Address correspondence and reprint requests to Dr Betty Soliven, Department of Neurology MC2030, The University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637, USA. E-mail: firstname.lastname@example.org
Glial cell line-derived neurotrophic factor (GDNF), a known survival factor for neurons, has recently been shown to stimulate the migration of Schwann cells (SCs) and to enhance myelination. GDNF exerts its biological effects by activating the Ret tyrosine kinase in the presence of glycosylphosphatidylinositol-linked receptor, GDNF family receptor (GFR) α1. In Ret-negative cells, the alternative transmembrane coreceptor is the 140-kDa isoform of neural cell adhesion molecule (NCAM) associated with a non-receptor tyrosine kinase Fyn. We confirmed that GDNF, GFRα1 and NCAM are expressed in neonatal rat SCs. We found that GDNF induces an increase in the partitioning of NCAM and heparan sulfate proteoglycan agrin into lipid rafts and that heparinase inhibits GDNF-signaling in SCs. In addition to activation of extracellular signal-regulated kinases, and phosphorylation of cAMP response element binding protein, we found that cAMP-dependent protein kinase A and protein kinase C are involved in GDNF-mediated signaling in SCs. Although GDNF did not promote the differentiation of purified SCs into the myelinating phenotype, it enhanced myelination in neuron–SC cocultures. We conclude that GDNF utilizes NCAM signaling pathways to regulate SC function prior to myelination and at early stages of myelin formation.
Glial cell line-derived neurotrophic factor (GDNF) was first identified as a survival factor for dopaminergic neurons of the midbrain (Lin et al. 1993). GDNF is a member of a family of growth factors that bind glycosylphosphatidylinositol-anchored family receptor (GFR) α1–4 (Lin et al. 1993; Kotzbauer et al. 1996; Baloh et al. 1998; Milbrandt et al. 1998). GFRα1, GFRα2, GFRα3 and GFRα4 preferentially bind GDNF, neurturin, artemin and persephin, respectively, whereas Ret tyrosine kinase mediates the transmembrane signaling by these growth factors (Airaksinen and Saarma 2002). GDNF has recently been shown to signal in a Ret-independent manner via a Src family tyrosine kinase Fyn. Downstream events of GDNF signaling in Ret-negative cells include phosphorylation of phospholipase Cγ, extracellular signal-regulated kinases (ERKs) and cAMP response element binding protein (CREB), and induction of immediate-early genes such as c-fos and mGIF (Poteryaev et al. 1999; Trupp et al. 1999; Pezeshki et al. 2003). Since GFRαs do not have an intracellular domain, coupling of GFRα1 with Src family kinases would require a coreceptor, recently identified as the 140 kDa isoform of neural cell adhesion molecule (NCAM) (Paratcha et al. 2003).
Schwann cells (SCs), the source of GDNF in the PNS, express GFRα1, NCAM, but not Ret (Naveilhan et al. 1997; Trupp et al. 1997, 1999; Paratcha et al. 2003). Given that GDNF and GFRα1 transcripts are up-regulated in distal segments of lesioned sciatic nerves, it has been proposed that lesion-induced GFRα1 capture and present GDNF to activate Ret on axons, thereby facilitating axonal regeneration (Trupp et al. 1997). Administration of high doses of GDNF to adult rats has also been shown to promote SC proliferation and myelination of normally unmyelinated small axons (Höke et al. 2003). Whether the latter observations are mediated by an autocrine action of GDNF on SCs or by indirect mechanisms via SC–axon interaction is uncertain.
The goal of this study was to further investigate signal transduction mechanisms of GDNF in SCs. We confirmed that GDNF-signaling in SCs is Ret-independent and involves activation of ERK1/2 and phosphorylation of CREB. The effect of GDNF was greatly attenuated by heparinase III indicating that heparan sulfate proteoglycan is required for GDNF-signaling in SCs. In addition, treatment with GDNF led to an increase in NCAM in lipid rafts, which was inhibited by heparinase III. Agrin, a potential candidate heparan sulfate proteoglycan, also increased in lipid rafts after GDNF application, suggesting that a dynamic association of NCAM and agrin in lipid rafts is important for the response to GDNF. In addition to the ERK1/2 pathway, we found that cAMP-dependent protein kinase A (PKA) and protein kinase C (PKC) are involved in GDNF signaling in SCs. PKA also participates in the regulation of GDNF transcript levels in SCs, providing a positive feedback loop in GDNF autocrine signaling in SCs.
