Two potato cultivars, salt-tolerant (Min 08085008, M5008) and salt-sensitive (Zhongshu D516, D516), were used in this study. Prior to this study, these two culti-vars were screening from 18 candidate cultivars (including Long 201205-6, Zhongshu D516, Favorita, Guinongshu No.1, Zhongshu N88, A7, Yunshu 506, Zihua 851, Xingjia No.2, Zhongshu No.3, Huasong 66, Minshu No.1, Heijingang, 08CA0687, Min 08085008, Black Rose No.1, and Zhongshu 157) based on our results of saline test in Fuqing, Fuzhou, Fujian, China (25.55° N, 119.50° E, elevation 24 m) and 16 different saline regions test (4 locations in Fuqing, Fuzhou, Fujian, 3 locations in Changle dis-trict, Fuzhou, Fujian [25.88° N, 119.60° E], 5 locations in Puxia County, Ningde, Fujian [26.91° N, 120.22° E], 3 locations in Xiuyu district, Putian, Fujian [25.31° N, 119.22° E], and 1 location in Research Center, Fujian Agriculture and Forestry University [26.05° N, 119.13° E]) in 2015. Two potato cultivars were selected: a salt-tolerant variety (M5008) and a salt-sensitive variety (D516). Tubers of both cultivars were planted in a substrate mixture (vermiculite:perlite:nutrient soil = 1:1:1) and grown for one month in a greenhouse under a 16/8-hour light/dark cycle at 25°C. Seedlings of uniform growth (15–20 cm in height) were transplanted into 6-liter black containers filled with 5 liters of 1/20× Murashige & Skoog (MS) nutrient solution. The nutrient solution was renewed every two days. After seven days, morphologically similar seedlings were selected and transferred to a hydroponic system using perforated foam boards in containers, with water replenished every 24 hours. For each cultivar, uniformly grown seedlings were divided into two groups: the salt-treated group received a nutrient solution supplemented with 150 mmol/L NaCl, while the control group was maintained in 1/20× MS nutrient solution. Each treatment included three independent biological replicates, with each replicate consisting of a pooled sample of roots from six individual seedlings.Root samples were collected at 0, 12 hours post-treatment, immediately frozen in liquid nitrogen, and stored at -80°C for subsequent proteomic and transcriptomic analyses. In addition, fresh potato roots (0.2 g each) were chopped and placed into pre-cooled mortars. Then, 2 ml of 0.05 M PBS (pH 7.0, containing 1% PVP) buffer was added, and the tissues were ground under icebath conditions. The homogenate was transferred into a 10 ml centrifuge tube, and the mortar was rinsed twice with 3 ml of PBS buffer. The combined solution was centrifuged at 4°C and 4000 r·min⁻¹ for 15 min. The resulting supernatant was stored at 4°C for subsequent assays of superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) activities. SOD activity was determined by colorimetric method, POD activity by the guaiacol method, CAT activity by ultraviolet absorption method, and malondialdehyde (MDA) content by thiobarbituric acid (TBA) colorimetric assay.

