Pathways to defense metabolites and evading fruit bitterness in genus Solanum evolved through 2-oxoglutarate-dependent dioxygenases

Abstract

The genus Solanum comprises three food crops (potato, tomato, and eggplant), which are consumed on daily basis worldwide and also producers of notorious anti-nutritional steroidal glycoalkaloids (SGAs). Hydroxylated SGAs (i.e. leptinines) serve as precursors for leptines that act as defenses against Colorado Potato Beetle (Leptinotarsa decemlineata Say), an important pest of potato worldwide. However, SGA hydroxylating enzymes remain unknown. Here, we discover that 2-OXOGLUTARATE-DEPENDENT-DIOXYGENASE (2-ODD) enzymes catalyze SGA-hydroxylation across various Solanum species. In contrast to cultivated potato, Solanum chacoense, a widespread wild potato species, has evolved a 2-ODD enzyme leading to the formation of leptinines. Furthermore, we find a related 2-ODD in tomato that catalyzes the hydroxylation of the bitter α-tomatine to hydroxytomatine, the first committed step in the chemical shift towards downstream ripening-associated non-bitter SGAs (e.g. esculeoside A). This 2-ODD enzyme prevents bitterness in ripe tomato fruit consumed today which otherwise would remain unpleasant in taste and more toxic.

Fig. 1

A simplified scheme of core steroidal glycoalkaloid modification in tomato and potato species. In tomato, during transition from green to red fruit, the core α-tomatine is bitter steroidal glycoalkaloid (SGA) and converted to the non-bitter esculeoside A and additional SGAs. The main SGAs in cultivated potato are α-chaconine and α-solanine. Both cultivated potato and S. chacoense share a common pathway up to the biosynthesis of α-chaconine and α-solanine. These SGAs are hydroxylated to form leptinines (leptinine I and leptinine II) and subsequently leptines (leptine I and leptine II) that are unique to the wild potato, S. chacoense. The known SGAs biosynthetic genes are shown in blue color. GAME31 and GAME32 genes characterized in this study are marked in green. Dashed and solid arrows suggest multiple or single enzymatic steps in the pathway, respectively. A detailed tomato SGAs biosynthetic pathway is presented in Supplementary Fig. 1. GAME GLYCOALKALOID METABOLISM, SGT STEROL ALKALOID GLYCOSYLTRANSFERASE, Ac Acetoxy, Glu Glucose, Gal Galactose, Xyl Xylose, Rha Rhamnose. α-tomatine-derived SGAs are marked in red while dehydrotomatine-derived SGAs are shown in black

Fig. 2

LC-MS-based screening of hydroxytomatine and acetoxytomatine SGAs in the BILs and ILs. a, b Levels of hydroxytomatine (a) and acetoxytomatine (b) detected following screening of leaf tissues (n = 1, single replicate obtained using the leaf-dipping method) of BIL and IL population (total 671 lines). SGA levels presented in the plots were arranged in ascending order. Top five BILs/ILs showing the highest SGA content are marked by a pink spot. Refer to Supplementary Data 1 for details regarding SGA level across the BIL and IL population. c, d Validation of hydroxytomatine (c) and acetoxytomatine (d) content in extracts of ground leaf tissue from selected topmost BILs (from a, b) along with S. lycopersicum (cv. M82) and wild species (S. pennellii) parents. Values indicate means ± standard error mean (n = 4 for parental lines and n = 3 for selected BILs and IL2-1 line). Asterisks indicate significant changes compared to S. lycopersicum samples as calculated by Student’s t test (*P value < 0.05; **P value < 0.01; ***P value < 0.001). LC-MS was used for targeted SGA profiling. The source data of Figs. 2c and 2d are provided as a Source Data file

