Building a body is not simple. Fish, frogs and people all start from a single cell that becomes, seemingly against many odds, a highly organized, very complicated creature. Fertilized eggs split into two cells that become four, then eight, 16 and—within a matter of weeks—tens of thousands of cells. By this point the original spherical ball has rearranged itself into an elongated shape, bulging rounder and thicker at one end, with a shallow furrow running along its length. Soon another astonishing cellular ballet begins. The furrow deepens, and the cells that make up its walls begin to lean toward one another until they touch and stick together, forming a long, hollow tube that will eventually give rise to the brain at the bulging end and the spinal cord at the other.

To assemble so precisely, these and other cells in the embryo must sense where they are in relation to the rest of the organism. Each cell needs to know where an animal's front, back, top and bottom are located. Each cell also must figure out which direction is closer to or farther from the rest of the body. We and other developmental biologists have spent the past few decades trying to understand how this cellular orientation system works. As part of this larger quest, we have discovered a key component that contains several proteins that function together as a miniature compass within each cell. Without this compass, the heart, lungs, skin and other organs could not develop properly. In humans, when one of these proteins is altered by mutation, serious birth defects are the result.

Although there is much that we still do not understand about how this orientation system functions, what we have discovered so far sheds new light on fundamental processes of development across the animal kingdom. So far we have learned the most about how the compass works in epithelial cells, which typically cover a tissue surface like flagstones on a sidewalk, forming layers that are just one cell in thickness. If the cotton sheet on a bed were made up of epithelial cells, the proteins that we and others have found would allow any given cell in the sheet to sense which of its sides is closer to the head or the foot of the bed.

Organisms with cells that know where they are within the body benefit from a distinct evolutionary advantage: their complex tissues no longer need to be symmetrical in all directions; different parts can specialize. The hairlike cilia at one end of the cochlear duct of the ear, for example, distinguish high-frequency sounds; those at the other end detect low-frequency sounds. Scientists refer to the ensuing asymmetry of the tissue layer as planar polarity because opposing poles can be seen through the plane of the tissues.

Once animals invented a tool that worked, they stuck with it. Like the genes that code for many regulatory proteins, the genes that code for planar polarity proteins are very similar among evolutionarily distant species. For example, the versions present in mammals are quite similar to those in insects. Not surprisingly, these genes are also ancient—having evolved more than 500 million years ago with the rise of the animal kingdom.

Insects Lead the Way

Much of what we know about planar polarity stems from studies with insects that began in the mid-20th century. For convenience, these experiments focused on the easily accessible hard outer shell, or cuticle, found on most adult insects and not on internal organs. This hard outer layer is secreted by a layer of softer epidermal (skin) cells, which lies just underneath the cuticle.

When viewed through a microscope, the outer surface of the cuticle reveals a well-ordered landscape of ridges and scales dotted at regular intervals with hairs and bristles. Some of these protrusions are sensitive to changes in pressure or in the concentration of chemicals and thus help the creatures respond to their environment. Moreover, nearly every hair or bristle lines up in parallel with its nearest neighbors, so that all their tips tend to point in the same direction. On the wings, the hairs point away from the body. On the body itself, hairs and bristles point away from the head. Like the walls of a newly forming neural tube, these cells seem to know which way is back and front. They also appear to know which direction is closer to or farther from other tissues (proximal or distal, respectively).

Cells appear to share this directional information with one another, as demonstrated by Peter Lawrence of the University of Cambridge, the late Michael Locke of the University of Western Ontario and others in a series of pioneering experiments conducted more than 40 years ago. These scientists carefully cut out tiny squares of skin from the epidermal layer that gives rise to the exoskeleton in kissing bugs (the genus Rhodnius) and milkweed bugs (Oncopeletus). They then turned the squares 180 degrees and reimplanted them in the epidermis on the host insects' abdomen.

One might simply expect that the ridges or bristles on the cuticle that eventually formed from the turned-around graft to point in the opposite direction from the ridges or bristles surrounding it. But after the next molt, when the insects had shed their old exoskeleton and synthesized a new one, the researchers observed a striking change. Instead of lining up in opposite directions, the structures formed beautiful swirls across the borders of the transplanted square. The pattern of swirls suggested that neighboring cells had adjusted their orientations to minimize the differences between them. Clearly, the cells were able to communicate with one another about which direction their ridges and bristles should point. But how?

