562 J.Xu et aL Plant Physiology and Biochemistry 109 (2016)561-570 proteins were further classified into two categories involved in 2.3.RNA-seg library preparation and illumina sequencing stress defense and photosynthesis (Li et al.,2009).Similarly.Yang et al.(2012)found 40 differentially expressed proteins in bottle Leaves from 6 RG or SG watermelon seedlings were pooled as gourd rootstock-grafted watermelon seedlings compared to the one single biological replicate.The experiment was repeated to self-grafted plants under salt exposure.These studies suggested the obtain three biological replicates.These pooled leaf samples were possibility that rootstocks could mediate gene expression patterns used for the RNA-seg analysis.Total RNA was isolated using TRIzol in scions under stress.However,studies on transcriptomic changes reagent (Invitrogen,Carlsbad,CA,USA)and digested using DNase I in rootstock grafted plants responding to chilling stress are lacking at 37 C for 30 min to remove any possible genomic DNA and are urgently required. contamination.The quality and concentration of each sample were In the present work,the squash rootstock-grafted watermelon determined using an Agilent 2100 Bioanalyzer.Six ug aliquots of seedlings (RG)were found to be more tolerant of chilling stress total RNA from each sample were purified using oligo(dT)magnetic than the self-grafted watermelon seedlings (SG).To explore the bead adsorption.Purified mRNAs were fragmented with endonu- differences in the gene expression between RG and SG watermelon clease and ligated with adaptors to generate libraries with unique seedlings under chilling stress,digital gene expression(DGE)based the 5'and 3'tags.After 15 cycles of linear PCR amplification,105 bp on Illumina sequencing was applied to identify differentially- PCR products were quantified and purified.Denatured molecules expressed genes(DEGs)between RG and SG.The transcriptome were then fixed onto an Illumina Sequencing Chip(Flow Cell)for data presented here provide straight forward information sequencing.RNA-seq libraries were sequenced on an Illumina regarding the molecular state of grafted plants challenged by HiSeqTM 2000 System by BGI-Tech(Shenzhen,China) chilling stress,which is important for understanding the tran- scriptomic changes of grafted watermelon plants in response to 2.4.Functional annotation of differentially-expressed genes (DEGs) chilling. Raw sequence reads were filtered through the lllumina pipeline by BGI Shenzhen,China.Gene expression levels were calculated 2.Methods using the RPKM (Reads Per kb per Million reads)method (Mortazavi et al.2008).Differentially expressed genes (DEG) 2.1.Plant material and chilling stress treatment among the samples were identified using the novel NOlseq method (Tarazona et al.,2011)with a probability>0.8 and an absolute value The watermelon line MW022 was used as the scion and the of the log2 Ratio>1 as the threshold to evaluate the significance squash Jingxinzhen NO.4 (JX)was used as the rootstock.An of gene expression differences.In brief,NOISeg method computes "insertion grafting"procedure described by Lee and Oda (2003) differential expression described as bellow:first,gene expression of was used in this study.Watermelon plants grafted onto their own sample in each group was used to calculate log2 (fold change)M roots were used as controls.Grafted seedlings were grown in a and absolute different value D of all pair conditions(gene expres- growth chamber at 28C/18C(16/8 h)day/night temperatures,a sion value will be substituted by 0.001 if it doesn't express in some relative humidity of 70%,and a photon flux destiny of sample).Second,average expression value of each gene standing 400 umol m-2 s-1.After the full development of the third true for replicates will be used to calculate M and D.Two replicates in leaves,the grafted plants were treated with a low temperature at one of the experimental conditions is sufficient to run the algo- 10 C at the same relative humidity and illumination intensity. Leaves from 6 rootstock-grafted (RG)or self-grafted (SG)water- rithm:Mi log2 and Di =Then,all these M/D values melon seedlings were pooled as one biological replicate at 0 d and are pooled together to generate the noise distribution.If gene i 1 d after the chilling treatment,respectively.Three biological rep- differentially expresses between two groups,we set Gi=1,other- licates were collected.All collected samples were immediately wise set Gi =0,and give a definition for probability of gene i frozen in liquid nitrogen and stored at-70C until use. differentially expressing as following formula P(G=1,)-P(G=1M=m,D=d） 2.2.Assessment of chilling damage index(Cl)and measurement of malonyldialdehyde (MDA)content -P(MI<m,D'<d） Eighteen rootstock-grafted(RG)or self-grafted (SG)watermelon When P is greater than threshold value,its corresponding gene seedlings were moved to a climate chamber at a temperature of is thought to differentially express between groups. 