Monosialoganglioside GM1 De ciency Inhibits the Neurotrophic Effects of GDNF by Disrupting Lipid Raft Assembly


 Recent studies have shown that monosialoganglioside GM1 deficiency can inhibit the signal transduction process of glial cell line-derived neurotrophic factor (GDNF), which plays an important role in the pathogenesis of Parkinson's disease (PD). However, its specific mechanism still needs to be explored. We inhibited the expression of GM1 by treating cells with D-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP). CCK-8 assay, EdU cell proliferation assay and Western blot assay were used to evaluate the effect of GM1 deficiency on the proliferation and differentiation of SH-SY5Y cells induced by GDNF and on the GDNF-RET signaling pathway. Lipid rafts were isolated by Triton X-100 solubilization and OptiPrepTM density gradient centrifugation. The alterations of lipid raft assembly and the translocation of RET into lipid rafts were evaluated after PDMP treatment. We found that PDMP treatment inhibited the proliferation and differentiation of SH-SY5Y cells induced by GDNF and reduced the phosphorylation of RET and its downstream signaling molecules Erk and Akt. In addition, after PDMP treatment, caveolin-1 and flotillin-1, the prototypical markers of lipid rafts, diffused from lipid rafts to non-lipid raft microdomains, and GDNF-induced RET translocation into lipid rafts was also reduced. These alterations could be partially reversed by adding exogenous GM1. Our results suggest that ganglioside GM1 deficiency could compromise the neurotrophic effects and signals downstream of GDNF by altering the assembly of lipid raft membrane microdomains.


Introduction
Parkinson's disease (PD) is a chronic degenerative disease of the central nervous system that seriously damages human health [1] . The main pathological change in PD is the progressive death of dopaminergic neurons in the substantia nigra pars compacta (SNpc), but the speci c mechanism is unclear. Recent research suggests that changes in the relative abundance of monosialotetrahexosylganglioside (GM1) may play a crucial role in the development and progression of PD [2,3] . Research data show that the GM1 content decreases with age in the brain. In people over 60, the total amount of gangliosides in the substantia nigra of the midbrain decreases, and this reduction is most pronounced in GM1 and GD1a [4] .
The expression of the ganglioside synthesis-related genes B3galt4 and St3gal2 is conspicuously absent in PD patients, and the GM1 level is decreased in the substantia nigra, prefrontal cortex and various peripheral tissues [5,6] . In B4galnt1 knockout mice, mice with ganglioside GM1 de ciency show the characteristic symptoms of PD, such as reduced dopamine levels in the striatum, loss of dopaminergic neurons in the SNpc, aggregation of α-synuclein, and motor dysfunction. Supplementation with the exogenous GM1 derivative LIGA20 can signi cantly alleviate the changes described above [7] . GM1 de ciency can not only cause neuropathological features in the substantia nigra striatum and motor disorders of PD but also lead to gastrointestinal autonomic nervous dysfunction and the non-motor symptoms of the brain cognitive system of PD [8] . All these results suggest that de ciency of endogenous ganglioside GM1 may be an initiator of the pathological process of PD and play an important role in it, but the speci c mechanism is not clear.
Recent studies have shown that GM1 de ciency can inhibit the signal transduction process of glial cell line-derived neurotrophic factor (GDNF) and trigger neurodegeneration [7] . GDNF is a nutrient factor identi ed and puri ed by Lin et al. from the serum-free medium of the rat glial cell line B49, which can promote the survival and morphological differentiation of dopaminergic neurons in the embryonic midbrain [9] . The protective and restorative effects of GDNF on dopaminergic neurons have been con rmed by a large number of experiments, and GDNF is an important target of medical intervention for neuroprotection and regeneration in the treatment of PD [10,11] . As an extracellular factor, GDNF needs to transmit signals into cells through membrane receptors. GDNF has a unique receptor system that consists of the transmembrane signal transduction receptor RET and the ligand binding receptor GFRα1 [12] . Lipid rafts play a vital role in the transmission of GDNF signals [13][14][15] . Lipid rafts are highly dynamic membrane microdomains enriched in cholesterol and spingolipids, and they are in liquid-ordered (Lo) phases with lower membrane uidity than surrounding cell membranes [16,17] . We and others showed that in the absence of GDNF, the signal transduction receptor RET is mainly located outside of lipid rafts [18] . After stimulation with GDNF, RET is translocated into lipid rafts, and multiple downstream effectors are recruited to activate RAS/Erk, PI3K/Akt and other signal transduction pathways involved in the regulation of cell survival, differentiation, migration, growth and other biological processes [13,14] .
As lipid rafts are in highly dynamic Lo phases, the lipid and protein composition of lipid rafts is not xed.
Current data suggest that changes in lipid composition can regulate the assembly and characteristics of lipid rafts [19][20][21] . Recently, studies of various model membrane systems have indicated that GM1 is also involved in the regulation of lipid raft assembly [22][23][24] . However, it is not clear whether GM1 can in uence the signal transduction process of GDNF-RET by altering lipid raft assembly in dopaminergic cells. To explore this problem, we used D-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP) to inhibit GM1 expression in SH-SY5Y cells [7,25] . We then investigated the effect of GM1 de ciency on the role of GDNF in promoting proliferation and differentiation, and regulating the GDNF-RET signaling pathway. Lipid rafts were isolated by Triton X-100 solubilization and OptiPrep TM density gradient centrifugation and the alterations of lipid raft assembly and the localization of RET were determined after PDMP treatment. Our results showed that PDMP treatment inhibited the proliferation and differentiation of SH-SY5Y cells induced by GDNF and reduced the phosphorylation of RET as well as the activation of its downstream signaling molecules Erk and Akt. In addition, after PDMP treatment, caveolin-1 and otillin-1, the representative markers of lipid rafts, diffused from lipid rafts to non-lipid raft microdomains, and GDNF-induced RET translocation into lipid rafts was also reduced. These alterations could be partially reversed by adding exogenous GM1. Our results suggest that ganglioside GM1 de ciency might compromise the neurotrophic effects of GDNF by altering the assembly of membrane microdomains and promote the progression of PD.

