Introduction

The chrysanthemum plants belonging to the family Asteraceae have been widely cultivated for horticulture, gardening, floriculture, herbal medicine and insecticide (Hitmi et al. 2000; Wu et al. 2010). Polyploidy and hybridization have been known as major factors for the speciation and genetic diversity of the genus Chrysanthemum (Liu et al. 2012; Yang et al. 2006). So far, numerous commercial chrysanthemum cultivars with polyploidy have been developed for cut flowers and pot plants using hybridization (Ding et al. 2008; Li et al. 2013).

Several pathogens including viruses, viroids, fungi and phytoplasma cause serious damages on the production of chrysanthemum cultivars. In particular, the chrysanthemum is the host for at least nine viruses including Chrysanthemum B virus (CVB), Cucumber mosaic virus (CMV), Potato virus Y, Potato virus X (PVX) Tobacco mosaic virus, Tomato aspermy virus, Tomato spotted wilt virus (TSWV) and Turnip mosaic virus (Verma et al. 2003) as well as two viroids like Chrysanthemum stunt viroid and Chrysanthemum chlorotic mottle viroid (Cho et al. 2013). To control viruses and viroids with chemical treatments and culturing approaches are still challenging in chrysanthemum plants. To make chrysanthemum cultivars resistant to viruses and viroids, it is important to identify chrysanthemum cultivars or to generate genetically modified (GM) chrysanthemums resistant to viral stresses.

Development of genetically modified (GM) chrysanthemums is one of effective ways to introduce new traits into the chrysanthemum. For instance, GM chrysanthemums with white flowering variety have been developed using chimeric chalcone synthase (CHS) gene (Courtney-Gutterson et al. 1994). Therefore, it is necessary to identify and characterize useful chrysanthemum genes for the GM chrysanthemum. Recently, several studies have identified chrysanthemum genes related to flowering and abiotic stresses (An et al. 2014; Fu et al. 2014; Liu et al. 2014; Song et al. 2014a, b; Tong et al. 2009; Yang et al. 2014). Of them, the overexpression of chrysanthemum dehydration responsive element binding factor 1 (DREB1) showed delayed flowering and strong resistance against freezing and drought stresses (Tong et al. 2009). Another chrysanthemum BBX24, which is a zinc finger transcription factor gene, has been characterized to function flowering and abiotic stresses (Yang et al. 2014). Both DREB1 and BBX24 genes can be usefully applied for GM chrysanthemum plants with delayed flowering and tolerance to abiotic stresses.

Although the chrysanthemum plants are one of economically important flower plants, biochemical and molecular genetic studies for the chrysanthemum are limited. One of main obstacles for chrysanthemum genetic and genome study is polyploidy. Therefore, transcriptome based approach has been recently performed to provide expressed sequence tags (ESTs) from chrysanthemum inflorescence (Chen et al. 2009). Moreover, next generation sequencing (NGS) has provided a large number of chrysanthemums mRNAs (Jo et al. 2014; Wang et al. 2013, 2014). In addition, NGS can be used to examine gene expression profiles of chrysanthemum under dehydration stress (Xu et al. 2013) and Alternaria tenuissima inoculation (Li et al. 2014). Furthermore, the generated sequencing data has been used for development of simple repeat markers (Wang et al. 2013). Furthermore, cDNA-amplified fragment length polymorphism (AFLP) technology has been applied to find genes associated with floral development (Ren et al. 2013) and salt stress (He et al. 2012).

Several genome wide transcriptome analysis have been performed for model plants in response to diverse plant viruses. For instance, PVX infection in Nicotiana benthamiana (García-Marcos et al. 2009), for CMV infection in Arabidopsis (Marathe et al. 2004; Whitham et al. 2003), N. tabacum (Lu et al. 2012), and tomato (Lang et al. 2011), for TSWV infection in tomato (Catoni et al. 2009) have performed to establish expression profiles upon virus infection. However, study on gene expression profiles of the non-model plants such as chrysanthemum is limited.

In this study, we for the first time established a microarray system to examine gene expression profiles of chrysanthemum in response to different plant viruses. For that, we selected three viruses which infect a wide range of plants including many commercially important vegetables. Finally, two positive single strand (ss) RNA viruses such as PVX in the genus Potexvirus and CMV in the genus Cucumovirus as well as one negative ssRNA virus such as TSWV in the genus Tospovirus were chosen. Expression profiles using a chrysanthemum microarray revealed commonly and differentially regulated chrysanthemum genes in response to three different viruses. The identified chrysanthemum genes will be good candidates for functional characterization in association with virus infection and they can be applied to develop GM chrysanthemum plants resistant to various plant viruses in near future.

