P. lobata seeds were collected from Anhui Province of China and grown in soil under normal controlled conditions as described previously (Li J. et al., 2014). G. max seeds were provided by Prof. Jian Zhao at the Huazhong Agricultural University of China. The chemical standards prunetin and biochanin A were purchased from BioBioPha Co. Ltd (Kunming, China), 4′,7-dimethoxy-5-hydroxyisoflavone were from Tianjin Sigma Technology Co. Ltd (Tianjin, China), 4′,7-dimethoxyisoflavone was from Tianjin Yifang Technology Co. Ltd (Tianjin, China) and all the other chemical standards were purchased from Shanghai Source Leaf Biological Technology Company (Shanghai, China). All organic solvents used for high-performance liquid chromatography (HPLC) and liquid chromatography–mass spectrometry (LC–MS) were of HPLC grade and they were purchased from Wuhan Dingguo Biotechnology Co. Ltd (Wuhan, China). All the restriction endonucleases were from Takara Company (Takara, Dalian, China). All the nucleotides used for the PCRs of this study are shown in Supplementary Table S1.

A transcriptome derived from the roots and leaves of P. lobata was constructed by our group and its raw RNA sequences have been deposited at the sequence read archive (SRA) of the National Center for Biotechnology Information (accession number: SRX480408). The coding region sequences of PlOMTs were discovered from the transcriptome, amplified by RT-PCRs from the P. lobata roots, and cloned into the pMD18-T vector (Takara, Dalian, China) for a sequencing confirmation. The thermocycling conditions for the PCRs were used as follows: at 95°C for 3 min for one cycle; at 95°C for 15 s, then at 55°C for 30 s and at 72°C for 2 min for 35 cycles, and a final elongation step at 72°C for 10 min.

For the phylogenetic analysis, the deduced amino acid sequences of the OMT candidates from P. lobata were aligned with other plant OMTs with known functions using the CLUSTAL X Version 2.0 program (Larkin et al., 2007). The phylogenetic tree was constructed by the neighbor-joining method of the MEGA 6.0 program (Tamura et al., 2013). The information of the sequences used in the phylogenetic analysis is summarized in Supplementary Table S2.

For the expression in yeast cells, the coding regions of PlOMTs were amplified and sub-cloned into the yeast expression vector pESC-HIS (Stratagene, La Jolla, CA, USA) to give the constructs pESC-HIS-PlOMTs. The plasmids pESC-HIS-PlOMTs and empty vector pESC-HIS were introduced separately into the yeast WAT11 strain (Pompon et al., 1996). Yeast cultures were grown in His dropout liquid medium (Clontech, Palo, CA, USA) containing 2% (w/v) glucose. After 24 h in a shaker maintained at 250 rpm, the yeast cells were collected by centrifugation, washed with sterile water for three times, and re-suspended to an OD₆₀₀ of 0.8 in 5 ml of His dropout liquid medium containing 2% (w/v) galactose supplemented with the substrates such as daidzein, genistein, formononetin, isoformononetin, prunetin, and biochanin A, separately, at a final concentration of 100 μM. After incubation at 30°C for 48 h, the yeast cultures were extracted with an equal volume of ethyl acetate, the acetate extracts were air dried at room temperature and re-dissolved in methanol for HPLC analysis.

