SEARCH

SEARCH BY CITATION

Keywords:

  • ephrin;
  • Eph receptors;
  • signaling;
  • animal models

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

Eph receptors and ephrins have captured the interest of the developmental biology community in recent years for their pleiotropic functions during embryogenesis. Loss-of-function studies using various animal models have demonstrated the involvement of Ephs and ephrins in many aspects of embryogenesis including segmentation, neural crest cells migration, angiogenesis, and axon guidance. An essential property of this signaling pathway is the ability of both Ephs and ephrins to behave as receptors or ligands and their consequent cell autonomous and nonautonomous mode of action. While many reports did not discriminate between Eph autonomous signaling (forward) and ephrin autonomous signaling (reverse), recent genetic and in vivo studies have shown that both forward and reverse signaling play important roles during embryogenesis. Developmental Dynamics 232:1–10, 2005. © 2004 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

Eph receptors and ephrins are membrane-bound proteins that function as a receptor-ligand pair. While 13 Eph receptors and 8 ephrins have been identified in mammals (Tuzi and Gullick, 1994; Orioli and Klein, 1997; Pasquale, 1997), only one Eph receptor and one ephrin are found in Drosophila melanogaster and one Eph receptor and 4 ephrins are found in C. elegans (George et al., 1998; Scully et al., 1999; Wang et al., 1999; Bossing and Brand, 2002). Eph receptors and ephrins fall in two classes based on sequence homology and binding preferences. It was thought until recently that EphA receptors bound preferentially to ephrin-As (with the exception of EphA4) while EphB receptors had a preference for ephrin-Bs. However, new data suggests that interactions across classes occur and are functional in specific contexts (Himanen et al., 2004). Both classes of ephrins exhibit important structural differences since ephrin-As are tethered to the membrane by means of a glycosylphosphatidyl inositol anchor, whereas ephrin-Bs are transmembrane proteins with a cytoplasmic domain. Although interactions across classes are limited, within a class they are promiscuous, with multiple Eph receptors binding to a given ephrin and vice versa. This promiscuity is probably the basis for the functional redundancy that has been observed in vivo for both Eph receptors (see for instance Orioli et al., 1996; Feldheim et al., 2000). Eph receptors belong to the superfamily of receptor tyrosine kinases and as such, they auto-phosphorylate upon binding to their cognate ephrin ligands and subsequently activate downstream signaling cascades (forward signaling). While neither class of ephrins possesses a catalytic activity, both can activate signal transduction pathways after interaction with Eph receptors (reverse signaling). Reverse signaling activated by transmembrane ephrins includes tyrosine phosphorylation of their cytoplasmic tail and interaction with various signaling molecules. The exact mechanism by which GPI-linked ephrins stimulate downstream signaling is still unclear, however, they have been shown to activate a member of the Src-family kinases and to impinge on integrin signaling. Oligomerization and clustering of Eph receptors and ephrins at the cell surface is essential for their signaling function and might be regulated by localization in membrane microdomains (for reviews on Eph/ephrin signaling, see Cowan and Henkemeyer, 2002; Kullander and Klein, 2002).

At the cellular level, both forward and reverse signaling through Eph receptors and ephrins regulate motility. Depending on cell type and on the members of the Eph/ephrin family involved, the outcome of the interaction can be either increased adhesion (attraction), or decreased adhesion (repulsion). These functions are achieved by means of functional and/or physical interaction with adhesion proteins such as integrins as well as multiple proteins implicated in cytoskeletal organization (Murai and Pasquale, 2003).

At the level of the whole organism, Eph receptors and ephrins are involved in an ever growing number of processes during embryogenesis as well as in the adult and in some pathological conditions (Palmer and Klein, 2003). Many of the analyses of Eph receptors and ephrins in vivo are loss-of-function studies that abolish both forward and reverse signaling (either through gene targeting to create a “knock-out” allele, or functional interference using soluble forms of receptors or ligands; Fig. 1). In the past few years, however, studies on Eph receptors and ephrins have attempted to establish whether both forward and reverse signaling are required for particular developmental processes by designing strategies that interfere in vivo with either one or the other (but not both) signaling cascades. These strategies are based on the use of “signaling” mutant forms of both Eph receptors and ephrins. The most commonly used method to selectively disrupt forward signaling or reverse signaling involves the use of targeted mutations that remove the majority of the Eph or ephrin cytoplasmic region, respectively (Fig. 1). Studies using such strategies in various animal models have demonstrated that reverse signaling is required during embryogenesis and have revealed that in some cases it can even act independently of forward signaling (Table 1). This indicates that Ephs act as bona fide ligands for ephrins in specific developmental processes. In addition, these analyses have revealed that reverse signaling, like forward signaling, can be either attractive or repulsive. They have not, however, shed light on the mechanisms underlying the switch between attractive and repulsive behaviors. This review focuses on such studies, particularly as they relate to axon guidance, segmentation, cell migration and angiogenesis during development. The majority of the studies discussed concern transmembrane ephrins, however, reverse signaling downstream of GPI-linked ephrins is discussed in a separate section. Ephrin reverse signaling also has been implicated in adult physiological processes, such as synaptic plasticity (Grunwald et al., 2001; Palmer and Klein, 2003), but will not be discussed here.

