Rapid de Novo Preparation of 2,6-Dideoxy Sugar Libraries through Gold-Catalyzed Homopropargyl Orthoester Cyclization

Abstract

A flexible de novo route capable of producing libraries of 2,6-dideoxy sugars is described. We have found that Au(JackiePhos)SbF₆MeCN promotes the conversion of homopropargyl orthoesters into functionalized 2,3-dihydro-4H-pyran-4-ones in good to excellent yields (71–90%). These latter compounds can be easily converted into a number of otherwise difficult to access 2,6-dideoxy sugars.

Graphical Abstract

It is estimated that nearly 20% of all natural products possess glycans, many of which contain 2,6-dideoxy sugars. The glycans can play a critical role in modulating natural product bioactivity.¹ In extreme cases, deoxy sugars may comprise a significant portion of a natural product, such as in the antibiotic saccharomicin B (Figure 1).^2-4 Altering the composition of these sugars through glycorandomization can dramatically affect the bioactivity of a natural product, including increasing potency or mitigating toxicity.^5-9 While glycorandomization holds promise as a new tool for drug discovery,^10-12 a major hurdle to its implementation is the lack of availability of deoxy sugars.

Figure 1.

Heptadecasaccharide antibiotic saccharomicin B.

Numerous approaches to the synthesis of deoxy sugars have been developed.^13-15 In many cases, deoxy sugar monosaccharide construction relies on de novo synthesis.^16-38 While there are a myriad of approaches to de novo synthesis, the majority of routes that have been reported are target specific. A more flexible approach relies on carrying out modifications on glycals,^39-48 which permits the generation of multiple different sugars from the same starting material. However, this strategy is cost-effective only when the glycal in question can be generated from readily available sugars, which is often not the case. Furthermore, modification of glycals often involves transforming them into the corresponding 2,3-dihydro-4H-pyran-4-ones, a process that requires harsh oxidations that do not always proceed to completion.^49-59 Thus, a general approach to deoxy sugar library construction remains elusive.

To address this issue, we sought an approach that would allow us to access all four possible diastereomers of suitably functionalized 2,3-dihydro-4H-pyran-4-ones. To this end, we drew inspiration from prior work by the Rhee lab, who demonstrated that homopropargyl acetals could be cyclized into the corresponding tetrahydro-4H-pyran-4-ones using gold(I) catalysts.^60,61 We envisioned that applying this cyclization chemistry to homopropargyl orthoesters 1–3 would provide access to all four possible diastereomers of 2,3-dihydro-4H-pyran-4-one [4–6 (Scheme 1)]. These latter compounds could then be converted into an array of 2,6-dideoxy sugars, such as 7–12.

Scheme 1.

Proposed Synthesis and Derivatization of Enone Precursors into Hard To Access Deoxy Sugars

To test the key gold-catalyzed cyclization, 1,2-anti- and 1,2-syn-substituted homopropargyl orthoesters 1–3 were prepared. The synthesis of the 1,2-anti-substituted homopropargyl orthoester (−)-1 began with protecting methyl (R)-(+)-lactate as a tert-butyldimethylsilyl ether (TBS) (Scheme 2),⁶² followed by reduction using DIBAL-H⁶³ to afford aldehyde (+)-14 in 78% yield over two steps. Alkynylation under conditions described by Marshall afforded (−)-15 in 82% yield as a 16:1 mixture of isomers when (S)-BINOL was used as the chiral ligand.⁶⁴ After separation of the diastereomers, we were pleased to find that benzylation of the newly formed hydroxyl under standard Williamson ether conditions proceeded smoothly with concomitant deprotection of the alkyne to give (−)-16 in 84% yield. Removal of the TBS group using TBAF afforded alcohol (−)-17 in 93% yield. Finally, treating (−)-17 with trimethyl orthoformate and magnesium chloride provided the key orthoester substrate (−)-1 in 86% yield.⁶⁵ To test the general applicability of the proposed cyclization, we also synthesized orthoester (+)-2 possessing the more labile naphthylmethyl (Nap) ether.⁶⁶ Gratifyingly, the synthesis of this substrate proved to be uneventful, and we were able to obtain (+)-2 through a similar sequence using (R)-BINOL to set the stereochemistry of the propargylic alcohol (+)-15. Finally, the 1,2-syn diastereomers (+)-3 and (−)-3 were synthesized, albeit with a lower dr, through a similar sequence (see the Supporting Information).⁶⁷

Scheme 2.

Synthesis of Orthoesters (−)-1, (+)-1, (+)-2, (−)-3, and (+)-3

With the requisite orthoesters in hand, we turned our attention to the key gold-catalyzed cyclization reaction. Our initial attempts at cyclization used AuSbF₆ with the JohnPhos ligand as described by Rhee and co-workers (Table 1, entry 1).⁶⁰ Unfortunately, these conditions resulted in decomposi-tion of the orthoester to afford alcohol (−)-17 as the major product. Reasoning that the loss of the orthoester was due to trace amounts of acid, we conducted the reaction in the presence of 2,4,6-tri-tert-butylpyrimidine (TTBP). Adding the acid scavenger afforded desired ketone (+)-4 in a modest 11% yield. Increasing the catalyst load to 5% did not have a major impact on the reaction (Table 1, entry 3). We next examined electron deficient ligands to determine if they would further improve the yield. While tris(pentafluorophenyl)phosphine⁶¹ failed to improve the yield of the reaction, the use of the JackiePhos ligand⁶⁸ led to the formation of the desired pyranone (+)-4 in excellent yield (Table 1, entry 5). Further optimization of the reaction revealed that the catalyst load could be reduced to 3.5% without any deleterious effect on the reaction. Smaller catalyst loads led to a corresponding decrease in the yield.

