Synthesis of “All-Cis” Trihydroxypiperidines from a Carbohydrate-Derived Ketone: Hints for the Design of New β-Gal and GCase Inhibitors

Pharmacological chaperones (PCs) are small compounds able to rescue the activity of mutated lysosomal enzymes when used at subinhibitory concentrations. Nitrogen-containing glycomimetics such as aza- or iminosugars are known to behave as PCs for lysosomal storage disorders (LSDs). As part of our research into lysosomal sphingolipidoses inhibitors and looking in particular for new β-galactosidase inhibitors, we report the synthesis of a series of alkylated azasugars with a relative “all-cis” configuration at the hydroxy/amine-substituted stereocenters. The novel compounds were synthesized from a common carbohydrate-derived piperidinone intermediate 8, through reductive amination or alkylation of the derived alcohol. In addition, the reaction of ketone 8 with several lithium acetylides allowed the stereoselective synthesis of new azasugars alkylated at C-3. The activity of the new compounds towards lysosomal β-galactosidase was negligible, showing that the presence of an alkyl chain in this position is detrimental to inhibitory activity. Interestingly, 9, 10, and 12 behave as good inhibitors of lysosomal β-glucosidase (GCase) (IC50 = 12, 6.4, and 60 µM, respectively). When tested on cell lines bearing the Gaucher mutation, they did not impart any enzyme rescue. However, altogether, the data included in this work give interesting hints for the design of novel inhibitors.


Introduction
Lysosomal storage disorders (LSDs) are a group of more than 70 inherited orphan diseases caused by specific mutations in genes encoding lysosomal enzymes and characterized by progressive accumulation of substrates within the lysosomes, leading to organ dysfunctions [1][2][3][4].
Defective activity of lysosomal β-galactosidase (β-Gal), which is responsible for the hydrolytic removal of a terminal β-galactose residue from several glycoconjugates, leads to two different types of rare LSDs, the sphingolipidosis GM1-gangliosidosis and the mucopolysaccharidosis IVB (MPS IVB, also known as Morquio disease type B) [5,6]. Typical substrates of β-Gal are GM1-gangliosides, glycoproteins, oligosaccharides, and glycosaminoglycan keratan sulfate, which accumulate inside the lysosomes due to the malfunctioning enzyme, which usually presents a mutation in its natural amino acid sequence. To date, no pharmacological treatment is available for these severe diseases.
Pharmacological chaperone therapy (PCT) is emerging as a promising therapeutic approach for the treatment of LSDs, especially for those that cannot be treated with enzyme replacement therapy (ERT). PCT is based on pharmacological chaperones (PCs), small molecules that can selectively bind mutant enzymes in the endoplasmic reticulum (ER) and stabilize their correct three-dimensional conformation when used at subinhibitory concentrations, thus improving lysosomal trafficking and rescuing enzymatic activity. PCs are reversible inhibitors and can be replaced by the natural substrate of the enzyme inside the lysosomes. Their main pros include oral administration, broad body distribution (PCs have the potential to cross the blood-brain barrier), and minor side effects [7][8][9][10][11].
Nitrogen-containing glycomimetics, such as iminosugars (carbohydrates analogues with a nitrogen atom replacing the endocyclic oxygen), are the most investigated class of PCs for LSDs [12,13].
Recently, the first commercially available PC for treating lysosomal Fabry disease, Galafold TM (1-deoxygalactonojirimycin, DGJ, 1, Figure 1), has been marketed in Europe [14]. Previous studies have shown that the N-alkylated derivative of DGJ N-nonyl-DGJ (2, Figure 1) is able to rescue the intracellular activity of mutant β-Gal in GM1-gangliosidosis patient fibroblasts, thus highlighting the potential of PC therapy for patients with brain pathologies [15,16]. In addition, several azasugars, carbohydrate analogues in which nitrogen formally replaces the anomeric carbon, have shown interesting biological activities, as recently reviewed by Simone and coworkers [17].
In particular, 4-epi-isofagomine (3, Figure 1) showed moderate inhibition of human lysosomal β-Gal (IC50 = 1 µM) [18]. Introduction of an alkyl chain as in compounds 4 and 5 ( Figure 1) enhanced the inhibitory potency up to IC50 = 10 and IC50 = 0.4 nM, respectively. These compounds were able to rescue mutant β-Gal activity in fibroblasts from GM1-gangliosidosis patients [19,20]. Moreover, the "all-cis" trihydroxypiperidines 6 and 7 ( Figure 1) were able to increase β-Gal activity in GM1 gangliosidosis patient fibroblasts up to 2-6 fold [21]. A commonly employed strategy for the design of a potential PC is to modify a natural inhibitor of the compromised enzyme with the aim of increasing its efficacy and selectivity. However, since the inhibitory potency of a synthetic analogue does not always correlate with its performance as a PC, a reliable structure-activity relationship (SARs) for PCs is not easy and needs to be demonstrated by experimental evidence [21]. In particular, 4-epi-isofagomine (3, Figure 1) showed moderate inhibition of human lysosomal β-Gal (IC 50 = 1 µM) [18]. Introduction of an alkyl chain as in compounds 4 and 5 ( Figure 1) enhanced the inhibitory potency up to IC 50 = 10 and IC 50 = 0.4 nM, respectively. These compounds were able to rescue mutant β-Gal activity in fibroblasts from GM1-gangliosidosis patients [19,20]. Moreover, the "all-cis" trihydroxypiperidines 6 and 7 ( Figure 1) were able to increase β-Gal activity in GM1 gangliosidosis patient fibroblasts up to 2-6 fold [21].
