A Biostimulant Obtained from the Seaweed Ascophyllum nodosum Protects Arabidopsis thaliana from Severe Oxidative Stress

Abiotic stresses cause oxidative damage in plants. Here, we demonstrate that foliar application of an extract from the seaweed Ascophyllum nodosum, SuperFifty (SF), largely prevents paraquat (PQ)-induced oxidative stress in Arabidopsis thaliana. While PQ-stressed plants develop necrotic lesions, plants pre-treated with SF (i.e., primed plants) were unaffected by PQ. Transcriptome analysis revealed induction of reactive oxygen species (ROS) marker genes, genes involved in ROS-induced programmed cell death, and autophagy-related genes after PQ treatment. These changes did not occur in PQ-stressed plants primed with SF. In contrast, upregulation of several carbohydrate metabolism genes, growth, and hormone signaling as well as antioxidant-related genes were specific to SF-primed plants. Metabolomic analyses revealed accumulation of the stress-protective metabolite maltose and the tricarboxylic acid cycle intermediates fumarate and malate in SF-primed plants. Lipidome analysis indicated that those lipids associated with oxidative stress-induced cell death and chloroplast degradation, such as triacylglycerols (TAGs), declined upon SF priming. Our study demonstrated that SF confers tolerance to PQ-induced oxidative stress in A. thaliana, an effect achieved by modulating a range of processes at the transcriptomic, metabolic, and lipid levels.


SF Protects Arabidopsis from Oxidative Stress
Oxidative stress was induced by spraying twenty-three-day-old A. thaliana Col-0 plants with 15 µM PQ ( Figure 1). Only PQ-treated plants developed large visible necrotic lesions within one day after treatment ( Figure 1A). The presence of cell death was demonstrated by trypan blue staining ( Figure 1B). To investigate the protective role of SF, Arabidopsis plants were pre-treated or primed with 0.1% SF six times on two consecutive days prior to the day of the PQ treatment. The SF-primed plants subjected to the subsequent PQ treatment did not develop any necrotic lesion on the leaves ( Figure  1A), and trypan blue staining did not detect dead cells ( Figure 1B), demonstrating that SF fully protects the plants from PQ-induced oxidative cell death. Cell death was further quantified by monitoring the electric conductivity of solutions in which rosette leaves from treatment groups and untreated controls were immersed ( Figure 1C; Supplementary Materials Data S1). Severe ion leakage was observed in PQ-treated plants, likely due to the ROS-induced oxidative damage [36]. However, when SF-primed plants were challenged with PQ, no increase in ion leakage was observed. Our results demonstrate that SF priming improved cell membrane integrity, reducing cell death induced by PQ.  Mean values are averages of three independent experiments with each point on the graph representing individual experimental replicates. In each experiment, ten rosettes were analyzed per treatment. Mean values were compared between groups by applying one-way ANOVA followed by Tukey's multiple comparisons test. An absence of letter sharing among the treatment groups denotes a statistically significant difference among those groups (p < 0.0001). Error bars denote standard deviation (SD). Full data are given in Supplementary Materials Data S1.

RNA-Seq Identifies Genes and Pathways Associated with SF-Induced Oxidative Stress Protection
By comparing gene expression in the different samples, we identified a total of 2192 significantly differentially expressed genes (DEGs) (Supplementary Materials Data S2; see Materials and Methods for details). To identify changes at the gene expression level, differential expression analysis was performed for a number of pairwise combinations (Table S1). In general, when PQ was applied alone, significant changes were observed in the expression of genes compared to the untreated control (H 2 O + H 2 O); the vast majority of the genes affected by PQ treatment (>1400) were upregulated (>1200). Pre-treatment with SF ("priming") prior to the PQ application resulted in a substantially lower number of DEGs (317) compared to untreated controls. Importantly, SF-primed PQ-stressed plants exhibited only a very small number of significant changes compared to plants treated with SF alone (six genes). Application of SF alone led to significant changes of only 163 genes compared to untreated controls. Our results suggest that SF primes plants in a manner which prevents PQ from inducing large-scale molecular or phenotypic changes (see Figure 1 for comparison). While SF induces only limited molecular or phenotypic changes when applied to unstressed plants, the biological effects of SF are striking when examined in the context of applying an oxidative stress-inducing agent such as PQ.

Cluster Analysis of DEGs
The DEGs from all pairwise comparisons were combined and clustered into 20 clusters using k-means clustering ( Figure S1). These 20 coordinated expression patterns revealed gene modulation and key transition states due to the treatments. Clusters 4, 7, 19, and 20 consisted of genes downregulated due to the SF priming in PQ-stress and non-stress conditions. Cluster 6 consisted of genes upregulated upon SF priming in PQ-stress and non-stress conditions. The remaining clusters exhibited upregulated genes (clusters 2, 3, 5, 8, 9, 10, 13, 16, 17, and 18) or downregulated genes (clusters 1, 11, and 15) due to the PQ induced oxidative stress. Interestingly, genes belonging to clusters that were up-or downregulated by PQ were not affected when plants were primed with SF prior to PQ application. Thus, treatment with SF prevented alteration of these genes that would otherwise occur due to the PQ application ( Figure S1). Expression levels and log 2 fold changes among treatments for the DEGs in each cluster are given in Supplementary Materials Data S2.

