Spatiotemporal changes in cell adhesiveness during vertebrate limb morphogenesis


  • Naoyuki Wada

    Corresponding author
    1. Applied Biological Science, Faculty of Science and Technology, Tokyo University of Science, Chiba, Japan
    • Applied Biological Science, Faculty of Science and Technology, Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, Japan
    Search for more papers by this author


During vertebrate limb development, various molecules are expressed in the presumptive limb field or the limb bud in a spatiotemporal-specific manner. The combination of these molecules regulates cellular properties that affect limb initiation and its morphogenesis, especially cartilage formation. Cell adhesiveness of the limb mesenchyme is a key factor in the regulation of cell distribution. Differential adhesiveness of mesenchymal cells is first observed between cells in the presumptive limb field and flank region, and the adhesiveness of the cells in the limb field is higher than that of cells in the flank region. In the limb bud, the adhesiveness of mesenchymal cells shows spatiotemporal difference, which reflects the positional identity of the cells. Position-dependent cell adhesiveness is also observed in blastema cells of the regenerating limb. Therefore, local changes in cell adhesiveness are observed during limb development and regeneration, suggesting significant roles for cell adhesiveness in limb morphogenesis. Developmental Dynamics 240:969–978, 2011. © 2011 Wiley-Liss, Inc.


During vertebrate limb development, the mesenchymal cells in the limb bud differentiate into cartilaginous elements with region-specific shape and arrangement. For understanding the mechanisms underlying region-specific cartilage morphogenesis, the regional difference of the mesenchymal cells in the limb bud has been investigated. During the past two decades, the expression pattern of various molecules in the limb bud has been analyzed, and many of these molecules are expressed in a region- and tissue-specific manner. Great advances have been achieved in the understanding of the roles of these molecules in pattern determination during limb morphogenesis (Towers and Tickle,2009a; Zeller et al.,2009). On the other hand, the cellular basis of region specific cartilage morphogenesis is still unclear, although it is an important issue to address limb morphogenesis.

Prior to cartilage differentiation, mesenchymal cells of the limb bud aggregate to form cellular condensation (Stott et al.,1999; Hall and Miyake,2000). The arrangement of the condensation in the limb bud represents the prepattern of the limb skeleton (Shubin and Alberch, 1986). Therefore, the regulation of condensation size, shape, and timing is key in the acquisition of the formation of cartilage morphology. The process of condensation is regulated by several cellular properties (Hall and Miyake,2000), one of the most important of which is cell adhesiveness.

Changes in cell adhesiveness in developing tissues other than the limb bud have been analyzed by engulfment of cell aggregates derived from different tissues, or by sorting out cells from different tissues (Townes and Holtfreter,1955; Moscona,1956). From these experiments, it has been understood that cells from different tissues possess distinct cell adhesiveness (Steinberg,1970), suggesting an important role for adhesiveness in the unique morphogenesis of each tissue. In addition, the results also suggested that the distinct cell adhesiveness between adjacent tissues regulate tissue arrangement by restricting cell distribution, thereby influencing subsequent morphogenesis.

Since condensation of the limb mesenchyme represents the prepattern of the limb skeleton, it is likely that cell adhesiveness of the limb mesenchyme is spatiotemporally different so that it may form the prepattern in the limb bud. The role of cell adhesiveness of limb mesenchyme has been studied mainly for understanding the regulation of chondrogenesis (Ede,1983; Hall and Miyake,2000). However, several experiments that were performed in the 1970s or earlier showed other roles for cell adhesiveness in limb initiation (Heintzelman et al.,1978), and precise patterning of the limb cartilage (Ede and Agerbak,1968; Ede and Flint,1975). These results highlight the significance of spatiotemporal regulation of the cell behavior of limb mesenchyme during limb initiation and morphogenesis. Nonetheless, our understanding of regulatory mechanisms still needs elucidation.

In this review, the focus is on the importance of the spatiotemporal regulation of cellular behavior during limb initiation and morphogenesis. In that sense, several changes of cellular properties that are the basis of the process and, in particular, on the adhesiveness of the mesenchyme, are delineated. As the limb bud begins to form, the adhesiveness of the mesodermal tissues of the limb field increases, implying the connection between limb budding and the change in cell adhesiveness. Once the limb bud is formed, mesenchymal cells, in turn, show distinct adhesiveness in a spatiotemporal-specific manner, and the adhesiveness appears to reflect the positional identity of each cell. Position-dependent cell adhesiveness is also observed during limb regeneration. Therefore, it is possible that distinct cell adhesiveness generally contributes to position-dependent morphogenesis during limb development and regeneration.


