Chemoenzymatic Synthesis of Sialoglycopolypeptides As

538

Bioconjugate Chem. 2009, 20, 538–549

Chemoenzymatic Synthesis of Sialoglycopolypeptides As Glycomimetics to Block Infection by Avian and Human Influenza Viruses Makoto Ogata,4,† Kazuya I. P. J. Hidari,4,‡ Takeomi Murata,† Shizumi Shimada,‡ Wataru Kozaki,§ Enoch Y. Park,† Takashi Suzuki,‡ and Taichi Usui*,† Department of Bioscience, Graduate School of Science and Technology, Shizuoka University, Ohya 836, Suruga ward, Shizuoka 422-8529, Japan, Department of Biochemistry, University of Shizuoka, School of Pharmaceutical Sciences and Global COE Program, and Department of Applied Biological Chemistry, Faculty of Agriculture, Shizuoka University. Received October 23, 2008; Revised Manuscript Received December 31, 2008

We designed a series of γ-polyglutamic acid (γ-PGA)-based glycopolypeptides carrying long/short R2,3/6 sialylated glycans to act inhibitors of the influenza virus. As an alternative design, sialoglycopolypeptides carrying longspacer linked glycans were engineered by replacement of the N-acetyllactosamine (LN) unit by an alkyl chain. The structure-activity relationship of the resulting sialoglycopolypeptides with different glycans in the array has been investigated by in vitro and in vivo infection experiments. The avian viruses specifically bound to glycopolypeptides carrying a short sialoglycan with higher affinity than to a long glycan. In contrast, human viruses, preferentially bound not only to long R2,3/6 sialylated glycan with LN repeats in the receptors, but also to more spacer-linked glycan in which the inner sugar has been replaced by a nonsugar structural unit such as a pentylamido group. Taken together, our results indicate that a spaced tandem/triplet pentylamido repeat is a good mimetic of a tandem/triplet LN repeat. Our strategy provides a facile way to design strong polymeric inhibitors of infection by avian and human influenza viruses.

INTRODUCTION Influenza viruses infect host cells through the binding of viral hemagglutinins (HAs) to sialoglycoproteins or sialoglycopeptides of the receptors on the host cell surfaces (1-3). Because the molecular recognition process leads to the adhesion of host cell and virus, molecules having a high affinity for the viral HA act as potent inhibitors of infection by influenza viruses (4-6). Host cell specificity of the virus is dependent not only on linkage of sialic acids (SAs) to the penultimate galactose, but also on the number of SA residues and the precise nature of the core structure (7-10). The recognition ability of glycol signals is also known to be greatly enhanced by clustering the sialooligosaccharides on macromolecular scaffolds such as the synthetic glycopolymers (6, 11-16). The multivalent interactions have been amplified by the so-called glycoside cluster effects (17). Various factors are known to influence the affinity and specificity of multivalent binding events including the structure of the individual saccharide residues (8-10). In this respect, a variety of synthetic glycopolymers carrying multivalent sialooligosaccharides that target HAs have been produced for the influenza virus inhibitors using polyacrylamide (11, 17-22), poly(acrylacid) (14, 23), polystyrene (12), and dendrimers (24) as polymer backbone. However, synthetic glycopolymers potentially pose several problems for in vivo use, such as low solubility, significant cytotoxicity, and immunogenicity. We have already reported that sialoglycopolypeptides with R- or γ-PGA backbones can be used as a scaffold in the synthesis of multivalent inhibitors of the influenza viruses (6, 9, 16). Glycopolypeptides possessing either multivalent SAR2,3Gal or * Corresponding author. Fax: +81-54-238-4873; e-mail: actusui@ agr.shizuoka.ac.jp. 4 These authors contributed equally to this work. † Shizuoka University. ‡ University of Shizuoka and Global COE Program. § Faculty of Agriculture, Shizuoka University.

SAR2,6Gal have aided our understanding of human adaptation of the H5N1 influenza virus (9). The length of the asialo portion in glycopolypeptides also increased binding affinity to the viral HA in a glycan length-dependent manner (8). Recent studies using cocrystal structures with glycan microarrays also showed that human viral HAs specifically bound to long R2,6 sialylated glycans with a length beyond trisaccharide, whereas avian viral HAs bound to short R2,3 sialylated glycans (25). These findings indicate that glycan length of the receptors is another determinant of human viral HAs binding to the terminal R2,6 linked sialoside in the core glycan structures. Thus far in these macromolecular approaches, the use of ligands has been limited to small carbohydrate units containing sialodi- or sialotrisaccharide. Spatial arrangement of the side chains is not usually considered during the design of the ligand. Our strategy of molecular design is to construct an amphiphilic structure, taking into account the contribution of the asialo portion of the glycan chain, by arranging multivalent short or long glycans attached to a polypeptide backbone. We therefore planned to chemoenzymatically synthesize a series of γ-PGAbased sialoglycopolymers carrying glycans or spacers of different lengths in the array, in order to identify the specific SA motifs recognized by the influenza virus. Using these glycopolymers, the structure-activity relationship was elucidated by measurement of the inhibition of infection by avian and human influenza viruses.

EXPERIMENTAL SECTION Materials. γ-PGA-Na (MW 990 000) from Bacillus subtilis was a kind gift from Meiji Food Materia Co. UDP-GlcNAc, UDP-Gal, and CMP-Neu5Ac were gifts from Yamasa Corporation (Chiba, Japan). Bovine milk β1,4-galactosyltransferase (β4GalT), R2,3-(N)-sialyltransferase (R2,3SiaT, rat recombinant, Spodoptera frugiperda), and R2,6-(N)-sialyltransferase (R2,6SiaT, rat recombinant, S. frugiperda) were purchased from Calbiochem-Novabiochem (San Diego, CA). β1,3-N-Acetylglucosami-

10.1021/bc800460p CCC: $40.75  2009 American Chemical Society Published on Web 02/12/2009

Chemoenzymatic Synthesis of Sialoglycopolypeptides

Bioconjugate Chem., Vol. 20, No. 3, 2009 539

Table 1. Synthesis of Asialoglycopolypeptides with Different Degrees of Polymerization of Glutamic Acid Residues products

compound

PGAa (mg)

BOPb (mg)

HOBtc (mg)

amino-sugard (mg)

yield (mg)

Poly[LNβ-O(CH2)5NHs/γ-PGA] Poly[LNnTβ-O(CH2)5NHs/γ-PGA] Poly[(LN)2β-O(CH2)5NHs/γ-PGA] Poly[LNβ1,3LNnTβ-O(CH2)5NHs/γ-PGA] Poly[(LN)3β-O(CH2)5NHs/γ-PGA] Poly[LNβ-O(CH2)5NHCO(CH2)5NHs/γ-PGA] Poly{LNβ-O[(CH2)5NHCO]2(CH2)5NHs/γ-PGA}

16 17 18 19 20 21 22

15 15 15 15 15 35 35

118 118 118 118 118 276 276

15 15 15 15 15 34 34

46 39 83 58 119 79 95

19 22 32 28 39 42 59

DSe (%) 40 35 48 31 44 47 49

a γ-PGA: MW ) 990 000. b Benzotriazol-1-yloxvtris-(dimethylamino)phosphonium hexafluorophosphate. c 1-Hydroxybenzotriazole hydrate. 5-AP-β-LN for 16, 5-AP-β-LNnT for 17, 5-AP-β-(LN)2 for 18, 5-AP-β-LNβ1-3LNnT for 19, 5-AP-β-(LN)3 for 20, 5-(5′-aminopentanecarboxamido)pentyl β-LN for 21, and 5-[5′-(5′′-aminopentanecarboxamido)pentanecarboxamido]pentyl β-LN for 22. e Degree of substitution of sugar derivatives based on DP of g-PGA as 100%. Calculated from 1H NMR data at 25 °C. d

