Author: David Metcalfe
Institution: Warwick Medical School, Coventry, UK
Date: April 2007
Abstract
The vertebrate limb is patterned by a number of interrelated molecular pathways which ultimately determine the musculoskeletal and soft tissue organisation of the completed limb. Once the limb field has been established, the proximodistal axis is primarily determined by fibroblast growth factors (Fgfs), dorsoventral by a Wnt system and anteroposterior by Sonic hedgehog (Shh). Although still relatively infrequent, congenital limb deformities may have profound physical and psychosocial effects on children and their families. Although various limb reconstructive procedures exist, a more agreeable situation would seek to prevent factors which result in developmental abnormalities in utero. A greater understanding of the pathophysiology of congenital limb abnormalities may help further this ambition. Work in vertebrate model organisms such as the chick and mouse have suggested that the most profound limb deformities may result from the interruption of sensitive molecular interactions in the early patterning of the limb. This fits comfortably with our understanding of heritable limb disorders and the function of teratogenic compounds. This article explores our current understanding of the molecular patterning of the vertebrate limb and considers the developmental events which have been associated with disorders such as amelia, meromelia and digit disturbances. Further molecular research in this area has the potential to elucidate ways in which congenital limb deformities may be avoided in the future.
Introduction
Congenital limb deformities occur in many forms. Children may be born with complete absence of a limb (amelia), partial absence (meromelia), duplication of fingers or toes (polydactyly) or failure of these digits to separate (syndactyly). Treatments may include one or a combination of prosthetic limbs, orthotics, surgery and rehabilitation (Esquenazi 2004). Although congenital anomalies are relatively infrequent, limb reconstruction following injury is known to have profound implications for the psychological health and wellbeing of afflicted individuals (Scott et al. 2004).
Causes of congenital limb abnormalities may result from maternal pathology such as diabetes, poor nutrition, genetic factors or exposure to teratogenic compounds in utero. The development of the vertebrate limb involves a complex cascade of biochemical pathways in order to determine the three major axes and to establish the various structures which must appear at specified positions along the limb (Gilbert 2005). Genetic abnormalities and teratogens typically disrupt this developmental process at the molecular level. This article explores what is currently understood about the development of the vertebrate limb within the context of congenital limb abnormalities. Previous reviews of vertebrate limb development have neglected clinical correlations (Tickle 2003) or are somewhat dated (Tickle and Eichele 1994) (Johnson and Tabin 1997) (Robertson and Tickle 1997). It is hoped that this article will encourage further research into the molecular basis of limb development with a view to minimising the future incidence of limb deformity.
Figure 1: Conserved structures of the chick and mouse limb pattern. Adjacent to the shoulder, the stylopod will form the humerus in the forelimb or the femur in the hindlimb. The zeugopod will form the radius and ulna or tibia and fibula in the forelimb and hindlimb respectively. The distal autopod will give rise to the wrist and fingers or ankle and toes. Image modified from Niswander 2003.
In the absence of human research subjects, chicks and mice are used in the study of limb development, largely for the ease with which developing limbs can be modified in vivo (Tickle 2004). The molecular basis of limb development is thought to be largely conserved between vertebrates. This will become evident later in the article when genetic causes of limb deformity are considered alongside animal studies. The basis of this conservation is indicated by X-rays (Figure 1) which show that all vertebrate limbs are composed of a conserved proximal stylopod, a central zeugopod and a distal autopod. In humans, the stylopod will give rise to the upper arm or thigh dominated by the humerus and femur respectively. The zeugopod will ultimately contain two major bones the ulna and radius in the forearm and tibia and fibula in the lower leg (Moore and Dalley 2006). The autopod shows more variation between vertebrates with the mouse being the model organism most superficially similar to humans. Again, in humans, the autopod should give rise to the carpal bones, metacarpals and the phalanges of each digit (Moore and Dalley 2006). These distinct structures are organised spatially by pattern formation. Patterning is the process whereby cells differentiate according to their relative position and, in doing so; facilitate arrangement of the limb structures. The limb structures in turn are arranged along three axes, namely anteroposterior, proximodistal and dorsoventral (Gilbert 2005). This article is similarly organised along these axes with a fourth section which briefly considers some points of communication between the three pathways.
Discussion
Limb fields are specified by Hox genes up-regulated by retinoic acid
Limbs arise from limb fields at discrete positions on the embryo. These are localities of pluripotent cells which might ultimately give rise to a limb (Gilbert 2005). Outgrowth of the limb bud is specified by factors such as retinoic acid, as shown by inhibition of retinoic acid synthesis with citral which retards limb growth in axolotls (Scadding 1999). Retinoic acid is secreted by the organiser Hensen's node and appears to function by up-regulating Hox gene expression (Gilbert 2005). Evidence for this comes from tadpoles which regenerate legs when tails are amputated and the stumps treated with retinoic acid (Scadding 1999).
