De Novo Asymmetric Achmatowicz Approach to Oligosaccharide Natural Products

Abstract

The development and application of the asymmetric synthesis of oligosaccharides from achiral starting materials is reviewed. This de novo asymmetric approach centers around the use of asymmetric catalysis for the synthesis of optically pure furan alcohols in conjunction with Achmatowicz oxidative rearrangement for the synthesis of various pyranones. In addition, the use of a diastereoselective palladium-catalyzed glycosylation and subsequent diastereoselective post-glycosylation transformation was used for the synthesis of oligosaccharides. The application of this approach to oligosaccharide synthesis is discussed.

Graphical Abstract

1.0. Introduction

Of the various classes of natural products, the carbohydrate containing ones have seen the least attention from the synthetic and medicinal chemistry communities.¹ This lack of interest is primarily due to a lack of synthetic ability rather than a lack in significant biological activities. In fact, the sugar portions of carbohydrate containing natural products play a crucial role in both the mechanisms of action as well as bioavailability (e.g., target binding, solubility, tissue targeting, and membrane transport). For example, the corresponding aglycons of natural products are often devoid of biological activity.² Medicinal chemists have long desired synthetic methods to vary the carbohydrate structures of natural products,³ which has inspired many de novo asymmetric approaches to monosaccharides.^4,5 This synthetic need is particularly apparent when it comes to the synthesis and medicinal chemistry studies of oligosaccharide containing natural structural motifs.^1,6,7

In this context, we developed a de novo asymmetric synthetic approach to carbohydrates,⁸ which relies upon asymmetric catalysis to set the D-/L-stereochemistry via the synthesis of a furan alcohol and an Achmatowicz rearrangement/carbonate formation to set the α-/β-anomeric stereochemistry.⁹ Finally, a Pd-catalyzed glycosylation^10,11,12 is used to stereospecifically transfer the pyranone to various glycosyl acceptors (e.g., N-, O-, C-nucleophiles).^13,14,15 The reaction occurs rapidly and in high yields for all four stereoisomers, with the enone acting as an anomeric directing group (via a Pd-π-allyl).¹⁶ This approach mechanistically stands in contrast to the use of silver and Au, which appears to occur via a Lewis acid glycosylation mechanism.¹⁷ From a stereochemical perspective this approach to hexose sugars is the simplest of all the possibilities. That is to say that it reduces the hexo-pyranoside down from 32 possible stereoisomers to four. This, in theory, should provide access to all the possible hexose stereoisomers. In practice however, we have found ways to make many but not all possible stereoisomers.¹⁸ Herein, we review our successful effort to use this approach for the synthesis of many carbohydrate-based natural products with medicinally relevant biological activity. Key to this approach is the use of Boc-pyranone as Pd-glycosyl donors and in turn oligosaccharide building blocks. The feasibility of this de novo asymmetric approach to carbohydrates is evident by its ability to engender medicinal chemistry studies (vide infra).¹⁹

1.1. Background

In this regard, we have been developing new asymmetric sequences for the de novo asymmetric synthesis of carbohydrates.⁷ While we and others have developed many enantioselective approaches to monosaccharides from achiral starting materials,^1,3,18 it is fair to say that our Achmatowicz approach is the only approach that uses asymmetric catalysis to control the installation of stereochemistry at both the monosaccharide and oligosaccharide level of complexity.²⁰ Herein we review our development and use of the Achmatowicz approach to carbohydrates natural products with oligosaccharide motifs (vide infra).

2.0. De novo Asymmetric Approaches to Chiral Furan Alcohols

Key to the success of this Achmatowicz approach is the ease at which furan alcohols can be prepared in enantiomerically pure form from achiral furans (e.g., 1 and 2).²¹ While we have explored many asymmetric approaches to furan alcohols, two preferred methods are the Noyori hydrogen transfer reduction of acylfurans (1 to 3) and the Sharpless dihydroxylation of vinylfurans (2 to 3) (Scheme 1).^6,22 Application of the Sharpless asymmetric aminohydroxylation (AA)²³ of vinylfurans provides access to azasugars²⁴ via the aza-Achmatowicz reaction.^25,26 Both routes are quite practical in terms of scalability (> 100 g) and catalyst loading (< 1 % catalyst). The Sharpless route is somewhat limited to the synthesis of hexoses with a C-6 hydroxy group (e.g., 3e, R = CH₂OTBS), whereas the Noyori protocol distinguishes itself in its environmentally benign reductant (e.g., i-PrOH, HCO₂H)²⁷ and flexibility to virtually any substitution at the C-6 position (e.g., 3a-e).²⁸ In addition, the acylfuran (1) starting materials are either commercially available or readily available from furan and the corresponding carboxylic acids.

Scheme 1:

De novo Asymmetric Synthesis of Chiral Furan Alcohols

2.1. Achmatowicz Approaches to Hexoses

Key to the overall Achmatowicz approach to carbohydrates is the recognition that furan alcohols like 3(S) can be converted into pyrans like 4(α-L) and 4(β-L) with L-sugar stereochemistry (Scheme 2). Analogously furan alcohols with (R)-stereochemistry can be converted into D-sugars. The enantio-divergent approach to furan alcohols of type 3(S)/3(R) enables a highly stereoselective approach to all the possible pyranone stereoisomers 4(α-L), 4(β-L), 4(α-D), and 4(β-D) by means of an Achmatowicz rearrangement and diastereoselective carbamate formation. The Achmatowicz rearrangement is an oxidative hydration/rearrangement of furfuryl alcohols into 2-substituted 6-hydroxy-2H-pyran-3(6H)-ones. When the pyranone 6-hydroxy group is converted to an acetate or carbonate type Pd-π-allyl leaving group, it becomes the basic pyranone building block for our de novo carbohydrate synthesis. The t-butyl carbonate formation can selectively give the α-pyranones 4(α-L) and 4(α-D) at –78 °C. At higher temperatures (e.g., rt), a 1:1 ratio of the α- and β-Boc protected enones was produced. This procedure is easily scaled for the production of multigram quantities of both α- and β-pyranones in either D- or L-enantiomeric form. The resulting sugar products can exist with variable substitution and stereochemistry. The pyranone subunits 4 can be viewed as a protected form of various monosaccharides or a component to various oligosaccharides (vide infra), via highly stereoselective Pd-glycosylation and post glycosylation transformations.

