n-Propyl 6-amino-2,6-dideoxy-2,2-difluoro-β-d-glucopyranoside is a good inhibitor for the β-galactosidase from E. coli

A convenient route has been developed for the synthesis of novel 6-amino-2,2-(or 3,3-difluoro)-2-(or 3),6-dideoxy-hexopyranoses. Biological screening showed these compounds as good inhibitors for several glycosidases. Especially n-propyl 6-amino-2,6-dideoxy-2,2-difluoro-β-d-glucopyranoside (8) was an excellent competitive inhibitor for the β-galactosidase from E. coli holding a Ki of 0.50 μM.

An extended search in literature as of 2020, October, revealed there are some 30,600 structures containing the substructure of a 6-amino-6-deoxy-hexopyranose (or -pyranoside). Monofluorinated analogs, however, have scarcely been prepared (some glycosyl fluorides and a couple of 2-fluoro and 3-fluoro analogs; the latter-by and large-being part of modified kanamycin derivatives) [31][32][33][34][35][36][37]-but to the best of our knowledge, there are no 2,2-or 3,3-difluoro-6-amino-2 (or 3), 6-dideoxy-hexopyranoses (respectively -hexopyranosides). Some molecules containing the structural element of a 6-amino-hexose have been proposed in-silico as possible lead structures for the therapy of COVID-19 diseases [38][39][40]. Therefore, we set out for the first synthesis of these targets and to test their ability to act as glycosidase inhibitors. It is reserved for later studies to have a look at their properties regarding lysosomal storage diseases or antiviral activity.
Partial deprotection [44] of 11 gave a mixture of 12 and 13. Main component 12 was iodinated as described above to afford 14 whose subsequent nucleophilic displacement reaction (→ 15) and hydrogenation gave glycoside 16.
An in vitro evaluation of these compounds to act as an enzyme inhibitor using a panel of commercially available glycosidases was performed (Table 1) using p-nitrophenolate [50] assays.
Compounds 8 and 16 were competitive inhibitors for all enzymes. Thereby, compound 8 was a nonselective inhibitor, and its highest activity was found for the ß-galactosidase from E. coli. Glycoside 16 was a weaker inhibitor than 8.
For a better interpretation of these results additional molecular modeling calculations were carried out. Thereby, the protein crystal structures were selected on their genetic similarity to the enzymes employed in the in vitro experiments. An additional focus was set to a good resolution of the enzyme structure. This was easily achieved for several of the enzymes, while for the β-glucosidase from almonds and the α-galactosidase (from green coffee beans) an additional search for an enzyme from a different organism was called for. This was performed utilizing UniProt.org eventually finding enzymes of high similarity. Thus, miglitol and compounds 8 and 16 were evaluated in molecular docking studies against the enzymes α-glucosidase (Geobacillus sp.; PDB: 2ZE0, Baker's yeast; PDB: 3AXI), β-glucosidase (from rice due to its similarity with the β-glucosidase from almonds; PDB: 1UAS), α-galactosidase (from white clover due to its similarity to the enzyme from green coffee beans; PDB: 1CBG), and β-galactosidase (E. coli; PDB: 1JYW). From the docking studies the binding affinity was estimated.
A rough blind docking of 50 individual runs showed that for compounds 8 and 16 their main binding site at the enzymes is similar to the binding site of miglitol in each case. Therefore, the search space was limited to the active site of the respective enzyme. As a result, five top poses were obtained being very closely located to each other holding H-bond interactions with closely related amino acids.
Thus, the 2,2-and 3,3-difluorinated miglitol analogs are efficient competitive inhibitors of the αand β-glucosidase as well as of the galactosidases. Miglitol, however, showed lowered inhibition for yeast's and green coffee bean's αglucosidase and galactosidase, while for compound 8 low K i values for all tested enzymes were established. This is most likely due to favored interaction of the difluorinated compounds in the active sites leading eventually to better binding affinities for compounds 8 and 16 with yeast αglucosidase and green coffee beans α-galactosidase (cf. Tables 2 and 3). Table 3 shows the main interactions of the compounds with the amino acids of the respective enzyme.