The identity of fore and hind limbs is established by T-box transcription factors
Acting downstream of the Hox genes are the T-box transcription factors; Tbx4 and Tbx5. The spatiotemporal expression of these proteins suggests a role in establishing identity of the fore and hindlimbs (Gilbert 2005). Tbx4 is, for example, up-regulated in regions destined to produce hindlimbs and Tbx5 in those destined to produce forelimbs. Indeed, the forced expression of Tbx4 or Tbx5 in early chick limb field cells is known to specify leg' or wing' identity respectively (Rodriguez-Esteban et al. 1999). Both Tbx4 and Tbx5, then, appear to provide an environment in which other factors can induce limb bud outgrowth. These components do not work in isolation, however, and analysis of Pitx1 -/- mice suggests this gene may also have a role in establishing hindlimb identity (Tickle 2003). Tbx5 has further been implicated in the congenital limb disorder Holt-Oram syndrome. In this disorder, sufferers typically exhibit a variety of limb abnormalities (often involving the carpal bones and thumbs) in addition to heart problems such as atrial septal defect (ASD) and ventricular septal defect (VSD). This was determined by analysis of Holt-Oram sufferers found to exhibit a premature stop codon in the Tbx5 gene. In situ hybridisation in human tissues has demonstrated raised Tbx5 expression in heart and limbs which further suggests a role for this gene in development of these structures (Quan et al. 1997). Similarly, Tbx5 haploinsufficient mice exhibit abnormally elongated phalanges in the first forelimb digit (Bruneau et al. 2001).
Human mutations in Tbx4 have been shown to cause small patella syndrome (SPS) which is characterised by patellar hypoplasia or aplasia as well as deformities of the feet and pelvis. As with the connection between Tbx5 and Holt-Oram syndrome, this has been demonstrated using genotype analyses of SPS sufferers and genetic analysis of Tbx4 -/- mice (Bongers et al. 2004).
Induction of the limb bud is signalled by Fgf10
The limb bud itself is derived from the lateral plate mesoderm within the limb field and its overlying ectoderm. Proliferation of mesenchyme cells from the somites and somatic layer of the mesoderm results in a circular outgrowth known as the limb bud (Gilbert 2005). Fgf10 is secreted by lateral plate mesoderm cells and its role in induction of the limb bud is suggested by its spatiotemporal expression pattern in the embryo. Further evidence comes from Fgf10 -/- mouse embryos which exhibit truncated limbs (Sekine et al. 1999). The strongest evidence, however, comes from artificial limb induction by exogenous provision of Fgf10 on beads placed under the flank ectoderm (Gilbert 2005). Fgf10 has been shown to function downstream of Tbx4 and Tbx5 as Fgf10-null mice can initiate limb buds which however remain flat and fail to grow (Tacheuchi 2003). Interestingly ablation of Fgf10 expression in the intermediate mesoderm does not prevent bud initiation or limb growth. Studies of mice deficient in both Fgf8 and Fgf4 however indicate that the latter may partially compensate in the absence of the former (Boulet 2004).
Fgf10 expression becomes restricted to the limb field by Wnt proteins Fgf10 is typically expressed throughout the lateral plate and intermediate mesoderm (Gilbert 2005). During limb bud induction, however, expression is restricted to the regions from which the bud will form. This varying expression has been shown by cloning a chick Fgf10 gene and detecting expression using In situ hybridisation (Ohuchi 1997). Fgf10 appears to be localised by the action of Wnt proteins, namely Wnt2b and Wnt8c (Gilbert 2005). Indeed, ectopic provision of Wnt2b and Wnt8c may induce Fgf10 expression in the chick forelimb and hindlimb respectively (Kawakami 2001). Fgf10 signals through a cascade of proteins
Limb bud initiation may however occur in Fgf10 knockout mice and this suggests the involvement of other factors (Takeuchi et al. 2003). Indeed, Fgf10 acts through Wnt3a to up-regulate Fgf8 expression which, in turn, stabilises other proteins promoting mesoderm outgrowth such as Sonic hedgehog (Crossley et al. 1996). This relationship between Fgfs was established by detecting Fgf8 mRNA in the adjacent ectoderm when exogenous Fgf10 was applied to the embryo flank (Ohuchi et al. 1997).
Establishment of the dorsoventral axis
Events establishing dorsoventral polarity begin early in development
Dorsoventral patterning is regulated by signalling from ectoderm flanking the limb bud (Tickle 2003) and grafting experiments have suggested that mesoderm has dorsoventral polarity as early as embryological stage 12 (Gilbert 2005). These experiments involved reversing the polarity of presumptive limb mesoderm and observing that the resulting limb had a reversed dorsoventral axis (Chen and Johnson 1999). The spatiotemporal expression of Wnt7a early in the dorsal ectoderm implies a role in establishing polarity (Gilbert 2005). Indeed mice deficient in Wnt7a exhibited ventral footpads on both paw surfaces a phenotype consistent with the hypothesis that this protein is dorsalising (Parr and McMachon 1995). Wnt7a is restricted to the dorsal ectoderm because of repression by the protein Engrailed-1 (En-1) in the ventral ectoderm. En-1 in turn is maintained through bone morphogenetic protein (Bmp) signalling (Niswander 2003).
