Reagent Controlled Direct Dehydrative Glycosylation with 2-Deoxy Sugars: Construction of the Saquayamycin Z Pentasaccharide

Abstract

The first synthesis of the pentasaccharide fragment of the angucycline antibiotic saquayamycin Z is described. By using our sulfonyl chloride mediated reagent controlled dehydrative glycosylation, we are able to assemble the glycosidic linkages with high levels of anomeric selectivity. The total synthesis was completed in 25 total steps, and in 2.5% overall yield with a longest linear sequence of 15 steps.

Graphical Abstract

Oligosaccharides often play critical roles in a number of biological processes, including modulating the bioactivity of natural products.^1,2 While methods have been developed for the efficient construction of oligosaccharides composed of common sugars such as glucose and galactose,^3,4 the construction of structures composed of unusual and uncommon sugars still presents a daunting synthetic challenge. One particular class of unusual sugars are 2-deoxy sugars, which are frequently found in bioactive secondary metabolites of microbial origin.⁵ These deoxy sugar-containing oligosaccharides play critical roles in modulating the antibacterial and anticancer properties of natural products, such as digitoxin^6,7 and the landomycins.⁸ Despite several decades of intense research by the synthetic community, and calls for furthering the development of carbohydrate synthesis as a field, a general approach to the efficient and selective construction of deoxyglycoside oligosaccharides has yet to emerge.⁹

A major challenge with the synthesis of deoxy-glycosides is controlling the selectivity in glycosylation reactions with deoxy-sugar donors. This is due to both their increased reactivity and lack of functionality for appending directing groups on the donor that can bias the selectivity of glycosylation reactions.^10,11 In recent years this has led several research groups to examine new methods for the stereo-selective construction of 2-deoxy glycosides.^11–36 Our own group has had a longstanding interest in developing glycosylation methodologies for deoxy glycoside synthesis wherein the stereochemical outcome of the reaction is determined by the promoter.^37–42 Most recently, we have demonstrated that either tosyl chloride or triisopropylbenzene-sulfonyl chloride (Trisyl chloride) can efficiently activate deoxy-sugar hemiacetals for highly selective glycosylation reactions with a number of 2,6-dideoxy- and 2,3,6-trideoxy-sugar donors.³⁷ Having established that this chemistry was useful for the construction of disaccharide linkages, we sought to determine its utility in the synthesis of more complex oligosaccharides.

In order to demonstrate the utility of our approach in a complex setting, we chose to apply it to the construction of the pentasaccharide fragment of saquayamycin Z (Figure 1). First isolated in 2005 by Ströch and co-workers, saquayamycin Z was shown to possesses activity against various Gram-Positive bacteria (P. aurenscens, B. subtilis, S. aureus, MIC = 0.1–1 μM) as well as against several human cancer cell lines (HM02, MCF7, and HepG2, GI₅₀ < 650 nM).^43,44 It is also the largest and most structurally complex of the saquaymycin family, uniquely possessing tetra- and pentasaccharide subunits.^8,45–47 The pentasaccharide fragment of this molecule is composed of di- and trideoxy-sugars with alternating α- and β-glycosidic linkages, providing an ideal target to examine the utility of our chemistry in a more intricate system. Here we report the first synthesis of this pentasaccharide.

Figure 1.

Saquayamycin Z, with the highlighted pentasaccharide of interest in red.

Retrosynthetically, we envisioned that the pentasaccharide 1 would arise from l-rhodinose 3 and tetrasaccharide 2 (Scheme 1). This latter compound could in turn result from a [2 + 2] glycosylation between selectively deprotected derivatives of d-olivose-(β₁→4)-l-rhodinose disaccharide 4, which could ultimately be constructed from rhodinose 6 and olivose 5.^48,49 These monosaccharides would arise from l- and d-rhamnal, respectively.

Scheme 1.

Retrosynthesis of the Saquayamycin Z Pentasaccharide

Our initial studies focused on construction of the requisite rhodinose derivatives 3 and 6. The synthesis of l-rhodinose derivative 6 was adapted from an approach described by Sulikowski (Scheme 2A).^50,51 To this end, allylic alcohol 7 was prepared from l-rhamnal in 70% yield using a Ferrier-like Mitsunobu reaction.⁵⁰ Inversion of the C4 position was accomplished via a second Mitsunobu reaction with 4-nitrobenzoic acid,⁵² followed by saponification of the resulting ester to afford allylic alcohol 8 in 67% yield over two steps. Finally, while catalytic hydrogenation over palladium on carbon gave a poor yield of 6 due to cleavage of the anomeric PMP group, hydrogenation over rhodium on alumina⁵³ led to the production of 6 in 80% yield.

Scheme 2.

