Suppressed Methionine γ-Lyase Expression Causes Hyperaccumulation of S-Methylmethionine in Soybean Seeds.

Several soybean ( Glycine max L.) germplasms, such as Nishiyamahitashi 98-5 (NH), 55 have an intense seaweed-like flavor after cooking because of their high seed 56 S -methylmethionine (SMM) content. In this study, we compared the amounts of amino 57 acids in the phloem sap, leaves, pods, and seeds between NH and the common soybean 58 cultivar Fukuyutaka (FY). This revealed a comparably higher SMM content alongside a 59 higher free methionine (Met) content in NH seeds, suggesting that the 60 SMM-hyperaccumulation phenotype of NH soybean was related to Met metabolism in 61 seeds. To investigate the molecular mechanism behind SMM hyperaccumulation, we 62 examined the phenotype-associated gene locus in NH plants. Analyses of the 63 quantitative trait loci in segregated offspring of the cross between NH and the common 64 soybean cultivar Williams82 indicated that one locus on chromosome 10 explains 65 71.4% of SMM hyperaccumulation. Subsequent fine-mapping revealed that a 66 transposon insertion into the intron of a gene, Glyma.10g172700 , is associated with the 67 SMM-hyperaccumulation phenotype. The Glyma.10g172700- encoded recombinant 68 protein showed Met- γ -lyase (MGL) activity in vitro , and the transposon-insertion 69 mutation in NH efficiently suppressed Glyma.10g172700 expression in developing 70 seeds. Exogenous administration of Met to sections of developing soybean seeds 71 resulted in transient increases in Met levels, followed by continuous increases in SMM 72 concentrations, which was likely caused by Met methyltransferase activity in the seeds. 73 Accordingly, we propose that the SMM-hyperaccumulation phenotype is caused by 74 suppressed MGL expression in developing soybean seeds resulting in transient 75 accumulation of Met, which is converted into SMM to avoid the harmful effects caused 76 by excess free Met.


2009). 119
Based on accumulating studies of Met regulation in plant tissues, numerous 120 attempts to improve Met levels in crops have been made through genetic engineering 121 particularly of the Asp family pathway (Hacham et al., 2008, Hanafy et al., 2013, Song 122 et al., 2013, Cohen et al., 2014, Kumar & Jander, 2017, Amir et al., 2019. Acceleration 123 of SMM transport from non-seed tissues to seeds was also attempted to increase Met 124 levels in seeds (Lee et al., 2008, Cohen et al., 2017a. As such, attempts to increase seed 125 Met levels are becoming more successful but are sometimes disturbed by abnormal 126 phenotypes of the plants (Krishnan & Jez, 2018, Amir et al., 2019. Severe growth 127 retardation was observed in potato (Solanum tuberosum) plants overexpressing the 128 feedback-insensitive CGS to form more Met and β-zein to store Met (Dancs et al., 129 2008). Tobacco (Nicotiana tabacum) plants overexpressing CGS and with elevated free 130 Met levels also had increased sensitivity to oxidative stress (Hacham et al., 2017). In 131 addition, Arabidopsis seeds overexpressing a mutant CGS accumulated Met to 2.5-fold 132 higher levels, and these conditions were associated with increased expression of 133 stress-related transcripts (Cohen et al., 2014, Cohen & Amir, 2017. These studies 134 suggest that Met levels are tightly controlled in plant tissues and that excessive free Met 135 is deleterious to plant health. Yet, the mechanisms by which Met levels are regulated in 136 some tissues remain poorly understood. 