's something that's happened eons ago and you're picking up the fingerprints of it.

Greg Fournier: Yes, I think that's exactly right. It's something that, even when you study this for 20 years, it's hard to wrap your head around just how much time we're talking about.

Sally Kornbluth: Exactly. I've heard kids, they'll look at some old book and say, "Wow, this book must be 10,000 years old." Not really understanding what the actual history of human time is, never mind the history of microbial time. It's really interesting to think about that.

I'm told you just came back from a geological road trip. Saw some incredible sights out west, including one called Craters of the Moon. What were some of the impressive and exciting things you saw out there?

Greg Fournier: I had a chance to do some driving around out west a few weeks ago. I went to a few national parks and got to see some really amazing geology and geomorphology. I had been to Yellowstone before so I skipped that one on this trip. But that let me go to some of these much less frequently visited parks, like Craters of the Moon. It's one of the few places in the United States where you can still see lava fields and cinder cones and this volcanic landscape. Then right around the bend, there will be grasslands and mountains and fields and deer. Some very ordinary things. The abrupt changes of the landscape you see out there are really amazing. I was able to go to a few other parks, like Wind Cave which is in the southwest corner of South Dakota. That has one of the largest cave networks in the United States. But aside from that, it also has a really great landscape of hills and fields and hiking trails. On one of the loop hikes I was planning to do, I went around a corner and there was a bison.

Sally Kornbluth: Wow.

Greg Fournier: It wasn't going anywhere. I just had a chance to have a nice staring contest with a bison for a while and then I decided to turn around and go back the way I came, which is always the right thing to do.

Sally Kornbluth: Probably wise.

Greg Fournier: Those kind of unplanned experiences or encounters were just great and the trip was full of them. It was really a nice opportunity.

Sally Kornbluth: Do you use your geologist mind when you're doing this sort of hiking? I might go out and say, "That's kind of a pretty rock." But I'm not thinking about the origins and how things got there and what sort of seismic forces might have led to the landscape. Do you reflect on these things? How do you think about some of the things that you saw in that respect?

Greg Fournier: In some ways I do. For me, especially out west, it's always interesting to think about the fact that this used to be at the bottom of an ocean. Coming back to Wind Cave, if you go into the cave network and you look on the ceiling, you can see little shells and fossils of little sea creatures just everywhere.

Sally Kornbluth: Oh, wow.

Greg Fournier: That's all in the limestone of the cave, which has been dissolved away over time leaving behind all of these structures and in some cases, you can see the fossils that were on the bottom of this ocean during the Mesozoic Era.

Sally Kornbluth: Oh, that's really cool. That's really cool. I have to go visit there. I've never been there.

Greg Fournier: You really are just surrounded by direct evidence that there's a continual set of changes happening on Earth. This was not always the top of a mountain, this was not always a plain, there were not always bison here. Intellectually, it's one thing to understand that. But to be 300 feet underground in South Dakota and say, "Oh, yes, that’s a seashell," is really an amazing experience.

Sally Kornbluth: That's really interesting because I think we tend to think of the ground we're standing on as a solid, immutable thing. Obviously, over geological time that's not true. It's really interesting to reflect on that and having really visual, clear markers of change. We don't notice it in our everyday life but the notion of seeing seashells on the ceiling of a cave is a great illustration. Some of your stories here recounting your trips, and your interest in long scales of time in evolution, bring me to ask how that relates to your own work. When you look, for instance, at microbe evolution, how does that impact our understanding of Earth's history, of mass extinctions? What do we learn from the science of microbes and the history of genomes that actually relates to the bigger picture that we see?

Greg Fournier: Even in talking about how ancient all of these geological features are, they're all quite young compared to the age of the Earth itself. Whether we're talking about the Rocky Mountains — which are relatively young in terms of mountains — or the inland sea that used to cover a lot of North America, these are all Mesozoic. When we reconstruct the evolutionary history of microbes, we can go far beyond that. In fact, we probably are reaching earlier than there are any rocks at all on the Earth. The very oldest rocks, deposited sedimentary rocks on Earth, are maybe 3.8-3.9 billion years old. The very oldest scraps of material we have are little crystals — zircons — that are a little older than 4 billion years old. We don't know when life originated on Earth, but it was very likely at least as old as 3.8 billion. When we reconstruct the relationships of genomes between all of the groups on the tree of life, we're actually creating these threads that are reaching back to even before the oldest rocks.

Sally Kornbluth: That's very interesting.

Greg Fournier: Yes. In a way, it's the oldest record we have of processes or events on the early Earth. Trying to figure out what tiny scraps of information might still be in there, and how we can use that to understand the history of life in the Earth, is really one of the most important things that I think we work on.