Protein extraction was conducted using trichloroacetic acid/acetone precipitation with phenol extraction methods as described in a previous study (Wu et al., 2014). The roots (3.0 g) of potato seedlings treated with 0.15 M NaCl or without treatment for 12 h were collected and pulverized in mortar with liquid nitrogen and appropriate amount poly(vinylpolypyrrolidone) (PVPP). After adding 6 mL of protein extraction buffer (pH 8.0, containing 50 mM Tris-HCl, 25 mM EDTA, 0.5 M thiourea, and 0.5% (w/v) β-mercaptoethanol), the samples were transferred to a 50-mL tube, vortexed thoroughly, and let stand for 1 h at 4°C, followed by centrifugation at 15,000 g for 15 min at 4°C. The supernatant was removed to a new tube and precipitated by added 7–8 volumes of cooled acetone (pH 8.0, containing 0.07% (w/v) β-mercaptoethanol) and incubated at -20°C overnight. The precipitate was collected by centrifugation at 15,000 g for 8 min at 4°C, and washed twice with 10 mL of acetone, followed by dissolved in 5 mL of 0.1 M Tris-HCl buffer (pH 8.0, containing 100 mM Tris-HCl, 50 mM EDTA, and 2% (w/v) β-mercaptoethanol) for 10 min at 4°C, and extracted by adding 5 mL of Tris-phenol for 15 min under continuous shaking at room temperature. The phenol phase was separated by centrifugation at 15,000 g for 10 min at 4°C, and precipitated by 7–8 volumes of a 0.1 M ammonium acetate/methanol solution, incubated at -20°C overnight. The pellets were collected by centrifugation at 15,000 g for 10 min at 4°C, and then washed 2–3 times by using a 10 mL 0.1 M ammonium acetate/methanol solution, and again 2–3 times by adding 10 mL of acetone. The resulting pellets were dried under vacuum, dissolved in lysis buffer (containing 7 M urea, 2 M thiourea, 4% (w/v) CHAPS, and 40 mM DL-dithiothreitol (DTT)), Protein concentration was measured accurately using a 2D Quant Kit (GE, USA) and was adjusted to 1 μg·μL^-1 with lysis buffer. Then, 100 μL of the protein sample was transferred to a new tube, 400 μL of methyl alcohol was added, and the mixture was carefully vortexed. Afterward, 100 μL chloroform and 300 μL Milli-Q water were added and mixed with a vortex. The sample was centrifuged for 10 min at 20,000 g and 4°C, and the top organic phase was removed carefully. Next, 300 μL methyl alcohol was added and mixed carefully by vortexing. Proteins were finally precipitated by centrifugation (10 min, 20,000 g at 4°C), and the precipitate was vacuum-dried. The protein pellet was dissolved in 20 μL of 50 mM NH₄HCO₃ at room temperature and subsequently reduced in 50 mM DTT for 30 min at 56°C. After cooling to room temperature, the sample was alkylated by incubating in 50 mM iodoacetamide for 30 min in the dark. After adding 50 μL 100 mM NH₄HCO₃, 10 μL 0.1μg·μL^-1 trypsin (Promega, USA) was added to the protein sample. The protein sample was digested by incubation at 37°C on a rocking shaker for 16 h in the dark, and 10 μL 20% formic acid was added to acidize the sample. Sample desalination was performed by Sep-Pak C18 columns (Waters, USA), and the samples were washed with 300 μL Milli-Q water and eluted in 50 μL methyl alcohol. The elution was dried under vacuum and stored at -80°C for further use. iTRAQ labelling of peptides was performed using an iTRAQ 4-plex kit (Applied Biosystems, USA) according to the manufacturer’s instructions. After the peptides were labelled with the isobaric tags, they were incubated at room temperature for 1 h. The peptide mixtures of different samples were subsequently pooled, dried by vacuum centrifugation and stored at -80°C until LC-MS/MS analysis were performed, and then stored at -80°C. A total of 100 µg of proteins were reduced, alkylated, trypsin digested, and labeled with iTRAQ regents (Applied Biosystems, USA) according to the manufacturer’s instructions.

The LC-MS/MS and data analysis of the peptides were performed on a TripleTOF 5600 MS (AB SCIEX, Concord, ON, CA, USA) according to the method as described by Zhong et al. (2014). Briefly, the dried peptides were dissolved in 2% ACN containing 0.1% FA and analyzed. Through a Nanospray III source (AB SCIEX), a splitless Ultra 2D Plus (Eksigent, Dublin, CA, USA) system was coupled to the TripleTOF 5600. HPLC gradients were delivered at 300 nL/min, and 60-min gradient was employed for each peptide fraction. Peptides were separated on a fused silica capillary emitter (New Objective, Woburn, MA, USA) packed in-house with 5 m C18 resin, and analyzed in the positive ion mode by ESI (spray voltage =2.4 kV). Survey scans were acquired in 250 ms and 20 product ion scans were collected in 50 ms per scan. For obtaining better iTRAQ reporter ion signal, adjust collision energy (adjust CE) was selected. The data have been submitted to the iProX platform and are available for sharing through this repository. The PXD accession number is PXD068921.

The overall experimental strategy for proteomic analysis is illustrated in Figure 1. To identify and quantify the protein in ‘D516’ and ‘M5008’ cultivars, the results of MS/MS data were searched against the Solanum tuberosum from the Uniport database (Bairoch and Apweiler, 2000) using MaxQuant (Tyanova et al., 2016). The specified cleavage enzyme and false discovery rate (FDR) were set as trypsin/P and 0.01, respectively. Other parameters, such as mass error for precursor ions and fragment ions, fixed modifications, variable modifications, and minimum peptide length, were set as default values. Proteins with at least one unique peptide were retained for further quantification and analysis. iTRAQ-plex was selected for the ‘D516’ and ‘M5008’ samples as the quantification method.

