Preventing Influenza A Virus Infection by Mixed Inhibition of Neuraminidase and Hemagglutinin by Divalent Inhibitors

Divalent inhibitors of the neuraminidase enzyme (NA) of the Influenza A virus were synthesized with vastly different spacers. The spacers varied from 14 to 56 atoms and were relatively rigid by way of the building blocks and their connection by CuAAC. As the ligand for these constructs, a Δ4-β-d-glucoronide was used, which can be prepared form N-acetyl glucosamine. This ligand showed good NA inhibitory potency but with room for improvement by multivalency enhancement. The synthesized compounds showed modest potency enhancement in NA activity assays but a sizeable potency increase in a 4-day cytopathic effect assay. The demonstration that the compounds can also inhibit hemagglutinin in addition to NA may be the cause of the enhancement.


■ INTRODUCTION
The Influenza A virus (IAV) is a notable cause of flu. The disease can take on deadly forms as exemplified by the socalled Spanish flu in 1918 with millions of victims, 1 while IAVs continuously pose a serious threat for future pandemics. 2 Of the two envelop proteins, hemagglutinin (HA) is responsible for viral attachment to cells by sialoglycan binding, while neuraminidase (NA) cleaves off sialic acids from sialoglycan receptors, thereby enabling the release of virions from (decoy) receptors and virion mobility. 3,4 A balance between HA and NA has been identified as important for viral virulence. 3,5,6 The main prophylactic intervention against an IAV infection is the use of vaccines. Antigenic variation of seasonal IAVs is a challenge, however, this requires frequent vaccine updates and may cause mismatches with viruses in the field. 7 NA inhibitors (NAIs) have shown their value as therapeutic intervention. Potent NAIs such as oseltamivir or zanamivir are applied to reduce the illness symptoms and infectivity. 8 However, resistance to the NAIs 9 greatly hampers the effectiveness of the therapy.
Difluorosialic acids have shown promise against NAIresistant NAs, 10 but this is also true for multivalent NAIs. 11 Multivalent NAIs also showed intriguing features besides activity against resistant NAs, such as activity at much lower concentrations than zanamivir itself. Furthermore, divalent zanamivir stays in tissues much longer than monovalent. 12 None of these aspects are currently well understood.
Considering the tetrameric composition of NA proteins and the presence of ca. 40−50 copies of them on a single virion, 13 an enhancing effect of multivalent ligands does not seem surprising. In early studies, attaching a spacer to the 7-hydroxyl of zanamivir was introduced as the preferred method to maintain inhibitory activity. 14 Linking zanamivir to flexible spacers or spacers of various lengths showed that there was a clear preference for a 16 atom spacer. Both longer and shorter spacers were less effective. Considering that the distance between the four catalytic sites within an NA tetramer is typically between 40 and 50 Å, the short dimers are not bridging between catalytic sites but bridging between tetramers or even between NAs on different virions. The most striking effects were a 2000-fold enhanced infection inhibition and a ca. 100-fold enhanced lung retention of the divalent inhibitors. 15 The results were confirmed in related studies and even in animal studies, 16−20 but interestingly, no multivalency effects were observed in the inhibition of the NA enzymatic activity by a monovalent MUNANA (4-methylumbelliferyl N-acetyl-α-D-neuraminic acid) probe. An exception was a study involving higher valent versions of difluorinated zanamivir, where a 145fold enhanced NA inhibition was observed. 