Apical Ectodermal Ridge

The earliest limb bud begins to form before an AER is present, but soon a thickened AER appears along the border between dorsal and ventral limb ectoderm. Molecular studies have shown that the position of the AER corresponds exactly to the border between dorsal ectoderm, which expresses the signaling molecule radical fringe, and ventral ectoderm, which expresses the transcription factor Engrailed-1 (En-1) (see Fig. 10.17A).

Although the AER had been recognized for years, its role in limb development was not understood until it was subjected to experimental analysis. Removal of the AER results in an arrest of limb development, thus leading to distal truncation of the limb (Fig. 10.8). In the limbless mutant in chickens, early limb development is normal; later, the AER disappears, and further limb development ceases. If mutant ectoderm is placed over normal limb bud mesoderm, limb development is truncated, whereas combining mutant mesoderm with normal ectoderm results in more normal limb development. These findings suggest that the ectoderm is defective in this mutant.

Further analysis has shown that in the limbless mutant, the entire ectoderm of the limb bud displays a dorsal character; that is, radical fringe and other “dorsal” molecules are expressed in dorsal and ventral ectoderm. Correspondingly, En-1 is not expressed in ventral ectoderm. In the absence of the juxtaposition of ectoderm with dorsal and ventral properties, an AER cannot be maintained.

The power of the AER is shown by experiments or mutants that result in the formation of two AERs on the limb bud. This situation leads to a supernumerary limb, as is illustrated by the mutants eudiplopodia in chickens and diplopodia in humans (Fig. 10.9).

The outgrowth-promoting signal produced by the AER is FGF. In the earliest stages of limb formation, the lateral ectoderm begins to produce FGF-8 as it thickens to form an AER. As the limb bud begins to grow out, the apical ridge also produces FGF-4, FGF-9, and FGF-17 in its posterior half. If the AER is removed, outgrowth of limb bud mesoderm can be supported by the local application of FGFs. Other studies have shown that in mutants characterized by deficient or absent outgrowth of the limb, the mutant ectoderm fails to produce FGF. The effects of the FGF produced by the apical ectoderm on the underlying mesoderm are discussed later in this chapter.

The mesoderm of the early limb bud consists of homogeneous mesenchymal cells supplied by a well-developed vascular network. The mesenchymal cells are embedded in a matrix consisting of a loose meshwork of collagen fibers and ground substance, with hyaluronic acid and glycoproteins prominent constituents of the latter. There are no nerves in the early limb bud.

It is impossible to distinguish different cell types within the early limb bud mesenchyme by their morphology alone. Nevertheless, mesenchymal cells from several sources are present (Fig. 10.10). Initially, the limb bud mesenchyme consists exclusively of cells derived from the lateral plate mesoderm. These cells give rise to the skeleton, connective tissue, and some blood vessels. Mesenchymal cells derived from the somites migrate into the limb bud as precursors of muscle and endothelial cells. Another population of migrating cells is that from the neural crest; these cells ultimately form the Schwann cells of the nerves, sensory nerves, and pigment cells (melanocytes).

Mesodermal-Ectodermal Interactions and the Role of Mesoderm in Limb Morphogenesis

Limb development occurs as the result of continuous interactions between the mesodermal and ectodermal components of the limb bud. The apical ectoderm stimulates outgrowth of the limb bud by promoting mitosis and preventing differentiation of the distal mesodermal cells of the limb bud. Although the AER promotes outgrowth, its own existence is reciprocally controlled by the mesoderm. If an AER from an old limb bud is transplanted onto the mesoderm of a young wing bud, the limb grows normally until morphogenesis is complete. If old limb bud mesoderm is covered by young apical ectoderm, however, limb development ceases at a time appropriate for the age of the mesoderm and not that of the ectoderm.

Similar reciprocal transplantation experiments have been used to show that the overall shape of the limb is determined by the mesoderm and not the ectoderm. This is most dramatically represented by experiments done on birds because of the great differences in morphology between the extremities. If leg bud mesoderm in the chick embryo is covered with wing bud ectoderm, a normal leg covered with scales develops. In a more complex example, if chick leg bud ectoderm is placed over duck wing bud mesoderm, a duck wing covered with chicken feathers forms. Such experiments, which have sometimes involved mosaics of avian and mammalian limb bud components, show that the overall morphology of the limb is determined by the mesodermal component and not the ectoderm. In addition, the regional characteristics of ectodermal appendages (e.g., scalp hair versus body hair in the case of mammals) are also dictated by the mesoderm. Cross-species grafting experiments show, however, that the nature of the ectodermal appendages formed (e.g., hair versus feathers) is appropriate for the species from which the ectoderm was derived.

