Two soybean accessions, “Zhongdou27” with high isoflavone content and “Dongong8004” with low isoflavone content, were crossed to develop RILs, using the single seed descent (SSD) method (Oldach et al., 2014). An expanded population with 119 F_5:18 families was derived and used to construct a linkage map and detect the QTLs associated with isoflavone content.

For fine-mapping of the identified QTL (qISO19-1), a BC₂F₃ population of 500 lines was constructed by backcrossing “Zhongdou27” (donor parent) and “Dongnong8004” (receptor parent). Heterozygosity at qISO19-1 was detected using recurrent selection in the BC₂F₂ population, and isoflavone contents were measured in the BC₂F₃ seeds produced by selected BC₂F₂ plants.

In preliminary experiments, “Zhongdou27” (male parent) and “Dongnong8004” (female parent) were planted in chernozem soils in Xiangyang (45°72′ N, 126°68′ E) and Hulan (45°9′ N, 126°58′ E) to confirm differences in isoflavone content. Subsequently, the RIL population was planted in Xiangyang and Hulan in 2018. For fine mapping, the BC₂F₃ population (as a transient generation) was planted in Xiangyang only (latitude 45°80′ N, 126°53′ E) in 2019. For candidate identification, a set of soybean germplasms consisting of 43 accessions were planted at Xiangyang in 2019 (Supplementary Table 1). Field trials were conducted in single-row plots (between 3 m long and 0.65 m rows) using a randomized complete block design, with three replicates per tested environment. Field management practices were typical, and were identical across environments. From all soybean materials, 20 seeds were collected from three plants per plot for isoflavone content measurement at the R8 stage (full maturity).

Isoflavone contents in the seeds collected from the parental lines, the RILs, the BC₂F₃ population, and 43 soybean germplasms were measured using high-performance liquid chromatography (HPLC) as described previously (Vyn et al., 2002; Zeng et al., 2009). Briefly, 20 seeds from each soybean line were ground into a powder. Isoflavones were extracted by adding 0.1 g of this powder to 10 mL of 80% (v/v) ethanol in a 15 mL falcon tube. The mixture was slowly vortexed for 1 h and then left overnight. Next, the extraction mixture was hydrolyzed using 2 mL HCl solvent (2 mol/L). The mixture was filtered through a nylon membrane filter (0.22 μm; Thermo Fisher Scientific, United States), and isoflavone content was measured in 1.5 μl of the filtrate using an HPLC (1290 Infinity II; Agilent, United States) with reversed-phase HPLC columns (ZORBAX SB-C18, Agilent, United States; 4.6 mm × 250 mm; 5 μm). Solvent A was double-distilled water (ddH₂O), and solvent B was methanol (chromatographic purity). The ratio of solvent A to solvent B ratio was 1:1, the solvent flow rate was 0.8 mL/min, and the temperature of the column was maintained at 50°C. Using an Agilent 1290 DAD detector, UV spectra were measured at 254 nm, and area responses were integrated using Agilent OpenLAB Control Panel software. The three major isoflavone components (daidzein, genistein, and glycitein) were identified and quantified based on standards purchased from the Chengdu Manster Biotechnology Co., Ltd. (China). Total isoflavone content was equivalent to the sum of the daidzein, genistein, and glycitein contents.

Genomic DNA for the RIL population and the parental lines were prepared as described By Qi et al. (2014). Sequencing libraries for these samples were constructed and sequenced on an Illumina HiSeq2500 sequencing platform, following the manufacturer’s instructions. The sequencing reads for the RIL population and the parental lines were aligned to the soybean reference genome (assembly Glycine_max_v2.1) (Schmutz et al., 2010) using Short Oligonucleotide Alignment Program 2 (SOAP2) (Li et al., 2009). GATK (McKenna et al., 2010) was used to identify polymorphic SNPs between the RIL population and the parental lines.

Co-segregating SNPs were separated into bins, and a bin map was constructed based on the recombinant breakpoints of the RIL population with HighMap (Liu et al., 2014). Genetic distances among markers were calculated using the Kosambi mapping function (Kosambi, 1944). Linkage groups were discriminated at a log-likelihood threshold of 3.0. QTL mapping were performed using IciMapping v4.1 (Meng et al., 2015). Putative QTLs were identified based on LOD threshold of 2.5.

