High Mannose-binding Lectin with Preference for the Cluster of α1–2-Mannose from the Green Alga Boodlea coacta Is a Potent Entry Inhibitor of HIV-1 and Influenza Viruses*

The complete amino acid sequence of a lectin from the green alga Boodlea coacta (BCA), which was determined by a combination of Edman degradation of its peptide fragments and cDNA cloning, revealed the following: 1) B. coacta used a noncanonical genetic code (where TAA and TAG codons encode glutamine rather than a translation termination), and 2) BCA consisted of three internal tandem-repeated domains, each of which contains the sequence motif similar to the carbohydrate-binding site of Galanthus nivalis agglutinin-related lectins. Carbohydrate binding specificity of BCA was examined by a centrifugal ultrafiltration-HPLC assay using 42 pyridylaminated oligosaccharides. BCA bound to high mannose-type N-glycans but not to the complex-type, hybrid-type core structure of N-glycans or oligosaccharides from glycolipids. This lectin had exclusive specificity for α1–2-linked mannose at the nonreducing terminus. The binding activity was enhanced as the number of terminal α1–2-linked mannose substitutions increased. Mannobiose, mannotriose, and mannopentaose were incapable of binding to BCA. Thus, BCA preferentially recognized the nonreducing terminal α1–2-mannose cluster as a primary target. As predicted from carbohydrate-binding propensity, this lectin inhibited the HIV-1 entry into the host cells at a half-maximal effective concentration of 8.2 nm. A high association constant (3.71 × 108 m−1) of BCA with the HIV envelope glycoprotein gp120 was demonstrated by surface plasmon resonance analysis. Moreover, BCA showed the potent anti-influenza activity by directly binding to viral envelope hemagglutinin against various strains, including a clinical isolate of pandemic H1N1-2009 virus, revealing its potential as an antiviral reagent.

Red algal lectin ESA-2 from Eucheuma serra is structurally and evolutionarily related to the cyanobacterial lectin OAA (15). Both lectins exclusively recognize high mannose-type N-glycans with extremely high affinity (association constant (K A ) ϭ ϳ10 8 M Ϫ1 ) but do not recognize monosaccharides or small oligomannoses (12,15). They also inhibit the HIV entry into the host cells with EC 50 values of low nanomolar range by directly binding to envelope gp120 (12).
In addition to inhibiting HIV, some high mannose-binding lectins (e.g. CV-N) show a broad range of antiviral activity against influenza virus (16), Ebola virus (17), human herpesvirus 6 (18), and hepatitis C virus (19). GRFT has been demonstrated to inhibit cytotoxic effects of the corona virus that causes severe acute respiratory syndrome (20). Currently, the emergence of two influenza virus strains, swine-origin influenza virus (H1N1-2009) and the highly pathogenic avian influenza virus (H5N1), has become a global threat to public health. Therefore, new anti-influenza agents are in great demand to confront the emergence of highly pathogenic mutants that acquired the ability to transmit human to human.
Previously, a novel lectin (BCA, previously declared as boonin) was isolated from the green alga Boodlea coacta, and its biochemical features were partially characterized (21). Interestingly, hemagglutination activities of BCA were strongly inhibited by glycoproteins with high mannose-type N-glycan but not by the monosaccharides tested. Here, we efficiently obtained high purity BCA using the yeast mannan-Cellulofine affinity column, and we clarified the full-length sequence of BCA by protein sequencing and cDNA cloning. Detailed oligosaccharide binding specificity of BCA and its antiviral activity against two global viruses, HIV and influenza virus, were evaluated.

EXPERIMENTAL PROCEDURES
Materials-The specimen of B. coacta was collected on the coast of Kagoshima, Japan. The algal sample was immediately transferred to the laboratory, washed, lyophilized, and ground on a ball mill to a powder. The powdered alga, which had been kept at Ϫ20°C, was used for purification of the lectin BCA. A small portion of the alga was stored at Ϫ20°C in RNAlater (Invitrogen) until used for the RNA extraction. Lysyl endopeptidase (Lys-C) and endoproteinase Asp-N were obtained from Takara Bio (Kyoto, Japan). Pyridylaminated (PA-) oligosaccharides were prepared as described previously (15).
Purification of B. coacta Lectin-The powdered alga (25 g) was stirred at 4°C overnight with 10 volumes (w/v) of 20 mM phosphate buffer (pH 7.0) containing 0.15 M NaCl (PBS). The mixture was centrifuged at 13,000 ϫ g for 30 min, and the supernatant was recovered. The residues were extracted once more with 150 ml of PBS in the same way. To both extracts combined, solid ammonium sulfate was added to attain a final concentration of 75% saturation. The mixture was kept overnight at 4°C and centrifuged at 13,000 ϫ g for 30 min. The resulting precipitate was dissolved in a small amount of distilled water and then dialyzed thoroughly against PBS. After the nondialysate was centrifuged at 13,000 ϫ g for 30 min, the supernatant was collected as a salting-out fraction.
