Cystathionine γ-synthase expression in seeds alters metabolic and DNA methylation profiles in Arabidopsis

Abstract

Arabidopsis (Arabidopsis thaliana) seeds expressing the feedback-insensitive form of cystathionine γ-synthase (AtD-CGS), the key gene of methionine (Met) synthesis, under the control of a seed-specific phaseolin promoter (SSE plants) show a significant increase in Met content. This elevation is accompanied by increased levels of other amino acids (AAs), sugars, total protein, and starch, which are important from a nutritional aspect. Here, we investigated the mechanism behind this phenomenon. Gas chromatography–mass spectrometry (GC-MS) analysis of SSE leaves, siliques, and seeds collected at 3 different developmental stages showed high levels of Met, AAs, and sugars compared to the control plants. A feeding experiment with isotope-labeled AAs showed an increased flux of AAs from nonseed tissues toward the developing seeds of SSE. Transcriptome analysis of leaves and seeds displayed changes in the status of methylation-related genes in SSE plants that were further validated by methylation-sensitive enzymes and colorimetric assay. These results suggest that SSE leaves have higher DNA methylation rates than control plants. This occurrence apparently led to accelerated senescence, together with enhanced monomer synthesis, which further resulted in increased transport of monomers from the leaves toward the seeds. The developing seeds of SSE plants, however, show reduced Met levels and methylation rates. The results provide insights into the role of Met in DNA methylation and gene expression and how Met affects the metabolic profile of the plant.

Introduction

Methionine (Met) is a sulfur-containing essential amino acid (AA) found at low levels in vegetative tissues and seeds, limiting the nutritional value of a plant-based diet for humans and animals (Amir 2010; Amir et al. 2012). Met plays a key role in protein synthesis and mRNA translation in plant cells, in addition to regulating cellular processes through its main catabolic product, S-adenosyl-Met (SAM). SAM is a donor of methyl groups that participate in the synthesis of many secondary metabolites, as well as DNA, RNA, and proteins such as histones, lipids, chlorophyll, alkaloids, ethylene, polyamines, vitamins, and cell-wall-related substances (Roje 2006; Sauter et al. 2013). Met synthesis is mainly controlled by the first unique enzyme in its biosynthetic pathway, cystathionine-γ-synthase (CGS) (Amir 2010). Overexpression of the Arabidopsis (Arabidopsis thaliana) CGS (AtCGS) led to substantially higher levels of Met in various species of transgenic plants (reviewed by Hesse et al. 2004; Amir 2010).

By using seed-specific promoters that express feedback-insensitive A. thaliana CGS (AtD-CGS) (Hacham et al. 2006) in tobacco (Nicotiana tabacum cv. Samsun NN), soybean (Glycine max L. cv. Zigongdongdou, and cv. Jilinxiaoli) and A. thaliana, we detected significantly higher levels of Met in transgenic seeds (Matityahu et al. 2013; Song et al. 2013; Cohen et al. 2014; Cohen, Shir, et al. 2016). In these studies, the Met level in the transgenic seeds was compared to control wild-type (WT) seeds or plants having an empty vector (EV). All these transgenic seeds with a higher Met content showed an unexpected increase in the levels of primary metabolites, including other AAs and sugars. Consequently, they also had high contents of total protein and starch (Matityahu et al. 2013; Song et al. 2013; Cohen et al. 2014; Cohen, Shir, et al. 2016), which are important from a nutritional aspect.

To better understand the reasons behind this phenomenon, Arabidopsis seeds expressing AtD-CGS under the control of the phaseolin promoter (called SSE plants; Cohen et al. 2014) were further analyzed. Along with having high levels of soluble and total Met (which incorporate proteins), SSE seeds showed increased accumulation of total soluble AAs (up to 3.6-fold), with a 40% and 10% increase in overall total protein-bound AAs and total proteins, respectively (Cohen et al. 2014). In addition to AAs, these SSE seeds had higher contents of sugar acids (up to 5.5-fold), polyamines (2.2-fold), polyols (2.5-fold), and different sugars (up to 6-fold). The high level of sugars most probably supported the accumulation of starch, whose level increased by 15% in SSE seeds compared to the control (Cohen et al. 2014).

These findings encouraged us to reveal the source of these metabolites, i.e. whether they are mainly de novo synthesized during seed development or mainly transported from vegetative tissues toward the seeds, since both of these processes take place during seed development (Baud et al. 2008). As reported for Arabidopsis, their seed formation is an intricate process that requires substantial spatiotemporal metabolic and transcriptomic rearrangements (Fait et al. 2006; Baud et al. 2008; Santos-Mendoza et al. 2008). Three fundamental stages were suggested during seed development: (i) embryo growth, cell division, and determination of morphogenesis. At this stage, nutrients (mainly sugars) are transported from the canopy leading to starch accumulation (Baud et al. 2008); (ii) seed maturation, where massive accumulations of storage reserves occur. Starch was mainly converted at this stage into AAs and fatty acids, utilized for synthesizing seed-storage proteins and oil (Fait et al. 2006; Amir et al. 2018); and (iii) seed desiccation and entrance to dormancy. At this stage, the whole metabolic activity gradually decreases, leading to a quiescent state; however, the synthesis of the seed-storage proteins continues while the oil contents tend to decrease (Baud et al. 2008). Some specific sugars and AAs increase at this stage to protect the membranes and proteins, contributing to the acquisition of tolerance desiccation (Hoekstra et al. 2001). It also suggests that this accumulation is required during imbibition to enable this process prior to storage reserve degradation and the beginning of germination (Angelovici et al. 2010).

Apart from the de novo synthesis of AAs within the seed, studies highlighted the important contribution of AAs synthesized in vegetative tissues. These AAs, as well as other soluble metabolites, are transported toward the developing seeds that serve as a strong sink. The massive fluxes of AAs are mainly associated with the leaves' senescence when the leaves degrade their proteins and more AAs are released (Watanabe et al. 2013).

In the current study, we aim to learn more about the reasons behind the elevations in AAs, sugars, and other metabolites in transgenic SSE seeds. We assume that more insights into the processes leading to this metabolic phenotype will establish a framework for further improving the nutritional value of seeds, especially regarding total proteins and starch.

Results

Arabidopsis developmental stage definition and sampling

To distinguish between the 2 possibilities that AAs synthesized de novo during the early stages of seed development or were imported from nonseed tissues, we collected rosette leaves (RL), siliques, and seeds from SSE and control plants having an EV at 3 developmental stages. The seeds started to harvest 12 d after flowering (DAF) when the phaseolin promoter became active in Arabidopsis seeds (Fait et al. 2011). The developmental stages were as follows: Stage I—plants at 42 d after germination (DAG) having developing seeds of 12 and 16 DAF; Stage II—plants at 49 DAG having seeds of 12, 16, and 21 DAF; and Stage III—plants at 56 DAG having seeds of 21, 26, and 30 DAF. The 30 DAF seeds were brown and dry and were therefore not analyzed since the metabolic and transcriptomic analyses of dry seeds have already been described (Cohen et al. 2014). The Stage III plants represent the latter stage of plant development when the irrigation was stopped and the leaves began to senescence. At this stage, most of the metabolites are transferred from the nonseed tissues toward the seeds. No substantial differences were detected in the morphological phenotype between SSE and EV plants during the 3 developmental stages (Supplemental Fig. S1).