Polydactyly is a condition characterized by supernumerary digits and exists as a mutant in birds. Reciprocal transplantation experiments between mesoderm and ectoderm have shown that the defect is inherent in the mesoderm and not the ectoderm. Polydactyly in humans (Fig. 10.11) is typically inherited as a genetic recessive trait and is commonly found in populations such as certain Amish communities in the United States in which the total genetic pool is relatively restricted (see Clinical Correlation 10.1 for further details).

Zone of Polarizing Activity and Morphogenetic Signaling

During experiments investigating programmed cell death in the avian limb bud, researchers grafted mesodermal cells from the posterior base of the avian wing bud into the anterior margin. This manipulation resulted in the formation of a supernumerary wing, which was a mirror image of the normal wing (Fig. 10.12). Much subsequent experimentation has shown that this posterior region, called the zone of polarizing activity (ZPA), acts as a signaling center along the anteroposterior axis of the limb. The signal itself has been shown to be sonic hedgehog (shh) (see Fig. 10.16), a molecule that mediates a wide variety of tissue interactions in the embryo (see Table 4.4). As seen later in this chapter, shh not only organizes tissues along the anteroposterior axis, but also maintains the structure and function of the AER. In the absence of the ZPA or shh, the apical ridge regresses.

Cross-species grafting experiments have shown that mammalian (including human) limb buds also contain a functional ZPA. A transplanted ZPA acts on the AER and elicits a growth response from the mesenchymal cells just beneath the part of the ridge adjacent to the transplanted ZPA. As few as 50 cells from the ZPA can stimulate supernumerary limb formation. Other structures, such as pieces of Hensen’s node, notochord, and even feather germs have been shown to stimulate the formation of supernumerary limbs if grafted into the anterior margin of the limb. Since these experiments were conducted, all the effective implanted tissues have been found to be sources of shh.

The ZPA is already established by the time the limb bud begins to grow out from the body wall. There is evidence that in the forelimb the position of the ZPA is determined by the location of the highest concentration of Hoxb8 expression along the body axis. Experiments have shown that in response to the localized application of retinoic acid along the anterior margin of the forelimb bud, Hoxb8 expression is induced within 30 minutes. This suggests a cascade beginning with retinoic acid signaling, leading to Hoxb8 expression, which determines the location of the ZPA.

Shh induces the expression of the signaling molecule gremlin, which has two inhibitory functions. Gremlin inhibits the action of mesodermal bone morphogenetic protein-2 (BMP-2), which in itself inhibits the expression of FGF-4 in the AER. Such an inhibition of a BMP inhibitor is reminiscent of the sequence of events involved in primary neural induction (see p. 84). In contrast, gremlin, which is localized in the posterior part of the limb bud, inhibits the action of Gli-3 so that Gli-3 functions only in the anterior part. Within the anterior part of the limb bud, Gli-3 inhibits the expression of shh. In Gli-3 mutants, shh becomes expressed ectopically in the anterior limb bud, and preaxial polydactyly results.

As the limb bud elongates, the ZPA becomes translocated more distally, and it becomes surrounded by an increasingly large zone of formerly shh-producing cells that were derived from the ZPA. Later, these cells become heavily involved in the formation of digits and in events leading to the termination of limb development.

Molecular Signals in Limb Development

As discussed earlier, initial development of the limbs involves the establishment of a limb field by the actions of a Hox gene combinatorial code that acts through yet unidentified axial signals to stimulate the expression of Tbx5 in the area of the future forelimb and Tbx4 in the hindlimb. Even in later development, Tbx5 is expressed exclusively in the forelimb, and Tbx4 is expressed exclusively in the hindlimb (Fig. 10.15). Because of these exclusive expression regions, it was originally assumed that these two genes determined the identity of the forelimb and hindlimb. More recent research has shown this not to be the case, however, and the search for the factors that determine limb identity continues. Pitx-1, which is also expressed in the hindlimb, may play a more important role than Tbx-4 as a determinant of hindlimb identity. The main functions of Tbx-4 and Tbx-5 appear to be the initiation of development in a limb-specific manner.