A commonly employed strategy for the design of a potential PC is to modify a natural inhibitor of the compromised enzyme with the aim of increasing its efficacy and selectivity. However, since the inhibitory potency of a synthetic analogue does not always correlate with its performance as a PC, a reliable structure-activity relationship (SARs) for PCs is not easy and needs to be demonstrated by experimental evidence [21]. Scheme 1. Compounds synthesized in this work through the functionalization of ketone 8 via: (i) reduction to alcohol and Williamson reaction to ether 9, (ii) reductive amination with dodecyl amine to access 10, (iii) addition of organolithium derivatives to finally obtain compounds 11-15 .

Synthesis
Ketone 8, precursor of all the new compounds, was synthesized as reported from aldehyde 16, derived in turn from inexpensive D-mannose in four steps with a high overall yield (85%). The piperidine skeleton of 17 was obtained through a double reductive amination procedure (DRA) [25][26][27] followed by protection of the endocyclic nitrogen atom with a tert-butyloxycarbonyl (Boc) group. Oxidation of 17 with Dess Martin periodinane (DMP) gave ketone 8, which was diastereoselectively reduced to the "all-cis" alcohol 18 with NaBH4 in EtOH (Scheme 2) [25]. The "all-cis" ether 19 was obtained by Williamson synthesis following a recently reported procedure on a similar substrate [28]. Treatment of alcohol 18 with sodium hydride in dry DMF, alkylation with 1-nonyl bromide gave ether 19 with a 53% yield. Concomitant deprotection of acetonide and Boc groups under acidic conditions (aqueous HCl in MeOH), followed by treatment with the strongly basic resin Ambersep 900 OH, gave ether 9 with a 70% yield (Scheme 3). In order The "all-cis" ether 19 was obtained by Williamson synthesis following a recently reported procedure on a similar substrate [28]. Treatment of alcohol 18 with sodium hydride in dry DMF, alkylation with 1-nonyl bromide gave ether 19 with a 53% yield. Concomitant deprotection of acetonide and Boc groups under acidic conditions (aqueous HCl in MeOH), followed by treatment with the strongly basic resin Ambersep 900 OH, gave ether 9 with a 70% yield (Scheme 3). In order to investigate the role of different configurations at C-3 on biological activity, the diastereomeric alcohol 17 was treated similarly to give protected ether 20 with a 54% yield. Deprotection of 20 as above yielded ether 21 quantitatively (Scheme 3). to investigate the role of different configurations at C-3 on biological activity, the diastereomeric alcohol 17 was treated similarly to give protected ether 20 with a 54% yield. Deprotection of 20 as above yielded ether 21 quantitatively (Scheme 3).

Scheme 3.
Synthetic strategies to obtain ethers 9 and 21 and the amine 10.
Ketone 8 was also converted to the amine 22 through reductive amination employing dodecyl amine. The reaction was performed in dry MeOH under catalytic hydrogenation on Pd(OH)2/C and gave, after purification by flash column chromatography (FCC), the protected piperidine 22 with a 55% yield. Final deprotection under acidic conditions followed by treatment with Ambersep 900 OH resin gave the free amine 10 with an 83% yield (Scheme 3). The "all-cis" configuration in compound 22 was established by comparison of its 1 H-NMR spectrum with those of similar compounds with a different alkyl chain at the exocyclic nitrogen atom [29]. The broad singlet at 4.41 ppm observed for 4-H in the 1 H-NMR spectrum of 22 is consistent with eq-ax relationships with both 3-H and 5-H occurring in its chair conformations.
The synthesis of "all-cis" trihydroxypiperidines alkylated at C-3 by addition of organometallic reagents to the carbonyl group of 8 was then addressed. Ketone 8 was first reacted with Grignard reagents (octylMgBr, ethylMgBr, and methylMgBr, Scheme 4) under conditions which had proved successful for Grignard additions to aldehyde 16 and a nitrone derived thereof [23,30].
The addition of ethyl magnesium bromide to a simpler N-Boc-protected 4-piperidone has been previously reported [31]. However, Grignard additions to ketone 8 proved difficult and sluggish. Variation of reaction conditions (temperature ranging from −78 °C to room temperature, Grignard equivalents ranging from 1 to 1.8, see the Supplementary Materials for further details) did not lead to substantial improvements, and the desired alcohols 23, 24, or 25 were obtained with yields of 12% or less, in complex mixtures difficult to purify. In case of 23 and 24, a single diasteromeric adduct was isolated (as assessed by ESI-MS and 1 H-NMR), while with methyl magnesium bromide a mixture of two diastereoisomers (about 12% yield) was observed in the 1 H-NMR spectrum. The low yields can be ascribed to formation of the reduction product 18. Reduction of ketones by Grignard reagents through a β-hydride addition is known as an undesired side reaction that can occur with particularly encumbered substrates as a consequence of the reduced rate of nucleophilic addition [32,33]. Formation of 18 was attested by analysis of the 1 H-NMR and ESI-MS spectra and confirmed after Scheme 3. Synthetic strategies to obtain ethers 9 and 21 and the amine 10.