The good binding affinities of compounds 8 and 16 as compared to miglitol allow no explanation for the low K i value of miglitol for the α-glucosidase from Geobacillus sp. (almond's β-glucosidase) and E. coli's β-galactosidase. We assume that these findings are caused by different residence times of the compounds. As a consequence of this dynamic effect, the hydroxyl groups of miglitol may help to hold this compound in the periphery of the binding site thereby increasing the residence time of miglitol as compared to the residence time of compounds 8 and 16. Comparing the   docking results from Table 2, compound 8 seems to be a slightly better inhibitor than miglitol (−0.47 kcal/mol on average), and compound 16 seems to be also a slightly better inhibitor than compound 8 (−0.16 kcal/mol on average). Table 3 summarizes the main binding residues thus allowing some insight into the different binding motifs of compounds 8, 16 and miglitol. For the α-glucosidase from Baker's yeast, Trp468 seems to be a common amino acid residue all evaluated compounds share H-bonding with, whereas other amino acid residues may vary. Furthermore compound 16 and miglitol share similar H-bonding motif; for example, docking of miglitol, 8, and 16 with the β-glucosidase from rice showed miglitol and 16 to share Lys58, His256, and Trp454 as amino acid residues with Hbonding interactions. The results for the β-galactosidase from E. coli revealed that miglitol and compound 8 only share Asn102, whereas 16 formed H-bonds to Asn102, Gly489, and Lys517 as also observed for miglitol. Figure 2 depicts some of the results from the modeling, the interaction of miglitol and of compound 16 in the active site of the β-galactosidase from E. coli.
Conclusion n-Propyl 6-amino-2,6-dideoxy-2,2-difluoro-β-D-glucopyranoside (8) as well as its 3,3-difluor-analog (16) were easily accessible from allyl glycosides 1 and 9 by a sequence of oxidation, difluorination, selective deprotection, nucleophilic displacement, and catalytic hydrogenation. While no significant cytotoxicity was observed in the SRB assays (EC 50 > 50 μM), screening of these compounds in p-nitrophenolate assays showed especially compound 8 as a good inhibitor for the β-galactosidase from E. coli. It also seems to be obvious for this novel class of compounds that the presence of a hydrophobic moiety in β-position has a stronger influence on the inhibitory activity with respect to β-galactosidase of E. coli than a corresponding hydrophobic moiety in γ-position.

Experimental
Melting points are uncorrected (Leica hot stage microscope), optical rotations were obtained using a Perkin-Elmer 341 polarimeter (1 cm micro cell), NMR spectra were recorded using the Varian spectrometers Gemini 2000 or Unity 500 (δ given in ppm, J in Hz, internal Me 4 Si or internal CCl 3 F), IR spectra (film or KBr pellet) on a Perkin-Elmer FT-IR spectrometer Spectrum 1000, MS spectra were taken on a Intectra GmbH AMD 402 (electron impact, 70 eV) or on a Thermo Electron Finnigan LCQ Table 3 Main interactions of the compounds with the amino acids of the respective enzyme (ascending order, non-H-bonding interactions are in italic; equal residues (bold) of binding of miglitol and 8 or 16, respectively) Geobacillus sp.
The cytotoxicity of the compounds was evaluated using the SRB (Kiton-Red S, ABCR) micro culture colorimetric assay using confluent cells in 96-well plates with the seeding of the cells on day 0 applying appropriate cell densities to prevent confluence of the cells during the period of the experiment. On day 1, the cells were treated with six different concentrations (1, 3, 7, 12, 20, and 30 μM); thereby, the final concentration of DMSO was always < 0.5%, generally regarded as nontoxic to the cells. On day 4, the supernatant medium was discarded; the cells were fixed with 10% trichloroacetic acid. After another day at 4°C, the cells were washed in a strip washer and dyed with the SRB solution (100 μL, 0.4% in 1% acetic acid) for about 20 min to be followed by washing of the plates (four times, 1% acetic acid) and air-drying overnight. Furthermore, tris base solution (200 μL, 10 mM) was added to each well and absorbance was measured at λ = 570 nm employing a reader (96-wells, Tecan Spectra, Crailsheim, Germany). The IC 50 values were averaged from three independent experiments performed each in triplicate calculated from semi logarithmic dose response curves applying a nonlinear fourparameter Hills-slope equation (GraphPad Prism5; variables top and bottom were set to 100 and 0, respectively).  