Materials and methods

Plant propagation and virus inoculation

Dendranthema grandiflorum Ramatuelle cultivar Shinma was obtained from the National Institute of Horticultural and Herbal Science in Korea and propagated in vitro in Murashige and Skoog (MS) solid medium containing 30 % sucrose and 0.1 % activated carbon. Five-week old chrysanthemum plants propagated by cuttage which were around 30 cm tall containing several leaves were used for virus inoculation. For virus inoculation, sap was prepared from approximately 0.5 µg of each virus infected tissues by homogenizing with the buffer containing 50 Mm potassium phosphate (pH 7.5). Mechanical inoculation was carried out on the leaves of chrysanthemum using carborundum. Mock samples were treated with inoculation buffer. After inoculation, plants were continually grown in growth chamber. Three different plants for each virus were inoculated. Three viruses were not co-inoculated. As a result, nine and three plants were used for virus and mock inoculation, respectively. After virus infection, no viral symptoms were observed for all virus inoculated chrysanthemum plants at 5 days after virus inoculation. We performed RT-PCR to confirm infection of each virus in inoculated leaves (Supplementary Fig. 1). In addition, we also inoculated same virus sap for PVX, CMV, and TSWV, respectively, on the leaves of Nicotiana benthamiana. We could detect viral RNAs but did not observe any virus symptoms on the inoculated leaves. All chrysanthemum plants used in this study were grown in growth chamber at 25 °C for 16 h/8 h. We collected virus inoculated leaves which did not display any viral symptoms and used for total RNA extraction.

Samplings of seedlings and total RNA extraction

To examine expression profiles in local symptoms, inoculated leaves were used. Each virus-infected and mock-treated samples were harvested at 5 days post infection and immediately frozen in liquid nitrogen. Total RNA was extracted using Qiagen RNeasy plant mini kit (Qiagen GmbH, Hilden, Germany) following manufacturer’s manual. To confirm virus infection in each sample, reverse-transcription polymerase chain reaction (RT-PCR) was performed. First strand cDNA was synthesized using GoScript™ Reverse Transcriptase (Promega Corp, Madison, USA) and PCR amplification was done using Takara Ex Taq™ Polymerase (Takara, Kusatsu, Japan) using virus specific primers (Supplementary Table S1). PCR products were analyzed by agarose gel electrophoresis (Supplementary Fig. S2).

Design of microarray probes

The 13,687 consensus sequences were obtained from 18,784 ESTs after removing redundant sequences. In addition, 85 sunflower chloroplast genes were also included (Timme et al. 2007). Seven probes with 60-nt long were designed for each gene started from 250 bp ahead to the end of stop codon by shifting 15 bp gradually. As a result, seven probes covered 150 bp in the 3′ region of each gene. Moreover, 50 mitochondrial genes and several selection markers encoding GFP, GUS, HYG, BAR, and KAN were included. In total, 92,448 probes were designed. The average size of probe was 60-nt long with adjusting its Tm value from 75 to 85 °C. The microarray chips were manufactured at NimbleGen Inc (http://www.nimblegen.com/). Random GC probes (40,000) to monitor the hybridization efficiency and four corner fiducial controls (225) were included to assist with overlaying the grid on the image.

