How to make a zone of polarizing activity: Insights into limb development via the abnormality preaxial polydactyly




Early in vertebrate limb development, a program initiates that polarizes the limb along the antero-posterior axis. The mesenchyme at the posterior margin is ultimately responsible for the asymmetry due to a region called the zone of polarizing activity (ZPA). The ZPA produces and secretes the molecule SHH, which coordinates the patterning of the resulting digits. Preaxial polydactyly (PPD) is a commonly occurring limb abnormality; investigating the genetic basis of this defect has provided insights into our understanding of digit patterning. PPD disrupts limb asymmetry by producing an ectopic ZPA at the opposite margin of the limb bud. Mutations in the long-range, limb-specific regulatory element of the Shh gene are responsible for the defect. Genetic analysis of this limb abnormality provides an important approach in understanding the mechanisms that control digit patterning.


The limb is one of the principal developmental models used to understand vertebrate development. The inherent processes that regulate growth, form and pattern are all exhibited by this highly accessible developmental system. Patterning along the antero-posterior (AP) axis of the developing limb bud specifies the identity and number of digits and has long been of interest to developmental biologists (Robert & Lallemand 2006; Tickle 2006). Identification of the zone of polarizing activity or the ZPA as the signaling center responsible for AP polarity laid the foundation for understanding the patterning processes during limb bud development. Generation of the ZPA along the posterior margin of the limb bud is the ultimate step in a process that polarizes the limb bud and as a result the limb bud is asymmetrical along the AP axis with a well defined posterior and anterior side. The signaling molecule sonic hedgehog (SHH) is produced specifically by the cells of the ZPA and plays a pivotal role in the process of pattern formation. A number of the molecular players have been identified and assigned supporting roles in antero-posterior axis formation.

The purpose of this review is to focus on the regulation of Shh in the limb bud. Since restricted expression of Shh is key to pattern formation, unraveling the mechanism that regulates Shh gene expression is central to understanding limb development. In addition misregulation of Shh underlies a common form of limb abnormality called preaxial polydactyly. In fact analysis of this limb malfomation has led to the identification of a major component of this regulation. The cis-acting regulatory element that drives Shh expression was identified and is the primary location of mutations that cause this common human limb abnormality. The aim here is to review the information concerning the actions of this cis-regulatory element in the context of the developing limb. The molecular components that activate and restrict expression of Shh in the posterior margin of the limb bud will be considered and an attempt to understand how this regulator is capable of causing abnormalities will be presented.

The ZPA expresses SHH in the limb

In the late 1960s the observations were initially published that laid the foundation for a large proportion of the limb developmental work that would follow. Saunders and Gasseling (Saunders & Gasseling 1968), working in the experimentally accessible system of chick embryos, showed that the limb is polarized along the antero-posterior axis. Mapping across the limb mesenchyme identified a region that possesses the activity that polarizes the limb along this axis. This mesenchyme at the posterior margin of the early limb bud when transplanted to the opposite, anterior margin of a second limb bud gave rise to extra digits displayed as mirror image duplications. This region of the limb bud containing the polarizing tissue was referred to as the zone of polarizing activity (ZPA). The properties of the ZPA led to the hypothesis that the polarizing activity was activated by a secreted molecule acting as a morphogen (Wolpert 1969). The expectations of this hypothesis were that a molecule was produced and secreted by the mesenchymal cells of the ZPA, which diffused to generate a concentration gradient. The identity of each digit was subsequently specified by the morphogen concentration. This early seminal work led to years of analysis with attempts to identify the substance(s) that imparts the polarizing activity.

Sonic hedgehog (SHH) was ultimately identified as the candidate morphogen (Riddle et al. 1993; Chang et al. 1994). The basic tenets necessary to classify SHH as the morphogen were met. First, SHH is a secreted molecule and second, it is expressed in a restricted focus in the posterior of the limb within the ZPA. Third, and perhaps more importantly, Shh autonomously transmits the activity of the ZPA in transplantations; in effect, cells expressing Shh when transplanted to the anterior margin of the limb induces mirror image duplications of the digits. Deletion of Shh by gene knockout confirmed the importance of this gene in antero-posterior polarization of the limb (Chiang et al. 1996; Chiang et al. 2001). Homozygous null mutations of Shh resulted in a severe limb phenotype with the loss of all digits in the forelimb and similarly in the hindlimbs with the retention of only digit one.

