Identification and characterization of GFRalpha-3, a novel Co-receptor belonging to the glial cell line-derived neurotrophic receptor family.

A new family of neuronal survival factors comprised of glial cell line-derived neurotrophic factor (GDNF) and neurturin has recently been described (Kotzbauer, P. T., Lampe, P. A., Heuckeroth, R. O., Golden, J. P., Creedon, D. J., Johnson, E. M., Jr., and Milbrandt, J. (1997) Nature 384, 467-470). These molecules, which are related to transforming growth factor-beta, are important in embryogenesis and in the survival of distinct neuronal populations. These molecules signal through a novel receptor system that includes the Ret receptor tyrosine kinase, a ligand (i.e. GDNF or neurturin), and an accessory glycosyl-phosphatidylinositol-linked molecule that is responsible for high affinity binding of the ligand. Two accessory molecules denoted GDNF family receptor 1 and 2 (GFRalpha-1 and GFRalpha-2) have been described that function in GDNF and neurturin signaling complexes. We have identified a novel co-receptor belonging to this family based on similarity to GFRalpha-1, which we have named GFRalpha-3. GFRalpha-3 displays 33% amino acid identity with GFRalpha-1 and 36% identity with GFRalpha-2. Despite the similarity of GFRalpha-3 to GFRalpha-1 and GFRalpha-2, it is unable to activate Ret in conjunction with GDNF, suggesting that there are likely additional undiscovered ligands and/or Ret-like receptors to be identified. GFRalpha-3 is anchored to the cell membrane by a phosphatidylinositol-specific phospholipase C-resistant glycosyl-phosphatidylinositol linkage. GFRalpha-3 is highly expressed by embryonic day 11 but is not appreciably expressed in the adult mouse. In situ hybridization analyses demonstrate that GFRalpha-3 is located in dorsal root ganglia and the superior cervical sympathetic ganglion. Comparison of the expression patterns of GFRalpha-3 and Ret suggests that these molecules could form a receptor pair and interact with GDNF family members to play unique roles in development.

Glial cell line-derived neurotrophic factor (GDNF) 1 was ini-tially discovered because of its ability to promote the survival of embryonic ventral midbrain dopaminergic neurons in culture (1). It is also capable of supporting the survival of rat superior cervical and dorsal root ganglion neurons in vitro as well as a variety of other neuronal populations in the central and peripheral nervous systems (2)(3)(4)(5)(6). Recently, a related molecule, neurturin (NTN), was purified based on its ability to maintain rat superior cervical ganglion neurons in culture (7). Together these molecules have established the existence of a new family of neuronal survival factors displaying similarity to transforming growth factor-␤ (20% sequence identity). Structurally, GDNF and NTN contain a Cys knot motif that is also present in nerve growth factor, platelet-derived growth factor, and transforming growth factor-␤ family members. This motif consists of six Cys residues that form three intramolecular disulfide linkages between parallel ␤-sheets (8). The high degree of sequence conservation between GDNF and NTN (42% sequence identity) as well as the conservation of the Cys knot motif suggests that these molecules would serve as ligands for related receptors. In fact, it was demonstrated that both GDNF (9 -12) and NTN (12)(13)(14)(15) can activate the Ret receptor utilizing different co-receptors for high affinity binding.
The c-ret proto-oncogene (16) codes for a receptor tyrosine kinase with a Cys-rich extracellular domain, a single transmembrane domain, and a cytoplasmic tyrosine kinase domain (17). Mutations in the Cys-rich region of the extracellular domain can lead to familial medullary thyroid carcinoma or multiple endocrine neoplasia 2A (18,19), whereas a mutation at amino acid 918 in the kinase domain of the receptor causes multiple endocrine neoplasia 2B, a more aggressive form of the disease (20). Inactivating mutations throughout the gene lead to Hirschsprung's disease, which is characterized by loss of innervation of the lower intestinal tract (21). In the developing mouse, RET is specifically expressed in a number of tissues including the ureteric bud of the developing kidney and the neural crest cells that innervate the lower intestine. Mice that lack functional RET receptors have a Hirschsprung's like phenotype and fail to develop kidneys (22). Although it was surprising that a tyrosine kinase could be activated by a member of the transforming growth factor-␤ family, mice that lack GDNF displayed a phenotype similar to the Ret knockout and provided the first clue that GDNF and Ret were a receptor/ ligand pair (23).