Gene Set Enrichment Analysis of DEGs
To gain insights into the molecular mechanisms involved in SF-mediated stress tolerance, we performed a Gene Ontology (GO) enrichment analysis of the DEGs (Supplementary Materials Data S3). In non-stress conditions, 128 genes were significantly downregulated upon SF priming versus the untreated control with the following GO terms being enriched: "trehalose metabolism in response to stress" (GO:0070413), "regulation of proteolysis" (GO:0030162), "response to chitin" (GO:0010200), "response to oxidative stress" (GO:0006979). Thirty-five DEGs were significantly upregulated in SF-primed versus unprimed control plants (Table S2), including several ERF/AP2 transcription factors such as ERF15, a positive regulator of ABA responses [37], ERF34, a transcriptional regulator of primary cell wall biosynthesis [38], BBX30, a B-box-type zinc finger protein involved in developmental regulation in long day-grown plants [39], and TINY2, a potential regulator of genes in the response to environmental stresses [40]. Also, CIPK20, which encodes a CBL-interacting protein kinase, was specifically upregulated in SF-primed plants. CIPK20, also known as PKS18, is involved in abscisic acid (ABA) signaling; its inhibition by RNA interference leads to ABA insensitivity [41]. Other genes specifically upregulated upon SF priming are: WAG1 encoding a serine/threonine-protein kinase involved in auxin signaling and the regulation of differential growth, tissue patterning, and organogenesis [42], and SPX1, a gene encoding a protein involved in responses to environmental cues and internal regulation of nutrition homeostasis [43]. A gene with unknown function (AT2G19650) coding for a putative cysteine/histidine-rich C1 domain family protein and several other unknown proteins encoded by genes AT3G44450, AT5G66740, AT2G15020, AT2G44940, AT5G11590, AT3G45210, AT1G04570, AT5G61570, and AT3G04140 were also specifically upregulated upon SF priming compared to PQ-treated plants and untreated controls (Supplementary Materials Data S2).
The top five enriched GO terms from each pairwise comparison, based on FDR (false discovery rate) values, were then selected and compared between the pairwise combinations. This analysis demonstrates that enrichment and transition of key biological processes and functions occurred during stress and upon SF priming ( Figure 2). The abiotic stress-related GO terms "response to oxidative stress", "response to wounding", and "response to toxic substance" were upregulated after PQ stress. In stark contrast, SF priming resulted in the downregulation of these stress-associated GO terms when subsequently exposed to PQ. Priming with SF specifically upregulated GO terms related to photosynthesis, hormones, and plant growth such as "response to red light" (GO:0010114), "protein-chromophore linkage" (GO:0018298), "photosynthesis" (GO:0015979), "light harvesting in photosystem I" (GO:0009768), "light harvesting in photosystem II" (GO:0009769), "response to auxin" (GO:0009733), and "regulation of growth" (GO:0040008) when exposed to PQ stress. It is also interesting that biotic stress and defense-related responses such as "response to chitin" (GO:0010200), "response to bacterium" (GO:0009617), "defense response to fungus" (GO:0050832), and "ethylene-activated signaling pathway'" (GO:0009873) terms were downregulated after SF priming followed by subsequent exposure to PQ. This supports our previous findings that a significant degree of specificity exists for extracts of A. nodosum in terms of stress reduction with efficacy clearly related to temperatures involved in the extraction process, i.e., high-temperature derived extract (SF) is effective in enhancing growth in abiotic stress condition compared to other lower-temperature derived extracts [29].
To obtain a complete view of significantly altered pathways at the transcriptional level during PQ treatment and upon SF priming, we performed a KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis [44] for DEGs from different pairwise comparisons ( Figure 3). The following pathways were significantly upregulated in PQ-induced stress conditions: "protein processing in endoplasmic reticulum" (ko04141), "plant-pathogen interaction" (ko04626), "phenylpropanoid biosynthesis" (ko00940), "phenylalanine, tyrosine, and tryptophan biosynthesis" (ko00400), "MAPK signaling pathway-plant" (ko04016), and "glutathione metabolism" (ko00480). Notably, these pathways were significantly downregulated in SF-primed PQ-stress condition. On the other hand, SF priming followed by PQ treatment specifically upregulated pathways for "plant hormone signal transduction" (ko04075) and "photosynthesis" (ko00195). activated signaling pathway'" (GO:0009873) terms were downregulated after SF priming followed by subsequent exposure to PQ. This supports our previous findings that a significant degree of specificity exists for extracts of A. nodosum in terms of stress reduction with efficacy clearly related to temperatures involved in the extraction process, i.e., high-temperature derived extract (SF) is effective in enhancing growth in abiotic stress condition compared to other lower-temperature derived extracts [29]. To obtain a complete view of significantly altered pathways at the transcriptional level during PQ treatment and upon SF priming, we performed a KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis [44] for DEGs from different pairwise comparisons ( Figure 3). The following pathways were significantly upregulated in PQ-induced stress conditions: "protein processing in endoplasmic reticulum" (ko04141), "plant-pathogen interaction" (ko04626), "phenylpropanoid biosynthesis" (ko00940), "phenylalanine, tyrosine, and tryptophan biosynthesis" (ko00400), "MAPK signaling pathway-plant" (ko04016), and "glutathione metabolism" (ko00480). Notably, these pathways were significantly downregulated in SF-primed PQ-stress condition. On the other hand, SF priming followed by PQ treatment specifically upregulated pathways for "plant hormone signal transduction" (ko04075) and "photosynthesis" (ko00195). Overall, the transcriptome analysis in this study demonstrated that PQ-induced oxidative stress causes extensive transcriptional reprogramming as evidenced by the upregulation of many stress marker genes, downregulation of many photosynthesis-and growth-related genes, and induction of genes involved in cell death. Upregulation of stress genes and downregulation of photosynthesis and growth genes are hallmarks of severe stress and cell death [45,46]. Such an extensive level of transcriptional reprogramming was absent in the SF-primed plants exposed to PQ, with no induction of stress genes and no downregulation of photosynthesis and growth genes. The overall pattern of differential gene expression between untreated, PQ stressed, SF priming, and SF priming PQ-stress treatments are represented by a heatmap (Figure 4). The heatmap demonstrates that at the transcriptome level, plants subjected to PQ stress after pre-treatment with SF resemble SF-primed and unstressed plants. Overall, the transcriptome analysis in this study demonstrated that PQ-induced oxidative stress causes extensive transcriptional reprogramming as evidenced by the upregulation of many stress marker genes, downregulation of many photosynthesis-and growth-related genes, and induction of genes involved in cell death. Upregulation of stress genes and downregulation of photosynthesis and growth genes are hallmarks of severe stress and cell death [45,46]. Such an extensive level of transcriptional reprogramming was absent in the SF-primed plants exposed to PQ, with no induction of stress genes and no downregulation of photosynthesis and growth genes. The overall pattern of differential gene expression between untreated, PQ stressed, SF priming, and SF priming PQ-stress treatments are represented by a heatmap (Figure 4). The heatmap demonstrates that at the transcriptome level, plants subjected to PQ stress after pre-treatment with SF resemble SF-primed and unstressed plants. Int  Full data are given in Supplementary Materials Data S1 and S2.