137 The soybean SMM-hyperaccumulation phenotype, like that of NH, was 138 assumed to be attributable to a genotype related to Met metabolism in seeds. We 139 investigated the genotype of SMM-hyperaccumulating soybean plants and identified the 140 gene that is responsible for SMM accumulation in soybean. Through analyses of gene 141 function and NH phenotypes, we propose a mechanism underlying SMM 142 hyperaccumulation. between the FY and NH seeds (Supplemental Table S1). The contents of SMM and free Met were determined with four F1 seeds after reciprocal 184 crossing of the FY and NH cultivars. The hyperaccumulation of SMM was only evident 185 in self-pollinated NH seeds, and the maternal and paternal genotypes did not play a 186 significant role in the accumulation of SMM in seeds (Fig. 4). Moreover, F2 seeds of the 187 FY × NH cross were segregated into high SMM/low SMM at a ratio of 3/17 with a 188 consistent segregation ratio of 1:3 (Chi-squared test, P = 0.30; Supplemental Fig. S2A). 189 These data indicate that hyperaccumulation of SMM is essentially regulated by a single 190 recessive allele. In order to identify the gene responsible for hyperaccumulation of 191 SMM, we crossbred NH to Williams 82 (WI) cultivar. WI was used because the SMM 192 contents in WI seeds were as low as those in FY seeds (see below) and also because the 193 reference genome sequence was produced with WI (Schmutz et al., 2010). A total of 194 156 F5 recombinant inbred lines (RILs) were generated from the cross between NH and 195 WI, and SMM levels in their mature seeds were determined (Supplemental Fig. S2B). ). Moreover, the protein sequence has a motif (Ser237-Xaa-Xaa-Lys240) that is 245 conserved in pyridoxal-5'-phosphate (PLP) enzymes of the γ subfamily and associates 246 with the cofactor PLP (Martel et al. 1987, Sato & Nozaki, 2009 (Supplemental Fig. S5). 247 Tyr142, Asp216, and Arg410 residues are also involved in substrate binding and 248 catalysis at appropriate positions (Goyer et al. 2007), and Gly144 is conserved as in 249 Arabidopsis and melon MGLs that retain restricted substrate specificity for L-Met 250 (Gonda et al. 2013) (Supplemental Fig. S5). 251 Glyma.10g172700 cDNA was cloned using RNA that was extracted from 252 developing WI seeds. We expressed the recombinant protein as an N-terminal 253 His-tagged protein and purified it using Ni 2+ -affinity chromatography (Fig. 7A). 254 Subsequently, L-Met reacted with the recombinant protein in the presence of PLP, and 255 the products were converted into their 3-methyl-2-benzothiazolinone hydrazone 256 derivatives. This derivatization resulted in increased absorption at 320 nm ( Fig. 7B), 257 suggesting the formation of an aliphatic carbonyl compound (Esaki & Soda, 1987, 258 Inoue et al., 1995. To confirm its structure, the reaction product that was extracted 259 using ethyl acetate was reacted with N,O-bis(trimethylsilyl)trifluoroacetamide and was 260 then analyzed using GC-MS. A peak at the retention time of 9.5 min was assigned as 261 trimethylsilylated 2-ketobutyric acid by comparing its MS profile and retention time 262 with that prepared from a standard compound ( Fig. 7C and 7D). Accordingly, we 263 concluded that Glyma.10g172700 encodes MGL that catalyzes γ-elimination of L-Met.

