The majority of prokaryotic drugs are produced in glycosylated form, with the deoxygenation level in the sugar moiety having a profound influence on the drug's bioprofile. Chemical deoxygenation is challenging due to the need for tedious protective group manipulations. For a direct biocatalytic de novo generation of deoxysugars by carboligation, with regiocontrol over deoxygenation sites determined by the choice of enzyme and aldol components, we have investigated the substrate scope of the F178Y mutant of transaldolase B, TalB(F178Y), and fructose 6-phosphate aldolase, FSA, from E. coli against a panel of variously deoxygenated aldehydes and ketones as aldol acceptors and donors, respectively. Independent of substrate structure, both enzymes catalyze a stereospecific carboligation resulting in the D-threo configuration. In combination, these enzymes have allowed the preparation of a total of 22 out of 24 deoxygenated ketose-type products, many of which are inaccessible by available enzymes, from a [3×8] substrate matrix. Although aliphatic and hydroxylated aliphatic aldehydes were good substrates, D-lactaldehyde was found to be an inhibitor possibly as a consequence of inactive substrate binding to the catalytic Lys residue. A 1-hydroxy-2-alkanone moiety was identified as a common requirement for the donor substrate, whereas propanone and butanone were inactive. For reactions involving dihydroxypropanone, TalB(F178Y) proved to be the superior catalyst, whereas for reactions involving 1-hydroxybutanone, FSA is the only choice; for conversions using hydroxypropanone, both TalB(F178Y) and FSA are suitable. Structure-guided mutagenesis of Ser176 to Ala in the distant binding pocket of TalB(F178Y), in analogy with the FSA active site, further improved the acceptance of hydroxypropanone. Together, these catalysts are valuable new entries to an expanding toolbox of biocatalytic carboligation and complement each other well in their addressable constitutional space for the stereospecific preparation of deoxysugars.

Aldol reactions constitute a powerful methodology for carbon-carbon bond formation in synthetic organic chemistry. Biocatalytic carboligation by aldolases offers a green, uniquely regio- and stereoselective tool with which to perform these transformations. Recent advances in the field, fueled by both discovery and protein engineering, have greatly improved the synthetic opportunities for the atom-economic asymmetric synthesis of chiral molecules with potential pharmaceutical relevance. New aldolases derived from the transaldolase scaffold (based on transaldolase B and fructose-6-phosphate aldolase from Escherichia coli) have been shown to be unusually flexible in their substrate scope; this makes them particularly valuable for addressing an expanded molecular range of complex polyfunctional targets. Extensive knowledge arising from structural and molecular biochemical studies makes it possible to address the remaining limitations of the methodology by engineering tailored biocatalysts.

We have evaluated different strategies for the one-pot synthesis of D-fructose 6-phosphate and its 1-deoxy and 3-deoxy analogs from readily available starting materials, by building up ''artificial metabolisms'' in vitro. The first consisted of an aldol cleavage-aldol formation cascade, in which glyceraldehyde 3-phosphate as the central intermediate is generated from fructose 1,6-bisphosphate and consumed in situ for a consecutive carboligation step catalyzed by fructose 6-phosphate aldolase (FSA). The second approach consisted of an aldolase-kinase coupling, in which the unphosphorylated ketose was produced in situ by FSA-catalyzed carboligation, followed by a hexokinase-catalyzed phosphorylation step. While both approaches profited from the high stereoselectivity and stability of the aldolases used, the first approach proved to be the most practical, effective and economical, whereas the second strongly depends on the substrate specificity of hexokinase, which shows inferior catalytic efficiency with the 1-deoxy substrate. Phosphorylation of 3-deoxyfructose failed because, contradictory to a literature report, this compound was found not to be acceptable as a substrate of yeast hexokinase.

Straightforward methods for the efficient, systematic preparation of libraries of the title compound classes have been evaluated. A general and efficient modular route to dihydroxyacetone monoethers was developed based on trityl glycidol, which, through epoxide opening, oxidation, and deprotection, provided variously alkylated ethers by three routine operations in good overall yields (eight examples, 24–59 %). The preparation of structurally related 1‐hydroxyalkanones depends on the availability of the most economic starting materials and on their physicochemical properties. Thus, the most practical one‐step approaches consisted of the sec‐selective oxidation of short‐chain 1,2‐diols (≤ C6) using NaOCl, and the direct ketohydroxylation of 1‐alkenes (≥ C6) using buffered stoichiometric KMnO4 or catalytic RuO4 with reoxidation by oxone, for which mostly good overall yields were achieved on a multigram scale (nine examples, 15–78 %).