Annotation of Genes Induced by SF for Oxidative Stress Protection
Based on the enrichment analysis, DEGs were selected from the various treatments (Table 1) for an in-depth literature review of the known functions of these genes. This investigation revealed genes that could be directly or indirectly involved in SF priming-induced oxidative stress tolerance. It also identified potential adaptions possibly underlying the stress tolerance phenotype.

Annotation of Genes Induced by SF for Oxidative Stress Protection
Based on the enrichment analysis, DEGs were selected from the various treatments (Table 1) for an in-depth literature review of the known functions of these genes. This investigation revealed genes that could be directly or indirectly involved in SF priming-induced oxidative stress tolerance. It also identified potential adaptions possibly underlying the stress tolerance phenotype.  In a previous study, we identified five genes that are hallmarks for oxidative stress [47]; each of them was highly induced by PQ treatment (Table 1). Besides the ROS inducible/marker genes [47], Table 1 contains the genes most regulated by PQ or/and SF that represent pathways and processes such as ROS dependent PCD and senescence, ROS detoxification, photosynthesis, carbohydrate metabolism and cellulose synthesis, growth and hormone signaling, autophagy, lipid metabolism, transcription factors, and stress-related genes. In addition, 140 out of 177 genes, known to be ROS responsive [48], were induced by PQ treatment (Supplementary Materials Data S4). Our results therefore indicate that PQ treatment caused prominent oxidative stress similar to previous reports [47,49,50]. In contrast, none of these genes were upregulated in SF-primed PQ-treated plants (Supplementary Materials Data S4).
Reactive oxygen species can induce programmed cell death (PCD) and metacaspases play a key role in several types of ROS-induced PCD [51]. Two metacaspase-encoding genes, AtMC2 and AtMC8, were highly induced by PQ in the unprimed plants but not in the SF-primed plants. Moreover, PROMOTION OF CELL SURVIVAL 1 (PCS1), encoding an aspartic protease that regulates cell death and promotes cell survival, was induced in both SF-primed and primed/stressed plants (Table 1). PCS1 plays an important role in reproductive gametogenesis and its loss-of-function mutation causes excessive cell death of embryonic tissues [52]. The above results demonstrate that SF prevents the activation of the ROS-triggered PCD pathway. In addition, we observed an upregulation of SENESCENCE-SUPPRESSED PROTEIN PHOSPHATASE (SSPP) and a downregulation of the transcription factor RD26 as well as of SENESCENCE-ASSOCIATED GENE 13 (SAG13) in SF-primed and primed/stressed plants (Table 1). SSPP functions in sustaining leaf longevity and its overexpression significantly delays leaf senescence in Arabidopsis [53]. The transcription factor RD26 (ANAC072) is a positive regulator of stress-and darkness-induced senescence [54,55]. As we recently demonstrated, an enhanced expression of RD26 triggers the accumulation of free amino acids (in particular Gln and Asn), enhances the accumulation of tricarboxylic acid (TCA) cycle intermediates and represses the accumulation of γ-aminobutyric acid (GABA; [54]). SAG13 is induced by stresses such as darkness, drought, wounding, and pathogen challenge in leaves [56] and is involved in the breakdown of cellular components during stress-induced senescence. Expression of SAG13 is directly controlled by RD26 [54].
The induced expression of both, RD26 and SAG13 by PQ and repression of these genes by SF priming, therefore, is likely to contribute to the absence of cell death in SF-treated plants ( Table 1).