264
We denoted the gene GmMGL1. This reaction had optimal activity at pH 7.0 and 265 followed Michaelis-Menten kinetics, with Km and Vmax values of 7.72 mM and 0.55 266 µmol mg −1 min −1 , respectively (Supplemental Fig. S6). 267 sequences found in several plant species indicated that GmMGL1 is located in a clade 278 different from the one GmMGL2 and GmMGL3 belong to (Supplemental Fig. S7). 279 The RT-qPCR analyses of the FY seeds showed that GmMGL1 mRNA 280 expression was enhanced at the early stage of seed maturation (from stages one to two) 281 and remained constant thereafter until the matured green stage (stage five; Fig. 8A). 282 However, GmMGL1 expression was considerably lower in NH seeds than in FY seeds 283 throughout seed development and differed little between developmental stages. 284 GmMGL2 and GmMGL3 expression levels were transiently induced during stage three, 285 but only in NH seeds, and they were not significantly different between FY and NH 286 cultivars at the other stages. The MGL activity in crude protein extracts prepared from 287 developing seeds (at stage four) of NH (5.86 ± 0.81 nmol h -1 g -1 ) was significantly 288 lower than that detected in FY seeds (12.1 ± 2.28 nmol h -1 g -1 ) (P < 0.05, Student's t-test, 289 n=4). GmMGL1 expression was significantly lower in the leaves, stems, and roots of 290 NH plants than in the leaves, stems, and roots of FY plantlets at the leaf-expansion stage 291 before flowering (Fig. 8B). Among these, the transcript levels of GmMGL2 and 292 GmMGL3 were highest in the leaves and did not differ significantly between the 293 soybean cultivars. 294 Because the genes of Met metabolism are regulated coordinately (Liao et al. 295 2012), we examined the effects of GmMGL1 suppression on the expression of 296 cystathionine γ-synthase (CGS), which catalyzes a key regulatory step of the Met 297 biosynthetic pathway (Hesse & Hoefgen, 2003), and of Met methyltransferase (MMT) 298 and homocysteine methyltransferase (HMT), which are directly involved in the 299 formation and decomposition of SMM ( Fig. 1; Cohen et al., 2017a). In SoyBase 300 BLAST searches using AtCGS (At3g01120), AtMMT (At5g49810), and AtHMT1 301 (At3g25900) as queries, two CGS homologs (Glyma.18g261600 and Glyma.09g235400; 302 referred to as GmCGS1 and GmCGS2, respectively), two MMT homologs 303 (Glyma.12g163700 and Glyma.16g000200; GmMMT1 and GmMMT2, respectively), 304 and three HMT homologs (Glyma.08g261200, Glyma.19g158800, and

Administration of Methionine Causes Accumulation of S-Methylmethionine in 311
Developing Seeds 312 313 Suppression of GmMGL1 expression in developing soybean seeds might lead to the 314 accumulation of Met, which would otherwise be catabolized to ammonia, methanethiol, 315 and 2-ketobutyric acid. One of the alternative fates of free Met is the formation of SMM 316 via the activity of MMT (Fig. 1), which is likely to occur in developing soybean seeds 317 because of the substantial expression levels of GmMMT1 and -2 (Supplemental Fig. S8). 318 To examine whether MMTs are active in developing soybean seeds, we conducted a 319 Met-feeding experiment. We fed free Met solution onto slices of immature green 320 soybean seeds of the FY and NH cultivars and determined SMM and Met contents using 321 LC-MS/MS (Fig. 9). Inclusion of 1 or 5 mM Met in the solution covering the cut 322 surfaces of the FY seeds yielded incremental increases in the SMM contents, and after 323 24 h of treatment, the SMM levels increased up to 37.9 and 135 µg g -1 for the 1 and 5 324 mM Met solutions, respectively. The SMM levels of the NH seeds also showed similar 325 incremental increases, but in a more prominent manner, and after 24 h, the SMM levels 326 increased up to 214 and 316 µg g -1 for the 1 and 5 mM Met solutions, respectively. The 327 SMM level in the NH seeds treated only with water also significantly increased to 80.0 328 µg g -1 after 24 h. No significant difference in the Met levels was observed for FY and 329 NH seeds treated with 1 mM Met in comparison