Physiological and Biochemical Traits of Two Major Arabidopsis Accessions, Col-0 and Ws, Under Salinity

Salinity affects plant growth and development as shown with the glycophyte model plant, Arabidopsis thaliana (Arabidopsis). Two Arabidopsis accessions, Wassilewskija (Ws) and Columbia (Col-0), are widely used to generate mutants available from various Arabidopsis seed resources. However, these two ecotypes are known to be salt-sensitive with different degrees of tolerance. In our study, 3-week-old Col-0 and Ws plants were treated with and without 150 mM NaCl for 48, 72, or 96 h, and several physiological and biochemical traits were characterized on shoots to identify any specific traits in their tolerance to salinity. Before salt treatment was carried out, a different phenotype was observed between Col-0 and Ws, whose main inflorescence stem became elongated in contrast to Col-0, which only displayed rosette leaves. Our results showed that Col-0 and Ws were both affected by salt stress with limited growth associated with a reduction in nutrient uptake, a degradation of photosynthetic pigments, an increase in protein degradation, as well as showing changes in carbohydrate metabolism and cell wall composition. These traits were often more pronounced in Col-0 and occurred usually earlier than in Ws. Tandem Mass Tags quantitative proteomics data correlated well with the physiological and biochemical results. The Col-0 response to salt stress was specifically characterized by a greater accumulation of osmoprotectants such as anthocyanin, galactinol, and raffinose; a lower reactive oxygen detoxification capacity; and a transient reduction in galacturonic acid content. Pectin degradation was associated with an overaccumulation of the wall-associated kinase 1, WAK1, which plays a role in cell wall integrity (CWI) upon salt stress exposure. Under control conditions, Ws produced more antioxidant enzymes than Col-0. Fewer specific changes occurred in Ws in response to salt stress apart from a higher number of different fascilin-like arabinogalactan proteins and a greater abundance of expansin-like proteins, which could participate in CWI. Altogether, these data indicate that Col-0 and Ws trigger similar mechanisms to cope with salt stress, and specific changes are more likely related to the developmental stage than to their respective genetic background.


INTRODUCTION
Salt stress, which results from natural salt accumulation and inappropriate agricultural practices, is a worldwide environmental issue that affects plant growth and development (Munns and Tester, 2008;FAO, 2018). It impairs plant growth by first inducing osmotic stress, due to the excess of salt in the soil resulting in an increased soil osmotic potential, thus disrupting water uptake by roots (Roy et al., 2014). The second phase is ion imbalance, caused by the accumulation of toxic ions such as sodium (Na + ) and chloride (Cl − ), leading to calcium (Ca 2+ ) and potassium (K + ) uptake deficiency within the plant (Munns and Tester, 2008;Sun et al., 2009;Almeida et al., 2017). Furthermore, salt stress is accompanied by an oxidative stress characterized by an accumulation of reactive oxygen species (ROS) (Miller et al., 2010). ROS accumulation affects the apoplast, membrane integrity by lipid peroxidation, and the redox regulation of proteins and may even damage DNA (Apel and Hirt, 2004;Møller et al., 2007;Triantaphylidès et al., 2008). All the major plant physiological and biochemical processes are impaired, including photosynthesis mainly characterized by an alteration of chloroplast structure, loss of chlorophyll, and reduction of CO 2 uptake due to stomatal closing (Chaves et al., 2009;Stepien and Johnson, 2009;Gao et al., 2019).
The changes induced by salt have consequences on carbon partitioning and allocation Dong et al., 2018). For instance, in Arabidopsis thaliana (Arabidopsis), the overall allocation is inhibited, and carbon partitioning is reduced into starch and increased into sugars (Kempa et al., 2008;Dong et al., 2018). Soluble sugars can act as osmoprotectant metabolites, but also as signaling molecules that modulate the expression of genes important for salt tolerance (Seifert, 2004;Gong et al., 2005;Kempa et al., 2008;Sun et al., 2013). Most of the carbon fixed by plants during photosynthesis is incorporated into cell wall polysaccharides (Seifert, 2004). The cell wall that is mostly composed of cellulose, hemicelluloses, and pectin (Carpita and Gibeaut, 1993) is also altered during exposure to salt stress as it accumulates salt when vacuoles cannot store it anymore, thereby inhibiting growth (Le Gall et al., 2015). The loss of function of several cell wall-related genes [UDP-arabinose 4epimerase 1 (AtMUR4), pectin methylesterase 31 (AtPME31)] or the overexpression of an expansin (AtEXP3) or a xyloglucan endotransglucosylase-hydrolase (AtXTH30) gene increases salt sensitivity (Kwon et al., 2008;Yan et al., 2018Yan et al., , 2019Zhao et al., 2019). Meanwhile, the loss of function of AtXTH30 leads to salt tolerance. The maintenance of cell wall integrity (CWI) during salt stress is important in plant salt tolerance. The module composed of leucine-rich repeat extensins (LRX3/4/5), rapid alkalinization factors (RALF22/23), and the Feronia plasma membrane receptor kinase (FER) participate in CWI under salt stress (Feng et al., 2018;Zhao et al., 2018).
Environmental pressures lead plants to local adaptation and to generate natural phenotypic and genetic variations. Because of its widespread geographic distribution, Arabidopsis experiences a broad range of climatic conditions (Mitchell-Olds and Schmitt, 2006). Therefore, many Arabidopsis accessions are available, and some of them are salt-tolerant (Bu-5, Bur-0, Ll-1, Wl-0, Ts-1, and Tsu-1) when compared to Col-0 (Rus et al., 2006;Baxter et al., 2010;Katori et al., 2010;Julkowska et al., 2017). However, Arabidopsis is a glycophyte (Kosová et al., 2013) and, surprisingly, is widely explored as a model plant to investigate the molecular mechanism of salt stress tolerance. Indeed, little is known about salinity tolerance of the main reference accessions Wassilewskija (Ws) and Columbia (Col-0), albeit a study reporting that Ws was more salt-tolerant than Col-0, and this was associated with an increase of the expression level of AtAVP1 encoding a vacuolar H + -translocating pyrophosphatase involved in regulating the movement of Na + (Jha et al., 2010). That is surprising as most of the Arabidopsis studied mutants are in these well-known genetic backgrounds and available from different seed stock centers.
The aim of this study was to unravel the main differences existing between Col-0 and Ws in their strategy to cope with salt stress. Although root is important to reduce sodium transport into the shoot tissue (Duan et al., 2013;Geng et al., 2013;Julkowska et al., 2014), we have focused our study on shoots because the regulation of salt accumulation in aboveground organs is essential for plant survival.