Materials And Methods
Cell culture SH-SY5Y human neuroblastoma cells were seeded into 25 cm 2 culture asks (NEST) with Dulbecco's minimum essential medium (DMEM)/F12 (KeyGen BioTech) containing 10% fetal bovine serum (Bovogen) and 1% penicillin/streptomycin (VICMED). All cells were cultured in a standard carbon dioxide incubator (5% CO 2 , 37°C; Thermo Fisher Scienti c). One day after seeding, the medium was changed to mixed medium with 10 µM retinoic acid (RA; Aladdin), and the cells were incubated for 4 days before experiments. The medium was changed every 2 days.
Cell proliferation assay For the EdU staining assay, SH-SY5Y cells were seeded in a 48-well plate at a concentration of 1×10 4 cells/well. When cells reached 70% con uence, cells were treated with PDMP and GM1 as described above. Then, the culture medium was replaced with fresh serum-free DMEM/F12 containing GDNF (50 ng/mL, R&D Systems) and 10 µM 5-ethynyl-2′-deoxyuridine (EdU). After 24 h incubation in the presence of EdU, cells were xed with 4% paraformaldehyde and permeabilized in 0.5% Triton X-100 in PBS for 10 min at room temperature. Then, the cells were incubated in 1×Apollo® staining reaction liquid for 30 min at room temperature in the dark, and nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI, KeyGen BioTech). The cells were examined and photographed with a uorescence microscope.
After indicated treatment, SH-SY5Y cells were placed on ice and washed twice with ice-cold PBS. The cells were then detergent extracted by adding detergent-resistant membrane (DRM) lysis buffer (TBS containing 1% Triton X-100 and 1% protease inhibitors) and rocking the plates gently at 4°C for 20 min. The crude lysates were then collected and spun at 16,200 g in a 4°C microfuge for 10 min. The supernatants were rapidly removed from the pellets and placed into new tubes. The pellets, which contained the DRMs, were washed quickly with ice-cold DRM lysis buffer and centrifuged again, and the supernatants were rapidly removed. Both the insoluble pellets and the supernatants were then treated with 2x sample buffer and boiled for 5 min.
A second method used to biochemically analyze lipid rafts was density gradient centrifugation with Optiprep TM reagent (Sigma). For these experiments, SH-SY5Y cells were maintained on 100 mm culture dishes precoated with collagen (NEST). After indicated treatment, cells were washed twice with ice-cold PBS. Subsequently, the cells were scraped of the dishes on ice, and this mixture was transferred to centrifuge tubes and homogenized by ultrasonication. Crude homogenates were mixed with 60% Optiprep™ containing 1% Triton X-100 to adjust the concentration of Optiprep™ to 45%, and the mixture was then placed at the bottom of an ultracentrifuge tube (Beckman). Next, 2 ml layers of 35%, 30%, 25%, 20% and 5% OptiPrep™ reagent mixed with isolation buffer were added to ll the tube. These gradients were centrifuged at 260,000 g, 4°C for 4.5 h, and each layer was carefully removed from the tube. Extracts were collected from each fraction, sample buffer was added and boiled for 5 min, and the extracts were analyzed by SDS-PAGE and immunoblotting.
Denatured protein extracts were subjected to SDS-PAGE and blotted onto PVDF membranes (Immobilon P; Millipore). Membranes were incubated in 8% milk for 0.5 h before an overnight incubation with the primary antibody at 4°C. The immunoblots were washed three times with TBST and then incubated with a secondary antibody in TBST for 1 h. After washing three times in TBST, the blots were visualized using