In this study, all identified proteins were blasted into known public databases, such as COG, GO, KEGG, KOG, Pfam, Swissport, TrEMBL, and NR (Ashburner et al., 2000; Bairoch and Apweiler, 2000; Tatusov et al., 2000; Kanehisa et al., 2002; Pruitt et al., 2007; El-Gebali et al., 2019) to obtain the detail function and annotation for further analysis. Differentially expressed proteins were identified using a combination fold change (≥ 1.5 or ≤ 0.66) and P values (< 0.05) calculated using the DESeq R package (Anders and Huber, 2010). GO enrichment analysis and KEGG pathway enrichment analysis were performed using the bingo of Cytoscape (Maere et al., 2005) and cluster profiler (Yu et al., 2012). The interaction network of proteins was obtained by the STRING database (http://string-db.org/), and then the functional cluster of the network was identified by the MCODE (Bader and Hogue, 2003) of Cytoscape.

Total RNA was extracted from potato roots treated with either 150 mM NaCl or water (control) for 12 hours using the Tiangen RNA extraction regent kit according to the manufacturer’s instructions. cDNA synthesis was performed using the PrimerScript™ RT Reagent Kit with gDNA Eraser (Takara, Shiga, Japan). qRT-PCR was performed using the SYBER premix ExTaq™ Kit (Takara, Shiga, Japan) and the 7500 Real-Time PCR system (Applied Biosystems). Gene-specific primers (listed in Supplementary Table S5) and SYBR Premix EX Taq. Actin was applied as an internal reference gene. It was used as the reference gene. Three replicates per treatment were performed.

As expected, the ‘M5008’ cultivar exhibited a higher salt resistance trait (Figure 2). Compared to the control, the leaves of ‘D516’ cultivar turned yellow while the leaves of ‘M5008’ cultivar had a stay-green phenotype at 12 h post salinity treatment.

As illustrated in Figure 3, the two potato varieties exhibited fundamentally distinct physiological responses to salt stress across all measured parameters. The salt-tolerant variety M5008 demonstrated a coordinated and robust antioxidant defense system, whereas the salt-sensitive variety D516 displayed progressive dysfunction in its oxidative stress management mechanisms.

Superoxide dismutase (SOD) activity revealed marked varietal differences (Figure 3A). M5008 showed significantly enhanced SOD activity during the 4–20 h stress period (p < 0.001), with a peak increase of 33.7% at 8 h (465.4 U in TR vs. 348.1 U in CK), indicating efficient scavenging of superoxide radicals. In contrast, D516 exhibited significant suppression of SOD activity between 8 and 20 h (p < 0.05–0.01), most pronounced at 20 h, where activity decreased by 12.0% (304.0 U in TR vs. 345.6 U in CK).

Analysis of peroxidase (POD) activity (Figure 3B) further highlighted inter-varietal differences in hydrogen peroxide detoxification capacity. M5008 displayed significantly elevated POD activity at 8–12 h (p < 0.05), particularly at 12 h (2367.1 U in TR vs. 1771.7 U in CK), supporting effective H₂O₂ clearance during critical phases of stress. In contrast, D516 exhibited inconsistent POD responses accompanied by high intra-group variability, most evident at 8 h, where TR values ranged from 1312.5 to 2877.7 U compared to the stable D516-CK value of 2784.4 U, suggesting inadequate regulatory control over this enzymatic pathway.

Catalase (CAT) activity profiles (Figure 3C) revealed the most pronounced divergence between varieties. M5008 maintained relatively stable CAT activity throughout the stress period, with transient upregulation during early stress (4–8 h), reflecting functional resilience. Conversely, D516 exhibited severe and sustained suppression of CAT activity across all time points (p < 0.001), with enzyme levels declining to only 14–18% of CK values at 16–20 h—indicative of a near-complete breakdown of this key antioxidant mechanism.