Scheme 2:

Pyranones Form Furan Alcohols via Achmatowicz reaction.

2.2. Pd-Catalyzed Glycosylation

The de novo Achmatowicz approach to hexoses has great potential for preparing various D- and L-sugars because the starting 6-t-butoxycarboxy-2H-pyran-3(6H)-ones (4) can easily be prepared from optically pure furfuryl alcohols 3 (either (R) or (S) enantiomers)²⁹ by a one or 2-step procedures (Scheme 2). Key to the assembly of these building blocks is the use of a Pd-glycosylation reaction.^15,30 This highly diastereoselective process occurs with retention of configuration converting pyranones 4(α-L), 4(β-L), 4(α-D), and 4(β-D) with a C-1 BocO-leaving group into C-1 O-glycosides 5(α-L), 5(β-L), 5(α-D), and 5(β-D), with an anomeric alkoxy-group. This approach has also been extended to a Pd-cyclitolization (5a-carbasugar glycosylation).³¹

The Pd-reaction occurs via a double inversion/net retention mechanism (i.e., 4(α-L) to 5(α-L) via 6(β-L) and i.e., 4(β-L) to 5(β-L) via 6(α-L); Scheme 3). The Pd-π-allyl reaction occurs rapidly and in high yields for both the α- and β-diastereomers (e.g., 4(α-L/D) to 5(α-L/D) systems and works best when Pd₂(dba)₃·CHCl₃ is used as the Pd(0) source with triphenylphosphine as the ligand in a 1:2 Pd/PPh₃ ratio. Various carboxylate leaving groups (e.g., AcO, BzO, PivO) work in the Pd-glycosylation reaction, however, the t-butyl carbonate group (BocO) performs the best. The BocO-group is the best leaving group and serves as the base to form t-BuOH, after decarboxylation. The t-BuOH that is formed is a poor nucleophile in the Pd-glycosylation reaction, due to its steric hinderance. When performed in conjunction with the Achmatowicz reaction and diastereoselective carbonate formation, the Pd-glycosylation enables a 3-step stereo-divergent pyranone synthesis from chiral furan alcohols. Key to the success of the de novo asymmetric Achmatowicz approach is the practical access to all four possible pyranone diastereomers 5(α-L), 5(β-L), 5(α-D), and 5(β-D) from either furan alcohol enantiomers 3(R) or 3(S).

Scheme 3:

The Mechanism of the Pd-Catalyzed Glycosylation

The resulting approach to sugars has become a viable de novo asymmetric alternative to traditional carbohydrate routes. While the comparison of these de novo routes to traditional carbohydrate routes is difficult, these de novo routes have a clear advantage in terms of D-/L-sugar variability and for rare sugars.³² In addition, when these de novo routes are applied to oligosaccharide targets, they often are executed with the use of fewer protecting groups and sometimes no protecting groups. In our view, the best metric for this evaluation is in terms of synthetic utility and variability, which is highlighted in the following synthetic endeavors (Schemes 4–24).¹³ The Pd-glycosylation reaction, which at first glance does not look like a typical glycosylation reaction, has the ability to use the enone functionality as precursors to mono-, di- and triol products. With various post-glycosylation reactions, we are able to transform the enones into the functional equivalent of variously substituted alcohol/polyol and hence function as atom-less protecting groups (vide infra).

Scheme 4:

Synthesis of Digitoxin, Mono-, Di- and Tri-Saccharides

3.0. De Novo Synthesis of Oligosaccharides

This Achmatowicz-based de novo approach to carbohydrates has had success in its application to various monosaccharide targets,³³ however, its full potential is best revealed in its application to oligosaccharide targets.³⁴ This potential can be measured in terms of its overall synthetic efficiency (i.e., number of steps, the minimal use of protecting groups) and complexity of the targets achieved. The complexity can be measured by the range and variability of the oligosaccharide motifs that can be prepared, as well as the biological and medicinal chemistry structure-activity relationship studies they enable. Of the many ways to compare these de novo routes, clearly the best metric is the scope of the synthetic and biological applications. Herein we demonstrate the potential of these de novo asymmetric approaches by reviewing their application to the synthesis of various mono-, di-, tri- tetra-, and heptasaccharide motifs.

3.1. Synthesis of Digitoxin

The first significant application of the Achmatowicz approach to oligosaccharides was its application toward the synthesis of the cardiac glycosides (Scheme 4). The cardiac glycosides, like digitoxin, emanates from a long folk medicine tradition (~1500 BC). The cardiac glycosides are produced in toxic plants (e.g., Digitalis purpurea, Digitalis lanata) and found in amphibians that feed on them. In addition to its cardiotonic effect, the cardiac glycosides have potential as anticancer and antiviral agents.³⁵ Studies have found that the carbohydrate portion of the cardiac glycosides controls the selectivity for cancer cells/infected cells over the cardiotonic effects.^36,37

The structure-activity relationship (SAR)-studies of the cardiac glycosides have been limited by the number of naturally occurring cardiac glycosides that are available (e.g., digitoxin, digoxin, and oleandrin). To address these concerns, we have developed a carbohydrate medicinal chemistry SAR-method that allows us to map the carbohydrate oligosaccharide space around the cardiac glycoside structural motifs.³⁸ These efforts began with the synthesis of natural product digitoxin 14 from its aglycon digitoxigenin (DigOH), which also involved a 4-step synthesis of the digitoxin monosaccharide 10. In addition, this led to a 10-step synthesis of digitoxin disaccharide 12b and ultimately a 15-step synthesis of digitoxin 14.

