Neurotrophic effects of GM1 ganglioside, NGF, and FGF2 on canine dorsal root ganglia neurons in vitro

Dogs share many chronic morbidities with humans and thus represent a powerful model for translational research. In comparison to rodents, the canine ganglioside metabolism more closely resembles the human one. Gangliosides are components of the cell plasma membrane playing a role in neuronal development, intercellular communication and cellular differentiation. The present in vitro study aimed to characterize structural and functional changes induced by GM1 ganglioside (GM1) in canine dorsal root ganglia (DRG) neurons and interactions of GM1 with nerve growth factor (NGF) and fibroblast growth factor (FGF2) using immunofluorescence for several cellular proteins including neurofilaments, synaptophysin, and cleaved caspase 3, transmission electron microscopy, and electrophysiology. GM1 supplementation resulted in increased neurite outgrowth and neuronal survival. This was also observed in DRG neurons challenged with hypoxia mimicking neurodegenerative conditions due to disruptions of energy homeostasis. Immunofluorescence indicated an impact of GM1 on neurofilament phosphorylation, axonal transport, and synaptogenesis. An increased number of multivesicular bodies in GM1 treated neurons suggested metabolic changes. Electrophysiological changes induced by GM1 indicated an increased neuronal excitability. Summarized, GM1 has neurotrophic and neuroprotective effects on canine DRG neurons and induces functional changes. However, further studies are needed to clarify the therapeutic value of gangliosides in neurodegenerative diseases.

certain growth factors 23 . Gangliosides are sialic acid-containing glycosphingolipids that can be found in different cellular membranes. In the CNS gangliosides account for up to 10% of the total lipid content in neurons 24 . They are involved in intercellular communication, cellular differentiation, neuronal development, and regeneration [25][26][27][28] due to their capacity to modulate Ca 2+ influx and their influence on the function of receptors for muscarinic acetylcholine, serotonin, glutamine, neurotransmitters, and neurotrophic factors 29 . Cellular gangliosides concentrate in lipid microdomains termed lipid rafts, which are signaling platforms rich in cholesterol and glycosphingolipids and contain amongst others high affinity tropomyosin-related kinase (Trk) and low affinity neurotrophin receptors (p75 NTR ) [30][31][32] . These receptors are activated by nerve growth factor (NGF) and other neurotrophic factors and have a strong impact on neuronal development, maintenance, and survival as well as memory formation and storage 33,34 . Anti-G M1 antibodies modulate the membrane-associated sphingomyelin metabolism by altering neutral sphingomyelinase activity thereby inhibiting NGF action on rat pheochromocytoma (PC12) cells 35,36 . G M1 and NGF protect primary cultured rat embryonic dorsal root ganglia (DRG) and spinal cord neurons from glutamate-induced excitotoxicity 37,38 . Both molecules may function by modulating Ca 2+ homeostasis, maintaining normal mitochondrial membrane potential or by promoting the mRNA expression of neuronal proteins such as growth associated protein 43 and neurofilaments (NFs) 38,39 . G M1 also protects rats against high altitude cerebral edema by suppressing oxidative stress and production of the pro-inflammatory cytokines IL-1β, IL-6 and TNF-α 40 . Nevertheless, another study using an ischemia/reperfusion model in rats demonstrated that G M1 neuroprotection seems to be dependent on p75 NTR 41 .
Exogenous gangliosides also directly interact with growth factors such as basic fibroblast growth factor (FGF2) and inhibit FGF receptor binding 42,43 . Moreover, G M1 can inhibit FGF2-mediated effects by acting on the same intracellular signaling molecules. For instance, FGF2 stimulates the activity of glycogen synthase kinase 3 (GSK3) in proliferating adult rat hippocampal progenitor cells, whereas G M1 prevents GSK3 activation in organotypic hippocampal slice cultures 44,45 . In contrast, cell-associated G M1 has been described to act as a functional co-receptor for FGF2 43 . Despite its name, FGF2 plays an essential role in neurogenesis, differentiation, axonal branching, and neuron survival in various types of brain and peripheral nerve lesions 46,47 . FGF2 also stimulates the proliferation of neuronal precursor cells isolated from postnatal murine DRGs 48 . The ability of adult rat ganglion cells to regrow axons in vitro can be influenced by FGF2 and G M1 49 . Neurotrophic effects of FGF2 seem to be also mediated by soluble mediators released from glial cells 50 . Summarized, NGF, FGF2 and gangliosides such as G M1 can exert neurotrophic effects, which might be exploited in the treatment of traumatic and degenerative diseases of the nervous system. In addition, the recent development of semisynthetic and potent G M1 compounds as well as solid lipid nanoparticles as drug delivery systems motivates the therapeutic use of G M1 51,52 . G M1 levels in rodent brains increase with age 53 , whereas an age-associated decrease in G M1 content has been demonstrated in the human brain 54 . Moreover, G M1 can be metabolized by an alternative asialo degradation pathway in mice but not in humans 55 and dogs 56 . In general, humans and dogs seem to have a similar G M1 metabolism [57][58][59] indicating that dogs represent an appropriate animal model to characterize synergistic neurotrophic effects of gangliosides in human neuronal cells. Nevertheless, a detailed in vitro study in canine neurons analyzing the influence of G M1 on neurite outgrowth, axonal transport systems, and neuronal functions including interactions with the neurotrophic growth factors NGF and FGF2 has not yet been performed. Consequently, the aim of the present study was to characterize neurotrophic effects of G M1 , NGF, and FGF2 with special emphasis on cytoskeletal protein expression and electrophysiological changes in canine DRG neurons as suitable translational approach 2,10,60 . In order to observe the effect of G M1 in a model mimicking neurodegenerative conditions due to disruptions of energy homeostasis occurring in pathological conditions such as traumatic spinal cord injury 61 , neurons were additionally cultured in a hypoxia glove-chamber (1% O 2 ) with and without G M1 supplementation.