Sally Kornbluth: Tell me a little bit about your sort of work studying the Permian–Triassic mass extinction. I think people also have real trouble wrapping their minds around the notion of mass extinctions aside from the sort of movie hype. How do you think about this and what have you learned?

Greg Fournier: Moving forward in time, from thinking about the earliest origins of microbes, all the way into the Phanerozoic era — the "modern era" — but still before the dinosaurs, before the Mesozoic, you had one of the largest of the five mass extinctions of complex life in Earth's history. That happened at the end of the Permian, about 250 million years ago. It was definitely, by far, the largest catastrophe that complex marine and terrestrial life has ever faced on Earth. We've understood this from the paleontological record for some time. It's correlated with the emergence of large igneous provinces in Siberia. Huge lava fields that cover thousands of square miles that are associated with the tectonic events that likely caused this mass extinction. Also, we see evidence in the fossil record of the vast majority of species just vanishing. The organisms that do survive and diversify, in many cases, are quite different afterward. That's the beginning of the Mesozoic era.

Sally Kornbluth: I guess I never thought about tectonic changes as instruments of mass extinction. In other words, in the popular lore, it's always a meteor or something like that, rather than intrinsic forces, volcanic activity, etcetera. That's really interesting.

Greg Fournier: I think it's especially interesting when you realize just how large these forces are. Over tens of thousands of years, when the Earth releases a tremendous amount of volcanic material, in terms of magma and gasses and especially carbon dioxide, what you see is a radical change of the climate. Processes that we observe happening today, like ocean acidification, for example, were the main drivers of this greatest mass extinction. What our work was trying to do is examine the causes in a different way by looking at the evolutionary history of microbes that may have evolved in response to these changes.

Sally Kornbluth: If I extrapolate from that, not to be doomsday, but if we think about on a much longer timescale, the impact of climate change and changing CO₂ in the atmosphere, we could imagine some of the gradual extinctions we're seeing. If you play that out over a much longer time, it's like it was sped up with these massive events. Is that a correct way to think about it? If the answer is no, say no and I'll ask a different question.

Greg Fournier: Interestingly, it seems to be the opposite. The changes that we're seeing now are likely much faster.

Sally Kornbluth: Really? Okay. This is really interesting, and scary.

Greg Fournier: It is scary. At the same time, there's so much uncertainty. Because we understand these mass extensions looking back over millions of years at the accumulated consequences. But during the event itself, it's very hard to say what those future consequences are going to look like. We don't know how much disruption the climate or biogeochemical cycles or food webs can take before you have perhaps runaway collapses that look like a mass extinction.

Sally Kornbluth: Your comment that it's faster is based on some snapshot of time, the time we've seen in terms of species extinction versus what we can extrapolate from the fossil record. In other words, how do you compare those rates?

Greg Fournier: It's especially tricky because the fossil record is very sparse. The sampling of organisms that we see in the fossil record is biased by what tends to die in an environment that is likely to be preserved. Organisms that are the most sensitive to extinction may already be ones that have smaller population sizes, so less likely, statistically, to be preserved in the record. What's concerning is more about the rate of change in the carbon cycle. The increased rate of CO₂ production through anthropogenic processes is much faster than the modeled increase rate of CO₂ production during these mass extinction events. Still, it accumulates over a very long period of time and we've only been really doing this for 200 years. That's where the uncertainty comes in.

Sally Kornbluth: I see. That's really, really interesting. I've heard that you'd like to use the local landscape to educate your students on environmental diversity. Where does that take you here in Massachusetts? I hear there's a place called Purgatory Chasm, which I want to assure our listeners is not at MIT. Maybe you can tell me a little bit about that.

Greg Fournier: Purgatory Chasm is a rock formation in Sutton, Massachusetts, about an hour southwest of here. It is likely formed by glacial meltwater, which tore a hole through this part of the landscape and resulted in about this mile-long miniature canyon that you can hike and explore in. You can still see the scrape marks of boulders that were likely moved by these glacial meltwaters and some rock formations that just show you the force with which these very recent processes occurred that shaped our landscape here in New England. It's just a really nearby, dramatic, accessible example of those kinds of processes.

Sally Kornbluth: Oh, very cool. I'm definitely going to have to go and take a look. That sounds really interesting. Are you seeing artificial intelligence as a tool in your work at all? Does that impact any of the things that you're doing now?

Greg Fournier: Interestingly, it did for the very first time only about a month ago.

