Beyond Darwin: Ways to Evolve New Functions

ALTHOUGH DARWIN PAINTED IN BROAD STROKES how evolution has prodded organisms to develop new forms and functions, investigators in several different scientific fields are exploring how organisms evolve new functions in a much more detailed way. New experimental methods and tools are expected to greatly aid those explorations by enabling the quick, inexpensive and complex analyses that are needed for laboratory investigations of evolution.

These “high-throughput” methods conduct millions of tests simultaneously and include genomics -- the deciphering of all the genes and their activity or alterations (mutations) in a tissue. It also includes microfluidics, which uses nanotechnology to create a microscopic network of pipes in which numerous simultaneous tests can be run, resulting in what some people call a lab on a chip. Microfluidics is also enabling genetic or other tests on a single cell. Another newly emerging field that promises to be relevant to uncovering how new functions evolve is the engineering of new proteins using various genetic and biochemical techniques.

The synergy of all these fields should lead to a better understanding of how complex new structures, such as the eye or even the entire nervous system evolved and enabled new functions. These findings are also likely to further advances in directed evolution, with such practical applications as improved vaccines, or bacteria engineered to produce oil from sugar or to carry out other useful new functions.

To explore this emerging field of evolution of new functions, 19 experts in a diverse range of fields, including evolutionary developmental biology, physics, biochemistry, genetics, and bioengineering, presented talks at a Kavli Futures Symposium held at the Aspen Center for Physics March 6-9, 2011. The Kavli Futures Symposia bring together small groups of scientists from different fields to discuss common future trends, challenges, and opportunities in science.

Three of the experts that attended the Kavli Futures Symposium on evolution of new functions were then brought together in a teleconference hosted by the Kavli Foundation to discuss what they got out of the symposium, and what the future might hold for the study of the evolution of new functions. The experts were:

• MICHAEL BRENNER, Professor of Applied Mathematics and Applied Physics at the School of Engineering and Applied Sciences and the Kavli Institute for Bionano Science and Technology, Harvard University
• STEPHEN QUAKE, Professor of Applied Physics and Bioengineering at Stanford University, Investigator, Howard Hughes Medical Institute
• MARK MARTINDALE, Researcher, Kewalo Marine Laboratory, University of Hawaii

The following is an edited transcript of the teleconference on the symposium.

The Kavli Foundation (TKF): What inspired you to have a Futures Symposium on the evolution of new functions?

BRENNER: The symposium was the brainchild of a conversation with Boris Shraiman -- a theoretical physicist at the Kavli Institute for Theoretical Physicis at the University of California, Santa Barbara -- myself and a couple of others. We felt there was an interesting nexus between the development and complexity of biology, and modern developments in biotechnology. It was our perspective there were just enormous opportunities becoming available in biotechnology, given that Steve Quake and others were making experiments possible that were never possible previously. We thought that we could stimulate a conversation about the future if we got a group of people together who had very different perspectives on the developmental biology side, the protein biology and molecular side, and the biotechnology side. It was a risky meeting to put together because we brought together people who normally would never talk to each other, and you don’t know what’s going to happen when you do that. But it was an enormously successful meeting and everyone really learned from each other.

TKF: One focus of the symposium was the promise of microfluidics and other lab-on-a-chip technologies to more efficiently and thoroughly analyze complex evolutionary phenomena. Is that correct?

Michael Brenner, Glover Professor of Applied Mathematics and Applied Physics in the Harvard School of Engineering and Applied Sciences (Courtesy: M. Brenner)

BRENNER: That’s part of it. These advancements allow you to do things that were never before possible. In evolutionary biology, the most difficult part of the research is things have to be repeated, so there’s tedium to all these repetitious experiments. If you want to do things over a long time scale under many different conditions, it’s really extraordinarily difficult. But technology is evolving to make that happen in a much easier and quicker way to understand biological evolution, and to be inspired by the ways that biological evolution has occurred in the construction of novel functions. At the same time, we can use this new understanding to figure out how to make better ways of doing microfluidic experiments. They go hand in hand, and this conference was an effort to study that.

MARTINDALE: What we want to do is compress evolutionary time. We can’t wait 500 million years to figure out if some novelty is going to appear or some new interaction is going to evolve. So we need to compress evolutionary time into a time-frame such that we can actually understand and observe it and experimentally manipulate it. The high-throughput approaches provide a new opportunity to do vast numbers of combinatorial kinds of experiments in a short period of time and let you try to detect the kind of novel interactions or new processes that may lead to some new and more complex traits that are more than the sum of the parts. So that’s what the attraction was to evolutionary biologists—to see how they can harness this high-throughput technology in some way, just like microbial geneticists use the very short generation time of bacteria and other microbes to recreate evolution in the test tube.

