Involvement of sulfur assimilation in the low β subunit content of soybean seed storage protein revealed by comparative transcriptome analysis

Xi Zhng, Ruixin Xu, Wei Hu, Wn Wng, Dezhi Hn, Fn Zhng,Yongzhe Gu, Yong Guo, Jun Wng,*,Lijun Qiu,**

aCollege of Agriculture,Yangtze University,Jingzhou 434025,Hubei, China

bNational Key Facility for Gene Resources and Genetic Improvement, Institute of Crop Sciences, Chinese Academy of Agricultural Sciences,Beijing 100081,China

cSoybean Research Center,Heihe Institute,Heilongjiang Academy of Agricultural Sciences,Heihe 100142,Helongjiang, China

Keywords:Soybean β subunit Transcriptome analysis Sulfur-containing amino acids Sulfur assimilation

A B S T R A C T The β subunit of soybean[Glycine max(L.)Merr.]seed storage protein is of great significance in sulfur-containing amino acid balance and soybean processing properties. The objective of this study was to elucidate the relationship between the β subunit and sulfur-containing amino acid composition, and the potential regulatory mechanism. The β subunit was independently accumulated in comparison with other major subunits (α/α′, acidic, basic,and A3)during seed filling,and a low level of β subunit content(BSC)was formed during the accumulation process. In low-BSC mature seeds, crude protein, oil content, and fatty acid composition were not changed, but sulfur-containing amino acids (Cys + Met) in the low-BSC seeds increased significantly (by 31.5%), suggesting that an internal regulatory mechanism within seed might be responsible for the rebalance of seed protein composition and that sulfur assimilation might be deeply involved in β subunit accumulation.Transcriptomic analysis revealed that genes involved in anabolism of cysteine,methionine,and glutathione were up-regulated but those involved in the catabolism of these compounds were down-regulated, suggesting a relationship between the elevation of methionine and glutathione and low BSC.Our study sheds light on seed composition in low BSC lines and on the potential molecular regulatory mechanism of β subunit accumulation,broadening our understanding of soybean seed protein synthesis and its regulation.

1.Introduction

As a crop with high protein (～40%) and oil content (18%),soybean [Glycine max (L.) Merr.] is used both in animal feed and for direct human consumption.The salt-soluble globulins 11S glycinin and 7S β-conglycinin,the most abundant storage proteins,account for 70%of the total seed proteins[1].The 7S β-conglycinin consists of α′, α, and β subunits, and 11S glycinin contains mainly acidic and basic subunits [2].Soybean contains all eight amino acids essential for human nutrition, though relatively small amounts of methionine(Met) and cysteine (Cys). Thus, the sulfur-containing amino acids are a limiting nutritional factor in soybean[3].Generally,glycinin contains more sulfur-containing amino acids than βconglycinin. Among these three major subunits of βconglycinin, the nutritional values rank in the order α′ ＞α ＞β, with the β subunit lacking both Met and Cys [4].Despite the adverse nutritional role of the β subunit, it plays an important role in modulating the composition of seed storage protein. A negative correlation was observed [5,6]between the contents of 7S β-conglycinin and 11S glycinin and also between the contents of β subunit and 11S glycinin.A low content of β subunit results in the elevation of 11S glycinin content, possibly increasing the overall sulfurcontaining amino acid content [6]. The β subunit content(BSC) is thus of great significance to seed quality as determined by sulfur-containing amino acids. For soybean processing, the β subunit is beneficial for emulsification and thermostability[7,8].

