Automated, Multistep Continuous-Flow Synthesis of 2,6-Dideoxy and 3-Amino-2,3,6-trideoxy Monosaccharide Building Blocks

Abstract

An automated continuous flow system capable of producing protected deoxy-sugar donors from commercial material is described. Four 2,6-dideoxy and two 3-amino-2,3,6-trideoxy sugars with orthogonal protecting groups were synthesized in 11-32% overall yields in 74-131.5 minutes of total reaction time. Several of the reactions were able to be concatenated into a continuous process, avoiding the need for chromatographic purification of intermediates. The modular nature of the experimental setup allowed for reaction streams to be split into different lines for the parallel synthesis of multiple donors. Further, the continuous flow processes were fully automated and described through the design of an open-source Python-controlled automation platform.

Graphical Abstract

A continuous flow platform that is able to concatenate multiple steps for rapid, multi-gram production of orthogonally protected deoxy-sugars is described. Intermediate purification of compounds is avoided by the ability to run telescoped sequences of 2-4 chemical transformations. Through the use of Mechwolf, an open-source, Python-controlled program, automation of the flow processes was achieved.

The growing recognition of the importance of glycomics in biomedical research has led to increased interest in developing automated platforms for the production of homogenous oligosaccharides. While many elegant systems have been developed for oligosaccharide synthesis, throughput is limited because the coupling partners must be produced through time-consuming multi-step manual synthesis.^1–3 Although one-pot protection schemes have shortened the production of a small set of monosaccharides obtained from commercially available substrates,^4,5 the “unusual” monosaccharides that are common components of many bioactive natural products and microbial antigens remain challenging to obtain.⁶ Readily-reproducible processes are needed for the rapid production of glycosylation-ready monosaccharides from inexpensive commercial materials.

Recently, our labs reported on the challenges associated with converting batch processes into machine-assisted syntheses⁷ and using continuous flow conditions to carry out protecting group manipulations (Scheme 1).^8,9 By taking advantage of the many benefits of flow chemistry,^10–15 such as increased reaction efficiency^16,17 and avoiding unwanted exotherms,¹⁸ the continuous flow processes resulted in significantly faster reaction times than was possible under standard batch-phase conditions. Despite these faster reaction times, the isolation and purification of substrates was required after every reaction. Recognizing that continuous flow processes have been successfully utilized for multi-step synthesis,^19–27 we envisioned that this technology could be employed for the production of highly-functionalized sugar-coupling partners through telescoping several reactions in a sequence. Given the modular nature of continuous flow processes this approach could be used to access scaffolds that are not commercially available, further opening up the chemical space available for glycomics research. Because continuous flow platforms can be controlled through the use of open-source Python-based programs such as MechWolf, processes could be fully automated and described for straightforward reproducibility.^28–35 Here we describe the development of the first automated continuous flow platform for the production of fully protected 2,6-dideoxy and 3-amino-2,3,6-trideoxy monosaccharides that are suitable for use in chemical glycosylation reactions.

Scheme 1.

Previously reported syntheses of 2 and 3 from l-rhamnal 1.

To begin the study, 2,6-dideoxy sugars 4–8 and 3-amino-2,3,6-trideoxy sugars 9 and 10 were targeted due to their prevalence in several bioactive natural products and interest from the synthetic community (Figure 1).^36–51 In our previous work, the synthesis of 6-deoxy glycals 2 and 3 commenced from commercially available diacetyl-l-rhamnal 1. While efficient, the cost of 1 led to investigation of a more affordable and scalable route. To this end, a peracetylation, bromination, and Fischer-Zach sequence starting from commercially available l-rhamnose 11 was explored.^52,53 After optimization of each reaction (SI, section 2.2), telescoping the sequence was achieved by utilizing acetyl bromide to perform a tandem acetylation/bromination to generate a glycosyl bromide. Following bromination, excess HBr was quenched by flowing a solution of sat. NaHCO₃ (aq.) for an in-line quench, followed by funneling the reaction mixture into a beaker of sat. NaHCO₃ (Scheme 2). Upon mechanical extraction of the organic phase, the reaction stream was mixed with solution of 4.17 M aq. NaH₂PO₄ and passed through a packed bed reactor of zinc. Employing these conditions on 5 grams of l-rhamnose 11 afforded glycal 1 in 64% yield over two steps in 28.5 minutes of total residence time at a rate of 7.53 g/h. A similar process was used for synthesis of more difficult to access l-fucal 13 beginning from l-fucose 12. EtOAc was used in this setup owing to the low solubility of the fucosyl bromide in CH₂Cl₂ (SI, section 2.3). This reaction proceeded more slowly and required recycling the product stream through the packed bed reactor three times. Through this process, l-fucal 13 was produced in 56% yield over the two steps in 66.5 minutes of total reaction time.

Figure 1.

2,6-dideoxy and 3-amino-2,6-dideoxy sugar targets 4-10.

Scheme 2.

Synthesis of l-rhamnal 1 and l-fucal 13 through a two-step telescoped sequence.

With a scalable route to 1 and 13 developed, telescoping protecting group manipulations to access glycal precursors 2 and 14 was investigated. Following optimization of the individual reactions (SI Section 2.2), telescoping the deacetylation, C3 TBS protection, and C4 benzyl ether installation proceeded smoothly (Scheme 3). Through this three-step sequence, 2 was produced in 62% yield over the three steps in 29.5 minutes of total reaction time from glycal 1. Notably, no exotherms were observed in this process, avoiding the need for active cooling of the reactors. Similarly, application of the telescoped process to glycal 13 enabled isolation of 14 in 59% yield over 3 steps in 29.5 minutes of residence time.