Wnt7a signals through Lmx-1b
Wnt7a from the dorsal ectoderm appears to up-regulate expression of Lmx-1b in the dorsal mesenchyme (Gilbert 2005). This has been shown by providing Wnt7a on beads and detecting Lmx-1b mRNA by In situ hybridisation. In contrast to other Wnt proteins, Wnt7a does not appear to signal through β-catenin/Lef as disruptions of these proteins fail to affect dorsoventral polarity of the limb (Kengaku et al. 1998). The product of Lmx-1b, a transcription factor, appears to activate genes involved in dorsoventral patterning. Indeed fusion of an En-1 repressor domain to the Lmx-1 homeodomain in chick dorsal mesenchyme results in ventral muscle patterning (Rodriguez-Esteban et al. 1998). Since ventral effects observed in Wnt7a -/- mice are limited to paws, other signals may up-regulate Lmx-1b in the proximal limb bud (Tickle 2003).
The apical ectodermal ridge is induced by Radical fringe
Fgf10 secreted my limb field mesenchyme cells induces formation of the apical ectodermal ridge (AER). The position of the AER is specified by Radical fringe (R-fng) which is a factor secreted by the dorsal ectoderm (Gilbert 2005). This has been demonstrated by grafting dorsal limb bud ectoderm onto the ventral side of another limb bud which resulted in an additional AER (Rodriguez-Esteban et al. 1997). Ventral ectoderm does not, however, express R-fng which, like Wnt7a, is repressed here by En-1. It is this dorsoventral border which marks the boundary along which the AER will form (Gilbert 2005). This has been shown by retroviral ectopic expression of R-fng which altered existing borders and so prevented AER formation. Similarly addition of exogenous R-fng on beads to ventral limb bud cells creates a new border and consequently an additional AER (Gilbert 2003).
R-fng up-regulates AER-specific genes Wnt3a transcripts are up-regulated in the chick limb field ectoderm in the presence of high R-fng concentrations (Gilbert 2005). Embryos in which Wnt3a is misexpressed exhibit irregular or absent AER formation a phenotype reminiscent of R-fng misexpression (Kengaku et al. 1998). This suggests that Wnt3a acts downstream of R-fng. Wnt3a then appears to activate AER-specific genes such as Bmp2, Fgf4 and Fgf8. Indeed misexpression of Wnt3a results in ectopic production of these proteins (Kengaku et al. 1998). Fgf4 and Fgf8 were shown to be necessary for AER function by using CRE-LOX to generate a conditional knockout of these genes in cells of the AER (Niswander 2003). Up-regulation of AER-specific genes causes the limb field dorsal ectoderm to thicken and become AER. Although Wnt3a is not found in all vertebrates, homologues will activate β-catenin/Lef as Wnt3a does in the chick. That Wnt3a signalling is mediated through β-catenin and Lef has been established by eliminating each of these proteins which consequently disrupts AER formation (Kengaku et al. 1998). The events necessary for formation of the AER are summarised by Figure 2.
Figure 2: The boundary between high and low R-fng expression specifies the position of the AER. The molecular details have yet to be fully established but, in Drosophila, Fgf signalling activates Notch in order to cause thickening of the ectoderm.
Establishment of the proximodistal axis
The proximodistal axis was thought to be specified by the progress zone
The way in which the proximodistal axis is specified is currently under review. Until recently the reigning explanation was the progress zone (PZ) model (Figure 3). This postulates a zone behind the AER in which cells proliferate and are fated to become progressively distal depending on the number of mitoses undergone and time spent in the PZ (Gilbert 2005). Mesenchyme underlying the AER, then, is held in a proliferative phase and induced to secrete FGF8 by AER-secretion of FGF10. The resulting positive feedback loop between FGF8 and FGF10, according to this model, is necessary for limb outgrowth (Ohuchi et al. 1999). When cells leave the PZ, their internal clock stops' and their ultimate position in the limb is determined (Dudley et al. 2002). That the AER sustains proliferation of limb field ectoderm has been shown by grafting on additional AERs which results in growth of superfluous distal limb structures (Gilbert 2005). Indeed early removal of the AER restricts limb development to the humerus whereas later removal permits the radius and ulna to form (Ohuchi 1999). This suggests that the AER programs adjacent cells, first to become the stylopod then zeugopod and finally autopod.
Figure 3: According to the progress zone model, mesenchymal cells proliferate and exit the PZ proximally. At this point their autonomous clock' will stop having already specified the cell fate. Image modified from Saunders 2002.