Sally Kornbluth: Oh, really? Interesting. Tell me about that.

Greg Fournier: One of the projects we've been working on for some time is reconstructing the most ancient evolutionary events within protein families that are conserved across the entire tree of life. We're specifically interested in these proteins called aminoacyl-tRNA synthetases. These enzymes are responsible for enforcing the genetic code. These proteins are found in every cell across the whole tree of life. Two main groups of them are related to each other. But if they're present in all life, it means that they must have diversified from one another before the single last common ancestor lineage of all life on Earth. When we reconstruct their relationships, we're actually reaching back to before the last common ancestor of all life on Earth, to look at the evolutionary events that were likely directly involved with establishing the rules of evolution as we understand them.

Sally Kornbluth: When you say that this divergence occurred before life as we know it, you're saying that they were evolving in cellular life forms that are no longer known to us?

Greg Fournier: In a way, yes. The way to think about it is, life evolves by this branching tree where species give rise to other species and then go extinct. If we work backwards by looking at living things that exist today, we create this tree, but we can only go as far back as the common ancestor of all of the groups that survived until the present day. We know by comparing the similarities of all of this diversity of life that this last common ancestor was already a modern-looking cell that had all of the machinery that modern cells have. But that machinery itself must have evolved from even simpler states.

Sally Kornbluth: Presumably if you compare tRNA synthetases to other molecules, other families, you can see things that also might have diverged before common ancestors, or things that had much later divergence will give you insight into the fact that those organisms must have existed.

Greg Fournier: Yes. There are similarities between very distantly related proteins in terms of how their secondary structures fold. We can see those superclasses of protein families being very diverse. That must have been established at some time. But the aminoacyl-tRNA synthetases, if we trace back to their common ancestor, it was likely involved in the same thing that synthetases are today. Which is making sure the correct amino acid goes out to tRNA so that the genetic code can be translated in a faithful manner. The difficulty with recovering this history is proteins, in a way, are like rocks. In that, over billions of years, so many processes and forces can change them that they get altered. More recent evolutionary events can overprint and replace older changes. Over time, the distance, the dissimilarity between these proteins, becomes so great that it becomes very difficult to determine what parts share a common ancestry with what other parts.

Sally Kornbluth: I see. Because it may have been  overwritten many, many times.

Greg Fournier: Exactly. Almost certainly in many cases. This is a problem that we call alignment. For more closely related proteins, or proteins that evolve very slowly, it's more or less a solved trivial problem. But for highly divergent proteins, we have to use some special tricks. One of those special tricks is to compare structure. Because the three-dimensional structure of the proteins changes much more slowly than their particular amino acid sequence. These structures are understood by X-ray crystallography. But the problem is not all proteins can be crystallized. Even for proteins that can, for example, there are hundreds of aminoacyl-tRNA synthetases proteins in protein structure databases. But if you look at them, there are regions of those proteins that are not in the crystal structure. Because they're the floppy parts, usually on the end of the protein, that probably contain a lot of information but are so variable and they don't crystallize well. Usually, you just cut those parts off.

Coming back to AI, we wanted to extract as much information as we could out of these protein sequences so that we could get a robust evolutionary signal to align and generate their deepest evolutionary relationships. If we can't do it on sequence alone, and we don't have the crystal structures because of those limitations, it turns out that there are AI-based tools now that will solve protein three-dimensional structures. We applied some of those tools to these proteins that had uncrystallizable regions. They gave us structures that were conserved across different versions we tried and allowed us to detect and extract these very divergent aligned sequence regions. In one case, one protein, it's involved in loading valine onto tRNA, we discovered that there's actually a duplication in this protein region. That duplication was confusing all the alignment programs.

In a way, we used AI to fill a gap in our understanding so that we could apply more traditional established methods to extracting as much information as we can from this very, very ancient signal, and it worked fantastically.

Sally Kornbluth: That's cool. Basically, you're using the similarities in 3D structures without having to go through the step of crystallizing the proteins. AI can essentially extrapolate the 3D structure and then make the comparison.

Greg Fournier: In this case, yes.

Sally Kornbluth: I know it's not that simple and we're not putting all X-ray crystallographers out of business, but that's really interesting. Coming back to an earlier thing that occurred to me. When you look at, let's say, a protein family, and you, as you said, believe the common ancestor is before the cells that we know of now, I guess these proteins are too complex to make the claim that there would have been independent origin, correct? In other words, they had to have come from a common ancestor.

Greg Fournier: Yes, even for relatively small proteins. There are 20 different amino acids.

Sally Kornbluth: The combinatorial, yes.

