Dynamic diversification of lignan metabolism in sesame via coordinated oxygenation and glucosylation across germination Authors: Erisa Harada, Yukie Ohba, Eiichiro Ono, Jun Murata, Hiromi Toyonaga, Akira Shiraishi, Toshiaki Azuma, Toshiyuki Waki, Yuto Uegaki, Eri Okamoto, Atsushi Hoshino, Toru Nakayama, Tatsuya Wakasugi, Masayuki P. Yamamoto, Manabu Horikawa Published: 2026 ABSTRACT SignificanceLignan metabolism in sesame (Sesamum indicum) serves as a valuable model for investigating how plants adapt their specialized metabolism. Although previous studies have focused on seed development, metabolic changes occurring during germination remain poorly characterized. Herein, we identify cytochrome P450 enzymes and UDP-glycosyltransferases that coordinate oxidative and glucosylation steps in lignan transformation during germination. Results demonstrate how structurally similar but functionally distinct enzyme paralogs expand lignan diversity through broad substrate recognition and regio-specific modifications. The finding that sesame uses distinct enzymatic systems to achieve same metabolic outcomes in seeds and seedlings underscores evolutionary flexibility of plant metabolism. This work offers molecular insights into how plants regulate specialized metabolites in response to developmental cues and environmental changes. Sesame (Sesamum indicum L.) is among the earliest domesticated crops, with a cultivation history dating back over 5,000 y. Valued for its high oil content (~55% of the seed), sesame has been utilized for millennia as an agricultural resource (1, 2). In addition to oil, seeds are rich in lignan compounds (3–5), notably (+)-sesamin (SE), (+)-sesamolin (SN), and (+)-sesaminol triglucoside (SL-TG), which are well known for their antioxidant activity and associated health benefits (6–8). Lignan metabolism during S. indicum seed development has been intensively investigated over the past two decades (Fig. 1). The central precursor, (+)-pinoresinol, is a key entry point in the lignan biosynthetic pathway. It is synthesized via the highly stereoselective dimerization of coniferyl alcohol, a reaction precisely directed by dirigent (DIR) proteins that ensure the exclusive formation of the (+)-enantiomer (2, 9–11). Subsequently, (+)-pinoresinol is converted into SE through methylenedioxy bridge formation catalyzed by the cytochrome P450 monooxygenase (CYP) SiCYP81Q1 (12, 13). Further oxidation of SE by SiCYP92B14 occurs via direct aromatic hydration to produce (+)-sesaminol (SL), or via oxidative rearrangement to yield either SL or SN (14, 15). Through these alternative oxidation modes, SiCYP92B14 contributes to the structural diversification of specialized lignans. The resulting SL undergoes glycosylation by the uridine diphosphate (UDP)-sugar-dependent glycosyltransferase SiUGT71A9 to form (+)-sesaminol monoglucoside (SL-MG) (16). The glycosylation of SL-MG proceeds in a stepwise manner through the action of glycoside-specific glycosyltransferases (GGTs), and SL is ultimately converted into SL-TG (Fig. 1) (16, 17). Dynamic transition of lignan metabolism from seed development to germination. Seed development (upper part) and germination (lower part) stages are shown. Newly identified pathways in this study are highlighted by pink and red arrows, including (+)-sesamin (SE) -oxidizing CYPs (SiCYP706V12–V14) and (+)-episesaminone (EO) -glucosylating UGTs (SiUGT73E4 and SiUGT73CH10). During seed development, (+)-sesamin (SE) is derived from coniferyl alcohol, and UGTs involved in SL glycosylation (16, 17) are also indicated. In contrast, information on lignan metabolism during seed germination remains quite limited (18, 19). One of its most distinctive features is the unique metabolic transition, in which lipophilic lignans like SE and SN are entirely depleted to form water-soluble glycosylated structures. Preliminary studies have identified several glycosylated lignans derived from SE or SN, including SL diglucoside and SL triglucoside, (+)-episesaminone (EO) diglucoside and EO triglucoside, and (+)-sesamolinone diglucoside in germinating seeds (20). The detection of (+)-episesaminone-β(1→2)-diglucoside (EO-DG2) in defatted seeds (21) further suggests that EO or EO-derived glycosides are already present during seed development. However, the metabolic enzymes responsible for these lignan transformations during germination remain largely unexplored. Notably, a sesame mutant lacking functional SiCYP92B14 shows a selective reduction in the accumulation of SN, while maintaining SL-TG levels comparable to those of the control line in mature seeds. Since SiCYP92B14 exhibits catalytic activity in producing SN and SL from SE in vitro, this implies the presence of an additional (+)-sesaminol synthase gene in the sesame genome other than SiCYP92B14 (14). Recent genomic analyses of both cultivated and wild sesame have expanded genomic resources, providing valuable information for improving oil quality and opening a comparative framework that is increasingly informative for exploring the evolutionary diversification of lignan metabolism (22–25). In this study, we performed an integrated analysis of lignan metabolism by combining metabolite profiling with RNA sequencing and in vitro enzyme assays, leading to the identification of CYP and UGT proteins involved in the extensive modification of lignans during germination. By exploring these largely uncharacterized biosynthetic pathways through biochemical characterization of multiple enzymes, we further our understanding of the regulatory mechanisms governing specialized lignan metabolism from seed development through germination, and established a basis for future in vivo functional studies. Results Sesame Lignans are Highly Glycosylated Upon Germination. Given the lineage-specific lignan diversity in Sesamum spp., we reexamined the lignan profile of the modern sesame cultivar (cv. Masekin) using LC–MS and NMR to identify uncharacterized compounds and monitor compositional changes during germination. In mature seeds, we detected the major lignans (SE, SL, and SL-TG), along with reported glucosides such as (+)-sesaminol diglucosides (SL-DG2 and SL-DG6), (+)-sesamolinol 4’-O-β-D-glucosyl-(1→6)-O-β-D-glucoside (SNol-DG6), and (+)-episesaminone triglucoside (EO-TG), which has previously been reported mainly in germinating seeds. Additionally, we newly identified sesaminone triglucoside (SO-TG), a putative stereoisomer of EO-TG (SI Appendix, Fig. S1B). Germination induced a marked shift in lignan composition: levels of SE and SN decreased, while water-soluble lignan glucosides (20)—EO-TG, SL-TG, EO-DG2, SL-DG2, and SNO-DG2—increased (Figs. 2 A and B; SI Appendix, Fig. S1). Sesaminolin diglucoside (SLN-DG2) was also newly identified via NMR analysis. Among these, the diglucosides EO-DG2, SL-DG2, and SNO-DG2 increased from day two and plateaued by day three. In contrast, the triglucosides (EO-TG and SL-TG) peaked on day one and declined thereafter. These metabolic transitions likely reflect the sequential oxidation and glycosylation of SE and SN during germination (26). Lignan composition and transcriptional changes