When the limb bud takes shape, its further development depends to a great extent on the actions of three signaling centers, one for each of the cardinal axes of the limb. As already discussed, outgrowth along the proximodistal axis is largely under the control of the apical ectodermal ridge and the FGFs that it produces. FGF-8 is produced along the entire length of the AER, and FGF-4 is produced only along the posterior half. FGF-4, in particular, is an integral part of a feedback loop linking the growth center in the AER to that of the ZPA.

The second major signaling center, this time along the anteroposterior axis, is the ZPA, and the signaling molecule is shh (Fig. 10.16). Although shh is a diffusible molecule, it functions through its effects on BMP-2 and the inhibitor of BMP-2, gremlin (Fig. 10.17). Gremlin has two major functions. First, it antagonizes Gli-3, confining Gli-3 activity to the anterior part of the limb bud, where it represses the expression of posterior limb genes. As mentioned earlier, gremlin also inhibits the inhibitory action of BMP-4 on the AER, thus promoting the activity of FGF-4. FGF-4 is necessary for maintaining the activity of the ZPA.

Organization of the dorsoventral axis of the limb begins when the dorsal ectoderm produces the signaling molecule, Wnt-7a, which stimulates the underlying limb bud mesenchyme to express the transcription factor, Lmx-1b, a molecule that imparts a dorsal character to the mesoderm underlying the dorsal ectoderm. Ventral ectoderm produces En-1, which represses the formation of Wnt-7a and consequently the formation of Lmx-1b in what will become ventral limb mesoderm, possibly by a default pathway (see Fig. 10.17A). The AER marks the border between dorsal and ventral limb bud ectoderm, and this border is characterized by an expression boundary between another signaling factor, radical fringe, secreted by dorsal ectoderm, and the En-1 formed in the ventral ectoderm. Further ventral spread of radical fringe expression is held in check by En-1.

All three axial signaling centers (Table 10.1) interact in the early limb bud. Wnt-7a from the dorsal ectoderm has a stimulating effect on the ZPA (see Fig. 10.17B), whereas shh from the ZPA is required for the production of FGFs from the AER that provide additional positive feedback to the ZPA.

Simultaneous with the establishment of the ZPA, an orderly sequence of the homeobox-containing genes Hoxd9 to Hoxd13 (Fig. 10.18) and certain of the Hoxa genes occurs in the early limb bud. This sequence represents a second wave of Hox gene expression after that involved in the initiation of limb development. Shh stimulates the expression of the Hox genes in the limb, and Gli-3 is involved in confining Hox gene expression to the more posterior parts of the limb bud. The Hox genes are involved in patterning the proximodistal axis of the limb (Fig. 10.19). Studies on mice and the analysis of certain human mutants have shown that certain defects in limb regions correspond to absent expression of specific Hox gene paralogues. For example, mutations of Hoxa 13 and Hoxd13 cause characteristic reduction defects of the digits that result from shortness of the phalanges (Fig. 10.20).

An interesting, but little explored question in the development of many structures is what causes development to cease. In the case of the limb, the answer may lie in the relationship between shh-producing cells in the ZPA and the expression of gremlin, which depends on exposure to shh. As the limb develops, a zone of cells that had produced shh, but no longer do, forms around the ZPA (Fig. 10.21). These cells are not themselves able to produce gremlin. As more of these cells accumulate, the distance between the shh-producing cells of the ZPA and the cells that can express gremlin increases, ultimately to the point at which these cells no longer receive a sufficient stimulus to produce gremlin. When this happens, the gremlin-based maintenance of FGF-4 production of the AER ceases, and the entire feedback system between the ZPA and AER winds down; this results in cessation of limb development. If the intervening wedge of formerly shh-producing cells is removed from the distal tip of the mature limb bud, the more anteriorly located mesodermal cells are again exposed to above-threshold concentrations of shh, and they can again produce gremlin. This reconstitutes the ZPA-AER axis through a regulative mechanism, and limb development continues past the point at which it usually ceases. The result is the formation of digits with more than the normal number of phalangeal segments.

