Department of Pharmacology, The Ohio State University College of Medicine, Columbus, Ohio, USA
Department of Psychiatry, Division of Molecular Psychopharmacology, The Ohio State University College of Medicine, Columbus, Ohio, USA
Address correspondence and reprint requests to Norton H. Neff, Department of Pharmacology, Graves Hall, College of Medicine, The Ohio State University, 333 West 10th Ave, Columbus, Ohio 43210, USA. E-mail: firstname.lastname@example.org
It has been proposed that GM1 ganglioside promotes neuronal growth, phenotypic expression, and survival by modulating tyrosine kinase receptors for neurotrophic factors. Our studies tested the hypothesis that GM1 exerts its neurotrophic action on dopaminergic neurons, in part, by interacting with the GDNF (glia cell-derived neurotrophic factor) receptor complex, Ret tyrosine kinase and GFRα1 co-receptor. GM1 addition to striatal slices in situ increased Ret activity in a concentration- and time-dependent manner. GM1-induced Ret activation required the whole GM1 molecule and was inhibited by the kinase inhibitors PP2 and PP1. Ret activation was followed by Tyr1062 phosphorylation and PI3 kinase/Akt recruitment. The Src kinase was associated with Ret and GM1 enhanced its phosphorylation. GM1 responses required the presence of GFRα1, and there was a GM1 concentration-dependent increase in the binding of endogenous GDNF which paralleled that of Ret activation. Neutralization of the released GDNF did not influence the Ret response to GM1, and GM1 had no effect on GDNF release. Our in situ studies suggest that GM1 via GFRα1 modulates Ret activation and phosphorylation in the striatum and provide a putative mechanism for its effects on dopaminergic neurons. Indeed, chronic GM1 treatment enhanced Ret activity and phosphorylation in the striatum of the MPTP-mouse and kinase activation was associated with recovery of dopamine and DOPAC deficits.
It has been proposed that the ganglioside GM1 promotes neuronal growth, phenotypic expression, and survival by modulating tyrosine kinase receptors for neurotrophic factors. We provide evidence that the GM1 enhances the activity of Ret tyrosine kinase receptor for glia cell-derived neurotrophic factor (GDNF) in the striatum in situ and in vivo, and propose that this might be a mechanism for GM1's neurotrophic actions on dopaminergic neurons. Ret activation is followed by Tyr1062 and Tyr981 phosphorylation and recruitment of PI3-K/Akt, Erk, and Src signaling. GM1 apparently acts by increasing the binding of endogenous GDNF to GFRα1 co-receptor, which is required for the GM1 effect on Ret.
In the CNS, the GM1 ganglioside displays neurotrophic actions in vitro and in vivo and promotes growth, survival, phenotypic expression, and function restoration of multiple neuronal populations, including cholinergic, dopaminergic, serotonergic, and noradrenergic phenotype (Hadjiconstantinou and Neff 1998, 2000). Many theories of GM1's bioactivity have been proposed, though a complete understanding of its mechanism(s) is still missing. The prevailing view is that GM1 interacts with neurotrophic signaling. In this regard, there is convincing evidence that GM1 modulates the activation and phosphorylation of the Trk kinase receptors for neurotrophins in the brain and initiates the Erk (extracellular signal-regulated kinase) and PI3-kinase (PI3-K)/Akt signaling pathways (Duchemin et al. 2002, 2008; Mo et al. 2005). The exact mechanism(s) of this interaction remains unclear and there are conflicting results and theories (Mutoh et al. 1995; Rabin et al. 2002; Duchemin et al. 2002, 2008; references and discussion here in).
A plethora of studies have shown that GM1 restores dopaminergic function in animal models of Parkinson's disease and normative aging (Hadjiconstantinou and Neff 1998, 2000; Goettl et al. 1999, 2003). For example, in the MPTP (1-methly-4-phenyl-1,2,3,6-tetrahydropyridine)-mouse GM1 improves the morphology of TH+ neurons in the substantia nigra pars compacta (SNc); corrects pre-synaptic dopaminergic markers; and alleviates motor and somatosensory impairments (Hadjiconstantinou et al. 1986, 1989; Hadjiconstantinou and Neff 1988; Weihmuller et al. 1988). The actions of GM1 on nigrostriatral dopaminergic neurons are reminiscent of those of GDNF (glia cell-derived neurotropic factor), a potent neurotrophic factor necessary for the survival and differentiation of midbrain dopaminergic neurons (Lin et al. 1993; Sullivan and Toulouse 2011; Ibanez 2013). This raises the possibility of a GM1 interaction with the GDNF receptor, analogous to that observed for Trk (Duchemin et al. 2002, 2008; Mocchetti 2005 and references herein). GM1 and other gangliosides are important components of lipid rafts and play a crucial role for their formation and stability (Sonnino et al. 2006). It is in the lipid rafts where GM1 is thought to interact with various membrane associated proteins and modify cellular signaling. The observation that ligand-induced activation and phosphorylation of the GDNF receptor kinase Ret occurs following translocation to lipid rafts (Tansey et al. 2000) suggests a spatial coincidence of GM1 and GDNF receptor complex and an opportunity for interaction.
