Dn Liu*,Yzeng ChengMin Gong,Qing ZhoCihong JingLirui ChengMin RenYunying WngAiguo Yng*

a Key Laboratory for Tobacco Gene Resources,Tobacco Research Institute,Chinese Academy of Agricultural Sciences,Qingdao 266101,Shandong,China

b Maize Research Institute,Sichuan Agricultural University,Chengdu 611130,Sichuan,China

Keywords:Cucum ber m osaic virus Tobacco Resistance Transcriptome Differential gene expression

A B S T R A C T Cucum ber m osaic virus(CMV)is one of the m ost severe viral diseases transm itted by aphids infecting Solanum crops in China,causing great losses of crop yields and incom e in rural com m unities.The tobacco cultivars NC82 and Taiyan 8 are closely related but differ in resistance to CMV.NC82 is susceptible to infection and Taiyan 8 is resistant,but the m echanism s underlying this difference in resistance are not clear.In this study,w e conducted RNA sequencing to analyze changes in gene expression induced in the leaves of Taiyan 8 and NC82 upon system ic infection w ith CMV,com pared w ith gene expression in the leaves of m ock-inoculated plants.Leaves w ere sam pled at one,three,eight,and 15 days after infection.In total,3443 and 747 differentially expressed genes were identified in Taiyan 8 and NC82,respectively.Gene ontology and pathw ay enrichm ent analyses revealed that the different responses to CMV infection betw een cultivars w ere based on m icrotubulebased processes,pentose and glucuronate interconversions,plant-pathogen interaction,and hormone signal transduction pathways.Genes encoding pathogenesis-related proteins,disease-resistance proteins,lipoxygenase,cellulose synthase,an auxin response factor,and an ethylene receptor show ed different expression patterns.The differen ces in gene expression follow ing CMV infection likely contributed to the different resistance levels of these two tobacco cultivars.The com prehensive transcriptom e dataset described here,w hich includes candidate response genes,w ill serve as a resource for further studies of the m olecular m echanism s associated w ith tobacco defense responses against CMV.

1.Introduction

Cucumber mosaic virus(CMV)is globally distributed and has a host range of＞1200 plant species,including common tobacco,Nicotiana tabacum L.[1].In N.tabacum,CMV can cause severe system ic m osaic sym ptom s such as leaf distortion and chlorosis,leading to severe yield losses every year in m any areas of China[2].Because CMV is very difficult to control solely w ith exogenous chem icals,genetic m anipulation,w hich is m ore eco-friendly,is under consideration as an alternative pathogen control m ethod.Thus,a better understanding of the molecular mechanisms of CMV resistance in tobacco w ould help w ith developing new strategies to im prove tobacco tolerance to CMV and to establish a theoretical basis for breeding resistant varieties of tobacco.

Plants have evolved a series of mechanisms to defend against viral infections.When a virus invades a plant,pathogenassociated molecular patterns(PAMPs),including bacterial flagellin and cellulose-binding elicitor proteins,are recognized by plant transmembrane receptors,which first activate basal defenses[3].In addition,plants have evolved a second line of defense,know n as effector-triggered im munity(ETI),to counter pathogens that have evaded the basal defenses[4].Genes encoding the specific determinants associated w ith ETI are know n as resistance(R)genes,w ith most R genes containing a nucleotide-binding site(NBS)and leucine-rich repeats(LRRs)[3,4].In Arabidopsis thaliana,the RCY1 gene,which encodes a coiled coil-NBS-LRR-type protein,confers resistance to the yellow strain of CMV(CMV-Y)w ith a resistance pathw ay that requires salicylic acid(SA)and ethylene signaling[5,6].In common bean(Phaseolus vulgaris)another NBSLRRgene,RT4-4,activates a resistance-related response,systemic necrosis,against CMV that is elicited by the CMV 2a gene product[7].Despite these responses in other plant species,few of the R genes identified thus far confer resistance to CMV in N.tabacum.In addition to R genes,RNA silencing is an important cellular pathw ay for anti-virus defenses[8].Host RNA-dependent RNA polymerases(Rd RPs)are components of the RNA silencing pathway[9]and are hypothesized to play a role in antiviral responses[8].For example,a mutation in AtRDR6,an Rd RPin A.thaliana,results in hypersusceptibility to CMV[10].