Cell Death and Development of Digits

Although it may seem paradoxical, genetically programmed cell death (apoptosis) is important in the development of many structures in the body. In the forelimb, it is prominently manifested in the anterior limb margin, in the future axillary region, between the radius and ulna, and in the interdigital spaces (Fig. 10.22). Experiments on avian embryos showed that, to a certain stage, mesodermal cells scheduled to die could be spared by transplanting them to areas in which cell death did not normally occur. After a certain time, however, a “death clock” was set (an example of determination), and the cells could no longer be rescued.

As limb development proceeds, changes become apparent in the AER. Instead of remaining continuous around the entire apex of the limb, the ridge begins to break up, leaving intact segments of thickened ridge epithelium covering the emerging digital rays (cartilaginous models of the digital bones). Between the digits, the ridge regresses (see Fig. 10.22A). As the digital primordia continue to grow outward, cell death sculpts the interdigital spaces (see Fig. 10.22C). BMP-2, BMP-4, and BMP-7 and the transcription factors Msx-1 and Msx-2 are strongly expressed in the interdigital spaces. The exact mechanism of interdigital cell death is still unclear, but the BMPs, especially BMP-4 acting under the mediation of Msx-2, are the prime movers in initiating interdigital cell death. The FGFs produced by the AER seem to play a dual role in interdigital cell death. Although FGF-2 antagonizes the death-inducing effects of the BMPs, FGFs promote the production of Msx-2, which cooperates with the BMPs in inducing interdigital cell death.

If interdigital cell death does not occur, a soft tissue web connects the digits on either side. This is the basis for the development of webbed feet on ducks and the abnormal formation of syndactyly (Fig. 10.23A) in humans. BMP is not found in the interdigital mesoderm in developing duck feet, although it is found in other regions of cell death in the duck limb.

There is more, however, to the development of digits than simply sculpting the interdigital spaces by cell death. Well before cell death becomes evident, other events specify the nature of each digit. A future digit is first recognizable as a longitudinal condensation of mesenchyme, which soon begins to lay down a precartilaginous matrix. The early digital ray then undergoes segmentation (see Fig. 10.26) to form specific phalangeal segments. Each digit develops its own character, as determined by the number of phalangeal segments or its specific size and shape. The underlying basis for the development of digital form is just becoming understood.

How individual digits form has long remained a mystery, but new research findings are beginning to clarify some aspects of the process. It is now evident that the identity of individual digits is not fixed until relatively late in limb formation. The driving force for the specification of most digits is shh. The exception to this rule is the first digit (thumb), which forms even in shh^−/− mutants. The identity of the remainder of the digits is determined by the concentration and the duration of exposure of their cells to shh. Digit 2 is formed from cells that have been exposed to shh, but have not themselves produced this signaling factor. Digits 3 to 5 arise from cells that have produced shh. Digit 3 is actually a hybrid. The anterior half consists of cells that have been exposed to, but have not produced shh, whereas the posterior half of that digit is composed of shh-producing cells that have been exposed to shh for the shortest amount of time. A longer period of exposure and a greater concentration of shh is required to form digit 4, and digit 5 requires the longest exposure time and concentration of shh. Growth of individual digital primordia is maintained by the production of FGF-8 by the remnants of the AER overlying the tips of the digital primordia, while BMP-mediated cell death is occurring in the interdigital mesenchyme.

All human digits contain three phalangeal segments except for the first digits (thumb and great toe), which consist of only two segments. Rarely, an individual is born with a triphalangeal thumb (see Fig. 10.23B). Why digits have different numbers of phalangeal segments is still not understood.

Skeleton

The skeleton is the first major tissue of the limb to show signs of overt differentiation. Its gross morphology, whether normal or abnormal, closely reflects the major pattern-forming events that shape the limb as a whole. Formation of the skeleton can be first seen as a condensation of mesenchymal cells in the central core of the proximal part of the limb bud. Even before undergoing condensation, these cells are determined to form cartilage, and if they are transplanted to other sites or into culture, they differentiate only into cartilage. Other mesenchymal cells that would normally form connective tissue retain the capacity to differentiate into cartilage if they are transplanted into the central region of the limb bud.

The ectoderm of the limb bud exerts an inhibitory effect on cartilage differentiation, so cartilage does not form in the region just beneath the ectoderm. On the dorsal side of the limb bud, mesenchymal cells are prevented from differentiating into cartilage by Wnt-7a, produced by the ectoderm.

