Introduction

Plants have evolved multiple and inducible mechanisms to defend themselves from biotic stress. While some of these mechanisms are common to all types of assaults, such as tissue reinforcement and the production of anti-microbial metabolites, others are tailored to combating specific types of pathogen interactions (McDowell and Dangl 2000). The most well studied mechanism is directed toward non-necrotrophic pathogens and is mediated by the specific recognition of pathogen avirulence (Avr) protein by a corresponding receptor encoded by a plant resistance (R) gene. This leads to an immediate local defense response, known as the hypersensitive response (HR), which is followed by down-stream signaling that activates defense response genes, subsequently resulting in a persistent, broad-spectrum resistance termed systemic acquired resistance (SAR; Rathjen and Moffett 2003). A different form of systemic resistance, referred to as induced systemic resistance (ISR), is mediated by interactions with non-pathogenic organisms such as Pseudomonas fluorescens (Pieterse et al. 1998; Feys and Parker 2000). Sclerotinia sclerotiorum is a necrotrophic fungal pathogen, which causes severe disease in most agronomical important crops, such as oilseed (Brassica napus), soybean, and sunflower. The defense responses mounted against this pathogen appear to be multi-factorial and remain poorly understood.

Three distinct, yet interconnected signaling pathways, mediated by salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) are involved in the plant disease response (Dong 1998; Feys and Parker 2000; McDowell and Dangl 2000; Martinez et al. 2001). These signaling pathways are differentially regulated and act individually, synergistically, or antagonistically depending on the pathogen involved (Glazebrook 2005; Kachroo and Kachroo 2007). The coordinate regulation of defense responses has been revealed using microarray analysis and examining Arabidopsis mutants in specific signaling pathways (Maleck et al. 2000; Cheong et al. 2002). Although JA/ET-dependent pathways were reported to be involved in the plant response against necrotrophic pathogens, the gene expression changes and the corresponding signal transduction pathways associated with S. sclerotiorum infection have not been reported.

Microarray analysis is a powerful technology that allows the simultaneous assessment of the expression of thousands of genes, even entire genomes (e.g., dissection of signal transduction pathways involved in plant defense). The transcriptome of Arabidopsis has been examined during SAR using a variety of microarray technologies (Maleck et al. 2000; Schenk et al. 2000). The quantitative nature of gene expression in Arabidopsis during compatible and incompatible interactions with the bacterial pathogen Pseudomonas syringae has revealed a common set of responsive genes (Tao et al. 2003). Marathe et al. (2004) reported that the “resistome” of Arabidopsis in response to Cucumber Mosaic Virus infection involved 80 genes commonly induced by both viral and bacterial pathogens.

The advances that have been made in Arabidopsis genomics have great potential to be used in closely related (phylogenetically) plant species. Brassica and Arabidopsis belong to the same family Brassicaceae and the levels of genome synteny regions between Arabidopsis and Brassica species have been reported (Cavell et al. 1998; Parkin et al. 2005). Girke et al. (2000) used an Arabidopsis cDNA microarray to investigate gene expression in the seeds of B. napus for the purpose of assessing cross-species use of microarray technology. Similarly, Horvath et al. (2003) reported that an Arabidopsis cDNA microarray may also be used in a more distantly related plant species such as leafy spurge, poplar, and wild oat. In these cases, the arrays comprised only a portion of the Arabidopsis genome. Also, cDNA microarray may not be able to distinguish among members of multigene families, related genes, and differentially spliced genes. Another limitation of using cDNA microarrays is that arrayed cDNA clones may not match their respective EST sequences found in databases due to errors in sample tracking.

Synthetic oligonucleotides (oligos) offer an attractive alternative to cDNA amplicons due to the lack of amplification and the errors associated with sample tracking. The uniform length and concentration of oligos provide more precise estimation of gene expression levels (Lee et al. 2004). Oligos can be synthesized to match any known or predicted gene sequence. Specifically, 70-mer oligos for all ∼26,000 predicted genes within the Arabidopsis genome have been developed (http://www.operon.com/arrays/omad.php). Lee et al. (2004) compared the Arabidopsis oligo-arrays with a cDNA array, and found that oligos have potentially more sensitivity for detecting changes in gene expression. More importantly, they found that oligo arrays designed from Arabidopsis gene sequences are effective for use in Brassica species (B. oleracea).