Interaction of GDNF with Ret requires an additional protein, GFR␣-1. GFR␣-1 was the first co-receptor isolated using an expression cloning strategy to identify high affinity binding proteins for GDNF (24,25). This molecule is attached to the cell surface via a glycosyl-phosphatidyl-inositol (GPI) linkage. Recently, an additional co-receptor, GFR␣-2, was isolated by virtue of its sequence similarity with GFR␣-1 (13)(14)(15). As mentioned previously, GFR␣-1 and GDNF form a high affinity complex that is capable of activating Ret (24,25). However, GFR␣-1 has also been reported to bind NTN and activate Ret (14). Similarly, GFR␣-2 can bind to both GDNF and NTN and activate Ret but appears to bind with higher affinity to NTN (14,15). This is unique in that the specificity of the tripartite signaling complex can be controlled either by the choice of co-receptor or by the local concentration of a particular ligand.
Here we report the identification and characterization of a new GFR␣ family member that we have named GFR␣-3. Expression profiles of this molecule suggest that it is very important in embryogenesis, particularly in the nervous system. The existence of several family members and their differential expression patterns raises the possibility that this tripartite receptor/ligand system is a paradigm utilized in many tissues during development.

MATERIALS AND METHODS
Sequence of GFR␣-3-A mouse expressed sequence tag data base clone encoding GFR␣-3 (Image consortium; accession number AA050083) was sequenced and analyzed. A mistake in the image consortium clone that generated a premature stop codon in the coding sequence was identified by sequencing multiple polymerase chain reaction products obtained from a mouse fetal brain library using primers based on the expressed sequence tag sequence. This base was corrected using the QuikChange site-directed mutagenesis kit (Stratagene). Alignment was performed using the Clustal W program (MacVector).
Cell Culture and Transient Transfection-COS and 293 cells were grown in a 5% CO 2 environment using Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, glutamine (293 cells only), penicillin, and streptomycin. Subconfluent 293 or COS cells in 10-cm dishes were transfected with 8 g of the DNA construct of interest using LipofectAMINE reagent (50 l) as described by the manufacturer (Life Technologies, Inc.).
NB41A3 cells were grown in a 6% CO 2 environment using Ham's F-10 supplemented with 15% heat-inactivated horse serum and 2.5% fetal bovine serum, penicillin, and streptomycin. Subconfluent NB41A3 cells in 35-mm dishes were transfected using a total of 2 g of DNA and 4 l of LipofectAMINE reagent for 10 h according to the manufacturer's instructions.
Luciferase Assays-48 h post-transfection, cells were harvested by scraping, pelleted at 250 ϫ g for 10 min, and resuspended in 400 l of 25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 4 mM EGTA, 1 mM dithiothreitol, and 0.1% Triton X-100. The extracts were cleared of cell debris by centrifugation at 12,000 ϫ g for 5 min. The luciferase assay was performed as described by Brasier et al. (26) using the same amount of protein for each sample.
Western Analyses-NB41A3 cells were grown in 60-mm dishes and transfected with 6 g of vector (pcDNA3) or GFR␣-1 and GFR␣-3 pcDNA3 expression constructs using 12 l of LipofectAMINE as described above. Samples were treated with 100 ng/ml GDNF for 5 min at room temperature prior to harvesting. Immune precipitation with the Ret polyclonal antibody was performed as described (29). Samples were run on a 10% SDS-PAGE gel, transferred to Immobilon (Millipore), and immunoblotted with phosphotyrosine antibody (1:10,000; Upstate Biotechnology, Inc.) or Ret antibody (1:500) as described (29).
