RNA Sequencing Reveals That Both Abiotic and Biotic Stress-Responsive Genes Are Induced during Expression of Steroidal Glycoalkaloid in Potato Tuber Subjected to Light Exposure

Steroidal glycoalkaloids (SGAs), which are widely produced by potato, even in other Solanaceae plants, are a class of potentially toxic compounds, but are beneficial to host resistance. However, changes of the other metabolic process along with SGA accumulation are still poorly understood and researched. Based on RNA sequencing (RNA-seq) and bioinformatics analysis, the global gene expression profiles of potato variety Helan 15 (Favorita) was investigated at four-time points during light exposure. The data was further verified by using quantitative Real-time PCR (qRT-PCR). When compared to the control group, 1288, 1592, 1737, and 1870 differentially expressed genes (DEGs) were detected at 6 h, 24 h, 48 h, and 8 d, respectively. The results of both RNAseq and qRT-PCR showed that SGA biosynthetic genes were up-regulated in the potato tuber under light exposure. Functional enrichment analysis revealed that genes related to PS light reaction and Protein degradation were significantly enriched in most time points of light exposure. Additionally, enriched Bins included Receptor kinases, Secondary metabolic process in flavonoids, Abiotic stress, and Biotic stress in the early stage of light exposure, but PS Calvin cycle, RNA regulation of transcription, and UDP glucosyl and glucoronyl transferases in the later stage. Most of the DEGs involved in PS light reaction and Abiotic stress were up-regulated at all four time points, whereas DEGs that participated in biotic stresses were mainly up-regulated at the later stage (48 h and 8 d). Cis-element prediction and co-expression assay were used to confirm the expressional correlation between genes that are responsible for SGA biosynthesis and disease resistance. In conclusion, the expressions of genes involved in PS light reaction, Abiotic stress, and Biotic stress were obviously aroused during the accumulation of SGAs induced by light exposure. Moreover, an increased defense response might contribute to the potato resistance to the infection by phytopathogenic microorganisms.


Introduction
Potato (Solanum tuberosum L.) is one of the most important staple crops for direct and processed consumption in many countries around the world. It serves as major, inexpensive low-fat food source, providing carbohydrates and containing high-quality proteins as well as antioxidative polyphenols, vitamins, and minerals [1]. Potato tubers possess small quantities of naturally occurring steroidal glycoalkaloids (SGAs), which are explained as stress metabolites or phytoalexins for protecting the potato plant from insect pests and herbivores [2]. However, SGAs can be hazardous for human health Genes 2019, 10, 920 3 of 17

Biological Materials and Treatment
The potato cultivar 'Helan 15' (Favorita cultivar, light yellow, and smooth skin, with small and shallow bud eyes) (Figure 1) was used for light exposure, which was obtained from Dingxi Academy of Agricultural Sciences (Gansu, China). Helan 15 was introduced by Crop Detoxification Technology Development Center of Minhe County (Qinghai, China) from the Institute of Plant Protection, Tianjin Academy of Agricultural Sciences (Tianjin, China), and its deposition number is Qingshenshu 2007001 (NCBI BioSample ID: SAMN11865697). The tubers with similar weight (72 ± 5 g) and free of visible damage were placed in a growth cabinet at 25 • C under constant white fluorescent light (3000 Lux) for 6 h, 24 h, 48 h, and 8 d, defined as T1, T2, T3, and T4, respectively. For control samples, the tubers were kept in the dark at the same time points and defined as C1, C2, C3, and C4, respectively ( Figure 1). For each sample, tissues with 1.5 mm thick in the surface of three tubers with the same treatment were harvested, immediately frozen in liquid nitrogen, and then stored at −80 • C. Each treatment included three biological replicates.

Quantitative Real-Time PCR
Quantitative Real-Time PCR (qRT-PCR) was used to further validate twelve DEGs that were involved in enriched 'Bins'. For cDNA synthesis, 500 ng of total RNA was transcribed to cDNA by using the PrimeScript™ RT reagent Kit with gDNA Eraser (Code No. RR047A, TAKARA Biotechnology Co., Ltd., Dalian, China). Primer design was performed by using the online software program Primer 3 (http://primer3.ut.ee/). The primer sequences are shown in Additional file S1: Table  S1. PCR conditions were 95 °C for 15 min., followed by 40 cycles of 95 °C for 10 s, and annealing/extension at 60 °C for 30 s. The melting curve was determined for each sample. Relative gene expression was calculated while using the cycle threshold (Ct)2 −ΔΔCt method, as described by Zuo et al. (2017) and Livak and Schmittgen (2001) [26,33]. Data from qRT-PCR analysis were expressed as means ±SD of three independent replicates.