SSE plants show a higher accumulation of Met and total soluble metabolites in RL, siliques, and seeds

A tissue-specific gas chromatography–mass spectrometry (GC-MS) analysis was performed to determine the levels of soluble Met and AAs in SSE and EV plants. Compared to EV, SSE seeds at 12 and 16 DAF from Stage I showed higher levels of soluble Met by 1.9- and 2.3-fold, respectively. Unexpectedly, the Met level was also higher in siliques of 12 DAF by 1.9-fold but was not altered in RL (Fig. 1A). The levels of total soluble AAs also increased in SSE seeds by 2- and 2.6-fold, respectively, compared to EV, as previously reported in dry SSE seeds (Cohen et al. 2014). Higher levels of total AAs, 2.5- and 3.6-fold, respectively, were also detected in siliques (Fig. 1B). A similar trend was also observed at Stages II and III, except that at these stages, the Met level also increased significantly in the RL of SSE (Fig. 1). Overall, our results indicate that high Met levels in the 3 organs correspond with high levels of total AAs.

Figure 1.

The levels of soluble Met A, C, E) and the total soluble AAs B, D, F) in RL, siliques, and developing seeds (DAF). The results taken during 3 developmental stages of transgenic Arabidopsis plants seed-specific expressing the AtD-CGS (SSE) and plants having the EV. The analysis was done by using GC-MS, normalized to norleucine internal standard. Data shown are means ± Sd of 4 replicates. Significance was calculated using the 2-way ANOVA test of P ≤ 0.05 and identified by different letters. Significant differences (P ≤ 0.05) using t-test are marked by asterisks.

Stage III was the most interesting since, at this stage, the RL of SSE showed increased levels of total soluble AAs, which correlate to metabolite abundance in seeds. Thus, this stage was selected for further analysis. To investigate how the higher Met content affected the accumulation of other metabolites, the primary metabolites profile was measured by GC-MS. The analysis identified 42 metabolites in SSE and EV (Supplemental Tables S1 to S3). A principal component analysis (PCA) showed that leaves, siliques from 21 and 26 DAF, and their seeds were clustered separately between SSE and EV (Fig. 2), suggesting remarkable metabolic differences. In general, the higher levels of Met and AAs in the 3 tested organs were accompanied by differing levels of many other primary metabolites (including sugars, organic acids, fatty acids, polyols, and others). To determine if this metabolic phenotype is unique to Stage III, the analysis was performed on Stages I and II to reveal that most of the metabolites also have higher levels in SSE organs than in EV (Supplemental Tables S1 to S3 and Figs. S2 to S4).

Figure 2.

Principal component analysis (PCA). The PCA was based on 42 annotated metabolites detected by GC-MS from plants seed-specific expressing the AtD-CGS (SSE) and those having the EV. The data are presented 4 biological replicates (except siliques that have 3 repetitions). Variance explained by each component is indicated in brackets.

The increased Met levels in RL of SSE were unexpected since phaseolin is a seed-specific promoter (Li et al. 1998; Fait et al. 2011). It might be that AtD-CGS expression has increased (through yet unknown regulatory mechanisms) in RL. To explore this assumption, the expression level of heterologous AtD-CGS was measured using reverse transcription quantitative PCR (RT-qPCR) in RL and in the developing seeds at Stage III. The AtD-CGS expression level increased during seed development as previously reported (Fait et al. 2011), but it was also found to increase in RL in SSE (by 8-fold) compared to EV (Fig. 3).

Figure 3.

The expression level of AtD-CGS in RL and seeds. The data taken from RL and seeds of 21 and 26 DAF of plants seed-specific expressing the AtD-CGS (SSE) and those having the EV at Stage III. Samples were normalized to the transcript level of phosphatase 2A subunit A3 (AtPP2A-A3). The data are presented as the mean ± Sd of 4 biological replicates. Significance was calculated using the 2-way ANOVA test of P ≤ 0.05 and identified by different letters. Significant differences (P ≤ 0.05) using t-test are marked by asterisks.

AAs flux from RL to the developing seeds increased in SSE plants

The higher level of metabolites in SSE seeds (Supplemental Tables S1 to S3) could be the result of either de novo synthesis or higher efflux from nonseed tissues. Developing seeds serve as a strong sink, mainly at the senescence stage when polymers undergo degradation to form monomers in vegetative tissues. To determine if SSE plants have a higher efflux from nonseed tissues, we performed an in planta feeding analysis using [¹⁵N] Asp or [¹⁵N] Glu. These 2 AAs belong to the most transported long-distance AAs in plants. The labeled AAs were supplemented to RL or siliques of EV and SSE plants at Stage III. After 10 h, the siliques of 21/26 DAF were harvested, and the accumulation of labeled Asp/Glu was measured by GC-MS in their seeds. The results from the silique feeding showed an increase in ¹⁵N/¹⁴N ratio in SSE seeds of 21 DAF when the siliques were fed with [¹⁵N]Asp (2.7-fold) and 1.6-fold when [¹⁵N]Glu was applied (Fig. 4A). At 26 DAF, the values were 6.4- and 2.8-fold when [¹⁵N] Asp or [¹⁵N]Glu was applied, respectively (Fig. 4A). However, when applying [¹⁵N]Asp and [¹⁵N]Glu to RL, the ¹⁵N/¹⁴N ratio did not change significantly in 21 DAF seeds but was 1.8- and 2.6-fold higher for [¹⁵N]Asp and [¹⁵N]Glu, respectively, in 26 DAF SSE seeds compared to EV (Fig. 4B). Together, these results suggest that SSE seeds at 26 DAF showed increased sink activity compared to 21 DAF seeds. They also strongly propose that RL and siliques of SSE serve as important sources of nutrients for the developing seeds of SSE.

Figure 4.

Feeding analysis done by using GC-MS. [¹⁵N] Asp or [¹⁵N] Glu used to feed siliques A) or to RL B) of plants seed-specific expressing the AtD-CGS (SSE) and those having the EV. Ten hours after the feeding, the seeds were collected and analyzed by GC-MS, and the ratio of ¹⁵N/¹⁴N was calculated. The data are presented as the mean ± Sd of 4 biological replicates. Significance was calculated using the 2-way ANOVA test of P ≤ 0.05 and identified by different letters. Significant differences (P ≤ 0.05) using t-test are marked by asterisks.

Transcriptome analysis reveals reprogramming of leaves and seeds in SSE at the late developmental stage

The results described above showed a strong link between Met content in SSE to other soluble metabolites. To further investigate this link, transcriptome profiling using 3′-end TRANSeq RNA sequencing (RNA-seq) (Tzfadia et al. 2018) was performed on RL and seeds (21 and 26 DAF) from Stage III. Since the quality and quantity of RNA extracted from siliques were low, we omitted these tissues from the analysis. Due to the metabolic phenotype of SSE, we sought genes involved in the metabolism of Met, AAs, and sugars, as well as in the transport, synthesis of soluble monomers, and catabolism of polymers.