Cleavage of the GPI linkage by Phosphatidylinositol-specific Phospholipase C-293 cells were transiently transfected with Flag-tagged GFR␣-1 and GFR␣-3 expression constructions as described above. Cells recovered for 48 h, were harvested in Dulbecco's modified Eagle's medium containing 2% bovine serum albumin, 20 mM Hepes, pH 7.6, and treated with 1 unit of phospholipase C (Boehringer Mannheim) for 30 min at 37°C. After pelleting, the medium was collected (1 ml) for immune precipitation, and the cells were lysed in Dulbecco's phosphatebuffered saline (Life Technologies, Inc.), 1% Triton X-100, 0.1 mg/ml phenylmethlysulfonyl fluoride. Both were subjected to immune precipitation as described using Flag M2 affinity gel (Eastman Kodak Company) (11). The resulting samples were analyzed by 10% SDS-PAGE and transferred to Immobilon. Western analysis was performed with Flag M2 monoclonal antibody (1:2500; Eastman Kodak Company) in Tris-buffered saline (10 mM Tris, pH 8.0, 150 mM NaCl, 5 mM MgCl 2 ), 0.1% Tween 20, and 5% nonfat dry milk. Horseradish peroxidase protein A (Amersham Corp.) was used at a 1:5000 dilution in the above buffer, and the antibody complexes were visualized by enhanced chemiluminescence (National Diagnostics).
Biotinylation of GFR␣-3-COS cells were transfected with the Flagtagged GFR␣-3 construct as described above. Two days post-transfection, cells were washed three times with Ringer's solution (10 mM Hepes, pH 7.4, 154 mM NaCl, 7.2 mM KCl and 1.8 mM CaCl 2 ) and incubated with 2 ml of 200 g/ml EZ-Link sulfo-NHS-Biotin (Pierce) for 30 min at 4°C. Following the biotin incubation, the cells were washed five times with Tris saline (10 mM Tris, pH 7.4, 120 mM NaCl), and the cells were lysed in phosphate-buffered saline, 1% Triton X-100, and 0.1 mg/ml phenylmethlysulfonyl fluoride. Immunoprecipitations were performed using the Flag M2 antibody (4 g) followed by incubation with protein A-agarose (Life Technologies, Inc.). The resulting samples were analyzed by 10% SDS-PAGE and transferred to Immobilon (Millipore). Western analysis was performed using horseradish peroxidase-conjugated streptavidin (1:500; Upstate Biotechnology Inc.) in Tris-buffered saline, 0.1% Tween 20, and 2% bovine serum albumin. Horseradish peroxidase protein A (Amersham Corp.) was used at a 1:5,000 dilution in the above buffer, and the antibody complexes were visualized by enhanced chemiluminescence (National Diagnostics).
Radioisotopic Labeling Procedure-293 cells (2 ϫ 10 6 cells) were transfected as described above with constructs encoding either Flagtagged GFR␣-1, a transmembrane version of Flag-tagged GFR␣-1, or Flag-tagged GFR␣-3. After an overnight recovery period, cells were incubated with [2-14 C]ethan-1-ol-2-amine HCl (100 Ci/4 ml, 50 -62 mCi/mmol, CFA 329, Amersham Corp.) for 18 h in Dulbecco's modified Eagle's medium and 10% dialyzed fetal bovine serum. The cells were harvested and subjected to immune precipitation as described above. A portion of the samples was used for Western analysis as described previously, and the remainder was analyzed by SDS-PAGE followed by fixing in 10% acetic acid and fluorography with Amplify (Amersham Corp.). The film was exposed at Ϫ80°C for 24 h.
