Scalable Synthesis of Versatile Rare Deoxyamino Sugar Building Blocks from d-Glucosamine

We report the syntheses of 1,3,4-tri-O-acetyl-2-amino-2,6-dideoxy-β-d-glucopyranose and allyl 2-amino-2,6-dideoxy-β-d-glucopyranoside from d-glucosamine hydrochloride. The potential of these two versatile scaffolds as key intermediates to a diversity of orthogonally protected rare deoxyamino hexopyranosides is exemplified in the context of fucosamine, quinovosamine, and bacillosamine. The critical C-6 deoxygenation step to 2,6-dideoxy aminosugars is performed at an early stage on a precursor featuring an imine moiety or a trifluoroacetamide moiety in place of the 2-amino group, respectively. Robustness and scalability are demonstrated for a combination of protecting groups and incremental chemical modifications that sheds light on the promise of the yet unreported allyl 2,6-dideoxy-2-N-trifluoroacetyl-β-d-glucopyranoside when addressing the feasibility of synthetic zwitterionic oligosaccharides. In particular, allyl 3-O-acetyl-4-azido-2,4,6-trideoxy-2-trifluoroacetamido-β-d-galactopyranoside, an advanced 2-acetamido-4-amino-2,4,6-trideoxy-d-galactopyranose building block, was achieved on the 30 g scale from 1,3,4,6-tetra-O-acetyl-β-d-glucosamine hydrochloride in 50% yield and nine steps, albeit only two chromatography purifications.


■ INTRODUCTION
Carbohydrates are ubiquitous cell surface components. They play major roles in a myriad of complex biological processes governed by cell−cell or cell−environment interactions including host−pathogen recognition. Going far beyond the structural diversity seen within the human glycome, the prokaryote glycome is highly varied. 1,2 Diversity stems for a large part from the large number of unique monosaccharides composing its alphabet. It expands as novel monosaccharides are revealed paralleling the structural analysis of natural glycans of increasing complexity. 3 Among distinct monosaccharides not found in mammalian glycans are several rare deoxyamino sugars often present in glycans from pathogenic bacteria but essentially absent from the human microbiota. 4,5 Whether as components of zwitterionic polysaccharides (ZPSs), 6,7 substrates for selective metabolic labeling, 4 key biosynthetic intermediates, and potential targets for novel antibiotics 8 or in the context of epitope mapping and vaccine design, 7,9 2-amino-2,6-dideoxyhexoses (ADDH) and 2,4-diamino-2,4,6-trideoxy-hexoses (DATDH) are the subject of wide interest. 10 Mostly identified in ZPSs, AAT (D-FucpNAc4N) is among the most studied DATDH. Being α-linked in glycans from the well-established immunomodulators Streptococcus pneumoniae 1 (Sp1) 11 and Bacteroides fragilis (PS A1), 12 in the enterobacterial common antigen 13 and in several other ZPSs, it is β-linked in the surface polysaccharides from Shigella sonnei, 14 Plesiomonas shigelloides, 15 and other bacteria. 16−19 Otherwise, it is occasionally found in both forms within the same glycans such as in the S. pneumoniae lipoteichoic acid (Sp LTA). 20 AAT also occurs in various 4-acylamino forms. 21 It may even be present as both 4amino and 4-acetylamino residues within selected strains. 22 Obviously, the different N-substitutions at C-2 and C-4 increase the synthetic complexity.
The growing interest in large homogeneous segments of ZPSs 6,11,47 has underlined the importance of robust and scalable versatile strategies to orthogonally protected ADDH. Herein, going beyond the existing paths, we propose two handy 2amino-2,6-dideoxy scaffolds easily achievable in high yield on a large scale from the naturally abundant D-glucosamine hydrochloride. We exemplify versatility through their conversion into diverse protected ADDH and DATDH. Moreover, we demonstrate robustness for the 30 g scale synthesis of a key orthogonally protected AAT brick used in the assembly of S. sonnei glycans of interest for vaccine development. 47

■ RESULTS AND DISCUSSION
Starting from D-glucosamine hydrochloride, the limiting C-6 deoxygenation step is commonly performed on a 2-azido or a 2phthalimido intermediate ( Figure 1E,F). 38 Instead, we have reported the 2-trichloroacetamide 2, easily obtained from 1,3,4,6-tetra-O-acetyl-β-D-glucosamine hydrochloride 1 (Scheme S1), as a precursor to the AAT building block 4 (Scheme 1). 45 In this work, the C-6 reduction step of the 6-iodo precursor 2 into trichloroacetamide 6 met with issues. Variations around the initial conditions did not fulfill the efficiency criteria either (Scheme S2). In particular, the use of NaBH 3 CN/CuX (X = Cl, Br, or I) was found less effective. 51 Radical-mediated reduction by means of (TMS) 3 SiH/AIBN, 52 Bu 3 SnH/Et 3 B, 53 or (TMS) 3 SiH/Et 3 B did not overcome limitations. 54 Concomitant hydrodechlorination yielding dichloroacetamide 7 could not be avoided, and chloroacetamide 8 was also observed under certain conditions. Similar observations stemmed from the Seeberger group 55 and the Codeé group, 56 which reported that the thioland phosphine-mediated azide reduction of unrelated 2trichloroacetamido derivatives had generated the corresponding 2-dichloroacetamido side products, respectively. We have previously achieved dichloroacetamide 5 from trichloroacetamide 4 (Scheme 1). 47 Herein, the more straightforward synthesis of the AAT acceptor 5 from precursor 3 also met with limitations. In particular, chloroacetamide 8 and the 6chloro derivative 9 were observed repeatedly upon radical 51,53 or hydride-mediated reduction, 45 respectively (Schemes 1 and S3). Having reconsidered the interdependence between the five key actions identified to build ADDH and DATDH from Dglucosamine ( Figure 1D), we now propose two original alternatives to the existing paths. The critical C-6 deoxygenation step is performed on a precursor featuring either an imine moiety at position 2 in place of the starting 2-amino group (generic route 1) or a 2-trifluoroacetamide moiety to mask that same amino group (generic route 2).