Approximately 87% of the sequencing reads were mapped to the Arabidopsis genome, and ∼94% of the uniquely mapped reads were found in exons. Comparing SSE to EV, we identified 663, 734, and 649 differentially expressed genes (DEGs) in RL, 21 DAF seeds and 26 DAF seeds, respectively (|log₂fold change| > 1.5, P-value < 0.05). Gene ontology (GO) enrichment analysis was carried out to suggest the biological pathways that were altered in each of the organs examined using PANTHER and KOBAS GO enrichment tools (Supplemental Table S4). An analysis of DEGs from RL and 26 DAF seeds using PANTHER showed enrichment of pathways associated with primary metabolism, nitrogen-containing compounds, and macromolecule metabolic processes in RL, whereas 26 DAF seeds showed enrichment related to macromolecule methylation. However, no enrichment was detected for DEGs in 21 DAF seeds. Further, the KOBAS tool provided indications of primary and secondary processes enrichment in RL and 21 DAF seeds. Since the GO enrichment analysis pointed to an effect on primary metabolism but did not bring conclusive insights into the biological processes that were altered in SSE plants, so we decided to analyze the transcriptome data of each tissue manually and grouped DEGs into different clusters according to the Uniport website and available literature (Supplemental Tables S5 to S8).

Influence of AtD-CGS expression on genes related to Met biosynthesis and methylation in 21 DAF SSE seeds

To study the effect of high Met in 21 DAF seeds, we sought DEGs related to the Met metabolism. Six genes related to Cys synthesis (the thiol group donor for Met synthesis) and sulfur assimilation pathway were upregulated (Supplemental Table S6). These include a sulfate transporter, adenylyl-sulfate kinase, sulfate deficiency-induced gene, sdi1 (induced by sulfur starvation; Aarabi et al. 2021), and sulfotransferase, which utilizes 3′-phospho-5′-adenylyl sulfate (PAPS) as a sulfonate donor to other metabolites (Hell and Wirtz 2011). The results suggest that an increased demand exists for Cys in SSE seeds. Inversely, genes encoding thioredoxin, glutathione transferase, and the sulfur–iron cluster were significantly downregulated (Supplemental Table S6 and Figs. S5 and S6). These metabolites most probably compete with Met for Cys as a substrate. Met is also transported in the form of S-methylMet (SMM) from RL toward the seeds (Kocsis et al. 2003). However, the level of homocysteine S-methyltransferase 3 (HMT3) that forms Met from SMM inside the seeds (Kocsis et al. 2003; Cohen et al. 2017) was downregulated in SSE. HMT also controls the SAM levels (Kocsis et al. 2003). While the Met synthesis increased, the results suggest that SSE is reprogramming to reduce the level of Met by upregulation of genes associated with Met catabolism or usage. These include methyl transferases forming sterol, thiamine, and glucosinolates (Supplemental Table S6). These results are in accordance with the higher levels of several Met catabolic products, such as polyamines and glucosinolates, in dry SSE seeds (Cohen et al. 2014). In addition, ligase genes that combine the tRNA of Met and 5-formyltetrahydrofolate cycloligase, the precursor for formyl-Met-tRNA synthesis, were downregulated, suggesting the limitation of incorporating Met into proteins. Together, the results indicate that, on the one hand, most of the Cys is used for Met synthesis induced by AtD-CGS activity. On the other hand, opposing pathways operate in the seeds in order to decrease Met levels.

To further determine if the higher AAs and other soluble metabolites that accumulate in the dry SSE seeds (Cohen et al. 2014) are formed by de novo synthesis in seeds or are transferred mostly from the vegetative tissues, genes related to AA transport and metabolism were identified. The results show that only 4 genes related to AAs metabolism were upregulated in SSE seeds (Supplemental Table S6), suggesting that most of the AAs were derived from nonseed tissues. The low expression of AAs permease and other key genes in AAs biosynthesis suggests that the seeds try to reduce the synthesis and import of AAs. In addition, the levels of 10 transporters that import soluble metabolites were significantly reduced (Supplemental Table S6).

To our surprise, 1 of the largest clusters of genes that was changed in SSE seeds was of genes related to methylation processes. Met, through SAM, is a donor for methyl groups required for methylation processes (Roje 2006). Eleven genes related to methylation were upregulated in SSE compared to EV (Table 1 and Supplemental Fig. S6). Among them are gene-encoding proteins that affect the structure and function of DNA and chromatin, such as histone H3K4 methylation, RNA-directed DNA methylation (RdDM), an antisilencing factor that prevents gene repression, DNA methylation, siRNA-mediated gene silencing, and response to DNA hypermethylation. In addition, genes related to RNA methylation and protein methyl transferase, as well as genes associated with primary and secondary metabolites having methyl groups, were also upregulated. Genes that were downregulated included those related to the synthesis of secondary metabolites, DNA methylation, structure, and function (Supplemental Table S6). Together, these results indicate that a higher Met level in young developing SSE seeds strongly affects Met/SAM metabolism and methylation processes.

Table 1.

Genes related to the methylation of DNA in the seeds of 21 and 26 DAF of SSE vs. EV