A 3-ml portion (23.2 mg of protein) of the salting-out fraction was applied to a yeast mannan-immobilized column (10 ϫ 100 mm, Vt ϭ 7.85 ml, 1.19 mg of ligand/ml of gel) equilibrated with PBS. The affinity column was prepared as described previously (22). The column was thoroughly washed with 20 mM phosphate buffer (pH 7.0) containing 1 M NaCl and then eluted with absolute ethylene glycol. The flow rate of 0.2 ml/min was maintained during the chromatography. Fractions of 1 ml were collected and measured for absorbance at 280 nm and for hemagglutination activity. The active factions showing the hemagglutination activity (Ͼ2 5 ) were pooled, thoroughly dialyzed against distilled water, and further applied to a YMC PROTEIN-RP column (6.0 ϫ 250 mm) (YMC, Kyoto, Japan) equilibrated with 10% acetonitrile in 0.05% trifluoroacetic acid (TFA). The column was washed with the starting solvent and then eluted at a flow rate of 1.0 ml/min by a linear gradient (10 -70%) of acetonitrile in 0.05% TFA. The eluate was monitored by absorption at 280 nm and hemagglutination activity. The active fractions were pooled and dialyzed against distilled water.
Analytical Methods-Protein concentration was quantitated by the Lowry method (23) using bovine serum albumin as the standard. SDS-PAGE (24) was performed using a 15% (w/v) gel. Staining for carbohydrate was carried out using the G. P. Sensor, a carbohydrate-detection kit (J-OIL MILLS, Tokyo, Japan), as described previously (25), except that fetuin was used as a reference glycoprotein.
Hemagglutination Assay-Hemagglutination assay was performed using a 2% (v/v) suspension of trypsin-treated rabbit erythrocytes, as described previously (21). Briefly, rabbit blood preparation was washed three times with 50 volumes of saline, and the packed cells were suspended in saline to give a 2% (v/v) suspension of native erythrocytes. One-tenth volume of 0.5% trypsin in saline was added to a 2% native erythrocyte suspension, and the mixture was incubated at 37°C for 60 min. After washing three times with saline, a 2% trypsin-treated erythrocyte suspension was prepared in saline. Hemagglutination activity was expressed as a titer, the reciprocal of the highest 2-fold dilution exhibiting positive hemagglutination, or as a minimum hemagglutination concentration, the protein concentration of the highest lectin dilution exhibiting positive hemagglutination.
Amino Acid Sequence Analysis-The N-terminal amino acid sequences of intact protein and peptides generated by enzymic digestion were determined by an automated protein sequencer (Applied Biosystems 477A) connected to the 3-phenyl-2-thiohydantoin-amino acid analyzer (120A) (Applied Biosystems).
Molecular Weight Determination of Protein and Peptides-The molecular weights of native BCA, PE-BCA, and peptide fragments were determined by Finnigan LCQ electron spray ionization (ESI)-mass spectrometry (MS) (Finingan, CA).
cDNA Cloning of BCA-Total RNA was extracted from the RNAlater-treated alga using the plant RNA isolation reagent (Invitrogen). mRNA purification from the total RNA was performed using a NucleoTrap mRNA purification kit (Macherey-Nagel, Düren, Germany). Full-length cDNAs were synthesized from 200 ng of mRNA using a GeneRacer kit (Invitrogen) according to the manufacturer's instruction. The first PCR for rapid amplification of the cDNA 5Ј end (5Ј-RACE) was initiated by adding to each 0.2 l of 10-fold diluted synthesized cDNA to 8 tubes of a 9.8-l solution containing 6 pmol of GeneRacer_5Ј_Primer, 50 pmol of a degenerated primer BCA_5Ј_RACE_R1 designed from the partial BCA sequence (see supplemental Table 1 for the primer sequences), 1 l of 10ϫ Blend Taq buffer (Toyobo, Osaka, Japan), 2 nmol each of dNTP, and 0.25 units of Blend Taq DNA polymerase (Toyobo). The reaction with T Gradient Thermocycler (Biometra, Göttingen, Germany) consisted of denaturation at 94°C for 3 min, followed by 35 cycles consisting of denaturation at 94°C for 30 s, annealing at gradient temperature of 50 -64°C (2°C increments) for 30 s, and extension at 72°C for 1 min, and the final extension step at 72°C for 5 min. The PCR products in 8 tubes were pooled and then diluted to 100-fold. The nested PCR was performed by the same method, except that 0.2 l of the dilution was used as a template and 2 pmol of GeneRacer_5Ј_Nested_Primer and 50 pmol of a degenerated primer BCA_5Ј_RACE_R2 as a primer pair (supplemental Table 1). Nested PCR products were subcloned into pGEM-T Easy vector (Promega). Cycle sequencing reaction was performed using a BigDye Terminator cycle sequencing kit version 3.1 and ABI 3130xl DNA sequencer (Applied Biosystems). 3Ј-RACE was performed in the same way as 5Ј-RACE as described above, except the use of 2 pmol each of GeneRacer_3Ј_Primer and BCA_F1 (supplemental Table 1) designed from the sequence obtained by the 5Ј-RACE. At last, following the manufacturer's instruction, the full-length BCA cDNA was amplified using a high fidelity DNA polymerase KOD FX Neo (Toyobo) and a primer pair of BCA_5Ј_End_F and BCA_3Ј_End_R (supplemental Table 1), which were designed from the 5Ј-and 3Ј-terminal sequences of BCA cDNA obtained by 5Ј-and 3Ј-RACE and subcloned into pCR-Blunt II-TOPO vector (Invitrogen).
