Molecular Regulation of Limb Growth

The ingenious work of Riddle et al.¹, Tickle², and others in manipulating limbs in chick embryos provided the basic understanding of limb development. It was such work that defined the terms as well as the role of the apical ectodermal ridge, the progress zone, and the zone of polarizing activity in patterning early limb development. Embryology has moved beyond the primitive understanding of which structures form when, into the molecular realm of developmental biology and genetics. As the signaling pathways for limb differentiation become well understood at a molecular level, morphological anomalies in limbs are seen as patterning errors and offer clues to the role of both genetic and epigenetic effects.

Investigators now have more sophisticated tools of molecular genetics, such as microarray chips, which can simultaneously search for abnormalities in the expression of thousands of genes, and techniques that can snip and splice as well as amplify and analyze extremely small quantities of DNA. One of the most revolutionary tools is the “knockout” animal, bred specifically to answer the question “What happens if this particular gene is missing?”

With the availability of these molecular genetic tools, and the compilation of the human genome, we now have a good, but not perfect, understanding of what should happen in the normal processes of limb development. Before it can be determined whether there is a genetic cause of a congenital malformation, it is necessary to have a baseline understanding of normal limb development—what starts it, what regulates it, and what stops it.

From this understanding, we gain insights into uncontrolled growth, including congenital overgrowth conditions and the dysregulation of growth that causes the malignant tumors that occur throughout life. The interest of pediatric orthopaedic surgeons in these conditions stems from our limited control over the growth, ultimate size, and especially the function of limbs that have been affected by growth abnormalities.

Because the scientific process of ascertaining the effect, the interactions, the order of cascading steps, and the feedback mechanisms is so very meticulous, a single scientist or team must focus on only one part of the puzzle. Much has been learned in a very short time and it is impossible to stay abreast of all that is now accepted as scientific fact. This paper delves into the regulation and patterning of limb growth, and in particular, the osteocartilaginous elements in the limb.

Early Patterning of the Limb Field

The upper-limb bud appears in the human embryo approximately twenty-eight days after fertilization along the symmetrical lateral mesodermal plates known as the “Wolff crest.” The early limb bud has two major components: a core of loose mesenchymal cells derived from the lateral plate mesoderm, and an outer layer of epithelial cells derived from the ectoderm. The skeletal elements and connective tissues (cartilage, bone, tendon, and vasculature) are derived from this mesenchyme. Limb muscles and nerves have separate lineages; musculature is formed from myogenic precursor cells that originate in mesodermal somites and migrate into the limb bud. The peripheral nerves, which arise from the neural crest, migrate and extend their axons later in response to trophic cues, such as ephrins, produced by the muscles and connective tissues³.

This rapid proliferation of an undifferentiated cellular substrate is orchestrated at the distal end of the limb bud in a region known as the progress zone. Proceeding from proximal to distal, this cellular matrix begins its differentiation by the condensation of cells that will form the cartilage templates of individual bones. The condensations that give rise to proximal limb elements form first. Cells at the tip of the limb bud remain undifferentiated. As the bud continues to enlarge, more distal skeletal elements differentiate sequentially until the complete set of condensations is laid down². The formation of these condensations in a precise proximal (early) to distal (late) manner is controlled by the apical ectodermal ridge, a thickening in the limb ectoderm that forms at the distal tip of the growing limb bud. The apical ectodermal ridge produces fibroblast growth factors (FGFs) that promote proliferation and inhibit condensation and differentiation in mesenchymal cells nearest the apical ectodermal ridge. As the limb grows outward, the proximal mesenchymal cells that are farthest from the apical ectodermal ridge and destined to become skeletal elements undergo condensation and initiate chondrogenesis.