We used 500 lines in the BC₂F₃ family to construct a local saturation map within qISO19-1 based on SSR markers. The SSR primers used for fine-mapping are given in Supplementary Table 2. We measured isoflavone contents in the seeds produced by these lines as described above. Recombinants were identified in the fine-mapping population based on 12 polymorphic SSR markers, and the QTLs significantly associated with isoflavone content (LOD > 2.5) were identified using IciMapping v4.1 (Meng et al., 2015). Student’s t-test was used to identify significant differences between lines of BC₂F₃ with high and low content of isoflavone in soybean seed.

For candidate gene identification, seeds of the 43 soybean germplasms described above were sampled at the R7 stage (yellow ripening stage); three replicate seeds were collected per accession for RNA extraction and isoflavone content measurement, to determine the correlation between isoflavone content and transcript abundance.

To analyze the relative expression dynamics of GmMT1 between high- and low- isoflavone soybean cultivars, developing seeds of “Zhongdou27” and “Dongnong8004” were sampled every 7 days from developmental stage R5 to R8 (three replicate seeds per cultivar were sampled at every time point). Plants were grown in a greenhouse under a 16 h light/8 h dark photoperiod at 25–26°C until sampling.

To determine the involvement of GmMT1 in the stress response, “Zhongdou27” was exposed to drought and salt stress. First, the roots of 3 week old “Zhongdou27” seedlings were immersed in quarter-strength (1/4) Murashige and Skoog (MS) liquid medium. We then supplemented the MS media of five seedlings with 150 mM NaCl, and the MS media of five seedlings with 8% (w/v) PEG6000 for the salt- and drought-stress response tests, respectively. The remaining five seedlings were kept in unsupplemented MS media as controls. Half of the uppermost fully extended leaf per seedling was sampled at 0, 1, 2, 4, 6, 8, 12, and 24 h after supplementation. During this period, plants were maintained in a greenhouse under a 16 h light/8 h dark photoperiod at 25–26°C. Harvested leaves were immediately frozen in liquid nitrogen and stored at −80°C. Leaves from three of the five treated plants were used for qRT-PCR.

qRT-PCRs were performed to determine the transcript abundance of GmMT1 in soybean seeds or leaves. Total RNA was isolated from leaves or seeds using RNAprep pure Plant Kits (DP432, Tiangen). First-strand cDNA was synthesized from total RNA using TIANScript RT Kits (KR104, Tiangen). qRT-PCRs were performed on an ABI 7500 Fast platform using SuperReal PreMix Plus (SYBR Green) Kits (FP205, Tiangen). Each qRT-PCR (20 μl) included 2 μl cDNA, 10 μl 2 × SuperReal PreMix Plus, 0.4 μl 50 × ROX Reference Dye^Δ, 5 μl of each forward and reverse primer, and 6.6 μl ddH₂O₂. The qRT-PCR amplification conditions were 95°C for 2 min, followed by 40 cycles of 95°C for 10 s, 60°C for 30 s, and 72°C for 30 s. The primers used are given in Supplementary Table 3. Three technical replicates were performed per sample, and the relative levels of transcript abundance were calculated using the 2^–ΔΔCT method (Livak and Schmittgen, 2001). The housekeeping gene GmActin4 (GenBank accession no. AF049106) was used as the internal standard. Student’s t-test was used to identify significant differences in transcriptional abundance of GmMT1.

We predicted the 3-D structure of the putatively encoded protein GmMT1 using Phyre 2 (Kelley et al., 2015). We identified DNA and protein sequences homologous to GmMT1 in soybeans and 15 other plant species in the Phytozome database². We aligned these methyltransferases using DNAMAN (version 7.212, Lynnon Corp., QC, Canada). We then constructed a phylogenetic tree based on this alignment in MEGA 5 (Tamura et al., 2011).