The coding regions of PlOMT2, PlOMT9, and HI4′OMT (Deavours et al., 2006) were amplified, and sub-cloned into the Escherichia coli expression vector pET28a in frame with an N-terminal His₆ tag, yielding the plasmids pET28a-PlOMT2, pET28a-PlOMT9, and pET28a-HI4′OMT. These constructs were then transformed into E. coli BL21 (DE3) cells using heat shock at 42°C and the expression of the recombinant OMTs was induced with 0.5 mM IPTG at 16°C for 14 h. Cells were centrifuged at 6,000 rpm (4,629 × g) at 4°C for 10 min. The resulting cell pellets were re-suspended in 20 ml of lysis buffer consisting of 20 mM sodium phosphate (pH 7.4), 300 mM sodium chloride, and 10 mM imidazole. Cells were lyzed with lysozyme on ice for 30 min followed by sonication on ice by 1-s pulse for 30 min. The cell-free extracts containing recombinant OMTs were obtained by centrifugations at 4°C for 30 min. The recombinant OMTs were then purified using HisPur^TM Ni-NTA Resin (Thermo, Waltham, MA, USA) following the provided protocol. The purity of the recombinant OMTs was checked by SDS-PAGE and their concentrations were determined by Bradford assays using BSA as the protein standard. For the peptide analysis of recombinant PlOMT2, its corresponding stained band was excised from the SDS-PAGE gel, digested with trypsin, and analyzed by MALDI-TOF/TOF mass spectrometry according to the method previously described (Li et al., 2012).

Unless otherwise stated, the in vitro enzyme assays were carried out in a 200 μl reaction mixture consisting of 0.5–5μg of the purified OMT, 50 mM Tris–HCl (pH 8.0), 2 mM DTT, 2 mM S-adenosyl methionine (SAM), and 100 μM substrates. The reactions were performed at 37°C for 20 min and stopped by adding 200 μl of methanol. The reaction time of 20 min was determined to be in the linear range for the activities by preliminary experiments. The 15 μl of the reaction mixture was then directly used for HPLC and LC—MS analysis.

For the assays with 2,7,4′-trihydroxy-isoflavanone, the reaction mixture consisted of 200 μg of the yeast microsomes expressing P. lobata IFS (GenBank accession number KC202929) prepared as previously described (Liu et al., 2003), 5 μg of the purified recombinant OMTs, 50 mM Tris–HCl (pH 8.0), 2 mM NADPH, 2 mM DTT, 2 mM SAM, and 100 μM liquiritigenin. The reaction was allowed to proceed at 30°C for 60 min, stopped by 200 μl of methanol, and was directly applied for HPLC and LC–MS analysis.

Plant materials (roots, stems, and leaves) of P. lobata grown at open field were collected, powdered under liquid nitrogen, and stored at –80°C for the measurement of methylated isoflavone formononetin and the PlOMTs gene expression. The increment of isoflavonoid biosynthesis in P. lobata in response to methyl jasmonate (MeJA) elicitation was previously reported (Thiem and Krawczyk, 2010; Li Z.B. et al., 2014). For the MeJA treatment in this study, a stock solution of MeJA was first dissolved in ethanol, and added into liquid 1/2 MS medium where one-month-old P. lobata seedlings were grown. The final concentration of MeJA in the MS medium was 100 μM and the concentration of ethanol in the medium was 0.001% (v/v). For the mock controls, the same concentration of 0.001% ethanol was added into the liquid MS medium. Roots were collected after 6 h treatment for the gene expression analysis, and after 10-day treatment for the measurement of methylated isoflavones. For gene expression analysis, total RNA was isolated and treated with DNase I. Transcript levels were analyzed by quantitative RT-PCR (q RT-PCR). For the measurement of methylated isoflavones, 20 mg of plant materials was extracted with 1 ml of ethanol with three biological repeats. The ethanol extracts were air dried at room temperature and re-dissolved in methanol for HPLC and LC–MS analysis. Two-tailed t test (confidence interval: 95%) was performed using GraphPad Prism 5 software for the statistical analysis of the data.

The coding region of PlOMT9 was PCR amplified and fused with the open reading frame of green fluorescent protein (GFP) in the plant expression vector pCAMBIA1302 (CAMBIA, Canberra, Australia) by enzyme digestions and ligations, yielding the plasmid pCAMBIA1302-PlOMT9. The Agrobacterium rhizogenes strain K599 was separately transformed with pCAMBIA1302-PlOMT9 as well as the empty vector pCAMBIA1302 as a control.