Quantitative trait loci (QTL) for resistance to S. sclerotiorum were previously mapped in a segregating population of B. napus derived from the partially resistant parent, RV289 (Zhao et al. 2006). In this study, we used an oligonucleotide microarray representing 26,000 genes from Arabidopsis for the purpose of investigating changes in gene expression in the partially resistant line RV289 and a highly susceptible line Stellar. The objectives of our study were to examine gene expression changes of B. napus at different time points after infection, to identify commonly and specifically expressed genes in the two genotypes and to classify signal transduction pathways during the early stage of infection. Certain genes that were up-regulated early and specifically expressed in the partially resistant line co-localized with the map positions of previously identified QTL. Our results provide a better understanding of the complex interactions between plants and necrotrophic pathogens, such as S. sclerotiorum, and the identification of candidate genes for further investigation.

Materials and methods

Plant materials

Two B. napus genotypes were used, the partially resistant line RV289 and the highly susceptible cultivar Stellar. RV289 is a single plant selected from a Chinese winter line Hua dbl-2, provided by Dr. J. Meng (Huazhong Agricultural University, China). A doubled haploid line derived from Stellar, a Canadian spring cultivar, was provided by Dr. R. Scarth (University of Manitoba, Winnipeg, Canada). These two lines have been used to build molecular linkage maps (Ferreira et al. 1994; Udall et al. 2005) and to locate QTLs for resistance to S. sclerotiorum (Zhao et al. 2006).

Plant growth, inoculation, and tissue harvest

Seeds were planted in 8-cm pots containing Scott’s Metro-mix 336p soil (Scott’s-Sierra Horticultural Products Co., Marysville, OH, USA) in flats of 24 plants and arranged as a randomized complete block design. The plants were grown in a controlled growth chamber under 450 μmol m−2 s−1 light intensity with 16-h photoperiod provided by fluorescent and incandescent lamps. The temperature was maintained at a constant 21°C. Plants were watered automatically with 1/4 Hoagland’s solution daily for 11 min and relative humidity was maintained at 60%.

Inoculation was performed on 3-week-old seedlings as previously reported (Zhao et al. 2004). Briefly, petioles of the third fully expanded leaf were severed 2.5 cm from the stem, and the open-end of the petiole was infected with S. sclerotiorum mycelium. The inoculated petiole and a 1-cm section of stem adjacent to the inoculation point were harvested using a razor blade at 12, 24, and 48 h post-inoculation (hpi) with mock-inoculated plants as a control. Eight individual plants were pooled for one RNA sample. Another RNA sample was collected with separated sets of growing plants and inoculated with separated cultured colonies as a second biological replicate. Harvested tissues were immediately placed in liquid nitrogen and stored at −80°C.

RNA preparation, probe labeling, slide hybridization, and scanning

Total RNA extraction, mRNA purification and probe preparation were conducted according to Wang et al. (2005). Labeled probes were hybridized to slides at 58°C for ∼16 h in the dark. After hybridization, the slides were washed four times at room temperature; once with 2 ×  SSC, 0.2% SDS for 4 min, followed by two washes in 0.2 ×  SSC for 2 min each and a final wash in 0.05 ×  SSC for 4 min. The slides were rinsed in 0.05 ×  SSC several times and dried by centrifugation at 850 rpm for 5 min at room temperature. A Genepix 400B scanner (Axon Instruments Inc., Union City, CA, USA) was used to scan and quantify the intensity of each feature using GenePix Pro5.0 software.

Arabidopsis 70-mer oligo microarray and hybridization design

The 26,090 oligonucleotides (see Wang et al. 2005 for oligo design) were spotted on SuperAmine slides (TeleChem International Inc., Sunnyvale, CA, USA) using an OmniGrid Accent microarrayer (Genemachines, San Carlos, CA, USA). Each slide contained 27,648 features (26,090 oligo-features plus controls). The printed slides were UV cross-linked, baked, and post-treated according to Wang et al. (2005). Gene names and GenBank accession numbers of the 26,090 oligos and their corresponding cDNA sequences can be found at http://www.operon.com/arrays/omad.php. Duplicated dye swaps were used in our experiments to compare gene expression changes. Each dye-swap consisted of two slide hybridizations: (1) two samples labeled with Cy3-dCTP and Cy5-dCTP, respectively, and (2) the same two samples reversely labeled with Cy3-dCTP and Cy5-dCTP. Further details of the experimental design and analysis methods can be found in Craig et al. (2003) and Wang et al. (2005).