The full-length cDNA sequence of GmMT1 was amplified from the developing seeds of the high-isoflavone cultivar “Zhongdou27” using RT-PCR. RT-PCRs were performed using the KOD One PCR Master Mix (Code No. KMM-201; Toyobo (Shanghai) Biotech Co., Ltd., China), following the manufacturer’s instructions. The primers used were GmMT1-F and GmMT1-R, which were designed based on sequences flanking GmMT1 in the Phytozome database (Goodstein et al., 2012), with the 5′ ends modified to include BglII and BstEII restriction sites (Supplementary Table 3). The RT-PCR cycling conditions were as follows: 5 min at 94°C; 35 cycles of 30 s at 94°C, 30 s at 60°C, and 45 s at 72°C; and a final 10 min at 72°C. The purified PCR products were ligated into the pGM-T vector (VK207, Tiangen). Positive clones expressing the correct sequence were further inserted into pCAMBIA3301 vector using double digestion and ligation. The bar gene was used as a selection marker in the pCAMBIA3301 vector. Two expression vectors (35S:GmMT1 or 35S:bar) were constructed for transformation using the recombinant pCAMBIA3301 plasmid.

The full-length coding region of GmMT1 was inserted into the pCAMBIA1302 vector under the control of the 35S promoter to generate a GFP-fused GmMT1 vector (35S:GmMT1-GFP). This recombinant vector and the control vector (35S:GFP) were transfected into separate groups of Arabidopsis protoplasts following Yoo et al. (2007). The transfected cells were examined and imaged under a confocal laser scanning microscope (DMi8, Leica, China). The primer sequences GmMT1-GFP-F and GmMT1-GFP-R were used for subcellular localization (Supplementary Table 3).

To verify that GmMT1 expression was associated with isoflavone content in soybeans, we overexpressed this gene in Arabidopsis and soybean plants. We selected “Dongnong50,” a low-isoflavone soybean cultivar, and A. thaliana Col-0 as the recipients of genetic transformation. The A. thaliana Col-0 mutant SALK_012168C was obtained from the Arabidopsis Biological Resource Center (ABRC). Transgenic soybean plants were grown in a greenhouse under a 16 h light/8 h dark photoperiod at 25–26°C; T0–T3 transgenic Arabidopsis plants were grown in a growth chamber with a photoperiod cycle of 16 h light/8 h dark at 22°C.

To develop transgenic Arabidopsis, the 35S:GmMT1 construct was transferred into Agrobacterium tumefaciens EHA105, and then transformed into two Arabidopsis strains (the Col-0 ecotype and the mt1 mutant) using on the floral dip method (Clough and Bent, 1998). Transgenic Arabidopsis lines were selected using phosphinothricin; bar and GmMT1 gene expression in the selected plants was verified by PCR amplification using specific primers (3301-bar-F/R and 3301-35S-GmMT1-F/R; Supplementary Table 3). We generated three independent T2 transgenic lines per strain. Isoflavone concentrations in the T2 transgenic plants were measured using HPLC as described above. T2 transgenic plants with no separation were used in subsequent experiments.

To test whether GmMT1 overexpression increased isoflavone production in the roots of low-isoflavone soybean cultivars, the 35S:GmMT1 recombinant plasmid was transformed into Agrobacterium rhizogenes strain K599, and transformed in to the hairy roots of cultivar Donong50 as previously described (Cao et al., 2009). After 2 weeks of cultivation, once the hairy roots had appeared, we removed ∼1 cm sections from the root tips of 205 plants. Using PCR and specific primers (3301-bar-F/R and 3301-35S-GmMT1-F/R; Supplementary Table 3), we confirmed bar and GmMT1 gene expression in roots. Isoflavone concentrations in the transgenic roots were measured as described above.

To determine whether the expression of GmMT1 influenced isoflavone content in soybean seeds, we then developed transgenic soybeans by introducing the recombinant plasmid into Agrobacterium tumefaciens strain EHA105, and performing stable transformation into the cotyledon nodes of soybean cultivar “Dongnong50” following Paz et al. (2004). The expression of GmMT1 in leaves of T2 transgenic soybean plants was verified using PCR amplification (3301-bar-F/R and 3301-35S-GmMT1-F/R) (Supplementary Table 3), western blotting, and qRT-PCR. Isoflavone concentrations in seeds of the T2 transgenic plants were measured as described above. T2 plants expressing GmMT1 were used for all subsequent experiments.

We tested the effects of salt and drought stress on the transgenic Arabidopsis and soybean plants. To test the effects of salt and drought on Arabidopsis, we planted wild-type (Col-0), mt1 mutant, and T3 transgenic (GmMT1-ox and GmMT1-ox mt1) Arabidopsis on MS agar plates (five plates per strain). We then treated three plates per strain with 100 Mm NaCl, and three plates per strain with 300 mM mannitol. The remaining untreated plates were used as controls. All plants were kept at 4°C for 3 days in the dark, and then transferred to a 22°C environment with a photoperiod cycle of 16 h light/8 h dark. We calculated the germination rate on the 4th day after treatment and measured the root growth of all plants using a Vernier caliper.