21 An interesting study reported tetravalent zanamivir with different lengths of the flexible poly(ethylene glycol) (PEG) spacer arms. 11 No enhancement was observed in the MUNANA assay with N2 and N9, but a 6-fold enhancement per ligand with a resistant variant of N2 was observed. A surface plasmon resonance assay revealed a 60-fold and 1400-fold binding enhancement for NA and a resistant variant, respectively. The fact that no major effects were seen in a cytopathic effect (CPE) assay, whereas an in vivo assay showed full protection, makes this system hard to comprehend. The combination of all mentioned results paints an intriguing picture with strong and useful effects of short dimers presumably by bridging between tetramers or whole viruses, while the major binding enhancements for large tetramers show likely chelation within a tetramer. A compound that combines these effects may be even more potent and could be a long-lasting chelator that could act in synergy with related HA inhibitors, such as those we recently reported. 22 Here, we report on a series of divalent NAIs with vastly different spacer lengths. The spacers used are rigidified with equatorially linked 1,4-glucose moieties, triazoles, and 1,4substituted phenyl groups. These building blocks were previously successfully applied in divalent galactose inhibitors of the Pseudomonas aeruginosa lectin LecA. 23 Instead of using zanamivir, we used an oseltamivir carboxylate mimic (OCM, Scheme 1), a Δ 4 -β-D-glucoronide, as the monovalent starting point. 24 This compound binds strongly to NA proteins but not as strong as oseltamivir carboxylate (OC, Scheme 1), the hydrolyzed version of the prodrug oseltamivir, although a direct comparison was not made. The weaker binding allows multivalency enhancements to be more easily determined, without entering sub-nano-molar potencies. A nice feature of OCM is that it can be synthesized from cheap glucosamine. By looking at NA X-ray structures, 25 attaching a spacer to the 3pentanol unit of OCM would not disrupt the binding. To this end, compound 12 and its diastereomer 13 were designed, synthesized, linked to four different spacers, and evaluated. One of the dimers was shown to inhibit infection much better than OCM and even better than OC by enhanced NA binding but surprisingly also by HA binding.

■ RESULTS
The proposed ligand 12 was synthesized as shown in Scheme 1. First, donor 4 was synthesized in three steps. The 2,2,2trichloroethoxycarbonyl (Troc) group was selected to enhance the glycosylation reaction. Using, the enantio-pure alcohol S-5, we found that glycosylation yielded either β-isomer 6 or the α-isomer, depending on the temperature of the reaction. At −78°C , it was possible to isolate the desired β-isomer 6 in 65% yield, which was converted to 8, in which the Troc group was replaced with an acetyl group. Deprotecting the hydroxyl groups under Zempleń conditions yielded 9. Next, a three-step procedure of C(6) oxidation, ester formation by MeI, and alcohol acetylation yielded 10. 1,8-Diazabicyclo(5.4.0)undec-7-ene (DBU)-mediated β-elimination gave 11, and after ester hydrolysis, 12 was obtained. Using the same procedures, the other stereoisomers 13−15 were prepared and characterized. Compound 13 was prepared by using rac-5 instead of S-5 in the glycosylation, followed by diastereomer separation at the stage of compound 8. Performing the remaining steps led to 13. Similarly, using rac-5 in the glycosylation of 4 under αisomer-producing conditions at room temperature, a diastereomeric mixture of 7a and 7b was obtained, presumably yielding the more thermodynamically stable product, which was subsequently converted to separate isomers 14 and 15, whose stereochemistry of the tail was not deciphered.
Recombinant soluble N9 protein (N9 Spain) stabilized as a tetramer using a tetrabrachion oligomerization domain 26 was used to assess the ability of the compounds (12−15) as well as the parent structure OCM to inhibit NA activity using the MUNANA substrate. The activity in the absence of inhibitory compounds was set at 100% ( Figure 1). Clearly, β-compounds 12 and 13 displayed much more inhibitory activity than αcompounds 14 and 15, with 12 being the most active compound, besides the control parent OCM. Extension of the original pentanol tail results in some reduction of potency, 6fold in this case. Nevertheless, 12 retained enough inhibitory potency; thus, it was selected for conjugation to spacers to induce multivalency effects.
In order to get an idea about the distances that spacer systems would need to cover, docking studies were performed. of zanamivir ( Figure S3) and spaced over ca. 47 Å (Figure 2), measured between the 3-pentanol tails.