cDNA library preparation and microarray hybridization

Twelve different total RNAs were subjected for cDNA library preparation. For the synthesis of double strand cDNAs, RevertAid™ H Minus First Strand cDNA Synthesis Kit (Fermentas, Ontario, Canada) was used. Briefly, 1 µl of oligo dT primer (100 mM) and 10 µl (10 µg) of total RNA were mixed and denatured at 70 °C for 5 min and renatured by cooling the mixture on ice. The 4 µl of 5X First Strand Buffer, 1 µl of RiboLock™ Ribonuclease Inhibitor, 2 µl of 10 mM dNTP mix and 1 µl of RevertAid™ H Minus M-MuLV Reverse Transcriptase enzyme were mixed together and used for First strand DNA synthesis by incubating at 42 °C for 1 h. To stop the reaction, mixture was heated at 70 °C for 10 min. To synthesize the second strand, 66.7 ml of nuclease free water, 5 µl of 10X reaction buffer for DNA Polymerase I (Fermentas), 5 µl of 10X T4 DNA ligase buffer (Takara, Kyoto, Japan), 3 µl of 10 U/µl DNA Polymersase I (Fermentas), 0.2 µl of 5U/µl Ribonuclease H (Fermentas) and 0.1 µl of 350 U/µl T4 DNA ligase (Takara) were added to the first strand reaction mixture and the reaction was proceeded at 15 °C for 2 h. Double stranded cDNA mixture was purified using MinElute Reaction Cleanup Kit (QIAGEN, Valencia, USA). For the synthesis of Cy3-labeled target DNA fragments, 1 mg of double strand cDNA was mixed with 30 µl (1 OD) of Cy3-9mer primers (Sigma-Aldrich, St. Louis, USA) and denatured by heating at 98 °C for 10 min. The reaction was further proceeded by adding 10 µl of 50X dNTP mix (10 mM each), 8 µl of deionized water, 2 µl of Klenow fragment (50 U/ml, Takara) and incubating at 37 °C for 2 h. DNA was precipitated by centrifugation at 12,000×g after adding 11.5 µl of 5 M NaCl and adding 110 µl of isopropanol. Precipitated samples were rehydrated with 13 µl of water. The concentration of each sample was determined by using spectrophotometer. 10 µg of DNA was used for microarray hybridization. The sample was mixed with 19.5 µl of 2X hybridization buffer (Nimblegen, Delaware, USA) and finalized to 39 µl with deionized water. Hybridization was performed with MAUI chamber (Biomicro, Salt Lake City, USA.) at 42 °C for 16–18 h. After the hybridization, the microarray was removed from MAUI Hybridization Station and immediatly immersed in the shallow 250 ml Wash I (Nimblegen) at 42 °C for 10–15 s with gentle agitation and then transferred to the second dish of Wash I and incubated for 2 min with gentle agitation. The microarray was transferred into dish of Wash II and further washed in Wash III for 1 min with agitation. The microarray was dried in a centrifuge for 1 min at 500 g and scanned using GenePix scanner 4000B (Axon, Union City, USA)

Microarray data analysis

The microarray was scanned with Genepix 4000 B (Axon) preset with a 5 µm resolution and for Cy3 signal. Signals were digitized and analyzed by Nimblescan (Nimblegen). The grid was aligned to the image with a chip design file, NDF file. The alignment was checked by ensuring that the grid’s corners are overlaid on the images corners. This was further checked by uniformity scores in the program. An analysis was performed in a two-part process. First pair reports (.pair) files were generated in which sequence, probe, and signal intensity information for Cy3 channel were collected. Data-based background subtraction using a local background estimator was performed to improve fold change estimates on arrays with high background signal. The data was normalized and processed with cubic spline normalization using quantiles to adjust signal variations between chips (Workman et al. 2002). Probe-level summarization by Robust Multi-Chip Analysis (RMA) using a median polish algorithm implemented in NimbleScan was used producing calls files. The method identifies probes that are outliers in the overall behavior of the expression measured for a given gene and the contribution by those outliers is reduced in the reported gene expression level. This improves the sensitivity and reproducibility of microarray results (Irizarry et al. 2003). Raw data were deposited in NCBI’s GEO database with an Accession Number GSE55985.

Multiple analysis was performed with limma package in R computing environment (Smyth 2004). The package adopts the linear modeling approach implemented by lmFit and the empirical Bayes statistics implemented by eBayes. Genes of which adjusted p value or false discovery below 0.05 were collected and further selected for those gene expressions were higher than 1 or less than −1 at least at the virus infected sample compared to that of mock sample. Multivariate statistical tests such as clustering, principal component analysis, and multidimensional scaling were performed with Acuity 3.1 (Axon Instruments). Hierarchical clustering was performed with similarity metrics based on squared Euclidean correlation and average linkage clustering was used to calculate the distance of genes using genesis (Sturn et al. 2002).

Gene enrichment analysis

We divided differentially expressed gene (DEG) into down-regulated and up-regulated DEG at each virus infected condition. To identify enriched GO terms in each group containing lists of DEG, we used the Gossip package (Blüthgen et al. 2005) implemented in Blast2GO (Conesa et al. 2005) which uses the Fisher’s Exact Test and corrects for multiple testing. For chrysanthemum annotation, all EST sequences used for microarray probes were subjected for Blast2GO analysis and the obtained gene ontology was used as the reference file. To identify enriched GO terms, GO terms with FDRs less than 0.05 were collected.