The development of this proximal-distal axis is integrated with the development of the anterior-posterior axis (in human development this corresponds to the radial-ulnar axis in the upper limb, and tibial-fibular axis in the lower limb). This axis is controlled by the zone of polarizing activity, a region of limb mesenchyme located on the posterior side of the limb bud near its junction with the flank mesenchyme. The zone of polarizing activity is the source of the secreted protein Sonic hedgehog (Shh)¹, which controls the radial-ulnar or tibial-fibular patterning of limbs and specifies both digit number and digit identity². A key mediator of these effects of Shh signaling in the limb is the conversion of the intracellular transcription factor Gli3 from a repressor to an activator. In the absence of Shh, Gli3 represses digit formation. Mutations that lead to ectopic Shh expression, or mutations in Gli3 that convert it to an activator, have been associated with some forms of polydactyly. This finding demonstrates the ability of the zone of polarizing activity and Shh to control digit number. In accordance with the fact that Shh acts through Gli3, mutations in Gli3 itself that cause syndromic polydactyly have been described in Greig cephalopolysyndactyly syndrome and Pallister-Hall syndrome in humans⁴; these syndromes are listed in the human genetic database Online Mendelian Inheritance in Man (OMIM) as entry numbers 175700 and 146510, respectively. (For conditions referenced in this paper, the reader is encouraged to go to http://www.ncbi.nlm.nih.gov/sites/entrez?db=omim and enter the OMIM identifier number in the search field.) Mutations may also occur in distal regulatory regions upstream of the Shh gene that lead to ectopic Shh expression; these have been associated with other forms of preaxial polydactyly in humans^5,6. In addition to the control of digit identity by graded expression of Shh along the anterior-posterior or the radial-ulnar axis of the limb bud, there is evidence that locally acting bone morphogenetic proteins (BMPs), produced in the interdigital mesenchyme, are required for normal differentiation⁷.

Beyond its role in specifying digit number and identity, Shh is required for proximal-distal outgrowth. The requirement for Shh in regulating outgrowth is seen in Shh-deficient mice, in which the limbs are truncated at the most proximal skeletal element. This aspect of Shh function is a reflection of the reciprocal requirement for Shh to maintain the expression of FGFs in the apical ectodermal ridge. Thus, loss of either Shh or FGFs in the limb bud leads to limb truncation and therefore loss of differentiation distally. BMPs and Gremlin, a secreted inhibitor of BMPs, also participate in this pathway; the BMPs produced throughout the limb mesenchyme and in the apical ectodermal ridge inhibit production of FGFs by the apical ectodermal ridge. A key role of Shh is to activate expression of the BMP antagonist Gremlin, thereby preventing this inhibition of FGF expression. Hence, overexpression or underexpression of FGFs, Shh, or BMPs in the early limb bud leads to phenotypes associated with limb truncations or patterning defects (alterations in digit identity and number), but not to limb overgrowth (Fig. 1).

Fig. 1.

Positive and negative-feedback loops control limb outgrowth and cessation of growth. This model explains how the two loops are used first to promote and then to terminate signals. Arrows indicate activation; the T-shaped red line indicates inhibition. Dashed lines represent diminished regulation. During promotion of outgrowth, the positive regulatory loop increases all signals; Shh produced in the ZPA (green zone) promotes FGF expression in the AER (blue zone). This effect of Shh is mediated through its ability to induce Gremlin. Gremlin in turn antagonizes bone morphogenetic proteins (BMPs), which inhibit FGF action (not shown). The overall effect is thus to promote FGF expression in the AER. The trigger to cessation of growth occurs when AER-FGFs are produced at a high-enough level that they repress expression of Gremlin (represented by a T-shaped line in the figure). By this stage of development, the limb mesenchyme has grown, which leads to a larger domain of Gremlin expression (red zone). Once Gremlin expression declines, cessation of limb growth occurs (last panel in figure). Gremlin is no longer present to repress BMP signals, and thus BMPs are able to repress FGF expression. Loss of FGF leads to an inability of FGF to maintain Shh expression. Thus, growth along all limb axes ceases. (FGFs = fibroblast growth factors, Shh = Sonic hedgehog, AER = apical ectodermal ridge, ZPA = zone of polarizing activity, and Grem = Gremlin.)

Regulation of Endochondral Ossification

The earliest overt sign of chondrogenesis is the aggregation of limb-bud mesenchymal cells into precartilage condensations, the sizes and shapes of which are specified by the actions of the Shh and FGF gradients described previously. The exact mechanism of chondrogenesis is almost entirely unknown; however, members of the Hox family of transcription factors are involved in this process. Hox genes are expressed in nested patterns along the proximal-distal axis of the limb bud, and mutations of these genes lead to alterations in the sizes and shapes of precartilage condensations in mice and humans⁸. For example, loss of Hox11 paralogs (Hoxa11, Hoxc11, and Hoxd11), which are normally expressed in a region of the limb that will give rise to the radius and ulna, leads to a severe reduction of these elements⁹. Because very few targets of Hox gene activity have been identified in the limb, the mechanisms by which Hox genes control the sizes and identities of condensations are unknown. However, recent studies have shown that Hox genes directly regulate the expression of BMPs, which play a vital role in the formation of skeletal condensations¹⁰.