QUAKE: It provides a way to dissect heterogeneity, which is crucial in development. You watch the developmental program across a broad range of interesting evolutionary features. It’s all about the ability to figure out what’s happening in particular cells as they morph and change identity. To really take that apart and dissect it in a deep fashion, you need the ability to do single-cell genomic analysis, and that’s really what microfluidics brings to the table. It’s been a great window in the study of these problems.

TKF: Why do single-cell analysis as opposed to averaging what you get from multiple cells?

QUAKE: Because cells are all different. If you think about the developmental hierarchy of any tissue or organ or even organism – from stem cell to progenitor to mature, differentiated cell – if you mix all those cells together and try to study, for example, their gene activity (expression) patterns, you get a very confused picture. That’s because each of these cells are doing very specific things and you are mixing them all together and averaging the results. Whereas when you are able to look at each cell individually, you can figure out there are different processes going on in different cells. One class of cells will do one thing, and the other class will do something else, and they are not really doing both at the same time.

TKF: Is this important because mutations that will occur and be relevant to evolution occur on the individual cell level?

Stephen Quake, Professor of Applied Physics and Bioengineering at Stanford University, Investigator, Howard Hughes medical Institute (Courtesy: S. Quake)

QUAKE: Yes. For example in cancer, you have a huge pool of mutations, each distributed among various individual cells, and when you treat a tumor with a drug, the drug will kill the cells – except for the one that has the drug-resistant mutation. That cell can grow back and create a whole new tumor. So this genomic heterogeneity really matters in a number of practical situations.

TKF: When it comes to the evolution of new functions, what kinds of new questions are you exploring? And how is this different from evolution research performed in the past?

MARTINDALE: We’re at the dawn of a new era where one could argue the only kind of real theory of development is “differential gene expression”, i.e. that different organisms develop differently because their genes have different activity patterns. This is a concept brought about in the late 60’s or early 70’s that doesn’t really have much explanatory power. One of the real reasons that we haven’t been able to make any solid theoretical progress is that we haven’t really been able to quantify genetic variation, especially at the single-cell level. Another component that would put developmental biologists on the map from an evolutionary view would be to say with certainty what kinds of quantifiable combinations of genes are actually operating in a particular cell type that you can define. The kind of technology Steve and other people are developing will be essential for us to make any of those kinds of quantitative, theoretical models for how the cells and tissues change state, make decisions, and eventually evolve new functions.

TKF: You mean the technology will enable you to describe how a network of genes operate together to create a new function?

MARTINDALE: Yes, and in a more quantifiable way than to just say ‘A turns on B and B turns on C.’ We need to know how much gene expression is required in order to activate the second gene, if there are multiple feedback loops, etc. That’s the kind of detailed information you need to generate a model that has any predictive power.

QUAKE: Everything that Mark is saying about development is also true about evolution, broadly speaking. There are various aspects of evolution that should have a real quantitative mathematical framework, and you can describe them and make predictions that can be falsified. Part of the reason that people didn’t do that in the past was because it was difficult to test them. Now that these powerful new microfluidic sequencing tools let us measure evolution in ways that haven’t been done before, there is an opportunity to come up with the real qualitative theoretical aspects of evolution that can be accessed experimentally.

TKF: Which talks at the symposium seemed to stir the most interest?

BRENNER: I really thought it was the synergy of all the talks. It was such an unusual set of topics to put together. For example, both Mark's and Steve’s talks were fabulous, but they were even more fabulous when put into the context of all the other talks, because there was a synergy. What Steve and Mark have been talking about, with respect to the technological advances allowing new insights into the way that development happened, and the way that evolution happens, is clearly important. But to think about how biology actually made things is also very stimulating if you are interested in material science and you want to make new things. So I wouldn’t pick out one talk or another. For me it was the collection.

Mark Martindale, Researcher, Kewalo Marine Laboratory, University of Hawaii. (Courtesy: M. Martindale)

MARTINDALE: Probably because I know the least about it, I found particularly interesting the talks on protein evolution and protein folding. These talks revealed the number of theoretical morphological spaces that proteins could occupy and how few they actually do occupy. I appreciated the talks that showed some of the high-throughput approaches to understanding protein evolution and protein folding, because understanding that is where the rubber meets the road in terms of biological processes. If you could get some kind of an understanding of how to manipulate proteins and protein-protein interactions by varying everything randomly--by trying every possible combination that evolution might not have ever been able to throw together--then there might be some real traction into understanding how complex traits appear from the molecular level up. I found that particularly exciting.

