Structure-Activity Relationship Study of the Neuritogenic Potential of the Glycan of Starfish Ganglioside LLG-3 ‡

LLG-3 is a ganglioside isolated from the starfish Linchia laevigata. To clarify the structure-activity relationship of the glycan of LLG-3 toward rat pheochromocytoma PC12 cells in the presence of nerve growth factor, a series of mono- to tetrasaccharide glycan derivatives were chemically synthesized and evaluated in vitro. The methyl group at C8 of the terminal sialic acid residue was crucial for neuritogenic activity, and the terminal trisaccharide moiety was the minimum active motif. Furthermore, the trisaccharide also stimulated neuritogenesis in human neuroblastoma SH-SY5Y cells via mitogen-activated protein kinase (MAPK) signaling. Phosphorylation of extracellular signal-regulated kinase (ERK) 1/2 was rapidly induced by adding 1 or 10 nM of the trisaccharide. The ratio of phosphorylated ERK to ERK reached a maximum 5 min after stimulation, and then decreased gradually. However, the trisaccharide did not induce significant Akt phosphorylation. These effects were abolished by pretreatment with the MAPK inhibitor U0126, which inhibits enzymes MEK1 and MEK2. In addition, U0126 inhibited the phosphorylation of ERK 1/2 in response to the trisaccharide dose-dependently. Therefore, we concluded that the trisaccharide promotes neurite extension in SH-SY5Y cells via MAPK/ERK signaling, not Akt signaling.


Introduction
Gangliosides, a complex family of sialylated glycosphingolipids, are abundant in the vertebrate nervous system and play an important role in the development of the central nervous system. There have been many reports indicating that gangliosides can induce neuronal differentiation. A ganglioside mixture extracted from bovine brain stimulated neurite outgrowth and neuronal differentiation of the SH-SY5Y cultured human neuroblastoma cell line [1,2]. SH-SY5Y cells differentiate into adrenergic, cholinergic, or dopaminergic neurons under stimulation by various differentiation-inducing factors such as retinoic acid, phorbol ester (12-O-tetradecanoylphorbol-13-acetate), platelet-derived growth factor (PDGF), brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), basic fibroblast growth factor, insulin-like growth factor (IGF), and dibutyryl cyclic adenosine monophosphate [3][4][5][6][7]. These differentiation-inducing factors stimulate phosphorylation of Akt and extracellular signal-regulated kinase (ERK) 1/2 during neuronal differentiation [8][9][10][11][12]. Gangliosides can modify the effects of growth factors by enhancing or inhibiting their actions [13]. Mammalian GM1 ganglioside enhances the effect of NGF by binding to Tropomyosin receptor kinase A/NGF receptors in rat pheochromocytoma PC12 cells [14]. The action of epidermal growth factor (EGF) is inhibited by GM3 ganglioside binding to EGF receptors [15,16], whereas GD1a ganglioside enhances it [17]. In many cases, these effects were observed only when micromolar concentrations of gangliosides were added to the cultured cells. Micromolar levels of exogenous gangliosides might affect membrane fluidity and stability by being incorporated into the membrane, thus interfering with receptors and signaling proteins localized in glycolipid-enriched and raft membrane microdomains [18][19][20]. However, there is evidence showing that just the oligosaccharide portion of gangliosides can evoke biological responses in vitro. Oligosaccharides derived from GT1b or GM2 gangliosides could activate calmodulin-dependent protein kinase II and protein kinase A, resulting in neurite elongation when they were applied to primary cultured neurons at nanomolar levels [18,21,22]. These reports strongly suggest the presence of specific glycoreceptors on the cell surface.

