Interview with Jack Szostak
September 1, 2005
CI: Some of the research you’ve been doing on the origin of life on Earth is fairly radical. How did you get started?
JS: I worked for a long time on yeast genetics, investigating certain aspects of the yeast cell cycle and genetic behavior. At the time I thought that the scientific community had determined all of the components of a chromosome. I took an engineering perspective, assuming that since we had the components and thought we knew how they worked, we should be able to put them together and make something similar to a chromosome. We took that synthetic approach, and the cool thing was that it didn’t work. There was something else going on, something we didn’t understand. In the process of not succeeding, we understood to a far greater extent what it actually takes to build a chromosome. That kind of approach subsequently inspired me. What we’re trying to do now is more ambitious and more risky. We’re trying to put together the pieces of a really simple cell and then see if we can make it functional. We expect to learn a lot from the failures along the way.
CI: Given your training, how and when did you first get drawn to the question of the origin of life?
JS: That began a little over twenty years ago. I’d been working on yeast genetics for twelve years. I started to feel—even at that time—that there were more labs working on yeast than there were genes. Even if we discovered something new, in a few months there’d be five or ten other labs working on it. I wanted to find a project with interesting questions that fewer people were working on.
CI: So you were brave enough to venture that far out into the unknown?
JS: At that point in my career, it felt like it was more important to try something interesting than to just keep doing the same thing. The new thing that was really exciting was the discovery of ribozymes by Tom Ceoh, Sid Altman, Norm Pace, and a few other people. It was a really tremendous advance that revolutionized the way people thought about the origin of life. I was surprised that more people weren’t jumping into the area. With my background in molecular biology, it seemed reasonable that I could contribute something. My lab started working on some of the naturally occurring ribozymes, trying to understand their catalytic activity and harness it to carry out replication-related reactions.
CI: It’s such a big subject that it seems important to break it down. What are the critical steps towards the development of a cell? Maybe you could start with the significance and limitations of all the Miller-Urey type experiments.
JS: Well, the first step in developing life is the synthesis of small molecules or building blocks from simple primordial stuff, which is why the Miller-Urey experiments were such a huge conceptual advance. You could summarize the work since then by saying that now we can at least make all of the components of biological systems. The problem is that you get those components as a small fraction of complicated mixtures. We don’t know how to make just the building blocks without contamination from a lot of other garbage. We also don’t know how to make them assemble the right way into the macromolecules we need.
CI: So Miller-Urey experiments only yield small amounts of amino acids and simple lipids?
JS: Yes, but I don’t even consider it a yield problem because there are always concentration mechanisms. At first glance, certain reactions are great—you can start with formaldehyde and make every sugar. But if you if you want to get to the RNA world and just make ribose, you don’t want to have ribose and 99% other stuff. We need to know how to channel the reactions to give us a specific set of compounds to work with.
CI: You seem to be using two conceptually different methods to approach the problem. In the first, you try to replicate the natural processes that formed the end product—a kind of reverse engineering. In the other approach, you’re exploring whether there are other possible processes that would give the same results, given that your simple experiments use a lot of specificity when it comes to selecting amino acids.
JS: The hope is that as more work gets done, a plausible continuous pathway will emerge. Some set of naturally occurring conditions that might have happened in nature that would have given us the sugar that we want—whether it’s ribose or some other building block. Or maybe another set of conditions that could have given us the right nucleobase subset. Hopefully someone will come along and figure out how they could be put together in the right way. There are a lot of ideas. As more people get involved and more creative ideas come along, the missing links shrink and we’ll eventually have a continuous pathway.
CI: Talking about the gaps reminds me of the Intelligent Design debate. Intelligent Design is a far more subtle anti-scientific stance than pure creationism, because evolution and Old Earth theories are acknowledged. Their tactic is to attack the gaps of our knowledge—the fact that there are still things that we don’t know about the first half-billion years of the Earth’s history. But you seem confident that connecting the dots and missing pieces doesn’t represent any fundamental lack of understanding. What about Hoyle’s famous comment that evolution is as likely as a tornado blowing through a junkyard and assembling a Boeing 747 jumbo jet? Can you briefly say why that simplistic perspective is flawed?
JS: True—you didn’t just start with a bunch of water and methane and have a tornado come through and end up with RNA. You have to look at the bigger picture; adjust your timescale. Think of it as a huge transition that can be explained by breaking it down into a series of small steps that occur in different environments over time. Once you examine these smaller steps you can understand how the end product is possible.