GDNF is a member of the GDNF family of ligands (GFLs) that also includes neurturin, artemin, and persefin (Airaksinen and Saarma 2002; Wang 2013). GFLs signal via a multicomponent receptor complex that consists of the glycosylphosphatidylinositol (GPI)-anchored to lipid rafts co-receptor GDNF-family receptor α1-4 (GFRα1-4) serving as the ligand binding site, and the common transmembrane receptor tyrosine kinase Ret (Durbec et al. 1996; Trupp et al. 1996) serving as the signal transducer. GFLs display differing affinities for the GFRα co-receptor, with the GDNF binding with high affinity to GFRα1 (Jing et al. 1996; Treanor et al. 1996). GFRα1 and Ret are highly expressed in midbrain dopaminergic neurons (Nosrat et al. 1997; Trupp et al. 1997; Golden et al. 1998; Sarabi et al. 2001), whereas GDNF is produced in the dopaminergic projection fields of striatum (Trupp et al. 1997; Golden et al. 1998). Optimal GDNF signal induction depends on ligand-induced recruitment of Ret into lipid rafts which may occur in cis (membrane anchored GFRα1) or trans (released GFRα1) (Tansey et al. 2000; Paratcha et al. 2001). The ligand-receptor complex triggers Ret dimerization/activation with subsequent autophosphorylarion of specific tyrosine residues and recruitment of several intracellular signaling cascades (Airaksinen and Saarma 2002; Ibanez 2013). The PI3-K, jun N-terminal kinase, mitogen-activated protein kinase, phospholipase C-γ, and Src signaling pathways contribute to the propagation of GDNF signaling and shape its action on neuronal growth, differentiation, and survival (Airaksinen and Saarma 2002; Ibanez 2013). Some of these pathways are engaged by GM1 and their involvement in its neurotrophic effects has been proposed (Duchemin et al. 2002, 2008; Mo et al. 2005).
Our laboratory has a long-standing interest in the actions of exogenous GM1 on brain dopaminergic neurons (Hadjiconstantinou et al. 1986; Hadjiconstantinou and Neff 1988; Goettl et al. 1999, 2003). The overwhelming evidence that GM1 mimics the GDNF effects on dopaminergic neurons and modulates receptor tyrosine kinases for neurotrophic factors (Hadjiconstantinou and Neff 1998, 2000; Mocchetti 2005) prompted us to explore the possibility of a GM1 interaction with the GDNF receptor complex in the brain. The activation and phosphorylation of Ret by exogenous GM1 was evaluated in striatal slices in situ, a system utilized to demonstrate the modulatory effect of GM1 on Trk and its signaling (Duchemin et al. 2002, 2008). Our studies indicate that GM1 enhances Ret signaling in striatum in situ and this effect requires the presence of GFRα1. To determine the functionality of the in situ findings, Ret activity and phosphorylation were evaluated in the MPTP-mouse model of Parkinson's disease. Chronic GM1 treatment increased the activity and phosphorylation of Ret in striatum and enhanced Ret signaling was associated with attenuation of dopamine deficits. Preliminary results of these studies have been reported (Newburn et al. 2005, 2007; Hadjiconstantinou et al. 2010).
Animals and treatments
Male Swiss-Webster mice, 30–35 g, were used for the studies (Harlan Laboratories, Indianapolis, IN, USA), which were performed in accordance with the Guide for Care and Use of Laboratory Animals as adopted by the National Institutes of Health, USA, after approval by the Ohio State University Institutional Laboratory Animal Care and Use Committee. For the in vivo studies, mice were treated with MPTP, 30 mg/kg, ip, or saline daily for 7 days. Twenty-four hour later, mice were administered GM1, 30 mg/kg, ip, or saline daily for 15 days (Hadjiconstantinou and Neff 1988) and used for experimentation 1, 3, 12, and 24 h post-treatment. For the in situ studies, naive mice were used. Animals were killed by decapitation, striatum, and midbrain dissected and immediately used for the preparation of slices or lysates.
Striatal Slice preparation and treatments
Slices were prepared as we have described (Duchemin et al. 2002). Based on time- and concentration–response studies (Fig. 1a and b), GM1, 100 μM, (Transbussan, Geneva, Switzerland) and 5 min incubation at 37°C were routinely used. GDNF, 50 ng/mL, (Jing et al. 1996; Trupp et al. 1996; EMD Millipore, Billerica, MA, USA) served as positive experimental control. To determine GM1 specificity, slices were treated with 100 μM of C2 ceramide (Enzo Life Sciences, Plymouth Meeting, PA, USA), sulfatide (Sigma-Aldrich, Saint Louis, MO, USA), asialo GM1 (Alexis Biochemicals, San Diego, CA, USA), and N-acetylneuraminic acid (NANA) (Sigma-Aldrich) for 5 min. The Ret kinase inhibitors, PP2 and PP1 (Enzo Life Sciences), 1 μM, the non-active PP2 form PP3 (EMD Millipore), the Trk inhibitors K252a, 250 nM, and AG879, 100 μM, (Sigma-Aldrich) and phosphatidylinositol-specific phospholipase C (PI-PLC; Sigma-Aldrich), 0.6 U/mL, were added 30 min prior to treatment. Slices were pre-incubated with anti-GDNF neutralizing antibodies (MAB 212; R&D Systems, Minneapolis, MI, USA), 10 μg/tube, 45 or 90 min prior to GM1 or GDNF. Under our experimental conditions, the brain slice preparation is viable for longer than 2 h (Duchemin et al. 2002).