The condensed cells that make up the precartilaginous aggregates express BMP-2 and BMP-4. As skeletal development continues, their expression becomes progressively restricted to the cells that become the perichondrium or periosteum surrounding the bones. BMP-3 transcripts are first seen in cartilage, rather than precartilage, but this growth factor also ultimately becomes located in the perichondrium. The translocation of expression of these BMP molecules to the perichondrium reflects their continuing role in the earliest phases of differentiation of skeletal tissues.

In contrast, BMP-6 is expressed only in areas of maturing (hypertrophying) cartilage within the limb bones. Indian hedgehog, a molecule related to shh, is also expressed in the same regions of hypertrophying cartilage (which is also marked by the presence of type X collagen), and this signaling molecule may induce the expression of BMP-6.

Differentiation of the cartilaginous skeleton occurs in a proximodistal sequence, and in mammals the postaxial structures of the distal limb segments differentiate before the preaxial structures. For example, the sequence of formation of the digits is from the fifth to the first (Fig. 10.24). The postaxial skeleton of the arm is considered to be the humerus, ulna, digits 2 through 5, and their corresponding carpal and metacarpal elements. The preaxial portion of the limb bud becomes progressively reduced during limb outgrowth and contributes only to the radius and possibly the first digital ray. Certain limb defects, sometimes called hemimelias, are characterized by deficiencies of preaxial or postaxial limb components (Fig. 10.25).

The development of the limb girdles remains incompletely investigated, but experimental work on the chick has shown that the blade of the scapula is derived from cells of the dermomyotome, whereas the remainder of the scapula arises from lateral plate mesoderm. The three bones of the pelvis all arise from lateral plate mesoderm, with no known contribution from the somites. Each of the bones of the pelvis, as well as the two developmentally different components of the scapula, is characterized by a different molecular signature. How the bones of the appendages are patterned to connect with their respective girdles is still poorly understood, but studies of mutants suggest that the transcription factors Pbx-1 and Pbx-2 play an important upstream role.

A characteristic feature in differentiation of the limb skeleton is the formation of joints. Joint formation occurs by the transverse splitting of precartilaginous rods, rather than by the apposition of two separate skeletal elements. Joint formation is first apparent when transverse strips of highly condensed cells cross a precartilaginous rod (Fig. 10.26). Formation of the zone of cell density is induced by Wnt-14, which stimulates the formation of growth differentiation factor-5, a member of the BMP family, in the region of the future joint. BMP activity, which is strongly involved in cartilage formation, must be excluded from the region of the developing joint. Noggin, an antagonist of BMP, plays an important role in joint formation because in its absence, BMP is expressed throughout the region where the joint should form, and the digital rays develop into solid rods of cartilage without joints. The roles of noggin and BMP in joint formation are very similar to those seen in the formation of the sutures between the cranial bones (see p. 177).

Condensation is followed by cell death in the interphalangeal joint regions and hyaluronin secretion and matrix changes in the region of the future joint. Then the skeletal elements on either side of the joint form articular cartilage, and a fluid-filled gap is created between them. Additional condensations of mesenchymal cells form the joint capsule, ligaments, and tendons. During later development, muscular activity is required to maintain the integrity of the joint, but early joint development is completely independent of muscular activity. A well-known mutant family, called brachypodism, involves a shortening of the limb and the lack of development of certain joints, specifically the interphalangeal joints. There are five major groups of brachydactylies, each of which contains several subtypes.

Musculature

The musculature of the limb is derived from myogenic cells that migrate into the very early limb bud from the ventral part of the dermomyotome of the somite. Each somite in the limb region contributes 30 to 100 migratory precursor cells to the future limb musculature. These cells are stimulated to leave the somite and migrate toward the limb through the stimulus of scatter factor (hepatic growth factor), which is produced by the proximal cells of the limb-forming area. Before migrating, the premuscle cells in the somite express c-met, which is a specific receptor for scatter factor. The premyogenic cells, which are morphologically indistinguishable from the other mesenchymal cells, express Pax-3 and spread throughout the limb bud. In the splotch mutant, which is characterized by the absence of Pax-3 expression, muscle cells do not populate the limb bud. Migrating premuscle cells also express the cell adhesion molecule N-cadherin, which is important in correctly distributing them throughout the limb bud mesenchyme. The migrating premuscle cells keep pace with the elongation of the limb bud, although cells expressing characteristic muscle molecules (e.g., MyoD) are not seen in the distal mesenchyme. Some experimental evidence suggests that premyogenic cells are not present in the distal mesenchyme. The reason may be the high concentrations of BMP in the distal mesenchyme, which block the proliferation of myoblasts and can even cause the death of these cells. Actual differentiation of the premuscle cells into muscle within the limb requires signals from the ectoderm, principally Wnt-6. If the ectoderm of the limb bud is removed, cartilage and connective tissue, but not muscle, differentiate.