Based on the above, 12 was selected as the monovalent ligand to be conjugated to divalent scaffold molecules. To cover the distance between bound OCM molecules of ca. 47 Å, four conjugates 17, 19, 21, and 24 were synthesized (Scheme 2). The number of atoms between the terminal carbons of the 3-pentanol parts is 14, 28, 42, and 56 atoms, respectively. Considering that as a crude estimation a rigid spacer may be as long in angstrom as it contains atoms, 28 this range should see some selectivity if a chelation mechanism should play a role.
The shortest spacered compound 17a was assembled from diazido-glucoside 16 28 and 12 by CuAAC conjugation. Deprotection of the central glucose unit yielded 17b. Bisazide 18 23 was also linked to 12 and yielded the longer divalent 19. Similarly, the extended bis-azide 20 was coupled to 12 to yield divalent 21. To make the longest divalent ligand of the series, a different strategy was applied. First, bis-azide 18 was monofunctionalized with 12 to yield mono-azide 22. Subsequently, 22 was coupled to bisalkyne 23 to yield the divalent 24 with the longest spacer. The synthesis of bisazido 20 is described in Scheme 3. 1-Azido galactose was converted to 26 by tert-butyldimethylsilyl (TBDMS) of the primary alcohol, followed by selective benzoylation of the equatorial hydroxyls. Mono-CuAAC conjugation with 1,4-diethynylbenzene yielded 27. CuAAC conjugation with bis-azide 28 28 yielded the symmetric 29. Introduction of azide groups, followed by sugar deprotection, yielded bis-azido spacer 20.
Compounds were first tested for their ability to inhibit NA activity using recombinant soluble tetrameric N1 and N9 proteins 26,29 and the MUNANA substrate. As the two proteins gave similar results ( Figure S2), in line with other mentioned reported systems with multivalent zanamivir units, IC 50 values were determined for the two data sets combined (Table 1). It was clear that OC was a ca. 600-fold more potent inhibitor than our monovalent 12 with these recombinant NA enzymes, while it was ca. 100-fold more potent than inhibitor 21. Next, we analyzed the inhibitory activity of the different compounds on virus particles rather than recombinant NA proteins as the source of NA activity (Table 1 and Figure 3). While the IC 50 values of all compounds were reduced in this assay, this was particularly the case for divalent 19 and 21, with multivalency enhancements of 20−30-fold in comparison with monovalent 12. As a result, OC was only ca. 25-fold more potent than compounds 19 and 21, while the potency difference between monovalent 12 and OC remained ca. 600-fold.
To assess the ability of the compounds to inhibit virus infection, a 4-day CPE assay was set up. MDCK cells in a 96well format were infected with 50 TCID 50 units of the H1N1 virus in the presence of a dilution range of the different compounds, and the lowest concentration of the compounds that could prevent cytopathogenic effects and killing of the cells resulting from virus replication was determined. Strikingly, compounds 19, 21, and 24 displayed comparable effectivity to OC in the CPE assay. The amount of OC needed to prevent cell killing was ca. 250-fold higher than the IC 50 value as determined using the MUNANA assay with the whole virus, while the difference was much smaller for compounds 19, 21, and 24 (33-, 4-, and 0.2-fold, respectively).
These results suggested that compounds 21 and 24 might have some additional activity, unlike OC, that contributes to the inhibition of virus replication. Therefore, we analyzed the ability of the different compounds to interfere with virusreceptor binding, a measure of HA inhibition. This was done using biolayer interferometry (BLI) as previously reported as a method that shows distinct activity of a viral HA protein and also its inhibition. 22,30 OC, which does not interfere with HAreceptor binding, was present to completely inhibit NA activity. 3 All compounds tested were able to inhibit virusreceptor binding with 21 having the largest effect at 10 μM (Figures 4 and S3). At this concentration, compounds 19 and 24 did not have increased inhibitory activity compared to monovalent 12.