Ketone 8 was also converted to the amine 22 through reductive amination employing dodecyl amine. The reaction was performed in dry MeOH under catalytic hydrogenation on Pd(OH) 2 /C and gave, after purification by flash column chromatography (FCC), the protected piperidine 22 with a 55% yield. Final deprotection under acidic conditions followed by treatment with Ambersep 900 OH resin gave the free amine 10 with an 83% yield (Scheme 3). The "all-cis" configuration in compound 22 was established by comparison of its 1 H-NMR spectrum with those of similar compounds with a different alkyl chain at the exocyclic nitrogen atom [29]. The broad singlet at 4.41 ppm observed for 4-H in the 1 H-NMR spectrum of 22 is consistent with eq-ax relationships with both 3-H and 5-H occurring in its chair conformations.
The synthesis of "all-cis" trihydroxypiperidines alkylated at C-3 by addition of organometallic reagents to the carbonyl group of 8 was then addressed. Ketone 8 was first reacted with Grignard reagents (octylMgBr, ethylMgBr, and methylMgBr, Scheme 4) under conditions which had proved successful for Grignard additions to aldehyde 16 and a nitrone derived thereof [23,30]. Given the poor results obtained with alkyl Grignard reagents, we investigated the addition reaction with less bulky sp 2 and sp organometals. The reaction of 8 with vinyl magnesium bromide was less sluggish and a single diastereoisomeric adduct 26 (see below for structural assignment) was obtained with a 51% yield (Scheme 4). Hydrogenation of 26 in the presence of Pd/C under acidic conditions (aqueous HCl in EtOH) allowed complete deprotection and reduction of the alkene. Final treatment with Ambersep 900 OH and purification by FCC gave the saturated azasugar 13 with a 52% yield (Scheme 4). Encouraged by this result, we turned our attention to the addition of alkynyl organometallic reagents, namely, lithium acetylides generated in situ by treating terminal alkynes with butyl lithium. Compared to many C-nucleophiles, acetylides are less basic and less sensitive to steric congestion [34]. The results of the addition of structurally differentiated lithium acetylides to ketone 8 are reported in Table 1   The addition of ethyl magnesium bromide to a simpler N-Boc-protected 4-piperidone has been previously reported [31]. However, Grignard additions to ketone 8 proved difficult and sluggish. Variation of reaction conditions (temperature ranging from −78 • C to room temperature, Grignard Molecules 2020, 25, 4526 5 of 23 equivalents ranging from 1 to 1.8, see the Supplementary Materials for further details) did not lead to substantial improvements, and the desired alcohols 23, 24, or 25 were obtained with yields of 12% or less, in complex mixtures difficult to purify. In case of 23 and 24, a single diasteromeric adduct was isolated (as assessed by ESI-MS and 1 H-NMR), while with methyl magnesium bromide a mixture of two diastereoisomers (about 12% yield) was observed in the 1 H-NMR spectrum. The low yields can be ascribed to formation of the reduction product 18. Reduction of ketones by Grignard reagents through a β-hydride addition is known as an undesired side reaction that can occur with particularly encumbered substrates as a consequence of the reduced rate of nucleophilic addition [32,33]. Formation of 18 was attested by analysis of the 1 H-NMR and ESI-MS spectra and confirmed after acetylation of the free OH group (see the Supplementary Materials) (Scheme 4) in case of octyl magnesium bromide addition.
Given the poor results obtained with alkyl Grignard reagents, we investigated the addition reaction with less bulky sp 2 and sp organometals. The reaction of 8 with vinyl magnesium bromide was less sluggish and a single diastereoisomeric adduct 26 (see below for structural assignment) was obtained with a 51% yield (Scheme 4). Hydrogenation of 26 in the presence of Pd/C under acidic conditions (aqueous HCl in EtOH) allowed complete deprotection and reduction of the alkene. Final treatment with Ambersep 900 OH and purification by FCC gave the saturated azasugar 13 with a 52% yield (Scheme 4).
Encouraged by this result, we turned our attention to the addition of alkynyl organometallic reagents, namely, lithium acetylides generated in situ by treating terminal alkynes with butyl lithium. Compared to many C-nucleophiles, acetylides are less basic and less sensitive to steric congestion [34]. The results of the addition of structurally differentiated lithium acetylides to ketone 8 are reported in Table 1 . obtained with a 51% yield (Scheme 4). Hydrogenation of 26 in the presence of Pd/C under acidic conditions (aqueous HCl in EtOH) allowed complete deprotection and reduction of the alkene. Final treatment with Ambersep 900 OH and purification by FCC gave the saturated azasugar 13 with a 52% yield (Scheme 4). Encouraged by this result, we turned our attention to the addition of alkynyl organometallic reagents, namely, lithium acetylides generated in situ by treating terminal alkynes with butyl lithium. Compared to many C-nucleophiles, acetylides are less basic and less sensitive to steric congestion [34]. The results of the addition of structurally differentiated lithium acetylides to ketone 8 are reported in Table 1  The alkynes were treated with BuLi (1.5 equiv.) at −78 • C for 30 min, followed by the addition of ketone 8 at −78 • C. The reaction mixture was allowed to warm to room temperature and left to react for 2.5-3 h, then worked up. In all reported cases, FCC of the crude mixtures afforded good yields (65-88%) of adducts with both simple and functionalized alkenes (Table 1). A single diastereoisomer was obtained in all cases, which was ascribed the (S) configuration at the newly formed C-3 stereocenter on the basis of the following considerations. Alkynyl lithium derivatives are small nucleophiles, which typically prefer an axial rather than an equatorial attack on cyclohexanones [35][36][37] in order to avoid torsional strain [38,39]. In Scheme 5, the two more stable chair conformations of ketone 8 are depicted. Only in the 6 C 3 conformation, nucleophilic axial attack at C-3 experiences a stabilizing interaction by the low-lying energy σ* orbital of the antiperiplanar C-O bond at C-4, according to a favorable Felkin-Anh model [40,41]. Thus, based on stereoelectronic and steric considerations, we assumed that nucleophilic attacks occurred selectively at the Re face of ketone 8, giving the "all-cis" 3, 4, 5-trihydroxypiperidines (Scheme 5). A careful analysis of the 1 H NMR spectra of derivatives 26-31 and the products of their transformations supports this structural assignment (see below).