Quantitative real-time RT-PCR

The primers used for real-time PCR were designed using PrimerSelect (DNASTAR, Madison, USA) following the manual and were listed in Supplementary Table 1. Two fragments for a actin gene (Accession Number JN638568.1) was used as a reference gene based on the previous study which demonstrated that the actin gene among eight tested reference genes was generally stable in various stress conditions for quantitative real-time RT-PCR in chrysanthemums (Gu et al. 2011). cDNAs of each sample were synthesized by SuperScript® III (Invitrogen, Carlsbad, USA) reverse transcriptase mixture using oligo (dT 20) (Invitrogen) as reverse primers. Real-time PCR was performed with a Bio-Rad CFX384 Real-time PCR system (Bio-Rad, Hercules, USA) in Bio-Rad iQTM SYBR® Green Supermix (Bio-Rad) reagents according to manufacturer protocols. Briefly, real-time PCR amplification reaction for the each genes was performed in 10 µL volume including 5 µL of iQTM SYBR® Green Supermix, 10 ng cDNA and 10 pm forward and reverse primers, respectively. The PCR cycling condition consisted of an initial 3 min 95 °C cycle followed by 40 cycles of 30 s at 95 °C, and 30 s at 55 °C. Data analysis was carried out by Bio-Rad CFX Manager software (Version 3.1).

Results

Generation of a chrysanthemum microarray and transcriptome analysis in response to three different viruses

To examine changes of the chrysanthemum transcriptome in response to diverse viruses, we generated a microarray based on previously published ESTs data derived from flower and leave tissues (Chen et al. 2009; Jo et al. 2014). The microarray probes were consisted of chrysanthemum 13,251 nuclear genes as well as 85 chloroplast-encoded genes of the sun flower belonging to the family Asteraceae (Timme et al. 2007). One month-old chrysanthemum plants were prepared by cuttage. Three different viruses were independently infected by the direct rub-inoculation with sap and mock treated plants were used as controls. A total of 12 samples as well as 12 DNA chips were used for microarray analysis (Supplementary Table 2). Microarray and data analyses were performed as described in materials and methods.

Identification of differentially expressed genes in response to three different viruses

We compared expression profiles of each virus infected condition to those of mock-treated condition to get expression ratios (fold changes) and respective p values by averaging data from three different biological replicates (Supplementary Table 3). To identify DEG in response to each virus infection, we applied a fold-change cutoff of 2 and p value threshold of 0.05 (Supplementary Table 4). To examine global gene expression patterns, volcano plots were generated by using fold changes and p values (Fig. 1a). CMV and PVX caused dramatic transcriptional changes of the chrysanthemum transcriptome than TSWV. Only less than 1 % of genes were identified as DEGs. In final, we identified 127 DEGs (CMV), 72 DEGs (TSWV), and 125 DEGs (PVX) (Supplementary Table 4). Interestingly, the number of up-regulated genes was more than two times as compared to that of down-regulated genes by CMV and PVX, respectively (Fig. 1b).

Fig. 1
figure 1

Identification of differentially expressed chrysanthemum genes in response to infection of three different viruses. a Volcano plots display log2 converted fold changes and p values. To identify differentially expressed genes, each virus infected sample was compared to mock treated sample. Obtained fold changes and p values were converted to log2 to generate volcano plots using ggplot2 program implemented in R program. More than twofold changes and p values less than 0.05 were applied to identify DEG indicated by blue-colored dots. b The number of DEG in response to individual virus infection. Blue and red colored bars indicate down- and up-regulated DEGs, respectively, in each virus infected sample

Full size image

Next, we compared the number of DEG among three conditions. Many genes displayed virus specific expression. For example, 46, 39 and 26 genes were specifically expressed by CMV, TSWV and PVX, respectively (Fig. 2a). Thirty-three genes were commonly identified in three conditions (Table 1; Fig. 2a). Of them, 22 genes showed up-regulation while 11 genes displayed down-regulation (Fig. 2b, c). In addition, 35 genes (CMV), 6 genes (TSWV), and 25 genes (PVX) exhibited virus specific up-regulation (Fig. 2b). Moreover, 11 genes (CMV), 20 genes (TSWV), and 13 genes (PVX) showed virus specific down-regulation (Fig. 2c).