Understanding the mechanism of limb field determination and limb bud initiation has been expanded by identification of several molecules that are expressed in the limb field (Ohuchi et al.,1997; Kawakami et al.,2001). Yet, little is known about how these molecules affect cellular activity in the limb field and induce limb outgrowth. Changes in cellular property during limb initiation were first reported by Heintzelman et al. (1978). In this study, the surface properties of tissue fragments derived from the wing bud, leg bud, or flank region were analyzed. Mesodermal fragments derived from either region at stage 17–18 were aggregated with another fragment derived from the heart or liver in hanging drop cultures. The relative distribution of each tissue within the combined fragments is an indicator of the adhesiveness intensity of each tissue. From this analysis, it was concluded that the adhesiveness of limb mesodermal tissue was stronger than that of the flank tissue, and that the leg bud mesodermal tissues adhere more tightly than the wing bud mesoderm of the wing bud (Heintzelman et al.,1978). This suggests that cell–cell and/or cell–matrix adhesiveness is increased during limb bud formation, especially in the leg bud.

To investigate whether increased adhesiveness of individual cells in the limb field is associated with limb bud initiation and outgrowth, the adhesiveness of dissociated cells derived from the limb field or the flank region was compared by rotation culture (Fig. 1). Dissociated cells from the distal region of the wing bud at stage 20 formed multiple spherical cellular aggregates in rotation culture, whereas mesodermal cells of the flank region were unable to form any apparent spherical aggregates under the same culture condition (Fig. 1A and B). Similarly, cells from the distal region of the wing bud at stage 25 also formed cellular aggregates, albeit larger than aggregates derived from stage-20 cells (Fig. 1C). Thus, individual cells in the limb bud have stronger adhesiveness as compared to mesodermal cells of the flank region. These observations are consistent with those of Heintzelman et al. (1978), and overall suggest that cells in the initial stage of limb outgrowth adhere more tightly to each other than cells of the flank region, and that increased adhesiveness may play a role in limb initiation and outgrowth.

Figure 1.

Aggregate formation by mesodermal cells of the wing bud or flank tissue. Mesodermal cells were isolated from the flank tissue at stage 20, the distal region of the wing bud at stage 20, or the distal region of the wing bud at stage 25. Dissociated cells from each region were separately incubated in rotation culture at 70 rpm for 24 hr. A: Mesodermal cells from the flank tissue at stage 20 form small clusters with various sizes, but no apparent spherical aggregates are observed. B: By contrast, cells of the distal region of the stage-20 wing bud form round aggregates under the same culture condition as A. C: Distal cells of the stage-25 wing bud make larger spherical aggregates than cells of the stage-20 wing bud.

An interesting comparison between the mechanical surface tension of cellular aggregates derived from the limb bud mesenchyme or the flank mesoderm has been recently reported (Damon et al.,2008). As expected, the tension of the limb mesodermal cells was higher than that of the mesoderm of the flank region, and these results also supported the previous work by Heintzelman et al. (1978). One possible role of increased adhesiveness is that cells in the limb field adhere tightly to each other as a result of increased adhesiveness, thereby allowing the cells of the field to behave independently from cells of the flank region. This possibility is supported by the effect of fibroblast growth factor (FGF)-8b on cell adhesiveness of the flank mesoderm. FGF-8b enhances the surface tension of the flank mesoderm to the level of the limb bud mesoderm (Damon et al.,2008). During normal limb development, Fgf-8 is expressed in the ectoderm of the limb field immediately before initiation of limb budding (Crossley et al.,1996; Vogel et al.,1996). In addition, exogenous application of FGF-8 into the flank mesoderm induces ectopic limb bud (Cohn et al.,1995; Ohuchi et al.,1995). Taken together, it is possible that Fgf-8 expressed in the ectoderm of the presumptive limb field enhances the adhesiveness of mesodermal cells beneath the ectoderm, ultimately promoting cell proliferation and limb outgrowth.