thumbnail image

Figure 1. Schematic representation of transmembrane ephrins, Eph receptors and how truncation of these proteins affect forward or reverse signaling, respectively. YYKV is a motif that is recognized by PDZ-containing proteins. Such a motif is also present at the C-terminal tail of Eph receptors. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Table 1. Summary of the Developmental Processes Requiring Ephrin-Induced Reverse Signalinga
 ProcessMechanismSignalingOrganismReference
  • a

    ac, anterior commissure; NCC, neural crest cells; RGC, retinal ganglion cells; acP, lateral projection of the anterior commissure; acA, anterior commissure.

acPAxon guidanceRepulsionUnidirectionalMiceHenkemeyer et al., 1996; Cowan et al, 2004
acP/acAAxon guidanceAttractionUnidirectionalMiceKullander et al., 2001b
RGCAxon guidance (optic disc)RepulsionUnidirectionalMiceBirgbauer et al., 2000
RGCAxon guidance (retinotectal)AttractionUnidirectionalXenopusMann et al., 2002
Vomeronasal axonsAxon guidanceAttractionUnidirectionalMiceKnoll et al., 2001
HindbrainCell sortingRepulsionBidirectionalZebrafishXu et al., 1999; Mellitzer et al., 1999
SomitesBoundary formationRepulsion?UnidirectionalZebrafishDurbin et al., 1998
NCCMigration?UnidirectionalMiceDavy et al., 2004
Urorectal regionMigrationAdhesionBidirectionalMiceDravis et al, 2004
EpidermisMigrationAttraction?BidirectionalC. elegansChin-Sang et al., 1999 Wang et al., 1999

AXON GUIDANCE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

Eph receptors and ephrins were first studied for their role in axon guidance. Their reported mutually exclusive and graded expression patterns in navigating axons and their targets were consistent with an important role in topographic mapping of axonal projections. It was shown that cells expressing Eph receptors avoided territories expressing ephrins, thus providing necessary cues to guide axons to their appropriate target (O'Leary and Wilkinson, 1999). It has been recognized more recently that Eph receptors and ephrins can also regulate axon pathfinding through attractive interactions (Knoll et al., 2001; Kullander et al., 2001b; Hindges et al., 2002; Mann et al., 2002; Eberhart et al., 2004) and that ephrins can act as receptors on navigating axons (see below).

The pioneer study that led to the concept of bidirectional signaling focused on the biological function of EphB2 in mice (Henkemeyer et al., 1996). Along with a null allele, the authors generated an allele that lacked the tyrosine kinase domain and C-terminal domain of the EphB2 receptor, but maintained its extracellular, transmembrane and juxtamembrane domains, fused to β-galactosidase. Histological examinations of brains from adult EphB2 null mice revealed specific defects in axon guidance of the lateral projection of the anterior commissure (acP). The second axon tract of the anterior commissure (acA), projecting to the olfactory bulbs, was unaffected in homozygous null mutants. By using dye-tracing experiments, the authors showed that in the mutants, cortical neurons were unable to extend axons toward the midline. Unexpectedly, when brains from mutant mice carrying the kinase inactive form of the EphB2 receptor were analyzed, no defect was observed in the acP tract, indicating that the catalytic activity of the receptor is dispensable for proper guidance of these particular nerve fibers. Expression studies demonstrated that EphB2 must act non-cell autonomously on the acP tract because it is not expressed by these neurons, but instead is detected in the cells underlying the acP. In addition, because in wild-type but not EphB2 null animals acP axons never invade the area where EphB2 is expressed, the authors concluded that EphB2 acts as a repulsive cue to guide acP fibers toward the midline. Because ephrins were detected in the acP tract, the authors proposed that EphB2 is acting as a ligand to activate an ephrin-induced reverse signaling and direct the migration of ephrin-expressing axons (Henkemeyer et al., 1996). This hypothesis was confirmed very recently with the demonstration that ephrin-induced reverse signaling is required to regulate proper guidance of the acP tract (Cowan et al., 2004). In this study, it was reported that mice carrying a deletion of the cytoplasmic domain of ephrin-B2 exhibit a defect in the acP tract similar to the EphB2 null animals, indicating that the intracellular domain of ephrin-B2 is required for normal pathfinding of these nerve fibers (Fig. 2A). Because ephrin-B2 is expressed in the acP tract (Cowan et al., 2004), it can be inferred that ephrin-B2 acts cell-autonomously in this tract by activating a signaling cascade upon interaction with EphB2 receptor expressed on adjacent cells (Fig. 3). It should be noted, however, that the rescue of the acP tract formation by expression of a truncated form of EphB2 was observed only in certain genetic backgrounds which might indicate a more complicated situation.