with the levels in seeds treated with 330 water except those after 24 h with NH seeds; however, following feeding with 5 mM 331 Met solution, the Met levels increased significantly in both FY and NH, with more 332 prominent increases in NH. The highest Met level for FY was 9.71 µg g -1 at 8 h and for 333 NH was 27.2 µg g -1 at 24 h. 334 To confirm substrate-product relationships, we performed feeding 335 experiments using 13 C-Met (C5, 99 atom %) with NH seeds. Subsequent LC-MS/MS 336 analyses of SMM in the extract confirmed that it was predominantly formed from 337 DISCUSSION 343 344 Using a molecular genetic approach to locate the allele responsible for 345 hyperaccumulation of SMM, we found that a transposon insertion into the intron of 346 GmMGL1 is strongly associated with SMM hyperaccumulation in soybean seeds. 347 Expression of the GmMGL1 gene and, accordingly, MGL activity in seeds were 348 suppressed due to the transposon insertion. Under these conditions, Met catabolism 349 would be low in seeds, leading to Met accumulation. Because excess Met levels have 350 been associated with various adverse effects in plant tissues, we hypothesized that 351 surplus Met in soybean seeds with MGL deficiencies was converted to the 352 better-tolerated compound SMM by MMT activity (Fig. 10). In line with this hypothesis, 353 Met-feeding experiments showed that surplus Met was efficiently converted into SMM 354 in green mature soybean seeds and that the conversion was more prominent with 355 MGL-deficiency. Alternatively, in a previous study of soybeans, transposons in or near a gene were 368 related to increased CHG/CHH methylation, and consequently, lower expression levels 369 (Kim et al., 2015). Hence, epigenetic regulatory mechanisms likely play roles in the 370 present repression of GmMGL1. As such, analyses of DNA methylation should be one 371 of the next priority research areas to reveal more details about the mechanisms of this 372 type of gene suppression. 373

Suppression of GmMGL1 Accounts for SMM Hyperaccumulation
In plant tissues, Met levels are tightly regulated through biosynthesis and catabolism 377 ( Fig. 1). Higher free Met contents (in addition to SMM contents) in NH seeds than in 378 FY seeds prompted us to assume that GmMGL1 participated in controlling free Met 379 levels in seeds. If this is the case, in the absence of substantial MGL activity as found 380 with NH, free Met levels should increase, and surplus Met could be converted into 381 SMM by the MMT activity in seeds. This scenario showed no fundamental 382 inconsistency in the results obtained in this study about the Met metabolism of NH. The 383 function of MGL to adjust free Met levels has been demonstrated with Arabidopsis, in 384 which knock-out of the AtMGL gene increased free Met contents in leaves, flowers, and 385 seeds (Goyer et al., 2007, Joshi & Jander, 2009. Notably, the Arabidopsis knock-out 386 mutant contained 4.5-fold higher SMM contents in leaves than its parental wild type; 387 therefore, it is presumed that conversion of surplus Met to SMM is common among 388 plants. In support of this hypothesis, SMM accumulation has been reported in multiple 389 transgenic plants with high free Met levels (Kim et al., 2002, Hacham et al., 2008, 390 Hacham et al., 2017. Taken together, it is suggested that GmMGL1 was involved in 391 controlling free Met levels in developing soybean seeds. SMM-hyperaccumulation is 392 likely to be a consequence of suppressed GmMGL1, and a subsequent "fail-safe" 393 system employing MMT activity to avoid the adverse effect of excess Met. 394 One of the MGL products, 2-ketobutyric acid, is partly converted to Ile in 395 Arabidopsis, especially under drought stress (Rébeillé et al., 2006, Joshi & Jander, 396 2009). However, we found no significant difference in either free or total Ile content 397 between NH and FY seeds, suggesting that GmMGL1 accounted little for Ile formation 398 in developing seeds. On the contrary, dry NH seeds had higher free Thr, Phe, His, and 399 homoserine levels in addition to increased free Met and SMM levels. Therefore, the 400 MGL deficiency is likely to cause pleiotropic effects on the metabolism of other amino 401 acids. Accumulation of free amino acids was often observed in several transgenic plants 402 generated to enhance Met levels (Hanafy et al., 2013, Cohen et al., 2014, Hacham et al., 403 2017, Huang et al., 2014. Accordingly, it is suggested that the adverse effect of surplus 404 Met in NH induced a stress response that led to higher Thr, Phe, His, and homoserine 405

