Design and Synthesis of a Novel Ganglioside Ligand for Influenza A Viruses †

A novel ganglioside bearing Neuα2-3Gal and Neuα2-6Gal structures as distal sequences was designed as a ligand for influenza A viruses. The efficient synthesis of the designed ganglioside was accomplished by employing the cassette coupling approach as a key reaction, which was executed between the non-reducing end of the oligosaccharide and the cyclic glucosylceramide moiety. Examination of its binding activity to influenza A viruses revealed that the new ligand is recognized by Neuα2-3 and 2-6 type viruses.


Introduction
Influenza viruses cause a substantial number of deaths during annual epidemics and occasional pandemics [1,2]. Based on the antigenicity of their internal proteins the viruses are divided into three types, A, B, and C, of which either type A or B viruses cause seasonal influenza in humans. When the viruses bind to the host cell, hemagglutinin (HA) on their cell surface plays a significant role in the infection process. The HA protein recognizes sialoglycoconjugates expressed on the plasma membrane of the host cell, for example, sialoglycoproteins and gangliosides (sialoglycosphingolipids), as cellular ligands. Furthermore, HA can also recognize specific linkages between sialic acid (Neu5Ac/Gc) and lactosamine (LacNAc: Gal1-4GlcNAc) residues, which are found at the terminal end of glycoconjugates [3,4]. The structure and distribution of sialoglycans are crucial for viruses to determine their host animals, and two major linkage types, that is, Neu5Ac2-3LacNAc and Neu5Ac2-6LacNAc, are essential for viral transmission. Human and swine viruses predominantly recognize the Neu5Ac2-6LacNAc sequence, while avian and equine viruses bind preferentially to the Neu5Ac2-3Gal (including Neu5Ac2-3LacNAc) moiety. Swine are considered as intermediate hosts between humans and birds since they possess an abundance of both Neu5Ac2-3LacNAc and Neu5Ac2-6LacNAc structures as receptor carbohydrate determinants. The simultaneous infection of an intermediate host, such as swine, with avian and human viruses could lead to genetic recombination between the viruses, resulting in the generation of a new pathogenic virus that could potentially cause severe pandemics. However, the exact natural ligand for influenza A viruses in an intermediate host, such as pigs, remains unclear. Therefore, in this study, we focused on identifying a new carbohydrate ligand that was not only highly recognized by influenza A viruses but also functions as a natural receptor for viral HA. For this purpose, a ganglioside bearing both the Neu5Ac2-3 and Neu5Ac2-6LacNAc sequences was designed ( Figure 1). It was hypothesized that the designed ganglioside 1 could be recognized by human-and avian-derived viruses because it contains two types of sialoglycan in a single molecule. We report the chemical synthesis of ganglioside 1 and its binding activity to influenza A viruses.

Chemical Synthesis
It was envisaged that the efficient synthesis of 1 could be achieved using the cassette approach between the non-reducing end of the oligosaccharide and the glucosylceramide, which was recently developed by our group [5][6][7][8][9][10]. Furthermore, it was thought that the construction of the non-reducing end of the heptasaccharide moiety, which includes two types of sialoside, Neu5Ac2-3/2-6Gal, should be executed through a convergent synthetic approach. For the convergent synthesis of a relatively large oligosaccharide such as 1, the design of the building blocks often affects the efficiency of the total synthesis as well as the overall yield. Our preliminary experiment on the synthesis of a ganglioside similar to 1 gave a significant finding that monosaccharyl (GlcN) units are more useful as glycosyl donors than oligosaccharyl (Neu5Ac2-3/2-6Gal1-4GlcN) donors for the formation of branched structure at the 3-and 6-positions of the inner galactose residue (data not shown). Therefore, target 1 was divided into four major components, from which each building block (Units A-D) was designed ( Figure 1).

Figure 1. Structures of the target ganglioside 1 and the designed building blocks (Units A-D).
