Megalin and the neurodevelopmental biology of sonic hedgehog and retinol

Megalin (gp330/LRP-2) is a member of the low-density lipoprotein (LDL)receptor family (Fig. 1). It is expressed exclusively in epithelial cells of the embryo and the adult, in which it mediates the endocytosis of an array of ligands(Barth and Argraves, 2001). Considering the diversity of ligands that can interact with megalin, it is not surprising that mice lacking megalin display a range of embryonic abnormalities and that most die perinataly(Willnow et al., 1996). Studies of megalin^–/– survivors have provided significant insights into the role of megalin in adult physiology. For example, renal megalin is critical for vitamin homeostasis in adults,mediating proximal tubule uptake of complexes of vitamins and their carriers,including steroid 25-(OH) vitamin D3 bound to vitamin D binding protein,vitamin A (retinol) bound to retinol-binding protein and vitamin B₁₂ bound to transcobalamin(Christensen and Willnow,1999). In megalin-deficient mice these vitamins are excreted in the urine. The resulting deficiency in vitamin D leads to bone disease(Nykjaer et al., 1999). In contrast to its role in adult physiology, we know little about how megalin so profoundly influences embryonic development. Megalin-deficient embryos display numerous craniofacial abnormalities, including an absence of olfactory bulbs,absence of the corpus callosum and fusion of forebrain hemispheres hallmarks of holoprosencephaly (HPE) (Willnow et al., 1996) (Table 1). The HPE phenotype indicates that megalin is essential for the formation of the nervous system. Here we review recent findings that provide new insights into the possible role(s) of megalin in neurodevelopment.

Megalin is expressed on trophectoderm cells in the developing mouse embryo as early as the blastocyst stage (Buc-Caron et al., 1987). Later in development (8 days post coitum, dpc), it is highly expressed by cells of the extraembryonic visceral endoderm of the yolk sac and several other polarized epithelia, including the amnion, cardiac endothelium and the neural tube (Buc-Caron et al., 1987). As development progresses, megalin expression persists on apical surfaces of neural epithelia, including the neural canal,choroid plexus and ependymal cells lining the ventricles(Kounnas et al., 1994; Willnow et al., 1996). Additionally, megalin is prominently expressed by epithelial cells of the developing ear, thyroid, lung, renal proximal tubules and intestine(Kounnas et al., 1994). In all cases megalin expression is limited to the apical surfaces of the plasma membrane.

Megalin binds to and endocytoses complexes of retinal-binding protein and retinol (Christensen et al.,1999). Furthermore, this is critical for transepithelial transport of retinol by renal-proximal tubule epithelial cells(Christensen et al., 1999; Marino et al., 2001). Given these findings and the well established importance of retinol in early development (Means and Gudas,1995), it is possible that megalin is a regulator of retinol metabolism critical for embryonic development. There are at least two possible mechanisms in which megalin might play a role. Firstly, megalin might directly mediate endocytosis of retinol by cells that synthesize retinoic acid (RA). Secondly, megalin may mediate transepithelial transport of retinol and thereby facilitate delivery of retinol to cells that synthesize RA.

With respect to the first possibility, megalin would be expected to be expressed by cells that also express retinaldehyde dehydrogenases (Raldh),enzymes that oxidize retinal to RA. However, there is evidence that this is not the case. In the mouse, there are three Raldh genes, Raldh1-Raldh3. The pattern of cells in the developing mouse embryo that express Raldh1-3 (Mic et al.,2002) does not closely overlap with that of megalin expression(Kounnas et al., 1994). Raldh1 is expressed in the ectoderm of the dorsal retina of the eye (E10.5)(Mic et al., 2002) and the substantia nigra of the developing brain (E15.5)(Smith et al., 2001). Raldh2 is highly expressed in the mesoderm of the trunk region of murine embryos(E8-E9) but absent in the developing brain(Smith et al., 2001). Smith et al. have proposed that RA produced by this trunk mesoderm diffuses both dorsally and medially to developing hindbrain tissue(Smith et al., 2001). Raldh3 is expressed in surface ectoderm of the optic vesicle (9-10.5 dpc)(Mic et al., 2002; Smith et al., 2001). In contrast to these patterns of Raldh gene expression, megalin is initially expressed in the floor plate and along apical margins of the neural tube(McCarthy et al., 2002) and later in the choroid plexus and ependymal layer of the brain(Kounnas et al., 1994). The fact that megalin and the known Raldh genes do not seem to be expressed in the same tissues argues against megalin playing a role as a mediator of retinol uptake by cells that synthesize RA. However, a recent study in which Raldh2-deficient embryos were conditionally rescued shows sites of RA synthesis that do not correspond to sites of expression of the known Raldh genes (Mic et al.,2002) but show some overlap with sites of megalin expression. For example, novel sites of RA synthesis include the floorplate and neural tube,which are known to also express megalin(McCarthy et al., 2002). Therefore, the possibility that megalin mediates uptake of retinol by a subset of RA-synthesizing cells, particularly neural epithelial cells, must still be considered.