Measurements of malondialdehyde (MDA) content (Figure 3D) provided direct evidence of differential membrane integrity under salt stress. M5008 experienced only transient lipid peroxidation, with rapid recovery following an initial spike at 8 h (5.56 μmol/g FW in TR vs. 4.10 μmol/g FW in CK; p < 0.001). In contrast, D516 suffered extensive and persistent membrane damage, as reflected by a 45.9% increase in MDA at 8 h (7.12 μmol/g FW in TR vs. 4.10 μmol/g FW in CK; p < 0.001) and significantly elevated levels from 8 to 24 h (p < 0.01–0.001).

A comprehensive evaluation of these physiological parameters indicates that salt tolerance is primarily determined by the synergistic functionality of the antioxidant network rather than the performance of individual enzymes. In M5008, balanced coordination among SOD, POD, and CAT enables effective mitigation of reactive oxygen species (ROS) accumulation and preservation of cellular membrane integrity. In contrast, D516 exhibits a progressive failure of antioxidant defenses—initiated by impaired SOD activity, exacerbated by erratic POD regulation, and culminating in the collapse of CAT function—resulting in uncontrolled oxidative stress and irreversible membrane deterioration.

These physiological findings are consistent with and functionally validate the proteomic data presented in the main manuscript. The constitutively prepared cellular state and stable expression of ROS-scavenging enzymes in M5008 align with its robust and well-coordinated antioxidant response, enabling effective containment of oxidative damage. Conversely, D516’s high-amplitude but dysregulated induction of antioxidant enzymes proves insufficient to counteract progressive oxidative injury, underscoring fundamental differences in antioxidant strategy between salt-sensitive and salt-tolerant potato cultivars.

Therefore, the distinct salt tolerance between the two cultivars renders them ideal models for exploring the underlying mechanisms of potato salt tolerance. Then, three independent samples of D516 and M5008 were collected for protein sequencing using comparative iTRAQ analysis (Table 1), and a total of 4505 proteins were identified (Supplementary Table S1). These identified proteins were blasted into different databases to obtain the annotation. As indicated, most proteins (4498) had homologous proteins in the NR proteins database. Furthermore, 4125 (91%) proteins had significant matches in the GO database, and 2015 (45%) proteins had matches in the KEGG database.

The results of the PCA (Figure 4) analysis indicate that the transcriptomic profiles of potato root samples are strongly influenced by both experimental treatment and cultivar-specific responses.

The tight clustering of the three biological replicates within each group (e.g., T-TR-1, 2, 3; S-TR-1, 2, 3) reflects high experimental reproducibility and data reliability. This consistent grouping validates the quality of the RNA-seq data and supports the robustness of the observed gene expression patterns.

The most prominent pattern is the clear separation of samples along the PC1 axis, which accounts for 28.68% of the total variance and primarily corresponds to the treatment effect (salt stress vs. control).

Samples on the right side of PC1 (positive values) include all control groups (CK), regardless of cultivar (S-CK and T-CK), which cluster closely together. This indicates that under non-stress conditions, the global gene expression profiles of the two cultivars are highly similar, with minimal baseline transcriptional divergence.

Conversely, all salt-stressed samples (TR), from both cultivars, are positioned on the left side of PC1 (negative values). This pronounced shift demonstrates that salt stress induces a substantial and genome-wide reprogramming of the transcriptome in potato roots. The magnitude of this change underscores that the imposition of salt stress is the primary source of variation in the dataset, overshadowing inherent genetic differences between the cultivars.

While treatment is the dominant factor, the second principal component (PC2), explaining 18.76% of the variance, captures key genotypic differences in the transcriptional response to stress.

The salt-sensitive ‘D516’ under stress (S-TR) is located in the bottom-left quadrant, with its three replicates forming a distinct cluster. In contrast, the salt-tolerant ‘M5008’ under stress (T-TR) occupies the top-left quadrant. The separation between the S-TR and T-TR groups along PC2 provides clear evidence that the two cultivars employ fundamentally different molecular strategies in response to the same stress stimulus. This differential regulation likely contributes to the superior physiological performance of ‘M5008’.

Notably, the T-TR group exhibits a more compact cluster compared to S-TR along both PC1 and PC2 axes. This tighter aggregation suggests that the transcriptional response in the tolerant cultivar is more consistent, coordinated, and tightly regulated. In contrast, the relatively greater dispersion of the S-TR replicates may reflect a more variable, less synchronized, or potentially dysregulated response in the sensitive cultivar, which could underlie its reduced capacity to cope with salt stress.