The PZ model has recently come under review
However recent data has led one authority to claim of the progress zone that it is not a zone as such and that neither is it progressive (Normile 2001). These criticisms have been sustained by injecting virus markers at different depths into the limb bud which failed to spread outwards as predicted by the PZ model (Dudley et al. 2002). Similarly removal of the PZ mesenchyme from Fgf8 and Fgf4 deficient mice sometimes failed to affect distal elements (Gilbert 2005). Of the models proposed to replace the PZ, the early allocation and progenitor expansion model has become most popular (Saunders 2002). This states that cell fates in the limb bud are specified early and that mitoses simply expand these cell populations. During these divisions, however, mesenchyme becomes fixed to a progressively limited range of proximodistal fates (Dudley et al. 2002). Similarly, this new model cannot readily explain the loss of proximal elements when chick limb buds are X-ray irradiated (Wolpert et al. 1979).
Limb outgrowth is specified by Fgfs secreted by the AER
Whichever model is ultimately vindicated, both accept that limb outgrowth is required for proximodistal polarity (Tickle, 2003). Additionally, classical ablation experiments have shown that the AER is necessary for limb development (Gilbert 2005). The AER appears to induce outgrowth by secreting Fgfs which have been investigated by analysing knockout and double knockouts in mice (Tickle 2003). Initially the AER is positioned and induced by Fgf10 and Fgf8. Once formed, AER overlying the lateral plate mesoderm expresses Fgf8 but in the posterior ridge of the limb bud this is later followed by Fgf4, Fgf9 and Fgf17. By the time digit primordia are evident, the posterior Fgf4-secreting part has expanded to occupy the entire AER (Tickle 2003). Indeed, addition of Fgf4 may rescue limb development even after the AER has been surgically removed (Niswander et al. 1993). Application of exogenous Fgf4 has additionally proven to cause proliferation of the distal mesenchyme (Moon et al. 2000). Fgf9 expression however persists longer than Fgf4, suggesting that each developmental phase is characterised by a different pattern of Fgf expression (Tickle 2003). Each Fgf signal alone is, however, functionally redundant and gene targeting in mice has shown that inactivation of any single Fgf gene fails to completely retard limb development (Sun et al. 2000). In the PZ model, the oscillations recorded by the autonomous clock of cells in the PZ are thought to be related to Fgf signalling (Tickle 2003).
Fgfs represent a family of 22 proteins involved in a range of developmental processes (Gilbert 2005). In particular, mutations in human Fgf receptors have been tentatively associated with disorders such as Apert syndrome and Pfeiffer syndrome (Ibrahimi et al. 2001). These pathologies are characterised by numerous developmental defects including severe syndactyly. Although this association has begun to gain some experimental support, further research is required in this area before firm conclusions may be drawn (Wilkie et al. 2002).
Proximodistal patterning appears to be influenced by Shh and retinoic acid
As will be shown in the next section, Fgf4 secretion by the AER is a result of Sonic hedgehog (Shh) expression. Evidence for this comes from Shh -/- mice in which distal elements are reduced but proximal structures remain unaffected (Riddle et al. 1993). In this way among others, proximodistal polarity may be specified by events involved in anteroposterior patterning. Already involved in establishment of the limb field, retinoic acid appears to have an additional role in proximodistal patterning. Indeed one of these roles is the up-regulation of Shh. Raldh2, the gene encoding retinaldehyde dehydrogenase 2 which oxidises retinal to retinoic acid, is expressed early in the lateral plate mesoderm. Similarly Cyp26 which encodes a product required for retinoic acid metabolism is expressed later in the distal ectoderm (Tickle 2003). Limb outgrowth in Raldh2 -/- mice can be rescued by the provision of retinoic acid in the maternal diet (Niederreither et al. 2002). Shh is also known to be involved in patterning of other skeletal elements such as separation of the cerebral hemispheres. Indeed, Shh mutations are found in some 37% people inheriting holopresencephaly which is characterised by anomalies of the forebrain and midface (Nanni et al. 1999).
Retinoic acid appears to signal through Hox and Meis genes
One explanation for the PZ model is that the autonomous clock is related to Hox gene expression which provides a measure of position along the proximodistal axis (Tickle 2003). The current model of Hox gene involvement is based on spatiotemporal expression patterns and gene targeting in mice (Gilbert 2005). The 5' Hoxd genes are expressed in the early lateral plate mesoderm whereas Hoxa genes are activated between the lateral plate mesoderm and bud stages (Tickle 2003). Targeted deletion of Hoxa-11 and Hoxd-11 caused lack of ulna and radius whereas deletion of i]Hoxa-13 and Hoxd-13 resulted in loss of the autopod (Gilbert 2005). Hox gene expression patterns, then, change with outgrowth of the limb bud. This has been shown by isolating Hox gene clones expressed during outgrowth and characterising their expression patterns by In situ hybridisation. In this way Hox expression was split into three phases (Figure 4) which correspond with development of the stylopod, zeugopod and autopod (Nelson et al. 1996).