Fig. 3

Discovery of the GAME31 candidate gene in hydroxytomatine-associated QTL region on tomato chromosome 2. a, b Hydroxytomatine (a) and acetoxytomatine (b) levels in 77 core IL population including parental lines. SGA content was determined from ground leaf tissue extracts (n = 3). The values represent the means of three biological replicates ± standard error mean (n = 3). LC-MS was used for targeted SGAs analysis. c Schematic representation of chromosomal regions in selected BILs (i.e. #6304, #6603, #6664, #6694, #6518) and the IL2-1 line carrying introgressions from the wild species S. pennellii. An overlapping QTL (between selected BILs and IL2-1 line) region located on chromosome 2 affecting the content of hydroxytomatine and acetoxytomatine revealed 17 annotated genes. A putative candidate gene termed GAME31 belonging to the 2-oxoglutarate dependent dioxygenase (2-ODD) family is depicted with a yellow block arrow along with three other 2-ODDs in the region. The remaining 11 genes are marked in green color, while the first and last genes in the region are shown with a white block arrow. Details regarding these 17 genes are provided in Supplementary Data 2. The source data of Figs. 3a and 3b are provided as a Source Data file

Fig. 4

GAME31 catalyzes the hydroxylation of the bitter flavor α-tomatine in tomato fruit. a GAME31 expression levels in leaves of selected BILs, IL2-1 line, and parents [S. lycopersicum (cv. M82) and S. pennellii] as determined by quantitative real-time PCR assay. The values indicate means of three biological replicates ± standard error mean (n = 3). Asterisks indicate a significant difference from S. lycopersicum samples calculated by Student’s t test (*P value < 0.05; **P value < 0.01; ***P value < 0.001). b, c Hydroxylation of α-tomatine (b) and dehydrotomatine (c) by the recombinant tomato GAME31 enzyme produced in E. coli cells (marked in red). Extracted ion chromatograms were obtained by LC-MS analysis (short 17-min run). The control reaction (shown in black) was performed with the respective substrate using extracts from E. coli cells transformed with an empty pET28 vector. m/z is shown for each substrate and hydroxylated product. Refer to Supplementary Fig. 11 for more information on the GAME31 enzyme assay with α-tomatine as a substrate. Details regarding the identification of enzyme assay products are provided in Supplementary Figs. 23 and 24. The structure of hydroxytomatine observed in the enzyme assay was elucidated using NMR analysis (refer to Supplementary Table 1). m/z mass to charge, E. coli Escherichia coli. d Red, ripe fruit of GAME31-silenced plants (RNAi; GAME31i) accumulated high levels of α-tomatine and dehydrotomatine as compared to wild-type fruit. Extracted ion chromatograms are shown. See Supplementary Fig. 15 for detailed quantitative analysis of SGAs in GAME31 RNAi and GAME31 co-suppression transgenic lines. e GAME31-overexpressing leaves (GAME31-Ox) accumulate hydroxytomatine isomers and acetoxytomatine, while these SGAs are present in minor quantities in wild-type leaves. Precursor SGAs, α-tomatine and dehydrotomatine, were reduced substantially in GAME31-Ox lines compared to wild type. Total ion chromatograms are shown. Lines #1 and #2 are two independent GAME31-Ox transgenic lines (#1 is shown here as a representative transgenic line). See Supplementary Figs. 16 and 18 for quantitative SGA analysis in GAME31-Ox transgenics with statistical information. The source data of Fig. 4a are provided as a Source Data file