The synthesis began with a Pd-catalyzed glycosylation between digitoxigenin (DigOH) and the β-D-glycosyl donor 4a to afford the β-D-pyranone 7. While not too surprising, it was gratifying to see the selective glycosylation of the axial secondary alcohol in the steroid A-ring over the bridgehead tertiary alcohol. An unselective sodium borohydride 1,2-reduction of the ketone in 7,³⁹ afforded a mixture of allylic alcohols 8. Fortunately, the lack of stereoselectivity did not hurt the throughput, as both diastereomers could be used in the subsequent Myers reductive rearrangement of 8. Thus, exposure of both diastereomers of 8 to the Mitsunobu conditions with NBSH (o-NO₂PhSO₂NHNH₂), cleanly provides alkene 9 as a single regio- and diastereo-isomer. Finally, an Upjohn dihydroxylation⁴⁰ of alkene 9 (OsO₄/NMO) exclusively gave the digitoxin monosaccharide 10 in good yield and as a single diastereomer. In addition to providing access to the desired monosaccharide 10, the synthesis demonstrated the compatibility of the Pd-glycosylation and post-glycosylation reaction sequence with the butenolide functional group.

The synthesis of the digitoxin disaccharide 12b utilized a similar sequence after regioselective protection. To accomplish this, we turned to orthoester chemistry.^41,42 Specifically, the regioselective acylation of the syn-3,4-diol in 10 occurred via a one-pot orthoester ((MeO)₃CCH₃) formation and acid-catalyzed hydrolysis (TsOH/H₂O) to afford axial acetate 11. Then the Pd-glycosylation of equatorial alcohol in 11 with 4a began a 4-step synthesis to give disaccharide 12a. A LiOH hydrolysis of the axial alcohol in 12a gave the deprotected disaccharide 12b. Once again, the same Pd-glycosylation and post-glycosylation sequence can be used for the synthesis of digitoxin 14. Thus, a regioselective acylation of 12a gave the diacetate 13, which via a familiar 5-step sequence was converted it into digitoxin 14.

3.1.1. Synthesis of Analogs

The synthetic approach used for the synthesis of digitoxin was also used in the synthesis of its neo-glycoside analogues (Scheme 5).⁴³ Additionally, this approach was used to prepare the three reducing sugars with a free alcohol at the anomeric position (digitoxin monosaccharide 20, disaccharide 21, and trisaccharide 22), which were used in the synthesis of the corresponding neo-glycosides 24a-c.⁴⁴ The synthesis began with the β-glycosylation of benzyl alcohol, followed by an unselective reduction and Myers reductive rearrangement to give alkene 15. A diastereoselective dihydroxylation of 15 gave the digitoxin β-monosaccharide 16 with a benzyl anomeric protecting group. As before, orthoester chemistry was used to regioselectively acylate the C-4 axial alcohol in 16 ((MeO)₃CCH₃, then TsOH/H₂O) to afford the protected monosaccharide 17. Simply repeating the 5-step Pd-glycosylation, Luche/Myers reduction/dihydroxylation and orthoester acylation converted 17 into the disaccharide 18a with both sugars with a C-3 axial acetate. The two acetates could easily be removed to form 18b with aqueous LiOH.

Scheme 5:

Mono-, Di-, and Tri-Digitoxin Neo-Glycoside Syntheses

Once again, exposing 18a was transformed via a similar 4-step sequence (Pd-glycosylation, Luche/Myers reduction, and dihydroxylation) followed by a LiOH ester hydrolysis to give the trisaccharide 19 with a benzyl group at the anomeric position. The two acetates could easily be removed to form 18b with aqueous LiOH. A Pd-C catalyzed hydrogenolysis was used to cleanly convert Bn-protected monosaccharide 16, disaccharide 18b, and trisaccharide 19 into monosaccharide 20, disaccharide 21, and trisaccharide 22 with minimal purification to remove the catalyst. Then the three reducing sugars monosaccharide 20, disaccharide 21 and trisaccharide 22 were converted into the corresponding neoglycoside by exposure to digitoxigenin with a methoxyamine in the A-ring by using the neoglycosylation condition developed by Thorson (AcOH in MeOH/CHCl₃). The three O-glycosides (10, 12b, and 14) and corresponding neoglycosides (24a, 24b, and 24c) were evaluated for their anticancer activity. These studies revealed two trends. The first being that the monosaccharides (10 and 24a) were more active than the disaccharides (12b and 24b), which were more active than the trisaccharides (14 and 24c). The second trend was that the three O-glycosides (20, 21, and 22) were more active than the corresponding neoglycosides (24a-c).^44,45