CI: Does the sheer time span and chemical real estate required for your research make it hard to replicate early origin scenarios under lab conditions? Is that viewed as a fundamental problem?
JS: No, quite the opposite. Let me explain the way I think about it. There are really two distinct, but related things that motivate our work. First, there’s the challenge of trying to assemble the simplest living, evolving, replicating system based off everything we know about chemistry and physics. Constructing this is informed by models that already exist. That’s completely distinct from prebiotic chemistry and the origin of life on Earth. We’d just like to find some way of building an artificial cell. I think that’s one of the huge challenges of modern chemistry, so it’s an exciting area to work in.
The second objective does have to do with the origin of life. Basically, I’d like to find a reasonable explanation for how life could have originated on this planet. I think the two motivations are complimentary. There might be particular steps of one problem or the other that we have no idea about. So we just forget about the early Earth for a while and think of a way that we can solve the same problem from chemistry and physics or vice versa. Your perspective changes and you recognize that the problem is not as difficult. When a problem isn’t insurmountable, you can think of many ways of solving it.
CI: One exciting thing about your first motivation is the possibility that similar things have happened in other habitable places throughout the universe. If you’re addressing fundamental mechanisms that adhere to the universal laws of physics and chemistry, then you’ve potentially established the basis for how life might be both similar and different elsewhere.
JS: Yes, that’s a very exciting aspect of the project. It would also be wonderful to experiment with creating really different forms of life in the laboratory.
CI: Since the experimentation so far is at the simple biochemical level, which aspects of the biochemistry seem to be inevitable or unique to the way life happened on Earth? And which seem to allow the possibility of something quite different?
JS: We think that the design and structure of a simple cell should have two main components. One component would be some kind of genetic apparatus for transmitting inherited information. The second would be a type of compartment boundary structure. In terms of the genetic apparatus, as far as we know the only way that it works is with some kind of nucleic acid. There are now hundreds of variants of RNA and DNA that have been developed by synthetic chemists. A lot of pieces in those variants are different than their original RNA and DNA templates, but the differences tend to be rather small in the scheme of things. There’s no reason why these variant polymers shouldn’t work in principle, but we don’t have any examples yet.
The compartment boundary issue is more open. We’ve concentrated our work on simple kinds of membranes for continuity with modern biological systems. But I think there might be more ways to handle that issue.
CI: In very general terms, why does life need a compartment?
JS: A compartment allows you to hold the relevant molecules together locally in space. You need to have some way of differentiating the cell from the rest of the environment. It’s a critical aspect of Darwinian evolution because the genetic molecule has to stay physically linked to whatever it is that’s expressing the selected phenotype. One easy way of accomplishing that is with membranes, which are universally used in biology. But at least conceptually, you can do it with water droplets and oil, mineral particles, or even porous gels. Anything that restricts diffusion will theoretically work.
CI: How important is the mobility in a liquid? Because you might imagine that the porous space of solid rock would be a pretty good place to localize chemicals.
JS: You still need to have a way of bringing in matter and energy, whether that means high energy chemical compounds or mechanical energy. In that sense, the cell can’t be completely cut off from the environment. Matter and energy are going to be transformed and stuff has to leave the system as well, so there’s a necessary balance that needs to be established.
CI: The porous space in rocks is also rather rigid; your work has already shown that you can grow compartments in ambient medium.
JS: True. The reason that we’re putting most of our effort into the membrane system is because it’s workable.
CI: What about the other part of your simple cell—the genetic apparatus? As you said, there are many potential variants of replicating molecules. They may have different error rates or different attributes as far as life goes, but is there any reason to believe that there’s a unique, convergent process in the evolution of biochemistry?
JS: That’s a good question. You can make a lot of arguments for and against RNA and DNA. Ultimately, that’s something that has to be answered experimentally by synthesizing other systems and looking at their properties Albert Eschenmoser has given us the preeminent example of that kind of work. For a long time people looked at RNA and DNA and said, “Wow, these are really wonderful molecules. Their base pairing is fantastic! Maybe biology has converged on the solution for information storage and transfer through these two molecules.”
But then Albert synthesized some variants to RNA with different sugars and utilized different methods for connecting the atoms to see how their base pairing compared to RNA and DNA. What he found was that base pairing is not at all unique to either. There are quite a number of related nucleic acids that possess excellent base pairing systems. In fact, some of them show much stronger base pairing. Others show much weaker pairing, and some show different geometries than the double helix. It really changes the way to think about the question. You can’t just assume that biology chose a particular configuration because the base pairing had the right strengths or the right context for getting accurate replication. It delves to a more subtle level.