Because com plex m echanism s are involved in plant resistance to CMV,several studies have investigated the genome-wide expression profiles of plants after CMV infection,as w ell as the m olecular m echanism s in the pathogens that are associated w ith host infection.Microarray analyses revealed that genes encoding kinases and phosphatases,protein degradation m achinery/proteases,and transcriptional regulators show ed significant transcript regulation during Arabidopsis-CMV-Y interactions [11]. In addition, a m icroarray-based transcriptome analysis of the response Chrysanthemum spp.to CMV indicated that genes associated w ith chitin response,horm one-m ediated viral stress response,and DNA m odification w ere involved in the establishment of host resistance to CMV[12].

RNA sequencing(RNA-seq)is the m ost pow erful tool for transcriptom e characterization in investigations of disease resistance in plants.RNA-seq allow s gene discovery and global gene expression profiling,w hich helps to identify key com ponents of resistance pathw ays.In N.tabacum,RNA-seq w as used[13]to exam ine changes in gene expression associated w ith system ic sym ptom developm ent.Many biological processes,such as photosynthesis,pigm ent m etabolism,and plant-pathogen interactions,w ere found to be involved in system ic sym ptom developm ent.Changing levels of gene expression in dark green-and light green tissues w ere associated w ith leaf m osaic sym ptom,likely reflecting differences in the expression of genes encoding expansins,indole-3-acetic acid(IAA),cytokinin,and Dicer protein in the tw o tissue types[14].

Although global transcriptom e profiles have been used to investigate differential responses of resistant and susceptible tobacco genotypes to infections by Alternaria sp.and Phytophthora nicotianae[15,16],this approach has not been used to investigate tobacco-CMV interactions during early stages of infection.To date,studies on tobacco resistance to CMV have focused mainly on inheritance analyses[17,18]and gene expression patterns during disease developm ent[13,14].Taiyan 8 is a tobacco variety that is putatively resistant to CMV,w ith its resistance reported[18]to be a quantitative trait controlled by m ultiple genes.How ever,the resistance m echanisms are unknow n.In this study,w e exam ined Taiyan 8 and N.tabacum NC82(hereafter referred as NC82),w hich is susceptible to CMV.Systemically infected leaves from both cultivars w ere sam pled one,three,eight,and 15 days after CMV inoculation.The gene expression profiles in the leaves w ere then analyzed using RNA-seq analysis.Com parisons of the RNA-seq datasets from both resistant and susceptible genotypes revealed significant differences in the expression of genes involved in plant horm one signal transduction,plant defense pathw ays,virus trafficking,and DNA m odification.These data provide new insights into the genetic mechanisms of resistance m echanism s against CMV.

2.Materials and m ethods

2.1.Plant growth conditions and virus inoculation

Taiyan 8(CMV-resistant)and NC82(susceptible)seeds were obtained from the National Infrastructure for Crop Germplasm Resources(Tobacco,Qingdao,Shandong,China).Seeds w ere surface-sterilized w ith 3.0%NaClO for 5-10 min,then placed on plates with Murashige and Skoog m edia containing 3.0%sucrose and 0.6%agar.Seeds w ere germ inated under a controlled 16 h-light/8 h-dark cycle at 25°C.The seedlings w ere transferred to soil and grow n in a cham ber under a light intensity of 70-100μm ol m-2s-1,and with 14 h of light at 26°C and 10 h of darkness at 20°C.

Uniform ly grow n plants of each cultivar at the six-leaf stage w ere assigned to tw o groups for virus and mock inoculations,respectively.Tw o true leaves from the bottom w ere m echanically inoculated w ith CMV-SD.Virus sap w as prepared by hom ogenizing approxim ately 0.5 g of infected leaf tissue w ith 50 m m ol L-1potassium phosphate buffer(p H 7.5)[12].Mock inoculations w ere perform ed using inoculation buffer only.The topm ost fully expanded leaf w as collected at one,three,eight,and 15 days post-inoculation(dpi).To reduce individual variation among plants,three sam ples taken from three different plants w ithin the sam e treatment were pooled into a replicate,and three replicates w ere used from each treatm ent at each tim e point.

2.2.Assessment of NC82 and Taiyan 8 resistance to CMV

Approximately 100 plants of each cultivar were inoculated to assess the resistance of NC82 and Taiyan 8 to CMV.Disease severity in the inoculated plants w as assessed at 40 dpi,and CMV symptoms were rated using the following scale(disease grade):0,no symptoms;1,vein clearing in new leaves or mosaic symptoms just beginning;3,a third of the leaves on the plant show mosaic sym ptom s,but no deformation is observed;5,half of the leaves on the plant show m osaic symptoms,w ith some deformation observed;7,two thirds of the leaves on the plant show m osaic sym ptom s and deform ation;and 9,all leaves on the plant show mosaic symptoms and severe deformation.Disease severity w as calculated as follows:

Disease severity=∑(disease grade×num ber of plants in each grade)/(total num ber of plants×highest disease grade)×100.