Overall, the PCA plot reveals a two-tiered transcriptional response to salt stress in potato:Tier 1 – A Shared Stress Response (PC1): Both cultivars exhibit a strong, common transcriptional reaction to salt stress, clearly distinguishing stressed samples from controls. This represents a core stress-responsive program activated across genotypes.Tier 2 – Genotype-Specific Regulatory Divergence (PC2): Within this shared response framework, the two cultivars diverge significantly in their transcriptional reprogramming. The tolerant ‘M5008’ activates a distinct and well-coordinated set of genes—or modulates their expression to differing extents—constituting an effective adaptive mechanism that is either absent or inadequately executed in the sensitive ‘D516’.

To identify the DAPs between two potato cultivars, the detected proteins were further analyzed. A total of 397 DAPs were identified, based on combination fold change (≥ 1.5 or ≤ 0.66) and P values (< 0.05), including 288 proteins that were up-regulated and 109 proteins that were down-regulated significantly in the ‘M5008’ cultivar compared to the ‘D516’ cultivar (Figure 5A).

The functions of all 397 DAPs were classified into three different categories using gene ontology (GO) classification, with 14 terms in the cellular component categories, 16 terms in the molecular function categories, and 19 terms in the biological process category (Figure 5B). In cellular component categories, cell part (58 DAPs), cell (58 DAPs), organelle (33 DAPs), membrane (17 DAPs), and membrane part (10 DAPs) were the top six terms. Catalytic activity (79 DAPs), binding (46 DAPs), and biological regulation (18 DAPs) were the top three terms in molecular function categories. Metabolic process (88 DAPs), cellular process (65 DAPs), response to stimulus (23 DAPs), and cellular component organization or biogenesis (13 DAPs) were the top four terms in biological process categories (Figure 5B).

GO annotation identified that 60 DAPs associated with oxidation reduction and 22 DAPs associated with glucose metabolic processes in the ‘M5008’ cultivar were significantly higher than those in the ‘D516’ cultivar (Figure 5C). Among these DAPs, Q6R608 (annotated as 4-alpha-glucanotransferase) and Q9FEN7 (annotated as dihydrolipoyl dehydrogenase) proteins in M5008 were 5.9-fold and 4.3-fold compared to D516, respectively (Supplementary Table S2). However, six DAPs associated with negative regulation of catalytic activity in the ‘M5008’ cultivar were remarkably lower than that in the ‘D516’ cultivar (Figure 5D). The expression of P20347 protein (annotated as cysteine protease inhibitor 1) was 2.9-fold compared to the ‘D516’ cultivar (Supplementary Table S2). These results suggested that metabolic differentiation is one of the reasons for salt-tolerance differentiation between ‘M5008’ and ‘D516’ cultivars.

In addition, all DAPs between two potato cultivars were further grouped using KEGG pathway annotation and enrichment analysis (Figure 5E). The result showed that nine pathways were remarkably enriched with DAPs, including phenylpropanoid biosynthesis, glycolysis/gluconeogenesis, stilbenoid, diarylheptanoid, and gingerol biosynthesis, carbon fixation in photosynthetic organisms, flavonoid biosynthesis, carbon metabolism, glutathione metabolism, phenylalanine metabolism, and terpenoid backbone biosynthesis. For example, K7WNY2 in the ‘M5008’ cultivar, coding glyceraldehyde-3-phosphate dehydrogenase involved in plant development, provided substrates for the phosphorylated pathway of serine biosynthesis (Munoz-Bertomeu et al., 2009) and had a higher expression than that in the ‘D516’ cultivars.

To determine DAPs in these two potato cultivars response to salt stress, three independent samples of ‘D516’ and ‘M5008’ cultivars with salt-treated (0.15 M NaCl for 12 h) and mock-treated were collected for protein sequencing using comparative iTRAQ analysis. For the ‘D516’ cultivar, a total of 511 DAPs, including 249 proteins with increased expression and 262 proteins with reduced expression, were defined in response to salt treatment. While, for the ‘M5008’ cultivar, a total of 456 DAPs, including 95 proteins with increased expression and 361 proteins with reduced expression, were identified after being exposed to salt conditions (Figure 5A; Supplementary Table S3).