The synthetic method for the terminal Neu5Ac2-3Gal unit A has been already established by our group. The coupling of the 5-N-Troc-protected sialyl donor 2 and galactosyl acceptor 3, carrying a p-methoxyphenyl (MP) group at the anomeric position, generated the Neu5Troc2-3Gal disaccharide in good yield. The isolation of -sialoside from the reaction mixture was easily accomplished by recrystallization [11]. The obtained disaccharide was readily converted into the corresponding trichloroacetimidate donor as Unit A [12]. Similarly, the other terminal Neu5Ac2-6Gal unit (B) was prepared efficiently according to the synthetic procedure for Unit A (Scheme 1). The sialylation of the diol galactosyl acceptor 5 was performed in the presence of NIS and TfOH [13,14] in a mixed solvent system, propionitrile-dichloromethane (5:1), at −30 °C [15,16]. This mixed solvent system was used because of the poor solubility of the acceptor 5 in acetonitrile. In addition, temperatures lower than −30 °C led to a significant decrease in the yield, possibly because of the observed precipitation of 5 during the reaction. As a result of optimization, the desired -glycoside 6 was obtained in 69% yield along with a 14% yield of the -isomer. Purification of the -glycoside 6 by silica gel column chromatography was troublesome compared with that of the regioisomer, Neu5Troc2-3GalMP, which has benzyl groups on the O-2 and O-6 positions of its galactose residue, which can be isolated easily by recrystallization from an EtOAc/n-hexane system [12]. The selective deprotection of the Troc group with Zn-Cu [17]  The inner core trisaccharide structure, Unit C, was prepared starting from the 2-N-Troc protected glucosamine derivative 13 (Schemes 2 and 3). First, the glucosaminyl donors 16 and 17 were prepared as shown in Scheme 2. Removal of the acetyl groups from 13 and the subsequent formation of cyclic benzylidene acetal between O-4 and O-6 afforded 14 in good yield. The following benzylation step in the presence of a Troc group was conducted under reductive conditions. Optimization of this reductive benzylation with benzaldehyde, TESOTf, and triethylsilane [20] revealed that the use of toluene as a solvent could increase the yield. The successive reductive opening of the benzylidene group by treatment with BF 3 etherate and triethylsilane [21] gave 15 in 78% yield over two steps from 14. The obtained alcohol 15 was transformed into two types of glucosaminyl donors, namely, 16 and 17, via the introduction of a levulinoyl (Lev) and monochloroacetyl (ClAc) group to the hydroxyl group at C-4, respectively.  the diol 23 at an almost quantitative yield. A second round of glucosaminidation was conducted between 17 and 23 under the same conditions as those of the initial glucosaminidation between 16 and 21. As a result, the desired trisaccharide 24 was obtained as Unit C in a moderate yield of 49%. In this reaction, a non-negligible amount of the tetrasaccharide 25, in which both hydroxyl groups were glucosaminylated, was observed as a byproduct ( Figure 2). An attempt at using a lower temperature to increase the selectivity failed due to the poor solubility of acceptor 23 in CH 2 Cl 2 . Furthermore, changing the other factors for glycosylation, for example, the leaving group (using trichloroacetimidate) and how the donor was added, did not improve the yield of 24. It is of importance that the generation of the tetrasaccharide 25 during the reaction was faster than the complete consumption of the acceptor 23. In addition, the trisaccharide 26, which was glucosaminylated at C-4 of the galactose residue, was not detected among the by-products. These findings suggested that the newly formed trisaccharyl alcohol 24 was preferred to the disaccharyl acceptor 23 as a glycosyl acceptor. This phenomenon might be explained by the poor solubility of the disaccharyl alcohol 23 in CH 2 Cl 2 compared with the trisaccharyl alcohol 24 ( Figure 2). Next, the acetylation of 24 with acetic anhydride and DMAP in THF [23] was carried out to protect the free hydroxyl group, affording 27 in 96% yield (Scheme 3). The monochloroacetyl group on 27 was then unblocked using DABCO in ethanol [24] with an excellent yield, providing the inner core trisaccharide acceptor 28, which was ready for the next glycosylation step.   