nyltransferase (β3GnT) was prepared by our previously described methods (26). 5-Trifluoroacetamidopentyl β-lactoside (5-TFAP-β-Lac), 5-TFAP-β-LN, and 5-aminopentyl β-LacNAc (5-AP-β-LN) were synthesized by our previously described methods (6). The human and avian influenza virus strains used in this study were propagated and purified as described previously (2, 27). Viral hemagglutination units (HAUs) of purified viruses were determined as described previously (28). All other reagents were of the highest quality commercially available and were used without further purification. Enzyme Assay. β3GnT activity was assayed as follows. UDP-GlcNAc (40 mM), LNβ-pNP (20 mM), and MnCl2 (16 mM) were dissolved in 150 mM Tris-HCl (pH 6.8) followed by 100 µL of enzyme solution (total vol 200 µL). The reaction was started by addition of β3GnT. At each sample time, 10 µL of the reaction mixture was added to 190 µL of distilled water, followed by immediate boiling for 5 min. After filtration through a 0.45 µm nitrocellulose filter (Millipore, Bedford, MA), the filtrates were analyzed by HPLC using a TSK-gel ODS 80TsQA (4.6 × 250 mm, TOSOH Co.) column. Reaction products were eluted with 10% MeOH containing 0.1% TFA and detected at the absorbance of 300 nm. HPLC was done at 40 °C with a flow rate 1.0 mL/min. One unit of enzyme activity is defined as the amount of enzyme capable of catalyzing the transfer of 1 µmol of GlcNAc per minute. β4GalT activity was assayed as follows. UDP-Gal (40 mM), GlcNAcβ1,3LNβ-pNP (20 mM), and MnCl2 (16 mM) were dissolved in 150 mM Tris-HCl (pH 6.8) followed by 100 µL of enzyme solution (total vol 200 µL). The reaction was initiated by addition of β4GalT. At each sample time, 10 µL of the reaction mixture was added to 190 µL of distilled water, followed by immediate boiling for 5 min. After filtration through a 0.45 µm nitrocellulose filter (Millipore, Bedford, MA), the filtrates were analyzed by HPLC using a TSK-gel ODS 80TsQA (4.6 × 250 mm, TOSOH Co.) column. Reaction products were eluted with 10% MeOH containing 0.1% TFA and detected at the absorbance of 300 nm. HPLC was done at 40 °C with a flow rate 1.0 mL/min. One unit of enzyme activity is defined as the amount of enzyme capable of catalyzing the transfer of 1 µmol of Gal per minute. HPLC Determination of Neutral Sugar Units in Asialoglycopolypeptides. The artificial asialoglycopolypeptides (0.02 µM, 250 µL) were hydrolyzed with 4 M trifluoroacetic acid (250 µL) at 100 °C for 4 h to release sugar units and the hydrolysates were then freeze-dried. In brief, to dry samples were added water (10 µL) and ethyl 4-aminobenzoate (ABEE) solution (40 µL). Derivatization was carried out at 80 °C for 1 h. The reaction mixture of ABEE derivatization in 10 µL portions was directly injected into a Honenpak C18 column (4.6 × 75 mm) and eluted using 10% CH3CN containing 0.02% TFA. A JASCO LC-2000 plus HPLC system equipped with an UV-2070 plus UV detector (305 nm), operating isocratically at 1.0 mL/min at a column temperature of 45 °C, was used (Table 2) (29).

HPLC Determination of Sialic Acids in Sialoglycopolypeptides. The artificial sialoglycopolypeptides (0.1 µM, 5 µL) were hydrolyzed with 0.05 M HCl (45 µL) at 80 °C for 1 h to release sialic acid units. Next, the samples were added to a 1,2-diamino4,5-methylenedioxybenzene (DMB) solution (200 µL). Derivatization was carried out at 50 °C for 2.5 h in the dark. All labeled compounds were kept at 4 °C in the dark to prevent degradation. The reaction mixture of DMB derivatization in 10 µL portions was directly injected into a Shodex ODP2 HP-4E column (4.6 × 250 mm, Shodex) and eluted with water/methanol/acetonitrile (84:7:9, v/v/v). A JASCO LC-2000 plus HPLC system equipped with an FP-2020 plus fluorescence detector (excitation, 373 nm; emission, 448 nm), operating isocratically at 0.5 mL/min at a column temperature of 40 °C, was used (Table 2) (30). Analytical Methods. HPLC analysis was carried out using an Asahipak NH2-P50 4-E column (4.6 × 250 mm, Shodex) with a JASCO Intelligent system liquid chromatograph and detection at 210 nm. The bound material was eluted with 80% CH3CN at a flow rate of 1.0 mL/min at 40 °C. FAB-mass analysis was carried out in positive ion mode using a JEOL JMS DX-303HF mass spectrometer coupled to a JEOL DA800 mass data system. An accelerating voltage of 10 kV and a mass resolution of 1000 were employed. A sample in distilled water (1 µL) was loaded onto a probe tip and mixed with glycerol (1 µL) as a matrix. 1H and 13C NMR spectra were recorded on a JEOL JNM-LA 500 spectrometer at 25 °C. Chemical shifts are expressed in δ relative to sodium 3-(trimethylsilyl) propionate as an external standard. The amount of protein was determined using a Bio-Rad protein assay kit. Synthesis of LN-Repeating Tetra- And Hexasaccharide Glycosides. 5-Trifluoroacetamidopentyl β-lacto-N-neotetraoside (5-TFAP-β-LNnT, 3) was synthesized by the alternative addition of β1-3 linked GlcNAc and β1-4 linked Gal to 5-TFAP-βLac (1), using two kinds of glycosyltransferase. Compound 1 (300 mg, 0.56 mmol) and UDP-GlcNAc (725 mg, 1.11 mmol) were dissolved in 150 mM Tris-HCl buffer pH 6.8 (50.3 mL) containing MnCl2 (88.2 mg) and 1% (w/v) NaN3 (0.56 mL), followed by addition of 540 mU (4.8 mL) of partially purified β3GnT preparation from the cell culture supernatant. The mixture was incubated for 168 h at 37 °C, and the reaction was terminated by boiling for 5 min. UDP-Gal (679 mg, 1.11 mmol) was dissolved the mixture, followed by addition of 4 U (0.2 mL) of β4GalT from bovine milk. The mixture was incubated for 124 h at 37 °C, and the reaction was terminated by boiling for 5 min. The precipitate was removed by centrifugation (8000 g, 20 min) and the supernatant was loaded onto a ODS column (3.5 × 60 cm) equilibrated with 10% MeOH at flow rate of 2.0 mL/min. After washing the column with 1500 mL of 10% MeOH, absorbed portion was eluted with 20% MeOH and a fraction size of 25 mL/tube. The eluate was monitored by measuring the absorbance at 210 nm using a spectrometer. An aliquot from fractions 14-28 was then concentrated and lyophilized: compound 3 was obtained in a total yield of 90% (451 mg) based on the acceptor substrate 1. [R]D30 + 7.0° (c

540 Bioconjugate Chem., Vol. 20, No. 3, 2009

Ogata et al.

Table 2. Glycopolypeptides Carrying LN-Repeating and Long-Spacer Linked Aminoalkyl Oligosaccharide Glycosides for Inhibition of Binding by Influenza Viruses DS (%) sugar moiety

spacer

a

compound

NS d

Siab e

kDac

LNβLNnTβ(LN)2βLNβ1,3LNnTβ(LN)3βLNβLNβ-

-O(CH2)5NHs -O(CH2)5NHs -O(CH2)5NHs -O(CH2)5NHs -O(CH2)5NHs -O(CH2)5NHCO(CH2)5NHs -O[(CH2)5NHCO]2(CH2)5NHs

16 17 18 19 20 21 22

40 (33) 35 (30) 48 (45) 31 (28) 44 (44) 47 (46) 49 (40)

s s s s s s s

2100 2700 3500 3300 4300 2700 3100

SAR2,3LNβSAR2,3LNnTβSAR2,3(LN)2βSAR2,3LNβ1,3LNnTβSAR2,3(LN)3βSAR2,3LNβSAR2,3LNβ-


23 24 25 26 27 28 29

0 0 0 0 0 0 0

40 (37) 35 (32) 48 (45) 31 (33) 44 (40) 47 (48) 49 (46)

2800 3400 4400 3900 5200 3600 4000

SAR2,6LNβSAR2,6LNnTβSAR2,6(LN)2βSAR2,6LNβ1,3LNnTβSAR2,6(LN)3βSAR2,6LNβSAR2,6LNβ-


30 31 32 33 34 35 36

0 0 0 0 0 0 0

40 (39) 35 (39) 48 (48) 31 (33) 44 (42) 47 (44) 49 (47)

2800 3400 4400 3900 5200 3600 4000

a Neutral sugar derivatives substituted. b Sialyl sugar derivatives substituted ()Sia contents). c Calculated kDa. d Degree of substitution of asialo- or sialo-sugar derivatives based on DP of γ-PGA as 100%. Calculated from 1H NMR data at 25 °C. e Degree of substitution of asialo- or sialo-sugar derivatives based on DP of γ-PGA as 100%.Calculated from two type HPLC analyses (ABEE and DMB methods). (All data normalized to those of 1H NMR.)

