Genetics on the wings

Natural History, Feb, 1997 by Sean Carroll

A few years ago, while visiting Duke University to deliver a talk on the genetics of fruit fly bristle patterns, I took a stroll across campus with butterfly expert Fred Nijhout. He asked me whether the sorts of genetic rules my colleagues and I were finding could perhaps explain the spectacularly diverse patterns found on butterfly wings. Intrigued by the question (but knowing next to nothing about butterflies), I read all I could over the next few months, including Nijhout's own outstanding book, The Development and Evolution of Butterfly Wing Patterns, which appeared soon after our encounter.

When I saw the beauty and diversity of these animals through his eyes, I was determined to try to answer Fred's question. With his help, my colleagues and I set up a colony of buckeyes--handsome brown Nymphalid butterflies with prominent eyespots on both fore and hind wings. For the last five years, I have been exploring the genetic "toolbox" of this and other species and, in the process, learning much about how evolution tinkers (to use molecular biologist Francois Jacob's evocative term), assembling new patterns from existing genetic materials.

Accustomed as I was to the colorless, drab wings of fruit flies, butterfly wing patterns seemed chaotic to me at first. They may have big spots, small spots, lots of stripes or no stripes, frilly borders, and different colors running in all directions. Fore- and hind-wing patterns frequently differ, as do top and bottom surfaces. As I became more familiar with them, however, I came to think of the wing markings as intricately detailed pointillist paintings. Tens of thousands of tiny scales, each one a single color and the product of a single cell, are the individual brush strokes that, when viewed at a distance, blend together in the whole design.

Earlier this century, scientists recognized that each butterfly wing pattern is a variation on a common theme. Every design is a composite of several discrete elements--spots, stripes, and borders--whose relative positions on the wing remain largely constant, even though the size, shape, color, and number of these elements vary independently of one another. Some of the thousands of living butterfly species--swallowtails, for example sport a rich assortment of design elements on their wings. Others--such as the sulphurs--have just a few. Butterfly wing markings have many functions, such as camouflage, mimicry, and perhaps attracting a mate. The role of eyespots appears to be to deceive predators: any bird about to snap up a juicy morsel is likely to pause when confronted by a startling flash of big "eyes" or at least to be confused and focus its attack on the wings, where damage is less serious than a direct hit on the soft body.

Every butterfly wing starts out as a flat disk of cells in the caterpillar. The disk grows during the caterpillar stage, and by the time the larva encloses itself in the chrysalis a blueprint of the future wing pattern has been drafted. Color development, however, takes place in the chrysalis. Fifteen years ago, Nijhout showed in an elegant series of microsurgery experiments that the position of eyespots in the buckeye is decided just before the caterpillar forms a chrysalis, while the colored rings around the eyespot are painted many days later, just before the adult emerges.

My hope was to identify some of the genes responsible for the eyespots and to catch them in action in the growing caterpillar wing disk. Many researchers, in my laboratory and elsewhere, have been involved in this work. One of our most important "partners" has been the tiny fruit fly. For the past eighty years or so, Drosophila has been the geneticist's workhorse, helping to solve universal questions about heredity and gene structure. (Fruit fly inspired work on animal development garnered a 1995 Nobel Prize for three pioneering geneticists.) Since these flies are distantly related to butterflies, my colleagues and I figured that some of the now well-known rules guiding wing formation in flies would also apply to butterflies and, if we were lucky, might lead us to genes responsible for the development of the colorful aspects of butterfly wings.

Using a collection of fruit fly genes as bait, my colleagues Jim Williams, Jane Selegue, Julie Gates, David Keys, and Grace Panganiban fished out corresponding genes from the thousands of genes in butterfly DNA. (Radioactively labeled fly genes can be used to locate and isolate their counterparts from butterfly DNA.) Finding a gene, however, is not the same as demonstrating its function. Using molecular probes marked with fluorescent chemicals and a high-powered microscope, we next looked to see which cells glowed when the probes stuck to the disks, revealing where genes were "turned on," or activated. Genes are activated in patterns: "on" in some cells; "off" in others. These patterns of gene activity are the earliest sign in animal tissue of future morphological changes. If any of our candidate genes had a role in determining butterfly wing patterns, we hoped this technique would enable us to catch them red-handed.