Detoxification of PQ-induced ROS
PQ produces superoxide radicals, followed by a cascade of reactions producing hydrogen peroxide and hydroxyl radicals [57,58]. Accumulation of these toxic free radicals will result in a decline of photosynthesis, an increase of lipid peroxidation and the destruction of cell membranes, leading to chlorosis and necrosis of plant tissues [58,59]. With respect to enzymatic antioxidants, SF-primed plants had higher levels of Fe SUPEROXIDE DISMUTASE 1 (FSD1) and ASCORBATE PEROXIDASE 1 and 4 (APX1 and APX4) compared to plants stressed with only PQ (Table 1). Increased expression of these genes can assist the detoxification of superoxides and peroxides [60][61][62]. In addition, higher levels of AtOXR, which encodes transmembrane OXR protein, was observed in SF-primed plants ( Table 1). Overexpression of AtOXR in Arabidopsis results in ascorbate accumulation, concomitant with an increased tolerance to oxidative stress [63]. Notably, expression of 1-myo-inositol-1-phosphate synthase encoding MIPS1 was elevated in SF-primed plants. MIPS catalyzes the rate-limiting step in the synthesis of myo-inositol which is critical for maintaining cellular levels of ascorbic acid [64]. In addition to transcriptome profiling, we also analyzed the primary metabolic profiles using GC-MS (see below). We found an increased level of threonate in PQ-stressed plants, an effect that was entirely mitigated when plants were primed with SF prior to the PQ treatment ( Figure 5). Threonate typically accumulates due to the degradation of ascorbate during stress, which appears to be suppressed by SF. In accordance with the strong priming effect of SF, the gene coding for L-ascorbate oxidase (AO) that converts ascorbate to dehydroascorbate (DHA) had higher expression in SF-primed plants ( Table 1). DHA is transported from the apoplast to the cytosol in exchange with ascorbate (reduced form), to ensure a constant apoplast redox flux [65][66][67][68]. The antioxidant ascorbate is one of the key components for reducing ROS. It also acts as a redox buffer and regulates the response to the extracellular environment in the apoplast [64]. The above observations suggest that priming by SF treatment induces a ROS detoxification pathway that involves the antioxidant ascorbic acid.
A gene coding for an iron-sulfur domain containing protein, NEET, was also upregulated in SF-primed plants (Table 1). This gene has a role in regulating redox reactions and the control of apoptosis. In Arabidopsis, T-DNA knockdown and RNA interference NEET lines accumulated higher levels of ROS than wild-type control plants [69]. NEET is involved in iron (Fe) homeostasis, supporting a proper ROS balance and photosynthesis in cells [70,71]. Notably, several cytochrome P450 encoding genes (CYP76C1, CYP71B35, and CYP706A) were also more strongly expressed in SF-primed plants ( Table 1). Cytochrome P450 enzymes have previously been shown to be involved in herbicide metabolism and detoxification [72,73]. A gene coding for an iron-sulfur domain containing protein, NEET, was also upregulated in SFprimed plants (Table 1). This gene has a role in regulating redox reactions and the control of apoptosis. In Arabidopsis, T-DNA knockdown and RNA interference NEET lines accumulated higher levels of ROS than wild-type control plants [69]. NEET is involved in iron (Fe) homeostasis, supporting a proper ROS balance and photosynthesis in cells [70,71]. Notably, several cytochrome P450 encoding genes (CYP76C1, CYP71B35, and CYP706A) were also more strongly expressed in SFprimed plants (Table 1). Cytochrome P450 enzymes have previously been shown to be involved in herbicide metabolism and detoxification [72,73].

Photosynthetic Genes
Most of the photosynthesis-related genes were significantly downregulated by PQ in unprimed plants. In fact, this is the largest functional group of genes that was repressed by PQ. The group

