Soybean plants (variety Daewon) were grown in pots inside a growth chamber at 30^oC/14 h and 25^oC/10 h, 60 – 70% relative humidity and at a sodium lamp light intensity of 1,000 μEm²/. The soybean leaves at the vegetative growth stages V1 (20 days), V3 (35 days), and V5 (56 days), and mature leaves at the reproductive stages R2 (77 days), R4 (98 days), R6 (112 days), and R7 (126 days) were used for the experiments. For the external application of coumestrol, soybean leaves at V1 growth stage were cut using sterilized scissors and immediately placed on a filter paper inside 125 × 125 × 20 mm square petri dishes (SPL, South Korea). Leaves were treated by supplying 5 ppm coumestrol [in 60% DMSO (dimethyl sulfoxide)] on the filter paper and incubated at 30/25°C (day/night) for 10 days. To avoid drying of samples, 5 ml of the solution was added after every 2 days (2, 4, 6, and 8 days). The samples were collected on the 1st, 3rd, 5th, and 10th day and stored at −80°C.

For phytohormone application, 20 days-old soybean plant leaves were cut and placed inside petri dishes as described above and the phytohormones SA (Sigma Aldrich), methyl jasmonate (MeJA) (Sigma Aldrich), and ethephon (Chaksak wang, farmhannong) were added at a concentration of 5 ppm alone and/or in different combinations. The samples were collected after 1, 3, 5, and 10 days after phytohormone applications. Control treatments were supplied with water.

Total protein from soybean leaves was extracted using the Mg/NP-40 buffer, which was fractionated with PEG 4000, following the method described by Kim et al. (2001, 2003). Each sample (150 μg) was mixed with isoelectric focusing (IEF) sample buffer and loaded onto an 18-cm IEF gel. The second-dimension separation was conducted on S sodium dodecyl sulfate poly acrylamide gel electrophoresis (SDS-PAGE) using 12% polyacrylamide gels. 2-DE gels were silver-stained (Kavran and Leahy, 2014), scanned (PowerLook III, UMAX) and exported as TIFF files from the scanner, after which gel spots were detected using ImageMaster 2D Platinum software (Amersham Biosciences). Differentially expressed protein spots were identified using MALDI-TOF-MS. Gel spots digested with trypsin were analyzed using a Voyager-DE STR (matrix-assisted laser desorption ionization time-of-flight) and MALDI-TOF mass spectrometer (PerSeptive Biosystems). Briefly, individual protein spots were isolated and remelted with digestion mixtures [93:5:2 water, acetonitrile, and trifluoroacetic acid (TFA)]. Then, the samples were sonicated for 5 min and centrifuged; 2-μl sample was added to 2-μl peptide sample solution (the matrix solution, the nitrocellulose solution, and isopropanol were mixed 100: 50: 50 (Kim et al., 2004), and 1 μl of this was placed on the MALDI plate and left for 5 min, after which the samples were washed with 0.1% v/v TFA. Des-Arg1-bradykinin (m/z 904.4681) and angiotensin 1 (m/z 1296.6853) were used as internal standards for calibration. For data analysis, the PerSeptive-Grams software was utilized. Database searches for protein identification were performed using Protein Prospector.¹

For secondary metabolites analysis, 1 g of soybean leaves were extracted with 20 mL of 70% ethanol for 12 h and centrifuged at 3,000 rpm for 5 min. One milliliter of supernatant was filtered through a 0.2 μm PTFE filter and subjected to metabolite analysis. Chromatographic separation was performed using a UPLC system (Waters Corp., Milford, MA, United States) equipped with a binary solvent delivery system and a UV detector with absorbance at 254 nm. Aliquots (2.0 μl) of each sample were then injected into a BEH C18 column (2.1 × 100 mm, 1.7 μm) at a flow rate of 0.4 ml/min and eluted using a chromatographic gradient of two mobile phases (A: water containing 0.1% formic acid; B: acetonitrile containing 0.1% formic acid). A linear gradient program was followed: 0 min, 10% B; 0–7 min, 10–15% B; 7–11 min, 15–30% B; 11–16 min, 30–50% B; 16–17 min, 80–100% B; 17–18 min, 100% B, 18.3–20 min, back to 10% B.