Potato Tubers with a Difference in Phenotypes
Time-dependent exposure experiments were performed to investigate the change of potato under light. Light exposure followed by phenotypic observation showed that the distinct green color was dependent on the duration of light exposure (Figure 1). A greener color became obvious by the eye when the tubers were exposed to light for 48 h, and when the light exposure reached eight days, the potato skin turned the greenest. Potatoes in the dark had no obvious change in color. Except skin color, no significant difference was found between treated and non-treated tubers. Figure 1. General appearance of potato tubers subjected to light exposure. Tubers were exposed to constant white fluorescent light in a growth cabinet for the time points indicated (6 h, 24 h, 48 h, and 8 d). Additionally, for the control, potatoes were placed in the dark in the same cabinet.

Quality of Data and Differentially Expressed Genes
A total of 24 RNA samples, which were collected from three biological replicates of with or without light exposured tubers at four time points (6 h, 24 h, 48 h, and 8 d) were subjected to RNAseq. The results are shown in Table 1. After filtering the raw sequence reads, more than 40 million clean reads were obtained from each sample. Genome and gene mapping ratio ranged from 74.38% to 80.43% and from 61.32% to 65.58%, respectively (Additional file S2: Table S2), which indicated that the sequences were appropriate for further analysis. From each sample, we detected approximately twenty thousand expressed genes, which indicated that our data could be expected to identify most of the genes expressed under each condition. When compared to samples in the control group at the Figure 1. General appearance of potato tubers subjected to light exposure. Tubers were exposed to constant white fluorescent light in a growth cabinet for the time points indicated (6 h, 24 h, 48 h, and 8 d). Additionally, for the control, potatoes were placed in the dark in the same cabinet.

RNA Extraction, RNA-Seq Library Construction and Sequencing
For all the samples, total RNAs were extracted while using the Plant RNAout kit (160906-50, Tiandz Inc., Beijing, China) according to the manufacturer's instructions. Extracted RNA was quantified and qualified by using a NanoDrop2000 Spectrophotometer (Thermo Scientific, MA, Waltham, United States) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), respectively. Poly(A) mRNA was purified by using oligo(dT) magnetic beads and then digested into short fragments (approximately 200 bp). The RNA-Seq library preparation method were performed according to our previous method that was described by Zuo et al. (2017) [26] and it included three replicates for each treatment. The amplified fragments were sequenced while using an Illumina HiSeq TM 2000. The produced reads were paired-end 2 × 150 bp. All the data pertaining to the present study has been included in the tables/figures of the manuscript and the raw data of RNA sequencing has been deposited at the Sequence Read Archive (SRA) database of NCBI (SRS4823473). The authors are pleased to share the data upon request.

Analysis of Sequencing Data
RNA-Seq bioinformatics analysis was used to analyze the transcriptome raw sequences. The original images were converted into sequence data by base calling to provide the raw reads. Subsequently, dirty reads were removed before data analysis to acquire the clean reads. The clean reads were mapped to the genome sequence (v4.04) of S. tuberosum [27]. Gene expression levels were quantified by using the RSEM v1.2.20 with BAM file alignment by Bowtie2 v2.2.3 (https://doi.org/10.5281/ZENODO.572271) [28] and default RSEM parameters (RNA-Seq by Expectation Maximization) [29]. Gene identification (gene ID), length, log2 ratio, and false discovery rate (FDR) were obtained. An FDR value ≤0.01 and an absolute value of log2 ratio ≥2 were both used as threshold to judge the significant differences in gene expression. After that, the expression data of DEGs at different time points were submitted to software Mapman [30]. Sixteen most enriched 'Bins' with a level equal to '2' from each treatment were extracted. The expressional patterns of DEGs involved in enriched Bins were extracted by gene accession. Heat maps were constructed by using a Multiple Array Viewer software MeV 4.9.0 (http://www.tm4.org/mev.html).