Results
NGF and G M1 ganglioside induce the formation of neurites. Neurons grown without supplements (Sato's medium with 1% BSA) showed a mean number of 2.1 βIII tubulin-positive processes at 2 days post seeding (dps). In order to study dose-dependent effects of G M1 on neurite outgrowth, neurons were incubated with increasing ganglioside concentrations. The mean number of processes per neuron significantly increased at a G M1 ganglioside concentration of 50 µM (4.7), 80 µM (5.2), and 100 µM (3.4) (Fig. 1). A ganglioside concentration of 80 µM was used for all further sets of in vitro experiments.
In order to analyze further growth promoting factors, the neurotrophic properties of NGF and FGF2 alone or in combination with G M1 ganglioside were analyzed ( Fig. 2; Suppl. Fig. 1). Culture of DRG neurons without NGF resulted in a significantly decreased mean number of βIII tubulin-positive processes per neuron (2.1 processes per neuron) when compared to neurons cultured with NGF (3.8 processes per neuron). Supplementation of neurons with G M1 ganglioside lead to a significant increase in the mean number of processes in cultures both with and without NGF supplementation (5.5 and 3.9 processes per neuron, respectively). Furthermore, neurons supplemented with both NGF and G M1 possessed significantly more processes than neurons cultured with FGF2 alone (3.1) or a combination of FGF2 and G M1 (2.9) In contrast, G M2 and G M3 supplementation did not affect neurite outgrowth and the glucosylceramide synthase inhibitor D-PDMP blocked neurite outgrowth (Suppl. Figs. 2 and 3). Microtubules formed by tubular polymers of tubulins represent an important part of the cytoskeleton, which is known to be affected by neurotrophin signaling pathways [62][63][64] . Consequently, the observed effects of G M1 ganglioside, NGF, and FGF2 on DRG neurons instigated the analysis of their impact on additional components of the neuronal cytoskeleton focusing on microtubular proteins and NFs. NGF and FGF2, respectively. Significantly more Tau1-positive cells were also found in NGF/G M1 ganglioside (87%) and FGF2/G M1 ganglioside (91%) containing media but G M1 supplementation alone (73%) did not have a significant influence on neuronal Tau1 expression ( Fig. 2; Suppl. Figure 1). These results demonstrated that NGF and FGF2 but not G M1 ganglioside supplementation affect Tau1 expression. Consequently, we also investigated the effect of these neurotrophic factors on MAP2 expression and NF phosphorylation status using antibodies directed against non-phosphorylated NF (nNF) and phosphorylated NF (pNF).
NGF and G M1 ganglioside stimulate the formation of synaptophysin accumulations in neuronal processes and disturb axonal transport mechanisms. Accumulations of synaptophysin, dynein, and kinesin in neuronal processes were observed in all culture conditions (Figs. 3, 4; Suppl. Fig. 4). 34% of neurons cultured without supplements demonstrated synaptophysin accumulations in their neurites. There was a significant increase in this percentage after addition of NGF (54%), G M1 (71%), FGF2/G M1 (75%), and NGF/G M1 Figure 3. Synaptophysin, dynein, and kinesin expression in canine dorsal root ganglia neurons. The neurons were grown without growth factors or G M1 ganglioside (no supplements) or supplemented with fibroblast growth factor 2 (FGF), nerve growth factor (NGF), G M1 , FGF and GM 1 , or NGF and G M1 and stained 2 days post seeding. Shown are representative pictures of neurons cultured without supplements and neurons supplemented with NGF and G M1 . Note accumulations of synaptophysin, dynein, and kinesin most likely within neuronal processes (arrows). Bars = 100 µm. The graphs show the percentages of neurons with processes containing respective accumulations for each condition (single values of 3 dogs with means). *Statistically significant differences (P < 0.05) compared to neurons cultured without supplements. # Statistically significant differences (P < 0.05) of FGF/G M1 and NGF/G M1 compared to FGF and NGF, respectively. + Statistically significant differences (P < 0.05) of FGF/G M1 and NGF/G M1 compared to G M1 only. No suppl. = no supplements.
