Piperidine Azasugars Bearing Lipophilic Chains: Stereoselective Synthesis and Biological Activity as Inhibitors of Glucocerebrosidase (GCase)

We report a straightforward synthetic strategy for the preparation of trihydroxypiperidine azasugars decorated with lipophilic chains at both the nitrogen and the adjacent carbon as potential inhibitors of the lysosomal enzyme glucocerebrosidase (GCase), which is involved in Gaucher disease. The procedure relies on the preparation of C-erythrosyl N-alkylated nitrones 10 through reaction of aldehyde 8 and primary amines 13 followed by oxidation of the imines formed in situ with the methyltrioxorhenium catalyst and urea hydrogen peroxide. The addition of octylMgBr to nitrone 10e provided access to both epimeric hydroxylamines 21 and 22 with opposite configuration at the newly created stereocenter in a stereodivergent and completely stereoselective way, depending on the absence or presence of BF3·Et2O. Final reductive amination and acetonide deprotection provided compounds 14 and 15 from low-cost d-mannose in remarkable 43 and 32% overall yields, respectively, over eight steps. The C-2 R-configured bis-alkylated trihydroxypiperidine 15 was the best ligand for GCase (IC50 = 15 μM), in agreement with MD simulations that allowed us to identify the chair conformation corresponding to the best binding affinity.


INTRODUCTION
Iminosugars [e.g.,deoxynojirimycin,DNJ (1), Figure 1] are among the most fascinating monosaccharide analogues in which a nitrogen atom replaces the endocyclic oxygen. 1 Together with azasugars, [e.g. isofagomine,IFG (2), or 1,5dideoxy-1,5-iminoxylitol, DIX (3), Figure 1], which are characterized by a nitrogen atom replacing the anomeric carbon of monosaccharides, iminosugars have been extensively investigated in the last thirty years as glycosidase 2 and glycosyltransferase inhibitors. 3 More recently, imino-and azasugar-based glycomimetics became attractive as potential therapeutic agents toward lysosomal storage disorders (LSDs), following the observation of their counter-intuitive effect in enhancing the enzyme activity, thus acting as chaperones. In the pharmacological chaperone therapy (PTC) of LSDs, these glycomimetics are employed at sub-inhibitory concentration to favor the mutated enzyme correct folding in the endoplasmic reticulum (ER), facilitate its translocation to the lysosomes, and recover some hydrolytic activity, compromised as a consequence of diverse genetic mutations. 4 Gaucher disease (GD), the most common LSD, is determined by mutations in the GBA gene, which encodes for the lysosomal enzyme glucocerebrosidase (GCase). The GBA mutations provoke partial deficit of GCase, with consequent loss or reduction of its hydrolytic activity [i.e., hydrolysis of glucosyl ceramide, GlcCer (4), Figure 1, to ceramide and glucose] in the lysosomes. Accumulation of glucosylceramide then leads to organ dysfunctions and severe impairment. Several good candidates behaving as pharmacological chaperones (PC) for GCase have been found, but no drugs are on the market yet. 5 The presence of an alkyl chain either at the nitrogen atom or at the adjacent carbon of imino-and azasugars was found to improve their pharmacokinetic properties, furnishing better PCs for GD. For instance, contrary to DNJ (1), N-nonyl-DNJ (5) (Figure 1) is a potent inhibitor of lysosomal GCase (IC 50 = 1 μM) which showed a twofold increase in the activity of the N370S mutant enzyme in fibroblasts at 10 μM concentration. However, it did not enhance the intracellular activity of the L444P variant. 6 With IFG and DIX derivatives, better results in terms of PC properties were obtained by shifting the alkyl chain from nitrogen to the adjacent carbon. Indeed, 6-nonyl IFG (6) ( Figure 1) displayed a remarkable GCase inhibition and PC activity (IC 50 = 0.6 nM, 1.5-fold enzyme activity enhancement at 3 nM in N370S GD fibroblasts) 7 and α-1-C-nonyl-DIX (7) showed an IC 50 of 6.8 nM toward GCase and a 1.8-fold enzyme activity increase in N370S mutated fibroblasts at 10 nM. 8 Among the most straightforward synthetic strategies to afford polyhydroxylated piperidine imino-and azasugars 9 are the reductive amination (RA) ring-closure starting from nitrogen-containing carbohydrate-based precursors or the double reductive amination (DRA) of dicarbonyl carbohydrate derivatives with an external nitrogen source. 10 By applying the latter strategy on the "masked" dialdehyde 8 derived from Dmannose, we have synthesized the N-alkyl trihydroxypiperidine 9 (Scheme 1), 11 which showed IC 50 = 30 μM toward GCase and 1.25-fold activity increase in N370S-mutated fibroblasts at 100 μM. 12 The former RA strategy, instead, was applied to the Cerythrosyl nitrone 10a, 13 obtained in turn by condensation of the same "masked" dialdehyde 8 with N-benzyl hydroxylamine. 14 Stereoselective Grignard additions to this key nitrone in the presence or absence of a suitable Lewis acid 14,15 afforded, in a stereodivergent manner, both epimers of a series of 2-alkylated trihydroxypiperidines, among which the octyl derivatives 11 and 12 (Scheme 1) showed the most promising biological properties. 16 In particular, the (2R) diastereoisomer 12 showed remarkable PC properties toward fibroblasts bearing the N370S/RecNcil mutation (1.9-fold enzyme recovery at 100 μM) and, more importantly, proved to be responsive with the homozygous L444P mutation (1.80-fold enzyme recovery at 100 μM), which is refractory to most PCs. Remarkably, both compounds 9 and 12 performed PC tasks toward wild-type fibroblasts (1.5-fold at 50 μM and 1.4-fold at 100 μM, respectively), which is an important factor for targeting sporadic forms of Parkinson disease. 17 Given the relevance of the presence of a long alkyl chain for inducing good enzyme recognition, and in consideration of the structure of the natural substrate GlcCer (4), we speculated that compounds possessing two alkyl chains might show improved biological properties.