Shortly after the condensations of the skeletal elements take shape, the myogenic cells themselves begin to coalesce into two common muscle masses: one the precursor of the flexor muscles and the other giving rise to the extensor muscles. The transcription factor Tcf-4 is expressed throughout the connective tissue associated with the muscle masses. This is the connective tissue that determines the morphology of the individual muscles as they take shape.

The next stage in muscle formation is the splitting of the common muscle masses into anatomically recognizable precursors of the definitive muscles of the limb. Little is known about the mechanisms that guide the splitting of the common muscle masses, although more recent evidence suggests that the pattern of blood vessels defines the future sites of cleavage through the secretion of platelet-derived growth factor and its action on the formation of connective tissue sheaths around the forming muscles. The fusion of myoblasts into early myotubes begins to occur during these early stages of muscle development.

Considerable evidence suggests that myogenic precursor cells do not possess intrinsic information guiding their morphogenesis. Rather, the myogenic cells follow the lead of connective tissue cells, which are the bearers and effectors of the morphogenetic information required to form anatomically correct muscles. In experiments in which the somites normally associated with a limb bud are removed and replaced by somites from elsewhere along the body axis, myogenic cells are morphogenetically neutral. Muscle morphogenesis is typically normal even though the muscle fiber precursors are derived from abnormal sources.

A later function of the T-box transcription factors Tbx-5 and Tbx-4, which play earlier important roles in initiating the development of the forelimbs and hindlimbs, respectively (see p. 193), is the regulation of muscle patterning. Mutations of these genes result in abnormal limb muscle patterning.

Depending on the specific muscle, the migration, fusion, or displacement of muscle primordia may be involved in the genesis of the final form of the muscle. In one case, genetically programmed cell death, apoptosis, is responsible for the disappearance of an entire muscle layer (the contrahentes muscle) in the flexor side of the human hand. The myogenic cells differentiate to the myotube stage; they then accumulate glycogen and soon degenerate. The contrahentes muscle layer is preserved in most of the great apes. The reason it degenerates in the human hand at such a late stage in its differentiation is not understood.

Although limb muscles assume their definitive form in the very early embryo, they must undergo considerable growth in length and cross-sectional area to keep up with the overall growth of the embryo. This growth is accomplished by the division of the satellite cells (see p. 183) and the fusion of their progeny with the muscle fibers. The added satellite cell nuclei increase the potential of the muscle fiber to produce structural and contractile proteins, which increase the cross-sectional area of each muscle fiber. Accompanying this addition to the nuclear complement of the muscle fibers is their lengthening by the addition of more sarcomeres, usually at the ends of the muscle fibers. The formation of new muscle fibers typically ceases at or shortly after birth. Although the muscles are capable of contracting in the early fetal period, their physiological properties continue to mature until after birth.

Tendons

To function properly, muscles must attach to bones through the formation of tendons. A tendon is a band of dense fibrous connective tissue that is attached to the muscle through the myotendinous junction and to the bone through the enthesis, a complex structure with four zones forming a gradient from type I collagen to fibrocartilage and cartilage and, finally, an actual osseous union with the bone.

Early experiments showed that when the somites adjoining the limb-forming regions were removed, the limbs developed without muscles, but rudimentary tendons did appear, although they later degenerated. These experiments showed that muscle fibers arise from somitic mesoderm, whereas tendons originate from lateral plate mesoderm. Further research has shown that all tendons are not equal. The tendons in limbs, axial structures, and the head require different conditions for their formation.

Overall, three phases are involved in tendon formation: (1) induction by FGFs, (2) early organization through the action of transforming growth factor-β, and (3) consolidation and differentiation, which require the expression of scleraxis (Scx). Tendons of the proximal limb arise from limb mesoderm located just under the lateral ectoderm, where they are induced by FGFs emanating from that ectoderm. Muscle is not needed for their early formation, but interactions with muscle are required for the later differentiation of tendons. The long tendons leading to the digits are more independent from muscle influences during the earlier stages of their formation than are the proximal tendons. Tendons in the head arise from cranial neural crest mesenchyme but, like limb tendons, are independent of muscles in the early stages of their formation. Conversely, the tendons of the axial muscles arise from the syndetome compartment of the somites and require an inductive influence from the myotome in order to form.