The hairy roots of G. max were generated using these transgenic A. rhizogenes strains, following previously described methods (Cho et al., 2000). After around 3–4 weeks from the inoculation, hairy roots appeared on infection sites. When these roots were at least 2 cm long, the positive transgenic hairy roots were identified by genomic PCRs. A segment from each positive root was excised and sub-cultured in 1/2 MS liquid medium with antibiotics in the dark. After 2 weeks of the sub-culture, those transgenic hairy root cultures were treated with MeJA at a final concentration of 100 μM. As a mock control, the same volume of ethanol was added into the cultures. The transgenic roots were used for metabolite analysis after 10 days of the treatments. For the metabolite analysis, 36 positive roots per each construct were cultured with 6 roots per 150 ml flask in a shaker, leaving that the half of the root cultures (18 roots in three flasks as 3 biological replicates) was used for the MeJA-mediated elicitation and the other half of the cultures was treated as the mock controls. Hairy roots from each flask were collected, powdered together, and extracted with ethanol for the quantification of relevant metabolites, which includes daidzein, daidzin, formononetin, and ononin.

Total RNAs were isolated from roots, stems, and leaves of P. lobata using Trizol (Invitrogen) and treated with DNase I (Thermo) to remove contaminated genomic DNA. First-strand cDNA was prepared using superscript III reverse-transcriptase (Invitrogen). The P. lobata actin gene (GenBank accession number HO708075) was used as an internal standard to normalize the variation in each cDNA preparation. The qRT-PCR was performed on an ABI 7500 Fast Real-Time PCR Detection System with FastStart Universal SYBR Green Master mix (Rox; Roche, Mannheim, Germany). The PCR conditions were as follows: 10 min of initial denaturation at 95°C, followed by 40 cycles of 95°C for 15 s and then 60°C for 1 min. All the RT-PCRs were performed in three biological repeats. Two-tailed t test (confidence interval: 95%) was performed using GraphPad Prism 5 software for the statistical analysis of the data. Gene-specific primers were used for the qRT-PCRs and are listed in Supplementary Table S1.

HPLC analysis was performed on an LC-20AT instrument (Shimadzu, Kyoto, Japan) using an inertsil ODS-SP reverse-phase column (250 mm × 4.6 mm, 5 μm). The column temperature was set at 30°C. Samples from the in vitro enzyme assays were eluted in the mobile phase, 0.4% phosphoric acid (solvent A) and HPLC-grade methanol (solvent B), with 65% B for 45 min for the reactions with genistein and 60% B for 40 min for the other reactions (with quercetin, kaempferol, and luteolin) at a flow rate of 0.8 ml min^-1. For other analyses, 0.1% (v/v) formic acid (A) and acetonitrile (B) were used as the mobile phase and samples were separated at a flow rate of 0.8 ml min^-1 in a stepped gradient mode as follows: 0–30 min, 25–90% B; 30–35 min, 90–25% B; 35–40 min, 25% B; and the flow rate was 0.8 ml min^-1. The detection wavelength was set at 260 nm for daidzein, genistein, prunetin, formononetin, isoformononetin, and biochanin A; at 280 nm for liquiritigenin; at 350 nm for apigenin and luteolin; at 370 nm for quercetin, kaempferol, and isoliquiritigenin. A standard curve was acquired from different concentrations (5–100 μg ml^-1) of each chemical standard for the quantification.

LC–MS/MS analysis was performed on an Accela LC system coupled with TSQ Quantum Access Max mass spectrometer (Thermo Scientific, USA) and electrospray ionization source. The column and the analysis method were the same as the HPLC analysis described above. The MS data were recorded in a positive ion mode with ranges of m/z 50–500.

