A New Class of Bi- and Trifunctional Sugar Oximes as Antidotes against Organophosphorus Poisoning

Abstract

Recent events demonstrated that organophosphorus nerve agents are a serious threat for civilian and military populations. The current therapy includes a pyridinium aldoxime reactivator to restore the enzymatic activity of acetylcholinesterase located in the central nervous system and neuro-muscular junctions. One major drawback of these charged acetylcholinesterase reactivators is their poor ability to cross the blood–brain barrier. In this study, we propose to evaluate glucoconjugated oximes devoid of permanent charge as potential central nervous system reactivators. We determined their in vitro reactivation efficacy on inhibited human acetylcholinesterase, the crystal structure of two compounds in complex with the enzyme, their protective index on intoxicated mice, and their pharmacokinetics. We then evaluated their endothelial permeability coefficients with a human in vitro model. This study shed light on the structural restrains of new sugar oximes designed to reach the central nervous system through the glucose transporter located at the blood–brain barrier.

Introduction

The recent use of organophosphorus nerve agents (OPNAs) such as VX against Kim Jong-Nam in Malaysia in 2017¹ or Novichoks in march 2018 against a former Russian spy, Sergei Skripal and his daughter Yulia,² demonstrates that these compounds are still a real threat for the civilian and military populations. Commonly, nerve agents affect the cholinergic neurotransmission by phosphylation of the catalytic serine residue of acetylcholinesterase (AChE, EC 3.1.1.7), a key enzyme for nerve impulse termination, located in the central nervous system (CNS) and at the neuromuscular junctions. This covalent binding leads to the irreversible inhibition of the enzyme, accumulation of the neurotransmitter acetylcholine, and subsequently to a cholinergic crisis characterized by nausea, dyspnea, seizure, and death if not treated rapidly.³ The current therapy for OPNA poisoning associates a muscarinic antagonist drug (e.g., atropine), an anticonvulsant drug (e.g., diazepam), and a pyridinium aldoxime reactivator (pralidoxime, trimedoxime, obidoxime, HI-6),⁴ which are able to remove the phosphyl group attached to the catalytic serine residue. These permanently charged quaternary oximes present an adequate pKa for the nucleophilic oxime residue increasing its nucleophilic character and an electron deficient heteroaromatic moiety responsible for their binding affinity and their positioning close to the phosphylated serine residue in the enzyme catalytic site. However, these charged oximes are known to cross the blood–brain barrier (BBB) poorly. Thus, if this medical countermeasure can limit the deadly peripheral cholinergic crisis, they do not reactivate cholinesterases of the CNS, causing long-lasting neurological disorders and side effects.⁵ Only 4 to 10% of the oxime present in the plasma cross the BBB and are subsequently available for brain protection against OP poisoning.⁶ However, in vitro and in vivo evaluation on OP-exposed mice has demonstrated the potential of zwitterionic oxime RS194B with surprisingly high protective indices despite their low accumulation in the brain of guinea pigs and low permeation in a parallel artificial membrane permeability assay (PAMPA).⁷⁻¹⁰ A recent study described novel substituted phenoxyalkylpyridinium oximes as a potential brain-penetrating AChE reactivator.¹¹ The linker alkyl chain between the pyridinium ring and the substituted phenoxy group increases the lipophilicity of these oximes and therefore assists in counterbalancing the positive charge of the quaternary ammonium. Animal experiments showed a 24 h survival to a lethal OP exposure in rats, a clear improvement compared to 2-PAM, and a reduction of the seizure-like behavior suggesting CNS activity of the molecules. The best reactivator is a poor substrate to P-glycoprotein and therefore can accumulate in the brain and reactivate the OP-inhibited AChE. Despite the synthesis and evaluation of numerous new oximes over the past years, no broad-spectrum oxime able to afford protection against all the main OPNAs has been identified.¹²

In order to facilitate reactivator crossing of the BBB and improve the reactivation efficacy of new oximes in the CNS, many strategies and new compound designs have been explored. Among the different strategies developed recently, two have drawn our attention.^13,14 Reactivators devoid of a permanent charge have been proposed to penetrate more efficiently the BBB and therefore is more centrally active. In vitro results show an increased efficacy of these non-quaternary reactivators.¹⁵⁻¹⁷ An in vivo study of oxime JR595 in mice showed a rapid absorption into blood and up to 40% of its blood concentration in the brain, which is a significant improvement. Despite the fast elimination of this oxime (1.0 to 1.5 h post injection), this uncharged reactivator still provides protection against multiple LD50 of VX and sarin.¹⁸ The efficient BBB crossing efficacy of some of these non-quaternary reactivators has been established.¹⁹ However, in vivo experiments with two other uncharged reactivators GM113 and GM508 did not show a better protection in the mouse model compared to HI-6.²⁰ Another strategy to improve BBB penetration has implied the synthesis of sugar-oxime conjugates.²¹ Indeed, the glucose transporter GLUT-1 located at both sides of the BBB has been shown to potentially facilitate crossing of this physiological barrier of sugar conjugates.²² When applied to AChE reactivators, reports have demonstrated that glucose conjugated-2-PAM attenuates paraoxon-induced hypothermia in rats, suggesting a central effect.^21,23 Structural modifications of these permanently charged glucose conjugated-2-PAM oximes have improved the reactivation kinetics of VX-and GB-inhibited AChE from human red blood cells, but BBB crossing assays showed an asymmetric transfer of the oxime through the MDR1-MDCK cell line, resulting in high oxime concentration trapped inside the cells.²⁴ In addition, a study reported IC₅₀ values ranging from the high μM to low mM range for phosphorylated/thiophosphorylated glucosyl-1,2,3-triazole AChE inhibitors, indicating the possible binding of molecules composed of a sugar moiety conjugated to a triazole.²⁵

Based on these previous results, herein, we report the evaluation of the combination of these two strategies through the development and evaluation of a new class of uncharged oximes designed to reach inhibited AChE of the CNS through the glucose transporter GLUT-1 located at the BBB. Among the uncharged oximes evaluated so far, 6- substituted 3-hydroxypyridinaldoximes have shown the best in vitro AChE reactivation efficacy. We thus synthesized multi-functional molecules bearing this 3-hydroxypyridinaldoxime moiety as a reactivator function attached in position 6 of the pyridine to a sugar (glucose or ribose) aiming at GLUT-1 facilitated BBB crossing of the glycoconjugate. One of the prerequisites for efficient AChE reactivation is the ability of the reactivator to bind phosphylated AChE near the phosphylated serine residue. The 3-hydroxypyridinaldoxime moiety displaying a limited affinity for the AChE active site and structural studies prompted us to increase the affinity of the sugar-oximes for the phosphylated AChE through the introduction of a triazole heterocycle between the sugar and the 3-hydroxypyridinaldoxime. Accordingly, docking experiments and previously obtained structures of different ligands bound to AChE have shown that such a triazole moiety could improve binding to the gorge of AChE by stacking with aromatic amino acids of the gorge.^26,27 Binding of the triazole should thus (1) prevent the sugar moiety to interfere with the reactivation process and (2) increase the reactivator efficacy toward inhibited enzymes thanks to an increased binding affinity. In this study, we performed an exhaustive evaluation of these newly designed sugar-oximes. In order to guide the synthetic efforts, we first investigated the binding properties of the newly designed molecules by molecular docking. Next, the most promising candidates were synthesized, and we evaluated their reactivation efficacy on recombinant hAChE inhibited by various OPNAs and determined the crystallographic structures of the complexes formed between two of the synthesized sugar-oxime conjugates and hAChE. Then, we determined the protective index of selected sugar-oximes on mice exposed to OPNAs and finally, we checked the BBB crossing abilities of these new sugar-oxime molecules with an in vitro model mimicking the human BBB. Analysis of these findings allowed us to shed new light on the different issues associated with BBB crossing by such chemical counter-measures against OPNA poisoning and opens the way to the development of a new family of reactivators.

Results

Chemistry

In order to evaluate the requirement of an additional triazole moiety to increase reactivator affinity with phosphylated AChE, a first series of 3-hydroxypyridinealdoxime-glucose conjugate was synthesized without the triazole ring. The sugar moiety was attached from its anomeric position (β isomer was chosen according to the literature describing efficient GLUT-1 mediated BBB crossing using such glycoconjugates) to the position 6 of the pyridine ring by an alkyl chain of four, three, or six carbons, yielding compounds 1, 2, and 15 in 48, 28, and 23% yield, respectively, over four steps (Scheme 1). Peracylated glucose bearing a terminal alkyne moiety at its anomeric position of different lengths was obtained as previously described.²⁸ A Sonogashira cross-coupling reaction was used to link the glycosyl moiety and a pre-functionalized pyridine ring using a standard procedure (CuI, Pd(PPh₃)₄) followed by hydrogenation in the presence of a Pearlman palladium catalyst.¹⁸ Finally, the acetals were removed by a mild and efficient procedure (LiCl in DMSO/H₂O),²⁹ forming the corresponding 3-hydroxypiconilaldehyde, which was converted to the aldoxime through reaction with hydroxylamine in methanol. Preparative RP-HPLC purified final compounds, and their purity (see Experimental Section: Chemistry) was determined by RP-HPLC.

Scheme 1. Alkyl Chain-Containing Aldoxime Synthetic Pathway.

Aldoximes 3, 4, 4′, and 5 (Scheme 2) bearing a triazole heterocycle between the sugar moiety and the position 6 of the pyridinaldoxime moiety were obtained using Huisgen cycloaddition between an alkyne chain linked to the glycosyl and an azidoalkynylpyridine. Backed by the docking experiments, a 4 carbon atom linker was found sufficient between the triazole moiety and the sugar to keep the sugar moiety away from AChE peripheral site. Docking experiments (vide infra) showed that the best binding affinities were obtained with a 3 or 4 carbon atom distance between the oxime and the triazole ring; thus, the β-glucoconjugates 3 and 4 were first targeted. In addition, α-glucoconjugate 4′ and β-riboconjugate 5 were also synthesized to evaluate the role of the sugar moiety (glucose vs ribose) and of the stereochemistry at the anomeric position (α-glucose vs β-glucose) in the hexose series. Particular attention was paid to the Huisgen cycloaddition with the azidopropynylpyridine, which required the use of 5 mol % of [CuCl(SIMes)]-4,7-dichloro-1,10-phenanthroline³⁰ to avoid by-product formation. The same sequence was used to form the oxime moiety (overall yields over the four final steps: 57, 76, 64, and 60% for 3, 4, 4′, and 5, respectively).

Scheme 2. Triazole-Containing Aldoxime Synthetic Pathway.

The glycosylation steps were carried out in the presence of either 3-butyn-1-ol, 5-hexyn-1-ol as a glycol acceptor, or the corresponding glycosyl donor in the form of peracetylated glucopyranose or ribofuranose (Scheme 3).

Scheme 3. Triazole and Alkyne Glycoconjugate Synthetic Pathway.

