Local Fitness Landscape Mapped Out For Green Fluorescent Protein

It’s Just Getting Worse

As we have discussed many times, proteins are a show-stopper for evolution. Proteins consist of dozens, hundreds and even thousands of amino acids and, like most machines, they don’t work very well until most of the parts (amino acids in this case) are in place. Half of the amino acids don’t give you half the function of a protein. You can read more about this here and here. Now, a new paper reinforces the problem of protein evolution.

One approach to studying how evolution could create new protein designs is to start with some sort of random sequence of amino acids, see how well it works, and try to evolve it to obtain a protein. This is difficult because the protein design space is astronomically huge and proteins are sparse within that space. Any random sequence of amino acids will merely give you junk. Furthermore, the fitness landscape is flat and doesn’t provide the guidance evolution needs to move toward functional proteins.

Another approach is to start at the end and work backwards. In other words, start with the finished product—a functional protein—and see what the fitness landscape looks like as you swap in different amino acids. This is difficult because, unfortunately for evolution, the fitness landscape drops off precipitously as you move away from the native protein design. Modifying only a few percent of the amino acids leads to a rapid loss of function.

The new paper takes this second approach. It uses a bioluminescent protein known as the green fluorescent protein, taken from the jellyfish, Aequorea victoria. It is a wonderful study that systematically mapped out the protein’s function (as measured by the protein’s fluorescence) for a total of 51,715 different protein sequences that are nearby the native sequence.

The results confirmed what earlier studies had indicated: the protein function drops off dramatically with only a relatively small number of substitutions. But the study also explored the effect of multiple substitutions. It is well known that the effect of two substitutions, for example, are not always simply the sum of their individual effects. They can interact with each other in either positive or negative ways. This is referred to as epistasis.

The new study found that negative epistasis was strong and prevalent. As one of the researchers explained:

We were really surprised when we finally had a chance to look at exactly how the interactions between mutations occur. We also did not expect that almost all the mutations that are only slightly damaging on their own can destroy fluorescence completely when combined together.

It was well understood that evolving a protein is an astronomically unlikely event, and these results indicate it is even more difficult. Those negative results, however, were not reported in the paper. Instead, the paper discussed possible ways that one green fluorescent protein, found in one particular species, may have evolved into other green fluorescent proteins, found in other species. The implications for the initial evolution of a protein were ignored.