was obtained in all cases, which was ascribed the (S) configuration at the newly formed C-3 stereocenter on the basis of the following considerations. Alkynyl lithium derivatives are small nucleophiles, which typically prefer an axial rather than an equatorial attack on cyclohexanones [35][36][37] in order to avoid torsional strain [38,39]. In Scheme 5, the two more stable chair conformations of ketone 8 are depicted. Only in the 6 C3 conformation, nucleophilic axial attack at C-3 experiences a stabilizing interaction by the low-lying energy σ* orbital of the antiperiplanar C-O bond at C-4, according to a favorable Felkin-Anh model [40,41]. Thus, based on stereoelectronic and steric considerations, we assumed that nucleophilic attacks occurred selectively at the Re face of ketone 8, giving the "all-cis" 3, 4, 5-trihydroxypiperidines (Scheme 5). A careful analysis of the 1 H NMR spectra of derivatives 26-31 and the products of their transformations supports this structural assignment (see below).

Scheme 5.
Stereochemical outcome of the addition of lithium acetylides to ketone 8.
Compounds 27, 28, and 30 were subjected to acidic treatment for the concomitant removal of acetonide and Boc-protecting groups, leading to the "all-cis" trihydroxypiperidines 32, 33, and 34 with good yields (51-71%) after treatment with the strongly basic resin Ambersep 900 OH. The triple bond was subsequently reduced by catalytic hydrogenation in the presence of Pd(OH)2/C in EtOH to give azasugars 11, 12, and 14 (39-98%) (Scheme 6). Due to the presence of the additional acid-labile acetal moiety, adduct 29 was expected to give complications arising from possible interactions of the free amine with the aldehyde and was not subjected to acid-induced deprotection. Unsuccessful deprotection of 31 with different acids Compounds 27, 28, and 30 were subjected to acidic treatment for the concomitant removal of acetonide and Boc-protecting groups, leading to the "all-cis" trihydroxypiperidines 32, 33, and 34 with good yields (51-71%) after treatment with the strongly basic resin Ambersep 900 OH. The triple bond was subsequently reduced by catalytic hydrogenation in the presence of Pd(OH) 2 /C in EtOH to give azasugars 11, 12, and 14 (39-98%) (Scheme 6).
was obtained in all cases, which was ascribed the (S) configuration at the newly formed C-3 stereocenter on the basis of the following considerations. Alkynyl lithium derivatives are small nucleophiles, which typically prefer an axial rather than an equatorial attack on cyclohexanones [35][36][37] in order to avoid torsional strain [38,39]. In Scheme 5, the two more stable chair conformations of ketone 8 are depicted. Only in the 6 C3 conformation, nucleophilic axial attack at C-3 experiences a stabilizing interaction by the low-lying energy σ* orbital of the antiperiplanar C-O bond at C-4, according to a favorable Felkin-Anh model [40,41]. Thus, based on stereoelectronic and steric considerations, we assumed that nucleophilic attacks occurred selectively at the Re face of ketone 8, giving the "all-cis" 3, 4, 5-trihydroxypiperidines (Scheme 5). A careful analysis of the 1 H NMR spectra of derivatives 26-31 and the products of their transformations supports this structural assignment (see below).

Scheme 5.
Stereochemical outcome of the addition of lithium acetylides to ketone 8.
Compounds 27, 28, and 30 were subjected to acidic treatment for the concomitant removal of acetonide and Boc-protecting groups, leading to the "all-cis" trihydroxypiperidines 32, 33, and 34 with good yields (51-71%) after treatment with the strongly basic resin Ambersep 900 OH. The triple bond was subsequently reduced by catalytic hydrogenation in the presence of Pd(OH)2/C in EtOH to give azasugars 11, 12, and 14 (39-98%) (Scheme 6). Due to the presence of the additional acid-labile acetal moiety, adduct 29 was expected to give complications arising from possible interactions of the free amine with the aldehyde and was not subjected to acid-induced deprotection. Unsuccessful deprotection of 31 with different acids Due to the presence of the additional acid-labile acetal moiety, adduct 29 was expected to give complications arising from possible interactions of the free amine with the aldehyde and was not subjected to acid-induced deprotection. Unsuccessful deprotection of 31 with different acids (CF 3 COOH, CH 3 COOH, and HCl) was ascribed to the higher reactivity of its triple bond. Therefore, hydrogenation of the alkyne under neutral conditions (H 2 , Pd/C in EtOH) was first carried out to give piperidine 35, which was subsequently deprotected with aqueous HCl in MeOH. Final treatment with the strongly basic resin Ambersep 900 OH gave trihydroxypiperidine 15 with a 56% yield (Scheme 6).

