Simulating evolution helped scientists design a better virus

It sounds like an arcane superpower. It boils down to random mutation and selection

Viruses hold a special place in theories about the origins of life. They are incredibly simple things, just consisting of a protein coat (called a capsid) and some RNA (or DNA), and so intuitively they could be some of the earliest and simplest forms of life. On the other hand, viruses are obligate parasites  –  they cannot reproduce outside of a more complex host like a bacterium, plant, or animal. So the question remains, which came first: the parasite or the host?

Conventionally, biologists have assumed that the parasitic nature of viruses meant that they evolved after more advanced life forms. More recent discoveries of giant, complex viruses, however, have called this assumption into question. A new “Virus World” hypothesis suggests that viruses evolved before cellular life-forms and then guided the latter's evolution. In this light, understanding how the first viruses came about becomes key to understanding the evolution of all modern life.

Now, a study published only this month in Nature promises answers to some of the more vexing questions about how viruses evolved. And amazingly, the researchers behind the study began with a very different goal : creating a better way of delivering vaccines.

Viral vaccines

Most viruses are very efficient at packaging all their genetic material into a small protein coat, which can then infect specific types of cells. This ability has made them a key target in the hunt for better means of delivering powerful drugs and vaccines into human cells. However, this very efficiency means that it’s hard to tamper with virus proteins, for example to make them attack only cancer cells. This is why the researchers, from the University of Washington in Seattle, decided to try and make virus-like particles (called synthetic nucleocapsids) from scratch.

They began with computers, designing two proteins that could fit together to make up a 30-sided capsid with 20 triangular faces (a common virus shape, called an icosahedron), which had enough empty space inside to carry cargo. And of course, like a natural virus, the cargo in their case was the very genetic material (RNA) that produced the coat protein. But this was just the beginning  –  a true virus has to not only package its genetic material effectively, but also to protect it from enzymes, present in our cells and blood, that can destroy it.

'Directed evolution'

Here, the scientists again took cues from nature and decided to employ evolution to design a better protein. “Employing evolution” sounds like some kind of arcane power, but all it boils down to is random mutation and selection. First they created a library of proteins, each differing randomly in nine positions that they thought might help better bind RNA. They then made different strains of bacteria express these mutant proteins, purified the resulting capsids, and subjected each to the harsh treatment of an enzyme that degrades RNA. They then analyzed the sequence of the RNA that survived best, and used that to develop further variants of their initial protein.

In the next round of directed evolution, they raised the stakes by creating proteins that differed in every single position (that’s thousands of different versions of the same protein) and tested them against the RNA-degrading enzyme. They then heated the proteins to human body temperature and finally washed them with mouse blood. By studying the surviving RNA "genome," they were then able to create six new and improved proteins – which were again subjected to more evolution (this time using a live mouse as a test bed) to find new proteins that could last longest in the bloodstream.

Eventually their process led to a new capsid that, in its ability to bind RNA, could out-compete a naturally occurring virus that had been bioengineered. And even though the process was enabled by random chance, the researchers could trace back the evolutionary steps that converted a simple protein assembly (a "blank slate," in their words) into a potential drug transporter with some virus-like properties. 

This is a remarkable achievement, both in its immediate application for therapeutics, and because it outlines the possible evolutionary events that might have first formed viruses billions of years ago. The authors do acknowledge, however, that their creation is still a far cry from natural viruses. It cannot replicate, for instance, or even invade living cells the way a virus can. But it still represents one more piece in the puzzle behind how viruses, and by extension, their hosts, including us, came to be.

Conventional biological research is usually focused on investigating natural phenomena reductively, for example by paring down a life form to its basic parts, molecules like DNA and proteins. And throughout the last century, this curiosity-driven process has solved many of life's mysteries, with the discoveries of new technologies and drugs seeming like a happy bonus. Recently, however, new research in synthetic biology, like this latest attempt to manufacture a synthetic virus-like capsid, has turned this paradigm on its head. 

Some scientists are trying to understand the rules behind how human genomes are made by trying to build one from scratch. Others have uncovered new mechanisms controlling the color of flowers by making chrysanthemums blue. And, often, like the team from UW, synthetic biologists start out trying to make useful products and then stumble across key discoveries along the way. The hope is that perhaps one day we will finally work out how life evolved on a barren world by creating it anew in an empty test tube.

Peer Commentary

Feedback and follow-up from other members of our community

This is one of those research topics that I'm really happy people are working on, but I'm also really nervous about the implications. If we can design "dream" viruses, their potential isn't just for medicine – they could also be weaponized. Then again, if we want to defend against biological warfare with viruses, it's better to know which characteristics of these viruses are key to invading and propagating. So, let's just keep this information in the right (i.e. carefully gloved, morally strong) hands, eh?

Devang Mehta

Absolutely, there are biosecurity risks with this kind of research, but we have to be careful about overstating them. In this study, the ‘virus’ was not infectious and, in my opinion, that’s an acceptable limit. There are other cases where ‘gain of function’ experiments are of more concern (see this recent piece).

Any study that tracks the process of natural selection in a live “population” is really useful for our understanding of how these processes occur in nature. Another potential use for these data would be to measure the relative influence of certain types of mutations on fitness, because a lot of work so far in this field is theoretical, not empirical. 

For example, it would be interesting to see if mutations to certain codon positions or amino acid positions are more successful in this selective scenario. There are tons of other questions that come to mind – almost enough to make me want to go into the field of experimental evolution of viruses!

Devang Mehta

Couldn't agree more!

Daniel Bear

The last couple paragraphs really get at how I think science is beginning to change. In fields like synthetic biology or artificial intelligence, it’s kind of a put-up-or-shut-up moment: “if our theories of these horrifically complex phenomena are good, we should be able to at least take baby steps in creating systems with similar behaviors.” 

The more I see it, the more I have hope we’re moving out of the “everything is complicated so let’s collect more data” phase and into the model-testing phase. 

On the subject itself: I love how well this story illustrates the idea of evolution as algorithm. Only now instead of natural selection determining what the algorithm should optimize, we do. 

As Ashley said, having a way to optimize means we have to be extremely careful with the goals — since choosing those even slightly wrong could end up optimizing for unintended phenomena (from the standpoint of a benevolent scientist). Though in reality I’m not too worried about designer bioweapons.