Statistical Analysis
The results are presented as the mean ± SD of a minimum of three independent sets of experiments. All statistical analyses were performed in Graph Prism 5.0. One-way analysis of variance (ANOVA) or Student's t test was used as indicated. P < 0.05 was considered statistically signi cant.

Results
Exogenous GM1 can partially restore GM1 content reduced by PDMP treatment To reduce the content of GM1 in the plasma membrane, SH-SY5Y cells were treated with 20 µM PDMP, an inhibitor of glucosylceramide synthesis, which can partially reduce the level of GM1. To examine the effect of PDMP on cell viability, cells were treated with various concentrations of PDMP in 96-well plates (0µM, 10µM, 20µM, 30µM, 40µM, 50µM, 60µM, 70µM, 80µM, 90µM, 100µM, 120µM, 160µM). We found that the cell proliferation decreased with increasing PDMP doses, and that PDMP had no signi cant effect on cell viability at 20 µM (Fig. 1A).
To con rm the inhibitory effect of PDMP on GM1 content, CtxB-Alexa 647 was used to label GM1. Treatment with 20 µM PDMP signi cantly reduced the quantity of GM1. Notably, treatment with 100µM exogenous GM1 restored the GM1 content to almost to the basal level ( Fig. 1B, C). This result is consistent with the report by Martino Calamai et al [26] .

GM1 de ciency inhibits GDNF-induced cell proliferation
To investigate the effect of GM1 de ciency on the neurotrophic effects of GDNF, CCK-8 assays and EdU staining assays were performed to measure the proliferation of SH-SY5Y cells induced by GDNF after PDMP treatment. Cells gradually proliferated upon GDNF treatment, and the proliferation was signi cantly enhanced after 36 hours of treatment. In contrast, starting at 24 hours of treatment, cells treated with PDMP+GDNF proliferated signi cantly less than those treated with GDNF alone, and the change was reversed after the addition of exogenous GM1 ( Fig. 2A). EdU staining results showed that after PDMP treatment, the proportion of EdU-positive cells induced by GDNF decreased signi cantly. The addition of 100 µM GM1 signi cantly restored the effect of GDNF on the proliferation of SH-SY5Y cells, even to a degree comparable to those treated with GDNF alone (Fig. 2B, C).
In conclusion, reducing the content of GM1 can inhibit the effect of GDNF on proliferation, and exogenous GM1 can reverse this change.