To deduce the biological processes and mechanisms associated with salt stress in two potato cultivars, GO enrichment and KEGG pathway analysis were further performed. During abiotic stress conditions, hydrogen peroxide is produced and plays a crucial role in acclimating plants to the stress condition (Choudhury et al., 2017). According to the GO analysis result, the signal pathway associated with hydrogen peroxide had been activated in the ‘D516’ cultivar (Figure 5B). For example, the expression of M0ZNG4, M1CCK1, and M1C911, which are annotated as peroxidase, were upregulated to 6.2, 10, and 10.6 times in the ‘D516’ cultivars after being exposed to salt conditions (Figures 5C, D; Supplementary Table S3), while these proteins in the ‘M5008’ cultivar only showed a slight difference. These results suggested that after being exposed to salt conditions for 12 h, excess ROS was accumulated in the ‘D516’ cultivar, which might contribute to oxidative stress damaging plant growth and development. For the ‘M5008’ cultivar, perhaps redox homeostasis was maintained by the enzymatic components and the nonenzymatic compounds, including glutathione metabolism (Figure 5B; Supplementary Table S3) and vitamin (Figure 5C; Supplementary Table S3).

Sugars and proteins have multiple roles under salt stress, such as structural components, osmolytes, signaling molecules, and supplements of energy (Massa and Melito, 2019). The GO analysis results showed that many DAPs in D516 were enriched in response to sugar signaling, including response to fructose, sucrose, hexose, monosaccharide, glucose, and disaccharide stimulus (Figures 5B, D; Supplementary Table S3), suggesting that sugar stimulus activated downstream target gene expression to acclimatize salt stress. However, for M5008, the DAPs were enriched into sugar metabolic and catabolic processes and energy metabolic processes, such as glucose catabolic and metabolic processes, hexose biosynthetic and metabolic processes, and NADP metabolic processes (Figures 5B, D; Supplementary Table S3). These processes might provide energy and components to acclimate plants to salt conditions, such as maintaining redox homeostasis. Besides, soluble sugar could also be produced through these processes to maintain cellular osmotic pressure. According to the KEGG pathway analysis, protein biosynthesis and metabolism were found in both M5008 and D516, including glycine, serine, threonine, proline, tryptophan, glutathione, and lysine (Figures 5C, D; Supplementary Table S3). These amino acid metabolisms may provide the components and soluble protein content to protect the tolerant cells.

Auxin is required to mediate plants responses to salt stress (Ribba et al., 2020). Many genes involved in auxin biosynthesis were induced by salt treatment, and auxin signaling also functions in tolerance to salinity stress (Ribba et al., 2020; Verma et al., 2022). GO analysis of M5008 showed that many DAPs were enriched into indolalkylamine biosynthetic processes and auxin-mediated signaling pathways (Figure 5C; Supplementary Table S3), suggesting that auxin biosynthesis and signaling pathways might be a strategy for the ‘M5008’ cultivar to enhance tolerance to salinity stress.

To further explore the potential and functional modules under salt conditions in potato, all different expressed proteins induced by salt treatment, comprised by 511 and 456 in both ‘D516’ and ‘M5008’ cultivars (Figure 6A), were superimposed into protein-protein interaction (PPI) network from the STRING and then identified functional cluster using the MCODE (Bader and Hogue, 2003) of Cytoscape. In total, the largest six distinct clusters with at least 10 nodes were identified and indicated with different colors, in which the proteins with similarly consistent trend after salt treatment in D516 and M5008 were labeled with yellow colors (Figure 6B).