Figure 4: The pattern of Hox gene expression changes along the promimodistal axis of the limb. Hox expression can be divided into three phases', each of which correspond to a limb element: stylopod, zeugopod or autopod. Image from Gilbert 2005
During stylopod formation, Hoxd-9 and Hoxd-10 mRNAs can be detected in the distal mesenchyme where Meis1 and Meis2 interact to specify proximal cell fates (Gilbert 2005). As outgrowth occurs, Meis expression is progressively repressed in distal cells by Fgf signalling from the AER (Tickle 2003). During zeugopod formation, all Hoxd genes are expressed in the posterior while only Hoxd-9 is expressed in the anterior (Gilbert 2005). As the autopod begins to form, Hoxd-9 expression ceases. Indeed, Hoxa-13 expression marks the beginning of the autopod and other Hox genes define anterior and posterior elements by their pattern of expression. Hoxa-12, Hoxa-11, Hoxd-10, Hoxd-11 and Hoxd-12, for example, are expressed in the posterior while Hoxa-13 and Hoxd-13 are restricted to the anterior (Nelson et al. 1996). Hox expression may be regulated by an enhancer which determines which gene is expressed at any one time. This process appears to restrict digit formation to the distal end of the limb (Gilbert 2005).
Mutations in human Hox genes may have dramatic effects on distal limb development. Two Hox genes in particular have been shown to determine the status of major abnormalities in the human limb (Goodman and Scambler 2001). Mutations in Hoxd-13 have, for example, been associated with synpolydactyly in which sufferers exhibit both duplicated digits (polydactyly) and incomplete separation (syndactyly). Indeed, one team of researchers has shown that the expansion size of a polyalanine tract within Hoxd-13 correlates closely with the severity of the human synpolydacyly phenotype (Goodman et al. 1997). The second well-characterised Hox mutation in human limb development is that of Hoxa-13 which has been linked to Hand-Foot-Genital syndrome (Mortlock and Innis 1997). Families carrying this autosomal dominant mutation typically exhibit developmental anomalies of the distal limbs and genitourinary tract. It has been noted that many additionally carry nonsense or missense mutations in Hoxa-13. Once again, the extent of the mutation has been tentatively correlated with the severity of the anomalous phenotype (Goodman et al. 2000).
Establishing the anteroposterior axis
The zone of polarising activity is defined by Shh
Saunders and Gasseling (1968) transplanted a posterior necrotic zone onto the anterior margin of new wing buds and found that this affected digit patterning. This mesodermal tissue was renamed the zone of polarising activity (ZPA). The ZPA is formed by the action of Hoxb-8 and retinoic acid on the posterior mesoderm which induces the formation of this region (Crossley et al. 1996). The involvement of retinoic acid has been shown by inhibition with citral which retarded ZPA induction (Scadding, 1999). A gradient of retinoic acid is thought to activate Hoxb-8 in the forelimb which increases the competence of the presumptive ZPA for Fgf8. A similar mechanism is thought to exist in the hind limb which may involve Hoxb-3 (Lu et al. 1997). Sonic hedgehog, homologous to the segment polarity gene hedgehog in Drosophila, was ultimately detected by In situ hybridisation in the ZPA (Riddle et al. 1993). Exogenous provision of Shh on beads is now known to be sufficient for polarising activity (Gilbert 2005).
Shh expression is restricted to the posterior limb bud mesenchyme
Shh expression is induced by Fgf8 secreted by the AER. Cells of the posterior limb bud mesenchyme are however more susceptible to Fgf8 than those of the anterior. The transcription factors dHAND and Hoxb-8 in the forelimb are thought to explain this asymmetry. Both dHAND and Hoxb-8 are detected in the posterior but not anterior limb bud mesenchyme (Gilbert 2005). Indeed induced expression of Hoxb-8 with retinoic acid is spatiotemporally consistent with its role as an up-regulator of Shh and inducer of the ZPA. Once established, however, the ZPA appears to negatively impact on Hoxb-8 expression. This has been shown by grafting ZPA cells to the anterior of another limb bud which confers resistance to Hoxb-8 expression as induced by retinoic acid (Stratford et al. 1997). Exogenous provision on beads of both dHAND and Hoxb-8 in mice results in an additional ZPA and digit duplication (Gilbert 2005). Both factors, then, appear to specify Fgf8 competence in the posterior limb bud mesenchyme.
Shh is tightly regulated by a signal transduction pathway
Unlike dHAND and Hoxb-8, Gli3 is detected only in the anterior limb bud mesenchyme. Indeed Gli3 and dHAND mutually exclude one another (Niswander 2003). Figure 5 illustrates the Shh signal transduction pathway. Although these interactions are necessary for autopod morphogenesis and digit identity, they are not required for stylopod or zeugopod patterning (Riddle et al. 1993). Gli3 -/- mouse strains for example exhibit polydactyly and sporadic digit patterning (Stratford et al. 1997) but mice deficient in Shh continue to exhibit stylopod and zeugopod development (Riddle et al. 1993).
Figure 5: Sonic hedgehog (Shh) binds its receptor Patched (Ptc1) which activates the previously repressed transmembrane domain Smo. Derepression of Smo inhibits the proteolytic cleavage of activator Gli3. This prevents direct repression (and indirect repression through dHAND) of Shh.