To further explore the dual role of both HA and NA inhibition, the BLI experiment was modified. The inhibitory potency of the most potent compound 21 was studied in the presence and absence of OC. In the presence of OC alone, virus binding is observed (Figure 5a), similar to the previous experiment. In agreement with the previous experiment, 21 inhibited virus-receptor binding in the presence of OC. In the absence of any inhibitory compounds including OC, a low level of virus binding is observed, which decreased with time, resulting from the virus being released from the sensor surface in an NA-dependent manner. Interestingly, the presence of 21 alone is sufficient to prevent apparent viral binding to the  sensor, even more so than when OC is additionally present, suggesting that inhibition of HA-receptor binding is stronger when binding of 21 to NA is not in competition with OC. To confirm whether NA activity was also affected in this experimental setup, sensors were regenerated, thereby removing all but the biotinylated glycoprotein receptor. NA activity of viruses bound in the first round was then monitored by a new virus binding experiment, this time in the presence of OC (Figure 5b). When sensors had been subjected to virus binding in the first round in the absence of NAIs, virus binding in the second round was reduced as a consequence of the reduction of sialoglycan receptors in the first round by NA activity. The sensor regenerated after the action of 21 ( Figure  5a) is completely capable of full virus binding, indicating that 21 was able to inhibit the NA in that previous experiment. Collectively, these data confirm that 21 in contrast to OC inhibits not only NA activity but also HA-receptor binding.

■ DISCUSSION AND CONCLUSIONS
We successfully synthesized a series of derivatives of OCM, extended for the first time from their 5-pentanol tail, a structural feature of importance in OC. Of the four diastereomers tested, exclusively the β-isomers were able to inhibit NA, with a small preference for the S-configuration of the newly created stereocenter, that is, compound 12. The next step was to make the compound divalent. A multivalent NAI based on OC has previously been reported, but the OC moiety was linked through its carboxylate and effects were modest in NA inhibition, likely due to the fact that the carboxylate plays a role in binding to NA. 31 Conjugation of 12 to a series of four spacers with vastly different lengths but similar chemical features and rigidities also proved possible. The motif of alternating glucose, triazole, and occasional benzene rings was previously shown to lead to greatly enhanced divalent binding versus a flexible PEG structure. 28 The spacer remains mostly linear, especially for the shorter cases, as indicated by modeling 28 and from X-ray structures. 32 In the present study, the divalent conjugates of 12 showed inhibition in the MUNANA assay by both free NA protein tetramers or by the whole virus. The latter is more strongly inhibited by compounds 19 and 21. The difference between the recombinant NA and full virus assay may be explained by the different experimental conditions as the compounds may be able to interact with multiple NA tetramers on virus particles or even with HA proteins on different virions. More striking results came from the MDCK CPE assay. We were able to turn a monovalent NA ligand that is a ca. 770-fold weaker NAI than OC into a ca. 2-fold better CPE inhibitor than OC by making it divalent with a specific rigid spacer (compound 21). The MUNANA assays indicated a moderate up to 30-fold enhancement of a divalent OCM ligand versus its monovalent counterpart. However, a strong enhancement in the 4-day CPE assay was observed particularly for 21 and 24, that is, the compounds with the longest spacers, relative to OC in comparison with the MUNANA assays. Interestingly, in the two most potent compounds, the two NA ligands are separated by 42 and 56 atoms in a relatively rigid spacer. Compound 19 with 28 atoms in the linker showed high efficacy in the MUNANA assay but somewhat less so in the infection assay. Previous work with flexible spacers showed that 16 atoms was optimal. 15 Our shortest compounds 17 and 17a with 14 atoms in the spacer were clearly not optimal.
Looking for answers, HA inhibition was studied by BLI, which revealed significant HA binding by monovalent 12 and its bivalent derivatives, with 21 being the most effective. While this type of binding was not anticipated, it should not be too surprising. Compound 12 has features in common with sialic acid. Interestingly, the inhibition of HA by OCM (10 μM be observed in the presence of OC (10 μM). In our previous work on HA inhibition, low micromolar inhibition was shown for a divalent ligand in which its terminal sialic acids were separated by a similar spacer as in 21. 22 Thus, the enhanced infection inhibition of 21 may be caused by both NA and HA inhibition of the same compound. Although chelation-type binding is a major challenge when covering distances approaching 50 Å as is the case for both NA and HA, the fact that the longest spacers showed the best inhibition of infection indicates that divalent binding may be at work in our case.  It is not the first time that multivalent ligands were able to inhibit both NA and HA. S-sialosides linked to albumin showed good HA inhibition but only weak NA inhibition and were not further studied in more biological assays. 33 The principle of such dual inhibition may have potential. The present study could provide a new impetus to aim for this dual inhibition with purposely designed compounds. These could possibly take advantage of multivalency and could prove effective and yield potent anti-infective and long-lasting activity with reduced resistance.