Does Shh play the classical role expected of a morphogen? In the limb, a concentration dependent action of Shh was suspected, but direct proof has been elusive. Indeed recent evidence argues against a classic gradient-dependent morphogen model, suggesting that timing and spatial activity is more important (Ahn & Joyner 2004; Harfe et al. 2004). Using elegant genetic analyses to both investigate Shh epistasis and to follow the lineage of Shh expressing cells, the mechanism of Shh action has become clearer. Patterning of the digits requires a combination of both Shh and Gli3 working in conjunction (Litingtung et al. 2002; te Welscher et al. 2002). This is clear in Shh/Gli3 double mutant mice in which polydactylous limb buds form with no recognizable differences between the digits. Gli3 exists as a processed shortened form acting as a repressor (GLI3R) (Wang et al. 2000) and a full length form acting as an activator (GLI3A) (Wang et al. 2007) (see Fig. 1). In the presence of Shh the processing of GLI3 is inhibited leading to higher levels of GLI3R at the anterior of the limb. In the absence of Shh only the most anterior digit (digit 1) forms (Chiang et al. 2001) suggesting that GLI3R is responsible for anterior identity. In contrast, specification of digits two to five are dependent on Shh (Ahn & Joyner 2004; Harfe et al. 2004). Only digit two requires SHH signaling by long-range diffusion. Digits three to five are specified according to time spent in the presence of Shh, with the cells that constitute digit five expressing Shh for the longest, and those constituting digit three the shortest.

Figure 1.

Figure 1.

Representation of the complex of gene interactions that gives rise to antero-posterior (AP) polarity in the limb bud. The diagram illustrates the gene expression required to induce Shh expression at the posterior margin (top right) and at the ectopic, anterior margin (bottom right). The patterns of gene expression responsible for the prepattern (labeled as initiation of limb asymmetry) are represented to the left. Here, the gradient of Gli3 (presumably GLI3R) and Hand2 gene expression is represented. Tbx2 and Tbx3 are expressed at both margins and are induced by the ectoderm proximal to the apical ectodermal ridge (AER) (not shown). At the anterior margin Alx4 presumably functions with GLI3R to inhibit the 5′Hoxd genes and further downstream, Shh. At the posterior margin in the presence of Tbx2/3, Hand2 and the 5′Hoxd, the Shh gene will be induced. In the limb with wild type zone of polarizing activity (ZPA) (no ZPA regulatory sequence mutations [ZRS]), Shh is activated at the posterior margin and at high concentrations this prevents the processing of GLI3 to produce the activator form (Gli3A). The GLI3 gradient in combination with Shh patterns the digits. The Gremlin expression prevents Bmp inhibition of the Fgfs in the AER. Cells that express Shh and their descendents do not express Gremlin eventually leading to the turn-off of Shh. In the mutant limb with the ZPA + Pt mutation (ZRS with a causative point mutation) many of the posterior functions are duplicated at the anterior margin. The point mutation enables Shh to be expressed in the presence of repression activity by Alx4 and GLI3R but may be responsive to activation by anterior Tbx. GLI3 processing is inhibited at both margins to produce GLI3A and Gremlin, 5′Hoxd and the Shh/Fgf positive regulatory loop (these genes not shown for sake of clarity) are induced on the anterior side and thus, set up the limb for producing extra digits.

Preaxial polydactyly and an ectopic ZPA

Congenital limb abnormalities are frequent in the population. Preaxial polydactyly (PPD) has a number of genetic causes but the condition that maps to human chromosome7q36 (near the telomere) (Heutink et al. 1994; Tsukurov et al. 1994; Hing et al. 1995; Heus et al. 1999; Zguricas et al. 1999), for example, occurs in about one in every 2000 births. This form of polydactyly includes a broad range of conditions that affect the digits on the preaxial or anterior side of the hands and feet (Fig. 2) (the anterior border of the limb bud will give rise to the thumb or big toe). Depending on the presentation, the defect is called PPD type II, triphalangial thumb (TPT), or triphalangial thumb-polysyndactyly (TPTPS). Triphalangial thumb in many cases appears as the conversion of the thumb to a finger; (i.e. three bone elements substituting for the normal two of the thumb and rotation of the first digit to the same plane as the fingers) (see Fig. 2B). In more severe cases the addition of an extra one or two anterior digits occurs, sometimes accompanied by long bone involvement such as tibial hypoplasia. In one exceptional patient, the hands appeared as mirror images with the reduplication of the posterior three digits to the anterior side of the hand (Lettice et al. 2002).