2.3.RNA isolation,cDNA library preparation,and sequencing

Total RNA w as isolated from virus-and mock-inoculated sam ples w ith TRIzol reagent(Invitrogen Corp.,Carlsbad,CA,USA),according to the m anufacturer's instructions.RNA purity and concentration w ere estim ated using a Nano Drop 2000 spectrophotometer(Thermo Scientific,Waltham,MA,USA),and RNA integrity w as assessed using the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system(Agilent Technologies,Santa Clara,CA,USA).

For library construction,3μg of total RNA per sample was used,and messenger RNA(m RNA)was enriched from total RNA using magnetic beads w ith oligo(d T).The enriched m RNA w as then fragmented using divalent cations under elevated temperature in a fragmentation buffer(Illum ina Inc.,San Diego,CA,USA).First-strand complementary DNA(cDNA)w as synthesized using a random hexam er prim er,and second-strand cDNA w as then synthesized using DNA polym erase Iand RNase H.After end repair and adenylation of the 3′ends of DNA fragments,sequencing adapters w ere ligated to prepare for hybridization,and 150-200 bp fragments w ere size-selected by gel purification.Finally,polymerase chain reaction(PCR)amplification was performed,and libraries w ere sequenced using the paired-end sequencing module(100 bp at each end)of the Illumina HiSeq 2000 platform(Illum ina Inc.).

2.4.De novo transcriptome assembly

To obtain clean data(clean reads),reads containing adapters and poly-N sequences and low-quality reads w ere removed from the raw reads.All dow nstream analyses w ere based on high-quality clean reads.Denovo transcriptom e assem bly w as accom plished using Trinity[19],w ith m in_kmer_cov set to 2 and all other param eters set as default.Assem bled contigs w ere then hierarchically clustered using Corset[20]w ith shared reads and expression data.The longest transcripts in the cluster units were assigned as unigenes to eliminate redundant sequences and com bined to produce the final assem bly used for gene annotation.

2.5.Gene functional annotation

Functional annotation of all assembled unigenes w as perform ed using seven public databases.All unigenes w ere searched against National Center for Biotechnology Information(NCBI)nonredundant protein(Nr),NCBI non-redundant nucleotide(Nt),Protein family(Pfam),Eu Karyotic Orthologous Groups(KOG)of proteins,Sw iss-Prot protein,Kyoto Encyclopedia of Genes and Genomes(KEGG)Ortholog(KO),employing BlastX(v.2.2.28+)with an E-value of less than 1e-5.Gene Ontology(GO)annotations were analyzed using Blast2GO(v.2.5)[21].

2.6.Quantification of gene expression and analyses of differential expression

Gene expression levels for each sam ple w ere estim ated using RSEM[22].Clean reads w ere m apped back onto the assembled transcriptom e and read counts for each gene w ere then obtained from m apping results.Analyses of differential expression were performed using the DESeq R package[23],w hich provides statistical routines for determ ining differential expression based on digital gene expression data using a negative binom ial distribution m odel.The resulting P-values were adjusted with the Benjamini-Hochberg approach to control the false discovery rate(FDR).Genes w ith FDR＜0.05 and an absolute value of log2(ratio)≥1 w ere considered differentially expressed.

2.7.GO and KEGGpathw ay enrichment analysis

Gene ontology enrichm ent analyses of differentially expressed genes(DEGs)w ere performed using Web Gene Ontology(WEGO)[24].The KEGG[25]database w as used to identify high-level functions and utilities of biological systems(e.g.,cells,organisms,and ecosystems)based on m olecular-level inform ation,particularly large-scale m olecular datasets generated by genom e sequencing and other highthroughput experim ental technologies(http://w w w.genom e.jp/kegg/).The KOBAS[26]softw are w as used to test the statistical enrichment of DEGs in KEGGpathw ays.