In vitro kinase assays
Striatal slices or one striatum were lysed in buffer containing 20 mM Tris, pH 7.4, 150 mM NaCl, 1% igepal (NP-40), 10% glycerol, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride and 1 μg/mL aprotinin, centrifuged and supernatant (tissue lysates) removed. Tissue lysates, 1 mg protein, were immunoprecipitated with 4 μg of anti-Ret antibodies (C-20; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and protein G PLUS-agarose (Santa Cruz Biotechnology) (Duchemin et al. 2008; Tsui-Pierchala et al. 2002a; with modifications; Supporting Information). Ret immunoprecipitates were washed with kinase assay buffer (50 mM HEPES, pH 7.4, 10 mM MgCl2, 10 mM MnCl2) and each divided into two samples for estimating Ret activity and monitoring immunoprecipitation efficiency, respectively. The amount of immunoprecipitated Ret used for the kinase assay was chosen based on preliminary protein response studies and was in the linear portion of the obtained curve. Immunoprecipitates were incubated in kinase buffer containing 2.5 μCi [γ-32P] ATP (3000 Ci/mmol; Amersham, Piscataway, NJ, USA) and myelin basic protein (MBP), 2 μg (Sigma-Aldrich) for 20 min at 23°C (Alberti et al. 1998; Qiao et al. 2001). Proteins were separated by 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and after exposure to X-ray film, membranes were stained with Ponceau Red and bands corresponding to MBP excised and 32P incorporation measured by liquid scintillation counting. The density of MBP band on the X-ray film was estimated (MetaMorph, Molecular Devices, Sunnyvale, CA, USA) and data expressed as percent of basal in the same experiment (Duchemin et al. 2002, 2008).
PI3-K activity was estimated as we have described (Duchemin et al. 2008). Immunoprecipitates were incubated with kinase buffer containing phosphatidylinositol, and 10 μCi [γ-32P]ATP (3000 Ci/mmol; Amersham) for 30 min at 22°C. Lipids were extracted and separated by TLC. The PI3-K inhibitor wortmannin served to identify assay specificity. Signal density of the PI(3)P band on the X-ray film was estimated (MetaMorph) and data expressed as percent of basal in the same experiment.
Lysate samples were incubated with anti-Akt agarose conjugated antibody (Santa Cruz Biotechnology), and Akt activity estimated in kinase buffer containing 10 μCi[γ-32P]ATP (3000 Ci/mmol, Amersham) and histone-2B. Proteins were separated by 15% SDS–PAGE, and phosphorylated histone-2B visualized by autoradiography (Duchemin et al. 2008).
Striatal slices or one striatum were lysed in 20 mM Tris, pH 7.4, 150 mM NaCl, 1% igepal (NP-40), 1 mM EDTA, 1 mM EGTA, 10 mM NaF, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride and 1 μg/mL aprotinin and supernatant (tissue lysate) cleared. Proteins were separated by 7.5 or 10% SDS–PAGE and immunoblots were performed and quantified as described (Duchemin et al. 2009; Supporting Information). Dilutions of primary antibodies were as follows: anti-Ret (1 : 200; Santa Cruz Biotechnology, C-20), anti-p-Ret (1 : 200; Santa Cruz Biotechnology, Tyr1062-R), anti-p-Ret-PY1062 (1 : 500; gift by Dr Brian Pierchala), anti-GFRα1 (1 : 1000; Santa Cruz Biotechnology, H-70), anti-p-Tyr (1 : 1000; EMD Millipore, 4G10), anti-GDNF (B-20, 1 : 500; Santa Cruz Biotechnology); anti-Src (1 : 1000; Santa Cruz Biotechnology, B-12), anti-Src[pY418] (1 : 1000; BioSource International, Inc, Camarillo, CA, USA), and anti-tubulin (B-7, 1 : 20 000; Santa Cruz Biotechnology). Proteins were identified based on a number of control conditions, co-blotted MW markers, and positive protein controls. Band density was estimated (Meta Morph) and following appropriate correction protein levels expressed as percent of basal in the same experiment.
Detection of GDNF and GM1 in GFRα1 immunoprecipitates
Lysate samples, 500 μg protein, were immunoprecipitated with 2 μg of anti-GFRα1 antibodies (Santa Cruz Biotechnology, H-70) and protein A/G PLUS-agarose (Santa Cruz Biotechnology). Proteins were separated by 15% SDS–PAGE, and western blots performed with anti-GDNF antibodies (1 : 500; Santa Cruz Biotechnology, B-20). Membranes were reprobed with anti-GFRα1 antibodies (1 : 1000; Santa Cruz Biotechnology, H-70). The density of GDNF and GFRα1 bands was estimated, the GDNF/GFRα1 ratio determined, and data expressed as percent of basal in the same experiment. The presence of GM1 in GFRα1 immunoprecipitates was evaluated with a horseradish peroxidase-conjugated cholera toxin B subunit (1 : 25 000; List Biological Laboratories, Campbell, CA, USA), and detergent resistant membrane lysates served as positive control for GM1 detection (Duchemin et al. 2008).
GDNF was estimated by ELISA (Emax(R)ImmunoAssay Systems, Promega, Madison, WI, USA) in striatal slices media in situ and in tissue lysates prepared from striatal slices in situ and striatum in vivo. GDNF released in the media was calculated as percent of total GDNF content (tissue and media).
Dopamine and DOPAC estimation
Dopamine and DOPAC (3,4-dihydroxyphenylacetic acid) were estimated with HPLC-ED (Hadjiconstantinou and Neff 1988).
For the statistical analysis of the data, the non-parametric Kruskal–Wallis test followed by a Dunn's post hoc test for multiple group comparisons was used. Independent Mann–Whitney test or t-test was used as appropriate. Analysis was performed with (GraphPad Software Inc., La Jolla, CA, USA), and a level of p < 0.05 was considered as statistically significant.