Analysis of the transcriptome database derived from the roots and leaves of P. lobata (GenBank accession number: SRX480408; Wang et al., 2015) revealed the presence of 108 unigenes annotated as OMTs (Supplementary Table S3), of which only 42 OMT unigenes are in a length of more than 1 kb. We decided to select the OMT candidates from these 42 unigenes. The criteria to select the OMT candidates are as follows: (i) the target genes would be transcribed at relatively higher levels in P. lobata roots than its leaves, given the fact that the isoflavonoids of P. lobata mainly accumulate in its roots (Kirakosyan et al., 2003; Matkowski et al., 2003); (ii) the OMT candidates would be annotated as enzymes related to (iso)flavonoid metabolisms by BLAST searches against public databases. Finally, nine OMT candidates were identified. Full-length cDNAs for these nine OMTs (termed PlOMT1-9, GenBank accession numbers, KP057884–KP057892) were presented in the dataset and their deduced amino acids share 21–81% identities with each other. The relationship of the OMTs candidates with other plant OMTs was assessed through phylogenetic analysis of deduced polypeptide sequences (Figure 2). PlOMT1, PlOMT6, and PlOMT8 clustered with flavonoid 8-O- and 4′-O-methyltransferases. PlOMT2 occurred in the same branch with HI4′OMTs whose functions are known to be responsible for the 4′-methylation in formononetin biosynthesis at the isoflavanone stage (Akashi et al., 2003; Deavours et al., 2006). PlOMT5, PlOMT7, and PlOMT9 could be grouped into the clade with I7OMTs. Enzymes in this subclass are believed to mediate the 7-O-methylation of isoflavones, such as daidzein (He et al., 1998; Akashi et al., 2003; Deavours et al., 2006). PlOMT3 and PlOMT4 showed a closer relationship to flavonoid 3′-O-methyltransferases, which indicates possible roles for PlOMT3 and PlOMT4 in 3′-O-methylation of (iso)flavonoids. As a focus on the investigation on the 4′- and/or 7-O-methylation processes here, all the PlOMTs described above, except for PlOMT3 and PlOMT4, were investigated in this study.

To screen the biochemical activities of the OMT candidates, the PlOMT cDNAs were inserted into the vector pESC-HIS (Stratagene, La Jolla, CA, USA) for expression in Saccharomyces cerevisiae WAT11 (Pompon et al., 1996), the yeast transformed with the empty vector pESC-HIS served as the control. The activities of PlOMTs were tested by adding various substrates (daidzein, genistein, formononetin, isoformononetin, prutein, or biochanin A) to yeast cultures that express recombinant PlOMTs, followed by testing the conversion of the substrates using HPLC analysis. Notably, yeast cells expressing PlOMT9 was able to metabolize part of the substrates (Figure 3), and no measurable activities were observed for the other PlOMTs (data not shown). Given the closer relationship of PlOMT9 to I7OMTs in the phylogenetic tree (Figure 2), PlOMT9 was initially presumed as an I7OMT. Unexpectedly, the in vivo assays showed that yeast expressing PlOMT9 efficiently methylated the 4′-hydroxy groups of genistein, daidzein, prunetin, and isoformononetin, forming their respective 4′-O-methylated isoflavones biochanin A (peak 1), formononetin (peak 3), 4′,7-dimethoxy-5-hydroxyisoflavone (peak 5), and 4,7-dimethoxyisoflavone (peak 6), respectively (Figures 3A–D). These data indicated that PlOMT9 primarily functions as an I4′OMT but not as an I7OMT as expected. Interestingly, when genistein or daidzein was fed, the cells bearing PlOMT9 also produced trace amounts of 4′, 7-di-O-methylated products (peak 2 and peak 4; Figures 3A,B), suggesting that PlOMT9 retains trace of I7OMT activity. To clarify this, the substrates whose 4′-OH is already methylated, formononetin (4′-methoxy-daidzein) and biochanin A (4′-methoxy-genistein), were fed to the PlOMT9-expressed cultures, tiny peaks representing the products 4′, 7-dimethoxyisoflavone (7-O-methylated formononetin, (peak 7; Figure 3E) and 4′,7-dimethyoxy-5-hydroxyisoflavone (7-O-methylated biochanin A, peak 8; (Figure 3F) were observed. Therefore, our data clearly suggested that PlOMT9 primarily functions as an I4′OMT but retains the trace I7OMT activity. The identities of all the catalytic products were confirmed by comparisons with their respective authentic standards using LC–MS analysis (Supplementary Figure S1).