To reveal the underlying cellular and molecular machinery required a change in tactics—from surgical manipulations to a genetic approach. And when it comes to genetics, the best understood insect is the common fruit fly (Drosophila melanogaster), which has been studied in detail since 1910.

Starting in the 1980s, researchers, including one of us (Adler), began investigating tissue polarity in fruit flies. Our general approach was to identify and study mutant fruit flies with defects in the polarity system to deduce how it worked typically. We knew, for example, that the hairs on a Drosophila wing, like those on the abdomens of kissing and milkweed bugs, point in a uniform direction, in this case toward its farthest edge. Mutations in a gene called frizzled, however, made it look as though the fruit fly was having a bad hair day, with many hairs pointing in the wrong direction; changes in another gene, called dishevelled, caused a similar pattern, as its name suggests. This similarity was a clue that these different genes were part of a single system that controlled cell orientation.

Two groups—one led by David Gubb and Antonio García-Bellido, both then at the Autonomous University in Madrid, and the other by Adler—systematically studied how frizzled and dishevelled and other mutations affected the orientation of various parts of the fruit fly cuticle. Eventually we and others determined that in Drosophila six different genes code for proteins that serve as the key components of the polarity system. Two of these six, which Adler isolated in 1998, acted a lot like frizzled. Mutations in either of these genes resulted in a series of swirls that reminded him of the brushstrokes in Vincent van Gogh's paintings. So he named one gene van Gogh and the other starry night.

Another step in understanding the cellular basis for planar polarity in Drosophila came a few years earlier, when Lily Wong, then a graduate student in Adler's laboratory, examined developing wings to see how the array of hairs were formed and how mutations in tissue polarity genes altered that process. Wong found that each epithelial cell formed a hair at its most distal edge and that mutations that altered polarity were associated with a shift in the site of hair formation. This result led Wong and Adler to hypothesize that polarity proteins are part of a pathway that regulates the architecture of the cytoskeleton, the meshwork of polymerized proteins that controls cell shape and movement.

Charles R. Vinson, also then a graduate student in the Adler lab, demonstrated local cell-to-cell signaling by creating small patches of frizzled mutant cells during the development of an otherwise normal wing. The mutant cells caused neighboring nonmutant cells farther away from the body to reorient their hairs approximately 180 degrees so that the hairs pointed back toward the mutant patch. The orientations of nonmutant cells that were at a greater distance from the mutant patch remained unaffected. Vinson and Adler interpreted this result to mean that the polarity system controls cell orientation with short-range signals and that there may be no need for a precise signal distributed over long distances—as might occur with a chemical gradient—to determine proper orientation.

An Attractive Model

The idea that polarity proteins might regulate the formation of the cytoskeleton led various researchers to try to figure out exactly where in the cell these proteins are distributed. It turns out that the polarity proteins are not evenly distributed, and thus they can affect different sides of the cell in different ways. By 2005 Tadashi Uemura of Kyoto University in Japan, Jeffrey Axelrod of Stanford University, Marek Mlodzik of the Icahn School of Medicine at Mount Sinai, and David Strutt and Helen Strutt of the University of Sheffield in England had revealed a series of striking patterns. For example, in the single layer of cells that forms the surface of a fruit fly wing, van Gogh proteins accumulate predominantly on the side of each cell closest to the body. In contrast, frizzled proteins accumulate predominantly on the side closer to the end of the wing. Starry night proteins are found on both sides of each cell.

The asymmetrical patterns suggested to us and others a working model for how the directional system works. The model postulates two types of interactions between the van Gogh and frizzled proteins—one that attracts them toward one another, and a second one that repels them away from one another. Van Gogh proteins found on the side of a wing cell closest to the body, for example, appear to attract frizzled proteins on the adjacent surface of a neighboring cell. Meanwhile, within each cell, frizzled and van Gogh proteins repel one another, so that they end up on opposite sides of the cell. At present, we do not know the mechanisms of the hypothesized attractive and repulsive forces, and this remains an area of intense investigation.