■ EXPERIMENTAL SECTION
Chemistry. Unless stated otherwise, chemicals were obtained from commercial sources and were used without further purification. Compounds 1, 2, 34 3, 35 4, 36 and 5 37 were synthesized following the literature procedure. Solvents were purchased from Biosolve (Valkenswaard, The Netherlands). All moisture-sensitive reactions were performed under a nitrogen atmosphere. Anhydrous tetrahydrofuran (THF) was dried over Na/benzophenone and freshly distilled prior to use. All the other solvents were dried over molecular sieves (4 Å). Thin-layer chromatography (TLC) was performed on Merck precoated Silica 60 plates. Spots were visualized by UV light, 10% H 2 SO 4 in EtOH, and triphenylphosphine in THF, followed by ninhydrin. Microwave reactions were carried out in a Biotage microwave Initiator (Uppsala, Sweden). The microwave power was limited by temperature control once the desired temperature was reached. Sealed vessels of 2−5 and 10−20 mL were used. Gel filtration chromatography was performed with columns packed with Bio-gel P-2 Fine (Bio-Rad) and Bio-gel P-6 Fine (Bio-Rad) and eluted with water. Water was purified using a Milli-Q Gradient A10 Water Purification System. Lyophilization was performed on a Christ Alpha 1-2 apparatus. Analytical liquid chromatography−mass spectrometry (LC−MS) was performed on an Agilent 6560 Ion Mobility Q-TOF LC/MS using a Waters XBridge HILIC column (5 μm, 250 × 4.6 mm) at a flow rate of 0.6 mL/min. The used buffers were 50 mM formic acid in H 2 O (buffer A, pH 4.4) and CH 3 CN (buffer B). Also, UV absorption was measured at 254 nm. Purification using preparative high-performance liquid chromatography (HPLC) was performed on a Shimadzu 20A HPLC system with a Waters XBridge BEH Prep Amide column (5 μm, 250 × 10 mm) at a flow rate of 3.0 mL/min. Runs were performed using a standard protocol: 80−30% gradient buffer B in 60 min, with the same buffers as described for the analytical LC−MS. Also, analytical HPLC runs were performed on a Shimadzu automated HPLC system with a reversedphase column (Alltech, C 18 , 90 M, 5 mm, 250 L, 4.6 mm, Deerfield, IL, USA) that was equipped with an evaporative light scattering detector (PL-ELS 1000, Polymer Laboratories, Amherst, MA, USA) and a UV/vis detector operating at 220 and 250 nm. Preparative HPLC runs were performed on an Applied Biosystems workstation. Elution was effected by using a linear gradient of 5% MeCN/0.1% TFA in H 2 O to 5% H 2 O/0.1% TFA in MeCN. 1 H NMR spectra were recorded on a 400, 500, or 600 MHz spectrometer. 13     Journal of Medicinal Chemistry pubs.acs.org/jmc Article compounds (i.e., 17a, 17b, 19, 21, and 24) were >95% pure by HPLC. Compound 6. Bromide 4 (10.0 g, 18.4 mmol), 4.0 g of 4 Å powdered sieves, 8-nonyn-3-ol 5 (0.85 g, 6.04 mmol, 1.1 equiv), and 100 mL of CH 2 Cl 2 were added to a flask under argon and cooled to −78°C. AgOTf (5.70 g, 22.2 mmol) was added in portions over 40 min. The reaction was allowed to proceed for 3 days at −78°C. The reaction mixture was then quenched with iPr 2 NEt (1.44 mL, 1.50 equiv), and the solution was filtered through celite after 10 min. The crude solution was washed with concentrated Na 2 S 2 O 3 (2 × 20 mL), with concentrated Na 2 CO 3 (2 × 20 mL), and once with brine and dried over Na 2 SO 4 . Chromatography with EtOAc/petroleum ether (1:3) provided 6 as a white foam (7.18 g, 65%). 