T-box protein 5 (Tbx5) is crucial for initiation of the forelimb bud, and deletion of Tbx5 in mice leads to failure of forelimb formation (Rallis et al.,2003; Minguillon et al.,2005). In addition, a possible role for Tbx5 in cell adhesiveness of the limb mesenchyme has been proposed (Ahn et al.,2002). During zebrafish development, tbx5 is exclusively expressed in the pectoral fin bud. Knockdown of tbx5 in the zebrafish affects the formation of the pectoral fin bud (Ahn et al.,2002). The affected pectoral fin bud formation is rescued by transplantation of fin bud cells from wild-type embryos but not by cells previously injected with a tbx5 morpholino. The latter are not able to populate and contribute to fin bud formation (Ahn et al.,2002), suggesting that tbx5-expressing cells tend to aggregate in the fin field. Taken together with the increased adhesiveness of mesodermal cells of the limb field, it is likely that Tbx5 regulates the adhesiveness and/or motility of a subset of mesodermal cells that populate the presumptive limb field as limb precursor cells.

Besides cell adhesiveness, cell migration also contributes to limb bud outgrowth. Oriented migration of mesodermal cells into the limb bud is observed in mouse, chick, and zebrafish embryos (Wyngaarden et al.,2010). Mesodermal cells of the flank region migrate into the limb field, where they condense before limb formation. It is possible that increased adhesiveness of mesodermal cells of the limb field contribute to this process. Migration of mesenchymal cells toward the limb bud is regulated by Wnt5a (Wyngaarden et al.,2010). In the developing limb bud, Wnt5a is expressed in the distal mesoderm and the apical ectodermal ridge (AER) (Dealy et al.,1993; Kawakami et al.,1999). Deletion of Wnt5a inhibits distal outgrowth of the limb bud and leads to truncation of the distal limb skeleton (Yamaguchi et al.,1999), suggesting that Wnt5a regulates limb outgrowth by modulating proliferation and polarity of the mesenchyme in the distal region. Wnt5a also acts as a chemoattractant for the lateral plate mesoderm during limb initiation (Wyngaarden et al.,2010). In Wnt5a-deficient mice, cells in the lateral plate mesoderm fail to migrate in an oriented and coordinated fashion into the limb field. In addition, implantation of a bead containing Wnt5a in the limb field alters the direction of mesodermal cells, and many of the cells actually migrate toward the bead (Wyngaarden et al.,2010). Taken together, the oriented migration regulated by Wnt5a leads to accumulation of cells in the limb field and, in cooperation with the increased adhesiveness, to outgrowth of the limb bud.

The FGF family is also involved in the regulation of cell migration during limb development. Fgf-4, which is expressed in the posterior half of the AER, stimulates distal migration of cells that lie beneath the AER (Li and Muneoka,1999), indicating a role of FGF-4 as a limb mesenchyme chemoattractant. Although the role of FGF-8 on limb mesenchyme movement has not been reported, it otherwise acts as a chemoattractant for the cranial mesenchyme (Creuzet et al., 2005). Therefore, it is possible that Fgf-8 expressed in the limb field regulates migration, adhesiveness, and proliferation of mesodermal cells in the limb field in order to promote limb initiation and outgrowth.


Cell Property Changes Affect Limb Pattern Formation

One of the main issues in limb development research is the understanding of how the undifferentiated mesenchymal cells of the limb bud form cartilaginous elements that have individual morphology (reviewed by Towers and Tickle,2009b). As described in the Introduction section, the regulation of condensation is a significant step in limb morphogenesis. The developmental changes in cellular properties are key in the regulation of condensation. Several reports have focused on cell adhesiveness from the viewpoint of cartilage morphogenesis of the limb bud. For example, adhesion and movement of limb mesenchymal cells have been studied in cells derived from the limb bud of normal and talpid3 (ta3) mutant embryos, the latter of which exhibit a polydactyl phenotype with unrecognizable digit morphology. Mesenchymal cells from ta3 limb bud adhere tightly to each other and form spherical cellular aggregates in rotation culture, whereas cells from wild-type limb bud adhere loosely (Ede and Agerbak,1968; Ede and Flint,1975), suggesting that ta3-mutant cells have stronger adhesiveness than wild-type cells. In addition, the motility of mesenchymal cells in ta3 limb bud is decreased as compared with that of cells in normal limb buds (Ede and Agerbak,1968; Ede and Flint,1975). These results suggest that regional differences in cellular properties affect cartilage morphogenesis by regulating the size, place, and number of mesenchymal condensation in the limb bud. Differences in the limb mesenchyme were confirmed by cell sorting (Ide et al.,1994; Wada and Ide,1994).