10 C with a 16 h light/8 h dark cycle.After 12 d of the chilling Gene Ontology(GO)was used to analyze biological functions by treatment,the chilling damage index was measured according to mapping all DEGs to GO databases (http://www.geneontology.org/ Yang et al.(2008).The degrees of chilling tolerance were measured )GO terms meeting a threshold of corrected p-value<0.05 were with 6 grades as follows:level 0:no symptom;level 1:chlorosis or defined as significantly enriched.The DEGs were also mapped to crinkled at the edge of old leaves;level 2:chlorosis or crinkled at the Kyoto Encyclopedia of Genes and Genomes (KEGG)database the edge of less functional leaves;level 3:chlorosis or crinkled at (Kanehisa et al.,2008)to identify significantly enriched metabolic the edge of functional leaves with good new leaves;level 4:chlo- pathways or signal transduction pathways. rosis or crinkled and wilting of functional leaves with damaged new leaves;level 5:severe damage of new leaves,plants wilting or 2.5.Quantitative real-time PCR (qRT-PCR)analysis death.The chilling indices(CI)of RG and SG seedlings was tabu- lated according to the following formula:Cl =(each Leave samples used for gRT-PCR were the same as those used for level x number of plants with the corresponding level)/total sequencing.Total RNA from watermelon leaves sampled at 0 d and number of measured plants. 1 d after chilling treatment was extracted using the RNApure Plant The MDA content was measured using 2-thiobarbituric acid as Kit with DNase I (CWBiotech,Beijing.P.R.China).The BU- described by Hodges et al.(1999). Superscript RT Kit was used for first-strand cDNA generation withproteins were further classified into two categories involved in stress defense and photosynthesis (Li et al., 2009). Similarly, Yang et al. (2012) found 40 differentially expressed proteins in bottle gourd rootstock-grafted watermelon seedlings compared to the self-grafted plants under salt exposure. These studies suggested the possibility that rootstocks could mediate gene expression patterns in scions under stress. However, studies on transcriptomic changes in rootstock grafted plants responding to chilling stress are lacking and are urgently required. In the present work, the squash rootstock-grafted watermelon seedlings (RG) were found to be more tolerant of chilling stress than the self-grafted watermelon seedlings (SG). To explore the differences in the gene expression between RG and SG watermelon seedlings under chilling stress, digital gene expression (DGE) based on Illumina sequencing was applied to identify differentially￾expressed genes (DEGs) between RG and SG. The transcriptome data presented here provide straight forward information regarding the molecular state of grafted plants challenged by chilling stress, which is important for understanding the tran￾scriptomic changes of grafted watermelon plants in response to chilling. 2. Methods 2.1. Plant material and chilling stress treatment The watermelon line MW022 was used as the scion and the squash Jingxinzhen NO.4 (JX) was used as the rootstock. An “insertion grafting” procedure described by Lee and Oda (2003) was used in this study. Watermelon plants grafted onto their own roots were used as controls. Grafted seedlings were grown in a growth chamber at 28
C/18
C (16/8 h) day/night temperatures, a relative humidity of 70%, and a photon flux destiny of 400 mmol m2 s 1 . After the full development of the third true leaves, the grafted plants were treated with a low temperature at 10
C at the same relative humidity and illumination intensity. Leaves from 6 rootstock-grafted (RG) or self-grafted (SG) water￾melon seedlings were pooled as one biological replicate at 0 d and 1 d after the chilling treatment, respectively. Three biological rep￾licates were collected. All collected samples were immediately frozen in liquid nitrogen and stored at 70