2.8. Quantitative reverse-transcription PCR (q RT-PCR)validation

Tw elve genes found to be differentially expressed in the tw o cultivars w ere chosen for quantitative reversetranscription PCR(q RT-PCR).The q RT-PCR analysis w as perform ed using 48 RNA sam ples,w hich w ere used to prepare sequencing libraries.Relative gene expression w as calculated using the full-quantification m ethod w ith NtEF1α(Gen Bank accession ID NM001326165)as the internal control gene.Gene-specific prim ers(Table S1)w ere designed using Prem ier Prim e 5.0(http://w w w.prem ierbiosoft.com/crm/jsp/com/pbi/crm/clientside/ProductList.jsp).Approxim ately 1μg of total RNA from each sam ple w as reverse-transcribed using the Prim eScript RT reagent kit(Takara,Dalian,China).For each gene,q RT-PCR experim ents w ere replicated three tim es.The PCR protocol included pre-denaturation at 95°C for 30 s follow ed by 40 cycles at 95°C for 5 s and 60°C for 30 s;a m elt-curve analysis(60-95°C in 0.5°C increm ents for 5 s per step)w as then conducted to test amplicon specificity.The 2-ΔΔCTm ethod[27]w as used for relative quantification.

3.Results

3.1.Symptoms in leaves of N.tabacum inoculated with CMV

To evaluate the resistance of NC82 and Taiyan 8 to CMV,symptom development was assessed at multiple time points after inoculation.Although the leaves of both cultivars appeared unaffected at 1 and 3 dpi,NC82 leaves show light leaf vein clearing beginning at 8 dpi and m arked leaf vein clearing at 15 dpi.Taiyan 8 leaves,in contrast,showed no obvious symptoms through all four sampling time points(Fig.1-A).At 40 dpi,clear differences in the symptom s induced by CMV w ere observed betw een the tw o cultivars.The leaves of NC82 developed severe m osaic symptoms that included the formation of discrete regions of dark green tissues,w hereas Taiyan 8 show ed only slight leaf chlorosis(Fig.S1).Mean disease severity at 40 dpi w as 75.78 for NC82 and 12.33 for Taiyan 8(Fig.1-B).To identify the presence of CMV virions in the systemically infected leaves,the read numbers derived from CMV w ere com pared.No virions w ere detected in leaves of either cultivar at 1 and 3 dpi.At 8 and 15 dpi,how ever,NC82 show ed m ore reads of CMV than Taiyan 8(Fig.1-C).

Fig.1-Sym ptom s of Cucumber mosaic virus(CMV)infection and virus detection in system ically infected leaves of Nicotiana tabacum L.(A)Sym ptom s of CMV infection NC82 and Taiyan 8 at 0,1,3,8,and 15 d ays post-inoculation(dp i).The red rectangle indicates vein clearing.(B)Disease severity in Taiyan 8 and NC82 at 40 d pi.Error bars represent the stand ard error of the m ean of three replicates;*P＜0.05.(C)Exp ression levels(FPKM)of CMV-associated genes in system ically infected leaves of NC82(NA)and Taiyan 8(TA).Cluster-14949.79513 is pred icted to encod e the CMV strain CTL segm ent RNA1;cluster-14949.77700 is p redicted to encode the CMV isolate RP10 segm ent RNA2;cluster-14949.80413 is p redicted to encode the CMV 3a gene for m ovem ent p rotein and the cp gene for coat p rotein.

3.2.Sequencing and transcriptome assembly

Forty-eight sequencing libraries w ere generated for the total RNA sam ples of both virus-and m ock-inoculated NC82 and Taiyan 8 leaves sampled at the four time points(two cultivars×tw o treatm ents×three replicates per tim e point).After quality filtering,1,592,796,496 clean reads(159.28 Gb)w ere generated for NC82 and 1,646,442,498 clean reads(164.64 Gb)w ere generated for Taiyan 8(Table 1).Clean data w ere subm itted to the NCBI Sequence Reads Archive(SRA)database(Accession num ber:SRP126464).In addition,36 sequencing libraries w ere generated for the inoculated leaves collected at 1,3,and 5 dpi.These libraries provided another set of clean data that w ere subm itted to the NCBI SRA(Accession num ber:SRP126702).All clean reads from the 84 sequencing libraries w ere used for transcriptom e assem bly,and 377,547 transcripts and 359,112 unigenes w ere generated.Unigene lengths ranged from 201 to 16,749 bp,w ith an average of 1177 bp and an N50 length of 1960 bp(Table 2).To evaluate the reliability of the assem bly,the clean reads w ere m apped back onto the assem bled transcriptom e.At least 78.21%of the clean reads m apped back successfully(Table S2)for each sample,demonstrating the high quality of the transcriptom e assem bly.

