During sherry wine production, a thick, foamy layer of yeast develops on the surface of the wine. These "flor" yeasts are able to thrive at concentrations of ethanol that kill most other organisms, partly because they are able to float on the surface of the wine and thus gain access to the oxygen in the air above the wine:
(Look here for a neat little movie illustrating this phenomenon.) A fun paper published in the Proceedings of the National Academy of Sciences identifies the specific genetic changes that occurred when this feature of flor yeasts evolved. It's a fun paper because it's conceptually simple, it uses only basic techniques of molecular genetics, and it reports a nice illustration of how simple molecular changes can drive adaptive evolution. Simple changes like these most likely drive much adaptation and specialization of microorganisms.
How did flor yeasts gain the ability to float? Researchers from Seville, Spain, found that this ability is due primarily to changes in a single gene, called FLO11. FLO11 produces a protein that is anchored to the outside of the yeast cell. This protein is heavily glycosylated, that is, covered with various sugar chains, and it affects the physical characteristics of the cell surface. Yeast cells regulate the production of Flo11 proteins in response to different nutrient environments, and thereby enter different growth modes that allow the yeast to make the best of the current environment. The Flo11 protein is similar to human proteins called mucins - these mucins seem to play a role in the ability of cancer cells to invade other tissues, essentially enabling cancer cells to adapt to changing nutrient conditions in their environment, much like yeast.
The Spanish group found that flor version of the FLO11 gene, when placed in a non-floating yeast strain, gave that yeast strain the ability to float. They also found that if they deleted FLO11 from the flor strain, that strain could no longer float. Here is a picture of a non-floating and a floating strain (circled in red):
Interestingly the FLO11 gene has one of the longest promoters, or control regions, of any yeast gene. A gene's promoter lies upstream of the gene itself, and this is where various proteins attach themselves to the DNA and regulate how much the gene gets expressed. In the case of FLO11, the long promoter means that this gene is regulated by many different signals, each one leading to some combination of proteins that sit on the DNA and regulate how much Flo11 is made.
The researchers in Seville found that if you took the flor version of just the promoter and put it in front of the FLO11 gene in a non-floating strain, that strain could then float. It turns out that the flor version of the promoter really cranks up the expression of the FLO11 gene, and this alone can confer some ability to float. The flor version of the promoter is missing one of the "off switches" found in the non-floating version - this could explain why the FLO11 gene gets cranked on so much.
That's not all there is to the story. There are also differences in the FLO11 gene itself - certain regions have expanded, allowing the final protein to be more heavily coated in sugar chains. This change in the gene alone can't confer the ability to float, but when combined with the flor version of the promoter, it produces a better floating ability than just the flor promoter alone. Flor yeasts produce more Flo11 protein, and the Flo11 they make has more sugar chains attached. This changes the surface properties of the cell and enables it to float. (In the figure below, each colored box represents a stretch of DNA.)
So what's the evolutionary story here? Inside a barrel of sherry, life is hard for yeast. Alcohol levels are high (after fermentation, sherry is fortified with brandy to about 15% alcohol), and available fermentable sugar is low. A good place to be is on top of the liquid, where yeast have more access to oxygen and can thus more efficiently metabolize what sugars are available. A few hundred years ago, a yeast in a non-floating strain lost one of its "off switches" in its FLO11 promoter through a random mutational event. (The molecular details of how such a random deletion event can occur are fairly well known.) That random mutation alone was enough to allow this yeast to float to the surface, where it was able to reproduce much better than its non-floating relatives. Subsequent changes in the FLO11 gene improved on this ability to float. This type of evolutionary scenario is probably typical of much of the adaptive evolution that occurs among mircoorganisms. Over millions and billions of years of mutation and selection in changing environments, this adds up to a lot of evolutionary change.