Developmental Change in the Cell Surface Properties of the Limb Mesenchyme

Limb mesenchymal cell adhesiveness is a crucial factor in the regulation of the mesenchymal condensation that precedes chondrogenesis. Cell adhesiveness in chick limb bud is spatiotemporally distinct. Cells derived from the distal region of different developmental stages are segregated from each other (Fig. 2A and B) (Ide et al.,1994; Wada and Ide,1994). A similar segregation between two groups of cells is observed when cells from different positions along the proximal–distal (PD) axis are mixed (Fig. 2C). The adhesiveness of distal cells at early stages is similar to that of cells of the proximal region of later-stage limb buds (Wada and Ide,1994). The process of sorting out ends within 20 hr from the beginning of the culture, suggesting that the process is independent of chondrogenesis, which begins 48 hr later. The process of cell sorting was reanalyzed later by means of a live imaging technique (Barna and Niswander,2007). The distinct adhesiveness of the limb mesenchyme along the PD axis has also been reported in mouse (Stadler et al.,2001) and Xenopus (Koibuchi and Tochinai,1998; Ohgo et al.,2010), indicating that the position-specific cell adhesiveness during limb development is a conserved mechanism across species. In addition, distinct cell adhesiveness is also observed in blastema cells of the regenerating limb, as discussed later (Nardi and Stocum,1983). Therefore, it is possible that the position-specific adhesiveness is a general mechanism underlying the regulation of both limb development and regeneration.

Figure 2.

Mesenchymal cells from the limb bud have spatiotemporally distinct adhesiveness. Two groups of mesenchymal cells were prepared from various regions or stages of the chick limb bud. Cells of one group were labeled with PKH26, a red fluorescent dye, mixed with unlabeled cells of the other group, and incubated after seeding on a culture dish. Twenty-four hours after incubation, the distribution of labeled cells was observed. A: Labeled cells (red) are evenly distributed when they are mixed with unlabeled cells derived from the same stage and position of the limb bud (black). B–D: Labeled cells are sorted out from unlabeled cells derived from different developmental stage (B), or different position (C, D) of the limb bud.

Position-dependent cell affinities of the limb mesenchyme are also observed along the anterior–posterior (AP) axis of the chick limb bud (Wada and Ide,1994), suggesting that limb mesenchymal cells divide into several subgroups of cells exhibiting unique adhesiveness properties. The differential adhesiveness is maintained until later stages. This is supported by the observation that cells from different digit primordia of the wing bud sort out from each other in vitro (Fig. 3) (Ide et al., unpublished data). The result suggests that one digit primordium in the chick limb bud might be formed from one subgroup of mesenchymal cells that exhibit similar adhesiveness characteristics, and that the individual morphology of each digit might reflect the adhesiveness of each primordium. In the mouse limb, digit 2 and 3 are formed from common cartilaginous aggregate, and digit 4 and 5 also emerge from one aggregate (Zhu et al., 2008), suggesting that each digit is firstly formed from a common aggregate, but gradually separated from each other through the alteration of cell adhesiveness in each primordium. Sorting of cells derived from different positions along the PD and AP axes of the limb bud also occurs in recombinant limb buds (Wada et al.,1993; Omi et al.,2002), suggesting that the position-dependent cell–cell recognition and subsequent cell sorting are not in vitro artifacts, but are rather part of a general event that occurs during in vivo morphogenesis.

Figure 3.

Cells from different digit-forming regions have distinct adhesiveness to reflect digit identity. Mesenchymal cells were prepared from each digit-forming region of the wing bud, and mixed with cells from other regions. In this experiment, chick-derived cells were mixed with quail-derived cells, and chick cells were detected with anti-chick-specific antibody. Chick cells (white) are distributed evenly when quail cells (black) are prepared from the same digit-forming region. On the other hand, chick cells are segregated when mixed with quail cells from a different digit-forming region.