(82%). FGF2 alone (43%) did not change the percentage of synaptophysin accumulations in neuronal processes ( Fig. 3; Suppl. Fig. 4). Moreover, dynein and kinesin accumulations were observed in neurites of 72% and 59% of neurons cultured without supplements, respectively (Fig. 4). G M1 ganglioside and growth factor supplementation did not have a significant impact on the presence of accumulations of dynein or kinesin in neurites compared to untreated conditions ( Fig. 3; Suppl. Fig. 4). Nevertheless, there was a significant increase in the percentage of neurons with dynein accumulations in their processes after G M1 supplementation (G M1 : 80%; FGF2/G M1 : 80%; NGF/G M1 : 80%) compared to the addition of FGF2 alone (67%). Moreover, the highest percentage of neurons with kinesin accumulations (78%) was observed after NGF/G M1 supplementation. These results indicate that a combination of NGF and G M1 ganglioside affects the distribution of the synaptic vesicle protein synaptophysin as well as anterograde and retrograde axonal transport mechanisms. These transport mechanisms are also critically involved in trafficking of neurotrophin receptor such as p75 NTR 67 , which prompted the analysis of p75 NTR expression. G M1 , NGF, and FGF2 synergistically induce the internalization and/or down-regulation of p75 ntR . The p75 NTR expression was significantly higher in neurons supplemented with FGF2/G M1 (46%) or NGF/G M1 (49%) compared to neurons cultured without supplements (72%), whereas addition of FGF2 (59%), NGF (62%), or G M1 ganglioside (65%) alone had no significant effect ( Fig. 5; Suppl. Fig. 5). Consequently, only the combination of G M1 ganglioside and growth factors affected the expression of p75 NTR , which binds all neurotrophins including NGF.
Depending on its interactions with co-receptors, p75 NTR can induce apoptosis in DRG neurons or generate survival signals [68][69][70][71] . The impact of G M1 ganglioside and growth factors on cell survival was investigated by studying cleaved caspase 3 expression, which plays a central role in the execution phase of apoptosis. G M1 , NGF and FGF2 favor neuronal survival. A significant decrease in the number of neurons expressing cleaved caspase 3 was present in neurons supplemented with G M1 , NGF and FGF2 (FGF2: 13%; NGF: 15%; G M1 : 12%; FGF2/G M1 : 11%; NGF/G M1 : 10%) compared to neurons cultured without supplements (23%) underlining their neurotrophic activity and neuroprotective function ( Fig. 6; Suppl. Fig. 6). In contrast, G M2 and G M3 supplementation had no significant effect on the survival of canine DRG neurons (Suppl. Figs. 2 and 3). Depletion of gangliosides by D-PDMP induced neuronal cell death as detected by a Trypan blue dye exclusion assay (Suppl. Fig. 3).
Exogenously administered G M1 accumulates in non-raft membrane fractions of neuroblastoma cells. Lipid raft isolation and analysis of primary DRG neurons did not reveal concentrations of G M1 above the limit of detection due to limited numbers of isolated cells (data not shown). Therefore, N1E-115 murine neuroblastoma cells were used to determine the cellular localization of exogenously administered G M1 . Separation of glycolipids on a thin layer chromatography (TLC) plate revealed the presence of G M1 in non-raft (NR) membrane fractions in G M1 treated cells (Suppl. Fig. 7). Confirmation of the presence of G M1 in the NR fraction was determined through the addition of 500 ng G M1 to the same sample. A clear increase in the intensity of the band at the same height of the standard was seen when compared to the untreated NR G M1 treated sample. The difference . Immunofluorescence double-labeling of adult canine dorsal root ganglia neurons. Neurons were supplemented with NGF and G M1 and stained 2 days post seeding for neuronal class III β tubulin (green) and synaptophysin (red) or kinesin (green) and dynein (red). Note synaptophysin and dynein accumulations (arrowheads) as well as kinesin accumulations (arrows) in neuronal processes. Bars, 50 μm. Figure 5. Expression of the p75 neurotrophin receptor (p75 NTR ) in canine dorsal root ganglia neurons. The neurons were grown without growth factors or G M1 ganglioside (no supplements) or supplemented with fibroblast growth factor 2 (FGF), nerve growth factor (NGF), G M1 , FGF and GM 1 , or NGF and G M1 and stained for neuronal class III β tubulin (green) and p75 NTR (red) 2 days post seeding. Shown are representative pictures of neurons cultured without supplements and neurons supplemented with FGF and G M1 as well as NGF and G M1 . Note p75 NTR positive neurons (arrows) and p75 NTR negative neurons (arrowheads). Bars = 100 µm. The graph shows the percentage of p75 NTR positive neurons for each condition (single values of 3 dogs with means). *Statistically significant differences (P < 0.05) compared to neurons cultured without supplements. + Statistically significant differences (P < 0.05) of FGF/G M1 and NGF/G M1 compared to G M1 only. No suppl. = no supplements. G M1 supplementation of neurons leads to an increased density of cytoplasmic multivesicular bodies. Neurons supplemented with 80 µM G M1 ganglioside displayed a higher density of cytoplasmic multivesicular bodies compared to neurons cultured without supplements (0 µM G M1 : 0.07/µm 2 ; 80 µM G M1 : 0.28/ µm 2 ; P < 0.0001; Fig. 7). Nodular enlargements of neurites containing accumulations of mitochondria, small vesicular structures, and/or neurofilaments were found in neurons with and without G M1 ganglioside supplementation (0 µM G M1 : 0.176/µm; 80 µM G M1 : 0.155/µm; Fig. 7). The percentage of nodular enlargements containing mitochondria increased from 24% (5 nodular enlargements with mitochondria/21 total number of nodular enlargements) in neurites of neurons cultured without supplements to 35% (13/37) in neurites of neurons supplemented with 80 µM G M1 . However, this increase was not statistically significant (P = 0.5557). G M1 supplementation leads to an elevated resting potential, a reduced action potential current threshold, and a slowing down of depolarization speed. To probe whether ganglioside supplementation affects the functional properties of cultured canine large DRG neurons, the membrane properties at rest were quantified and the action potential generation was analyzed. From three dogs 18 neurons cultured in Sato's medium with NGF only and 17 neurons additionally treated with 80 µM G M1 were recorded by analyzing one well-plate without and one with G M1 treatment from each dog. Resting membrane properties, such as resting potential, input resistance, membrane time constant and effective cell capacity were extracted from a small 120 ms long hyperpolarization 68,69 induced by a −25 pA current injection (Fig. 8A). The resting potential was determined before the hyperpolarization and was significantly different (P = 0.008) between the non-treated (−58.3 ± 1.2 mV) and G M1 -treated (−53.4 ± 1.2 mV) DRG neurons (Fig. 8B). The membrane time constant, measured from a mono-exponential fit to the onset of the hyperpolarization (Fig. 8A) remained unaffected by the treatment (non-treated: 22.7 ± 0.3 ms; G M1 -treated: 19.5 ± 0.3 ms; P = 0.514; Fig. 8C). The input resistance was calculated according to Ohm's law (Fig. 7A,D). In current and in voltage clamp the input resistance was unchanged between non-treated (current clamp: 119.7 ± 22.6 MOhm; voltage clamp: 92.9 ± 16.2 MOhm) and G M1 -treated cells (current clamp: 134.5 ± 19.3 MOhm; voltage clamp: 119.9 ± 15.6 MOhm; current clamp P = 0.624; voltage clamp P = 0.24; Fig. 8E). The effective capacitance, the membrane surface that can be charged from the soma was on average 471.2 ± 80.8 pF for non-treated cells and 324.9 ± 34.2 pF for G M1 -treated cells (P = 0.112; Fig. 8F).