Sequence Data Processing-Homologous sequences were identified with the basic local alignment search tool program. The internal tandem repeat regions were compared with each other using ClustalW2 (26). Signal peptide region was predicted with SignalP 3.0 (27).
Centrifugal Ultrafiltration-HPLC Method-The oligosaccharide binding activity of BCA was determined using a centrifugal ultrafiltration-HPLC method as described previously (15). Briefly, 90 l of 500 nM BCA in 50 mM Tris-HCl (pH 7.0) and 10 l of 300 nM PA-oligosaccharide were mixed and kept at room temperature for 1 h. Subsequently, unreacted PA-oligosaccharides were recovered by centrifugation (10,000 ϫ g, 30 s) with Nanospin Plus (Gelman Science, MI). An aliquot of the filtrate was applied to TSKgel ODS-80TM column (4.6 ϫ 150 mm, Tosoh) and eluted with 10% methanol in 0.1 M ammonium acetate buffer at a flow rate of 1.0 ml/min in a column oven (40°C). The eluate was monitored at an excitation wavelength of 320 nm and an emission wavelength of 400 nm, and then unbound PA-oligosaccharide (O unbound ) was quantified. The amount of bound PA-oligosaccharide (O bound ) was obtained by following formula: O bound ϭ O added Ϫ O unbound , where O added represents the amount of added PA-oligosaccharide determined from the filtrate of reaction solution without a lectin. The binding activity (O bound /O added ) was calculated as a ratio of the amount of bound PA-oligosaccharide to that added.
Anti-HIV Activity of BCA-In vitro evaluation of anti-HIV activity of BCA was performed by a colorimetric assay as described previously (28). Briefly, 3-(4,5-dimethylthiazol-2-yl)-3,5-diphenylformazan (MTT, Sigma) was used to detect the viability of both HIV-1 (HTLV-IIIB strain)-and mock-infected MT-4 cells in the presence of a test compound at various concentrations.
Interaction of BCA with gp120-Direct interaction of BCA with the HIV envelope gp120 was analyzed by surface plasmon resonance analysis using a BIAcore 2000 system (GE Healthcare) as described previously (12), except that the recombinant glycosylated HIV-1 IIIB gp120 (baculovirus, ImmunoDiagnostics) was immobilized to give 300 resonance units on a carboxymethylated dextran-coated sensor chip (CM5, GE Healthcare). The data were fit globally to a simple Langmuir 1:1 binding model with local R max (maximum response) using BIAevaluation 3.1 software.
Anti-influenza Activity of BCA-Evaluation of anti-influenza activity was performed by the neutral red (NR) dye uptake assay. Various concentrations of the lectin were prepared with DMEM containing 10 g/ml trypsin in a 96-well microplate. To each well, virus was added as a multiplicity of infection of ϳ0.001 infectious particles per cell. After incubating at 37°C for 48 h, 100 l of NR dye (150 g/ml in DMEM) was added and further incubated for 2 h. NR dye incorporated into the cells was extracted by the addition of 100 l of 1% acetic acid, 50% ethanol. The color intensity of the dye absorbed by and subsequently eluted from the cells was measured at 540 nm with a microplate reader (1420 multilabel counter, PerkinElmer Life Sciences) as a factor of surviving from the virus infection.
Immunofluorescence Microscopy-Immunofluorescence staining was performed to visualize and evaluate BCA inhibition of influenza virus infection as described previously (29). Briefly, MDCK cells grown on cover glass were infected with A/Udorn/72 at a multiplicity of infection of ϳ0.001 infectious particles per cell, in the presence or absence 1 M BCA in DMEM containing 10 g/ml trypsin. After 24 h post-infection, the infected cells were fixed and visualized by incubating with mouse monoclonal anti-hemagglutinin antibody (HyTest, Turku, Finland) at 37°C for 1 h followed by fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG antibody (Anticorps Secondaires, Compiègne, France) at 37°C for 1 h. The cells were mounted using Vectashield with 4Ј,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA) and were observed under a fluorescence microscope (OLYMPUS BX51, Olympus, Japan).
