At the distal tip of the developing limb buds , at the interface between dorsal and ventral ectoderm, a ridge of thickened ectoderm forms in response to signals from the underlying mesoderm. This thickening develops into the apical ectodermal ridge (AER), the signaling center for the PD axis (Fig. 2). Signaling from the AER is critical for limb bud outgrowth and elongation, directly inducing formation of upper limb structures in a proximal (early) to distal (late) manner. These structures may be viewed as three distinct segments achieved by patterning along the proximodistal axis: the stylopod (upper arm), zeugopod (forearm), and autopod (wrist and hand). This PD patterning appears to occur very early when the AER is established in the initiating limb bud and is dependent on FGF signaling from the AER (Mariani et al. 2008; Sun et al. 2002).

The importance of the AER is demonstrated by limb differences along the PD axis if AER function is impaired. Disruption results in limb truncations at a level corresponding with the stage of development when the disruption occurred (Summerbell 1974). The later the removal or disruption of the apical ectodermal ridge in development, the more distal the resulting truncation. Structures proximal to the level of truncation remain intact. These truncations can be rescued by grafting an AER from a different chick embryo, defining the AER as both necessary and sufficient to promote outgrowth.

The molecular cue from the AER was isolated and identified to be one of several growth factors from the fibroblast growth factor (Fgf) family. In the developing limb, members of the Fgf family exhibit varying degrees of functional redundancy. Of the Fgfs, Fgf10 alone is both necessary and sufficient to produce an intact limb (Duboc and Logan 2011b) and is normally expressed by the mesenchyme underlying the AER. The AER itself specifically expresses Fgf4, Fgf8, Fgf9, and Fgf17 (the AER-FGFs). Mice deficient in Fgf4, Fgf9, or Fgf17 retain normal limb development, likely rescued by the functional redundancy of the Fgfs. In contrast, Fgf8 is critical as loss of Fgf8 caused impaired limb outgrowth and significantly smaller limbs (Mariani et al. 2008).

Fgf10 from the sub-AER mesenchyme and Fgf8 from the AER are intrinsically related in a positive feedback loop that forms the core signaling required for growth along the PD axis (Ohuchi et al. 1997). Early expression of Tbx5 during upper limb bud initiation first induces expression of Fgf10 from the sub-AER mesenchyme (Agarwal et al. 2003). Fgf10 signals to the overlying AER to express Wnt3a, which in turn drives Fgf8 expression. Fgf8 from the AER then signals to the sub-AER mesenchyme to maintain Fgf10 expression from the underlying mesenchyme, resulting in an Fgf8-Fgf10 positive feedback loop located at the distal end of the developing limb. This Fgf/Wnt signaling loop is critical for limb development – loss of Wnt3 (tetra-amelia, MIM 273395) (Niemann et al. 2004) or Fgf10 (Sekine et al. 1999) results in amelia, or absence of limb formation.

The importance of the distal end of the limb bud in limb outgrowth was initially conceptualized in a “progress zone” model to explain proximodistal development. The progress zone model posits that there is a zone of undifferentiated mesenchymal cells within the sub-AER mesenchyme with an intrinsic timing mechanism. The cells in this zone are maintained by Wnt3a and Fgf8 from the AER, which maintains a pool of progenitor cells by stimulating proliferation and inhibiting differentiation (ten Berge et al. 2008). A timing mechanism would provide progenitor cells with positional cues based on how long the cells reside in the progress zone. Mesenchymal stem cells that exit relatively early develop into proximal structures such as the humerus. Cells that exit late end up in distal locations as the limb elongates and develop into distal structures such as the forearm and hand. The time-dependent mechanism of the progress zone model provided a mechanistic explanation for the time-dependent transverse deficit phenotype produced by removal of the AER. The later the AER excision, the more distal the defects due to loss of progenitors with longer residence in the progress zone.

Subsequent experiments, however, demonstrated that the progress zone model is inaccurate. Lineage tracing in X-irradiation experiments to induce phocomelia in chick embryos demonstrated that proximodistal patterning was unaffected despite radiation-induced defects in proximal structures (Galloway et al. 2009). Similarly, fate mapping in chick limb buds demonstrated that proximodistal cell fates were established early, with limb truncations occurring due to apoptosis of these fated cells rather than defects in PD patterning (Dudley et al. 2002).