G M1 supplementation results in increased neurite outgrowth and viability in neurons cultured
under hypoxic conditions. When neurons were cultured under hypoxia (1% O 2 ) for 6 days directly after seeding in Sato's medium containing NGF, the viability as determined by Trypan blue dye exclusion assay was significantly reduced (7.9%) when compared to neurons cultured under normoxia (33.6%). Treatment with 80 µM G M1 resulted in a significant increase in viability in neurons both cultured under hypoxia and normoxia (hypoxia G M1 -treated: 20.0%; normoxia G M1 -treated: 46.1%) when compared to neurons cultured in Sato's medium containing NGF only (control). Cultures supplemented with G M1 under hypoxia showed a significantly lower viability than G M1 -treated neurons cultured under normoxia (Fig. 9). There were no significant differences in cleaved caspase 3 expression between the groups of this experiment (hypoxia control: 9.67%; hypoxia G M1 : 8.40%; normoxia control: 16.02%; normoxia G M1 : 4.28%; Suppl. Fig. 8).
As neurons cultured under hypoxia for 6 days directly after seeding showed an almost complete lack of neurite-outgrowth, it was not possible to determine the impact of G M1 treatment upon neurite-outgrowth under hypoxic conditions. Therefore, a second set of experiments was performed. Neurons were first cultured under normoxia for 6 days and afterwards transferred to the hypoxia glove-chamber for 48 hours. G M1 treatment during the hypoxia-period of 48 hours resulted in a significant increase in mean neurite-number per neuron (2.7 neurites/neuron) as compared to control neurons cultured in Sato's medium containing NGF only (1.4 neurites per neuron; Fig. 10). When culturing neurons for 6 days under normoxia followed by 48 hours of hypoxia, there was no significant impact of G M1 treatment upon the viability as determined by Trypan blue dye exclusion assay (control: 55,47%; G M1 : 53.06%; Suppl. Fig. 9). Cleaved caspase 3 expression was slightly reduced in hypoxic neurons treated with G M1 (35.95%) compared to control neurons (39.77%), albeit it did not reach significance (Suppl. Fig. 10).

Discussion
There is accumulating evidence of a beneficial effect of G M1 application upon different pathological conditions of the nervous system, like spinal cord injury, high altitude cerebral edema, and neurodegenerative diseases 19,40,72 . Furthermore, age-dependent alterations in the amount and distribution of gangliosides in the CNS seem to be involved in the pathogenesis of neurodegenerative diseases due to their impact on neuronal maintenance, stability, and regeneration 28,53,[73][74][75] . However, effects of ganglioside application on neuronal protein expression and especially functional activities have not been investigated in a species with a ganglioside metabolism similar to humans so far. The present study revealed that G M1 , but not G M2 and G M3 ganglioside has neurotrophic effects on canine DRG neurons such as inducing neurite outgrowth and suppressing apoptosis. Interestingly, even though neurodegenerative changes induced by NGF withdrawal in rodent neurons are mostly reported to affect embryonal neurons [76][77][78][79] , the culture of DRG neurons from adult dogs without NGF supplementation in the present study lead to a significantly reduced number of neurites, an increased apoptotic rate and decreased synaptophysin-accumulations. Therefore, in contrast to previous studies focusing on rodent neurons, it seems that NGF withdrawal continues to have a negative impact on canine DRG neurons cultured in vitro during later stages of development. This negative effect of NGF withdrawal was more than compensated by the supplementation with G M1 , while the combined treatment with both NGF and G M1 appeared to be even more beneficial.   www.nature.com/scientificreports www.nature.com/scientificreports/ Moreover, G M1 ganglioside supplementation affected the neuronal cytoskeleton including the phosphorylation status of NFs, synaptophysin expression, axonal transport mechanisms, cell metabolism, and biophysical properties. A previous study described that gangliosides increased the expression of NF proteins in embryonic chicken DRG neurons 80 . G M1 also potentiated NGF activity on neurite outgrowth and NF expression in rat PC12 cells 81 . In addition, NGF induced an increase in neuronal processes in adult rat DRG neurons in vitro and promoted neurite length and arborization in bovine DRG neurons 82,83 . Likewise, G M1 triggered an increased neurite outgrowth in canine DRG neurons characterized by higher numbers of neuronal processes positive for βIII tubulin and NFs. G M1 also enhanced MAP2 expression in these processes, which might be related to neurite initiation 84 . No effect of G M1 was found on Tau1 expression, whereas NGF and FGF2 induced its expression in canine DRG neurons. FGF2 also induced Tau1 expression in proliferating adult rat hippocampal progenitor cells 45 . Anti-apoptotic effects of G M1 , NGF, and also FGF2 on canine DRG neurons were substantiated by a decreased neuronal cleaved caspase 3 expression. Similarly, pro-survival effects were also described for NGF 85 and G M1 86 in rodents, whereas apoptosis was induced by FGF2 in murine DRG neurons after sciatic nerve injury 87 . G M1 also counteracted lead-induced apoptosis by decreasing the expression of Bax and cleaved caspase 3 and by increasing the level of Bcl-2 in the developing rat hippocampus 88 .