Configuration Assignment
Relevant chemical shifts and coupling constants are reported in Tables 2 and 3 for H-4, H-5, and H-6 in 1 H-NMR spectra in selected compounds of the two series of protected (in CDCl 3 ) and deprotected (in CD 3 OD) trihydroxypiperidines, respectively. These values show regularities (the same applies to H-2 signals), which allowed us to ascribe the same configuration at C-3 for all compounds (see the Supplementary Materials for the full Table). Moreover, the shape of the signals and their coupling constants, where detectable, are consistent with the (S) absolute configuration tentatively assigned (see above) on the basis of mechanistic considerations. Indeed, the signal of H-5 appears as a broad singlet (or as a narrow multiplet), which is in agreement with its equatorial position in a preferred chair conformation, which places the R substituent equatorially, i.e., in the (3S) configuration ( 6 C 3 alcohol in Scheme 5). The lack of large ax-ax coupling constants is confirmed by signals of H-6. For example, in piperidine 13 (R = ethyl), the two hydrogens at C-6 display vicinal coupling constants J = 3.4 and J = 2.4 Hz, typical for ax-eq and eq-eq relationships, respectively. The same applies to the other derivatives when the signals are well resolved, as in compounds 11, 12, 14, and 15. (CF3COOH, CH3COOH, and HCl) was ascribed to the higher reactivity of its triple bond. Therefore, hydrogenation of the alkyne under neutral conditions (H2, Pd/C in EtOH) was first carried out to give piperidine 35, which was subsequently deprotected with aqueous HCl in MeOH. Final treatment with the strongly basic resin Ambersep 900 OH gave trihydroxypiperidine 15 with a 56% yield (Scheme 6).

Configuration Assignment
Relevant chemical shifts and coupling constants are reported in Tables 2 and 3 for H-4, H-5, and H-6 in 1 H-NMR spectra in selected compounds of the two series of protected (in CDCl3) and deprotected (in CD3OD) trihydroxypiperidines, respectively. These values show regularities (the same applies to H-2 signals), which allowed us to ascribe the same configuration at C-3 for all compounds (see the Supplementary Materials for the full Table). Moreover, the shape of the signals and their coupling constants, where detectable, are consistent with the (S) absolute configuration tentatively assigned (see above) on the basis of mechanistic considerations. Indeed, the signal of H-5 appears as a broad singlet (or as a narrow multiplet), which is in agreement with its equatorial position in a preferred chair conformation, which places the R substituent equatorially, i.e., in the (3S) configuration (6C 3 alcohol in Scheme 5). The lack of large ax-ax coupling constants is confirmed by signals of H-6. For example, in piperidine 13 (R = ethyl), the two hydrogens at C-6 display vicinal coupling constants J = 3.4 and J = 2.4 Hz, typical for ax-eq and eq-eq relationships, respectively. The same applies to the other derivatives when the signals are well resolved, as in compounds 11, 12, 14, and 15. H-6 in 1 H-NMR spectra in selected compounds of the two series of protected (in CDCl3) and deprotected (in CD3OD) trihydroxypiperidines, respectively. These values show regularities (the same applies to H-2 signals), which allowed us to ascribe the same configuration at C-3 for all compounds (see the Supplementary Materials for the full Table). Moreover, the shape of the signals and their coupling constants, where detectable, are consistent with the (S) absolute configuration tentatively assigned (see above) on the basis of mechanistic considerations. Indeed, the signal of H-5 appears as a broad singlet (or as a narrow multiplet), which is in agreement with its equatorial position in a preferred chair conformation, which places the R substituent equatorially, i.e., in the (3S) configuration (6C 3 alcohol in Scheme 5). The lack of large ax-ax coupling constants is confirmed by signals of H-6. For example, in piperidine 13 (R = ethyl), the two hydrogens at C-6 display vicinal coupling constants J = 3.4 and J = 2.4 Hz, typical for ax-eq and eq-eq relationships, respectively. The same applies to the other derivatives when the signals are well resolved, as in compounds 11, 12, 14, and 15.   The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.

Biological Screening
Compounds 9-15, 21, 32, and 33 were first evaluated as human lysosomal β-Gal inhibitors at 1 mM in human leukocyte homogenates and the results are shown in Table 4 and compared to previously published data. Unfortunately, none of the tested compounds strongly inhibited β-Gal (only a moderate 22% inhibition was found for the "all-cis" ether 9). These data demonstrate that both the alkylation of the hydroxy or the exocyclic amine group and the introduction of a substituent at C-3 of the trihydroxypiperidine skeleton dramatically affect β-Gal inhibition.
However, the screening on a panel of 12 commercial glycosidases (see the Supplementary Materials), showed that only trihydroxypiperidine 10 (bearing a dodecyl chain connected at C-3 The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.

Biological Screening
Compounds 9-15, 21, 32, and 33 were first evaluated as human lysosomal β-Gal inhibitors at 1 mM in human leukocyte homogenates and the results are shown in Table 4 and compared to previously published data. Unfortunately, none of the tested compounds strongly inhibited β-Gal (only a moderate 22% inhibition was found for the "all-cis" ether 9). These data demonstrate that both the alkylation of the hydroxy or the exocyclic amine group and the introduction of a substituent at C-3 of the trihydroxypiperidine skeleton dramatically affect β-Gal inhibition.  The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.