associated with metabolic transition during seed germination. (A) Temporal changes in lignan content during germination. The yellow line shows the sum of SE and SN; the blue line indicates the combined levels of EO-TG, SL-TG, EO-DG2, SL-DG2, SNO-DG2, and SLN-DG2; and the green line represents their total. Values represent the mean ± SD of the summed lignan contents across five biological replicates. (B) LC chromatograms of seed extracts from 0 to 3 d after imbibition (DAI), detected at 280 nm. Peak identities correspond to those shown in Fig. 1. A chromatogram of a standard mixture (Std. mix) of lignan metabolites is shown at the Bottom. RNA-seq–based expression profiles of representative CYP (C) and UGT (D) genes associated with lignan metabolism during germination. Expression levels are shown as TPM (transcripts per million). SiCPR1 serves as an internal reference. SiCYP81Q1 and SiCYP92B14 are included as references for genes predominantly expressed during seed development. Subsequently, we aimed to determine the complete structural features of the lignan glucosides, as the sugar linkage patterns had not been fully elucidated in prior studies (20). EO-TG, SL-TG, EO-DG2, SL-DG2, SNO-DG2, and SLN-DG2 were purified from three-day-old sesame seedlings for NMR analysis (SI Appendix, Figs. S34–S39). The NMR spectra for SL-TG, SL-DG2, and EO-DG2 were consistent with previous reports. EO-TG shared its sugar linkage pattern with SL-TG, while SNO-DG2 and SLN-DG2 matched those of EO-DG2 or SL-DG2. The structure of SLN-DG2 was characterized in this study. We confirmed that all of these lignans are retained in the cotyledons, hypocotyls, and radicles, and are not translocated to the true leaves after germination (SI Appendix, Fig. S2). The observed chemical diversity of lignans in germinating sesame seeds suggests the involvement of multiple lignan-modifying enzymes during germination. Screening of CYP and UGT Enzyme Genes Involved in Lignan Metabolic Transition Upon Germination. To identify the enzymes responsible for these metabolic transformations, we performed RNA-seq analysis at multiple germination stages. The decline in lipophilic SE and SN levels coincided with the accumulation of lignan glucosides two days after imbibition (Fig. 2 A and B; SI Appendix, Fig. S1), suggesting that CYP and UGT genes involved in this conversion are coinduced during this timeframe. Based on this observation, candidate CYP and UGT genes were selected according to coexpression with lignan accumulation during seed germination. Additional genes were included to ensure sufficient coverage, including SiCYP706V14 and SiUGT73CH10 based on genomic context, as well as UGT genes selected based on expression patterns during the later stages of seed development. This analysis identified six CYP genes (SiCYP706V12, SiCYP706V13, SiCYP706V14, XM011092788, XM011092789, and XM011092791) and five UGT genes (SiUGT73E4, SiUGT73CH10, SiUGT86C3, SiUGT88A5, and XM011102154) as candidates (Fig. 2 C and D; SI Appendix, Fig. S3). In addition, UGT proteins previously associated with seed development (17)–including those implicated in SL-TG biosynthesis (SiUGT71A9, SiUGT94AA2, SiUGT94D1, and SiUGT94AG1), as well as SiUGT94AB2, whose true substrate remains unidentified—were also expressed during germination (SI Appendix, Figs. S3–S5). In total, six CYPs and ten UGTs were selected for further biochemical characterization. Tandemly Arranged SiCYP706V12–V14 Genes Encode Enzymes Producing (+)-Sesaminol (SL) or (+)-Episesaminone (EO) from (+)-Sesamin (SE). To evaluate the enzymatic activity of the candidate CYP proteins, we performed bioconversion experiments using yeast as an expression system. Among six CYP candidates, three CYP proteins (designated as SiCYP706V12, SiCYP706V13, and SiCYP706V14) exhibited oxidation activity toward SE; SiCYP706V12 primarily produced SL, while SiCYP706V13 and SiCYP706V14 generated EO, as determined by LC–MS analysis (Fig. 3B; SI Appendix, Fig. S6). These three genes encode proteins that share approximately 80% amino acid identity and are located in tandem on Chr. 2 (22) (Fig. 3A). The SiCYP706V12–V14 genes belong to the CYP71 clan, the largest CYP clan in plants (27, 28). Phylogenetic analysis showed that SiCYP706V12–V14 are distinct from the previously identified sesame CYPs (SiCYP81Q1 and SiCYP92B14) but closely related to IrCYP706V proteins involved in diterpene-type oridonin biosynthesis in Isodon rubescens, a member of the same order, Lamiales (29) (SI Appendix, Fig. S7). Notably, the lignan-related CYP genes—SiCYP81Q1, SiCYP92B14, and SiCYP706Vs—do not form a biosynthetic gene cluster in the sesame genome but are located on different chromosomes (Chr. 4, Chr. 8, and Chr. 2, respectively) (30). Functional characterization of sesame CYP706V enzymes and UGT enzymes. The UGTs are involved in (+)-episesaminone-targeting glycosylation in sesame. (A) Genomic organization of Sesamum indicum CYP genes. Functionally characterized members, including SiCYP706V12–V14, are highlighted in black. (B) Oxidative reaction profiles of SiCYP706V12–V14 proteins. Yeast cells coexpressing each CYP with SiCPR1 were incubated with SE (Left) or SN (Right). HPLC chromatograms of enzyme reactions are shown (monitored at 280 nm). For each substrate (SE or SN), the Bottom chromatogram shows a standard mixture (Std. mix) containing the substrate and representative oxidation products. The corresponding chemical structure of these compounds are shown below the chromatograms. The oxidation profile is altered by the I215L/F245L/L377I mutations in SiCYP706V12. (C) Genomic layout of S. indicum UGT genes, with newly identified functional UGTs highlighted in black. (D) LC–MS analysis of recombinant SiUGT73E4, SiUGT73CH10, and SiUGT71A9 proteins acting on EO. EO-MG was detected as the primary product. (E) LC–MS analysis of recombinant SiUGT94AA2, SiUGT94AB2, and SiUGT94D1 proteins acting on EO glucosides. Multiple glucosylated products, including EO-DG2, EO-DG6, and EO-TG (peak 10), were detected. All reactions in panels (D and E) were monitored at 280 nm and compared with the substrate (control) and authentic standards. The chemical structures of the oxidation products were further characterized by NMR. SiCYP706V12 catalyzes oxidation at the C6 (or C6′) position on the aromatic ring, leading to the formation of SL (3). It also oxidized the symmetrically equivalent carbon on the opposite aromatic ring of SL, resulting in compound 18 (SE(OH)2) during the reaction (Fig. 3B). In contrast, SiCYP706V13 and SiCYP706V14 (SiCYP706V13–V14) targeted the C7 carbon of the furofuran ring, forming EO (4). Due to tautomerization, this compound gave two distinct peaks (4-1 and 4-2), corresponding to (+)-episesaminone (keto form) and 7-hydroxysesamin (lactol form), respectively. Collectively, these results demonstrate that SiCYP706V proteins act as (+)-sesamin-oxidizing enzymes, with SiCYP706V12 functioning as a (+)-sesaminol synthase, while SiCYP706V13 and SiCYP706V14 function as (+)-episesaminone synthases. SiCYP706V Proteins Display Activity Toward Multiple Sesame Lignans. Although both SiCYP706V proteins and SiCYP92B14 