CI: Once you’ve shown that there are other modes of base pairing and replication, your projection of the vast set of potential translational proteins must be a sizeable area to explore.
JS: Once you get into translation you have similar questions on an even larger scale. We’ve started to do some work on that. First, you have to ask why we have twenty amino acids that are used more or less universally. There are several possibilities and questions to answer. Which amino acids were available earlier? Were they utilized first with others adding to the code to exploit them? Do these twenty amino acids compose an ideal set, or would any comparable set of amino acids work just as well? We haven’t had the tools to answer these questions until very recently, but now quite a number of labs have been working on expanding the genetic code and developing technology for synthesizing r-chains that contain different amino acids.
CI: Talk about the special role of water in all of this. It’s often talked about as an essential ingredient for life but I’ve heard you say that water is also a nasty fluid at some level.
JS: It’s the most toxic compound in our environment. It’s always hydrolyzing our DNA and RNA, which generates mutations. Cells spend a huge amount of energy and machinery correcting those errors. In fact, all macromolecules are hydrolyzed by water.
People go on and on about how wonderful and special water is, and water does provide some properties that can be taken advantage of. But other properties are really disadvantages. The question of whether you could have life in some other solvent is really interesting, and at some point I would like to start a project on that.
CI: What would your next favorite solvent be?
JS: I think to make it interesting it should be something really different from water. Since water is a polar solvent, I think it would be interesting to play with a really apolar solvent like hexane.
CI: Should we accept the primacy of carbon as the basis for information-carrying macromolecules?
JS: I think so—no one’s come up with a really good counter-example.
CI: The interesting part of the early life on Earth story is that it’s not a lab experiment that’s performed in a vacuum. The profound interaction between the environment and the entities that emerge seems pretty hard to model.
JS: It’s hard to model in the sense that we don’t have very good constraints on the environment of the early Earth. Any time we build a system in the lab we’re giving it a particular environment that we more or less decide upon. If we build a proto-cell, its interaction with the environment a critical factor in determining how it obtains new building blocks for growth, and what kind of energy sources it utilizes.
CI: Stephen J. Gould creates a hypothetical situation in which a hundred Earths are switched on under identical primeval conditions and left alone to evolve. He argues that you would be amazed to find mammals on most of them and humans on any of them except one. Going back to the first billion years on those hundred Earths, what level of diversity would you expect to find from one planetary experiment to the next?
JS: My intuition is that the answer would lie somewhere in between the extreme defined by Gould, and the other extreme as propounded by Simon Conway Morris, who overwhelmingly believes in the power of convergence. As you said, chemistry and physics are universal. So anything that evolves has to be consistent with chemical and physical laws. There are not an infinite number of ways to solve a problem. There may be only one way of solving a problem, or there may be one way that is the simplest and will therefore evolve first. We see this in our laboratory evolution experiments. When we take a population of RNA molecules and try to select molecules that can do a certain job, like catalyze a reaction, or bind to a certain target, we see in some cases that there is one, best, simplest, way of doing it. We get the same answer again and again in independent experiments in different labs, which is convergence. In other cases, however, we see lots of different solutions. So I think the answer is going to be somewhere in between.
CI: In your mind, what’s the minimal level of complexity required for the mechanism of natural selection can kick in?
JS: I see it coming in at the stage where you first bring together replicating genetic material and replicating compartment structures. Once those things start to come together, Darwinian behavior emerges spontaneously. How complex of a structure does that require? Depending on your point of view, I think it’s really simple compared to any modern cell, or really complicated compared to any simple chemical reaction.
CI: For people not familiar with the research that would probably be surprising. Most people think of natural selection in terms of fairly complex organisms and don’t imagine it applying to biochemical systems.
JS: Right. I think this is part of the problem that many people
still have when thinking about the origin of life. They become distracted by
how complicated modern living systems are. Comparatively, the first living
cells were extremely simple. What makes the whole thing possible or plausible is
that you can subtly shift from complicated chemistry to simple biology. Once
you have simple biology,
CI: How does the natural selection mechanism in this simpler regime drive towards complexity?
JS: That’s easy. Just imagine the first cell, either on Earth or in my lab. This cell is going to be under extreme pressure to survive. It’s going to be relying solely on the chemistry in the environment, and barely able to replicate. There’s constantly a risk of extinction from perturbations in the environment. The end result is an incredible selective pressure that drives the evolution of complexity.