Materials and methods
Schwann cell culture
Primary SCs were prepared as previously described (Nagano et al. 2001), with some modifications. Sciatic nerves from P3–P4 rats were stripped free of epineurium and then mechanically dissociated following enzymatic treatment with 0.25% trypsin and 0.03% collagenase for 25 min. The resulting cell suspension was plated onto polylysine-coated 100-mm dishes and maintained in Dulbecco's modified Eagle's medium supplemented 10% fetal bovine serum and 1% penicillin/streptomycin. The next day, 10 µm cytosine arabinoside was added for 48 h to eliminate proliferating fibroblasts. Culture medium was then switched to Dulbecco's modified Eagle's medium + 10% fetal bovine serum + 1% penicillin/streptomycin containing 2 µm forskolin and 20 µg/mL bovine pituitary extract to stimulate SC proliferation. For further expansion, confluent SC cultures were detached using trypsin and seeded at a density of 1.0 × 106 cells per 100-mm polylysine-coated dish. SCs in passages 3 and 4 were used in experiments, after removal of forskolin and bovine pituitary extract from culture medium for at least 3 days prior to experiments.
In a subset of experiments, the spontaneously immortal SC clone (iSC) (a generous gift from Dr E. M. Shooter, Stanford University) was used. The iSC (derived from adult rat sciatic nerves) was maintained in Dulbecco's modified Eagle's medium + 10% heat-inactivated horse serum + 1% penicillin/streptomycin. These cells express S-100 and Ran-1 antigen (Bolin et al. 1992). Confluent cultures were detached with trypsin-EDTA and replated into polylysine-coated dishes or coverslips in 10% serum for 24 h, then switched to serum-free medium for experiments.
Cultured SCs or iSCs in subconfluent condition were incubated in serum-free medium for 3–4 h, followed by pretreatment with various agents for 30 min prior to exposure to GDNF (10 ng/mL) at indicated times at 37°C, except for 2 h treatment duration for anti-GDNF (R & D Systems, Minneapolis, MN, USA) and anti-NCAM antibodies (Chemicon, Temecula, CA, USA). Phosphatidylinositol-specific phospholipase C (PIPLC) and staurosporine were obtained from Sigma (St Louis, MO, USA); other inhibitors (PP2, PD98059, H89, calphostin C, SU6656) were from EMD Biosciences (La Jolla, CA, USA).
In vitro myelination assay
Dorsal root ganglia (DRG) from neonatal rat pups were dissociated and plated onto matrigel-coated chamber slides in neuron medium which consists of neurobasal medium, B27 supplements, 0.5 mm glutamine and NGF7.5S (500 ng/mL), as described by other investigators (Svenningsen et al. 2003). Purified neuronal cultures were obtained by treating DRG cultures four times with fluorodeoxyuridine (10 µm) and uridine (10 µm) for 48 h in alternate feedings. Prior to seeding of purified SCs onto DRG cultures, immunopanning with anti-thy-1 antibody (Chemicon) was performed to remove fibroblasts remaining after treatment with 10 µm cytosine arabinoside. Approximately 60 000 SCs were seeded onto each DRG culture and allowed to proliferate and align along the axons for 4–5 days. To initiate myelination, DRG–SC cocultures were fed with 50 µg/mL ascorbic acid in 10% fetal bovine serum with or without GDNF (10 ng/mL) for 10 days. Immunofluorescence studies were done using anti-MBP (myelin basic protein) antibody (1:1000) (Sternberger Monoclonals, Lutherville, MD, USA) and anti-PGP9.5 (1:300) (Chemicon) to label myelin fragments and neurons, respectively. SC nuclei were labeled with 4′, 6-diamidino-2-phenylindole (DAPI) included in the anti-fade mounting medium (Vectashield H-1200, Vector Laboratories, Burlingame, CA, USA).
