Two hundred and twenty-three tomato accessions, consisting of 2 Solanum cheesmaniae (wild), 16 S. pimpinellifolium (PIM), 65 S. lycopersicum var. cerasiforme (CER), 105 S. lycopersicum (BIG), and 35 Guangxi accessions ( Supplementary Table S1 ), were collected by our laboratory at Wuhan National Key Laboratory for Germplasm Innovation and Utilization of Horticultural Crops, China, in spring 2018. Four-week-old seedlings were placed in a constant temperature room at 4°C. After 7 days, the symptoms of cold damage on the plants were observed and the images were recorded. Cold damage was graded to five levels: level 0 indicated no symptoms and no sensitivity to the current low temperature; level 1 was characterized by injury spots on the edge of true leaves; level 2 was characterized by large-area injury spots on true leaves; level 3 was characterized by large-area injury spots on true leaves with leaf curling; and level 4 was characterized by wilting of plants and even necrosis of apical meristem (Wang et al., 1996). Three biological replicates were analyzed, and each replicate consisted of 15 individual plants.

Four-week-old seedlings were subjected to cold stress treatments at 4°C for 48 h. Twenty cold-sensitive and 20 cold-tolerant seedlings were selected. After treatment, the third fully expanded leaf from the top of each seedling was collected respectively, frozen in liquid nitrogen, and then stored at −80°C. The leaves were sent to Beijing Novogene Technology Co., Ltd. (Wuhan Branch) for library construction and sequencing. We used the average FPKM (expected number of fragments per kilobase of transcript sequence per million base pairs sequenced) value as a measure of gene expression.

The cold-tolerant A57 (S. lycopersicum cv. Ailsa Craig) was used to study the expression of SUS3. Four-week-old seedlings were subjected to abiotic stress treatments, namely, ABA, ethylene, drought, cold, and salt stress treatments. The methods corresponding to these abiotic stress treatments were described previously (Song et al., 2016). The soil on the surface of the plants was washed off, and the residual water on the surface was absorbed. The plants were divided into five groups for different treatments. Plants in group 1 were placed in a dry environment at room temperature to induce drought stress. The roots of plants in group 2 were immersed in 400 mM of NaCl solution to induce salt stress; plants in group 3 were placed in a 4°C cold chamber to induce cold stress. The plants in group 4 and group 5 were sprayed with 100 μM of ABA and 200 μM of ethylene solutions for hormone induction treatment, respectively. Then, 0, 1, 3, 6, 12, and 24 h after each treatment, the second leaf from the apical meristem was collected, snap-frozen in liquid nitrogen, and stored at −80°C for RNA extraction.

Total RNA was extracted using the TRIzol reagent (Invitrogen, USA) and further reverse-transcribed into cDNA using a HiScript II 1st Strand cDNA Synthesis Kit (Vazyme, China). The relative transcript levels of specific genes were quantified using real-time PCR (qRT-PCR), which was conducted using a QuantStudio™ 6 Flex System (ABI, USA). The expression level of the actin gene (Solyc11g005330) was used as an internal control. For each biological replicate, the relative expression level of SUS3 was quantified using three technical replicates and calculated using the 2^−ΔΔCT method. The related primer sequences can be found in Supplementary Table S2 .

Cold-sensitive M82 (S. lycopersicum) was selected for transgenic tomato plants. For overexpression constructs, the complete open reading frame (ORF) of SUS3 was amplified from M82 using SUS3-OE primers ( Supplementary Table S2 ), and these constructs were cloned into pHELLSGATE8 vectors. An RNA interference (RNAi) construct, a 412-bp fragment from the SUS3 gene, was amplified from cDNA that was prepared from M82 using the SUS3-Ri primers ( Supplementary Table S2 ) and cloned into pHGRV using Clonase BP (Invitrogen, USA). M82 was transformed with all of the recombinant plasmids using Agrobacterium (strain GV3101)-mediated transformation. The cotyledons of tomato were used for transformation. Positive transgenic plants were screened via PCR with a CaMV35S promoter primer and a gene-specific primer. After PCR and qRT-PCR analyses, several independent homozygous transgenic lines expressing abnormal levels of SUS3 (OE1-2, OE2-4, OE3-2, Ri1-1, Ri8-3, and Ri17-2) and WT (M82) were selected for further analysis.

Six-week-old SUS3-OE lines, SUS3-RNAi lines, and WT were placed in a 4°C constant temperature chamber for 7 days for phenotype observation and photographing. Three replicates were set for each treatment, with 15 seedlings per replicate. The third fully expanded leaf from the apical meristem was harvested and snap-frozen in liquid nitrogen and then stored in a −80°C freezer for physiological and biochemical parameters.