GM1 de ciency inhibits GDNF-induced cell differentiation
To examine the effects of GM1 on the pro-differentiation effect of GDNF, SH-SY5Y cells were treated under four conditions (Ctrl, GDNF, PDMP+GDNF and PDMP+GM1+GDNF), and microtubule-associated protein 2 (MAP2) immuno uorescent staining was performed. The axon length and the number of SH-SY5Y cells with axons were determined using ImageJ software. As shown in Fig. 3A, GDNF treatment increased the average axon length and the number of cells with axons, indicating the pro-differentiation effect of GDNF. Pretreatment with PDMP reduced the average axon length and the number of cells with axons induced by GDNF. Quantitative analysis revealed that exogenous GM1 increased the axon length at least by 0.1 inch compared to the PDMP treatment group (Fig. 3B). The number of cells with axons was also signi cantly increased by exogenous GM1 (Fig. 3C). In addition to the longer protrusions and increased number of cells with protrusions, the cells cultured with exogenous GM1 also maintained high tyrosine hydroxylase (TH) expression (Fig. 3D), which is speci cally expressed by dopaminergic cells. In conclusion, reducing the content of GM1 can inhibit the effect of GDNF on differentiation induction, and exogenous GM1 can reverse this change.

GM1 de ciency impairs GDNF-RET signaling
Stimulation with 50 ng/mL GDNF increased the phosphorylation of the primary GDNF signaling molecules RET, Erk and Akt. Pretreatment with PDMP reduced GDNF-RET signaling. Fig. 4 demonstrates that p-RET, p-Erk and p-Akt induced by GDNF were reduced to basal levels when GM1 was depleted. The crucial role of GM1 in these changes was further illustrated by the increase in the levels of p-RET, p-Erk and p-Akt when GM1 was exogenously supplied to the cells pretreated with PDMP.

GM1 de ciency disturbs the assembly of lipid rafts
To investigate the alteration of lipid raft assembly, we performed Triton X-100 solubilization and OptiPrep TM density gradient centrifugation to isolate lipid rafts and examined the alterations in lipid raft markers in each sample by Western blotting. Immunoblotting of cell extracts from Triton X-100 solubilization samples revealed that treatment with 20 µM PDMP largely dispersed caveolin-1 and otillin-1, the markers of lipid rafts, from the DRMs, also known as lipid rafts, to the detergent-soluble, non-raft membrane domains, and that this could be partially reversed by the addition of exogenous GM1. However, the total protein levels of CD71, a non-lipid raft marker, were not different between the GM1de cient cells and the GM1-supplemented cells (Fig. 5A, B).
Fractions from OptiPrep TM density gradient centrifugation were also analyzed by Western immunoblotting. Not surprisingly, the results were consistent with those described above (Fig. 5C). Caveolin-1 and otillin-1 in SH-SY5Y cells were mainly localized in lipid raft fractions, while CD71 was distributed in non-lipid rafts. After pretreatment with PDMP, caveolin-1 and otillin-1 tended to disperse from lipid raft to non-raft fractions, which could be partially reversed by the addition of GM1. At the same time, the localization of CD71 in non-lipid rafts was not changed by PDMP treatment. These results suggested that GM1 de ciency could result in changes in lipid raft compositions.