0.1, water). FAB-mass: m/z 889 [M + H]+ (matrix: glycerol). 1 H NMR (D2O, 500 MHz): δ 4.70 (d, 1H, J1′′,2′′ 8.6 Hz, H-1′′), 4.48 (d, 1H, J1′,2′ 8.0 Hz, H-1′), 4.47 (d, 1H, J1,2 8.0 Hz, H-1), 4.43 (d, 1H, J1′′′,2′′′ 7.6 Hz, H-1′′′), 4.16 (1H, H-4′), 3.99-3.94 (2H, H-6b, H-6′′b), 3.93 (1H, H-4′′′), 3.91 (1H, H-Rb), 3.87-3.63 (15H of sugar protons), 3.60-3.56 (4H, H-Ra, H-5′′, H-5, H-2′), 3.54 (dd, 1H, J1′′,2′′ 8.6, J2′′,3′′ 10 Hz, H-2′′), 3.34 (t, 2H, H-ε), 3.29 (t, 1H, J1,2 8.0, J2,3 8.0 Hz, H-2), 2.03 (s, 3H, CH3CONH′′-), 1.66 (q, 2H, H-β), 1.61 (q, 2H, H-ω), 1.40 (q, 2H, H-γ). 13C NMR (D2O, 500 MHz): δ 177.7 (CH3CONH′′-), 161.6 (CF3CO), 118.8 (CF3CONHs), 105.8 (C-1′′′), 105.7 (C1′), 105.6 (C-1′′), 104.8 (C-1), 84.9 (C-3′), 81.2 (C-4), 81.0 (C4′′), 78.2 (C-5′′′), 77.7 (C-5′), 77.6 (C-5), 77.4 (C-5′′), 77.2 (C3), 75.6 (C-2), 75.3 (C-3′′′), 75.0 (C-3′′), 73.8 (C-2′′′), 73.1 (CR), 72.8 (C-2′), 71.4 (C-4′′′), 71.2 (C-4′), 63.8 (C-6′′′, C-6′), 62.9 (C-6), 62.7 (C-6′′), 58.0 (C-2′′), 42.5 (C-ε), 31.1 (C-β), 30.2 (C-ω), 25.2 (C-γ), 25.0 (CH3CONH′′-). 5-TFAP-β-(LN)2 (4) was synthesized by the alternative addition of GlcNAc and Gal residue to 5-TFAP-β-LN (2) in a similar manner. Compound 4 was obtained in a total yield of 72% (356 mg) based on the acceptor substrate 2. [R]30 D -12.3° (c 0.1, water). FAB-mass: m/z 930 [M + H]+ (matrix: glycerol). 1H NMR (D2O, 500 MHz): δ 4.71 (d, 1H, J1′′,2′′ 8.3 Hz, H-1′′), 4.52 (d, 1H, J1,2 7.6 Hz, H-1), 4.48 (d, 1H, J1′′′,2′′′ 8.0 Hz, H-1′′′), 4.46 (d, 1H, J1′,2′ 7.9 Hz, H-1′), 4.16 (1H, H-4′), 4.04-3.97 (2H, H-6b, H-6′′b), 3.94 (1H, H-4′′′), 3.90 (1H, H-Rb), 3.87-3.66 (16H of sugar protons), 3.60-3.56 (4H, H-Ra, H-5′′, H-5, H-2′), 3.55 (dd, 1H, J 1′′,2′′ 8.3, J2′′,3′′ 10 Hz, H-2′′), 3.33 (t, 2H, H-ε), 2.04 (s, 3H, CH3CO), 2.03 (s, 3H, CH3CONH′′-), 1.62-1.56 (4H, H-β, H-ω), 1.35 (2H, H-γ). 13C NMR (D2O, 500 MHz): δ 177.7 (CH3CONH′′-), 177.2 (CH3CO), 161.6 (CF3CONHs), 118.8 (CF3CO), 105.72 (C-1′), 105.68 (C-1′′′), 105.6 (C-1′′), 103.8 (C-1), 84.9 (C-3′), 81.3 (C-4), 80.9 (C-4′′), 78.2 (C-5′′′), 77.7 (C-5′), 77.6 (C-5), 77.4 (C-5′′), 75.3 (C-3′′′, C-3), 75.0 (C-3′′), 73.8 (C-2′′′), 73.1 (C-R), 72.8 (C-2′), 71.4 (C-4′′′), 71.1 (C-4′), 63.8 (C-6′′′, C-6′), 62.9 (C-6), 62.7 (C-6′′), 58.0 (C-2′′), 57.9 (C-2), 42.5 (C-ε), 31.0 (C-β), 30.2 (C-ω), 25.2 (C-γ), 25.0 (CH3CONH′′-), 24.9 (CH3CO). 5-TFAP-β-(LNβ1,3LNnT) (5) was synthesized by the alternative addition of GlcNAc and Gal residue to 3 in a similar

manner. Compound 5 was obtained in a total yield of 75% (153 mg) based on the acceptor substrate 3. [R]30 D + 7.2° (c 0.1, water). FAB-mass: m/z 1255 [M + H]+ (matrix: glycerol). 1H NMR (D2O, 500 MHz): δ 4.70 (d, 2H, J1′′′′,2′′′′ 8.6, J1′′,2′′ 8.6 Hz, H-1′′′′, H-1′′), 4.48 (d, 2H, J1′′′,2′′′ 7.0, J1′,2′ 7.0 Hz, H-1′′′, H-1′), 4.47 (d, 1H, J1,2 8.3 Hz, H-1), 4.44 (d, 1H, J1′′′′′,2′′′′′ 7.7 Hz, H-1′′′′′), 4.16 (2H, H-4′′′, H-4′), 3.99-3.90 (5H, H-6b, H-6′29b, H-6′′′′b, H-4′′′′′, H-Rb), 3.87-3.53 (30H of sugar protons and aglycon R-methylene protons), 3.34 (t, 2H, H-ε), 3.30 (t, 1H, J1,2 8.3, J2,3 8.3 Hz, H-2), 2.04 (s, 6H, CH3CONH′′′′-, CH3CONH′′-), 1.66 (q, 2H, H-β), 1.62 (q, 2H, H-ω), 1.41 (q, 2H, H-γ). 13C NMR (D2O, 500 MHz): δ 177.7 (CH3CONH′′′′-, CH3CONH′′-), 161.6 (CF3CONHs), 118.8 (CF3CO), 105.8 (C-1′′′′′), 105.7 (C1′′′, C-1′), 105.6 (C-1′′′′, C-1′′), 104.8 (C-1), 84.9 (C-3′′′, C-3′), 81.2 (C-4), 81.0 (C-4′′′′, C-4′′), 78.2 (C-5′′′′′), 77.7 (C-5′′′, C-5′), 77.6 (C-5), 77.4 (C-5′′′′, C-5′′), 77.2 (C-3), 75.6 (C-2), 75.3 (C-3′′′′′), 75.0 (C-3′′′′, C-3′′), 73.8 (C-2′′′′′), 73.1 (C-R), 72.8 (C-2′′′, C-2′), 71.4 (C-4′′′′′), 71.2 (C-4′′′, C-4′), 63.9 (C-6′′′′′), 63.8 (C-6′′′, C-6′), 62.9 (C-6), 62.7 (C-6′′′′, C-6′′), 58.0 (C2′′′′, C-2′′), 42.5 (C-ε), 31.1 (C-β), 30.2 (C-ω), 25.2 (C-γ), 25.0 (CH3CONH′′′′-, CH3CONH′′-). 5-TFAP-β-(LN)3 (6) was synthesized by the alternative addition of GlcNAc and Gal residues to 4 in a similar manner. Compound 6 was obtained in a total yield of 85% (232 mg) based on the acceptor substrate 4. [R]D30-10.7° (c 0.1, water). FAB-mass: m/z 1296 [M + H]+ (matrix: glycerol). 1H NMR (D2O, 500 MHz): δ 4.71 (d, 2H, J1′′′′,2′′′′ 8.2, J1′′,2′′ 8.2 Hz, H-1′′′′, H-1′′), 4.52 (d, 1H, J1,2 7.4 Hz, H-1), 4.48 (d, 1H, J1′′′,2′′′ 7.3 Hz, H-1′′′), 4.47 (d, 1H, J1′′′′′,2′′′′′ 7.0 Hz, H-1′′′′′), 4.46 (d, 1H, J1′,2′ 8.0 Hz, H-1′), 4.16 (2H, H-4′′′, H-4′), 4.00-3.94 (3H, H-6b, H-6′′b, H-6′′′′b), 3.93 (1H, H-4′′′′′), 3.90 (1H, H-Rb), 3.87-3.66 (23H of sugar protons), 3.60-3.56 (5H, H-Ra, H-5′′′′, H-5′′, H-5, H-2′′′, H-2′), 3.55 (dd, 2H, J1′′′′,2′′′′ 8.2, J2′′′′,3′′′′ 10, J1′′,2′′ 8.3, J2′′,3′′ 10 Hz, H-2′′′′, H-2′′), 3.33 (t, 2H, H-ε), 2.04 (s, 6H, CH3CONH′′′′-, CH3CONHs), 2.03 (s, 3H, CH3CONH′′-), 1.62-1.57 (4H, H-β, H-ω), 1.36 (2H, H-γ). 13C NMR (D2O, 500 MHz): δ 177.7 (CH3CONH′′′′-, CH3CONH′′-), 177.2 (CH3CONHs), 161.6 (CF3CONHs), 118.8 (CF3CONHs), 105.7 (C-1′′′′′, C-1′′′, C-1′), 105.6 (C-1′′′′, C-1′′), 103.8 (C-1), 84.9 (C-3′′′, C-3′), 81.3 (C-4), 81.0 (C-4′′′′, C-4′′), 78.2 (C-5′′′′′), 77.7


