Jaws of the Fates

In his poem about the lonely “Maldive Shark,” [HN1] Herman Melville describes the daunting jaws of a serious meat-eater, which serve as an asylum for the sleek little pilot fish [HN2], azure and slim, hiding in his “jaws of the Fates.” The “jaws of the Fates” may act rather unpredictably in the uncharted oblivion of the Indian Ocean, but Depew et al. [HN3] (1) report a remarkable catch—the genes that dictate the fates of jaws—on page 381 of this issue. These authors turn lower jaws into upper jaws by simultaneously inactivating the homeobox genes Dlx5 and Dlx6 [HN4] of mice. Such a spectacular transformation of “jaw identity” unveils a family of genes that are crucial for directing formation of the vertebrate face. This gene family may have been subject to profound modifications during vertebrate evolution and in certain human congenital diseases [HN5]. The Dlx5 and Dlx6 genes are now implicated in the elaboration of vertebrate lower jaws, from the ferocious feeding machinery of the great white shark to the sophisticated hearing system of mammals.

A complex series of cellular and molecular interactions underlies the assembly of the vertebrate face [HN6]. Most structures are formed by the neural crest [HN7], a tissue that emanates from the early embryonic brain and populates the so-called branchial arches [HN8] (2). Branchial arches are a segmental series of bulges in the embryonic head and are predecessors of all facial elements. Within the first (mandibular) branchial arch, jaw elements develop from three bulges: the mandibular, maxillary, and frontonasal processes that are filled by neural crest cells from different origins (midbrain and hindbrain) (3). Widespread mixing between them supports the notion that jaw neural crest is exposed to instructive signals from its environment that establish a proximodistal axis [HN9] to the jaw-forming branchial arch (3). Such external cues are translated into a code of neural crest proximodistal “identity,” leading to the precisely orchestrated formation of skeletal and muscular elements. Much evidence implicates the Hox homeobox genes [HN10] as the encoders of “rostrocaudal identity” in all branchial arches posterior to the jaw-forming arch. However, until now no genes have been proven to act as true selector genes for proximodistal identity of neural crest cells in branchial arches.

• Download high-res image
• Open in new tab
• Download Powerpoint

Jawed vertebrates have three pairs of Dlx homeobox genes—Dlx1/2, Dlx5/6, and Dlx3/7—that are expressed in restricted domains across the proximodistal axis of the branchial arches (4). Their nested expression within the branchial arches and the fact that their Drosophila homolog distalless [HN11] is a master regulator of distal leg identity make the Dlx genes excellent candidates for encoding distal identity in vertebrates. In all bilateral organisms, distalless genes appear to be involved in controlling the outgrowth of body appendages (5). Thus, the idea that the vertebrate Dlx homologs serve a similar function is attractive. Rather disappointingly, mice missing a single Dlx gene exhibit only piecemeal changes in the identities of isolated skeletal elements and teeth. This finding suggested that Dlx genes act as “micromanagers” rather than as “master regulators.” Now, Depew and colleagues report the striking phenotype of the Dlx5/6 double mutant mouse [HN12] (1). They provide evidence that Dlx5 and Dlx6 are indeed the selectors of distal branchial arch identity. Their work suggests that the absence of a clear phenotype in mice lacking one Dlx gene is due to compensation by other coexpressed Dlx genes. Thanks to their discovery, the concept of a proximodistal molecular identity code is alive and well.

The phenotype of the Dlx5/6 double mutant mouse is complex but remarkably clearcut. All of the skeletal elements below the primary jaw joint (the joint between the malleus and incus of the mammalian middle ear [HN13]) are missing. Even more intriguing, in the region of the lower jaw, the mutant mice possessed a second complete set of bona fide upper jaw elements. During the early evolution of mammals, the major upper jaw element (the so-called palatoquadrate) became fragmented. Parts of the palatoquadrate became fused to the braincase—the “tennis racket”-shaped alisphenoid bone in figure 3E and supplementary figure 3D of the Depew et al. paper—or gave rise to elements of the mammalian pharynx (pterygoid). Other parts such as the quadrate turned into the mammalian incus, and the hyomandibula became the third middle ear bone (the stapes)—all elements of a new hearing apparatus (6). [HN14] In the Dlx5/6 double mutant mice all of these elements are duplicated, resulting in a symmetrical instead of an asymmetrical mouth.

If Dlx5 and Dlx6 are distal selector genes, then where does the initial patterning information for the branchial arch proximodistal axis come from? A recent elegant paper by Couly et al. [HN15] sheds light on this difficult question (7). By transplanting pharyngeal endoderm [HN16] into different locations in the developing chick head, Couly et al. generated jaw duplications that are remarkably similar to those obtained by Depew et al. (1). A growing body of evidence suggests that initial cues from the pharyngeal endoderm impose a first proximodistal patterning axis onto the adjacent branchial arch neural crest cells. This prepattern is then “interpreted” by neural crest cells, resulting in the nested expression of Dlx gene pairs, Dlx5/6 and Dlx3/7, by these cells. If this is the case, what does the Dlx5/6 mutant phenotype tell us about the evolution of jaws?

Although upper and lower jaw elements have never been completely symmetrical during vertebrate history, early jawed vertebrates related to the ancestors of bony fish (such as acanthodians [HN17]) experimented with the shape and symmetry of their jaws. Acanthodians, for example, sometimes displayed remarkable symmetry between their upper and lower jaw elements (see the figure, part A, green) as well as in their dentition (see the figure, part B, yellow) (8). Such symmetrical features, which disappeared later in evolutionary history, now reappear in the Dlx5/Dlx6 double mutant mice.

The Depew et al. work suggests that lower jaw patterning that is dependent on Dlx5/6 expression may have been elaborated and embellished between the phylogenetic nodes of jawed vertebrate and bony fish ancestors. Going back one step further in evolutionary history, jawless vertebrates such as lampreys [HN18] only have four Dlx genes with unclear homologies to their jawed vertebrate counterparts (9). All lamprey Dlx genes are expressed in branchial arches, but a nested expression pattern appears to be the invention of the jawed vertebrates (10). Our knowledge of the enhancer organization that controls this nested Dlx gene expression in jawed vertebrates is still rudimentary. Comparative genomic and functional studies of the regulatory elements controlling Dlx gene expression in lampreys, sharks, bony fish, coelacanths, and tetrapods will reveal the molecular evolution of the proximodistal code that underlies the shapes and fates of jaws.