levels. 406
It has been reported for several plant species, including Arabidopsis and wheat 410 (Triticum aestivum), that SMM formed in vegetative tissues is transported to seeds 411 through the phloem (Bourgis et al., 1999, Lee et al., 2008, Frank et al., 2015, Cohen et 412 al., 2017b. Therefore, it was assumed that the phloem transportation of SMM formed in 413 vegetative tissues to seeds could also be accountable for hyperaccumulation of SMM in 414 NH seeds. The level of SMM in the phloem exudate collected from leaves of NH was 415 higher than that found for FY plants. Therefore, transportation of SMM through the 416 phloem toward the seeds is likely to be at least partly accountable for the 417 hyperaccumulation of SMM in NH seeds. The SMM levels in the pods of both NH and 418 FY showed a tendency to decrease during maturation; thus, the transportation of SMM 419 from the pods to the seeds should also be taken into consideration. However, it was 420 remarkable that the SMM levels found in the phloem exudate of NH plants were 421 291-fold lower than that in Arabidopsis phloem exudate. Concordant with the fact that 422 amino acid levels in soybean phloem exudate were 8-fold lower than those in 423 Arabidopsis and wheat (Bourgis et al., 1999), our observation of low levels of SMM in 424 soybean phloem exudate prompted us to consider that the contribution of phloem 425 transport of SMM toward seeds for SMM hyperaccumulation is not negligible, but is 426 limited. Furthermore, the results of the reciprocal crossing of NH and FY indicated that 427 maternal as well as paternal genotypes did not play a substantial role in determining the 428 seed phenotype of SMM hyperaccumulation, which was caused only when the genotype 429 of the seeds was homozygous for mgl1, and thus, the involvement of vegetative organs 430 in SMM hyperaccumulation in the seeds of NH is likely limited. 431 The extensively lower levels of SMM and Met in the soybean phloem, compared to 432 levels in the Arabidopsis phloem, are noteworthy because the levels of free and total 433 amino acids including Met are more than 10-fold higher in soybean seeds than in 434 The addition of Met at 1 mM onto developing seeds of FY had only a slight effect 442 on Met or SMM concentrations, probably because Met, supplied exogenously, was 443 appropriately catabolized in part by intrinsic GmMGL1 activity in FY seeds. This

Determination of SMM contents 491
Seed coats were carefully removed and soybean seeds containing hypocotyls were then 492 powdered using a multi-beads shocker (PM2000, Yasui Kikai, Osaka, Japan) equipped 493 with stainless metal cones (MC-0316S, Yasui Kikai) and operated at 2500 rpm for two 494 30-s periods with a 10-s interval. Powder samples of 20 mg were then mixed with 1 mL 495 of distilled water containing 50 µg mL −1 L-methionine-S-methyl d6 sulfonium chloride 496 (d6-SMM; Toronto Research Chemicals Inc. Toronto, Ontario, Canada) and were then 497 placed in a water bath sonicator (US-2, SND Co. Ltd., Suwa City, Nagano, Japan) for 498 Chromatography was conducted using a Discovery HS F5 column (15 cm × 2.1 mm, 3 508 µm; Supelco, Bellefonte, PA, USA), and HPLC and MS analyses were performed using 509 previously described conditions (Morisaki et al., 2014). To quantify 13 C-labeled SMM  Table S2. 519 To collect phloem exudate, fully expanded leaves of NH and FY plants at 520 their seed-filling stage were detached with the base of the petiole under a solution of 20 521 mM EDTA (pH 7.0) and immersed into 0.2 mL of the same solution in 0.5-mL 522 microtubes, and placed in humid chambers in the dark at 25°C (Urquhart & Joy, 1981). 523 After 5 h, the EDTA solution was collected and used for the LC-MS/MS analyses as 524 described above to estimate the concentration of SMM and free Met. As a comparison, 525 fully expanded rosette leaves of Arabidopsis (ecotype Ws-0) were used at its 526 seed-filling stage. 527 528