1   Compound 7a and 7b Mixture. Bromide 4 (10.0 g, 18.4 mmol), 4.0 g of 4 Å powdered sieves, and 8-nonyn-3-ol (rac-5) (0.85 g, 6.04 mmol, 1.1 equiv) were added; 100 mL of CH 2 Cl 2 was added to a flask under argon at r.t., and AgOTf (5.70 g, 22.2 mmol) was added in portions over 40 min. The reaction was allowed to proceed for 6 h at 25°C. The reaction was then quenched with iPr 2 NEt (1.44 mL, 1.50 equiv), and the solution was filtered through celite after 10 min. The crude solution was washed with concentrated Na 2 S 2 O 3 (2 × 20 mL), concentrated Na 2 CO 3 (2 × 20 mL), and 1 × brine and dried over anhydride Na 2 SO 4 . Chromatography with EtOAc/petroleum ether (1:3) provided Compounds 7a and 7b as a white foam (7.65 g, 69%).  13  Compound 8. Compound 6 (5 g, 8.3 mmol) was dissolved in 20 mL of Ac 2 O, 3 g of zinc dust was added, and the reaction mixture was stirred overnight. TLC showed completion, and the reaction mixture was filtered through celite, concentrated, and azeotroped (2 × 50 mL) of toluene. The crude material was flash-chromatographed with EtOAc/petroleum ether (3/2), which provided compound 8 as a white foam (2.9 g, 75%). 1 13  Compound 9. To a solution of compound 8 (3 g, 6.4 mmol) in anhydrous MeOH (30 mL) at 0°C under Ar was added a solution of NaOMe (1 M in MeOH, 6 mL). The reaction mixture was initially stirred for 10 min at 0°C, then warmed to r.t. naturally, and monitored by TLC analysis. After 2 h, the reaction mixture was neutralized with Amberlite IR-120(H+) resin. The resin was filtered off and washed with MeOH (2 × 20 mL); then, the filtrate was evaporated under reduced pressure to give a light-yellow syrup. This was purified by column chromatography (DCM/MeOH = 9/1) to furnish 9 (2.17 g, 99%) as a white solid. R f 0.2 (DCM/MeOH = 9/ 1). 1  Compound 10. To an ice-cooled solution of compound 9 (0.343 g, 1 mmol, 1.0 equiv) and PhI(OAc) 2 (0.21 g, 0.65 mmol, 2.5 equiv) in a mixture of CH 2 Cl 2 , t BuOH, and H 2 O (4:4:1, 10 mL) were added TEMPO (40 mg, 0.25 mmol, 1 equiv) and AcOH (3 drops). The resulting mixture was stirred vigorously overnight at 4°C, after which TLC analysis (DCM/MeOH, 9:1 v/v) indicated complete conversion of the starting material. The reaction mixture was then quenched with sat. aq Na 2 S 2 O 3 (25 mL), and aq H 3 PO 4 (5 mL, 1.0 M) was added. The mixture was evaporated under reduced pressure and then loaded on a C-18 column. Reverse-phase chromatography in water afforded a white solid after lyophilization. The obtained acid was dissolved in DMF (10 mL) under N 2 , and K 2 CO 3 (277 mg, 2 mmol, 2.0 equiv) was added to it, followed by MeI (185 μL, 3 mmol, 3.0 equiv). The reaction mixture was stirred in the dark at r.t. for 24 h. Acetic anhydride (0.6 mL, 5.1 mmol, 5 equiv) and 4-dimethylaminopyridine (DMAP) (13.4 mg, 0.11 mmol, 11%) were then added, and stirring was continued for another 12 h at r.t. Then, water was added, and the mixture was extracted with EtOAc (3 × 50 mL). The combined organic extracts were washed with H 2 O (2 × 10 mL) and brine and were dried (Na 2 SO 4 ), filtered, and concentrated under reduced pressure. The crude residue was purified by flash chromatography (EtOAc/PE 1:9 → 3:7) to give 10 (209 mg, over 3 steps, 46%) as a white fluffy solid. 