C until use. 2.2. Assessment of chilling damage index (CI) and measurement of malonyldialdehyde (MDA) content Eighteen rootstock-grafted (RG) or self-grafted (SG) watermelon seedlings were moved to a climate chamber at a temperature of 10
C with a 16 h light/8 h dark cycle. After 12 d of the chilling treatment, the chilling damage index was measured according to Yang et al. (2008). The degrees of chilling tolerance were measured with 6 grades as follows: level 0: no symptom; level 1: chlorosis or crinkled at the edge of old leaves; level 2: chlorosis or crinkled at the edge of less functional leaves; level 3: chlorosis or crinkled at the edge of functional leaves with good new leaves; level 4: chlo￾rosis or crinkled and wilting of functional leaves with damaged new leaves; level 5: severe damage of new leaves, plants wilting or death. The chilling indices (CI) of RG and SG seedlings was tabu￾lated according to the following formula: CI ¼ P(each level  number of plants with the corresponding level)/total number of measured plants. The MDA content was measured using 2-thiobarbituric acid as described by Hodges et al. (1999). 2.3. RNA-seq library preparation and illumina sequencing Leaves from 6 RG or SG watermelon seedlings were pooled as one single biological replicate. The experiment was repeated to obtain three biological replicates. These pooled leaf samples were used for the RNA-seq analysis. Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and digested using DNase I at 37
C for 30 min to remove any possible genomic DNA contamination. The quality and concentration of each sample were determined using an Agilent 2100 Bioanalyzer. Six mg aliquots of total RNA from each sample were purified using oligo (dT) magnetic bead adsorption. Purified mRNAs were fragmented with endonu￾clease and ligated with adaptors to generate libraries with unique the 50 and 30 tags. After 15 cycles of linear PCR amplification, 105 bp PCR products were quantified and purified. Denatured molecules were then fixed onto an Illumina Sequencing Chip (Flow Cell) for sequencing. RNA-seq libraries were sequenced on an Illumina HiSeq™ 2000 System by BGI-Tech (Shenzhen, China). 2.4. Functional annotation of differentially-expressed genes (DEGs) Raw sequence reads were filtered through the Illumina pipeline by BGI Shenzhen, China. Gene expression levels were calculated using the RPKM (Reads Per kb per Million reads) method (Mortazavi et al., 2008). Differentially expressed genes (DEG) among the samples were identified using the novel NOIseq method (Tarazona et al., 2011) with a probability0.8 and an absolute value of the jlog2 Ratioj  1 as the threshold to evaluate the significance of gene expression differences. In brief, NOISeq method computes differential expression described as bellow: first, gene expression of sample in each group was used to calculate log2 (fold change) M and absolute different value D of all pair conditions (gene expres￾sion value will be substituted by 0.001 if it doesn't express in some sample). Second, average expression value of each gene standing for replicates will be used to calculate M and D. Two replicates in one of the experimental conditions is sufficient to run the algo￾rithm: Mi ¼ log2 xi 1 xi 2 ! and Di ¼ xi 1 xi 2 . Then, all these M/D values are pooled together to generate the noise distribution. If gene i differentially expresses between two groups, we set Gi ¼ 1, other￾wise set Gi ¼ 0, and give a definition for probability of gene i differentially expressing as following formula: P Gi ¼ 1 xi 1; xi 2  ¼ P Gi ¼ 1 Mi ¼ mi ;Di ¼ di  ¼ P M* < mi ;D* <di  When P is greater than threshold value, its corresponding gene is thought to differentially express between groups. Gene Ontology (GO) was used to analyze biological functions by mapping all DEGs to GO databases (http://www.geneontology.org/ ). GO terms meeting a threshold of corrected p-value0.05 were defined as significantly enriched. The DEGs were also mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa et al., 2008) to identify significantly enriched metabolic pathways or signal transduction pathways. 2.5. Quantitative real-time PCR (qRT-PCR) analysis Leave samples used for qRT-PCR were the same as those used for sequencing. Total RNA from watermelon leaves sampled at 0 d and 1 d after chilling treatment was extracted using the RNApure Plant Kit with DNase I (CWBiotech, Beijing, P. R. China). The BU￾Superscript RT Kit was used for first-strand cDNA generation with 562 J. Xu et al. / Plant Physiology and Biochemistry 109 (2016) 561e570