Tissue	Gene ID	logFC (SSE-EVL)	P-value	Gene description (TAIR)	Predicted effect on DNA/histone methylation
Seeds_21	AT3G48670	1.06	0.049	Forms a complex with FDM1/IDNL1 and FDM2/IDNL2 that is required for RNA-directed DNA methylation (RdDM)	Up
	AT3G22590	1.95	0.031	Histone H3K4 methylation encodes PLANT HOMOLOGOUS TO PARAFIBROMIN (PHP)	Up
	AT5G53920	1.81	0.007	Protein methyltransferase. One target is PRPL11 which it methylates on Lys 109. PROTEIN METHYLTRANSFERASE A (PRMA)	Up
	AT4G02405	1.591	0.039	S-Adenosyl-L-methionine-dependent methyltransferases superfamily protein	Up
	AT5G40530	1.345	0.038	Methyltransferase required to silence Rdna (DNA sequence that codes for ribosomal RNA)	Up
	AT1G55250	1.319	0.025	Involved in monoubiquitination of histone H2B, and it is also a prerequisite for H3K4me and maybe H3K79me	Up
	AT5G13830	1.76	0.014	FtsJ-like methyltransferase family protein	Up
	AT3G12550	1.74	0.044	Belongs to a subgroup of SGS3-like proteins that act redundantly in RNA-directed DNA methylation	Up
	AT4G08590	1.35	0.046	ORTHRUS-like protein. Chromatin organization. DNA methylation on cytosine within a CG sequence. Maintenance of DNA methylation	Up
	AT1G66080	1.35	0.049	HLP1 is required in maintaining histone H3K acetylation and H3K4 methylation	Up
	AT2G27040	1.18	0.044	Chromatin silencing. AGO4 is a member of a class of PAZ/PIWIת-educed site-specific CpNpG and CpHpH methylation	Up
	AT5G58130	0.98	0.049	Protein REPRESSOR OF SILENCING 3. RNA-binding protein required for DNA demethylation	Down
	AT2G23740	0.91	0.049	Encodes a SET-domain protein SUVR5 that mediates H3K9me2 deposition and silencing	Down
	AT2G17970	2.21	0.001	RNA demethylase ALKBH9B, 2-oxoglutarate (2OG), and Fe(II)-dependent oxygenase superfamily protein	Down
	AT1G09060	−1.247	0.048	JMJ24 is appears to regulate basal levels of transcription of silenced loci by controlling methylation in heterochromatic regions.	Down
	AT3G54560	−1.757	0.006	Encodes HTA11, a histone H2A protein. Loss of all H2A.Z results in a reduction in DNA methylation of transposons	Down
	AT1G20870	1.42	0.043	Encodes an antisilencing factor that prevents gene repression and DNA hypermethylation. Increased DNA methylation 3	Down
	AT5G11470	1.36	0.031	Protein ANTI-SILENCING 1. SG1 i involved in CHG methylation within genebodies.	Down
	AT1G08130	1.18	0.024	Component of the active DNA demethylation machinery and is indispensable for cell viability	Down
	AT1G63020	−1.16	0.025	DNA-directed RNA polymerase IV subunit 1. Protein RNA-DIRECTED DNA METHYLATION DEFECTIVE 3.	Down
Seeds_26	AT2G22475	1.52	0.010	Histone H3K9 methylation. Encodes GL2-expression modulator (GEM).	Up
	AT5G63080	1.40	0.042	Histone H4R3 methylation	Up
	AT2G17900	1.19	0.018	Histone-lysine N-methyltransferase ASHR1. Homology Subgroup S-ET—Protein containing an interrupted SET domain.	Up
	AT1G20870	−1.41	0.038	Increased DNA methylation 3 acts as an antisilencing factor that prevents DNA hypermethylation and gene repression	Up
	AT1G80420	−1.83	0.008	DNA demethylation. Encodes a component of plant break excision repair and functions at several stages during active DNA demethylation	Up
	AT3G14890	1.01	0.041	DNA demethylation encodes a base excision repair protein	Down
	AT2G23740	−1.01	0.049	Encodes a SET-domain protein SUVR5 that mediates H3K9me2 deposition.	Down
	AT3G52030	−1.15	0.011	Histone H3K4 methylation. F-box family protein with WD40/YVTN repeat domain	Down
	AT5G49020	−1.23	0.008	Probable histone-arginine methyltransferase 1.4 (AtPRMT14) encodes a type I protein arginine methyltransferase.	Down
	AT3G21060	−1.24	0.039	Encodes a structural core component of a COMPASS-like H3K4 histone methylation complex	Down
	AT3G18990	−1.33	0.009	B3 domain-containing transcription factor VRN. Required for the methylation of histone H3	Down
	AT2G26680	−1.49	0.037	RNA-directed DNA methylation 4. FkbM family methyltransferase	Down
	AT2G30280	−1.51	0.012	Encodes RDM4, a transcriptional regulator functioning in RdDM and plant development. Chromatin silencing	Down
	AT1G15910	−1.72	0.023	Factor of DNA methylation 1 belongs to a subgroup of SGS3-like proteins that act redundantly in RNA-directed DNA methylation	Down
	AT2G40030	−1.14	0.014	DNA-directed RNA polymerase V subunit 1. Required for normal RdDM at non-CG methylation sites and transgene silencing.	Down
	AT5G10400	−1.44	0.044	Histone H3.2 (Histone H3.1)	Down
	AT2G19670	−2.00	0.003	Protein arginine methyltransferase 1A. chromatin organization	Down
	AT1G78190	−1.81	0.007	Multifunctional methyltransferase subunit TRM112 homolog B	Down
	AT4G28830	−1.09	0.045	S-Adenosyl-L-methionine-dependent methyltransferases superfamily protein	Down
	AT4G27910	−1.74	0.036	Histone-lysine N-methyltransferase ATX4 (EC 2.1.1.-). Encodes a SET domain containing protein, putative H3K4 methyltransferase	Down
	AT3G04380	−2.11	0.003	Encodes SUVR4, a nucleolar histone methyltransferase with preference for monomethylated H3K9.	Down

Influence of At-DCGS expression on genes in 26 DAF SSE seeds

The 26 DAF seeds undergo a desiccation process, and at this stage, the phaseolin promoter has its highest activity expressing the AtD-CGS transgene (Fig. 3) (Fait et al. 2011). As a response, the following genes related to Met synthesis were upregulated in SSE: (i) homoserine kinase, the enzyme that produces O-phosphohomoserine, the carboamino substrate for CGS; (ii) gamma-glutamyl hydrolase that forms the polyglutamyl tail on folate (the substrate of 5-methyltetrahydrofolate and the donor of methyl group to Met synthesis); (iii) methylenetetrahydrofolate dehydrogenase; and (iv) a folate transporter (Supplemental Table S7 and Fig. S5). To balance Met levels, the HMT3 gene, as well as the genes encoding tetrahydrofolate and another folate transporter, was downregulated. This again suggests that SSE seeds try to reduce Met/SAM content.

Regarding genes involved in the AA metabolism, tRNA synthetase was upregulated in SSE, while genes encoding branched-chain AA synthesis involving in aromatic AAs, AAs permease, and the 2-thiocytidine tRNA biosynthesis protein were downregulated (Shigi 2018). The reduction in permease and genes related to AAs degradation supports the assumption that the seeds have enough AAs. Additionally, most of the transporters related to other soluble metabolites were downregulated, suggesting that the seeds try to reduce their accumulation. The higher levels of soluble AAs apparently led to changes in the protein profile since the expression levels of 2 genes encoding albumin proteins were increased. This finding is in accordance with a previous report on dry seeds of SSE that had higher protein levels of several subunits of 12S-globulins and 2S-albumins, the main Arabidopsis seed-storage proteins, regardless of their Met contents in SSE (Cohen, Pajak, et al. 2016).

As detected in 21 DAF, the expression level of genes related to global methylation status was differently regulated in 26 DAF seeds. These include genes encoding sterols and pectin biosynthesis, methylation of metabolites, protein, and tRNA methylation. The number of genes involved in DNA and histone methylation whose expression levels were downregulated was about 3-fold higher than those that upregulated (Table 1). The list includes genes leading to lower methylation of H3K4me3, H3K9me2, and H3K9, which play a role during development, genes preventing hypermethylation, and those involved in lower methyltransferases, including genes in the RdDM process. These results suggest that 26 DAF SSE seeds try to avoid hypermethylation and additional changes in the chromatin structure (Tables 1 and S7).

High level of Met alters gene expression and metabolic profiles in SSE leaves

The Trans-seq analysis detected 664 genes that were significantly upregulated or downregulated in RL of SSE compared to EV, 152 of which were uncharacterized (Supplemental Table S8). The higher level of Met in RL at Stages II and III (Fig. 1) suggested that Met is synthesized in this organ. So, we first sought genes related to the Met metabolism. At Stage III, the expression level of a chloroplast-localized sulfate transporter was upregulated (3.1-fold), indicating that there is greater demand for sulfate. Also, genes related to glutathione S-transferase (GST) were downregulated, indicating that in RL, as in SSE seeds, most of the Cys is directed toward Met synthesis and less toward glutathione synthesis and its function. The other upregulated genes were related to Met degradation (Supplemental Table S8 and Figs. S5 and S6), suggesting that the RL try to counteract the high Met content like the seeds.

One possible explanation for the increased levels of AAs and metabolites in SSE leaves could be that senescence processes increased in SSE leaves, which led to a higher rate of polymer and protein degradation. The results indicate that senescence-associated genes were upregulated, including those involved in protein degradation, which could lead to high levels of soluble AAs (Supplemental Table S8). The high levels of soluble metabolites could also result from an increased synthesis rate that occurs in the RL of SSE in parallel to the catabolic processes. Indeed the results showed that genes encoding enzymes in the synthesis of AAs, in sugar metabolism, and in other primary metabolites were upregulated in the RL of SSE compared to EV (Supplemental Table S8). Overall, these results indicate metabolic reprogramming in SSE leaves. While some genes lead to polymer degradation, other genes encode enzymes in the biosynthesis of monomers and polymers. The high expression level of transporters (Supplemental Table S8) suggests that these monomers are transported from the leaves to the seeds.