In initial studies, immunoblotting was employed to identify the GDNF receptor complex components present in Ret and GFRα1 immunoprecipitates from striatal slices. Under control conditions, Ret immunoprecipitates contained GFRα1, whereas Ret was identified in GFRα1 immunoprecipitates (data not shown), suggesting Ret and GFRα1 complexing. Congruent with reports that Ret is associated with Src (Encinas et al. 2001), Src was present in striatal Ret immunoprecipitates (Fig. 3a and c). Immunoblots of Ret immunoprecipitates from striatum revealed a single band of ~ 170 kDa (Fig. 1, 2 and 7a; Figure S1b and S2a and b) corresponding to the fully glycosylated mature Ret protein (Takahashi et al. 1993), and hence our studies present changes in Ret 170 kDa (Ret).
Activation of Ret by GM1
When slices prepared from striatum or midbrain were incubated with GM1, 100 μM, or GDNF, 50 ng/mL, for 5 min, there was an increase in kinase activity associated with Ret immunoprecipitates demonstrated by increased phosphorylation of the exogenous substrate MBP in an in vitro kinase assay (Figure S2a). Across all studies GM1, 100 μM, caused an over 2-fold increase in kinase activity in striatal slices. Kinase activation required the whole GM1 molecule, as treatment with NANA, C2 ceramide, asialo GM1, and sulfatide had no effect (Figure S2b). Direct treatment of Ret immunoprecipitates with GM1, 100 μM, for 5 min had no effect on kinase activity (data not shown). The GM1-induced Ret activation in striatal slices was time [KW = 39.089, p < 0.0001] and concentration [KW = 33.361, p < 0.0001] dependent. Kinase activity was increased by 2 min after GM1, reached a maximal by 5 min, and remained elevated for over 30 min returning to basal levels by 60 min (Fig. 1a). At 5 min, 25–100 μM of GM1 produced a maximal response with increases in Ret activity being evident at 1 μM and becoming significant at 10 μM (Fig. 1b). Higher concentrations of GM1, > 500 μM, however, attenuated the response (Fig. 1b). The calculated EC50 for the GM1-induced activation of Ret was ~ 12 μM. GM1 had no effect on the Ret content, as there was no difference in the amount of immunoprecipitated Ret over time [KW = 3.220, p = 0.7808] or with various GM1 concentrations [KW = 3.147, p = 0.6773]. Similar results were obtained when Ret was estimated over time, 2–60 min, in striatal lysates by immunoblot (data not shown).
Ret response characterization
To ascertain the specificity of the Ret response the kinase inhibitors PP2 and PP1 were employed. Originally proposed as Src family kinase inhibitors (Hanke et al.1996; IC50 ~ 5 nM), PP2 and PP1 exhibit a broader kinase inhibition and inhibit Ret kinase with an IC50 of 40–100 nM (Carlomagno et al. 2002, 2003; Mologni et al. 2005). Because MBP, a commonly used Ret substrate (Methods), is not a known Src substrate (Brown and Cooper 1996; Martinez Ferrando et al. 2012) the PP2 and PP1 use as Ret inhibitors in the kinase assay was deemed appropriate. PP2 or PP1, 1 μM, had no appreciable effect on basal kinase activity (Fig. 2a–c), but higher concentrations, 2–10 μM, abrogated it in a concentration-dependent fashion (data not shown). Given that 1 μM of PP2 or PP1 inhibit about 80% of Src kinase activity (Bain et al. 2007) their displayed low potency in our kinase assay is consistent with Ret and not Src inhibition. PP2 or PP1, 1 μM, markedly inhibited the Ret activation by GM1 (Fig. 2a and b) or GDNF (Fig. 2c). PP3, the negative control for PP2, had no effect (data not shown). Pre-treatment with PI-PLC to cleave the GPI anchor and remove GFRα1 (Jing et al. 1996; Treanor et al. 1996; Trupp et al. 1999) prevented Ret activation by GM1 or GDNF (Fig. 2d).
GM1 enhances Ret phosphorylation
Ret immunoprecipitates were immunoblotted with an anti-phosphotyrosine antibody to evaluate kinase autophosphorylation. Following GM1, a band with MW of ~ 170 kDa displayed enhanced phosphotyrosine signal that was sensitive to PP2 and identified as Ret (Fig. 3a). Similar observations were made when lysates from GM1- or GDNF-treated striatal slices were immunobloted with anti-p-Ret antibodies recognizing phosphorylated Ret at Tyr1062 (Fig. 3b; Figure S1c). The antibodies detected a band of ~ 170–180 kDa that has been identified as phosphorylated Ret in sympathetic neurons (Tsui-Pierchala et al. 2002a). GM1 or GDNF treatment increased the levels of the pRet (~ 3- and 5-fold, respectively), and the response was abrogated by PP2 (Fig. 3b) signifying Tyr1062 as a Ret autophosphorylation site after GM1.
GM1 enhances Src phosphorylation
In addition to pRet, anti-phosphotyrosine immunoblots of Ret immunoprecipitates from GM1-treated striatal slices detected enhanced phosphorylation of a band of ~ 60 kDa, which was also sensitive to PP2 and identified as Src (Fig. 3a). To confirm this finding, Ret immunoprecipitates or tissue lysates were immunoblotted with a pSrc antibody recognizing Src phosphorylatated at Tyr418 which is required for full catalytic activity (Bjorge et al. 2000). GM1 and GDNF enhanced pSrc levels, ~ 70 and 90% over basal, respectively, and responses were blocked by PP2 and PI-PLC (Fig. 3c and d).