Data analysis

The background-corrected intensity values were used as data for the statistical analysis. A robust local regression (the lowess function in the R package; Cleveland 1979) was used to normalize the background corrected data. Analysis of variance (ANOVA) models were employed to detect differentially expressed genes using the normalized data. Two different analyses were conducted to detect significant genes. The first analysis assumes a common variance for all genes (Craig et al. 2003; Jiang 2004), while the second analysis is based on a per-gene-variance assumption (Jiang 2004).

The ANOVA model under the common variance assumption is

$$ {\text{log}}{\left( {y_{{ijkgr}} } \right)} = \mu + A_{i} + D_{j} + T_{k} + G_{g} + AG_{{ig}} + DG_{{jg}} + TG_{{kg}} + \varepsilon _{{ijkr}} , $$

where log(y ijkgr ) is the background corrected and normalized natural logarithm of the intensity of the r th replicate of gene g on array i, with dye j and treatment/condition k; μ is the average natural logarithm of gene intensity over all the genes, arrays and dyes. A, D, T, and G are the array, dye, treatment and gene main effects, respectively, and \( \varepsilon _{{ijkr}} \) is normally distributed with mean zero. Differential expression is tested using a null hypothesis that reflects the average treatment. Specifically, \( H_{{\text{0}}} :T_{k} + TG_{{kg}} = T_{{k'}} + TG_{{k'g}} \) is tested for each gene g.

When the variance of each unique gene is acknowledged, a per-gene-variance ANOVA model is employed:

$$ {\text{log}}{\left( {y_{{ijkgr}} } \right)} = \mu _{g} + A_{{ig}} + D_{{jg}} + T_{{kg}} + \varepsilon _{{ijkgr}} . $$

μ, A, D, and T are the gene specific average natural logarithm of the intensity, array, dye, and treatment effects, respectively. Since the subscript g in every term is suppressed, then we have the ANOVA model for a single gene g:

$$ {\text{log}}{\left( {y_{{ijkr}} } \right)} = \mu + A_{i} + D_{j} + T_{k} + \varepsilon _{{ijkr}} . $$

Our justification for using both the common gene and individual gene variance assumption is as follows: if all genes have the same variation across replicates, the common-variance approach is more powerful than the per-gene-variance approach as it has more degrees of freedom for testing. However, the assumption of a common variance across all genes does not hold, i.e., the variation is different across genes. Therefore, the per-gene-variance assumption and approach is more plausible than the common-variance assumption. Unfortunately, the degrees of freedom in the test for a per-gene-variance approach is smaller since the number of biological replicates is limited. As a compromise between a false common variance assumption and a test based on a small number of biological replicates per gene, both approaches are employed in independent analyses to detect the statistically significant differentially expressed genes.

Because we are testing more than 26,000 features for differential expression, it is necessary to adjust the Type-I error rate. False Discovery Rate (FDR; Benjamini and Hochberg 1995) was used for multiple comparison corrections. The significance level α was chosen as 0.05.

RT-PCR and Northern-blot analysis

Total RNA (12 μg), the same as that used in the microarray experiment, was fractionated in a 1.2% formaldehyde agarose gel, and then transferred to a Hybond-N+ membrane. Probes were labeled with 32P-dCTP using reverse-transcriptase polymerase chain reaction (RT-PCR) amplified cDNA fragments of B. napus with primers designed based on the conserved regions of Arabidopsis gene and had a product size of 200-500 nucleotides. Only a subset of the genes derived from the integration of differentially expressed genes in RV289 at 24 hpi and which were co-localized with positions of previously mapped QTL was selected for verification of the microarray. Primer pairs used in the RT-PCR were shown in Table 1. Nine probes derived from amplification of a single and distinct band from B. napus cDNA were used in the northern hybridization. The hybridization was performed at 60°C using Church buffer (0.5 M Na3PO4, 2.5% SDS). Blots were washed twice in 0.5 ×  SSC, 0.5% SDS at room temperature for 5 min and then twice in 0.2 ×  SSC, 0.2% SDS at 65°C each for 5 min. The blots were exposed to an X-ray film at −80°C for 1–3 days depending on the signal intensity.