Figure 2.

Figure 2.

(A) Composition of the genomic structure around the Shh gene. The position of the ZPA regulatory sequence (ZRS) within the Lmbr1 gene is indicated (yellow oval). The gene desert is indicated and varies in size depending on the species. Regulatory elements identified within the gene desert by Jeong et al. (2006) are also indicated (yellow ovals, marked SBE2-3, and SFPE1). The specificity of ZRS action is shown by the red arrows: the brick patterned rectangles indicate the block to the ZRS for more proximal genes. (B) Representation of the phenotype caused by abnormal ZRS activity. The first limb to the left is the result of no ZRS activity. The next three limbs represent predictions that would occur with increasing levels of Shh expression (i.e. larger domains of ectopic ZPA). Also represented is triphalangial thumb, which in some cases appears as a transformation of the thumb to the finger. With increasing ZRS activity the addition of an extra digit is predicted.

Interestingly, it was work on the mouse polydactyly models, mutations called Sasquatch (Ssq) (Sharpe et al. 1999; Lettice et al. 2002; Lettice et al. 2003) and hemimelic extra toe (Hx) (Sagai et al. 2004), that identified the underlying mechanism for PPD. However, it was genetic evidence from both mouse and human that contributed the initial pieces of the puzzle. The genetics of Hx placed the chromosomal location of the mutation to a region on mouse chromosome 5 containing a gene called Lmbr1 (Clark et al. 2000). Subsequently, the human PPD locus was found to map to a corresponding region on chromosome 7q36 also containing the LMBR1 gene (Heus et al. 1999). A human t(5,7)(q11,q36) translocation identified in a PPD patient located the breakpoint in intron 5 of the LMBR1 gene as crucial to the phenotype (Lettice et al. 2002). In accord, the Ssq mutation, which arose inadvertently during a transgenic experiment, implicated the corresponding intron in mouse Lmbr1.

Having identified Lmbr1 as central to the mutant phenotype, this gene became the prime candidate responsible for PPD. Alternatively, the Saunders and Gasseling experiments (Saunders & Gasseling 1968), discussed above, showed extra digits formed due to an ectopic ZPA in the anterior limb bud and in agreement with this prediction, Lmbr1 is physically linked to the Shh gene but resides at an extreme distance of approximately 1Mb. The Ssq mutant mouse was known to misexpress Shh at the anterior margin of the limb bud (Sharpe et al. 1999). Increasingly, long-range regulation became a valid explanation for the basis for PPD. To show that indeed the mutations were involved in Shh regulation, a variation on a genetic cis-trans test was devised relying on the dominance of the polydactyly mutation (Lettice et al. 2002). Mice heterozygous for the Ssq mutation and for the Shh null mutation on opposing chromosome 5 s were mated. Since the mutation and the Shh gene are 1Mb away, recombination was expected to be frequent and in fact, the frequency in these experiments was 1 in 50 mice. All recombinant mice that carried both the Ssq and Shh null mutations on the same chromosome displayed a wild-type phenotype showing that the mutations acted in cis. Cis-action was confirmed in a similar genetic test for the Hx allele (Sagai et al. 2004).

ZPA formation is dependent on a long-range enhancer

Molecular studies led to the identification of a highly conserved putative cis-acting regulatory element within intron 5 of the Lmbr1 gene (Lettice et al. 2003), Lmbr1 itself has no role in limb development (L. Lettice, pers. comm., 2004). This regulatory element has been assigned names such as MFCS (Sagai et al. 2004), and shARE (Dahn et al. 2007); here however, I will use the original designation of ZRS for ZPA regulatory sequence for the sake of simplicity (Lettice et al. 2003).