To analyze the substrate specificity of PlOMT9, the PlOMT9 fused with an N-terminal His₆ tag was purified from an expression in E. coli cells by Ni-affinity chromatography (Supplementary Figure S2), and assayed with a broad range of substrates whose structures are shown in Supplementary Figure S3. As shown in Table 1, using SAM as the methyl donor, the purified PlOMT9 recognized 4′-hydroxy isoflavones as substrates and was found to be inactive with other types of compounds. Of the 4′-hydroxy isoflavones, the best substrate for PlOMT9 is genistein (100% relative activity, 7670.6 pkat mg^-1) followed by daidzein (48.8% relative activity, 3742.9 pkat mg^-1). Other isoflavones with a C-4′ hydroxyl group (prunetin and isoformononetin) are also utilized by PlOMT9 as acceptors but are poor substrates (5.2–6.2% relative activity; Table 1). When daidzein was used as the substrate for the in vitro assay, formononetin was predominantly formed while isoformononetin and 4′,7-dimethoxyisoflavone were also produced (Supplementary Figure S4). Again, this result further confirmed that PlOMT9 mainly functions as an I4′OMT with a trace I7OMT activity. The identities of all the catalytic products from the in vitro assays were confirmed by comparisons with their respective authentic standards using LC–MS analysis (Supplementary Figure S5).

The intermediate 2,7,4′-trihydroxy-isoflavanone was previously proposed as a natural methylation acceptor for HI4′OMT from licorice (G. echinata; GenBank accession number BAC58011) and M. truncatula (GenBank accession number AAY18581). The 4′-methylation activity with 2,7,4′-trihydroxy-isoflavanone can be detected by the formation of the intermediate 2,7-dihydroxy-4′-methoxy-isoflavanone from in vitro enzyme assays (Akashi et al., 2003; Deavours et al., 2006). To examine the activity of PlOMT9 with 2,7,4′-trihydroxy-isoflavanone, 2,7,4′-trihydroxyisoflavanone was prepared by the incubation of the yeast microsome-expressing P. lobata IFS (GenBank accession number KC202929; Gou et al., 2013) incubated with a racemic mixture of 2R/S-liquiritigenin and NADPH (Akashi et al., 1999). The previously characterized M. truncatula HI4′OMT (Deavours et al., 2006) was expressed and purified from E. coli cells as a positive control. In consistent with the previous report (Deavours et al., 2006), the product 2,7-dihydroxy-4′-methoxy-isoflavanone (peak 1) was clearly observed in the incubation with HI4′OMT and part of 2,7-dihydroxy-4′-methoxy-isoflavanone had dehydrated to formononetin (peak 2; Figure 4). In contrast, in the reactions with PlOMT9, the intermediate 2,7-dihydroxy-4′-methoxy-isoflavanone was not detected at all (Figure 4), suggesting that PlOMT9 is not able to methylate the substrate 2,7,4′-trihydroxy-isoflavanone. Formononetin (peak 3) was formed in the reaction with PlOMT9 (Figure 4), which probably resulted from the activity of PlOMT9 with daidzein that was formed by the spontaneous dehydration of 2,7,4′-trihydroxy-isoflavanone. In addition, considering the closer relationship of PlOMT2 to HI4′OMTs in the phylogenetic tree (Figure 2), we also prepared a purified recombinant PlOMT2 by expression in E. coli cells (Supplementary Figure S2) and assessed its activity with 2,7,4′-trihydroxy-isoflavanone, no enzymatic products by PlOMT2 were observed as well (Figure 4).