In contrast to the other two axes, it is unclear whether cells occupying different positions along the dorsal–ventral (DV) axis have distinct adhesiveness. The cartilaginous aggregates are formed at the medial region along the DV axis of the limb bud. As a consequence, no apparent morphological boundary is formed along the DV axis of the limb skeleton. On the other hand, as assessed by clonal analysis of cell lineage in the mouse limb, cell intermingling between the dorsal and ventral halves does not occur, suggesting the existence of a boundary along the DV axis of the limb bud (Arques et al.,2007). LIM homeobox transcription factor 1b (Lmx1b) is a candidate protein for the regulation of cell migration along the DV axis of the limb bud. Expression of Lmx1b is restricted to the dorsal half of the limb bud with a clear boundary separating the ventral half (Riddle et al.,1995; Arques et al.,2007; Qiu et al.,2009). Moreover, the limbs of Lmx1b deleted mice show a biventral structure (Chen et al.,1998). Fate mapping analysis showed that Lmx1b-negative cells tend to segregate from Lmx1b-positive cells in vivo (Qiu et al.,2009), suggesting that Lmx1b regulates dorsal-specific cell adhesiveness in order to restrict cell intermingling and circumscribe the dorsal compartment. The different cellular properties of the limb mesenchyme of the dorsal and ventral half are also recognized by the differential distribution of motor neurons along the DV axis in the limb bud (Kania and Jeseell,2003). Lmx1b-knockout mice exhibit disrupted patterning of limb muscles and joints (Chen et al.,1998). In conclusion, it is likely that compartmentalization along the DV axis resulting from different cell adhesiveness characteristics has a role in proper tissue patterning during limb morphogenesis.

Molecular Basis of Position-Dependent Cell Adhesiveness

Transcription factors.

Various transcription factors are expressed in the developing limb bud in a region-specific manner. Therefore, it has been suggested that these transcription factors might be involved in position-dependent cell adhesiveness.

Hox genes are expressed in a region-specific manner in the limb bud, where they regulate growth and adhesion of mesodermal cells (reviewed by Pearson et al.,2005). Hoxa13 is expressed in the distal-posterior region of the limb bud at early stages. At later stages, the expression localizes to the presumptive autopod-forming region (Yokouchi et al.,1991; Nelson et al.,1996). As expected from its expression pattern, Hoxa13 is involved in autopod formation in cooperation with Hoxd13, the expression of which is partially overlapped with Hoxa13. Misexpression of Hoxa13 in the whole limb bud affects morphogenesis of proximal cartilaginous elements, such as the stylopod and zeugopod, that are normally formed from Hoxa13-negative cells (Yokouchi et al.,1995). Disruption of Hoxa13 in the limb bud causes hypoplasia of the autopod (Fromental-Ramain et al.,1996; Stadler et al.,2001). Notably, Hoxa13/d13 double knockout mice completely lack their autopods (Fromental-Ramain et al.,1996), indicating that Hoxa13, in cooperation with Hoxd13, functions in the regulation of autopod formation.

Hoxa13 regulates limb skeletal patterning by controlling cell growth and adhesiveness. Misexpression of Hoxa13 in the proximal region alters the adhesiveness of the local cells. Hoxa13-misexpressed cells are sorted out from Hoxa13-negative cells (Yokouchi et al.,1995). In addition, Hox13 paralogue genes coordinately regulate the expression of the cell surface receptor, EphA, during limb development (Stadler et al.,2001; Cobb and Duboule,2005; Salsi and Zappavigna,2006; Kawakami et al.,2009). Interactions of Eph receptors with their ephrin ligands lead to cellular recognition and subsequent cell repulsion. Therefore, it is possible that Hox13 genes regulate cell adhesiveness and migration by controlling the expression of Eph/ephrin proteins in the limb bud.

Another homeobox gene, Meis, is also involved in adhesiveness of the limb mesenchyme. In embryos of various species, both Meis1 and 2 are expressed in the proximal region of the limb bud (Mercader et al.,1999). The expression of Meis localizes to the presumptive stylopod-forming region, indicating a role for Meis in PD axis formation during limb development. Misexpression of Meis in the distal region of the limb bud proximalizes the positional identity of the limb mesenchyme. Expression of HoxA11 in the limb bud is expanded distally, whereas expression of HoxA13 is shortened along the PD axis (Mercader et al.,2000). In addition, the position of the zeugopod is shifted distally, and autopod formation is affected (Mercader et al.,1999,2009), indicating that a distal to proximal transformation occurs in the Meis-misexpressed limb bud. When Meis is misexpressed, the cellular adhesiveness of the limb mesenchyme is also proximalized (Mercader et al.,2000), suggesting that one of the roles of Meis in limb development is to specify the proximal identity of the limb mesenchyme by modulating cell adhesiveness.

Expression and function of Meis genes and their products are tightly connected with retinoic acid (RA) signaling. Recent reports have shown that RA plays a role in PD patterning. High levels of RA signaling determine proximal identity and contribute to the formation of proximal structures by activating Meis1/2, which in turn proximalize cellular identity (Mercader et al.,2000). It has been shown that CYP26 is involved in RA-mediated signaling (Yashiro et al.,2004). In addition, RA signaling affects cell adhesiveness, suggesting its involvement in the distinct adhesiveness of the limb mesenchyme. Pretreatment of the limb mesenchyme with RA interferes with segregation of cells along the PD axis (Tamura et al.,1997). This, in turn, also indicates the apparent relationship between cellular positional identity and cell adhesiveness.