catalyze the oxidation of SE, their catalytic roles differ in the range of additional lignan substrates they can accept and in the oxidation products they generate. Notably, SiCYP706V proteins can oxidize both major sesame lignans—SE and SN—whereas SiCYP92B14 is selective for SE (14). Bioconversion assays showed that SiCYP706V proteins oxidize SN similarly to SE (Fig. 3B). Specifically, SiCYP706V12 targets the C6 position on the aromatic ring to produce product SLN (11), while SiCYP706V13–V14 act on the C7 position of the furofuran ring, yielding products SNO (12) and its tautomer. Furthermore, NMR analysis using 2H-labeled SE as a substrate demonstrated that SiCYP706V12 catalyzes direct oxidation at the aromatic ring (SI Appendix, Fig. S8). Analysis using structural isomers of SE and related lignans (SL, (+)-piperitol, and (+)-pinoresinol)—revealed that SiCYP706V proteins are capable of oxidizing all tested lignans. SiCYP706V12 converts SL into SE(OH)2 (also formed from SE) (Fig. 3B) and generates products 26 and 27 from (+)-piperitol and (+)-episesamin, respectively (SI Appendix, Fig. S9), consistently targeting the C6 (or C6′) carbon of its substrates for oxidation. In case of (+)-episesaminone synthases, SiCYP706V13 and SiCYP706V14 exhibited similar yet distinct activities (SI Appendix, Figs. S10 and S11). They both oxidized SL and (+)-piperitol at the C7 position of the furofuran ring, producing keto and lactol compounds (30-1, 30-2 and 31-1, 31-2). In the oxidation of SL, they also oxidized at C7′ position of the furofuran ring, resulting in only the keto-form compound 29. SiCYP706V14 uniquely prefers to oxidize the C7′ position of (+)-piperitol, forming additional products (35-1 and 35-2) (SI Appendix, Fig. S11). SiCYP706V13, in contrast, exhibited broader specificity, acting on (+)-pinoresinol to yield a distinct new peak (SI Appendix, Fig. S10). These results support the idea that catalytic site divergence among SiCYP706V proteins drives regio-selective lignan oxidation in sesame. Mutations in (+)-Sesaminol Synthase SiCYP706V12 Alter Catalytic Outcomes. To investigate the basis of the distinct catalytic outcomes among SiCYP706V isoforms, mutations were introduced into SiCYP706V proteins. These mutations altered catalytic outcomes and demonstrated that regio-specificity in oxidation sites directly influences the resulting product structures. The two (+)-episesaminone synthases, SiCYP706V13 and SiCYP706V14, share 84% amino acid sequence identity and 81%–82% identity with SiCYP706V12, the (+)-sesaminol synthase (SI Appendix, Fig. S12A), suggesting that a small number of polymorphic residues may underlie functional divergence. To further explore this possibility, we generated homology-based structural models of the SiCYP706V proteins and identified 13 residues near the putative substrate-binding pocket for site-directed mutagenesis. (SI Appendix, Fig. S12B). Biochemical assays of SiCYP706V12 single-residue mutants revealed that two mutants, SiCYP706V12_I215L and L377I, produced a distinct new peak (4-1) corresponding to EO, and SiCYP706V12_F245L yielded it as a minor product. Notably, SiCYP706V12_L377I also generated lactol form of EO (4-2) (SI Appendix, Figs. S13 and S14). These observations were further validated through bioconversion experiments using SN and (+)-piperitol as alternative substrates (SI Appendix, Figs. S15 and S16). Together, these findings demonstrate that single amino acid substitutions—specifically I125L, F245L, and L377I in SiCYP706V12—are sufficient to enhance the production of EO, in addition to the original product, SL. We further examined the effect of multiple amino acid substitutions. Notably, the triple mutant SiCYP706V12_I215L/F245L/L377I produced a higher quantity of EO-type products compared to SL, suggesting that residues I215, F245, and L377 play key roles in shifting SiCYP706V12’s catalytic preference (Fig. 3B; SI Appendix, Fig. S17). In contrast, none of the single-residue mutants of SiCYP707V13 altered its catalytic behavior. Considering that ancestral enzyme functions are often inferred from functional retrogression, i.e., the loss of newly acquired activity (31, 32), these results indicate a directional shift in catalytic properties within the SiCYP706V family, with the catalytic features represented by SiCYP706V13/14 being less derived than those of SiCYP706V12. Sun et al. previously proposed, based on time-calibrated phylogenetic analysis of CYP706 genes in Lamiaceae, that IrCYP706V5 may represents the closest orthologous counterpart of the sesame CYP706V lineage (29). Consistent with this proposal, our protein-based phylogenetic analysis places SiCYP706V14 as the closest sesame CYP706V member to IrCYP706V5 within the CYP706V clade (SI Appendix, Fig. S7). In addition, we observed that IrCYP706V5 also exhibits (+)-sesamin oxidation activity (SI Appendix, Fig. S18). Together, these functional and phylogenetic observations support the view that the ability to synthesize SL represents a more recent catalytic specialization within the sesame CYP706V family. Paralogous SiUGT73E4 and SiUGT73CH10 are Glycosyltransferases for (+)-Episesaminone (EO). Generally, GGT proteins catalyze regio-specific glycosylation at either the 2-OH or 6-OH position of the sugar moiety in glycosides and fall into the UGT79, UGT91, or UGT94 families (17). Based on this classification, we excluded four of the ten candidate genes from consideration as potential glycosyltransferases for the EO aglycone. The remaining six genes included SiUGT71A9, SiUGT73E4, SiUGT86C3, and SiUGT88A5—previously cloned from mature sesame seeds (16)—and two newly identified ones (SiUGT73CH10 and XM011102154). To evaluate their glycosylation activity toward EO, Escherichia coli cells expressing these UGTs were subjected to bioassays using EO as the substrate. SiUGT73E4, SiUGT73CH10, and SiUGT71A9 displayed glucosylating activity, producing a new peak 8 (Rt = 5.1 min), identified as EO-MG by NMR analysis (Fig. 3D). Notably, SiUGT73E4 and SiUGT73CH10 appear to be tandemly arranged twin genes on Chr. 9 (Fig. 3C). In the UGT protein phylogeny, SiUGT73E4 and SiUGT73CH10 fall within the UGT73 clade together with Lamiales UGTs such as snapdragon flavonoid 7-O-glucosyltransferase (AmUGT73CH1) and Perilla chalcone 2-O-glucosyltransferase (PfUGT73CH3) (33), whereas SiUGT71A9 belongs to the distinct UGT71 clade (SI Appendix, Fig. S19). Despite their clear phylogenetic separation, SiUGT73E4/CH10 and SiUGT71A9 share the ability to glucosylate EO, indicating that EO-accepting activity arose independently in distinct UGT families. We subsequently assessed the substrate specificity of these UGTs against a panel of lignans (SI Appendix, Fig. S20). Despite sharing over 75% amino acid identity, SiUGT73E4 and SiUGT73CH10 exhibited markedly different substrate preferences. SiUGT73CH10 had broader substrate specificity, glycosylating SNO, (+)-sesaminone (SO), SL, SLN, and EO. In contrast, SiUGT73E4 catalyzed the glucosylation of EO, which was the only substrate converted. The substrate preference of SiUGT73CH10 for keto-containing lignans followed the order SO, SNO, and EO, suggesting that the equilibrium states in solution may affect its reactivity (SI Appendix, Fig. S20). Additionally, SiUGT71A9, previously known to glucosylate SL, also acted on SLN, EO, and SNO. Finally, we evaluated the kinetic parameters of SiUGT73E4, SiUGT73CH10, and SiUGT71A9 toward EO (Table 1; SI Appendix, Fig. S23). SiUGT73CH10 exhibited the highest catalytic efficiency (kcat/Km) for this substrate, followed by SiUGT73E4, with SiUGT71A9 showing the lowest efficiency. In contrast, SiUGT71A9 exhibited substantially higher efficiency for SL and SLN than for EO (Table 1). These findings provide valuable insights into the metabolic flux of specialized lignans, which are coordinately regulated by multiple, functionally redundant enzymes in sesame seeds. Table 1.Kinetic parameters of UGT proteinsSubstrateEnzymekcat (s−1)Km (µM)kcat/Km (s−1.