Quantitative Real-Time PCR
Quantitative Real-Time PCR (qRT-PCR) was used to further validate twelve DEGs that were involved in enriched 'Bins'. For cDNA synthesis, 500 ng of total RNA was transcribed to cDNA by using the PrimeScript™ RT reagent Kit with gDNA Eraser (Code No. RR047A, TAKARA Biotechnology Co., Ltd., Dalian, China). Primer design was performed by using the online software program Primer 3 (http://primer3.ut.ee/). The primer sequences are shown in Additional file S1: Table S1. PCR conditions were 95 • C for 15 min., followed by 40 cycles of 95 • C for 10 s, and annealing/extension at 60 • C for 30 s. The melting curve was determined for each sample. Relative gene expression was calculated while using the cycle threshold (Ct)2 −∆∆Ct method, as described by Zuo et al. (2017) and Livak and Schmittgen (2001) [26,33]. Data from qRT-PCR analysis were expressed as means ±SD of three independent replicates.

Potato Tubers with a Difference in Phenotypes
Time-dependent exposure experiments were performed to investigate the change of potato under light. Light exposure followed by phenotypic observation showed that the distinct green color was dependent on the duration of light exposure ( Figure 1). A greener color became obvious by the eye when the tubers were exposed to light for 48 h, and when the light exposure reached eight days, the potato skin turned the greenest. Potatoes in the dark had no obvious change in color. Except skin color, no significant difference was found between treated and non-treated tubers.

Quality of Data and Differentially Expressed Genes
A total of 24 RNA samples, which were collected from three biological replicates of with or without light exposured tubers at four time points (6 h, 24 h, 48 h, and 8 d) were subjected to RNA-seq. The results are shown in Table 1. After filtering the raw sequence reads, more than 40 million clean reads were obtained from each sample. Genome and gene mapping ratio ranged from 74.38% to 80.43% and from 61.32% to 65.58%, respectively (Additional file S2: Table S2), which indicated that the sequences were appropriate for further analysis. From each sample, we detected approximately twenty thousand expressed genes, which indicated that our data could be expected to identify most of the genes expressed under each condition. When compared to samples in the control group at the same time point, 1288, 1592, 1737, and 1870 differentially expressed genes (DEGs) were found from 6 h, 24 h, 48 h, and 8 d, respectively (Figure 2A; FDR ≤ 0.01 and log2 ratio ≥ 2). Approximately 75% of DEGs were up-regulated, which is an indication of metabolic processes that were aroused in potato during light exposure. Overlapping of the down-regulated DEGs among the samples from four time points showed that most of these were expressed in each sample. However, a great number of up-regulated DEGs (393) commonly existed in the four samples. The detailed expression data of all DEGs are presented at Additional file S3: Table S3.

Overview of Mapman Analysis
The most enriched Bins as a level equal to '2' in each sample are displayed in Figure 4. In total, sixteen Bins were enriched at least in one sample. The enriched Bins PS light reaction, PS Calvin cycle, Peroxidase, and Protein Degradation were commonly detected from at least three time points. Enriched Bins Receptor kinases and Cell wall degradation were only found from the first time point of light exposure, and most DEGs showed a down-regulation. Bins Abiotic stress, Biotic stress, and Secondary metabolic process in flavonoids were detected at both 6 h and 24 h. Additionally, RNA