The accumulation of the β subunit starts at ～30 days after flowering (DAF), and peaks at ～50 DAF. This accumulation pattern is similar to that of the A3 subunit, but distinct from those of the α′, α, acidic, and basic subunits [9]. The accumulation of the β subunit is affected by various factors including sulfur,nitrogen,nodulation,and plant hormones.β subunit content induced up to three fold under sulfur deficiency in comparison with normal sulfur supplementation, whereas the 11S subunit content was reduced by 40%[10]. During sulfur assimilation, O-acetyl-L-serine (OAS), the precursor of cysteine, was induced under sulfur deficiency[11,12], and exogenous application of OAS to immature cotyledons in vitro increased β subunit accumulation [13]. In vitro application of an excess of another intermediate product of sulfur assimilation, L-methionine, as a sulfur source strongly inhibited β subunit accumulation [14]. It is believed[13] that the influences of OAS and methionine on β subunit accumulation are independent and do not interact with each other. Glutathione, as a critical compound in sulfur assimilation, plays a significant role in response to biotic or abiotic stress. Its application inhibited the accumulation of the β subunit [15], and glutathione and OAS also acted independently to influence β subunit accumulation [16]. As the most essential nutrient for plant growth, nitrogen is also involved in the regulation of β subunit accumulation. Exogenous application of glutamine increased the mRNA expression of the β subunit [17]. When glutamine was used as the only nitrogen source,a transaminase inhibitor inhibited β subunit mRNA accumulation, whereas other amino acids including alanine, glutamate, and asparagine showed no effect on its accumulation. Nitrogen is involved not only in β subunit accumulation.It also plays a positive role in protein synthesis,but negatively regulates the accumulation of proteins rich in sulfur amino acids, such as Bowman-Birk protease inhibitor[18].Nodules in legumes fix atmospheric nitrogen and convert it into ammonia with the aid of symbiotic bacteria,supplementing the host plant with nitrogen [19]. Interestingly,seeds from non-nodulated soybean lines lack β subunit,whereas the α and α′ subunits of 7S conglycinin remain uninfluenced [20,21]. Sulfur and nitrogen are both important for plant growth and often work coordinately to regulate many physiological process. A higher sulfur:nitrogen ratio depressed the accumulation of β subunit [22,23]. However,although sulfur and nitrogen work interactively on plant growth to some degree, they tend to act individually to influence β subunit accumulation [24,25]. Abscisic acid(ABA), which plays an important role in seed maturation, is reported [11] to promote β subunit accumulation as well.Exogenous application of ABA increased sulfate deposition and OAS accumulation, and the authors proposed that the positive role of ABA in β subunit accumulation was more likely to be mediated by OAS than by sulfate, given that OAS could be induced under sulfur deficiency.

Despite the progress described above, the precise mechanism by which these factors influence β subunit accumulation has remained obscure, probably owing to the lack of available mutants in key regulatory pathways.We previously[6]reported a natural mutant with low BSC from the Chinese soybean core collection, which could be useful for the elucidation of BSC regulation. In this study, we evaluated the amino acid, FA composition, crude protein and oil content of low-BSC plants,and the potential relationship between BSC and sulfur assimilation was also investigated via transcriptome study.

2. Materials and methods

2.1. Plant materials

A low-BSC landrace (Yangyandou, YYD, accession ZDD17356)was identified in the Chinese soybean core germplasm collection (a collection of ～2% of the accessions representing 70%of the genetic diversity of all soybean germplasm)[26].F7residual heterozygous lines (RHL) were derived from a cross between YYD and Zhonghuang 13 (ZH13) [6]. Low-BSC and normal-BSC seeds, each from 20 RHL plants, were planted in 2016 in soil-filled pots in the Crop Planting Base of Yangtze University and grown under normal conditions. Experiments were performed in three independent replicates using a completely randomized design. Seeds of different developmental stages were collected for transcriptome studies and qRT-PCR using the sampling standard described in Fig.1.

2.2. Measurement of total protein and oil content

Crude protein and oil content was determined by nearinfrared reflectance (NIR) spectroscopy (DA7200, Perten Instrument, Huddinge, Sweden). Seeds from 10 plants each of the low-and normal-BSC groups were measured.Student's ttest was used for comparison of means.

Fig.1-Sampling standard and β subunit accumulation pattern.(A)Representative seed sizes at several developmental stages of ZH13.(B) Accumulation pattern of major subunits of 11S glycinin and 7S conglycinin during several seed developmental stages in seeds of low-BSC and normal-BSC lines as resolved by SDS-PAGE.

2.3. Separation of major subunits of soybean seed storage protein

Seeds of different developmental stages were ground to fine powder in liquid nitrogen with mortar and pestle and then dehydrated in a freeze dryer (lyophiliser CHRIST, ALPHA 1-2/LD Plus, Osterode am Harz, Germany). The major subunits of seed storage protein were extracted from 5 mg of seed powder with protein extraction buffer(0.05 mol L-1Tris base,5 mol L--1urea, 0.2 mg L-1sodium dodecyl sulfate, 0.025%bromophenol blue, pH 8.0, 0.1 mol L-1β-mercaptoethanol)and resolved by SDS-PAGE (12%). Bands were visualized by Coomassie brilliant blue(G-250)staining.