Shh mediates interactions between the ZPA and AER
Secreted by the ZPA, Shh signals the AER to secrete increasing quantities of Fgf4 (Zuniga et al. 1999). This has been shown by grafting cells expressing Shh into the vicinity of an AER and observing the increased expression of Fgf4. Furthermore, mesenchymal expression of Shh may be up-regulated by the addition of Fgf4 with retinoic acid (Niswander et al. 2002). This suggests the existence of a Shh/Fgf4 feedback system between the ZPA and AER.
Whole mount In situ hybridisation has shown that the Shh signal is relayed to the AER by Formin and Gremlin. Formin, for example, is transcriptionally up-regulated in mesenchyme adjacent to the polarising region. Anterior grafts of cells expressing Shh have also been shown to induce ectopic Formin expression in limb bud mesenchyme (Zuniga et al. 1999). Analysis of murine limb deformity (ld) mutants which lack Formin and exhibit defective Shh/Fgf4 feedback has shown that Gremlin lies downstream of Formin. Ectopic expression of Formin in wild type embryos, for example, up-regulates Gremlin although this is not observed in ld mutants (Zuniga et al. 1999). This suggests that Formin induces mesenchymal competence for Gremlin. Gremlin, in turn, has been identified by an expression-cloning screen in Xenopus as a Bmp antagonist (Hsu et al. 1998). Indeed analysis of ld mutants suggested that antagonising Bmp2 in particular induces the AER to secrete Fgf4. Eliminating the effects of Bmp2 with grafted cells secreting Noggin, another Bmp antagonist, for example, is sufficient for the AER to secrete Fgf4 (Zuniga et al. 1999). This forms a full feedback loop between Shh and Fgf4 secretion from the AER (Figure 6).
Figure 6: Secretion of Shh from the ZPA promotes forming expression which confers competence on mesenchymal cells for Gremlin. Gremlin then binds to, and antagonises, Bmp2 which facilitates secretion of Fgf4 from the AER. Fgf4 in turn up-regulates Shh.
Digit identity is regulated by Shh through Bmps, Hox and Gli3
The chick has three digits 2, 3 and 4 which arise at different locations and exhibit distinct morphologies (Drossopoulou et al. 2000). The identity of these digits is thought to be specified by Bmp2 and Bmp7 which are maintained in a gradient by Shh (Gilbert 2005). Shh provision on beads, for example, can drive ectopic expression of Bmp2. More digits were able to form in embryos in which the Shh-loaded beads were left for longer before being removed (Drossopoulou et al. 2000). This role of Shh in the anterior mesenchyme is contrary to its antagonism of Bmp when signalling between the ZPA and AER (Niswander 2003).
Shh prevents the processing of Gli3 into its repressor form (Figure 6). Since inhibition of processing occurs in a dose-dependent fashion, a gradient of Gli3 is formed (Niswander 2003). This gradient appears to specify digit identity but also to constrain the distal elements to pentadactyly (Gilbert 2005). Shh signalling will only cease once digit primordia has appeared. The posterior limb region at this point spans the width of the limb bud and Bmps will persist in the interdigital mesenchyme until freeing of the individual digits takes place (Parr and McMahon 1995). This occurs by virtue of the interdigital necrotic zone and apoptotic events in the so-called anterior and posterior necrotic zones further sculpt the limb. The signal for apoptosis is provided by Bmp2, Bmp4 and Bmp7, as shown by retroviral infection of PZ cells with dominant negative Bmp receptors. When Bmp signalling is blocked in this way, interdigital apoptosis is retarded (Gilbert 2005).
Coordinating the Axes
The events and signals specifying the three axes are related in both space and time (Gilbert 2005). Indeed new data suggests that the proximodistal and anteroposterior axes are inseparable. Fgf8 -/- mice, for example, have barely affected proximodistal fates but lack digits which is suggestive of a role in establishing anteroposterior polarity. Similarly when mesenchymal cells are removed, digit patterning along the anteroposterior axis is affected (Niswander 2003).
Figure 7: Signals responsible for axis specification are inter-related so that the limb can be properly formed in three dimensions. Image from Niswander 2003.
These inter-relationships are however inevitable. Positional information must be considered relative to all axes in order to ensure the correct limb pattern in three dimensions (Figure 7). Wnt7a signalling in the limb bud, for example, specifies dorsoventral polarity but also maintains Shh expression necessary for the anteroposterior axis (Tickle 2003). Indeed Wnt7a -/- mice often lack the most posterior digit (Parr and McMahon 1995). Shh in turn activates Fgf4 secretion from the AER which helps recruit cells to the PZ and has a role in maintaining Shh expression in the ZPA (Gilbert 2005). In this way, communication between the AER and ZPA maintains Shh and Fgf expression. Another significant interaction is the cleavage of Gli3 indirectly by Shh which creates a gradient of activator and repressor Gli3 (Niswander 2003).
