Reagent-Controlled Synthesis of the Branched Trisaccharide Fragment of the Antibiotic Saccharomicin B

Abstract

A concise synthesis of a branched trisaccharide, α-L-Dig-(1 → 3)-[α-L-Eva-(1 → 4)]-β-D-Fuc, corresponding to saccharomicin B, has been developed via reagent-controlled α-selective glycosylations. Starting from the D-fucose acceptor, L-epi-vancosamine was selectively installed using 2,3-bis(2,3,4-trimethoxyphenyl)cyclopropene-1-thione/oxalyl bromide mediated dehydrative glycosylation. Following deprotection, L-digitoxose was installed using the AgPF₆/TTBP thioether-activation method to produce the trisaccharide as a single α-anomer. This highly functionalized trisaccharide can potentially serve as both a donor and an acceptor for the total synthesis of the antibiotic saccharomicin B.

Graphical abstract

Saccharomicin B, a heptadecasaccharide antibiotic, was first isolated and characterized in 1998 by Kong and co-workers from the cultured broths of the rare actinomycete Saccharothrix espanaensis (Figure 1).¹ This class of antibiotics was found to possess broad-spectrum activity, both in vitro and in vivo, against a wide range of Gram-positive and Gram-negative bacteria, including multidrug resistant strains of staphylococci and enterococci.² Given the pressing need for the development of new classes of antimicrobials, saccharomicin B holds the potential to serve as a next-generation antibiotic, especially if a route could be developed that was flexible enough to provide both the natural material and analogues. The biosynthesis of saccharomicin B has yet to be fully elucidated,³ and the promiscuity of the enzymes involved in the synthesis is unknown. This leaves chemical synthesis as the only avenue for the production of analogues. Before such analogue synthesis will be possible, however, an efficient route to the parent compound needs to be devised.

Figure 1.

Structures of saccharomicin B, the target trisaccharide 1, and the key deoxy-sugar monosaccharide building blocks (2–5).

This daunting molecule possesses several structural elements which would pose a major challenge to any attempted synthesis. Saccharomicin B is composed of five different deoxy-sugars including the rare L-4-epi-vancosamine (Eva), D-saccharosamine (Sac), and L-digitoxose (Dig), which are difficult to synthesize and install in a stereocontrolled fashion. Historically, it has been extremely difficult to control the stereochemical outcome of glycosylation reactions using deoxysugar donors without recourse to lengthy sequences involving temporary prosthetic groups or de novo synthesis,⁴ which introduces several steps between each glycosylation reaction. As a consequence, only two reports of studies directed at fragments of saccharomicins have been reported to date, both from McDonald’s lab.^5,6 The first of these studies was directed at the synthesis of the fucose–aglycone terminus of the molecule. A later study was directed at the synthesis of both the antipodal fucose–saccharosamine and fucose–saccharosamine–digitoxose structures.

As part of a synthetic campaign directed at the total synthesis of saccharomicin B, we sought to target the highly congested trisaccharide 1 using our reagent-controlled α-selective dehydrative glycosylation strategy.⁷ The challenges involved in assembling the branched trisaccharide fragment 1 are, first, the synthesis of a set of rare, suitably functionalized deoxy-sugar building blocks 2–5, followed by the stereospecific construction of the two α-glycosidic linkages.

Synthesis of the trisaccharide began with the construction of L-4-epi-vancosamine. This is a rare, branched-chain 2,3,6-trideoxy-aminosugar with an amine-bearing C-3 tertiary center.^8–10 Synthesis of the L-4-epi-vancosamine hemiacetal building block 2 was effectively accomplished starting from vancomycin-hydrochloride (Scheme 1). To this end, the amino group of the vancosamine moiety on vancomycin-HCl was subjected to Cbz carbamate protection, followed by acidic methanolysis of the vancosaminyl sugar under anhydrous conditions¹¹ to give the corresponding methyl glycoside 6 in 87% yield.^8a,12 Inversion of configuration at the C-4 atom of 6 was achieved via a stereospecific oxidation–reduction sequence, as previously described by Kahne and Walsh.^8b Thus, pyridinium dichromate (PDC) mediated oxidation, in the presence of acetic acid, afforded the 4-ketosugar 7 in 90% yield, which upon subsequent reduction with sodium borohydride (NaBH₄) in a mixture of MeOH–CH₂Cl₂ (1:2) at −78 °C furnished selectively the 4-epi-vancosamine derivative 8 in 97% yield. Acid hydrolysis (4 M aq HCl, THF, 0 °C) of the methyl glycoside 8, followed by regioselective O-silylation of the anomeric hydroxyl group with triisopropylsilyl chloride (TIPSCl) and imidazole in dry DMF, gave the silyl derivative 9 as a single β-anomer (85% over two steps). Catalytic hydrogenolysis (Pd/C, H₂) of the N-Cbz carbamate produced a primary amine, which was readily transformed into the corresponding azide by the copper(II)-catalyzed diazotransfer from freshly prepared trifluoromethanesulfonyl azide (TfN₃) solution in toluene.^13,14 Subsequent benzylation under standard conditions (BnBr, NaH, DMF, 0 °C) afforded the azide derivative 10 (91% yield, over three steps). Finally, selective removal of the anomeric TIPS group from 10 with tetra-n-butylammonium fluoride (TBAF, 1 M in THF) in the presence of acetic acid at room temperature cleanly provided the desired L-4-epi-vancosamine hemiacetal 2 in 98% yield.