We therefore addressed the synthesis of 1,2-dialkyltrihydroxypiperidines and envisaged that a general synthetic strategy could involve the Grignard addition to C-erythrosyl N-alkyl nitrones 10 followed by RA (Scheme 1).
In this work, we report our results on this subject, which include a docking study of the putative 1,2-dialkyltrihydroxypiperidines in the GCase catalytic site; a general direct synthesis of nitrones 10 and their conversion to target compounds 14 and 15 and to derivatives containing three alkyl tails; the biological evaluation of the dialkylated trihydroxypiperidines toward commercial glycosidases and human lysosomal enzymes and their in vitro activity on cell lines; a molecular dynamics (MD) simulation of the new compounds within the GCase enzyme active cavity. With the aim of assessing the viability of target compounds 14 and 15 in GCase catalytic site, we carried out preliminary docking studies with acid-beta-glucosidase (PDB ID 2NSX). 18 The protonated forms of compounds 14 and 15 were considered, taking into account that a stereogenic nitrogen atom is formed (for details, see the Supporting Information section). For compound 14, the best pose corresponds to an (R)-configuration at the nitrogen atom adopting a 1 C 4 conformation (score −6.332). The observed orientation for the piperidine ring is different with respect to that observed in a known chaperone IFG (2), Figure 1], thus loosing interactions of hydroxyl groups with Asp127, Trp179, and Asn396 that now form new interactions with Ser237, Asp283, and Gln284. Although the aliphatic chain linked to the nitrogen atom shows hydrophobic interactions with Phe128, Trp179, Tyr313, Leu314, and Ala238, the other aliphatic chain is clearly exposed to the solvent (see the Supporting Information section). The (S)-isomer shows a close value in the docking score (−6.071) for the same conformation and, also in this case, the piperidine ring is oriented in the wrong way with respect to IFG. A similar situation was found for protonated 15. In this case, the (S)isomer in a 1 C 4 conformation showed the best docking value (−6.614) due to hydrophobic interactions of the aliphatic chains in the binding site, the (R)-isomer in the same conformation being very close (docking score = −6.354). Again, in both cases, the piperidine ring is oriented in the wrong way when compared with the original structure bound to IFG. For the (R)-isomer, the chain bonded to the nitrogen atom gives interactions with Phe128, Trp179, Phe246, and Trp381. The similar orientation of the piperidine ring to that observed for 14 gives rise to the same interactions of the hydroxyl groups with Ser237, Asp283, and Gln284 ( Figure 2). These preliminary results should be considered with caution due to the predicted orientations of the piperidine ring in a rigid protein. Even though the calculations predict a good binding to the enzyme, the particular interactions and orientations of the ligands will be confirmed (or rejected) through further molecular dynamics studies that will allow conformational changes in the protein (vide infra).
2.2. Synthesis and Structural Assignment. The aldehyde 8 was synthesized in four steps from D-mannose on gram scale as reported. 11,19 The synthesis of nitrones 10 from 8 by condensation in analogy to the reported 10a would require the corresponding N-monosubstituted hydroxylamines, compounds not readily available in general. However, nitrones can also be obtained in several alternative ways, 20 among which the oxidation of imines developed by some of us 21 was selected as the most appropriate for our purposes. This procedure was established first on preformed imines employing methyltrioxorhenium (MTO) as the catalyst and urea hydrogen peroxide complex (UHP) as a mild stoichiometric oxidant. The methodology has the great advantage over other oxidation methods (on secondary amines 22 or N,N-disubstituted hydroxylamines) 20a of forming a single nitrone, since the double bond is previously installed during the imine formation. Moreover, a convenient one-pot version has been subsequently implemented, where imines are formed in situ from inexpensive or readily available primary amines and aldehydes. 23 In this work, we employ the one-pot condensation/oxidation strategy to our carbohydrate-derived "masked" dialdehyde 8 with primary amines 13a−h for the synthesis of several new Ccarbohydrate N-alkyl nitrones 10 (Scheme 1). We also demonstrate that such nitrones undergo stereodivergent and highly stereoselective Grignard reagent additions in the presence or absence of a suitable Lewis acid. The following reductive amination (RA) of the formed adducts directly gives access to new C,N-dialkyl trihydroxypiperidines, as demonstrated with the synthesis of compounds 14 and 15 (Scheme 1). The target compounds were designed as to better mimic the two chains of the GlcCer (4) natural substrate.