Generic Route 1. Aiming to achieve 4 while avoiding interference of the TCA moiety during C-6 deoxygenation, we set to investigate a route featuring a late-stage suitable Nprotection step. We reasoned that a route encompassing an early deoxygenation step would offer a faster and more versatile access to intermediates bearing any protecting group combination. Satisfactorily, treatment of the upstream 6-iodo imine 11 with Bu 3 SnH/Et 3 B provided the 6-deoxy analogue 12 in high yield (Scheme 2). Varying the tosylation conditions had little influence (Scheme S4). In contrast, changing the tosyl group for the bulkier triisopropylbenzenesulfonyl moiety was beneficial. Acidic hydrolysis of the C-2 imine gave the versatile 1,3,4-tri-O-acetyl-β-D-quinovosamine (QuiN) 13 from Dglucosamine hydrochloride in five steps and 32% yield on a 5 g scale. Conversion of triacetate 13 into the 2-azido analogue 14 paved the way to a panel of AAT and QuiN donors ready for 1,2cis glycosylation. Otherwise, starting from imine 12, Nunmasking and subsequent trichloroacetylation of intermediate 13 furnished triacetate 15, in turn converted into allyl glycoside 6, a well-established AAT precursor. 45 This efficient three-step protocol provides an access to the known 6 in eight steps from Dglucosamine hydrochloride (Scheme 2A). The process was adapted to give the N-dichloroacetyl (7), N-trifluoroacetyl (19), and N-Troc (20) analogues with excellent anomeric control (Scheme 2B). Likewise, the thiophenyl QuiN 21 was easily achieved from imine 12 by means of trifluoroacetamide 17, demonstrating that aglycon diversification post-N-protection is Scheme 1. Initial Route to the 2-N-Trichloroacetyl and 2-N-Dichloroacetyl AAT (4 and 5) and Quinovosamine (QuiN, 6 and 7) Building Blocks a straightforward. Subjected to improving the path to iodide 11, imine 12 represents a highly versatile precursor to ADDH and DATDH.
Generic Route 2. Aiming at improving robustness, we identified the more advanced allyl 2,6-dideoxy-2-trifluoroacetamido-β-D-glucopyranoside 28 as a promising alternative to imine 12 and azide 14 (Scheme 4). The trifluoroacetyl moiety was envisioned as an easy-to-introduce participating Nprotecting group stable under harsh reductive conditions, albeit seemingly cleavable under relatively mild basic conditions. 58 The synthesis of diol 28 from 1 by means of 24 58 followed that reported for the trichloroacetamide analogue (Scheme 1). 45 The conversion of tetra-acetate 1 into the 6-iodo 27 went smoothly to deliver the latter intermediate in 77% yield over six steps (Scheme 4A). It is noteworthy that the key reduction step furnished diacetate 19 in excellent yield (96%, 40 g scale) when using an optimized Bu 3 SnH/Et 3 B combination (Scheme S6). Robustness was confirmed as reduction at C-6 of 27 and subsequent transesterification gave diol 28 in a 90−97% yield on a 20 g scale (Scheme S7). Moreover, investigation toward scaleup demonstrated that re-O-acetylation post-tosylation was not mandatory. This step and the related transesterification of diacetate 19 into diol 28 were removed from the process. As a reward, scaffold 28, serving as a versatile intermediate to protected ADDH and DATDH, was obtained in six steps and 78% overall yield from the commercially available 1 on a 150 mmol scale (Schemes 4B and S8). Modification at positions 2, 3, and 4 of diol 28 can follow multiple paths (Scheme 4A), in part exemplified below.