Most of these interactions are involved in coordinating the axes. Bmps for example both inhibit Wnt7a and signalling from the AER, as shown by inhibition of Bmps with Noggin which dramatically improves AER longevity (Zuniga et al. 1999). In this way these signals can retard patterning along all three axes (Gilbert 2005). This provides a means of communication between the axes so that the limb can be patterned appropriately in three dimensions.
Conclusion
Molecular patterning of the vertebrate limb occurs by virtue of biochemical pathways working in parallel while also communicating with one another. This sets the basis for the later development of complex anatomical structures such as soft tissue, nerves, vasculature and appropriate musculoskeletal organisation. Although congenital abnormalities may arise at any point along the developmental continuum, disruption of molecular components during the early stages of patterning clearly exert profound effects on the limb phenotype. This is evident from gene knockouts in mice which result in complete ablation of the relevant limb and associated structures. Although murine models of limb development do not necessarily reflect events in humans, discoveries in model organisms have on a number of occasions been followed by clinical observations. Certainly in all vertebrates, the nature of the resulting limb anomaly is essentially determined by the timing of the interruption and the precise molecular component disrupted. Further research into the molecular patterning of the vertebrate limb and the factors which can potentially disrupt this process will aid our understanding of the pathophysiology of limb abnormalities in humans.
References
Bongers, E. M. H. F. et al. (2004). Mutations in the human Tbx4 gene cause small patella syndrome. American Journal of Human Genetics 74, 1239-1248.
Boulet, A. M. et al. (2004) The roles of Fgf4 and Fgf8 in limb bud initiation and outgrowth. Developmental Biology 273, 361-372.
Bruneau, B. G. (2001). A murine model of Holt-Oram syndrome defines roles of the T-box transcription factor Tbx5 in cardiogenesis and disease. Cell 106, 709-721.
Chen, H. and L. R. Johnson (1999) Dorsoventral patterning of the vertebrate limb: a process governed by multiple events. Cell Tissue Research 296, 67-73.
Crossley, P. H. et al. (1996) Roles for FGF8 in the induction, initiation, and maintenance of chick limb development. Cell 84, 127-136.
Drossopoulou, G. et al. (2000). A model for anteroposterior patterning of the vertebrate limb based on sequential long- and short-range Shh signalling and Bmp signalling. Development 127, 1337-1348.
Dudley, A. T. et al. (2002) A re-examination of proximodistal patterning during vertebrate limb development. Nature 418, 539-544.
Esquenazi, A. (2004). Amputation rehabilitation and prosthetic restoration. From surgery to community reintegration. Disability & Rehabilitation 26, 831-836.
Gilbert, S. F. (2003) Pathway for the induction of the AER. Devbio: a companion to Developmental Biology, c16.
Gilbert, S. F. (2005) Developmental Biology, 7th ed. Published by Sinauer Associates Inc. p523-546.
Goodman, F. R. et al. (1997). Synpolydactyly phenotypes correlate with size of expansions of Hoxd13 polyalanine tract. Proceedings of the National Academy of Science USA 94, 7458-7463.
Goodman, F. R. et al. (2000). Novel Hoxa13 mutations and the phenotypic spectrum of hand-foot-genital syndrome. The American Journal of Human Genetics 67, 197-202.
Goodman, F. R. and Scambler, P. J. (2001). Human Hox gene mutations. Clinical Genetics 59, 1-11.
Hsu, D. R. et al. (1998) The Xenopus dorsalizing factor Gremlin identifies a novel family of secreted proteins that antagonize BMP activities. Molecular Cell 1, 673-683.
Hui, C. and A. Joyner (1993) A mouse model of grieg cephalopolysyndactyly syndrome: the extra toes mutation contains an intragenic deletion of the Gli3 gene. Nature Genetics 3, 241-246.
Ibrahami, O. A. et al. (2001). Structural basis for fibroblast growth factor receptor 2 activation in Apert syndrome. Proceedings of the National Academy of Science USA 98, 7182-7187.
Johnson, R. L. and Tabin, C. J. (1997). Molecular models for vertebrate limb development. Cell 90, 979-990.
Kawakami, Y. et al. (2001) WNT signals control FGF-dependent limb initiation and AER induction in the chick embryo. Cell 104, 891-900.
Kengaku, M. et al. (1998) Distinct WNT pathways regulating AER formation and dorsoventral polarity in the chick limb bud. Science 280, 1274-1277.
Lu, H. C. et al. (1997). Retinoid signalling is required for the establishment of a ZPA and for the expression of Hoxb-8, a mediator of ZPA formation. Development 124, 1643-1651.
Moon, A. M. et al. (2000) Normal limb development in conditional mutants of Fgf4. Development 127, 989-996.
Moore, K. L. and Dalley, A. F. (2006). Clinically orientated anatomy, 5th ed. Published by Lippincott Williams & Wilkins. p675-684.
Mortlock, D. P. and Innis, J. W. (1997). Mutation of Hoxa13 in hand-foot-genital syndrome. Nature Genetics 15, 179-180.