Scheme 1.

Synthetic Pathway for the L-4-epi-Vancosamine Building Block

Synthesis of the 2,6-dideoxy-sugar, L-digitoxose,¹⁵ building blocks 3 and 4 is outlined in Scheme 2. Zemplén de-O-acetylation of the commercially available 3,4-di-O-acetyl-L-rhamnal (11) using catalytic sodium methoxide, followed by regioselective 3-O-silylation of the resulting diol with tert-butyldimethylsilyl chloride (TBDMSCl) in the presence of imidazole at 0 °C¹⁶ and subsequent etherification with 2-naphthylmethyl bromide (NapBr) and sodium hydride in anhydrous DMF at 0 °C, gave the 4-O-Nap ether protected derivative 12 in excellent yield (92%, over three steps). Desilylation (TBAF/THF) of the 3-O-TBDMS group, followed by PDC/AcOH oxidation of the resulting allylic alcohol in ethyl acetate at room temperature, gave the hex-1-en-3-ulose 13 in 73% yield over two steps. Low-temperature cesium carbonate-promoted Michael-type addition of ethanthiol to the enone derivative 13 in THF¹⁷ produced the thioethyl 2-deoxy-3-uloside 14 in 86% yield as a single α-isomer. Sodium borohydride (NaBH₄) reduction of ketone 14 in a mixture of MeOH–CH₂Cl₂ (3:1) at −15 °C was achieved with excellent stereoselectivity. ¹H and ¹³C NMR spectra of the crude product showed a single product, containing an axial C3-OH, corresponding to the desired L-digitoxose stereoisomer. Subsequent benzylation of the crude alcohol furnished the thioglycoside 3 in 94% yield over two steps. Chemoselective hydrolysis of the anomeric phenyl thioether group with N-bromosuccinimide (NBS) in wet acetone (10:1 acetone–water)^18,19 at 0 °C smoothly afforded the L-digitoxose hemiacetal 4 in quantitative yield.

Scheme 2.

Synthetic Pathway for L-Digitoxose Building Blocks

Synthesis of the D-fucose acceptors (Scheme 3) commenced with the regioselective alkylation of the equatorial 3-hydroxyl group of diol²⁰ 15, via dibutylstannylene acetal formation under Dean–Stark conditions,^21,22 using 2-naphthylmethyl bromide (NapBr) to afford the 3-O-Nap ether building block 5 in 93% yield. Benzylation (BnBr, NaH, DMF, 0 °C) of 5 at the 4-position gave quantitatively the fully protected derivative 16, and subsequent oxidative cleavage of the Nap ether with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) in a mixture of CH₂Cl₂–H₂O (20:1) afforded 17 in 83% yield.

Scheme 3.

Synthetic Pathway for the D-Fucose Building Blocks

With the appropriate deoxy-sugar glycosyl donors and acceptors in hand, we turned our attention to disaccharide formation. To this end, 2,3-bis(2,3,4-trimethoxyphenyl)-cyclopropene-1-thione 18 was initially activated with oxalyl bromide and then treated with the epi-vancosaminyl hemiacetal donor 2 and 2,4,6-tri-tert-butylpyrimidine (TTBP) in a mixture of 1,1,2-trichloroethylene (TCE) and CH₂Cl₂, followed by the addition of the Fuc-C4-acceptor 5 in THF at room temperature. This dehydrative glycosylation condition afforded the desired disaccharide 19 with complete α-stereoselectivity, albeit in a modest 16% yield along with a substantial amount of trehalose (Table 1, entry 1). After rapid screening of different reaction temperatures and times (Table 1, entries 2–6), we were able to achieve a satisfactory isolated yield (65%), with no impact on selectivity, by performing all the additions at −10 °C and then maintaining the reaction temperature at −10 °C for 1 h before warming to room temperature (Table 1, entry 6). Increasing the donor-to-acceptor ratio from 2:1 to 3:1 did not offer any improvement to the yield (Table 1, entry 7). Interestingly, our best result for the α-1,4-selective glycosylation of 2 with 5 was obtained when the donor 2 was coactivated in the presence of the acceptor 5 (Table 1, entry 8), affording disaccharide 19 in 67% yield. The α-(1 → 4) glycosidic linkage in 19 was confirmed by the ¹H NMR spectrum, which showed a doublet (δ 5.14 ppm) with a vicinal coupling constant J_1,2 for H-1′ of 4.3 Hz.

Table 1.