Table 1 RT-PCR primers used in the verification of gene expression for the 12 candidate genes
Full size table

Results

Gene expression profiles in response to S. sclerotiorum

We used an Arabidopsis 70-mer oligo-microarray representing 26,000 annotated genes for the purpose of investigating the transcriptional differences of genes in B. napus after S. sclerotiorum infection. Two ANOVA models, each reflecting different model assumptions, were employed to estimate and test for differential expression between genotypes at each of three time points. The expression ratio was determined by mRNA abundance in mycelium-inoculated versus mock-inoculated plants to eliminate the circadian clock and plant growth stage effects, since the samples were collected at different times of day. Using the results from both statistical models we report the statistically significant differentially expressed genes detected by FDR, and examine the intersection (Table 2) of these two analyses. The statistically significant differentially expressed genes across genotypes at the same time point, as well as those statistically significant genes across time within a genotype were compared using the resulting gene lists.

Table 2 Number of statistically significant differentially expressed genes detected by common variance and per-gene-variance at significance level of 0.05 using FDR correction at three different time points post-inoculation in RV289 and Stellar
Full size table

Sclerotinia sclerotiorum infection altered the expression of ∼7% of the 26,090 genes (Table 2, Fig. 1). At the 12 hpi, we detected only one gene (At4g02380, which encodes a late embryogenesis abundant family protein) that showed a statistically significant differential expression between RV289 and Stellar. The changes of gene expression at this time point could be too low to be detected by the microarray experiment; therefore, only those genes differentially expressed at 24 and 48 h were reported. A total of 686 and 1,547 genes were differentially expressed in RV289 and Stellar, respectively. Most of the statistically significant genes exhibited a threefold to tenfold change (Fig. 2), although some showed changes in expression of over 100-fold. Complete lists of differentially expressed genes and their corresponding P-values at each time point within each genotype are available in the supplementary information (Supplementary Tables 1 to 4).

Fig. 1
figure 1

a and b Venn diagrams showing the number of overlapping and specific genes which had significantly differential expression changes at 24 and 48 h post-inoculation in Brassica napus. RV indicates the partially resistant genotype RV289 and ST denotes the susceptible genotype Stellar. Results were based on the number of significantly differentially expressed genes detected by ANOVA from eight slides involving two biological and two technical replicates. Number of genes specifically regulated within each genotype and time point was generated from a direct comparison of resulting gene lists from each genotype and time point. a Comparisons within each genotype at two time points. b Comparisons between genotypes at the same time point. A complete list of genes from all 26k oligo arrays with the ratios and corresponding P-values from the data analysis are accessible from the supplementary tables

Full size image
Fig. 2
figure 2

a and b Number of significantly differentially expressed genes and the levels of expression at 24 and 48 h post-inoculation in RV289 (a) and Stellar (b). Number of genes differentially regulated within each genotype and time point was generated from ANOVA analysis of eight slides involving two biological and two technical replications. Levels of expression were determined by the mRNA transcript abundance in mycelium-inoculated plants versus mock-inoculated controls

Full size image

Co-regulation patterns of significantly differentially expressed genes

Within each genotype, the number of differentially expressed genes increased over time (Fig. 1). In RV289, the number of genes that were significantly differentially expressed increased from 163 to 370 over the 24 to 48 hpi, whereas in Stellar the number increased from 53 genes at 24 hpi to 1,295 statistically significant genes at 48 hpi. A similar number of genes were found at both time points within each genotype (Fig. 1a); however, the expression levels increased with infection time for most of the genes that over-lapped in both time points in the partially resistant genotype RV289; while in the susceptible genotype Stellar most of the overlapping genes decreased their expression levels over time.