The present lipid analysis showed an increase in G M1 in the non-raft (NR) fraction in G M1 treated cells compared to the NR fraction of non-treated cells. Similarly, exogenous G M1 appears to be predominately partitioned into NR fractions in rat cerebellar granule cells 89 , whereas G M1 has been shown to primarily be localized within plasma membrane lipid rafts under normal conditions 90 . This could be the result of normal cellular trafficking pathways being bypassed leading to NR localization. However, this localization does not seem to prevent the influence of G M1 on neurotrophin interactions with their receptors. A current study indicated that despite belonging to separate membrane domains TrkA might interact with G M1 by laying down its extracellular portion onto the membrane, thereby approaching the oligosaccharide portion of G M1 91 .
Recent studies investigated the molecular basis of the interactions between G M1 and neurotrophins including NGF and brain-derived neurotrophic factor (BDNF) 91,92 . G M1 acts as a bridge able to increase and stabilize the interactions of NGF with its high affinity receptor TrkA, which promotes neurite formation in murine neuroblastoma (Neuro2a) cells 92 . G M1 also induces the release of BDNF from hippocampal neurons 93 and the amount of G M1 in the environment of the BDNF receptor TrkB can modulate its activity 94 . Increased interactions between neurotrophins and their receptors in response to G M1 supplementation likely stimulate several intracellular signaling pathways explaining the observed neurotrophic effects. Trk receptors activate the phosphatidylinositol-3 kinase (PI3K)-protein kinase B (AKT), RAS-mitogen-activated protein kinase (MAPK) and phospholipase C (PLC)-γ-protein kinase C (PKC) pathways 64,95 . The PI3K-AKT pathway has antiapoptotic activity and controls dendritic arborization together with the RAS-MAPK pathway [96][97][98] . The RAS-MAPK signaling cascade promotes neuronal differentiation including neurite outgrowth 64 . Extracellular signal-regulated kinase (ERK) and p38 are two MAPKs, which are activated by NGF to stimulate the phosphorylation of the cAMP response element-binding protein (CREB) and influence gene transcription 99,100 . For instance, BDNF regulates dendritic branching in a CREB-dependent manner by increasing the expression of the guanine deaminase cypin, which directly binds to tubulin heterodimers and promotes microtubule polymerization and formation of proximal dendrites 62,101 . Complexes of BDNF and TrkB also activate small GTPases of the Rho family including Cdc42/ Rac/RhoA to stimulate actin and microtubule synthesis and growth of neuronal fibers 62 . Furthermore, activation of PLC-γ by BDNF triggers Ca 2+ -and PKC-regulated pathways that promote synaptic plasticity [62][63][64]102 . BDNF not only modulates the number of synapses by regulating the morphology of the axonal tree but also promotes synapse formation and maintenance in vivo 69,103 . Moreover, TrkB signaling mediates survival of hippocampal and motor neurons after axotomy 104 . Neurotrophins including their precursor forms such as pro-BDNF can also bind to p75 NTR , which can modulate and even counteract neurotrophic effects mediated by Trk receptors 62 . This low affinity neurotrophin receptor acts on RhoA, JNK/c-jun and NF-κB signaling pathways, which promote neuronal growth cone development, apoptosis and, neuronal survival, respectively 63,64 . The PI3K-AKT, RAS-MAPK and PLC-γ-PKC pathways are also activated by FGF2, which can bind to four different signaling tyrosine kinase FGF receptors mediating FGF family functions in embryogenesis and organogenesis as well as in metabolism, tissue repair, and regeneration 105 . In the CNS, FGF signaling is involved in neuronal differentiation, migration and excitability, synaptogenesis, myelination and learning and memory formation 105 .