More molecular functions will make the cell more stable, allow for more effective reproduction, and will increase its ability to adapt to different environmental conditions. As new functions evolve that allow it to survive in different environments, it will gradually colonize different environments on the early Earth. As it runs out of its initial sources of compound food and nutrients, it will be under additional selective pressure to evolve. When you add up all the selective pressures, what you see is a cell that encodes a lot of genes which do a lot of different jobs. You’d see this relatively quickly, at least in terms of geological time. From that point you would pretty quickly get up to the complexity of a modern bacterial cell.
CI: In your mind, what is the most useful way to think about complexity? Is it base pairs? Genes? A biochemical network?
JS: We tend to think of complexity in terms of the amount of information required to specify a system in a given environment. Which is related to base pairs, but more specific, because it only applies to base pairs that are significant?
CI: Would having information that’s too efficiently coded be a disadvantage?
JS: That really depends on the environment and constraints that the organism finds itself in. We find some cells—especially viruses—that appear to be under very strong pressure to minimize the size of their genomes. They go to great lengths, like having overlapping genes, to get the most out of a small amount of DNA. On the other hand, there are organisms like us that have a huge amount of DNA. A lot of it has unknown functions and probably some doesn’t have any particular function.
CI: Does your work on levels of complexity necessary for early life give you any ideas or insights into why it took so long to evolve to multicellularity? Why did life seem to emerge extremely quickly on a very inhospitable planet but not become big and highly complex for a long time?
JS: Nothing that we’ve done really addresses that question. I listen to people talk about oxygen levels in the atmosphere and how the slow transformation that ultimately led to the spike in oxygen levels may have allowed larger, more complicated life to get going, but none of that’s anything we’ve done.
CI: You alluded to the complexity of life later on Earth—beyond the first billion years—and the fact that you have some quite successful miniature organisms that maintained a low level of complexity. Since we know that the Earth has microbes that have been nearly unchanged for three billion years, is movement towards greater complexity really a unique requirement of natural selection?
JS: I would say that there’s a window from the primordial one-gene cell up to something that has on the order of a thousand genes where there’s inevitably going to be very strong selection. A one-gene cell isn’t going to be stable in terms of evolutionary strategy, right? But once you get up to the complexity of bacteria which have genomes of one to two million base pairs, we know just by observation that they can survive at that level of complexity for a long time. But when you’re talking about the environmental constraints or transitions that lead to life, that’s much more complicated. There are many ideas, but I think it’s still an open question.
CI: Let me ask about one of these difficult pieces in the origins of life story. How do you get to an RNA world? What might the steps leading up to that look like?
JS: That goes back to what we were talking about with making the sugars from stuff like formaldehyde and nucleobases from simple stuff like cyanide. The first problem is getting a restrictive set of building blocks. I think that finding ways of channeling a network of reactions to give you a small set of compounds is one of the really big questions. After that, we need ways to put these compounds together in the right way with the right connectivity. There’s been incremental progress in terms of putting together nucleosides and nucleotides, but there’s definitely still work to be done.
There’s also the question of how to bring energy into the system. How do you activate the nucleotides and drive them to polymerize into polymers? There are ideas going back twenty or thirty years or more, but I think it’s still an area that would benefit from more work. That’s what I see as the main challenge of going up from simple building blocks to genetic polymers.
CI: Something I know you’ve had great success with is concentrating the RNA into a compartment. Can you talk about that?
JS: There’s a particular clay mineral montmorillonite which is a relatively common, abundant, clay that’s been known to have interesting catalytic sites on its surface. It’s used a lot as a catalyst in industry. Particular versions of this clay are known to catalyze the assembly of RNA from activated building blocks. We found that the same clay can also accelerate the assemblage of fatty acids into membranes and vesicles. That’s already a process that goes on spontaneously, but it happens a lot faster when you add these clay particles.
We experimentally put some RNA on the surface of the clay and then washed off the RNA that wasn’t bound. After that, we did a membrane assemblage reaction. We found beautiful, fluorescently colored RNA on the clay inside big lipid vesicles. It showed in a pretty visual experiment that these simple mineral particles can not only help assemble RNA and membranes, but can also physically bring them together. It’s very attractive as an example of what could have happened in nature.
CI: How long were the RNA strands that you were working with?
