Identification and Transcriptome Analysis of Genes Related to Membrane Lipid Regulation in Sweet Sorghum under Salt Stress

Sweet sorghum has strong stress resistance and is considered a promising energy crop. In the present study, the effects of salt on the membrane lipid metabolism of two sweet sorghum inbred lines (salt-tolerant M-81E and salt-sensitive Roma) were analyzed. After treatment with 150 mM NaCl, higher levels of fresh weight and chlorophyll fluorescence, as well as lower levels of malondialdehyde (MDA) were found in salt-tolerant M-81E. Concomitantly, 702 and 1339 differentially expression genes (DEGs) in M-81E and Roma were identified in response to salt stress. We determined that most DEGs were related to glycerophospholipid metabolism, glycerolipid metabolism, and other membrane lipid metabolisms. Under NaCl treatment, the expression of the membrane-associated phospholipase A1 was down-regulated at the transcriptional level, along with an increased content of phosphatidylcholine (PC) in both cultivars. The inhibition of triacylglycerol (TAG) mobilization in M-81E delayed salt-induced leaf senescence. Furthermore, enhanced levels of glycerol-3-phosphate acyltransferase (GPAT) expression contributed to improved salt resistance in M-81E. The results of this study demonstrate membrane the role of lipid regulation in mediating salt-defensive responses in sweet sorghum and expand our understanding of the relationship between changes in membrane lipid content and salt resistance.


Introduction
Salinity is one of the main unfavorable environmental factors that restrict plant growth and has led to a decline in global crop production [1][2][3][4][5][6]. The injurious effect of salinity stress on plant growth takes place in two major ways: osmotic stress and ionic toxicity [7][8][9]. High salt ion concentrations cause disturbances in photosynthesis, cellular metabolism, and nutritional balance [10]. In addition, a large number of studies have shown that salt stress is the causal factor of modifications in lipid bilayer composition and membrane permeability [11][12][13]. Such modifications can directly induce the alteration of membrane fluidity and H + -ATPase activity, influencing the passive influx of potentially toxic ions such as Na + and Cl − [13,14]. As a biological barrier, the cell membrane can protect the cells and organelles against various stresses, which is an important mechanism that produces salt tolerance in plants [15,16].
Plant membrane lipids are divided into three categories: sphingolipids, sterols, and glycerolipids [17]. Sphingolipids have a ceramide backbone, which affects the integrity and permeability of the cell membrane [18]. Sterols mainly have structural and regulatory effects. Changes in membrane sterol composition might be important for protecting membranes

Effects of Salt Stress on Plants
Our results showed that salt stress produced significantly lower fresh weights for both cultivars when compared with non-saline conditions. However, the M-81E cultivar had a higher fresh weight under salt conditions than the Roma cultivar ( Figure 1A). Malondialdehyde (MDA) content was increased by salt stress, but the increase in Roma under salt stress was higher than that in M-81E ( Figure 1B). The maximal quantum yield of Photosystem II (Fv/Fm) for M-81E leaves showed no significant difference between salt stress and control conditions, but in Roma the ratio of Fv/Fm significantly decreased in response to salt stress ( Figure 1C). Under the salt environment, the photochemical quenching coefficient (qP) of all lines was markedly reduced and Roma had lower qP values when compared with M-81E ( Figure 1D).

Identification of Differential Expression Genes (DEGs) Involved in Salt Stress
Under control conditions, we identified 2690 genes differentially expressed in the M-81E group when compared with the Roma group. Of these, 2632 were DEGs in the presence of salt (Figure 2A). In the gene expression profiles of M-81E, 702 DEGs were identified when comparing the control with the salt treatment. Of these genes, 289 were up-regulated ( Figure 2B). In the Roma cultivar, 1339 genes were differentially expressed between control and salt-stressed plants, of which 634 genes were up-regulated under salt stress ( Figure 2C). DEGs related to membrane lipid metabolism were selected for further analysis.

Identification of Differential Expression Genes (DEGs) Involved in Salt Stress
Under control conditions, we identified 2690 genes differentially expressed in the M-81E group when compared with the Roma group. Of these, 2632 were DEGs in the presence of salt ( Figure 2A). In the gene expression profiles of M-81E, 702 DEGs were identified when comparing the control with the salt treatment. Of these genes, 289 were upregulated ( Figure 2B). In the Roma cultivar, 1339 genes were differentially expressed between control and salt-stressed plants, of which 634 genes were up-regulated under salt stress ( Figure 2C). DEGs related to membrane lipid metabolism were selected for further analysis.