From the N-Trifluoroacetyl QuiN 28 to AAT and D-FucNAc Building Blocks: 3-O-Protection and C-4 Inversion. Applying the diisopropylethylamine (DIPEA)-triggered self-catalyzed regioselective acetylation conditions 59 to diol 28 provided acetate 29 together with its regioisomer (Scheme 5, entries 1 and 2). Aiming at avoiding formation of the latter, diol 28 was treated with acetyl chloride under conditions previously established for the trichloroacetamide analogue. 45 This more demanding protocol delivered the expected 29 with an improved regioselectivity (entry 3). This option was adopted on a multigram scale. Otherwise, the 3-O-benzoyl 30 was achieved smoothly (77%) in the presence of Me 2 SnCl 2 (entry 4). 60 In contrast, tert-butyldimethylsilylation required attention to furnish the expected 31. The 4-O-tert-butyldimethylsilyl isomer was always present (entries 5 and 6). Moreover, the 4-Obenzyl 32 was formed preferentially while the reaction was stopped before completion to avoid extensive N-benzylation in the presence of sodium hydride (entry 7). To our knowledge, the observed regioselectivity was not reported previously for Nprotected QuiN derivatives. Further attempts at achieving a better regioselectivity by use of a diaryl boronic acid catalyst 61 or an iron catalyst 62 were unsuccessful.
Alcohol 29 was next elaborated into D-FucNAc 33 in high yield via triflate-mediated inversion using the nitrite-mediated Lattrell-Dax method. 63 Alternatively, azide-mediated substitution of the intermediate triflate readily generated upon reaction of 29 with triflic anhydride in the presence of pyridine delivered the fully protected AAT 34 in excellent yield (Scheme 6). Moreover, scale-up toward a robust synthesis of AAT was in agreement with expectations. In particular, the key intermediate

■ CONCLUSIONS
The proof of concept is established for two versatile strategies to rare D-deoxyamino sugars from the naturally abundant Dglucosamine hydrochloride and its more advanced commercially available 1,3,4,6-tetra-acetate equivalent. Key features include a temporary 2-N-protection in the form of an imine or a trifluoroacetamide, enabling a high yielding early stage C-6 deoxygenation step. The tosylation−iodination−reduction at the primary hydroxyl was achieved in high yield on the 3,4,6-triol substrate. Subsequent diversity-oriented chemical manipulation provided orthogonally protected QuiN, FucNAc, and AAT intermediates equipped for 1,2-cis and/or 1,2-trans glycosylation. Versatility is illustrated in terms of protecting group selection and sequential introduction to deliver multiple ADDH and DATDH bricks. Some examples are closely related to known compounds as the disclosed routes may be envisioned as alternatives to published syntheses. The novel 6-deoxy-Dglucosamine derivatives, diacetate 19 and its diol equivalent 28, both of which feature a 2-trifluorocatamide moiety, were shown to be easily achievable on the multidecagram scale. While these allyl glycosides are the main model scaffolds in this study, the proposed strategies are easily applicable to other substrates, especially thioglycosides, thus enabling greener C-6 reduction conditions. Beyond novelty, this study adds to previous demonstrations of the potential of fine-tuned generic scaffolds for the synthesis of rare deoxyamino sugars, the interest for which is manifest. Synthesis robustness and scalability are demonstrated for the pivotal QuiN diol 28 (30 g, six steps, 73%, no chromatography purification) and the more advanced AAT Compounds were visualized using UV (λ = 254 nm) and/or orcinol (1 mg·mL −1 ) in 10% aq H 2 SO 4 with charring. Flash column chromatographies were carried out using silica gel (25 or 40−63 μm particle size). Reactions that required heating were run in flasks equipped with an air flux condenser using a heat-on block equipped with an external temperature probe and filled in with sand whenever necessary to ensure proper heat transfer. NMR spectra were recorded at 303 K on a Bruker AVANCE spectrometer equipped with a BBO probe at 400 MHz ( 1 H) and 100 MHz ( 13 C). Spectra were recorded in CDCl 3 , CD 3 CN, CD 3 OD, and DMSO-d 6 . Chemical shifts are reported in ppm (δ) relative to residual solvent peaks at 7.28/77.0 ppm for CDCl 3 , 1.39/ 1.32 ppm for CD 3 CN, 3.33/49.0 ppm for CD 3 OD, and 2.50/39.5 ppm for DMSO-d 6 for the 1 H and 13 C spectra, respectively. Coupling constants are reported in hertz (Hz). Elucidation of chemical structures is based on 1 H, COSY, DEPT-135, HSQC, decoupled HSQC, 13 C, decoupled 13 C, and HMBC spectra. Signals are reported as s (singlet), d (doublet), t (triplet), dd (doublet of doublet), q (quartet), dt (doublet of triplet), dq (doublet of quartet), ddd (doublet of doublet of doublet), m (multiplet), and broad (prefix br). Of the two magnetically nonequivalent geminal protons at C-6, the one resonating at a lower field is denoted as H-6a, and the one at a higher field is denoted as H-6b. HRMS spectra were recorded in the positive-ion electrospray ionization (ESI + ) mode on a WATERS QTOF Micromass instrument or on a Q exactive mass spectrometer (Thermo Fisher Scientific) equipped with a H-ESI II Probe source. Solutions were prepared using 1:1 MeCN/H 2 O containing 0.1% formic acid. In the case of sensitive compounds, solutions were prepared using 1:1 MeOH/H 2 O to which was added 10 mM ammonium acetate.