Fig. 2
figure 2

Comparison of the number of DEGs among three different gene sets. Venn diagrams display the number of DEGs from three different gene sets. A total of 127, 72, 125 DEGs from CMV, TSWV, and PVX, respectively, were compared. Total, up and down indicate identified total number of DEGs, only up-regulated DEGs and only down-regulated DEGs, respectively

Full size image
Table 1 The 33 chrysanthemum DEGs commonly regulated by three different viruses
Full size table
Table 2 Significantly enriched GO terms in up- and down-regulated gene sets by TSWV infection
Full size table

Of identified DEGs, genes encoding dehydrin (ESTC008904), nuclease HARBI1-like protein (ESTC000886), and 9-cis-epoxycarotenoid dioxygenase chloroplastic-like protein (ESTC000686) were strongly up-regulated by CMV (Supplementary Table 4). TSWV infection strongly up-regulated genes encoding nematode resistance HSPRO2-like protein (ESTC000887), ethylene responsive factor (ERF)/AP2 transcription factor (TF) (ESTC004608) and ERF109-like protein (ESTC001650). By PVX infection, genes encoding nuclease HARBI1-like protein (ESTC000886), alcohol dehydrogenase-like 3-like protein (ESTC001297) and nematode resistance HSPRO2-like protein (ESTC000887) were strongly up-regulated. Twenty-two genes such as genes encoding calcium-binding protein, glycosyltransferase family protein and an ERF TF were up-regulated by all three viruses while 11 genes including genes encoding xyloglucan-specific fungal endoglucanase inhibitor protein precursor and ribonucleoside-diphosphate reductase small chain-like protein were down-regulated by three viruses.

Go enrichment analysis to identify important functions associated with virus infection

To reveal function of DEGs in response to each virus, we conducted gene ontology (GO) enrichment analysis implemented in Blast2GO program. A total of 132 GO terms were significantly enriched in three conditions (Supplementary Table 5, Fig. 3a). Only 8 and 39 terms were identified from CMV and PVX, respectively while 85 GO terms were obtained from TSWV. Six GO terms were commonly identified in three conditions (Fig. 3b). They are related to endogenous stimulus, chitin, organonitrogen compound response, and DNA unwinding function according to biological process (Fig. 3c) and MCM complex was solely identified according to cellular component (Fig. 3d).

Fig. 3
figure 3

Comparison of identified enriched GO terms from three different gene sets. a The number of identified GO terms by gene enrichment analysis according to three categories. b Comparison of identified enriched GO terms among three different gene sets. Venn diagram illustrating the number of overlapped and specific GO terms for each condition. Commonly identified GO terms according to biological process (c) and cellular component (d). A detailed information for enriched GO terms is available in Supplementary Table 3. The directed acyclic graph (DAG) illustrates the relationships of the GO terms, which are highly over-represented in the chrysanthemum transcriptome in response to three virus infection. Violet colored box indicates identified enriched GO terms

Full size image

CMV infection up-regulated expression of genes involving in stresses as well as beta-carotene 15-monooxygenase and serine/threonine phosphatase activities (Supplementary Table 5). TSWV infection resulted in up-regulation of genes related to external or internal stimulus (TSWV) (Supplementary Fig. 2A). Of them, genes responding to chitin including as CCR4-associated factor, nematode resistance HSPRO2 and protein phosphatase 2C (PP2C) were identified as major DEGs (Table 2). In particular, up-regulation of three ERF TF genes indicated association of ethylene mediated signaling pathway for plant defense response against TSWV infection. Down-regulated genes in response to TSWV were related to DNA metabolic process (Supplementary Table 5 and Supplementary Fig. 2B). For example, these genes were involved in DNA replication, chromatin organization and cytokinesis by cell plate formation. The genes including mitotic spindle checkpoint protein, ribonucleoside-diphosphate reductase, prolifera, kinesin, cyclin-dependent kinase and DNA replication licensing factor are required for DNA-dependent DNA replication (Table 2). Four genes encoding histone required for chromatin organization were down-regulated by TSWV. These genes including kinesin-1, cyclin-dependent kinase b2-2, DNA replication licensing factor MCM3 homolog 1 are required for assembly of chromatin, nucleosome, and MCM (eukaryotic genome DNA replication) complex. In addition, genes required for methylation like histone methylation and phosphorylation were also up-regulated.

Up-regulated genes by PVX were also involved in response to various stresses such as chitin, nitrogen compound and oxygen-containing compound (Supplementary Table 5 and Supplementary Fig. 2C). Those genes were also up-regulated by PVX and were related to defense response to establish host immune system (Table 3). In addition, genes required for nucleic acid processing function such as mRNA 3′-end processing, nuclear-transcribed mRNA poly(A) tailing shortening, nuclear-transcribed mRNA catabolic process and 3′-5′ exonuclease activity were up-regulated by PVX infection (Supplementary Table 5). The genes involving in mRNA 3′-end processing are identified as four chrysanthemum CCR-associated factor 1 gene (Table 3). Furthermore, PVX infection up-regulated ethylene response related genes encoding CCR-associated factor 1 and ERF TFs (Table 3).