thumbnail image

Figure 2. Different uses of signaling mutants in vivo. A: The ephrin signaling mutant is expressed in place of the wild-type protein. If the signaling mutant rescues the phenotype associated with loss of ephrin, then reverse signaling is not required for normal function. On the other hand, if the signaling mutant does not rescue the phenotype, this suggests that reverse signaling is required for function. B: The signaling mutant is overexpressed in a tissue expressing the wild-type protein. If a phenotype is associated with overexpression of the ephrin signaling mutant, it can be due to a dominant negative effect of the mutant inhibiting normal reverse signaling by sequestering available Eph receptors. However, control experiments have to be performed to ascertain that the phenotype is not due to overstimulation of the forward signaling. C: In a situation where Eph receptors are present, if expression of wild-type ephrin and signaling mutants induce a similar phenotype, this suggests that reverse signaling is not required for this function. On the contrary, if the signaling mutant is unable to induce the phenotype, then reverse signaling is essential. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

thumbnail image

Figure 3. Reverse signaling is involved in guiding AcP axons. A: In wild-type animals, ephrin-B2-expressing AcP axons avoid the EphB2-expressing territory. B: In EphB2 knock-out (K/O) animals, AcP axons improperly invade this territory. C: Expression of a truncated, β-galactosidase conjugated form of ephrin-B2 phenocopies the EphB2 null mutation. This could indicate either that the cytoplasmic tail of ephrin-B2 is required for the guidance decision or that the mutant ephrin-B2 does not activate the forward signaling properly. D: Expression of a truncated, β-galactosidase conjugated form of EphB2 rescues the EphB2 null phenotype, demonstrating that forward signaling is dispensable for the guidance decision and establishing that reverse signaling is required for this process. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Another Eph receptor, EphA4 (which can bind ephrin-B2 and ephrin-B3 in addition to ephrin-As), has also been shown to act as a ligand to control the formation of the anterior commissure tract (Kullander et al., 2001b). Unlike what was observed in the EphB2 null mice, loss of EphA4 affected both the acA and the acP tracts in a non-cell autonomous manner (Dottori et al., 1998; Kullander et al., 2001b). This phenotype was rescued by expression of kinase defective mutant receptors (harboring a point mutation in the kinase domain that abolishes catalytic activity), suggesting that EphA4 acted as a ligand to regulate pathfinding of ephrin-expressing axons (Kullander et al., 2001b). Of interest, the authors postulated that EphA4 might provide an attractive cue to ephrin-expressing axons because nerve fibers of the anterior commissure migrate away from the EphA4 expression domain in the mutant animals (Kullander et al., 2001b). The fact that EphA4 seems to be attractive for ephrin-expressing axons, while EphB2 appears to induce repulsion of the same axons indicates that they might not be acting through the same ephrin. Because ephrin-B2 and EphB2 seem to be acting together and ephrin-B2 mutant mice do not phenocopy EphA4 mutant mice in the anterior commissure (Kullander et al., 2001b; Cowan et al., 2004), ephrin-B2 is not a likely candidate for responding to EphA4. Ephrin-B3 is a known ligand for EphA4, however no defects of the anterior commissure were reported in the ephrin-B3 null mice (Kullander et al., 2001a), raising the possibility that EphA4 is acting as a ligand for one of the ephrin-As. Consistent with this, GPI-linked ephrins are capable of inducing a reverse signaling cascade that increases cell adhesion (Davy et al., 1999; Davy and Robbins, 2000; Huai and Drescher, 2001) and they have also been shown to act as receptors in axonal guidance (Knoll et al., 2001).