TKF: But isn’t the reason why there’s this unused pool of mutations -- in part -- because so many mutations are lethal?

MARTINDALE: They are lethal in an organism sense but not in the ability to generate a novel structure.

BRENNER: There were talks on how mutations in proteins can lead to novel characteristics. The argument of Jesse Bloom was that two mutations combined could negate the deleterious effect of a single mutation, and people talked about the complexity of the entire process.

MARTINDALE: It was interesting to hear about Sue Lindquist’s work on the heat shock protein 90, which has the ability to stabilize proteins altered by mutation and how it acts as a sort of capacitor of evolutionary change--acts as a buffering mechanism that allows tinkering to happen. I was excited about getting a new understanding of protein-protein interaction and understanding it at a level at which we might make better predictions--understanding the co-evolution of proteins in a complex, how you can have these incredibly stable complexes made up of 50 different proteins and what the implications are of being able to manipulate one cog in the complex. Also, I was really jazzed about the high-throughput microfluidics that Dave Weitz talked about, and the opportunities it might have for the evolution of chemistry and biochemistry.

TKF: Are you trying to figure out a way of overcoming the limitations that are out there in nature to some degree?

BRENNER: People are just trying to figure out what happened, and what is happening. Jesse Bloom’s talk was in the context of understanding data in influenza and mutations and how flu resistance occurs.

An adult sea anemone with its ring of tentacles surrounding its mouth opening. These animals have about the same number of genes as flies and nematode worms, and yet have far fewer morphologically distinct cell types than these more advanced animals. This finding questions the previously assumed tie between genetic complexity and morphological complexity. (Courtesy: M. Martindale)

QUAKE: I would like to see evolution explored, not just in the conventional context, but also in areas like cancer and the immune system, which are very practical applications of evolution. Your body generates diversity in creating antibodies, and it selects the ones that have good functions. All of the same principles and concepts that apply to studying evolution over the hundred-million-year time scale should also describe what goes on in your immune system over the course of much briefer periods—years, months, weeks. I’m very excited about trying to take general concepts and apply them to areas that haven’t previously been explored as evolutionary models.

MARTINDALE: That’s evolution in a time scale that we can comprehend.

TKF: So if you understand, from an evolutionary standpoint, what goes on as your immune system is battling various invaders, then in theory, you might be able to improve things like vaccines?

BRENNER: Absolutely. Every method people have for thinking about how to combat disease or anything else is developed under an intellectual paradigm. If one could invent new concepts for how evolutionary change occurs, then they could really change the way you think about those problems. At this conference, it was really mind blowing to see what is actually happening. For example, Mark’s talk on sponges and the evolution of related organisms. The notion that the essential elements of the nervous system were actually there in their primordial components before there was a nervous system is quite shocking--there were pre-synaptic proteins that were around before there were synapses. Still, I’m not sure what we’re supposed to do with this information. We’re struggling to figure out the questions and put them into an actual framework.

MARTINDALE: The original paradigm of evolutionary developmental biologists is that evolution is about tinkering. So now what’s exciting for us, in this genomic era, is that we can ask more rigorous questions about what was the same or what was different--what was really new at these particular time points in the evolution of new forms and functions in organisms. There could be really important inventions or quantum level increases in the ability for the parts to interact to lead to new complex traits.

BRENNER: The notion of evolution as a tinkerer and not an engineer implies that you take parts from everywhere and you shove them together and somehow new biological functions happen in a way as to be completely unquantifiable. But it can’t be that bad. There must be some conceptual basis for these processes.

QUAKE: People felt the same way about statistical physics back in the day. How would you quantify randomness and disorder? Eventually a framework was found to do that, and it was quite different than the deterministic, Newtonian mechanics. People are casting around for that in evolution because it doesn’t obviously map on to the way we think about a lot of physical sciences.

BRENNER: Yes, our understanding of mechanics was a complete mess before people understood how the random motions of molecules were contributing to the properties of gases and liquids and solids. Then statistical mechanics led to a conceptual framework for people to make scientific advances, which were mostly qualitative. That’s what seems to be missing here, and the advent of new technologies and data analyses makes it seem possible that we can find a new framework or paradigm to explain how new functions evolve.

TKF: So the advantage of being able to quantify something is that it enables you to make predictions and engineer as opposed to tinker?