Table 3 Significantly enriched GO terms in up-regulated gene sets by PVX infection
Full size table

Gene expression of TFs and chloroplast genes

Using known Arabidopsis TF database, at least 610 chrysanthemum TFs were present on microarray chips. To identify differentially expressed TFs, hierarchical clustering was performed to identify three groups of TFs (Fig. 4a). Forty-three TFs in the group A were down-regulated while 24 genes in group B and 88 genes in group C were up-regulated (Fig. 4b). In particular, eight genes in group B were members of the ERF TF family. In case of chloroplast encoded genes, most chloroplast genes were down-regulated by three viruses (Fig. 4c). However, some chloroplast genes encoding photosystem II and cytochrome b6/f complexes were slightly up-regulated by CMV.

Fig. 4
figure 4

Expression profiles of chrysanthemum TFs and chloroplast genes. a Three group of chrysanthemum TFs which were differentially expressed. To identify chrysanthemum TFs differentially expressed by infection of three different viruses, Log2 converted fold changes of a total of 610 chrysanthemum TFs were subjected for hierarchical clustering using the genesis program. Detailed information for TFs can be found in Supplementary Table 4. b Average expression views of TFs in each group showing three distinct expression patterns. c Expression of 85 chloroplast encoded genes in response to infection of three different viruses

Full size image

Validation of microarray data by real time RT-PCR

To validate results of microarray analysis, we performed quantitative real time RT-PCR with six selected genes encoding ethylene-responsive transcription factor 4, phosphatase 2c 25, ribonucleoside-diphosphate reductase small chain, basic 7s globulin, GDSL esterase lipase and a unknown function protein (Supplementary Table 1). Completely independent biological samples were prepared for real time RT-PCR. Two fragments for an actin gene were used as a reference gene. Two repeated real time RT-PCR results displayed reliable standard deviations (Fig. 5a). Except one gene encoding basic 7s globulin, all five genes showed strong correlation between microarray results and RT-PCR results (Fig. 5b). For instance, fold changes of the PP2C 25 gene (ESTC000379) as compared to mock by RT-PCR were higher than microarray data.

Fig. 5
figure 5

Quantitative real time RT-PCR to validate microarray data. a Normalized expression of six selected genes with standard errors. To normalize expression, two actin genes were used. b The detailed fold changes of six genes by microarray and real time RT-PCR. Red and green color indicate up- and down-regulation as compared to mock sample. The numbers are log2 converted fold changes

Full size image

Comparison of chrysanthemum microarray data with the previous Arabidopsis microarray data in response to RNA viruses

Previously, several studies performed transcriptome analyses in response to various plant viruses (Ascencio-Ibáñez et al. 2008; Marathe et al. 2004; Whitham et al. 2003). It might be of interest to compare DEGs between two different hosts upon same virus infection. Of them, we selected a study which performed a comprehensive Arabidopsis transcriptome analysis in response to five viruses including CMV and PVX using Arabidopsis GeneChip microarrays (Affymetrix) (Whitham et al. 2003). To compare two studies, all chrysanthemum DEGs were converted to corresponding Arabidopsis gene loci based on sequence similarity. However, many chrysanthemum DEGs were chrysanthemum specific genes. Unexpectedly, only a gene out of 94 non-redundant Arabidopsis DEGs in response to five RNA viruses was overlapped with the list of DEGs upon PVX infection (Supplementary Fig. 3). The gene (AT2G22500) is one of the mitochondrial dicarboxylate carriers (Palmieri et al. 2008). However, members of some gene families such as protein kinase, phosphatase, ubiquitin, heat shock protein, ethylene responsive factor and cytochrome P450-like protein were commonly identified in both studies.

Discussion

In this study, for the first time, we generated a chrysanthemum microarray based on EST data and demonstrated its application to examine expression profiles of chrysanthemum transcriptome in response to three different viruses. In addition, infection of three different viruses in a same host led to identify virus specific genes and genes commonly regulated by three viruses. In general, the early stage of virus infection does not display any disease symptoms in the infected host plants, but the host initiates to establish the host immune system by manipulating transcriptional machinery via several signaling pathways. Thus, it is interesting to monitor transcriptional changes of the host regulated by different viruses at early stage. Based on microarray data, CMV and PVX led to regulate more genes than TSWV at transcriptional level. However, gene enrichment analysis revealed the number of enriched GO terms for TSWV infection was much higher than those of CMV and PVX. This result indicates that the number of DEGs is not correlated with the functional diverse of identified DEGs. Although several functions associated with viral infection were identified in this study, we discuss more on genes associated with defense/stress related gene, hormone signaling pathway and DNA modification.