To a solution of 2 (300 mg, 0.72 mmol) in dry DCM (6.0 mL), DAST (417 μL 3.03 mmol) was added dropwise under argon, and the mixture was stirred at 25°C for 5 days. Methanol (1.0 mL) was carefully added, and the solvents were removed under diminished pressure. The oily residue was dissolved in DCM (90 mL) and washed with water (50 mL). The aqueous phase was re-extracted with DCM (3 × 100 L); the organic layers were combined, and the solvent was evaporated under reduced pressure. The remaining residue was subjected to chromatography (silica gel To an ice-cold solution of 3 (3.00 g, 7.17 mmol) in dry ether (30 mL) and THF (30 mL), lithium aluminum hydride (490 mg, 12.91 mmol) was added in several portions, and the suspension was stirred for 15 min at 25°C. Then, the suspension was heated to reflux, and a solution of dry aluminum chloride (1.63 g, 12.19 mmol) in dry ether (30 mL) was added dropwise followed by stirring under reflux for another 48 h. The suspension was cooled to 25°C, methanol (30 mL) was carefully added, and stirring was continued for another 30 min. After usual aqueous workup followed by chromatography (silica gel, n-hexane/ethyl acetate, 5:3) 4 (450 mg, 15%) and 5 (1.50 g, 50%) were obtained each as a colorless oil. To a solution of 5 (1.50 g, 3.57 mmol) in toluene (30 mL) containing triphenylphosphane (2.06 g, 7.85 mmol) and imidazole (1.09 g, 16.06 mmol), iodine (1.81 g, 7.14 mmol) was added in several portions. After stirring at 90°C for 2 h, the reaction mixture was decanted, and the remaining oil was washed with ether (3 × 100 mL). The combined organic layers were evaporated, and the remaining residue was subjected to chromatography (silica gel, n-hexane/ethyl acetate, 85:15) to afford 6 (900 mg, 52%) as a colorless oil; [α] D = −25.32°(c 0.44, CHCl 3 ); R ƒ (n-hexane/ethyl acetate, 85:15) = 0.56; analysis calcd for C 23  Allyl 6-azido-3,4-di-O-benzyl-2,6-dideoxy-2,2-difluoro-β-Darabino-hexopyranoside (7) To a solution of 6 (940 mg, 1.77 mmol) in dry DMF (19 mL) a solution of lithium azide (20% in water, 2.17 mL, 8.86 mmol) was added at 25°C, and the solution was stirred at this temperature for 4 days. The solvents were removed under diminished pressure, and the remaining residue was dissolved in DCM (100 mL) and water (50 mL). The aq. phase was extracted with DCM (3 × 100 mL), and the combined organic layers were dried (Na 2 SO 4 ). The solvent was removed under reduced pressure, and the remaining residue was subjected to chromatography (silica gel, hexane/ethyl acetate, 85:15) to afford 7 (700 mg, 89%) as a colorless oil; To an ice-cold solution of 11 (4.21 g, 10.06 mmol) in dry ether (40 mL) and THF (40 mL) lithium aluminum hydride (687 mg, 18.111 mmol) was added in several portions, and the suspension was stirred for 15 min at 25°C. Then, the suspension was heated under reflux, and a solution of dry aluminum chloride (2.28 g, 17.11 mmol) in dry ether (30 mL) was added followed by stirring under reflux for 72 h. The suspension was cooled to 25°C, methanol (30 mL) was carefully added, and stirring was continued for another 45 min. Usual aqueous workup followed by chromatography (silica gel, n-hexane/ethyl acetate, 5:3) gave 12 (490 mg, 12%) and 13  Allyl 2,4-di-O-benzyl-3, 6-dideoxy-3,3-difluoro-6-iodo-β-Dribo-hexopyranoside (14) To a solution of 12 (1.82 g, 4.33 mmol) in toluene (40 mL) containing triphenylphosphane (2.50 g, 9.52 mmol) and imidazole (1.33 g, 19.48 mmol) iodine (2.20 g, 3.66 mmol) was added in several portions. After stirring at 90°C for 1 h, the reaction mixture was decanted and the remaining oil was washed with ether (4 × 100 mL). The combined organic layers were evaporated, and the remaining residue was subjected to chromatography (silica gel, n-hexane/ethyl acetate, 85:15) to afford 14 (1.80 g, 78%) as a colorless oil; [α] D = +23.79°(c 0.36, CHCl 3 ); R ƒ (n-hexane/ethyl acetate, 85:15) = 0.51; analysis calcd for C 23  Allyl 6-azido-2,4-di-O-benzyl-3,6-dideoxy-3,3-difluoro-β-Dribo-hexopyranoside (15) To a solution of 14 (1.45 g, 2.73 mmol) in dry DMF (29 mL) a solution of lithium azide (20% in water; 3.35 mL, 13.67 mmol) was added at 25°C, and the solution was stirred at this temperature for 4 days. The solvents were removed under diminished pressure, and the remaining residue was dissolved in DCM (100 mL) and water (50 mL). The aq. phase was extracted with DCM (3 × 100 ml), and the combined organic layers were dried (Na 2 SO 4 ). The solvent was removed under reduced pressure, and the remaining residue was subjected to chromatography (silica gel, n-hexane/ethyl acetate, 85:15) to afford

Data availability
The data sets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.