Table 1-Statistics from Illum ina sequencing.

Although 267,674 unigenes(74.53%)w ere annotated in at least one database,only 25,730 unigenes(7.16%)w ere annotated in all seven databases.Based on sequence sim ilarity,196,003(54.57%)and 246,215(68.56%)unigenes w ere annotated in the Nr and Nt databases,respectively,and 123,347(34.34%)unigenes were annotated in the Sw iss-Prot database(Table 3).

Table 2-Statistics of assem bled d ata.

3.3.Identification of DEGs at different time points

To elim inate the effects of genetic differences and developm ent stage,DEGs w ere obtained by com paring gene expression levels betw een CMV-and m ock-inoculated samples of the same cultivar at the same time-points.In Taiyan 8,3443 DEGs w ere identified betw een virus-and m ockinoculated sam ples over the four different tim e-points,w hereas the same analysis in NC82 revealed only 747 DEGs(Table S3).In Taiyan 8,the num ber of DEGs increased w ith increasing dpi:123 DEGs at 1 dpi,181 DEGs at 3 dpi,204 DEGs at 8 dpi,and 2965 at 15 dpi.The m arked increase in DEGs at 15 dpi w as due to increased num bers of up-and dow nregulated DEGs,but m ostly to dow nregulated DEGs,w hose num ber w as three tim es that of upregulated genes(707upregulated versus 2258 dow nregulated;Table 4).Conversely,there w ere tw ice as m any upregulated genes as dow nregulated genes in NC82 at 15 dpi(Table 4).In NC82,there w ere 91,198,155,and 313 DEGs at 1,3,8,and 15 dpi,respectively.There w ere nine tim es as m any DEGs at 15 dpiin the resistant cultivar Taiyan 8 as in the susceptible cultivar NC82.

Table 3-Num bers and percentages of non-redundant unigenes annotated in public databases.

3.4.Quantitative reverse-transcription PCR(qRT-PCR)validation of DEGs

To validate the results obtained from RNA sequencing,five genes involved in disease defense and horm one signal transduction,w hich show ed opposite expression patterns in the tw o cultivars in response to CMV infection,as w ell as seven random ly selected genes that w ere differentially expressed in both cultivars,w ere chosen for q RT-PCR analysis.Although the fold changes in transcript abundance determ ined by RNA-seq and qRT-PCR did not match exactly,all 12 q RT-PCR analyses indicated that expression trends(as up-or dow nregulation)w ere sim ilar to those from RNA-seq.Expression profiles obtained from q RT-PCRand RNA-seq w ere m utually consistent,thus validating the reliability of the results generated by RNA-seq(Fig.2).

3.5.GO functional classification and pathw ay enrichment analyses of DEGs

To identify differences in GO functional classifications betw een the tw o cultivars after CMV inoculation, all DEGsannotated with GO term s were analyzed at different tim e points using the WEGO softw are.The GO term s w ere significantly different at P＜0.05,w ith the P-value based on the proportion of associated genes in the total num ber of analyzed DEGs.Under the category“biological processes”,term s w ith significant differences w ere“m acromolecule m etabolic process”and“regulation of m etabolic process”.The percentage of genes associated w ith these tw o terms w as higher for NC82 than for Taiyan 8.How ever,Taiyan 8 show ed a higher gene percentage associated w ith the term s“cell division”,“m icrotubule-based process”,“m ulticellular organism al developm ent”,and“m ulticellular organism al process”.The term“nucleic acid binding”w as predom inant in the m olecular function category,follow ed by “transcription regulator”,“transcription factor”,and“cofactor binding”.With the exception of“cofactor binding”,these term s w ere m ore frequently associated with genes of NC82 than w ith those of Taiyan 8.Only tw o term s in the“cellular com ponents”category,“m em brane-enclosed lum en” and “organelle lum en”,w ere associated at significantly different frequencies betw een the tw o cultivars(Fig.3;Table S4).

Table 4-Differentially expressed genes(DEGs)in Taiyan 8 and NC82 cultivars.

Fig.2-Valid ation of RNA-seq analysis using quantitative reverse-transcription polym erase chain reaction(q RT-PCR).Error bars rep resent the stand ard deviation of q RT-PCRsignals(n=3).TA,Taiyan 8 sam p les;NA,NC82 sam p les.