Enzyme-linked Immunosorbent Assay (ELISA)-Direct interaction between BCA and viral hemagglutinin was analyzed using an ELISA as described previously (29). BCA (5 g/ml) in carbonate buffer (pH 9.6) was immobilized on ELISA plates (BD Biosciences). Following washing with phosphate-buffered saline (pH 7.4), containing 0.1% Tween 20 (PBST), the wells were blocked with 3% skim milk for 1 h at 37°C. After a further PBST washing, the wells were incubated with various concentrations of an influenza vaccine preparation (Astellas, Tokyo, Japan) enriched for virus hemagglutinin for 1 h at 37°C. After washing with PBST, the wells were incubated with mouse antihemagglutinin monoclonal antibody (HyTest) for 1 h at 37°C followed by incubation with horseradish peroxidase (HRP)conjugated goat anti-mouse IgG antibody (GE Healthcare) for 1 h at 37°C. After washing with PBST, 3,3Ј,5,5Ј-tetramethylbenzidine substrate (Sigma) was added. The reaction was stopped using 3,3Ј,5,5Ј-tetramethylbenzidine stop reagent (Sigma), and absorbance at 450 nm was measured using a microplate reader (1420 multilabel counter, PerkinElmer Life Sciences). The same ELISA was performed to test the inhibitory effect of yeast mannan on the interaction between BCA and hemagglutinin, except that yeast mannan was added to the plate coated with BCA prior to incubation with hemagglutinin (3 g/ml).

RESULTS
Purification of BCA-The lectin of B. coacta was extracted with buffer and effectively recovered as a precipitate with 75% saturated ammonium sulfate (supplemental Table 2). Total hemagglutination activity of the precipitate was higher than that of the extract, implying that some unknown inhibitor(s) coexisting in the extract, which form complexes with the lectin as seen in some plants (30), may be removed during the procedure by salting-out. In affinity chromatography of the ammonium sulfate precipitate, the active component was adsorbed to the column and eluted as a single peak with ethylene glycol, although some activity was also detected in a nonadsorbed fraction (Fig. 1A). The major active peak eluted with ethylene glycol gave a single protein band of 14 kDa, whereas the nonadsorbed fraction gave several protein bands, including 16 kDa (Fig. 1B). The protein yield of the major active peak was 4.2 mg from 25 g of the powdered alga (supplemental Table 2). The active peak was further purified by reverse-phase chromatography on a YMC protein-RP column (Fig. 1C). The finally purified lectin thus obtained was named BCA and was used for further exam-ination. The molecular masses of BCA and PE-BCA determined by ESI-MS were 13,812 and 13,919 Da and resembled that estimated by SDS-PAGE. The difference in molecular masses of BCA and PE-BCA suggests that the lectin protein contains a single cysteine residue in the molecule. BCA was negative for carbohydrate staining, indicating that it has no sugar moiety (data not shown).
Nucleotide and Amino Acid Sequences of BCA-To elucidate the complete amino acid sequences of BCA, cDNA cloning for BCA was performed. The full-length cDNA encoding BCA was isolated by 5Ј-and 3Ј-RACEs using the degenerated primers designed based on the partial sequences determined by Edman degradation of PE-BCA and its peptide fragments. The cloned nucleotide sequence of BCA and its deduced amino acid sequence are shown in Fig. 2A. The deduced amino acid sequence was almost comparable in the partial sequences determined by Edman degradation, as shown in Fig. 2B. The peptide fragments produced by proteolytic digestions (Lys-C and Asp-N) of PE-BCA were separated by reverse-phase HPLC and designated L1-L16 and A1-A15 (data not shown). The N-terminal amino acid sequences and molecular masses of these fragments were determined as well as PE-BCA, which resulted in elucidation of the partial sequences that correspond to 118 amino acid residues of the lectin molecule (Fig. 2B). In the sequencing, some peptides (A1, A4, A5, L4, and L12) were generated by nonspecific cleavage. The 30 N-terminal amino acid sequence of PE-BCA was determined as GAF(Q/K)AI-SGESGKYLSHAFAKIWLQNGYQGL, at the fourth residue of which two amino acids, glutamine and lysine, were identified, suggesting the presence of the isoforms. Comparison of cDNA and peptide sequences of BCA elucidated that the general stop codons of TAA and TAG, but not TGA, encoded glutamine  (Fig. 2B). The three tandem-repeated domains, each of which consisted of 40 -43 residues, showed sequence identity of 47.6 -62.8% to each other (Fig. 3A). No homologous gene candidates that showed high sequence similarities with BCA were found by in silico search. BCA showed no significant sequence similarity with other high mannose-type glycan-binding lectins such as CV-N, OAA, and ESA-2. Strikingly, however, BCA partially possessed the carbohydrate-binding motif of GNA-related lectins, despite having no overall sequence similarity. The sequence alignment of both BCA and GNA shows obvious disruption of GNA-subdomain III in BCA, whereas the sequences of BCA at GNA-subdomains I and II were partially conserved (Fig. 3B).