The one-pot condensation/oxidation was investigated starting from aldehyde 8 with structurally diversified primary  Table 1. The reaction afforded the corresponding N-alkyl nitrones 10a−h with good yields and in an operationally very simple manner. The procedure was optimized using benzylamine (13a) for the preparation of the known nitrone 10a. The best results were obtained by stirring a solution of the aldehyde 8 with 1.2 equiv of the appropriate amine 13 in MeOH at room temperature in the presence of anhydrous Na 2 SO 4 , until disappearance of the starting aldehyde 8 at an 1 H NMR control (3−5 h, depending on the amine), which attested the formation of the corresponding imine. After cooling of the reaction mixture at 0°C, addition of UHP (3 equiv) and MTO (4 mol %) caused the solution to turn yellow, indicating the formation of the catalytically active peroxorhenium species. 24 Upon completion of the oxidation at room temperature, a simple work-up consisting of solvent removal under reduced pressure followed by CH 2 Cl 2 addition allowed filtering off the undissolved urea and Na 2 SO 4 to give, after evaporation of the solvent, the crude nitrones which were purified by flash column chromatography on silica gel. Alternative addition of UHP and MTO since the beginning of the reaction (prior to imine formation) was found to be unpractical and afforded complex mixtures of products deriving from competitive oxidation of the primary amines. The scope listed in Table 1 demonstrates the versatility of this method, which gives comparable results with complete conversions and good yields (70−80%) for benzylamines 13a−c (entries 1−3), homobenzylamine 13d (entry 4), and linear aliphatic amines 13e−f (entries 5−6). Decrease in reaction yields with the α-branched (methyl)benzyl 13c (entry 3) and aliphatic amines 13g and 13h (entries 7−8) suggests that the oxidation reaction is somewhat sensitive to steric effects. However, the corresponding nitrones 10c and 10g−h were still obtained in overall satisfactory yields (60−75%) through the two-step process. All the nitrones 10 were obtained exclusively as the more stable Z-diastereoisomer, in agreement with previous results (see also the Supporting Information section for 1D NOESY experiments on nitrone 10f). 21 Nitrones 10 turned out to be excellent starting materials for the straightforward synthesis of N-substituted trihydroxy piperidines. Indeed, treatment of the nitrones 10e, 10f, and 10h in MeOH under H 2 atmosphere (balloon) over Pd/C or Pd(OH) 2 /C as the catalyst afforded the N-substituted piperidines 16, 11 17, and 18, respectively, in excellent yields (Scheme 2). The overall efficiency of this one-pot process is remarkable, considering that it results from a cascade of several reactions, consisting of nitrone reduction, debenzylation, condensation with sugar aldehyde and iminium or enamine reduction (RA). Application of the same procedure to nitrone 10a afforded the N-unsubstituted piperidine 19 11 in quantitative yield.
It is worth to note that the yields obtained using this strategy were by far superior to those previously reported for the synthesis of piperidines 16 and 19 employing condensation of aldehyde 8 with primary amines and subsequent reduction to amine and RA-cyclization step, or, in alternative, employing a DRA strategy with NaBH 3 CN as the reducing agent on aldehyde 20, in turn derived from 8. 11 (Scheme 3). For comparison, N-dodecyl trihydroxypiperidine 17, obtained in 90% yield from nitrone 10f (Scheme 2) was obtained also via DRA of 20 with dodecylamine in a modest 38% yield (Scheme 3).
The synthetic strategy based on the addition of Grignard reagents to nitrone 10e followed by intramolecular RA provided access to the desired 1,2-octyl trihydroxypiperidines 14 and 15 (Scheme 1). The length of the chains (C8) was chosen on the basis of the data previously obtained from biological assays. We had previously observed that the addition of Grignard reagents to the carbohydrate-derived nitrone 10a proceeded with opposite diastereofacial preference in the presence or absence of a suitable Lewis acid. The addition reaction, followed by RA, allowed the introduction of different alkyl groups with opposite configurations at the piperidine C-2, affording two diastereomeric series of 2-alkyl trihydroxy piperidines. 14,16,17 The addition of octylMgBr to nitrone 10e was first carried out in THF at −78°C for 3 h without any Lewis acid. The reaction gave smoothly the hydroxylamine 21, with the S absolute configuration at the newly formed stereocenter accordingly to our prior work in good yield (75%) and excellent stereoselectivity (>98%) (Scheme 4). The configuration at the newly created stereocenter was unambiguously confirmed by 1 H NMR and 1D NOESY spectra at a later stage of the synthesis (vide infra). The addition of octylMgBr to nitrone 10e in the presence of BF 3 ·Et 2 O at −30°C for 2 h resulted in a complete reversal of the diastereoselectivity in favor of the epimeric hydroxylamine 22, which was obtained with excellent stereoselectivity (>98%) and good yield (70%) with an R configuration at the newly formed stereocenter (Scheme 4).