Modern attempts to understand PD patterning emphasizes the conceptualization of limb patterning in the context of dynamic interactions between molecular events (Tabin and Wolpert 2007). More recently, a “two-signal” model was proposed to explain proximodistal determination of limb structures. Specifically, the two opposing signals were retinoic acid (RA) for induction of proximal structures, with the FGFs from the AER determining formation of distal structures (Cooper et al. 2011; Roselló-Díez et al. 2011). Supporting this model, ectopic introduction of RA to the distal limb bud resulted in proximalization of the distal limb (Mercader et al. 2000). Additionally, the AER-FGFs were demonstrated to establish distal structures by repression of Meis1/2, homeobox genes that establish the proximal limb (Mariani et al. 2008).

The role of retinoic acid appears to be permissive rather than actively establishing PD patterning (Cunningham et al. 2013). Fgf8 expression by the developing heart suppresses limb bud initiation due to inhibition of Tbx5 expression. Retinoic acid from the trunk functions to block cardiac Fgf8, enabling limb bud expression of Tbx5 and Meis1/2 and limb bud initiation. For proximodistal patterning itself, RA is unnecessary. Interestingly, Tbx5 itself is needed for cardiomyocyte differentiation (Holt-Oram syndrome, MIM 142900), the mutation of which is characterized by upper limb and cardiac defects.

PD patterning remains incompletely understood and is an evolving field. It is also not known the extent to which these mechanisms are conserved in human limb development.

Despite our lack of understanding, the clinical implications have not changed. Defects along the proximodistal axis, such as limb truncations and longitudinal defects, remain intrinsically related to the AER and FGF function. Insult to the AER during development will lead to clinically observed truncations at variable stages depending on the timing of the insult. Frequently, isolated limb cases suspected to be secondary to insults to the AER will predominantly be mechanical and nonheritable due to the extensive role FGFs play in other biological systems, resulting in severe systemic defects that may render limb defects lower in priority.

Arguably, the most studied axis of limb patterning, and perhaps the most clinically relevant, is the AP axis. The anteroposterior axis distinguishes the radial (or preaxial) from the ulnar (or postaxial) side. This axis was initially discovered when sections of posterior (ulnar) limb bud tissue were excised at varying stages of chicken limb development resulting in limbs that developed longitudinally but that did not develop radioulnar-based digit identities. Excision of radial tissue did not produce the same effect. It was discovered that a small region of posterior cells, at the junction of the limb paddle and the trunk, was a signaling center for AP development termed the zone of polarizing activity (ZPA). This zone of tissue polarized the limb along the AP axis. When ZPA tissue was grafted onto the radial side, a mirror image of the limb along the AP plane was produced (Tickle 1981).

The key signaling molecule of the ZPA is Sonic Hedgehog (SHH). This ligand was named for its molecular resemblance to the drosophila molecule Hedgehog, which is important for fly segmentation, thus leading to its similar name despite different phyla. SHH is a diffusible signaling molecule with expression restricted to the posteriorly located ZPA (Fig. 3) and forms a posterior (high) to anterior (low) gradient of SHH signaling. Functionally, SHH is critical for regulating AP patterning and growth of the zeugopod (forearm) and autopod (hand).

Restriction of SHH expression to the posterior-located ZPA is accomplished by Gli3, a transcription factor that pre-patterns the early limb bud along the AP axis before SHH expression is activated (te Welscher et al. 2002). Gli3 exists in two forms: GLI3A (activator) and GLI3R (repressor). By default, Gli3 is modified to form the Gli3 repressor form. On the posterior (ZPA) side, SHH inhibits this modification to allow for production of the GLI3A activator form, producing a gradient of high GLI3A:GLI3R ratio posteriorly to low GLI3A:GLI3R anteriorly. Manipulation in chick embryos to produce high GLI3A:GLI3R ratios throughout results in polydactyly with posterior digit identities of all extraneous digits (Litingtung et al. 2002), reflecting the role of high SHH signaling in specifying posterior structures. Of note, the stylopod (upper arm) is patterned independently of SHH, presumably accomplished by Gli3-mediated pre-patterning (Niswander 2002). The elbow represents the transition from SHH-independent to SHH-dependent limb development, which may have clinical implications in such clinical phenotypes as below-elbow truncations.

The interactions between SHH and Gli3 result in a gradient of SHH signaling and GLI3A:GLI3R ratio along the AP axis that directs specification and development of autopod ulnar-sided to radial-sided structures (Fig. 4). In humans, defects in the function of either SHH or Gli3 may cause limb defects along the AP axis, frequently manifesting clinically with digit abnormalities such as polydactyly and syndactyly (Anderson et al. 2012). Mutations of GLI3, for example, result in various preaxial and postaxial polydactylies in Greig cephalopolysyndactyly syndrome (MIM 175700) (Hui and Joyner 1993).