Northern Analyses-For the cell lines, 20 g of total RNA was run on a 1% denaturing agarose gel, transferred to Hybond N (Amersham Corp.) and probed with the appropriately labeled cDNA. Both the multitissue Northern and developmental Northern were purchased (CLON-TECH) and probed with the appropriately labeled DNA.
In Situ Hybridization Analysis-A partial cDNA fragment of the mouse RET proto-oncogene was cloned by reverse transcription-polymerase chain reaction from Neuro 2A RNA using oligonucleotides corresponding to the murine cDNA sequence between nucleotide 361 and nucleotide 690. The fragment was subcloned into the pGEM-T vector (Promega), and the plasmid was linearized with NcoI. Mouse GFR␣-3 was obtained from the image consortium, and a PstI fragment was cloned into pJCC (modified pBluescript, Stratagene). The construct was linearized with NotI. For all of the above constructs, sense and antisense probes were made using T7 or T3 RNA polymerase (Life Technologies, Inc.) and [ 33 P]UTP (NEN Life Science Products) (27). For the hybridization, embryos of E14.5 were frozen in 2-methylbutane and stored at Ϫ80°C. Frozen embryos were cryostat sectioned (20 m), and in situ hybridization was performed as described previously (27). (28) of the expressed sequence tag data base performed using blocks of GFR␣-1 amino acid sequences that contained multiple Cys residues. We reasoned that the placement of these Cys residues would be highly conserved among GFR␣-1-related molecules. The expressed sequence tag clone (Image consortium; accession number AA050083) obtained was derived from 13.5-14.5-day-old mouse embryos and was sequenced in its entirety (Fig. 1). The full-length cDNA contains an open reading frame of 397 amino acids that contains a hydrophobic signal peptide as well as a stretch of hydrophobic amino acids at its COOH terminus that comprises a putative GPI linkage sequence or membrane attachment sequence. It also contains three potential N-linked glycosylation sites that are represented by the sequence Asn-Xaa-(Thr/Ser).

GFR␣-3 Is a Member of the GFR␣ Family-GFR␣-3 was identified by a BLAST search
GFR␣-3 shares lower amino acid identity with GFR␣-1 and GFR␣-2 (33 and 36%, respectively) than that seen between GFR␣-1 and GFR␣-2 (48%) (Fig. 2). In addition to the lower amino acid identity, GFR␣-3 contains only 28 Cys residues as compared with 31 and 30 Cys residues for GFR␣-1 and GFR␣-2, respectively. However, the positions of all 28 Cys residues are conserved among the other family members. This analysis suggests that although GFR␣-3 belongs to the GFR␣ family, it is a more distant relative.
GFR␣-3 Cannot Activate Ret Using GDNF as the Ligand-Because both GFR␣-1 and GFR␣-2 can form functional signaling complexes with Ret upon addition of GDNF, we sought to determine whether GFR␣-3 and GDNF could form a functional signaling complex with Ret. Previously we determined that activated Ret signals via the mitogen-activated protein kinase pathway using the Gal-Elk/GAL-Luc reporter system (29). NB41A3 mouse neuroblastoma cells express Ret but require the addition of either GFR␣-1 or GFR␣-2 in conjunction with GDNF to activate Ret (12). When GFR␣-1 is transiently introduced into these cells with the Gal-Elk reporter system, GDNF is able to elicit a 4.5-fold induction of reporter activity. Neither vector alone (pcDNA3) nor GFR␣-3 is able to increase reporter activity (Fig. 3A). In addition, transient expression of GFR␣-1 in NB41A3 cells treated with GDNF results in increased tyrosine phosphorylation of Ret, whereas expression of GFR␣-3 does not (Fig. 3B). Therefore, despite the conservation of this co-receptor family, only GFR␣-1 and GFR␣-2 are able to use GDNF to activate Ret. However, GFR␣-3 can be co-immunoprecipitated with Ret when both molecules are transiently expressed in 293 cells (data not shown). This implies that GFR␣-3 and Ret most likely will form a functional signaling complex in the presence of the appropriate ligand.