(C-5′′′, C-5′), 77.6 (C-5), 77.4 (C-5′′′′, C-5′′), 75.3 (C-3′′′′′, C-3), 75.0 (C-3′′′′, C-3′′), 73.8 (C-2′′′′′), 73.1 (C-R), 72.8 (C-2′′′, C-2′), 71.4 (C-4′′′′′), 71.1 (C-4′′′, C-4′), 63.8 (C-6′′′′′, C-6′′′, C-6′), 62.9 (C-6), 62.7 (C-6′′′′, C-6′′), 58.0 (C-2′′′′, C-2′′), 57.9 (C-2), 42.5 (C-ε), 31.0 (C-β), 30.2 (C-ω), 25.2 (C-γ), 25.0 (CH3CONH′′′′-, CH3CONH′′-), 24.9 (CH3CONHs). Synthesis of More Spaced Disaccharide Glycosides. Compound 9 (180 mg, 0.39 mmol), HOBT (208 mg, 1.54 mmol) and 6-trifluoroacetamidohexanoic acid (350 mg, 1.54 mmol) were dissolved in DMF (7.0 mL). Then, DIPCD (240 µL, 1.54 mmol) was added to the solution with continuous stirring for 24 h at room temperature. After the reaction mixture was concentrated to a solid, it was dissolved in 5 mL of CHCl3/ CH3OH/H2O ) 7.5/2.5/0.35 and then loaded onto a silica gel 60N column (2.0 × 30 cm). The column was developed with the same solvent at a flow rate of 10 mL/min and a fraction size of 20 mL/tube. Fractions 26-66 were pooled and concentrated: 5-[5′-(trifluoroacetamido) pentanecarboxamido]pentyl β-LN (7) was obtained in a total yield of 59% (154 mg). [R]D30-2.9° (c 0.1, water). FAB-mass: m/z 678 [M + H]+ (matrix: glycerol). 1H NMR (D2O, 500 MHz): δ 4.52 (d, 1H, J1,2 7.0 Hz, H-1), 4.47 (d, 1H, J1′2′ 8.0 Hz, H-1′), 3.98 (dd, 1H, J5,6b 1.2, J6a,6b 12.2 Hz, H-6b), 3.93 (1H, H-4′), 3.90 (m, 1H, H-Rb), 3.83 (dd, 1H, J5,6a 4.6, J6a,6b 12.2 Hz, H-6a), 3.78-3.75 (2H, H-6′a, H-6′b), 3.73-3.70 (4H, H-5′, H-4, H-3, H-2), 3.67 (dd, 1H, J2′,3′ 10, J3′,4′ 4.9 Hz, H-3′), 3.61 (m, 1H, H-Ra), 3.60 (1H, H-5), 3.54 (dd, 1H, J1′,2′ 8.0, J2′,3′ 10 Hz, H-2′), 3.32 (t, 2H, H-ε′), 3.17 (t, 2H, H-ε), 2.24 (t, 2H, H-R′), 2.03 (s, 3H, CH3CONHs), 1.64-1.54 (6H, H-β′, H-β, H-ω′), 1.51 (m, 2H, H-ω), 1.32 (4H, H-γ′, H-γ). 13C NMR (D2O, 500 MHz): δ 179.5 (-CONHs), 177.2 (CH3CONHs), 161.8 (CF3CONHs), 117.6 (CF3CONHs), 105.7 (C-1′), 103.8 (C-1), 81.3 (C-4), 78.2 (C5′), 77.6 (C-5), 75.32 (C-3′), 75.28 (C-3), 73.8 (C-2′), 73.1 (CR), 71.4 (C-4′), 63.8 (C-6′), 62.9 (C-6), 57.9 (C-2), 42.4 (C-ε′), 42.0 (C-ε), 38.4 (C-R′), 31.0 (C-β), 30.7 (C-ω), 30.2 (C-ω′), 28.1 (C-γ′), 27.8 (C-β′), 25.3 (C-γ), 25.0 (CH3CONHs). 5-{5′[5′′-(Trifluoroacetamido) pentanecarboxamido]pentanecarboxamido}pentyl β-LN (8) was synthesized by the addition of spacer parts to 7 in a similar manner. Compound 8 was obtained in a total yield of 64% (260 mg). [R]D30+8.3° (c 0.1, water). FAB-mass: m/z 791 [M + H]+ (matrix: glycerol). 1H NMR (CDCl3, 500 MHz): δ 4.40 (d, 1H, J1,2 7.9 Hz, H-1), 4.38 (d, 1H, J1′2′ 7.3 Hz, H-1′), 3.87 (dd, 1H, J5,6b 2.5, J6a,6b 12.5 Hz, H-6b), 3.84-3.81 (3H, H-Rb, H-4′, H-6a), 3.74 (dd, 1H, J5′,6′b 7.7, J6′a,6′b 11.7 Hz, H-6′b), 3.67 (dd, 1H, J5′,6′a 4.3, J6′a,6′b 11.7 Hz, H-6′a), 3.64-3.59 (4H, H-5′, H-4, H-3, H-2), 3.54-3.48 (2H, H-3′, H-2′), 3.46 (m, 1H, H-Ra), 3.41 (m, 1H, H-5), 3.26 (t, 2H, H-ε′′), 3.16-3.08 (4H, H-ε′, H-ε), 2.16 (4H, H-R′′, H-R′), 1.96 (s, 3H, CH3CONHs), 1.63-1.52 (8H, H-β′′, H-β′, H-β, H-ω′′), 1.46 (4H, H-ω′, H-ω), 1.31 (6H, H-γ′′, H-γ′, H-γ). 13C NMR (CDCl3, 500 MHz): δ 175.7 (-CONH′-), 175.6 (-CONHs), 173.5 (CH3CONHs), 158.8 (CF3CONHs), 115.6 (CF3CONHs), 104.1 (C-1′), 102.0 (C-1), 79.9 (C-4), 76.3 (C-5′), 75.6 (C-5), 73.8 (C3′), 73.3 (C-3), 71.9 (C-2′), 70.4 (C-R), 69.5 (C-4′), 61.9 (C-6′), 61.2 (C-6), 56.0 (C-2), 40.2 (C-ε′′), 39.9 (C-ε), 39.8 (C-ε′), 36.63 (C-R′), 36.55 (C-R′′), 29.5 (C-β), 29.4 (C-ω, C-ω′), 28.8 (C-ω′′), 26.9 (C-γ′′), 26.7 (C-β′′), 26.1 (C-γ′), 25.9 (C-β′), 23.8 (C-γ), 23.0 (CH3CONHs). Synthesis of Various Aminoalkyl Di-, Tetra- And Hexasaccharide Glycosides. 5-AP-β-LN 9, 5-AP-β-LNnT 10, 5-APβ-(LN)2 11, 5-AP-β-(LNβ1,3LNnT) 12, 5-AP-β-(LN)3 13, 5-(5′aminopentanecarboxamido)pentyl β-LN 14, and 5-[5′-(5′′aminopentanecarboxamido)pentanecarboxamido]pentyl β-LN 15 were synthesized from 2-8 by alkaline hydrolysis according to our previously reported method (6).