Immunoprecipitation and western blot analysis
Primary SCs were washed with phosphate-buffered saline and scraped off in a lysis buffer (10 mm Tris-HCl pH 7.5, 137 mm NaCl, 0.5% sodium dodecyl sulfate, 2 mm EDTA, 10% glycerol, 1 mm phenylmethyl sulfonylfluoride, and 1 mm sodium orthovanadate). Supernatants obtained after centrifugation of the whole cell lysates at 700 g were used as samples. Protein concentration was determined using the bicinchoninic acid (BCA) assay (Pierce, Rockford, IL, USA). The proteins (30 µg per lane) from the samples were resolved by 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and blotted to polyvinylidene difluoride membranes. The non-specific binding was blocked with 5% bovine serum albumin in Tris-buffered saline (20 mm Tris-HCl pH 7.6, 137 mm NaCl) containing 0.1% Tween 20 for 2 h at room temperature. Blots were then incubated overnight at 4°C with primary antibodies, then washed briefly with Tris-buffered saline plus 0.1% Tween 20 followed by incubation with goat anti-rabbit or anti-mouse horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 2 h at room temperature. The following primary antibodies were used in this study: anti-pERK1/2 (1:1000), anti-pCREB (1:1000), anti-CREB (1:1000), anti-pAkt (1:1000), anti-pMet (1:1000), anti-Met (Cell Signaling, Beverly, MA, USA); anti-NCAM (1:1000), anti-Ret (1:1000), anti-GFRα1 (1:1000) (Santa Cruz Biotechnology); anti-ERK2 (1:1000), anti-pPKARII (5 µg/mL) (Upstate Biotechnology, Waltham, MA, USA); anti-agrin m247 (1:500) (EMD Biosciences), and pRet antibodies (1:1000). Anti-PY905Ret, anti-PY1015Ret, anti-PY1062Ret, and anti-PY1096Ret antibodies were generously provided by Dr E. M. Johnson Jr. (Washington University, MO, USA). The immunoreactive bands were detected using enhanced chemiluminescence (ECL) detection reagents (Amersham), then scanned and quantified using the NIH Image analysis programs. All experiments were repeated at least three times with similar results. Representative data are shown. Results are expressed as percentage control values by normalization of relative density of pERK2/total ERK2, and pCREB/total CREB of test conditions to controls of each experiment.
For immunoprecipitation of NCAM, cells were lysed in 150 mm NaCl, 50 mm Tris, 4% Triton X-100, pH 8.0, 1.25 mm phenylmethyl sulfonylfluoride, 40 µm leupeptin, 10 µg/mL aprotinin, 1 mm orthovanadate) for 30 min on ice. Cell lysates were frozen and thawed thrice and then centrifuged for 10–20 min at 14 000 g at 4°C. The supernatants (200 µg) were precleared by incubation (1 h, 4°C with rocking) with 50 µL of protein A Sepharose beads (Pierce) diluted 1:1 in lysis buffer. The samples were centrifuged and incubated with 5 µg of polyclonal NCAM antibody (Chemicon) for 16 h at 4°C, followed by incubation with protein A bead suspension (10% v/v in lysis buffer) for another hour. The immunoprecipitates were washed three times with lysis buffer and resolved by 7.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis.
Cells were lysed on ice with an ice-cold lysis buffer (1% Triton X-100, 10 mm Tris-HCl pH 7.5, 137 mm NaCl, 2 mm EDTA, 10% glycerol, 1 mm phenylmethyl sulfonylfluoride, and 1 mm sodium orthovanadate) (Triton X lysis buffer) for 30 min. After centrifugation at 700 g for 5 min, the supernatant was saved as the Triton X-soluble fraction. The pellet was resuspended on ice in ice-cold Triton X lysis buffer plus 60 mm octyl-β-glucoside (Pierce) for 30 min, followed by centrifugation at 6000 g for 3 min. The supernatant from the last centrifugation was saved as the Triton X-insoluble fraction (detergent-resistant membrane fraction). Five micrograms of protein of the insoluble fraction was used for western blotting.
RNA preparation and reverse transcription–polymerase chain reaction
Total RNA was extracted from iSCs and primary SCs according to the manufacturer's instruction (Qiagen, Valencia, CA, USA). First strand cDNA was synthesized from total RNA by RT reaction. The reaction mixture (20 µL) contained 2–3 µg total RNA, 2.5 µL H2O, 0.5 µg of oligo(dT) primer, 40 U RNasin, 4 µL of the 5 × buffer, 2 µL of 10 mm dNTP mixture, and 40 U reverse transcriptase AMV (Roche, Basel, Switzerland). The reaction was carried out at 37°C for 60 min, heated for 10 min at 70°C, chilled on ice and stored at −20°C. The cDNA was amplified by PCR in a reaction mixture containing 2 µL of cDNA template, 10 mm Tris-HCl, 1.5 mm MgCl2, 50 mm KCl, pH 8.3, 2.0 mm MgCL2, 200 µm of each dNTP, 0.5 µm each sense and antisense primers of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or GDNF and 2.5 U Taq DNA polymerase in reaction, added water to the final volume of 50 µL. Negative control using water (instead of DNA template) was included in each PCR reaction. The amplification profile for GAPDH or GDNF detection included 1 cycle of initial denaturation at 94°C for 5 min, 35 cycles of denaturation at 94°C for 30 s, primer annealing at 58°C for 30 s, and extension at 72°C for 30 s, then 1 cycle for final extension at 72°C for 7 min. The expression of GAPDH was used as a control, whereas non-reverse transcriptase-treated RNA samples were used to exclude contaminating genomic DNA as a source of amplified products. The PCR products were subjected to electrophoresis on a 1.5% agarose gel stained with 0.5 µg/mL ethidium bromide. Data were quantified by scanning the bands followed by analysis using the NIH Image program. Optical densities of GDNF bands were normalized to that of GAPDH band.