Carbohydrate Binding Specificity of BCA-The carbohydrate binding specificity of BCA was investigated by a centrifugal ultrafiltration-HPLC method. Fig. 4 shows the structure of PAoligosaccharides used in this study. Of the 42 kinds of PA-oligosaccharides tested, BCA selectively recognized 9 carbohydrate structures (oligosaccharides 14 -22) that are categorized into high mannose-type N-glycans (oligosaccharides 13-24) as shown in Fig. 5. The complex type N-glycans (oligosaccharides 1-12), hybrid-type N-glycans (oligosaccharides [25][26][27], the core structures of N-glycans (oligosaccharides 28 and 29), and oligosaccharides from glycolipid (oligosaccharides 30 -37) were incapable of binding to BCA. The binding ability of BCA to high mannose-type N-glycans differed depending on the structure of branched carbohydrate moiety. BCA showed the preference for the oligosaccharides bearing terminal ␣1-2-linked mannose(s), and the activity was increased in proportion to the increased number of ␣1-2 substitutions at nonreducing end. This observation was evident from the fact that BCA bound completely to the oligosaccharides 15 and 19 bearing three fully exposed ␣1-2-linked mannoses (Fig. 5). As for the oligosaccharides having two terminal ␣1-2-mannoses, two closely proximate ␣1-2-mannoses were likely to be preferred for BCA compared with two sterically distant ␣1-2-mannoses, because BCA completely bound to oligosaccharide 20, and the activities for oligosaccharides 16 and 21 were somewhat decreased (about 80%). Activity toward oligosaccharide 17 was much lower (25.4%) despite having two terminal ␣1-2-linked mannoses. With the oligosaccharides having only one terminal ␣1-2-mannose (oligosaccharides 14, 18, and 22), the binding activity of BCA was further decreased showing the activity around 30 -47%. In contrast, this lectin did not interact with high mannose-type N-glycans that are devoid of terminal ␣1-2-mannose as seen in oligosaccharides 13, 23, and 24. These results indicate that primary targets of BCA are the terminal ␣1-2-linked mannose(s). Moreover, no interaction was observed in mannobioses 38 -40, mannotriose 41, and mannopentaose 42 that are the main constituents of branched moiety of high mannose-type N-glycans. Interestingly, even ␣1-2linked mannobiose 38 lacks the binding ability to BCA. The results suggest that BCA exclusively recognizes nonreducing terminal ␣1-2-linked mannose and the clustering of those residues might contribute to enhance BCA affinity.
Anti-HIV Activity of BCA-In vitro evaluation of potent anti-HIV activity of BCA was determined by the conventional MTT assay using MT-4 cells. BCA inhibited the HIV-1 infection dose-dependently, with an EC 50 of 8.2 nM (Fig. 6A). Cell viability was not affected up to 100 nM, the highest dose in this experiment.
Direct Interaction of BCA with gp120-To evaluate the molecular basis of anti-HIV activity of BCA, we tested the direct interaction of BCA with a recombinant HIV envelope glycoprotein gp120 by surface plasmon resonance analysis. As shown in Fig. 6B, BCA dose-dependently bound to the gp120 immobilized on a sensor chip CM5. From the kinetic analysis, the association constant (K A ) of BCA-gp120 interaction was calculated to be 3.71 ϫ 10 8 M Ϫ1 .
Anti-influenza Virus Activity of BCA-The anti-influenza virus activity of BCA was investigated with respect to inhibiting replication in MDCK cells by the NR dye uptake assay. Ten influenza A virus strains, including laboratory-adapted strains and one influenza virus B strain, were tested for their BCA sensitivity (Fig. 7A). As shown in Table 1, BCA showed antiviral activity for most of influenza virus strains tested, even though the EC 50 varied depending on the strains. Philippines/2/82 (H3N2) was the most sensitive strain to BCA. Interestingly, BCA showed stronger inhibition against H3N2 subtypes at EC 50 values of 18.8 -74.2 nM, whereas it was much weaker for H1N1 subtypes showing EC 50 values of 79.3-1590.2 nM. The clinical isolates of the recent pandemic strain, swine-origin influenza virus, A/Oita/OU1 P3-3/09 (H1N1), was also susceptible to BCA, although the degree of susceptibility (EC 50 of 820 nM) was much lower compared with the H3N2 strains. A laboratory-adapted strain, A/PR8/34 (H1N1), was the only strain that is insensitive to BCA. This lectin was also effective against influenza B strain B/Ibaraki/2/85 at a moderate level.
Immunofluorescence Microscopy-To explore the mechanism of inhibition of influenza virus infection by BCA, the pres- ence of viral antigen in the infected cells was observed in the presence or absence of BCA using immunofluorescence microscopy. After 24 h postinfection with A/Udorn/72 (H3N2), viral antigens were detected with the specific anti-hemagglutinin antibody. Fig. 7B shows that BCA efficiently inhibited influenza virus entry into the cells, whereas the virus invaded and proliferated in the host cells in the absence of BCA. The viral ion channel (M2 protein) inhibitor, Amantadine, did not prevent virus invasion into the cells.
Direct Interaction of BCA with Influenza Viral Hemagglutinin-To examine whether BCA directly binds to the oligosaccharide on enveloped glycoprotein (hemagglutinin), ELISAs were performed using an influenza vaccine preparation, which contains hemagglutinin of A/California/7/09 (H1N1), A/Victoria/210/09 (H3N2), and B/Brisbane/60/08. As shown in Fig. 7C  (left panel), obvious binding of hemagglutinin to the immobilized BCA was demonstrated. In contrast, hemagglutinin did not bind to a reference glycoprotein (yeast mannan). The interaction between hemagglutinin and BCA was significantly inhibited by the presence of yeast mannan bearing high mannose glycans (Fig. 7C, right panel), indicating that BCA actually binds to the hemagglutinin through high mannose glycans.