Fig. 5

GAME32 hydroxylates the main potato SGAs to form CPB resistance-associated leptinines. a Schematic presentation of the 2-ODD family members GAME32 (black), GAME32-like (blue), and 2-ODD34 (yellow) genes in the S. chacoense (Sc) genome (accession M6). Truncated 2-ODDs are shown in white block arrows. b, c The recombinant ScGAME32-M6 enzyme converts α-chaconine (b) and α-solanine (c) to leptinine I (plus isomer) and leptinine II, respectively. Red: reaction assay with the respective substrate and the recombinant S. chacoense-M6 GAME32 protein (produced in E. coli); black: control reaction with the respective substrate and the protein extracts of empty vector-transformed E. coli. Mass to charge (m/z) is shown for all substrates and assay products. MS-MS analysis and identification of leptinines metabolites are provided in Supplementary Figs. 29–32. d De novo production of leptinines in cultivated potato hairy roots overexpressing ScGAME32-8380-1-1 (in red) or ScGAME32-8380-1-2 (in blue) as compared to control (in black) hairy roots (generated with an empty vector). e SGA profiles in leaves of wild type (non-transformed) and ScGAME32-8380-1-1 or ScGAME32-8380-1-2 overexpressing stably transformed cultivated potato plants determined by LC-MS. As observed for hairy roots, accumulation of leptinines was detected in the leaves of stably transformed cultivated potato plants. Extracted ion chromatograms are presented in all cases. Refer to Supplementary Fig. 22c, d for quantitative data on SGA analysis

Fig. 6

Genomic organization of 2-ODD family genes in selected Solanum species associated with SGA metabolism. Four 2-ODD genes, namely, GAME31 (red block arrow), GAME32 (black block arrow), GAME32-like (blue block arrow), and 2-ODD34 (yellow block arrow), were selected for functional characterization in this study. We proposed a function for GAME31 and GAME32 genes based on in vivo and/or in vitro experiments. While GAME31 and 2-ODD34 homologs are present in all Solanum species examined, the leptinine-producing GAME32 gene appears to be unique to S. chacoense, wild potato species that possess genetic CPB pest resistance. Cultivated potato lacks the GAME32 gene, resulting in absence of leptinines and leptine SGAs, and is therefore susceptible to CPB. Truncated 2-ODDs are shown in white block arrows. The arrowheads represent direction of transcription

Fig. 7

Pathways to unique defense metabolites and for evading fruit bitterness evolved through 2-ODD activity. a GAME proteins (GAME31 and GAME32) carrying out typical hydroxylation reactions in Solanum SGA metabolism and other putative candidates (GAME32-like and 2-ODD34) form distinct clades in the 2-oxoglutarate-dependent dioxygenase (2-ODD) family. Sequences from the following species were represented: cultivated tomato [S. lycopersicum, (Sl)], cultivated potato [S. tuberosum, (St)], wild tomato [S. pennellii, (Sp)], Capsicum annuum (Ca), cultivated eggplant [S. melongena, (Sm)], and wild potato [S. chacoense, (Sc)]. GAME31, GAME32, GAME32-like, and 2-ODD34 proteins forming separate clades are depicted in red, black, blue, and brown colors, respectively. Amino acid sequences used in the phylogenetic analysis are provided in Supplementary Data 3. Red dot denotes common ancestor shared between GAME32 and 2-ODD34 clade proteins. b A model presenting the impact of GAME31 and GAME32 proteins activity on tomato fruit flavor (bitter/sweet) and fruit toxicity as well as resistance of potato plants to CPB insect pest. Activity of GAME31 is required to modulate bitterness and toxicity in ripe tomato fruit. Green tomato fruit accumulates a bitter and toxic α-tomatine SGA. As the fruit matures reaching the red stage, the entire α-tomatine pool is converted to non-bitter and less toxic esculeosides. In cultivated tomato varieties, GAME31 performs the first reaction step (i.e. hydroxylation) that ensures degradation of α-tomatine and results in non-bitter and harmless ripe fruit. On the contrary, red ripe fruit of certain wild tomato species (e.g. S. lycopersicum var. cerasiforme) are bitter in flavor due to high accumulation of α-tomatine. These wild species are apparently altered in the process of α-tomatine catabolism. GAME32 activity in S. chacoense wild potato is vital for its genetic resistance against Colorado potato beetle (CPB). Through the action of the GAME32 enzyme, α-chaconine and α-solanine are first converted to leptinines and further to downstream leptines, respectively, in S. chacoense, which makes this wild species naturally resistant to CPB. The leptinine-producing GAME32 gene is absent in the cultivated potato genome, thus preventing formation of further leptines and this likely makes cultivated potato CPB susceptible