Functional Classification by Gene Ontology (GO)
DEGs were evaluated using GO analysis to identify their functional information [36]. These DEGs could be categorized into three main functional classes-biological processes, molecular functions, and cellular components-using GO categories. In M-81E and Roma, 1344 and 2788 transcript sequences, respectively, were assigned to 50 s-level GO categories. These were summarized using the functional classes as follows: 15 cell components, 10 molecular functions, and 25 biological processes. For M-81E and Roma, "nucleus" was the main term of the cell component category. In terms of molecular function, "molecu-

Functional Classification by Gene Ontology (GO)
DEGs were evaluated using GO analysis to identify their functional information [36]. These DEGs could be categorized into three main functional classes-biological processes, molecular functions, and cellular components-using GO categories. In M-81E and Roma, 1344 and 2788 transcript sequences, respectively, were assigned to 50 s-level GO categories. These were summarized using the functional classes as follows: 15 cell components, 10 molecular functions, and 25 biological processes. For M-81E and Roma, "nucleus" was the main term of the cell component category. In terms of molecular function, "molecu-lar_function" and "protein binding" were significantly richer terms. In terms of biological process categories, "biological_process" and "regulation of transcription, DNA-templated" were significantly enriched terms ( Figure 3A,B).

Functional Classification by KEGG
A Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was also performed on the DEGs to explore metabolic pathways in the transcriptome of sweet sorghum [37]. Through the analysis of DEGs, it was found that the KEGG pathway related to membrane lipids was mainly concentrated on the glycerolipid metabolism and glycerophospholipid metabolism pathways, which indicated that salt stress regulated membrane lipid metabolism mainly by affecting glycerolipid metabolism and glycerophospholipid metabolism ( Figure 4).

Functional Classification by KEGG
A Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was also performed on the DEGs to explore metabolic pathways in the transcriptome of sweet sorghum [37]. Through the analysis of DEGs, it was found that the KEGG pathway related to membrane lipids was mainly concentrated on the glycerolipid metabolism and glycerophospholipid metabolism pathways, which indicated that salt stress regulated membrane lipid metabolism mainly by affecting glycerolipid metabolism and glycerophospholipid metabolism ( Figure 4).

DEGs Related to Membrane Lipid Metabolism under Salt Stress
Salt stress can cause membrane lipid changes (including in membrane lipid content and signal lipid activity) and ultimately change the biological characteristics of membrane lipids [38]. Changes in plant membrane lipids cause lipid biosynthesis and metabolic enzyme regulation [39]. In this study, 25 genes were mapped to two pathways related to membrane lipid metabolism, of which 11 were mapped to glycerolipid metabolism while the other 14 were mapped to glycerophospholipid metabolism ( Figure 5, Figure S1). It can be seen in Table 1 that in the glycerol membrane lipid metabolism pathway, M-81E had six DEGs encoding three enzymes and Roma had five genes encoding four enzymes. In the glycerophospholipid metabolism pathway, 6 DEGs in M-81E encoded 5 enzymes and 10 DEGs in Roma encoded 7 enzymes.

DEGs Related to Membrane Lipid Metabolism under Salt Stress
Salt stress can cause membrane lipid changes (including in membrane lipid content and signal lipid activity) and ultimately change the biological characteristics of membrane lipids [38]. Changes in plant membrane lipids cause lipid biosynthesis and metabolic enzyme regulation [39]. In this study, 25 genes were mapped to two pathways related to membrane lipid metabolism, of which 11 were mapped to glycerolipid metabolism while the other 14 were mapped to glycerophospholipid metabolism ( Figures 5 and S1). It can be seen in Table 1 that in the glycerol membrane lipid metabolism pathway, M-81E had six DEGs encoding three enzymes and Roma had five genes encoding four enzymes. In the glycerophospholipid metabolism pathway, 6 DEGs in M-81E encoded 5 enzymes and 10 DEGs in Roma encoded 7 enzymes.

qRT-PCR Validation of DEGs
The analysis of the DEGs and the qRT-PCR verified that the fragments per kb per million (FPKM) method could be used to calculate gene expression levels. To verify the RNA-seq data, 12 DEGs related to membrane lipid regulation were selected for quantitative real-time PCR analysis. These included genes related to glycerolipid regulation and glycerophospholipid regulation. Using qRT-PCR analysis, the expression of these genes was found to show a similar pattern to that of genes produced using FPKM values sequenced under corresponding treatments ( Figure 6). These results indicated that the RNA-seq data were reliable.