GM1 promotes the binding of endogenous GDNF to GFRα1
That Ret activation entails GDNF binding to GFRα1 as a first step (Jing et al. 1996; Treanor et al. 1996) and GFRα1 is required for the GM1 action on Ret (Fig. 2d these studies) raises the possibility that GM1 influences GDNF binding. This was explored by following the amount of GDNF bound to GFRα1 immunoprecipitates by immunoblot. The levels of the immunoprecipitated GFRα1 were not different from basal following GDNF or GM1 treatment [K = 0.1681, p = 0.9194], and the same observation was made following GM1 concentration, 1–100 μM, response studies [KW = 2.449, p = 0.6453]. Accordingly, the bound GDNF to GFRα1 was computed as GDNF/GFRα1 ratio. Basal GDNF binding to GFRα1 was low in agreement with the observed basal Ret activation and phosphorylation (Figs 1-3). GDNF treatment dramatically increased, 3- to 4-fold over basal, the GDNF immunoreactivity associated with GFRα1 (Fig. 4a and b), indicating enhanced binding of the factor. Likewise, more GDNF immunoreactivity was detected in the GFRα1 immunoprecipitates after GM1, ~80–100% over basal (Fig. 4a and b), suggesting increased binding of endogenous GDNF. Endogenous GDNF binding was GM1 concentration dependent (Fig. 4c) with a calculated EC50 ~ 13 μM, which is comparable to that for Ret kinase activation ~ 12 μM. There was a slight GDNF binding at 1 μM of GM1 that gradually increased and reached a plateau at 50–100 μM, implying saturability.
GM1 could increase endogenous GDNF binding to GFRα1 in situ indirectly by stimulating its release in the media, or directly by promoting the binding of tissue GDNF. GM1 had no effect on tissue or media GDNF content, and GDNF release was unaltered (Fig. 5a). The same was true for midbrain (data not shown). To exclude the likelihood that GM1 promotes the binding of the GDNF released in the medium during the assay, striatal slices were incubated with GDNF neutralizing antibodies prior to adding GM1, 100 μM, or GDNF, 50 ng/mL, and pRet estimated in tissue lysates. The dose of GDNF antibodies, 10 μg/500 μl/tube, was calculated from its Neutralization Dose50, approximately 1–3 μg in the presence of 10 ng/mL of rhGDNF (manufacturer's information). This antibody dose was deemed sufficient to completely neutralize the activity of the released GDNF, estimated to 0.21 ng/500 μl/tube. Two time periods of antibody pre-incubation were tested, 45 and 90 min, with similar results. As shown in Fig. 5b, the neutralizing antibodies reduced by about 42% the pRet levels induced by exogenous GDNF, but had no effect on the basal or GM1-enhanced kinase phosphorylation. Efforts to develop a GDNF biochemical binding assay with striatal membranes have been unsuccessful and we have not been able to investigate the effect of exogenous GM1 on GDNF binding.
GM1 is not associated with GFRα1
To ascertain whether association of GM1 with GFRα1 was necessary for the GM1-stimulated binding of endogenous GDNF GFRα1 immunoprecipitates were blotted with cholera B toxin (Fig. 4d). No cholera B toxin band was detected in GFRα1 immunoprecipitates under basal, GM1 or GDNF treatment conditions, suggesting that neither exogenous (GM1 treatment) nor endogenous (basal) GM1 is associated with GFRα1. Because GFRα1 immunoprecipitates contain Ret (see 'Co-immunoprecipitations') and GDNF (Fig. 4a–c), the findings imply lack of GM1 association with any putative GDNF receptor complex, such as GFRα1, GDNF/GFRα1, GFRα1/Ret or GDNF/GFRα1/Ret, under basal or stimulated conditions.
Ret contributes to the PI3-K activity enhancement by GM1
We have shown that GM1 enhances the activity of PI3-K and its target Akt in striatal slices in situ, an effect partially owing to Trk activation (Duchemin et al. 2008). Ret Tyr1062, which was found phosphorylated after GM1 in these studies, provides a site for diverse signaling recruitment including PI3-K/Akt cascade (Airaksinen and Saarma 2002). A reasonable question, hence, is whether Ret contributes to the GM1 activation of PI3-K and Akt. As we have reported (Duchemin et al. 2008), treating striatal slices with GM1 increased PI3-K (Fig. 6a) and Akt activity (Fig. 6d), which was inhibited, in part, by PP2 (Fig. 6a–d). PP2 did not prevent the activation of PI3-K following nerve growth factor (NGF), 50 ng/mL, treatment (Basal 100 ± 4; PP2 96 ± 13; NGF 223 ± 27*; NGF + PP2 199 ± 25*% of Basal ± SEM. *p < 0.05 vs. basal. n = 8–10). To exclude a PP2 effect through Src, herbimycin, a selective Src family kinase inhibitor (Uehara et al. 1989), was utilized. Herbimycin, 50 μM, failed to bock the GM1-induced increase of PI3-K activity (Fig. 6b and c).