Loss of EphA4 also affected the formation of another major axon tract: the corticospinal tract (CST). A significant number of CST axons crossed back across the spinal cord midline aberrantly in the mutant mice (Dottori et al., 1998; Kullander et al., 2001b). This process was dependent on the forward signaling activated downstream of EphA4 since mice expressing the kinase defective mutants alleles of EphA4 exhibited the same CST defects as the null mice (Kullander et al., 2001b). Because EphA4 was shown to be expressed on CST fibers (Kullander et al., 2001b), the authors concluded that EphA4 acts cell autonomously in these neurons to regulate pathfinding. In a complementary study, a different group demonstrated that ephrin-B3 acts as the midline repulsive cue preventing EphA4-expressing CST fibers from recrossing to the ipsilateral side of the spinal cord independently of reverse signaling (Yokoyama et al., 2001). Indeed, ephrin-B3 is expressed at the spinal cord midline and loss of ephrin-B3 causes a CST phenotype similar to the EphA4 mutations (Kullander et al., 2001a; Yokoyama et al., 2001). This phenotype could be rescued by expression of a truncated version of ephrin-B3 lacking its cytoplasmic domain, indicating that forward signaling is sufficient to prevent CST fibers from recrossing the spinal cord midline (Yokoyama et al., 2001).

Ephrin-induced reverse signaling has also been implicated in retinal axon pathfinding. Mice lacking both EphB2 and EphB3 receptors exhibit retinal ganglion cell (RGC) axon pathfinding errors characterized by aberrant defasciculation of RGC axons as they migrate toward the optic disc (Birgbauer et al., 2000). In some mutants, these axons never reached the optic disc, while in other cases, axons grew into the opposite side of the retina. Because this phenotype could be rescued by a kinase-dead mutant EphB2 receptor, the authors concluded that EphB2 (and possibly EphB3) might be acting as ligand for ephrin-expressing axons (Birgbauer et al., 2000). Consistent with this, EphB2 and EphB3 seem to be acting non-cell autonomously because dorsal RGC axons, where Eph receptor expression is low, are more affected than ventral RGC axons in the mutant mice (Birgbauer et al., 2000). However, expression patterns of transmembrane ephrins in the retina are not consistent with a straightforward role for one or several ephrins acting as receptor(s) on RGC axons. Indeed, ephrin-B1 and ephrin-B3 are expressed uniformly in the RGC layer throughout the retina, while ephrin-B2 is not expressed in RGCs. A possible explanation for the difference in dorsal vs ventral axons despite uniform ephrin expression is that co-expression of both Eph receptors and ephrins in the same cell (i.e., ventral RGC) might affect in cis the outcome of the Eph/ephrin interaction in trans. Such a mechanism has previously been described in RGCs for EphA receptors. It was shown that co-expression of ephrinA ligands in EphA-expressing RGCs decreased the ability of the receptors to respond to their ligands on adjacent cells (Hornberger et al., 1999). One can thus speculate that co-expressed Eph receptors might also modulate the ability of ephrins to respond to cues on adjacent cells possibly by a mechanism involving constitutive tyrosine phosphorylation (Hornberger et al., 1999). Another possibility is that ephrins are rapidly turned over at the cell surface by endocytosis in presence of high levels of Eph receptors (endocytosis of ephrins after interaction with Eph receptors has been reported in vitro, see Davy and Robbins, 2000; Martson et al., 2003; Zimmer et al., 2003). These mechanisms would create a functional gradient of ephrin-B1 and ephrin-B3 in the retina in place of an expression gradient, which would account for the dorso-ventral asymmetry observed in mutant mice. However, it should be noted that no RGC axon pathfinding defect has been reported in the ephrin-B3 and ephrin-B1 mutant mice to date.

In addition to guiding RGC axons toward the optic disc, ephrin-induced reverse signaling also participates in topographic mapping of the retinotectal system in Xenopus (Mann et al., 2002). Perturbation of Eph/ephrin interaction causes RGC axons to shift away from the Eph receptor-expressing domain of the tectum. A similar phenotype is seen after expression of a dominant negative ephrin-B ligand (lacking the cytoplasmic domain) in dorsal RGC, indicating that the function of ephrin-Bs in retinotectal pathfinding is associated with its cytoplasmic domain (Fig. 2B; Mann et al., 2002). The same mutation had no effect when expressed in ventral RGC, however, expression of wild-type ephrin-B in these RGC caused them to shift aberrantly toward the Eph-expressing domain in the tectum. Based on these results, the authors concluded that ephrin-expressing RGC axons are attracted toward the Eph-expressing tectal region (Mann et al., 2002).

SEGMENTATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

In addition to their roles in axon guidance, Ephs and ephrins were recognized early on for their role in segmentation (Wilkinson, 2000). Initial expression studies of these proteins showed that several members of both the Eph receptor family and the ephrin family are expressed in a segmented pattern in the hindbrain and in somites, suggesting that Ephs and ephrins could have a role in segmentation during embryogenesis (Gale et al., 1996). A role for both forward and reverse signaling in hindbrain segmentation was established using zebrafish and Xenopus as a model. In the hindbrain, Eph receptors and ephrins are expressed in alternating rhombomeres (Eph receptors are expressed in odd-numbered rhombomeres and ephrins are expressed in even-numbered ones). In an initial study, it was shown that expression of a dominant negative form of EphA4 (lacking the kinase domain) in the hindbrain disrupted segmental expression of certain genes (Xu et al., 1995). In a following study, Xu et al. demonstrated that Eph receptors and ephrins are required for the proper sorting of cells at rhombomere boundaries (Xu et al., 1999). The authors reported that mosaic expression of ephrin-B2 led to the sorting of ephrin-B2-expressing cells at the boundary of r3/r5 rhombomeres while these cells were found scattered within r2/r4/r6 territories. This suggests that cells expressing ephrin-B2 are excluded from the Eph receptor-expressing rhombomeres, presumably after activation of the receptors. The sorting out process in the rhombomeres appears to involve forward signaling because a dominant negative mutant of ephrin-B2, lacking the cytoplasmic domain, was still competent to induce the sorting phenotype (Fig. 2C; Xu et al., 1999). Additionally, reverse signaling was shown to induce sorting of EphA4 expressing cells at the boundary of r2/r4/r6 rhombomeres, indicating that unidirectional forward and reverse signaling promote sorting out and restrict cell intermingling, presumably by a repulsion mechanism.

Using the same dominant negative constructs in a series of animal cap assays, it was shown that bidirectional, but not unidirectional, signaling through Eph/ephrin restricts cell intermingling in adjacent cell populations (Mellitzer et al., 1999). Based on these findings, the authors proposed that bidirectional signaling was inducing repulsion of both cell populations whereas with only unidirectional signaling, the population of cells not receiving a signal is still able to mingle.

Similar to what has been described in the hindbrain, perturbation of either forward or reverse signaling in zebrafish leads to defects in somite formation (Durbin et al., 1998). In somites, Eph receptors and ephrins are expressed in mutually exclusive anterior and posterior domains, respectively. By taking advantage of fused somites (fss−/−) zebrafish mutants in which disruption of somite formation is associated with loss of EphA4 and uniform ephrin-B2 expression in paraxial mesoderm, it was shown that reverse signaling is required for the formation of boundaries during somite morphogenesis (Barrios et al., 2003). Indeed, injection of donor cells ectopically expressing wild-type EphA4 into a fss−/− host led to formation of ectopic boundaries between donor and host cells. Similarly, injection of donor cells ectopically expressing a kinase-dead version of EphA4 induced boundary formation, demonstrating that ephrin-induced reverse signaling is sufficient for this process (Barrios et al., 2003). Forward signaling was shown to be responsible for epithelialization of the somite, in an autonomous and non-autonomous manner. Intriguingly, dominant negative ephrin does not interfere with this process suggesting that reverse signaling is not required and that Eph receptors might activate an ephrin-independent pathway on adjacent cells (Barrios et al., 2003). Whether repulsion is the mechanism by which Eph/ephrin promote boundary formation in paraxial mesoderm is still unclear. Also unclear is the reason why both in rhombomere and somite formation, unidirectional signaling was sufficient to promote cell sorting and boundary formation, while in animal cap assays bidirectional signaling was required. One possibility is that Eph/ephrin signaling regulates cell sorting in concert with adhesion molecules, which would be available or functional in regions of endogenous Eph/ephrin requirement (hindbrain and somites) but unavailable in cells which do not normally engage in Eph/ephrin interaction (animal cap cells). These cells, therefore, would require both forward and reverse signaling to compensate for the loss of an adhesive interaction.

