Overexpression and RNAi-mediated Knockdown of Two 3β-hydroxy-Δ5-steroid dehydrogenase Genes in Digitalis lanata Shoot Cultures Reveal Their Role in Cardenolide Biosynthesis

Abstract

Abstract3β-hydroxy-Δ5-steroid dehydrogenases (3βHSDs) are supposed to be involved in 5β-cardenolide biosynthesis. Here, a novel 3βHSD (Dl3βHSD2) was isolated from Digitalis lanata shoot cultures and expressed in E. coli. Recombinant Dl3βHSD1 and Dl3βHSD2 shared 70% amino acid identity, reduced various 3-oxopregnanes and oxidised 3-hydroxypregnanes, but only rDl3βHSD2 converted small ketones and secondary alcohols efficiently. To explain these differences in substrate specificity, we established homology models using borneol dehydrogenase of Salvia rosmarinus (6zyz) as the template. Hydrophobicity and amino acid residues in the binding pocket may explain the difference in enzyme activities and substrate preferences. Compared to Dl3βHSD1, Dl3βHSD2 is weakly expressed in D. lanata shoots. High constitutive expression of Dl3βHSDs was realised by Agrobacterium-mediated transfer of Dl3βHSD genes fused to the CaMV-35S promotor into the genome of D. lanata wild type shoot cultures. Transformed shoots (35S:Dl3βHSD1 and 35S:Dl3βHSD2) accumulated less cardenolides than controls. The levels of reduced glutathione (GSH), which is known to inhibit cardenolide formation, were higher in the 35S:Dl3βHSD1 lines than in the controls. In the 35S:Dl3βHSD1 lines cardenolide levels were restored after adding of the substrate pregnane-3,20-dione in combination with buthionine-sulfoximine (BSO), an inhibitor of GSH formation. RNAi-mediated knockdown of the Dl3βHSD1 yielded several shoot culture lines with strongly reduced cardenolide levels. In these lines, cardenolide biosynthesis was fully restored after addition of the downstream precursor pregnan-3β-ol-20-one, whereas upstream precursors such as progesterone had no effect, indicating that no shunt pathway could overcome the Dl3βHSD1 knockdown. These results can be taken as the first direct proof that Dl3βHSD1 is indeed involved in 5β-cardenolide biosynthesis.

Four enzymes control natural variation in the steroid core ofErysimumcardenolides

“…In this study, we identified and biochemically characterized four cardenolide biosynthetic enzymes from E. cheiranthoides. Of these, Ec3βHSD and EcP5βR2 belong to enzyme families that have been speculated, or shown, to be involved in cardenolide biosynthesis in other species (13,30,31,45,47). However, the mutant lines generated in this study provide critical in vivo evidence for their role in cardenolide synthesis, and the lack of a cardenolide phenotype for EcP5βR1 knockouts highlights the potential disparity between in vitro activity and in vivo function.…”

“…Enzymes of this type are thought to be required for cardenolide biosynthesis (30)(31)(32). as yellow rectangles.…”

“…All species examined had more than one Ec3βHSD-like gene, with the exception of E. cheiranthoides. The three orthologs from E. crepidifolium (44), two from D. lanata (30)(31)(32), and two from A. thaliana, AtSDR5 (AT2G47140) and AtSDR3 (AT2G47130) (10), have been shown to accept cardenolide intermediates in vitro, but in planta evidence for involvement in cardenolide biosynthesis only exists for Dl3βHSD1 (30) (Figures 6A, S15). Of the species included in the KSI gene tree, only E. cheiranthoides contained more than one copy (Figures 6B, S16).…”