µM−1)For aglycon (+)-sesaminol (SL)SiUGT73CH100.093 ± 0.0016.6 ± 0.50.014SiUGT71A94.3 ± 0.623 ± 6.50.19 (+)-sesaminolin (SLN)SiUGT73CH100.088 ± 0.0045.3 ± 1.20.019SiUGT71A90.71 ± 0.078.2 ± 1.40.088 (+)-episesaminone (EO)SiUGT73CH100.024 ± 0.0019.1 ± 1.20.0027SiUGT71A90.0082 ± 0.003154 ± 960.000053SiUGT73E40.0053 ± 0.00027.6 ± 1.30.00073 (+)-sesamolinone (SNO)SiUGT73CH100.032 ± 0.00023.9 ± 0.290.0083For lignan glucoside EO-MGSiUGT94AB283 ± 13.3 ± 0.3125SiUGT94AA235 ± 231 ± 2.61.1SiUGT94D10.85 ± 0.023.8 ± 0.440.19 EO-DG6SiUGT94AB23.4 ± 0.531 ± 8.60.12 EO-DG2SiUGT94AA216 ± 249 ± 110.35SiUGT94D12.6 ± 0.2159 ± 200.016 SLN-MGSiUGT94AG10.58 ± 0.0270 ± 7.60.0087SiUGT94D12.4 ± 0.0062.6 ± 0.130.91 SNO-MGSiUGT94AB235 ± 0.96.9 ± 0.215.1SiUGT94AA22.7 ± 0.154.7 ± 5.50.050SiUGT94D10.64 ± 0.051.37 ± 0.140.047Concentrations of a countersubstrate: [UDP-glucose] = 1 mM. Kinetic parameters of UGT proteins Concentrations of a countersubstrate: [UDP-glucose] = 1 mM. Regio-Specific Glycosylation Activities for (+)-Sesamin-Derivative Glucosides by SiUGT94 Proteins. Next, we assessed the glycosylation activity of four GGT proteins—SiUGT94AB2, SiUGT94AA2, SiUGT94D1, and SiUGT94AG1—toward EO glucosides using UDP-glucose and lignan glucosides as potential cosubstrates. SiUGT94AB2 converted EO-MG (8, Rt = 5.2 min) into EO-DG2 (9, Rt = 4.4 min) and EO-DG6 (23, Rt = 4.5 min) into EO-TG (10, Rt = 3.8 min) (Fig. 3E); however, it did not glycosylate EO-DG2. These results indicate that SiUGT94AB2 mediates β(1→2)-O-glucosylation, but not β(1→6)-O-glucosylation, of EO glucosides. Although SiUGT94AG1 is known to catalyze β(1→2)-O-glucosylation in SL-MG and SL-DG6 reactions, it showed no activity toward EO-MG, EO-DG2, or EO-DG6 (SI Appendix, Fig. S21), suggesting that the aglycone structure substantially affects substrate recognition, emphasizing the role of UGT94AG1 in SL-TG biosynthesis. Conversely, SiUGT94AA2 and SiUGT94D1 glycosylated EO-MG and EO-DG2 to yield EO-DG6 (23) and EO-TG (10), respectively (Fig. 3E). However, neither SiUGT94AA2 nor SiUGT94D1 acted on EO-DG6. Collectively, these results confirm that both GGTs mediate sugar–sugar β(1→6)-O-glucosylation of EO glucosides, in addition to their activity on SL glucosides. This conclusion is further supported by molecular phylogenetic analysis of the UGT94 family (SI Appendix, Fig. S19). SiUGT94AG1 and SiUGT94AB2, tandemly arranged on Chr. 6, share high similarity and exhibit regio-specific β(1→2)-O-glucosylation activity, while SiUGT94AA2 (Chr. 12) and SiUGT94D1 (Chr. 7) and are located on different chromosomes (Fig. 3C) and catalyze regio-specific β(1→6)-O-glucosylation. During germination, SE and SN are metabolized by SiCYP706V proteins into two distinct lignans types: SL and EO-type products. Accordingly, we assessed the substrate specificities of the four GGT proteins using lignan glucosides featuring either furofuran-ring or open-chain keto structures (SI Appendix, Fig. S22). SiUGT94AB2 (β(1→2)-O-glucosylation) exhibited strong activity toward open-ring EO-type glucosides such as EO-MG, EO-DG6, and SNO-MG, generating EO-DG2, EO-TG, and SNO-DG2, respectively. SiUGT94AG1, another β(1→2)-O-glucosylationing enzyme, preferentially acted on furofuran-ring-containing substrates such as SL-MG, SLN-MG. In contrast, SiUGT94AA2 and SiUGT94D1, catalyzing β(1→6)-O-glucosylation, accepted both ring type substrates, indicating broader substrate specificity. They glycosylated SL-MG, EO-MG, EO-DG2, and SNO-MG to form SL-DG6, EO-DG6, EO-TG, and SNO-DG6, respectively. Additionally, SiUGT94D1 recognized SLN-MG and produced SLN-DG6. These patterns contrast with the regio-specific β(1→2)-O-glucosylating GGTs (SiUGT94AB2 and SiUGT94AG1), which exhibit distinct substrate preferences. These data reveal that sesame employs functionally complementary GGTs with distinct regio- and substrate preferences to construct diverse lignan glycosides during germination. To estimate the dominant metabolic flux in sesame seeds, we calculated the kinetic parameter kcat/Km values for each GGT recombinant protein using lignan glucosides (Table 1), excluding undetectable products (SLN-DG2, SLN-DG6, SNO-DG2, and SNO-DG6). SiUGT94AB2 showed the highest catalytic efficiency for EO-MG, indicating that it is the most efficient GGT enzyme for this substrate. For EO-DG2, SiUGT94AA2 exhibited a kcat/Km value an order of magnitude higher than that of SiUGT94D1, indicating it is the primary enzyme. Regarding SNO-MG, SiUGT94AB2 again had the highest efficiency. For SLN-MG, SiUGT94D1 had a kcat/Km value two orders of magnitude higher than that of SiUGT94AG1, suggesting it is the most efficient enzyme for SLN-MG. However, the absence of SLN-DG6 in germinating sesame seeds suggests that SiUGT94AG1 plays primarily glycosylates SLN-MG in vivo (SI Appendix, Fig. S23). SiCYP706V12 Is the (+)-Sesaminol Synthase During Seed Development. We previously reported that SL glucosides continued to accumulate in mature seeds of CYP92B14-defective lineages (12), suggesting the presence of another (+)-sesaminol synthase in sesame. Since SL and EO are synthesized by SiCYP706V proteins, we speculate that these proteins function not only during germination but also throughout seed development. To validate this observation, we performed RNA-seq analysis using newly generated data from developing seeds sampled at defined stages (14–49 d after flowering, DAF). RNA-seq analysis revealed that SiCYP706V12 is expressed during seed development, while SiCYP706V13 and SiCYP706V14 are not. Expression of SiCYP706V12 is upregulated at later stages of seed development (35–42 DAF) and temporally coexpressed with UGT genes involved in SL-TG biosynthesis (SI Appendix, Fig. S4), supporting its role as an alternative (+)-sesaminol synthase in sesame seeds. To explore this further, we compared lignan profiles with RNA-seq data from SL-TG and SL-TG-deficient lineages (SI Appendix, Table S1). In both types, we observed a coordinated upregulation of SiCYP706V12 and UGT genes associated with SL-TG biosynthesis at the late developmental stage. Notably, there was a temporal difference of approximately one-week in gene upregulation between SL-TG-containing lineages (#1381, #4294) (DAF 35–49) and SL-TG-deficient lineages (#4125, #30631) (DAF 42–49) (Fig. 4A; SI Appendix, Figs. S24–S26). Importantly, we