PCR primers were designed according to the rat mRNA sequences from GenBank. Sequences of oligonucleotides were as follows: GAPDH: sense 5′-CTGACAGCCCCAGAGTGTGT-3′, antisense 5′-CATCACCCCATTTGATGTTAGC-3′; GDNF: sense 5′-GAAGACCACTCCCTCGGCCACC-3′, antisense 5′-GTCGTACATTGTCTCGGCCGC-3′.
For quantification of proliferating cell nuclear antigen (PCNA)+ cells, SCs or iSCs on coverslips were fixed with acetone for 15 min, followed by methanol for 15 min at 4°C. Then, these cells were labeled with a monoclonal antibody α PCNA (1:200)(Santa Cruz Biotechnology) and rabbit α S100 antibody (1:200)(Sigma) followed by fluorescein isothiocyanate-conjugated and rhodamine conjugated-secondary antibodies (1:400). Scoring of PCNA+ nuclei was accomplished by examining eight randomly selected, non-overlapping fields (300–400 cells on each coverslip).
Data from PCNA experiments, western blot and RT–PCR analysis were expressed as mean ± SD. Statistical analysis was performed by anova followed by Student's t-test or Bonferroni method for multiple comparisons. For data with unequal variance, Dunnett's T3 was used. Statistical probability of p < 0.05 was considered significant.
Glial cell line-derived neurotrophic factor-signaling in Schwann cells is Ret-independent
Western blot analysis was carried out initially to confirm the presence of GFRα1 in primary SCs. The band at 55–60 kDa disappeared after treatment with PIPLC 1.0 IU/mL, which cleaves glycosylphosphatidylinositol-linked proteins such as GFRαs from the cell membrane (Fig. 1a, upper panel). Pretreatment of SCs with PIPLC (1.0 IU/mL) abolished GDNF-dependent phosphorylation of ERK1/2 and CREB. GDNF (10 ng/mL) increased the phosphorylation of ERK2 and CREB to 171 ± 15% (n = 3) and 182 ± 21% (n = 3) of control values, respectively. In SCs treated with GDNF plus PIPLC, pERK2 and pCREB levels were 104 ± 13% (n = 3) and 65 ± 15% (n = 3) of control values, respectively (p < 0.005 for GDNF vs. GDNF + PIPLC for both pERK2 and pCREB). Data shown in this figure and all subsequent figures are representative of at least three independent experiments.
Next, we searched for possible receptor tyrosine kinases that can associate with GFRα1 in SCs. Although a Ret-like immunoreactive band was initially detected in SCs, subsequently experiments failed to detect tyrosine phosphorylation of immunoprecipitated Ret (Fig. 1a, lower panel). In addition, functional Ret was not detected using antibodies specific to Ret phosphorylated sites, Tyr905, Tyr1015, Tyr1062, and Tyr1096 with or without GDNF application (data not shown). The same antibodies were effective in detecting Ret autophosphorylation in other cell types (Tsui-Pierchala et al. 2002). GDNF has recently been reported to activate Met receptor tyrosine kinase, the receptor for a SC mitogen hepatocyte growth factor, in MDCK cells (Krasnoselsky et al. 1994; Popsueva et al. 2003). We did not observe activation of Met by GDNF in SCs (data not shown).
We examined the concentration dependence and time course of GDNF action in primary SCs. GDNF increased the phosphorylation of ERK1/2 in a dose-dependent manner with a maximal effect at 10 ng/mL. The response was eliminated by concomitant treatment with the GDNF neutralizing antibody (Fig. 1b, upper panel). Based on these results, 10 ng/mL GDNF was used in all subsequent experiments. GDNF-induced activation of ERK1/2 began at 2 min, peaked at 5–10 min, followed by a decline to the basal levels by 30 min. GDNF also promoted rapid phosphorylation of CREB, which was observed at 5 min after GDNF stimulation, peaked within 15 min, followed by a return to basal levels within 30 min (Fig. 1b, lower panel). Similar results on the phosphorylation of ERK1/2 and CREB were observed in cultured immortalized SCs (iSCs) (data not shown). On the other hand, the serine-threonine kinase Akt, a downstream target of phosphatidylinositol-3-kinase, was not activated by GDNF in primary SCs or in iSCs (data not shown).