Albeit the observed stereochemical outcome of the additions of octylMgBr to the nitrone 10e in the absence or presence of BF 3 ·Et 2 O is in agreement with the results obtained previously in the additions of several Grignard reagents to nitrone 10a, we were delighted to see that in this case a complete diastereoselectivity was reached. Instead, in the Grignard additions to 10a, we had observed diastereomeric ratios in the range 1.4−5.6:1 without Lewis acids and 3−9:1 with BF 3 ·Et 2 O in favor of the (S) and (R) configured adducts, respectively. 16,17 A magnesium Cram-chelate transition state (TS) may account for the preferred nucleophilic attack to the Si diastereoface of nitrone 10e in the absence of Lewis acid, resulting in the exclusive formation of the hydroxylamine 21 with the S configuration at the newly formed stereocenter ( Figure 3A). Conversely, when an equimolar amount of BF 3 · Et 2 O is added, chelation is prevented, thus favoring the nucleophilic attack to the more accessible Re face of nitrone 10e, which leads to the hydroxylamine 22 with the R configuration at the newly formed stereocenter ( Figure  3B). 14,17 The hydroxylamines 21 and 22 showed high tendency to oxidize spontaneously in air to the corresponding aldonitrones 23 and 24 (Scheme 5), according to the behavior observed for other hydroxylamines obtained by addition of Grignard reagents to the N-benzyl nitrone 10a; notwithstanding the CN bond does not benefit of conjugation in the present case. 16,17 In particular, spontaneous partial oxidation of 21 and 22 occurred with 50% conversion, as evaluated by 1 H NMR spectroscopy. The exclusive formation of aldonitrones instead of ketonitrones was providential, since it did not trigger any loss of configurational integrity at the newly formed stereocenter and was not detrimental for our synthetic purposes (see below).
However, a certain regioselectivity in favor of the aldonitrones was expected based on our previous studies.
Due to the low stability of hydroxylamines 21 and 22, they were characterized only by 1 H NMR and MS analyses immediately after purification by column chromatography. Characterization of pure nitrones 23 and 24 was carried out after complete oxidation of the hydroxylamines 21 and 22 with the hypervalent iodine reagent 2-iodoxy benzoic acid (IBX), which is the reagent of choice to promote regioselective oxidation of N,N-disubstituted hydroxylamines in favor of the corresponding aldonitrones. 13,25 In this case, it provided the corresponding nitrones 23 and 24 in excellent yields and with complete regioselectivity (Scheme 5).
The final reductive amination step was performed on the hydroxylamine/nitrone mixtures by employing 2 equiv of acetic acid under an H 2 atmosphere (balloon) and Pd/C as the catalyst in MeOH (0.015 M solution), followed by treatment with a strongly basic resin to give the free amines 25 and 26 in excellent 90 and 80% yield, respectively (Scheme 6).
Careful analysis of the 1 H NMR, 2D NMR (COSY, HSQC), and 1D NOESY spectra of piperidines 25 and 26 allowed us to establish unambiguously their configuration, thus validating the stereochemical outcome of the additions of octylMgBr reported in Scheme 4. The two chair conformations of each compound are reported in Figure 4. The 1 H NMR spectrum of (2S) piperidine 25 showed a small coupling constant between 3-H and 4-H ( 3 J 3−4 = 4.0 Hz), suggesting a high preference for its 1 C 4 conformation where both 3-H and 4-H are equatorial. The 1 C 4 is expected to be the absolute minimum energy conformation for the S-configured piperidine at C-2, with the octyl chain lying in equatorial position ( Figure 4). Accordingly, the 1D NOESY spectrum of 25 did not show the nOe correlation peak between 2-H and 4-H. For the diastereomeric piperidine 26, the opposite R configuration at C-2 shifted the equilibrium to a preferred 4 C 1 conformation in order to accommodate again the octyl chain in an equatorial position, as Final deprotection of the acetonide protecting groups under acidic conditions (aqueous HCl in MeOH) followed by treatment with the strongly basic resin Ambersep 900-OH gave the target 1,2-dioctyl trihydroxypiperidines 14 and 15 as free amines in excellent yields (Scheme 6).
We envisaged that the intriguing formation of the aldonitrones 23 and 24 by easy and regioselective oxidation of hydroxylamines 21 and 22 would furnish the chance for further extending our study. Indeed, another alkyl chain might be introduced via iteration of the Grignard addition to these nitrones to provide final piperidines possessing three lipophilic tails. This hypothesis was proven with nitrone 24. In order to avoid formation of a further stereogenic center resulting in a mixture of diastereoisomers, heptyl magnesium bromide was chosen as the appropriate Grignard reagent, but the addition turned out to be sluggish. In the absence of the promoter at different temperatures (−30°C, 0°C, rt), in THF, it failed to give any product. Addition of BF 3 ·Et 2 O (1.0 equiv) resulted in the formation of the desired hydroxylamine unstable and presumably underwent mainly rapid oxidation to the corresponding ketonitrones 28 and 29, as suggested by the presence of two TLC spots with very similar R f and a peak at 638.55, corresponding to [M + Na] + , in the ESI-MS spectrum (Scheme 7).