2.4. Amino acid content determination

Mature soybean seeds were ground to fine powder with a grinder.Dry powder of ～100 mg was then hydrolyzed for 22 h at 110 °C with 6 N HCl under a nitrogen atmosphere. Amino acids were determined on an amino acid analyzer (L-8500,Hitachi High-Technologies Corporation,Tokyo,Japan)according to the user manual. Determination was performed for three replicates of each sample.Two biological replicates each of low-BSC and normal-BSC seeds were analyzed.

2.5. Fatty acid composition determination

The fatty acid (FA) composition of soybean dry seeds was quantified following Wang et al. [19] with modifications. A sample of 500-1000 mg of soybean seed powder was supplemented with 100 mg pyrogallic acid, 2 mL ethanol (95%), and 2.0 mL methyl undecanate (C11:0) (used as internal standard for quantification), and the mixture was heated at 80 °C in a water bath for 40 min. FAs in the hydrolyzed mixture were then extracted from the aqueous into the organic phase three times with 50 mL petroleum ether each time. The FAs dissolved in the organic phase were then collected by rotary evaporation. Gas chromatography (GC) was performed on an Agilent 7890A(United States) GC-MS system equipped with a DB-23 capillary column (0.25 μm thickness, 30 m in length,and 0.25 μm diameter) and a flame ionization detector.Nitrogen was used as carrier gas and the temperature program started at 180 °C followed by a 3 °C min-1ramp-up to 230 °C, holding for 5 min. The peak area of each FA was calculated for the determination of absolute FA content. The relative contents of 12 FAs: C14:0, C15:0, C16:0, C17:0, C17:1,C18:0, C18:1, C18:2, C18:3, C20:0, C22:0, and C24:0, were calculated as the ratio of each FA peak area to the sum of peak areas of all FAs identified. Three biological replicates each of low- and normal-BSC were analyzed.

2.6. Transcriptomic analysis

Total RNA from immature seeds was extracted with TRIzol(Thermo Fisher Scientific, 15596026) according to the user manual. RNA concentration was measured with a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, USA).RNAs of S3-S7 from low- and normal-BSC RHL of each biological replicate were first combined in equal amounts,and then RNAs of different biological replicates were combined to form mixed pools of low BSC and normal BSC. The two RNA pools were then used for library construction and sequencing.A NEBNext Ultra RNA Library Prep Kit for Illumina(NEB, USA) was used to construct sequencing libraries following the manufacturer's recommendations. Briefly,poly-T oligo-attached magnetic beads were used to purify mRNA from total RNA. Then fragmentation was performed using divalent cations at elevated temperature. cDNA was then synthesized in two steps:first-and second-strand cDNA,using random hexamer primer, M-MuLV Reverse Transcriptase (RNase H free), and DNA polymerase I, RNase H respectively. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities.Adenylation of 3′ ends of DNA fragments was performed by ligating an adapter with a hairpin loop structure to prepare for hybridization.cDNA fragments preferentially with 150-200 bp in length were selected and then enriched by PCR with highfidelity DNA polymerase to create final cDNA libraries. The libraries were sequenced on an Illumina HiSeq 2000 platform and paired-end reads were generated.

2.7. qRT-PCR analysis

For RNA sequencing validation, real time qRT-PCR was performed for four genes using gene-specific primer sets(Table S1). First-strand cDNA was synthesized with a PrimeScript 1st Strand cDNA Synthesis Kit (TAKARA,D6110A). qRT-PCR was performed using SYBR Premix Ex Taq(Tli RNaseH Plus) (TAKARA, RR420A) by QuantStudio 6 Flex(ABI). The geometric mean of cons6/7/15 [34] was used as internal reference, and the relative expression level of each gene was determined by the delta-delta-cycle threshold (Ct)method [35]. Primers designed for qRT-PCR are described in Table S1.

2.8. Analysis of high-throughput data

For the raw RNA-seq data obtained by sequencing, a quality score calculated by Phred [27] was used for quality control.The raw RNA-seq reads were then aligned to the reference genome using TopHat2 [28]. Aligned fragments were then joined with Cufflinks [29] and compared with preliminary annotation results using Cuffcompare [30]. According to the location of mapped reads on the genome, the count of each fragment was indexed by fragments per kilobase of transcript per million fragments mapped (FPKM) and determined using Cuffquant and Cuffnorm implemented in Cufflinks based on the beta negative binomial distribution [30,31]. EBSseq [32]was used to perform normalization and differential expression analysis among different data sets. Differentially expressed genes (DEGs) were identified with normalized expression fold change ＞2 with P ＜0.01 and Benjamini-Hochberg adjusted P ＜0.01 as false discovery rate (FDR).Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed with Blast2GO[33]and KOBAS 2.0[36].