Measurement of Fresh and Dry Mass and Shoot Water Content
During the time course of the experiments, shoots were harvested from five plants for each genotype. The fresh mass (FM) was immediately determined. The samples were dried in an oven at 65 • C for 3 days to measure the dry masses (DMs). The shoot water content (SWC) and the tolerance index in the shoots (TI) were calculated using the following formula (Jha et al., 2010;Julkowska et al., 2016): SWC (g water/g DM) = (shoot FM − shoot DM)/shoot DM; TI = shoot DM salt/shoot DM control.

Pigment Content
Photosynthetic pigments were extracted from 10 mg of freezedried shoot powder and quantified according to Duran Garzon et al. (2020). The pigments were separated by reverse-phase highperformance liquid chromatography (Prominence, Shimadzu Co., Kyoto, Japan) using a Zorbax Eclipse PAH column (4.6 × 150 mm, 3.5-µm pore size; Agilent Technologies, Santa Clara, CA, United States) and a UV diode array detector (SPD-M20A; Prominence, Shimadzu Co.) as described by Edelenbos et al. (2001). Commercial pigment standards (neoxanthin, violaxanthin, antheraxanthin, lutein, zeaxanthin, chlorophyll a and b, and β-carotene) from DHI-Water and Environment (Hørsholm, Denmark) were used for calibration. Canthaxanthin (Sigma-Aldrich, St. Louis, MO, United States) was used as an internal standard for quantification of each pigment, except β-carotene, where a standard curve was performed. Total anthocyanin content was assessed according to Kim et al. (2017). Briefly, 20 mg of frozen leaves was homogenized in 600 µL of an acidified methanol buffer (methanol/HCl 99/1 vol/vol) and extracted under gentle agitation overnight at 4 • C in the dark. The anthocyanin extract was purified using 800 µL of water/chloroform (50/50 vol/vol) and centrifuged at 10,000g for 5 min. The absorbance of the upper phase was first read at 530 nm and further at 657 nm using a microplate reader (Powerwave; Biotek, Colmar, France). The total anthocyanin content was calculated using the formula A 530 − 0.25 A 657 , and the result was expressed as corrected A 530 · g −1 FM (Rabino and Mancinelli, 1986).

Antioxidant Enzyme Activity Assays
Antioxidant enzyme activity assays are described in Supplementary Material.

Organic Polar Primary Metabolites Analysis
Primary polar metabolites were extracted from 10 mg of freezedried shoot powder as described by Pontarin et al. (2020). Briefly, the sample was homogenized in 400 µL of prechilled methanol supplemented with 200 nmol of ribitol as internal standard and mixed for 10 min at 70 • C. Chloroform (200 µL) was added to the reaction and mixed again at 70 • C for 5 min. Distilled water (400 µL) was added to the mixture (vortexed for 20 s) before centrifugation (10 min, 10,000g). An aliquot of the methanol/water supernatant (50 µL) that constitutes the crude extract was speed-vacuum dried and resuspended in 40 µL of 20 mg · mL −1 methoxyamine hydrochloride in pyridine for 2 h at 37 • C. N-trimethylsilyl-N-methyl trifluoroacetamide (70 µL) was added to the sample and heated at 37 • C for 30 min. Gas chromatography-mass spectrometry (GC-MS) analysis was performed on a system composed of a TRACE GC ULTRA gas chromatograph and a DSQII quadrupole mass spectrometer (Thermo Fisher Scientific, Waltham, MA, United States).

Cell Wall Sugar Composition
Plant cell wall material was prepared from 150 mg of frozen shoot powder as described by Duran Garzon et al. (2020). Dry cell wall material was digested with amylase according to Fleischer et al. (1999). After digestion, the dry cell wall (2 mg) was hydrolyzed, and the samples were injected onto a CarboPac-1 column (Dionex) HPAEC, separated, and detected by PAD. Data were collected and processed using the Chromeleon 6.50 software (Dionex Corp.) (Duran Garzon et al., 2020).

Proteomics Analysis
Proteins were extracted from the shoots using a global protein extraction method adapted from Ma et al. (2016) and described by Leschevin et al. (2021). Proteins were identified and quantified using a Tandem Mass Tags (TMTs)-based comparative proteomics analysis method (Leschevin et al., 2021). These experiments were performed on three biological replicates. Data were processed using Proteome Discoverer 2.4 (Thermo Fisher Scientific) before being run against The Arabidopsis information Resource (TAIR10_pep_20101214), as well as against Swissprot 1 . Proteins were annotated using the ProtAnnDB (San Clemente et al., 2009) and Uniprot 2 databases. The MS/MS data (raw data, identification, and quantification results) are available at ProteomeXchange with identifier PXD022441 3 . To identify differentially accumulated proteins (DAPs), the results were classified according to the ratio of amount of proteins in saltstressed shoots vs. control shoots for a given ecotype at T48, T72, and T96. We have defined the DAPs as those with a fold change >1.7 or <0.6 at a p < 0.05 between two comparison groups (T48, T72, and T96).