The measurements of coumestrol and plant hormones were performed as described by Pan et al. (2010) and Ng et al. (2016) with minor modifications using flight (LC-ESI-QTOF) tandem mass spectrometry. Soybean leaves (1 g) were extracted with 80% MeOH aq. (10 mL) and shaken for 24 h at room temperature. The filtered solution was analyzed using Liquid Chromatography Electrospray-Ionization Quadrupole Time-of-flight (LC-ESI-QTOF) Tandem Mass Spectrometry. The hormones were analyzed using the method described by Pan et al. (2010). Extraction solutions I, II, and internal standards were prepared. Extraction solvent 1 [2-propanol/H2O/0.5% TFA (Trifluoroacetic acid)]. Extraction solution II; dichloromethane. A working solution of internal standards was prepared by diluting the combined stocks with methanol. Frozen plant tissues (100 mg) were ground and transferred into an E-tube, then 1-ml extraction solution I was added and shaken at 4°C for 30 min. After that, 1-ml extraction solution II was added and kept at 4°C for 30 min, after which it was centrifuged at a revolution speed of 13,000 rpm at 4°C for 5 min. The aqueous phase was transferred to a new tube, and 600-μl isopropanol was added to the sample and mixed gently. The samples were incubated at room temperature for 10 min and centrifuged at 13,000 rpm for 10 min at 4°C. The supernatant was discarded. The pellet was washed with 1-ml 75% EtOH. The samples were air-dried, and the pellet was resuspended in 50-μl nuclease-free water to dissolve the pellet.

Soybean leaf metabolites were identified using MS (accurate mass in negative mode) or MS/MS spectra (fragmentation pattern), UV/Vis spectra, and in-house library comparison as described by Yuk et al. (2011b). Calculations were based on the area of HPLC analysis. Values, reported as relative contents vs. the amount of individual coumestrol at R4 stage, are the mean measurements carried out on three independent samples analyzed three times.

For quantification of coumestrol, we used external standard measurement. Four different concentrations (10, 100, 1,000, and 10,000 ppm) of coumestrol (27,885, Sigma-Aldrich, 95%, HPLC) were used and generated standard curve. The value from the standard were then used to calculate concentration of coumestrol in samples. All the steps were carried out using a preinstalled software (MassLynx 4.1).

The total RNA was extracted using Trizol TRI Reagent Solution (ThermoScientific) according to the manufacturer’s standard protocol, cDNA was synthesized according to the manufacturer’s instructions from the total RNA using the reverse transcription BioFact^TM RT Kit. The real-time polymerase chain reaction was conducted using an Eco^TM real-time PCR system (Illumina) using 2× Real-Time PCR Master Mix, including SYBR Green I (Biofact) and 10 pmol of each primer, processed in a two-step PCR program. All analyses were conducted in triplicate, and gene expression data were analyzed using ECO Study program (Illumina). Relative expression levels were determined by testing treated and control plants, and normalized to GmActin.

Soybean leaves are known to turn yellow upon maturity as the development and maturity of pods begin. Cutting the supply of food to several yellowing leaves helps divert and accumulate nutrients in the pods for future generation. Leaf yellowing and plant maturity also relate to gradual variations in the day and night length, which significantly affects soybean senescence. Previously, it was reported that mature soybean leaves with yellow color inhibit the α-glucosidase activity due to high coumestrol levels (Yuk et al., 2011a). The color of soybean leaves begins to turn yellow from R2 onward, and completely yellow leaves develop at the mature or senescent R7 stage (Supplementary Figure 1A). The color change or senescence is accompanied by a concomitant loss of the total chlorophyll content. Our results indicated that significant increase in the chlorophyl content occurs up to V5 and then starts to reduce, with the least chlorophyl content at the R7 stage (Supplementary Figure 1B). To verify the correlation between leaf color changes or senescence and leaf metabolites, soybean leaves at different growth stages were collected after sowing i.e., 20 days (V1), 35 days (V3), 56 days (V5), 77 days (R2), 98 days (R4), 112 days (R6), and 126 (R7) days after sowing and the metabolite profile of the leaves were analyzed at the above-mentioned growth stages. HPLC analysis showed diverse and significant differences in the chemical composition of the leaves at the various growth stages, as reflected by detecting diverse chemical compounds in the leaves collected at various growth stages (Figure 1). More specifically, we identified three major chemical groups that were significantly abundant in mature, yellow leaves of the R7 growth stage (Figure 1, red box, peaks 1, 2, and 3). Although these three groups were also detected in the earlier growth stages, their quantities were significantly lower, as reflected by the smaller peaks in the chromatogram). Additionally, at least eight other chemical compounds (Figure 1, peaks 5–12) detected had a significant quantitative difference between fresh and mature leaves as significantly lower quantities were detected in mature leaves (R7). Among these, compounds representing peaks 5–9 were identified as flavonoid glycosides, while those representing peaks four, ten, eleven, and twelve were identified as glycosyl-oxy-isoflavones. These results are parallel with the previous work of Yuk et al. (2016). The chemical compounds representing peaks one, two, and three at the R7 stage were identified as coumestans. These three chemical compounds were identified as Isotrifoliol (peak 1), Coumestrol (peak 2), and Phaseol (peak 3). This indicates that coumestans play a key role in soybean leaf maturity. Additionally, among these three compounds, coumestrol (peak 2) was the most abundant chemical compound at the R7 stage.