GM1 enhances Ret kinase activity and phosphorylation in vivo
To validate our in situ findings and provide a measure of functionality for the GM1-induced Ret activation/phosphorylation in striatal slices in situ, we investigated the effect of GM1 on Ret in the MPTP-mouse. As mentioned afore, in this animal model of Parkinson's disease chronic GM1 treatment restores TH+ cell morphology in SNc; attenuates dopamine and metabolite deficits in striatum; and improves motor and somatosensory impairments (Hadjiconstantinou et al. 1986, 1989; Hadjiconstantinou and Neff 1988; Weihmuller et al. 1988). In these studies, the effect of chronic GM1 on Ret protein, activity and phosphorylation was assessed. Ret immunoblots of tissue lysates were prepared from striata in vivo revealed two bands of ~ 150 and ~ 170 kDa (supplier's information; see Ret immunoblots in Fig. 7b and Figure S1a). Because the detection of the ~ 150 kDa band was inconsistent (Supporting Information) and to be in line with our in situ studies only the ~ 170 kDa Ret was evaluated. Fifteen days after MPTP, there was a reduction in Ret protein in the striatum (Control 100 ± 3; MPTP 58 ± 8*% of Control ± SEM. *p < 0.002 vs. control. n = 15) calculated across all time-points. Diminished Ret protein was reflected by attenuated kinase activity (Control 100 ± 4; MPTP 70 ± 3*% of Control ± SEM. *p < 0.0001 vs. control. n = 20) and phosphorylation at Tyr1092 (Control 100 ± 7; MPTP 66 ± 5*% of Control ± SEM. *p < 0.0001 vs. control. n = 20) calculated across all time-points. GM1 treatment had no effect on the Ret levels in the intact [KW = 4.230, p = 0.2376] or MPTP [KW = 3.730, p = 0.2922] mice (data not shown, see Fig. 7b Ret immunoblots in intact and MPTP mice), but it increased Ret activity and phosphorylation (Fig. 7a and b). Ret activity increased at 1 h post-GM1 in intact and MPTP mice, about 2-fold when values were expressed as% of intact control mice, lasted over 3 h and reached intact control levels by 12 h; [KW = 21 982, p < 0.0001] for intact mice, and [KW = 30 699, p < 0.0001 for MPTP mice. The phosphorylation of Ret at Tyr1092 increased also in both treatment groups after GM1; [KW = 19 414, p < 0.0002] for intact mice, and [KW = 29 332, p < 0.0002] for MPTP mice. Notably, the magnitude and time kinetics of the pRet response were parallel to those for kinase activity. In the GM1-treated MPTP mice Ret activity and pRet levels were still above the MPTP control values at 12 h (Fig. 7a and b), a difference that was no longer seen 24 h post-treatment (data not shown). GFRα1 and GDNF protein in striatum was within control levels in all treatment groups and time-points studied (data not shown). Treatment with GM1, partially corrected the decreased dopamine and DOPAC content in the striatum of the MPTP mice, but had no effect in intact mice (Fig. 7c), as published (Hadjiconstantinou et al. 1986; Hadjiconstantinou and Neff 1988).
Despite ample evidence that GM1 displays neurorestorative effects on dopaminergic neurons following an injury and during normative aging (Hadjiconstantinou et al. 1986, 1989; Hadjiconstantinou and Neff 1988, 1998, 2000; Goettl et al. 1999, 2003) the mechanism(s) underlying the GM1 actions is largely unknown and speculative. Suggestions for a role of brain-derived neurotrophic factor have not materialized, as the GM1-induced phosphorylation of Trk B in brain is low and occurs primarily in frontal cortex and hippocampus (Duchemin et al. 2002). These studies convincingly demonstrate that GM1 activates and phosphorylates the cell membrane associated 170 kDa Ret protein (Takahashi et al. 1993) with subsequent recruitment of Src and PI3-K signaling cascades in striatal slices in situ; an action shared with GDNF. Incubation of striatal slices with GM1 enhanced the kinase activity associated with Ret immunoprecipitates, as evidenced by the increased phosphorylation of the exogenous substrate in an in vitro kinase assay. Under the same experimental conditions, GDNF increased the activity of Ret confirming the validity of our in situ system. PP2 and PP1, tyrosine kinase inhibitors with Ret efficacy (Carlomagno et al. 2002, 2003; Carniti et al. 2003; Mologni et al. 2005; Knowles et al. 2006), completely blocked the GM1 and GDNF effect establishing the specificity of the kinase assay and GM1 response. Moreover, Ret activation required the whole GM1 molecule, as C2 ceramide, sulphatide, NANA and asialo GM1 had no effect in line with previous reports on Trk (Duchemin et al. 2002).
Following GM1, Ret activation occurred rapidly and was maintained for over 30 min returning to basal levels by 60 min, a temporal pattern similar to that reported for GDNF-induced kinase phosphorylation (Jing et al. 1996; Trupp et al. 1996; Coulpier et al. 2002). The kinase response was GM1 concentration dependent with an estimated EC50 ~ 12 μM, and bell-shaped with high, > 500 μM, concentrations losing effectiveness. A similar response pattern has been reported for Trk and its downstream signaling (Duchemin et al. 2002, 2008; Nishio et al. 2004) with the suggestion that GM1 might have concentration-dependent kinase enhancing or suppressive effects. Ret activation by GM1 was accompanied by enhanced kinase autophosphorylation demonstrated by pTyr immunoblots of Ret immunoprecipitates and pRet level measurements in tissue lysates. GM1- and GDNF-induced Ret signaling in striatal slices in situ requires the presence of the GPI-anchored GFRα1 co-receptor suggesting recruitment of Ret by a cis mechanism (Treanor et al. 1996; Tansey et al. 2000; Paratcha et al. 2001).