Discussion
In the present investigation, the two sweet sorghum inbred lines showed different physiological responses to salt stress. Upon exposure to salt stress, total fresh weight was found to be significantly higher in the salt-tolerant M-81E than in the salt-sensitive Roma, which could help M-81E to perform physio-biochemical processes more efficiently under the effects of water deficit. Lipid peroxidation measured as the amount of MDA is considered to be an indicator of oxidative damage from stress [40,41]. MDA content was increased by NaCl treatments in both genotypes. However, salt-tolerant M-81E had less MDA content than Roma. From these findings, it can be inferred that a higher membrane stability index and water retention capacity might have imparted salt stress tolerance to M-81E.
It can be seen from the KEGG pathway of glycerophospholipid metabolism that phosphatidylcholine (PC) is an important intermediate phospholipid and is the main component of the membrane. Lipid metabolism has been found to contribute significantly to the ability of plants to survive under abiotic stresses [42]. Furthermore, PC levels have been found to significantly increase in callus cultures of the halophyte Spartina patens and the roots of salt-tolerant Plantago cultivars [43,44]. The addition of choline, a key substrate for PC biosynthesis, resulted in enhanced salt resistance in wheat, which was evidence for the positive regulatory role of PC [45]. In our study, the DEGs encoding phospholipase A1 that were responsible for catalyzing the acidolysis of PC were both significantly downregulated in Roma and M-81E. Increased PC accumulation seems to be an indispensable adaptive response of plants to osmotic stress caused by salt damage.

Discussion
In the present investigation, the two sweet sorghum inbred lines showed different physiological responses to salt stress. Upon exposure to salt stress, total fresh weight was found to be significantly higher in the salt-tolerant M-81E than in the salt-sensitive Roma, which could help M-81E to perform physio-biochemical processes more efficiently under the effects of water deficit. Lipid peroxidation measured as the amount of MDA is considered to be an indicator of oxidative damage from stress [40,41]. MDA content was increased by NaCl treatments in both genotypes. However, salt-tolerant M-81E had less MDA content than Roma. From these findings, it can be inferred that a higher membrane stability index and water retention capacity might have imparted salt stress tolerance to M-81E.
It can be seen from the KEGG pathway of glycerophospholipid metabolism that phosphatidylcholine (PC) is an important intermediate phospholipid and is the main component of the membrane. Lipid metabolism has been found to contribute significantly to the ability of plants to survive under abiotic stresses [42]. Furthermore, PC levels have been found to significantly increase in callus cultures of the halophyte Spartina patens and the roots of salt-tolerant Plantago cultivars [43,44]. The addition of choline, a key substrate for PC biosynthesis, resulted in enhanced salt resistance in wheat, which was evidence for the positive regulatory role of PC [45]. In our study, the DEGs encoding phospholipase A1 that were responsible for catalyzing the acidolysis of PC were both significantly downregulated in Roma and M-81E. Increased PC accumulation seems to be an indispensable adaptive response of plants to osmotic stress caused by salt damage.
Glycerolipids are the most abundant lipids in higher plants. They include phospholipids, glycolipids, oils, and extracellular lipids, which are widely involved in different biological processes [46]. The synthesis of glycerolipid catalyzes the lipid acylation of sn-glycerol-3-phosphate (G3P) in glycerol-3-phosphate acyltransferase (GPAT) to form lysophosphatidic acid (LPA), and this is followed by the formation of phosphatidic acid PA under the catalysis of 1-acyl-sn-glycerol-3-phosphate acyltransferase (LPAAT) [47,48]. The PA produced in eukaryotic cells can be dephosphorylated in the endoplasmic reticulum to produce DAG, and then be catalyzed by diacylglycerol acyltransferase (DGAT) to form triacylglycerol (TAG) [49] (Figure 7). The reaction process that occurs in the endoplasmic reticulum through GPAT, LPAAT, and DGAT successively catalyzes the sn-1, sn-2, and sn-3 lipid acylation of G3P to form TAG, also known as the Kennedy pathway [49,50]. In the process of glycerolipid metabolism, SORBI_3003G360700 encoding GPAT was upregulated in M-81E in this study. The overexpression of SsGPAT in Arabidopsis can prevent chlorophyll content from decreasing and help maintain the level of unsaturated fatty acids. At the same time, it reduces the photoinhibition of PSII and PSI, protects the photosynthesis equipment, and maintains the membrane's resistance to salt stress. In contrast, all these factors have been found to be reduced in SsGPAT t-DNA insertion mutants under salt conditions [51]. The expression of GPAT involved in photosynthetic activities possibly explains the higher levels of chlorophyll fluorescence that were obvious in M-81E subjected to salt stress. When a plant is subjected to salt stress, it will induce ion stress and osmotic stress, which will produce ionic, water, and oxidative imbalance signals. These, in turn, will induce membrane lipid regulation through metabolic pathways for glycerophospholipids and glycerol membrane lipids.

