Synthesis of D-manno-heptulose via a cascade aldol/hemiketalization reaction

A [4 + 3] synthesis of D-manno-heptulose is described. The cascade aldol/hemiketalization reaction of a C4 aldehyde with a C3 ketone provides the differentially protected ketoheptose building block, which can be further reacted to furnish target D-manno-heptulose.

Introduction D-manno-Heptulose is a rare naturally occurring seven-carbon sugar first isolated from avocado [1], which exhibited promising diabetogenic effects through suppression of the glucose metabolism and insulin secretion via competitive inhibition of the glucokinase pathway [2][3][4][5][6]. Accordingly, ketoheptoses and fluorinated ketoheptoses were considered to be potential therapeutic agents for hypoglycemia and cancer as well as diagnostic tools for diabetes [7][8][9][10][11][12]. Amino-and azido-group-containing ketoheptoses were also synthesized for the development of novel antibiotics and the evaluation of carbohydrate-lectin interactions by conjugation with fluorescent quantum dots via click chemistry [13,14]. Besides, differentially protected D-manno-heptulose building blocks could serve as valuable precursors for the synthesis of C-glycosides [15,16].
The known synthesis of D-manno-heptulose mainly rely on the use of rearrangements and chain elongation reactions [17]. Rearrangement reactions such as the Lobry de Bruyn rearrangement and the Bilik rearrangement employ unprotected aldoses as substrates, usually yielding an equilibrium mixture of aldoses and ketoses [18,19]. In addition to chain elongations of aldoses employing the Henry reaction, the aldol reaction, and the Wittig reaction for the preparation of ketoheptoses [20][21][22], sugar lactones were also often utilized for the synthesis of D-mannoheptulose via reactions with C-nucleophiles or conversion into exocyclic glycals followed by dihydroxylation [10][11][12][13][23][24][25][26][27]. Remarkably, Thiem et al. reported the highly efficient synthesis of D-manno-heptulose from D-mannose in 59% overall yield over five steps [26]. However, the synthesis of D-manno-heptu- lose and its derivatives from the common differentially protected ketoheptose building block is still attractive due to the versatile functionalization possibilities of the building block into various derivatives of D-manno-heptulose. A de novo synthesis has proved to be an attractive strategy to produce orthogonally protected carbohydrate building blocks from simple precursors [28][29][30][31][32][33][34][35][36][37][38][39]. Here, we report a [4 + 3] approach to access differentially protected ketoheptose building blocks, which enables the synthesis of D-manno-heptulose. As depicted in Scheme 1, D-manno-heptulose (1) could be obtained by global deprotection of the differentially protected ketoheptose building block 2. The ketoheptose 2 can be further divided into C 4 aldehyde 3 and C 3 ketone 4 via a cascade aldol/hemiketalization pathway.