Synthesis of Asialoglycopolypeptides with γ-PGA Backbones. The amino functions of oligosaccharide glycosides (9-15) were coupled with the carboxyl group of γ-PGA by a condensation reaction. DSs of the substituted residues in glycopolypeptide were adjusted to 31-49% (Table 1). Poly[LNβO(CH2)5NHs/γ-PGA] 16, poly[LNnTβ-O(CH2)5NHs/γ-PGA] 17, poly[(LN)2β-O(CH2)5NHs/γ-PGA] 18, poly[LNβ1,3LNnTβO(CH2)5NHs/γ-PGA] 19, poly[(LN)3β-O(CH2)5NHs/γ-PGA] 20,poly[LNβ-O(CH2)5NHCO(CH2)5NHs/γ-PGA]21,andpoly{LNβO[(CH2)5NHCO]2(CH2)5NHs/γ-PGA} 22were prepared from 9-15, respectively, according to our previously reported method (6). Synthesis of Sialoglycopolypeptides with γ-PGA Backbones. R2,3-Sialoglycopolymers 23-29 were synthesized from 16-22, respectively, using rat recombinant R2,3-(N)-sialyltransferase. In addition, R2,6-sialoglycopolymers 30-36 were also synthesized from 16-22, respectively, using rat recombinant R2,6-(N)-sialyltransferase according to our previously reported method (6). Focus-Forming Assay. Virus titers were determined by focus-forming assay using Madine-Darby canine kidney (MDCK) cells. MDCK cells were seeded at 1.5 × 104 cell/well in 96well plates and cultured in MEM supplemented with 10% fetal bovine serum at 37 °C. After removal of medium, virus solutions were serially diluted with serum-free DMEM and inoculated onto the plates. The cells were then incubated for 30 min at 34.5 °C. After removing the virus solution, overlay medium (DMEM containing 0.5% (w/v) tragacanth gum (Wako, Osaka) and 0.2% (w/v) bovine serum albumin) was added, and the plates were incubated at 34.5 °C for 20-24 h. The cells were then fixed with methanol. Infectious foci were detected with mouse antinucleoprotein (NP) monoclonal antibody (clone H16L-10-4R55, ATCC) as the primary antibody and HRPconjugated goat antimouse immunoglobulin as the secondary antibody. Detected foci were counted under a light microscope. Virus infectivity was determined as focus-forming units (FFU). Inhibition of Influenza Virus Infection by Glycopolypeptides. Influenza viruses were mixed on ice with or without glycopolypeptides at the indicated concentrations. The virusglycopolypeptide mixtures (50 µL) were then inoculated for 30 min at 34.5 °C onto MDCK cells grown in 96-well plates. After washing three times with serum-free DMEM, overlay medium was added and plates were incubated at 34.5 °C for 20-24 h. Virus infectivity was then determined by focus-forming assay as described above. The optimal titer of inoculated virus was predetermined such that more than 80-100 foci appeared per well. Each experiment was performed in triplicate wells. Significant analysis of means of focus numbers between control and treatment was performed by t test. Solid-Phase Binding Assay. The assay measuring direct binding activity of influenza virus to glycopolymers was performed by our method (9, 31). Glycopolymers were briefly immobilized on polystyrene Universal-Bind microplates (Corning) under an ultraviolet irradiation method. After blocking with PBS containing 2% bovine serum albumin, the plates were incubated in solutions containing influenza virus in PBS (27 HAU of A/WSN/33) at 4 °C overnight. After five washes with ice-cold PBS, the plates were incubated in a substrate solution containing 40 µM of 2′-(4-methylumbelliferyl)-R-D-N-acetylneuraminic acid in PBS at 37 °C for 1 h. The reactions were terminated by the addition of a 500 mM carbonate buffer (pH 10.2). Fluorescence was measured at 355 nm (excitation) and 460 nm (emission). The direct virus-binding activity was determined by the quantity of 4-methylumbelliferon released by viral neuraminidases. Influenza Virus Infection in Mice. Influenza A/WSN/33 virus was used in this study to evaluate antiviral effects of

542 Bioconjugate Chem., Vol. 20, No. 3, 2009 Scheme 1. Enzymatic Synthesis of LN-Repeating Tetra- And Hexasaccharide Glycosides

synthetic glycopolymers as described previously (31). Briefly, groups of six female BALB/c mice (five weeks old, 14-18 g) were inoculated intranasally with 25 µL of premixes containing viral suspension (2.0 × 104 pfu/mL) and glycopolymers at the indicated concentrations. Control was treated with 25 µL of viral suspension and asialo-glycopolymers. Mice were observed for 13 days after infection. Anti-influenza efficacy was determined by prevention of loss of body weight and protection against lethality caused by influenza virus infection. The significance of differences in the ratios of survival mice between groups treated with asialo- and R2,3-sialoglycopolymers was analyzed by Log-rank test (32) on Kaplan-Meier survival curves (33). Animal care and experiments were performed in accordance with the guidelines for the care and use of laboratory animals of the Shizuoka University. Hemagglutination Inhibition Assay. The hemagglutination inhibition (HI) assay was carried out using 96-well microtiter plates as described previously (22). Phosphate-buffered saline (PBS, pH 6.5) was used as a dilution buffer. Virus suspension (22 HA units in 0.025 mL of PBS) was added to each well containing the synthetic glycopolypeptides (200 to 0.024 nM) in a 2-fold serial dilution in dilution buffer. After incubation for 1 h at 4 °C, 0.05 mL of 0.6% (v/v) guinea pig suspension erythrocytes was added to the plates, and allowed to settle for 2 h at 4 °C. The maximum dilution of the samples showing complete inhibition of hemagglutination was defined as the HI titer.

RESULTS Convenient Synthetic Route to a New Type of Artificial Glycopolymer. Synthesis of Oligosaccharide Glycosides Carrying LN Repeats. We have recently reported that a condensation reaction between lactose (Lac) or LN and 5-trifluoroacetamido-1-pentanol can be catalyzed by cellulase from Trichoderma reesei to obtain 5-TFAP-β-Lac (1) and 5-TFAPβ-LN (2), respectively (6). O-Linked oligosaccharide glycosides carrying LN repeats were then derived from the resulting products (Scheme 1). Thus, consecutive additions of β1,3 linked GlcNAc and β1,4 linked Gal to respective 1 and 2 acceptors led to the synthesis of tetrasaccharide glycosides 3 and 4 carrying LNnT and tandem LN repeats, respectively, by using either β3GnT or β4GalTI. GlcNAc and Gal residues can then be added to the resulting glycosides 3 and 4 in a similar way to