Determinations of amino acid and protein contents 529
Soluble amino acids were extracted from dry seed flour (20 mg) as described by Hanafy 530 et al. (2013). Flour samples were suspended in 240 µL of 3% (w/v) sulfosalicylic acid 531 and were suspended with vigorous shaking for 30 min. After centrifugation at 12,000 × 532 g for 10 min at 25°C, precipitates were extracted two more times as described above. 533 Combined supernatants were then filtered and analyzed using LC-MS/MS with an ESI 534 interface (Tomita et al., 2016) as detailed above. An Intrada amino acid column (100 × 3 535 mm i.d., 3 µm; Imtakt, Kyoto, Japan) was used with a column temperature of 40°C. 536 Mobile phases were applied at 0.4 mL min − 1 and comprised solvents A 537 (acetonitrile/formic acid at 100/0.3, v/v) and B (0.1 M acetonitrile/ammonium formate 538 at 20/80, v/v) at 15% B for 10 min, followed by a linear increase from 15% B to 60% B 539 over 15 min, then from 60% B to 100% B over 5 min, and then 100% B for 10 min. The 540 injection volume was 4 µL. The MS system was operated in MRM mode using positive 541 ESI with a capillary voltage of 3000 V, a source temperature at 550°C, a curtain gas of 542 35 (arbitrary units), ion source gases 1 and 2 of 80 and 60 (arbitrary units), respectively, 543 a declustering potential of 16 V, and an entrance potential of 5 V. MRM transitions of 544 the precursor to product ions used for the quantification and collision energy are 545 summarized in Supplementary Table S2. Quantification was done using calibration 546 curves constructed using amino acid mixture standard solution (Wako Pure Chemicals, 547 Osaka, Japan) supplemented with homoserine and homocysteine (Wako Pure 548 Chemicals). HCl and served to LC-MS/MS analysis, as described above. To analyze total Cys and 563 Met, the flour was oxidized with 1 mL of performic acid at 0°C for 16 h to give cysteic 564 acid and methionine sulfone prior to acid hydrolysis. 565 566

Analyses of quantitative trait loci for S-Methylmethionine contents 567
RILs, including 155 F5 lines were developed from a single seed descendant of the cross 568 between NH and WI. Total DNA extraction and linkage map construction by simple 569 sequence repeat markers (WSGP ver. 2) were performed as described previously (Fujii 570 et al., 2018). SMM contents in each RIL were quantified using bulked F6 seeds that 571 were derived from the F5 individual. Because SMM contents varied widely among RILs, 572 QTL analysis was conducted with the common logarithm (log) value for contents. QTL analyses were performed using composite interval mapping, as 574 implemented in QTL Cartographer 2.5 software (Wang et al., 2005). The genome was 575 scanned at 1-cM intervals. One thousand permutation tests were conducted to determine 576 the threshold value of the limit of detection score.  Table  600 S3). The resulting PCR products were cloned into pGEM T-easy vectors (Promega, chloramphenicol (30 µg mL −1 ) at 37°C to an optical density of 0.6-0.8 at 600 nm. After 606 chilling the cultures on ice for 15 min, isopropyl β-D-1-thiogalactosylpyranoside was 607 added to a concentration of 1 mM, and cells were then cultured at 30°C for 16 h. 608 Cells from 50-mL cultures were recovered by centrifugation at 4,000 × g for 609 20 min at 4°C and were resuspended in 5 mL of 100 mM potassium phosphate buffer 610 (pH 7.5) containing 0.01% (w/v) dithiothreitol and 1.3 mM pyridoxal phosphate (PLP). 