An analysis of the RL transcriptome showed that the following 5 genes related to SAM were upregulated: a mitochondrial transporter of SAM; the 1-aminocyclopropane-1-carboxylate synthase 4/1-aminocyclopropane-1-carboxylate synthase (ACS), the key enzyme for ethylene synthesis; spermine synthase that uses SAM as a precursor; and methylesterases required for riboflavin and lignin syntheses. In addition, 3 genes encoding SAM-dependent methyltransferase were upregulated. Two other genes of SAM-dependent methyltransferase were downregulated (Supplemental Table S8 and Fig. S5). These suggest that RL try to balance the levels of SAM, the main catabolic product of Met.

In addition to methyltransferases, other genes related to methylation were upregulated in RL of SSE (Supplemental Table S8), including those related to histone methylation, RdDM, and pectin methylesterase, and in the metabolism of secondary metabolites. Also, genes related to protein methylation, methionyl-tRNA synthetase, tRNA methyltransferase, and RNA methylation were upregulated. However, some genes of histone methylation, DNA demethylation HSP20-like chaperone, and those involved in stable transcriptional silencing were downregulated (Tables 2 and S8).

Table 2.

Genes related to the methylation of DNA in RL of SSE vs. EV

Tissue	Gene ID	logFC (SSE-EVL)	P-value	Gene Description (TAIR)	Predicted effect on DNA/histone methylation
Leaf	AT4G16570	4.84	0.001	Arginine methyltransferase 7 mediates the symmetric dimethylation of histone H4 “Arg-3” to form H4R3me2s	Up
	AT3G42670	4.79	0.002	Gene silencing by RNA-directed DNA methylation encodes a nuclear localized SNF domain containing protein involved in RNA silencing	Up
	AT4G30860	3.72	0.015	Histone-lysine N-methyltransferase contains a SET domain which is known to be involved in modification of histone by methylation	Up
	AT2G44150	1.85	0.009	Histone-lysine N-methyltransferase ASHH3	Up
	AT3G49660	1.84	0.010	A structural core component of a COMPASS-like H3K4 histone methylation complex.	Up
	AT1G55970	2.09	0.010	Histone acetyltransferase HAC4. Acetyltransferase enzyme. Acetylates histones, giving a specific tag for transcriptional activation.	Up
	AT2G27840	1.69	0.040	Histone deacetylase HDT4.	Up
	AT2G25880	1.66	0.020	Histone phosphorylation	Up
	AT3G06930	1.46	0.024	Probable histone-arginine methyltransferase 1.3 encodes a type I protein arginine methyltransferase.	Up
	AT4G28830	1.71	0.021	S-Adenosyl-L-methionine-dependent methyltransferases superfamily protein, methyltransferase activity, nucleic acid binding	Up
	AT4G18470	3.24	0.02	Negative regulation of histone H3K4 methylation.	Up
	AT4G16570	4.84	0.001	Arginine methyltransferase 7 mediates the methylating histone H4 to form H4R3me2s.	Up
	AT3G20010	1.48	0.037	Helicase-like transcription factor CHR27 encodes a member of the SNF2 family involved in RNA-directed DNA methylation	Up
	AT1G66080	1.37	0.042	HLP1 is required in maintaining histone H3K acetylation and H3K4 methylation marks at the promoters of heat shock protein genes in providing thermotolerance/thermomemory response	Up
	AT1G76440	−0.71	0.040	Positive regulation of DNA demethylation HSP20-like chaperones superfamily protein	Up
	AT4G34060	4.41	0.002	Encodes a protein with 5-meC, preference for CpG and CpHpG sequences. Involved in maintaining methylation marks	Down
	AT4G18470	3.24	0.020	Negative regulation of histone H3K4 methylation	Down
	AT1G21920	−1.01	0.025	Histone H3K4-specific methyltransferase SET7/9 family protein	Down
	AT2G24740	−1.13	0.019	Histone-lysine N-methyltransferase, H3 lysine-9 specific SUVH8. Encodes a SU(VAR)3-9 homolog, a SET domain protein	Down
	AT1G02580	−2.94	0.010	Encodes the imprinted gene MEA that belongs to Polycomb Repressive Complex 2 (PRC2) and has a SET domain for methyltransferase activity and is involved in the stable transcriptional silencing of target genes.	Down

The higher levels of genes related to histone methylation are associated with genes involved in DNA integrity and function. Genes related directly to DNA integrity changed significantly, suggesting DNA remodeling. Changes were also observed in the expression of transcription factors (23 and 12 were upregulated or downregulated, respectively), most of which are yet uncharacterized for their function. Changes were also detected in genes of RNA editing, RNA polymerization, and RNA binding and in ribosome assembly and structure (Supplemental Table S8). The proportion between upregulated and downregulated transcripts suggests that protein translation increased in RL. Overall, the results indicate that a high level of Met in RL leads to the induction of genes related to methylation, chromatin structure, and transcription, and they also provide indications of translation induction in SSE leaves (Supplemental Table S8).

Global and specific changes in DNA methylation observed in SSE leaves and seeds

Our transcriptome data suggest that a relatively large number of genes are associated with DNA, RNA and histone methylation in SSE leaves and seeds (Tables 1 and 2). Taking into consideration the role of Met in methylation through SAM, these results led us to the assumption that the DNA methylation rate changed in SSE. DNA methylation can influence gene expression, transposon mobility, genome integrity, and chromatin structure (Chwialkowska et al. 2017). For an initial indication of the occurrence of changes in DNA methylation, we used a methylation-sensitive amplification polymorphism (MSAP) analysis, which is one of the most commonly used methods for assessing changes in DNA methylation in plants (Chwialkowska et al. 2017). The method involves cleaving PCR fragments from selectively amplified DNA using methylation-sensitive restriction enzymes McrBC, HpaII, and MspI. We selected the transposable elements AtSN1 and AtLINE, as well as 5SrDNA ribosomal DNA, as specific targets for testing DNA methylation levels. McrBC is used as a strong marker to determine a high or low methylation state. The results obtained from this assay suggest that an increase occurs in the DNA methylation pattern in RL, while a decrease in DNA methylation was detected in the 26 DAF seeds (Fig. 5A). This is in accordance with the finding that 26 DAF seeds show a high number of methylation-related genes that were downregulated, while most of the genes related to methylation in RL were upregulated (Table 1).

Figure 5.

Determination of global DNA methylation level at RL and at developing seeds. The analysis determined plants seed-specific expressing the AtD-CGS (SSE) and those having the EV. A) MSAP using 3 methylation-sensitive restriction enzymes. The approach is made on seeds 26 DAF and on leaves. The data represent 5 replicates. All the samples in each line were run on the same gel; B) percent of the methyl cysteine in seeds of 21/26 DAF and RL of EV and SSE by using a colorimetric assay “MethylFlash Global DNA Methylation.” The data are presented as the mean ± Sd of 4 biological replicates. Asterisks indicate statistical significance between EV and SSE by Tukey–Kramer honestly significant difference (HSD) (P ≤ 0.05).