DISCUSSION
High mannose-binding algal lectins are ones from the expected and promising compounds for anti-HIV or antiviral agents such as microbicides, as they exhibit exclusive specificity toward certain oligosaccharide(s) on the surface of the virus with high affinity, thereby inhibiting virus entry into the host cells. Here, we report that BCA is the first HIV-and influenza virus-inhibiting protein from the green algae, which shows strict specificity for high mannose oligosaccharide but unprecedented mode of oligosaccharide recognition. We also elucidated the primary structure of BCA and its distinctive structural features.
Several green algal species belonging to the orders Dasycladales and Siphonocladales, to which B. coacta belongs, have been found to use a noncanonical genetic code where TAA and TAG encode glutamine instead of translation termination (31)(32)(33), as well as oxymonads (34,35), diplomonads (36,37), and some ciliates (38,39). We found that BCA cDNA uses the noncanonical genetic code as shown in Fig. 2. To confirm that B. coacta itself uses the unusual genetic code, the full-length cDNA encoding the translation elongation factor-1␣ (EF-1␣), which was found to use the noncanonical genetic code in some green algae (33), was cloned from B. coacta (data not shown; DDBJ/EMBL/GenBank TM accession number AB604604). However, we did not find TAA and TAG codons in its ORF. Then we nonspecifically surveyed other genes that may use the unusual genetic code in the full-length cDNA pool derived from B. coacta (see supplemental Fig. 1 for details). As a result, we found another gene, ribosomal protein L37a, that uses the noncanonical genetic code in B. coacta (supplemental Fig. 1; DDBJ/ FIGURE 5. Binding activities of BCA to PA-oligosaccharides. Binding activity was expressed as a ratio (%) of the amount of a bound oligosaccharide to that of an added oligosaccharide. The assay was performed in duplicate for each PA-oligosaccharide, and the activity is expressed as the average value from duplicate assays. The assays were reproducible without any significant difference.
EMBL/GenBank TM accession number AB604605). Although we did not find TAA and TAG codons in putative ORF of almost all genes cloned in this experiment, they used the TGA codon as a termination, but not TAA and TAG, as well as B. coacta EF-1␣ gene. Thus, it is strongly suggested that B. coacta uses a noncanonical genetic code, and the algal species using the unusual genetic code may exist in the order Siphonocladales.
The molecular mass (13,812 Da) of BCA determined by ESI-MS was distinct from the calculated one (17,561 Da) from the deduced amino acid sequence of BCA cDNA, which consisted of 161 amino acids, excluding the signal peptide sequence. Concerning this discrepancy, the molecular mass (13,812 Da) calculated from the portion (20 -144 amino acids) of three tandem-repeated sequences coincides well with the determined mass by ESI-MS. In the sequence analyses of the peptide fragments of PE-BCA, none of fragments derived from the C-terminal region (145-180 amino acids) deduced from BCA cDNA was obtained. This suggests that the precursor of BCA may be post-translationally modified for its C-terminal truncation as seen in the other plant lectins (40), including the GNA-related lectin family (41,42), resulting in synthesis of the mature BCA consisting of repeated domains alone. The occurrence of three internal repeats and both the N-and C-terminal potential truncation regions of BCA led us to survey the similarity of this protein to GNA, which also consists of three subdomains, signal peptide, and C-terminal propeptide. Interestingly, BCA has sequence motifs similar to the carbohydratebinding site of GNA-related lectins. Nevertheless, the overall sequence similarity was quite low between BCA and GNA-related lectins at an undetectable level by the normal homology search. It should be noted that both the signal peptide and the C-terminal propeptide of GNA are necessary for trafficking to the vacuole (42). Moreover, the importance of the C-terminal propeptide of GNA to temporarily inactivate the carbohydratebinding ability in the endoplasmic reticulum has been proposed (42). The presence of a putative vacuolar targeting motif of BCA and the partially conserved GNA-like carbohydrate-binding site suggests the similar physiological role(s) of BCA.
GNA-related lectins have been distinguished either as singledomain lectins with an exclusive specificity toward mannose/ oligomannosides or as two-domain lectins that acquired a diverse carbohydrate specificity (2). From the aspects of a simple monomeric structure consisting of a single GNA-like domain and the strict specificity for high mannose oligosaccharides of BCA, this lectin more closely resembles the single-domain GNA-related lectins. Of the three subdomains with consensus amino acid sequences (QDNVY), each of which corresponds to characteristic mannose-binding sites of GNA-related lectins, putative carbohydrate-binding sites of BCA were partially conserved at subdomains I and II of GNA-related lectins, but it underwent a significant change at subdomain III. This might reflect the different specificity of BCA from other typical GNA-related lectins, but the mode of oligosaccharide recognition of BCA should be clarified by the structural analyses of BCA-oligosaccharide complexes. It is noteworthy that BCA has been predicted to have three ␣-helices within its monomeric molecule, each located in tandem repeat domains as deduced from the secondary structure (supplemental Fig. 2). This suggests that the three-dimensional structure of BCA might be distinct from the typical GNA-related lectins, which exhibit a ␤-prism structure built up of three subdomains, each consisting of four strands of antiparallel ␤-sheets (43). Moreover, it is likely that plant GNA-related lectins may have evolved through the process unrelated from BCA, because prokaryotic proteins with the GNA domain(s) share a high sequence similarity with plant GNA-related lectins (44,45) but not with BCA. Thus, the biosynthetic process of this algal lectin is an interesting target to be further investigated, including the identification of its C-terminal amino acid.