Scheme 3.

Synthesis of 2 and 14 through a three-step telescoped sequence.

With a telescoped sequence for regioselective C3 protection established, efforts shifted towards developing a parallel process for installing orthogonal protecting groups at the C4 position. To this end, either a benzyl or naphthyl ether was introduced at the C4 position by utilizing a Y-splitter to divert the stream following alkoxide formation through the use of a commercially available solution of NaHMDS in THF. Further, a silyl ether removal step was integrated to the telescoped sequence by using a commercially available solution of TBAF.⁵⁴ Following optimization of reagent stoichiometry, the telescoped sequence afforded 14 in 52% yield over 4 steps and 16 in 49% over 4 steps in 34.5 minutes of total reaction time (Scheme 4a). The general applicability of this method was tested by applying the telescoped sequence to glycal 13. The four-step process enabled synthesis of 17 in 51% yield over 4 steps in 34.5 minutes of total residence time (Scheme 4b). The reactions proceeded smoothly at ambient temperature (23-25 °C) with no exotherms observed.

Scheme 4.

(a) Synthesis of 15 and 16 using a Y-splitter in a four-step telescoped sequence. (b) Synthesis of 17.

Next, efforts were focused on derivatizing the glycals into corresponding deoxy sugars. This was achieved by using a Y-splitter to divert a mixed stream of glycal and PPh₃·HBr for the addition of either H₂O or 4-methylbenzenethiol (Scheme 5a).^55,56 Using the Y-splitter, simultaneous synthesis of l-olivose hemiacetal 4 in 83% yield in 16 minutes and l-olivose thioglycoside 5 in 64% yield as a 3:1 α:β mixture anomers in 11 minutes was achieved. Furthermore, hydration of glycal 14 produced l-oliose 6 in 77% yield (Scheme 5b).

Scheme 5.

(a) Synthesis of hemiacetal 4 and thioglycoside 5 using a Y-splitter. (b) Synthesis of l-oliose 6.

Following synthesis of l-olivose 4 and 5, the construction of more difficult to access donors was investigated. To this end, an oxidation, 1,4-conjugate addition, reduction pathway was used to invert the C3 hydroxyl group on glycals 15 and 17. After optimization of individual steps (SI, section 2.11), a telescoped sequence of oxidation, 1,4-conjugate addition, and reduction resulted in the formation of α-l-digitoxose 7 thioglycoside as a single diastereomer in 39% yield in 33.5 minutes of total residence time (Scheme 6a). Importantly, the telescoped process was run at ambient temperatures and scaled to one-gram of 15, allowing for production of l-digitoxose 7 at a rate of 1.06 g/h. Extending this telescoped sequence to glycal 17 provided β-l-boivinose 8 thioglycoside in 45% yield as a single diastereomer over the three-step telescoped sequence in 33.5 minutes of total residence time.⁵⁷

Scheme 6.

Synthesis of l-digitoxose 7 and l-boivinose 8 through a three-step telescoped sequence.

With telescoped routes to four 2,6-dideoxy sugars optimized, efforts shifted towards amino sugar synthesis. C3 amino functionality was introduced by subjecting 16 to an oxidation, 1,4-conjugate addition, and reductive amination sequence using NH₄OAc and NaBH₃CN. To facilitate the reductive amination, the base from the 1,4-conjugate addition was neutralized by a packed bed reactor of IR-120 resin (SI, Section 2.14).⁵⁸ With this setup, l-ristosamine 9 was furnished in 34% yield over three steps in 73.5 minutes of total residence time (Scheme 7). Analogously, l-megosamine 10 was obtained in 35% yield using this telescoped sequence in 73.5 minutes of total residence time.

Scheme 7.

Synthesis of l-ristosamine 9 and l-megosamine 10 through a three-step telescoped sequence.

Finally, the single reactions and telescoped sequences were automated by using our python based open source MechWolf program. Notably, no observable decreases in yield or further optimizations were required when using the automation platform. For example, automating the three-step telescoped sequence from glycal 1 to 2 proceeded in 60% overall yield in 29.5 minutes of total residence time. Features built into the Mechwolf program, such as a self-updating stoichiometry table, tube length calculator, and reactor visualization function allow for greater reproducibility between runs (SI, Section 3). Furthermore, the open-source nature of Mechwolf enables users to save and export the code for each online notebook page. This allows for the synthetic community to continue to build a repository and standardize monosaccharide building block synthesis.

In conclusion, an automated, continuous flow platform for the rapid production of otherwise difficult to access deoxy-sugars has been developed. Starting from commercially available l-rhamnose 11, l-olivose hemiacetal 4, l-olivose thioglycoside 5, l-digitoxose 7, l-ristosamine 9, and l-megosamide 10 were produced in 32%, 24%, 13%, 11%, and 11% overall yields, respectively, in three telescoped sequences. Similarly, starting from l-fucose 12, l-oliose 6 and l-boivinose thioglycoside 8 were synthesized in 25% and 12% overall yields, respectively. By concatenating 5–9 chemical transformations, the deoxy-sugars were produced in total reaction times ranging from 74–131.5 minutes (compared to several days using batch conditions), with each sugar requiring only three total chromatographic steps. Notably, no exotherms were observed and all telescoped sequences were able to be run at room temperature. The efficiency of this process is further increased through the use of Y-splitters that permit the simultaneous production of a variety of targets in a single run. Through our open-source Python-controlled program, the flow processes were successfully automated. Extension of the automation platform to other sugar coupling partners for use in automated oligosaccharide synthesis^59,60 is underway.