Identification of regulatory elements is at best problematic due to the fact that no obvious genomic landmarks identify their location. Comparative genomic-based approaches have been useful in the identification of gene regulatory sequences, and are based on the notion that the conservation of non-coding elements across diverse species indicates positively selected, functional sequences (Boffelli et al. 2004). Studies imply that deep conservation in vertebrates as well as ultra-conservation (sequences 200 bp in length and 100% conserved among human and rodents) (Bejerano et al. 2004) are useful indicators of sequences with a high likelihood of demonstrating gene regulatory activity. Recent analysis of highly conserved vertebrate sequences in a mouse transgenic assay showed that of the approximately 170 sequences assayed 45% functioned as tissue-specific enhancers at E11.5 in development (Pennacchio et al. 2006). A comparative genomic approach was instrumental in identifying the ZRS, which proved to be conserved among highly diverse vertebrates including non-tetrapod species. The ZRS contains a highly conserved domain of approximately 800 bp conserved from mammals to fish.

Other genomic features of interest that have an influence on developmental gene expression are the so-called ‘gene deserts’. These are large expanses of hundreds of kilobases scattered throughout the genome that contain no recognizable encoding genes (Nobrega et al. 2003). A large proportion of the highly conserved non-coding sequences are found in clusters in gene deserts in and around developmental regulatory genes (Ovcharenko et al. 2005) and reside at distances of 100kb or more from the gene. The Shh gene is preceded by a large 600–800kb (in mouse and human) gene desert and regulatory elements other than the ZRS have been identified located within the desert (Fig. 2). Using a transgenic assay designed to identify long range regulatory elements, P1-derived artificial chromosomes (PACs) were modified to carry a reporter gene to identify other Shh regulatory sequences (Jeong et al. 2006). In combination with comparative sequence analysis, cis-acting regulatory elements were identified in the Shh desert region, which are responsible for expression in the notochord, floor plate, hindbrain, ventral forebrain and other regions of the central nervous system (Fig. 2). Shh is probably only one of a number of developmental genes with this genomic constitution and represents a paradigm for the complex regulation that controls spatial and temporal embryonic expression.

Transgenic analysis in mouse embryos confirmed that the ZRS contained all of the regulatory information required to drive reporter gene expression in the posterior margin of the limb bud (Lettice et al. 2003). Deletion of the endogenous ZRS from the genome showed that the ZRS is both necessary and sufficient to drive expression of Shh in the limb bud (Sagai et al. 2005). The animals carrying the heterozygous deletion were normal suggesting that the PPD mutations were not a result of inactivation of the ZRS. In contrast, the homozygotes lost Shh expression specifically in the limb buds; the remainder of the animal was normal. The overt phenotype appeared very similar to that of the limbs found in Shh null animals (Chiang et al. 2001).

Such a genomic assembly raises a number of questions about the control of expression. The ZRS appears to be a self contained element comprising much of the information for initiation and spatial expression in the limb bud. This is not the case for all Shh regulators; for example, at least two discrete elements regulate the Shh expression in the notochord (Jeong et al. 2006). The ZRS also does not lie within the gene desert but resides within an intron of a gene in a gene-rich region and must operate over the gene desert to initiate limb-specific activity. Therefore regulation of Shh expression is complex requiring multiple elements that reside throughout 1Mb of DNA. The transfer of regulatory information over long distances is a common but poorly understood event in development.

Single point mutations generate an ectopic ZPA

Having identified a functional regulatory element, it was subsequently shown that this enhancer participated in the mutant phenotype. Mutation analysis in four different PPD families identified point mutations residing in four different positions within the most highly conserved 800 bp of the ZRS (Lettice et al. 2003). Two different mutations in three other families have also been reported and these were all point mutations in this regulatory element (Gurnett et al. 2007). In mouse, point mutations were found for the Hx mutation and for two mouse lines derived from an ethyl nitrosourea (ENU) mutagenesis screen (Lettice et al. 2003; Sagai et al. 2004; Masuya et al. 2007). Subsequently, transgenic assays showed that the ZRS carrying either of two different mouse mutations were capable of driving ectopic reporter gene expression. In these assays expression occurred at both the anterior and posterior margins of the limb bud similar to that seen for Shh expression in the mutants (Maas & Fallon 2005; Masuya et al. 2007). Thus, single nucleotide changes re-direct Shh expression to an ectopic, anterior site of the limb bud. The ZRS knockout in mouse suggests that PPD does not result from a loss-of-activity or haploinsufficiency; whereas, the transgenic assay suggests the point mutations are sufficient to drive new additional expression. Taken together these observations strongly suggest that PPD results from dominant mutations. The point mutations enable Shh regulation to bypass the normal developmental programming that occurs at the posterior margin and to generate a novel ZPA at the anterior margin.