The lack of regioselectivity in the oxidation of 27 is a major drawback in view of our synthetic target, since the ketonitrone 29 has lost the stereochemical information to be installed at C-2 in the final piperidine. In order to prevent the rapid oxidation of the hydroxylamine 27, the adduct 30 was reduced in situ to the corresponding amine 31 by direct addition to the crude reaction mixture of indium powder in slightly acidic aqueous ethanol. 26 This one-pot Grignard addition/reduction afforded the amine 31 in 50% yield without any loss of stereochemical integrity (Scheme 7). The RA of amine 31 with H 2 as the reducing agent in the presence of catalytic Pd/C and acetic acid (2 equiv) in MeOH provided the partially protected trihydroxy piperidine 32 in 55% yield (Scheme 7 No remarkable inhibitory activity was found at this concentration for compounds 14 and 15 toward any of these enzymes apart from a 30% inhibition of β-glucosidase from almonds by compound 14 (see the Supporting Information section). Even if moderate, this value was encouraging, since we have previously noticed compounds with a low inhibitory activity toward this commercial enzyme that turned out to be stronger inhibitors of human lysosomal GCase. 12,17 2.3.2. Inhibitory Activity of GCase. Compounds 14,15, and 33·HCl were then tested at 1 mM for GCase inhibition in human leukocyte homogenates. The percentages of inhibition, together with the corresponding IC 50 values, are shown in Table 2. The results obtained were compared with data of previously reported compounds 9, 11, and 12. 12,16,17 Our results show that both the newly synthesized trihydroxypiperidines 14 and 15 were able to strongly inhibit GCase, imparting 100% inhibition of the enzyme at 1 mM concentration (Table 2, entries 4−5). However, a remarkable difference emerged between the two compounds from measurement of their IC 50 , which followed the same trend observed for the C-2 mono-alkylated azasugars 11 and 12. Indeed, the 1,2-dioctyl azasugar 15 with the R configuration at C-2 was a stronger inhibitor than its S-configured diaster-Scheme 7. Addition of heptylMgBr to Nitrone 24, One-Pot Addition/Reduction to Amine 31, and Synthesis of the Three-Tailed Trihydroxypiperidine 33·HCl The Journal of Organic Chemistry pubs.acs.org/joc Article eoisomer 14 (IC 50 = 15.0 μM vs IC 50 = 100 μM, Table 2 entry 5 vs 4). In this latter case, the difference was even more pronounced than for the corresponding monoalkylated congeners 12 and 11. It is also worth to note that the dialkylated piperidine 15 is twofold more active than its monoalkylated counterpart 12. Apparently, the presence of a second octyl chain at the nitrogen atom imparted beneficial interactions within the enzyme active site only in the case of the 2R-configured pair. Conversely, the 1,2-dialkylated azasugar 33·HCl, although maintaining the octyl chain at C-2 with R configuration is a much less potent GCase inhibitor than 15 (IC 50 = 130.0 μM vs IC 50 = 15.0 μM, Table 2 entry 6 vs 5), suggesting that the presence of a further alkyl chain is detrimental. 2.3.3. Molecular Dynamic Studies. We carried out MD simulations for (R) and (S) protonated forms of compounds 14 and 15, considering both conformations 4 C 1 and 1 C 4 as starting points and taken into consideration the best poses obtained from the docking studies and also the best ones in which the piperidine ring is correctly oriented as IFG. In all cases, MD converged to stable complexes which were reached after 10 ns of simulation and continued to be stable for 250 ns (all MD simulations were replicated four times). The complexes showed the ligand in the binding site establishing different interactions with key residues Asp128, Trp180, Asn235, Glu236, Tyr314, and Glu341 and with the correct orientation of the piperidine ring (see the Supporting Information section). 28 The overall RMSD for the protein system appeared to have reached equilibrium in the first ns, and the stabilization of the protein−ligand complex after 20 ns keeping the interactions of the complexes constant during the rest of the simulation. The ligands showed a complete stability of chair conformations and tend to arrange the aliphatic chains toward the external part, exposed to the solvent, in order to keep interactions of the hydroxyl groups of the piperidine ring. The presence of the alkyl chains induces some visible conformational change of the protein in the area close to the binding site. Interestingly, similar protein conformations were observed for (R)-14H + 4 C 1 and (S)−14H + 1 C 4 , whereas a loop formed by residues 345−349 is partially opened with (R)-14H + 1 C 4 and completely opened with (S)-14H + 4 C 1 (see the Supporting Information section). In the case of compound 15H + , the protein shows a higher flexibility with the loop formed by residues 345−349 adopting different orientations. The mobility of the loop, rich in hydrophobic residues (Met347, Phe348, and Trp349) can be attributable to hydrophobic interactions with alkyl chains that stabilizes the complex as expected. The sole inspection of the interactions in the binding site shows that (R)-15H + 4 C 1 is the complex having the highest stability ( Figure 5).