Role of neural cell adhesion molecule, heparan sulfate proteoglycans and lipid rafts in glial cell line-derived neurotrophic factor-mediated signaling
NCAM has recently been identified as a coreceptor for GDNF (Paratcha et al. 2003). We found that pretreatment with anti-NCAM antibodies (5 µg/mL) did not prevent GDNF-stimulated ERK1/2 and CREB phosphorylation. Furthermore, an increase in the phosphorylation of these proteins was observed in SCs treated with these antibodies alone (Fig. 2a). As NCAM signaling elicited by anti-NCAM antibodies is thought to mimic NCAM homophilic binding (Beggs et al. 1997; Schmid et al. 1999), our results show that NCAM homophilic binding does not preclude GDNF signaling via GFRα1-NCAM heterophilic binding and vice versa.
As lipid rafts are enriched in effectors of signal transduction, we examined NCAM localization in lipid rafts and its response to GDNF using Triton X-insoluble or the so-called detergent-resistant membrane (DRM) fractions. We found that all three NCAM isoforms were expressed with a predominance of NCAM140 and NCAM120 in SCs. Though NCAM is localized both in Triton X-soluble and -insoluble fractions, GDNF induced an increase of NCAM in the DRM fraction, which peaked at 15 min followed by a return to control levels at 30 min. As shown in Figs 2(b) and (c), GDNF-induced accumulation of NCAM in the DRM fraction was attenuated by heparinase III (0.5 mIU/mL), an enzyme which cleaves side chains of heparan sulfates. In addition, the effect of GDNF on ERK1/2 and CREB phosphorylation was inhibited by heparinase III, indicating that heparan sulfates are required for the Ret-independent GDNF signaling in SCs. These results are summarized in bar graphs shown in Fig. 2(d).
Among the heparan sulfate proteoglycans, a potential candidate is agrin, which binds to NCAM in other cell types (Storms et al. 1996). Of the agrin isoforms, the B/Z– isoform is inactive till deglycosylated and is involved in lipid raft aggregation in lymphocytes (Khan et al. 2001). Using an anti-agrin m247 antibody in western blot analysis, a sharp immunoreactive band at ∼220 kDa corresponding to the active agrin B/Z– isoform was detected in SCs. We found that GDNF increased the partitioning of active agrin B/Z– into the DRM fraction with similar kinetics as that of NCAM, and both were attenuated by heparinase III (Fig. 3a). These blots were stripped and reprobed with anti-actin antibody to demonstrate equal sample loading. The levels of agrin in the DRM fraction were 344 ± 39% in SCs treated with GDNF (10 ng/mL) for 15 min, and 76 ± 16% of control values in those treated with GDNF + heparinase III (n = 4 each) (p < 0.001 for GDNF vs. GDNF + heparinase III). Furthermore, we found that agrin co-immunoprecipitated with NCAM and GFRα1 in GDNF-treated SCs (Fig. 3b). These results suggest that active agrin-B/Z– may be involved in promoting NCAM association with lipid rafts.
Glial cell line-derived neurotrophic factor activated pathways in Schwann cells: role of protein kinase A and protein kinase C
We examined further post-receptor signaling mechanisms activated by GDNF in SCs with pharmacologic inhibitors of protein kinases. We found that GDNF-stimulated ERK1/2 and CREB phosphorylation was suppressed by PP2 (5 µm), and by SU6656 (1 µm), inhibitors of Src family kinases (Figs 4a and 5c). In contrast, the inhibition of CREB phosphorylation by PD98059, a MAPK kinase (MEK) inhibitor, was incomplete. These findings indicate possible activation of other pathways by GDNF in SCs. It is recognized that CREB phosphorylation is regulated by other kinases, including PKA, PKC and Ca2+/calmodulin-dependent kinases (Gonzalez et al. 1989; Sheng et al. 1991; Xie and Rothstein 1995). To evaluate the contribution of the PKA pathway, SCs were pretreated with a PKA inhibitor H89 (5 µm). Treatment with H89 prevented GDNF-stimulated CREB phosphorylation, but not ERK1/2 phosphorylation (Fig. 4a). Results from PP2, PD98059 and H89 experiments are summarized in bar graphs shown in Fig. 4(b). Activation of PKA by GDNF was also verified indirectly by increased phosphorylation of PKA regulatory subunit (PKARII) (data not shown).
