A cellulose synthase-derived enzyme catalyses 3-O-glucuronosylation in saponin biosynthesis

Abstract

Triterpenoid saponins are specialised metabolites distributed widely in the plant kingdom that consist of one or more sugar moieties attached to triterpenoid aglycones. Despite the widely accepted view that glycosylation is catalysed by UDP-dependent glycosyltransferase (UGT), the UGT which catalyses the transfer of the conserved glucuronic acid moiety at the C-3 position of glycyrrhizin and various soyasaponins has not been determined. Here, we report that a cellulose synthase superfamily-derived glycosyltransferase (CSyGT) catalyses 3-O-glucuronosylation of triterpenoid aglycones. Gene co-expression analyses of three legume species (Glycyrrhiza uralensis, Glycine max, and Lotus japonicus) reveal the involvement of CSyGTs in saponin biosynthesis, and we characterise CSyGTs in vivo using Saccharomyces cerevisiae. CSyGT mutants of L. japonicus do not accumulate soyasaponin, but the ectopic expression of endoplasmic reticulum membrane-localised CSyGTs in a L. japonicus mutant background successfully complement soyasaponin biosynthesis. Finally, we produced glycyrrhizin de novo in yeast, paving the way for sustainable production of high-value saponins.

Fig. 1. Proposed biosynthetic pathways of oleanane-type triterpenoid saponins catalysed by characterised enzymes in Glycine max, Glycyrrhiza uralensis and Lotus japonicus.

The yellow arrow indicates the cyclisation reaction catalysed by β-amyrin synthase (bAS), green arrows show oxidation reactions catalysed by cytochrome P450 monooxygenases (P450s) and red arrows indicate glycosylation reactions catalysed by UDP-dependent glycosyltransferases (UGTs). The reaction sites are indicated by the corresponding colours. MtCYP72A63 is in parentheses because its catalytic reaction is represented by the results of enzyme assays conducted with exogenous substrates. Bold red arrows indicate glycosylation reactions catalysed by the novel cellulose-synthase-derived glycosyltransferase (CSyGT) characterised in this paper. Gm G. max; Gu G. uralensis, Lj L. japonicus, Mt Medicago truncatula, Ara arabinose, Gal galactose, Glc glucose, GlcA glucuronic acid, Rha rhamnose, Xyl xylose. *The glycosylation pattern of R is identical to that of group B soyasaponins. ^†Unpublished data.

Fig. 2. Gene co-expression analyses of CSyGTs in G. max, G. uralensis and L. japonicas.

All heatmaps are in hierarchical clustering of expression profiles and expression levels are normalised across conditions. a G. max expression profiles retrieved from Phytozome. b UniGene expression profiles of G. uralensis. Library 1 extracted from roots of 308–19 (high glycyrrhizin-producing) strain in June; Library 2 from roots of 308–19 strain in December; Library 3 from roots of 87–458 (low glycyrrhizin-producing) strain in June; Library 4 from leaves of 308–19 strain in June. c Gene expression profiles of L. japonicus retrieved from Lotus Base.

Fig. 3. Functional characterisation of CSyGTs in yeast and in planta.

a Overlays of LC–MS chromatograms obtained by selected-ion monitoring (SIM) of the theoretical m/z values of the compounds of interest. Chromatograms of monoglucuronides produced in vivo in transformed glycyrrhetinic acid-producing strains (GA), soyasapogenol B-producing strains (SB) and oleanolic acid-producing strains (OA) with GuCSyGT, LjCSyGT and GmCSyGT1. b Soyasaponin Bb accumulation in the roots of wild-type (Gifu) and LjCSyGT mutant lines (Supplementary Fig. 2). c Soyasaponin Bb accumulation in the roots of wild-type and LjCSyGT mutants transformed with GuCSyGT (Gu), LjCSyGT (Lj), GmCSyGT1 (Gm1) or empty vector (E). Three biologically independent replicates were performed. Error bars represent SDs (n = 3).

Fig. 4. Subcellular localisation of LjCSyGT.

a Confocal image of LjCSyGT-RFP co-expressed in hairy roots of the LjCSyGT mutant lines with an endoplasmic reticulum (ER) marker and a Golgi marker. Scale bar, 50 µm. LjCSyGT-RFP shows the characteristic web-like pattern of the ER network, suggesting localisation to the ER. b Soyasaponin Bb accumulation in transgenic roots of LjCSyGT mutant lines expressing RFP-LjCSyGT, LjCSyGT-RFP or RFP-CCaMK (calcium/calmodulin-dependent protein kinase is known to localise to the nucleus) as a control.

Fig. 5. Phylogenetic analyses of the cellulose-synthase superfamily in sampled dicots.

Best-scoring maximum likelihood (ML) tree constructed using RAxML. Numbers are the bootstrap values (%) from 1,000 replicates. *CSyGTs are clustered together, expanding from other CslMs. At Arabidopsis thaliana, Cq Chenopodium quinoa, Gm Glycine max, Gu Glycyrrhiza uralensis, Lj Lotus japonicus, Pg Panax ginseng.

Fig. 6. In vivo assays of CSyGT and GmCslM activity using a yeast-expression system.

All overlays of chromatograms were analysed by LC–MS with selected-ion monitoring (SIM) of the theoretical m/z values of the compounds of interest. Signals were compared to authentic standards (standard) if available. a Chromatograms of in vivo-produced monoglucuronides by transformed triterpenoid aglycone-producing strains with GmCSyGT1–3, GmCslM1 or GmCslM2. b Chromatograms of products of transformed glycyrrhetinic acid-producing strains (GAs), selected based on the theoretical m/z values of 631.4 (left) and 645.4 (right) of glycyrrhetinic acid monoglucoside and monoglucuronide, respectively. c Chromatogram of in vivo-produced glucoglycyrrhizin (GLU, GA–GlcA–Glc) and intermediates by transformed glucoglycyrrhizin-producing platform strains (GLU). d Results of in vivo substrate-feeding assays of CSyGTs and GmCslMs. Structures of the substrates and full-length LC–MS chromatograms are shown in Supplementary Fig. 6. GA glycyrrhetinic acid (m/z 469.3), GA–GlcA glycyrrhetinic acid-3-O-monoglucuronide (m/z 645.4), GA–Glc glycyrrhetinic acid-3-O-monoglucoside (m/z 631.4), GL glycyrrhizin (m/z 821.4), GLU glucoglycyrrhizin (m/z 807.4), SBMG soyasapogenol B-3-O-monoglucuronide (m/z 633.4), OA oleanolic acid (m/z 631.4).

Fig. 7. De novo production of glycyrrhizin in yeast.

a Overlay of LC–MS chromatogram showing in vivo production of glycyrrhizin in the yeast strain GL0–3. Signals were compared to authentic standards. Chromatograms were selected based on the theoretical m/z values of glycyrrhetinic acid (GA, 469.3), glycyrrhetinic acid-3-O-monoglucuronide (GA–GlcA, 645.4) and glycyrrhizin (GL, GA–GlcA–GlcA, 821.4). b Glycyrrhizin production by GL0–3. The amount of glycyrrhizin was measured separately from collected cells (circles) and medium (dashed circles) 5 days after induction. Error bars represent SDs (n = 3).