Photosynthetic Genes
Most of the photosynthesis-related genes were significantly downregulated by PQ in unprimed plants. In fact, this is the largest functional group of genes that was repressed by PQ. The group included genes encoding subunits of photosystem II, genes involved in chlorophyll biosynthesis, such as protochlorophyllide oxidoreductase, genes from the Calvin-Benson cycle such as RuBisCO, and several others (Table 1). SF mitigates against the detrimental effect of PQ and enables plants to photosynthesize even when threated with PQ. None of these genes were repressed in SF-primed plants subsequently treated with PQ. On the contrary, a number of these genes had even higher levels of expression in the SF-primed plants, suggesting that SF actually induces their expression in both unstressed and PQ-stressed plants (Table 1). Upregulation in this manner may enhance photosynthetic efficiency in the SF-primed plants and also protect against stress, consistent with the healthy phenotype observed for all SF-treated groups.
Gene AT2G23670, which encodes thylakoid protein YCF37, was upregulated in SF-primed plants compared to both PQ-stressed and untreated control plants. YCF37 is likely involved in photosystem 1 (PSI) assembly and oligomerization; its gene is induced by high-light stress and the protein has a likely role in repairing PSI during light stress [74]. A disrupted photosynthetic machinery and impaired photosynthesis result in the generation of various ROS, including superoxide radicals (O 2 •-) and singlet oxygen ( 1 O 2 ). Studies have also shown that to prevent high light-induced 1 O 2 accumulation and oxidative damage, it is essential to minimize the steady state concentration of chlorophyll biosynthetic intermediates in day light plants [75]. Here, we observed an increased level of PROTOCHLOROPHYLLIDE OXIDOREDUCTASE C (PORC) gene expression in SF-primed plants.

Carbohydrate Metabolism
Several genes related to carbohydrate metabolism were upregulated in the SF-primed plants ( Table 1). GALACTINOL SYNTHASE (GOLS) encodes a key enzyme in the synthesis of raffinose family oligosaccharides (RFOs). Raffinose protect plants against PQ-induced oxidative stress [76]. Hence, upregulation of this pathway in SF-primed plants likely contributes to the enhanced tolerance to PQ observed in this study.
Two other genes, GLYCOSYL HYDROLASE 9C2, also known as ENDO-1,4-BETA-GLUCANASE 6, and TRICHOME BIREFRINGENCE-LIKE 27 (TBL27), were upregulated in SF-primed plants and may have functions in cellulose synthesis, cell wall deposition and O-acetylation of cell wall polysaccharides [77,78]. Glycosyl hydrolases are involved in cellulose biosynthesis, cell wall deposition and integration of polysaccharides [77]. O-acetylation of xyloglucans is possibly accomplished by the enzyme encoded by TBL27 (also known as ALTERED XYLOGLUCAN 4, AXY4) [79]. Acetylation of cell wall polysaccharides is essential for maintaining the structural integrity of the leaf; it is also important for exerting a global impact on plant stress responses [80].

Growth and Hormone Signaling
Several growth-related genes are repressed by PQ, including genes encoding expansins (EXPA8, EXPA11, and EXPB1), auxin-responsive genes such as IAA6, PIN7, and SAUR-like genes. In contrast, these genes are not repressed in SF-PQ treated plants. On the contrary, many growth-and auxin-related genes are induced by SF in both PQ-stressed and unstressed SF-primed plants (Table 1). Gibberellin-regulated GASA6 is also repressed by PQ but induced in SF-primed plants (stressed or unstressed), while GIBBERELLIN 2-OXIDASE 6 (that inactivates GA) is induced only in PQ-treated unprimed plants. Taken together, these data show that while PQ represses growth, this effect of PQ is completely eliminated by SF. Furthermore, there is evidence that SF promotes growth (induction of growth-related genes as well as auxin and GA pathways), providing a molecular explanation of its action as a biostimulant/plant strengthener.

Autophagy
Oxidative stress can activate autophagy, an energy-dependent cellular process of recycling cellular components [81]. The WRKY33 transcription factor plays a role in regulating autophagy by interacting with autophagy-related (ATG) proteins [82,83]. Expression of WRKY33 is highly induced by PQ while its transcript levels are even lower in SF-treated plants than controls ( Table 1). Overexpression of ATG8 (different genes including ATG8E) in Arabidopsis leads to an increase in the number of autophagosomes concomitant with an activated autophagy [84]. Similarly, experimental evidence was obtained to indicate that elevated expression of ATG8H increases autophagy during abiotic stress conditions [85]. Expression of both, ATG8E and ATG8H is repressed by SF (Table 1). Collectively, our observations suggest activation of autophagy by PQ and repression of autophagy by SF.

Lipid Metabolism
Genes related to lipid degradation including PHOSPHOLIPASE A 2A (PLA2A) and MYZUS PERSICAE-INDUCED LIPASE1 (MPL1) [86] are strongly upregulated by PQ treatment, but not in PQ-treated plants previously primed with SF (Table 1). On the other hand, one gene encoding an oleosin-B3 like protein is downregulated by PQ treatment but upregulated in SF-primed plants.
Oleosins are relatively small hydrophobic proteins (with a molecular weight of 15-26 kDa) associated with oil bodies, i.e., cellular organelles for the storage of triacylglycerols in plants. A knockout mutant of a gene coding oleosin-B3-like protein is hypersensitive to salt stress [87]. Lipid reconfigurations occur during oxidative stress balancing energy metabolism and mitigating the oxidative damage [23,88].