Synthesis of LLG-3 Analogues
To determine the minimum motif in LLG-3 (1) that potentiates neurite outgrowth of PC12 cells in the presence of NGF, a series of mono-to tetrasaccharide glycan derivatives were obtained by disconnecting the glycosidic bonds within LLG-3 (1) from the reducing end (2, 4, 5, and 6) ( Figure 1). In addition, demethylated tetrasaccharide derivative 3 was also designed to examine the effect of the methoxy group at the C8 position of the sialic acid residue on the neuritogenic activity.
The tetrasaccharide sequence in 2 was constructed by the glycosylation of glucosyl acceptor 8 [46] with trisaccharyl imidate donor 7, which was developed in the previous study of the total synthesis of LLG-3 [40], under mild acidic conditions, giving tetrasaccharyl glycoside 9 in 81% yield (Scheme 1). Finally, global deprotection was performed as previously reported [47] to deliver LLG-3 tetrasaccharide 2.  To synthesize demethylated tetrasaccharide 3, the outer trisaccharide moiety was constructed, and then it was combined with a glucose unit (Scheme 2). Thus, sialyl glycolic acid derivative 10 [48] was condensed with 5-amino-sialyl galactoside derivative 11 [40] in the presence of EDC•HCl and HOBt in MeCN to afford trisaccharide 12 in 71% yield. Next, trisaccharide 12 was converted to suitably protected glycosyl donor 15, which was analogous to 7, and 15 was then coupled with glucosyl acceptor 8, producing tetrasaccharide 16 in 78% yield. After removal of the benzyl groups from 16, a global deprotection procedure similar to that used for 9 delivered demethylated tetrasaccharide 3.
For the synthesis of trisaccharide 4, we attempted to obtain 2-(trimethylsilyl)ethyl (SE) glycoside by glycosidating trisaccharyl donor 7 with SE-OH under conventional reaction conditions (Scheme 3). However, the small amount of the stereoisomer (α-glycoside) that was also generated during the reaction showed similar mobility to the desired β-glycoside by TLC analysis, and the isomers could not be separated by chromatographic methods. To circumvent this problem, in the synthetic route in Scheme 3 suitably protected galactose 19 [49], which contained the β-SE glycoside, was used as the glycosyl acceptor in the glycosidation of sialic acid donor 18 [50], and disaccharide 20 was obtained in moderate yield. Next, selective removal of the Troc group with zinc and AcOH in MeCN and the ensuing coupling reaction using 8-Me-sialyl glycolic acid 22 [40] produced protected trisaccharide 23 in high yield. Finally, trisaccharide 23 underwent stepwise deprotection, including de-N-acetylation, demethylation, and basic ester hydrolysis, to afford 4. The synthesis of disaccharide 5 started with the glycosidation of sialyl donor 18 with SE-OH in the presence of NIS and TfOH [51] at −40 °C in EtCN, which was used as a stereo-directing reaction media [52] (Scheme 4). This reaction produced α-SE glycoside 25 with high stereoselectivity (α/β = 6.2/1), which was then converted to the amine derivative and condensed with 22 to afford protected disaccharide 27. Finally, 27 was treated in a similar way to trisaccharide 23, furnishing target compound 5. In the final part of the syntheses of the LLG-3 glycan analogues, sialyl glycoside 25 was transformed into target compound 6 based on our method for the synthesis of 8-O-methyl sialic acid-containing molecules [40] (Scheme 5). First, 25 was converted into 8-OH derivative 29 via regioselective 8O to 5N migration of the acetyl group upon treatment with zinc under acidic conditions. Then, 8-OH protection with the chloroacetyl group gave 30, and it was further modified to diacetylimide 31 by reaction with isopropenyl acetate in the presence of acid. Next, the chloroacetyl group was selectively cleaved by selenocarbamoylpiperidine [53], and the retrieved OH was methylated by Meerwein's reagent, giving 32 in 78% yield over two steps. Finally, global deprotection produced monosaccharide 6. Scheme 5. Synthesis of LLG-3 monosaccharide 6. (a) CAc2O, DMAP/THF, RT, quant.; (b) NH2NH2•AcOH/THF, RT, 80%; (c) i. LiCl/Pyr, reflux; ii. 0.1 M NaOH aq., RT, 48% (2 steps). CAc = chloroacetyl, IPA = isopropenyl acetate, Ts = p-toluenesulfonyl, SCP = 1-selenocarbamoylpiperidine, TTBP = 2,4,6-tri-tert-butylpyrimidine.