Plant Material and Salt Treatment
According to our previous research, M-81E was deemed to be a salt-tolerant inbred line, while Roma was considered to be a salt-sensitive inbred line [59]. Sweet sorghum M-81E and Roma seeds were washed with flowing water for at least 8 h, then planted in plastic basins and irrigated with running water. The germinated seeds were irrigated with a half-strength Hoagland's nutrient solution containing 2500 μM Ca(NO3)2·4H2O, 2500 μM KNO3, 1000 μM MgSO4·7H2O, 500 μM KH2PO4, 23.125 μM H3BO3, 4.57 μM MnCl2·H2O, 0.38 μM ZnSO4·H2O, 0.16 μM CuSO4·5H2O, 0.055 μM H2MoO4·H2O, 10.02 μM FeSO4·7H2O, and 10.005 μM Na2EDTA·2H2O [60]. The seedlings were placed in an artificial intelligence incubator and cultivated at 28 ± 3 °C (day/night), with an illumination intensity of 600 μmol m −2 s −1 and 70% relative humidity (15 h photoperiod). Previous experimental investigation showed that 150 mM NaCl is a suitable concentration [59]. When sweet sorghum had grown to the three-leaf stage, seedlings were treated with a nutrient solution of between 0 and 150 mM NaCl. The NaCl content was increased by 50 mM every 12 h until it reached 150 mM. After 48 h of NaCl treatment, the total fresh weight and chlorophyll fluorescence profiles of each sample were measured immediately. The roots and leaves of three-leaf seedlings were then sampled and used for subsequent analyses.

Determination of MDA and Chlorophyll Fluorescence
The MDA content of the leaves was determined using the thiobarbituric acid reaction described by Heath and Packer [61]. After the leaf was darkened for 20-30 min, fluores- When a plant is subjected to salt stress, it will induce ion stress and osmotic stress, which will produce ionic, water, and oxidative imbalance signals. These, in turn, will induce membrane lipid regulation through metabolic pathways for glycerophospholipids and glycerol membrane lipids.
In higher plants and microalgae, TAG biosynthesis occurs through acyl-CoA-dependent or acyl-CoA-dependent pathways. Alternatively, TAG can be formed through the catalysis of membrane-bound diacylglycerol acyltransferase (PDAT) through an acyl-CoAindependent pathway. In addition to having a strong relationship with the accumulation of lipids in tissues, PDAT also plays a role in lipid metabolism during plant seed germination and leaf senescence [52]. SORBI_3006G221500, which encodes LAPPT, and two DEGs that encode PDAT (SORBI_3010G001700, SORBI_3004G286700), were significantly up-regulated in Roma, and two other DEGs encoding PDAT (SORBI_3010G270700, SORBI_3002G067000) were significantly down-regulated in M-81E. These DEGs expressed enhanced TAG abundance in salt-sensitive Roma leaves but reduced abundance in salt-tolerant M-81E. These findings are consistent with the results of Silva et al. (2003), who found that leaves of the salt-sensitive sweet potato cultivar Xu 32 maintained a higher abundance of TAG under saline conditions [53]. Elevated TAG in vegetative tissues can act as an energy storage reservoir during stressful periods [54,55]. However, a massive accumulation of TAG is frequently observed in the senescing leaves of plants, which reflects the low salt resistance of plants [56,57]. The down-regulation of TAG is considered to be a strategy adopted by salt-tolerant plants to ameliorate the adverse effects of salt stress [58].

Plant Material and Salt Treatment
According to our previous research, M-81E was deemed to be a salt-tolerant inbred line, while Roma was considered to be a salt-sensitive inbred line [59]. Sweet sorghum M-81E and Roma seeds were washed with flowing water for at least 8 h, then planted in plastic basins and irrigated with running water. The germinated seeds were irrigated with a halfstrength Hoagland [60]. The seedlings were placed in an artificial intelligence incubator and cultivated at 28 ± 3 • C (day/night), with an illumination intensity of 600 µmol m −2 s −1 and 70% relative humidity (15 h photoperiod). Previous experimental investigation showed that 150 mM NaCl is a suitable concentration [59]. When sweet sorghum had grown to the three-leaf stage, seedlings were treated with a nutrient solution of between 0 and 150 mM NaCl. The NaCl content was increased by 50 mM every 12 h until it reached 150 mM. After 48 h of NaCl treatment, the total fresh weight and chlorophyll fluorescence profiles of each sample were measured immediately. The roots and leaves of three-leaf seedlings were then sampled and used for subsequent analyses.