611 After the addition of 5-µL aliquots of 100 mM phenylmethane sulfonyl fluoride and 50 612 mg mL −1 lysozyme, suspensions were kept on ice for 15 min, and cells were then 613 disrupted using a tip-type ultrasonic disruptor (UD-211, Tomy Seiko, Tokyo, Japan). 614 After centrifugation at 12,000 × g for 10 min, supernatants were directly applied to a 615 column (2 mL) of Ni-NTA agarose (Nacalai Tesque, Kyoto, Japan) that had been 616 equilibrated with 100 mM potassium phosphate buffer (pH 7.5) containing 0.01% (w/v) 617 dithiothreitol and 10 µM pyridoxal phosphate (PLP). The column was then washed with 618 10 mL of the same buffer containing 10 mM imidazole, and His-tagged recombinant 619 proteins were eluted with 10 mL of the same buffer containing 250 mM imidazole. 620 Active fractions were finally combined and desalted using a PD-10 column (GE 621 Healthcare, Chicago, IL, USA). 622 623

(w/v) 3-methyl-2-benzothiazolone hydrazine hydrochloride (MBTH). Reaction tubes 632
were then tightly closed and incubated at 50°C for 40 min. The MGL product 633 2-ketobutyric acid was quantified according to absorbance at 278 nm, which is derived 634 from the MBTH derivative of 2-ketobutyric acid. The absorbance at 0 min was 635 subtracted from later measurements, and a calibration curve was generated using 636 authentic 2-ketobutyric acid (Sigma-Aldrich). The structure of 2-ketobutyric acid was 637 confirmed using GC-MS after converting the acid into a trimethylsilylated product To determine MGL activity in developing soybean seeds, the seeds of NH and 656 FY at seed developmental stage four (cf. the photo in Fig. 8) were homogenized with 657 four volumes of 50 mM sodium phosphate buffer (pH 7.5) containing 5% (w/v) sorbitol, 658 10 mM dithiothreitol, 5 mM sodium metabisulfite, and 2.5 µM PLP. After centrifugation 659 at 20,000 × g for 20 min at 4°C, the cleared supernatant (0.5 mL) was mixed with 0.25 660 mL of 0.2 M Met in a buffer (50 mM sodium phosphate, pH 7.5 containing 2.5 µM PLP) 661 in a total volume of 4.5 mL, and incubated at 30°C with a shaking water bath for 17 h. 662 Genomic DNA was isolated according to Hanafy et al. (2013). Total RNA was isolated 672 using the Qiagen RNeasy Plant Mini Kit according to the manufacturer's instruction. (At3g01120), AtMMT (At5g49810), and AtHMT1 (At3g25900) as queries, respectively. 685 Primers for genomic PCR and RT-qPCR are shown in Supplemental Table S3. 686 687

Met-feeding 688
Pods harboring the seeds of developmental stage two (cf. the photo in Fig. 8) were 689 collected and were gently removed with their seed coats. Thin sections of 1-mm 690 thickness were excised at the short axis using a razor blade, and they were immediately 691 placed on a sheet of Parafilm (Bemis Flexible Packaging, Chicago, IL, USA) in a glass 692 Petri dish. The inner surface of the Petri dish was covered with a moistened paper towel. Nishiyamahitashi 98-5 (NH) seeds from stages one to five (refer to Fig. 8  shown as means ± standard errors (SE) of four replicates. Significant differences were 780 identified using a two-way analysis of variance (ANOVA) after Box-Cox 781 transformation and Fisher's least significant difference test (LSD; P < 0.05). Different 782 lowercase letters indicate significant differences between the developing stages (P < 783 0.05, Tukey's HSD test after two-way ANOVA). Different capital letters indicate 784 significant differences between all treatments (P < 0.05, Tukey's HSD test after 785 two-way ANOVA). ***: P < 0.001, **: 0.001 < P < 0.01, NS: 0.05 < P (simple main 786 effect test after two-way ANOVA). shown as µg per g of mature dry seeds. The data are shown as means ± SE of three 791 replicates. Significant differences were identified using Student's t-test after Box-Cox 792 transformation (**: P < 0.01, *: 0.01 < P < 0.05, n.s.: 0.05 < P). HomoSer: homoserine, 793 HomoCys: homocysteine, n.d.: not detected.  