Neurite Outgrowth Evaluation in PC12
To evaluate the neuritogenic activity of the glycan moiety of LLG-3, the mean total neurite lengths per cell were measured in rat PC12 cells ( Figure 2). Although LLG-3 tetrasaccharide 2 showed activity after 10 nM addition of 5 ng/mL NGF, tetrasaccharide 3 did not ( Figure 2B,C).
This result clearly indicates that the methoxy group at the C8 position of sialic acid residue affects the neuritogenic activity. Furthermore, to evaluate the minimum length of the glycan moiety of LLG-3 for neuritogenic activity, trisaccharide 4, disaccharide 5, and monosaccharide 6 were compared ( Figure 2). Disaccharide 5 and monosaccharide 6 showed no neurite growth activity ( Figure 2E,F). However, trisaccharide 4 induced substantial neuritogenic activity ( Figure 2D), suggesting that the trisaccharide is the minimum essential glycan moiety of LLG-3 for neuritogenic activity in PC-12 cells.

Trisaccharide 4 Stimulated Neurite Extension in SH-SY5Y cells
We also examined the neuritogenic activity of trisaccharide 4 in human neuroblastoma SH-SY5Y cells. Trisaccharide 4 elongated SH-SY5Y neurites cultured in low serum-containing medium. The maximum neurite length was attained when 1 nM of trisaccharide 4 was added to the cells and the increase was statistically significant (p < 0.05, Dunnett's test) (Figure 3). The neurite length increased up to 1 nM of trisaccharide 4 in a dose-dependent manner, and then decreased at higher concentrations ( Figure 3). Many researchers have reported that neuritogenesis of SH-SY5Y is often accompanied by activation of the MAPK/ERK and phosphatidylinositide 3-kinase (PI3K)/Akt signaling cascade after stimulation with growth factors, such as NGF, BDNF, and retinoic acid [8][9][10][11][12], and the signal transductions mediated by MAPK/ERK and PI3K/Akt are thought to be important for cell survival and neuronal differentiation. Therefore, we investigated whether trisaccharide 4 also promotes phosphorylation of ERK 1/2 and Akt. Phosphorylation of ERK 1/2 was rapidly induced by addition of 1 or 10 nM of trisaccharide 4. The ratio of phosphorylated ERK (p-ERK) to ERK reached a maximum 5 min after stimulation, and then decreased gradually ( Figure 4A); however, the ratios varied considerably between experiments when 10 nM of trisaccharide 4 was added ( Figure 4A). Although 1 nM of trisaccharide 4 showed a slightly lower value against 40 ng/mL of NGF (157.3 ± 16.9% vs. 216.0 ± 15.8%), the increase in the relative ratio of p-ERK to ERK (p-ERK/ERK) at 5 min was statistically significant (p < 0.05, Dunnett's test) compared with 0 min ( Figure 4A).  Trisaccharide 4 also induced phosphorylation of ERK after 5 min incubation dose-dependently. The relative ratio of p-ERK/ERK increased in a dose-dependent manner and reached a maximum value at 1 nM (p < 0.05, Dunnett's test), but decreased at higher concentrations ( Figure 4B). The dose dependency of ERK phosphorylation corresponded well to the results in Figure 3. However, trisaccharide 4 did not induce Akt phosphorylation significantly, although 40 ng/mL of NGF increased the relative ratio of phosphorylated Akt to Akt (p-Akt/Akt) 5 min after stimulation (547.7 ± 100.7%) ( Figure 4C).