To determine whether PKC is involved in GDNF-mediated signaling, SCs were pretreated with staurosporine, a PKC inhibitor. Staurosporine (10–100 nm) strongly inhibited both ERK1/2 and CREB phosphorylation (Figs 5a and b). Similar results were obtained with another PKC inhibitor, calphostin C (0.5 µm) (Fig. 5c). In contrast, pretreatment of SCs with KN62 (10 µm), a specific inhibitor of Ca2+–calmodulin-dependent protein kinase II (CaMKII), did not block GDNF-dependent ERK1/2 or CREB phosphorylation (Fig. 5a). Results from staurosporine and KN62 experiments are summarized in bar graphs in Fig. 5(b). In calphostin C experiments (n = 3), pERK2 levels were 196 ± 21% in GDNF-treated SCs, and 86 ± 11% of control values in SCs treated with GDNF + calphostin C (p < 0.001). pCREB levels were 175 ± 17% for GDNF and 79 ± 15% of control values for GDNF + calphostin C (p < 0.005).
Both PKC and calcium have been reported to up-regulate GDNF expression in SCs, whereas elevation of cAMP with forskolin (fsk) has no effect (Kinameri and Matsuoka 2003). We re-examined the effect of fsk (20 µm) on GDNF mRNA in primary SCs and iSCs. Under our employed PCR conditions, increasing number of PCR cycles leads to a linear increase in the amount of reaction products. As shown in Fig. 6(a), treatment of iSCs with fsk led to an increase in the transcript levels of GDNF, which was evident at 6 h and 24 h (n = 3). The effect of fsk on GDNF mRNA in both iSCs and SCs was completely inhibited by H89 (5 µm). Expressed as percentage of GAPDH mRNA, GDNF transcript at 35 cycles was 27.9 ± 0.03% (n = 3) in untreated ones, and 67.5 ± 0.1% in iSCs treated with fsk for 6 h (n = 3 each, p < 0.01). In primary SCs, the GDNF mRNA level was 69.7 ± 0.07% in untreated ones, and 96.5 ± 0.08% in those treated with fsk for 6 h (n = 3 each, p < 0.02).
Functional outcome of glial cell line-derived neurotrophic factor action in Schwann cells
Activation of ERK1/2 and CREB is generally associated with cell survival, proliferation or differentiation (Sato-Bigbee and DeVries 1996; Bonni et al. 1999; Lee et al. 1999; Kaplan and Miller 2000). We found that treatment with GDNF (10 ng/mL) for 48 h in serum-free medium stimulated the proliferation of iSCs. The percentage of PCNA+/S100+ cells was 19.7 ± 3.3 (n = 5) in untreated iSCs, and 32.7 ± 4.3 (n = 5) in GDNF-treated ones (p < 0.003). In contrast, there was no increase in the percentage of PCNA+/S100+ cells when primary SCs were treated with GDNF, either in serum-free medium or in 10% serum (n = 4, data not shown). Furthermore, treatment of primary SCs with GDNF for 3 days did not induce the expression of MBP, as shown in Fig. 7(a). Lysates from acutely dissociated SCs from P10 animals were used as a positive control for western blot analysis. GDNF also did not induce the expression of myelin-associated glycoprotein (MAG) level (data not shown). Immunofluorescence studies of DRG–SC mixed cultures revealed no increase in MBP+ or MAG+ cells in cultures treated for 5–7 days with GDNF (10 ng/mL) in the absence of ascorbic acid and serum (n = 3). On the other hand, treatment of DRG–SC cocultures with GDNF (10 ng/mL) for 10 days after seeded SCs had aligned along axons led to enhanced myelination in the presence of ascorbic acid and serum (Fig. 7b). Without GDNF, myelination in DRG–SC cocultures was still sparse at 10 days (n = 3 each).
In the present study, we have characterized GDNF-mediated Ret-independent signaling in neonatal rat SCs. We found that GDNF signaling involves GFRα1, Src family kinase activity, activation of ERK1/2, and phosphorylation of CREB. In contrast, GDNF did not activate the phosphatidylinositol-3-kinase pathway based on the lack of increase in Akt phosphorylation. We have also defined other GDNF signaling pathways, specifically the involvement of PKA and of PKC. Studies directed at proximal events have confirmed that Ret-independent signaling in SCs involves NCAM, as reported by Paratcha and coworkers. In their study, complex formation between GFRα1 and NCAM leads to inhibition of homophilic NCAM interactions and NCAM-mediated cell adhesion (Paratcha et al. 2003). We found that GDNF signaling in SCs was not precluded by NCAM homophilic binding induced by anti-NCAM antibody. It is possible that Ret-independent GDNF signaling occurs through a multimeric complex containing GDNF/GFRα1/NCAM molecules with a stoichiometry similar to that described for Ret-dependent signaling [(GDNF)1(GFRα1)2(Ret)2] (Jing et al. 1996).