In order to gain a better understanding of the glycoconjugate trafficking according to the sugar derivative and the glycosyl stereochemistry, several intermediates bearing simpler loads (alkyne or methyltriazoles) were synthesized (Scheme 3). The alkyne chain glycoconjugates 6, 7, 12, and 13 were synthesized as a single diastereoisomer in two steps from the corresponding peracetylated sugar in 57, 70, 36, and 6% yields, respectively, by a glycosylation reaction followed by the Zemplén procedure to study the influence of possible inhibiting interaction. The triazole glycoconjugates, 8, 9, 10, and 14 were synthesized, as pure diastereoisomers, in three steps and in 12, 1, 22, and 8% overall yields, respectively, using Huisgen cycloaddition with 1.1 equivalent of trimethylsilylmethyl azide followed by TBAF treatment and Zemplén transesterification procedure to uncover the N-methyltriazole glycoconjugates as free hydroxyl groups.

Compound 11, bearing the triazole alkyl chain on position 6 of the sugar, was also synthesized to investigate the impact of the position of the triazole-comprising chain on the BBB permeability (Scheme 4). Compound 11 was obtained in 5 steps and with 24% overall yield from glucose.

Scheme 4. Position 6-Substituted Alkyl-Triazole Glycoconjugate.

The chemical structures of the molecule tested in this study are summarized in Figure 1.

Figure 1.

Sugar oximes and synthetic product synthesized and tested.

The stability of bi-functional oxime 2 and tri-functional oxime 3 in PBS buffer was studied by LC–MS (Supporting Information S45). The analysis of compound 2 shows a drop from 98.7 to 97.8% purity after 4 h of incubation to reach 91.4% purity after 52 h. For compound 3, a decrease from 99.2 to 96.7% after 4 h and to 90.9% after 52 h of incubation has been noticed. In vitro experiments were carried out for a maximum of 1 h, and therefore, 2 and 3 are considered to be stable during the experiment.

Molecular Docking

To determine if the candidate oximes 1, 2, 3, and 4 can bind in a productive way to the OPNA phosphylated hAChE close to the catalytic OP-serine adduct, we performed flexible molecular docking on hAChE phosphylated by either VX or GB (sarin) as previously described.³¹ The side chain of aromatic residues (Trp286, Tyr72, Tyr341, Tyr124, Tyr337) of the active site of hAChE as well as the alkoxy chain of the nerve agent were allowed to span freely from their native position. The binding energy determined by the scoring function of Autodock Vina and the distances between the phosphorus atom of nerve agents and the oxime oxygen atom are reported in Table 1.

Table 1. Binding Affinity and Distances Between the Phosphorus Atom of VX and GB-Inhibited hAChE and the Oxime Oxygen Atom of the Best Molecular Docking Pose of Selected Oximes. Binding Affinity in kcal/mol and P–O Distances in Angstrom.

	VX		GB
oxime	ΔG (kcal/mol)	distance P–O (Å)	ΔG (kcal/mol)	distance P–O (Å)
1	–9	10.7	–8.2	9.2
2	–8.6	4.2	–7.7	6.3
3	–10.1	10.2	–8.7	9.6
4	–9.7	10.8	–8.1	11

Results for bifunctional sugar-oximes 1 and 2 showed a uniform low binding affinity for all OPNAs tested, indicating a poor positioning of the oximes inside the gorge of the hAChE. The sugar moiety of compound 1 does not interact with the aromatic residues of the gorge to give a possible stabilizing interaction, as shown in Figure 2. Interestingly, the sugar moiety of compound 2, which has a 3 carbon atom linker, is able to bind between Tyr341 and Tyr337 of the VX-inhibited hAChE. The resulting 4.2 Å distance observed between the oxime oxygen atom and the phosphorus atom of the phosphylated hAChE is sufficiently short to allow a possible reactivation of the enzyme by compound 2. As expected, the binding affinity observed for tri-functional sugar oximes 3 and 4 appeared to be higher than those observed for 1 and 2 (Table 1). The triazole moiety of compounds 3 and 4 acts as a ligand of the peripheral site by forming a π stacking interaction with Tyr341, stabilizing the molecule in a productive way. Overall, the sugar moiety of these new oximes appears to be localized most of the time at the entrance of the gorge without strong interaction with the peripheral site. In this position, the sugar moiety cannot interfere with the binding of the oxime moiety down into the gorge, allowing the oxime oxygen atom to reach a short and productive distance to the phosphorus atom.

Figure 2.

Molecular docking of selected oximes in the active site of VX and GB-inhibited human acetylcholinesterase (respectively top and bottom panel). The binding energy determined by the scoring function of Autodock Vina and the distance between the phosphorus atom of VX or GB and the oxime oxygen atom are indicated in the top left corner of each docking pose.

In Vitro Reactivation of OPNA Phosphylated hAChE

We first determined the half-maximal inhibitory concentration (IC₅₀) for compounds 1, 2, 3, and 4 (Table 2). Results showed a generally low affinity of the new sugar-oximes for recombinant uninhibited hAChE. Noteworthy, oximes 1 and 2, which differ only by one carbon atom in the alkyl chain between the 3-hydroxypyridinaldoxime and the glucose, displayed a high binding affinity difference, implying that the length of the linker (i.e., four carbons for oxime 1vs 3 carbons for oxime 2) is a key parameter for the binding of the reactivators to their target. The higher flexibility allowed by the four carbon alkyl chain of oxime 1 may prevent the molecule to bind tightly to the enzyme. Oximes 3 and 4 also differ by one carbon atom in the length of the alkyl chain between the oxime moiety and the triazole (3 carbons for oxime 3 and 4 for oxime 4). The binding affinity of oxime 3 is approximately three times lower than the affinity of oxime 4, which may indicate that the four carbon linker of compound 4 allows a more effective binding of the molecule inside the active site of hAChE. Altogether, these results displaying a low affinity for native hAChE clearly suggest that oximes 1–4 can be used at a high concentration without inhibiting hAChE.

Table 2. Half Maximal Inhibitory Concentration (IC₅₀) for hAChE.

oxime	IC50 μM
1	97% at 5 mM
2	140 ± 20
3	1600 ± 200
4	580 ± 10
HI-6	55 ± 5
2-PAM	560 ± 30
obidoxime	640 ± 70

We then determined the reactivation kinetics constants for the selected oximes, as shown in Table 3. The α anomer of compound 4 (compound 4′) and the riboconjugate of compound 4 (compound 5) were also tested for their reactivation efficacy. Surrogates of VX (NEMP), sarin (NIMP), and tabun (NEDPA) as well as an organophosphorus pesticide (paraoxon) were used for hAChE inhibition, as they give the same phosphyl moiety on the serine residue as their corresponding OPNA.

Table 3. Reactivation Kinetics of hAChE Inhibited by Nerve Agent Surrogated and Paraoxon by Selected Oximes^a.

OP	oxime	kr min–1	KD mM	kr2 mM–1·min–1
NEMP	1	0.4 ± 0.1	4 ± 1	0.1
	2	0.3 ± 0.01	1 ± 0.09	0.2
	3	0.5 ± 0.07	1.5 ± 0.3	0.3
	4	1 ± 0.1	1 ± 0.3	1
	4’	1 ± 0.02	1 ± 0.04	1
	5	2.5 ± 0.2	2.4 ± 0.4	1.0
	HI-6	0.65 ± 0.03	0.07 ± 0.008	9.3
	2-PAM	0.1 ± 0.01	0.2 ± 0.06	0.5
	obidoxime	0.2 ± 0.007	0.6 ± 0.04	0.3
NIMP	1	0.2 ± 0.006	1 ± 0.08	0.2
	2	0.2 ± 0.02	2 ± 0.4	0.1
	3	0.2 ± 0.01	1.3 ± 0.2	0.15
	4	0.7 ± 0.1	2 ± 0.5	0.4
	4’	0.4 ± 0.09	2 ± 0.7	0.2
	5	nd	nd	0.2
	HI-6	1 ± 0.2	0.09 ± 0.03	11
	2-PAM	0.24 ± 0.02	0.2 ± 0.03	1.2
	obidoxime	0.2 ± 0.02	0.5 ± 0.09	0.4
NEDPA	1	0.1 ± 0.01	1.3 ± 0.2	0.1
	2	0.2 ± 0.02	1.2 ± 0.3	0.2
	3	0.2 ± 0.08	3 ± 1.4	0.07
	4	0.6 ± 0.02	0.9 ± 0.1	0.6
	4’	0.5 ± 0.04	0.6 ± 0.1	0.8
	5	0.8 ± 0.2	3 ± 1	0.3
	HI-6	0.05 ± 0.005	0.2 ± 0.04	0.2
	2-PAM	0.15 ± 0.025	0.8 ± 0.2	0.2
	obidoxime	0.6 ± 0.04	0.3 ± 0.07	2
PARAOXON	1	0.03 ± 0.005	0.5 ± 0.2	0.05
	2	nd	nd	0.2
	3	0.2 ± 0.08	2 ± 0.4	0.2
	4	0.6 ± 0.1	2 ± 0.7	0.3
	4’	nd	nd	0.8
	5	0.6 ± 0.1	2.7 ± 0.7	0.2
	HI-6	0.09 ± 0.009	0.8 ± 0.14	0.1
	2-PAM	0.05 ± 0.006	0.4 ± 0.08	0.1
	obidoxime	0.3 ± 0.02	0.5 ± 0.09	0.6

^and: If [reactivator] ≪ K_D, then there is a linear dependence between k_obs and [reactivator]: k_obs = (k_r/K_D)[reactivator]. In this case, k_r and K_D cannot be determined, but k_r2 = k_r/K_D is the slope of the line and can be directly obtained by fitting.

The results, displayed in Table 3, showed a generally low efficacy (k_r2) of the sugar oximes in the same range as 2-PAM in all cases, and in the same range of HI-6 (but lower than obidoxime) for NEDPA and paraoxon phosphylated hAChE. The low affinity (K_D), in the millimolar range, of these compounds for the phosphylated hAChE is clearly responsible for the poor reactivation efficiency. However, compounds 4, 4′, and 5 performed better than 2-PAM and obidoxime for the NEMP-inhibited hAChE, mainly due to the high reactivation rates observed. Regarding the NIMP-inhibited hAChE, the sugar oximes are outperformed by HI-6, 2-PAM and obidoxime, and they are slightly better than HI-6 and 2-PAM for NEDPA-inhibited enzyme. Compound 4′ appeared to reactivate the paraoxon-inhibited hAChE better than HI-6, 2-PAM, and obidoxime.

X-ray Structures of Compounds 3 and 4 in Native hAChE

In order to gain insight into the binding position of the sugar oximes in the gorge of hAChE, we solved the structure of the complexes formed by native hAChE and compounds 3 (PDB 7P1P) and 4 (PDB 7P1N). Data collection and refinement statistics are summarized in Table S2. hAChE crystallizes in spacegroup P6₁ in our condition, resulting in crystals of commonly poor quality, diffracting to 3 Å resolution, with a very high average B-factor for the protein, near 90 ų, and noisy Fo–Fc electron density maps.³² However, careful examination of the initial Fo–Fc electron density maps allows to unambiguously identify the presence of the sugar oximes in the active site gorge of monomer A of the dimeric structure (Figure 3). Compound 3 has an RSZD value of 1.6 and an RSZO value of 0.6 as computed with EDSTATS, and compound 4 has an RSZD value of 1.9 and an RSZO of 0.6, reflecting some missing electron density.