GFR␣-3 Membrane Anchor Analysis-Both GFR␣-1 and GFR␣-2 possess the characteristic carboxyl-terminal features of GPI-linked proteins in that they have a COOH-terminal hydrophobic domain separated by a hydrophilic linker region from a cleavage consensus sequence for GPI linkage, i.e. Ala-Ser-Ser for GFR␣-1 and Gly-Ser-Asn for GFR␣-2 (13,30). Because the carboxyl terminus of GFR␣-3 varies significantly from the other family members, we sought to determine whether this molecule is secreted or attached to the extracellular surface of the plasma membrane via its hydrophobic tail or a GPI linkage. A common method to demonstrate whether a protein is linked to the membrane by a GPI linkage is its ability to be cleaved by PIPLC. In the following experiments, the co-receptors have been tagged with the Flag epitope on their amino termini for easy detection. As can be seen in Fig. 4A, GFR␣-3 unlike GFR␣-1 cannot be removed from the membrane by treatment with PIPLC. GFR␣-1 undergoes an apparent shift in molecular weight upon cleavage with PIPLC. This is because GFR␣-1 expressed in 293 cells is present as a doublet that is not well separated in this gel system. The higher molecular weight species is the molecule that is concentrated in the medium after PIPLC cleavage. The lower molecular weight species is presumably in the process of being transported to the cell surface. To demonstrate that GFR␣-3 is attached to the extracellular surface, cells expressing GFR␣-3 were treated with EZ-Link sulfo-NHS-Biotin as described under "Materials and Methods." The transfected cells contained a biotinylated protein immunoprecipitated with Flag antibody corresponding to the expected size of GFR␣-3, demonstrating that this protein is indeed attached to the extracellular surface of the cells. However, these experiments do not allow us to predict whether it is attached via a GPI linkage or by insertion of its hydrophobic tail because resistance to PIPLC has been demonstrated for several GPI-linked proteins including erythrocyte acetylcholinesterase. This protein contains a modified inositol group in its anchor conferring resistance to PIPLC (31). Chemical analyses of membrane anchors have determined that they are glycophospholipid units containing ethanolamine, mannose, glucosamine, phosphatidylinositol, and occasionally galactose. Fatemi et al. (32) have demonstrated that the GPI-linked protein Thy-1 can be effectively labeled with [ 3 H]ethanolamine. When cells transfected with Flag-tagged GFR␣-1, a transmembrane form of GFR␣-1, and GFR␣-3 are incubated in the presence of [ 14 C]ethanolamine, GFR␣-1 and GFR␣-3 are both effectively labeled (Fig. 4C). All three proteins are expressed to equivalent levels as depicted by the Western analysis of these proteins shown in Fig. 4D. Therefore, GFR␣-3 is a GPI-linked protein that is insensitive to cleavage by PIPLC. This does not rule out the possibility that other lipases such as phosphatidylinositol-specific phospholipase D may be used to remove this molecule from the membrane. Resistance to PIPLC could result from inaccessibility of the cleavage site in situ as is the case for the GPI-linked variant surface glycoproteins on intact trypanosomes (33). Alternatively, as described above, the anchor can be modified to prevent cleavage. The biological significance of this unusual anchoring mechanism for this family of co-receptors is not known. However, because GFR␣-1 (25) and GFR␣-2 (13) can presumably be severed from the membrane with PIPLC, the reversibility of this mode of attachment may be critical for their biological function. Because GFR␣-3 cannot be efficiently removed from the membrane using this enzyme, it may signal via an alternative mechanism (see Fig. 7).
