Randomness often does more harm than good. Take the phrase: "nothing in biology makes sense except in the light of evolution." I replaced a random word in this sentence with a random, different word, and got: "nothing in biology makes sense except in doctorishness light of evolution." Not only did this random replacement turn Dobzhanksy's famous phrase into a confusing statement on doctors and evolution, the entire sentence also became grammatically incorrect.

Random mutations in proteins have similar effects, except that they introduce molecular errors, rather than grammatical ones. A random amino acid that takes the place of an established one (amino acids are the building blocks of proteins, just like words are the building blocks of sentences) often distorts its local surroundings. Sometimes these small, local changes ripple through the protein, destabilizing its entire structure.

This doesn't mean that a mutated protein is lost beyond hope, however. In a recent paper, Joe Thornton and two colleagues describe an ancient, destabilizing mutation that paved the way for the evolution of a protein with new function. Some proteins first have to be broken, before they can be fixed.

The proteins that Thornton and his team studied were the glucocorticoid receptor and the mineralocorticoid receptor (GR and MR). Both these proteins are activated by steroid hormones and can switch other genes on or off. Their effects on the body are different, however. MR is responsible for maintaining the balance of water and salt in the body, while GR orchestrates the stress response. Amongst other things, activation of GR eventually leads to an increase in blood sugar and repression of the immune system. The two receptors also differ in their sensitivity to hormones. MR can be activated by a variety of different hormones at low doses, whereas GR only responds to high concentrations of the stress hormone cortisol.

GR and MR both evolved from the single ancestral gene AncCR (for 'Ancestral Corticosteroid Receptor'). This gene was duplicated 450 million years ago in the common ancestor of cartilaginous fish and bony vertebrates. Thornton reconstructed this extinct AncCR earlier and found that it was sensitive to a variety of steroid hormones, just like the modern MR. He concluded that after the initial duplication, MR was the gene that retained AncCR's original properties while GR was free to evolve a new function.

GRs in cartilaginous fish, such as the little skate depicted here, respond to multiple steroid hormones. They are not specific for cortisol, like our GRs are.

The evolutionary path that the GRs took after this duplication is unclear, although some circumstantial clues illuminate the way. In cartilaginous fish, GR still responds to the same broad set of hormones as MR and AncCR, although it is not as sensitive to them. Based on this finding Thornton reasoned that GR evolved in two distinct and independent steps: GR first became insensitive to hormones (in the common ancestor of bony animals and cartilaginous fish), before it evolved the specific response to cortisol (in the ancestor of bony vertebrates).

To test this scenario, Thornton and his team 'resurrected' the ancestral GR (AncGR) from the sequences of hundreds of GRs in modern animals. The activity of this reconstructed AncGR was similar to that of to the modern GR of cartilaginous fish: they both respond to the same hormones as AncCR, but only when they are present in high concentrations. This observation is in line with the two-step hypothesis.

Both MRs and GRs evolved after a single ancestral gene (AncCr) was duplicated. GR first became insensitive to most hormones (single rectangle) before becoming specific for cortisol in bony vertebrates (double rectangle).

The sequence of the reconstructed AncGR gave a clue as to how this reduced sensitivity evolved. Thornton and colleagues saw that AncGR had acquired 36 different mutations since it first parted ways with AncMR. To study how these mutations affected AncGR's sensitivity to hormones, the researchers engineered some of these mutations back into the original AncCR. Two mutations had particular dramatic effects. They decreased hormone sensitivity more than 100 times when they were introduced by themselves, and 10,000 times when they were present together.

The structure of AncGR revealed that these mutations didn't mess up the part of the protein that binds hormones, as you might expect. Instead, they destabilized the entire protein. One mutation (V43A) carved a hole in the otherwise tightly packed protein, whereas the other (R116H) disrupted molecular interactions that were present in the original protein. These changes are far from subtle. They degraded the original structure and function of the entire protein, much like the glaring grammatical error that destroys the meaning of a sentence. That AncGR remained functional at all was due to a different mutation (C71S), that partly buffered the effects of the other two.

While these two mutations are harmful at first glance, in hindsight you could say that they opened up a new evolutionary opportunity. After all, it was only because GR became insensitive to hormones that it could find a distinct niche for itself. It remains to be seen how common this 'bad mutation turns good'-scenario really is. After all, a mutation that had impaired GR too much would have eliminated all function, dooming the protein to decay over time. The balance between creation and destruction is a delicate one indeed.


Little skate by Andy Martinez

GR phylogeny from reference.


Carroll SM, Ortlund EA, & Thornton JW (2011). Mechanisms for the evolution of a derived function in the ancestral glucocorticoid receptor. PLoS genetics, 7 (6) PMID: 21698144