Figure 3.

Views of oximes 3 (PDB 7P1P) and 4 (PDB 7P1N) in complex with hAChE. Top: overall view of oximes 3 and 4 location inside the gorge of hAChE defined by the solvent accessible surface showing the sugar moiety outside the gorge. Bottom: closer view of oximes 3 and 4 inside the gorge of hAChE. A 1.0-σ feature-enhanced map is represented as a blue mesh. Key peripheral (Trp286) and active site (Trp86) tryptophan residues are represented in sticks with carbons in cyan. H-bonds are represented in black dashes and water molecules as red spheres.

No ligand is observed in the active site gorge of monomer B due to the occupation of the peripheral site by Arg493 of monomer A, as already observed for structures with ligands binding to the peripheral site.³² Unexpectedly, a second molecule of compound 4 was identified bound at the surface of monomer A, the sugar dislodging His123 from a narrow groove, the triazole in close contact with Pro194, and the 3-hydroxypyridine aldoxime moiety stacked in between Pro111 and the main chain of Arg90 and Glu91 of monomer B of a symmetry mate (not shown). This binding mode at the interface of two symmetry partners is most likely artifactual and is not unexpected with the 2 mM concentration of compounds used for soaking the crystals. An overall view of the complexes shows that 3 and 4 bind into hAChE with the 3-hydroxypyridine aldoxime moiety at the bottom of the gorge, close to the catalytic serine, and thus in a favorable orientation considering reactivation. The hydroxyl of compounds 3 and 4 is nested in the oxyanion hole, strongly interacting with the hydroxyl of catalytic Ser203 (at 2.3 Å distance) and within hydrogen bond distance from the main chain nitrogen of Gly121 and Gly122 (2.8 to 3.0 Å). The oximate of compounds 3 and 4 points toward Trp86 at 3.9 and 3.4 Å, respectively, from the indole ring. The triazole ring of both compounds can be satisfactorily modeled. It is stabilized by π-stacking interactions in the middle of the gorge in between the aromatic rings of Tyr341 and Trp286. Moreover, a water molecule bridges a nitrogen of the triazole to the mainchain nitrogen of Phe295 with H-bonds (2.6 Å interatomic distances). Almost no electron density is visible for the sugar part of 3, which was refined with partial occupancy. The average B-factor is 111 ų for the sugar moiety compared to 93 ų for the rest of the ligand. These high B-factor values are in accordance with the Z scores for the whole ligand reflecting missing electron density. This indicates that the sugar moiety is spanning freely in the solvent in the absence of possible stabilizing interactions with the amino acid at the gorge entrance. By contrast to 3, electron density is present corresponding to the sugar moiety of 4, so that it could be modeled with full occupancy, likely because the molecule appears to enter slightly deeper into the gorge, so that the sugar moiety is closer to Trp286, allowing stabilizing van der Waals interactions (3.7 Å in between the closest atoms). Yet, the average B-factor for the sugar moiety remains higher than the rest of the ligand (143 ų compared to 122 ų), suggesting that the interaction remains weak. These X-ray structures show that compounds 3 and 4 can obviously bind deeper into the gorge when the catalytic serine is free, differing from molecular docking simulations in VX- and GB-hAChE. However, these structures confirm some general predictions from the docking: the 3-hydroxypyridine aldoxime bind deep into the gorge and thus in a favorable location for reactivation; the triazole ring contributes to the binding by interacting mainly with aromatic residues in the middle of the gorge; the sugar moiety was predicted to weakly interact with peripheral site residues and thus be localized at the rim of gorge. The 3-carbon methylene chain between the 3-hydroxypyridine aldoxime and the triazole of 3 appears optimal for binding to the native enzyme but notably too long for the phosphylated enzyme as it appears folded in the molecular docking prediction.

In Vivo Protective Index Assessment

In order to determine the in vivo efficacy of the new oximes, protective indexes (PI) were determined for compounds 4 and 4′ at 100 μmol/kg and compared to 2-PAM, HI-6, and obidoxime. Experiments of the up-and-down procedure performed on paraoxon, NIMP, and NEMP exposure are presented in Tables S3–S5, respectively, and summed up in Figure 4. LD₅₀ values of paraoxon, NIMP, and NEMP were established to 818, 605 and 350 μg/kg, respectively. The PI of a 100 μmol/kg i.p. 2-PAM treatment 1 min after paraoxon exposure was previously assessed to be 2.58.³³ Alone, 100 μmol/kg of obidoxime gave the most interesting protective spectrum against the three studied OPs even if HI-6 showed a higher PI than obidoxime in the case of treatment of NEMP exposure (3.00 vs 2.06); the lowest protective efficacy of HI-6 against paraoxon exposure (PI = 1.6) was crippling. It should be noted that oximes 4 and 4′ presented an almost identical and moderate in vivo protection against the three OP exposures.

Figure 4.

Radar representation of protective index values of 100 μmol/kg intraperitoneal treatment of 2-PAM, HI-6, obidoxime, oximes 4 and 4′ 1 min after paraoxon, and NIMP or NEMP subcutaneous exposure. Protective indexes were determined by the up-and-down method.

Pharmacokinetic Study

Both oximes (4 and 4′) exhibited a similar enzymatic reactivation profile with a low reactivation percentage at the peak (react max = 2.5 ± 0.8% and 3.3 ± 0.7%, respectively). The peak is reached more rapidly for oxime 4 than for oxime 4′ (T_max = 7.0 ± 0.0 vs 10.0 ± 0.0 min after injection) as shown in Figure 5. Oxime 4 also persisted for a shorter time than oxime 4′ in mice plasma as demonstrated by the MRT value (15.0 ± 0.0 vs 20.0 ± 0.0 min) (Table 4).

Figure 5.

Reactivation of VX-inhibited hAChE by oximes 4 and 4′ in mice plasma. The same dose of 100 μmol/kg of oxime 4 and 4′ was administered intraperitoneally to mice (n = 7). Blood samples were drawn at various time points (0, 2, 5, 10, 15, 30, 60, and 180 min) after treatment, and the levels of reactivation of VX-phosphylated hAChE were determined. Values are presented as percentages of maximum reactivation and points are means ± SEM. Fitting was performed on GraphPad Prism software.

Table 4. Pharmacokinetic Data of Oximes 4 and 4′^a.

oxime	MRT (min)	Tmax (min)	react max (%)	Cmax (μM)
oxime 4	15.0 ± 0.0	7.0 ± 0.0	2.5 ± 0.8	96.7
oxime 4′	20.0 ± 0.0	10.0 ± 0.0	3.3 ± 0.7	145.7

^aMRT: mean residence time, T_max and React max respectively x and y coordinates of the peak of reactivation of VX-phosphylated hAChE by oximes 4 and 4’ in mice plasma presented in Figure 5, and C_max the peak concentration of oximes 4 and 4’ calculated from standard reactivation curves (Figure S1).

Blood–Brain Barrier Permeability Tests

One aim was to improve the transport of the oximes across the BBB endothelial cells to reactivate the central AChE. In order to determine if our sugar oximes are able to reach the CNS, we evaluated the endothelial permeability coefficients (Pe) of some gluco- and riboconjugate oximes (compounds 1, 2, 3, 4, 4′, 5, and 15) as well as representative simpler glycoconjugate molecules (compounds 6, 7, 8, 9, 10, 11, 12, 13, and 14) using the human BBB in vitro model. No toxicity was observed in our culture conditions (data not shown).

Results showed that all the compounds, the conjugate oximes or the simpler glycoconjugate molecules, had a lower rate of transport through BBB endothelial cells than 2-PAM (Pe_2-PAM = 2.89 ± 1.81 × 10^–3 cm·min^–1) (Figure 6). Only one glycoconjugate without an oxime moiety (compound 13) exhibited a permeability coefficient in the same range (Pe₁₃ = 2.83 ± 0.52 × 10^–3 cm·min^–1). Most of the molecules showed Pe values between those of HI-6 (Pe = 0.23 ± 0.01 × 10^–3 cm·min^–1) and obidoxime (Pe = 0.72 ± 0.08 × 10^–3 cm·min^–1) that are described to slowly cross the BBB, which is confirmed in our in vitro model study (Figure 6).

Figure 6.

Endothelial permeability coefficients (Pe) of control oximes (2-PAM, HI-6, obidoxime), glycoconjugate oximes (compounds 1, 2, 3, 4, 4′, 5, and 15), glycoconjugate intermediates (compounds 6, 7, 8, 9, 10, 11, 12, 13, and 14), and ¹⁴C d-glucose. Values are means ± SD, n = 3–9.

The rate of transport of glycoconjugate oximes (compounds 1, 2, 3, 4, 4′, 5, and 15) has also been compared to that of ¹⁴C d-glucose, which is taken up by a transporter (GLUT-1) so as to facilitate its crossing through membranes of the endothelial cells. The Pe values of glycoconjugate oximes (Pe from 0.34 to 0.58 × 10^–3 cm·min^–1, Table S6) were two- to three-folds lower than that of glucose (Pe _glucose = 1.43 ± 0.23 × 10^–3 cm·min^–1). At this stage, to understand at which level of the oxime glycoconjugates were no longer supported by the facilitated diffusion, the Pe values of the simpler glycoconjugate molecules were also evaluated and compared to that of glucose. None or slight differences were observed between the endothelial permeability coefficients of the ¹⁴C d-glucose and those of the ribose (compounds 12 and 13) or glucose (compounds 6 and 7) connected with an additional alkyne moiety bearing carbon chain (3 or 5 additional carbons branched at the anomeric oxygen atom) showing that these molecules crossed the BBB endothelium in the same way as glucose. The strategy used here is the first step to improve the transport of molecules through the BBB. However, when a triazole moiety has been connected to the additional carbon chain of ribose (compound 14) or of glucose (compounds 8, 9, 10, 11, and 12), Pe values showed a three- and two-fold decrease, respectively, demonstrating that the rate of transport through the endothelial cells was slowed down in the same range as for the gluco- and riboconjugate oximes (Figure 6, Table S6). These results showed that in the process of the glycoconjugate synthesis, the triazole moiety, required for the binding of the oximes to hAChE, has a detrimental effect on BBB crossing ability, as this moiety has been shown to restrain the facilitated diffusion through glucose transporters.

Discussion and Conclusions

Reactivation of the central OP-phosphylated AChE is one of the aims of the newly synthesized oximes, which have been described lately in the literature. An efficient crossing of the BBB appears to be one major limitation for these numerous newly designed oxime reactivators. If efficient reactivation of peripheral phosphylated hAChE is the key for survival after OPNA exposure, inhibition of central AChE may induce long-term side effects and detrimental neurological disorders. There are many design strategies, among which the use of uncharged or glucoconjugated reactivators has been evaluated to overcome this limitation. In this study, we intended to evaluate the combination of these two strategies and evaluated the ability of a new family of uncharged glucoconjugated 3-hydroxypyridinaldoxime to reactivate inhibited hAChE with the goal in mind to take advantage of the glucose transporter system located at the BBB in order to reach the centrally OP-phosphylated AChE. We thus synthesized simple bi-functional glycoconjugated uncharged reactivators 1 and 2, which displayed, as expected, low affinity for the phosphylated hAChE and a poor reactivation profile. We also synthesized tri-functional glycoconjugated uncharged oximes 3, 4, 4′, and 5 bearing an additional triazole moiety in order to increase the affinity of these oximes for the phosphylated AChE through binding of the triazole moiety to the peripheral site of the enzyme.