Cluster 1 contained 40 different expressed proteins in response to salt in the ‘D516’ and ‘M5008’ cultivars. Additionally, five of all proteins displayed a similarly expressed pattern in both cultivars. For instance, the protein M1AZE4, coding glucose-6-phosphate isomerase, which was proven to promote the synthesis of starch in leaves (Yu et al., 2000), was down-regulated after salt treatment in ‘D516’ and ‘M5008’ cultivars (Figure 6B; Supplementary Table S4). Cluster 2, composed of 59 proteins, was the largest cluster, including 7 proteins, and exhibited low expression in both cultivars after treatment with salt (Figure 6B; Supplementary Table S4). A striking example is M1CXP4, the glycine cleavage system P protein, which affects cotyledon stage Arabidopsis (Engel et al., 2007). The 10 proteins in ‘D516’ and ‘M5008’ cultivars expressed a similar inclination response to salt stress in cluster 3 with 54 nodes (Figure 6B; Supplementary Table S4). For example, the up-regulated protein M1D2W7 coding nucleoside diphosphate kinase 1 has been reported to play a role in response to reactive oxygen species (ROS) stress (Fukamatsu et al., 2003). The 50, 13, and 26 different expressed proteins were constructed in clusters 4, 5, and 6, respectively, in which 4, 1, and 5 proteins in D516 and M5008 appeared in the expression pattern after treatment (Figure 6B; Supplementary Table S4). Of these 10 similarly expressed proteins, M1AJK1 provided electron transfer to heme oxygenase and cytochrome B5 (Louërat-Oriou et al., 1998) by coding NADPH-cytochrome P450 reductase.

On the other hand, GO enrichment analysis for the six clusters applying the biological process terms was performed. The different specific biological functions were enriched in each cluster, such as glucose metabolic (Cluster 1), translational initiation (Cluster 2), small molecule biosynthetic (Cluster 3), ubiquitin-dependent protein catabolic (Cluster 4), intracellular transport (Cluster 5), and phosphatidylcholine metabolic (Cluster 6) (Figure 6C). It was interesting that Cluster 1 also enriched oxidation reduction, consistent with a previous study that found salt stress could alter the expression of several ROS-scavenging enzymes and ROS-responsive regulatory genes (Rodrigo-Moreno et al., 2013). Taken together, these results suggest that the salt-related pathway in D516 and M5008 might be a concordantly regulated network of multiple proteins using interactions between proteins and functional units at a molecular level.

To determine if the candidate DAPs were associated with their transcriptional alteration, qRT-PCR analysis was performed. The results showed that the gene expression levels of G6PD1, P5CSA, and PIP2A2 in the ‘D516’ cultivar were down-regulated significantly after salt treatment. In the ‘M5008’ cultivar, two genes, TPS1 and HEXO1, were up-regulated and one gene, GAPCP1, was down-regulated in response to salt stress (Figure 7A). Further analysis indicated that the abundance of these six candidate proteins was consistent with their transcription level, indicating a good correlation between the transcription level and the corresponding protein abundance in ‘D516’ and ‘M5008’ cultivars.

Osmotic stress and ionic toxicity are the main primary effects of salt stress (Arif et al., 2020). Under salinity, Na⁺ accumulates promptly in tolerant cell cytosols through the non-selection cation channel (NSCCs) and K⁺ transporters (HKT1), with a consequent activation of the Ca²⁺ channel (Massa and Melito, 2019). To exclude Na⁺ from the cytosol to the extracellular, the cytosol Ca²⁺ activates the SOS signaling pathway, which is indeed one of the crucial mechanisms to counteract the injurious effects of ion toxics (Köster et al., 2019). SOS3 recognizes the salt-induced Ca²⁺ signal and activates and recruits SOS2 to the cytomembrane, causing the stimulation of the cytomembrane Na⁺/H⁺ antiport (SOS1) activity and further excreting the excess Na⁺ out of the cells (Ishitani et al., 2000). In our study, many genes involved in the SOS pathway were remarkably up-regulated in the ‘M5008’ cultivar but down-regulated in the ‘D516’ cultivar (Figure 8D). Additionally, numerous genes involved in the synthesis and metabolism of sugar, such as glucose and hexose (Figures 8B, D), were significantly activated in response to salt stress in the ‘M5008’ cultivar, contributing to the adjustment of osmotic stress and the demand for energy and components used for salt response. These findings suggested that the ‘M5008’ cultivar possessed a higher ability to maintain ion homeostasis and adjust osmotic stress.