In addition to the described neurotrophic effects, supplementation of canine DRG neurons with G M1 and NGF amplified accumulations of synaptophysin in their processes. This effect is possibly related to an increase in endocytosis and/or retrograde transport of synaptic vesicles or anterograde transport of synaptophysin molecules. The transfer of synaptophysin from the soma to neurites can be induced by NGF in cultured newborn rat trigeminal ganglion neurons, which promotes activity-dependent maturation of synaptic connections 106 . The present synaptophysin accumulations might also be linked to an increased synapse formation, which could be further enhanced by the increased number of neuronal processes resulting in an enlarged contact area of the DRG neurons. Similarly, previous studies demonstrated that gangliosides have an impact on the synaptic plasticity and function of rat hippocampal neurons and thereby influence long term potentiation and learning ability 107 . Formation and maintenance of synapses are dependent on various retrograde messengers including neurotrophins 108 . Consequently, an increased binding, internalization, and retrograde transport of NGF receptors in signaling endosomes might promote synaptogenesis [109][110][111] . Similarly, the stimulation of canine DRG neurons with NGF and FGF2 most likely results in an internalization of p75 NTR proteins due to ligand binding and thereby in a decrease of p75 NTR expression on the cell surface. In addition, G M1 augmented the formation of dynein accumulations in canine DRG neurons indicating changes in axonal transport such as increased retrograde transport. Likewise, the transport rate of tubulin and actin in sensory fibers of adult Sprague-Dawley rats can be stimulated by ganglioside treatment, which could be caused by molecular interactions between cytoskeletal elements and integral membrane glycolipids 112 . (2020) 10:5380 | https://doi.org/10.1038/s41598-020-61852-z www.nature.com/scientificreports www.nature.com/scientificreports/ Interestingly, one study described accumulations of synaptophysin as a reliable immunohistochemical marker for axonal damage in CNS lesions 113 . Moreover, the present study demonstrated that G M1 and NGF supplementation results in an increased expression of pNF and nNF within neuronal processes, which was accompanied by a decrease in the pNF/nNF ratio possibly reflecting changes in the NF phosphorylation status. Normal axons are characterized by highly phosphorylated NFs, whereas damaged axons found in different inflammatory, degenerative and traumatic CNS diseases [114][115][116] and disturbed axonal transport 117,118 contain nNFs. Thus, supplementation of canine DRG neurons with G M1 and NGF might also cause degenerative changes in neuronal processes. To allow a more in-depth analysis of this effect, an ultrastructural analysis was performed revealing that G M1 induced the formation of cytoplasmic multivesicular bodies. These intracellular endosomal organelles are characterized by a single outer membrane containing a varying number of internal vesicles. They were initially described as prelysosomal structures but actually participate in various endocytic and trafficking functions, including protein sorting, recycling, transport, storage, and release, and represent a general indicator of neuronal stress 119 . The increased number of multivesicular bodies found in neurons from aged animals seems to be associated with age-related neurodegenerative changes 119 . Nevertheless, several studies also demonstrated that multivesicular bodies contain signaling molecules and their receptors allowing their recruitment for membrane insertion, which might affect synaptic plasticity [120][121][122] . Interestingly, multivesicular bodies located within the neuronal soma and dendrites accumulate NGF, whereas endosomes mediate the retrograde axonal transport of growth factors such as NGF 123 . Thus the G M1 -dependent formation of multivesicular bodies in the soma of the present canine DRG neurons might represent a sign of neuronal stress and degenerative changes and/or result from an enhanced trafficking of membrane receptors such as p75 NTR 124,125 .
To clarify the impact of the described morphological changes on functional properties of canine DRG neurons, an electrophysiological analysis was performed. G M1 supplementation induced an elevated resting potential, a reduced action potential current threshold, and a slowing down of the depolarization speed. The elevated membrane potential might have partially caused the reduction of the current threshold and possibly the speed of depolarization. The elevation of the membrane potential approximates the action potential threshold and thus less current might be needed to stimulate the neuron. Moreover, the depolarization entails an increased inactivation of sodium channels in DRG neurons 126,127 . This reduced availability of sodium channels might result in a slowing down of the action potential during the depolarization phase. However, considering the age-dependent decrease in G M1 content in the human brain the effects of G M1 supplementation on neuronal electrical excitability might be exploited to reduce impaired functional capacities of aged brains.
In summary, the present study showed neurotrophic and neuroprotective effects of G M1 gangliosides on DRG neurons isolated from adult dogs, which were characterized by increased neuronal survival and neurite outgrowth and possibly even enhanced synaptic density. Supplementation of G M1 was even able to more than compensate negative effects on neurite outgrowth, apoptosis and synaptophysin accumulations that were observed when culturing DRG neurons without NGF. Beneficial effects were most pronounced when DRG neurons were treated with both NGF and G M1 . Additionally, G M1 supplementation lead to a significant increase in viability and neurite outgrowth of canine DRG neurons cultured under hypoxic conditions, which occur during spinal cord lesions caused by disc herniation, spinal stenosis, tumors, and spine trauma 61 . G M1 also increased the formation of multivescicular bodies indicating changes in cell metabolism. Moreover, G M1 might stimulate axonal transport mechanisms and increase the electrical excitability of these sensory neurons. These findings motivate further investigations upon the efficacy of ganglioside treatment in canine models of different traumatic, age-dependent and degenerative diseases of the nervous system.