An in silico study by molecular docking of bi-functional oximes 1 and 2 and tri-functional oximes 3 and 4 showed a general low binding affinity between −7.7 and −10.1 kcal/mol for VX or GB inhibited hAChE. Except for compound 2, the results showed P–O distances between 9.2 and 11 Å. All together, these in silico results suggested that oximes 1, 2, 3, and 4 may reactivate yet with moderate efficiency the VX or GB-phosphylated hAChE. We then evaluated the in vitro reactivation efficiency of oximes 1, 2, 3, 4, and 4′ (the α anomer of compound 4) and oxime 5 (the riboconjugated analogue of oxime 4). The results show a general low binding affinity (K_D) in the millimolar range of the sugar oximes for the phosphylated hAChE. This low binding affinity negatively affected the overall reactivation efficacy (k_r2). It may be due to the presence of the sugar moiety on the oximes, which does not allow the compound to penetrate deep enough inside the gorge of the phosphylated enzyme for an efficient reactivation.

Crystal structures of compounds 3 and 4 have been solved in complex with hAChE (PDB 7P1P and 7P1N). Data show that both oximes were able to enter the gorge of the hAChE with the 3-hydroxypyridinaldoxime moiety located near the catalytic site, the triazole moiety interacting by π-stacking interactions with Tyr341 and Trp286, and the glucose moiety at the rim of the gorge without favorable interactions with amino acids of the peripheral site. These results showed the importance of the triazole ring to orient compounds 3 and 4 inside the gorge in a productive way for the reactivation of phosphylated hAChE. The importance of the triazole-containing linker was previously showed for AChE inhibitors TZ2PA5 and TZ2PA6, based on a phenylphenanthridinium peripheral site ligand and a tacrine active site ligand synthesized by click-chemistry inside the active gorge of mouse acetylcholinesterase (mAChE). Crystallographic analysis of TZ2PA5- TZ2PA6-mAChE complexes revealed van der Waals interactions of the side chains of Phe297 and Tyr341 with the triazole moiety. Upon binding, conformational changes of Tyr337, Tyr341, Trp286, and His447 were observed. Moreover, a Tyr337Ala mutant of mAChE showed the key role of the triazole in the proper orientation of these mAChE inhibitors.³⁴⁻³⁶ Such large conformational changes upon binding of oxime 3 and 4 inside the gorge of hAChE are not observed in the crystallographic structure solved in our study, most likely because the affinity of the oximes is orders of magnitude weaker compared to that of the inhibitors cited above (mM vs fM) and thus does not provide a sufficient energy gain to trigger conformational adaptation of the gorge residues. The presence of the triazole moiety on oximes 3, 4, 4′, and 5 was predicted to improve the affinity of the sugar oximes for the inhibited enzymes, but results showed a similar low binding affinity in the same range as compounds 1 and 2. However, a high enough reactivation kinetics k_r allows oximes 4, 4′, and 5 to be more efficient than 2-PAM and obidoxime to reactivate the NEMP-phosphylated hAChE as well as oximes 4 and 4′ to reactivate NEDPA-phosphylated hAChE. The α anomer of the oxime 4 (named 4′) is the only compound to perform better than 2-PAM, HI-6, and obidoxime for paraoxon phosphylated hAChE.

Our goal was to design new uncharged oximes that could reach the CNS more efficiently through the glucose transporter located at the BBB. In order to evaluate our hypothesis, we determined protective indexes for compounds 4 and 4′ on mice exposed to NIMP, NEMP, and paraoxon. Our results showed that oximes 4 and 4′ display a similar protection profile. The use of the α anomer of compound 4 did not improve the protective index. In vivo results showed that both oximes protect as well as HI-6 in mice exposed to paraoxon and appeared to be more effective than 2-PAM against NIMP and NEMP. The pharmacokinetic study demonstrated, after an intraperitoneal injection, the presence of compounds 4 and 4′ in mice blood during the experiments but showed that only a low percentage of VX-phosphylated enzymes could be reactivated. In vitro reactivation studies have shown the low affinity of the sugar-oximes for the phosphylated hAChE and, therefore, their moderate reactivation capabilities. We can hypothesize that the low protective indexes recorded in this study may be improved by administering a higher dose of oxime in adequacy with the compound properties; interestingly, contrary to HI-6, these oximes poorly inhibit native hAChE as demonstrated by IC₅₀ values and can potentially be used at a higher dose if no other type of toxicity is observed in vivo.

To evaluate the ability of our sugar oximes conjugates to cross the BBB, we determined the endothelial permeability coefficients of gluco- and riboconjugate oximes (1, 2, 3, 4, 4′, 5, 15) as well as simpler glycoconjugated molecules (compounds 6, 7, 8, 9, 10, 11, 12, 13, 14) using the human BBB in vitro model (Figure 10). Our results showed that the designed sugar oximes do not cross the BBB at the same transport rate as glucose, implying a structural restrain from our compounds. 2-PAM has the highest transport rate than any other evaluated oximes, but the Pe for oximes 1, 2, 3, 4, 4′, and 5 is higher than the Pe for HI-6 and slightly lower than the Pe for obidoxime. However, we were disappointed to observe that the presence of the triazole moiety on tri-functional oximes 3, 4, 4′, and 5, incorporated to increase the affinity of these compounds for phosphylated hAChE, and its presence in simpler glycoconjugated molecules such as 8, 9, 10, 11, and 14 reduces the transport rate through the BBB, which may negatively impact the protective indexes determined in our in vivo experiments.

In summary, we designed a new family of uncharged oxime glycoconjugates to combine two strategies to reach the CNS more efficiently through the use of non-permanently charged oximes and take advantage of a glucose transport system to improve the protection of oxime-based reactivators against neurotoxic organophosphorus poisoning. Our study showed that our compounds perform roughly as the oximes currently used across the world but highlights the structural restraints, which apply to oximes targeting the glucose transport system: addition of an additional aromatic moiety to improve binding of the oximes to phosphylated hAChE negatively impacts their BBB crossing ability. Based on this work, new sugar-oximes could be designed using other moieties than triazoles to bind efficiently phosphylated hAChE and probably longer side chains between the first two components of these tri-functional reactivators (sugar/AChE binding moiety/oxime), synthesized in order to use efficiently this glucose transport system of the BBB to reach the central OP-phosphylated AChE.

Experimental Section

Chemicals

2-PAM and HI-6 were obtained from Pharmacie Centrale des Armées (Orléans, France). NIMP (4-nitrophenyl isopropyl methylphosphonate), NEMP (4-nitrophenyl ethyl methylphosphonate), and NEDPA (4-nitrophenyl ethyl dimethylphosphoramidate) were obtained from UMR CNRS 7515 ICPEES (Strasbourg, France). Obidoxime, heparin, DTNB, acetylthiocholine (ATC), and paraoxon were purchased from Sigma-Aldrich (Saint-Quentin-Fallavier, France). HI-6, 2-PAM, obidoxime (chlorhydrate salts), and oximes 4 and 4′ were dissolved into physiological serum (0.9% NaCl) to obtain the concentration of 10 mM for in vivo experiments.

Chemistry

General Procedures

Solvents were purified by a dry solvent station MB-SPS-800 (MBraun, Garching, Germany) immediately prior to use. Triethylamine was distilled from KOH. All reagents were obtained from commercial suppliers (Sigma-Aldrich, St Quentin Fallavier, France; Acros Illkirch, France; TCI Europe, Paris, France) unless otherwise stated. The melting points were recorded on a Stuart SMP30 apparatus (Stuart, Staffordshire, UK). Column chromatography purifications were performed with prepacked SI-HP (30 μm) or SI-HC (15 μm) columns from Interchim (Montluçon, France). Preparative normal phase chromatography was carried out on an automated flash purification apparatus, either Biotage Isolera One (Biotage, Uppsala, Sweden) or Interchim 420 PuriFlash. Thin-layer chromatography (TLC) was carried out on Merck DC Kieselgel 60F-254 aluminum sheets (Merck, Darmstadt, Germany). Compounds were visualized by UV irradiation and/or spraying with a solution of vanillin followed by smooth heating. ¹H and ¹³C NMR spectra were recorded with a Bruker DPX 300 spectrometer (Bruker, Wissembourg, France) and are presented in the Supporting Information (S9, NMR spectra). Chemical shifts are expressed in parts per million (ppm) from CDCl₃ (δH = 7.26 ppm, δC = 77.16 ppm). J values are expressed in hertz. Mass spectra were obtained with a Finnigan LCQAdvantage MAX (ion trap) apparatus equipped with an electrospray source (Thermo Electron Corporation, Waltham, MA). High-resolution mass spectra were obtained with a Varian MAT 311 spectrometer (Varian MAT, Bremen, Germany) using electrospray analysis. HPLC quality grade acetonitrile and Milli-Q purified water were used for analytical and preparative HPLC. Preparative HPLC run was carried out with an Interchim 4250 apparatus (Interchim, Montluçon, France) with an Interchim puriflash C18 column (Interchim, Montluçon, France), 30 × 250 mm, C18AQ-5 μm. Analytical HPLC was performed on a ThermoFisher UHPLC Ultimate 3000 instrument (Thermofisher, Waltham, MA) equipped with a PDA detector under the following conditions: Syncronis C18 column (3 μm, 3 × 100 mm) with MeCN and 20 mM NH₄OAc as eluents [using a gradient from 100% 20 mM NH₄OAc to 100% MeCN over 30 min] at a flow rate of 0.5 mL/min with UV detection at 254 nm. Method A: Syncronis C18 column (3 μm, 3 × 100 mm) with MeCN and 20 mM NH₄AcO as eluents [100% 20 mM NH₄OAc to 100% MeCN over 30 min] at a flow rate of 0.5 mL/min and UV detection at 254 nm; method B: Syncronis C18 column (3 μm, 2.1 × 100 mm) with MeCN and analytical H₂O with 0.1% formic acid as eluents [95% H₂O with 0.1% formic acid to 100% MeCN over 15 min] at a flow rate of 0.25 mL/min and UV detection at 254 nm; or method C: Waters Acquity BEH C18 column (1.7 μm, 2.1 × 100 mm) with MeCN and analytical H₂O with 0.1% formic acid as eluents [95% H₂O with 0.1% formic acid to 100% MeCN over 10 min] at a flow rate of 0.4 mL/min and UV detection at 254 nm. LC–MS was performed on a ThermoFisher ISQ EC mass spectrometer with a heated electrospray ionization.

Sonogashira Cross-Coupling

Bromopyridine (1 equiv) and alkyne (1 equiv) was dissolved in TEA (0.2 M) and DCM (0.1 M). The reaction mixture was degassed for 5 min before Pd(PPh₃)₄ (5 mol %) and CuI (10 mol %) were added. The reaction mixture was stirred overnight at room temperature under argon. Solvents were removed, and the residue was purified by flash chromatography on silica gel.