Insights into the range of limb phenotypes in PPD can be inferred from the mechanisms by which Shh operates in the ZPA. For example, triphalangial thumb frequently occurs in PPD patients. In effect, to convert the thumb to a finger, ectopic SHH is predicted to inhibit the processing of anterior GLI3, resulting in reduced levels of GLI3R (Wang et al. 2000) and loss of anterior specificity (Fig. 1). Also levels of ectopic SHH may determine the number of extra digits that are duplicated. Based on transplantation experiments in the chick wing there is a direct correlation between the number of cells transplanted from the ZPA and the extent of digit duplications at the anterior margin (Tickle 1981). The severity of the phenotype; (i.e. the number of digits), may therefore be directly dependent on the extent of the ectopic Shh expression. The identity of the duplicated digit is quite often difficult to determine since digits two, three and four are very similar. However, based on work described above, the identity of the duplicated digit may be predicted to relate to the time of exposure to SHH (Yang et al. 1997; Harfe et al. 2004). Hence, the wide spectrum of abnormalities displayed by PPD patients is due to a number of factors that may include the level of Shh ectopic expression and the time of exposure. It is not clear whether phenotypic variability is stochastic, since families reportedly show a broad range of phenotypes (Heus et al. 1999), or is dependent on the position of the point mutation within the ZRS. PPD is a rare example of point mutations modifying enhancer activity, and further examination of the point mutations should prove valuable in identifying crucial parameters of enhancer function and misexpression.

Long range regulation in disease

The ZRS lies 1Mb from the gene that it regulates, buried inside an intron in a gene rich region. Although enhancers have long been known to work over great distances, an upper ceiling that limits the enhancer function might be expected, but here the ceiling has been raised considerably. Both the position of the enhancer within another gene and at such a distance, present an intriguing series of strategic problems. To regulate the expression of the Shh gene, the enhancer must bypass one gene and traverse an 800kb desert (Fig. 2). The long distance position of the ZRS is quite likely crucial since its location in the intron of Lmbr1 has been retained for hundreds of millions of years, having been found in vertebrates including teleosts and most recently reported in sharks (Dahn et al. 2007).

Analysis of other key regulators of vertebrate limb development has also revealed cis-regulatory regions that operate on their target genes from a considerable distance. These regulators have been called global control regions (GCR). The regulator in the limb deformity (ld) locus sits inside a-3′ intron of the Formin-1 (Fmn1) gene (Khokha et al. 2003; Zuniga et al. 2004) and when deleted causes the ld mutation. A role for this GCR is the regulation of Gremlin, an antagonist of BMP, in the limb. Gremlin sits at least 350kb away from the regulator and appears to be the primary target of the regulator. However the Fmn gene is also regulated in a similar pattern in the limb. A second example is the GCR, which regulates the 5′ Hoxd genes and is at least 250kb away from the gene cluster (Spitz et al. 2003). This region is intragenic but two genes are situated between the GCR and the Hoxd cluster. These two genes, Lunapark and Evx2, are expressed in the limb under the influence of this GCR.

All of these regulators meet similar problems in operating over vast chromosomal landscapes in search for promoters. Unlike the GCRs however, the ZRS does not appear to regulate other nearby genes. The Rnf32 gene lies between the ZRS and Shh (Fig. 2A), sits in the reverse orientation to Shh but is testis-specific (van Baren et al. 2002). The Lmbr1 gene is expressed at low levels throughout the embryo and is not up-regulated in the posterior limb (Clark et al. 2000). The ZRS regulatory mechanism favors the distally located Shh gene and ignores the proximal genes that are an order of magnitude closer. If the ZRS is not required for regulation of genes in the local genomic landscape, then the question remains what aspect of the ZRS regulatory mechanism requires such a distance to operate? Also the size and complexity of the ZRS raises questions as to whether all of the 800 bp is required to regulate Shh in the limb. Are there factor binding sites that regulate gene activity combined with sequences that guide ‘search and find’ missions that locate the correct promoter?