Expression Patterns of the Steroidal Glycoalkaloid (SGA)-Related Genes
A previous study has demonstrated that HMG1 and HMG2 were the primary metabolism genes, while SGT1, SGT2, and SGT3 were the secondary metabolism genes, which were demonstrated to be directly involved in SGA biosynthesis, and the expression of PSS1 and PVS1 were associated with the accumulation of SGAs [21,22]. The overexpression of HMG1 in potato plant increased SGA content in potato leaves [22]. Here, we extracted expression patterns of seven SGA-related genes from the current RNA-seq data ( Figure 3A). Four of these genes were obviously up-regulated, including HMG1 (PGSC0003DMP400024174), SGT1 (PGSC0003DMP400020829), SGT2 (PGSC0003DMP400030574), and SGT3 (PGSC0003DMP400020813). After light exposure, the log 2 ratio values of HMG1, SGT1, SGT2, and SGT3 were recorded as 3.51, 6.43, 2.56, and 7.21 at 48 h, and as 6.02, 7.58, 2.59, and 7.34 at 8 d, respectively. qRT-PCR detection further showed similar trends for each gene between two methods ( Figure 3B). The results presented herein showed that light could induce the expression change of these genes that are involved in the generation of SGAs.  For the qRT-PCR data, the means ± SD for the three replicates are represented. Error bars represent the range of relative expression (qPCR fold change) calculated by 2 −(∆∆Ct±SD) (n = 3). The asterisks indicate a significant difference as compared with the group in the dark at each time point (* p < 0.05 and ** p < 0.01).

DEGs Involved in Abiotic Stress Were Differentially Expressed in Response to Light Exposure
A comprehensive illustration of DEGs that are related to Abiotic stress in potato tuber, which was influenced by light exposure, is presented in Figure 5. In total, 98 DEGs were differentially expressed at least in one time point, 42 of these encoding heat shock protein, seven germin-like protein subfamily, six chaperone protein, universal stress protein and wound-responsive protein, and four abscisic acid receptor, kirola-like protein, and MLP-like protein, etc. We noticed that the upregulated DEGs mainly encoded heat shock protein, chaperone protein, wound-responsive protein, universal stress protein, abscisic acid receptor, and kirola like protein, whereas the down-regulated DEGs encoded germin-like protein and MLP-like protein. These results indicated that a multiple abiotic response in potato tuber had been activated during light exposure, especially in the early

Overview of Mapman Analysis
The most enriched Bins as a level equal to '2' in each sample are displayed in

DEGs Involved in Biotic Stress Were Differentially Expressed in Response to Light Exposure
DEGs expressional patterns that were involved in enriched Bin Biotic stress were investigated. As shown in Figure 6, we identified six representative clusters, comprising a total of 15 DEGs, which were found from potato tuber in the early stage of light exposure (both 6 h and 24 h), most of which were down-regulated (93.3% and 73.3% at 6 h and 24 h, respectively). Very importantly, we discovered that DEGs in the Bins displayed stronger transcription at the later stage (48 h and 8 d). It is especially relevant to note that many of the bins associate with plant disease, including disease resistance protein, late blight resistance protein, pathogenesis-relate (PR) protein, and tobacco mosaic virus (TMV) resistance protein N, and most of these genes (53.3% disease resistance protein, 62.5% late blight resistance protein, 75% PR protein, 66.7% TMV resistance protein) were highly expressed at 8 d. These results suggested that the expression of DEGs that were involved in 'biotic stress' were only up-regulated in the later stage (48 h and 8 d) of light exposure. All of the DEGs that were involved in PS light reaction, PS Calvin cycle, PS photorespiration, and Secondary metabolic process in flavonoids displayed an up-regulation pattern, while DEGs that belonged to Biotic stress, Peroxidase, Receptor kinases, Receptor like kinase, Storage protein, and Cell wall degradation and Simple phenols were mainly down-regulated. For all of the other enriched Bins, most involved DEGs showed an up-regulated pattern. We noticed that expression of DEGs involved in both Biotic stress and Abiotic stress was influenced by light exposure at 6 h and 24 h.

DEGs Involved in Abiotic Stress Were Differentially Expressed in Response to Light Exposure
A comprehensive illustration of DEGs that are related to Abiotic stress in potato tuber, which was influenced by light exposure, is presented in Figure 5. In total, 98 DEGs were differentially expressed at least in one time point, 42 of these encoding heat shock protein, seven germin-like protein subfamily, six chaperone protein, universal stress protein and wound-responsive protein, and four abscisic acid receptor, kirola-like protein, and MLP-like protein, etc. We noticed that the up-regulated DEGs mainly encoded heat shock protein, chaperone protein, wound-responsive protein, universal stress protein, abscisic acid receptor, and kirola like protein, whereas the down-regulated DEGs encoded germin-like protein and MLP-like protein. These results indicated that a multiple abiotic response in potato tuber had been activated during light exposure, especially in the early stage.