identified two single-nucleotide polymorphisms (SNPs) in the SiCYP706V12 gene between these lineages (SI Appendix, Fig. S27). One SNP led to an amino acid change at Cys454, a residue highly conserved for heme coordination, likely impairing catalysis by disrupting the heme-coordinating cysteine (34, 35). As expected, the SiCYP706V12_C454Y mutant failed to oxidize SE, confirmed by a bioassay using a yeast heterologous expression system (Fig. 4B; SI Appendix, Fig. S28). Since SL-TG-deficient lineages still retain functional SiCYP92B14 for SN biosynthesis, these results underscore that SiCYP706V12, rather than SiCYP92B14, serves as the (+)-sesaminol synthase responsible for SL-TG accumulation during seed development. Role of SiCYP706V12 in lignan metabolism across sesame seed development and germination. (A) LC chromatograms of sesame seeds at 42 d after flowering from lines with (#1381, #4294) and without (#4125, #30631) SL-TG accumulation. The Bottom chromatogram shows a standard mixture (Std mix). (B) Enzymatic assays of wild-type and C454Y mutant SiCYP706V12 expressed in yeast using (+)-sesamin as substrate. The C454Y variant was identified through SNP analysis. The Bottom chromatogram shows a standard mixture (Std. mix). (C) Protein–protein interactions between lignan-related CYP proteins and SiUGT71A9 analyzed by split-ubiquitin yeast two-hybrid assay. Yeast cells coexpressing SUC–X–Cub–LexA–VP16 and NubG-Y fusions were grown on selective media (–W, –L, –H, –Ade, +7 mM 3AT), where “X” represents one of the five CYP proteins (SiCYP706V12, SiCYP706V13, SiCYP706V14, SiCYP92B14, or SiCYP81Q1) or no protein, and “Y” represents one of the following: PC (positive control, Ost1-NubI), no protein, or SiUGT71A9. (D) Schematic summary of yeast two-hybrid–detected protein–protein interactions among sesame lignan-related CYP, UGT, and CPR proteins. Arrows indicate detected interactions; absence of arrows indicates no interaction was observed under the conditions tested. Proteins are color-coded as follows: CYPs, yellow; UGTs, purple; CPR1, gray. (E) Dynamic transition of lignan metabolism from seed development to germination in sesame seeds. Schematic summary of enzymatic reactions involved in lignan oxidation and glucosylation across developmental stages. During seed development, (+)-sesamin (SE) is partially oxidized by SiCYP92B14 to form SN and SL (14). During germination, residual SE and SN are fully oxidized by SiCYP706V12–V14, followed by glycosylation by UGT enzymes. CYPs and UGTs are represented as dark and light gray rectangles, respectively. Arrow thickness indicates relative metabolite abundance. Key lignans are abbreviated as follows: SN, (+)-sesamolin; SL, (+)-sesaminol; EO, (+)-episesaminone; SLN, (+)-sesaminolin; SNO, (+)-sesamolinone; “derivs” denotes derivatives. The formation of metabolons—enzyme assemblies mediating sequential reactions—has been proposed as one mechanism that can facilitate the efficient processing of reactive or unstable intermediates, as demonstrated in the biosynthesis of cyanogenic glycosides (36) and lignins (37). In sesame, we previously proposed a putative enzyme assembly involving SiCYP81Q1 and SiUGT71A9/94AA2/94AG2 for SL-TG biosynthesis, based on protein–protein interaction data using a yeast two-hybrid system (17). Notably, SiUGT71A9 physically interacted with SiCYP706V12 (Fig. 4C), whereas no interaction was detected with SiCYP92B14, consistent with our previous report (17). This interaction pattern suggests that SiCYP706V12 may participate in a specialized enzyme assembly associated with SL-TG biosynthesis, as summarized in Fig. 4D based on the interaction data shown in Fig. 4C and SI Appendix, Figs. S32 and S33, thereby distinguishing this putative assembly from the sesamolin/sesaminol pathway mediated by SiCYP92B14. Furthermore, the observation that heterologously expressed SiCYP706V12 generates a minor amount of EO alongside the predominant product, SL (SI Appendix, Fig. S29), implies that a small amount of EO may also be formed from SE by SiCYP706V12 during seed development. Discussion Our study reveals a dynamic, developmentally regulated metabolic cascade in sesame, wherein lipophilic lignans undergo structural diversification through oxidation and glycosylation. This cascade operates across both seed development and germination and is mediated by distinct but partially redundant CYP and UGT enzymes (Figs. 1 and 4E). These findings underscore how metabolic plasticity in sesame is driven by enzyme duplication, functional divergence, and cooperative interaction within the metabolic network. In particular, the emergence of SiCYP706V12–14 as germination-specific (+)-sesamin oxidases with complementary regio-selectivities reflects an adaptive expansion of catalytic capacity. Among these, SiCYP706V12 is functionally distinct in also acting during seed development, as demonstrated by its expression profile and its essential role in SL-TG biosynthesis, confirmed through loss-of-function analysis (Figs. 4 A and B). Its physical interaction with SiUGT71A9—an enzyme previously linked to SL glycosylation in developing seeds—suggests that these enzymes may function as part of a specialized assembly, potentially facilitating the efficient processing of intermediates through a dedicated pathway. Beyond oxidation, lignans undergo sequential glycosylation by UGT enzymes with distinct expression patterns and substrate specificities. During seed development, SL-TG biosynthesis involves SiUGT71A9 and GGTs such as SiUGT94AG1 (β(1→2)-O-glucosylation) and SiUGT94AA2/SiUGT94D1 (β(1→6)-O-glucosylation). The role of SiUGT94AG1 is supported by mutant (S6-32) analysis revealing SL-DG6 and EO-TG accumulation in SiUGT94AG1–deficient (17) lines (SI Appendix, Fig. S30). During germination, newly induced UGTs—SiUGT73E4 and SiUGT73CH10—glucosylate the keto-type aglycone EO, which is produced by SiCYP706V13/V14. A physical interaction between SiCYP706V13/V14 and SiUGT73CH10 was detectable in yeast two-hybrid assays, as summarized in Fig. 4D. Despite keto-hemiacetal tautomerism, glycosylation occurs specifically at the keto form, indicating enzymatic discrimination that stabilizes reactive intermediates. Although SiUGT73CH10 displays higher catalytic efficiency, SiUGT73E4 likely plays a substantial role when coexpressed with the corresponding CYP genes (SI Appendix, Fig. S31). Both enzymes work in coordination with GGTs—SiUGT94AB2, SiUGT94AA2, and SiUGT94D1—that catalyze region-specific modifications of the sugar moiety. The latter two GGTs act on both SL (17) and EO glucosides, illustrating substrate versatility and pathway convergence across developmental stages, as integrated in the overall metabolic framework shown in Fig. 1 (see also SI Appendix, Fig. S23). Similar enzymatic multifunctionality is observed in other plants, such as CsUGT73AC15 in tea flavonol triglycoside biosynthesis (38). The accumulation profiles of lignan glycosides suggest that glycosylation regulates both biosynthesis and turnover. Diglucosides such as SL-DG2 and EO-DG2 persist throughout germination, whereas triglucosides like SL-TG and EO-TG peak