For the ZRS, not only the correct regulatory response but also the effects of single nucleotide changes must be considered. These are mutations that activate expression at an ectopic site without inactivating the normal activity. Interestingly, the mutations are scattered over much of the 800 bp, so that not just a single transcription factor binding site is affected. To understand the congenital defect a mechanism is needed that takes into account the response to normal developmental signals with nominal affects but activates and re-directs expression to an ectopic site.

How to make a ZPA

Since SHH autonomously reproduces the polarizing activity of mesenchymal cells from the ZPA, production of SHH could be considered the ultimate task of the ZPA. Hence, understanding how Shh expression is regulated in the limb mesenchyme may reveal the molecular mechanism that underlies the specification of the ZPA and leads to the speculation that generating polarizing activity in the limb can be reduced to the action of a single regulatory element. However, regulating Shh expression in the limb bud is complex and here is broken down into three stages: (i) the initiation; (ii) maintenance; and (iii) termination stage. As shown in transgenic experiments, the ZRS appears to contain the information required for at least the first two stages (Lettice et al. 2003).

The initiation phase occurs during the period referred to as the limb prepattern (Chiang et al. 2001; te Welscher et al. 2002); i.e. the initial establishment of antero-posterior polarity in the early limb bud. This prepattern appears to be requisite for the induction of Shh. Gli3 and HAND2 are the major players in this process (te Welscher et al. 2002). HAND2 (also called dHAND) becomes localized to the posterior third of the limb mesenchyme very early during limb bud induction. This restricted expression resolves from an initial broader domain, which becomes restricted in response to expression of Gli3. Hand2 and Gli3 expression resolve into two complementary patterns (Fig. 1). Due to mutual down-regulation, Hand2 is restricted to the posterior half and Gli3 to the anterior half, playing a prominent role in positioning the ZPA to the posterior margin. Loss of Hand2 from this prepatterning process results in posterior spread of the Gli3 gene and no induction of Shh. Thus induction of Shh on the posterior side of the limb bud is dependent on both the presence of dHAND and the lack of GLI3. Misexpression of HAND2 in the anterior limb margin of mouse and chicken is sufficient to induce Shh expression with the consequent mirror image duplications (Charite et al. 2000; Fernandez-Teran et al. 2000). Although Shh is downstream of HAND2, the protein portion of HAND2 that binds DNA is not required to induce Shh expression (McFadden et al. 2002). This raises doubts as to a role of dHAND directly regulating Shh expression at the ZPA and suggests that a mechanism independent of DNA binding is important in the intiation of Shh expression.

Other genes that are crucial to establishing limb polarity are the Hox genes. The Hox genes, in particular those of the HoxA and HoxD cluster, are responsible for various aspects of limb development (Wellik & Capecchi 2003). Combined Hoxa/Hoxd mutations specifically affect the formation of different segments of the limb and deletions of both HoxA and HoxD clusters lead to a phenotype similar to the Shh null limb (Kmita et al. 2005). The 5′ Hoxd genes, Hoxd10–13, are expressed early in limb bud development in a posteriorly restricted domain (Nelson et al. 1996; Zakany et al. 2004) and occurs independent of SHH. A specific role for these 5′Hoxd genes in localized activation of Shh has been unraveled (Zakany et al. 2004). Using elegant genetics means, Hoxd11 and Hoxd13 were ectopically expressed in the anterior domain of the limb, which led to misexpression of Shh in an overlapping domain. Restricted Shh expression was shown to rely on multiple HOX proteins controlled by quantitative and qualitative combinations of the paralogous groups 10–13 (Tarchini et al. 2006). The control of Shh expression was at two levels, determining both the degree and the position of expression. Interestingly, Shh expression was non-responsive to the genes of the HoxD cluster at the 3′ side of Hoxd10. This transition in responsiveness is important since these 3′ genes are expressed throughout the whole of the limb at early stages. The early posterior restriction of Hox gene products participate in the AP prepattern (Fig. 1) and it has been suggested that Gli3R may play a role in restricting the 5′ Hox genes (Tarchini & Duboule 2006). The 5′ Hoxd genes may be acting directly to bind ZRS as recently shown by chromatin immunoprecipitation analysis (Capellini et al. 2006).