As the differentiating tendon approaches the developing bone, Scx in the tendon cells stimulates the production of BMP-4, which, in turn, stimulates the bony outgrowth or bone ridge on which the tendon attaches. The molecular basis for the formation of the myotendinous junction remains obscure. Once the muscle begins to function and exerts mechanical force through its contractions, final differentiation of the body of the tendon and of the enthesis occurs.

Innervation

Motor axons originating in the spinal cord enter the limb bud at an early stage of development (during the fifth week) and begin to grow into the dorsal and ventral muscle masses before these masses have split up into primordia of individual muscles (Fig. 10.27). Tracing studies have shown a high degree of order in the projection of motoneurons into the limb. Neurons located in medial positions in the spinal cord send axons to the ventral muscle mass, whereas neurons located more laterally in the spinal cord supply the dorsal muscle mass. Similarly, a correlation exists between the craniocaudal position of neurons in the cord with the anteroposterior pattern of innervation of limb muscles within the common muscle masses. For example, the rostralmost neurons innervate the most anterior muscle primordia.

Local cues at the base of the limb bud guide the entering pathways of nerve fibers into the limb bud. If a segment of the spinal cord opposite the area of limb bud outgrowth is reversed in the craniocaudal direction, the motoneurons change the direction of their outgrowth and enter the limb bud in their normal positions (Fig. 10.28). If larger segments of spinal cord are reversed, and the neurons are at considerable distances from the level of the limb bud, their axons do not find their way to their normal locations in the limb bud. The muscles themselves apparently do not provide specific target cues to the ingrowing axons because if muscle primordia are prevented from forming, the main patterns of innervation in the limb are still normal.

Sensory axons enter the limb bud after the motor axons. Similarly, neural crest cell precursors of Schwann cells lag slightly behind the outgrowth of motor axons into the limb bud. Cells of the neural crest surround motor and sensory nerve fibers to form the coverings of the nerves in the limbs. By the time digits have formed in developing limbs, the basic elements of the gross pattern of innervation in the adult limb have been established.

Vasculature

The earliest vasculature of the limb bud is derived from endothelial cells arising from several segmental branches of the aorta and the cardinal veins and, to some extent, from angioblasts (endothelial cell precursors) arising from the somites or endogenous to the limb bud mesoderm. Initially, the limb vasculature consists of a fine capillary network, but soon, some channels are preferentially enlarged, resulting in a large central artery that supplies blood to the limb bud (Fig. 10.29). From the central artery, the blood is distributed to the periphery via a mesh of capillaries and then collects into a marginal sinus, which is located beneath the AER. Blood in the marginal sinus drains into peripheral venous channels, which carry it away from the limb bud.

Even in the earliest limb bud there is a peripheral avascular zone of mesoderm within about 100 µm of the ectoderm of the limb bud (Fig. 10.30A). The avascular region persists until the digits have begun to form. Angioblasts are present in the avascular zone, but they are isolated from the functional capillaries. Experimental studies have shown that the proximity of ectoderm is inhibitory to vasculogenesis in the limb bud mesoderm. If the ectoderm is removed, vascular channels form to the surface of the limb bud mesoderm, and if a piece of ectoderm is placed into the deep limb mesoderm, an avascular zone forms around it (Fig. 10.30B). Degradation products of hyaluronic acid, which is secreted by the ectoderm, seem to be the inhibitory agents.

Just before the skeleton begins to form, avascular zones appear in the areas where the cartilaginous models of the bones will take shape. Neither the stimuli for the disappearance of the blood vessels nor the fate of the endothelial cells that were present in these regions is understood at this time.

The pattern of the main vascular channels changes constantly as the limb develops, probably from the expansion of preferred channels within the capillary network that perfuses the distal part of the developing limb. With the establishment of the digital rays, the apical portion of the marginal sinus breaks up, but the proximal channels of the marginal sinus persist into adulthood as the basilic and cephalic veins of the arm (see Fig. 10.29C).