3. Results

3.1. Accumulation pattern of β subunit in low-BSC seeds

In normal-BSC seeds the β subunit started to accumulate at the S4 stage and was abundant in the S5 and S6 stages, and accumulation ceased at the S7 stage (Fig. 1-B). Only trace amounts of the β subunit accumulated during the S4-S7 stages in low-BSC seeds (confirmed by LC-MS, data not shown). These results associated the difference in β subunit levels between low-BSC and normal-BSC seeds with developmental stages S4-S6, providing clues for further transcriptomic study. In general, α/α′ subunits accumulated at S2 in low-BSC and S3 in normal-BSC seeds, with both of these stages preceding β subunit accumulation.The accumulation of acidic and basic subunits of 11S glycinin started between the time course of α/α′ and β subunit accumulation.The A3 subunit and β subunit showed similar accumulation patterns in general (Fig. 1-B). The accumulations of α/α′,acidic, basic and A3 subunits all showed similar patterns in low-BSC and normal-BSC immature seeds, suggesting that low BSC did not influence the accumulation time course of other major components of seed storage protein.

3.2.Protein content,oil content,and amino acid and fatty acid profiles of low-BSC and normal-BSC seeds

Given that the β subunit is devoid of sulfur-containing amino acids,the relationship between β subunit and the seed amino acid profile remained undetermined. In the present study,seventeen amino acids in both low-BSC and normal-BSC seeds were identified, with contents ranging from 1.33 to 75.05 g kg-1(Fig. 2-A). As expected, the lowest contents of these amino acids were those of Cys and Met. The content of sulfur-containing amino acids(Cys + Met)was 31.5%higher in low- than in normal-BSC seeds (Fig. 2-A, Student's t-test,P ＜0.01).Cys and Met were accumulated to respectively 12.0%and 48.3%higher levels in low-than in normal-BSC seeds,but only the Met difference was significant(Fig.2-A).Notably,the content of glutamate in low-BSC seeds(63.9 g kg-1)was 14.9%lower than that in normal-BSC seeds (75.05 g kg-1), although not significantly (Fig.2-A).

To quantify the influence of BSC alteration on total protein and oil content,we measured crude protein and oil content of low- and normal-BSC seeds. The crude protein contents of low-BSC and normal-BSC seeds were 43.91 and 43.98%respectively, and showed no significant difference (Fig. S1,n = 10, P ＞0.05). Likewise, low-BSC and normal-BSC seeds showed no significant difference in oil content. The mean oil contents were respectively 17.30% and 17.51% for low- and normal-BSC seeds (Fig. S1, n = 10, P ＞0.05). There were no significant difference in the(protein+oil)content as well(Fig.S1,n = 10,P ＞0.05).Of the 12 FAs measured,oleic acid(C18:1)and linoleic acid(C18:2)accounted for respectively 36.26%and 42.55%, or 78.81% of total FAs (Fig. 2-B). Small but nonsignificant differences in C18:1, C18:2, C20:0, and C22:0 between low- and normal-BSC seeds were observed (Fig. 2-B). Thus,BSC alteration did not change the overall crude protein content,oil content,or fatty acid composition.

Fig.2-Amino acid and fatty acid profiles in mature seeds with low BSC and normal BSC.(A)Content of 17 amino acids in lowand normal-BSC seeds.* indicate a significance level of 0.05 by Student's t-test,n = 2.BSC,β subunit content;Asp, aspartic acid;Thr,threonine; Ser,serine;Glu,glutamic acid;Gly,glycine;Ala,alanine;Cys,cysteine;Val,valine;Met,methionine; Ile,isoleucine; Leu,leucine;Tyr,tyrosine;Phe,phenylalanine;Lys,lysine;His,histidine;Arg,arginine;Pro,proline.(B)Relative contents of 12 fatty acids in low-BSC and normal-BSC seeds,n = 3.The fatty acids were myristic(C14:0),pentadecanoic(C15:0),palmitic(C16:0),margarc acid(C17:0),lignoceric(C17:1),stearic(C18:0),oleic(C18:1),1inoleic(C18:2),1inolenic(C18:3),arachidic(C20:0),behenic(C22:0),and lignoceric (C24:0)acids.