Malondialdehyde (MDA), relative electrolyte leakage levels, and proline were determined as previously described (Campos et al., 2003). The anthrone colorimetry method (Fukao et al., 2006) was adopted to determine the content of soluble sugar. Three biological replicates were analyzed for each treatment.

Extraction of enzyme solution: A 0.2-g sample, which had been stored in a −80°C freezer, was weighed and combined with precooled phosphate buffer (pH 7.8, with 0.1 mM of EDTA) and 1.8 mL of 1% PVP. After mixing, the resulting mixture was subjected to centrifugation at 4°C and 12,000 rpm for 15 min. The supernatant obtained was considered to be the enzyme extract. The levels of superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX), and peroxidase (POD) were quantified through determination as described in previous studies (Jimenez et al., 1997; Morohashi, 2002; Al-aghabary et al., 2005).

To investigate the localization of SUS3, the full-length CDS of SUS3 was cloned into the dual vector pGDG between XhoI and ApaI from the cDNA of A57. To fuse the target gene with the reporter gene, we introduced two additional base pairs CT after the XhoI restriction site. The target gene was fused with the C-terminus of a green fluorescent protein (GFP) reporter gene and expressed under the control of the 35S promoter ( Supplementary Table S2 ) for tomato protoplasts. After 18 h of incubation, the image of GFP and red fluorescent protein (RFP) fluorescence was captured by confocal laser scanning microscopy (Leica SP8, Germany).

In 223 natural populations, 147 were not sensitive to cold stress and no damage was observed on their leaves, while 76 showed sensitivity to cold stress and showed varying degrees of damage. Therefore, based on the degree of damage, the impact of cold damage can be graded to different levels ( Figure 1A ). To study the effect of cold stress on whole-genome gene expression patterns, RNA-seq was performed in 20 cold-sensitive and 20 cold-tolerant accessions. There were 475 differentially expressed genes (DEGs), consisting of 176 upregulated genes and 299 downregulated genes ( Figure 1B ). The Gene Ontology (GO) terms of DEGs included heme binding, oxidoreductase activity, tetrapyrrole binding, and iron ion binding ( Figure 1C ). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of DEGs included starch and sucrose metabolism, biosynthesis of nucleotide sugars, amino sugar and nucleotide sugar metabolism, plant–pathogen interaction, phenylpropanoid biosynthesis, and MAPK signaling pathway ( Figure 1D ). We detected the expression of 475 DEGs, and a total of 34 significant expression (padj ≤ 0.01) were identified ( Supplementary Table S3 ). Thirty-four DEGs have a higher expression level by RNA-seq ( Figure 1E ). In order to verify the accuracy of RNA-Seq, 34 DEGs were selected to quantify the expression with M82 (cold-sensitive) and A57 (cold-tolerant) by qRT-PCR ( Supplementary Table S4 ). The findings were largely consistent with RNA-seq. The expression levels of Solyc07g042550 were the highest ( Figure 1F ), which indicates that SUS3 is a major candidate for having an important influence on cold tolerance ( Supplementary Table S3 ).

We analyzed and predicted the structural features and function of SUS3 at the molecular level. The length of the gene is 2,955 bp, which includes an open reading frame (ORF) of 2,418 bp, starting at position 273 with ATG and ending at position 2690 with TAA. It encodes a protein with a length of 806 amino acids and a molecular weight of 92.59 kDa ( Supplementary Figure S1A ). SUS3 had eight helical segments that had 10–70, 100–250, 270–330, 350–450, 480–500, 550–610, 620–660, and 690–780 amino acid sequences and did not contain a signal peptide, and it was a stable non-transmembrane and hydrophilic protein ( Supplementary Figures S1B–D ).

In order to verify the protein function of SUS3, the subcellular localization of tomato protoplast confirmed that the GFP was irregularly dotted around the cells ( Figure 2A ). SUS3 was specifically expressed in the cytoplasm and cell membrane. This result suggests that the SUS3 is mainly located in the cytoplasm and cell membrane. To further investigate the evolutionary relationship of SUS3 and its families, 24 family genes from strawberry (FaSUS1-1~FaSUS1-10, FaSUS2-1~FaSUS2-7, FaSUS3-1~FaSUS3-7), 6 genes from Arabidopsis (AtSUS1~AtSUS6), 13 genes from soybean (GmSUS1-1~GmSUS1-4, GmSUS2-1~GmSUS2-3, GmSUS3-1~GmSUS3-6), 7 genes (OsSUS1~OsSUS7) from rice, 5 genes (ZmSUS1-1~ZmSUS2-1, ZmSUS2-1~ZmSUS2-2, ZSUSm3) from maize, and 22 genes (NtSUS1-1~NtSUS1-13, NtSUS2-1~NtSUS2-4, NtSUS3-3, NtSUS3-5) from tobacco were analyzed. The results showed that the SUS family genes and tobacco SUS genes had high homology, implying that the SUS families and tobacco had similarities in biological functions and growth regulatory mechanisms. Meanwhile, SUS3 has the farthest evolutionary relationship with the other five SUS family genes, and it is speculated that it may exercise different biological functions ( Figure 2B ).