To explore whether isoflavone accumulation, driven by GmMT1 overexpression, improved soybean resistance to P. sojae, we measured GmMT1 transcript abundance in transgenic and wild-type hairy roots, as well as the reaction of transgenic and wild-type hairy roots to P. sojae infection. Before P. sojae infection, we removed ∼1 sections from the hairy root tips of wild-type and GmMT1-overexpressing “Donong50” that had been cultured for 15 days. The relative expression levels of GmMT1 were detected in these samples. Then, to test the effects of P. sojae infection, well-propagated cultures of P. sojae were cut into small pieces and placed on new carrot agar (CA) plates. The hairy roots were spread on the CA medium with the mycelia, and were cultured in a growth chamber under a 16 h light/8 h dark cycle at 25°C. After ∼14 days of incubation, once the hairy roots had appeared, we imaged the hairy roots to check for signs of infection.

Two soybean varieties (Glycine max (L.) Merr), one with high isoflavone content (“Zhongdou27”), and one with low isoflavone content (“Dongnong8004”), were planted in two locations GmMT1 Effects Soybean Isoflavone Content (Xiangyang and Hulan, China) to verify the difference in isoflavone content between the accessions. In both locations, isoflavone content was significantly higher in “Zhongdou27” than in “Dongnong8004” (P < 0.001) (Supplementary Table 4). Thus, these accessions were suitable for map-based QTL analysis. We therefore crossed “Zhongdou27” (male parent) with “Dongnong8004” (female parent) to derive an RIL mapping population of 119 F_5:18 families. When the RIL were planted in Xiangyang and Hulan, isoflavone contents varied widely with location and population. Across the RIL population, total isoflavone content ranged from 1211.99 μg/g to 5683.43 μg/g (Supplementary Table 4). All traits of interest (i.e., total isoflavone content and content of each individual isoflavone) were continuously distributed (Supplementary Figure 1).

Genome resequencing was conducted to genotype the parental lines (“Zhongdou27” and “Dongnong8004”) and the 119 RILs in the mapping population. For “Zhongdou27,” we generated 26.18 GB of raw data, and 78.88% of the reads were successfully aligned to the soybean reference genome with an average depth of 24.54-fold; for “Dongnong8004,” we generated 25.58 GB of raw data, and 78.30% of the reads were successfully aligned to the soybean reference genome with an average depth of 22.86-fold (Supplementary Table 5). A total of 343,907 high-quality SNPs were identified between the two parents. Across all RILs, we generated 441.7 GB of raw data, with an average sequencing depth of 3.65-fold. For each of the 119 RILs, we generated an average of 3.46 GB of raw data (Supplementary Table 5). A total of 353 million SNPs were identified among the 119 RILs; all SNP sites in the RILs were integrated as recombination bin units. Finally, a genetic linkage map with 4231 bin markers was constructed along the 20 chromosomes (Supplementary Table 6). The total length of the bin map was 2172.98 centimorgans (cM), with a mean interval between markers of 0.82 cM (Supplementary Figure 2 and Supplementary Table 6).

Using linkage mapping, we identified 15 QTLs associated with daidzein (DZ), genistein (GC), glycitein (GT), and total isoflavone contents (TI), covering 10 of the 20 soybean chromosomes. Almost all QTLs were detected in both locations (Xiangyang and Hulan) or as multiple-effect QTLs, controlling the abundance of two or more isoflavones; the one exception was QTL qGC1-2, which was only associated with glycitein content in seeds from plants grown in Xiangyang (Table 1 and Supplementary Figure 3). Four of the 15 QTLs have been previously reported (Table 1). The remaining 11 QTLs were novel (Table 1). One novel locus, qISO19-1, which was detected between markers Bin2464 and Bin2465 on chromosome 19, was identified both by the inclusive composite interval mapping method (ICIM) (Figure 1A and Table 1). This QTL controlled the contents of daidzein, genistein, and total isoflavone (Figure 1B and Table 1). The genetic contribution of this QTL to each of these traits was more than 10%, indicating that qISO19-1 was an important QTL controlling individual and total content of soybean isoflavone. Thus, we selected qISO19-1 for further study.