Synthesis of the Starting Monsaccharides: Rhodinose Acceptor 6 (A), Rhodinose Donor 3 (B), and Olivose Donor 5 (C)

Unfortunately, attempts to convert acceptor 6 into donor 3 through acetylation followed by oxidative removal of the anomeric PMP group with ceric ammonium nitrate (CAN) were largely unsuccessful, forcing us to examine an alternative route to this molecule. A viable pathway was found by adapting a synthesis of 6-deoxy-l-talose described by Banaczek and coworkers (Scheme 2B).⁵² To this end, a BF₃·OEt₂-mediated Ferrier rearrangement on rhamnal using benzyl alcohol as a nucleophile followed by deacetylation afforded 10 as a mixture of anomers (α:β = 7:1) in 88% yield.^52,54 Subjecting this molecule to a three-step sequence of Mitsunobu inversion, saponification, and acetylation of the resulting alcohol afforded 11 in 86% yield over three steps. Finally, catalytic hydrogenation over palladium on carbon resulted in both reduction of the alkene and cleavage of the anomeric benzyl ether to provide donor 3 in 72% yield.

The d-olivose donor was synthesized from d-rhamnal 13 (Scheme 2C), which was in turn prepared from commercially available d-glucal 12 by a procedure adapted from Takahashi (see Supporting Information).³¹ From rhamnal, dibutyltin oxide mediated regioselective protection of the allylic alcohol in 13 with 2-(bromomethyl)naphthylene (NapBr) afforded 14 in 76% yield with complete regioselectivity. The remaining C4 hydroxyl was then protected as a benzyl ether under Williamson ether synthesis conditions to afford 15 in 98% yield. Finally, hydration of the glycal using catalytic triphenylphosphine hydrobromide afforded hemiacetal donor 5 in 83% yield.^49,55 With all the necessary monosaccharides in hand, we turned our attention to their assembly into the pentasaccharide target.

We first focused on the construction of disaccharide 4, which would serve as a precursor for both the donor and acceptor disaccharides needed to construct the penultimate tetrasaccharide. Adapting conditions from our previously optimized tosyl chloride promoted glycosylation, we initially examined a 30 min activation time of hemiacetal 5 with TsCl prior to addition of the nucleophile.³⁷ Under these conditions we were able to obtain the desired β-linked product 4 in 55% yield as a single diastereomer (Table 1, entry 1). While extending the activation time led to a mild decrease in yield (entries 2–3), we found that a shortened activation time led to a slight increase in yield to 59% without loss of selectivity (entry 4). Substitution of 2,6-di-tert-butyl-4-methylpyridine (DTBMP) as an additive in place of 2,4,6-tri-tert-butylpyrimidine (TTBP) also did not lead to improvement of the yield (entries 5–7). Attempts to make the acceptor more nucleophilic and more reactive through the addition of 18-crown-6 also led to similar yields (entry 8). Finally, we examined the use of other non-nucleophilic bases as proton scavengers; however, this led to losses in both yield and selectivity (entries 9–11). Moving forward, we opted to stay with glycosylation conditions from entry 7 as the most efficient route to disaccharide 4.

Table 1.

Optimization of Disaccharide 4

entrya	donor/acceptor	additive	activation time	yield	α:βb
1	2:1	TTBP	30 min	55%	β-only
2	2:1	TTBP	180 min	52%	β-only
3	2:1	TTBP	120 min	48%	β-only
4	2:1	TTBP	15 min	59%	β-only
5	2:1	DTBMP	15 min	35%	1:5
6	2:1	DTBMP	30 min	55%	β-only
7	1.5:1	DTBMP	30 min	56%	β-only
8c	1.5:1	DTBMP	30 min	58%	β-only
9	2:1	–	30 min	39%	1:2.5
10	1.5:1	TTBPy	30 min	27%	1:10
11	1.5:1	PEMP	30 min	46%	1:9
12d	1.5:1	DTBMP	60 min	54%	1:7

^aAll reactions performed with 0.5 g of acceptor 6.

^bAnomeric ratio measured by ¹H NMR.

^c18-crown-6 added during acceptor metalation.

^dAcceptor scale 1 g of TTBP = 2,4,6-tri-tert-butyl-pyrimidine, DTBMP = 2,6-di-tert-butyl-4-methylpyridine, TTBPy = 2,4,6-tri-tert-butylpyridine, PEMP = 1,2,2,6,6-pentamethylpiperidine.

Having established a route to disaccharide 4, we turned our attention to the construction of donor 17 and acceptor 16. To this end, removal of the naphthylmethyl (Nap) ether using DDQ in the presence of the proton scavenger β-pinene afforded acceptor 16 in 79% yield without affecting the anomeric PMP.⁴⁸ Additionally, selective removal of the anomeric PMP protecting group without affecting the Nap ether was accomplished using CAN to afford hemiacetal donor 17 in 93% yield (Scheme 3).⁵⁶

Scheme 3.