Determination of MDA and Chlorophyll Fluorescence
The MDA content of the leaves was determined using the thiobarbituric acid reaction described by Heath and Packer [61]. After the leaf was darkened for 20-30 min, fluorescence levels (Fo and Fm) were measured using an FMS-2 portable modulated fluorometer (Hansatech, King's Lynn, United Kingdom). A qP coefficient was then calculated using the saturation pulse method.

Construction and High-Throughput Sequencing of Transcriptome Libraries
Trizol reagent (Invitrogen, California, USA) was used to extract total RNA following the steps provided by the manufacturer. Total RNA quantity and purity were analyzed using a Bioanalyzer 2100 and an RNA 6000 Nano LabChip Kit (Agilent, Santa Clara, CA, USA). Ribosomal RNA was depleted using approximately 10 ug of total RNA representing a specific adipose type. After purification, poly(A)-or poly(A) + RNA fractions were broken into small pieces. Then the cleaved RNA fragments were reversely transcribed to create the final cDNA library, with an average insertion length of 300 bp (±50 bp) for the paired-end libraries. We used Illumina Hiseq 4000 (Lc-Bio, Hangzhou, China) to perform paired-end sequencing following the vendor's recommendation.

Transcriptome Assembly and Identification
Cutadapt was used to obtain clean reads [62]. The quality of sequenced genomes of sweet sorghum was verified using the FastQC, Bowti2, and TopHat2 methods. We then combined all transcripts from the sweet sorghum samples and used Perl scripts to rebuild an overall transcriptome [63,64]. StringTie and Ballgown were used to assess the expression levels of all transcripts [65,66].

Identification and Functional Annotation of DEGs
StringTie was used to assess mRNA expression levels by calculating FPKM. Differential expression of mRNAs was screened using the R package Ballgown with|log2(fold change)| > 1 and with statistical significance (p-value < 0.05) [64,65]. To obtain highly similar sequences, single-gene sequences were aligned with those in the NR, KEGG, and Swiss-Prot databases [37,67,68]. We used the BLAST tool to obtain COG, and BLASTN to obtain the nucleotide database NT [69]. The KEGG database can analyze gene products. The online KEGG web server was used to distribute KEGG pathways to the assembly sequence.

Quantitative Real-Time PCR Analysis
Real-time fluorescence quantitative PCR was used to verify RNA-seq results. The Beacon Designer software (Palo Alto, California, USA) was used to design the primers of 12 genes (Table S1). RNA samples of M-81E and Roma plants were prepared using a total RNA extraction kit (Huayueyang, Beijing, China) [59]. We used tap root tissue to isolate total RNA and a Nanodrop-ND-1000 spectrophotometer (Wilmington, DE, USA) to quantify RNA. The housekeeping gene β-actin (GenBank ID: X79378) from Streptococcus bicolor was used as an internal standard. We carried out sample preparation according to the manufacturer's instructions. A real-time quantitative PCR instrument (Roche Diagnostics, Meylan, France) was used to perform real-time PCR. The relative expression level of each gene was calculated using the 2 − Ct method [70].

Statistical Analysis
All data were evaluated using one-way ANOVAs in SPSS, version 25.0 (IBM, Armonk, New York, NY, USA). A p-value of <0·05 was taken as statistically significant.

Conclusions
This study demonstrates the detailed changes in the membrane lipid metabolism of two sweet sorghum cultivars when subjected to salt stress. Key genes involved in glycerolipid metabolism or glycerophospholipid metabolism were identified by comparing control and salt-stressed plants. The adjustment of lipid metabolism in response to salt stress could contribute to an increase in salt tolerance, as displayed by the improved expression profile of GPAT, coupled with TAG mobilization performed cooperatively in mediating salt-defensive responses in sweet sorghum leaves. The results of this study provide new insight into the potential role of the membrane lipid regulatory networks underlying the mechanisms of salt tolerance in sweet sorghum.

Conflicts of Interest:
The authors declare no conflict of interest.