early and subsequently decline (Figs. 2 A and B; SI Appendix, Fig. S1). This pattern may reflect limited GGT expression (Fig. 2D) or selective degradation, with SL-DG2 possibly accumulating due to resistance to β-glycosidase (4). Thus, glycosylation patterns likely modulate the metabolic stability and fate of lignans during seedling establishment. Our results revealed that S. indicum possesses at least four CYP enzymes capable of oxidizing the lipophilic lignans, SE and SN, and three UGT enzymes initiating glucosylation (Figs. 1 and 4E; SI Appendix, Fig. S23). It is of evolutionary interest that structurally unrelated yet functionally redundant enzymes, namely CYP92B/CYP706V and UGT71A/UGT73E/UGT73CH, have been independently recruited into the same metabolic pathway. Rather than overall structural similarity, local structural affinity for the same metabolite within the substrate pockets of distinct enzymes may underlie the emergence and diversification of metabolic pathways. This redundancy among (+)-sesamin-oxidizing CYP enzymes resembles that in maize diterpenoid defenses (39), berberine (40), strigolactone (41, 42), and diosgenin biosynthesis (43). These examples support convergent evolution of metabolic functions among distinct CYP families (28). CYP706 proteins exhibit pronounced catalytic specialization—as seen in CYP706A3 (volatile sesquiterpene emission in Arabidopsis) (44), ZmCYP706A3 (45) and OsCYP706C2 (46) (strigolactone biosynthesis). In Isodon, CYP706V gene diversification was driven by tandem gene duplication of ancestral CYP706G/C genes (29), a process likely shared by sesame. Functional divergence between IrCYP706V2/7 (ent-kaurene) and SiCYP706V12/13/14 ((+)-sesamin derivatives) exemplifies CYP-driven metabolic innovation within Lamiales. In contrast, functional data for CYP92-related genes remains scarce, with recent findings limited to benzoxaninoids metabolism (47, 48). Uncovering the complete lignan metabolic pathway provides valuable insights into the physiological significance of these specialized metabolites. Germination, a metabolically dynamic phase (49, 50, 51), entails rapid water uptake, with seed moisture increasing from 5%–8% in mature seeds (52) to over 40% within 48 h of imbibition (53), coinciding with extensive changes in lignan glucosylation. Although similar hydration dynamics upon germination occur in legumes and rice (54), sesame converts lipophilic lignans entirely into glucosides. In contrast, soybean isoflavones (55) and monolignans in Archtium lappa (56) undergo transient interconversion during germination. This distinctive metabolic transition in sesame likely reflects multiple physiological roles: 1) hydrophobic lignans may enhance seed durability during maturation; 2) rapid removal of these compounds postgermination prevents potential toxicity to developing seedlings; and 3) lignan metabolites might be recycled as carbon sources via unidentified salvage pathways (57) or contribute to microbiome establishment in emerging seedlings (58). Glucosylation of specialized metabolites in plants is often associated with developmental or environmental responses (59), and sesame lignan profiles follow this paradigm. For example, drought stress during flowering and seed maturation shift the lignan profile toward glycosides (60), likely reflecting to oxidative signals. Several lignans also have allelopathic effects. SE and related compounds suppress germination or seedling development in Lactuca and ryegrass (61, 62). Lipid-storing seeds (e.g., cucumber, peanut) are more sensitive to SE and (+)-fargesin than carbohydrate-storing seeds (e.g., rice) (63). SE cause developmental defects in A. thaliana when ectopically coexpressed with steroleosin, which binds to (+)-sesamin (64). In mature sesame seeds, SE colocalize with lipids, possibly interfering with delta-5 desaturase activity in polyunsaturated fatty acid biosynthesis (65), suggesting a conserved inhibitory mechanism. The CYP706V proteins and multiple UGT enzymes play essential roles in lignan transformation, promoting structural diversification and facilitating adaption to changing water and metabolic demands during germination. Although seed maturation and germination are typically treated as distinct phases, our findings reveal a continuous metabolic cascade coordinated by CYP and UGT enzymes. Expansion of CYP (>300) and UGT (>200) genes in the S. indicum genome (SI Appendix, Supporting Methods) underpins this metabolic plasticity. The catalytic preferences of upstream enzymes shape downstream metabolism (66) and may promote tandem expansion of versatile genes—a hallmark of specialized metabolism (67). The evolution of analogous enzymes across divergent plant lineages underscores the dynamic nature of plant metabolic adaption (66). This study demonstrates how coordinated action between CYP and UGT enzymes integrates specialized metabolism with both developmental and environmental signals. Materials and Methods Plant Material. Plants of Sesamum indicum cv. Masekin were cultivated in a field in Osaka and also grown in a greenhouse at the University of Toyama. Immature seeds were harvested at defined stages of seed development (14–49 d after flowering, DAF), as described below. To induce germination, seeds of S. indicum cv. Masekin were placed on filter paper in 90 mm Petri dishes containing 1.5 mL of water and incubated in the dark at 28 °C. Germinated seeds were collected at several time points after imbibition. Sesame strains (#1381, #30631, #4125, and #4294) were grown in a greenhouse at the University of Toyama. Chemicals. (+)-Sesaminol (SL) and its glucosides were purchased from Nagara Science (Japan), while UDP-glucose was obtained from Nacalai Tesque (Japan). (+)-Sesamin (SE), (+)-sesamolin (SN) and (+)-2,2′-2H2-sesamin were prepared according to previously reported methods (14, 15). The 2H labeling yield at the C2 and C2′ positions of SE was 70%, determined by comparing the 1H NMR signal intensities of H-2,2′ and H-9a,9a′. (+)-Episesaminone (EO) was synthesized from SE via treatment with dimethyldioxirane in acetone for 20 h at room temperature, as detailed in the Supporting Information. EO-DG2 and EO-TG were isolated from sesame seeds, whereas EO-MG and EO-DG6 were enzymatically synthesized using recombinant SiUGT71A9 and SiUGT94AA2, respectively, expressed in an E. coli system. Detailed procedures are available in the SI Appendix, Supporting Methods. SNO-MG and SLN-MG were generated by bioconversion of SN using either SiCYP706V13, SiUGT73E4, and SiCPR1 or SiCYP706V12, SiUGT71A9, and SiCPR1, coexpressed in a yeast system, as described later. NMR Analyses of Stable Isotope-Labeled Compounds. NMR spectra were recorded at ambient temperature on Bruker AVANCE III HD spectrometers (Bruker Biospin AG, Switzerland) operating at 800 or 400 MHz, equipped with 5-mm TCI cryogenic and BBO probes, respectively. Samples (0.4 to 1 mg) were dissolved in CDCl3. Data were analyzed using Bruker TopSpin 3.2 software (Bruker Biospin AG, Switzerland). The acquisition parameters were as follows: number of data points, 65K; spectral width, 16,025 Hz; relaxation delay, 10 