If megalin plays a role in transport of retinol to cells that synthesize RA, there are two possibilities: one, megalin expressed by extraembryonic visceral endoderm (VE) mediates transport of maternally derived retinol to the embryo; and two, megalin expressed on apical surfaces of certain embryonic epithelia mediates transepithelial transport of retinol to centers of active RA biosynthesis. If the first possibility was correct, megalin-deficient embryos would display a phenotype consistent with a retinol deficiency. However, embryos lacking retinol (Kastner et al., 1995), as well as those lacking Raldh2(Niederreither et al., 2000),display neurodevelopmental phenotypes unlike those of megalin-deficient embryos (Willnow et al.,1996). Mice lacking dietary retinol or Raldh2 expression have severe hindbrain defects (Niederreither et al., 2000), whereas megalin-deficient mice display abnormalities in forebrain development (Willnow et al.,1996). The fact that megalin-deficient mice show no apparent hindbrain abnormalities indicates that VE transport of maternal retinol by megalin is not required to supply the embryo with retinol required for normal hindbrain patterning.

Early in development, embryonic requirements for retinol may be satisfied through simple diffusion-based transport; however, as embryonic epithelial integrity increases and/or epithelia become stratified, trans- and intra-epithelial transport mechanisms may be required. Since certain centers of RA expression are located adjacent to megalin-expressing epithelia, megalin might play a role in transport of retinol through or within stratified epithelial layers. In this way, it might mediate delivery of retinol to cells that express Raldh enzymes but are inaccessible to retinol. For example, sites of synthesis of RA by Raldh3 have been described in cells located deep in the basal layers of the ventral forebrain(Smith et al., 2001). Interestingly, suppression of Raldh3 expression in Pax6-deficient mice leads to microphthalmia and forebrain defects, including absence of olfactory bulbs(Suzuki et al., 2000), defects similar to those described for megalin-deficient mice(Willnow et al., 1996). It is thus possible that megalin expressed by cells of the apical layers of the forebrain and retinal epithelium (McCarthy et al., 2002) mediates transport of retinol to a subset of RA-synthesizing cells located in the basal regions of the developing forebrain and the retina. Similarly, megalin expressed on apical surfaces of the neural tube might facilitate the transport of retinol to subapical regions where retinoid-dependent neurons develop(Pierani et al., 1999).

The hedgehogs are a family of proteins that act as morphogens during development, controlling cell growth, cell fate determination and organ patterning. The mouse has three hedgehog proteins: Indian hedgehog (Ihh),desert hedgehog (Dhh) and sonic hedgehog (Shh). Hedgehog proteins are synthesized as ∼45 kDa precursors, which are cleaved to form N-terminal polypeptides (e.g. N-Shh) of ∼25 kDa that have signaling activity(Lee et al., 1994; Porter et al., 1995). Developmental roles for Ihh include regulation of proliferation of cells during vasculogenesis, bone and cartilage formation and hematopoiesis(Ingham and McMahon, 2001). Roles for Dhh include control of gonad and germ line development and peripheral nerve sheath formation (Clark et al., 2000; Parmantier et al., 1999). Shh is perhaps the best-characterized member of the family (Ingham and McMahon,2001). Concomitantly with autoproteolytic cleavage of Shh,cholesterol is covalently attached to its C-terminus, and palmitic acid is added to its N-terminus (Pepinsky et al.,1998; Porter et al.,1996). These lipid modifications are thought to be important for limiting the range of N-Shh signaling or targeting the protein to specific membrane domains (Bumcrot et al.,1995; Lee et al.,1994; Porter et al.,1995; Taylor et al.,1993). The current model of Shh signaling is that binding of N-Shh to patched (Ptc), a twelve-membrane-span protein containing a sterol-sensing domain, activates a second transmembrane protein, smoothened (Smo), which signals via a G-protein-coupled pathway(Ingham and McMahon,2001).