3.2. Synthesis of a PI-080 Trisaccharide

In addition to symmetrical linear trisaccharides like the digitoxins, we have also applied this de novo approach to the synthesis of trisaccharide with different sugars making up the oligomer. An example of this can be seen in our de novo asymmetric approach to the trisaccharide of PI-080 (32) (Scheme 6), which consists of a β-D-olivose, α-L-rhodinose, and α-L-aculose.⁴⁶ As with the digitoxin this began with the β-glycosylation of PMBOH with 4a(β-D) to stereo-specifically give β-D-pyranone 25. An unselective reduction of 25(β-D) gave a mixture of allylic alcohols which after a Myers reductive rearrangement gave alkene 26. Again, a highly diastereoselective dihydroxylation converted 26 into monosaccharide 27 with digitoxose stereochemistry. Then the digitoxose monosaccharide 27 was selectively converted in to a protected olivose sugar 28 by a regioselective and stereospecific Mitsunobu-like inversion (PPh₃/DEAD/p-NO₂BzOH). It worth noting that this combination of diastereoselective dihydroxylation and regioselective inversion of the axial alcohol results in an excellent solution to the problem of 1,2-trans-diequatorial addition of a cyclohexene (26 to 28). Then the C-4 alcohol in 28 was exposed to an α-glycosylation with 4a(α-L) to give the olivose/aculose disaccharide 29. A selective Luche reduction and Mitsunobu inversion were used to convert the aculose ring in disaccharide 29 to disaccharide 30 with one rhodinose sugar. Then, disaccharide 30 was exposed to LiOH to hydrolysis of the two nitrobenzoates and diimide to reduce the alkenes to give the olivose/rhodinose disaccharide 31. Finally, our Pd-catalyzed glycosylation conditions were used to regioselectively and stereospecifically glycosylate the diol in 31 with 4a(α-L) to give the PI-080 trisaccharide 32. The trisaccharide 32 displayed significant growth inhibition (GI₅₀ from 0.1 to 11 mM) and cytotoxicity (LC₅₀ from 5.1 to 100 mM) against a range of cancer cell lines.⁴⁷

Scheme 6:

Synthesis of the PI-080 Trisaccharide

3.3. Landomycin E Trisaccharide Synthesis

A similar approach as the one used for the PI-080 trisaccharide was used for a de novo asymmetric synthesis of the trisaccharide portion of Landomycin E 40b (Scheme 7).^48,49 The Landomycin E trisaccharide 40a consists of two β-olivose-sugars terminated by a α-rhodinose. As a result, the route used similar glycosylation and post-glycosylation chemistry as was used for the digitoxins and PI-080 (Schemes 4–6). The synthesis began with a β-glycosylation of benzyl alcohol with 4a(β-D) to give β-aculose monosaccharide 33. As above a three-step Luche/Myers reduction and dihydoxylation sequence was used to convert 33 into digitoxose 34. Another regioselective inversion of the C-3 axial alcohol (PPh₃/DEAD/p-NO₂BzOH) was used to convert the digitoxose stereochemistry in 34 into a protected olivose sugar 35. A β-Pd-glycosylation with 4a(β-D) of the C-4 alcohol in 35 followed by nitrobenzoate hydrolysis and TBS-protection was used to convert it into olivose/aculose disaccharide 36. Once again, the enone functionality in 36 was converted into a digitoxose sugar by a 3-step sequence (Luche/Myers reduction and dihydroxylation) to give the olivose/digitoxose disaccharide 37. The C-3/4 diol of the digitoxose sugar was regioselectively converted into an olivose sugar by a three-step Mitsunobu/TBS-protection and nitrobenzoate hydrolysis to give the bis-olivose disaccharide 38. A Pd-catalyzed α-glycosylation of the C-3 alcohol in 38 with 4a(β-D) followed by DibalH reduction was used to give trisaccharide 39. Finally, a TBS protection and alkene hydrogenation gave a protected form of the landomycin E trisaccharide 40a.

Scheme 7:

Synthesis of the Landomycin Trisaccharide

3.4. Synthesis of Cleistetroside

The power of the de novo Achmatowicz approach to natural occurring rare sugar oligosaccharides was demonstrated with the synthesis of digitoxin, PI-080⁴⁷ and Landomycin E trisaccharide.⁴⁸ We next look to expand the potential of this approach with its application to oligosaccharides that consisted of less rare sugars and the rhamnan set of oligosaccharides seemed ideal for this purpose. As will be revealed (cf., Schemes 8–19), the application of this approach to sugars with α-rhamno/manno-stereochemistry can be particularly formidable. Our first synthetic foray into this class of oligosaccharides was the synthesis of the partially acylated cleistrioside and cleistetroside class of natural product oligosaccharides.⁵⁰ This effort led to the successful synthesis of 2 members of the cleistriosides and 9 members of the cleistetrosides^51,52 For space reasons we have chosen to limit this discussion to the synthesis of cleistetroside-2 (Scheme 8). Key to the practicality of this approach will be its reliance on the minimal use of protecting groups (i.e., chloroacetate and acetonide).

Scheme 8:

Synthesis of the Cleistetrosides

Scheme 19:

Synthesis of the Mezzettiasides 5–7

The route to cleistetroside-2 began with a Pd-glycosylation between dodecanol and pyranone 4a(α-L). After a three-step sequence (Luche reduction, dihydroxylation, and acetonide protection), the protected monosaccharide 41 was obtained. A subsequent Pd-glycosylation of 41 with pyranone 4a(α-L) yielded the aculose/rhamnose disaccharide 42, with the C-2/3 diol of the rhamno-sugar protected as an acetonide and the C-2–4 positions of the enone in the aculose sugar serving as an atomlessly protected rhamno-sugar. Thus in only three steps (Luche reduction, acylation, and Upjohn dihydroxylation), the enone in 42 was readily converted into 43 with rhamno-sugar stereochemistry and the requisite C-4 acetate group already installed.