As depicted in Scheme 4, the coupling of the trisaccharide acceptor 28 with the Neu5Ac2-6Gal donor 12 promoted by TMSOTf was conducted in CH 2 Cl 2 at room temperature, affording the pentasaccharide 29 in 74% yield. During this glycosylation step, the generation of several by-products containing trichloroacetamide glycoside, which is occasionally formed as a by-product during glycosylation using trichloroacetimidate donors, made the purification process an arduous task. Column chromatography on silica gel followed by gel filtration was found to be useful for purification. Next, the conversion of the Troc carbamate at C-2 of both glucosamine residues into acetamide was achieved by treatment of alloyed zinc with copper in AcOH and followed by acetylation, giving the acetamide compound 31 in 61% yield over two steps. Finally, cleavage of the levulinoyl group by using hydrazine monoacetate in THF [25] released the 4-OH to provide the pentasaccharide acceptor 32 in 92% yield.  Scheme 5 shows the assembly of the non-reducing end heptasaccharide moiety. The Neu5Ac2-3Gal donor 4 was coupled with 32 in the presence of TMSOTf in CH 2 Cl 2 at room temperature to provide the heptasaccharide 33 in 62% yield. During this glycosylation step, the generation of the trichloroacetamide glycoside and the dimer of donor 4, which was formed by the nucleophilic attack of the hydrolyzed donor on the oxocarbenium species derived from the donor, as by-products, made the purification of the desired product 33 laborious. The structure of isolated 33 was elucidated based on its MS, 1 H, and 13 C-NMR spectra. For instance, the -configuration of the newly formed glycosidic linkage was evident from the coupling constant of the anomeric proton at  5.01 (J 1,2 = 7.5 Hz). Next, cleavage of the benzyl groups by hydrogenolysis (giving 34) followed by acetylation with conventional conditions afforded 35 in 89% yield over two steps. Selective exposure of the anomeric hydroxyl group was easily achieved by treatment with trifluoroacetic acid in CH 2 Cl 2 to yield 36. This was then converted to the corresponding trichloroacetimidate donor 37 in 95% yield over two steps from 35, which was then ready for cassette coupling with the glucosylceramide block 38 (Unit D). 12   We previously addressed the development of the cassette coupling approach between a non-reducing end oligosaccharide and a glucosylceramide (GlcCer) moiety for the synthesis of various glycolipids, particularly gangliosides [5][6][7][8][9][10]. This approach resulted in a solution to the inevitable low yield of sugar and ceramide fragments. Hitherto, we developed two types of GlcCer units: one is a cyclic type GlcCer tethered by succinic ester between the sugar and lipid portions [5,7,8], while the other is an acyclic type GlcCer [6,9,10]. In this study, we chose the cyclic type due to its ease of preparation. The reported cyclic GlcCer acceptor 38 [7] was subjected to glycosylation with the oligosaccharide donor 37 in the presence of TMSOTf in CHCl 3 at room temperature, affording the fully protected ganglioside 39 in a moderate yield of 49%. In this reaction, chloroform was employed as solvent instead of the conventional dichloromethane because of the somewhat poor solubility of 38 in CH 2 Cl 2 . Following the same protocol as that used for the other glycosylation reactions, the structure of 39 was elucidated. Next, cleavage of the p-methoxybenzyl (PMB) group by TFA in CH 2

Binding Assay
The synthesized ganglioside 1 (termed GSC-734) was then assessed for its binding activity to influenza viruses ( Figure 3). The binding assay showed that GSC-734 was recognized by Neu5Ac2-3 and 2-6 type viruses. Moreover, the binding activity of the Neu5Ac2-6 type virus (H3N2) was almost identical to that of the previously reported 2-6 sialylparagloboside [26]. This observation suggests that the branched structure of the sugar part does not potently influence its binding activity to the viruses.