Fig.3-Gene ontology(GO)term s associated w ith d ifferentially exp ressed genes(DEGs)in NC82 and Taiyan 8.Significant d ifferences in the percentages of genes associated w ith each term w ere observed betw een the tw o cultivars.Term s w ere grouped into the follow ing three categories:biological processes,cellular com ponents,and molecular functions.Red and blue bars represent percentage of associated DEGs in NC82(NA)and Taiyan 8(TA),resp ectively.Differences in percentages w ere considered significant at P≤0.05.

In addition to GO analysis,DEGs in the two cultivars were m apped to terms in the KEGG enrichment pathw ays,and encoded enzymes w ere assigned to 106 KEGG pathw ays in Taiyan 8 and 74 KEGG pathways in NC82(Table S5).The top 20 KEGG pathw ays w ith the highest representation of DEGs are show n in Fig.4.In NC82,the most significantly enriched pathw ay w as“protein processing in endoplasmic reticulum”(Fig.4-A).In Taiyan 8,most DEGs were involved in DNA replication and repair(ko03030,ko03430,ko03440,ko03420,ko03410),nucleotide metabolism(ko00240,ko00230),carbohydrate m etabolism(ko00040,ko00520,ko00500,ko00051),amino acid m etabolism(ko00270,ko00250,ko00460),signal transduction(ko04075),and transport and catabolism(ko04145)(Fig.4-B).

3.6.Identification of DEGs associated w ith plant-pathogen interaction

Based on KEGG analysis,13 DEGs involved in plant-pathogen interaction w ere identified in Taiyan 8 and 7 DEGs in NC82(Table 5).These DEGs encoded m itogen-activated protein kinases,cyclic nucleotide-gated ion channel proteins(CNGC),and heat shock proteins in both cultivars.Six DEGs involved in calcium signaling,including calcium-dependent protein kinases(CPKs),calmodulin,and calmodulin-like proteins were identified only in Taiyan 8 plants after inoculation,w hereas the expression of these genes in NC82 did not change(Table 5).One gene that encoded the receptor-like protein kinase FLS2 and a WRKY transcription factor that w as repressed only in NC82 at 8 dpi was retrieved(Table 5).

3.7.Identification of DEGs associated with defense against diseases

Several DEGs associated w ith disease defense and stress resistance w ere found in both cultivars,including pathogenesisassociated proteins (PRs), disease-resistance proteins,lipoxygenases(LOXs),and cellulose synthase.The DEGs encoding PRs show ed opposite expression patterns betw een Taiyan 8 and NC82.Three DEGs encoding PRs identified in Taiyan 8 were all downregulated,whereas those identified in NC82 were all upregulated.Another three DEGs that w ere predicted to encode disease-resistance proteins w ere upregulated in Taiyan 8 at 15 dpi,w hereas one(Cluster-14949.82640)w as downregulated in NC82 at the same time point,confirming the roles of these DEGs in protecting plants against biotic stress.One DEG(Cluster-14949.203419)encoding RSH1 w as expressed in Taiyan 8 at 1 dpi,whereas the same DEGremained unchanged in NC82 at the same time point and w as suppressed at 8 dpi.Four DEGs encoding LOXs w ere induced by CMV infection at 15 dpi in Taiyan 8,whereas LOXs in NC82 were downregulated at 1 dpi and 3 dpi.The genes encoding cellulose synthase also showed differential expression patterns.The expression of Cluster-14949.184199,w hich is predicted to encode cellulose synthase protein H1,w as induced in Taiyan 8 at 15 dpi but suppressed in NC82 at 1 dpi(Table 6).The early negative responses of defense-associated genes in NC82 suggest that this cultivar is m ore susceptible to CMV infection.

Fig.4-The top 20 Kyoto Encyclopedia of Genes and Genom es(KEGG)p athw ays w ith the highest representation of d ifferentially expressed genes(DEGs)in(A)NC82 and(B)Taiyan 8 cultivars.

Table 5 - The expression and annotation of DEGs associated with plant-pathogen interaction in Taiyan 8 and NC82 cultivars at different time points after inoculation.

Table 6 - Expression and annotation of DEGs associated with disease defense in Taiyan 8 and NC82 cultivars at different time points after inoculation.