DEGs Involved in Biotic Stress Were Differentially Expressed in Response to Light Exposure
DEGs expressional patterns that were involved in enriched Bin Biotic stress were investigated. As shown in Figure 6, we identified six representative clusters, comprising a total of 15 DEGs, which were found from potato tuber in the early stage of light exposure (both 6 h and 24 h), most of which were down-regulated (93.3% and 73.3% at 6 h and 24 h, respectively). Very importantly, we discovered that DEGs in the Bins displayed stronger transcription at the later stage (48 h and 8 d). It is especially relevant to note that many of the bins associate with plant disease, including disease resistance protein, late blight resistance protein, pathogenesis-relate (PR) protein, and tobacco mosaic virus (TMV) resistance protein N, and most of these genes (53.3% disease resistance protein, 62.5% late blight resistance protein, 75% PR protein, 66.7% TMV resistance protein) were highly expressed at 8 d. These results suggested that the expression of DEGs that were involved in 'biotic stress' were only up-regulated in the later stage (48 h and 8 d) of light exposure.

Cis-Elements in the Promoter Region of SGA-Biosynthetic and Disease-Resistant Genes
Light-responsive elements (L) were ubiquitous and relatively abundant on the promoters of SGA-biosynthetic and disease-resistant genes. Furthermore, we also found many hormones responsive elements, such as B (abscisic acid (ABA)), G (gibberellins (GA)), I (Auxin (IAA)), J (methyl jasmonate (MeJA)), and S (salicylic acid (SA)), and adversity (C: low temperature, D: drought, M: defense) responsive elements from several SGA biosynthetic genes and disease-resistant genes (Figure 7). For example, four ABA-responsive elements were detected from HMG1 (PGSC0003DMP400024174), five and eight from gene encoding disease-resistant protein RGA (PGSC0003DMP400051526) and PR (pathogenesis-related) protein (PGSC0003DMP400065212), respectively. Additionally, defense-and stress-responsive elements were discovered from the promoter regions of four SGA biosynthetic genes (HMG1, PSS1, PVS1, and SGT1) and two diseaseresistant genes (PGSC0003DMP400023176 and PGSC0003DMP400007994). The above results indicated that both SGA biosynthetic and disease-resistant genes might co-respond to signals from light, hormone, and stress, and their functions need to be further verified.

Cis-Elements in the Promoter Region of SGA-Biosynthetic and Disease-Resistant Genes
Light-responsive elements (L) were ubiquitous and relatively abundant on the promoters of SGA-biosynthetic and disease-resistant genes. Furthermore, we also found many hormones responsive elements, such as B (abscisic acid (ABA)), G (gibberellins (GA)), I (Auxin (IAA)), J (methyl jasmonate (MeJA)), and S (salicylic acid (SA)), and adversity (C: low temperature, D: drought, M: defense) responsive elements from several SGA biosynthetic genes and disease-resistant genes (Figure 7). For example, four ABA-responsive elements were detected from HMG1 (PGSC0003DMP400024174), five and eight from gene encoding disease-resistant protein RGA (PGSC0003DMP400051526) and PR (pathogenesis-related) protein (PGSC0003DMP400065212), respectively. Additionally, defenseand stress-responsive elements were discovered from the promoter regions of four SGA biosynthetic genes (HMG1, PSS1, PVS1, and SGT1) and two disease-resistant genes (PGSC0003DMP400023176 and PGSC0003DMP400007994). The above results indicated that both SGA biosynthetic and disease-resistant genes might co-respond to signals from light, hormone, and stress, and their functions need to be further verified.

Verification of DEGs Using qRT-PCR
The expression patterns of 12 representative DEGs involved in enriched Bins were confirmed using qRT-PCR assay to verify the reliability of the RNA-seq data ( Figure 8B). Three biotic responsive genes encoding PR protein (pgsc0003dmp400065212), late blight resistance protein (pgsc0003dmp400023176), and disease resistance protein (pgsc0003dmp400004056) were checked by qRT-PCR assay. Additionally, two genes that were involved in flavonoid metabolism (pgsc0003dmp400051588 and pgsc0003dmp400006441), five in abiotic stresses (pgsc0003dmp400055694, pgsc0003dmp400005812, pgsc0003dmp400056275, pgsc0003dmp400036046 and pgsc0003dmp400049458), and two receptor-like kinase genes (pgsc0003dmp400045105 and pgsc0003dmp400040156) were also detected. Similar changes in all detected genes were observed between RNA-seq ( Figure 8A) and qRT-PCR assays ( Figure 8B), as indicated the accuracy of the RNA-seq data.