RET translocation into lipid rafts is blocked in GM1-reduced SH-SY5Y cells
Many studies have shown that lipid rafts play a vital role in the transmission of GDNF-RET signals. Upon GDNF treatment, RET translocates into lipid rafts and activates RAS/Erk, PI3K/Akt and various other signal transduction pathways. To investigate the effect of lipid raft alterations induced by GM1 de ciency on GDNF-RET signals, we isolated lipid rafts and determined the translocation of RET into lipid rafts. First, we performed Triton X-100 solubilization to isolate lipid rafts. The results showed that, without GDNF treatment, RET was primarily located in detergent-soluble, non-raft membrane domains, and very little RET was detected in lipid raft fractions. After 20 minutes of treatment with 50 ng/mL GDNF, RET in lipid raft membrane domains increased. Pretreatment with PDMP inhibited the translocation of RET into lipid rafts induced by GDNF. In addition, GDNF-induced RET distribution in lipid rafts was partially restored following the restoration of GM1 content in PDMP pretreated cells (Fig. 6A).
To con rm these results, we used another biochemical method, OptiPrep TM density gradient centrifugation, to isolate lipid rafts. Consistent with the observations from Triton X-100 solubilization experiments, the majority of RET was located outside of DRMs without GDNF treatment, and after GDNF treatment, RET was translocated into lipid rafts. When GM1 was reduced, GNDF-induced RET translocation into lipid rafts was inhibited, which could be restored by GM1 addition (Fig. 6B).
GM1 is decreased in MPP + -induced PD cell model MPP + treatment is frequently used to establish a cell model of PD [27,28] . In this study, SH-SY5Y cells were treated with 2.5 mM, 5 mM, or 7.5 mM MPP + for 24 h, and cell viability was determined by CCK-8 assay.
The results showed that at 5 mM, MPP + signi cantly reduced cell viability to approximately 70% of that of the control group (Fig. 7A). Therefore, we used 5 mM MPP + to treat SH-SY5Y cells. To investigate the effect of 5 mM MPP + on GM1 content, we labeled GM1 with CtxB-Alexa 647 and quanti ed it by ImageJ software. We found that GM1 was signi cantly decreased after MPP + treatment (Fig. 7B, C). This change could be partially reversed after adding exogenous GM1.

Synergistic protective effect of GM1 and GDNF in an MPP +induced PD cell model
To determine the synergistic protective effect of GM1 and GDNF on the MPP+-induced PD cell model, we rst selected the appropriate concentration of GDNF. MPP + -injured SH-SY5Y cells were treated with various concentrations of GDNF (10 ng/mL, 20 ng/mL, 30 ng/mL, 40 ng/mL, 50 ng/mL, 60 ng/mL) for 36 h. Subsequently, cell viability was determined by CCK-8 assay. The results showed that 10 ng/mL GDNF had no signi cant protective effect on MPP + -injured SH-SY5Y cells (Fig. 8A). We selected this concentration of GDNF to investigate the synergistic protective effect of GM1 and GDNF.
SH-SY5Y cells were treated with 20 µM PDMP with or without GM1 for 24 h followed by incubation in serum-free medium for 4 h. Subsequently, cells were stimulated with GDNF for 36 h. Cell viability was determined by CCK-8 assay and the expression of TH protein was evaluated by Western blot assay. The results showed that treatment with GM1 alone at 40 µM or 100 µM had no protective effect on injured cells, while 10 ng/mL GDNF together with 40 µM or 100 µM GM1 increased cell viability and the expression of TH protein in MPP + -injured SH-SY5Y cells (Fig. 8B, C).