1 13 13 To a solution of compound 11 (0.10 g, 0.28 mmol) in MeOH:H 2 O (1:1) at 0°C was added aq NaOH (0.5 N) until pH 13. The reaction mixture was stirred at r.t. and monitored by TLC analysis (EtOAc/MeOH/H 2 O = 8/2/1). After 16 h, the reaction mixture was neutralized with Amberlite IR-120(H+) resin. The resin was filtered off and washed with MeOH (2 × 10 mL); then, the filtrate was evaporated under reduced pressure to give a lightyellow syrup, which was purified by column chromatography (EtOAc/MeOH/H 2 O = 10/2/1) to furnish 12 after lyophilization (0.08 g, 96%) as a white solid. Compound 17a. Alkyne 12 (5 mg, 0.0147 mmol) and compound 28 38 (2.3 mg, 0.006 mmol) were suspended in a mixture of 1:1 t BuOH/water (0.1 mL) and stirred magnetically. A freshly prepared solution of sodium ascorbate (3.8 mg, 0.019 mmol) in water was added, followed by addition of a freshly prepared aqueous solution of CuSO 4 ·5H 2 O (1.6 mg, 0.006 mmol). This heterogeneous mixture was stirred vigorously overnight at r.t. The mixture was evaporated under reduced pressure to give a crude product. The latter was purified by silica gel column chromatography using H 2 O/MeOH/EtOAc 1:2:10 as the eluent to give a colorless oil which was lyophilized to afford an amorphous solid (4.1 mg, 63%). The product was further purified by RP-HPLC (C 18 Column) and the lyophilized to give pure compound 17a as a white solid. 1 13  Compound 17b. Alkyne compound 12 (5 mg, 0.0147 mmol) and azide compound 16 23 (1.47 mg, 0.006 mmol) were suspended in a mixture of 1:1 t BuOH/water (0.1 mL) and stirred magnetically. A freshly prepared solution of sodium ascorbate (3.8 mg, 0.019 mmol) in water was added, followed by addition of a freshly prepared aqueous solution of CuSO 4 ·5H 2 O (1.6 mg, 0.006 mmol). This heterogeneous mixture was stirred vigorously overnight at r.t. The mixture was evaporated under reduced pressure to give a crude product. The latter was purified by silica gel column chromatography using H 2 O/MeOH/EtOAc 1:2:10 as the eluent to give a colorless oil, which was lyophilized to afford an amorphous solid (3.6 mg, 62%). The product was further purified by RP-HPLC (C 18 column) and lyophilized to give pure 17b as a white solid. 1  Compound 19. Alkyne 12 (5 mg, 0.0147 mmol) and azide compound 18 39 (3.5 mg, 0.006 mmol) were suspended in a mixture of 1:1 t BuOH/water (0.1 mL) and stirred magnetically. A freshly prepared solution of sodium ascorbate (3.8 mg, 0.019 mmol) in water was added, followed by addition of a freshly prepared aqueous solution of CuSO 4 ·5H 2 O (1.6 mg, 0.006 mmol). This heterogeneous mixture was stirred vigorously overnight at r.t. The mixture was evaporated under reduced pressure to give a crude product. The latter was purified by silica gel column chromatography using H 2 O/ MeOH/EtOAc 1:2:8 as the eluent to give a colorless oil, which was lyophilized to afford an amorphous solid (4.2 mg, 52%). The product was further purified by RP-HPLC (C 18 column) and lyophilized to give pure 19 as a white solid. 1  Compound 21. Alkyne 12 (5 mg, 0.0147 mmol) and azide 20 (5.7 mg, 0.006 mmol) were suspended in a mixture of 1:1 t BuOH/water (0.1 mL) and stirred magnetically. A freshly prepared solution of sodium ascorbate (3.8 mg, 0.019 mmol) in water was added, followed by addition of a freshly prepared aqueous solution of CuSO 4 ·5H 2 O (1.6 mg, 0.006 mmol). This heterogeneous mixture was stirred vigorously overnight at r.t. The mixture was evaporated under reduced pressure to give a crude product. The latter was purified by silica gel column chromatography using H 2 O/MeOH/EtOAc 1:3:8 as the eluent to give a light-yellow oil, which was lyophilized to afford an amorphous solid (4.8 mg, 41%). The product was further purified by RP-HPLC (HILIC Column) and lyophilized to give pure 21 as a white solid. 