Based on the functional studies described above, EphA4 and ephrin-B2 appear to be an important receptor/ligand pair involved in segmentation in Xenopus and zebrafish. In the mouse, EphA4 and ephrin-B2 show the same complementary expression domains. No defects in hindbrain segmentation have been reported in mice deficient for these genes, perhaps due to redundancy with other ephrins and Eph receptors (Dottori et al., 1998; Adams et al., 2001; Kullander et al., 2001b). In addition, somite morphogenesis in mice occurs in absence of ephrin-B2 and EphA4 (Wang et al., 1998; Adams et al., 1999), again suggesting a compensation mechanism by a different ephrin. Although one group reported that somite polarity appeared normal in ephrin-B2 deficient mice (Wang et al., 1998), we have observed abnormal somite development in our independent ephrin-B2 null mouse line (Davy and Soriano, unpublished observation). Because mutant mice expressing a truncated ephrin-B2 lacking the cytoplasmic domain do not present somite defects (Cowan et al., 2004), ephrin-induced reverse signaling does not appear to be necessary for establishment of somite polarity in mice.

CELL MIGRATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

Eph receptors and ephrins also regulate both cranial and trunk neural crest cell (NCC) migration (Holder and Klein, 1999; Wilkinson, 2000). The involvement of ephrin-induced reverse signaling in trunk NCC migration has not been assessed specifically because it is believed that ephrins expressed by the somites act non-cell autonomously on NCCs expressing Eph receptors (Krull et al., 1997; Wang and Anderson, 1997). In Xenopus, it was shown that Eph receptors and ephrins are expressed on adjacent streams of migrating branchial NCCs and that perturbation of Eph receptor function led to abnormal migration of NCCs (Smith et al., 1997). Expression of a dominant negative mutant of EphA4 induced scattering of NCCs and allowed them to invade improper territories. EphA4 acts in a cell autonomous manner, because only EphA4-expressing but not ephrin-expressing NCCs were affected by expression of this construct (Smith et al., 1997). Although this result implicates forward signaling in proper guidance of branchial NCCs in Xenopus, it does not rule out the possibility that reverse signaling is also involved.

Similarly, cell autonomous forward signaling has been shown to regulate branchial NCCs migration in the mouse (Adams et al., 2001). In this study, it was shown that ephrin-B2 null mice exhibit a defect in the second branchial arch, which appeared hypoplastic and abnormally vascularized. The abnormal development of the second branchial arch appeared to be due to a lack of migration of NCC into the arch. Because this phenotype was partially rescued by a mutant form of ephrin-B2 lacking the cytoplasmic domain, the authors concluded that forward signaling in Eph-expressing NCCs was necessary and sufficient for proper branchial arch development (Adams et al., 2001). To explain the specificity of the phenotype for the second branchial arch, it was postulated that ephrin-B2 expressed in the neural tube (in r4 and r6) might be involved in delamination of Eph-expressing NCCs, presumably through a repulsive mechanism (Adams et al., 2001).

Ephrin-B1 is also required for proper migration of branchial NCCs because mutant mice display a cleft palate, consistent with a defect in NCC (Davy et al., 2004). Ephrin-B1 mutant NCCs were found to invade improper territories, and these phenotypes are recapitulated when ephrin-B1 is deleted conditionally in NCC, as well as in mice expressing a mutated ephrin-B1 unable to bind PDZ-containing proteins (Davy et al., 2004). Taken together, these results indicate that ephrin-B1 acts cell autonomously in NCC to regulate their targeted migration, and that reverse signaling involving binding of PDZ-containing protein is required for this function. Surprisingly, the cleft palate observed in null animals was much more severe than when ephrin-B1 was deleted specifically in NCC, or when the PDZ binding site point mutant was expressed. This suggests that ephrin-B1 might be involved at multiple steps during NCC migration, activating both forward and reverse signaling in a cell autonomous and non-autonomous manner (Davy et al., 2004).

Very recently, forward and reverse signaling have been implicated in migration and adhesion of cells involved in septation of the urorectal region (Dravis et al., 2004). Mice expressing mutant forms of ephrin-B2 and EphB2 (β-galactosidase-conjugated truncations lacking the cytoplasmic domain and tyrosine kinase domain, respectively) presented severe anorectal malformations due to a defect in midline fusion. Interestingly, the truncated ephrin-B2 protein fused to β-galactosidase appears to act as dominant-negative for reverse signaling because heterozygotes for this lacZ allele exhibited urorectal malformations that are not seen in mice heterozygous for ephrin-B2 null alleles. Based on these results, the authors concluded that bidirectional signaling was required for proper urorectal development and that unidirectional signaling was detrimental for this process (Dravis et al., 2004). Because EphB2 and ephrin-B2 are co-expressed in the cells involved in the septation events, the authors proposed that simultaneous activation of forward and reverse signaling in the same cell leads to adhesion while unidirectional activation of either forward or reverse signaling leads to repulsion.