To strengthen the finding that 26 DAF seeds are in a state of low methylation while RL have higher methylation, the global DNA methylation rate was measured in leaves and seeds using the colorimetric assay MethylFlash Global DNA Methylation (5-mC) ELISA Easy Kit (EpiGentek). The results (Fig. 5B) show that SSE leaves have a higher, albeit insignificant, level of DNA methylation than EV. The methylation rate in SSE seeds of 21 DAF was similar for SSE and EV but was lower in the 26 DAF seeds. The results are in accordance with the results obtained from the transcriptome analysis (Supplemental Fig. S5 and Tables 1 and 2) and MSAP for specific targets. In line with previous studies (An et al. 2017; Kawakatsu et al. 2017), global DNA methylation tends to increase from 21 DAF seeds to 26 DAF seeds in both SSE and EV (Fig. 5B). Overall, these results suggest that Met induction could globally affect DNA methylation levels.

Discussion

The regulation of Met content in plants is controlled by complex regulatory networks that involve cellular, developmental, and stress response processes (Hesse et al. 2004; Amir 2010; Zierer et al. 2016). SAM, the main catabolic product of Met that controls the expression level of AtCGS, is involved in many regulatory processes, including numerous transmethylation reactions such as DNA and histone methylation (Meng et al. 2018). The current study was designed to investigate the mechanism that led to the metabolic phenotype of SSE. The key findings of this study are as follows: (i) the RL, siliques, and developing seeds of SSE have higher levels of Met, as well as other AAs, sugars, and soluble metabolites; (ii) the flux from nonseed tissues toward the seeds, as detected for 2 AAs, is substantially higher in SSE; and (iii) DNA methylation levels are affected in SSE. Twenty-six DAF seeds show low methylation status, whereas in RL, they tend to have high methylation.

Association between high levels of Met and increased levels of other AAs, sugars, and soluble metabolites in leaves and seeds

Previous studies have shown that the seeds of A. thaliana, 2 transgenic soybean accessions, N. tabacum, and Azuki bean (Vigna angularis) that express AtCGS under seed-specific promoters (Hanafy et al. 2013; Matityahu et al. 2013; Song et al. 2013; Cohen et al. 2014; Cohen, Shir, et al. 2016) not only have higher Met contents but also have higher levels of other AAs, sugars, and soluble metabolites. The elevation of these monomers leads to significantly higher levels of total protein and starch, which are very important from a nutritional aspect. Such dramatic metabolic phenotypes were not detected when the key genes in the biosynthesis of Lys, Thr, or Trp were induced by seed-specific promoters in Arabidopsis, N. tabacum, and soybean. The levels of other AAs and metabolites showed only minor changes in these seeds (Karchi et al. 1994; Angelovici et al. 2009; Kita et al. 2010). Based on the results obtained here, we would like to suggest that Met is unique in causing significant metabolic and transcriptomic changes that may be caused by changes in DNA methylation and chromatin structure, as discussed below. The observation that a high Met level affects other monomers was also detected in N. tabacum plants overexpressing the AtD-CGS cassette (Hacham et al. 2017). The higher expression level of AtD-CGS in seeds also affects the levels of other metabolites in Stage III (Supplemental Fig. S4). The contents of most of the primary metabolites increased in the RL of SSE, apparently due to the higher rate of polymer degradation because of senescence initiation. In the 21 DAF seeds, most of the metabolites were higher in SSE than in EV; however, their levels significantly decreased in the 26 DAF seeds. This reduction could be attributed to the massive synthesis of protein at this stage (Ruuska et al. 2002; Frank et al. 2015). Yet, several AAs and sugars were synthesized at this stage in both EV and SSE, as previously reported (Fait et al. 2006; Baud et al. 2008; Santos-Mendoza et al. 2008), although this was emphasized more in SSE.

The transcriptome analysis of the mature seeds (26 DAF) showed that many genes related to the synthesis and metabolism of different metabolites were mostly upregulated. We hypothesize that at this late stage of seed development, these RNAs are not translated to protein. Hence, it might be that these genes were translated during germination and helped the SSE seeds to germinate better under different stress conditions (Cohen et al. 2014). The transcriptome analysis of RL in the current study showed that many genes related to senescence and degradation of polymers were upregulated in SSE, enhancing the hydrolysis of polymers such as proteins, lipids, and starch, which cause an accumulation of monomers (Woo et al. 2013). Among those genes are the 5-meC and thymine-DNA glycosylase that initiate DNA demethylation (Ponferrada-Marín et al. 2009), release transcriptional silencing of a hypermethylated transgene, and also control many senescence-associated genes including a senescence regulator (Yuan et al. 2020). While a higher degradation rate occurred, the expression levels of 11 and 29 genes involved in the synthesis of AAs and in monomer/polymers of carbohydrate, respectively, increased in SSE leaves. This is an unusual phenomenon, indicating the complex SSE metabolism that leads to higher monomers in RL.

In planta feeding analysis reveals increased metabolic flux from nonseed tissues toward SSE seeds

The higher contents of Met and other primary metabolites in RL and siliques of SSE (Fig. 1) suggested that they could be a source of metabolites for the seeds. However, a high level of Met in RL would not necessarily push the metabolites into the seeds. When a feedback-insensitive Asp kinase, or AtD-CGS, was overexpressed in N. tabacum and rice (Oryza sativa), the leaves showed a higher level of Met, but the Met level in the seeds did not change (Hacham et al. 2008; Whitcomb et al. 2018). However, this is not the case in other transgenic plants. Overexpressing the AtD-CGS in soybean led to a significantly higher Met level in the seeds (up to 4.8-fold) (Yu et al. 2018). In addition, overexpressing genes for sulfur assimilation and Cys biosynthesis pathways also led to higher soluble and protein-incorporated Met in the seeds of transgenic maize (Zea mays) (Xiang et al. 2018), rice (Nguyen et al. 2012), and soybean (Kim et al. 2012).

The [¹⁵N]Asp/Glu feeding analysis (Fig. 4) indicated that the flux from the leaves and siliques toward the seeds of SSE is significantly high compared to EV. These findings are in accordance with the high expression levels of 20 transporters in RL. Among these transporters are genes that are involved in AAs transport, including the AAs transporter of AVT1J and nodulin MtN21-like transporter family protein, genes in sugars transport such as the fructose transporter, and genes in lipids and in ion/mineral-like ABC transporters. Based on the feeding analysis, metabolic profiling and the transcriptomic data of developing seeds and RL, we suggest that the metabolic flux toward the seeds increases following Met induction in SSE seeds. The low number of genes that were highly expressed in SSE seeds related to the synthesis of AAs and other metabolites (Supplemental Tables S6 and S7) also strengthens the assumption that most metabolites are not produced de novo inside the seeds but are derived from the nonseed tissues. It was previously suggested that at least 2 bottlenecks exist in the distribution of different metabolites between leaf and seed: the transporter-mediated loading of these metabolites into the phloem and the import into the developing seeds (Tegeder and Masclaux-Daubresse 2018). The results of the current study show that the expression levels of transporters were mainly upregulated in leaves and seeds of SSE, suggesting that these plants overcame these 2 bottlenecks, thus enabling better phloem and seed loading of metabolites that derived from the leaves.