Quantitative trait loci qISO19-1 was fine-mapped using 75 simple sequence repeat (SSR) markers near qISO19-1 (between markers Bin2464 and Bin2465; Figure 1A). Of these SSR markers, 12 (M1–M12) were polymorphic between the parental lines, as well as within the BC₂F₃ population. The local saturation map of qISO19-1 for daidzein and total isoflavone contents showed that the logarithm of the odds (LOD) scores were above the threshold from M4 (SSR_19_0116) to M5 (SSR_19_0123) (Figures 1B,C). Within the fine-mapped population, we identified five recombinants that were homozygous for “Dongong8004” alleles at M4 and M5, and five recombinants that were homozygous or heterozygous for the “Zhongdou27” at these alleles (Figure 2A). Levels of daidzein and total isoflavone in the seeds produced by the lines carrying the “Zhongdou27” alleles were significantly higher (P < 0.001) than those produced by the lines carrying the “Dongong8004” alleles (Figure 2B). Thus, the qISO19-1 QTL was narrowly defined to a 62.9-kb region between markers SSR_19_0116 and SSR_19_0123 on chromosome 19. Based on comparisons with the reference genome (G. max Williams 82) (Schmutz et al., 2010), this interval harbors five putative genes (Figure 1A): Glyma.19G017200, Glyma.19G017300, Glyma.19G017400, Glyma.19G017500, and Glyma.19G017700 (Supplementary Table 7).

Relative expression levels of these five genes were quantified for 43 soybean germplasms: 20 germplasms with high isoflavone content (3709—5970 μg/g) and 23 germplasms with low isoflavone content (1263—2042 μg/g) (Supplementary Table 1). Four genes, Glyma.19G017200, Glyma.19G017300, Glyma.19G017400, and Glyma.19G017500, were expressed in the seeds of all accessions during the late R6 stage (Supplementary Figure 4). Glyma.19G017700 was expressed at a relatively low level in 30 accessions (Supplementary Figure 4). The relative expression level of Glyma.19G017500 was significantly positively correlated with isoflavone content in the seeds of all accessions (Supplementary Table 8). Thus, we considered Glyma.19G017500, which is a methyltransferase (MT), a candidate gene at the qISO19-1 locus. This gene was named GmMT1.

The full-length coding sequence (CDS) of GmMT1 (Glyma.19G017500) from the high-isoflavone cultivar “Zhongdou27” was 957 bp long; this sequence was 99.16% consistent with the reference genome (G. max Williams 82; Schmutz et al., 2010; Supplementary Figure 5a). GmMT1 encoded a putative protein composed of 318 amino acid with a predicted molecular mass of 35.16 kDa. This protein harbored all 20 amino acids, with the most abundant being leucine and least abundant being tryptophan. The calculated instability and aliphatic indexes of this protein were 51.94 and 89.47, respectively, suggesting that the protein was unstable. The predicted 3-D structure suggested that GmMT1 encoded an m² G966 specific 16S rRNA methyltransferase (Lesnyak et al., 2007; Supplementary Figure 5b). Using the Phytozome database (Goodstein et al., 2012), we identified a second copy of GmMT1 on chromosome 13 (Glyma.13G066900); the amino acid sequence putatively encoded by this copy was 82.15% homologous with GmMT1 (Supplementary Figure 6). Phylogenetic analysis of GmMT1 and several other plant methyltransferases indicated that GmMT1 formed a well-supported clade with the methyltransferases of Phaseolus vulgaris, Medicago truncatula, and Trifolium pratense (Supplementary Figure 5c), suggesting that the functions of MT1 may be conserved across leguminous plants.

At developmental stage R7, GmMT1 was significantly upregulated in the seeds of the high-isoflavone cultivar “Zhongdou27” as compared to the seeds of the low-isoflavone cultivar “Dongnong8004” (Figure 3A). At all other developmental stages tested, GmMT1 expression level did not differ between cultivars. However, the stark difference in GmMT1 expression at stage R7 suggested that this gene may participate in isoflavone accumulation in soybean seeds. GmMT1 expression was also induced in the high-isoflavone cultivar “Zhongdou27” in response to salt and drought stress (Figures 3B,C). In both cases, GmMT1 was significantly upregulated at 12 h after stress initiation in comparison to the control group (Figures 3B,C).