Next, we conducted 2-D gel electrophoresis (2-DE) to separate various proteins extracted from leaves at two growth stages V1 and R7. Since the 2-DE separates proteins based on two properties in two dimensions, it was better suited to separate diverse proteins expressed at specific growth stages. The results indicated the detection of highly diverse proteins spread across both gels. Additionally, significant differences were observed in terms of the number, intensity, and positions of proteins/spots on the two gels, corresponding to the two growth stages, V1 and R7 (Figures 2A,B, respectively). For further detailed analysis, 28 different proteins were selected (represented by 28 spots on the gel). Among these, 15 proteins showed a significantly high accumulation (Figure 2, blue arrows) whereas 13 showed significant reduction (Figure 2, red arrows) in the gel corresponding to the R7 stage compared to the gel corresponding to the V1 stage. These proteins were further analyzed using MALDI-TOF-MS. A total of 19 proteins/spots were successfully identified, including 12 proteins that showed higher accumulation and seven proteins that showed a reduction at the R7 stage compared to V1. A list of the identified proteins is given in Supplementary Table 1. Enzymes, including malate synthase, alanine aminotransferase, and glutamate semialdehyde aminomutase, were identified among the proteins with lower accumulation in mature leaves at the R7 stage. However, different proteins involved in secondary metabolite biosynthesis were identified among the proteins with significantly higher accumulation in mature leaves.

According to the HPLC results shown in Figure 1, as aging in soybean plants progresses, the accumulation of three coumestans gradually increases. As described above, coumestrol was the most abundant among the three coumestan compounds at R7 stage, this indicates that coumestrol contributes to changes in leaf color and senescence besides several other metabolites that may be involved in leaf senescence. Additionally, we identified two intriguing proteins (corresponding to spots 5 and 13 in Figure 2) that showed a significantly high accumulation at R7 compared to V1. MALDI-TOF-MS analysis was used to identify the two proteins as 3,9 -dihydroxypterocarpan 6A-monooxygenase (CYP93A1) and isoflavone reductase homolog 2 (IFR2). Putative CYP93A1 protein is involved in the biosynthesis of the phytoalexin glyceollin, whereas IFR2 encodes an enzyme involved in the biosynthesis of glyceollins from daidzein (Graham et al., 1990; Oliver et al., 2003). Diadzein is either converted to coumestrol or the pterocarpan glyceollins. The protein identification results were obtained from Mascot database search using peptide mass fingerprint. Mascot search results with scores greater than 55 are significant at P < 0.05. We got a score of 56 for CYP31A1. However, the peptide sequence matches of the whole CYP93A1 sequence was 14% (Supplementary Figure 5). We then blasted these matched amino acids on NCBI and got hit on GLYMA_03G143700v4 (G. max) which is the soybean CYP93A1 (Supplementary Figures 6A,B). This confirmed that the protein in the spot is CYP93A1. Thus, results of Figure 2, led us to the conclusion that various coumestans, especially coumestrol, contributes to leaf maturity and yellowing in soybean along with several other metabolites that may be involved in leaf senescence. To confirm this, we externally applied coumestrol to V1-stage soybean leaves detached from the top of the plant (Supplementary Figure 2). Ten days after the application, the leaves showed dramatic phenotypic and physiological changes. After applying 5-ppm coumestrol, the leaves exhibited an interesting development of roots along with chlorosis or yellowing of the leaves (Figure 3). Similar results of root development were obtained when we applied coumestrol to mung bean leaves (unpublished). The development of roots from leaves is an unusual phenomenon and has not been studied in detail. Lozovaya et al. (2005) reported the accumulation of various isoflavones (including, diadzine, and genistin), and isoflavonoid phytoalexins, including coumestrol in hairy soybean roots following Fusarium solani f. sp. glycines infection. Additionally, Lee et al. (2012) reported the accumulation of coumestrol in root nodules. Moreover, recently, Xu et al. (2021) reported that the reduction of root nodulation and rhizobial infection upon exogenous ABA application is related to the reduction of isoflavonoid compounds, including coumestrol. As indicated in Figure 3B, the yellow color that appeared throughout the leaf caused by coumestrol application looked similar to the changes that appear in soybean leaves when naturally aged. This indicates that coumestrol may also regulate root development in soybean in addition to its role in senescence.