High Met levels alter the global methylation status in SSE plants

The changes detected in the metabolic and transcriptomic profiles might result from changes in the methylation rate in SSE organs due to an increase in Met levels. This assumption is based on the fact that Met through SAM is the precursor of the methyl groups (Roje 2006) and that changes are detected in the expression level of genes in the current study. The changes include genes related to methylation processes, such as DNA and histone methylation, that were upregulated in RL at Stage III. In addition, 1 of the largest clusters of genes with upregulated expression levels was related to DNA integrity and function. We assume that the higher Met content detected at Stage III in RL provided more substrate to the various methyl transferases, which caused the altered methylation phenotype to exhibit a trend of higher methylation. MSAP and global DNA methylation analyses strongly strengthened this assumption. The higher expression level of genes related to the methylation of histones can affect chromatin structure. Such changes might also be related to the unexpectedly higher expression level of the heterologous AtD-CGS in RL (Fig. 2). A previous study showed that the phaseolin promoter could be expressed in Arabidopsis leaves under special circumstances (Sundaram et al. 2013). This promoter controls the synthesis of most abundant seed-storage proteins in the common bean Phaseolus vulgaris and is strictly turned off in vegetative tissues due to a nucleosome positioned over its TATA regions (Li et al. 1998). Sundaram et al. (2013) showed that the expression of the P. vulgaris ABI3-LIKE FACTOR, in combination with the plant regulator abscisic acid (ABA) in Arabidopsis leaves, could lead to the activity of this promoter. These researchers also suggest that this activity is related to changes in histone methylation. However, by carefully examining the RNA-seq analysis in the current study, we did not find significant differences in the expression of ABI3 in SSE. Also, although some genes related to ABA synthesis and its responses were upregulated or downregulated in SSE leaves, we did not notice a clear trend suggesting that ABA content increased in SSE leaves (Supplemental Table S8). Therefore, further study is needed to reveal why this promoter begins to be active in SSE leaves.

The relationships between the levels of Met, SAM, and DNA methylation were previously studied when SAM levels were reduced in plants. In these plants, DNA methylation and the methylation of histone H3K4me3 were significantly reduced, linking Met metabolism with DNA and histone methylation (Li et al. 2011; Huang et al. 2016; Meng et al. 2018). In addition, mutants that are defective in SAM accumulation, such as fpgs1 (Zhou et al. 2013), hog1 (Rocha et al. 2005), and mthfd1 (Groth et al. 2016), show a reduction in DNA methylation and a release of chromatin silencing. However, whether higher Met/SAM levels would increase the methylation rate in plant cells has not yet been studied. An indication of this possibility is a report that overexpression of Met synthase (METS1) in Arabidopsis leads to a 20% increase in cytosine methylation. This manipulation led to impaired plant immunity and enhanced disease susceptibility (González and Vera 2019). In addition, high levels of Met in nutrition in animals affect DNA methylation and gene expression (Zhang et al. 2006).

While the RL tend to show a higher DNA methylation rate, the 21 DAF seeds did not present any change. This is also reflected by the similar number of genes that were downregulated or upregulated in these tissues (Table 1). However, in the 26 DAF seeds, most of the genes related to methylation processes were significantly reduced (by 3-fold compared to those that upregulated). Only 1 common gene was detected in the seeds of both 21 and 26 DAF, while all the other DEG related to the methylation process differed between these developmental stages. This common gene is the SET-domain protein SUVR5 (AT2G23740) that mediates H3K9me2, which functions in suppressing transposons and repetitive sequences, protecting plant genomes from genome instability during plant development and environmental stress. The results obtained from the MASP and global methylation analyses also showed a lower methylation rate in these seeds. Unlike SSE, EV seeds showed that the methylation rate tended to increase during seed maturation from 21 DAF to 26 DAF, as previously reported (An et al. 2017; Kawakatsu et al. 2017). This elevation is important since a high methylation level assists the seeds during the dormancy phase (An et al. 2017). The lower methylation level in 26 DAF SSE is unexpected since we assume that a higher level of Met would lead to a higher methylation rate. This phenotype can occur if these seeds have retarded development and thus lower methylation. However, we cannot detect genes that point to this possibility; moreover, the seed phenotype did not change compared to EV. The lower level of DNA methylation in SSE seeds may suggest that SSE activates genes that reduced DNA methylation to enable the seeds to function during late seed filing and germination. Overall, the results suggest that high expression levels of AtD-CGS enhance the level of substrates required for Met synthesis; however, the seeds try to reduce Met, SAM, and DNA methylation in its progenies.

The complex effect of Met induction in SSE plants—a general overview

This study shows that SSE organs have higher levels of Met and other soluble metabolites. We assume that the induced Met supports the activity of methyltransferases, which also affect DNA methylation and chromatin structure. This causes a change in gene expression, leading to the accumulation of monomers and soluble metabolites in these SSE organs. Such high levels of monomers could cause metabolic discomfort in the SSE plants. Therefore, both the RL and the seeds attempt to lower the levels of Met and SAM, as well as prevent a transition from the leaves to the seeds by lowering the expression levels of certain transporters. The upregulation of genes involved in the synthesis of ethylene, spermine, riboflavin, and SAM-dependent methyltransferase in RL, and of genes involved in the synthesis of glucosinolates, sterols, and thiamine in seeds, supports this observation. These results are in accordance with the higher levels of several Met catabolic metabolites, polyamines, and glucosinolates in dry SSE seeds (Cohen et al. 2014). They are also in accordance with N. tabacum plants having a higher Met level due to overexpression of AtCGS, which removes the excess of Met in the form of ethylene and dimethyl sulfide (2 catabolic volatile Met products) (Hacham et al. 2002).

The accumulation of soluble metabolites in leaves and siliques might trigger the higher flux toward the developing seeds (Fig. 4). Another option is that a high Met content in seeds increased the sink strength of the SSE seeds. The higher AAs and sugars in seeds support the higher synthesis of total protein and starch (Matityahu et al. 2013; Song et al. 2013; Cohen et al. 2014; Cohen, Shir, et al. 2016). These elevations are very important from a nutritional aspect. A better understanding of the mechanism behind these phenotypes, especially the link to DNA methylation, could be used in the future to improve the nutritional quality of seed crops and, therefore, should be studied further. Further studies are required to strengthen the link between high Met to chromatin structure that changes the expression level of the genes. This could include the whole-genome bisulfite sequencing of the leaves and seeds, as well as crossing the SSE with different mutants related to the epigenetic phenotype.

Materials and methods

Plant growth and sampling

The fifth generation of transgenic homozygous Arabidopsis (A. thaliana) ecotype Col-0 plants expressing a seed-specific feedback-insensitive form of CGS (AtD-CGS) (SSE) and the control EV were grown as described by Cohen et al. (2014). At the time of flowering, the flowers were tagged with twines of different colors to differentiate between the seeds at different developmental stages. Samples from ∼30 individual plants were collected and pooled to obtain independent biological replicates. RL and siliques were collected from each group, frozen in liquid nitrogen and lyophilized. The developing seeds and silique hulls were separated after lyophilization.

Soluble AA extraction and determination by GC-MS analysis

Ten milligrams of lyophilized samples were used for the extraction of free AAs. The tissues were homogenized with a Restch MM 301 homogenizer, as previously described (Cohen, Shir, et al. 2016). The crude extract was separated into polar and nonpolar phases with methanol/water/chloroform. The upper polar phase was vacuum dried and dissolved in 40 µl of 20 mg ml⁻¹ methoxyamine hydrochloride in pyridine for 2 h, followed by derivatization for 30 min in N-methyl-N-(trimethylsilyl)-trifluoroacetamide at 37 °C for 2 h with vigorous shaking (Cohen, Shir, et al. 2016). One microliter of sample was injected into a GC-MS system with a split ratio of 1:1, together with the AA standards. All analyses were carried out on a GC-MS system (Agilent 7890A) coupled with a mass selective detector (Agilent 5975c) and a Gerstel multipurpose sampler (MPS2; Cohen et al. 2014). Norleucine (2 mg ml⁻¹ in HPLC grade water) and ribitol (2 mg ml⁻¹ in HPLC grade water) were used as internal standards. The peak areas were calculated from the standard calibration curves and normalized to the internal controls (norleucine and ribitol) signal.