We expressed the green fluorescent protein (GFP) and the GmMT1-GFP fusion protein, both under the control of the 35S promoter, in separate Arabidopsis protoplasts. In protoplasts carrying 35S:GFP, GFP signal was dispersed throughout the cell (Figure 4). However, in protoplasts carrying 35S:GmMT1-GFP, the GFP signal was primarily observed in the chloroplasts (Figure 4), indicating that GmMT1 was a chloroplast-localized protein.

We used Agrobacterium rhizogenes-mediated transformation to overexpress GmMT1 in the hairy roots of soybean cultivar “Donong50,” a low-isoflavone cultivar. In the hairy roots overexpressing GmMT1 (Supplementary Figures 7a,b), total isoflavone concentrations were significantly greater than in the wild-type hairy roots (Supplementary Figure 7c).

We then used Agrobacterium tumefaciens-mediated transformation to co-overexpress GmMT1 and the selection marker gene bar in “Dongnong50,” a low-isoflavone cultivar that is also an excellent transgenic receptor, to generate three independent T2 transgenic soybean lines (Figure 5A). In the seeds of the T2 transgenic lines, GmMT1 was significantly upregulated as compared to non-transgenic “Dongnong50” (Figure 5B), and western blots confirmed the expression of the bar protein in the transgenic plants (Figure 5C). Isoflavone contents in the seeds of the GmMT1-overexpressing T2 transgenic lines were significantly greater than isoflavone contents in the seeds of the wild-type plants (increases of nearly 3.0-fold; Figure 5D). Thus, our results suggested that GmMT1 expression might be associated with isoflavone biosynthesis in both soybeans and Arabidopsis.

Using Agrobacterium-mediated transformation, we overexpressed GmMT1 in wild-type Arabidopsis thaliana Columbia-0 (Col-0) and in A. thaliana mt1, a mutant strain in which the Arabidopsis homolog of GmMT1 (At3G28460) was silenced, to generate GmMT1-ox and of GmMT1-ox mt1; three independent T3 lines of each transgenic strain were generated (Supplementary Figures 8a–c). Under control conditions, germination rate did not differ among wild-type (Col-0), mt1 mutant, and T3 transgenic (GmMT1-ox and GmMT1-ox mt1) Arabidopsis seeds (Figures 6A,B). When exposed to salt or drought stress, seed germination rate in the wild-type and the mt1 mutant decreased significantly (Figures 6A,B). In the GmMT1-ox plants, seed germination rate decreased in response to both salt and drought stress (significantly in the latter case), but remained significantly greater than those in stressed non-GmMT1-overexpressing plants (Figures 6A,B). In contrast, in GmMT1-ox mt1 plants, seed germination rates decreased significantly in response to both salt and drought stress, and seed germination was significantly greater than stressed non-GmMT1-overexpressing plants after salt stress, but not drought stress (Figures 6A,B). Root growth measurements returned similar results to the seed germination assays. That is, after salt or drought stress, the roots of transgenic wild-type and mt1 Arabidopsis plants were longer than those of plants not overexpressing GmMT1 (Figures 6C,D).

The relative growth rates of the hairy roots of low-isoflavone soybean cultivar Donong50 decreased significantly with respect to both length and weight after exposure to drought or salt stress (Figures 6E–G). However, the hairy roots of transgenic Donong50 overexpressing GmMT1 were significantly more tolerant of drought and salt stress (Figures 6E–G). Indeed, as compared to the non-transgenic cultivar, relative rates of root length and weight increase in the transgenic cultivar were respectively 12.13% and 11.78% greater after salt stress, and respectively 7.91% and 9.15% greater after drought stress (Figures 6E–G). Thus, GmMT1-driven increases in isoflavone concentrations might increase the resistance of plants to salt and drought stress.

Under control (uninfected) conditions, GmMT1 was significantly more highly expressed in the three transgenic Donong50 lines overexpressing GmMT1 than in the non-transgenic low-isoflavone soybean cultivar Donong50 (Figure 7A). Correspondingly, the symptoms of P. sojae infection, including watery and rotting lesions, were less obvious in the transgenic hairy roots as compared to the non-transgenic hairy roots after 2 weeks of treatment (Figure 7B). Thus, our results indicated that the overexpression of GmMT1 might increase the resistance of soybeans to P. sojae infection, and/or reduce the severity of the infection.