Next, we need to selectively protect the C-2 position and glycosylate the C-3 position of 43. The use of the chloroacetate equivalent of the previously demonstrated very successful ortho-ester chemistry did not work in this situation. Thus, we turned to the novel use of tin acetal chemistry to regioselectively direct a C-3 glycosylation. This was accomplished by the use of dibutyltin oxide to perform a C-2/3 tin acetal intermediate. The tin intermediate underwent a regioselective Pd-glycosylation reaction with pyranone 4a(α-L) to give the C-3 glycosylated product (C-3 to C-2, 7:1), which after chloroacylation gave trisaccharide 44. A 4-step post-glycosylation transformation protocol (Luche reduction, acylation, Upjohn dihydroxylation and orthoester acylation) was used to transform the aculose ring in 44 into 45 a rhamno-sugar with a C-2/4 diacetate selectively installed. Finally, a 6-step sequence was used to convert 45 into 46. This involved a glycosylation (Pd(0), 4a(α-L)), a 3-step post-glycosylation (Luche reduction, acylation, and Upjohn dihydroxylation) and a 2-step chloroacetate/acetonide deprotection (thiourea then H₃O⁺). The synthetic cleistetroside-2 (46) obtained by this route and its related cleistrioside and cleistetroside oligosaccharides were used in anticancer studies, which demonstrated the synthetic viability of these de novo asymmetric approaches for medicinal chemistry studies.

3.5. Synthesis of Anthrax Tetrassacharide

This asymmetric Achmatowicz approach was also applied to a more complex member of the rhamnan oligosaccharides. More specifically, a unique tetrasaccharide that was isolated from an exosporium glycoprotein (BC1A) from the cell wall of Bacillus anthracis spores.⁵³ Bacillus anthracis is one of the most well-known members of the Bacillaceae family. When this Gram-positive spore-forming bacterium is inhaled, it causes the fatal infectious disease called anthrax in humans and other mammals. The anthrax tetrasaccharide 66 (Scheme 11) consists of three L-rhamnose sugars and is terminated by a unique D-sugar, called anthrose. Importantly, the anthrose sugar is unique to Bacillus anthracis and not found in spores from other Bacillus species.⁵⁴

Scheme 11:

Synthesis of the Antrax Tetrassacharide (end game)

Our synthesis of the anthrax tetrasaccharide 66 began with the assembly of its tri-rhamnose rhamnan unit in the protected tris-L-rhamno-trisaccharide 54 (Scheme 9).^55,56 This began with a Pd-glycosylation between benzyl alcohol and pyranone 4a(α-L) which was followed by Luche reduction, dihydroxylation, and Ley spiro-ketal protection⁵⁷ to give the L-rhamno-sugar 47 with a Bn-protecting group at the anomeric position and a free C-2 alcohol. The use of the Ley spiroketal protecting group was critical as it allowed the oligosaccharides to be built with connectivity to the C-2 position of the sugar. The C-2 alcohol in 47 was then glycosylated with pyranone 4a(α-L), which provided the disaccharide 48. A diastereoselective Luche reduction and Upjohn dihydroxylation produced the rhamno-triol 49. Then a 3-step one pot C-2/C-4 bis-acylation of triol 49 was accomplished (CH₃C(OMe)₃ p-TsOH_(cat) then Ac₂O then AcOH/H₂O) to afford the 2,4-diacetate 50.^55,58 Another rhamno-sugar was introduced on the C-3 alcohol in 50 by a 3-step (Pd-glycosylation (50 + 4a(α-L), Luche reduction and dihydroxylation) to form trisaccharide 51. Once again, the 3-step orthoester bis-acylation of the C-2/C-4 alcohols of triol 51 was used to form monoalcohol 52. Unfortunately, we found the Ley spiro-ketal protecting group in 52 did not survive glycosylation with glycosyl-donors like trichloroacetimidate 55 (Scheme 10). To skirt this problem, we decided to protect the C-3 alcohol as a levulinate ester, and then the spiroketal was removed with TFA to give diol 53. The diol in 53 was then protected as a bis-acetate (Ac₂O/Py) and the levulinate was removed with hydrazine to give trisaccharide 54. Now without the acid-sensitive Ley-protecting group, we found the C-3 alcohol in 54 was able to react under the Schmidt type glycosylation condition (e.g., with 55 to form 56).

Scheme 9:

Synthesis of the Antrax Tetrassacharide

Scheme 10:

Synthesis of the Anthrose

With the synthesis of the rhamno-trisaccharide 54 accomplished and before it was converted into tetrasaccharide 56, we needed to develop an Achmatowicz approach to the anthrose portion of the tetrasaccharide (Scheme 10). This began with the Pd-glycosylation between PMBOH and pyranone 4a(α-D) to give α-D-aculose 25 with the anomeric position protected as a PMB group. We next look to install a C-4 azide. This was accomplished with a Luche reduction (NaBH₄/CeCl₃) to give allylic alcohol 57. The allylic alcohol was then converted into methyl carbonate 58 (ClCO₂CH₃/DMAP).⁵⁹

Using Pd(0) chemistry, the methyl carbonate group of 58 was regio- and stereo-specifically replaced with an azide group (TMSN₃, (allylPdCl)₂/dppb) to afford allylic azide 59 (93%).⁶⁰ An Upjohn dihydroxylation of 59 (OsO₄/NMO) was used to install the rhamno-stereochemistry in 61. Once again dibutyltin acetal chemistry (BnBr/Bu₂SnO) was used to regioselectively protect the diol to form benzyl ether 61.⁴¹ Then, the C-2 axial alcohol in 61 was inverted by a triflation (Tf₂O) and S_N2 displacement/hydrolysis (NaNO₂) to give 62 with gluco-stereochemistry.⁶¹ The C-2 alcohol in 62 was converted into a levulinate (LevOH/DCC/DMAP) and the PMB-group was removed to provide the anomeric alcohol 63. Then, the anomeric alcohol was converted into trichloroimidate 55. The final glycosylation step was accomplished with rhamno-trisaccharide 54 and trichloroimidate 55 and was catalyzed with TMSOTf to give the tetrasaccharide 56 (Scheme 9).