Expression of GFR␣-3 in Embryonic and Adult Mouse-The presence of an additional GFR␣-1-like receptor raises the question of whether these molecules are redundant in their role of

FIG. 3. Activation of Ret by GDNF and GFR␣-1 or GFR␣-3. A,
NB41A3 cells were transiently transfected with the Gal-Elk/Gal-Luc reporter system in conjunction with the indicated expression constructs. 24 h prior to harvest, 10 ng/ml GDNF was added. Luc activity was assayed as described. The data are the averages of two experiments and are presented as a fold increase calculated by division of the GDNF-treated samples by their untreated counterparts. B, NB41A3 cells grown in 60-mm dishes were transfected with the indicated expression constructs and treated with 100 ng/ml GDNF for 5 min at room temperature where indicated. Ret was immunoprecipitated as described under "Materials and Methods" and subjected to SDS-PAGE analysis. The proteins were transferred to Immobilon P and probed with antibodies against phosphotyrosine. Samples were visualized by an ECL exposure of 5 min. signaling. To address this question, we performed Northern analyses on cell lines known to express the RET receptor. RET is expressed in the neuroblastoma cell lines LA-N-1, LA-N-2, LA-N-5, SY5Y, and SK-N-SH but not in SK-N-MC cells (Fig.  5A). Neither GFR␣-1 nor GFR␣-3 was found to be expressed in these cell lines either by Northern analysis (Fig. 5A) or reverse transcription-polymerase chain reaction (data not shown). In contrast, GFR␣-2 was expressed in LA-N-1 and SK-N-SH cells and was particularly high in LA-N-2 and LA-N-5 cells (Fig. 5A). Low levels were expressed in SY5Y cells, which were detectable by reverse transcription-polymerase chain reaction (data not shown). Because RET is unable to bind GDNF in the absence of a GFR␣-1-like molecule (24,25), our ability to activate RET upon GDNF treatment in LA-N-1, LA-N-5 and SK-N-SH cells (11,29) suggests that GFR␣-2 acts as a GDNF receptor and, upon binding GDNF, activates the RET receptor. A recent publication by Baloh et al. (14) has demonstrated that this is indeed the case. GFR␣-2 functions as a low affinity receptor for GDNF or as a high affinity receptor for neurturin both signaling in conjunction with the Ret receptor. Surprisingly none of our Ret expressing cell lines expressed GFR␣-1 or GFR␣-3.
To further address the role of specific signaling by GFR␣-3, we analyzed the expression of this molecule compared with that of GFR␣-1 and GFR␣-2 during mouse development. In embryogenesis, RET expression is seen beginning at E9 of mouse development. It continues throughout fetal development but is almost entirely absent in the adult animal (34). Northern analysis reveals that both GFR␣-1 and GFR␣-3 are expressed highly at E11 and drop off to lower levels by E17 (Fig. 5B). GFR␣-2, in contrast, is expressed at low levels at E11 with increased expression levels throughout development culminating with strong expression in adult mouse brain, spleen, lung, and liver (Fig. 5C). GFR␣-1 displays a more limited adult expression pattern because it is expressed only in liver and kidney, whereas GFR␣-3 is not detected in any of the adult tissues tested. Ret expression in the adult mouse was reported in the brain, spinal cord, salivary gland, heart, spleen, and lymph nodes (35,36). The disparate expression patterns of these molecules in the adult indicate that Ret may function alone, use another co-receptor or interact with its co-receptor in trans (2). Alternatively, additional Ret-like receptors may interact with these co-receptors in a cell-specific manner. It is also possible that adult tissues could respond to trauma and injury by up-regulating GFR␣-3. In this way, the developmental state could be recapitulated, allowing regeneration of the afflicted tissues.