Cell surface molecules.

Cell adhesiveness is primarily regulated by cell surface molecules. It is possible that the transcription factors discussed above regulate the expression of cell recognition and/or cell adhesion molecules. Ephrin–Eph interactions and cadherins are candidates for the regulation of position-specific cell adhesiveness.

Ephrin are the cell surface ligands of the Eph tyrosine kinase-type cell surface receptors. Ephrins are classified into 2 classes: type A and B. Type A ephrins are produced as a glycosylphosphatidylinositol (GPI)-anchored cell surface protein and bind preferentially to EphA receptors, while type B ephrins are membrane-spanning proteins that bind to EphB receptors, with some exceptions (Holder and Klein,1999; Wilkinson,2001). Eph–ephrin interactions induce repulsive movement between neighboring cells by altering intracellular signaling (Wilkinson,2001; Pasquale,2005). These interactions regulate cell migration and distribution in various developmental systems, such as neural crest migration, retinotectal projection, and rhombomere formation, to settle cells into the proper position in the developmental field of the systems (Kullander and Klein,2002).

Ephrin-A2 and ephrin-A5 proteins are predominantly distributed in the proximal region of the chick limb bud (Wada et al.,2003). Ephrin-A2 is strongly distributed through the presumptive stylopod and zeugpod, but weakly distributed through the autopod region, where one of the ephrin-A2 receptors, EphA4, is highly expressed (Ohta et al.,1996; Patel et al.,1996). Misexpression of ephrin-A2 in the distal region of the limb bud alters cell adhesiveness, and ephrin-A2-misexpressed cells are sorted out from normal cells in vitro. In addition, ectopic cell sorting is also observed in vivo, and induces skeletal malformations of the limb such as partial duplication or fusion of digits (Wada et al.,2003). By contrast, misexpression of ephrin-A2 has no effect on skeletogenesis in the proximal region. Therefore, it is possible that different amounts of ephrin-A2 along the PD axis influence skeletal morphogenesis in the limb bud by affecting cell adhesiveness.

In addition to EphA4, other EphA receptors, such as EphA3 and EphA7, are also expressed in the distal region of mouse limb bud (Stadler et al.,2001; Cobb and Duboule,2005; Kawakami et al.,2009). EphA7 is expressed in the distal region of the limb bud, and restricted to presumptive digits. In vitro inhibition of EphA7 function by blocking antibodies causes incomplete cartilaginous nodule formation, suggesting the involvement of EphA7 in the mesenchymal condensation that precedes cartilage differentiation. The expression of EphA7 is decreased in Hoxa13-knockout limbs (Stadler et al.,2001), and is regulated by both Hoxa13 and Hoxd13 (Salsi and Zappavigna,2006). Thus, EphA7 acts as a downstream effecter of Hox13 genes to regulate cell migration and cartilage morphogenesis during limb development.

Treatment with PI-PLC, an enzyme that removes GPI-anchor type cell surface protein from the cell membrane, interferes with sorting of cells from different PD positions, suggesting the involvement of GPI-anchored protein in distinct cell adhesiveness along the PD axis of the limb bud (Wada et al.,1998). Since ephrin-A2 is a GPI-anchored protein, GPI-anchored protein-dependent cell sorting may be partially mediated by ephrin-A2. In addition, position-dependent tissue engulfment of the regenerating blastema is also inhibited by PI-PLC treatment (see next section) (da Silva et al.,2002), suggesting that GPI-anchored proteins are generally involved in position-dependent cell–cell recognition during limb morphogenesis.