Statistical Analyses
Statistical analyses were performed using Kruskal-Wallis test (Kruskal and Wallis, 1952). Multiple comparisons after the Kruskal-Wallis test were performed using post hoc Dunn test (Dunn, 1964). A significance threshold of 0.05 was applied in all tests. R version 4.0.2 was used for all analyses.

Physiological Assessment of Ws and Col-0 Salt-Treated Plants
To determine the natural variation between Col-0 and Ws, control and salt-treated plants were phenotyped by measuring different parameters including the FM of the shoots, the SWC, and the tolerant index of the shoots (TI). Under control conditions, Col-0 only showed rosette leaves, whereas Ws already produced the main inflorescence stem ( Figure 1A). A decrease in FM was observed in both ecotypes after a prolonged salt stress exposure (T72 and T96) (Supplementary Figure 1A). This reduction in shoot biomass was associated with a significantly

The Mineral Status in Col-0 and Ws Shoots Under Salt Stress
The macronutrients content including Na + , K + , Cl − , NO 3 − , and SO 4 2− was determined in the shoots of Ws and Col-0. Na + and Cl − concentrations quickly reached toxic levels in both ecotypes under salt treatment with a Na + increased level by nearly 33-fold (Ws) and 36-fold (Col-0) at T48, 62-fold (Ws), and 66-fold (Col-0) at T72, and 48-fold (Ws) and 45-fold (Col-0) at T96 ( Table 1), indicating that the capacity to accumulate Na + into their shoots was similar in both ecotypes. In a similar manner, the content of Cl − accumulated in both ecotypes in response to salt stress. However, this was in a range of 1.6-fold greater in Ws at T48 (95fold) and T72 (136-fold) as compared to Col-0 (59-fold at T48 and 82-fold at T72), before reaching a similar trend at T96 (67fold in Col-0 vs. 74-fold in Ws). This suggests that the Cl − uptake and its translocation to shoots are higher in Ws than in Col-0 under high salinity.
As salinity can affect nutrient uptake, we then measured the amounts of K + , NO 3 − , and SO 4 2− . Under control conditions, the other ions (K + , NO 3 − , and SO 4 2− ) showed a similar content in both ecotypes. When the plants were exposed to salt stress, a decreased level of K + by nearly twofold was observed at T96 in both ecotypes. The K + /Na + ratio greatly dropped under salt stress by 50-fold at T48 and nearly by 100-fold at T72 and T96 in both accessions, confirming that Na + reduces K + uptake in both ecotypes. The NO 3 − content was reduced by two-and threefold at T48 upon salt stress in Ws and Col-0, respectively, and this decrease reached 5.0-and 7.4-fold at T96 in Ws and Col-0, respectively. As a result, the NO 3 − /Cl − ratio dramatically dropped by 350-and 500-fold at T96 in Ws and Col-0, respectively. Furthermore, SO 4 2− uptake was reduced by 154-and 161-fold at T96 as shown by the SO 4 2− /Cl − ratio in Ws and Col-0, respectively. All these results showed that high levels of Na + and/or Cl − uptake in plants reduce NO 3 − and SO 4 2− translocation to shoots and may have a negative effect on sulfur and nitrogen metabolism in both ecotypes.

Effect of Salt Stress on Photosynthetic Pigment and Anthocyanin Contents in the Shoots of Ws and Col-0
Salt stress showed a visible leaf pigment alteration phenotype in both ecotypes, albeit stronger in Col-0 ( Figure 1A). That was correlated with a reduction in chlorophyll (chl) a by 1.5-fold in both ecotypes and in chl b more pronounced in Col-0 than in Ws (1.6-fold in Col-0 vs. 1.3-fold in Ws) (Supplementary Figure 2A). The carotenoids content decreased similarly in both Col-0 (1.5-fold) and Ws (1.4-fold) compared with control plants (Supplementary Figure 2B). The anthocyanin content was significantly higher in Col-0 (1.5-to 2.8-fold) than in Ws, under control conditions (Figure 2). Under salt exposure, an increase in anthocyanin was observed in both ecotypes. However, Col-0 produced more anthocyanin than Ws under salinity (Figure 2). Unlike at T48, where a similar increase by approximately fourfold was observed in both ecotypes, Col-0 accumulated more anthocyanin at T72 (sevenfold) and at T96 (fivefold) than Ws, where a two-and threefold increased level was observed at T72 and T96, respectively.

Antioxidant Enzyme Activities in Ws and Col-0 Shoots After Exposure to Salt Stress
Under control conditions, SOD, MDHAR, and DHAR activities had average values from 1.2-to 1.9-fold higher in Ws than in Col-0 ( Figures 3A-C), whereas APOX, GR, and CAT activity levels were similar in both ecotypes (Supplementary Figures 3A-C). These ROS-scavenging enzymatic activities significantly increased in response to salt stress in both ecotypes. However, unlike control conditions, most antioxidant enzymatic activities were significantly higher in Col-0 as compared to Ws. SOD activity strongly increased by 4-to 6.5-fold in Col-0, whereas it was only 1.5-to 2-fold greater in Ws as compared to control plants ( Figure 3A). MDHAR enzymatic activity increased more in Col-0 (2.8-fold) than in Ws (1.5-fold) to reach a similar activity level in both ecotypes ( Figure 3B). DHAR enzymatic activity was at least twofold higher in salt-treated Col-0 shoots during the time course of the experiment as compared to Ws where a 1.3-fold increased level was observed ( Figure 3C). CAT enzymatic activity was induced by more than twofold in both ecotypes, and APOX and GR enzymatic activities increased similarly in both ecotypes at T96 (Supplementary Figure 3).

