A biophysicist learns the art of hugging.

If there are two things I love, they are warm hugs and simple answers to long-standing questions. Why must proteins bend in order to bind to their partners? This bending, known as induced fit, is puzzling, because it costs elastic energy and makes the binding less tight. But all sorts of proteins show induced fit — such as antibodies that recognize viruses and regulatory proteins that embrace DNA — despite the fact that such processes would be more efficient if the protein and target fitted together like pieces of a jigsaw puzzle.

Enter two Israeli physicists, Tsvi Tlusty and Yonatan Savir. They used statistical mechanics to show that bending is a good idea if the goal is not to bind tightly but to avoid binding the wrong partner (Y. Savir and T. Tlusty PLoS ONE 2, e468; 2007).

Suppose that a protein needs to bind its target molecule 'A', and to avoid binding molecule 'B', which is a bit smaller than A but otherwise similar in shape. The protein would do well to make its binding pocket a little larger than A; it would then have to bend a little to embrace A, at a small energy cost, but bend a lot to bind B. Crucially, the elastic energy required for a protein to flex rises ever more steeply with the extent of bending, just as that needed for a spring to bend rises with the square of bending. So the energy difference between binding A and binding B is greater with induced fit than it is when A is a perfect fit.

Like all fruitful theories, this one makes testable predictions. It allows researchers to hypothesize what sort of imperfect fit might best serve a particular protein so that binding to non-target molecules is minimized. An antibody that attaches to virus proteins should be able to avoid similarly shaped human proteins, for instance. So perhaps, as with people, if you really want to know whether proteins are made for each other, it's in the hug.

Discuss this paper at http://blogs.nature.com/nature/journalclub