To test whether SUS3 might contribute to responses to other abiotic stresses, we examined the expression pattern of SUS3 in AC plants after subjecting them to ABA, ethylene, drought, cold, and salt stresses ( Figure 3 ). A mild induction of SUS3 expression was detected in response to ABA, ethylene, and salt stresses, and the SUS3 was strongly induced by drought and cold treatment. In the course of 24 h under these treatments, the expression of SUS3 showed an initial increase followed by a decrease. More specifically, the expression level of SUS3 continued to increase significantly within 6–12 h and reached the maximum.

To investigate the biological functions of SUS3, we transformed M82 with the 35S:SUS3 construct. Three independent OE lines (OE1-2, OE2-4, OE3-2) and three RNAi lines (Ri1-1, Ri8-3, Ri17-2) were chosen for further characterization ( Figure 4C ). We tested the cold tolerance of the SUS3 transgenic seedlings. After cold stress for 7 days, the leaves of the wild type showed wilting, and the stems became shriveled and bent and were unable to uphold upright. In contrast, the SUS3-OE lines in their leaves were capable of preserving regular upward growth ( Figure 4A ). However, the phenotype of the SUS3-RNAi lines was similar to that of WT under cold stress and did not demonstrate more wilting ( Figure 4B ). Therefore, it suggests that the loss of SUS3-RNAi lines might not significantly affect cold tolerance.

To evaluate cold tolerance, we analyzed the MDA, relative electrolyte leakage, and proline and soluble sugar contents between transgenic lines and WT plants under cold stress. Under optimal growth conditions, there is no difference between transgenic lines and WT in the above indexes. After 7 days of cold treatment at 4°C, MDA and relative electrolyte leakage increased in SUS3-OE lines and WT, but the levels were significantly lower in the SUS3-OE lines than in WT ( Figures 4D, F ). These results indicate that SUS3 can alleviate the damage to cellular membranes that occurs during cold stress. Under cold stress, the contents of both proline and soluble sugar were significantly higher. Moreover, the increase of sugar in leaves of the SUS3-OE lines was significantly higher than in WT.

Various abiotic stresses often result in the excessive accumulation of reactive oxygen species (ROS) in plants, causing substantial damage to multiple cellular structures (Gill and Tuteja, 2010). Therefore, we assessed the levels of superoxide radicals (O₂ ^−.) in SUS3-OE lines and WT. After treatment at 4°C for 7 days, the accumulation of O₂ ^−. increased in both SUS3-OE lines and WT. However, the accumulation in SUS3-OE lines was significantly lower in comparison to the WT ( Figure 5A ). This indicates that overexpression of SUS3 has the ability to prevent the accumulation of excessive O₂ ^−. free radicals, thereby reducing the damage to plants and enhancing their resistance to cold damage.

In order to determine whether overexpression of SUS3 can reduce the damage of plant cells under cold stress, we measured the activities of four key antioxidant enzymes (SOD, CAT, APX, and POD). The results showed that under optimal growth conditions, the activities of the four enzymes did not show significant differences between SUS3-OE lines and WT. However, after 7 days of cold treatment at 4°C, the activity of SOD and CAT was increased in SUS3-OE lines and WT, but the enzyme activity in SUS3-OE lines was significantly higher than in WT ( Figures 5B, D ). Under cold stress, APX activity in both SUS3-OE lines and WT also increased, but there was no significant difference ( Figure 5C ). The activity of POD increased in both SUS3-OE lines and WT, but the activity increased more significantly in the WT than in SUS3-OE lines ( Figure 5E ). Compared with the WT, overexpression of SUS3 increased the activities of SOD and CAT and decreased the activities of POD under cold stress. There was an accumulation of fewer ROS in SUS3-OE lines, thus reducing the oxidative stress injury. This may lead to less ROS accumulation in SUS3-OE lines, thus reducing oxidative stress injury.