This jelly fish-like creature, which is called a ctenophore, is evolutionarily the most ancient still-living animal to possess neurons and muscle cells. These unique cell types enabled the evolution of many different more complex animals millions of years ago. Researchers have recently deciphered the complete genetic makeup of the ctenophore. This genetic information will hold clues to how exactly novel cell types evolved. (Courtesy: M. Martindale)

BRENNER: That’s partly the case. The point of measuring what’s going on in every cell is partly to know [more] so you may be more willing to manipulate, but it’s also so that you can check to see if the qualitative concepts that are driving all thinking about development are correct. We had a talk from Daniel Fisher, a theoretical physicist, who always likes to say that the point of being quantitative is so you know that your qualitative arguments are correct. It gives conceptual frameworks.

MARTINDALE: You score your experiments somehow, so you need parameters that you can actually measure in order to know whether you are manipulating the system in the way that you predict. It’s important from a practical point of view, not just to say that you can do it, but it gives you opportunities to manipulate the system and know what you’re doing after you do the manipulation.

TKF: How does mathematical modeling add another level to this?

BRENNER: I’m in the business of trying to find something to calculate that’s useful. If there were a way to make a theory that explained how new functions evolve it wouldn’t just be spectacular, it would be revolutionary. Now I have no idea how to do that, but the only way, as a community, to figure out what it might mean are meetings like this one where we all talk to each other and muddle through it.

QUAKE: We are all grasping for the framework and the general principles. There’s got to be something unifying out there. That’s what we’re all casting around for.

TKF: Evolutionary developmental biology really looks backwards. You compare different organisms to uncover what changes in embryonic development in the past led to the emergence of new forms and functions over time—for example to see what changes in the development of the fish jaw led to the tiny bone in our inner ears that helps us hear. But it seems like some of the other people at the conference were looking forward and trying to learn from what happened in the past so as to better influence the future. Is that your impression as well?

MARTINDALE: There are two different ways of looking at this historically. One of which is to say that evolution is a historical phenomenon and ask if Stephen Jay Gould could wind the clock back and run evolution a thousand different times, how many different outcomes would you get? That is a philosophical question that even I was too scared to ask anyone at the symposium about because I figured they would just laugh at me, or that the answer was known but no one has told me yet. Then other people are very mechanistically oriented and want to be able to manipulate and understand the exact evolutionary phenomenon, whether it’s a genetic mutation or some kind of non-genetic means of inheritance. You have the evolutionary developmental biologists who are trying to piece together likely scenarios of what happened to give rise to complex traits, and then you have the fruit fly geneticists who are trying to drive the origin of new species by understanding exactly what genes get mutated, regardless of whether they are responsible for the origins of any kind of morphological variation. So those are the two camps that have historically been arguing with one another. This meeting skipped over that and got people thinking about different ways of approaching evolution from all these different points of view. That’s what the refreshing part of it was for me.

TKF: How will the findings of evolutionary developmental biologists and others trying to understand how new functions evolved in the past help those in the lab trying to direct evolution in a specific way, so as to develop new traits in organisms and vice-versa?

MARTINDALE: It depends on what your goals are and what you regard as progress. The people working in the lab have a huge advantage in being able to manipulate the system so as to generate specific new traits themselves, whereas the rest of us are trying to figure out what was the most likely scenario for generating the outcomes we see around us [in various organisms]. There’s still a big gulf between those two kinds of approaches.

QUAKE: Maybe one way to link them is that some of the experimental approaches allow you to explore a wider range of evolutionary behavior than what has been seen thus far. So you can calculate, of all the possible pathways, how many did evolution explore and use. It lets you get a sense of the depth and extent of the diversity that’s been tapped in the natural world.

BRENNER: We had at least two talks that touched on these issues from wildly different perspectives. Paul Rainey gave a fascinating talk. He studied a bacterial colony that can have either a smooth or a wrinkled appearance. He sequenced the genes responsible for wrinkling or smoothness and then mutated these genes in the bacteria of the smooth colony in a series of different ways to find out how many different mutation pathways in the genes of the smooth colony could cause the colony to become wrinkled. He found there were many different [mutational path]ways that did this. This is a completely different perspective about why something happens in evolution—it’s not just because there’s a single mutation that has to occur, there are a number of different mutations that take you down different pathways and can get you to the same result.

Eric Siggia gave a fascinating theoretical talk about the developmental pathways taken to produce the division of an organism’s body into a series of segments, such as head, body, and tail, or the repeating segments that comprise the vertebrae in our backbones. He essentially argued that the detailed structure of the [gene] network responsible for the development of body segments wasn’t very important in determining the kinds and numbers of segments an animal will have, but rather the various ways the proteins these genes produce can interact. He had a computer model that, by generating all of the possible protein interactions in the development of body segments, created enormous numbers of different possible worlds--that had animals with different body segmentation patterns. People would ask him “Is this possible world the same as the world we live in?” and he was dismissive of that question because he could generate lots of different possible worlds. But certain features of the worlds that he created occurred frequently and were consistent with what is seen during development. That was an example of an entirely new way of viewing how segmentation occurs—a new type of theory.

