Neurotrophic factors and their receptors

The best-studied member of this family is basic FGF (FGF-2) that is found in low- and high-molecular weight isoforms exhibiting tissue- and species-specific regulation (Meisinger and Grothe, 1997). FGF-2 isoforms are characterized by two unusual biological properties. First, they lack a conventional signal sequence for export out of the cell (Florkiewicz et al., 1991; Renko et al., 1990), but apparently exit the cell via non-classical mechanisms (the 18kDa isoform of FGF-2 has been reported to be excreted and stored in the extracellular matrix). Second, they translocate to the nucleus in vivo and in vitro (Clarke et al., 2001; Klimaschewski et al., 1999; Baldin et al., 1990). Nuclear accumulation of FGF-2 and FGF receptor (FGFR) expression is induced by stimuli that promote neurite outgrowth and neuronal differentiation of neural-crest derived adrenal medullary cells (Stachowiak et al., 1997). While the intranuclear functions of FGF-2 are poorly understood, signaling of FGF-2 through low-affinity heparan sulfate proteoglycans (HSPG) or membrane-bound tyrosine kinase receptors has been described in detail (Klint and Claesson-Welsh, 1999). The HSPGs are attached to the cell membrane or to extracellular matrix proteins and have been shown to be critical for FGFR function, i.e., they facilitate dimerization and activation of the receptors.

FGF-2 and its receptors are constitutively expressed in dorsal root ganglia (DRGs) and in the peripheral nerve. Furthermore, FGF-2 and FGFR3 are strongly and rapidly up-regulated in DRGs and in the peripheral nerve following axotomy. FGF-2 is induced in Schwann cells at the lesion site (Ji et al., 1995; Grothe et al., 1997) and prevents apoptosis of DRG neurons if applied to the transected sciatic nerve (Otto et al., 1987). Furthermore, FGF-1 and FGF-2 promote axonal growth as demonstrated by their ability to improve nerve regeneration across a collagen-filled nerve conduit (Aebischer et al., 1989; Danielsen et al., 1988). Channels filled with Schwann cells overexpressing the high molecular weight isoform of FGF-2 are particularly useful in promoting regeneration (Timmer et al., 2003). Due to their effects on mitogenesis of mesoderm- and neuroectoderm-derived cells, it was assumed that FGFs support axonal regeneration via increased proliferation of Schwann cells and enhanced angiogenesis (Aebischer et al., 1989). Yet, direct trophic effects of FGF-2 isoforms on adult DRG neurons seeded onto a laminin substrate were observed. In response to a preconditioning lesion, i.e., transection of the sciatic nerve one week before culture, FGF-2 isoforms enhance maximal axonal length when compared to untreated control neurons (Klimaschewski et al., 2004). The effects of FGF-2 isoforms on axonal growth by adult DRG neurons are completely blocked by SU5402, a specific FGFR antagonist, indicating that FGF-2 acts via cell surface receptors which are detected in all DRG neurons.

The FGFs bind in an overlapping pattern to four structurally related high affinity tyrosine kinase receptors. The high sequence similarity between the receptors together with the overlapping pattern of FGF binding – most FGFs bind to all four receptors – implies redundancy within this growth factor receptor family. Different FGF family members will activate FGFR subtypes to a different extent depending on their ability to bind with high affinity to each receptor subtype. FGF-2 binds to all four FGFR members with high affinity (Ornitz et al., 1996), but not to all splice variants of FGFR2 and FGFR3.

The general structure of FGFRs is highly conserved during evolution. They are characterized by immunoglobulin-like (Ig-like) domains, a heparin-binding region and an acidic box domain in the extracellular part of the receptor, whereas the intracellular region following the transmembrane domain harbors the split tyrosine kinase domain. FGFR1 and FGFR2, but not FGFR3 or FGFR4, are synthesized by peripheral sensory neurons (Oellig et al., 1995). Mice carrying null mutations in each of the FGFR genes reveal that FGFR1 and FGFR2 are absolutely required for early embryonic development, while animals lacking FGFR3 survive and show no obvious telencephalic defects (Deng et al., 1996).

FGFR1 is the most abundant FGFR in the nervous system and displays several functions in development and during regeneration in the adult (Stachowiak et al., 1996; Meisinger et al., 1996). One of the characteristics of the FGFR family is the occurrence of numerous receptor isoforms that are produced from alternative mRNA splicing (Klint and Claesson-Welsh, 1999). The original FGFR1 cloned from chicken contains three Ig-like domains in the extracellular domain and is termed ‘FGFR1 long’. In contrast, ‘FGFR1 short’ contains two extracellular Ig-like loops only. ‘FGFR short’ displays higher affinity to both FGF and heparin compared to ‘FGFR long’. The additional Ig-like loop and the acidic box of the long FGFR1 isoform seem to display an autoinhibitory function which would prevent FGF-independent activation by HPSGs.