3.8.Identification of DEGs involved in plant hormone signal transduction

Many of the DEGs identified in both cultivars w ere involved in plant horm one signal transduction pathw ays,including several genes encoding auxin-responsive proteins,auxin response factors,and ethylene receptors(Table 7).Am ong these DEGs,tw o genes associated w ith plant hormone response show ed contrasting expression patterns betw een Taiyan 8 and NC82.The gene Cluster-14949.211040,w hich encodes an auxin response factor,was downregulated in Taiyan 8 at 15 dpi,w hereas it w as upregulated in NC82 at 3 dpi.Cluster-14949.230569,w hich is predicted to encode an ethylene receptor,w as also dow nregulated in Taiyan 8 and upregulated in NC82(Table 7).

4.Discussion

To our know ledge,no tobacco germ plasm show s full resistance to CMV.Taiyan 8,a cultivar that show s m oderate resistance to CMV,displayed sym ptom s of CMV infection later and w ith less severity after inoculation than NC82(Fig.1-A and B).As the three tobacco varieties w ith available reference genom es w ere all susceptible to CMV[28],in this study,the leaf transcriptom es of tw o tobacco cultivars Taiyan 8 and NC82 after the inoculation of CMV w ere assem bled to identify candidate resistant genes de novo.The results indicated that differential regulation of gene expression w as linked to the different patterns in pathogen defense that w ere observed betw een the tw o cultivars.

GO analysis revealed that the percentage of genes associated w ith the“m icrotubule-based process”(GO:0007017)of the total DEGs analyzed in Taiyan 8 was approximately twice that in NC82(Table S4).Plant cytoskeletons contribute to defense against the penetration of pathogenic fungi and oom ycetes[29].How ever,in plant-virus interactions,the virus appears to use the plant cytoskeleton for intra-and intercellular m ovem ent[30].By increasing the size exclusion limit of plasmodesm ata using viral m ovem ent proteins(MPs),viruses can spread throughout the host plant and com m andeer plant m echanisms for the intra-and intercellular trafficking of m acrom olecules[31,32].Viral MPs,localized to plasm odesm ata,interact w ith m icrotubules and favor the cell-to-cell m ovem ent of viruses[31,33].Most(73/83)of DEGs associated w ith m icrotubule-based processes in Taiyan 8 w ere dow nregulated,whereas 5/8 of the DEGs associated w ith these processes w ere upregulated in NC82(Tables S4),indicating that m icrotubules responded differently in host defense against CMV between resistant and susceptible cultivars.This difference in m icrotubule response to CMV infection m ay affect CMV movement.In support of this hypothesis,the CMV read num bers at 8 and 15 dpiin the system ically infected leaves of Taiyan 8 w ere significantly sm aller than those in NC82(Fig.1-C).

The pentose and glucuronate interconversion pathw ay(ko00040)is involved in the biosynthesis of cell w all com ponents[34].In Taiyan 8,genes in this pathw ay that encode pectinesterase,w hich is essential for the formation of the D-galacturonate backbone of pectic acid[35],were downregulated.Only one DEG w as assigned to ko00040 in NC82,and it is predicted to encode a pectinesterase inhibitor.The expression of this gene w as significantly upregulated,leading to pectinesterase inhibition[36].Thus,the synthesis of pectin acid,the main cell wall com ponent,might be suppressed by CMV infection in both cultivars.How ever,the genes encoding the synthesis of cellulose,another cell w all component,responded differently to CMV infection.The gene Cluster-14949.184199,which encodes a cellulose synthase,w as upregulated in Taiyan 8 and dow nregulated in NC82(Table 6),and m ight thus play an important role in plant pathogenesis[37].

PAMP-triggered im m unity,w hich is the first layer of the plant innate im m une system,is prom pted by the recognition of PAMPs by pattern-recognition receptors for plant cell surfaces[38].The suppressed expression of the receptor-like protein kinase FLS2 and the WRKY transcription factor in NC82 at 8 dpi suggested that the basal defense system of the susceptible cultivar is m ore vulnerable to CMV infection.The second layer of defense is ETI,w hich is activated by the recognition of m icrobial effectors by plant resistant proteins[38].Several disease-resistance proteins w ere activated in Taiyan 8 but rem ained unchanged or suppressed in NC82 upon CMV infection.RPPs are mem bers of a large class of pathogen resistance genes encoding NBS-LRR dom ains[39].The RelA/Spot hom olog AtRSH1 in A.thaliana interacts specifically w ith AtRPP5,w hich is a pathogen R gene containing NBS-LRRthat confers resistance to dow ny m ildew[40].In Taiyan 8,tw o RSH1 genes and one RPP13-like gene(Cluster-14949.193737)w ere upregulated sim ultaneously(Table 6).The RSH1-RPP protein interaction m ay contribute to CMV disease resistance.In A.thaliana,LOXs play important roles in defense regulation and hypersensitive cell death responses to m icrobial pathogens,conferring basal resistance to a broad range of pathogens[41].Four DEGs encoding LOXs in Taiyan 8 were induced by CMV infection,w hereas the expression of tw o LOXs in NC82 w as suppressed.The enhanced expression of disease-resistance proteins and LOXs in Taiyan 8 m ay represent an attempt to trigger the plant immune system to m ount an active defense response to CMV infection.