Finally, tetrasaccharide 56 was converted into the anthrax tetrasaccharide 66 in five steps (Scheme 11). This began with the selective levulinate ester hydrolysis (H₂NNH₃OAc), and a silver (I) oxide promoted methylation (Ag₂O in MeI) of the anthrose pyranose in 56 to give 64. Then a one-pot global deprotection of the acetates in 64 along with azide reduction afforded a primary amine (PEt₃, LiOH_(aq)), which was selectively coupled with 3-hydroxy-3-methylbutanoic acid (HBTU/Et₃N, to give amide 65. The natural product synthesis was completed by a hydrogenolysis of both benzyl groups (H₂, Pd/C) to give anthrax tetrasaccharide 66.

3.6. (1→4),(1→6)-branched oligosaccharides

The asymmetric Achmatowicz approach to oligosaccharides has also been used for the synthesis of unnatural oligosaccharides. This has been successfully applied for the synthesis of 1,4- and 1,6-linked trisaccharides with rhamnose and amecitose stereochemistry. Outlined in Scheme 12 is the use of this approach to 1,4-/1,6-branched oligosaccharides.⁶² This began with the synthesis and bidirectional glycosylation of C-4 and C-6 diol 67. The synthesis of diol 67 was accomplished from Boc pyranone 4e(α-D) in 3 steps by a Pd-glycosylation of BnOH, Luche reduction, and TBS-deprotection. A Pd-catalyzed bis-glycosylation of diol 67 with two equivalents of pyranone 4e(α-D) followed by diastereoselective bis-1,2-reduction and bis-TBS-protection gave the trisaccharide 68. A tetra-Pd-glycosylation of 68 with excess pyranone 4e(α-D) gave heptasaccharide 69. Exposure of 69 to excess NaBH₄ reduced the four enone groups to give a tetra-allylic alcohol, which was per-dihydroxylation (OsO₄/NMO) to give 70. The all α-manno-hepta-saccharide 70 was prepared in only 9 steps from pyranone 4e(α-D). The route is equally amenable to the all-L-enantiomer as well as the various D-/L-stereoisomers and deoxy congeners. It is important to note the acid-sensitivity of the glycosidic bonds in deoxy hepta-saccharides 69 and 70, which might not survive the strongly acidic conditions of a traditional glycosylation.

Scheme 12:

De Novo Synthesis of Branched Oligosaccharides

3.7. Synthesis of Merremoside D

Our success with the glycosylated lipid oligosaccharides, like the cleistriosides and cleistetrosides, emboldened us to apply the asymmetric Achmatowicz approach to the resin glycoside class of macrolactone oligosaccharides.^63,64,65 We identified merremoside D as the ideal test case for our development of a synthetic approach. The merremoside D (92) is a member of the family of resin glycoside natural products, isolated by Kitagawa from the tuber of Merremia mammosa (Lour.) Hall. f. (Convolvulaceae).⁶⁶ The Indonesian plant, Merremia mammosa, has been used in traditional medicine for an array of illnesses. This structurally complex macrolactone/oligosaccharide consist of a bis-rhamnose disaccharide bridged by jalapinolic acid containing lactone at the C-1 and C-3’ position. The amphiphilic nature of the resin glycosides has been suggested to be the source of its ionophoretic activity (i.e., membrane transporter).⁶⁷

The convergent approach began with the synthesis of two disaccharide fragments, a disaccharide donor 77 and macrolactone/disaccharide acceptor 89 (Schemes 13 and 14). The synthesis of the donor disaccharide 77 began with the Pd-glycosylation between BnOH and pyranone 4a(α-L) and Luche reduction to give allylic alcohol 71. The allylic alcohol was dihydroxylated using the Upjohn procedure and acetonide protected to form the rhamno-pyranosides 72 with a free C-4 alcohol.^68,69 The alcohol in 72 was glycosylated with pyranone 4a(α-L) and deprotected (TFA) to give disaccharide 73. Using tin-acetal chemistry, the C-3 equatorial alcohol in 73 was acylated to give 74 with a C-3 i-butyrate ester (iba).⁷⁰ The remaining C-2 alcohol was chloroacylated and the enone was reduced and the resulting C-4 allylic alcohol was acylated to give disaccharide 75. A subsequent Upjohn dihydroxylation, and acetonide protection gave 76. Finally, the C-1 position was deprotected, and the anomeric alcohol was converted into a Schmidt trichloroacetimidate 77.⁷¹

Scheme 13:

Synthesis of the Merremoside D Disaccharide

Scheme 14:

Synthesis of the Merromoside D Lactone

The approach to the macrolactone fragment 89 began with the synthesis of the methyl jalapinolate 81, which is the aglycon portion of the macrolactone (Scheme 14). This de novo asymmetric aglycon synthesis began with the Noyori asymmetric reduction of achiral ynone 78 to give propargyl alcohol 79. An alkyne-zipper isomerization and TBS-protection were used to convert 79 into terminal alkyne 80. An oxidative alkyne cleavage and Fisher esterification with concomitant TBS-deprotection were used to prepare the methyl jalapinolate 81. A Pd-catalyzed glycosylation between 81 and pyranone 4a(α-L) gave monosaccharide 82. A Luche reduction of 82 gave allylic alcohol 83, followed by an Upjohn dihydroxylation to form rhamno-sugar 84. An acetonation of the C-2/3 diol of 84 gave 85 with a free C-4 alcohol, which was Pd-glycosylated with pyranone 4a(α-L) to give monosaccharide 86 after a Luche reduction. A protection of the C-4 alcohol in 84 followed by Upjohn dihydroxylation and ester hydrolysis gave seco-lactone 87. A Corey-Nicolaou macrolactonization of 87 gave a mixture of regioisomeric macrolactones 88a and 88b (1:4),⁷² which was unfortunately biased toward the wrong regioisomer 88b. It should be noted that Yang et al. generated a similar mixture.⁶⁴ The minor isomer could be separated from this mixture, and the major isomer could be equilibrated to afford a 1:2 ratio of 88a to 88b. The minor macrolactone, isolated from both of these mixtures, could be protected as a chloroacetate, and the Bn-group could be removed by hydrogenolysis to afford the desired fragment 89.