The Tbx2 and Tbx3 genes are also involved in Shh expression. However their absolute involvement is not clear and is confused by the observations that the roles of Tbx2 and Tbx3 in mouse appear to be reversed relative to that in the chick (Naiche et al. 2005). Both Tbx2 and Tbx3 are expressed early and along both the anterior and posterior margins of the limb bud. Data from mouse implicates Tbx3 as the gene having a role in Shh expression based on the deletion of the Tbx3 gene. The lack of Tbx3 results in the down-regulation of Shh. If Tbx3 is a Shh inducer the postulated role is one of restricting Shh to the posterior margin. Nissim et al. (2007) showed that a relationship exists between the limb ectoderm distal to the apical ectodermal ridge (AER) and Tbx marginal expression. Extrapolating from observations in chick that marginal expression of Tbx2 (which is predicted to be the functional equivalent to Tbx3 in mouse) is involved in the posterior restricted expression of Shh, the work elucidates a new signaling center identified as the dorso-ventral ectoderm, which lies proximal to the AER. This ectodermal signaling center is an important mediator of Shh expression operating via marginal Tbx expression. The signaling ectoderm and Tbx2/3 expression is symmetrical, being present on both anterior and posterior sides of the limb bud (Fig. 2B). This symmetry may be of importance for ectopic anterior expression of Shh in the PPD mutations (discussed below). In the wild type limb bud, however, asymmetric expression of Shh is achieved only in cells made competent by HAND2.

Once initiation of Shh expression is achieved a continuous series of events expands and/or maintains the expression of Shh depending on signaling from the AER. Here, the Fgfs appear to have a role through a Fgf/Shh positive feedback loop that maintains Shh expression (Martin 1998). Signaling from the AER to the ZPA via the FGFs is necessary for normal levels of Shh expression and in turn, SHH maintains expression of the Fgf genes. In the Shh knockout Fgf4 and Fgf8 are down-regulated; whereas, ectopic expression of Shh at the anterior margin results in up regulation of these Fgfs. On the other hand, the double conditional knockouts of Fgf4 and Fgf8 (Martin 1998) have extensive detrimental affects on limb size, shape and patterning. In the double mutants no Shh was detected, suggesting that the Fgfs may have an additional early role in the induction of the gene.

An important component of the regulatory loop is the Gremlin gene, which plays a role in termination of Shh expression. Gremlin is a secreted bone morphogenetic protein (BMP) antagonist and functions in the distal limb bud to prevent the down-regulation of FGFs in the AER (Merino et al. 1999; Zuniga et al. 1999; Khokha et al. 2003; Zuniga et al. 2004). SHH induces Gremlin production in the adjacent mesenchyme, which eventually initiates a self-regulative ‘beginning-of-the-end’ of the antero-posterior patterning process and in doing so, regulates limb bud size (Scherz et al. 2004). The role of Gremlin is to indirectly maintain Fgf4 (and presumably Fgf8, 9 and 17) expression in the AER by preventing BMP down-regulation. As limb bud development continues, the history of the ZPA cells becomes crucial in the loss of Shh expression. Cells that leave the ZPA, but no longer express SHH, spread medially to contribute to the posterior digits (digits 3, 4, & 5). Shh expressing cells and their descendants cannot express Gremlin, thus forming a barrier between the SHH signal and the Gremlin-expressing cells. This series of events disrupts the Fgf/Shh loop thereby attenuating Shh expression, offering the satisfying hypothesis that loss of the ZPA is a passive event directly linked to the growth of the limb bud.

How to make an ectopic ZPA

In addition to Ssq (Sharpe et al. 1999) and Hx (Blanc et al. 2002; Masuya et al. 1995), a number of other unrelated loci in mice cause preaxial polydactyly due to misexpression of Shh. These polydactyly mutations include Strong's luxoid (lst), luxate (lx), X-linked polydactyly (Xpl) (Masuya et al. 1997), dominant hemimelia (Dh) (Lettice et al. 1999), and Rim4 (Masuya et al. 1995). Embryos for each of the mutations were assayed for Shh expression and showed that ectopic expression along the anterior margin is a general phenomenon in creating preaxial polydactyly. Unfortunately, the initial predictions that these mutations would lead to a network of genes relevant to Shh expression has not come to fruition and to date only the gene for lst has been identified.