Metabolome Reconfiguration Induced by SF for Oxidative Stress Protection
To gain further insights into the effect of SF on plants, we determined the metabolic profiles under various conditions using GC-MS. We observed substantial differences in primary metabolites of oxidatively stressed plants when compared with untreated plants (  (Foyer and Noctor, 2011 [89]). The increase of threonate may suggest an impairment of the ascorbate-glutathione cycle for ascorbate regeneration in PQ-treated plants ( Figure 5).
Interestingly, we observed an accumulation of various free amino acids in PQ-stressed plants, including branched chain amino acids (BCAAs), such as valine and isoleucine, as well as other amino acids such as alanine, glycine, cysteine, lysine, and asparagine. The amino acids most likely accumulate in the stressed plants as a result of protein degradation caused by PQ-induced disruption of regular metabolism rather than de novo amino acid biosynthesis, a conclusion supported by the fact that PQ causes cell death (Figure 1). Of note, an accumulation of amino acids during abiotic stresses has been observed in many different studies [21,[90][91][92], and it has been suggested that metabolites accumulating under stresses might be used as building blocks for new cellular components to support recovery of growth after stress adaptation [21]. Moreover, one of the glutamate-derived compounds linked to oxidative stress responses is γ-aminobutyric acid (GABA) which is involved in pH regulation, stress responses, signaling, and other processes in plants [21,91,93]. Here, we observed that GABA over accumulates in PQ-stressed plants (Figures 5 and 6). Metabolic and mechanical disruptions during stress cause cytosolic acidification and induce an acidic pH-dependent activation of GABA synthesis [93]. A stress-specific pattern of accumulation is consistent with a physiological role of GABA in stress mitigation [21,91,[93][94][95]. Interestingly, some of the metabolic effects observed in PQ-stressed plants are not only characteristic of oxidative stress, but also occur in senescing plant tissues [96]. An important result obtained here is that the metabolic changes induced by PQ are almost completely abolished when plants are pre-treated with SF and then subjected to PQ treatment. Several TCA cycle intermediates, in particular, fumarate and, to some extent, citrate and malate, increased in SF-primed plants, effects which might support cellular respiration and photosynthesis [97,98]. It has previously been suggested that an accumulation of citrate under oxidative stress conditions may control the production of ROS and, therefore, increase the plant's tolerance to stress [99,100]. Moreover, the levels of free amino acids were reduced following SF application which may indicate their utilization during the process of inducing stress tolerance and potentially growth ( Figure 6). In contrast, raffinose accumulated after SF treatment (Figures 5 and 6) which might increase oxidative stress tolerance as suggested previously [76]. The soluble sugar maltose accumulated in SF-primed and primed/stressed plants (Figures 5 and 6). Maltose at physiologically relevant concentrations may protect proteins, membranes, and the photosynthetic electron transport chain from stress-induced damage [101]. Metabolic and mechanical disruptions during stress cause cytosolic acidification and induce an acidic pH-dependent activation of GABA synthesis [93]. A stress-specific pattern of accumulation is consistent with a physiological role of GABA in stress mitigation [21,91,[93][94][95]. Interestingly, some of the metabolic effects observed in PQ-stressed plants are not only characteristic of oxidative stress, but also occur in senescing plant tissues [96]. An important result obtained here is that the metabolic changes induced by PQ are almost completely abolished when plants are pre-treated with SF and then subjected to PQ treatment. Several TCA cycle intermediates, in particular, fumarate and, to some extent, citrate and malate, increased in SF-primed plants, effects which might support cellular respiration and photosynthesis [97,98]. It has previously been suggested that an accumulation of citrate under oxidative stress conditions may control the production of ROS and, therefore, increase the plant's tolerance to stress [99,100]. Moreover, the levels of free amino acids were reduced following SF application which may indicate their utilization during the process of inducing stress tolerance and potentially growth ( Figure 6). In contrast, raffinose accumulated after SF treatment (Figures 5 and 6) which might increase oxidative stress tolerance as suggested previously [76]. The soluble sugar maltose accumulated in SF-primed and primed/stressed plants (Figures 5 and 6). Maltose at physiologically relevant concentrations may protect proteins, membranes, and the photosynthetic electron transport chain from stress-induced damage [101].