The two fragments were stitched together by means of a convergent TMSOTf-catalyzed Schmidt glycosylation (Scheme 15). This was accomplished by treating a 2:1 mixture of disaccharide glycosyl donor 77 and disaccharide glycosyl acceptor 89 with TMSOTf (12%, CF₃SO₃Si(CH₃)₃) to afford tetrasaccharide 90 with complete α-selectivity via the anchimeric assistance of the C-2 chloroacetate (Scheme 15). The tetrasaccharide 90 was deprotected by a 2-step procedure. Specifically, the acetonide of 90 could be removed with TFA to furnish tetraol 91. Finally, the two chloroacetate groups in 91 were removed with thiourea to provide merremoside D (92). While the preponderance of evidence suggests that the synthetic material was consistent with the natural material (specific rotation, melting point and HRMS), the comparison of the spectral data was complicated by the limited NMR data reported for the isolated merremoside D.⁶⁶

Scheme 15:

Synthesis of the Merromoside D (end game)

3.8. Synthesis of Mezzettiaside

Our final application of this asymmetric Achmatowicz approach to oligosaccharides was one aimed at the family of natural products known as the messettiasides.⁷³ Like the cleistriosides and cleistetrosides, the messettiasides make up a family of acetylated rhaman type oligosaccharides (Schemes 16–19).^74,75 In our approach to this class of natural products was aimed at an efficient synthesis of the entire class of the natural products, which minimizes the use of protecting groups and the total number of steps to the all the member of the natural product family. Our approach to the mezzettiaside family of natural products was based upon a modular strategy with the identification of three key building blocks, disaccharide 99, trisaccharide 106 and tetrasaccharide 111.

Scheme 16:

Synthesis of the Mezzettiaside Core Disaccharide

The synthesis of the key disaccharide 99 began with the Pd-catalyzed glycosylation of n-octanol with pyranone 4a(α-L) to give aculose sugar 93 (Scheme 16). A 1,2-reduction of enone 93 under the Luche condition gave allylic alcohol 94. This allowed for the selective installation of an n-hexanoate group at C-4, followed by an Upjohn dihydroxylation to give diol 95 with rhamno-stereochemistry. As with our cleistrioside/cleistetroside syntheses, the regio- and stereo-selective glycosylation/chloroacylation of monosaccharide 95 could be accomplished using stochiometric tin acetal chemistry (e.g., Bu₂Sn=O).⁵¹ Without the use of tin acetal chemistry, the wrong regioisomer was produced as the major product. Specifically, a 1:2 mixture of regioisomer C-3 vs C-2 (i.e., 96:97) was produced when the Pd-glycosylation was performed on diol 95. Although it was difficult to reproduce, we have found that when the tin acetal of diol 95 was glycosylated under our typical Pd-glycosylation conditions, the desired regioisomer 96 was selectively produced (~7:1 of 96:97).⁷³ A practical alternative to this reproducibility problem emerged when we turned to the use of the Taylor catalyst (Ph₂BOCH₂CH₂NH₂)⁷⁶ over the stochiometric tin acetal chemistry. The resulting dual nucleophilic/electrophilic catalytic system resulted in a much more reactive B/Pd-catalyst system. Presumably, this increased reactivity is due to the anionic nature of the borinate complex make them an ideal coupling partner with the cationic Pd-π-allyl complex. Our optimized conditions using the Taylor catalyst (3% Pd₂(dba)₃·CHCl₃/4PPh_3, 15 mol% Ph₂BOCH₂CH₂NH₂ in CH₂Cl₂) typically used a 3:1 ratio of boron to palladium. Using these conditions, CH₃CN/THF (10:1) as solvents gave a 8:1 ratio of 96 to 97. The C-2 alcohol of the mixture could be chloroacylated ((ClAc)₂O, 10 mol% DMAP in Py) to form a similar mixture of regioisomer 98. Then the enone could be reduced under the Luche condition to form the key disaccharide 99, at this point the minor regioisomer could be removed chromatographically.

At this stage, the key disaccharide 99 served as a lynch pinpoint in the synthesis providing access to three disaccharide mezzettiasides, specifically, mezzettiaside-9, 10, and 11 (Scheme 17). This began a combination of allylic alcohol acylation, de-chloroacylation, and dihydroxylation. For the synthesis of mezzettiaside-9 (100) the transformation was accomplished in four steps and started with the chloroacetylation of 99 at the C-4 position ((ClAc)₂O, 10 mol% DMAP in Py), followed by an Upjohn dihydroxylation (OsO₄/NMO(aq)), bis-acetylation (Ac₂O/Py) and finally bis-deprotection of the two chloroacetate groups (thiourea, NaHCO₃ and n-Bu₄NI) to give mezzettiaside-9 (100). By simply not including the chloroacetylation step, the sequence resulted in a three steps mezzettiaside-11 (101) synthesis. Using a slightly different sequence, the regioisomeric diacetate mezzettiaside-10 (103) was prepared in four steps, C-4 acetylation, dihydroxylation, and C-2 ortho-ester regioselectively acetylation to form 102. Then, a selective removal of the C-2 chloroacetate group gave mezzettiaside-10 (103).