Leaf senescence may be caused by various factors, such as the phytohormone, drought, and oxidative stress. Ethylene and cytokinin are well-known phytohormones connected with plant senescence. It was reported that coumestrol levels increase following insect attack to prevent plant damage (Loper, 1968). JA is a key hormone that helps plants protect themselves against insects or necrotrophic pathogens. There is a high possibility of the interaction between coumestrol and phytohormones. Thus, we investigated changes in coumestrol levels following the application of MeJA, SA, and ethylene. Following the application of these phytohormones, soybean leaves did not exhibit any significant phenotypic changes for up to 5 days (Figure 4). However, after 10 days of application, clear changes in leaf color was observed. The leaves treated separately with MeJA and SA turned light yellow with a concomitant increase in coumestrol content. Similar results were observed in leaves treated with MeJA and SA with an additive or synergistic effect, as indicated by the significant increase in coumestrol contents compared with water or MeJA and SA alone. Notably, ethephon (ethylene releasing compound) had an even stronger and highly significant effect than the other phytohormones. Leaves treated with ethephon alone or combined with MeJA started to turn yellow after 5 days of treatment and resulted in a significantly higher accumulation of coumestrol. Ethephon appeared to have the highest significant effect as the color of the leaves changed within 5 days of the application with more than ten times increase in the coumestrol content compared to the control plants. Ethephon and MeJA had an additive or synergistic effect on the coumestrol content after 10 days of treatment (Figure 4). Leaf chlorosis was observed after 5–10 days of phytohormone application. Leaves treated separately with MeJA and SA turned light yellow with a concomitant increase in coumestrol content. MeJA and SA, when combined, had an additive or synergistic effect on leaf color and coumestrol content as indicated by the significant increase in coumestrol contents compared to water, or MeJA and SA alone. The ethylene releasing compound, Ethephon, had an even stronger and highly significant effect than the other phytohormones. Leaves treated with ethephon alone or combined with MeJA started to turn yellow after 5 days of treatment and resulted in a significantly higher accumulation of coumestrol. Ethephon and MeJA had an additive or synergistic effect on the coumestrol content after 10 days of treatment.

Next, we checked the expression levels of genes involved in coumestrol biosynthesis (Figure 5) following the application of various hormones alone or in combination as described above. Two genes, CYP81E28 (Gm08G089500) and CYP81E22 (Gm16G149300), are known to be involved in the hydroxylation of daidzein during coumestrol biosynthesis (Reinprecht et al., 2017). Overall, we did not find any statistical difference between the expression of CYP81E28 in plants treated with any of the phytohormone alone or in combination after 3, 5, or 10 days. However, compared to the control plants treated with water only, a significant increase in expression was recorded after 5 and 10 days of SA application and 3 days of MeJA + Ethephon (Figure 5A). Generally, the phytohormones increased the expression of CYP81E22 after 3, 5, and 10 days of phytohormone application compared to the control plants treated with water only (Figure 5B). More specifically, no significant difference was observed in the expression of CYP81E22 after 3, 5, or 10 days of SA application. Ethephon treatment significantly increased the expression of CYP81E22 after 5 days whereas, a significantly higher expression was recorded when it was applied along with MeJA (Figure 5B). The genes GmIFS1 and GmIFS2 function upstream of coumestrol biosynthesis and are involved in converting liquiritigenin to daidzein (Sohn et al., 2021). When compared to the control plants, we found a significant increase in the expression of GmIFS1 after phytohormone application. MeJA induced a significantly high expression of GmIFS1 after 10 days of application (Figure 5C). However, changes in the expression of GmIFS2 were largely non-significant following phytohormone treatment. However, the highest significant expression was recorded after 3 days of the combined treatment of MeJA and ethephon (Figure 5D). The highest change in expression after phytohormone treatment was recorded for GmCYP93A1, which is involved in the final step of coumestrol biosynthesis. All the phytohormones significantly increased the expression compared to the control plants treated with water only. A significant increase in expression after 3, 5, and 10 days was recorded following MeJA treatment. The highest GmCYP93A1 expression was recorded after 3 days of ethephon treatment, which increased by twofolds when ethephon was applied along with MeJA (Figure 5E).