Greg Fournier: It explodes to the point where if you take the average-size protein, the probability of two of those arising by chance is, like, one over the number of subatomic particles in inverse squared by itself. It's just hyper astronomical. There are no numbers.

Sally Kornbluth: Fair enough, fair enough. I asked that as a naïve question, or sounding naïve question. I kind of knew that but I wanted to tease that out for the audience, because that kind of question does come up. Particularly when people are discussing evolution, which sometimes can be hard to think about on the long time scales.

Greg Fournier: In a way, it is a trick question. Because the reality is, evolution doesn't need to find an exact sequence to do something.

Sally Kornbluth: That's exactly right. When you think about microbial evolution and how things might move over time, thinking about things like horizontal gene transfer and how microbes pass genetic information back and forth, maybe you can tell our listeners a little bit about what that is, and how that helps you understand evolution of the microbes around us.

Greg Fournier: Sure. It's really important. For the most part, complex life acquires this genetic information through parent cells and then it passes them on to progeny. However, microbes don't necessarily need to rely on that. They don't need to only rely on mutations or changes that were acquired by their direct ancestors and passed on. They can acquire genetic material — DNA — directly from the environment. Sometimes this happens through viruses. Sometimes this happens just through taking up pieces of DNA as food. Microbes or prokaryotes, they don't have a nucleus, so their DNA is just floating around in a tightly packed bag of proteins and everything else. Some of that DNA can recombine into their genome. If it's not particularly harmful, it won't be removed and it won't kill the cell. If you're especially lucky, it may actually be expressed and have a function that increases fitness in the long run. It doesn't matter if this is an incredibly rare event because there are a lot of microbes.

Sally Kornbluth: Right, right. You've been at MIT for a while. I understand that in your academic career, you've somehow kept circling back to MIT. It may not have been totally intentional, but can you talk a little bit about your experience at MIT, the intellectual community, how that's influenced your work?

Greg Fournier: It is true that I seem to keep on ending up back here. I was a technician for a few years in a few different labs right after undergrad. Then I went to graduate school at the University of Connecticut. Then afterwards got the NASA postdoctoral fellowship and ended up being hosted by a lab here in Course 1 at MIT. Then was fortunate enough to get a position in the EAPS [Earth, Atmospheric, and Planetary Sciences] department.

I think one of the best things about MIT, and why it works so well for the kind of work I want to do, is people here, first and foremost, want to study interesting problems and come up with interesting solutions. That matters a lot more than discipline or department or field or methodology or approach. If you have an idea and you reach out and talk to somebody, they will likely be excited about it. If they don't want to work on it or can't work on it, they will certainly help you find someone that does. That kind of collaboration and interdisciplinary or cross-disciplinary approach or vision, is, I think, fantastic for the kind of work that I want to do. I also think it's really good for students, because in the real world this is how things get done.

Sally Kornbluth: Every time I think of some interesting question or something that strikes me while I'm here and I look to see if anybody at MIT is doing whatever it is, the answer's pretty much always yes. You can find somebody who's working on almost any area of science and engineering. With your example, if I were in a lab, I could reach out for help with the project or to think about things. It's really amazing.

Greg Fournier: It really is. I think people here really do care about helping each other do the best work they can and about exploring things that they really care about and are excited about.

Sally Kornbluth: If someone is just starting a career in academia, I think the impression is that PIs [Principal Investigators] who are here really had success at every step of the way. From my own experience, I know that's not true. You're always taking turns.

I'll give you an example from my own career, which is when I was originally working on cell proliferation and cell cycle. That's what I started my lab on. We set up this really incredibly complicated experiment. Every time we set up this experiment, the nuclei would essentially explode. We were looking under the microscope, “What is that? What is that?” Then we suddenly realized, from some pictures I'd seen elsewhere in a journal, that we had essentially been looking at apoptosis, or cell death, in the microscope. Someone else had actually reported an in vitro reconstitution of that as well. But we had been looking at completely different stimuli. I was like, "Oh, that's why the way we set up this experiment happened." It actually led to more than half of my lab working on cell death. But it started out as months of failed experiments. Then a light bulb went off, “Oh, when you do X, Y, and Z, the cell dies. When you do X, Y, and Z, the nuclei fall apart.” It looked like an error. It was all really frustrating. Then you go, “Well, maybe there's actually an intelligent interpretation of this.”

Are there any moments you might share where you had any doubts or when something went wrong in an unexpected way or opened new doors?