Materials and Methods
Tissues used. DRGs of cervical, thoracic, and lumbar spinal cord segments from 19 healthy Beagle dogs (16 weeks to 2 years old) were used for cell culture experiments, transmission electron microscopy (TEM), and electrophysiology. The dogs were euthanized in the context of other studies, conducted in compliance with the law of animal welfare of Lower-Saxony and North Rhine-Westphalia, Germany and approved by Lower Saxony State Office for Consumer Protection and Food Safety, permission numbers: 33.9-42502-05-14A443, and 33.19-42502-05-16A044; and Ministry for Climate Protection, Environment, Agriculture, Conservation and Consumer Protection of the State of North Rhine-Westphalia, permission numbers: 501/A79, and T100525-3. In addition, cell culture experiments were also performed with DRGs of morphologically unchanged cervical, thoracic, and lumbar spinal cord segments obtained from one euthanized female adult dog (7 years old) during a routine necropsy at the Department of Pathology of the University of Veterinary Medicine Hannover. For the experiments in the hypoxia chamber, cryopreserved canine DRG neurons were used as previously described 14 due to the lack of fresh tissue samples.
The first two sets of experiments were performed in triplicates, whereas quadruplicates were used in the third experiment. All experiments were analyzed using immunofluorescence as described 129 . Antibodies and dilutions used are shown in Table 1. The number of neuronal class III β-tubulin (βIII tubulin) or NF-positive processes per neuron were counted. The number of neurons with and without immunoreaction for Tau1, p75 NTR , and cleaved caspase 3 was determined. Immunostainings for microtubule-associated protein (MAP) 2, nNF, and pNF were evaluated by counting the number of neurons with and without immunopositive processes to characterize the effects of NGF, FGF2, and G M1 ganglioside on neuronal cytoskeleton. In addition, due to the detection of accumulations of synaptophysin and of the axonal transport proteins dynein and kinesin in neuronal processes the number of neurons with and without respective accumulations in their processes was determined. Moreover, a Trypan blue dye exclusion assay (C.I. 23850; VWR International GmbH, Hannover, Germany) was used to investigate neuronal death in the third set of experiments. Finally, respective percentages were calculated and statistically evaluated.
For lipid analysis, N1E-115 murine neuroblastoma cells (ATCC, CRL-2263) were differentiated using 1% DMSO for two days and cultured in either DMEM medium with 10% FCS, 1% penicillin-streptomycin and 1% sodium-pyruvate (S11-003; PAA Laboratories GmbH, Cölbe, Germany) containing 80 µM G M1 or without for an additional two days 130 . Thereafter, lipid rafts were isolated and analyzed by TLC using a previously published protocol with minor amendment 89,131 . Briefly, cells were washed in PBS, lysed in 1 ml 1% Triton X-100 in PBS using a 21 G syringe and rotated for 2 hours at 4 °C. Thereafter, cell debris was removed by centrifugation at 17,000 × g for 20 min at 4 °C, followed by a second centrifugation at 100,000 × g for 90 min at 4 °C. Lipids were isolated using a chloroform, methanol and water based method and separated by using TLC. The lipids were visualized using orcinol monohydrate staining (O1875-10G; Sigma-Aldrich) and compared to a standard run on the same TLC plate (TLC Silica gel 60 F254 plates, 1.05729.0001; Merck, Darmstadt, Germany). The limit of quantification was determined using three times the y-intercept and given as 230 ng and the limit of detection at 50 ng. www.nature.com/scientificreports www.nature.com/scientificreports/ Transmission electron microscopy. Light microscopical changes characterized by immunofluorescence were most prominent in DRG neurons supplemented with 30 ng/ml NGF and 80 µM G M1 ganglioside. To characterize the changes induced by G M1 ganglioside at the ultrastructural level, DRG neurons supplemented with 30 ng/ml NGF in combination or without G M1 ganglioside (80 µM) were compared using TEM. For this purpose, 2240 neurons/well were seeded on 6 well plates, fixed for 24 h in 2.5% glutaraldehyde solution, rinsed with 1% sodium cacodylate buffer (pH 7.2), post-fixed in 1% osmium tetroxide, and embedded in EPON 812 (Serva, Heidelberg, Germany). Sections were stained with lead citrate and uranyl acetate and investigated using an EM 10 C (Carl Zeiss Jena GmbH, Oberkochen, Germany) and a slow-scan 2K-CCD camera (TRS Tröndle, Moorenweis, Germany) 132 .
The number of multivesicular bodies in the soma of all neurons present in the investigated sections was determined (0 µM G M1 : n = 16 neurons; 80 µM G M1 : n = 18 neurons). In addition, the length of the neurites and the number of nodular enlargements with and without mitochondria were counted in all neurites (0 µM G M1 : n = 17 neurites; 80 µM G M1 : n = 27 neurites), that were present in the investigated sections. This was performed by using photos and analysis software (analySIS 3.1 software package; Soft Imaging system, Münster, Germany).

Electrophysiology.