General Methods for Chemical Synthesis
All reactions were carried out under a positive pressure of argon, unless otherwise noted. All chemicals were purchased from commercial suppliers and used without further purification, unless otherwise noted. Molecular sieves were purchased from Wako Chemicals Inc. (Osaka, Japan) and dried at 300 °C for 2 h in a muffle furnace prior to use. Solvents as reaction media were dried over molecular sieves and used without purification. TLC analysis was performed on Merck TLC (silica gel 60F254 on glass plate, Darmstadt, Germany). Compound detection was either by exposure to UV light (2536 Å) or by soak in a solution of 10% H 2 SO 4 in ethanol followed by heating. Silica gel (80 mesh and 300 mesh) manufactured by Fuji Silysia Co. (Kasugai, Japan) was used for flash column chromatography. Quantity of silica gel was usually estimated as 100 to 150-fold weight of sample to be charged. Solvent systems in chromatography were specified in v/v. Evaporation and concentration were carried out in vacuo. 1 H-NMR and 13 C-NMR spectra were recorded with JEOL ECA 400/500/600 spectrometers. Chemical shifts in 1 H-NMR spectra are expressed in ppm (δ) relative to the signal of Me 4 Si, adjusted to δ 0.00 ppm. Data are presented as follow: Chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, dd = double of doublet, dt = double of triplet, m = multiplet and/or multiple resonances), integration, coupling constant in Hertz (Hz), position of the corresponding proton. COSY methods were used to confirm the NMR peak assignments. MALDI-TOF mass spectra were run in a Bruker Autoflex (Billerica, MA, USA) and CHCA was used as the matrix. Highresolution mass (ESI-TOF MS) spectra were run in a Bruker micrOTOF. Optical rotations were measured with a 'Horiba SEPA-300' high-sensitive polarimeter (Kyoto, Japan). (methyl 4,7,8,9- (6). To a mixture of 2 (691 mg, 0.964 mmol) and 5 (300 mg, 0.643 mmol) in EtCN/CH 2 Cl 2 (5:1, 9.6 mL) was added 3 Å molecular sieves (991 mg) at r.t. After stirring for 1 h and then cooling to −30 °C, NIS (324 mg, 1.44 mmol) and TfOH (12.7 μL, 0.144 mmol) were added to the mixture. After stirring for 45 min at the same temperature as the reaction was monitored by TLC (1:3 EtOAc-toluene, twice development), the reaction was quenched by the addition of triethylamine. The precipitate was filtered through Celite. The filtrate was evaporated to remove EtCN and then diluted with CHCl 3 , washed with satd aq Na 2 S 2 O 3 and brine. The organic layer was subsequently dried over Na 2 SO 4 , concentrated and the residue was purified by silica gel column chromatography (1:5 EtOAc-toluene) to give 6 (473 mg, 69%) along with its -isomer (96 mg, 14%  5-acetamido-4,7,8,9- (8). To a solution of 6 (3.77 g, 3.51 mol) in AcOH/CH 2 Cl 2 (3:2, 70 mL) was added Zn/Cu couple (18.9 g) at r.t. The reaction mixture was heated to 40 °C and was stirred for 45 min at the same temperature as the reaction was monitored by TLC (4:1 toluene-EtOAc). The precipitate was filtered through Celite and the filtrate was co-evaporated with toluene. The obtained residue was exposed to high vacuum for 6 h.   (10). To a solution of 8 (3.34 g, 3.41 mol) in 1,4-dioxane (34 mL) was added Pd(OH) 2 /C (3.34 g). After stirring for 45 min at r.t. under a hydrogen atmosphere as the reaction was monitored by TLC (15:1 CHCl 3 -MeOH), the mixture was filtered through Celite. The filtrate was concentrated and the obtained crude residue was roughly purified by silica gel column chromatography. The obtained product was exposed to high vacuum for 24 h. The residue was then dissolved in pyridine (34 mL). Benzoic anhydride (3.09 g, 13.6 mmol) and DMAP (20.8 mg, 0.171 mol) were added to the mixture at 0 °C. After stirring for 9 h at r.t. as the reaction was monitored by TLC (15:1 CHCl 3 -MeOH), the reaction was quenched by the addition of MeOH at 0 °C. The mixture was co-evaporated with toluene and the residue was then diluted with CHCl 3 , and washed with 2 M HCl, H 2 O, satd aq NaHCO 3 and brine. The organic layer was subsequently dried over Na 2 SO 4 , and concentrated. The resulting residue was purified by silica gel column chromatography (1:1 toluene-EtOAc) to give 10 ( (15). To a solution of 14 (4.89 g, 9.15 mmol) in THF/toluene (1:4, 9.0 mL) was added TESOTf (4.1 mL, 18.3 mmol) at −20 °C. After stirring for 45 min at −20 °C, benzaldehyde (4.7 mL, 45.7 mmol) and triethylsilane (2.2 mL, 13.7 mmol) were added to the mixture. After stirring for 2 h at −20 °C as the reaction was monitored by TLC (1:4 EtOAc-toluene), the reaction was quenched by satd aq Na 2 CO 3 . Dilution of the mixture with EtOAc provided a solution, which was then washed with satd aq Na 2 CO 3 and brine. The organic layer was subsequently dried over Na 2 SO 4 and concentrated. The obtained residue was exposed to high vacuum for 24 h. The resulting residue was dissolved in CH 2 Cl 2 (183 mL) and cooled to 0 °C. BF 3 ·OEt 2 (4.7 mL, 18.3 mmol) and triethylsilane (14.6 mL, 91.5 mmol) were added to the solution at 0 °C and the mixture was then stirred for 1 h at 0 °C as the reaction was monitored by TLC (1:4 EtOAc-toluene). The reaction was quenched by the addition of satd aq Na 2 CO 3 at 0 °C and then diluted with CHCl 3 , and washed with satd aq Na 2 CO 3 and brine. The organic layer was subsequently dried over Na 2 SO 4 and concentrated. The     (21). To a solution of 20 (2.18 g, 3.36 mol) in AcOH/CH 2 Cl 2 (3:2, 33.6 mL) was added Zn/Cu couple (6.00 g) at r.t. The reaction mixture was stirred for 1 h at r.t. as the reaction was monitored by TLC (30:1 CHCl 3 -MeOH). The precipitate was filtered through Celite and the filtrate was co-evaporated with toluene. The residue was diluted with CHCl 3 and washed with satd aq NaHCO 3 and brine, dried over Na 2 SO 4 , and concentrated. The obtained residue was purified by silica gel column chromatography (4:1 EtOAc-n-hexane) to give 21 ( (23 5-acetamido-4,7,8,9- 5-acetamido-4,7,8,9- 62 (t, 1 H, H-3gax . The virus-binding score was expressed as mean score ± SD.

Conclusions
We efficiently synthesized a novel ganglioside designed as a ligand for influenza A viruses employing the cassette coupling approach between the heptasaccharyl sugar part and the cyclic glucosylceramide moiety. The present study revealed that the cassette approach can be applied to the synthesis of lacto-series gangliosides as well as ganglio-series gangliosides. This success will expand the applicability of the cassette approach to the synthesis of other glycolipids. In addition, we examined the binding activity of the synthesized ganglioside ligand to influenza A viruses. It was found that the synthetic ligand is recognized by Neu2-3 and 2-6 type viruses, suggesting that a glycan structure containing both the Neu2-3Gal and Neu2-6Gal sequences in a single molecule could exist as a natural ligand for influenza A viruses. To identify an actual natural ligand for influenza viruses, the synthesis of a series of gangliosides with both the Neu2-3Gal and Neu2-6Gal sequences in a single molecule is currently underway.