Greg Fournier: That's a good example. There's a project that we're still currently working on that is about the evolutionary history of genes and microbes that get energy by oxidizing iron. All living things use iron. It's one of the most abundant elements on Earth. But it is actually found in very, very low concentrations in the ocean. Most environments are hard to get reduced iron because we have oxygen everywhere. If there's iron and oxygen, you get rust, and that precipitates out. Hanging onto iron is something that life had to evolve and learn how to do. It's very, very good at doing it. We were hoping that we could just find the genes involved in metabolisms for oxidizing iron and they would reach back billions of years and tell us about how iron worked in the Archaean ocean before there was oxygen and iron was freely available to all for energy metabolism. But every time we reconstructed the histories of these genes, they didn't reach back deep in time. They were only found within relatively narrow groups on the tree of life.

Sally Kornbluth: Yet you knew other organisms had to be handling iron somehow.

Greg Fournier: Right. We probably looked at 30 or 40 genes. We tried so many different enzymes and functions that are even tangentially related to these iron metabolisms. No matter what we looked at, it was the same answer. There was no record. This is not what we hoped to find. It's a bit frustrating. After discussing it with our students and postdocs in the lab and thinking about it, we're like, "Well, okay, maybe if we flip this on its head, it's actually telling us something very interesting." Which is, maybe there are some ancient metabolic processes and microbes that are invisible. Because what's happening is, evolution is throwing them out and replacing them so frequently. Or different groups on the tree of life are replacing one another, which is something that we know is the case for complex life, but really isn't how we think of microbes for the most part. If that's the case, then we're only seeing the snapshot of relatively modern microbial life that's doing iron oxidation. This niche was probably way bigger three billion years ago, but the microbes doing it have either changed jobs or the genes have been replaced or changed. Maybe there's some instability in this niche where it's being overwritten or overprinted with more recent diversity, the way we see after a mass extinction or the way we see in complex life. Maybe this is a hint that in some ways microbial life actually works a little bit more like complex life than we really previously appreciated. Where there are turnovers or funnel secessions of different ecotypes that replace, even down to individual enzymes, older lineages, even though they were doing the same job.

Sally Kornbluth: It's funny, my PhD thesis advisor used to always say, “You can learn as much from the experiments that didn't work if you really think about them.” Now, of course, that assumes you didn't mess up the experiment. But when you've done something legitimately correctly and you get a really unexpected result or a failure in what you anticipate, it can open up whole new doors in terms of how to think about things. I think students, when they're doing their graduate work, that can just be seen as a long frustration. Because obviously it can take a long time. But sometimes it really does open really interesting new doors.

Greg Fournier: It really does. It also helps teach us how to formulate hypotheses instead of, “We failed to prove the hypothesis or we did not.” Instead, to think of it as, “Here are multiple scenarios.” Then we'll see what the data and analysis says. Which one of these scenarios is more likely? Ideally, all of them are interesting in their own way. Instead of accepting or rejecting a hypothesis, it's about, “Do our observations and data and analysis support one narrative over another one?” Then tell that story. Because there's always an interesting story.

Sally Kornbluth: Exactly. I still remember this really well because there's a friend of mine who's at the University of Connecticut, Bruce Mayer, he was across the bench from me in graduate school. We had been working, all the lab had been working, on cancer-causing genes that were part of a family called tyrosine kinases that attach phosphates to tyrosines. He had just been sequencing this cancer-causing gene. We assumed it was going to be another one of these enzymes. It had the same structure at the end terminus but then there was no kinase domain. It turned out to be a whole new class of signaling proteins that he had not anticipated. At first, it was like, “What's wrong with this sequence?” Then it was like, "Haha, there's nothing wrong with the sequence. This is what it looks like." That was kind of cool.

Greg Fournier: That's really interesting, considering how conserved those domains are.

Sally Kornbluth: Right. It was SH2, SH3 domains, and no kinase domain. These little modular SH2 and SH3 signaling molecules are known to be modular. I think if you think about long-term evolution, a lot of these little pieces of protein find themselves in lots of different contexts.

Maybe a last thing. When you're training new young scientists, do you have any important words of advice for them?

Greg Fournier: I think the advice I find myself giving students repeatedly is to try a lot of different things. Then follow up on the ones that seem to be going somewhere and that you're excited about.

Sally Kornbluth: You have just talked about evolution. In other words, try a whole bunch of different things, and then the selective pressure of your experimental findings should force you along a pathway that leads to your successful outcome. Anyway, thank you so much for being with us here today.

Greg Fournier: Thank you very much. It was a pleasure.

Sally Kornbluth: To our audience, thanks again for listening to Curiosity Unbounded. I very much hope you'll join us again. I'm Sally Kornbluth. Stay curious.