With the aim of obtaining more accurate information on the stability of the complexes (rather than the qualitative observation of the interactions), we carried out additional calculations to evaluate the binding energy. For the purpose of comparison both MM/GBSA and MM/PSBA calculations were performed. 29 In both cases, higher values were obtained for compound 15, the best values being found for (R)-15H + 4 C 1 (E binding = −57.9 kcal/mol with MM/GBSA and −49.4 kcal/mol with MM/PBSA) corroborating the observed interactions and in good agreement with experimental results (see the Supporting Information section).
2.3.4. Preliminary Biological Screening toward Human Lysosomal Glycosidases. In order to avoid undesired inhibitory side effects of a potential new drug, it has to be selective for a given target. The selectivity of the newly synthesized compounds 14, 15, and 33·HCl toward GCase was then investigated, evaluating their inhibition at 1 mM concentration toward six other lysosomal glycosidases (namely, α-and β-mannosidases, αand β-galactosidases, α-fucosidase, and α-glucosidase) in cell homogenates (leucocytes or lymphocytes) isolated from healthy donors. The results,   Close-up view of GCase in complex with compounds (R)-15H + 4 C 1 (cyan) with (R) configuration at C-2 and (R)-14H + 4 C 1 (green) with (S) configuration at C-2. Dashed yellow and magenta lines indicate H-bond interactions of (R)-15H + 4 C 1 and (R)-14H + 4 C 1 , respectively. It can be appreciated that the former has more polar contacts with the residues of the binding site than the latter. 2.3.5. Pharmacological Chaperoning Activity. The ability of compounds 14 and 15 to enhance the activity of GCase after incubation (4 days) with fibroblasts bearing the selected mutations was evaluated. The experiments were performed using fibroblasts derived from Gaucher patients bearing the N370S/RecNcil and L444P/L444P mutations (see the Supporting Information section). Unfortunately, no enzymatic activity rescue was observed after incubation with increasing concentrations of compounds 14 and 15 from 5 nM to 10 μM. The assay could not performed at higher concentration (50 or 100 μM), since the low cell viability observed hampered the measurement of the enzymatic activity.
These data show that the simultaneous presence of an octyl chain at both the endocyclic nitrogen and C-2 of the piperidine skeleton makes these compounds too cytotoxic at the highest concentrations, while at the lowest concentrations, no activity rescue was observed.
The low viability detected above 50 μM concentration may be ascribed to interactions of 14 and 15 with lipids and proteins of the cell membranes which lead to cell lysis, as it was previously reported for other amphiphilic compounds when tested in vitro. 30 Expecting an even higher impact on cell viability from compound 33·HCl, which has an additional lipophilic chain, a preliminary MTT test was carried out on wild-type fibroblasts (see Supporting Information). Since a remarkable cytotoxicity was observed for prolonged incubation at 100 and 50 μM concentrations, compound 33·HCl was co-incubated in Gaucher patients' fibroblasts (bearing the N370S/RecNcil mutation) in lower concentrations, ranging from 5 nM to 30 μM. Unfortunately, also in this case, no enzymatic activity rescue was observed after 4 days of incubation.

CONCLUSIONS
Following our interest in the discovery of new inhibitors and/ or pharmacological chaperones for the GCase enzyme, we investigated the condensation/oxidation reaction of carbohydrate-derived aldehyde 8 with several primary amines employing urea hydrogen peroxide as the stoichiometric oxidant and methyltrioxorhenium as the catalyst. This strategy afforded in a simple and straightforward way several new C-carbohydrate Nalkyl nitrones 10a−h. Further ring-closure reductive amination provided a series of new precursors of trihydroxypiperidine azasugars bearing different substituents at the nitrogen atom much more effectively than previously reported.
The reaction of the C-erythrosyl N-octyl nitrone 10e with octylMgBr in the presence or absence of BF 3 ·Et 2 O provided access to both epimeric hydroxylamines with opposite configuration at the newly created stereocenter in a stereodivergent and completely stereoselective way. Final reductive amination and acetonide deprotection provided compounds 14 and 15 from low-cost D-mannose in remarkable 43 and 32% overall yields, respectively, over eight steps. A third alkyl chain was introduced by iteration of the organometal addition to a nitrone easily obtained by oxidation with complete regioselectivity after the first addition, followed by the one-pot hydroxylamine/amine reduction mediated by indium metal.
This strategy provided the azasugar 33·HCl bearing three alkyl chains.
However, once tested in fibroblasts derived from Gaucher patients bearing the N370S/RecNcil and L444P/L444P mutations, no GCase activity enhancements were observed, demonstrating that the addition of one (14 and 15) or two alkyl chains (33·HCl) strongly increases the cytotoxicity of the compounds and it is detrimental for the pharmacological chaperoning activity.