Cyanobacterial lectins, CV-N (N. ellipsosporum) and scytovirin (S. varium), show the strong anti-HIV activity at EC 50 values at a picomolar level. A red algal lectin GRFT is the strongest HIV entry inhibitor with the EC 50 of 40 pM and has The assay was performed in triplicate, and the activity was expressed as the average value from triplicate assays. B, interaction of BCA with a recombinant HIV envelope glycoprotein gp120. The interaction was analyzed by surface plasmon resonance (SPR). Each sensorgram represents the BCA binding to gp120 on sensor chip CM5. Ninety l of BCA solutions (15.6, 31.3, 62.5, 125, and 250 nM) were injected into the flow cells at 30 l/min for 3 min. The response in resonance units (RU) is plotted against time (seconds). Binding kinetics of the interaction between BCA and gp120 were calculated by fitting the data to Langmuir model for 1:1 binding. k a , association rate constant; k d , dissociation rate constant; K A , association constant; and K D , dissociation constant.
broad spectrum activity against various HIV clades (14). The cyanobacterial lectin, OAA (O. agardhii), and the red algal lectin, ESA-2 (E. serra), which are members of a new lectin family recently discovered, also inhibit HIV entry into the cells at EC 50 values at nanomolar levels (12). BCA inhibits the HIV replication at an EC 50 of 8.2 nM with the stronger activity than OAA Nuclei within the cells were stained with DAPI (ϫ200 magnification). C, interaction of BCA with an influenza envelope glycoprotein hemagglutinin. The interaction was analyzed using ELISA. BCA or a reference glycoprotein, yeast mannan, was immobilized onto the plate and incubated with an influenza vaccine that contains hemagglutinin from a mixture of influenza viruses as follows: A/California/7/09 (H1N1), A/Victoria/210/09 (H3N2), and B/Brisbane/60/08. The bound hemagglutinins were detected by the specific anti-hemagglutinin antibody as described under "Experimental Procedures." To examine the inhibitory effect of yeast mannan on interaction between BCA and hemagglutinin, the plate coated with BCA was incubated with yeast mannan prior to the incubation with influenza vaccine (3 g/ml) and was assayed the same way as above. and ESA-2. All of these lectins inhibit HIV infection, based on the same mechanism that is by the recognition of high affinity binding to the high mannose glycans of gp120 on the virus surface. Recent studies have shown that the pattern of glycosylation of HIV viral particles depends largely on the derived host cell lines rather than the strain difference and affects the HIV infectivity (46). Therefore, susceptibility of different host-derived HIV strains to BCA remains to be clarified. Besides the reason of glycomic distinctions in viral particles, the relative difference of antiviral potency among lectins might be somewhat ascribed on their diversified mode of high mannose oligosaccharide recognition. Fig. 8 shows the schematic diagram of high mannose oligosaccharide recognized by BCA. This lectin primarily targets the ␣1-2-linked mannose at the nonreducing end, and clustering of the ␣1-2-mannose significantly increases the binding affinity for BCA. It is likely that BCA recognizes the nonreducing terminal ␣1-2-linked mannose(s) rather than the internal one, because the binding activity of BCA was not altered between oligosaccharides 16 and 21, in which the only structural difference is the presence of additional ␣1-2-linked mannose at the D1 terminal (Fig. 4). This is also applicable in comparison between oligosaccharides 14 and 18, where internal ␣1-2-mannose at the D1 arm did not affect the binding activity.
It is known, for instance, that high mannose-binding proteins such as human HIV-neutralizing antibody 2G12 or CV-N prefer the D1 and D3 arm rather than the D2 arm (47,48). Actinohivin, an anti-HIV lectin from the actinomycete, which binds high mannose oligosaccharide, prefers the nonreducing terminal ␣1-2-mannose at the D1 arm and shows the highest affinity for the oligosaccharide having a combination of D1 and D3 ␣1-2-mannoses (49). MVL specifically recognizes 4 -5 units of the oligomannose core (50), whereas scytovirin prefers the D3 arm of Man 9 GlcNAc 2 (51). Both ESA-2 and OAA prefer the exposed ␣1-3-mannose in the D2 arm (12,15). In contrast to these lectins, BCA did not show preference for certain arms but rather showed the preference for clustering of ␣1-2-mannose residues at the nonreducing terminus regardless of the attached arm position. This unique oligosaccharide-binding property of BCA, the exclusive recognition of the ␣1-2-mannose cluster, might be explained somewhat by the "cluster effect" of multivalent ligands. Multivalent ligands sometimes display significant increases in functional affinity for lectins. For instance, the hepatic Gal/GalNAc receptor on the surface of mammalian hepatocytes has been shown to discriminate the cluster of galactose residues effectively (52). The number of Gal residues/ cluster and the distance between the Gal residues are critical determinants of the binding affinity. Studies on the effect of multivalent presentation of mannose ligands for concanavalin A showed the dramatic enhancement in functional affinity of mannose-containing polymers relative to the monovalent derivatives (53). Thus, in biological systems, binding affinity and specificity of lectins can sometimes be enhanced by using the multivalent saccharide ligands, resulting in the successful recognition in many biological events such as cell-cell interactions. Unfortunately, however, the biological ligand(s) of BCA have not been identified like other algal lectins. Very recently, one of the two oligomannose-binding sites of cyanobacterial lectin MVL has shown to have a catalytic activity to cleavage of chitin fragments to GlcNAc (54). Therefore, it would be important to survey the catalytic activity of algal lectins, including BCA, to gain insight into their biological significance.