Oxidative stress is one of the secondary effects of salt stress (Zulfiqar and Ashraf, 2021). Under salinity conditions, the cytosolic Ca²⁺ activates NADPH oxidase, producing a high level of apoplastic ROS (Kurusu et al., 2015). The ROS influx into the cytoplasm is perceived by the membrane-localized histidine kinase, inducing the activation of the MAPK signaling cascade and further regulating the growth and development of the whole plant system (Massa and Melito, 2019). According to our study, the synthesis and signal pathway of ROS were significantly activated in the ‘D516’ cultivar (Figure 8B). Redox homeostasis is maintained by enzymatic components, such as SOD, CAT, and glutathione-S-transferase, and nonenzymatic molecular compounds, including proline, GSH, and α-tocopherol (Rajput et al., 2021). From our proteomic data, numerous genes in response to oxygen metabolism and genes involved in the oxidoreduction coenzyme metabolic process in the ‘M5008’ cultivar were activated (Figure 7B), maintaining the redox homeostasis in the ‘M5008’ cultivar.

Overall, under salinity conditions, the ‘M5008’ cultivar was endowed with a higher capacity than that in the ‘D516’ cultivar for sensing a saline environment and prompting a reaction to change its physiological status, including ionic equilibrium, osmoregulation, and redox homeostasis maintenance (Figure 7B).

Our physiological results clearly show that ‘M5008’ maintains a more robust and stable antioxidant system (Figure 2). The proteomic data provide a mechanistic basis for this observation: while ‘D516’ exhibited a high-amplitude but transient induction of specific peroxidases (Figures 8B, D), ‘M5008’ displayed constitutive and stress-induced enrichment of a broad suite of redox homeostasis components. This included not only enzymes such as SOD and POD but also pathways such as glutathione metabolism (Figures 5E; 8C). This proactive, multi-component strategy likely prevents the damaging accumulation of reactive oxygen species (ROS), as evidenced by the controlled MDA dynamics in ‘M5008’ (Figure 2D). These findings align with the emerging consensus that salt tolerance is determined not merely by ROS scavenging capacity but by the ability to maintain redox signaling and homeostasis under stress (Choudhury et al., 2017; Kesawat et al., 2023). Furthermore, the significant enrichment of DAPs in central carbon metabolism—particularly glycolysis and gluconeogenesis—and energy-related processes (Figure 5E) in ‘M5008’ suggests a strategic metabolic reprogramming to generate ATP, NADPH, and carbon skeletons essential for osmotic adjustment and the biosynthesis of protective compounds (Zhang et al., 2023).

PPI network analysis revealed that salt tolerance involves highly organized functional modules. The identification of “translational initiation” (Cluster 2) and “ubiquitin-dependent protein catabolic process” (Cluster 4) as key clusters (Figures 6B, C) indicates that ‘M5008’ precisely regulates its proteome. It can selectively translate stress-responsive mRNAs while simultaneously degrading damaged or regulatory proteins via the ubiquitin-proteasome system—a critical mechanism for cellular maintenance under stress, as highlighted in recent studies on plant protein turnover (Wang et al., 2024). Additionally, the enrichment of “auxin-mediated signaling pathways” in ‘M5008’ (Figure 8B) implicates phytohormone regulation in its adaptive strategy. Auxin acts as a central regulator of root architecture and growth under stress, and its role in salt tolerance is supported by evidence showing that auxin signaling mutants often exhibit increased sensitivity to salinity (Ribba et al., 2020., Verma et al., 2022).

The enrichment of “phosphatidylcholine metabolic process” in Cluster 6 (Figure 6C) underscores the importance of membrane lipid remodeling in salt tolerance. Adjustments in phospholipid composition are crucial for maintaining membrane fluidity and functionality under stress and may contribute to the generation of lipid-derived signaling molecules (Hou et al., 2023) The more coordinated response of ‘M5008’, evident from the tighter clustering of its biological replicates in the PCA plot (Figure 4), suggests superior integration of stress sensing and signaling, leading to a consistent and effective proteomic response.

In conclusion, our findings support a model in which the salt tolerance of ‘M5008’ is conferred by the integrated operation of several key mechanisms (Figure 7B):

① Energetic and Metabolic Flexibility: Redirection of carbon flux to fuel defense responses and osmotic adjustment.

② Robust Redox Management: Utilization of a balanced, multi-component antioxidant system to prevent oxidative damage.

③ Proteomic Precision: Coordinated regulation of protein synthesis and degradation to maintain proteome integrity and adaptability.

④ Enhanced Signaling and Infrastructure: Engagement of hormone signaling (e.g., auxin) and membrane remodeling to facilitate adaptive growth and preserve cellular integrity.