BMPs are required for the formation, survival, proliferation, and differentiation of prechondrogenic cells in condensations into chondrocytes; when BMP receptors 1a and 1b are deleted from precartilage condensations in double-mutant mice¹¹, the skeleton does not develop beyond the condensation stage. The expression of transcription factor Sox9 is maintained by BMPs, and Sox9 mediates BMP action and is required so that cells within cartilaginous condensations can differentiate into chondrocytes. Loss of Sox9 in humans leads to campomelic dysplasia (OMIM 114290)¹². Thus, Sox9 is activated by signals controlling limb-bud outgrowth, and Sox9 in turn is directly required for the expression of the major extracellular matrix proteins in cartilage, including type-II collagen and aggrecan.

The differentiation of prechondrogenic cells under the control of BMPs and Sox9 is essential for the next stage of endochondral development, which is the formation of the growth plate (Fig. 2)¹³. The proliferation of chondrocytes in the growth plate is precisely controlled by a feedback loop comprised of the secreted signals Indian hedgehog (Ihh) and parathyroid hormone-related protein (PTHrP)¹⁴. The growth plate consists of a resting zone of stratified layers of slowly dividing epiphyseal chondrocytes that produce small amounts of extracellular matrix. This is followed by more rapidly dividing epiphyseal chondrocytes in the proliferation zone. Under the influence of unknown signals, the most distal epiphyseal chondrocytes undergo hypertrophy and reorientation of the plane of cell division, leading to the formation of chondrocyte columns. These chondrocytes form the columnar zone. They are the most rapidly dividing and biosynthetically active cells in the growth plate. The epiphyseal and columnar zone cells are maintained in a proliferative state by the action of PTHrP, which diffuses or is transported through the growth plate by an unknown mechanism¹⁴ and which inhibits production of Ihh by cells in the columnar zone. As the columns elongate, the cells at the distal ends of the columns exit the cell cycle, become prehypertrophic, move beyond the influence of PTHrP, and express Ihh, which in turn diffuses through the growth plate to activate PTHrP expression. The essential roles of these factors in the development of all skeletal elements that develop through the endochondral pathway have been elucidated by the phenotypes of knockout mice lacking PTHrP, or carrying activating mutations in the PTHrP receptor.

Fig. 2.

The growth plate feedback control loop involves PTHrP, Ihh, and fibroblast growth factor receptor 3 (FGFR3) (see text for details). (Reprinted, with permission, from: Oji GS, Gomez P, Kurriger G, Stevens J, Morcuende JA. Indian hedgehog signaling pathway differences between swarm rat chondrosarcoma and native rat chondrocytes. Iowa Orthop J. 2007;27:9-16.)

Knowledge gained to date about the role of these genes, growth factors, and secreted proteins in limb formation often occurs in the relative isolation of basic science research and oversimplifies the complexity of the regulation of growth and differentiation. Clinical conditions often defy easy explanations despite great enthusiasm and high expectations. For example, germline mutations in genes that regulate the FGF, BMP, Shh, and PTHrP pathways discussed above have not been shown to be associated with generalized or widespread disproportionate growth of limb segments in hemihypertrophy conditions (Fig. 3), or in the hamartomatous overgrowth characteristic of Proteus and Proteus-like syndromes (OMIM 176920). As another example, ectopic expression of Shh due to a mutation in a regulatory loop that leads to overexpression has been linked to polydactyly rather than overgrowth in humans⁶. In addition, individuals with overproduction of BMP due to the heterozygous mutation in the gene for Noggin (Noggin is a potent secreted antagonist of BMP signaling) exhibit defects in joint morphogenesis; homozygous loss of Noggin leads to massive skeletal overgrowth and synarthrosis, but no evidence of hamartomas^15,16. Similarly, gain-of-function and loss-of-function mutations in FGF receptors lead to skeletal dysplasias such as Apert (OMIM 101200), Carpenter (OMIM 201000), and Crouzon syndromes (OMIM 123500) but not to hamartomas or overgrowth. Moreover, mutations in each of the above genes lead to alterations in the rate of growth, yet this appears to be discordant with findings that the rate of skeletal growth is not altered in a condition such as Proteus syndrome¹⁷. Exactly how gene expression is regulated and how these different signaling molecules interact have yet to be elucidated.

Fig. 3.