Preliminary docking studies suggested compound 15 as the best ligand although the preferred conformation and orientation were not envisaged. Further MD simulations identified the protonated derivative of 15 with (R) configuration at the nitrogen atom and adopting a 4 C 1 conformation as the best ligand. The different isomers and the corresponding conformers induce a substantial change in the conformation of a loop, composed by residues 345−349, to accommodate the aliphatic alkyl chains. MM/GBSA calculations corroborated the preference by (R)-15H + 4 C 1 with a binding energy of −57.9 kcal/mol.  13 C NMR spectra were recorded at 50 MHz or at 100 MHz. Chemical shifts are reported relative to CDCl 3 ( 1 H: δ = 7.27 ppm, 13 C: δ = 77.0 ppm). Integrals are in accordance with assignments, coupling constants are given in Hz. For detailed peak assignments, 2D spectra were measured (g-COSY, g-HSQC) and 1D-NOESY. The following abbreviations were used to designate multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, quin = quintuplet, sext = sextet, sept = septet, br s = broad singlet, and dd = double-doublet. IR spectra were recorded with an IRAffinity-1S Shimadzu spectrophotometer. ESI-MS spectra were recorded with a Thermo Scientific LCQ fleet ion trap mass spectrometer. Elemental analyses were performed with a Thermo Finnigan FLASH EA 1112 CHN/S analyzer. Optical rotation measurements were performed on a JASCO DIP-370 polarimeter.
To a stirred solution of aldehyde 8 (200 mg, 0.72 mmol) and anhydrous Na 2 SO 4 (380 mg) in MeOH (9 mL) at room temperature, the appropriate amine 13a−h (1.2 equivalents) was added and the resulting mixture was stirred at room temperature under nitrogen atmosphere until 1 H NMR control attested the formation of imine intermediates (3 h). The reaction mixture was cooled at 0°C and urea−hydrogen peroxide (UHP, 203 mg, 2.16 mmol), and methyltrioxorhenium (MTO, 7.2 mg, 0.29 mmol) were added sequentially. The reaction mixture was stirred at room temperature The Journal of Organic Chemistry pubs.acs.org/joc Article for 16 h, when a TLC control (PEt/EtOAc 1:1) attested the disappearance of the starting material and the presence of a new UV visible spot; then, the solvent was removed under reduced pressure. CH 2 Cl 2 was added to the crude mixture, and the undissolved urea was filtered off. Solvent removal under reduced pressure afforded the crude product, which was purified by flash column chromatography on silica gel. 4.1.3. Synthesis of (Z)-Benzyl 2,3-O-(1-methylethylidene)-5deoxy-N-benzyl-D-lyxofuranosylamine N-Oxide (10a). 14 Application of the general procedure to 200 mg (0.72 mmol) of 8 with 94 μL (0.86 mmol) of benzylamine (13a) furnished, after purification by column chromatography (PEt/EtOAc 3:1), 221 mg (0.56 mmol, 80%) of 10a as a white solid (R f = 0.40, PEt/EtOAc 2:1).
4.1.15.2. Procedure with Lewis Acid. To a stirred solution of nitrone 10a in dry THF (0.03 M) at room temperature, boron trifluoride diethyl etherate (0.82 mmol, 100 μL) was added, and the resulting mixture was stirred at room temperature under nitrogen atmosphere for 15 min. The reaction mixture was cooled at −30°C and 2.0 M solution of octylmagnesium bromide in diethyl ether (1.48 mmol, 800 μL) was slowly added. The reaction mixture was stirred at −30°C under nitrogen atmosphere for 2 h, when a TLC control (Hex/EtOAc 2:1) attested the disappearance of the starting material. A saturated ammonium chloride solution (10 mL) and Et 2 O (10 mL) were added to the mixture at 0°C and stirred for 20 min. The two layers were separated, and the aqueous layer was extracted with Et 2 O (2 × 10 mL). The combined organic layers were washed with brine (2 × 30 mL), dried with Na 2 SO 4 , and concentrated under reduced pressure to give a mixture of hydroxylamines 21 and 22 (22 > 98%). The crude mixture was purified by silica gel column chromatography (gradient eluent from Hex/EtOAc 12:1 to 10:1) to give 298 mg (0.57 mmol, 70%) of 22 (R f = 0.20, Hex/EtOAc 12:1) as a colorless oil. The secondary hydroxylamine 22 spontaneously oxidizes to the corresponding nitrone 24, so we could only perform the 1 H NMR and MS-ESI spectra immediately after their purification by column chromatography.