During early development, Shh is prominently expressed in the notochord,the floorplate of the neural tube and the zone of polarizing activity in the limb (Marti et al., 1995). Numerous studies have demonstrated that Shh activity is critical for the induction, proliferation and survival of various embryonic cell types,including ventral neural cells in the developing brain, cranial neural crest cells and ventral neural retinal cells(Ingham and McMahon, 2001). For example, mice lacking Shh have anomalies of midline structures such as the notochord and floorplate of the early brain(Chiang et al., 1996) and later display an absence of ventral neuronal cells and cranial motor neurons(Litingtung and Chiang, 2000)(Table 1). The most common structural anomaly of the developing forebrain in Shh knockout mice is HPE. Mice lacking other components of the Shh signal transduction pathway also display neurodevelopmental abnormalities. For instance, mice lacking Ptc display embryonic lethality, but partial rescue of Ptc-null embryos results in severe exencephaly (Milenkovic et al.,1999) and a high incidence of cerebellar medulloblastomas characteristic of basal cell nevus syndrome(Goodrich et al., 1997). Smo-null zebrafish embryos show increased apoptosis of neural tube cells and other neural tube defects that include absence of secondary motor neurons,synopthalmia and ventral forebrain defects(Chen et al., 2001; Varga et al., 2001)(Table 1). Recent findings demonstrate that mice lacking dispatched (Disp), a protein critical to the secretion and long-range signaling of N-Shh, also display embryonic defects similar to those of Shh^–/– and Smo^–/– embryos (Ma et al., 2002). These include incomplete separation of the optic vesicles and a lack of a floorplate, resembling early anomalies in forebrain morphology that lead to HPE.

Since the spectrum of neurodevelopmental defects that make up the Shh-,Smo- and Disp-deficient phenotypes includes many key features of the megalin-deficient phenotype (Table 1), Herz and Bock have speculated that megalin is a component of the Shh signaling pathway (Herz and Bock,2002). This speculation was fueled by studies showing that other members of the megalin family function as receptors for factors that influence neurodevelopment. For example, LRP6 functions as a co-receptor with frizzled to mediate signal transduction by the secreted signaling factor Wnt/Wg(Pinson et al., 2000; Tamai et al., 2000; Wehrli et al., 2000). Similarly, very-low-density lipoprotein receptor (VLDLR) and apolipoprotein E receptor 2 (ApoER2) function as receptors for reelin, a protein that influences the positioning of dentate granule cells during brain development(Trommsdorff et al., 1999). Both megalin and LRP bind to midkine, a neurotrophic factor important for survival of embryonic neurons (Muramatsu et al., 2000). LRP has also been shown to mediate the endocytosis as well as the nuclear targeting of midkine, which is a requirement for its ability to promote cell survival (Shibata et al., 2002). Recently, McCarthy et al. have found that megalin binds to the N-terminal fragment of Shh (N-Shh)(McCarthy et al., 2002). Furthermore, they have demonstrated that megalin functions as an endocytic receptor for N-Shh in cultured cells(McCarthy et al., 2002). These findings strengthen the idea that megalin influences Shh signaling in developing neural tissues, perhaps explaining the basis for many of the neurodevelopmental abnormalities observed in megalin-deficient embryos.

Thus far, a direct relationship between megalin-mediated endocytosis of Shh and regulation of Shh signaling has not been demonstrated. There are, however,several possible consequences of the interaction between megalin and Shh. The binding of N-Shh might elicit direct signaling by megalin, similar to the response of reelin binding to VLDLR and ApoER2, which activates phosphoinositide 3-kinase (PI 3-kinase) signaling(Beffert et al., 2002). Recent yeast two-hybrid screening studies demonstrate the potential of numerous cytoplasmic adaptor proteins to associate with the cytoplasmic tail of megalin(Gotthardt et al., 2000; Patrie et al., 2001; Petersen et al., 2003). This raises the possibility that megalin influences a variety of cell-signaling pathways. Interestingly, many of the adaptor proteins (e.g. Dab-1, JIP-1,MAGI-1 and PSD-95) have roles in neurogenesis and/or neuronal functions related to synaptic processes such as neurotransmitter receptor clustering and signaling complex assembly. Furthermore, several of the megalin-binding cytosolic proteins (e.g. JIP-1, Dab-2 and SEMCAP-1) regulate signal transduction pathways (Gotthardt et al.,2000). Recently, a novel cytosolic adaptor protein, MegBP, has been shown to bind to the PxxP motifs in the megalin cytoplasmic tail(Petersen et al., 2003). Since MegBP interacts with several transcriptional regulators (e.g. the SKI-interacting protein SKIP and TGF-β-stimulated clone 22 homologous gene (THG-1), it has been proposed that the megalin interaction with MegBP regulates the sequestration and release of MegBP-bound transcription factors(Petersen et al., 2003).