Several tyrosine residues have been identified as autophosphorylation sites in Ret and their binding partners and downstream signaling pathways characterized (Airaksinen and Saarma 2002; Sariola and Saarma 2003; Encinas et al. 2004). Tyr1062, shown to be phosphorylated by GM1, engages the PI3-K/Akt and Erk pathways in neurons (Trupp et al. 1999; Besset et al. 2000; De Vita et al. 2000; Hayashi et al. 2000; Paratcha et al. 2001), which are activated by GM1 in striatal slices in situ (Duchemin et al. 2002, 2008). Conducted kinase inhibitor studies support Ret involvement in the GM1-induced activation of the PI3-K and its target Akt (Duchemin et al. 2008; these studies) and Erk (Duchemin et al. 2002; unpublished observations). The Ret-dependent activation of Src by GM1 provides indirect evidence for phosphorylation of Ret Tyr981, the primary residue responsible for Src activation (Encinas et al. 2004). The enhanced Src Tyr419 phosphorylation seen in Ret immunoprecipitates indicates interaction between the two kinases (Encinas et al. 2001) and Ret-dependent Src activation in striatal slices in situ (Trupp et al. 1999). Interestingly, NGF had no effect on Src activity in striatal slices, suggesting that GM1-induced Src activation is mostly associated with Ret expressing neuronal terminals (Nosrat et al. 1997; Trupp et al.1997).
Together our published and present studies indicate that GM1 recruits overlapping signaling cascades in the striatum via Trk (Duchemin et al. 2002, 2008) and Ret (these studies). In striatum, GM1 induces a robust activation of TrkA and C while its effect on TrkB is quite small (Duchemin et al. 2002, 2008), suggesting that TrkB signaling is a minor contributor to GM1's effects in the nucleus. Based on these observations and taken into consideration that in striatum TrkA is expressed in intrinsic cholinergic neurons (Sobreviella et al. 1994; Holtzman et al. 1995) and TrkC and Ret are present predominantly in dopaminergic efferents (Lamballe et al. 1994; Nosrat et al. 1997; Trupp et al. 1997; Numan and Serrogy 1999), we propose that GM1 is capable of initiating (i) TrkA dependent activation of Erk and PI3-K signaling in striatal intrinsic cholinergic neurons; (ii) TrkC-dependent Erk and PI3-K signaling in mesencephalic dopaminergic neurons; and (iii) Ret dependent PI3-K, Erk, and Src signaling in mesencephalic dopaminergic neurons. The ability of GM1 to activate Trk and Ret signaling in the striatum with similar kinetics is of special interest. Neurons respond synergistically to multiple neurotrophic factors and GM1 may act as a pleiotropic neurotrophic factor modulator in the same neuron. The rapid kinetics of Trk (Duchemin et al. 2002, 2008) and Ret activation by GM1 in striatal slices in situ preclude a Trk-mediated transactivation of Ret and a Trk/Ret cross-talk, as it has been reported for TrkA (Tsui-Pierchala et al. 2002b; Pierchala et al. 2007) and TrkB (Esposito et al. 2008) in cell cultures. In agreement, the Trk inhibitors K252a and AG879 (Duchemin et al. 2002, 2008) failed to prevent the GM1-induced Ret activation in striatal slices (data not shown). Because the Trk/Ret cross-talk has slow kinetics and requires hours for Ret phosphorylation by the neurotrophins (Tsui-Pierchala et al. 2002b; Pierchala et al. 2007), we cannot rule out the likelihood of GM1 eliciting such a kinase interplay in cell systems or animals in vivo under appropriate conditions.
Despite convincing evidence that exogenous gangliosides and GM1 modulate the activation and phosphorylation of receptor tyrosine kinases for trophic factors the underlying mechanisms are conjectural and subject to debate (Posse de Chaves and Sipione 2010). Our observation that GM1 modulates Ret activation and autophosphorylation provides new clues as to how GM1 may effect trophic signaling. Our studies clearly demonstrate that (i) GM1 activation of Ret requires GPI-anchorage of GFRα1, and (ii) GM1 at concentrations that increase Ret activity increases the binding of endogenous GDNF to GFRα1. Enhanced GDNF binding is GM1 concentration dependent and saturable, and parallels the Ret response with an estimated EC50 comparable to that for Ret activation, 13 μM and 12 μM, respectively. The possibility that GM1 acts by promoting the binding of GDNF released in the media is unlikely as (i) the release of GDNF was not stimulated by GM1, and (ii) neutralization of the released GDNF had no effect on the GM1-induced Ret phosphorylation. It should be noted, that the estimated GDNF release under basal conditions was low, to 0.42 ng/mL, and its neutralization had no measurable effect on basal pRet levels implying that the detected GDNF in the media was not a major contributor to Ret activity.
The modulatory effect of exogenous GM1 on the binding of endogenous GDNF to GFRα1 may be related to GM1's ability to be taken up by neuronal cells (Shwarzmann 2001) and interact with GPI-anchored proteins (Kasahara and Sanai 2000; Loberto et al. 2003; Prioni et al. 2004). In previous studies, we explored the fate of exogenous GM1 in striatal slices in situ. Using identical experimental conditions, we found that following addition to striatal slices exogenous GM1 rapidly binds to the plasma membranes and accumulates over time (Duchemin et al. 2008). At the time of Ret maximal activation, 5–10 min, an estimated 3% of the ganglioside is bound to the plasma membranes of which roughly 25% is inserted (Duchemin et al. 2008). The rapidity and temporal pattern of the acute GM1-induced Ret signal induction in striatal slices indicate that insertion of the ganglioside in the plasma membrane is not required, a conclusion made for Trk as well (Duchemin et al. 2008 and discussion), and that the GM1/GFRα1 interaction occurs is trans (ganglioside bound on the plasma membrane surface but not inserted; Fantini and Barrantes 2009). This supposition is further supported by the fact that as more exogenous GM1 is incorporated into the plasma membrane over the time (Duchemin et al. 2008) the Ret response declines (these studies).