Primary Metabolic Changes in Col-0 and Ws Shoots Under Salt Stress
To monitor the metabolic variations in both ecotypes under salt stress, metabolomics analysis was performed to identify differentially accumulated organic polar metabolites. In total, 33 primary metabolites (14 amino acids, 2 polyamines, 11 soluble sugars, and 6 organic acids) were reproducibly identified in the shoots of Ws and Col-0 in both control and salt-treated plants. The results are presented as the log2(salt/control)-fold change and integrated into a metabolites map (Figure 4 and Supplementary Table 1) as described by Wu et al. (2013). Most primary metabolites were significantly affected upon salt stress exposure. Nitrogen metabolism was altered with a higher content of amino acids. Pro, which is the most frequently Data are means ± standard deviation of three biological replicates. Means followed by the same letter in the same graph are not significantly different according to Dunn test. encountered compatible solute during an osmotic stress, was the most accumulated metabolite with an average value of nearly 8.5-fold in both ecotypes upon salt stress exposure. Contents of other amino acids, such as serine, leucine, isoleucine, and valine, increased by an average value of three-to fourfold in both ecotypes. Phenylalanine, derived from the shikimate pathway, increased greater in Ws (2.7-fold) than in Col-0 (2.2-fold) under salt stress. However, the highest accumulation of Pro, methionine, threonine, valine, glutamate, and glutamine occurred earlier in Col-0 (T48) as compared to Ws (T72 and T96). The level of non-protein amino acid, γ-aminobutyric acid, increased similarly at T48 in both ecotypes and continued to accumulate by 2.6-fold in Ws at T72 as compared to Col-0 where it started to decrease. The content in polyamines including putrescine and spermidine decreased in both ecotypes.
The carbohydrate metabolism also changed upon salt stress in both ecotypes as shown by an alteration of sugar content. Sucrose, a well-known osmolyte, was accumulated by an average value of fourfold in salt-treated plants. The amount of trehalose and maltose increased by more than twofold at T96 in both ecotypes, indicating a role of these disaccharides as a compatible solute and a starch remobilization under salt exposure. Raffinose, which is derived from the raffinose family of oligosaccharides (RFOs) biosynthesis pathway using myo-inositol and galactinol, was the most accumulated sugar in Col-0 with an average value of nearly fivefold vs. 3.7-fold in Ws upon salt treatment. Galactinol content increased in both ecotypes and was 1.5-fold greater in Col-0 than in Ws salt-treated plants.

Sugar Composition of Non-cellulosic Cell Wall Polysaccharides in Ws and Col-0 Salt-Treated Shoots
As carbohydrate metabolism was altered in both Col-0 and Ws under salinity, the sugar composition of non-cellulosic cell wall polysaccharides was determined ( Table 2 and Supplementary  Table 2). Under control conditions, Ws contained more mannose (D-Man, 1.4-fold) and xylose (Xyl, 1.8-fold) residues than Col-0. Galactose (Gal) and galacturonic acid (GalUA) content was 1.3-fold higher in Col-0 than in Ws. Salt stress induced cell wall remodeling in both ecotypes. Pectin was altered in Col-0 as shown by a transient reduction by 1.4-fold in GalUA residues amount at T72. Additionally, salt treatment increased Gal content in Ws and Col-0. That was greater in Col-0 (1.5-fold) than in Ws (1.3-fold) at the end of the treatment, and changes occurred earlier in Col-0 as compared to Ws. Similarly, Ara content increased by 1.3-fold in both ecotypes, but that occurred earlier in Col-0 (T72) than in Ws (T96).

Quantitative Proteomics Data Related to the Studied Traits in Col-0 and Ws Shoots
A TMTs-based comparative proteomic analysis was performed on total soluble proteins from the shoots of control and salttreated plants. All the data described in this paragraph are available in Supplementary Tables 3, 4. Altogether, upon salt stress exposure, there were 542 DAPs in Ws and 558 in Col-0. The distribution of overaccumulated (316 Ws vs. 331 Col-0) and underaccumulated proteins (226 Ws vs. 227 Col-0) was similar in both ecotypes (Figures 5A,B). There were common 75 underaccumulated and 159 overaccumulated proteins between Ws and Col-0 (Figures 5C,D). Based on our physiological and biochemical data, we focused the following analysis on DAPs involved in photosynthesis, primary and secondary metabolisms, ROS detoxification, and cell wall.
Among the DAPs, 53 (9.5%) and 66 (12%) were related to photosynthesis in Col-0 and Ws, respectively (Figure 6). The number of overaccumulated proteins was lower (13) than that of underaccumulated proteins (40) in Col-0, whereas in Ws, the numbers of proteins overaccumulated (28) and underaccumulated (32) were similar (Supplementary  Tables 3, 4). Additionally, Ws contains six DAPs (petD, FD3, CURT1A and CURT1B, THF1, P19), which were either overaccumulated or underaccumulated, depending on the time points. The DAPs involved in photosynthetic pigment biosynthesis were more numerous in Col-0 (23) than in Ws (11). The amount of a pheophorbide a oxygenase (PaO) involved in chlorophyll catabolism nearly doubly increased at T96 in both Col-0 and Ws. Meanwhile, proteins involved in chlorophyll biosynthesis including chelatases (four in Col-0 vs. three in Ws), protochlorophyllide reductases (two in Col-0), and other chlorophyll biosynthesis-related proteins (nine in Col-0 vs. three in Ws) were underaccumulated upon salt stress treatment. The abundance of proteins related to carotenoid biosynthesis pathway as LUT2 and PSY1 was also reduced in both Col-0 and TABLE 2 | Monosaccharide distribution of non-cellulosic cell wall polysaccharides in Ws and Col-0 at T48, T72, and T96 under control and 150 mM of NaCl conditions. l-Arabinose (Ara), d-galactose (Gal), d-xylose (Xyl), and l-galacturonic acid (GalUA) content. The complete non-cellulosic sugar composition profile can be found in Supplementary Table 2. Individual sugars are expressed in mole% of the total non-cellulosic cell wall polysaccharides. Data are means ± standard deviation (SD) of three biological replicates. Means followed by the same letter in the same series of values are not significantly different according to Dunn test.