Zemplén Deacetylation

To a solution of acetylated pyranose or furanose (1 equiv) in anhydrous MeOH (0.1 M) was added 0.5 M NaOMe solution in MeOH (0.1 equiv). The reaction mixture was stirred under an inert atmosphere for 1 h and quenched with Amberlite IR-120 resin filtered through celite. The filtrate was concentrated in vacuo to afford the deprotected pyranose or furanose.

Huisgen Cycloaddition

Azido compound (1.0 equiv) and alkyne (1.1 equiv) [in the case of coupling with (azidomethyl)trimethylsilane, 1.1 equiv of azido compound and 1.0 equiv alkyne were used] were dissolved in TEA (25 equiv), and DMF (0.1 M). CuI (1.0 equiv) was added in the reaction mixture before being stirred at room temperature for 18 h. Volatiles were evaporated, and AcOEt was added before the solution was washed with brine (×2). The organic layer was dried (MgSO₄) and evaporated in vacuo. To a solution of crude triazole in THF (0.1 M) was added 1 M in THF of TBAF (2.0 equiv) at 0 °C. The solution was stirred at this temperature for 1 h and then at room temperature for 2 h. A saturated aqueous solution of NaHCO₃ was added, followed by extraction with AcOEt (×3). The organic phase was washed with brine, dried over MgSO₄, filtered, and concentrated in vacuo. The crude material was purified by flash chromatography on silica gel.

Hydrogenation

Benzylated pyridine (1.0 equiv) was dissolved in a mixture of MeOH (0.02 M) and AcOEt (0.04 M). The reaction mixture was evacuated and back-filled with nitrogen three times before palladium hydroxide on charcoal 20% (50% by weight) was added in the reaction mixture and then evacuated and back-filled with hydrogen three times and stirred between 30 min and 2 h (as monitored by TLC analysis) under a H₂ atmosphere. The reaction mixture was evacuated and back-filled with nitrogen before being filtered on Celite and concentrated in vacuo. The crude material was purified by flash chromatography on silica gel.

Acetal Deprotection

A mixture of the acetal (1 equiv), LiCl (6 equiv) and H₂O/DMSO (0.05 M, 1:1) was stirred for 3 h at 95 °C. The reaction mixture was allowed to cool to room temperature, diluted with AcOEt (50 mL), and rinsed with H₂O (2 × 50 mL), dried (Na₂SO₄). The solvent was removed, and the residue obtained was purified by flash chromatography or reversed phase HPLC or flash chromatography.

Oxime Formation/Deacetylation

A total of 0.5 M solution of NaOMe in MeOH (3.1 equiv) and NH₂OH.HCl (2 equiv) were added to a solution of glucoside (1.0 equiv) in MeOH (0.1 M) at rt, then stirred at rt for 2 h. The solvent was removed, and the residue obtained was purified by flash chromatography or reversed phase HPLC or flash chromatography.

The synthesis of the 15 compounds tested is described below. The general chemistry, experimental information, and syntheses of all other compounds are supplied in the Supporting Information. The purity of all final compounds as determined by HPLC analysis is ≥95%. The canonical SMILES of all compounds tested are provided in Table S1.

(E)-3-Hydroxy-6-(4-(((2R,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)butyl)picolinaldehyde Oxime (1)

General procedure of oxime formation/deacetylation applied on aldehyde 47 (496 mg, 0.94 mmol). The residue was purified by preparative HPLC (0% for 5 min then 0% to 30% MeCN in H₂O over 25 min then 30 to 100% over 5 min, 40 mL/min, Interchim puriflash prep C18AQ, 30 × 250 mm, 5 micro, PF5C18AQ-250/300) to afford the title compound as a solid (137 mg, 39%). mp = 47–53 °C. [α]_D²⁰ = −22.8 (c 0,50 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.31 (s, 1H), 7.29 (d, J = 8.5 Hz, 1H), 7.19 (d, J = 8.5 Hz, 1H), 4.27 (d, J = 7.8 Hz, 1H), 3.96 (dt, J = 9.6, 6.4 Hz, 1H), 3.88 (dd, J = 11.9, 2.0 Hz, 1H), 3.72–3.64 (m, 1H), 3.59 (dt, J = 9.6, 6.3 Hz, 1H), 3.41–3.25 (m, 3H), 3.19 (dd, J = 8.9, 7.7 Hz, 1H), 2.81–2.72 (m, 2H), 1.87–1.73 (m, 2H), 1.73–1.61 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 153.3, 152.4, 151.4, 134.8, 124.7, 124.0, 103.0, 76.7, 76.5, 73.7, 70.3, 69.0, 61.4, 36.1, 28.8, 26.3. HRMS (ESI⁺): m/z calculated for [C₁₆H₂₅N₂O₈]⁺ 373.1611, found 373.1608.

(E)-3-Hydroxy-6-(3-(((2R,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)propyl)picolinaldehyde Oxime (2)

General procedure of oxime formation/deacetylation applied on aldehyde 46 (336 mg, 0.98 mmol). The residue was purified by preparative HPLC (0% for 5 min then 0 to 40% MeCN in H₂O over 25 min then 30 to 100% over 5 min, 40 mL/min, Interchim puriflash prep C18AQ, 30 × 250 mm, 5 micro, PF5C18AQ-250/300) to afford the title compound as an oil (143 mg, 41%). [α]_D²⁰ = −18.0 (c 0,50 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.32 (s, 1H), 7.29 (d, J = 8.5 Hz, 1H), 7.22 (d, J = 8.5 Hz, 1H), 4.27 (d, J = 7.7 Hz, 1H), 3.94 (dt, J = 9.6, 6.1 Hz, 1H), 3.91–3.84 (m, 1H), 3.69 (dd, J = 11.9, 5.1 Hz, 1H), 3.58 (dt, J = 9.8, 6.4 Hz, 1H), 3.43–3.25 (m, 4H), 3.22 (dd, J = 8.9, 7.7 Hz, 1H), 2.85 (dd, J = 8.6, 6.7 Hz, 2H), 2.07–1.91 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 152.9, 152.4, 151.5, 134.9, 124.7, 124.2, 103.1, 76.7, 76.5, 73.8, 70.3, 68.4, 61.4, 32.8, 29.9. HRMS (ESI⁺): m/z calculated for [C₁₅H₂₃N₂O₈]⁺ 359.1454, found 359.1458.

(E)-3-Hydroxy-6-(3-(4-(4-(((2R,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)butyl)-1H-1,2,3-triazol-1-yl)propyl)picolinaldehyde Oxime (3)

General procedure of oxime formation/deacetylation applied on aldehyde 48 (677 mg, 1.07 mmol). The residue was purified by normal phase flash chromatography (10 to 30% MeOH in DCM over 30 min, 12G SIHC) then reversed-phase flash chromatography (5 to 60% MeCN in H₂O over 30 min, 80 G, C18-15 micro, Interchim) to afford an off-white solid (384 mg, 75%). Rf = 0.12 (DCM/MeOH 90/10, v/v). [α]_D²⁰ = −15.3 (c 0,53 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.28 (s, 1H), 7.73 (s, 1H), 7.25 (d, J = 8.4 Hz, 1H), 7.12 (d, J = 8.5 Hz, 1H), 4.87 (s, 7H), 4.41 (t, J = 6.9 Hz, 2H), 4.25 (d, J = 7.8 Hz, 1H), 3.94 (dt, J = 9.6, 6.3 Hz, 1H), 3.86 (dd, J = 11.9, 1.9 Hz, 1H), 3.67 (dd, J = 11.9, 5.3 Hz, 1H), 3.58 (dt, J = 9.6, 6.2 Hz, 1H), 3.39–3.32 (m, 1H), 3.32–3.24 (m, 1H), 3.28–3.25 (m, 1H), 3.18 (dd, J = 8.9, 7.7 Hz, 1H), 2.76–2.71 (m, 2H), 2.71 (t, J = 8.0 Hz, 2H), 2.29 (p, J = 6.9 Hz, 2H), 1.84–1.69 (m, 2H), 1.69–1.56 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 152.5, 151.6, 151.4, 147.7, 135.2, 124.6, 124.0, 121.9, 103.0, 76.7, 76.5, 73.7, 70.3, 68.9, 61.4, 49.3, 33.2, 29.8, 28.7, 25.6, 24.5. HRMS (ESI⁺): m/z calculated for [C₂₁H₃₁N₅O₈Na]⁺ 504.2070, found 504.2072.

(E)-3-Hydroxy-6-(4-(4-(4-(((2R,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)butyl)-1H-1,2,3-triazol-1-yl)butyl)picolinaldehyde Oxime (4)

General procedure of oxime formation/deacetylation applied on aldehyde 49 (146 mg, 0.22 mmol). The residue was purified by normal phase flash chromatography (10 to 30% MeOH in DCM over 30 min, 12G SIHC) then reversed-phase flash chromatography (5 to 100% MeCN in H₂O over 30 min, 80 G, C18-15 micro, Interchim) to afford an oil (112 mg, 100%). Rf = 0.14 (DCM/MeOH 9/1, v/v). [α]_D²³ = −14.5 (c 0,43 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.26 (s, 1H), 7.74 (s, 1H), 7.26 (d, J = 8.4 Hz, 1H), 7.13 (d, J = 8.5 Hz, 1H), 4.39 (t, J = 6.8 Hz, 2H), 4.26 (d, J = 7.7 Hz, 1H), 3.93 (dt, J = 9.5, 6.2 Hz, 1H), 3.86 (dd, J = 12.1, 1.8 Hz, 1H), 3.67 (dd, J = 11.9, 4.8 Hz, 1H), 3.57 (dt, J = 9.6, 6.2 Hz, 1H), 3.44–3.11 (m, 3H), 3.18 (t, J = 8.3 Hz, 1H), 2.74 (t, J = 7.7 Hz, 2H), 2.71 (t, J = 7.7 Hz, 2H), 2.00–1.82 (m, 2H), 1.82–1.57 (m, 6H). ¹³C NMR (75 MHz, MeOD): δ 153.9, 153.8, 152.7, 149.1, 136.3, 126.1, 125.4, 123.2, 104.3, 78.0, 77.8, 75.1, 71.6, 70.3, 62.7, 51.0, 36.9, 30.7, 30.0, 27.8, 27.0, 25.9. HRMS (ESI⁺): m/z calculated for [C₂₂H₃₄N₅O₈]⁺ 496.2407 found 496.2413.