Selective Deprotections of Disaccharide 4

With both the disaccharide donor and acceptor in hand, we turned to the construction of tetrasaccharide 2. Our approach was based on previous work from our lab, wherein we found that a 4-O-acetyl protected d-amecitose hemiacetal underwent α-selective glycosylation when using 2,4,6-triisopropylbenzenesulfonyl chloride (trisylCl) as a promoter. We hypothesized that such a system would prove to be effective here as well, owing to the similarity between amicetose and rhodinose.^37,57 Attempts to couple 16 and 17 using trisyl chloride afforded the desired tetrasaccharide in 65% yield, but as an inseparable 4:1 (α:β) mixture of isomers (Table 2, entry 1). Using tosyl chloride as a promoter led to a complete loss of selectivity (entry 2). Both shortening and extending the activation time led to losses in selectivity (entries 3–4). Ultimately maintaining the reaction at −78 °C, using freshly recrystallized trisyl chloride, and substituting DTBMP for TTBP led to a slight increase in selectivity to 5:1 (α:β) accompanied by a small decrease in yield (entry 6). These results led us to hypothesize that the previously observed α-selectivity was attributable to the C-4 acetate protecting group. As such, we wanted to take a closer look at this coupling.

Table 2.

Optimization of Tetrasaccharide 2

entry	scavenger	promoter	activation time	yield	α:βa
1	TTBP	Trisyl-Cl	60 min	65%	4:1
2	TTBP	TsCl	60 min	46%	1:1
3	TTBP	Trisyl-Cl	120 min	50%	1:1
4	TTBP	Trisyl-Cl	30 min	55%	2:1
5b	TTBP	Trisyl-Clc	60 min	65%	3.3:1
6b	DTBMP	Trisyl-Clc	60 min	53%	5:1

^aRatio determined by ¹H NMR.

^bReaction maintained at −78 °C.

^cTrisyl-Cl recrystallized from pentanes.

As a model system for this coupling we initially looked at glycosylation of p-methoxyphenol with 4-O-acetate-protected rhodinose donor 3. As expected, the glycosylation afforded the product 20 in moderate yield as a single α-anomer (Table 3, entry 1). A similar result was seen for the coupling between 3 and disaccharide acceptor 16, affording trisaccharide 21 as a single α-anomer (Table 3, entry 2). The necessity of the acetate for the selectivity of the reaction was further confirmed when we attempted to couple disaccharide acceptor 16 with naphthylmethyl protected rhodinose donor 19. In this case we observed an α:β ratio in trisaccharide 22 that was comparable to what we observed in the construction of the tetrasaccharide 2 (Table 3, entries 3 and 4). In hindsight, the role of this protecting group is not surprising, as the 4-O-acyl group has been previously demonstrated to be α-directing with both galactose^58–60 and fucose^61–63 glycosyl donors.

Table 3.

Test Glycosylations with Rhodinose Donors

entry	donor	R	acceptor	product	yield	α:βa
1	3	Ac		20	63%	α-only
2	3	Ac		21	59%	α-only
3	17			2	53%	5:1
4	19			22	46%	5:1b

^aRatio of anomers determined by ¹H NMR.

^bAnomeric ratio determined from isolated yields.

With the tetrasaccharide in hand, we next turned to completion of the pentasaccharide. Fortunately, we found that upon removal of the Nap ether the anomers 18α/β were readily separable (Scheme 4). With the desired α-linked tetrasaccharide acceptor 18α in hand, we turned our attention to the final glycosylation. To this end, a trisyl chloride promoted glycosylation between 4-O-acetate rhodinose donor 3 and tetrasaccharide acceptor 18α afforded pentasaccharide 23 in 66% yield as a single α-isomer. Removal of the benzyl groups using Raney nickel afforded the corresponding diol 24 in 60% yield.^57,64 Finally, removal of the acetate protecting group with potassium carbonate in methanol proceeded in quantitative yield to afford 1. Overall, we obtained the pentasaccharide in 2.5% total yield with 25 total steps and a longest linear sequence of 15 steps.

Scheme 4.

Synthesis and Global Deprotection of the Saquayamycin Z Pentasaccharide

In conclusion, we have completed the first synthesis of the pentasaccharide fragment of saquayamycin Z through a [2 + 2 + 1] glycosylation route. By taking advantage of our previously reported TsCl-mediated glycosylation chemistries, we were able to install the requisite olivose moieties in good yield and selectivity. Furthermore, we found that Trisyl chloride can effectively activate rhodinose donors for stereoselective glycosylation, although the selectivity of the reaction was dependent on the nature of the protecting group on the axial C-4 alcohol. Taken together, these studies are helping us to establish guidelines for the construction of deoxy-sugar oligosaccharides. Ultimately, we anticipate that such guidelines will permit the efficient and routine construction of these molecules.