Verification of DEGs Using qRT-PCR
The expression patterns of 12 representative DEGs involved in enriched Bins were confirmed using qRT-PCR assay to verify the reliability of the RNA-seq data ( Figure 8B). Three biotic responsive genes encoding PR protein (pgsc0003dmp400065212), late blight resistance protein (pgsc0003dmp400023176), and disease resistance protein (pgsc0003dmp400004056) were checked by qRT-PCR assay. Additionally, two genes that were involved in flavonoid metabolism (pgsc0003dmp400051588 and pgsc0003dmp400006441), five in abiotic stresses (pgsc0003dmp400055694, pgsc0003dmp400005812, pgsc0003dmp400056275, pgsc0003dmp400036046 and pgsc0003dmp400049458), and two receptor-like kinase genes (pgsc0003dmp400045105 and pgsc0003dmp400040156) were also detected. Similar changes in all detected genes were observed between RNA-seq ( Figure 8A) and qRT-PCR assays ( Figure 8B), as indicated the accuracy of the RNA-seq data.

Co-Expressional Networks of Genes between SGA and Disease Resistance
WGCNA is designed for constructing co-expression networks from microarray-based expression data and it not only considers the co-expression patterns between two genes, but also the overlap of neighboring genes. While using recently available R-gene database (http://prgdb.crg.eu/wiki), we constructed a gene co-expression network between SGA biosynthesis and disease resistance using WGCNA. Four distinct transcription modules were identified from the transcriptome data (Supplementary Figure S1). Different samples were correlated with four distinct modules, in which module 'blue' (MEblue) and module 'turquoise' (MEturquoise) were highly

Co-Expressional Networks of Genes between SGA and Disease Resistance
WGCNA is designed for constructing co-expression networks from microarray-based expression data and it not only considers the co-expression patterns between two genes, but also the overlap of neighboring genes. While using recently available R-gene database (http://prgdb.crg.eu/wiki), we constructed a gene co-expression network between SGA biosynthesis and disease resistance using WGCNA. Four distinct transcription modules were identified from the transcriptome data (Supplementary Figure S1). Different samples were correlated with four distinct modules, in which module 'blue' (MEblue) and module 'turquoise' (MEturquoise) were highly correlated with the samples (Supplementary Figure S2). After filtering, the MEturquoise module, including five SGAs biosynthetic genes, was used for constructing the co-expression network. As shown in Figure 9, the SGA biosynthesis genes (yellow marked in Figure 9) were correlated with the disease resistance genes (green marked in Figure 9) in different patterns (Figure 9). Some disease-resistant genes, such as IMPA1 (PGSC0003DMG400014989), SPK1B (PGSC0003DMG400006184), and WDR5B (PGSC0003DMG400019361), were co-expressed to multiple SGA biosynthetic genes (HMG1, SGT1, SGT2, and SGT3), while CIPK18 (PGSC0003DMG400020550) and At1g51550 (PGSC0003DMG400026513) were co-expressed to SGT1 and SGT2 (Figure 9 and Additional file S4: Table S4). Above all, we suggest that potato SGA biosynthesis positively regulates its disease resistance.  Figure S2). After filtering, the MEturquoise module, including five SGAs biosynthetic genes, was used for constructing the co-expression network. As shown in Figure 9, the SGA biosynthesis genes (yellow marked in Figure 9) were correlated with the disease resistance genes (green marked in Figure 9) in different patterns ( Figure 9). Some diseaseresistant genes, such as IMPA1 (PGSC0003DMG400014989), SPK1B (PGSC0003DMG400006184), and WDR5B (PGSC0003DMG400019361), were co-expressed to multiple SGA biosynthetic genes (HMG1, SGT1, SGT2, and SGT3), while CIPK18 (PGSC0003DMG400020550) and At1g51550 (PGSC0003DMG400026513) were co-expressed to SGT1 and SGT2 (Figure 9 and Additional file S4: Table S4). Above all, we suggest that potato SGA biosynthesis positively regulates its disease resistance. Figure 9. Co-expressional networks of genes responsible for SGA biosynthesis and disease resistance. In the photograph, the SGAs biosynthesis genes were marked in yellow, and the disease resistance genes were marked in green.

