Skip to main content
Student homeUCI News home
Story
5 of 50

The UCI Podcast: Directing biomedical evolution

Chang Liu's Science study outlines improvements to the gene evolution system his lab developed

Chang Liu, UC Irvine professor of biomedical engineering, is a leader in the field of directed evolution, an effort to train genes to evolve in a laboratory in order to perform desired functions. Through this method of gene engineering, researchers hope to create new biological substances for use in pharmaceuticals and as cheaper and more environmentally responsible industrial catalysts. The work is also helping the scientific community build genetic knowledge for the benefit of bioengineering.

Liu’s research group has a new paper out in the journal Science in which they discuss improvements to their OrthoRep gene evolution platform. The system allows mutations to happen a million times faster than natural evolutionary time frames. In this episode of The UCI Podcast, Liu speaks with UC Irvine communicator Brian Bell about the findings detailed in the paper.

As a musical accompaniment to this podcast episode, Liu plays Bach’s E Major Fugue from the Well-Tempered Clavier Book 2.

To get the latest episodes of The UCI Podcast delivered automatically, subscribe at:

Apple Podcasts Spotify.

Brian Bell:

From the University of California Irvine, this is The UCI podcast. I’m Brian Bell.

Life on Earth has become incredibly diverse and complex through evolution, the process by which characteristics of species undergo transformations through gene mutations happening over very long time periods. Scientists have been looking for ways to mimic evolutionary processes in their laboratories to come up with new molecules with beneficial properties. The hope is that these compounds will revolutionize pharmaceutical treatments and serve as industrial catalysts that are cheaper to make and much less toxic than current options.

For more than a decade, Chang Liu, professor of biomedical engineering who also holds a joint appointment in the Department of Molecular Biology and biochemistry, has been leading the charge at UC Irvine in the field of directed evolution. This week, the journal Science has published a paper by Professor Liu and his collaborators. The work outlines a directed evolution platform created in the Liu Laboratory that drastically increases gene mutation rates over natural evolutionary time frames. A conversation with Professor Chang Liu is up next.

Bell:

Professor Chang Liu, welcome to the The UCI Podcast.

Change Liu:

Thank you.

Bell:

So today we’re going to talk about some of your work in the field labeled broadly as directed evolution. Is that still an accurate name for the work that you’re doing?

Liu:

Yeah, it is directed evolution focuses on basically using evolutionary ideas and evolutionary algorithms and processes to improve molecules. And we do just that. And I suppose more precisely we have developed kind of more advanced forms of directed evolution where things can evolve autonomously inside organisms and that basically scales the process and speeds it up by quite a bit.

Bell:

How long have you been pursuing this field of research and what have been some of your major achievements?

Liu:

My group started in 2013 and even before that I was working on directed evolution and its applications in protein engineering. But in terms of the history of my group, uh, we had this idea that directed evolution could be done inside cells by accelerating the mutation rate of specific genes so that you don’t have to take them out and actually work with things in a test tube, but you can have the cell be the environment by which something evolves. Uh, and that idea started in 2013 and it led to our development of what we call an orthogonal replication system. That has been a major focus of my work since and over the years. We’ve basically improved the orthogonal replication system where we can now evolve genes of our interest at very high rates because those are encoded on a separate system inside a cell. And that separate system has the advantage that its mutations and its mutational parameters and mutational rate do not impact the genome. The genome still is happy and fine at its low mutation rate of the organism. So the organism is happy and fine, but our gene of interest is under this high pressure to mutate and evolve, and that’s what our architecture of orthogonal application has achieved. So that’s basically our, our main achievement in the lab and we’ve been advancing it in a number of dimensions and directions since we started in 2013.

Bell:

So this upgrade of the OrthoRep gene evolution system is the subject of your new paper in Science, correct?

Liu:

That’s right.

Bell:

In the science paper you stress that it takes eons for genes to evolve, but you’re trying to get genes to evolve in a much shorter time frame in your laboratory. How fast are we talking about?

Liu:

The answer is it depends, but I think in many realistic cases now we’re at a million times the rate that you can expect a gene to evolve in nature. And I’ll give you an example. So in this paper, the kind of technical achievement is that we have our orthogonal replication system now mutating at 1 million times the rate of the genome. So this means that if we have a gene and we encoded on the orthogonal replication system, it will accumulate mutations a million times faster than if that gene remained in the genome, right? So there’s a lot more sequence change that is available to it. And if you imagine a case where evolution is occurring kind of neutrally by this I mean that we are not changing the environment, but a particular gene just has to stay functional and operate at the normal ability that it has, then what is limiting sequence change is simply the mutation rate.

So in that case, we will have a million times acceleration in evolution. Now, I think it’s reasonable to ask, right, what is the point of evolving something if you’re not changing its function? And there is actually a major point which we emphasize in this paper, which is that you want to be able to have lots of sequence diversity that you can analyze because there’s not just gonna be one way that a protein or a gene can achieve its function or maintain its function, but there are probably going to be millions and billions and trillions of ways that that is possible. And if you can get more and more examples of those different ways by which a gene can maintain its function, you can learn a model of how the gene works. And this is exactly what we do in the paper. There’s kind of another aspect of evolution which we work on, which isn’t emphasized in this paper, but in other papers, which is of course to evolve things to have new functions that we want very quickly.

And there our million times increase in the mutation rate doesn’t necessarily translate to a million times increase in the evolution rate because it just depends on how hard it is to evolve that new function, right? If it’s really easy to evolve that new function and a beneficial mutation is very easy to come by, then you don’t need a million times increase in the mutation rate to access it. You can have a much lower increase in mutation rate and still access it and evolution will run with it. So I think in the case where we’re trying to develop new function, uh, we certainly get a dramatic acceleration, but it’s hard to pin whether it’s a million fold or much less. It’s somewhere in between. Right?