JS: They were short—on the order of twenty nucleotides. Basically, they were synthetic RNAs just used for visualization. People like Tim Ferris have built up RNAs on the order of thirty or fifty nucleotides on the clay surface.
CI: Is the mechanism going to be similar when you go up several orders of magnitude to millions?
JS: I think that the first cells would have actually started off with very short RNAs, maybe thirty or fifty. Then, as a result of evolution acting over time, there’d be selective pressure for longer sequences that coded for more and more functions.
CI: Would it be a lengthy process before you’d have anything that we’d call a gene?
JS: Not at all. I think that as soon as you have a genetically encoded function you have a gene. So to me, this first proto-cell is a one-gene cell.
CI: We now have a zircon that shows that Earth might have been habitable 4.4 billion years ago. Would you speculate as to a minimum timescale for the development of a one-gene system?
JS: I think that this could have happened fairly quickly. There are different parts of the process that probably have different timescales. To develop certain geological environment could take a significant time. It could be on the order of millions of years. On the other hand, molecules like sugars, nucleotides, and RNA don’t stay around for millions of years. The transition there had to be fast, and could have just been years.
CI: What do you mean they don’t stay around?
JS: Suppose you have a big pile of ribose just sitting on the planet. It’s not going to have a long shelf life. Something has to happen to it or it’ll be degraded. The same thing is true with nucleosides, nucleotides, and membrane compartments. The more complicated the building blocks are, the more fragile they are. It makes me think that once you have all the early environments in place, maybe all the chemical steps came together and happened between one and a thousand years. But to get up to that point might have taken millions of years. It certainly took tens of millions of years for the early Earth to cool down enough for this to happen and it might have taken additional tens of millions of years to have land and continents. You’d also need the basalt to weather enough to give you clays and evaporate deposits. To get all of the geology and microenvironments on the planet probably took tens of millions of years. But once all that was in place, I think the chemistry could have been fast.
CI: If we’re imagining other potential habitable environments where the basic chemistry is similar, would it surprise you if many of the same mechanisms we find on Earth would also apply?
JS: Unless we were really lucky and got to other planets at a really early stage, what I’d most expect to see would be life with the complexity of bacteria. You’d probably have to time your visit to just the right stage to be able to see simpler forms of early life. And of course at the other end of the spectrum, we don’t know if, and how, more complex life would have evolved. That makes it a pretty interesting question.
CI: And different versions of this chemistry could have played out in quite different environments. For instance, could you imagine vesicle formation occurring near deep sea hydrothermal vents? Is that plausible?
JS: There are some ideas on how the building blocks of membranes—the lipids—could have been made in those environments. I’m not sure they can be concentrated and assembled into vesicles there, but I really don’t know. To me it seems more plausible that they’d be made in one environment and then transported to a different environment where they could be concentrated.
Since these kinds of molecules tend to be found on surfaces, one idea is that they’d form a sort of oil slick on the surface of ponds and lakes, or even the ocean. Then wave and wind action could blow them off into aerosol droplets which could then get blown on land near shore. Eventually you’d get deposits of organic material and you’d have a really concentrated source of material that could be siphoned away by a nearby stream. Then you’d have a small pond with a really high concentration of these little membrane vesicles.
We need to get away from the idea of thinking of the Earth as a homogenous environment. Life didn’t start in the primordial ocean; it started by bringing together things from lots of different micro-environments.
CI: In the sense that Earth is just one biological experiment, could there be a range of biological experiments that might occur on other planets?
JS: Yes. Earth is composed of such an eclectic and complicated set of environments that it makes it hard to imagine what life would be like on a different planet. With a different planetary mass, different spectrum, and different orbital period around a star, I’m sure that you’d find any planetary environment complicated as well.
CI: But you’re finding mechanisms of sufficient generality and power that you wouldn’t be surprised to find them operating in a different cosmic context?
JS: I wouldn’t be surprised to find these kinds of things going on other planets, not at all.
CI: Now that you’re working with a lot of the pieces of this story in your lab, what’s the next frontier in your work?
JS: The hardest thing right now is getting a replicating nucleic acid system. We have a lot of new ideas and we’re putting a lot of energy into that. We’ve also made a lot of progress on the membrane systems. What we’d like to have would be a greater array of realistic examples relevant to the origin of life. Then we’d like to put them together. It’s not enough just to have a replicating nucleic acid and replicating membrane system. They have to work collectively under the same physical conditions. We’ve made some progress, but there’s much to be done.