Mutations in the aristaless-related homeobox gene Alx4 are responsible for lst (Qu et al. 1998; Takahashi et al. 1998); a role for Alx4 in limb development was initially identified in the gene knockout (Qu et al. 1997). Alx4 is expressed broadly in the anterior mesenchyme of the limb bud and initially was postulated to be a component of a signaling cascade that repressed expression of Shh. Alx4 therefore made an excellent candidate factor for operating as a ZRS repressor. However, analysis of Alx4 null embryos shows that in its absence a set of genes that function upstream of Shh are ectopically activated, arguing that Alx4 does not directly regulate Shh itself (Kuijper et al. 2005). Alx4 operates upstream of Shh repressing expression of Hoxd13 and components of the FGF4 feedback loop (Fig. 1) and therefore is part of a program that ensures the limb bud is asymmetric in the prepattern phase of limb bud development. As a result, genes that are normally expressed in the posterior half of the limb that activate Shh are repressed in the anterior half by the Alx4 gene. In addition expression of Alx4 itself is dependent on the genes that prepattern the limb. Gli3 in the posterior of the early limb bud is at least partially responsible for Alx4 expression, whereas Hand2 and at a later stage Shh restricts Alx4 expression anteriorly.

The ZRS point mutations that result in PPD operate at a different stage than Alx4. Activation of Shh expression is downstream and independent of Alx4 and as discussed above, the mutations act to directly initiate Shh misexpression (Fig. 1). Associated with the induction of Shh expression, the downstream 5′ Hoxd genes are up-regulated and the Fgf/Shh regulatory loop is established (Sharpe et al. 1999). Subsequently, well defined extra digits appear. The series of genes such as HAND2 and Hoxd10–13 that are crucial to prepattern the limb and induce the posterior expression of Shh are not required for ectopic expression; the ZRS point mutations override these requirements. Gli3 has been suggested to down-regulate Shh expression (Buscher et al. 1997) and again, the point mutations do not respond to this inactivation and as a response to Shh misexpression, Gli3 itself is down-regulated in the ectopic ZPA. Genes held in common between both margins of the limb and that are expressed symmetrically may be important when considering regulators of Shh misexpression. In this context the Tbx3 gene may be of importance. Based on the gene expression patterns, a likely scenario is that the point mutations within the ZRS enable Shh expression to compensate for the lack of prepattern genes while maintaining responsiveness to Tbx3 activation. These suggestions obviously require further experimental validation, but from the information at hand, it does show an unexpected ability for regulatory mutations to sidestep complex regulatory cascades. When this occurs for developmental genes, new and unusual phenotypes can be expected.

Evolution of the ZPA

The tetrapod limb was an essential morphological adaptation in vertebrate evolution and was crucial in the transition from an aquatic to a land-based lifestyle. Molecular and anatomical data supports the notion that in the fin-to-limb transformation, the evolution of the autopodium (i.e. the hands and feet) was the defining innovation and that the teleost fin appears to be homologous to the more proximal limb structures (Sordino et al. 1995; Wagner & Chiu 2001). Shh is expressed in the teleost fin bud at the posterior margin; however, in the zebrafish the expression does not reach as distally as in tetrapods. The ZRS sequence was identified in fish and used in a mouse transgenic assay. The fish sequence although approximately 70% homologous over < 600 bp still has the capacity to drive specific limb expression in mouse embryos (Lettice et al. 2003).

Recently the homologous regulatory element was found in chondrichthyans (cartilaginous fish, which includes sharks, skates and rays) and is located in the same intronic position of the Lmbr1 gene as all other vertebrates examined (Dahn et al. 2007). Hence the ZRS elements are deeply conserved in vertebrate phylogeny in structures that are homologous but substantially different in morphology. In the skate, the fin buds are very broad and Shh is found to be expressed along the posterior margin but much later than expected during bud development. The delayed onset of Shh expression argues against a role in regulating antero-posterior polarity, but suggests a later role in skeletal patterning but restricted to the most posterior fin cartilaginous elements. Regulation of Shh appears to be a constant in a vastly evolving structure. Shh has various roles from a nominal task in patterning in chondrichthyans, to a role in defining all the distal structures in the tetrapod. The autopod is alleged to be neomorphic, having arisen as a new structure during the tetrapod transition (Sordino et al. 1995). Hence with the basic regulatory mechanism for Shh expression in place, it appears that the role of Shh changed within the context of the evolving limb bud to eventually define digit number and pattern in a tetrapod specific structure.