Mole%
Frontiers in Plant Science | www.frontiersin.org Ws. LUT5, and the carotenoid cleavage dioxygenase 4, CCD4, followed a similar trend (Supplementary Tables 5, 6). The amount of proteins involved in starch catabolism (BAM5 and AMY1) strongly increased in Col-0 upon salt stress exposure (Supplementary Table 4). Other proteins related to starch degradation (SEX1/GWD1 and SEX4) were the most abundant in Ws at T96, whereas this occurred earlier in Col-0 (T48). Meanwhile, proteins involved in starch biosynthesis either overaccumulated (APL3, fourfold) in Col-0 and Ws shoots or underaccumulated (PTST3, twofold) in Col-0 at T96. Proteins related to starch/maltose to sucrose conversion (DPE2) or sucrose biosynthesis (SPS1) overaccumulated by 1.7-fold in Ws upon T72. The abundance of SUS1, which plays a role in directing the carbon flow to cell wall or starch biosynthesis, overdoubled in both Col-0 and Ws, but this occurred earlier in Col-0 (T48) than in Ws (T72). Sugar transport was also reduced in both ecotypes as shown with a lower abundance of SUC1 and STP1 transporters in Ws and Col-0, respectively.
Raffinose synthase (RFS5) overaccumulated in Ws, whereas RFS6 was doubly reduced in both Ws and Col-0. RFS6 returned to its original level in Col-0 at T96, whereas it continued to decrease in Ws (Supplementary Tables 3, 4). The myoinositol-1-phosphate synthase 1 (MIPS1) overaccumulated in both Col-0 and Ws, and that was twofold greater in Col-0 as compared to Ws at T48. Proteins involved in amino acid biosynthesis (ALS, VAT1, CGS1, LYSA1, CYSC1, HISN1A, SSU, ASN2, DHS2, and AK2) were underaccumulated in Ws and Col-0 shoots. The amount of delta-1-pyrroline-5carboxylate synthase 1 (P5CS1) involved in Pro biosynthesis increased by twofold in both Ws and Col-0. Similarly, proteins involved in polyamine metabolism (ARGAH2 and DELTA-OAT) overaccumulated in both Ws and Col0. NATA1, which is involved in the reduction of putrescine level, strongly overaccumulated in Col-0 upon salt stress (more than sevenfold), whereas it was not detected in Ws.
Proteins predicted to be secreted were well represented among the DAPs. They were 62 and 66 (i.e., approximately 12%) in Col-0 and Ws, respectively, with 90% of them already identified in at least one cell wall proteome (see WallProtDB 4 ). Twentythree DAPs were common to Ws and Col-0, whereas 43 were specific to Ws and 39 to Col-0 (Supplementary Tables 3, 4). These proteins belonged to the main functional classes of cell wall proteins previously defined (Jamet et al., 2008). However, Ws and Col-0 exhibited distinct patterns: (i) there were more DAPs acting on cell wall polysaccharides (16 vs. 11), more miscellaneous proteins (10 vs. 6), and proteins of yet unknown function (9 vs. 5) in Col-0 than in Ws; and (ii) there were more DAPs related to lipid metabolism (11 vs. 3), more proteins with interaction domains with proteins or polysaccharides (11 vs. 6), and more proteins possibly involved in signaling (5 vs. 2) in Ws than in Col-0. Some protein families deserve additional comments. Among the DAPs acting on cell wall polysaccharides, four glycoside hydrolase (GH) families were specifically represented in Col-0: GH9 (endoglucanases), GH16