1 13  The grid box around the NA tetramer was generated automatically and enlarged to incorporate the entire protein. For "cavity detection" slow and effective "molecular" method was used for detection of binding sites. Inner and outer contours of the grid box were also calculated automatically using "balanced" settings for "site shape potential" calculations. Docking was performed with OEDocking 3.3.1.2 using the hybrid program. 41 A hit list of top 100 ranked molecules was retrieved, and the best ranked hybrid-calculated poses were inspected visually and used for analysis and representation. The results were evaluated in visualization software VIDA 4.4.0 (OpenEye Scientific Software, Inc., Santa Fe, NM, USA).
MUNANA Assay. The inhibitory activities of different compounds were assessed by using the synthetic monovalent substrate MUNANA similarly as described previously. 43 Briefly, compounds were diluted in the reaction buffer (50 mM Tris-HCl, 4 mM CaCl 2 , pH 6.0) and subsequently serially diluted 1:2 in a flat-bottom 96-well black plate, followed by the addition of a similar volume of reaction buffer containing a fixed, non-saturated amount of NA protein or virus. Subsequently, 200 μM MUANANA diluted in the reaction buffer was added to each well to a final concentration of 100 μM, mixed well, and incubated at 37°C for 60 min. The reaction was terminated by addition of the stop solution (0.1 M glycine, 25% ethanol, pH 10.7). The fluorescence signal was immediately determined in relative fluorescence units by using a FLUOstar OPTIMA plate reader with the excitation and emission wavelengths at 340 and 490 nm, respectively.
CPE Assay Experiments. Compounds were serially diluted in Opti-MEM starting at 5 μM, followed by the addition of 50TCID 50 units of H1N1 pdm09. Subsequently, the virus and compound mixtures were incubated on a monolayer of MDCKII cells in a 96-well plate at 37°C and 5% CO 2 for 4 days. After the incubation, the cell cultures were visually inspected using a microscope for virally induced CPE and the inhibition thereof by the compounds. The lowest concentration for each compound that inhibits the formation of CPE was used to assess their inhibitory activities (N = 8−16).
BLI Binding Assay. All BLI experiments were performed using OctetRED384 (ForteBio) as described previously. 3,45 All the experiments were carried out in phosphate-buffered saline with calcium and magnesium (Lonza) at 30°C and with shaking of plates at 1000 rpm. In short, streptavidin sensors were loaded to saturation with biotinylated lysosomal-associated membrane glycoprotein 1 (LAMP1) containing increased levels of α2,6-linked sialic acids. 45 Subsequently, sensors were moved to wells containing a mixture of the virus and variable concentrations of the indicated compounds. When indicated, oseltamivir carboxylate (OC; 10 μM final concentration, gift from Roche) was added to this mixture to inhibit NA activity. As a control, binding was analyzed in the absence of inhibitory compounds or in the presence of OC only. For the virus rebinding assay, sensors bound with viruses were moved to wells carrying 0.1 M glycine (pH = 2) three times for 5 s in order to remove all bound viruses after a virus binding step as described above. Afterward, the sensors were dipped into virus-containing wells in the presence of 10 μM OC to check virus binding to the remaining sialoglycan receptors.