Defects in digit development are closely correlated with SHH function because SHH signaling specifies the number of digits as well as the identity of each digit. To achieve this, SHH is expressed in a gradient that varies both spatially and temporally (Harfe et al. 2004). Spatially, the concentration gradient exposes the mesoderm to varying concentrations of SHH, with the concentrations greatest on the posterior (ulnar) side and none on the anterior (radial) side. The mesenchymal progenitors of the fifth digit are exposed to the greatest SHH concentration whereas the thumb develops in the absence of SHH signaling. Temporally, the length of exposure to SHH determines digit identity. Shorter-term exposure of the mesenchyme is sufficient to specify anterior digits (second digit), whereas the fifth digit requires the longest SHH exposure for correct specification (Scherz et al. 2007). In fact, the posterior three digits contain SHH-expressing cells from the ZPA while the SHH contribution to the second digit derives from paracrine signaling (Harfe et al. 2004). Furthermore, SHH acts as a mitogen to produce the necessary progenitor pool for forming five complete digits (Malik 2014).

Digit identity is established by the SHH gradient from the ZPA , but digits initially develop as “webbed” fingers. Extraneous tissue between fingers must undergo apoptosis to form digits with interdigital separations. Bone morphogenetic proteins (BMPs), widely known for their role in chondrogenesis and osteogenesis, mediate apoptosis of the interdigital mesenchyme to produce separated fingers. The interdigital mesenchyme expresses BMP2, BMP4, and BMP7 which antagonizes the pro-survival effects of the Fgfs from the overlying AER (Pajni-Underwood et al. 2007; Suzuki 2013). In animals with webbed limbs such as bats, BMP antagonists block BMP signaling to produce persistent interdigital tissue (Oberg et al. 2010). Enhanced function or signaling of Fgfs can also prevent interdigital apoptosis by overcoming BMP-mediated inhibition, as occurs in the syndactyly observed in Apert syndrome (MIM 101200).

The factors governing formation of the digits are less well understood. Digit formation is currently thought to occur from the combination of the initial AP patterning established by SHH, BMP signaling from the interdigital mesenchyme, and phalangeal growth mediated by the phalanx-forming region (PFR) found in the sub-AER mesenchyme. Cells from the sub-AER mesenchyme are continuously incorporated into each of the PFRs and subsequently into the growing digit. Cells incorporated earlier form the proximal digit, undergoing condensation to form the proximal phalanges; cells incorporated later form the distal phalanges. BMP signaling from the interdigital mesenchyme promotes this process via its stimulatory effects on the PFR, in contrast to the inhibitory effect BMPs have on the AER-FGFs that produces interdigital apoptosis (Suzuki et al. 2008).

SHH is critical in various development processes, most notably is the development of the notochord. Mutations to the SHH gene proper, such as large deletions that produce a defective protein, occur but are unlikely to be encountered, due to either developmental lethality or serious neurologic defects that take precedence. More relevant, however, are mutations to regulators of SHH signaling. Several of these regulators have clinical importance and include ZRS mutations, GLI3 mutations, and ciliopathies.

The ZRS (ZPA regulatory sequence) is an enhancer sequence that is necessary and sufficient for regulating the spatiotemporal SHH activity in the developing limb (Anderson et al. 2012). This sequence is located at a distance (1 Mb upstream) from SHH, in an intron in the LMBR1 gene. Mutations to this region produces a range of phenotypes involving the upper limb including preaxial polydactyly (MIM 174500), isolated triphalangeal thumb (MIM 174500), syndromic triphalangeal thumb (MIM 174500), syndactyly (MIM 186200), and acheiropody (bilateral congenital amputations, MIM 200500). Triphalangeal thumbs and preaxial polydactyly arise secondary to point mutations that result in ectopic SHH expression, producing an ectopic ZPA at the anterior margin of the limb bud (Lettice et al. 2008). This ectopic SHH may respecify anterior digits into posterior digits (triphalangeal thumb or thumb-to-finger transformation) or induce the formation of a mirror hand (duplication of posterior digits in the anterior margin) (Fig. 5). Syndactyly and polysyndactyly may occur from overexpression of SHH, particularly in the interdigital mesenchyme, and are associated with mutations that increase the dosage of SHH such as ZRS duplications (Klopocki et al. 2008) or the adoption of a more widely expressed enhancer by SHH (Anderson et al. 2012). Acheiropodia, characterized by congenital upper and lower limb amputations and aplasia of the hands and feet, is associated with mutations to the LMBR1 gene not involving the ZRS sequence, raising the possibility of additional regulatory sites in addition to the ZRS (Ianakiev et al. 2001).