3.3. Comparative transcriptomic analysis in low-BSC and normal-BSC seeds

Although the difference in β subunit level between low- and normal-BSC seeds could be traced mainly to the S4-S6 stages and the sharp increase in Met content in low-BSC mature seeds, the mechanism underlying this phenomenon remained undetermined. From the RNA libraries, 20.28 GB of clean sequence data after removal of adapter sequences and short reads was obtained: 9.38 GB from the low-BSC and 10.9 GB from the normal-BSC library. The Q30 was ＞94.19%,indicating high sequencing quality. Of the clean reads, 83.9%were successfully mapped to the soybean reference genome(Wm82.a2.v1, https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Gmax) (Table 1).

Of 1835 DEGs identified with a log2(low BSC/normal BSC)expression ratio ≥1 and a false discovery rate (FDR) ＜0.01,1210 were down-regulated and 625 were up-regulated(Fig.S2).

3.3.1.Functional classification of DEGs

Of 1835 DEGs, 1238 were assigned GO numbers and classified into the three principal categories of biological process,cellular component, and molecular function. Of 20 main categories in the biological process, the most frequently represented were metabolic process, followed by cellular process, single-organism process, response to stimulus,biological regulation, and localization. In the cellular component category, DEGs were most frequent in cell part, cell organelle, membrane, and membrane part. In the molecular function category, catalytic activity and binding activity were most frequent(Fig.3-A).

In the KEGG pathway analysis,specific enrichment of DEGs was found for 19 pathways with a P-value threshold of 0.05(Fig.3-B,Table 2).Of these pathways,eight were enriched withP-value ＜0.01: glutathione metabolism, protein processing in endoplasmic reticulum, linoleic acid metabolism, cysteine and methionine metabolism, plant-pathogen interaction,alpha-linolenic acid metabolism, starch and sucrose metabolism,and galactose metabolism.These pathways are involved mainly in carbon and sulfur metabolism.

Table 1-Characterization of RNA-seq data.

Fig.3-GO classification and KEGG enrichment of DEGs in low-and normal-BSC soybean seeds.(A)GO classification of DEGs in low-and normal-BSC seeds.The left axis shows the percentage and the right axis the number of genes assigned to a given GO term.(B)KEGG pathway enrichment scatter diagram of DEGs in low-and normal-BSC seeds.The X axis shows the enrichment factor of a pathway,defined as the ratio of DEGs in that pathway to all annotated genes in the same pathway, with the-lg Pvalue shown on the Y axis.The significance levels 0.05 and 0.01 are shown with red and green dashed lines.

Table 2-The 20 most highly DEG-enriched KEGG pathways.

3.3.2. Expression patterns of genes involved in sulfur assimilation

The functional annotation of Wm82.a2.v1 revealed 17 genes involved in the cysteine and methionine synthesis pathways.For cysteine synthesis,the carbon/nitrogen backbone of cysteine is derived from O-acetylserine in a reaction catalyzed by serine acetyltransferase (SAT), and the thiol group is then transferred to O-acetylserine by O-acetylserine(thiol)lyase(also called cysteine synthase)[37].There were 12 serine acetyltransferase coding genes in soybean,one(Glyma.01G160000)of which was up-regulated 5.3 fold in low-BSC seeds in comparison with normal-BSC seeds, and the remaining genes were either undetectable or unchanged (Fig. 4, Table S2). Three of 18 Oacetylserine(thiol)lyase coding genes (Glyma.03G006700,

Glyma.20G148100, and Glyma.20G229000) were up-regulated,with the remaining 15 undetectable or unchanged. In methionine synthesis, the carbon/nitrogen skeleton is provided by phosphorylated homoserine,while the sulfur atom comes from cysteine[38,39].A novel gene(Soybean_newGene_1151)coding for methionine synthase and hypothesized to catalyze the committed step to methionine(also called homocysteine methyltransferase) was up-regulated, while the remaining 10 Met synthase homologs were undetectable or unchanged (Fig. 4,Table S2).Other genes coding for cystathionine β-lyase,cystathionine γ-synthase, and homoserine kinase, which catalyze the respective syntheses of homocysteine, cystathionine, and Ophosphohomoserine, were unchanged (Fig. 4, Table S2). The genes coding for enzymes responsible for catabolism of methionine, namely methionine γ-lyase, 1-aminocyclopropane-1-carboxylate synthase,and S-adenosylmethionine decarboxylase,were down-regulated in low-BSC seeds(Fig.4,Table S2).Two of 13 genes coding for aspartate aminotransferase and all three genes coding for asparagine synthetase were down-regulated in low-BSC seeds(Fig.4,Table S2),suggesting that the catabolism of aspartate and methionine was suppressed.