Results and Discussion
The synthesis of the C 4 aldehyde commenced with commercially available D-lyxose (5, Scheme 2). The reaction of 5 with ethanethiol in the presence of hydrochloric acid followed by selective protection of the 4,5-diol with 2,2-dimethoxypropane using pyridinium p-toluenesulfonate as the promoter gave the 4,5-O-isopropylidene derivative 6 in 71% yield over two steps [40]. Treatment of diol 6 with bis(tributyltin) oxide and subsequent exposure to p-methoxybenzyl (PMB) chloride in the presence of tetra-n-butylammonium bromide (TBAB) at 110 °C led to regioselective protection of the 3-OH with the PMB group, affording the 3-O-PMB protected alcohol 7 (55%) [41]. At this stage, we initially planned to synthesize the 2-OH-protected C 4 aldehyde for the assembly of the seven-carbon skeleton. Thus, acetylation of the 2-OH group in 7 with acetic anhydride and DMAP in dichloromethane provided ester 8 in 86% yield. The positions of the 2-acetyl and 3-PMB groups were determined by 1 H, 13 C and 2D NMR spectra of 8 (see Supporting Information File 1 for details). Cleavage of the isopropylidene acetal group in 8 under acidic conditions gave diol 9 (50%). However, oxidative cleavage of diol 9 with sodium periodate resulted in the unexpected formation of α,β-unsaturated aldehyde 10 in 71% yield, indicating that the 2-acetyl group might be prone to initiate the elimination reaction. The double bond of 10 was assigned to have Z-configuration based on the analysis of the NOEs between the olefinic hydrogen and the aldehyde hydrogen (see Supporting Information File 1 for details). In addition, when alcohol 7 was subjected to benzoyl chloride and DMAP in dichloromethane at room temperature or tert-butyldimethylsilyl chloride and imidazole in DMF at room temperature, no reaction occurred probably because of the steric hindrance between the 2-OH group and the surrounding functional groups.
To overcome the difficulties in the synthesis of the 2-OH-protected C 4 aldehyde and to improve the synthetic efficiency in the assembly of ketoheptose skeletons, we envisioned ketoheptoses could be assembled by a cascade aldol/hemiketalization reaction between 2-OH-unprotected C 4 aldehyde 3 and C 3 ketone 4. As such, the isopropylidene acetal group in 7 was cleaved under acidic conditions to produce triol 11 in 86% yield (Scheme 3). Cleavage of the resulting vicinal diol in 11 with sodium periodate led to the C 4 aldehyde 3 in nearly 60-70% yield. In this oxidative cleavage reaction, almost no elimination product was found based on TLC monitoring. Given that the C 4 aldehyde 3 was unstable upon purification by silica gel column chromatography, it was immediately used for the subsequent coupling after the extraction procedure. The aldol reaction of aldehyde 3 with the readily available ketone 4 [42,43] under the catalysis of L-proline at room temperature for three days proceeded sluggishly, leading to the desired product in a very low yield. Gratifyingly, when the L-proline-catalyzed aldol reaction was performed at 70 °C for one day, the TLC indicated the complete consumption of aldehyde 3, and the generated 4,5-anti-selective coupling intermediate 12 underwent in situ cyclization to provide hemiketal 13 as the major product in about 50-60% yield (35% overall yield from compound 11). Notably, trace amounts of a stereoisomer and a minor highly polar unknown byproduct were also observed in this cascade reaction. The excellent anti-selectivity for the L-proline-catalyzed aldol reaction can be explained by the Houk-List transition state model [43][44][45]. Compound 13 was then acetylated to afford differentially protected ketoheptose building block 2 in 83% yield. The structure of 2 was unambiguously confirmed by 1 H, 13 C, and 2D NMR spectra (see Supporting Information File 1 for details). The anomeric α-configuration of compound 2 was confirmed by analysis of the NOE effects between the C-1 hydrogen and the C-5 hydrogen.
With the ketoheptose building block 2 in hand, we turned our attention to the synthesis of D-manno-heptulose (1). Upon exposure to NBS in acetonitrile and water, the dithioacetal in 2 was cleaved to give the corresponding aldehyde [46,47], which was then reduced by potassium borohydride in a methanol and dichloromethane solvent mixture to produce alcohol 14 as the predominant product (84% over two steps, Scheme 4). In addition, a trace amount of the deacetylated product was also detected . DDQ-mediated oxidative cleavage of the PMB group in alcohol 14 produced only a moderate yield (≈50%) of the 5,7-diol probably due to the presence of the free 7-hydroxy group. We envisaged that protection of the free 7-hydroxy group in 14 followed by treatment with DDQ could yield the desired 5-hydroxy product in high yield. Indeed, acetylation of alcohol 14 with acetic anhydride delivered ester 15 in 91% yield. Removal of the PMB group in 15 with DDQ resulted in a very clean reaction, affording alcohol 16 in an excellent yield (91%). Saponification of all esters in 16 with potassium carbonate followed by acidic cleavage of the isopropylidene acetal group with aqueous acetic acid furnished D-mannoheptulose (1, 76% over two steps). The structure of 1 was found to be in good agreement with those reported for α-D-mannoheptulose (1) by comparison of the NMR spectra (see Supporting Information File 1 for details) [26].
The key step is a cascade aldol/hemiketalization reaction for the construction of the differentially protected ketoheptose building block, which was finally converted into D-manno-heptulose for subsequent biological evaluation. Although the synthesis of D-manno-heptulose (5% overall yield, 13 steps) is not so efficient as the Thiem's method (59% overall yield, 5 steps), the reported differentially protected ketoheptose building blocks may find further application in the preparation of structurally diverse D-manno-heptulose derivatives.

Supporting Information
Supporting Information File 1 Experimental details, characterization data, and NMR spectra of all new compounds.