Table 1.

Optimization of Gold Cyclization

entry	anion	additive	ligands	catalyst (mol %)	Yield (%)a
1	SbF6	none	JohnPhos, MeCN	3.0	0
2	SbF6	TTBP	JohnPhos, MeCN	3.0	11
3	SbF6	TTBP	JohnPhos, MeCN	5.0	6
4	Cl	DTBP, 4 Å MS	(C6F5)3P	7.0	5
5	SbF6	DTBP, 4 Å MS	JackiePhos, MeCN	7.0	86
6	SbF6	DTBP, 4 Å MS	JackiePhos, MeCN	3.5	87

^aBased on isolated weight.

Having established the optimal conditions for cyclization, we turned our attention to examining the scope of the reaction. We found that the 1,2-anti orthoesters (−)-1, (+)-1, and (+)-2 consistently gave excellent yields of the desired ulose (Table 2, entries 1–3, respectively). On the other hand, 1,2-syn orthoesters (−)-3 and (+)-3 reacted to afford the desired products in slightly attenuated yields (Table 2, entries 4 and 5, respectively). The lower yields with the syn isomers may be due to an unfavorable steric interaction between the axial benzyl ether and the substituents on C-2 of the pyranose ring initially formed in the reaction (Figure 2).^60,69 Importantly, the gold-catalyzed cyclizations could be run on multigram scale, providing access to significant quantities of all of the 2,3-dihydro-4H-pyran-4-one diastereomers needed for deoxy sugar construction.

Table 2.

Cyclization of Orthoester Substrates into Glycals

Entry	Substrate	Mol % Cat.	Product	Yield (%)a
1		3.5		87
2		3.5		86
3		3.5		90
4		3.5		71
5		3.5		74

^aBased on isolated weight.

Figure 2.

Possible steric interaction arising during cyclization of syn-homopropargyl orthoesters.

We next turned our attention to transforming the cyclization products into various d- or l-deoxy sugars. Our initial targets were l-digitoxose and l-olivose, which could be readily obtained from (−)-5 (Scheme 3). To this end, treating (−)-5 with cesium carbonate and ethanethiol afforded ketone (−)-22 as a 12.5:1 mixture of isomers.⁷⁰ Reduction of (−)-22 using sodium borohydride stereoselectively generated l-digitoxose (−)-7 in 80% yield.⁷¹ To access l-olivose (−)-8, the enone was selectively reduced under Luche conditions to provide allylic alcohol (−)-23.⁷² Acetate protection and hydration of the glycal with triphenylphosphine hydrogen bromide afforded l-olivose hemiacetal (−)-8 in 88% yield over two steps.

Scheme 3.

Derivatization of Enone (−)-5 into l-Digitoxose 7 and l-Olivose 8

A similar approach was used to obtain d-boivinose (+)-9 and d-oliose (+)-10 (Scheme 4). Treating enone (+)-6 with adamantanethiol, cesium carbonate, and triethyl amine cleanly afforded ketone (+)-25 as a single anomer.⁷⁰ The ketone was then reduced using sodium borohydride to afford d-boivinose 9.⁷³ To prepare d-oliose (+)-10, enone (+)-6 was subjected to a Luche reduction to stereoselectively generate alcohol (−)-26.⁴⁵ Acetate protection followed by hydration afforded d-oliose (+)-10 in 78% yield over two steps.

Scheme 4.

Derivatization of Enone (+)-6 into D-Boivinose 9 and d-Oliose 10

Having established scalable routes to the four diastereomeric 2,6-dideoxy sugars, we turned our attention to deoxy amino sugar synthesis. To install amino functionality, 1,4-conjugate addition of thiophenol to enone (−)-5 was followed by addition of O-benzylhydroxylamine to generate oxime (−)-28 (Scheme 5).^39,74,75 Refluxing (−)-28 in THF in the presence of 3 equiv of lithium aluminum hydride reduced the oxime to the corresponding amine to afford l-ristosamine (−)-11.^76,31

Scheme 5.

Derivatization of Enone (−)-5 into l-Ristosamine 11 and l-Saccharosamine 12

To synthesize l-saccharosamine (−)-12, we first unmasked the C-4 hydroxyl on (−)-5 because the Nap group can interfere with installation of the C-3 methyl group (Scheme 5).^39,77 The resultant enone was subjected to a 1,4-conjugate addition/oxime formation sequence to afford (−)-29 in 76% yield over three steps. Treatment of the oxime with methyl cerium under conditions previously described by Scharf afforded l-saccharosamine (−)-12 as a single isomer in 64% yield.³⁹

In summary, we have developed a flexible de novo route for preparing an array of difficult to access 2,6-dideoxy sugars. This process provides access to all four diastereomers of protected 2,3-dihydro-4H-pyran-4-ones starting from inexpensive d- or l-methyl lactate. The resulting pyranones can be readily transformed into a variety of orthogonally protected 2,6-dideoxy- and 2,3,6-trideoxy-3-amino sugars. We anticipate this approach will both greatly accelerate deoxy sugar oligosaccharide synthesis and be an invaluable tool for glycodiversification studies of bioactive natural products.