On the other hand, despite the involvement of GPI-anchored protein in cell adhesiveness, the role of GPI-protein-mediated adhesiveness in vivo remains unclear. The limbs in mice lacking PIGA, which encodes PIGA, an enzyme that is involved in the biosynthesis of the GPI-anchor, show severe chondrodysplasia phenotype, and delayed osteogenesis (Ahrens et al.,2009). Despite these skeletal defects, the skeletal pattern in the limb is barely affected, and a rough outline of skeletal pattern appears to be established without a GPI-anchored protein. Since GPI-anchored protein affects cell adhesiveness as shown above, it is possible that GPI-anchored protein-mediated adhesiveness may play a role other than initial skeletal patterning. Another possible role for GPI-anchored proteins in limb development is the formation of properly shaped cartilaginous nodules. The nodule formation of limb mesenchyme in vitro is disrupted by misexpression of ephrin-A2 (Wada et al.,2003). In this case, the expression of chondrogenic markers is maintained, suggesting the disruption of nodule formation was not a result of inhibition of chondrogenesis itself. In addition, inhibition of EphA7 function in vitro also disrupts nodule formation (Stadler et al.,2001), suggesting that interaction of EphA7 with GPI-ligand(s) regulates nodule formation. Moreover, treatment of limb mesenchyme with PI-PLC in vitro also disrupts nodule formation (Fig. 4). Therefore, GPI-anchored proteins expressed in the limb mesenchyme may regulate the proper cell–cell interaction, which is necessary for cartilaginous nodule formation. Disrupted morphogenesis of the limb cartilage observed in PIGA-deficient mice (Ahrens et al.,2009) may be due to interference of nodule formation.

Figure 4.

Nodule formation in vitro is affected by PI-PLC treatment. Mesenchymal cells from the distal half of stage-24 limb buds were prepared, seeded, and incubated for 3 days. A: The control culture forms round cellular aggregates, the cartilaginous nodules (arrowheads). B: Addition of PI-PLC to the culture medium inhibits cartilaginous nodule formation, and the morphology of the cells appears uniform.

Class-B ephrins are also expressed in the limb bud (Flenniken et al.,1996; Baker et al.,2001). The somatic cells from ephrin-B1 haploinsufficient female mice are composed of ephrin-B1-positive or ephrin-B1-negative cells after random inactivation of the X chromosome, because ephrin-B1 is mapped to the X-chromosome (Compagni et al.,2003; Davy et al.,2004). Segregation between ephrin-B1-positive and ephrin-B1-negative cells occurs in the limb bud, and skeletal morphogenesis of the limb bud is affected. This result suggests that class-B ephrins are also involved in cellular adhesiveness of the limb mesenchyme.

The involvement of the cadherin family of cell adhesion molecules in position-specific cell adhesiveness has also been suggested. Several cadherins are expressed in the limb bud in a region-specific manner (Kimura et al.,1995; Kitajima et al.,1999; Yajima et al.,1999,2002). The distribution of N-cadherin reflects the positional identity of the limb mesenchyme along the PD axis of the limb bud (Yajima et al.,1999,2002). N-cadherin is weakly distributed at the early stage of limb development. At later stages, its expression is still weak in the proximal region of the limb bud, but becomes strong in the distal region, and forms a gradient along the PD axis of the limb bud (Yajima et al.,1999,2002), suggesting that the amount of N-cadherin protein regulates position-specific adhesiveness. Position-dependent sorting of cells of the limb mesenchyme is affected by misexpression of N-cadherin or inhibition of N-cadherin protein, implying the involvement of N-cadherin in position-specific cell adhesiveness (Yajima et al.,1999,2002).

In addition to N-cadherin, other cadherins, such as cadherin-11 (Kimura et al.,1995) and PB-cadherin (Kitajima et al.,1999) are also expressed in the distal region of the limb bud, indicating that the total amount of cadherin proteins in the cells of the distal region is higher than that of the proximal region. This distribution possibly affects position-dependent cellular adhesiveness and might explain stage-dependent aggregate formation in rotation culture (Fig. 1). Cadherin-11 is involved in cell sorting of the limb mesenchyme in vitro (Kimura et al.,1995), and cadherin-dependent cell adhesion regulates chondrogenesis of the limb mesenchyme (Kim et al.,2009). The polarized distribution of cadherin proteins in the limb bud may affect position-dependent chondrogenesis and cartilage morphogenesis thorough modification of cell adhesiveness.