Genes detected at the same time point (24–48 h) were compared between genotypes and grouped as being either genotype-specific or common (Fig. 1b). At 24 hpi, RV289 had more specifically expressed genes than Stellar, while at 48 hpi the opposite was observed. We also found that 199 and 446 genes overlapped between RV289 and Stellar at 24 and 48 hpi, respectively (Fig. 1b). The overlapping genes had similar patterns of expression, i.e., were either activated or repressed in both genotypes and no significant difference in expression levels was observed between the two genotypes in a two-sample t-test (188 of 199 at 24 hpi, and 426 of 446 at 48 hpi).

Functional classification of the significantly differentially expressed genes

The differentially expressed genes detected at 24 hpi in RV289 were categorized into 13 groups according to the putative function of each gene, which was inferred from metabolic processes known to be related to each gene in Arabidopsis (Fig. 3). These categories included pathogenesis-related (PR) proteins, proteins involved in defense-related pathways, the oxidative burst, hormonal responses, transport, abiotic stress, cell maintenance and development, cell wall modification, protein degradation, secondary metabolism, and signal transduction (Table 3). A complete list of the functional groups of genes is shown in supplementary Table 5.

Fig. 3
figure 3

Functional classification of genes which displayed significantly differential expression in RV289 at 24 h post-inoculation with S. sclerotiorum

Full size image
Table 3 A subset of functional genes in RV289 at 24 hpi from the microarray data
Full size table

As expected, most of the genes (278 of 316) were up-regulated; in particular those involved in the defense-related pathways, hormonal responses, and molecule transporters. Most of the down-regulated genes included those encoding proteinases, proteins involved in cell maintenance and development. This may indicate that immediately after pathogen infection, defense-related genes leading to the generation of antimicrobial compounds to block pathogen growth are activated and that nutrient transport to the infection site to meet this increased energy demand is accelerated. Interestingly, some genes with similar functions were up-regulated while others were down-regulated; one obvious case being the protease inhibitors. Down-regulation of protease inhibitors might indicate that these genes were negative regulators in response to S. sclerotiorum.

We identified groups of genes involved in signal recognition and transduction pathways that were differently expressed in response to S. sclerotiorum infection. The first group included genes responsive to JA, ET, gibberellin, and auxin. Most of these genes were up-regulated, except for one gene (At1g45015) which is putatively involved in lipid recognition in the JA pathway in Arabidopsis was down-regulated. The second group of genes included proteases, phosphatases, kinases, and calcium-binding proteins which were shown previously to be involved in defense responses (Cheong et al. 2002; Ali et al. 2003; Marathe et al. 2004). Proteases such as aspartyl protease and cysteine proteases have been shown to mediate plant-pathogen interactions (Marathe et al. 2004; Xia et al. 2004); however, most genes in the protease pathway detected in our study were down-regulated. The third group contained genes encoding transcription factors, such as the AP2, bZIP (HY5), zinc finger, basic helix-loop-helix, myb (MYB55), WRKY, NAC, and homeobox-leucine zipper family proteins. These transcription factors accounted for half of the signal transduction proteins detected in our study. We also found other groups of genes encoding transporters, including nuclear, protein, sugar, monosaccharide, amino acid, ABC, and magnesium transporters.

Integration of the significantly differentially expressed genes with mapped QTLs

Several QTL with the resistance alleles from RV289 were mapped to a B. napus linkage map (Zhao et al. 2006). This allowed us to determine whether genes uniquely induced in the partially resistant line RV289 were located in the QTL regions. We used a comparative mapping method to align QTL locations in B. napus to the collinear regions in Arabidopsis (Parkin et al. 2005). Sequences of the flanking RFLP markers in QTL regions spanning 10–15 cM on linkage groups N2/N12, N5, and N14 were used to query the Arabidopsis genome (http://www.arabidopsis.org/Blast) using BLAST (Altschul et al. 1990) and were found to be collinear with Arabidopsis chromosomes 3, 5, and 2 (Fig. 4). Twelve genes located in these regions and which also displayed significantly differential expression changes between RV289 and Stellar at 24 hpi were identified. Annotations of the 12 genes were shown in Table 4.