Biological Screening
Compounds 9-15, 21, 32, and 33 were first evaluated as human lysosomal β-Gal inhibitors at 1 mM in human leukocyte homogenates and the results are shown in Table 4 and compared to previously published data. Unfortunately, none of the tested compounds strongly inhibited β-Gal (only a moderate 22% inhibition was found for the "all-cis" ether 9). These data demonstrate that both the alkylation of the hydroxy or the exocyclic amine group and the introduction of a substituent at C-3 of the trihydroxypiperidine skeleton dramatically affect β-Gal inhibition.
However, the screening on a panel of 12   The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.

Biological Screening
Compounds 9-15, 21, 32, and 33 were first evaluated as human lysosomal β-Gal inhibitors at 1 mM in human leukocyte homogenates and the results are shown in Table 4 and compared to previously published data. Unfortunately, none of the tested compounds strongly inhibited β-Gal (only a moderate 22% inhibition was found for the "all-cis" ether 9). These data demonstrate that both the alkylation of the hydroxy or the exocyclic amine group and the introduction of a substituent at C-3 of the trihydroxypiperidine skeleton dramatically affect β-Gal inhibition.
However, the screening on a panel of 12 commercial glycosidases (see the Supplementary Materials), showed that only trihydroxypiperidine 10 (bearing a dodecyl chain connected at C-3 through a nitrogen atom) was able to inhibit β-glucosidase from almonds. In particular, compound 10 showed an IC50 = 85 µM towards this enzyme, which prompted us to evaluate compounds 9-15,  The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.

Biological Screening
Compounds 9-15, 21, 32, and 33 were first evaluated as human lysosomal β-Gal inhibitors at 1 mM in human leukocyte homogenates and the results are shown in Table 4 and compared to previously published data. Unfortunately, none of the tested compounds strongly inhibited β-Gal (only a moderate 22% inhibition was found for the "all-cis" ether 9). These data demonstrate that both the alkylation of the hydroxy or the exocyclic amine group and the introduction of a substituent at C-3 of the trihydroxypiperidine skeleton dramatically affect β-Gal inhibition.
However, the screening on a panel of 12 commercial glycosidases (see the Supplementary Materials), showed that only trihydroxypiperidine 10 (bearing a dodecyl chain connected at C-3 through a nitrogen atom) was able to inhibit β-glucosidase from almonds. In particular, compound 10 showed an IC50 = 85 µM towards this enzyme, which prompted us to evaluate compounds 9-15, 21, 32, and 33 also on human lysosomal β-glucosidase (GCase) ( The observed upfield shift (0.3-0.5 ppm) of H-4 within the two series of compounds on turning from the alkynyl to the saturated substituents (see for instance 32 vs. 11), consistent with H-4 falling in the deshielding cone of the triple bond in the former derivatives when in a cis relationship, further supports this assignment.
These data overall show that GCase inhibition higher than 90% is guaranteed by the presence of a long alkyl chain (8, 9, or 12 carbon atoms, compounds 9, 10, 12, and 21). This parallels previous studies, which showed that alkylated imino-and azasugars are strong GCase inhibitors due to favorable interaction of the alkyl chain with the hydrophobic domain of the enzyme [24,44,45]. However, in terms of IC50 values, a remarkable difference was found between compounds 9 and 21, in agreement with previous observations by Compain and coworkers on differently configured ethers derived from 1,5-dideoxy and 1-5-imino-D-xylitol (DIX) [46]. Moreover, good GCase inhibitors can be identified among piperidines with only two free hydroxy groups (e.g., 9, 10, and 21), as previously observed with different azasugars [46].
We then assayed the strongest GCase inhibitors 9, 10, and 12 (IC50 lower than 100 µM) in human fibroblasts derived from Gaucher patients bearing the N370S mutation. None of the compounds gave enzyme rescue when tested at six different concentrations (10, 100, 1, 10, 50, and 100 µM) (see the Supplementary Materials). In particular, compound 10 showed remarkable toxicity at the highest concentrations (50 and 100 µM). Notably, the stronger inhibitory activity of 10 with respect to 36 (see Table 4) does not correspond to a higher chaperoning activity (indeed, compound 36 was able to rescue GCase activity of 1.5-fold at 100 µM) [22].
The same behavior is evident by comparing the C-2 alkylated trihydroxypiperidines 38 and 37: The best inhibitor was the dodecyl alkylated (compound 38) but the best chaperone was the octyl Conversely compound 33, bearing the triple bond and a longer aliphatic substituent, showed 65% inhibition, which increased considerably after hydrogenation to 12 (Table 4, entry 10 vs. 4). This latter compound showed a moderate IC50 = 60 µM, which is close to that observed for trihydroxypiperidines bearing an octyl chain at C-2, previously synthesized in our group (37 and 39, Table 4, entries 12 and 14) [23,24].
These data overall show that GCase inhibition higher than 90% is guaranteed by the presence of a long alkyl chain (8, 9, or 12 carbon atoms, compounds 9, 10, 12, and 21). This parallels previous studies, which showed that alkylated imino-and azasugars are strong GCase inhibitors due to favorable interaction of the alkyl chain with the hydrophobic domain of the enzyme [24,44,45]. However, in terms of IC50 values, a remarkable difference was found between compounds 9 and 21, in agreement with previous observations by Compain and coworkers on differently configured ethers derived from 1,5-dideoxy and 1-5-imino-D-xylitol (DIX) [46]. Moreover, good GCase inhibitors can be identified among piperidines with only two free hydroxy groups (e.g., 9, 10, and 21), as previously observed with different azasugars [46].