Discussion
Gangliosides are highly expressed in the nervous system of vertebrates [29] . In PD, the level of GM1 in dopaminergic neurons of the occipital cortex and SNpc is signi cantly lower than that of the control group at the same age, suggesting that GM1 de ciency may be a risk factor for PD [7] . Lack of dopaminergic neurons in the striatum and the SNpc and aggregation of α-synuclein, which are typical characteristics of PD, were observed in mice with GM1 de ciency [30] . However, the speci c mechanism of ganglioside GM1 de ciency leading to neurodegenerative changes is still unclear. Dopaminergic neurons require speci c neurotrophic factors for proper differentiation and maintenance in vivo [31,32] . As the most important neurotrophic factor in the midbrain dopamine system, GDNF can promote the survival and morphologic differentiation of embryonic midbrain dopaminergic neurons and enhance their uptake of dopamine [33] . Recent studies have shown that GDNF signal transduction is blocked in GM1-de cient mice. In our study, PDMP treatment inhibited GDNF-RET signal transduction and GDNF-induced proliferation and differentiation of SH-SY5Y cells, while the addition of exogenous GM1 partially reversed these negative effects. These observations are consistent with those of Ohmi et al [34] .
Lipid rafts are highly dynamic microdomains within cell membranes enriched in cholesterol and sphingolipids [16,35] . It has been shown that, in a variety of neurodegenerative diseases, such as PD, Alzheimer's disease and familial amyotrophic lateral sclerosis, lipid rafts appear to have abnormal lipid composition and functional de ciencies in the brain, which can promote the pathogenesis and development of these diseases [36][37][38] . Our previous study also showed that the lipid and protein components of lipid rafts were signi cantly altered in MPP + -induced PD cell models [39] . Furthermore, studies of various unit membrane models have shown that GM1 can affect the phase separation of lipid membranes and the assembly of microdomains [22,24,40,41] . However, it is not clear whether GM1 can affect the signal transduction process of GDNF by affecting lipid raft assembly in SH-SY5Y cells.
Our results showed that the lipid raft markers caveolin-1 and otillin-1 were redistributed from lipid rafts to non-lipid raft regions when GM1 was reduced. In addition, we also found that GDNF induced less RET translocation into lipid rafts and decreased RET phosphorylation in GM1-de cient cells. After the addition of exogenous GM1, GDNF-induced RET translocation to lipid rafts can be partially restored. These results indicate that lipid raft dysfunction caused by ganglioside GM1 de ciency affects GDNF-RET signaling.
Furthermore, our study also found that 5 mM MPP + could signi cantly decrease GM1 content, and that the addition of exogenous GM1 reversed the reduction. Treatment with 100 µM GM1 or 10 ng/mL GDNF alone had no signi cant effect on cell viability or TH protein expression in the MPP + -induced cell model of PD, while treatment with 100 µM GM1 combined with 10 ng/mL GDNF signi cantly increased cell viability and TH protein expression, suggesting that GM1 and GDNF have a synergistic protective effect on MPP + -induced cell injury.
Our results suggest that ganglioside GM1 de ciency might compromise the neurotrophic effects of GDNF by altering the assembly of lipid raft membrane microdomains and promote the progression of PD.

Declarations Ethics approval
Not applicable for that section.

Consent to participate
Not applicable for that section.

Consent for publication
Not applicable for that section.
Availability of data and materials: SH-SY5Y cells were purchased from Cell Bank of Chinese Academy of Sciences(SCSP-5014).

Competing interests
The authors have no competing interests to declare that are relevant to the content of this article.  50μM, 60μM, 70μM, 80μM, 90μM, 100μM, 120μM, 160μM). After 24 h of treatment, cell viability was determined by a CCK-8 kit (* represents the comparison between the experimental group and the control group: *P < 0.5, ***P < 0.001). (B, C) SH-SY5Y cells were treated with 20 μM PDMP with or without 100 µM GM1 for 24 h followed by washing and staining with CtxB-Alexa 647 (red uorescence). GM1 content under indicated conditions was quanti ed by ImageJ software. Scale bar = 50μm (* represents the comparison between the experimental group and the control group: ***P < 0.001).   Western blot, and densitometric analysis was performed by ImageJ software (* represents the comparison between the experimental group and the control group: ***P < 0.001; GDNF group versus PDMP+GDNF group: P < 0.001; PDMP+GDNF group versus PDMP+GM1+GDNF group: P < 0.001).  GM1 de ciency inhibits the translocation of RET to lipid rafts. SH-SY5Y cells were treated with 20 μM PDMP with or without 100 µM GM1 for 24 h followed by incubation in serum-free medium for 4 h.
Subsequently, cells were stimulated with 50 ng/mL GDNF for 20 min. (A) Lipid rafts were isolated by Triton X-100 solubilization. The expression of RET was assessed by Western blot analysis, and the band intensities of the Western blots were quanti ed using ImageJ software (* represents the comparison between the experimental group and the control group: ***P < 0.001; GDNF group versus PDMP+GDNF group: P < 0.001; PDMP+GDNF group versus PDMP+GM1+GDNF group: P < 0.001). (B) Lipid rafts were isolated by OptiPrepTM density gradient centrifugation. Localization of RET in the density gradient was analyzed by Western blot analysis using fractions separated by OptiPrepTM density gradient centrifugation. (red uorescence). GM1 content was quanti ed by ImageJ software. Scale bar = 50μm (* represents the comparison between the experimental group and the control group: ***P < 0.001). Western blot assay were used to assess cell viability and the expression of TH protein, respectively. (* represents the comparison between the experimental group and the control group: ***P < 0.001).