U0126 Inhibits Trisaccharide 4-Promoted ERK Phosphorylation and Neurite Extension
Trisaccharide 4 promoted neurite extension of SH-SY5Y significantly at 1 nM ( Figure 5A,B,F), and this effect was abolished by pretreatment with MAPK inhibitor U0126, which inhibits enzymes MEK1 and MEK2 ( Figure 5C,D,E). The inhibitory effect of U0126 on trisaccharide 4-induced neurite extension was dose-dependent ( Figure 5F). In addition, U0126 inhibited the phosphorylation of ERK 1/2 in response to trisaccharide 4 dose-dependently ( Figure 6A,B), and this response resembled that observed in the neurite extension inhibitory effect ( Figure 5F). Therefore, we can infer that trisaccharide 4 stimulates neuritogenesis in SH-SY5Y via activation of the ERK signal cascade, not the PI3K/Akt pathway.  The effect of trisaccharide 4 on neuritogenesis was slightly weaker than that of NGF in PC12 cells (Supplemental Figure 1). This may be accounted for by the fact that trisaccharide 4 activates only the MEK/ERK signaling pathway, unlike NGF, which can activate both MEK/ERK and PI3K/Akt signaling cascades to elongate neurites. Because trisaccharide 4 exerts a prompt effect on SH-SY5Y cells at concentrations as low as 1 nM (Figure 3 and Figure 4A,B), this suggests the presence of specific receptors that recognize trisaccharide 4 on the cell surface of SH-SY5Y. Alternatively, trisaccharide 4 may interact with other receptors specifically or non-specifically and modify their functions. In this scenario, the receptors may be growth factor receptors because fetal bovine serum used in cell culture will inevitably contain growth factors such as PDGF, IGF, EGF, insulin, fibroblast growth factor-2, and transforming growth factor beta 1. Trisaccharide 4 may exert its effect by mimicking, enhancing, or inhibiting these growth factors by interacting with their receptors. There are several reports suggesting the presence of glycoreceptors on the cell surface that recognize the oligosaccharide portion of the ganglioside and elicit biological responses [18,21,22]. Our results strongly support this idea, and we defined the precise sugar structure from which the neuritogenic activity of the LLG-3 ganglioside originated.

General Methods
All reactions were carried out under a positive pressure of argon, unless otherwise noted. All chemicals were purchased from Wako Chemicals Inc. (Miyazaki, Japan), Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), or Sigma-Aldrich Co. (St. Louis, MO, USA) and used without further purification, unless otherwise noted. Molecular sieves were purchased from Wako Chemicals Inc. (Miyazaki, Japan) and pre-dried at 300 °C for 2 h in a muffle furnace, and dried in a flask at 300 °C for 2 h in vacuo prior to use. Dry solvents for reaction media (CH2Cl2, toluene, THF, CH3CN, DMF, pyridine) were purchased from Kanto Chemical Co. Inc. (Tokyo, Japan) and used without purification. Other solvents for reaction media were dried over molecular sieves and used without purification. TLC analysis was performed on Merck TLC plates (silica gel 60F254 on glass plate). Compound detection was either by exposure to UV light (253.6 nm) or by soaking in a solution of 10% H2SO4 in ethanol followed by heating. Silica gel (80 mesh and 300 mesh; Fuji Silysia Co. (Aichi, Japan) ) was used for flash column chromatography. The quantity of silica gel was usually 100 to 200 times the weight of the crude sample. Solvent systems for chromatography were specified as v/v ratios. Evaporation and concentration were carried out in vacuo. 1 H-NMR and 13 C-NMR spectra were recorded on 400 MHz (JEOL ECX400), 500 MHz (Biospin AVANCE III, Bruker, Billerica, MA, USA) or 600 MHz (JEOL ECA600) spectrometers. Chemical shifts in 1 H-NMR spectra are expressed in ppm (δ) relative to the Me4Si signal, adjusted to δ 0.00 ppm. Data are presented as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, dd = double doublet, td = triple doublet, m = multiplet and/or multiple resonances), integration, coupling constant in hertz (Hz), and position of the corresponding proton. COSY methods were used to confirm the NMR peak assignments. Highresolution mass (ESI-TOF MS) spectra were obtained with a mass spectrometer (micrOTOF, Bruker, Billerica, MA, USA). Optical rotations were measured with a high-sensitivity polarimeter (SEPA-300, Horiba, Kyoto, Japan).