Data are presented as means ± SE (n = 3). Significant differences were identified using 853 (SMM) and methionine (Met) contents after treating with 0 (circle and white)-, 1 864 (square and gray), and 5 (triangle and black) mM Met. Data are shown as means ± SE 865 (n = 3). Significant differences were identified using a two-way analysis of variance 866 , and seeds (without seed coats) (C) harvested at the flowering stage (F), the immature green seed stage (IG; corresponding to stage two in Fig. 8), and the mature green seed stage (MG; corresponding to stage five in Fig. 8) are shown as means ± standard errors (SE) of four replicates. Significant differences were identified using a two-way analysis of variance (ANOVA) after Box-Cox transformation and Fisher's least significant difference test (LSD; P < 0.05). Different lowercase letters indicate significant differences between the developing stages (P < 0.05, Tukey's HSD test after two-way ANOVA). Different capital letters indicate significant differences between all treatments (P < 0.05, Tukey's HSD test after two-way ANOVA). ***: P < 0.001, **: 0.001 < P < 0.01, NS: 0.05 < P (simple main effect test after two-way ANOVA). Nishiyamahitashi 98-5 (NH; gray bar) are shown as µg per g of mature dry seeds. The data are shown as means ± SE of three replicates. Significant differences were identified using Student's t-test after Box-Cox transformation (**: P < 0.01, *: 0.01 < P < 0.05, n.s.: 0.05 < P). HomoSer: homoserine, HomoCys: homocysteine, n.d.: not detected.  www.plantphysiol.org on July 9, 2020 -Published by     Fig. 8). The Glycine max 20S proteasome subunit (Glyma.06g078500) was used as an internal control in RT-qPCR analyses. Transcript levels relative to the internal control are shown as multiples of the lowest value of 1. Data are presented as means ± SE (n = 3). Significant differences indicated with different lowercase letters were identified using Tukey's HSD tests after Box-Cox transformation (P < 0.05).   Figure 8. Expression of GmMGL1, GmMGL2, and GmMGL3 in soybean plants. Expression of GmMGL1, GmMGL2, and GmMGL3 in developing seeds (A) and in the leaves, stems, and roots (B) of Fukuyutaka (FY) and Nishiyamahitashi 98-5 (NH) soybean cultivars. Sizes of NH and FY seeds collected for RNA extraction (stages one to five) are shown in the upper panel. The Glycine max 20S proteasome subunit (Glyma.06g078500) was used as an internal control in RT-qPCR analyses. Transcript levels relative to the internal control are shown as multiples of the lowest value of 1. Data are presented as means ± SE (n = 3). Significant differences were identified using a two-way analysis of variance (ANOVA) after Box-Cox transformation. *** above the symbols in A indicate significant differences between plant lines in each developing stage (P < 0.001, simple main effect test after two-way ANOVA). ** indicate significant differences between plant lines in B (MGL1) (P < 0.05, simple main effect test after two-way ANOVA). Different lowercase letters above columns indicate significant differences between organs in B (MGL2 and MGL3) (P < 0.05, Tukey' HSD test after two-way ANOVA ).  Intensity (cps) B Figure 9. Absorption and conversion of exogenously supplied methionine to S-methylmethionine in a section of a developing soybean seed. A. Smethylmethionine (SMM) and methionine (Met) contents after treating with 0 (circle and white)-, 1 (square and gray), and 5 (triangle and black) mM Met. Data are shown as means ± SE (n = 3). Significant differences were identified using a two-way analysis of variance (ANOVA) after Box-Cox transformation. Different letters above the symbols indicate significant differences between Met concentrations in each hour (P < 0.05, Tukey' HSD tests after simple main effect tests). B. Mass spectrum of SMM extracted from the seed sections treated with 5 mM 13 C5-Met for 24 h (upper panel). Reaction of MMT from 13 C5-Met is shown as an inset. The positions of 13 C in Met and SMM are shown with asterisks. Mass spectrum of non-labelled SMM is shown (lower panel). Tentative assignments of molecular and fragment ions are also shown.

A B
www.plantphysiol.org on July 9, 2020 -Published by Downloaded from Copyright © 2020 American Society of Plant Biologists. All rights reserved.