The mechanisms by which high mannose-binding lectins inhibit HIV adsorption to its target cells are also applicable for other enveloped viruses. Indeed, CV-N showed a wide variety of antiviral activity for enveloped viruses such as influenza viruses, Ebola virus, human herpesvirus 6, and hepatitis C virus (16 -19). The emergence of the recent pandemic strain, swineorigin influenza virus (H1N1-2009) led us to address the potential anti-influenza virus activity of BCA, to explore the possibility as a novel antiviral agent. BCA showed potent anti-influenza  activity against most of all the influenza virus strains tested, except a laboratory-adapted strain PR/8/34 (H1N1). The recent pandemic strain, A/Oita/OU1 P3-3/09 (H1N1), was also inactivated by BCA, but the sensitivity was much lower (EC 50 of 800 nM) compared with other sensitive strains. It has been reported that the pandemic H1N1 viruses, which have only single N-glycosylation sequons on the head of hemagglutinin, were resistant to the innate immune proteins of the collectin superfamily, whereas the recent seasonal H1N1 possessing three to four sequons was sensitive to these proteins (55). The degree of glycosylation of viral hemagglutinin would be a key determinant for BCA sensitivity because the strain PR/8/34, which is devoid of N-glycosylation sequons, was resistant to BCA. Interestingly, the inhibitory activities of BCA for H3N2 subtypes were relatively higher than for H1N1 subtypes. It seems that H3N2 subtypes are more sensitive to ␣1-2-linked mannose-binding lectins than H1N1 subtypes, because this tendency was also observed for CV-N. CV-N has been reported to neutralize H3N2 subtype (A/Sydney/05/97) completely, whereas H1N1 subtype (A/Beijing/262/95) is still infectious by the same CV-N treatment (16). Similarity in virus-inactivating profiles between BCA and CV-N might be primarily ascribed on the selective recognition of the ␣1-2-linked mannose unit. As for influenza viruses, the hemagglutinin of glycoprotein appears to be a potential target for antiviral agents because certain glycosylation sites on the hemagglutinin of influenza A viruses are highly conserved and show site specificity for attached glycans (56). It has been demonstrated that the hemagglutinin 1 (HA1) subunit has a high mannose oligosaccharide at site 65, which contains almost exclusively Man 9 GlcNAc 2 , near the receptor-binding site.
As the number of amino acid sequences that encode the potential N-glycosylation site, referred to as sequons, in human influenza A hemagglutinin H3 subtype HA1 has increased over time, more recent strains might be more susceptible to lectins. This increase in the number of sequons is only found in the human H3 subtype of influenza A virus and not in H1 subtype (57). In this connection, we have observed the increased BCA sensitivity to Philippines/2/82 (H3N2) compared with the earlier isolated strain, Aichi/2/68 (H3N2). Similarly, H3N2 strains have become more sensitive to human surfactant protein D over time, in accordance with the increase of attached glycans on hemagglutinin (58). Hartshorn et al. (58) showed that sensitivity of H3N2 to surfactant protein D increased with multiplication of the glycosylation site, and the extent was greater when the glycosylation site was introduced in close proximity to the receptor-binding site. Further experiments will be needed to verify our results using more recent H3N2 strains, because the newly introduced glycosylation site does not always have high mannose oligosaccharides with ␣1-2-linked mannose units; in other words, increased sequons might not directly link to the BCA sensitivity.
Effective viricidal agents are continuously in high demand because the vaccine supply is sometimes not on time. Antibody-based medicine might have some defects as the virus mutates with high frequency by antigenic drift or antigenic shift. Furthermore, they usually could not be fully active against the different subtypes such as H3N2 or H1N1. At this point, carbohydrate moieties on the virus surface alternatively may be good targets for medicinal treatment because they exist in high abundance in most influenza strains, especially in the more recent strains. Some virus strains have been developing an increased number of glycosylation sites to evade antibody pressures by changing antigenicity (59). Therefore, it is advantageous to use lectins prophylactically, as they universally inactivate a wide range of virus strains and different subtypes. Further evaluation of BCA safety concerning cytotoxicity and inflammatory activity should be required, because some lectins such as CV-N show various side effects such as mitogenic activity and stimulation in the production of a wide variety of cytokines (60).