ANGIOGENESIS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

Several studies have implicated Eph receptors and ephrins in angiogenesis (Adams, 2002). Genetic studies in the mouse have demonstrated that deletion of ephrin-B2 and EphB4 result in identical phenotypes characterized by defective angiogenic remodeling (Wang et al., 1998; Adams et al., 1999; Gerety et al., 1999). In addition, it was shown that EphB4 is expressed on veins, while ephrin-B2 is restricted to arteries. This observation led to the proposal that this receptor/ligand pair might be involved in setting up arterial and venous identity of blood vessels, possibly by means of repulsion between ephrin-B2- and EphB4-expressing endothelial cells. Other vascular defects were reported in the ephrin-B2 null mice in addition to defective angiogenic remodeling, including partially penetrant defects in large vessel primordia and abnormal intersomitic vessels (Adams et al., 1999). The same authors subsequently provided evidence that ephrin-induced reverse signaling was required for blood vessel remodeling, because expression of a deleted form of ephrin-B2 lacking the cytoplasmic domain, was unable to rescue the angiogenesis defects associated with loss of ephrin-B2 (Adams et al., 2001).

The role of ephrin-induced reverse signaling in angiogenesis has recently become more controversial with the publication of a study reporting that mice expressing a mutant form of ephrin-B2 in which the cytoplasmic domain has been replaced by β-galactosidase do not present angiogenic defects. These mice, however, die postnatally from cardiac defects, indicating a novel role for ephrin-B2 reverse signaling in heart development (Cowan et al., 2004). It is possible that the drastic difference in phenotype observed in both studies is due to the type of mutation that was used. Indeed, the mutated proteins (HA-tagged lacking the cytoplasmic domain in the Adams study vs. replacement of the cytoplasmic domain by β-galactosidase in the Cowan study) might differ in their subcellular localization or ability to cluster at the cell surface which might impact their function. Nonetheless, the Cowan study demonstrated that angiogenesis proceeds normally in absence of ephrin-B2 cytoplasmic domain, inferring that forward signaling is sufficient for this process. The truncated HA-tagged version of ephrin-B2, which is able to function in some in vivo context (it could rescue the branchial arch defect associated with loss of ephrin-B2, see above and Adams et al., 2001), might be acting as a hypomorph allele in endothelial cells. Further resolution of forward and reverse ephrin-B2 signaling in angiogenesis will require the generation of knock-in mice carrying point mutations in the cytoplasmic domain of ephrin-B2 which should not interfere with the subcellular localization of the mutant proteins. Alternatively, mice expressing kinase-dead EphB4 receptors would be informative.

A role for EphB4-induced forward signaling in regulating angiogenic growth has also been demonstrated in Xenopus (Helbing et al., 2000). As in mice, Xenopus EphB4 and ephrin-B1 and ephrin-B2 are expressed in complementary patterns in the developing vasculature and somites respectively, and perturbation of Eph/ephrin interaction leads to defect in intersomitic vessel migration (Adams et al., 1999; Helbing et al., 2000). Overexpression of a mutant form of EphB4 lacking the tyrosine kinase domain, induced disorganization of the intersomitic vascular network, indicating that EphB4-induced forward signaling is required for normal intersomitic vessel development. Similarly, ectopic expression of truncated forms of ephrin ligands resulted in a similar phenotype. However, because ephrins are expressed in somites, but not in blood vessels, the authors concluded that ectopic activation of EphB4 is sufficient to induce the phenotype (Helbing et al., 2000).

GPI-LINKED EPHRINS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

As mentioned previously, it has been shown in vitro that ephrin-A ligands are able to activate a signaling cascade cell autonomously after interaction with a cognate Eph receptor (Davy et al., 1999; Davy and Robbins, 2000; Huai and Drescher, 2001). Because of their lack of cytoplasmic domain and because the mechanisms by which they activate downstream signaling cascades are still unclear, mutant forms of ephrin-A ligands that specifically disrupt reverse signaling have not been created. The importance of ephrin-A-induced reverse signaling in vivo can nevertheless be addressed by comparing the phenotypes of EphA protein null mutants with receptor mutants in which catalytic activity is specifically inhibited. This approach has been used in C. elegans where all four ephrins are GPI-linked to the membrane (Chin-Sang et al., 1999; Wang et al., 1999). It was shown that mutations in the kinase domain of the Eph receptor VAB-1 do not cause a complete loss-of-function phenotype, suggesting kinase-independent functions (George et al., 1998). In addition, loss-of-function mutations in ephrins synergize with the kinase mutation in VAB-1 (Chin-Sang et al., 1999; Wang et al., 1999). These results led the authors to conclude that GPI-linked ephrins could act as receptors and activate a reverse signaling pathway to regulate epidermal morphogenesis. In mice, mutations affecting specifically the kinase domain of EphA receptors have not been used extensively (with the exception of EphA4, see the axon guidance section), possibly because of the high degree of redundancy between members of the Eph receptor family which might require mutating more than one receptor to obtain a phenotype. However, GPI-linked ephrins have been proposed to act as receptors in a guidance decision affecting vomeronasal axons (Knoll et al., 2001; Knoll and Drescher, 2002). Indeed, vomeronasal axons expressing high levels of ephrin-A5 project to territories where EphA6 is high. Moreover, in stripe assays, these axons prefer to grow on Eph receptors rather than a control protein, suggesting that ephrin-A5 expressing axons are attracted toward EphA6 expressing territories (Knoll et al., 2001).