Optimization of α-Stereoselective Dehydrative Glycosylation for the Synthesis of Disaccharide 19^a

entry	temp (°C)	time (h)	yield (%)b
1	rt	2	16
2	0	2	39
3	−10	2	52
4	−20	18	26
5	−10 → rt	2	58
6c	−10 → rt	3	65
7c,d	−10 → rt	3	63
8c,e	−10 → rt	3	67

^aAcceptor (1 equiv), donor (2 equiv), thione (2.2 equiv), oxalyl bromide (2.2 equiv), and TTBP (6 equiv) unless otherwise noted.

^bYield of isolated product after purification.

^cReaction mixture was stirred at −10 °C for 1 h before being allowed to warm to rt.

^dAcceptor (1 equiv), donor (3 equiv), thione (3.3 equiv), oxalyl bromide (3.3 equiv), and TTBP (9 equiv).

^eDonor was coactivated in the presence of acceptor; THF was not used.

Having established the conditions for the construction of disaccharide 19, we turned our attention to the assembly of trisaccharide 1. To this end, unmasking of the Nap protecting group on 19 using DDQ afforded disaccharide 20 in 72% yield (Scheme 4). With both the disaccharide acceptor 20 and the L-digitoxose hemiacetal donor 4 in hand, we initially examined the dehydrative glycosylation to directly forge the target trisaccharide fragment. However, condensation of 20 with 4 under our cyclopropenium activation conditions proceeded in poor yield, favoring the formation of the β-isomer. In hindsight, this was not unexpected as digitoxose is known to preferentially undergo β-selective glycosylation reactions.²³

Scheme 4.

Dehydrative Glycosylation of Disaccharide Acceptor 20 with L-Digitoxose Hemiacetal Donor 4

Attempts to optimize our cycopropenium cation-mediated glycosylation on a model system failed to produce the desired anomer (see Supporting Information). Our inability to utilize our cyclopropenium method for the construction of 1 led us to consider alternative approaches. To this end, we were attracted to the mild activation of deoxy thioglycosides using silver hexafluorophosphate (AgPF₆) reported by Hirama et al.^24a This approach takes advantage of the different reactivity between 2-oxy- and 2-deoxy-thioglycosides toward AgPF₆. After evaluating the scope of the reaction with the two reported²⁴ non-nucleophilic bases (DTBMP and TTBP) using D-fucose thioglycoside 17 as a model C-3 acceptor (Table 2), we found that the AgPF₆/TTBP combination is the optimal glycosylation promoter system for the stereoselective formation of the L-Dig-α-(1 → 3)-D-Fuc linkage.

Table 2.

Optimization of AgPF₆-Mediated Glycosylation Using D-Fucose Thioglycoside 17 as a Model C-3 Acceptor^a

entry	base	yield (%)b
1	DTBMP	62
2c	DTBMP	44
3	TTBP	72

^aAcceptor (1 equiv), donor (1.5 equiv), AgPF6 (3 equiv based on donor), and base (4 equiv based on donor) unless otherwise noted.

^bYield of isolated product after purification.

^cAgPF₆ (3 equiv based on acceptor) and base (4 equiv based on acceptor).

Next, we focused on applying the optimized AgPF₆ thioether-activation method to effect the final glycosylation to assemble the target trisaccharide 1. To this end, a mixture of disaccharide acceptor 20 and glycosyl donor 3 in the presence of TTBP and 4 Å molecular sieves was treated with AgPF₆ at −10 °C for 3 h to give exclusively the desired trisaccharide 1 as a single α-anomer in 76% yield (Scheme 5). Again, selectivity was confirmed by ¹H NMR, which showed a doublet for the anomeric proton of the digitoxose residue (δ 4.78 ppm) with a vicinal coupling constant J_1,2 of 4.7 Hz. In addition, ¹³C NMR showed the anomeric carbon signals for the digitoxose and the epi-vancosamine residues at δ 98.6 ppm (J_C-1,H-1 168.7 Hz) and δ 97.2 ppm (J_C-1,H-1 173.0 Hz), respectively.

Scheme 5.

Synthesis of the Target Trisaccharide 1

In conclusion, we have developed an efficient strategy for the stereoselective synthesis of a branched trisaccharide 1, of potential value for a future synthesis of the antibiotic saccharomicin B, via reagent-controlled α-selective glycosylations of 2,6-dideoxy- and 2,3,6-trideoxy-sugars. The rare deoxy sugar building blocks were readily constructed from commercially available materials. Our cyclopropenium cation-mediated reagent-controlled dehydrative glycosylation between L-4-epivancosamine hemiacetal 2 and D-fucose C4-acceptor 5 gave 19 exclusively as the α-anomer. Following DDQ-mediated oxidative cleavage of the 3-O-Nap ether, glycosylation of the formed disaccharide acceptor 20 with the L-digitoxose thioglycoside 3, promoted by AgPF₆/TTBP, afforded the desired α-linked trisaccharide 1 as a single diastereomer. This highly functionalized trisaccharide can potentially serve as both a donor and an acceptor for the total synthesis of saccharomicin B, which is under investigation in our laboratory.