24: straw yellow oil.  1161,1209,1261,1496,1597,2237,2856,2927,3032,3066 (25). To a mixture of nitrone 23 and hydroxylamine 21 in dry MeOH (0.015 M), acetic acid (2 equivalents) and Pd/C (107 mg) were added under nitrogen atmosphere. The mixture was stirred at room temperature under hydrogen atmosphere (balloon) for 2 days, until a control by 1 H NMR spectroscopy attested the presence of the acetate salt of compound 25. The mixture was filtered through Celite, and the solvent was removed under reduced pressure. The corresponding free amine was obtained by dissolving the residue in MeOH; then, the strongly basic resin Ambersep 900-OH was added, and the mixture was stirred for 40 min. The resin was removed by filtration, and the crude product was purified on silica gel by flash column chromatography (CH 2 Cl 2 /MeOH/NH 4 OH (6%) 10:1:0.1) to afford 150 mg (0.38 mmol, 90%) of 25 (R f = 0.50, CH 2 Cl 2 /MeOH/ NH 4 OH (6%) 10:1:0.1) as a white solid.  (100) 11.91;N,3.52. Found: C,72.50;H,12.05;N,3.85. 4.1.19. Synthesis (2R,3R,4S,5R)-3-Hydroxy-4,5-O-(1-methylethylidene)-2-octyl-N-octyl-piperidine (26). To a mixture of nitrone 24 and hydroxylamine 22 in dry MeOH (0.015 M), acetic acid (2 equivalents) and Pd/C (125 mg) were added under nitrogen atmosphere. The mixture was stirred at room temperature under hydrogen atmosphere (balloon) for 2 days, until a control by 1 H NMR spectroscopy attested the presence of the acetate salt of compound 26. The mixture was filtered through Celite, and the solvent was removed under reduced pressure. The corresponding free amine was obtained by dissolving the residue in MeOH; then, the strongly basic resin Ambersep 900-OH was added, and the mixture was stirred for 40 min. The resin was removed by filtration, and the crude product was purified on silica gel by flash column chromatography (CH 2 Cl 2 /MeOH/NH 4 OH (6%) 10:1:0.1) to afford of the enzyme. Enzyme and inhibitor were pre-incubated for 5 min at rt, and the reaction started by addition of the substrate. After 20 min of incubation at 37°C, the reaction was stopped by addition of 0.1 mL of sodium borate solution (pH 9.8). The p-nitrophenolate formed was measured by visible absorption spectroscopy at 405 nm (Asys Expert 96 spectrophotometer). Under these conditions, the pnitrophenolate released led to optical densities linear with both reaction time and concentration of the enzyme.
4.2.2.2. IC 50 Determination. The IC 50 values of inhibitors against GCase were determined by measuring the initial hydrolysis rate with 4-methylumbelliferyl-β-D-glucoside (3.33 mM). Data obtained were fitted by using the appropriate Equation (for more details, see the Supporting Information section).
4.2.3. Preliminary Biological Screening toward Human Lysosomal Glycosidases. The effect of 1 mM concentration of 14, 15, and 33·HCl was assayed toward six lysosomal glycosidases other than GCase, namely, α-mannosidase, β-mannosidase, α-galactosidase, βgalactosidase, α-fucosidase from leukocytes isolated from healthy donors (controls) and α-glucosidase from lymphocytes isolated from healthy donors' flesh blood (controls). Isolated leukocytes or lymphocytes were disrupted by sonication, and a micro BCA protein assay kit (Sigma-Aldrich) was used to determine the total protein amount for the enzymatic assay, according to the manufacturer's instructions (for more details, see the Supporting Information section).
4.2.4. Pharmacological Chaperoning Activity. Fibroblasts with the N370S/RecNcil (or L444P/L444P) mutation from Gaucher disease patients were obtained from the "Cell line and DNA Biobank from patients affected by Genetic Diseases" (Gaslini Hospital, Genova, Italy). Fibroblast cells (15.0 × 10 4 ) were seeded in T25 flasks with DMEM supplemented with fetal bovine serum (10%), penicillin/streptomycin (1%), and glutamine (1%) and incubated at 37°C with 5% CO 2 for 24 h. The medium was removed, and fresh medium containing the compounds (14,15 and 33·HCl) was added to the cells and incubated for 4 days. The medium was removed, and the cells were washed with PBS and detached with trypsin to obtain cell pellets, which were washed four times with PBS, frozen, and lysed by sonication in water. Enzyme activity was measured as reported above. Reported data are mean S.D. (n = 2). 4.2.5. Cytotoxicity Test. The MTT test was carried out using the human fibroblasts wild type at different concentrations of compound 33·HCl. Fibroblasts were grown in the presence of Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), 1% glutamine, and 1% penicillin−streptomycin, at 37°C in controlled atmosphere with 5% CO 2 . For the experiments, cells were seeded at a density of 20,000 cells per well in 24-well plates and grown for 24 h (or 48 h) before adding the compound. The compound was dissolved in water; then, aliquots of these were diluted in the growth medium. To preserve sterility of solutions, these were filtered with 0.22 μm filters before adding to the dishes containing fibroblasts. Then, cells were incubated at 37°C in 5% CO 2 for 24 h (or 48 h). After this time, the media were replaced with medium containing 0.5 mg/mL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT); the cells were incubated for an additional 1 h at 37°C in 5% CO 2 . Finally, the number of viable cells was quantified by the estimation of their dehydrogenase activity, which reduces MTT to water-insoluble formazan. Growth medium was removed and substituted with 300 μL of DMSO to dissolve the formazan produced. The quantitation was carried out measuring the absorbance of samples at 570 nm with the iMark microplate absorbance reader (BIO RAD) in a 96-well format.
Copies of 1 H NMR and 13 C NMR spectra of all new compounds, 1D NOESY spectra for compounds 25 and 26; biological screening toward commercial glycosidases; percentage of GCase inhibition of compounds 14, 15, and 33·HCl toward human GCase. IC 50 for compounds 14, 15, and 33·HCl toward GCase; biological screening toward human lysosomal glycosidases; chaperoning activity assays on fibroblasts with the N370S/RecNcil and L444P/L444P mutations from Gaucher disease patients; and computational methods (docking and MD simulations), docked poses, final snapshots of MD simulations and full data of MMGBSA and MMPBSA data (PDF)