Involvement of position-dependent cell adhesiveness in limb morphogenesis is also observed during the process of regeneration of the urodele limb (Nardi and Stocum,1983; Crawford and Stocum,1988). In normal regeneration, the blastema tissue formed at the proximal level of the limb can sequentially form the proximal to distal structures of the limb (i.e., stylopod to autopod), while the tissue formed at the distal region can form only distal elements (i.e., only autopod), indicating that cells in the blastema maintain their positional identity along the PD axis. Positional identity of cells in the blastema is represented as their position-dependent adhesiveness. When the blastema at the distal level of the limb is transplanted next to that at the proximal level, the former is displaced distally until it reaches its original level and regenerates the distal structure. By contrast, no distal displacement occurs when the proximal blastema is grafted to the proximal region. Thus, cells in the grafted blastema can recognize their current position along the PD axis by contact-dependent cell–cell interaction. Thus, distinct adhesiveness of the cells has a key role in positional recognition. When the blastema tissue at the distal level is attached to that at the proximal level in hanging drop cultures, the latter tissue engulfs the former tissue (Nardi and Stocum,1983). Blastema tissues at the same levels do not show such engulfment, suggesting that cells in the blastema have position-dependent adhesiveness along the PD axis, and that cells of the distal blastema are more adhesive than cells of the proximal blastema. A recent experiment using GFP-transgenic axolotl also showed that the distal-cartilage-derived cells moved distally during regeneration when they were grafted to the proximal cartilage before amputation (Kragl et al.,2009), indicating that the position-dependent sorting out of blastema tissue is based on individual adhesiveness of the cells. Position-dependent cell adhesiveness of the regenerating limb blastema has also been reported in Xenopus (Ohgo et al.,2010).

The position-dependent adhesiveness of blastema cells is regulated by RA signaling, similar to that of cells in the developing limb bud. The adhesiveness of cells in the distal blastema is proximalized by RA treatment (Crawford and Stocum,1988), or by ectopic activation of the RA receptor (RAR) (Pecorino et al.,1996). Such proximalization of cell adhesiveness is consistent with the finding of RA-induced skeletal proximalization in the regenerating limbs (Maden,1983). These observations suggest that downstream molecules of the RA/RAR signaling pathway are involved in the proximal–distal positional identity of the regenerating blastema through regulation of cell adhesiveness. One candidate molecule is Prod-1, a GPI-anchored cell surface protein (da Silva et al.,2002). Prod-1 is strongly expressed in the proximal blastema, and its expression gradually decreases along the PD axis. Inhibition of Prod-1 function disrupts position-dependent engulfment of blastemal tissues, suggesting the involvement of Prod-1 in cell–cell recognition of blastema cells. In addition, treatment with PI-PLC also disrupts the engulfment, emphasizing the involvement of Prod-1 in position-dependent cell recognition of blastemal cells (da Silva et al.,2002). Moreover, overexpression of Prod-1 in the limb blastema induces displacement of cells to proximal regions, and affects distal structure formation (Echeverri and Tanaka,2005). Taken together, these results indicate the involvement of Prod-1 in proximal–distal pattern formation of the regenerating limb through position-specific cell adhesiveness.

Prod-1 had been initially isolated as an orthologue of CD59 in urodele (da Silva et al.,2002); however, subsequent analysis showed that Prod-1 is a similar but different molecule than CD59, and functional equivalent molecules to Prod-1 have not been reported in other species. Based on the 3-dimensional structure analysis, Prod-1 is a taxa-specific protein that has been identified only in Salamandridae and Ambystomidae, and Prod-1 is not found in the genomes in other animals, suggesting a tight connection of Prod-1 with the ability of regeneration (Garza-Garcia et al.,2010). However, it is still possible that the downstream molecules of Prod-1 signaling might be conserved in animals other than urodele, therefore, a better understanding of the Prod-1 signaling would provide hints for activation of limb regeneration ability in the animals.


As described above, regional differences in cell adhesiveness are tightly correlated with limb initiation and subsequent developmental events of the limb bud, especially region-dependent cartilage morphogenesis. In addition, adhesiveness may also influence patterning of tissues other than cartilage in the limb bud, such as muscles and neuronal axons, through cartilage morphogenesis. In this review, the focus has been exclusively on the regional changes in adhesiveness of limb mesenchyme. However, this change may, in turn, influence other cellular properties, such as cell migration or polarity of cell division, and it is possible that regional changes of cellular properties as a whole affect limb morphogenesis. Therefore, it would be of great importance to gather information on the spatiotemporal changes of these properties during limb development in order to grasp limb morphogenesis from the point of view of cell behavior.

It remains unclear whether the increased adhesiveness of mesenchymal cells during limb initiation contributes to limb field determination, to the subsequent limb budding process, or to both. Recent progress in molecular analysis has revealed the expression of several genes in the mesenchyme of the presumptive limb-forming region. A better understanding of the roles of such molecules in regulating cell adhesiveness in the limb bud and the lateral plate mesoderm will provide new angles for limb initiation research.


The results cited in this review were supported in part by Grants-in-aid from the Japanese Ministry of Education, Culture, Sports, Science and Technology.