Lipidome Level Changes Induced by SF for Oxidative Stress Protection
Lipids are the main components of biological membranes and they represent important carbon and energy storage compounds in plants [102,103]. Lipids may sense extracellular signals and trigger lipid-mediated signaling and have also been shown to accumulate during several stresses, e.g., when photosynthesis is inhibited [102,103]. Here, we observed major changes in lipid profiles of plants exposed to oxidative stress (Figure 7). In total, 155 annotated lipids from ten neutral and polar lipid classes were identified. Of those, 108 lipids were significantly changed in at least one of the tested conditions in comparison with the untreated control condition (Figure 7). In our analysis, we detected different classes of lipids including neutral lipids of the diacylglyceride (DAG) and triacylglyceride (TAG) classes, as well as polar lipids including digalactosyldiacylglycerol (DGDG), monogalactosyldiacylglycerol (MGDG), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), and sulfoquinovosyldiacylglycerol (SQDG). It has been shown that under adverse conditions, neutral lipids play an important role in intracellular homeostasis and energy balance [104].
A more detailed analysis of the 108 significantly changed lipids revealed that most of the neutral TAG lipids and polar chloroplast lipids, such as, for example, galactolipids (i.e., MGDG and DGDG) which constitute the bulk of the chloroplast membrane, increased in stressed plants, likely resulting from cell death and chloroplast degradation, while they were primarily decreased in SF-treated plants ( Figure 7).
Notably, the TAG class of neutral lipids (representing the main source of energy) showed divergent patterns in different conditions: the levels of some lipids (e.g., TAG 60:5 and TAG 60:6), increased while levels of others (e.g., TAG 54:0 and TAG 54:2) decreased in PQ-stressed plants. A contrasting pattern was observed in SF-primed plants. A large proportion of the observed alterations in lipid levels in stressed plants was reversed in stressed plants pre-treated with SF, which demonstrates the protective role of SF against oxidative stress. Most of the neutral lipids, namely, TAGs and DAGs, decreased when stressed plants were compared with stressed plants pre-treated with SF ( Figure 7). This observation is in line with previous reports which showed that plants under adverse stress condition, such as, for example, extended darkness, usually accumulate those lipid classes [88], an effect mitigated by application of SF, as shown here.  cluster closely together along the first component (PC1). In contrast, the unprimed but PQ-stressed group (H 2 O + PQ) was found to deviate significantly from the other treatment groups in all the omics approaches used ( Figure 8A). This result indicates a higher similarity of H 2 O-and SF-treated plants at the molecular level and suggests that SF treatment induces stress adaptations by inhibiting damaging stress responses such as those induced by PQ. Furthermore, the electrolyte leakage data show a similar trend to the mean PC1 values, supporting the earlier observation ( Figure 8B). This was not observed for PC2 ( Figure S2 and Figure S3). Finally, the results of the PCA were strongly correlated to the visible plant phenotypes obtained (Figure 1): the two SF-treated groups are phenotypically similar to the negative control with which it clusters along PC1, whilst the oxidatively damaged plants are distant from that cluster. Thus, many of the molecular changes induced by SF and PQ manifest phenotypically. Mean PC1 values were compared among groups using one-way ANOVA followed by Tukey's multiple comparisons test to correct for multiple testing. For both, PC1 and electrolyte leakage, a significant difference was observed between the H2O + PQ group (oxidatively stressed) and each of the three remaining groups. For PC1, the adjusted p-values for all three pairwise comparisons were p ≤0.0004 (***), <0.0001 (****), and <0.0001 (****) for transcriptome, metabolome, and lipidome, respectively (underlined in the graph; see Figure S2 for comparison of PC1 values by oneway ANOVA).). Horizontal error bars denote the standard deviation (SD) of PC1 values. Vertical error bars indicate the SD of electrolyte leakage. Full data are given in Supplementary Materials Data S1 and S7.

Plant Material, Stress Treatments, and RNA Extraction
Arabidopsis thaliana (Columbia-0) seeds were sown in soil and the pots were kept for four days for vernalization (4 °C in darkness) for uniform seed germination. Thereafter, the pots were kept in a growth chamber in a 12 h light (150 µE m −2 s −1 , 20 °C, 60% relative humidity)/12 h dark (16 °C, 70% relative humidity) cycle. At least 30 uniformly developed 23 day-old plants were used for the Mean PC1 values were compared among groups using one-way ANOVA followed by Tukey's multiple comparisons test to correct for multiple testing. For both, PC1 and electrolyte leakage, a significant difference was observed between the H 2 O + PQ group (oxidatively stressed) and each of the three remaining groups. For PC1, the adjusted p-values for all three pairwise comparisons were p ≤0.0004 (***), <0.0001 (****), and <0.0001 (****) for transcriptome, metabolome, and lipidome, respectively (underlined in the graph; see Figure S2 for comparison of PC1 values by one-way ANOVA).). Horizontal error bars denote the standard deviation (SD) of PC1 values. Vertical error bars indicate the SD of electrolyte leakage. Full data are given in Supplementary Materials Data S1 and S7.

Plant Material, Stress Treatments, and RNA Extraction
Arabidopsis thaliana (Columbia-0) seeds were sown in soil and the pots were kept for four days for vernalization (4 • C in darkness) for uniform seed germination. Thereafter, the pots were kept in a growth chamber in a 12 h light (150 µE m −2 s −1 , 20 • C, 60% relative humidity)/12 h dark (16 • C, 70% relative humidity) cycle. At least 30 uniformly developed 23 day-old plants were used for the experiments. The Ascophyllum nodosum seaweed extract, SuperFifty (SF), was produced by and obtained from BioAtlantis Ltd. (Tralee, County Kerry, Ireland). For physiological and molecular analyses, multiple applications of 0.1% SF were performed, three times a day, starting three hours after day light by two-hour intervals, on two consecutive days. This was followed by 15 µM PQ foliar treatment, 24 h after the last SF application. Uniform spraying was achieved by spraying from a 10 cm distance and one spray/plant-the approximate volume of each spray was around 750 µL. Rosette leaves were harvested 24 h after the PQ treatment from three biological replicates for phenotypic and transcriptome analyses and from seven biological replicates for primary metabolite and lipidome analyses. Total RNA was extracted from rosette leaves of the treated plants and controls by RNeasy Plant Mini Kit (Qiagen, Hilden, Germany). Each biological replicate was a pool of at least two rosette leaves.