Left arm of child with hemihypertrophy. (Photograph courtesy of Shriners Hospital for Children Northern California.)

Cessation of Limb Growth

In contrast to the large amount of information regarding the molecular details of limb growth, essentially nothing is known about the mechanisms underlying the cessation of limb growth. However, there is new evidence that negative-feedback pathways involving BMPs, FGFs, and Shh supply the signals to cease limb outgrowth¹⁸ (Fig. 1). In this model, low levels of FGF interact with Shh to promote limb outgrowth until an increased production of FGF by the apical ectodermal ridge crosses a threshold that triggers the negative-feedback pathway, which leads to repression of the expression of Gremlin, a BMP antagonist. As a result of reduced Gremlin expression, BMP signaling is unopposed, and this triggers reduced FGF expression. Once activated, this negative-feedback pathway leads to cessation of outgrowth. In this model, FGF, Shh, and Gremlin regulate and maintain outward growth of the limb, and, at a critical time, transition into the cessation of growth. However, as discussed above, delayed growth cessation does not appear to explain overgrowth, because the manipulation of levels of expression of FGFs, BMPs, and Shh in the developing limb leads to defects in patterning rather than disproportionate growth.

Some information regarding the normal control of cessation of limb growth can also be gleaned from studies of digit regeneration in mammals. The digit tips of very young children and rodents are known to regenerate at least partially following amputation¹⁹. The distal phalanx forms late in embryogenesis by endochondral ossification of the metaphysis and proximal epiphysis and intramembranous bone formation in the tuft. The distal phalanx continues to elongate until skeletal maturity at puberty. Amputation through the distal third of the terminal phalanx may allow regeneration in the young animal, but regenerative failure is observed following amputation of the distal two-thirds of the bone²⁰. Regeneration is characterized by the formation of a blastema of proliferating cells that appear undifferentiated and express the secreted protein BMP4 and the transcription factor Msx1, and these genes have been shown to be functionally required for the embryonic regeneration response^19-21. The regenerated digit attains normal size or remains slightly smaller, indicating that the extent of proliferation of blastemal cells is subject to general growth-control mechanisms. This ability to regulate the extent of regeneration is conserved in fish and salamanders; in these organisms, the rate of cell proliferation is higher in proximal amputations compared with distal ones, leading to attainment of the final size of the regenerated fins or limbs at the same time, regardless of the extent of growth required²². In these organisms, the final form and size of the regenerated limb appears to again be patterned by the same mechanisms extant in embryogenesis. Overgrowth either of parts or the whole does not occur.

The TSC-mTOR Pathway in the Control of Cell Growth

Control of growth of individual cells is an important process during embryogenesis and throughout life. A serine-threonine protein kinase, involved with growth, replication, and motility through regulation of DNA transcription and protein synthesis, is known as FK506-binding protein 12-rapamycin complex-associated protein 1 (FRAP1), or mammalian target of the immunosuppressive drug rapamycin (mTOR).

mTOR plays a central role in the regulation of cell growth²³, receiving input from multiple signaling pathways to stimulate protein synthesis. Dysregulated mTOR activity caused by mutations in genes that normally repress mTOR activity is associated with several hamartoma syndromes, including Cowden disease, Proteus syndrome and Proteus-like syndrome, and tuberous sclerosis complex (TSC). For example, phosphatase and tensin homolog (PTEN), a tumor suppressor gene involved in the regulation of cell-cycle length and in the initiation of apoptosis, negatively regulates mTOR activity. Homozygous loss of PTEN leads to embryonic lethality, whereas heterozygous loss in epithelial tissues leads to neoplasia²⁴. Germline mutations in PTEN (OMIM 601728) have been found in multiple overgrowth syndromes, including 85% of Cowden disease cases (OMIM 158350), 65% of Bannayan-Riley-Ruvalcaba syndrome cases (OMIM 153480), and 20% of Proteus syndrome cases (OMIM 176920)²⁵. Although there is controversy regarding the relevance of PTEN mutations as a specific cause of Proteus syndrome, the finding of mutations in this gene in multiple overgrowth syndromes is compelling, and it is likely that mutations in genes other than PTEN that also negatively regulate mTOR activity may underlie at least some cases of Proteus syndrome²⁴. If this turns out to be the case, then treatment with rapamycin, a negative regulator of mTOR activity, may be of benefit. Initial results indicate that this treatment ameliorates at least some aspects of overgrowth in Proteus syndrome²⁶.