(E)-3-Hydroxy-6-(4-(4-(4-(((2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)butyl)-1H-1,2,3-triazol-1-yl)butyl)picolinaldehyde Oxime (4′)

General procedure of oxime formation/deacetylation applied on aldehyde 50 (369 mg, 0.57 mmol). The residue was purified by normal phase flash chromatography (10 to 30% MeOH in DCM over 30 min, 12G SIHC) then reversed-phase flash chromatography (5 to 100% MeCN in H₂O over 30 min, 80 G, C18-15 micro, Interchim) to afford an oil (225 mg, 80%). [α]_D²⁶ = +65.1 (c 0,42 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.28 (s, 1H), 7.75 (s, 1H), 7.28 (d, J = 8.5 Hz, 1H), 7.14 (d, J = 8.5 Hz, 1H), 4.78 (d, J = 3.7 Hz, 1H), 4.40 (t, J = 6.9 Hz, 2H), 3.81 (dd, J = 11.7, 2.4 Hz, 2H), 3.86–3.71 (m, 1H), 3.67 (dd, J = 11.7, 5.7 Hz, 1H), 3.66 (dd, J = 9.7, 8.8 Hz, 1H), 3.58 (ddd, J = 9.9, 5.6, 2.4 Hz, 1H), 3.48 (dt, J = 9.7, 5.9 Hz, 1H), 3.40 (dd, J = 9.7, 3.7 Hz, 1H), 3.29 (dd, J = 9.8, 8.8 Hz, 1H), 2.76 (t, J = 7.7 Hz, 2H), 2.74 (t, J = 7.7 Hz, 2H), 1.99–1.87 (m, 2H), 1.84–1.61 (m, 6H). ¹³C NMR (75 MHz, MeOD): δ 153.9, 153.8, 152.9, 149.0, 136.4, 126.0, 125.3, 123.2, 100.1, 75.2, 73.7, 73.6, 71.9, 68.6, 62.8, 51.0, 37.0, 30.7, 29.9, 27.9, 27.2, 26.0. HRMS (ESI⁺): m/z calculated for [C₂₂H₃₄N₅O₈]⁺ 496.2407, found 496.2399.

(E)-6-(4-(4-(4-(((2R,3R,4S,5R)-3,4-Dihydroxy-5-(hydroxymethyl)tetrahydrofuran-2-yl)oxy)butyl)-1H-1,2,3-triazol-1-yl)butyl)-3-hydroxypicolinaldehyde Oxime (5)

General procedure of oxime formation/deacetylation applied on aldehyde 51 (398 mg, 0.69 mmol). The residue was purified by normal phase flash chromatography (5 to 30% MeOH in DCM over 30 min, 25G SIHC) and then purified by preparative HPLC (0% for 5 min, then 0 to 40% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as an oil (233 mg, 73%). Rf = 0.13 (DCM/MeOH 9/1, v/v). [α]_D²⁴ = – 21.1 (c 0,53 MeOH). ¹H NMR (300 MHz, MeOD): δ 8.14 (s, 1H), 7.59 (s, 1H), 7.13 (d, J = 8.4 Hz, 1H), 6.99 (d, J = 8.4 Hz, 1H), 4.73 (d, J = 4.0 Hz, 1H), 4.26 (t, J = 6.9 Hz, 2H), 3.92 (dd, J = 6.9, 4.7 Hz, 1H), 3.82 (td, J = 6.7, 3.5 Hz, 1H), 3.76 (dd, J = 4.7, 1.0 Hz, 1H), 3.71–3.57 (m, 1H), 3.60 (dd, J = 11.8, 3.6 Hz, 1H), 3.42 (dd, J = 11.7, 6.5 Hz, 1H), 3.28 (dt, J = 9.5, 6.2 Hz, 1H), 2.62 (t, J = 6.5 Hz, 2H), 2.59–2.51 (m, 2H), 1.78 (dd, J = 15.0, 7.5 Hz, 2H), 1.66–1.39 (m, 6H). ¹³C NMR (75 MHz, MeOD): δ 152.5, 152.4, 151.5, 147.6, 135.0, 124.6, 123.9, 121.8, 107.4, 83.4, 74.9, 71.4, 67.0, 63.7, 49.6, 35.6, 29.3, 28.7, 26.5, 25.8, 24.6. HRMS (ESI⁺): m/z calculated for [C₂₁H₃₂N₅O₇]⁺ 466.2302, found 466.2303.

(2R,3R,4S,5S,6R)-2-(Hex-5-yn-1-yloxy)-6-(hydroxymethyl)tetrahydro-2H-pyran-3,4,5-triol (6)

General procedure of Zemplén deacetylation applied on acetylated glucoside 20 (0.3 g, 0.7 mmol) to afford the title compound as a colorless oil (183 mg, quantitative). [α]_D²³ = + 33.8 (c 0,51 MeOH). ¹H NMR (300 MHz, MeOD): δ 4.26 (d, J = 7.8 Hz, 1H), 4.00–3.91 (m, 1H), 3.91–3.85 (m, 1H), 3.72–3.64 (m, 1H), 3.58 (dt, J = 9.7, 6.3 Hz, 1H), 3.34–3.25 (m, 3H), 3.23–3.14 (m, 1H), 2.29–2.17 (m, 3H), 1.75 (ttd, J = 7.2, 5.8, 5.3, 1.3 Hz, 2H), 1.69–1.56 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 102.9, 83.5, 76.7, 76.5, 73.7, 70.3, 68.8, 68.2, 61.4, 28.4, 24.9, 17.4. HRMS (ESI^–): m/z calcd for [C₁₂H₁₉O₆]⁻ 259.1182, found 259.1179.

(2R,3R,4S,5S,6R)-2-(But-3-yn-1-yloxy)-6-(hydroxymethyl)tetrahydro-2H-pyran-3,4,5-triol (7)

General procedure of Zemplén deacetylation applied on acetylated glucoside 22 (0.3 g, 0.76 mmol) to afford the title compound as a colorless oil (146 mg, quantitative). [α]_D²³ = – 17.9 (c 0,54 MeOH). ¹H NMR (300 MHz, MeOD): δ 4.31 (d, J = 7.7 Hz, 1H), 3.97 (dt, J = 9.7, 7.3 Hz, 1H), 3.88 (dd, J = 11.9, 1.8 Hz, 1H), 3.71 (dt, J = 9.6, 7.3 Hz, 1H), 3.74–3.62 (m, 1H), 3.36 (s, 4H), 3.31–3.26 (m, 2H), 3.19 (dd, J = 8.9, 7.7 Hz, 1H), 2.53 (td, J = 7.3, 2.7 Hz, 2H), 2.28 (t, J = 2.7 Hz, 1H). ¹³C NMR (75 MHz, MeOD): δ 103.0, 80.3, 76.60, 76.57, 73.6, 70.2, 69.2, 67.6, 61.3, 19.2. HRMS (ESI^–): m/z calcd for [C₁₀H₁₅O₆]⁻ 231.0869, found 231.0874.

(2R,3S,4S,5R,6R)-2-(Hydroxymethyl)-6-(2-(1-methyl-1H-1,2,3-triazol-4-yl)ethoxy)tetrahydro-2H-pyran-3,4,5-triol (8)

General procedure of Zemplén deacetylation applied on acetylated glucoside 26 (0.44 g, 0.95 mmol). The residue was purified by preparative HPLC (0%, 5 min, 0 to 30% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as a colorless oil (50 mg, 18%). [α]_D²² = – 22.9 (c 0,48 MeOH). ¹H NMR (300 MHz, MeOD): δ 7.78 (s, 1H), 4.27 (d, J = 7.7 Hz, 1H), 4.09 (dt, J = 9.8, 6.5 Hz, 1H), 4.02 (s, 3H), 3.83 (dd, J = 11.8, 1.7 Hz, 1H), 3.76 (dd, J = 9.8, 6.4 Hz, 1H), 3.68–3.58 (m, 1H), 3.34 (d, J = 8.9 Hz, 1H), 3.25 (d, J = 1.1 Hz, 1H), 3.24–3.21 (m, 1H), 3.15 (dd, J = 8.9, 7.7 Hz, 1H), 2.96 (t, J = 6.4 Hz, 2H). ¹³C NMR (75 MHz, MeOD): δ 146.4, 125.2, 104.4, 78.0, 78.0, 75.0, 71.6, 69.5, 62.7, 36.9, 27.1. HRMS (ESI⁺): m/z calcd for [C₁₁H₂₀N₃O₆]⁺ 290.1352, found 290.1357.

(2R,3S,4S,5R,6S)-2-(Hydroxymethyl)-6-(4-(1-methyl-1H-1,2,3-triazol-4-yl)butoxy)tetrahydro-2H-pyran-3,4,5-triol (9)

General procedure of Zemplén deacetylation applied on acetylated glucoside 27 (0.64 g, 1.32 mmol). The residue was purified by preparative HPLC (0%, 5 min, 0 to 3% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as a colorless oil (79 mg, 19%). [α]_D²² = + 79.3 (c 0,61 MeOH). ¹H NMR (300 MHz, MeOD): δ 7.68 (s, 1H), 4.73 (d, J = 3.7 Hz, 1H), 4.02 (s, 3H), 3.82–3.68 (m, 2H), 3.66–3.57 (m, 2H), 3.52 (ddd, J = 9.9, 5.6, 2.3 Hz, 1H), 3.43 (dt, J = 9.7, 5.9 Hz, 1H), 3.35 (dd, J = 9.7, 3.7 Hz, 1H), 3.24 (dd, J = 9.8, 8.8 Hz, 2H), 2.69 (t, J = 7.3 Hz, 2H), 1.81–1.68 (m, 2H), 1.68–1.55 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 149.1, 124.2, 100.1, 75.1, 73.7, 73.6, 71.9, 68.6, 62.7, 36.9, 29.9, 27.2, 25.9. HRMS (ESI⁺): m/z calcd for [C₁₃H₂₄N₃O₆]⁺ 318.1665, found 318.1672.

(2R,3S,4S,5R,6R)-2-(Hydroxymethyl)-6-(4-(1-methyl-1H-1,2,3-triazol-4-yl)butoxy)tetrahydro-2H-pyran-3,4,5-triol (10)

General procedure of Zemplén deacetylation applied on acetylated glucoside 28 (0.36 g, 0.75 mmol). The residue was purified by preparative HPLC (0%, 5 min, 0 to 30% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as a colorless oil (107 mg, 45%). [α]_D²² = – 17.7 (c 0,51 MeOH). ¹H NMR (300 MHz, MeOD): δ 7.66 (s, 1H), 4.22 (d, J = 7.7 Hz, 1H), 4.02 (s, 3H), 3.96–3.87 (m, 1H), 3.86–3.77 (m, 1H), 3.69–3.57 (m, 1H), 3.54 (dt, J = 9.5, 6.2 Hz, 1H), 3.35 (d, J = 8.7 Hz, 1H), 3.30–3.24 (m, 2H), 3.23 (p, J = 1.3 Hz, 1H), 3.14 (dd, J = 8.9, 7.7 Hz, 1H), 2.67 (t, J = 7.4 Hz, 2H), 1.81–1.64 (m, 2H), 1.69–1.53 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 149.2, 124.2, 104.3, 78.0, 77.8, 75.1, 71.6, 70.3, 62.7, 36.9, 30.0, 27.0, 25.8. HRMS (ESI⁺): m/z calcd for [C₁₃H₂₄N₃O₆]⁺ 318.1665, found 318.1663.