s; number of scans, 64; receiver gain 3.2. Lignan Profile Analyses. Sesame seeds or seedlings were freeze-dried, and 10 mg of dried material was ground into a fine powder using a TissueLyser II (Qiagen, Hilden, Germany). One milliliter of 70% acetonitrile aqueous solution was added to the homogenized samples. The extract was filtered after centrifugation. All extractions were performed using five independent biological samples. The resulting filtrate was analyzed using an ion-trap time-of-flight mass spectrometer (Shimadzu LCMS-IT-TOF, Shimadzu Corp., Japan) equipped with a photodiode array detector (Shimadzu, Japan). Chromatographic separation was performed on a YMC Triart C18 column using mobile phase A (10 mM aqueous ammonium acetate) and mobile phase B (acetonitrile or methanol). Four LC methods (A–D) were employed. Methods A and C used acetonitrile as mobile phase B, whereas Methods B and D used methanol. Methods A and B were carried out on a YMC Triart C18 column (TA12S03-1503PTH, 150 mm × 3 mm i.d., 3 µm) with a linear gradient program of 0–10–25–35–35.1–40 min. The gradient composition was 10–40–80–80–10–10% B for Method A and 30–50–90–90–30–30% B for Method B. The flow rate was 0.3 mL/min. Methods C and D were performed on a YMC Triart C18 column (TA12S03-L503PTH, 75 mm × 3 mm i.d., 3 µm) with a linear gradient program of 0–5–12.5–15–15.1–18 min. The gradient composition was 10–40–80–80–10–10% B for Method C and 30–50–90–90–30–30% B for Method D. The flow rate was 0.4 mL/min for Method C and 0.3 mL/min for Method D. Homology Modeling. Three-dimensional structural analyses were performed using Discovery Studio 4.0 (BIOVIA, CA). Initial structural models of SiCYP706V proteins were generated using the crystal structure of the steroidal inhibitor-bound form of human CYP17A1 as a template (PDB code: 3SWZ), selected through Discovery Studio’s modeling module based on the highest score. SE-bound structures of SiCYP706V proteins were created by replacing the template substrate with SE. These constructed structures were energy-minimized and optimized via molecular dynamics simulation. RNA-Seq Analysis. Total RNA from S. indicum cv. Masekin seeds during germination (1–5 d after imbibition, DAI) was isolated using the RNeasy Plant Kit (Qiagen, Hilden, Germany) and subjected to cDNA library construction using the Truseq mRNA sample preparation kit (illumina, CA) according to the manufacturer’s instructions. Single-end sequencing (101 cycles) was performed for each cDNA library using a Hiseq 1500 (illumina) in rapid mode. The resulting RNA-seq data were deposited in the NCBI BioProject under accession number PRJNA1421054. In addition, RNA from four sesame strains (#1381, #30631, #4125, and #4294) collected at multiple stages during seed development (14, 21, 28, 35, 42, and 49 d after flowering, DAF) was isolated using the RNeasy Plant Kit and subjected to cDNA library construction using MGIEasy RNA Directional Library Prep Set (MGI Tech, China) according to the manufacturer’s instructions. Paired-end sequencing (150 cycles) was performed using a DNBSEQ G-400 platform (MGI). The resulting RNA-seq data were deposited in the NCBI BioProject under accession number PRJNA1019945. RNA-seq reads were mapped to the S. indicum reference genome assembly (GCF_000512975.1) using HISAT2 version 2.2.0 (68), together with previously reported RNAseq data from S. indicum cv. Masekin seeds at six developmental stages (BioProject PRJNA350858) (14). Transcript abundance was quantified as transcripts per million (TPM) using StringTie version 2.2.1 (69). The criteria used for screening CYP and UGT candidate genes are described in the SI Appendix. Molecular Cloning and Vector Construction of SiCYP and SiUGT Proteins. cDNAs encoding CYP and UGT proteins, amplified from S. indicum cv. Masekin seed-derived cDNA, were used as PCR templates. As described in previous studies, amplified CYP and SiCPR1 cDNA fragments were subcloned into the yeast expression vectors pYE22m and pJHXSBP-His, respectively (12). For UGT proteins, amplified DNA fragments were inserted into the E. coli expression vector pET15b, which carries an N-terminal His6-tag (16, 17). Amplified DNA fragments of SiUGT71A9 and SiUGT73E4 were subcloned into the yeast expression vector pJHXSBP-URA. All vector-inserted genes were validated by sequencing both strands. The expression vector for SiUGT74CH10 was constructed by gene synthesis, optimized for heterologous expression in E. coli (Genscript Biotech Corp., Piscataway, NJ). Expression vectors for SiCYP706V12 and SiCYP706V13 mutants were generated from the wild-type sequences by PCR-based mutagenesis, as previously described (13, 70). All primer sets used for PCR are listed in SI Appendix, Table S2. Phylogenetic Analysis. Phylogenetic analyses of CYP and UGT proteins are described in the SI Appendix. Bioconversion Assay in Yeast Cells. All transformations were carried out using the yeast INVSc strain (Invitrogen, Carlsbad, CA) following standard protocols. Transformants coexpressing CYP and SiCPR1 proteins were cultured in synthetic defined liquid medium (Sigma-Aldrich, St. Louis, MO) at 30 °C. A 50 μL aliquot of stationary-phase culture was transferred into 500 μL of fresh medium in a 24-well plate containing 100 µM substrate. The optimal incubation time varied by substrate. All assays, including candidate gene identification, multiple-substrate testing, and mutant activity evaluation, were performed with two independent transformants (n = 2). After 5 to 16 h, an equal volume of acetonitrile (final concentration: 50%) was added, and the cultures were centrifuged. The filtered supernatants were analyzed by HPLC equipped with photodiode array and fluorescence detectors (Shimadzu LC Nexera X2, Shimadzu Corp., Japan), and the identities of the products were further confirmed by LC–MS using an ion-trap time-of-flight mass spectrometer (Shimadzu LCMS-IT-TOF, Shimadzu Corp., Japan) equipped with a photodiode array detector (Shimadzu, Japan). Collection of Oxidation and Glucosylated Products From Large-Scale Yeast Cultures. For NMR-based structure elucidation or the preparation of substrates for UGT kinetic assays, 150 to 450 mL of yeast culture was used. Fifty mL of overnight-grown yeast cells expressing CYP and SiCPR1 were harvested and resuspended in ~15 mL of fresh medium, then inoculated into 150 mL of liquid medium in a 1 L Erlenmeyer flask. The substrate was added to a final concentration of 50 µM. After 3 to 6 h (or up to 16 h for UGT assays), 100 µL of culture was sampled for HPLC-based monitoring for product formation. This large-scale assay was performed once for product isolation. For oxidation products, the reaction was terminated by the addition of ethyl acetate. The organic layer was dried over magnesium sulfate. After evaporation under reduced pressure, the residue was reconstituted in 2 mL of 50% acetonitrile and filtered. The resulting solution was purified by HPLC to isolate the target compounds. For glucosylated products, yeast cultures coexpressing UGT together with CYP and SiCPR1 were quenched by the addition of acetonitrile to a final concentration ~10%. Insoluble material was removed by filtration through Celite. The filtrate was subjected to preliminary purification using ODS column chromatography, followed by HPLC to yield the purified products. All final products were analyzed by MS and NMR. 2H-Labeled SE Oxidation in Yeast Cells. Yeast cells coexpressing SiCYP706V12 and SiCPR1 were incubated with 2H-labeled SE, and the resulting oxidation products were analyzed by NMR spectroscopy. Detailed NMR analyses are provided in SI Appendix, Fig. S8. Heterologous Expression of SiUGT Proteins in E. coli. The plasmids pET15b-SiUGT71A9, pET15b-SiUGT73E4, pET15b-SiUGT73CH10, pET15b-SiUGT94AA2, pET15b-SiUGT94AB2, pET15b-SiUGT94D1, and pET15b-SiUGT94AG1 were transformed into E. coli BL21(DE3) cells (New England Biolabs Inc., Ipswich, MA). Transformants were precultured at 37 °C for 3 to 4 h in LB medium supplemented with 50 μg/mL ampicillin. A 1 mL aliquot of the culture was transferred into 100 mL of Overnight Express Autoinduction medium (Novagen, Darmstadt, Germany) with ampicillin and incubated at 25 °C, shaking at 125 rpm for 20 to 24 h. Recombinant cells were harvested by centrifugation (9,500 × g, 10 min), washed with distilled water, transferred to a 15 mL conical tube, and centrifuged again (8,000 rpm, 5 min). Purification of SiUGT Proteins. UGT protein extraction followed previously reported methods with slight modifications (17). Harvested cells were suspended in BugBuster Master Mix (Merck KGaA, Darmstadat, Germany) containing 0.1% (v/v) β-mercaptoethanol and a protease inhibitor cocktail (Nacalai Tesque, Japan) at room temperature. The suspension was shaken for 15 min at room temperature. All subsequent steps were performed at 0 to 4 °C. Cell debris was removed by centrifugation (17,000×g, 15 min). Polyethyleneimine was added to the supernatant to a final concentration of 0.12% (v/v), and the solution was allowed to stand for 30 min before centrifugation (17,000×g, 15 min). Recombinant protein purification was performed using His-Spin Trap and HiTrap Q HP (5 mL) columns (Cytiva, Marlborough, MA). At each purification stage, eluted fractions were assayed for glucosyltransferase activity using an ion-trap time-of-flight mass spectrometer (Shimadzu LCMS-IT-TOF, Shimadzu Corp., Japan). Detailed protocols are available in the SI Appendix, Supporting Methods. Protein concentrations were determined using the Bradford assay with bovine serum albumin as the standard. Enzyme Assay of SiUGT Proteins. The standard reaction mixture (50 μL) contained 40 μM glucosyl acceptor, 1 mM UDP-glucose, 100 mM potassium phosphate buffer (pH 7.5), and UGT enzyme. The mixture, minus the enzyme, was preincubated at 30 °C for 10 min, and the reaction was initiated by enzyme addition. After 2 min, the reaction was quenched by adding 50 μL of acetonitrile. For activity screening, a UGT enzyme concentration of 500 nM was used to identify activities toward EO, and this experiment was performed once. In assays of EO and EO-related lignans, 100 nM of each UGT enzyme was used, and for glycosylation of glucosylated lignans by GGT proteins, the enzyme concentration was also 100 nM. Each of these reactions was performed in triplicate (n = 3). Reaction products were analyzed using an ion-trap time-of-flight mass spectrometer (Shimadzu LCMS-IT-TOF, Shimadzu Corp., Japan) on an ODS column (YMC Triart C18, 100 × 2.1 mm I.D. S-3 μm, 12 nm). Substrates and products were separated with a linear gradient of 20%–80% (v/v) acetonitrile containing 0.1% (v/v) formic acid at a flow rate of 0.3 mL/min and detected at 280 nm using a diode array detector. The same conditions were used to determine UGT kinetic parameters, except the sugar acceptor concentration was varied. Apparent Km and Vmax values for each acceptor were determined under saturating UDP-glucose levels. Initial velocity data were fitted to the Michaelis–Menten equation by nonlinear regression using Prism 8 software (GraphPad Software, Japan). For kinetic analyses, six sugar concentrations were used to obtain saturation curves, and each measurement was performed independently three times. Yeast Two-Hybrid Assays With the Split-Ubiquitin System. To assess interactions between SiCYP proteins (SiCYP706V12, SiCYP706V13, SiCYP706V14, SiCYP81Q1, and SiCYP92B14) and UGT proteins (SiUGT71A9, SiUGT73CH10, and SiUGT73E4), the split-ubiquitin membrane yeast two-hybrid system was employed using the DUALmembrane Kit 3 (Dualsystems Biotech, Zurich, Switzerland), as described previously (17, 71). Full-length CYP sequences (excluding stop codons) were cloned into the SfiI sites of the pBT3-SUC vector to generate N-terminal SUC-tagged fusions with Cub–LexA–VP16 at the C-terminus. UGT coding sequences were similarly inserted into the SfiI sites of the pPR3-N vector to produce N-terminal NubG fusions. Saccharomyces cerevisiae strain NMY51 was cotransformed with the constructs following the manufacturer’s polyethylene glycol–lithium acetate protocol. Protein–protein interactions were evaluated by monitoring yeast growth on synthetic dropout media lacking tryptophan (W), leucine (L), histidine (H), and adenine (Ade), and supplemented with 3-aminotriazole (3AT) where required, after incubation at 30 °C for 2 to 4 d. Defined Compounds 1, SE, (+)-sesamin; 2, SN, (+)-sesamolin; 3, SL, (+)-sesaminol; 4, EO, (+)-episesaminone; 5, SL-MG, (+)-sesaminol monoglucoside/(+)-sesaminol 2-O-β-D-glucoside; 6, SL-DG2, (+)-sesaminol β(1→2)-diglucoside/(+)-sesaminol 2-O-β-D-glucosyl-(1→2)-O-β-D-glucoside; 7, SL-TG, (+)-sesaminol triglucoside/(+)-sesaminol 2-O-β-D-glucosyl-(1→2)-O-[β-D-glucosyl(1→6)] β-D-glucoside; 8, EO-MG, (+)-episesaminone monoglucoside/(+)-episesaminone 9-O-β-D-glucoside; 9, EO-DG2, (+)-episesaminone β(1→2)-diglucoside/(+)-episesaminone 9-O-β-D-glucosyl-(1→2)-O-β-D-glucoside; 10, EO-TG, (+)-episesaminone triglucoside/(+)-episesaminone 9-O-β-D-glucosyl-(1→2)-O-[β-D-glucosyl(1→6)] β-D-glucoside; 11, SLN, (+)-sesaminolin; 12, SNO, (+)-sesamolinone; 13, SLN-MG, (+)-sesaminolin 2-O-β-D-glucoside; 14, SLN-DG2, (+)-sesaminolin β(1→2)-diglucoside/ (+)-sesaminolin 2-O-β-D-glucosyl-(1→2)-O-β-D-glucoside; 15, SNO-MG, (+)-sesamolinone 9-O-β-D-glucoside; 16, SNO-DG2, (+)-sesamolinone β(1→2)-diglucoside/(+)-sesamolinone 9-O-β-D-glucosyl-(1→2)-O-β-D-glucoside; 17, SNol-DG6, (+)-sesamolinol 4′-O-β-D-glucosyl-(1→6)-O-β-D-glucoside; 18, SE(OH)2, dihydroxysesamin; 19, PPL, (+)-piperitol; 20, PNL, (+)-pinoresinol; 21, SO-TG, sesaminone triglucoside/sesaminone 9-O-β-D-glucosyl-(1→2)-O-[β-D-glucosyl(1→6)] β-D-glucoside; 22, SL-DG6, (+)-sesaminol β(1→6)-diglucoside/(+)-sesaminol 2-O-β-D-glucosyl-(1→6)-O-β-D-glucoside; 23, EO-DG6, (+)-episesaminone β(1→6)-diglucoside/(+)-episesaminone 9-O-β-D-glucosyl-(1→6)-O-β-D-glucoside; 24, SLN-DG6, (+)-sesaminolin β(1→6)-diglucoside/(+)-sesaminolin 2-O-β-D-glucosyl-(1→6)-O-β-D-glucoside; 25, SNO-DG6, (+)-sesamolinone β(1→6)-diglucoside/(+)-sesamolinone 9-O-β-D-glucosyl-(1→6)-O-β-D-glucoside. Detailed structural information is provided in the SI Appendix. Supplementary Material Appendix 01 (PDF) Dataset S01 (TXT) Dataset S02 (TXT) Dataset S03 (TXT) Dataset S04 (TXT) Dataset S05 (TXT)