We then assayed the strongest GCase inhibitors 9, 10, and 12 (IC50 lower than 100 µM) in human fibroblasts derived from Gaucher patients bearing the N370S mutation. None of the compounds gave enzyme rescue when tested at six different concentrations (10, 100, 1, 10, 50, and 100 µM) (see the Supplementary Materials). In particular, compound 10 showed remarkable toxicity at the highest concentrations (50 and 100 µM). Notably, the stronger inhibitory activity of 10 with respect to 36 (see Table 4) does not correspond to a higher chaperoning activity (indeed, compound 36 was able to rescue GCase activity of 1.5-fold at 100 µM) [22].
The same behavior is evident by comparing the C-2 alkylated trihydroxypiperidines 38 and 37: The best inhibitor was the dodecyl alkylated (compound 38) but the best chaperone was the octyl However, the screening on a panel of 12 commercial glycosidases (see the Supplementary Materials), showed that only trihydroxypiperidine 10 (bearing a dodecyl chain connected at C-3 through a nitrogen atom) was able to inhibit β-glucosidase from almonds. In particular, compound 10 showed an IC 50 = 85 µM towards this enzyme, which prompted us to evaluate compounds 9-15, 21, 32, and 33 also on human lysosomal β-glucosidase (GCase) ( Table 4). Point mutations in the gene encoding this enzyme cause Gaucher disease, the most common autosomal recessive LSD [42,43].
Conversely compound 33, bearing the triple bond and a longer aliphatic substituent, showed 65% inhibition, which increased considerably after hydrogenation to 12 (Table 4, entry 10 vs. 4). This latter compound showed a moderate IC 50 = 60 µM, which is close to that observed for trihydroxypiperidines bearing an octyl chain at C-2, previously synthesized in our group (37 and 39, Table 4, entries 12 and 14) [23,24].
These data overall show that GCase inhibition higher than 90% is guaranteed by the presence of a long alkyl chain (8, 9, or 12 carbon atoms, compounds 9, 10, 12, and 21). This parallels previous studies, which showed that alkylated imino-and azasugars are strong GCase inhibitors due to favorable interaction of the alkyl chain with the hydrophobic domain of the enzyme [24,44,45]. However, in terms of IC 50 values, a remarkable difference was found between compounds 9 and 21, in agreement with previous observations by Compain and coworkers on differently configured ethers derived from 1,5-dideoxy and 1-5-imino-d-xylitol (DIX) [46]. Moreover, good GCase inhibitors can be identified among piperidines with only two free hydroxy groups (e.g., 9, 10, and 21), as previously observed with different azasugars [46].
We then assayed the strongest GCase inhibitors 9, 10, and 12 (IC 50 lower than 100 µM) in human fibroblasts derived from Gaucher patients bearing the N370S mutation. None of the compounds gave enzyme rescue when tested at six different concentrations (10, 100, 1, 10, 50, and 100 µM) (see the Supplementary Materials). In particular, compound 10 showed remarkable toxicity at the highest concentrations (50 and 100 µM). Notably, the stronger inhibitory activity of 10 with respect to 36 (see Table 4) does not correspond to a higher chaperoning activity (indeed, compound 36 was able to rescue GCase activity of 1.5-fold at 100 µM) [22].
The same behavior is evident by comparing the C-2 alkylated trihydroxypiperidines 38 and 37: The best inhibitor was the dodecyl alkylated (compound 38) but the best chaperone was the octyl derivative 37 [24]. The inefficacy of dodecyl alkylated compounds 10 and 38 as PCs can be ascribed to the cytotoxicity imparted by the 12-carbon atom alkyl chains. These data are also consistent with other reports on amphiphilic N-alkylated iminosugars, which suggest that cytotoxicity is strongly chain-length dependent and that potent inhibitors with chains longer than C 8 can be toxic when assayed in cell lines [47,48]. NaH (7 mg, 0.3 mmol, 60% on mineral oil) was added to a solution of 18 [25] (22 mg, 0.08 mmol) in dry DMF (1.2 mL) at 0 • C. The mixture was stirred at room temperature for 30 min, then 1-bromononane (54 µL, 0.28 mmol) was added, and the reaction mixture was stirred at room temperature for 72 h, until the disappearance of the starting material was observed via TLC (CH 2 Cl 2 /MeOH/NH 4 OH (6%) 10:1:0.1). Then, water was slowly added, and the reaction mixture was extracted with AcOEt (3 × 3 mL). The combined organic layer was washed with saturated NaHCO 3 and brine and concentrated after drying with Na 2 SO 4 . The crude residue was purified by flash column chromatography on silica gel (hexane/AcOEt 8:1) to give 17 mg of 19 (R f = 0.3, hexane/AcOEt 8:1, 0.04 mmol, 53%) as a colorless oil. 3.1.2. Synthesis of (3S, 4R, 5R)-4, 5-Dihydroxy-3-(Nonyloxy) Piperidine (9) A solution of 19 (15 mg, 0.04 mmol) in MeOH (3 mL) was left stirring with 12 M HCl (60 µL) at room temperature for 18 h. The crude mixture was concentrated to yield 9 as the hydrochloride salt. The corresponding free amine was obtained by dissolving the residue in MeOH (4 mL), then the strongly basic resin Ambersep 900 OH was added, and the mixture was stirred for 45 min. The resin was removed by filtration and the crude product was purified on silica gel by flash column chromatography (DCM/MeOH/NH 4 OH (6%) 10:1:0.1) to afford 7 mg of 9 (R f = 0.2, DCM/MeOH/NH 4 OH (6%) 10:1:0.1, 0.03 mmol, 70%) as a pale-yellow oil.  (20) NaH (11 mg, 0.46 mmol, 60% on mineral oil) was added to a solution of 17 [25] (57 mg, 0.21 mmol) in dry DMF (3 mL) at 0 • C. The mixture was stirred at room temperature for 30 min, then 1-bromononane (140 µL, 0.73 mmol) was added, and the reaction mixture was stirred at room temperature for 40 h, until the disappearance of the starting material was observed via TLC (CH 2 Cl 2 /MeOH/NH 4 OH (6%) 10:1:0.1). Then, water was slowly added and the reaction mixture was extracted with AcOEt (3 × 5 mL). The combined organic layer was washed with saturated NaHCO 3 and brine and concentrated after drying with Na 2 SO 4 . The crude residue was purified by flash column chromatography on silica gel (hexane/AcOEt 10:1) to give 45 mg of 20 (R f = 0.4, hexane/AcOEt 7:1, 0.11 mmol, 54%) as a colorless oil.  (21) A solution of 20 (54 mg, 0.14 mmol) in MeOH (6 mL) was left stirring with 12 M HCl (150 µL) at room temperature for 18 h. The crude mixture was concentrated to yield 21 as hydrochloride salt. The corresponding free amine was obtained by dissolving the residue in MeOH (5 mL), then the strongly basic resin Ambersep 900 OH was added, and the mixture was stirred for 45 min. The resin was removed by filtration to give 36 mg of 21 (0.14 mmol, 100% yield) as a pale-yellow oil.  (22) Ketone 8 [25] (63 mg, 0.23 mmol) and dodecylamine (65 mg, 0.35 mmol) were dissolved in MeOH (3 mL), and molecular sieves (3 Å pellets; 25 mg) were added. The reaction mixture was stirred at room temperature for 1 h and then Pd(OH) 2 /C (30 mg) was added. The mixture was further stirred at room temperature under hydrogen atmosphere for 51 h. The catalyst and the molecular sieves were removed by filtration, the obtained compound was washed several times with MeOH, and the solvent was evaporated under vacuum. The crude residue was purified by flash column chromatography on silica gel (gradient eluent from hexane/AcOEt 5:1 to 2:1) to afford 56 mg of 22 (R f = 0.4, hexane/AcOEt 2:1, 0.13 mmol, 55%) as a colorless oil. Vinyl magnesium bromide (288 µL, 0.29 mmol) was added to a dry THF solution (1 mL) of ketone 8 (52 mg, 0.19 mmol), dropwise at 0 • C under nitrogen atmosphere. The solution was stirred at 0 • C for 5 h when the disappearance of 8 was attested by a TLC control (hexane/AcOEt 2:1). A saturated aqueous NH 4 Cl solution was added at 0 • C, and the mixture was stirred for 10 min. The reaction mixture was extracted with AcOEt (3 × 3 mL). The combined organic layer was washed with water, saturated NaHCO 3 , and brine and concentrated after drying with Na 2 SO 4 . The crude residue was purified by flash column chromatography on silica gel (gradient eluent from hexane/AcOEt 5:1 to 2:1) to give 29 mg of 26 (R f = 0.2, hexane/AcOEt 5:1, 0.10 mmol, 51%) as a pale-yellow oil. 3.1.8. Synthesis of (3S, 4R, 5R)-3-Ethyl-3, 4, 5-Trihydroxypiperidine (13) Compound 26 (26 mg, 0.09 mmol) was dissolved in EtOH (5 mL) and HCl 12 M (150 µL) and Pd/C (13 mg) were added. The reaction mixture was stirred at room temperature under hydrogen atmosphere for 24 h. The catalyst was removed by filtration through Celite, and the filtrate was concentrated under vacuum to give the hydrochloride salt of 13. 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 45 min. The resin was removed by filtration and the crude product was purified on silica gel by flash column chromatography (gradient eluent from DCM/MeOH/NH 4 OH (6%) 10:1:0.1 to 1:1:0.1) to give 8 mg of 13 (R f = 0.2, DCM/MeOH/NH 4 OH (6%) 10:1:0.1, 0.05 mmol, 52%) as a colorless oil. To a dry THF solution (0.24 M) of alkyne (2 eq.), n-BuLi (1.5 eq.) was added dropwise over 5 min at -78 • C under nitrogen atmosphere. The solution was allowed to warm to 0 • C over 1 h and held at 0 • C for an additional 30 min. The solution was then recooled to -78 • C and ketone 8 (1 eq.) was added in one portion. The solution was allowed to warm to room temperature and stirred at room temperature until the disappearance of ketone 8 was attested by a TLC control (EtP/AcOEt 2:1). A saturated aqueous solution of NH 4 Cl was added and the reaction mixture was extracted with AcOEt. The combined organic layer was washed with water, saturated NaHCO 3 , and brine and concentrated after drying with Na 2 SO 4 . The crude compound was purified by flash column chromatography.