DISCUSSION
In this study, two Arabidopsis accessions, Ws and Col-0, were used to identify new traits that could discriminate the degree of salt tolerance between them. We have evaluated the growth; the mineral status; photosynthetic pigments; antioxidative defense systems, such as ROS-scavenging enzymes, anthocyanin, and osmolytes from primary metabolism; and cell wall changes. We have combined physiological approaches with proteomics and metabolomics analyses (Figure 7).
Salinity tolerance is related to the ability to preserve steady water content and an accumulation of dry biomass as reported by Ellouzi et al. (2011). We observed a growth reduction in both ecotypes after salt exposure (Supplementary Figure 1A). That was correlated with lower SWC and TI values ( Figure 1B and Supplementary Figure 1B). Aquaporins (PIPs and TIPs) play a key role in plant water relation and are involved in osmotic balance and nutrient homeostasis in response to salinity (Boursiac et al., 2005;Afzal et al., 2016;McGaughey et al., 2018). It has also been shown that PIPs could transport CO 2 under water limiting conditions for photosynthesis capacity (Uehlein et al., 2012;Boudichevskaia et al., 2015). Our proteomics data were consistent with the water status of the shoots, which was reduced similarly in Ws and Col-0 in response to salt stress. We could identify two aquaporins (TIP2 and PIP1A) in Ws and one (PIP1D) in Col-0 whose amounts were salt-reduced by twofold (Supplementary Tables 5, 6). Therefore, according to McGaughey et al. (2018), the lower SWC and aquaporin abundance observed in both ecotypes suggest a limited capacity of salt-treated Ws and Col-0 plants to withstand from salt during a short period of stress. The steady water content under salt stress is related to the ability to exclude sodium from the leaves or ion compartmentalization capacity (Almeida et al., 2017). Both Ws and Col-0 showed a strong accumulation of Na + ions in leaves, indicating an alteration of sodium homeostasis in plants (Table 1). We did not identify any Na + transporter as HKT1 nor SOS proteins in our proteomics study. That can be explained by our extraction method, which mainly extracts soluble proteins, among many abundant photosynthesis-related proteins overlaying less abundant proteins during MS analysis. Under salt stress, K + uptake is limited by the competition between Na + influx and K + efflux, which is stimulated by Na + accumulation into the cell (Assaha et al., 2017). In our study, the excess of Na + prevents K + uptake in both Col-0 and Ws (Table 1). Many studies have reported that maintaining cellular Na + /K + homeostasis is a key factor for salt tolerance (Mostofa et al., 2015;Singh et al., 2018). K + is an essential macronutrient involved in not only photoprotection and optimal photosynthetic chain electron transport (Dukic et al., 2019) but also regulation of intracellular and chloroplast osmolarity (Barragán et al., 2012;Tsujii et al., 2019). The amount of K + /H + antiporters (KEA-1, KEA-2, and KEA-3), which are located in chloroplast (Kunz et al., 2014) and may play an important role in the initial rapid Ca 2+ response to hyperosmotic stress (Stephan et al., 2016;Tsujii et al., 2019), remained unchanged in Ws, whereas the amount of KEA1 and KEA3 likely reduced (0.7-fold) at T96 in Col-0 (Supplementary Tables 5, 6). Although we observed a similar Na + accumulation in both ecotypes, it is possible that Col-0 limits excess of Na + by closing the K + channels KEA1 and KEA3. This indicates not only a stronger alteration of K + transport activity, but also a limited photosynthetic capacity more pronounced in Col-0 than in Ws.
Photosynthetic pigments are altered under salt stress (Esteban et al., 2015;Ben Abdallah et al., 2016). The photosynthetic pigments were reduced in both ecotypes and were stronger in Col-0 than in Ws (Supplementary Figure 2). These data correlated well with the proteomics data where proteins involved in the PaO pathway (i.e., chlorophyll degradation) and carotenoids biosynthesis pathway overaccumulated and underaccumulated, respectively, greater in Col-0 than in Ws upon salt stress (Supplementary Tables 3, 4). Additionally, more proteins related to the photosynthetic electron transport chain are underaccumulated in Col-0, whereas most of them are overaccumulated in Ws. All these data indicate that Ws seems to maintain its photosynthetic capacity at a higher level than Col-0 under salinity.
To safeguard their photosystems under salt stress, but also to maintain cellular homeostasis or carbon storage, plants accumulate osmoprotectants as soluble sugars (Nishizawa et al., 2008;Schneider and Keller, 2009;Knaupp et al., 2011;ElSayed et al., 2014), anthocyanin (Kim et al., 2017), or amino acids (Dong et al., 2018). Starch remobilization is important to provide carbon and energy when photosynthesis is limited under abiotic stress (Thalmann and Santelia, 2017). It provides organic acids, sugars, and sugar-derived osmolytes for osmotic adjustment (Monroe et al., 2014;Santelia and Lunn, 2017). Thalmann et al. (2016) showed that the reduction in starch accumulation in Arabidopsis plants exposed to high osmotic stress results from induced starch degradation, leading to the production of maltose, the major starch catabolite. We did not measure the starch content, but the level of maltose increased in both ecotypes and was twofold greater in Col-0 than in Ws at T72 (Supplementary Table 1). That was consistent with an overaccumulation of proteins involved in starch breakdown as AMY1, BAM5 (Col-0), GWD1 (Ws), and SEX4 (Col-0 and Ws) (Supplementary Tables 3, 4). Interestingly, APL3 (AGPase), a key enzyme of starch biosynthesis pathway, overaccumulated in both ecotypes and to a higher level in Col-0. This indicates that salt stress stimulates both starch catabolism and biosynthesis in Ws and Col-0 to mitigate salt stress (Kempa et al., 2008;Thalmann and Santelia, 2017). Sucrose increased by fourfold in both ecotypes under salinity. That was consistent with the abundance of the proteins SPS1 and DPE2 in Ws. SUS1 promotes sucrose degradation and has been shown to be possibly involved in cellulose and starch biosynthesis in Arabidopsis ( Baroja-Fernández et al., 2012). SUS1 overaccumulated in both Col-0 and Ws, but that was greater and earlier in Col-0 than in Ws. Furthermore, the decreased amount of a cell wall invertase (cwINV1), a sugar transport protein (STP1) in Col-0, and a sucrose transport protein (SUC1) in Ws suggests a reduction in carbon partitioning with consequences on sourcesink relationships (Thalmann et al., 2016).