Similar major changes occur in the arterial channels that course through the developing limb (Fig. 10.31). Preferred channels connected to the primary axial artery ultimately take ascendancy, especially in the forearm, thus leaving the original primary axial artery a relatively minor vessel (the interosseous artery) in the forearm.

Clinical Correlation 10.1 discusses limb anomalies.

Clinical Correlation 10.1   Limb Anomalies

Because they are so obvious, limb anomalies have attracted attention for centuries and have been subject to many systems of classification. Most earlier classification schemes were based on morphology alone, usually from the perspective of surgeons or rehabilitation specialists. Only in more recent decades has it been possible to assign genetic or mechanistic causes to some of the known limb malformations. Table 10.2 presents a summary of the common types of morphological limb defects. Many of the most common limb malformations are the result of still poorly understood developmental disturbances.

Table 10.2

Common Structural Types of Limb Malformations

Term	Description
Amelia (ectromelia)	Absence of an entire limb
Acheiria, apodia	Absence of hands, feet
Phocomelia	Absence or shortening of proximal limb segments
Hemimelia	Absence of preaxial or postaxial parts of limb
Meromelia	General term for absence of part of a limb
Ectrodactyly	Absence of any number of digits
Polydactyly	Excessive number of digits
Syndactyly	Presence of interdigital webbing
Brachydactyly	Shortened digits
Split hand or foot	Absence of central components of hand or foot

The number of genetic conditions known to underlie limb defects is vast (Table 10.3). For example, as of 2010, 310 clinical entities involving polydactyly had been described. Of these, 80 were associated with mutations in 99 genes. In most cases, the means by which the gene mutations are translated into defective development are not well understood. One example of the complexity of limb anomalies is the split hand–split foot malformation, sometimes known as ectrodactyly. This malformation is characterized by a reduced number of digits and a wide separation between the anterior and posterior digits (Fig. 10.32). Mutations of at least 15 genes have been associated with the split hand–split foot malformation, and this malformation is a component of at least 25 separate syndromes that affect different parts of the body. A common developmental pathway leading to the split hand–split foot malformation is disruption of the middle portion of the apical ectodermal ridge or its functions through various mechanisms late in limb formation.

Table 10.3

Some Genetic Conditions Causing Primary Patterning Defects of the Limb

Type of Molecule	Gene	Syndrome	Limb Defect
Transcription factor	GLI3	Greig’s	Polydactyly, syndactyly, cephalopolysyndactyly
	GLI3	Pallister-Hall	Posterior polydactyly
	TBX3	Ulnar-mammary	Upper limb deficiencies and posterior duplications
	TBX5	Holt-Oram	Upper limb deficiencies and anterior duplications
	HOXA13	Hand-foot-genital	Brachydactyly
	HOXD13	Synpolydactyly	Syndactyly, insertional polydactyly, brachydactyly
	PAX3	Waardenburg I and III	Syndactyly
	SOX9	Campomelic dysplasia	Bowing of long bones
Signaling protein	CDMP1	Hunter-Thompson	Brachydactyly
	CDMP1	Grebe’s	Severe brachydactyly
	SHH	Preaxial polydactyly	Extra digits
	SHH, FBLN1	Polysyndactyly	Interdigital webbing, extra digits
Receptor protein	FGFR1 or FGFR2	Pfeiffer’s	Brachydactyly, syndactyly
	FGFR2	Apert’s	Syndactyly
	FGFR2	Jackson-Weiss	Syndactyly, brachydactyly
	DACTYLIN, p63	Split hand–split foot, ectrodactyly	Distal syndactyly, fusion
	SALL1	Triphalangeal thumb	Additional segment in thumb

Adapted from Bamshad M and others: Pediatr Res 45:291-299, 1999.

Some of the most frequently encountered limb anomalies have causes that do not involve classic development mechanisms. Several have mechanical causes. Intrauterine amputations by amniotic bands, presumably caused by tears in the amnion, can result in the loss of parts of digits or even hands or feet (see Fig. 8.16). Other deformities, such as clubfoot (talipes equinovarus) and some causes of congenital dislocations, have been attributed to persistent mechanical pressures of the uterine wall on the fetus, particularly in cases of oligohydramnios (see Chapter 7).

A very rare limb deformity is macromelia (or macrodactyly, see Fig. 8-17), in which a limb or a digit is considerably enlarged over the normal size. Such abnormalities are sometimes associated with neurofibromatosis, and the neural crest may be involved in this defect.