Up-regulation of defense/stress genes in response to three viruses

It is natural most commonly identified DEGs were related to defense/stress response and those genes were preferentially up-regulated. Of identified defense/stress related GO terms, endogenous stimulus and organonitrogen compound were significantly enriched. More specifically, genes involving in response to chitin were significantly up-regulated. Chitin is a polymer of N-acetyl-d-glucosamine which an important structural component in the fungal cell walls as well as a well-known elicitor of plant immune system associated with pathogenic fungi (Sánchez-Vallet et al. 2014; Zipfel 2008). Plants do not possess chitin but produce chitin degrading enzymes named as chitinases which disrupt the integrity of fungal cell walls releasing chitin (Sánchez-Vallet et al. 2014). Several studies strongly demonstrated that plants recognize chitin by various pattern recognition receptors such as LysM receptor like kinases to induce pathogen-triggered immunity (PTI) (Miya et al. 2007; Wan et al. 2008). Thus, it seems that virus infection induces a series of host genes which are involved in chitin signaling and plant disease resistance. In addition, a previous study treated rice seedling with chitosan as a fungal elicitor, which is produced by deacetylation of chitin, and found the accumulation of pathogenesis-related proteins. These results showed that chitosan or chitin affect defense response of the rice plants (Agrawal et al. 2002). However, it seems that the genes involving in response to chitin are generally homolog to known many stress responsive genes. For example, several ERF TFs and a WRKY11 TF functioning in chitin response in our study have been frequently identified. Many previous studies demonstrated roles of those TF as diverse viral stress responsive genes (Catoni et al. 2009; García-Marcos et al. 2009; Lu et al. 2012). Of known TF families, many WRKY TFs function in a complex defense signaling acting as both positive and negative regulators in coping with various biotic and abiotic stresses (Pandey and Somssich 2009).

Among identified defense/stress response genes, many genes are involved in resistance against Pseudomonas syringae. Especially, the chrysanthemum WRKY11 like gene (ESTC004883), which was up-regulated by three viruses, was homolog to the loss of Arabidopsis WRKY11 gene led to susceptible to virulent P. syringae (Journot-Catalino et al. 2006). Another example is a CCR4-associated factor 1 (CAF1) homolog 11 gene. This gene was first characterized in the yeast to function in mRNA deadenylation (Tucker et al. 2001). In Arabidopsis thaliana, expression of two genes, AtCAF1a and AtCAF1b, was up-regulated by various stresses, and overexpression of AtCAF1a showed increased resistance against P. syringae infection (Liang et al. 2008). Similarly, the nematode resistance HSPRO2 has showed increased resistance against P. syringae, and expression of HSPRO2 was also up-regulated by P. syringae infection (Murray et al. 2007). Therefore, it seems the chrysanthemum WRKY11, CAF1 and HSPRO2 genes might play a role as regulators against various biotic stresses including viruses and bacteria.

Up-regulation of genes involved in hormone mediated stress signaling pathways in response to three viruses

TFs play an important role in gene expression in higher plants. Of known TFs, for instance, up-regulation of several ERF TFs by three viruses in our study suggests their involvement in virus infection and ethylene mediated signal pathway. It has known that members of AP2/ERF TF family are highly conversed in plant kingdom and many studies demonstrated their functional roles in mediating stress responses and plant developments (Licausi et al. 2013). Our study, at least three ERFs TFs such as ERF3-5 as well as three CAF1 homolog 11 genes are involved in ethylene mediated signaling pathway. Those genes are highly up-regulated in response to all three viruses indicating that this could be a common signaling pathway to establish a basal stress resistance against viral infection. In addition, the nematode resistance HSPRO2 was also involved in basal resistance to P. syringae by negatively regulating jasmonic and ethylene signaling (Murray et al. 2007).

The members of the PP2C family have known to function as regulators in various signal pathways including abscisic acid (ABA) signal transduction (Meyer et al. 1994; Rodriguez 1998). Our study as well as a previous study suggests the function of identified chrysanthemum PP2C homology gene might function in viral stress signaling pathway (Schweighofer et al. 2004). Many members of threonine-protein kinase families are involved in signal pathways in response to various stresses (Afzal et al. 2008). In our study, one of threonine-protein kinases name as OXI1 was up-regulated by three viruses. The expression of Arabidopsis OXI1 gene was induced by H2O2 generating stimuli and is required for active oxygen species processes by activating mitogen activated protein kinases (Rentel et al. 2004). Taken together, it is likely that expression of genes associated with diverse hormone mediated signaling pathways was up-regulated to establish basal immune system in response to diverse viruses. However, the functional role of such genes acting as positive or negative regulator against virus infection should be elucidated in further study.