Sugar osmolytes such as raffinose or galactinol are derivatives of sucrose and were more abundant in Col-0 than in Ws (Figure 4 and Supplementary Table 1). Surprisingly, we observed a strong underaccumulation of the raffinose synthase RFS5, in both Col-0 and Ws. However, the level of RFS5 was recovered in Col-0 at T96, and the amount of another raffinose synthase, RFS6, doubly increased in Ws upon salt stress. Raffinose biosynthesis could also arise from stachyose catabolism through raffinose a-galactosidase (AGAL) activity. We identified two AGALs (AGAL1 and AGAL2) in both ecotypes where AGAL1 increased by 1.6-fold upon T72 in Col-0, but not in Ws (Supplementary  Tables 5, 6). The higher levels of raffinose and galactinol measured in Col-0 may be important as osmoprotectants to reduce the oxidative damages.
Pro, another compatible solute, as well as other amino acids (serine, glycine, phenylalanine, valine, threonine, leucine, and isoleucine), accumulated in both ecotypes (Figure 4 and Supplementary Table 1). However, Pro, valine, leucine, and phenylalanine levels were higher in Col-0 at T48. Additionally, P5CS1, which contributes to stress-induced Pro accumulation, was twice more abundant under salt stress in both ecotypes, which agrees with a former study on atp5cs mutants exposed to salt stress (Funck et al., 2020). Underaccumulated proteins involved in amino acid biosynthesis were more abundant in Col-0 than in Ws, and the amount of overaccumulated proteins involved in amino acid catabolism was similar in both ecotypes. This suggests an active protein degradation in response to salt stress in both ecotypes, and these changes often occurred earlier in Col-0 than in Ws. Increased amounts of these amino acids can be considered as a salt tolerance mechanism to promote osmolytes or nitrogen allocation to produce non-protein amino acids or secondary metabolites (Kempa et al., 2008;Dong et al., 2018;Batista-Silva et al., 2019;Jia et al., 2019). Their complete oxidation could also provide a source of energy possibly required to prepare the recovery phase after salt stress (Hildebrandt et al., 2015).
Anthocyanin acts as an antioxidant molecule and is actively synthesized under salt stress (Chutipaijit et al., 2011;Zhang et al., 2012;Cheng et al., 2013;Landi et al., 2015;Kim et al., 2017). The overaccumulation of anthocyanin in response to stress reduces photosynthesis efficiency (Chalker-Scott, 1999;Gould et al., 2000). In our study, the anthocyanin content increased greater in Col-0 than in Ws (Figure 2). These results are supported by the work of Chunthaburee et al. (2016), which showed that salt-tolerant rice cultivars displayed smaller increase in anthocyanin content than salt-sensitive cultivars. Specific proteins involved in anthocyanin biosynthesis such as LDOX and one glycosyltransferase (UGT75C1) were more abundant in Col-0 than in Ws (Supplementary Tables 3, 4). Furthermore, the amount of proteins involved in the early steps of the phenylpropanoid pathway (F3H, CHS, and CHI) increased in Col-0. Therefore, the overaccumulation of anthocyanin observed in Col-0 likely appears more as a trait of sensitivity to salt tolerance.
Other mechanisms, such as ROS-scavenging enzymes, including SOD, CAT, and APOX, have been developed in plants to alleviate the oxidative damages caused by ROS (Mittler, 2002;Ahanger et al., 2017;Gupta et al., 2018). In our study, an increase in antioxidative enzymatic activities was observed in both ecotypes in response to salt stress (Figure 3 and  Supplementary Figure 3). Most of these activities were higher in Col-0 than in Ws (Figure 3). The lower increase in antioxidative enzymatic activity observed in Ws under salt stress could be explained by a greater activity under control conditions. SOD enzymatic activity, which acts as the first line of defense against oxidative damage in chloroplasts, was higher in Col-0 than in Ws, suggesting that a stronger oxidative stress was generated in Col-0. Surprisingly, the amount of the SOD MSD1 protein was greater in Ws in response to salt stress (Supplementary  Tables 3, 4). However, other SODs (FSD1, FSD2, FSD3, CSD1, and CSD2) were identified in both ecotypes, and the abundance of FSD3, CSD1, and 2 likely decreased in Ws, whereas they remained unchanged or slightly increased (CSD2) in Col-0, indicating a higher H 2 O 2 production in Col-0 (Supplementary Tables 5, 6). The correlation observed between SOD enzymatic activities and SOD protein abundance was also shown for the enzymes involved in H 2 O 2 detoxification. The detoxification of H 2 O 2 was favored in Col-0 by the ascorbate-glutathione cycle, whereas Ws promotes CAT activity. Therefore, the oxidative metabolism seems to be better regulated in Ws than in Col-0, with coordinated SOD and CAT activities allowing to keep a balance between ROS production and detoxification and to reduce oxidative load to ascorbate-glutathione cycle, while GR activity maintains the redox status (Mostofa et al., 2015).