Virus infection disturbs the equilibrium of plant horm one signal transduction.In Taiyan 8,m ost DEGs associated w ith auxins w ere suppressed by CMV infection,including one gene encoding an auxin response factor(ARF),w hich w as induced in NC82.Sim ilarly,three DEGs involved in the ethylene pathw ay w ere dow nregulated in Taiyan 8,one of w hich(ETR1,Cluster-14949.230569)w as upregulated in NC82(Table 7).Auxins and ethylene are both involved in sym ptom developm ent during virus infection,and the active of their signaling pathways have been reported to promote disease susceptibility to Pseudomonas syringae[42,43].ARFs activate auxin-responsive gene expression through interactions w ith

auxin-responsive elements in auxin signaling pathw ays,and are suppressed by a fam ily of repressors know n as AUX/IAA proteins[42].Previous study[44]has suggested that interaction betw een CMV replicase-and AUX/IAA selectively enhances virus pathogenicity in tissues of Solanum lycopersicum(w here AUX/IAA proteins accum ulate).The failure to suppress ARF3 and ARF4 w ith sm all interfering RNA(siRNA)in w iry m utants results in shoestring m osaic disease,w hich causes symptom s similar to those seen in phenotypes of tomato plants infected w ith CMV and tom ato m osaic virus[42].Ethylene is responsible for sym ptom developm ent in cucumber plants infected w ith CMV[45].Mutations in the ethylene pathw ay such as ein2(ethyleneinsensitive2)and erf106(ethyleneresponsive transcription factor 106)lead to resistance against crucifer-infecting tobacco m osaic virus(TMV-cg)[46].These results suggest that inhibition of auxin and the ethylene pathw ay in Taiyan 8 might inhibit symptom development resulting from CMV infection.SA is a putative signal m olecule that stim ulates responses to different biotic and abiotic stresses,and is essential for the establishm ent of local and systemic acquired resistance[42].Three DEGs encoding the nonexpressor of a pathogenesis-related protein(NPR)involved in the SA pathw ay w ere suppressed in Taiyan 8,but remained unchanged in NC82 after CMV infection(Table 7).NPRs suppressed PR gene expression and pathogen resistance in Arabidopsis[47],suggesting that the NPRin the SA pathw ay m ay be im portant for resistance against CMV.

Table 7 - Expression and annotation of DEGs involved in plant hormone signal transduction in Taiyan 8 and NC82 cultivars at different time-points after inoculation.

5.Conclusions

The patterns of gene expression in the leaves of Taiyan 8 and NC82 plants that w ere systemically infected with CMV provided a com prehensive overview of the transcriptom es of tw o tobacco cultivars w ith different disease responses.The m ain differences in responses betw een the two cultivars were that(1)intercellular virus spread and sym ptom development in the resistant cultivar m ay be inhibited by the suppression of genes associated w ith virus trafficking and ethylene and auxin signaling pathw ays;and(2)regulation of defenserelated genes and the SA pathw ay m ight activate pathw ays associated w ith defense responses in the resistant cultivar.Our results show that Taiyan 8 uses different defense pathw ays that comprise a complex resistance netw ork in response to CMV infection.Further studies should focus on the functional validation of candidate DEGs,to determ ine w hether these genes are responsible for the different responses to CMV infection observed between Taiyan 8 and NC82.The results of this study shed light on the m olecular m echanism s underlying resistance to CMV.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.11.008.

Acknow ledgm ents

This w ork w as funded by the Agricultural Science and Technology Innovation Program(ASTIP-TRIC01),the Science Foundation for Young Scientists of the Tobacco Research Institute of the Chinese Academ y of Agricultural Sciences(2016A04),the National Natural Science Foundation of China(31301678),Fundam ental Research Funds for Central Nonprofit Scientific Institution,and Tobacco Genom e Project of China National Tobacco Corporation (110201601028,110201402006,110201301009).