To see how this model works to spread directional signals among a group of cells, imagine looking down at a sheet made up of many rows of cells with planar polarity proteins that are more or less randomly distributed within each cell. Now place, on the proximal side of the sheet, a new row of cells in which the proteins are not randomly distributed; instead the frizzled proteins are lined up on the distal side, and the van Gogh proteins are lined up on the proximal side of the cells. The model predicts that the attractive forces between the frizzled proteins in the new first row of cells and the otherwise randomly distributed van Gogh proteins in their now second-row neighbors would pull more of the van Gogh proteins over to the proximal surface of the second row of cells.

Any frizzled proteins in the second row, however, would then begin to gather on the distal side of the cells, away from the van Gogh proteins accumulating on the proximal side. As the frizzled proteins would gather on the distal side of the second row of cells, they would attract van Gogh proteins on the adjoining proximal surface of the third row. Thus, the asymmetrical pattern of tissue polarity proteins would spread from one row of cells to the next throughout the sheet.

This model is consistent with a large body of experimental data. In particular, the model predicts that the patterns of protein asymmetry should be extremely stable because any wayward cell—that is, a cell with an incorrect pattern of polarity protein accumulation—will be nudged back into the right orientation by signals from its proximal and distal neighbors. In this way, each cell creates its own compass, which defines its orientation and also influences the orientations of its neighbors.

Variations on a Theme

Insects, of course, are not the only animals that exhibit planar polarity. Inspired by the Drosophila experiments of Gubb and Adler, researchers (including Nathans) began looking for planar polarity genes in vertebrates. These experiments, and subsequent large-scale sequencing studies of various genomes, uncovered remarkably similar polarity genes throughout the animal kingdom. Interestingly, there appear to be no similar genes in plants, implying that the beautiful patterns of flowers and other plant organs are programmed by entirely different polarity systems.

For reasons that remain unclear, mammals have multiple versions of each Drosophila polarity gene. For example, humans and other mammals have three different starry night genes, whereas fruit flies have only one. Frizzled and dishevelled genes come in multiple copies as well.

Nathans has been particularly interested in teasing out the details of the planar polarity system in mammals. As with the earlier insect experiments, different structures within the skin—in this case, hairs—proved to be the best and most accessible place to start.

In contrast to the fly wing, where each cell produces one hair, each mammalian hair emerges from a follicle that is composed of dozens to hundreds of cells. Moreover, unlike the neighboring cells on an insect's wings, mammalian hair follicles do not touch one another directly; neighboring hair follicles are usually separated by many dozens of skin cells. Despite these differences in surface structures between insects and mammals, the results of eliminating polarity genes are quite similar. In 2004 Nino Guo, then a graduate student in Nathans's laboratory, used genetic engineering methods to eliminate the Frizzled6 gene in mice. Guo and Nathans were surprised to see that the hair follicles on the mutant mice were no longer parallel to one another but had reoriented to create a series of whorls reminiscent of the patterns seen on the mutant Drosophila wings.

Perhaps the biggest surprise occurred, however, when Nathans's lab started looking at how neurons in the mammalian brain are connected to one another. The major pathways in this complex network are laid down during embryonic development—as individual neurons send out axons (the “wires” that mediate long-range communication in the brain) that grow along predefined routes toward their targets. Nathans and his colleague, Yanshu Wang of Johns Hopkins University School of Medicine, found that Frizzled3 plays an essential role in guiding axons through the maze of embryonic neural tissue. When the researchers produced mice that lacked a Frizzled3 gene, the axons could no longer find their way and began following aberrant trajectories. Nathans's group then decided to test whether the Frizzled6 gene, which was so important to hair patterns, could take the place of Frizzled3, and vice versa. Using genetically engineered mice, the team found that Frizzled3 was fully capable of replacing Frizzled6, resulting in normal hair patterns. Yet Frizzled6 could partially but not fully replace Frizzled3 in directing the growth of axons. Thus, the polarity systems found in the skins and brains of mice are similar but not identical.

The resulting polarity systems play an important role in the existence of all vertebrates (including humans), from the earliest days of embryonic life to our every breathing moment, when the cilia in our airways propel any accumulating mucus in just one direction—up and out of the chest. As researchers obtain ever greater insights into the ways that individual cells sense their place in the body plan, we are continually amazed by the beauty of embryonic development. Among the many genetic changes that gave rise to the incredible diversity within the animal kingdom was a group of polarity-signaling genes. This set of genes—and their associated proteins—proved so successful over the past half a billion years that complex animals have used them ever since to solve a wide variety of evolutionary challenges.