PERSPECTIVES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

The studies reviewed above, together with a host of biochemical studies, have clearly established reverse signaling as an important component of Eph receptor and ephrin activities. However, to fully understand the role of ephrins in vivo, several questions will have to be addressed. The molecular basis underlying the switch between attraction and repulsion is an important question, which also applies to forward signaling. How can ephrins induce attraction in one situation and repulsion in another? A possible explanation might be that ephrins activate different signaling cascades depending on their subcellular localization. Indeed, ephrins have been shown to be localized or translocate to lipid rafts, which are generally thought of as signaling platforms (Bruckner et al., 1999; Davy et al., 1999). A recent study of axon guidance in the chick lends support to this idea because it demonstrates that two distinct populations of EphA4-expressing axons respond to ephrin-A5 in an opposite manner (attraction vs. repulsion) and that Eph activation localizes to different subcellular compartments in each type of neuron (Eberhart et al., 2004).

It has recently been shown in vitro that both Eph receptors and ephrins are endocytosed after interaction and that this process is required for axon retraction (Mann et al., 2003; Martson et al., 2003; Zimmer et al., 2003). Endocytosis-defective mutant forms of Eph and ephrins have been described (Zimmer et al., 2003) and will be useful to study the importance of this process in vivo. Importantly, these studies raise the possibility that commonly used mutant forms of transmembrane ephrins lacking the cytoplasmic domain might have additional effects than simply abrogating reverse signaling. Because these mutants proteins are not endocytosed, forward signaling might be affected, for instance by preventing termination of the signal. Although such mutations are useful to establish a requirement for the cytoplasmic domain of ephrins in a particular process, based on these findings we now have to appreciate that they may not be sufficient to draw conclusion on reverse signaling (i.e., activation of a signal transduction cascade). A definitive approach to assess the role of ephrin-induced reverse signaling in vivo is to design point mutations that inhibit specific signaling pathways without affecting endocytosis or cell surface localization. In all cases, a combined use of both ephrin and Eph receptor signaling mutants and a clear understanding of expression pattern is the best approach to ensure that interpretation of the results is correct (Fig. 3).

Several studies have indicated that ephrin-induced signaling might be modulated by other receptor tyrosine kinases. Striking support for this type of cross-talk comes from studies in Xenopus showing that activation of fibroblast growth factor receptor (FGFR) induced tyrosine phosphorylation of ephrin-B1 and inhibited its ability to decrease cell adhesion (Chong et al., 2000). This cross-talk has recently been shown to be biologically significant in controlling morphogenetic movements of retinal progenitors in Xenopus (Moore et al., 2004). Other signaling pathways that might modulate ephrin-induced signaling include platelet derived growth factor receptor (Bruckner et al., 1997) and Wnts, by means of the cytoplasmic protein Dishevelled (Tanaka et al., 2003) and Axin (Cowan and Henkemeyer, 2001). In addition to assessing the significance of these interactions in vivo, it would be interesting to test whether ephrins, when engaged in a cross-talk with other signaling pathways, can function independently from Eph receptors. Despite the tremendous progress that has been made in comprehending the role of ephrin-induced signaling in vivo, many questions remain unanswered and these proteins will continue to capture the interest of the scientific community for years to come.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES

We thank Mark Henkemeyer for communicating results before publication, and our laboratory colleagues and Jon Cooper for critical comments on the manuscript. A.D. was supported by a fellowship from the Human Frontier Science Program. Work from the author's laboratory is supported by grants from the National Institute of Child Health and Human Development to P.S.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. AXON GUIDANCE
  5. SEGMENTATION
  6. CELL MIGRATION
  7. ANGIOGENESIS
  8. GPI-LINKED EPHRINS
  9. PERSPECTIVES
  10. Acknowledgements
  11. REFERENCES