To complete the analysis of G M1 -induced effects on canine DRG neurons, functional changes of neurons supplemented with 30 ng/ml NGF in combination (G M1 -treated) or without (non-treated) G M1 ganglioside (80 µM) were investigated using electrophysiology. Cultured DRG cells were placed on their coverslip under an upright BX51 WI Olympus microscope and continuously perfused with extracellular recording solution containing (in mM) NaCl 125, NaHCO 3 25, NaH 2 PO 4 1.25, KCl 2.5, D-Glucose 25, L-Ascorbic acid 0.4, Myo-Inositol 3, Na-pyruvate 2, MgCl 2 1, CaCl 2 2 at a pH 7.4 and was oxygenated with 95% O 2 and 5% CO 2 . Electrophysiological recordings were carried out between 26-28 °C with an EPC 10/2 amplifier (HEKA, Lambrecht/Pfalz, Germany). Stimulus generation and presentation was controlled by the PatchMaster software. Cells were visualized with CCD-cameras (TILL-Imago VGA, Retiga 2000DC) controlled by TILLvisION imaging system (FEI Munich GmbH, Munich, Germany). In general, large DRG neurons were selected. Recordings were performed in whole-cell configuration using an intracellular solution containing (in mM) K-gluconate 145, KCl 4.5, HEPES 15, Mg-ATP 2, K-ATP 2, Na 2 -GTP 0.3, Na2-phosphocreatine 7, K-EGTA 0.5, Alexa488 0.05. Data were acquired with 20 kHz, and filtered by 3 Hz. Access resistance was compensated in voltage clamp mode before switching into current clamp, where bridge balance was set to 100%. For determining the input resistances and the cell capacitance by voltage clamp recordings all filters and the clamp were removed. Data was not corrected for the liquid junction potential of ~15 mV.
Cells were challenged with a −25 pA current injection of 120 ms length with 50 repetitions. The average voltage response to this hyperpolarization was used to determine membrane decay time constant by a mono-exponential fit. Using Ohm's law the input resistance during steady state was calculated. To obtain a second estimate of the cells input resistance and to calculate the cells effective capacitance the average current of 20 repetitions of a 150 ms long −10 mV hyperpolarization was recorded in voltage clamp. The input resistance was again estimated following the Ohm's law from the steady state current. The cell capacitance was determined from the area under the current transients for a time frame of three times the decay time constant. To probe for action potentials properties a 1 ms square current injection was applied and incremented by 100 pA. The first supra-threshold response to this current injection was used to analyze the action potential properties. Data analysis was carried out in IgorPro6 (Wavemetrics).
Hypoxia chamber. In a first experiment, cryopreserved neurons were thawed according to a previously published protocol 14 and seeded in two 96 Half Area Well Microplates at a density of 150 neurons per well in Sato's medium containing 30 ng/ml NGF. Both titer plates contained one group of neurons treated with 80 µM G M1 and one group without G M1 supplementation. One of the two titer plates was directly transferred into a hypoxia glove-chamber (Coy Laboratory Products, Grass Lake, MI, USA), where it was cultured with 1% O 2 (7 mm Hg, 5% CO 2 ) for 6 days. The other titer plate was cultured under normoxia (21% O 2 ; 5% CO 2 ) for 6 days. The medium was changed on day 4 in both titer plates. After 6 days, immunofluorescence staining for cleaved caspase 3 and pan-neurofilament and a Trypan blue dye exclusion assay were performed.
Since cultivating canine DRG neurons under hypoxia for 6 days directly after seeding resulted in an almost complete absence of neurite-outgrowth, a second experiment was performed in order to evaluate the influence of G M1 treatment upon neurite outgrowth under hypoxic conditions. Cryopreserved neurons were again seeded in a 96 Half Area Well Microplate at a density of 150 neurons per well. Neurons were cultured under normoxia (21% O 2 ; 5% CO 2 ) for 6 days in Sato's medium supplemented with 30 ng/ml NGF and a medium-change was performed on day 4. After 6 days, the neurons were transferred to the hypoxia glove-chamber (1% O 2 , 7 mm Hg, 5% CO 2 ) to remain there for 48 hours according to a previously published protocol 133 . For the 48 hours under hypoxic conditions, the medium was changed again and one of the two groups of neurons on the titer plate was treated with 80 µM G M1 , while the other group remained in Sato's medium supplemented with NGF (control). Media used for this medium-change were pre-equilibrated in the hypoxia glove-chamber for 5 hours prior to treatment of the cells. On day 8 immunofluorescence staining for cleaved caspase 3 and pan-neurofilament and a Trypan blue dye exclusion assay were performed. All experiments were performed in quadruplicates and evaluated as described for the other cell culture experiments. Statistical analysis. Statistical analysis was performed using GraphPad software (Prism 6; GraphPad Software, Inc., La Jolla, CA, USA). Immunofluorescence data were evaluated using a one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparisons test (influence of G M1 on neurite outgrowth; comparison of supplemented groups versus control) or Tukey's multiple comparisons test (influence of G M1 , NGF, and FGF2 on protein expression; comparison of groups supplemented with G M1 , NGF, and FGF2 versus control and Scientific RepoRtS | (2020) 10:5380 | https://doi.org/10.1038/s41598-020-61852-z www.nature.com/scientificreports www.nature.com/scientificreports/ comparison of groups supplemented with NGF/G M1 and FGF2/G M1 versus G M1 , NGF, and FGF2 alone). TEM data (neurons supplemented with and without 80 µM G M1 ) were analyzed using Mann-Whitney tests (density of multivesicular bodies) and Fisher's exact tests (nodular enlargements with and without mitochondria). Fisher's exact tests were also used to analyze Trypan blue dye exclusion assays. Mean values are given in the description of the results, whereas figures also show single values. Electrophysiological data is shown as mean ± standard error of the mean (SEM) and assayed with a two-tailed unpaired t-test. Two-tailed unpaired t-tests were also used to analyze effects of G M2 , G M3 and D-PDMP supplementation on neurite outgrowth and cleaved caspase 3 expression. P values < 0.05 were considered statistically significant.

Data availability
The datasets generated and analyzed during the current study can be obtained from the corresponding author on reasonable request.