The synthesis of cysteine from glutathione involves two enzymatic steps, catalyzed respectively by glutamatecysteine ligase and glutathione synthetase. Glutamatecysteine ligase was up-regulated, whereas glutathione synthetase remained unchanged. The gene coding for hydroxyacylglutathione hydrolase, which catalyzes the formation of glutathione from (R)-S-lactoylglutathione, was upregulated, also contributing to the accumulation of glutathione. In contrast, genes coding for enzymes with glutathione as substrate, including glutathione peroxidase and glutathione S-transferase(GST),were all down-regulated(Fig.4,Table S2).

Interestingly, sulfate transporters and ATP sulfurylase,which are responsible for the transport of sulfur from other organs to seeds, are involved in the first essential step of sulfate reduction [40,41]. These two genes were both downregulated at the transcriptional level (Fig. 4, Table S2),indicating that sulfur transport was reduced in low-BSC seeds during seed filling.

ni semyzne gnidnopserroc rof gnidoc seneg dna selgnatcer ni detacidni era setilobateM.sdees CSB-wol ni skrowten msilobatem ruflus ni seneg fomargaid citamehcS-4.giF.ylevitcepser,kcalb dna,neerg,der ni detneserper era sdees CSB-lamron ot evitaler CSB-wol ni degnahcnu dna,detaluger-nwod,detaluger-pu setilobatem ro seneG .scilati.spets citamyzne elpitlum etacidni senil dehsaD.segnahc tnacifingis etacidni sworra drawpU

To validate the expression of these DEGs and characterize gene expression during seed filling,four genes involved in sulfur assimilation,namely serine acetyltransferase(Glyma.01G160000),methionine γ-lyase (Glyma.10G172700) and cysteine synthase(Glyma.03G006700/Glyma.20G229000), were chosen for qRT-PCR quantification during the S1-S7 stages. All four genes were mostly differentially expressed in low- and normal-BSC seeds during the S3-S7 stages (Fig. 5), a finding consistent with the accumulation time course of the β subunit.The genes coding for serine acetyltransferase and cysteine synthase were upregulated in low-BSC seeds (Fig. 5-A, B, D), whereas the methionine γ-lyase gene was down-regulated(Fig.5-C),a finding consistent with those from RNA-seq analysis and supporting the reliability of the DEGs identified by RNA sequencing.

4. Discussion

The β subunit accumulation pattern in which low-BSC seeds accumulated trace amounts of β subunit, with accumulation starting at the S4 stage as in normal-BSC seeds(Fig.1-B), and the agreement of the accumulation time courses of the other subunits (α/α′, acidic, basic and A3) with those reported previously [9], suggested that the β subunit is distinct from other major subunits and regulated independently of them.

BSC accumulation is influenced by various environmental factors[11,14,17,21,42].Because in the present study,low-and normal-BSC seeds were planted under identical conditions with the aim of avoiding differential environmental influence,the observed variation in BSC can be attributed mainly to genetic differences. Previously [6], low BSC involved an alteration of the relative contents of other subunits,including α, α′, acidic and basic subunits. In the present study, crude protein content was not affected by low BSC (Fig. S1). In several studies [43-48] a rebalanced proteome was observed after the silencing of conglycinin or glycinin-associated genes,with seed glycinin or conglycinin showing elevated accumulation compensating for the shortfall and thereby maintaining stable total protein and oil contents. There must be internal regulatory mechanisms underlying this phenomenon, possibly residing in the genetic background and regulated at the translational level [45,47]. The finding that sulfur-containing amino acids were more abundant in low-than in normal-BSC seeds (Fig. 2-A), agrees with the lack of both methionine and cysteine in the β subunit.