Discussion
In the current study, we investigated the gene expression profiles of the potato tuber between SGAs and biotic stress that is induced by light exposure. We confirmed the significant up-regulation of all seven SGA biosynthesis genes, indicating gene expression induction of SGA along with light exposure (Figures 2 and 3; [34]). Subsequently, we also found the expression of genes correlated with plant disease resistance were activated in the process. Cis-element prediction and co-expression networks further verified the relationship. To our knowledge, this is the first investigation regarding the correlation of gene expression between the stress response and SGA accumulation induced by light exposure.
Light is an important factor for plant development and it has crucial effects on the growth, production, and quality of potatoes [35]. Several authors have demonstrated that light could also increase the SGAs concentration to twice or three times when compared with the initial levels in Figure 9. Co-expressional networks of genes responsible for SGA biosynthesis and disease resistance. In the photograph, the SGAs biosynthesis genes were marked in yellow, and the disease resistance genes were marked in green.

Discussion
In the current study, we investigated the gene expression profiles of the potato tuber between SGAs and biotic stress that is induced by light exposure. We confirmed the significant up-regulation of all seven SGA biosynthesis genes, indicating gene expression induction of SGA along with light exposure (Figures 2 and 3; [34]). Subsequently, we also found the expression of genes correlated with plant disease resistance were activated in the process. Cis-element prediction and co-expression networks further verified the relationship. To our knowledge, this is the first investigation regarding the correlation of gene expression between the stress response and SGA accumulation induced by light exposure.
Light is an important factor for plant development and it has crucial effects on the growth, production, and quality of potatoes [35]. Several authors have demonstrated that light could also increase the SGAs concentration to twice or three times when compared with the initial levels in potato, which occurred either in the field, at harvest, or during storage [3,9,36]. Gull and Isenberg (1960) [37] and De Maine et al. (1998) [38] found that the content of the observed chlorophyll formation associated with greening and of solanine subjected to light exposure developed independently. In our present study, we found that during the light exposure, a green color in tuber peel became obviously by the eye (Figure 1), which was well in accordance with the previous research [3]. Transcriptome changes were compared to identify the genes underlying the light exposure ( Figure 2). When compared with the sample in darkness at the same time point, about 75% of DEGs were up-regulated ( Figure 2C), and the SGA biosynthesis-associated genes, especially SGT1, SGT2, and SGT3 (24 h, 48h, and 8 d, p < 0.01) were up-regulated in the process by the transcript profiling and qRT-PCR (Figure 3). During the up-regulation of photosynthesis and SGA biosynthesis-associated genes, many DEGs that are involved in stresses (biotic and abiotic) response and flavonoid metabolism were discovered (Figure 4). For abiotic stress, most up-regulated DEGs encoded heat shock protein, chaperone protein, wound-responsive protein, universal stress protein, and abscisic acid receptor ( Figure 5). SGAs are toxic compounds to insects, bacteria, and animals, but they have been suggested to have defensive functions for potato [39][40][41][42]. Several mechanisms of SGA toxicity are suggested, such as the disruption of the membrane fluidity and the inhibition of cholinesterase activity [42][43][44]. Although little investigation was involved in the correlation of gene expression between light-induced SGA biosynthesis pathway and potato defense responses, several stress treatments, such as wounding and light exposure, increased potato tuber SGAs level indicated that SGA biosynthetic pathway is probably associated with abiotic defense responses [45], which was consistent with our results ( Figure 5). Multiple abiotic stress-responsive proteins can be induced by one stress factor. For instance, the rapid accumulation of wound response proteins was identified from both light and water stress treated plant tissues [46]. Additionally, heat shock protein contributes to plant tolerance to multiple stress, such as heat, drought, etc. [47]. Although the current investigation suggests a probable relationship that is based on these experiments, further confirmation for the correlation between SGA biosynthesis and abiotic stress responses are needed in the future.
Additionally, most of the DEGs that are involved in biotic stress encoding disease protein, late blight resistance protein, PR protein, and TMV resistance protein N were down-regulated at 6 h and 24 h, but they showed an opposite trend at 48 h and 8 d (Figure 6). For a long time, opinions have existed for regarding relationship between SGA biosynthesis and disease resistance. In many cases, potato SGA accumulation was correlated with its resistance to Fusarium solani var. coeruleum, Fusarium sulphureum [16], Phytophthora infestans [6,7], and Clavibactermic higanensis ssp. Sepedonicus [48]. On the contrary, infection of P. infestans and Rhizoctonia solani showed no obvious influences on the SGA level in potato tuber [49], because these phytopathogenic microorganisms could overcome SGA toxicity and directly resist pathogen infection [42]. However, from our results, the expression of disease-resistant genes was reduced or it did not significantly change in the early stage of light exposure, but it was strongly induced in the later stage, where a common signal-transduction pathway might be activated at the later stage, which leads to PR-protein and other biotic stress protein accumulation. For a resistant wild potato S. arcanum to the early blight, the important roles of SGAs biosynthetic genes on its resistance have been verified by both non-targeted metabolomic and functional genomics analysis [50]. The results indicated that the key genes involved in SGA metabolism were positively regulated for potato disease resistance, which was consistent with our results (Figure 6).
PlantCARE was used to predict the upstream promoter elements of genes (upstream 1500-0 bp of the gene) and analyze the number of cis-elements related to hormones, adversity, and circadian rhythm to further explore the function of the candidate genes [51]. The results showed that, beside elements in response to light, a great number of hormone (ABA, GA, IAA, SA, ect.) and defense responsive elements were also detected ( Figure 7). To our knowledge, the plant hormone plays crucial roles in plant resistant responses against pathogenic bacterial and fungal attacking [52,53]. SA mainly induce plant immune responses to hemibiotrophic pathogens, whereas JA to biotrophic and necrotrophic pathogens [54]. GA usually suppresses plant resistance against to necrotrophic infection, while IAA has a contribution to resistance [55]. Therefore, we suggested that SGAs biosynthesis were closely related to abiotic and biotic stress. Subsequently, the RNA-Seq data were validated via qRT-PCR ( Figure 8). Additionally, WGCNA (weighted gene co-expression network analysis) has become a powerful practice for exploring gene-to-gene relationships and to uncover coordinately expressed gene modules [32,56]. In our study, DEGs were filtered and screened through multiple steps and criteria, and the networks were then constructed while using SGA biosynthetic genes and disease-resistant gene. Interestingly, we found SGA biosynthetic genes were closely related with disease-resistant genes in the context of gene co-expression networks (Figure 9), which can provide a theoretical basis for further research.