Exogenous feeding of ¹⁵N-labeled Asp and Glu into leaves and siliques and GC-MS analysis

For the feeding assay, ¹⁵N-labeled Asp or Glu was purchased from Cambridge Isotope Laboratories (https://shop.isotope.com). Tissue-specific labeling experiments were performed on RL and siliques. Application sites on the adaxial surface of leaves and siliques were gently abraded with fine silica powder to increase the uptake of labeled AAs. Ten hours after the exogenous application of 5 mM of ¹⁵N Asp and Glu, the labeled/nonlabeled RL and silique hulls were harvested, frozen immediately in liquid nitrogen, and stored at −80 °C until extraction for GC-MS analysis. The quantification of ¹⁵N/¹⁴N enrichment in Asp and Glu was based on the determination of the areas of characteristic peaks: the peak of mass M corresponding to the unlabeled [¹⁴N] AA and the peaks of mass M + n corresponding to labeled AAs, where n determines the [¹⁵N] atoms present naturally or during the labeling experiment. In all the experiments, the silique hulls were separated, and the corresponding seeds were used for analysis.

RT-qPCR

For transcript determination, total RNA was extracted from 50 mg of seeds and leaves using the Spectrum Plant Total RNA kit (Sigma) according to the manufacturer's instructions. One microgram of RNA was used for cDNA biosynthesis using the Verso cDNA biosynthesis kit (Thermo Scientific) according to the manufacturer's protocol. The expression of mRNA levels of AtD-CGS in EV and SSE transgenic seeds and leaves was quantified using RT-qPCR. To normalize variance among samples, protein phosphatase 2A subunit A3 (AtPP2A-A3) transcript level was used as an endogenous control (Cohen et al. 2014). The list of primers is presented in Supplemental Table S9.

RNA sequencing and transcriptome analysis

Total RNA was isolated from the RL and seeds (21 and 26 DAF) of SSE and EV plants using the Spectrum Plant Total RNA kit (Sigma) according to the manufacturer's protocol. The quality and integrity of isolated RNA samples were checked by a standard NanoDrop spectrophotometer and gel electrophoresis. RNA from the silique tissues was omitted due to poor quality and integrity. Three biological replicates per tissue were used for RNA sequencing. The libraries for Illumina high-throughput sequencing were prepared according to Tzfadia et al. (2018) using the TranSeq 3′-end sequencing protocol. The raw reads were mapped to the recent genome of Arabidopsis (TAIR10; Arabidopsis Information Resource; www.arabidopsis.org) to generate normalized values for each gene. PCA for TranSeq data was performed using MetaboAnalyst 4.0 software (http://metaboanalyst.ca/).

Identification and enrichment analysis of DEGs

The edgeR V3.12.0 and limma V3.26.1 R packages were used for preprocessing and normalization of the RNA-seq read counts and analysis of DEGs. Library sizes were scaled using the calcNormFactors function based on TMM normalization (limma), and log-CPM with prior 0.25 was used for further analysis (according to limma User's Guide). Multidimensional scaling (MDS) using the limma R package (Ritchie et al. 2015) enabled the separation between SSE and WT transcriptomes obtained from the leaves or seed tissues. A differential expression analysis was conducted with the limma package in R (for genes having at least 1 sample with ≥10 read) using linear models and empirical Bayes moderated t-statistics. Genes with P ≤ 0.05 and log₂fold change ≥ 0.585 (i.e. 1.5-fold change) were considered to be differentially expressed. GO enrichment analysis was performed using PANTHER (http://pantherdb.org/, (Mi et al. 2021) and KOBAS (http://kobas.cbi.pku.edu.cn/genelist/; Bu et al. 2021) tools.

Statistical analysis

PCA and a heat map of GC-MS data were conducted using the MetaboAnalyst 5.0 comprehensive tool (http://metaboanalyst.ca/; Xia et al. 2015) with auto-scaling manipulations. Statistical significance was evaluated using JMP software version 8.0 (SAS Institute Inc., Cary, NC, USA). Significant differences between samples were calculated according to the Turkey–Kramer HSD test (P < 0.05) or Student's t-test (P < 0.05).

Accession numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries (see the data in the tables and in the Supplemental Tables S4 to S8. The accession numbers of cystathionine γ-synthase is AT3G1120.

Acknowledgments

We want to thank Prof. Asaph Aharoni, Weizmann Institute of Science, whose lab performed the libraries for the Trans-seq analysis.

Author contributions

A.G., Y.H., M.L., and R.A.: experimental design. A.G., Y.H., S.D., S.P., and M.L.: conducted the experiments. A.G., Y.H., M.L., and R.A.: data analysis. R.A.: manuscript preparation. All the authors read, revised, and approved the manuscript.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Table S1. The level of primary metabolites in the leaves, seeds, and siliques of EV and SSE at developmental Stage I, as detected by GC-MS.

Supplemental Table S2. The level of primary metabolites in the leaves, seeds, and siliques of EV and SSE at developmental Stage II, as detected by GC-MS.

Supplemental Table S3. The level of primary metabolites in the leaves, seeds, and siliques of EV and SSE at developmental Stage III, as detected by GC-MS.

Supplemental Table S4. GO enrichment analysis of the DEGs in each tissue using the PANTHER and KOBAS tools.

Supplemental Table S5. Upregulated genes in 21 and 26 DAF seeds and leaves of SSE.

Supplemental Table S6. Differentially upregulated and downregulated transcripts in 21 DAF seeds of the SSE line compared to EV.

Supplemental Table S7. Differentially upregulated and downregulated transcripts in 26 DAF seeds of the SSE line compared to EV.

Supplemental Table S8. Differentially upregulated and downregulated transcripts in the SSE RL compared to EV.

Supplemental Table S9. Primers for RT-qPCR.

Supplemental Figure S1. The phenotypes of SSE and EV.

Supplemental Figure S2. Heat map analysis of primary metabolites from the leaves, siliques, and developing seeds of EV and SSE plants at Stage I.

Supplemental Figure S3. Heat map analysis of primary metabolites from the leaves, siliques, and developing seeds of EV and SSE plants at Stage II.

Supplemental Figure S4. Heat map analysis of primary metabolites from the leaves, siliques, and developing seeds of EV and SSE plants at Stage III.

Supplemental Figure S5. The changes detected in gene expression of 4 functional groups (metabolism of Met, sulfur, AAs, and methylation processes) between SSE and EV.

Supplemental Figure S6. Schematic diagram of the metabolic network of the Met/Cys metabolism, as well as the Asp family pathway and the sulfur assimilation pathway.

Funding

This work was supported by a grant from the Israel Science Foundation (ISF grant no. 1857/20).

Data availability

The data that support the findings of this study are available in the Supplemental data.

Author notes

Present address: Institute of Biological, Environmental and Rural Sciences (IBERS), Plas Gogerddan, Aberystwyth University, Aberystwyth, Ceredigion SY23 3EE, UK.

Present address: Department of Cell and Metabolic Biology, Leibniz Institute of Plant Biochemistry, Weinberg 3, Halle 06120, Germany.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/General-Instructions) is Rachel Amir.

Conflict of interest statement. None declared.