Ogata et al. Scheme 2. Synthesis of More Spaced Disaccharide Glycosides

form hexasaccharide glycosides 5 and 6 carrying LN repeats. The target products 3-6 were obtained in high yields of 90-75%, based on the acceptors. Synthesis of More Spaced Disaccharide Glycosides. The aglycon trifluoroacetamido group of 2 was deacylated to 5-aminopentyl β-LN (5-AP-β-LN) 9 by hydrolysis in an alkaline solution as shown in Scheme 2. The resulting amino function was coupled to 6-trifluoroacetamidohexanoic acid with DIPCD and HOBt in DMF to produce 7 with tandem alkylamido repeats in the aglycon moiety. The resulting glycoside 7 was then subjected to the condensation reaction to produce 8 with triplet alkylamido repeats. The longer spaced disaccharide glycosides 7 and 8 were easily purified with a silica gel column, in yields of 59% and 64% based on the corresponding glycosides, respectively. Synthesis of Asialoglycopolypeptides. The resulting oligosaccharide glycosides with different glycans in the array were similarly deacylated to the corresponding amino group by alkali treatment as mentioned above (Scheme 3). The amino function of the resulting oligosaccharide glycosides 9-15 was reacted with the carboxyl groups of γ-PGA in the presence of the condensation reagents BOP and HOBt, as described earlier. In the synthesis, γ-PGA of 990 kDa was used as the polypeptide base of the glycopolymers. The reaction solution was applied to a column of Sephadex G-25 M PD-10 to separate the glycosylated γ-PGA from the low molecular weight reactants. The degree of substitution (DS) in mole fraction of the substituted residues in the asialoglycopolypeptides 16-22 was calculated by two methods: 1H NMR and chemical analyses. The 1H NMR results were based on the relative intensities of signals due to peptide methylene protons and those due to spacer-linked methylene protons in the aglycon. The chemical analysis procedure used the method of Yasuno et al. (29). The DS in those asialoglycopolypeptides was limited to the extent of 30-50% by controlling the coupling conditions of the aminated spacer-linked glycosides with γ-PGAs (Tables 1 and 2). Synthesis of Sialoglycopolypeptides. A series of asialoglycopolypeptides with different glycans in the array were sialylated to 23-29 carrying SAR2,3Gal and 30-36 carrying SAR2,6Gal by R2,3- and R2,6-SiaTs, respectively (Scheme 3). After separation through a column of Sephadex G-25 M PD-10, the target glycopolypeptides were obtained. The structures of the synthesized glycopolymers were confirmed by 1H and 13C NMR analyses with reference to our previous report (6). The DS of sialyl sugar derivatives in the sialoglycopolypeptides was determined by 1H NMR and chemical analyses, such as those of asialoglycopolypeptides mentioned above. The structures of sialoglycopolypeptides 27 and 34 as representative examples were analyzed from the corresponding 1H NMR spectrum. In the 1H NMR spectrum of 27, characteristic signals at δ 2.76 (dd, 1H, J3′′′′′′ax, 3′′′′′′eq 11.6,



Scheme 3a

a (1) Chemoenzymatic synthesis of LN-repeating and long-spacer linked aminoalkyl oligosaccharide glycosides, (2) coupling of the resulting glycoside with γ-PGA, and (3) sialylation to highly water-soluble glycopolypeptides carrying clustered identical sialooligosaccharide segments.

J3′′′′′′eq, 4′′′′′′ 3.7 Hz, H-3′′′′′′eq) and δ 1.81 (t, 1H, J 3′′′′′′ax, 3′′′′′′eq 11.6, J3′′′′′′ax, 4′′′′′′ 11.6 Hz, H-3′′′′′′ax) were assigned to the H-3′′′′′′ proton. In 34, δ 2.67 (dd, 1H, J3′′′′′′ax, 3′′′′′′eq 12.2, J3′′′′′′eq, 4′′′′′′ 4.3 Hz, H-3′′′′′′eq) and δ 1.73 (t, 1H, J3′′′′′′ax, 3′′′′′′eq 12.2, J3′′′′′′ax, 4′′′′′′ 12.2 Hz, H-3′′′′′′ax) were assigned to the H-3′′′′′′ proton. It was possible to evaluate the extent of sialylation from the integration data of the resulting proton signals (Table 2). The properties of the synthesized glycopolypeptides are summarized in Table 2. The chemical analysis was performed using the method of Sato et al. (30). The degree of substitution of neutral sugar derivatives (NS) and sialyl sugar derivatives (Sia) was based on the DP of γ-PGA, which was arbitrarily given a value of 100% (Table 2). In the 13C NMR spectra of 27 and 34, the respective C-3′′′′′ and C-6′′′′′ signals were distinguished by a downfield position with chemical shifts at δ 78.0 and δ 66.2. These data indicate that the terminal Gal residues of the sugar chains are regiospecifically sialylated in the R2-3 and R2-6 linkages. Interaction of Synthetic Glycopolypeptides with Influenza Viruses. Inhibition of Influenza Virus Infection by Glycopolypeptides. To investigate the inhibitory effects of glycopolypeptides carrying different length glycans in the array on infection of MDCK cells by influenza virus, virus and test agents were inoculated concurrently. The inhibitory activity was determined by focus-forming assay. This experiment was performed to determine whether glycopolymers act directly on viruses in the early stages of viral infection, particularly adsorption. The results are summarized in Table 3. Infection of the avian strains, A/Duck/HongKong/313/4/78 (H5N3), A/Duck/

HongKong/47/5/76 (H3N2), and A/Duck/HongKong/47/5/76 (H7N2) were inhibited by sialoglycopolymers carrying the R2,3 sialylated glycan, but not by those carrying the R2,6 sialylated glycan. Inhibition by 23, 25, and 27 decreased in a glycan length-dependent manner. Compound 23, carrying a short glycan of trisaccharide, was shown to be the most effective inhibitor among the glycopolymers tested. The polymer showed 4-10fold higher inhibitory activity than 27 carrying triplet LN repeats. The virus appears to recognize a short R2,3 sialylated glycan in preference to the long glycan. In contrast, the human virus, A/WSN/33 (H1N1) was inhibited by glycopolymers carrying long R2,3/6 sialylated glycan with LN repeats. The inhibition clearly occurred in a glycan length-dependent manner. Viral infection was strongly inhibited by 27 and 34 carrying long glycans at a very low concentration (IC50 0.06 pM). Inhibition by 27 displayed 13 × 106-fold higher activity compared to that of 23, which carries a short glycan. In this case, the sugar elongation of the inner glycan for 27 resulted in a remarkable increase of inhibition activity to enhance the virus binding. IC50 of 27 and 34 in the present study is 7.5 × 105-fold stronger than R-PGA-based glycopolypeptide carrying the same LN repeat (31). The human virus, A/Aichi/2/68 (H3N2), which has high susceptibility to compounds 32 and 34 (IC50 0.1-0.2 pM), also showed a similar pattern of inhibitory effect. Taken together, our results indicate that the length of the sialylated glycan critically influences binding to the human virus, which is favored by long glycan


Ogata et al.

Table 3. IC50 of Sialoglycopolypeptides Carrying LN-Repeating Glycoside in Focus-Forming Assays of Human and Avian Influenza Viruses IC50a (Sia nM) R2,3-sialopolypeptides virus

subtype

b

c

23 (single )

25 (tandem)

R2,6-sialopolypeptides 27 (triplet)

30 (single)

32 (tandem)

34 (triplet)

Avian A/Duck/HongKong/313/4/78 A/Duck/HongKong/24/76 A/Duck/HongKong/47/5/76

H5N3 H3N2 H7N2

10 20 80

40 60 200

A/WSN/33 A/Aichi/2/68

H1N1 H3N2

800 2

0.1 (pM) 0.4

100 100 300

N.E.d N.E. N.E.

N.E. N.E. N.E.

N.E. N.E. N.E.

0.06 (pM) 0.04

0.02 3 (pM)

1 (pM) 0.1 (pM)

0.06 (pM) 0.2 (pM)

Human

a Concentration of Sia units. b Compound no. c Single, tandem, and triplet means respective LN repeats. d No effect means that inhibitory effect was not observed up to 300 µg/mL of glycopolymers.