Transcriptomic analysis of low- and normal-BSC seeds revealed 1210 up-regulated and 625 down-regulated DEGs(Fig.S2). These DEGs were most frequently represented in metabolic process and enriched in 19 pathways involved mostly in sulfur and carbon metabolism (Fig. 3). Sulfur assimilation is deeply involved in the regulation of β subunit accumulation[13,14,49,50]. Our study showed an up-regulation of serine acetyltransferase (SAT), O-acetylserine(thiol)lyase (OAS-TL),and homocysteine methyltransferase, which are responsible for the synthesis of OAS, cysteine, and methionine, and down-regulation of the genes for catabolism of methionine and S-adenosyl methionine (SAM) (Fig. 4). These processes may have resulted in the accumulation of methionine.Exogenous application of L-methionine inhibited β subunit accumulation [14]. These findings suggest that β subunit content in low-BSC seeds is regulated via the methionine pathway, although the detailed mechanism remains unknown.

The up-regulation in low-BSC seeds of genes involved in glutathione anabolism and down-regulation of genes for glutathione catabolism would be expected to result in the accumulation of glutathione, though the glutathione level was not measured in this study. Application of glutathione inhibited the accumulation of β subunit [15]. The role of glutathione in the regulation of BSC awaits further study.

The elevated content of sulfur-containing amino acids(Cys + Met) in low-BSC seeds suggested that sulfur was sufficient in these seeds (Fig. 2). However, this inference is contradicted by the observation that four sulfate transporters were down-regulated in low-BSC seeds (Fig. 4, Table S2). This finding could possibly be explained by previous findings[51-53] that ATP sulfurylase was suppressed by the accumulation of sulfur-compounds, especially cysteine and glutathione. The down-regulation of sulfate transporters and ATP sulfurylase is probably the result rather than the cause of the increased sulfur assimilation in low-BSC seeds.

The sulfur-containing amino acid content in soybean is a major nutritional limiting factor whose increase has long been emphasized as a breeding goal [3]. Owing to the low content of sulfur-containing amino acids in 7S conglycinin and 11S glycinin, seed storage protein composition alteration has received little attention.The major avenue for enhancing the sulfur containing amino acids is believed to increase the sulfur sink, and most efforts have been aimed at modifying critical steps in sulfur uptake and assimilation. Several strategies have been adopted, including overexpression of enzymes involved in committed steps of sulfur uptake and assimilation, knockdown of expression of proteins poor in sulfur-containing amino acids, and ectopic expression of artificially modified proteins rich in sulfur-containing amino acids,but only modest success has been achieved[3].In some cases, the sulfur-containing amino acid content were elevated, but unexpected side effects on plant growth or yield restricted further application[54,55].In the present study,the alteration of BSC resulted in an increase in sulfur-containing amino acids. Most importantly, low-BSC lines maintained their crude protein and oil content and did not show stunted plant growth.This finding suggests that the alteration of seed storage protein composition is an efficient way to increase sulfur-containing amino acid content. Combination of this strategy with the genetic modification of sulfur uptake might advance breeding practice.

5. Conclusions

We investigated the accumulation pattern of β subunit in low-BSC seeds and found that the depressed level of β subunit resulted from the accumulation process. In low-BSC seeds,sulfur-containing amino acids,especially Met,were increased but crude protein content, oil content, and fatty acid composition remained unchanged.Transcriptomic analysis revealed that genes involved in the synthesis of cysteine, methionine,and glutathione were up-regulated and genes responsible for the catabolism of these compounds were down-regulated.

Fig. 5 - Methionine metabolism gene expression patterns in low-BSC seeds. Relative expression levels of genes were determined during seed developmental stages in both low-BSC (red) and normal-BSC (green) seeds using qRT-PCR. Normalization was performed against the genomic mean of an internal standard, Cons6/7/15 [34]. (A) Expression patternof Glyma.01G160000 (serine acetyltransferase). (B) Expression pattern of Glyma.03G006700 (cysteine synthase); (C) Expression pattern of Glyma.10G172700 (methionine γ-lyase);(D) Expression pattern of Glyma.20G229000 (cysteine synthase). Three biological replications were measured for each gene (C and D), and significance levels of 0.05, 0.01 byStudent's t-test are indicated by * and **.

These findings suggested that methionine and glutathione metabolism is associated with β subunit accumulation.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.01.001.

Conflict of interest

The authors have no conflict of interest.

Acknowledgments

This study was supported by the National Key Research and Development Program of China (2016YFD0100201-14), National Natural Science Foundation of China (31401401) and Youth Fund of Heilongjiang Academy of Agricultural Sciences(2017XQ04).