(2R,3R,4S,5S,6R)-6-((4-(1-Methyl-1H-1,2,3-triazol-4-yl)butoxy)methyl)tetrahydro-2H-pyran-2,3,4,5-tetraol (11)

Benzylated glucoside 30 (440 mg, 0.64 mmol, 1.0 equiv) was dissolved in MeOH (13 mL, 0.05 M). The reaction mixture was evacuated and back-filled with nitrogen three times. Palladium on charcoal 10% (137 mg, 0.129 mmol, 0.2 equiv) was added in the reaction mixture, evacuated and back-filled with hydrogen three times, and stirred for 18 h under a H₂ atmosphere. The reaction mixture was evacuated and back-filled with nitrogen before being filtered on Celite and concentrated. The residue was purified by preparative HPLC (0%, 5 min, 0 to 30% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as an oil (172 mg, 84%). Rf = 0.18 (DCM/MeOH 9/1, v/v). ¹H NMR (300 MHz, MeOD): δ 7.66 (d, J = 4.2 Hz, 1H), 5.05 (d, J = 3.7 Hz, 1H), 4.42 (d, J = 7.7 Hz, 1H), 4.02 (s, 3H), 3.85 (ddd, J = 10.0, 5.1, 2.4 Hz, 1H), 3.69 (dd, J = 10.8, 2.0 Hz, 1H), 3.66–3.40 (m, 4H), 3.29–3.19 (m, 3H), 3.12–3.04 (m, 1H), 2.67 (t, J = 7.4 Hz, 2H), 1.76–1.63 (m, 2H), 1.63–1.51 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 149.2, 124.2, 124.2, 98.2, 93.9, 78.2, 76.9, 76.3, 74.9, 73.8, 72.2, 72.1, 72.0, 71.9, 71.8, 71.4, 36.9, 30.0, 27.2, 25.9. HRMS (ESI⁺): m/z calcd for [C₁₃H₂₄N₃O₆]⁺ 318.1665, found 318.1669.

(2R,3R,4S,5R)-2-(Hex-5-yn-1-yloxy)-5-(hydroxymethyl)tetrahydrofuran-3,4-diol (12)

General procedure of Zemplén deacetylation applied on acetylated riboside 23 (0.3 g, 0.84 mmol) to afford the title compound as a colorless oil (193 mg, quantitative). [α]_D²³ = – 42.4 (c 0.53 MeOH). ¹H NMR (300 MHz, MeOD): δ 4.86 (s, 1H), 4.05 (dd, J = 6.9, 4.7 Hz, 1H), 3.95 (td, J = 6.7, 3.5 Hz, 1H), 3.89 (dd, J = 4.7, 1.0 Hz, 1H), 3.78 (dt, J = 9.7, 6.3 Hz, 2H), 3.73 (dd, J = 11.7, 3.5 Hz, 1H), 3.56 (dd, J = 11.7, 6.5 Hz, 1H), 3.42 (dt, J = 9.6, 6.1 Hz, 1H), 2.27–2.15 (m, 3H), 1.75–1.63 (m, 2H), 1.63–1.52 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 107.3, 83.4, 74.9, 71.4, 66.8, 63.8, 28.3, 25.0, 17.3. Terminal alkyne carbon undetermined. HRMS (ESI^–): m/z calcd for [C₁₁H₁₇O₅]⁻ 229.1076, found 229.1079.

(2S,3R,4S,5R)-2-(Hex-5-yn-1-yloxy)-5-(hydroxymethyl)tetrahydrofuran-3,4-diol (13)

General procedure of Zemplén deacetylation applied on acetylated riboside 24 (0.32 g, 0.89 mmol) to afford the title compound as a colorless oil (205 mg, quantitative). [α]_D²³ = + 101.6 (c 0,53 MeOH). ¹H NMR (300 MHz, MeOD): δ 4.98 (d, J = 4.2 Hz, 1H), 4.03–3.91 (m, 3H), 3.80 (dt, J = 9.9, 6.4 Hz, 1H), 3.69 (dd, J = 12.0, 3.5 Hz, 1H), 3.61 (dd, J = 12.0, 4.2 Hz, 1H), 3.53 (dt, J = 9.9, 6.3 Hz, 1H), 2.27–2.18 (m, 3H), 1.83–1.70 (m, 2H), 1.63 (tdd, J = 9.3, 6.2, 4.8 Hz, 2H). ¹³C NMR (75 MHz, MeOD): δ 102.1, 85.0, 83.5, 71.7, 69.9, 68.2, 67.3, 61.9, 28.3, 25.0, 17.4. HRMS (ESI^–): m/z calcd for [C₁₁H₁₇O₅]⁻ 229.1076, found 229.1068.

(2R,3S,4R,5R)-2-(Hydroxymethyl)-5-(4-(1-methyl-1H-1,2,3-triazol-4-yl)butoxy)tetrahydrofuran-3,4-diol (14)

General procedure of Zemplén deacetylation applied on acetylated riboside 29 (0.61 g, 1.48 mmol). The residue was purified by preparative HPLC (0%, 5 min, 0 to 30% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as a colorless oil (123 mg, 29%). [α]_D²² = – 21.4 (c 0,49 MeOH). ¹H NMR (300 MHz, MeOD): δ 7.65 (s, 1H), 4.89–4.74 (m, 5H), 4.02 (s, 2H), 4.00 (d, J = 4.7 Hz, 1H), 3.90 (td, J = 6.7, 3.5 Hz, 1H), 3.85 (dd, J = 4.7, 1.0 Hz, 1H), 3.79–3.66 (m, 2H), 3.68 (dd, J = 11.6, 3.5 Hz, 2H), 3.50 (dd, J = 11.7, 6.5 Hz, 1H), 3.37 (dt, J = 9.6, 6.2 Hz, 1H), 3.31 (s, 2H), 2.66 (t, J = 7.4 Hz, 2H), 1.73–1.61 (m, 2H), 1.61–1.49 (m, 2H). ¹³C NMR (75 MHz, MeOD): δ 149.1, 124.1, 108.7, 84.7, 76.2, 72.7, 68.3, 65.1, 36.9, 30.0, 27.1, 25.9. HRMS (ESI⁺): m/z calcd for [C₁₂H₂₂N₃O₅]⁺ 288.1559, found 288.1563.

(E)-6-(6-(((2R,3R,4S,5R)-3,4-Dihydroxy-5-(hydroxymethyl)tetrahydrofuran-2-yl)oxy)hexyl)-3-hydroxypicolinaldehyde Oxime (15)

General procedure of oxime formation/deacetylation applied on aldehyde 52 (318 mg, 0.66 mmol). The residue was purified by normal phase flash chromatography (5% to 30% MeOH in DCM over 30 min, 25G SIHC) and then purified by preparative HPLC (5% to 50% MeCN in H₂O over 30 min, puriflash C18, 30 × 250 mm, C18AQ-5 micro) to afford the title compound as an oil (128 mg, 52%). Rf = 0.22 (90/10 DCM/MeOH, v/v). [α]_D²⁰ = −26.3(c 0,51 MeOH). ¹H NMR (300 MHz, CDCl₃): δ 8.29 (s, 1H), 7.27 (d, J = 8.5 Hz, 1H), 7.15 (d, J = 8.5 Hz, 1H), 4.83 (s, 1H), 4.03 (dd, J = 6.9, 4.7 Hz, 1H), 3.93 (td, J = 6.7, 3.5 Hz, 1H), 3.87 (d, J = 4.7 Hz, 1H), 3.79–3.65 (m, 2H), 3.54 (dd, J = 11.7, 6.6 Hz, 1H), 3.36 (dt, J = 9.3, 6.3 Hz, 1H), 2.71 (dd, J = 8.7, 6.7 Hz, 2H), 1.67 (h, J = 7.2 Hz, 2H), 1.54 (q, J = 6.6 Hz, 2H), 1.37 (h, J = 5.5, 4.5 Hz, 4H). ¹³C NMR (75 MHz, CDCl₃): δ 154.8, 153.8, 152.8, 136.2, 126.1, 125.3, 108.7, 84.7, 76.3, 72.8, 68.8, 65.2, 37.8, 31.3, 30.6, 30.1, 27.1. HRMS (ESI⁺): m/z calculated for [C₁₇H₂₇N2O₇]⁺ 371.1818, found 371.1824.

The stability in solution of bi-functional oxime 2 and tri-functional oxime 3 was monitored by HPLC (method C) at 25 °C during 52 h after dissolution in 0.01 M PBS buffer at 50 μM. An LC–MS (method C) analysis was carried out at the end of the experiment for each sample (see Supporting Information).

Molecular Docking

Flexible dockings with the newly designed sugar oxime conjugates have been performed using AutoDock Vina³⁷ as previously described³¹ with the receptor files for GB-hAChE and VX-hAChE. For each phosphylated hAChE, 10 poses per oxime were generated and the best identified poses were determined based on the minimal distance between the oxygen atom of the oxime and the phosphorus atom of the OP serine conjugated and by the higher binding affinity achieved.

Recombinant Human Acetylcholinesterase

Recombinant hAChE was produced and purified as previously described.³⁸

Phosphylation of Recombinant Human Acetylcholinesterase

Stock solutions of OPNAs surrogates (NIMP as sarin surrogate, NEMP as VX surrogate and NEDPA as tabun surrogate) or VX at 5 mM and paraoxon (POX) at 20 mM in isopropanol were used to inhibit the purified hAChE as previously described.³⁹ It is worth noticing that using the OPNA surrogates gives the same phosphyl residue compared to the real OP and thus the same phosphylated AChE. Briefly, a 10-fold excess of OPNA surrogates or VX was used to perform the inhibition of hAChE in a sodium phosphate buffer (100 mM, pH 7.4, 0.1% BSA) at 25 °C. Complete inhibition of hAChE was monitored by measuring the residual activity with a modified Ellman assay as previously described,⁴⁰ and excess of OPNA surrogates or VX was removed using a desalting PD-10 column (GE Healthcare).

IC₅₀ Measurements

Oximes were dissolved in methanol to prepare a 40 mM stock solution and subsequently diluted in water to reach the desired concentrations. Recombinant hAChE activity was measured in a buffer containing 0.1 M phosphate buffer pH 7.4/0.1% BSA/0.1 mg/mL DTNB/1 mM ATC and in the presence of various oxime concentrations using a modified Ellman assay⁴⁰ measuring the released thiophenol concentration through the UV/vis absorbance at 412 nm and 25 °C. Measurements were performed at least in duplicate for each tested concentration, and final methanol concentrations were kept below 5%. The compound concentration producing 50% inhibition was determined by nonlinear fitting with ProFit (Quantumsoft) using the standard IC50 equation:

In Vitro Reactivation of Phosphylated Human Acetylcholinesterase

The reactivation of OPNA-inhibited hAChE was performed as previously described.¹⁷ Briefly, the phosphylated enzymes were incubated at 37 °C with various concentrations of oximes in 0.1% BSA/0.1 M phosphate buffer pH 7.4. The final concentration of methanol was kept below 2% and had no effect on enzyme activity. Aliquots of the reactivation mixture were transferred at different time intervals to cuvettes containing 1 mM ATC in 1 mL of Ellman’s buffer (0.5 mM DTNB in 0.1 M sodium phosphate buffer, pH 7.4) to measure the hAChE activity through the evaluation of the amount of thiophenol obtained by thiocholine cleavage of DTNB through measurement of its UV–vis absorption at 412 nm and 25 °C. The reactivation constants k_obs, K_D, k_r, and k_r2 have been calculated by non-linear fitting of the standard oxime-concentration-dependent reactivation equation with ProFit (Quantumsoft) derived from the following scheme:

Human Acetylcholinesterase Crystallization

Recombinant hAChE crystals were grown using the hanging drop vapor diffusion method as described previously.³⁸ Crystals were soaked 60 min in the mother liquor containing 2 mM of each compound. Crystals were then washed with a cryoprotectant solution (1.6 M lithium sulfate, 100 mM HEPES pH 7.0, 60 mM magnesium sulfate, and 18% glycerol) and flash-cooled in liquid nitrogen.