Nanni, L. et al. (1999). The mutational spectrum of the Sonic Hedgehog gene in holoprosencephaly: SHH mutations cause a significant proportion of autosomal dominant holoprosencephaly. Human Molecular Genetics 8, 2479-2488.
Nelson, C. E.et al. (1996) Analysis of Hox gene expression in the chick limb bud. Development 122, 1449-1466.
Niederreither, K. et al. (2002). Retinaldehyde dehydrogenase 2 (RALDH2)-independent patterns of retinoic acid synthesis in the mouse embryo. Proceedings of the National Academy of Sciences (USA) 99, 16111-16116.
Niswander, L. et al. (1993) FGF-4 Replaces the Apical Ectodermal Ridge and Directs Outgrowth and Patterning of the Limb. Cell 75, 579-587.
Niswander, L. et al. (2002) A positive feedback loop coordinates growth and patterning in the vertebrate limb. Nature 371, 609-612.
Niswander, L. (2003) Pattern formation: old models out on a limb. Nature Reviews 4, 131-142.
Normile, D. (2001). International Congress of Developmental Biology: developmental progress fills the air in Kyoto. Science 293, 788-789.
Ohuchi, H. et al. (1997) The mesenchymal factor, FGF10, initiates and maintains the outgrowth of the chick limb bud through interaction with FGF8, an epical ectodermal factor. Development 124, 2235-2244.
Ohuchi, H. et al. (1999) FGF10 can induce Fgf8 expression concomitantly with En1 and R-fng expression in chick limb ectoderm, independent of its dorsoventral specification. Development, Growth and Differentiation 41, 665-673.
Parr, B. A. and A. P. McMahon (1995) Dorsalizing signal Wnt-7a required for normal polarity of D-V and A-P axes of mouse limb. Nature 374, 350-353.
Quan, Y. L. et al. (1997). Holt-Oram syndrome is caused by mutations in [I]TBX5[/I], a member of the Brachyury (T) gene family. Nature Genetics 15, 21-29.
Riddle, R. D. et al. (1993) Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75, 1401-1416.
Robertson, K. E. and Tickle, C. (1997). Recent molecular advances in understanding vertebrate limb development. British Journal of Plastic Surgery 50, 109-115.
Rodriguez-Esteban, C. et al. (1997) Radical fringe positions the apical ectodermal ridge at the dorsoventral boundary of the vertebrate limb. Nature 386, 360-366.
Rodriguez-Esteban, C. et al. (1998) Lhx2, a vertebrate homologue of apterous, regulates vertebrate limb outgrowth. Development 125, 3925-3934.
Rodriguez-Esteban, C. et al. (1999) The T-box genes Tbx4 and Tbx5 regulate limb outgrowth and identity. Nature 398, 814-818.
Saunders, J. W. (2002) Is the progress zone model a victim of progress? Cell 110, 541-543.
Scadding, S. R. (1999) Citral, an inhibitor of retinoic acid synthesis, modifies pattern formation during limb regeneration in the axolotl Ambystoma mexicanum. Canadian Journal of Zoology 77, 1835-1837.
Scott, S. R. H. et al. (2004). Psychological distress reported by patients undergoing limb reconstruction surgery: implications for psychological interventions. Behavioral Science 8, 1573-3572.
Sekine, K. et al. (1999) FGF10 is essential for limb and lung formation. Nature Genetics 21, 138-141.
Stratford, T. H. et al. (1997) Hoxb-8 has a role in establishing early anterior-posterior polarity in chick forelimb but not hindlimb. Development 124, 4225-4234.
Sun, X. et al. (2000) Conditional inactivation of Fgf4 reveals complexity of signalling during limb bud development. Nature Genetics 25, 83-86.
Takeuchi, J. K. et al. (2003) Tbx5 and Tbx4 trigger limb initiation through activation of the Wnt/Fgf signalling cascade. Development 130, 2729-2739.
Tickle, C. and Eichele, G. (1994). Vertebrate limb development. Annual Review of Cell Biology 10, 121-152.
Tickle, C. (2002) The early history of the polarizing region: from classical embryology to molecular biology. International Journal of Developmental Biology 46, 847-852.
Tickle, C. (2003) Patterning systems from one end of the limb to the other. Developmental Cell 4, 449-458.
Tickle, C. (2004) The contribution of chicken embryology to the understanding of vertebrate limb development. Mechanisms of Development 121, 1019-1029.
Wilkie, A. O. M. et al. (2002). FGFs, their receptors, and human limb malformations: clinical and molecular correlations. American Journal of Medical Genetics 112, 266-278.
Wolpert, L. et al. (1979). The effect of cell killing by x-irradiation on pattern formation in the chick limb. Journal of Embryology and Experimental Morphology 50, 175-193.
Zuniga, A. et al. (1999) Signal relay by BMP antagonism controls the SHH/FGF4 feedback loop in vertebrate limb buds. Nature 401, 598-602.