To examine the relevance of PlOMT9 to its methylated isoflavone biosynthesis in P. lobata, we measured the distribution of its in vitro enzymatic metabolite in various organs. As shown in Figure 5A, the formononetin (PlOMT9 product) was mostly detected in the root (58.4 ± 5.1 μg g^-1 dry weight), but was almost undetectable in its aerial organs (stem and leaf). Expression analysis by quantitative reverse transcription-PCRs revealed that PlOMT9 gene is specifically expressed in the root with extremely low levels in the stem and leaf (Figure 5B). Thus, in a spatial manner, the relative abundance of PlOMT9 transcripts matches the accumulation pattern of formononetin in P. lobata, supporting the proposed roles of PlOMT9 physiologically. The gene expression pattern of all the other PlOMTs was also investigated. As shown in Figure 5C, PlOMT5, PlOMT6, and PlOMT8 genes show preferential expression in the roots; the genes coding for PlOMT1 and PlOMT7 are transcribed at a higher level in the stem relative to the root and leaf while the PlOMT2 gene shows similar expressions in all the three organs.

To further test the correlation in an eliciting manner, one-month-old seedlings of P. lobata were treated with 100 μM MeJA and their roots were then harvested for metabolite and nucleic acid extraction. The roots of 10-day-treated plants were used for metabolite extraction. The concentration of formononetin in the MeJA-treated plants apparently increased in comparison with the controls (Figure 5D). To examine the gene expressions, the roots of 6-h-treated plants were collected for this purpose. Relative to the control plants, the transcript level of PlOMT9 was significantly elevated in response to the elicitation (Figure 5E). Taken together, the transcript abudance of PlOMT9 correlates well with the accumulation pattern of formononetin in P. lobata. Two-tailed t test was performed and a P value of 0.05 was considered to be statistically significant.

To assess whether PlOMT9 play a role for the formononetin biosynthesis in planta, P. lobata hairy roots were initially considered as the expression system for this purpose. However, our endeavor to establish P. lobata hairy roots failed although Liu et al. (2000) previously reported the success of P. lobata hairy root inductions. Upstream isoflavonoid biosynthesis pathway is conserved in legume species, and one of the legume species, G. max, also contains IFSs and its hairy roots physiologically provide daidzein (the in vitro substrate of PlOMT9 described above). Thus, the G. max hairy root was chosen as the expression host here as its hairy root system has been well established (Cho et al., 2000). The coding region of PlOMT9 was fused to GFP in the vector pCAMBIA1302 and transformed into the hairy roots of G. max. The free GFP alone in the pCAMBIA1302 was expressed in hairy roots as the control. The presence of the transgenes was confirmed by genomic PCRs, and positive hairy roots were sub-cultured and treated with or without MeJA signal for metabolite quantification by LC–MS analysis. In the roots untreated with MeJA, the content of daidzein and daidzin was reduced by 17.5 and 26%, respectively, in the PlOMT9-expressed lines with respect to the control (Figure 6). On the other hand, compared to that of the control lines, the average concentration of formononetin in the PlOMT9-expressed roots was slightly increased (PlOMT9 expressed-lines, 1.20 μg g^-1 DW; the control lines, 1.12 μg g^-1 DW), whereas its 7-O-glucoside, namely ononin (Figure 1), was increased by 57.7% (PlOMT9-expressed-lines, 10.53 μg g^-1 DW; the control lines, 6.68 μg g^-1 DW; Figure 6). When the roots were elicited with MeJA, the roots expressing PlOMT9 produced significantly increased levels of both formononetin and ononin (formononetin and ononin were increased by 111.2 and 940.9%, respectively) compared to the control roots. Moreover, this increment was accompanied with the decrease in the biosynthesis of daidzein and daidzin (daidzein and daidzin were decreased by 24.3 and 45.2%, respectively; Figure 6). Taken together, all these data corroborated that PlOMT9 is involved in the biosynthesis of formononetin in planta.