Ligand binding in cooperation with the accessory heparin sulfate proteaglycan leads to FGFR1 dimerization and autophosphorylation of their cytoplasmatic domains which recruites a number of signaling molecules relevant for axonal growth. The lipid-anchored 90 kD docking protein FRS2 serves as a linker between FGF activation and the downstream Ras/MAP kinase signaling pathway. FRS2 becomes tyrosine-phosphorylated upon FGF stimulation and associates with the Grb/SOS complex to relay activation of the downstream MAP kinase pathway. MAPK activation by FGFR leads to a sustained and robust signal which stimulates neuronal differentiation, for example by pheochromocytoma (PC12) cells.

FGFR activation is required for neurite outgrowth stimulated by CAMs (N-CAM, L1 or N-cadherin) in neurons and neuron-like cell lines suggesting the existence of CAM/FGFR1 complexes (Niethammer et al., 2002; Saffell et al., 1997; Lom et al., 1998). Neurite outgrowth by PC12 cells requires sustained MAPK activation. Moreover, PC12 cells overexpressing a dominant negative FGFR reveal reduced NGF-induced process formation and autophosphorylation of FGFR, while selective FGFR inhibitors or oligonucleotides that interfere with FGF-2 receptor binding completely block neurite outgrowth induced by NGF (Chevet et al., 1999). Further interaction between TrkA and FGFR signaling is indicated by the observation that NGF increases levels of FGFR1 mRNA in PC12 cells (Meisinger et al., 1996). Treatment of PC12 cells with bone morphogenetic protein 2 (BMP2), a TGF superfamily member, does not result in neuronal differentiation, but up-regulates FGFR1 on mRNA and protein level rendering PC12 cells responsive to subthreshold concentrations of FGF-2 (Hayashi et al., 2001). These results suggest that activated FGFR1 forms complexes with other growth factor receptors and probably represents a common signaling interface for various extracellular stimuli that regulate axon initiation, elongation and branching.

Besides their stimulatory functions, activated tyrosine kinase receptors initiate a cascade of events of negative signaling that decreases the amplitude of positive signals and modulate the level of growth factor stimulation. Hence, the same receptor simultaneously induces positive and negative signals that appear to be functionally connected by numerous feedback loops. Negative receptor signaling involves coordinated action of ubiquitin ligases like c-Cbl, adaptor proteins like Grb2, inhibitory molecules like Sprouty, cytoplasmatic kinases and phosphoinositol metabolites. These inhibitory signals are essential for normal cell functioning and their de-regulation may result in cancer or autoimmune diseases. In the case of FGF-stimulated axonal growth these normally protective negative mechanisms partially act as inhibitors of nerve regeneration.

Tyrosine kinase receptor activation is followed by rapid endocytosis and degradation of both the receptor and the ligand. Ligand binding induces receptor mono-ubiquitination by the ubiquitin ligase c-Cbl which functions as signal for sorting of the receptor into intraluminal vesicles of multivesicular endosomes and subsequent delivery to lysosomes followed by degradation via lysosomal enzymes. Receptor tyrosine kinases like FGFR, EGFR or PDGFR are mono-ubiqitinated at multiple sites (multi-ubiquitination), while cytoplasmatic phosphorylated protein tyrosine kinases are poly-ubiqitinated and degraded in the proteasome. In the case of FGFR signaling, c-Cbl does not directly bind to the receptor but catalyzes the ubiquitination of the receptor via interaction with FRS2 and Grb-2. c-Cbl competes with SOS for Grb2 und thus nullifies MAPK signaling. The adaptor protein FRS2 is therefore important in the assembly of both positive (SOS) and negative (Cbl) signals to mediate a balanced FGFR signal transduction.

Most of the knowledge about tyrosine kinase receptor degradation has been obtained from EGFR and, in spite of the large amount of work performed in the FGF and FGFR field, little is known about the intracellular trafficking of these proteins. Haugsten et al. (2005) showed that FGFR1, R2 and R3 are sorted for lysosomal degradation but that FGFR4 escapes into a recycling pathway. FGFR1 revealed the highest levels of ubiquitination and the fastest degradation indicating that different levels of ubiquitination of FGFRs determine their intracellular sorting. In HeLa cells cells transiently transfected with FGFR1 the colocalization of internalized FGF-1 with the endosomal protein EEA1 and the lysosome-associated membrane protein LAMP-1 was carefully analyzed by confocal micrsocopy.