Electrolyte Leakage Measurements and Trypan Blue Staining
Electrolyte leakage measurements were performed at the rosette stage by checking conductivity with a HI 8733 conductivity meter (Hanna Instruments, Woonsocket, RI, USA) as described in Reference [23]. Visualization of dead cells was done using detached rosette leaves employing a slightly modified version of trypan blue staining protocol explained in Reference [105]. In short, entire detached leaves were incubated for 30 min in hot trypan blue staining solution (80-90 • C). Thereafter, leaves were de-stained by rinsing them once in water, followed by incubation in 15 M chloral hydrate for two hours. The de-staining solution was renewed when needed (two-three times) before taking photos.

Transcriptome Sequencing and Data Analysis
Transcriptome sequencing was performed by LGC Genomics (Berlin, Germany) using the Illumina HiSeq 4000 platform to obtain 75 bp-long single-end reads. The quality of raw reads was evaluated with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Sequencing adaptors were trimmed using Cutadapt [106], and reads below 20 bp length were discarded. Ribosomal RNA contamination was assessed using SortMeRNA (v2.1) [107], and reads aligning to rRNA were filtered out. Processed reads were quantified using kallisto (v0.43.0; bootstraps: 100) [108] against Arabidopsis cDNA sequences (Araport11) [109] to obtain gene expression levels. Differential expression analysis was carried out using EdgeR package in R/Bioconductor [110]. The FDR cutoff < 0.001 and absolute log 2 fold change ≥ 1 were used to identify significantly differentially expressed genes. Generation of heatmaps and clustering of genes were performed with the pheatmap R-package [111], using k-means clustering. GO enrichment of Biological Process terms and KEGG pathway enrichment analyses were carried out using GOSeq [112] and clusterProfiler [113], respectively, with FDR cutoff < 0.05. The enriched GO terms and KEGG pathways were plotted using ggplot2 R-package [114].

Gas Chromatography-Mass Spectrometry (GC-MS) Analysis of Primary Metabolites
GC-MS was employed for the analysis of primary metabolites. Primary metabolites from 50 (±5) mg ground freeze-dried leaf material were extracted following extraction and derivatization procedure as described [115]. Peaks were annotated manually, and ion intensity was determined by the aid of TagFinder software [116], using a reference library from the Golm Metabolome Database [117] and following the recommended reporting format [118]. The relative abundances were normalized to fresh weight of relevant samples and were plotted as heatmap by Multiple Experiment Viewer (MeV; http://mev.tm4.org).

Lipid Profiling
Lipids were extracted from 50 (±5) mg ground freeze-dried leaf material using the methyl tert-butyl ether (MTBE) method [119]; data were analyzed as reported [23]. Raw lipid intensities were log 2 transformed to bring them closer to normal distribution and to exclude the dominant effect of extreme small/large values [120]. The heatmap of fold changes was drawn using MeV.

Analysis of Differential Behavior of Metabolites and Lipids
Analysis of differential behavior was conducted on the data after the preprocessing, described above. Differential behavior between pairwise conditions (i.e., SF + PQ versus H 2

Principal Component Analysis
Principal component analysis (PCA) from the R package pcaMethods [122] was applied to the log 2 transformed metabolite/lipid data; outlier samples were removed. For transcriptome data, the log 2 transformed expression levels were used for PCA analysis using the prcomp function in the R package stats.

Statistical Analysis
Statistical analysis for electrolyte leakage and PCA analysis was performed by GraphPad Prism 6.0 (GraphPad Software, San Diego, CA, USA) using one-way ANOVA followed by Tukey's multiple comparisons test to adjust the p-values.

Conclusions
Our transcriptome, metabolome, and lipidome analysis provided a system-wide level understanding of the protective role induced by the Ascophyllum nodosum biostimulant SF against oxidative stress in Arabidopsis. Comparative transcriptomics identified oxidative stress mechanisms affected by SF priming. Transcription factors and genes linked to stress adaptation mechanisms, such as autophagy and reactive oxygen detoxification, were identified. Moreover, upregulation of photosynthesis, hormone signaling, and growth-related genes provided evidence for stimulation of growth induced by SF. Increased levels of several primary metabolites in SF-primed plants, including maltose and raffinose, suggest a contribution to stress protection. Lipid profiling revealed that alterations in lipids were associated with reduction of cell death and chloroplast degradation and enhancement of intracellular energy balance in SF-primed plants during oxidative stress. Collectively, our data reveal that treatment of plants with SF prior to a stress may be employed as a climate-smart strategy to alleviate oxidative stress-induced damages in crops.
Supplementary Materials: Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/2/474/s1. Figure S1: Expression profiles of all 20 clusters obtained from k-means clustering using heatmap, Figure S2: Comparison of mean PC1 and PC2 values among treatment groups, Figure S3: Scatter plot of PC2 values versus mean electrolyte leakage (%), Table S1: Number of differentially expressed genes for each pairwise comparison, Table S2: List of 35 DEGs significantly upregulated upon SF-priming versus untreated control, Data S1: Data of Figure 1C and Figures S1 and S3, Data S2: Data of Figure 4 and Figure