NCAM belongs to the immunoglobulin superfamily and is expressed as three major isoforms (NCAM180, NCAM140, NCAM120) that differ in their C-terminals. Of the three isoforms, NCAM140 is the one implicated in Ret-independent GDNF signaling (Paratcha et al. 2003). In contrast to Ret, which is recruited to lipid rafts upon exposure to GDNF, NCAM140 is constitutively present in lipid rafts through palmitoylation (Beggs et al. 1997; Tansey et al. 2000; Niethammer et al. 2002). We found that GDNF stimulates the recruitment of NCAM to lipid rafts via a process that involves heparan sulfate proteoglycans. The latter has been reported to promote NCAM homophilic binding, and to play an important role in Ret-dependent GDNF signaling (Barnett et al. 2002; Cole et al. 1986). The work by Barnett and coworkers suggest that heparan sulfates assist Ret-dependent signaling by concentrating GDNF in the vicinity of its receptors, rather than by stabilizing multimeric ligand–receptor complexes similar to that described for fibroblast growth factor 1 (FGF1) signaling system (Pellegrini et al. 2000; Barnett et al. 2002). One of the heparan sulfate proteoglycans that copurify with NCAM from brain extracts is agrin (Cotman et al. 1999). Adult SCs express both agrin-B/Z+ and agrin-B/Z– isoforms, perhaps accounting for the ability of perisynaptic SCs to induce aggregation of acetylcholine receptors (Yang et al. 2001). Agrin-B/Z+ induces activation of the muscle-specific receptor tyrosine kinase MuSK, which is crucial for aggregration of acetylcholine receptors at the neuromuscular junction (Glass et al. 1996). In comparison, active deglycosylated agrin-B/Z– acts in an autocrine fashion and induces lipid raft clustering in the immunological synapse (Khan et al. 2001; Bezakova and Ruegg 2003). Our finding that GDNF increases the partitioning of both NCAM and agrin-B/Z– into the DRM fraction highlights the role of lipid rafts and its associated molecules in Ret-independent GDNF signaling in SCs.
NCAM-mediated signaling via lipid rafts is distinct from its signaling outside of lipid rafts. Within lipid rafts, NCAM140 constitutively associates with Fyn, leading to the recruitment of focal adhesion kinase, and subsequently activation of the Ras–Raf–ERK1/2 pathway (Beggs et al. 1997; Niethammer et al. 2002). Outside of lipid rafts, NCAM140 signaling is dependent on FGF receptor, which results in activation of phospholipase Cγ, and PKC, intracellular Ca2+ elevation and CaMKII activation (Niethammer et al. 2002; Povlsen et al. 2003). Simultaneous activation of both signal transduction pathways is required to induce neurite outgrowth via NCAM stimulation in PC12 cells (Kolkova et al. 2000). Our results showing the involvement of Src family kinases but not CaMKII suggest that the main pathway of NCAM-dependent GDNF signaling originates within lipid rafts. On the other hand, data from staurosporine experiments raise the possibility of concomitant GDNF-NCAM signaling outside of lipid rafts converging at the Ras-Raf complex, or simply indicate that an optimal level of PKC activity is required for a full activation of Raf, as shown by other investigators (Marais et al. 1998). It should also be noted that a direct association between Fyn and a PKC isoform PKCθ has been reported in T lymphocytes (Ron et al. 1999). Furthermore, PKC can phosphorylate CREB in vitro at multiple sites, including Ser133 (Yamamoto et al. 1988).