units such as LN repeats. As controls, asialoglycopolymers did not affect the virus binding. Direct Binding ActiVity of Influenza Virus to Glycopolypeptides. For further investigation, we directly determined the binding activity of glycopolymers to human virus by a solidphase virus binding assay. We have recently established a direct binding assay with artificial glycopolymers for human and avian influenza viruses (9, 31). In this study, we applied the assay for determination of direct binding dynamics of artificial glycopolymers carrying long glycan units such as LN repeats to human virus, A/WSN/33. The binding signals were detected when R2,3/6-sialoglycopolymers, but not asialoglycopolymers (data not shown), were used as ligands. As the concentration of sialoglycopolymers with LN repeats increased, the signals from human virus bound to glycopolymers increase in a dose-dependent manner regardless of the core carbohydrate structure. The inhibition by R2,3sialopolymers 23, 25, and 27 carrying single, tandem, and triplet LN repeats increased in a glycan length-dependent manner (Figure 1A). Also, the R2,6-sialopolymers 30, 32, and 34 with different LN repeats showed a similar pattern of inhibitory effect as those of the R2,3-sialopolymers (Figure 1B). These results were comparable with those obtained by in vitro infection experiments. Taken together, the human viruses bind exclusively to sialoglycopolymers carrying elongated core carbohydrates with higher affinity. Protection of Mice by Intranasal Administration of Glycopolypeptides from Death Induced by A/WSN/33 Infection. We examined the protective effects of sialoglycopolymers (23 and 27) with single and triplet LN repeats in a murine model of respiratory tract viral infection with A/WSN/33 strain. Asialoglycopolymer (20) was used as control to examine influenza infection. Groups of mice were administered glycopolymers at five weeks old. Compound 27 with triplet LN repeats, administered at a dose of 3 nmol/kg corresponding to approximately 15.6 mg/kg, reduced mortality below 0% on day 13 postinfection (Figure 2A). In contrast, compound 23 with a single LN, administered at 3 nmol/kg did not significantly affect lethality in mice induced by infection with A/WSN/33 as compared to administration of asialoglycopolymer 20 (Figure 2A). In mice treated with compound 27 with triplet LN repeat, but neither compound 23 with a single LN or asialoglycopolymer 20 at a dose of 3 nmol/kg, we also observed improvement in the weight gain of mice surviving to day 13 postinfection (Figure 2B). Thus, sialoglycopolymers with an elongated LN chain as core structure showed a significant protective effect in this murine model of influenza virus infection. The protective effect of this glycopolymer is comparable to that observed in vitro (Table 3). Glycopolymer-related toxicity was not observed in any of the groups examined. Hemagglutinin Inhibition Assay of the Glycopolypeptides Carrying Spacer-Linked Alkylamido Repeats as Sugar Mimet-

Figure 1. The direct binding activity of human influenza virus, A/WSN/ 33 (H1N1), to glycopolypeptides carrying a single, tandem, or triplet LN-repeating glycoside was determined by our method (9, 31). (A) SAR2,3-glycopolymers, (B) SAR2,6-glycopolymers.

ics. Various substances carrying sialooligosaccharide glycosides are known to inhibit hemagglutination by binding to viral HA (6). We tested the inhibitory effect of the glycopolypeptides with different glycans and spacers in the array on hemagglutination using human and avian influenza viruses. The hemagglutination inhibition activity toward glycopolypeptides showed a distinct difference between the avian and human strains as in Table 4. The avian viruses reacted with 23, 25, and 27 carrying R2,3 sialylated glycans (Table 4A and Figure 3A). Inhibition by the sialopolymers tended to decrease in a glycan lengthdependent manner. Furthermore, we prepared different length spacer linked polymers 28, 29, 35, and 36 carrying a trisaccharide R2,3/6 motif (SAR2,3/6LN) in order to elucidate whether the inner sugar should be replaced by a nonsugar structural unit, such as an alkylamido group. The avian virus A/Duck/HongKong/313/4/78 was not greatly affected to inhibition by replacement of alkylamido group (28 and 29) by sugar unit (the corresponding 25 and 27). Thus, the R2,3-sialoglycopolypeptides carrying tandem and triplet alkylamido repeats showed a similar pattern of inhibitory effect to those of the R2,3-


Figure 2. Effect of glycopolymer influenza virus (AWSN/33) infection in mice was performed as described in the Experimental Section. Graph (A) indicates Kaplan-Meier survival curves (*P < 0.01, Log-rank test). Mice were intranasally inoculated with viral suspension and asialo(20, triplet) or SAR2,3- (23, single; and 27, triplet) glycopolymers at a dose of 3 nmol/kg. Graph (B) shows body weight dynamics of mice.

sialoglycopolypeptides carrying tandem and triplet LN repeats (Table 4B and Figure 3A). In contrast, the human viruses preferentially bound not only to long R2,3/6 sialylated glycan 27 (26) and 34 (33) with LN repeats in the receptors, but also to more spacer-linked glycan (Table 4A). For the human virus, A/Aichi/2/68, inhibition by R2,6-sialopolymers 30, 31, and 33 carrying a single LN, LNnT, and LN-LNnT unit increased in a glycan length-dependent manner (Figure 3B). This result was consistent with in vitro and in vivo infection experiments, suggesting that enhancement of the binding affinity of glycopolymers with viral HA is accompanied with the extension of LN repeats as the core structure. Furthermore, inhibition by the human virus, A/Aichi/2/68, was also strongly inhibited by R2,6sialopolymers 35 (IC50 0.73 µM) and 36 (IC50 1.5 µM) carrying tandem and triplet alkylamido repeats with 4-8-fold higher affinity compared to 30 carrying a single alkylamido spacer (Table 4B). The inhibitory activity of 35 and 36 gave somewhat similar values to that of 31 and 33 carrying tandem and triplet LN repeats (Figure 3B). It indicates that a spaced tandem/triplet alkylamido repeat is a good mimetic of a tandem/triplet LN repeat. Indeed, 35 and 36 displayed 21-43-fold higher affinities for the viral hemagglutinins relative to the control fetuin (IC50 31.5 µM), which binds both avian and human types of influenza virus due to the presence of two different SA linkages, R2,3 and R2,6. As controls, asialoglycopolymers did not affect influenza virus-mediated hemagglutination (data not shown).

DISCUSSION The binding specificity of influenza viral HA involves both recognition of the SA linkages as well as the core glycan


sequence (8, 9, 25, 34, 35). We therefore designed a series of sialoglycopolypeptides to act as potential polymeric inhibitors of infection by avian and human influenza viruses in order to analyze the structure-activity relationship on the viral HA. Each of these compounds comprised three parts: glycan, spaced-linker, and polypeptide backbone. Here, we propose a convenient synthesis of asialo-type polymers with LN repeats starting LN or Lac. Schemes 1-3 represent the pathway via the following three steps: (1) enzymatic and chemical glycosylations of the respective spaced-disaccharide (LN/Lac) glycosides; (2) enzymatic sugar elongation of the LN unit to the resulting disaccharide glycosides by consecutive use of β3GnT and β4GalT to produce tetra- and hexasaccharide glycosides with tandem and triplet LN repeats; (3) coupling of the resulting glycoside to γ-PGA. High molecular-weight γ-PGA was used as a backbone for each glycopolymer. As an alternative design, the resulting disaccharide glycosides were converted to long spacerlinked disaccharide glycosides with tandem and triplet alkylamido repeats by consecutive coupling reactions. These units were then coupled to γ-PGA. The DS (%) of asialoglycan in the resulting glycopolypeptides was adjusted to a proportion of around 40% by controlling the coupling reaction (Table 1). Finally, a series of asialoglycopolypeptides were sialylated to obtain glycopolypeptides carrying R2,3/6 sialylated glycans in the array by the use of R2,3- and R2,6-SiaTs. We have already reported that γ-PGA-based glycopolypeptide carrying a single LN unit can be used as an acceptor of a CMP-Neu5Ac donor by utilizing rat recombinant R2,3/6-SiaTs (6). This method was applicable to the sialylation of the present asialoglycopolypeptides. The rat recombinant R2,3/6-SiaTs regiospecifically catalyzed sialylation to the terminal Gal residue in the short asialoglycan as well as the Gal in the long asialoglycan, such as LN repeats (36). As a result, sialylation of the asialo-portion of the glycopolymers was almost quantitative at around 40%, whose proportion is based on the optimum result on inhibition of viral infection (Table 2). To date, infections by influenza viruses have been successively inhibited by sialoglycopolymers carrying short glycans, such as sialyllactose and sialyl LN, no matter how short the glycan chain (12). However, recent studies have demonstrated that the length of the glycan beyond a trisaccharide critically influences viral HA binding contacts (9, 25, 34). In the present study, the contribution of the asialo portion in the sugar chains was examined by in vitro or in vivo infection experiments, solid-phase binding assay, and hemagglutination inhibition assay using glycopolypeptides carrying different lengths of glycan or spacer-linked glycan in the array. Five strains of avian and human influenza viruses were used as models for studies aimed at identifying the HA glycan binding specificity. First, the in vitro infection experiments (focusforming assay) examined that artificial glycopolypeptide effectively inhibited cellular infection by avian and human strains (Table 3). The avian viruses specifically recognized a short R2,3 sialylated glycan to an LN/Lac motif in a trisaccharide. In addition, the virus did not bind to the terminal SA-linked R2,6 of the LN/Lac motif. In contrast to this result, human viruses preferentially bound to a long R2,3/6 sialylated glycan with LN repeats that are at least pentasaccharide in length. These results were further supported by solid-phase binding assays that elongation of core carbohydrate portion markedly enhances binding affinity of R2,3/6-sialopolymers with the human viral HA (Figure 1). A possible explanation could lie with the difference in the conformation of the terminal trisaccharides resulting from sialic acid linkage in the cocrystal structures (25, 34). Chandrasekaran et al. have proposed that the cone-like and umbrella-like topologies are governed by conformation of


Ogata et al.