In Situ Hybridization Analysis of GFR␣-3-To get a better understanding of the potential partners for GFR␣-3 in the embryo, this molecule was localized by in situ analysis and compared with the expression pattern for Ret. Data in Fig. 6A are consistent with previously reported results demonstrating that at E14 RET is expressed prominently in cranial ganglia, myenteric ganglia of the gut, motor neurons in the spinal cord and the hindbrain, the dorsal root ganglia, and the growing tips of the renal collecting ducts (34,36). Interestingly, GFR␣-3 is expressed more selectively. At E14 GFR␣-3 is expressed in subpopulations of cells in the dorsal root ganglia (data not shown), selected cranial ganglia, the superior cervical sympathetic ganglia, and regions in the lower urogenital and digestive tracts (Fig. 6B). It is not present in detectable levels in the developing kidney. More similar to Ret, GFR␣-1 is expressed widely in the developing embryo. It is present in the ventral midbrain, spinal cord motor neurons, subpopulations of the A, receptor signaling with the co-receptor and receptor expressed on the same cell. B, signaling with the co-receptor and ligand expressed on the same cell. The yellow diamonds represent GDNF-like molecules that interact with a member of the GFR␣ co-receptor family (blue). The coreceptor/ligand pair is then able to activate a Ret-like receptor (red). dorsal root ganglia, the developing kidney, the mesenchyme of the developing gut, the retina, the pituitary, urogenital tract, and pancreatic primordium (24). Like GFR␣-3, GFR␣-2 is expressed less widely being found only in the developing and adult dorsal root ganglia and the superior cervical ganglion of the rat (14). Therefore, the co-receptors in the GFR␣ family appear to maintain distinct tissue-specific expression that does not always overlap with Ret expression. This suggests that other Ret-like receptors or alternative signaling methods may be involved in GFR␣-1-, GFR␣-2-, and GFR␣-3-dependent signaling. The ability of GFR␣-1 and GFR␣-2 but not GFR␣-3 to be removed from the membrane via cleavage of a GPI linkage also implicates unique signaling mechanisms (Fig. 7). In cases where Ret and the co-receptor are expressed on the same cell, soluble ligand can diffuse to the cell, bind to the co-receptor, and activate Ret (Fig. 7A). Alternatively, the ligand and coreceptor could be expressed on the same cell, be removed by cleavage of the GPI linkage, and travel as a pair to activate Ret on a nearby cell (Fig. 7B). Because GFR␣-3 cannot be removed by PIPLC cleavage, it may be expressed on the same cell as the Ret-like receptor to act as a co-receptor (Fig. 7A). It will be of interest to determine if one signaling paradigm prevails over the other in development as well as in the mature organism.
In conclusion, we have identified a novel member of the GFR␣ family of co-receptors and have demonstrated that a unique expression profile is observed for the GFR␣ family members. Based on the observations that the family of coreceptors is expressed in a nonredundant fashion, we propose that there are likely additional molecules related to RET and/or additional GDNF-like ligands. We also suggest that the tripartite receptor system exemplified by RET, GDNF, and GFR␣-1 is used throughout development and in the mature animal in a tissue-specific manner. These systems use the classical components of a receptor-mediated system including a signaling receptor and a soluble ligand. However, they are distinct in that they include a GPI-linked protein that is thought to modulate the sensitivity of RET activation. Similar to the GDNF, ciliary neurotrophic factor (CNTF), a survival factor for ciliary neurons, also requires a GPI-linked receptor, CNTFR␣, for functional ligand induced formation of its tripartite receptor complex (37). This receptor complex is comprised of the leukemia inhibitory factor receptor binding protein LIFR␤ and gp130, the signal transducer of interleukin 6 in conjunction with CNTFR␣ and CNTF. Interestingly, the bipartite leukemia inhibitory factor receptor is converted into a tripartite CNTF receptor by addition of the specificity conferring molecule, CNTFR␣. The expression of CNTFR␣ predominantly in the nervous system limits CNTF to neuronal actions. Therefore, the tripartite receptor complex may be a common paradigm used in the nervous system to increase the fidelity of the specific signals for survival of various neuronal populations. Further studies are in progress to understand these novel receptor-mediated signaling systems and their involvement in neuronal development and survival.