The majority of exogenously administered GM1 is associated loosely or tightly with the lipid rafts, of which is a major glycosphingolipid constituent (Prinetti et al. 1999; Loberto et al. 2003; Prioni et al. 2004), and this might be critical for understanding its actions on surface receptors and their signaling. GM1 could modulate the GFRα1 by altering the biophysical properties of the membrane and the spatial relations of the clustered receptors in the lipid rafts, changing the function of its GPI anchor, or interacting with the GFRα1 itself in the extracellular space (Fantini and Barrantes 2009; Posse de Chaves and Sipione 2010; Caré and Soula 2011). In either case, changes in receptor conformation might occur that affect endogenous ligand-binding kinetics and Ret signaling. There is substantial evidence in favor of a physical association between gangliosides and surface receptors, which involves glycans, oligosaccharide chains, and specific sphingolipid-binding domains (Fantini and Barrantes 2009; Posse de Chaves and Sipione 2010). Using cholera toxin B, we were not able to demonstrate GM1 binding to GFRα1 complexes under basal or stimulated conditions. This is in line with a trans GM1/GFRα1 interaction which is readily reversible. The possibility that under our experimental conditions GM1 may directly or indirectly affect GFRα1 receptor trafficking is rather unlikely, as GM1, 1–100 μM, had no effect on the receptor protein content in tissue lysates (sum of cytosolic and membrane receptors) or lipid rafts (membrane receptors; personal observation) at the time of the maximally increased binding of endogenous GDNF to GFRα1 and Ret phosphorylation which occurred 5 min after GM1. Taking into consideration the consensus that the final GDNF signal complex conforms to a GDNF2:GFRα12:Ret2 stoichiometry (Wang 2013), we hypothesize that GM1 may enhance the binding of endogenous GDNF to inactive GFRα1 and/or preformed GFRα1/Ret complexes present in lipid rafts, the existence of which has been proposed (Eketjall et al. 1999; Cik et al. 2000; Bespalov and Saarma 2007; Runeberg-Roos and Saarma 2007).
That GM1 increases Ret activity and phosphorylation in the striatum in vivo confirms our in situ findings and provides additional credence for the validity and utility of our in situ system. Indeed, chronic treatment with GM1 increased Ret activity and pRetTyr1092 levels in the striatum of the intact mice, a response that occurred soon after GM1 discontinuation and lasted for over 3 h. Ret protein, activity, and phosphorylation at Tyr1092 were reduced in the striatum of the MPTP-mouse and chronic GM1 enhanced kinase activity and phosphorylation without affecting protein levels. The activation and phosphorylation of Ret in MPTP mice paralleled that in intact mice implying a common mechanism(s). However, it appeared that kinase deactivation was slower in the injured animals. Given that MPTP causes an about 50–60% loss of TH+ neurons in SNc and GM1, despite their morphological improvement, does not change their number (Hadjiconstantinou et al. 1989), we surmize that the Ret modulation by the ganglioside occurs predominantly in the residual dopaminergic neurons. Enhancement of Ret signaling by GM1 in the MPTP-mouse is functional as it coincides with dopamine recovery in the striatum (Hadjiconstantinou et al. 1986; Hadjiconstantinou and Neff 1988; these studies) and improved motor and somatosensory impairments (Weihmuller et al. 1988). The phosphorylation of Ret Tyr1092 by GM1 together with reports that Erk and PI3-K are involved in the GDNF effect on injured nigrostriatal dopaminergic neurons (Pong et al. 1998; Sawada et al. 2000; Ugarte et al. 2003) support a role for Tyr1062 signaling in the neurotrophic effects of GM1 in the MPTP-mouse.
In summary, our studies provide evidence for Ret activation and phosphorylation in striatum in situ after GM1. The GM1 effect requires the presence of GFRα1 and involves enhanced binding of the endogenous GDNF. GM1-induced Ret activation contributes to the GM1 engagement of the PI3-K/Akt and Src signaling pathways and may underlie the neurotrophic actions of the ganglioside on striatal neurons. As the majority of Ret in striatum is expressed in dopaminergic projections (Trupp et al. 1997) and GM1 enhanced Ret in midbrain also, we assume that the GM1 effect on Ret signaling takes place mainly in mesencephalic dopaminergic neurons and propose that the data provide a mechanism for the well-established neurorestorative effect of GM1 on injured nigrostriatal dopaminergic neurons. Indeed, chronic GM1 treatment enhanced Ret activation and phosphorylation in the striatum of the MPTP-mouse, which coincided with attenuation of dopamine and metabolite deficits. GM1's effects on dopaminergic neurons may, also, involve Ret signaling by the GFL neurturin, which is a dopaminergic neurotrophic factor on its own (Kordower and Bjorklund 2013). Neurturin displays high affinity for GFRα2, but also binds to GFRα1 with lower affinity (Jing et al. 1997). The idea that GM1 may interact with the binding of other GFLs to their cognate GFRα receptors is attractive and needs further exploration. In closing, we would like to stress that although our studies focused on striatum and dopaminergic neurons, GM1 may enhance Ret signaling in other brain regions and spinal cord and include other neuronal phenotypes, such as cholinergic, serotonergic, noradrenergic, and motor neurons, which are targets of GDNF (Bohn 1999) and their function reportedly is improved by the ganglioside (Hadjiconstantinou and Neff 1998, 2000; Goettl et al. 1999, 2000, 2002, 2003).
Acknowledgment and conflict of interest disclosure
The research was supported by a Parkinson Foundation Development Fund, Department of Pharmacology, The OSU College of Medicine. The authors want to thank Dr Brian Pierchala for the generous gift of the anti-p-Ret antibodies and useful conversations. The authors have no conflict of interest to disclose.
All experiments were conducted in compliance with the ARRIVE guidelines.