Scheme 17:

Synthesis of the Mezzettiasides 9–10

The fully protected intermediate 102, from the synthesis of mezzettiaside-10 (103), was also the lynchpin for the syntheses of the larger trisaccharide mezzettiasides (Scheme 18). The approach to the trisaccharide mezzettiasides 2-4 and 8 (109, 108, 107, and 105) began with the Pd-catalyzed glycosylation of disaccharide 102 with pyranone 4a(α-L) to give trisaccharide 104. A 4-step sequence (reduction, dihydroxylation, C-2 acylation, and chloroacetate deprotection) was used to convert trisaccharide 104 into mezzettiaside-8 (105). A 3-step sequence (reduction, C-4 acylation, dihydroxylation) on trisaccharide 104 was used to convert it into pivotal trisaccharide intermediate 106, which could be used to prepare messettiasides 2-4. For the conversion to messettiaside-4 (109), all that was need was the deprotection of the chloroacetate with thiourea. A 2-step procedure was used (C-2 acetylation and chloroacetate deprotection) for the conversion to messettiaside-3 (108). Similarly, a 2-step procedure was used (C-3 acetylation and chloroacetate deprotection) for the conversion to messettiaside-2 (107).

Scheme 18:

Synthesis of the Mezzettiasides 2–4 and 8

The key tris-rhamno-trisaccharide intermediate 106 for the trisaccharide mezzettiasides syntheses was also used for the synthesis of the tetrasaccharide mezzettiasides (mezzettiaside 5-7) (Scheme 19). This was accomplished with a C-3 regioselective dual B-nucleophilic/Pd-electrophilic catalyzed glycosylation (2.5 mol% Pd₂(dba)₃·CHCl₃/4PPh_3, 15 mol% Ph₂BOCH₂CH₂NH₂ in THF/CH₃CN) to give tetrasaccharide enone 110. The enone in 110 was converted into a rhamno-sugar 111 by a Luche reduction and Upjohn dihydroxylation. A regioselective borinate-catalyzed C-3 acetylation followed by chloroacetate deprotection was used to convert 111 into mezzettiaside-5 (112). Similarly, an orthoester-mediated C-2 acetylation and chloroacetate deprotection were used to give mezzettiaside-6 (113). Finally, a direct chloroacetate deprotection of 111 provided the final tetrasaccharide natural product, mezzettiaside-7 (114).

3.9. Pd-glycosylations in oligosaccharide synthesis by others

The overall acceptance of this approach can be seen by its use by others in the synthesis of carbohydrate-based natural products and oligosaccharides (Scheme 20 and 21). An example of this is the Lowary synthesis of the C-3 O-Me-mannose trisaccharide 116. The synthesis began with the iterative Pd-glycosylation of n-octanol with 4f(α-D), Luche reduction to give enone containing trisaccharide 115. A subsequent Luche reduction of 115 followed by a per-dihydroxylation gave a tris-manno-trisaccharide, which was selectively methylated at the three C-3 positions to give 116.⁷⁷

Scheme 20:

Lowary Synthesis of 1,4-Oligomers

Scheme 21:

Tang Synthesis of 1,4-Oligomers

An alternative example of this approach can be seen in the Tang synthesis of the tetrasaccharide 120 (Scheme 21).³⁴ A novel feature of the Tang approach was the development of a chiral catalyst to give a reagent control enone reduction to install an amecitose or rhodinose sugar. The Tang synthesis of 120 began with the Pd-glycosylation of BnOH with 4a(β-L), followed by a Rh-catalyzed per-reduction of the enone to give the amecitose monosaccharide 117. Simply repeating this 2-step process gave disaccharide 118 and then again trisaccharide 119. Finally, the orthogonality of the reagent control was demonstrated by the introduction of a rhodinose sugar to form 120. This was accomplished with the Pd-glycosylation of 119 with 4a(β-L), followed by a Rh-catalyzed per-reduction with the enantiomeric ligand system to give the rhodinose sugar of tetrasaccharide 120.

4.0. Conclusions

For the last 25 years, we have been exploring the use of asymmetric catalysis in combination with Achmatowicz rearrangement for the “de novo asymmetric synthesis of carbohydrate”. We refer to these syntheses as “de novo asymmetric” when the synthesis begins with achiral starting materials and when asymmetric catalysis is used for the installation of all the asymmetry. As part of these efforts, we have developed two orthogonal de novo asymmetric approaches to hexoses: an iterative dihydroxylation strategy and an Achmatowicz strategy. While there have been numerous de novo asymmetric approaches to carbohydrates, their applications to oligosaccharides have been a noticeable deficiency. In contrast to the other approaches, the asymmetric Noyori/Achmatowicz approach to carbohydrates has shown great capacity for the synthesis of oligosaccharide targets. This is presumably due to the compatibility of the Pd-catalyzed glycosylation reaction to a myriad of functional groups. Key to the success of the approach is the coupling of asymmetric catalysis with the Achmatowicz rearrangement for the synthesis of stereochemically simplified D- and L-pyranones with α- and β-stereochemistry. Thus, simplifying the 32 possible hexose monosaccharides down to 4 stereoisomers. For this approach to be viable, it relies upon a growing set of highly diastereoselective post-glycosylation reactions for the installation of the rest of the desired monosaccharide stereochemistry. The overall efficiency of these approaches is the result of the strategic use of the enone functionality in the pyranone as atom-less protecting groups. While this use of atom-less protecting groups has been under-explored in carbohydrate chemistry, the success of the above approaches suggests that this strategy should be given more attention by the synthetic organic chemistry community. The overall practicality of these syntheses can be seen in their ability to provide material for medicinal chemistry studies, often in ways that are not readily available by traditional carbohydrate syntheses.^1,6