Bell:

With these millions of mutations means there’s kind of a data science problem on your hands. And in the paper, you talk about the use of machine learning. Can you describe how that plays into the research a little more?

Liu:

Yeah, exactly. So our work is very synergistic with machine learning. And this is because, you know, the success of machine learning comes from the fact that it, it is it’s knowledge base, right? You train machine learning models with lots of examples and then it learns something about the probability of a given function from those examples, right? You know, we see this with something we’re relatively familiar with, with, with image recognition or image generation. You know, if you give a machine learning network a, a billions of examples of images, it can learn to parse what the features are and what makes an image of a person and what makes an image of a dog, et cetera, right? So it’s all based on a lot of data. And in biology the most important type of data is evolutionary data with regard to machine learning. So I would say that the majority of successes in our modeling of genomes and of proteins through machine learning comes from the fact that there’s just billions of years of evolutionary data providing examples that you can train on.

And our contention is that even though nature has already provided a lot of that data, we would like more, right? We would like more, especially in cases where there aren’t that many examples of a particular protein function or if it’s a designed protein, there aren’t that many examples that nature has made that are similar. Or if it’s a protein that we want to operate in a specific environment, right, that nature has not addressed in detail, then we want the ability to generate that data. The fact that we can now accelerate the evolution of genes by so much and generate a lot more evolutionary data on demand means that we can interface with machine learning and better train these models. At least that’s the hope going forward. And in this paper, we give some example of that, where we show that the sequences we evolve for a given enzyme are not ones that, uh, leading machine learning models trained on natural data can predict suggesting that, you know, we are finding new spaces in the world of possible sequences that would benefit these models if they had access to.

Bell:

In the paper, you say that yeast is an important experimental platform in your work. Can you describe the role of yeast and some of the other tools that you use in your laboratory?

Liu:

The reason we focus on yeast is, you know, one major reason is that this orthogonal replication system that we developed is in yeast, and that’s for a number of kind of deliberate and historical reasons. But you know, that’s one of the main explanations that we’re in yeast because the system, the genetic system that we developed for rapid evolution is in yeast. But I can speak a little bit about the advantages of yeast. I mean, yeast is a very well understood organism. It’s very easy to manipulate. The genetics are well worked out, and so practically it grows very fast. So practically it’s a very good host to do evolution experiments. And scientifically yeast is a, you carry out, it is a good model of a lot of human biology. So what you can produce in yeast and what you can learn about biology in yeast often applies to higher organisms. And so it just is a natural model system by which we do our experiments.

Bell:

If you can rapidly produce specifically design genes in your laboratory, what are some uses for these molecules?

Liu:

One of the things that genes encode, of course, are proteins, right? And so a lot of our work is in the space of protein engineering, where we try to get proteins to do new things that are useful. And one area where proteins are very useful is in therapeutics. So a lot of drugs, a lot of modern biological drugs that you take are proteins and they need to have specific properties, they need to have specific abilities, and we evolve proteins to have those abilities. For instance, we have a large subgroup in the lab working on antibody engineering, where we’re trying to get therapeutic candidates that can neutralize viruses or target cancer antigens and address a number of diseases that way. Another area that proteins are useful for is in bio catalysis, where we want catalysts that can make new molecules, right? And even though there’s a whole ecosystem of organic chemistry methods to make new molecules, there are some downsides to those, such as the fact that they often rely on solvents that are, you know, not very environmentally friendly. And the reactions are sometimes inefficient. And if you can get enzymes to replace some of those reactions or even do new reactions, then you can make more molecules with more versatility. And so that’s another area that we work in where we use our evolutionary techniques to challenge enzymes to evolve new properties.

Bell:

How is your team organized in your laboratory? Do you work with undergraduate as well as graduate students? Postdoctoral scholars? How many people are in your lab?

Liu:

So right now my lab is, let’s see, nine PhD students, six postdocs. We have a lab manager and a project scientist. So you know, it’s hovering around like 17 people. And then in addition to that, we have a few undergrads who do their senior thesis or undergraduate research in the group. It’s a good sized lab. And the people there are very talented. They’re very engaged and they have diverse backgrounds spanning multiple disciplines. So it’s a very like lively environment to be in.

Bell:

How does it feel to have your work published in this prestigious journal Science this week?

Liu:

Yeah, it feels really good. I mean, we are very excited about this work and what it means for both our lab’s, future work and for the field. So I’m glad it will be featured in a prestigious journal where many people will see the, the fruits of our labor. Yeah, so it feels, it feels good. Yeah. I’m happy.

Bell:

Professor Chang Liu, thank you for joining us on the UCI Podcast today.

Liu:

Yeah, thank you for interviewing me.

Bell:

You can learn more about directed evolution in the laboratory of biomedical engineering Professor Chang Liu at the Henry Samueli School of Engineering website, engineering.uci.edu.

And incidentally, the music we we’re listening to right now is being played by Professor Liu on a keyboard he keeps in his office. What are we listening to?

Liu:

This is Bach’s E Major Fugue from the Well-Tempered Clavier, Part Two.

Bell:

That’s right. Music knowledge and cutting-edge science happening right here on the fourth floor of the Susan and Henry Samueli Interdisciplinary Science and engineering building on the UC Irvine campus.

The UCI Podcast is a production of Strategic Communications and public affairs at the University of California Irvine. I’m Brian Bell.

Thank you for listening.