We found that H89 decreases GDNF-induced CREB phosphorylation but not ERK1/2 phosphorylation, indicating that PKA is either activated downstream of or independent of the Ras-ERK1/2 pathway. The above finding may also reflect the complex relationship known to exist between these two pathways, depending on the presence or absence of rap1, which is inhibitory to Raf1 but stimulatory to B-Raf (Vossler et al. 1997). Activation of PKA by forskolin inhibits the phosphorylation of ERK1/2 in iSCs and attenuates glial growth factor-stimulated ERK1/2 activity in SCs (Kim et al. 1997; Nagano et al. 2001). Results were opposite in SCs treated with 8-bromo-cAMP (Mutoh et al. 1998). An initial study by Doherty and coworkers supports a role for G proteins but not PKA in NCAM-mediated signaling, but a subsequent study showed that NCAM-mediated neurite outgrowth is inhibited by a cAMP antagonist rp-cAMP (Doherty et al. 1991; Jessen et al. 2001). Exactly how the PKA pathway is activated by GDNF-NCAM signaling is unclear. Src but not Fyn has been reported to modulate the activity of G protein α subunits, which are linked to adenylate cyclase regulation (Hausdorff et al. 1992). We found that GDNF transcript level is up-regulated by the PKA pathway. Others have shown that PKC and calcium are involved in the regulation of GDNF expression, though there has been no consensus regarding the role of PKA (Grimm et al. 1998; Verity et al. 1998; Kinameri and Matsuoka 2003). Hence, a positive feedback loop exists in SCs where exposure to GDNF leads to activation of ERK1/2, PKA and PKC resulting in CREB activation, which in turn up-regulates the expression of GDNF that is then secreted to exert both paracrine and autocrine actions.
The functional outcome of autocrine GDNF signaling includes enhanced SC migration and myelination (Höke et al. 2003; Paratcha et al. 2003). The effect of GDNF on SC migration is perhaps not surprising given the recognized role of NCAM in cell migration (Ackley et al. 1997; Thomaidou et al. 2001). In addition, both ERK1/2 and PKA pathway have been shown to regulate cell migration, the former by enhancing the phosphorylation of myosin light chain and the latter by phosphorylation of RhoA on Ser188 thereby down-regulating RhoA activity (Lang et al. 1996; Klemke et al. 1997; Meintanis et al. 2001; Ellerbroek et al. 2003). We found that GDNF enhances the proliferation of iSCs, but not primary SCs. This apparent discrepancy may be related to the complex effects of time in vitro and prior exposure to serum and fsk on the response of primary SCs to mitogens (Dong et al. 1997). The Ras-ERK1/2 signaling has been associated with SC dedifferentiation while PKA is implicated in SC proliferation and differentiation (Morgan et al. 1991; Kim et al. 1997; Howe and McCarthy 2000; Harrisingh et al. 2004). Although GDNF does not enhance SC differentiation in purified SC cultures or DRG–SC mixed cultures, it increases myelination in DRG–SC cocultures when applied with ascorbic acid and serum. It is plausible that additional factors in the serum, basal lamina or axonal membranes synergize with GDNF in promoting myelination, or alternatively, the effect of GDNF is mediated indirectly via a reciprocal SC–axon interaction during the induction of myelination. Since NCAM is expressed primarily by non-myelinating and premyelinating SCs, one may argue that enhanced myelination from GDNF-NCAM signaling is a consequence of increased SC proliferation and migration towards axons. On the other hand, studies on temporal and spatial expression of NCAM, and findings from NCAM-deficient and P0–/–NCAM–/– mice support the concept that GDNF-NCAM signaling plays a significant role during myelin formation, in addition to its effect on SC proliferation or migration (Carenini et al. 1997; Martini and Schachner 1986; Carenini et al. 1999).
Based on our study and work by other investigators, we propose the following model for GDNF-signaling in SCs, as shown in Fig. 8. Within the lipid rafts, binding of GDNF, GFRα1, and NCAM together with agrin initiates the signal transduction and lead to the activation of Fyn, which results in the activation of ERK1/2 pathway. Activation of PKC may be due to GDNF-NCAM signaling outside of lipid rafts in association with another receptor such as FGF receptor, or due to a direct activation of PKC by Fyn. NCAM signaling also leads to cAMP elevation and PKA activation, possibly via modulation of G protein activity. The concomitant activation of these pathways leads to increased proliferation and migration of SCs. Our study does not exclude possible contribution to enhanced myelination by an indirect effect of GDNF via axons. It is likely that both GDNF-mediated autocrine signaling in SCs and paracrine signaling on neurons contribute to the repair of peripheral nerve injury or disorders where GDNF is known to be up-regulated (Hammarberg et al. 1996; Trupp et al. 1995; Höke et al. 2003). Peripheral nerve regeneration and remyelination may become impaired once there is a decline in GDNF expression, as shown in chronically denervated SCs (Höke et al. 2002).
We would like to thank Dr E. M. Johnson, Jr. (Washington University, St Louis, MO, USA) for providing us with anti-phosphoRet antibodies, and Shawna Cook for her assistance in some experiments. This work was supported by National Institute of Health grant RO1 NS39346, R21 NS49014, and in part by grants from the Brain Research Foundation, by a gift from Mr M. P. Miller, and by the Jack Miller Neuropathy Center.