TKF: Are you saying it wasn’t just lethal mutations in genes combined with environmental pressures that tended to narrow the options and lead organisms down a specific evolutionary pathway to develop a certain new function, and that many other possible pathways could be taken that could lead to the development of new functions yet to be found in nature?

BRENNER: Eric had a model of segmentation pathways that was at an entirely different level than mutations in genes. He invented a theoretical description that was at the level of pathways, and he found a way to formulate it so he could generate different worlds, and then he compared the results of those worlds to those generated by experiments.

I highlight this because it’s a completely different way of thinking about this problem than the way that anyone’s ever thought about it before. If some description of that sort became well-developed and accepted in a field, then it could have the same relationship to developmental and evolutionary biology as statistical mechanics did to the study of complex physical systems. It could completely change the perspective of what sorts of questions you are supposed to ask.

MARTINDALE: Ever since mathematicians tried to describe biological processes with their equations and models, people have been able to predict something very similar to what we see in the biological world. The physical properties of biological systems certainly put constraints on the kinds of outcomes that occur in the natural world; the types of forms and functions seen. But evolution is not an engineering phenomenon. Evolution is a series of accidents of sorts, so it’s no doubt that it operates in a different way than what you could predict with a simple theoretical model. That’s not to say that you can’t understand about how different parts of the system interact with one another. In terms of these different pathways for generating the same wrinkled looking bacterial colony, that is probably the result of all that hidden genetic variation that we never really knew existed in the population of bacteria. Secondary mutations are probably operating on this genetic variation, which is allowing different pathways to get to the same trait. That’s sort of shocking to us.

TKF: Aren’t you just talking about the ability of completely different and unrelated animals, such as bats and birds, to have the same trait, i.e. wings? This concept of convergent evolution has been known for quite some time.

MARTINDALE: Yes, but now we have more of a mechanism for it and actually not just one mechanism but several possible mechanisms for the same result.

TKF: So in addition to having this quantitative ability, now you can also explore mechanisms for the evolution of new functions. This is exciting, as before you didn’t know exactly how two very different animal species developed the same trait and which pathways led to the same place.

MARTINDALE: Yes; we don’t have to hand wave about what could have happened. We can actually say what did happen, because we know every gene that’s in there, we know where the mutation is, and we know there are multiple pathways running parallel that operate that stage of development. It’s much more powerful to be able to decipher what the absolute mechanisms might be. This was learned because people actually went out and figured out how it happened for each species, and did not rely on theoretical models for how it ‘could’ or ‘should’ have happened.

TKF: In the end, what does a meeting like this do for the field and yourselves as researchers. Are new collaborations possible?

QUAKE: It’s only a matter of time.

BRENNER: The meeting did a lot scientifically for me. If the goal is to try to find new ways of thinking about things that exist, then the only way to do it is to get people of diverse perspectives together and hope new insights come out of it. The history of science is filled with examples of meetings like this that led to vast changes in perspective. We want to catalyze changes in perspectives.

MARTINDALE: I agree. These were vastly different viewpoints on an incredibly complex problem. It’s a little too early to say that the bridges have all been built, the new model has been established, and we know what direction we need to all move, but this was a good start that forced us to go out of our comfort zones.

BRENNER: The most important question that we as scientists have to face every day is what question should we work on, and how should we think about it. If the goal is to really open up new ways of thinking, then the question always has to be, “Why is today the right day to put out new ways of thinking about this broad area?” And there are reasons for why, right now, this might be a time of change. That’s because of the increased knowledge about biology that has come from sequencing and other technologies. This has opened up new tools that allow us to look at old phenomena in totally new ways that may lead to enough insight for concepts to emerge. The meeting was an effort to try to bring that out and examine that hypothesis. I left feeling confident that something could happen.

MARTINDALE: Unfortunately, I do think it’s important to point out that there’s virtually no funding mechanism to allow any of us to pursue any of the things that we learned and thought about in this fantastic week. That’s because what we’re talking about is discovery and risk taking, which are not the things that the funding agencies that I normally rely on would likely let me explore. I feel depressed in a way. We have to do something new, but we don’t have the funding opportunities to support this. This is something that needs to be addressed.

- June 2011