Table 4 A. Inhibition of Human and Avian Influenza Virus Hemagglutination by Sialoglycopolypeptides Carrying LN-Repeating Glycoside IC50a (Sia µM) R2,3-sialopolypeptides subtype

23 (singlec)

A/Duck/HongKong/313/4/78 A/Duck/HongKong/24/76 A/Duck/HongKong/47/5/76

H5N3 H3N2 H7N2

99 8.8 35

145 5.7 46

A/WSN/33 A/Aichi/2/68

H1N1 H3N2

0.22 231

0.71 (nM) 188d

virus

25 (tandem)


30 (single)

32 (tandem)

34 (triplet)

Avian 317 14 54

NDb ND ND

ND ND ND

ND ND ND

0.34 (nM) 94e

4.3 (nM) 6

1.4 (nM) 2.7f

0.34 (nM) 1.2g

Human

B. Inhibition of Human and Avian Influenza Virus Hemagglutination by Sialoglycopolypeptides Carrying Long-spacer Linked Disaccharide Glycoside IC50a (Sia µM) R2,3-sialopolypeptides virus A/Duck/HongKong/313/4/78 A/Aichi/2/68

subtype H5N3 H3N2

23 (singlec)

28 (tandem)

99

Avian 120

231

Human 47

a

b


30 (single)

35 (tandem)

36 (triplet)

240

NDb

ND

ND

94

6

0.73

1.5

c

Minimum concentrations required for complete inhibition of hemagglutination. ND: not detected (No activity). Single, tandem, and triplet means respective LN repeats. Single, tandem, and triplet means respective alkylamido repeats. d Compound 24 was used. e Compound 26 was used. f Compound 31 was used. g Compound 33 was used.

Figure 3. Inhibitory effects of glycopolypeptides with different glycans (a and c) and spacers (b and d) on hemagglutination using two strains of influenza viruses: (A) Avian strain A/Duck/HongKong/313/4/78 (H5N3) and (B) human strain A/Aichi/2/68 (H3N2). (A) a and b: SAR2,3-polymer 23 carrying a short glycan and spacer, was shown to be the most effective inhibitor. (B) c: Inhibition by SAR2,6-polymers 30, 31, and 33 carrying single LN, LNnT, and LN-LNnT units increased in a glycan length-dependent manner. d: Inhibition by SAR2,6-polymers 30, 35, and 36 carrying single, tandem, and triplet alkylamido repeats increased in a spacer length-dependent manner.

the R2,3 and R2,6 sialylated trisaccharides, respectively (25). In the umbrella-like topology, longer R2,6 oligosaccharides

(at least pentasaccharide residues) favor this conformation (25). Thus, in the umbrella-like topology, which is unique


to R2,6, the specificity for a kink motif of a least five sugar residues long could indicate that a kinked sugar chain of comparable residues is able to reach into the binding locus on the HA surface. This concept was also the case for the present viral HA binding. Therefore, whereas the avian viral HA interactions with the glycans is characteristic of short R2,3 glycans, such as single LN/Lac, human viral HA interactions involve R2,6 and is typically adopted by long glycans with multiple repeating LN units. To examine the inhibitory activity of sialoglycopolymers carrying multiple LN repeats, we administered glycopolymers to mice as an infection model. After inoculation with A/WSN/ 33, virus invades and propagates in various tissues, particularly in the lung and brain. As a result, mice develop mortality and typical morbidity, such as loss of movement and food consumption, resulting in loss of weight. Compound 27 carrying sialylated triplet LN repeats, but not with compound 23 carrying sialylated single LN, effectively protected mice from lethal challenge of A/WSN/33. At 13 days postinfection, only mice administered 3 nmol/kg of compound 27 all survived. No toxic effect was observed in mice inoculated with the same concentration of glycopolymer alone. This observation was consistent with the results of in vitro infection experiments. In vitro and in vivo infection experiments clearly showed that the extended structure of the glycan enhanced the inhibitory activity. Furthermore, we tested the inhibitory effect of glycopolypeptides designed by replacement of nonsugar glycomimetics in the asialo portion of the molecule in order to identify the specific SA motifs recognized by influenza virus. The mimetics under discussion here, such as the spaced sugars, were anticipated to act as functional sugar mimetics, which typically display the activity of a carbohydrate. However, the relationship between the viral binding activity and the core determinant of the glycan is still unclear. Acyclic spacers were used to replace sugar units in a rather flexible way. Long spacer-linked polymers carrying alkylamido repeats linked to a terminal trisaccharide motif (SAR2,3/6LN) were prepared, based on the assumption that the seven-atom spacer of pentylamido group would be a mimetic for two glycosyl units. Interestingly, the mimetics turned out to be strong viral inhibitors against the human strain A/Aichi/2/ 68 by hemagglutinin inhibition assay. Thus, in the human virus, 35/36 carrying a tandem/triplet alkylamido repeat showed 2-4-fold higher inhibitory potency over 31/33 carrying a R2,6 sialylated long glycans. The length of the alkylamido chain linked to the terminal SAR2,3/6LN appeared to be critical for recognition by the human viral HA. A tandem alkylamido repeat constituted the minimum length of the bound spacer-linked glycan. Whitesides and colleagues reported that the crucial factors for influenza virus inhibition by conventional SA-conjugated glycopolymers are the cluster of SA units, the molecular weight, electrostatic considerations, presence of bulky/hydrophobic groups, and steric stabilization of the glycopolymer (17). Numerous types of synthetic multivalent glycoconjugates have been synthesized with regard to the multivalency principle occurring in the viral HA interactions with glycans. However, contribution of the asialo portion of the glycopolymer on inhibitory activity has not been extensively examined, because the length of glycan was limited to mono-, di-, and trisaccharides. Our study indicates that replacement of a part of the glycan by a nonsugar structural unit, such as pentylamido group, results in good inhibitory activity. This type of structure is suited to build up glycopolypeptides carrying more spacerlinked glycans by replacement of a long glycan. Conventional procedures for obtaining extended glycans, such as LN


repeats, are cumbersome and involve several steps. However, nonsugar glycomimetics in the asialo portion of the molecule simplifies the large-scale preparation of the glycopolypeptides. Furthermore, γ-PGA is an obvious starting point to prepare polymeric inhibitors, because it is a readily available and inexpensive polymer. Indeed, the γ-PGA backbone from B. subtilis is commercially available in large quantities. The physiological merits of R- and γ-PGAs as the backbone to a glycopolymer in terms of cytotoxicity and immunogenicity have already been investigated. We have demonstrated that sialoglycopolymers with a R-PGA backbone have relatively low immunogenicity (16). Generally, conventional glycopolymers with a backbone such as PA can be toxic (17, 20). In our previous study, acrylamide monomer, which is known to be a potent neurotoxin, and PAA-Na were found to be cytotoxic to MDCK cells under the conditions used for the tests (16). It has also been shown that sugar-bovine serum albumin (BSA) and PA conjugates act as antigens or immunogens (16).

CONCLUSION Artificial sialoglycopolymers with a γ-PGA backbone that possess a single type of SA linkage (either R2,3 or R2,6) were chemoenzymatically synthesized. These compounds were then used as a tool to study the molecular interactions between receptors and viruses. Replacement by nonsugar glycomimetics in the asialo portion of the molecule made it possible to generate strongly active glycopolypeptides carrying more spacer-linked short glycans that were suitable for large-scale synthesis. It should be emphasized that our PGA-based polymers have an extremely high solubility in water (>10% w/v), as compared with PA-based polymers (1

Chemoenzymatic Synthesis of Sialoglycopolypeptides As

Recommend Documents