Head-on collisions of proteins create mutations
Ars Technica » Scientific Method 2013-03-28
One of the central tennets of evolutionary theory is that mutations are random—you can't predict what the next one will be, or when it's going to happen. But it also turns out that mutations are probabilistic. Some of them are a bit more or less likely, depending on the chemistry of the DNA base and its location in the genome.
Now, researchers have identified a mechanism that makes certain types of mutation more probable. This mechanism is a head-on collision between proteins that involves the complex that copies DNA when a cell divides. Because of the mechanics of these collisions, there's a distinct bias towards mutations occurring on one of the two strands of DNA that make up a double helix. The researchers found that this bias is so fundamental that bacterial genes are arranged to take advantage of it, so that some key genes are kept safer from mutations, while others that are key to adaptation can mutate more often.
The problem with collisions arises from the structure of DNA itself. The sugars in the molecule's backbone have a distinctive top (the 5' carbon) and bottom (the 3' carbon). Even in a molecule that's millions of sugars long, every single one of those is oriented the same way. If you move down the strand in one direction, you'll always hit the 5' end first (if you go in the other direction, you'll always hit the 3' end).