Data Collection, Reduction, and Refinement

Diffraction data were collected at the European Synchrotron Radiation Facility (ESRF, Grenoble, France) at the ID23-2 beam line (wavelength = 0.873 Å), processed with XDS⁴¹ and scaled with XSCALE. The structure was solved by molecular replacement with PHASER⁴² using PDB 4EY4 as starting model, iterative cycles of model building using Coot,⁴³ and refinement using Phenix.⁴⁴ Data collection and refinement statistics are presented in the Supporting Information (Table S2). The quality of the 7P1P and 7P1N structure was investigated by comparing the accuracy of the model with the real space difference density Z-score (RSZD) and the accuracy of the model with the real space observed density Z-score (RSZO) for the ligands. These metrics were studied using EDSTATS⁴⁵ and MOLPROBITY⁴⁶ in the CCP4 suite.

Animals

We used 9 week old male Swiss mice (Janvier Labs, Le Genest-Saint-Isle, France), weighing 35–45 g at the experimentation time. The animals (3–4/cage) were housed for 14–18 days before the experiments in an environment maintained at 22 ± 1 °C with controlled humidity on a 12 h dark/light cycle with light provided between 7 a.m. and 7 p.m. They were given food and tap water ad libitum. All experiments were carried out in compliance with the European Directive on the protection of animals used for scientific purposes (2010/63/UE) and were approved by our Institutional Animal Care and Research Advisory Committee (approval n°239 of 10-09-2018).

Plasma Test Protocols: Monitoring of the Reactivability of Blood Samples

This protocol has been previously fully described.²⁰

Standard Reactivation Curves for Pharmacokinetics

Naive mice plasma from Janvier Labs (Le Genest-Saint-Isle, France) was heated at 56 °C for 30 min to inactivate endogenous cholinesterases. In a 96-well Greiner plate, VX-phosphylated hAChE solution in sodium phosphate buffer was incubated for 30 min at 37 °C in the presence of different concentrations of oximes diluted in heat-inactivated mice plasma (0, 1, 5, 10, 25, 50, 75, 100, 150, 200, and 500 μM). A mixture of 2 mM acetylthiocholine and Ellman’s buffer (0.5 mM DTNB in 0.1 M sodium phosphate buffer, pH 7.4, 25 °C) was then simultaneously added in all wells for measurement of the resulting hAChE activity at 412 nm every 5 s for 30 min with an SAFAS spectrophotometer (Monaco). The percentage of reactivated enzyme (%E_react) was calculated as the ratio of the recovered VX-phosphylated hAChE activity and HI-6 reactivated hAChE activity, considering that maximal reactivation (i.e., %E_react = 100%) was achieved by incubating VX-phosphylated hAChE with 200 μM HI-6 diluted in heat-inactivated plasma. Standard curves were fitted with GraphPad Prism software using a simple linear regression.

Blood Sampling

Twenty-four hours before the experiment, mice were anesthetized with isoflurane gas (Vetflurane, Virbac, France) allowing the shaving of their hind limbs after a 3 min-long application of a commercial depilatory cream. Then, mice were returned to their cages to allow recovery and complete anesthesia washout. At the day of experimentation, mice received intraperitoneal (i.p.) injection of oxime at 100 μmol/kg by analogy with the previous studies conducted in our department. At various times (0, 2, 5, 10, 15, 30, 60, and 180 min after oxime injection), the saphenous vein was drilled with a needle, and approximately 20 μL of blood were collected with a heparinized capillary tube and put in a collection tube containing 2 μL of sodium heparin (Choay, Sanofi, France). Plasma was next isolated from erythrocytes by centrifugation at 4 °C, 3000g for 10 min. Plasma samples were then heated 30 min at 56 °C and treated as previously described to obtain the percentage of reactivated enzyme (%E_react) with a one-compartment model. T_max corresponds to the time when the curve reaches the peak of %E_react. The areas under the percentage of the reactivation curve (AUC) and the first moment curve (AUMC) were calculated using the trapezoidal rule. Mean residence time (MRT) was calculated as the ratio of AUMC to AUC.⁴⁷

LD₅₀ Estimation and Protective Index Using the Up-and-Down Method

LD₅₀ was estimated using the improved method of Dixon’s up-and-down procedure described by Rispin et al.⁴⁸ This method uses an iterative dose-selection algorithm. It consists of a single ordered dose progression in which mice are dosed, one at a time, at 24 h intervals. The first animal received a dose a step below the level of the best estimate of the LD₅₀. If the mouse survives, the dose for the next animal is increased by 1.1-fold the original dose; if it dies, the dose for the next animal is decreased by the same factor. In our particular conditions, the testing stops when one of the following criteria is met: (1) three consecutive animals survive at the highest dose (which is normally 2000 mg/kg); (2) five reversals occur in any six consecutive animals tested; (3) at least four animals have followed the first reversal and the specified likelihood-ratios, which compare the maximum likelihood estimate for LD₅₀ with LD₅₀ values above and below exceed the critical value of 2.5. Profile likelihood methods are used to estimate confidence intervals. In practicing the stopping criteria, the resulting LD₅₀ and the corresponding confidence interval were determined using the AOT 425 Pgm software as recommended by OECD (7). Antidotal efficacy of the oximes is expressed as a protective index (PI) with 95% confidence interval. The PI corresponds to the ratio of LD₅₀ of the studied OP agent (either NIMP, NEMP or paraoxon) combined with oxime treatment on LD₅₀ of OP alone.

Blood–Brain Barrier Permeability Tests

Human Blood–Brain Barrier Model Setting Up

The blood–brain barrier (BBB) permeability studies were performed using the in vitro human BBB model previously described and detailed.⁴⁹ After the infant’s parents signed informed consents, endothelial cells were isolated and differentiated from cord blood CD34⁺-hematopoietic stem cells according to the protocol described by Pedroso et al.⁵⁰ and then frozen. The preservation and preparation protocol of these cells issued from the human cord blood were approved by the French Ministry of Higher Education and Research (CODECOH number DC2011-1321). After thawing in 100 mm Petri dishes (Corning, VWR, Switzerland), the endothelial cells derived from human stem cells reached the confluency and were then subcultured onto matrigel (BD Biosciences, Franklin Lakes, NJ, USA, 354230)-coated Transwell inserts in the presence of bovine pericytes seeded at bottom of the wells, on the other side of Transwell inserts, to induce the properties of the BBB. Renewal of the medium [ECM basal medium (Sciencell, Carlsbad, CA, USA) supplemented with 5% (v/v) fetal calf serum, 1% (v/v) EC growth supplement (Sciencell) and 50 μg/mL gentamycin (Biochrom AG, Berlin, Germany)] of the co-culture thus set up was carried out every other day. After 6 days under these culture conditions, the endothelial cells differentiated in human brain-like endothelial cells (hBLECs) reproduced characteristics of the in vivo BBB⁴⁹ and are widely used to predict molecule toxicity and passage to the CNS.^33,51−53

Endothelial Permeability Coefficient Evaluation

The permeability of the BBB to the different oximes and ¹⁴C d-glucose was evaluated by measuring the endothelial permeability coefficient (Pe), which represents the speed of diffusion through the BLEC monolayer.^54,55 Human BLEC monolayers that developed after 6 days of co-culture were transferred into new plates containing 1.5 mL per well (abluminal compartment) of HEPES buffered-Ringer’s solution (RH; 150 mM NaCl, 5.2 mM KCl, 2.2 mM NaCl₂, 0.2 mM MgCl₂, 6 mM NaHCO₃, 2.8 mM glucose, 5 mM HEPES). The medium in the apical chambers (luminal compartment) was replaced by 0.5 mL of RH containing either one glycoconjugate oxime or 2-PAM or HI-6 or obidoxime used as oxime control. All compounds were tested at 50 μM (dose checked as non-toxic for the human BLECs) for a diffusion duration of an hour at 37 °C. Then, the amount of each oxime in the luminal and abluminal compartments was measured by mass spectrometry with a TripleTOF 5600+ system (AB SCIEX, Concord, ON, Canada). Percentages of recovery were checked between 86 and 106%. The quantification of radiolabeled ¹⁴C d-glucose (PerkinElmer, Boston, MA, USA) was performed using a scintillation counter TriCarb 2100TR (PerkinElmer, USA). The clearance principle was used to calculate a concentration-independent permeability coefficient. The mean compound cleared volume was plotted against time, and the slope was estimated by linear regression. The permeability values of the inserts (PSf for inserts with a Matrigel coating only) and the inserts with hBLECs (PSt, Matrigel-coated inserts + endothelial cells) were taken into consideration by applying the following equation: 1/PSe = 1/PSt – 1/PSf. To obtain the endothelial permeability coefficient (Pe expressed in cm/min), the permeability value (PSe) was divided by the insert’s membrane surface area (1.13 cm²).

Glossary

Abbreviations Used

AChE

acetylcholinesterase

ATC

acetylthiocholine

BBB

blood–brain barrier

BSA

bovine serum albumin

CNS

central nervous system

DMSO

dimethyl sulfoxide

DTNB

5,5-dithio-bis-(2-nitrobenzoic acid)

GB

sarin

GLUT-1

glucose transporter 1

hAChE

human acetylcholinesterase

HI-6

asoxime

NEDPA

nitrophenyl ethyl dimethylphosphoramidate

NEMP

nitrophenyl ethyl methylphosphonate

NIMP

nitrophenyl isopropyl methylphosphonate

OP

organophosphorus

OPNAs

organophosphorus nerve agents

PAMPA

parallel artificial membrane permeability assay

RSZD

real-space difference density Z score

RSZO

real-space observed Z score

2-PAM

pralidoxime

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.1c01748.

• Additional synthesis procedures, stability study by HPLC of tested compounds, crystallographic data collection and refinement statistics, in vivo protective index assessment, pharmacokinetic study, blood–brain barrier permeability tests, and NMR spectrum (PDF)

• Molecular formula strings (CSV)

Accession Codes

PDB codes for hAChE with bound 3 is 7P1P and with bound 4 is 7P1N.

Author Present Address

^∥ Université de Paris, CNRS UMR 8038, INSERM U 1268, CiTCoM, 4, avenue de l’Observatoire, 75270 Paris Cedex 06, France

Author Contributions

Synthesis was done by N.P., P.W., R.B., L.J., and P.Y.R.; molecular docking, in vitro assays, and crystallography were done by O.D.S., C.C., A.J.G., F.N., and J.D.; blood brain barrier permeability tests were done by C.L., C.C., F.G., and M.P.D.; and in vivo protective index assessment and pharmacokinetic study were done by A.S.H., A.G.C., and M.T.

The authors declare no competing financial interest.