Conclusions
In this study, we proposed that stress gene expression, along with SGA biosynthesis, was induced by light exposure. In particular, the up-regulation of disease-resistant genes indicated that the response was correlated with SGA accumulation. The correlation between genes that are responsible for SGA biosynthesis and disease resistance was investigated by both cis-element prediction and co-expressional assay, in which IMPA1 (PGSC0003DMG400014989), SPK1B (PGSC0003DMG400006184), WDR5B (PGSC0003DMG400019361), CIPK18 (PGSC0003DMG400020550), and At1g51550 (PGSC0003DMG400026513) were the candidate genes for further functional study. Besides the toxicity of SGAs to phytopathogenic microorganisms, the aroused expression of disease-resistant genes during light-induced accumulation of SGAs also contributes to potato disease resistance. While our study provides some insight into the relationship between genes of SGAs and plant resistance, more efforts are required to validate and extend our findings. A critical functional experiment would be carried out to verify whether the observed patterns hold true.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/10/11/920/s1. Table S1, primers used for qRT-PCR in the present study. Table S2, summary of sequencing and mapping data from four indicated time points. Table S3, the detail expression data of all DEGs comparing control samples. Table S4, genes of SGAs biosynthesis and disease resistance used for Weighted gene co-expression network analysis (WGCNA). Figure S1, weighted gene co-expression network analysis (WGCNA) of potato. Figure S2, the correlation between modules and different samples.