Down regulation of DNA modification related genes in response to three viruses

It is of interest that TSWV infection led to down-regulate expression of genes involved in DNA metabolic process including DNA replication, chromatin organization, histone modification, cytokinesis, and chromatin and spindle assembly. They are genes encoding mitotic spindle checkpoint protein MAD2, syntaxin-related protein knolle, kinesin-1, cyclin-dependent kinase b2-2, histone, and DNA replication licensing factor MCM3 homolog 1 and they are preferentially targeted to nucleosome and MCM complex. Expression of those genes were strongly down-regulated by not only TSWV but also CMV and PVX. This result suggests their expression is commonly regulated by all three viruses.

A hexameric minichromosome maintenance (MCM) complex composed of six proteins (MCM2-7) are highly conserved in the eukaryotic organisms and play essential roles at the initiation step of DNA synthesis (Tye 1999). MCM proteins are also important for DNA unwinding which might be affected by external stimuli. In higher plants, MCM like protein referred as PROLIFERA (PRL) has been identified in Arabidopsis by gene trap tagging (Springer et al. 1995). The Arabidopsis PRL gene encoding MCM7 protein localizing to the nucleus is required for DNA replication and cytokinesis during early Arabidopsis development (Holding and Springer 2002; Springer et al. 2000). Another study has shown that the expression of pea MCM6 was up-regulated by salinity and cold stresses, and overexpression of MCM6 gene in tobacco plants displayed salinity tolerance (Dang et al. 2011). It is likely that two genes MCM3 (ESTC005732) and MCM7 (ESTC002782), which were strongly down-regulated by three viruses, might involve in viral infection, but detailed functional studies should be carried out.

The MAD2 gene (mitotic arrest deficient 2) encoding spindle assembly protein has been firstly identified in yeast (Li and Murray 1991) and highly conserved in eukaryotic organisms functioning in cell division (Caillaud et al. 2009). The chrysanthemum kinesin-1 gene (ESTC004117) is homologous to Arabidopsis ATK5 gene which is required for early spindle assembly during cell division (Ambrose and Cyr 2007). Strong down-regulation of DNA replication related genes such as MAD2 and kinesin-1 by ssRNA viruses indicates that viruses might inhibit transcription of genes involving in host DNA replication. By contrast, DNA viruses such as geminiviruses with ssDNA genomes replicate through double-stranded intermediates and induce transcripts associated with DNA replication machinery for their viral DNA synthesis (Hanley-Bowdoin et al. 1999). A previous Arabidopsis microarray analysis in response to Cabbage leaf curl virus (CaLCuV), which is a member of geminiviruses, revealed that expression of cell cycle-associated genes was changed by CaLCuV infection (Ascencio-Ibáñez et al. 2008). For instance, genes expressed during S and G2 stages were up-regulated while genes expressed in G1 and M were down-regulated by CaLCuV. This result provided strong evidence that transcriptional reprogramming plant cell cycle by a small ssDNA virus (Hanley-Bowdoin et al. 2013). Thus, the host transcriptional machinery associated with plant cell cycle is affected by both ssRNA and ssDNA viruses, however, transcriptional modulation by ssRNA and ssDNA viruses might be different from each other.

Recent studies suggest that plant gene expression regulated by various stresses sometimes relies on nucleosome histone post-translational modification including DNA methylation (Bruce et al. 2007; Chinnusamy and Zhu 2009). As shown in many previous studies, down-regulation of four chrysanthemum genes related to histone indicates suppression of chromatin organization and DNA methylation. A previous tomato transcriptome analysis upon TSWV infection also identified DEGs involved in chromatin organization and DNA packaging indicating involvement of viral stress in epigenetic changes (Catoni et al. 2009). It seems that TSWV infection causes similar impact of the host transcriptome regardless of host range.

Taken together, we present the first study of transcriptional changes of chrysanthemum plants by a microarray analysis. In addition, the changes of chrysanthemum transcriptome response to three different viruses including CMV, TSWV and PVX showed many novel interesting findings including virus specific gene expression and up-regulation of hormone mediated stress responsive genes and down-regulation of genes associated with DNA replication and histone modification.