The cell wall is the first interface between the plant cell and its environment and has important role in the perception of environmental stimuli, resulting in activation of cellular responses to biotic and abiotic stresses (Le Gall et al., 2015;Feng et al., 2018;Novaković et al., 2018;Zhao et al., 2018). We observed a transient reduction in pectin in Col-0 and an increase of arabinan and galactan in both ecotypes after salt exposure. Xyl content remained unchanged in both ecotypes under salt stress ( Table 2). That can be explained by the decreased abundance of a bifunctional β-L-arabinofuranosidase/β-D-xylosidase (BXL1) in Col-0 (Supplementary Table 4). A similar trend occurred in Ws (Supplementary Table 5). In Arabidopsis vascular tissues, BXL1 releases Xyl from xylan and arabinoxylan, but also arabinose from arabinoxylan and arabinan (Minic et al., 2004). The decrease in BXL1 abundance observed in Col-0 and to a certain extent in Ws could maintain arabinoxylan and arabinan content in our conditions. Additionally, a decrease in XTH4 and XTH6 (T48 and T72) abundance in Col-0 concomitantly to an increase in XTH24 (T96) could contribute to maintain the xyloglucan (XG) level in Col-0 (Supplementary Table 4). Besides, salt stress transiently reduced the amount of GalUA residues in Col-0 (T72), indicating pectin degradation ( Table 2). Pectin is essential for the CWI (Bethke et al., 2016;Feng et al., 2018). Feng et al. (2018) reports that the FER is a sensor of the CWI, which interacts with the salt-induced disruption of pectin and induces cell wall repair by triggering a cytoplasmic Ca 2+ signaling pathway. Zhao et al. (2018) showed that LRX3/4/5, RALF22/23, and FER are critical to maintain CWI under salt stress. We identified LRX3 in Col-0 but not in Ws, and the amount likely increased in response to salt stress in Col-0 (Supplementary Table 6). FER, rapid alkalinization factors (RALFs), and other LRXs, which are crucial players of the signaling pathway regulating plant growth under salt stress, might be synthesized earlier than at T48 in salttreated plants. We could also identify a cell wall-associated kinase WAK1 in Col-0, which can activate plant stress response after binding to pectin fragments (Brutus et al., 2010;Kohorn and Kohorn, 2012). The abundance of WAK1 doubly increased in Col-0 up to T72 (Supplementary Table 4). Thus, the reduction in GalUA residues concomitantly with the increase in WAK1 abundance observed in Col-0 suggests the existence of a cell wall signal to maintain CWI in Col-0. Arabinan and galactan increased in both ecotypes, but started earlier in Col-0. These polysaccharides are mainly found in rhamnogalacturonan I (RG-I) side chains, which play a role in the hydration status of the cell wall matrix for their high water-binding capability and ability to form gels that could preserve cells from damage (Leucci et al., 2008). Zhao et al. (2019) reported that Ara is important to maintain root growth under salt stress in Arabidopsis. Ara and Gal residues also decorate hydroxyproline-rich glycoproteins such as arabinogalactan proteins (AGPs) and extensins (Draeger et al., 2015). Olmos et al. (2017) showed that AGPs are released under salt stress to maintain CWI and suggests a role of AGP in Na + /Ca 2+ exchange. The increased amounts of Ara and Gal residues observed in both genotypes suggest rather an increase in AGPs to maintain CWI under salt stress. AtFLA4/SOS5, which is a fascilin-like AGP, positively regulates cell wall biosynthesis and root growth by modulating ABA signaling in response to salt stress (Seifert et al., 2014). Shi et al. (2003) have proposed that AtFLA4 is involved in ionic interactions, most likely with other FLAs, such that they form a network at either the plasma membrane surface, or in the cell wall that controls the rate of cell expansion. We identified one FLA (FLA18) in Col-0 whose amount doubled at T72 and five different FLAs (FLA1, FLA8, FLA9, FLA13, and FLA16) in Ws whose amounts nearly tripled at T72. Besides FLA16, other FLAs overaccumulated in Ws and could be detected in Col-0 but with no change. FLA16 has been shown to be involved in stem development (Liu et al., 2020) and was not identified in Col-0, which could be explained by its rosette leaves stage. Thus, the overaccumulated FLAs observed in Ws, except for FLA16, could be important in the maintenance of CWI. Similarly, FLA18, WAK1, and LRX3, which have been found in Col-0, could be good candidates for the maintenance of CWI in salt tolerance. Expansin is able to promote cell wall expansion under a low apoplastic pH (Cosgrove, 2000). AtEXLA1 and AtEXLB1 overaccumulated in both Ws and Col-0, but the abundance was higher in Ws than in Col-0 at T48. RALFs induced a rapid alkalinization of the apoplast upon salt stress (Zhao et al., 2019). Thus, the overaccumulation of expansins under salt stress would not favor cell wall expansion but rather participate to the salt stress signaling pathway. Several studies showed that overexpression of an expansin gene in Arabidopsis was shown to confer enhanced salt tolerance by lowering ROS accumulation and increasing antioxidant activity (Abuqamar et al., 2013;Lü et al., 2013;Chen et al., 2019;Jadamba et al., 2020). Thus, the higher abundance of FLAs associated with a higher overaccumulation of expansins observed in Ws could contribute to a better CWI in response to salt stress than Col-0, where only one FLA and WAK1 were overaccumulated. Most of these experiments showed a transient change at T72 in both ecotypes. The T72 kinetic point more than likely corresponds to a critical metabolism change point, which could determine subsequent salt adaptation or cell death.

CONCLUSION
In conclusion, our results showed that salt stress alters several metabolic pathways in Arabidopsis. Both ecotypes adopt a similar global strategy to cope with salt stress. However, Ws is less saltsensitive than Col-0, and the degree of salt tolerance could be rather related to the stage of development than to their respective genetic background.

AUTHOR CONTRIBUTIONS
CR, KP, and ML designed the experiments. ML performed most of the experiments. SB helped for hydroponics culture. MI performed anthocyanin content and helped to determine antioxidant enzymes activities. AQ carried out the metabolites and ions analysis with the help of ML. CR and RR analyzed cell wall sugar composition. PM, KP, and ML performed proteomics analysis. EJ and HS analyzed and organized the proteomics data. ML, MI, EJ, and CR wrote the manuscript. All authors read and approved the manuscript.