Although megalin might function as a Shh signaling receptor in its own right, the bulk of information on megalin function favors the hypothesis that it indirectly regulates Shh signaling through endocytosis of N-Shh. As stated above, megalin targets ligands to endosomes and transports ligands across epithelia. Several potential consequences of N-Shh interaction with megalin-expressing epithelial cells are depicted in Fig. 2. The interaction might not only mediate N-Shh uptake but also be part of an intracellular N-Shh trafficking system. A role for megalin in Shh trafficking is supported by evidence showing that the megalin–N-Shh interaction is resistant to dissociation at low pH and that N-Shh is not efficiently targeted to lysosomes for degradation in megalin-expressing cells(McCarthy et al., 2002). Avoidance of the lysosomes may then be part of a process by which N-Shh becomes targeted to specific cellular compartments, such as the basal or lateral surfaces of the epithelial cells for release. This would make N-Shh available to cells within the epithelium, thus extending the effective range of its signaling. Such a transcytotic, epithelial transport mechanism has been described for Wnt (Dierick and Bejsovec,1998), which binds to the megalin family member arrow/LRP6(Pinson et al., 2000; Tamai et al., 2000; Wehrli et al., 2000). This system appears to involve formation of Wnt-containing exovesicles called argosomes from basolateral membranes of imaginal disc epithelial cells and the subsequent endocytosis of these vesicles by adjacent epithelial cells(Greco et al., 2001). The process could establish a morphogen gradient by dispersing Wnt-containing membrane fragments over large distances through the Drosophilaimaginal disc epithelium.

Another way in which megalin might influence Shh signaling is by modulating Ptc and Smo trafficking (Fig. 2). Several recent studies indicate that there are low levels of Ptc on the surfaces of cells and that the bulk of Ptc protein exists in intracellular endosomal compartments (e.g. in imaginal disc epithelial cells)(Capdevila et al., 1994; Denef et al., 2000). Low cell surface levels of Ptc and increased endosomal accumulation may be the result of megalin-mediated endocytosis. This scenario would be analogous to the roles of LRP and VLDLR in mediating uptake of the urokinase receptor (uPAR) bound to urokinase-plasminogen activator inhibitor-1 complexes, as well as the uptake of the cell surface receptor tissue factor (TF) in complex with TF-pathway inhibitor and factor VIIa (Strickland et al., 2002). In each of these cases, the megalin family member regulates cell surface levels of the membrane receptors (uPAR and TF). Whether megalin can interact with Ptc is not known; however, it is possible that N-Shh serves as a ligand bridge similar to that in uptake of uPAR and TF by LRP and VLDLR. In support of this possibility are the findings of Incardona et al. showing that, after Shh binding, Ptc-Smo complexes are internalized in kidney epithelial cells (Incardona et al.,2002). Further, in wing disc epithelial cells that have responded to Shh, Ptc becomes concentrated in apical vesicles, whereas Smo becomes distributed basolaterally (Denef et al.,2000). These findings indicate that, following binding of N-Shh to epithelial cells, complex trafficking of Ptc and Smo occurs. Whether megalin orchestrates such trafficking in epithelial cells remains to be established.

Here, we have considered roles for megalin in neurodevelopment, emphasizing its involvement in the trafficking of and signaling by RA and N-Shh. Given that hindbrain development is dependent on maternally derived retinol, but that megalin-deficient embryos show no apparent hindbrain defects, a critical role for megalin in placental and VE uptake/transport of retinol required for proper neurodevelopment appears doubtful. One could test for the involvement of megalin in maternal-embryonic retinol transport by supplementing the diets of dams carrying megalin-deficient embryos with RA and looking for reduced severity in the neural phenotype of the null embryos. Perhaps the most plausible possibility is that megalin expressed by neuroepithelial cells plays a role in the transport of retinol to RA-generating sites located within the developing neuroepithelium that are involved with eye and forebrain development. Considering that embryos lacking megalin have neurodevelopmental abnormalities similar to those lacking genes of the Shh pathway, it is also reasonable to speculate that megalin is involved in epithelial trafficking of N-Shh. Such a role might be analogous to the epithelial trafficking of Wnt/Wg that is important for establishment of a signaling gradient in imaginal disc epithelium. In light of growing evidence that members of the LDL receptor family mediate signal transduction directly, the possibility that megalin mediates direct signaling in response to ligation with N-Shh must also be considered. Finally, given that several LDL receptor family members are known to influence surface levels of other membrane proteins, modulation of cell surface expression/trafficking of N-Shh receptors Ptc and Smo is yet another way in which megalin might impact N-Shh signal transduction during development.