Fig. 4
figure 4

Integration of the mapped QTL regions on linkage groups of N2, N12, N5, and N14 in B. napus and the collinear chromosome positions in Arabidopsis chromosomes 2 (At 2), 3 (At 3), and 5 (At 5). Black boxes indicate QTL intervals in corresponding linkage groups of B. napus. Slash marks on vertical lines indicate relative positions of RFLP marker loci flanking QTL regions. Alignments between RFLP marker loci and corresponding Arabidopsis genes are shown. Twelve candidate genes localized in these collinear regions are shown in black and italics

Full size image
Table 4 Annotation of 12 candidate genes identified from the integration of differentially expressed genes in RV289 at 24 hpi and mapped QTL
Full size table

Verification of the microarray data and the identification of candidate genes

Microarray analysis revealed hundreds of genes, which were statistically significant in differential expression. In order to narrow down the number of genes in the verification experiment, we only used 12 genes derived from the integration of gene expression and previously mapped QTL in the following Northern blot analysis (Table 1). Homologous fragments of the 12 genes from B. napus were amplified by PCR using primers designed from conserved regions of Arabidopsis. Nine of the twelve primer pairs produced a single band and were used in the subsequent verification experiments (At3g46080 did not amplify; At5g15890 and At5g62150 had more than one band). Northern blot analysis showed that seven of the nine genes were up-regulated in the inoculated samples compared with the control and that transcript abundance increased with time (Fig. 5, only five genes are shown, and the other two genes, At2g32190 and At5g13080, were hybridized in different blots). Transcription of these five genes was present at higher levels in RV289 than in Stellar at 24 hpi. Transcripts at 12 hpi were undetectable using Northern blot analysis, confirming the results from the microarray data.

Fig. 5
figure 5

Northern-blot analysis of selected genes revealed from the integration of microarray data and mapped QTL. Total RNA of 12 μg from RV289 and Stellar at 12, 24, and 48 hpi was separated on 1.2% formamide gel. Probes were cDNA fragments generated from PCR amplification of B. napus cDNA using primers designed from conserved regions of the target Arabidopsis genes revealed from microarray results. + mycelium-inoculated samples,—mock-inoculated controls. EtBr stained total RNA was shown as a loading control. Only RNA samples from one biological replicate was shown. The second biological replicate gave the same result

Full size image

The mRNA abundance of the gene homologous of At2g38470 was examined by quantitative RT-PCR analysis in RV289 and Stellar at 24 hpi. The expression ratio was higher in RV289 than in Stellar after corrected PCR efficiency and normalized using the gene Actin (AF111812) in the treatments relative to the mock-inoculated controls (data not shown).

Discussion

We used Arabidopsis arrays to examine gene expression changes in B. napus in response to S. sclerotiorum. RV289 is a partially resistant line and had revealed a significantly reduced disease symptom compared with Stellar (Zhao et al. 2004). In this study, we observed temporal and quantitative differences in gene expression between two genotypes of B. napus in response to S. sclerotiorum infection. RV289 tended to induce the expression of genes associated with defense mechanisms earlier (24 hpi) than the susceptible genotype, with many of these same genes being expressed later (48 hpi) in the susceptible line Stellar. This suggests that the difference in the defense response of the two genotypes to S. sclerotiorum may due to temporal and quantitative differences in gene expression. This seems plausible since our previous screening data indicate that the defense response is multi-genic and the partially resistance phenotype was observed as restricting lesion progression rather than preventing infection.

Several plant hormones are important molecules in regulating plant defense response. SA is involved in plant R gene-mediated disease resistance to biotrophic pathogens, whereas JA and ET are implicated in the response to necrotrophs (McDowell and Dangl 2000). Auxin was also reported to be involved in the plant response to various stresses and is interconnected with these disease resistance signaling pathways (Cheong et al. 2002; Dowd et al. 2004; Reymond et al. 2004). In our study, we detected few genes known to be specially regulated by SA but several that are specially regulated by JA and ET in the response to S. sclerotiorum infection. This indicates that response to S. sclerotiorum is SA-independent, and mainly JA and ET-dependent, although this hypothesis requires further examination. We also found that auxin-responsive genes were up-regulated after S. sclerotiorum infection, suggesting the concomitant activation of both auxin and JA/ET pathways in the response to S. sclerotiorum infection.

Transcription factors play important roles in gene regulation (Riechmann et al. 2000). AP2, WRKY and zinc finger proteins have been reported to be differentially regulated by various stresses (Chen et al. 2002). Arabidopsis transcription factor AtWRKY33 (At2g38470) was recently reported to be required for resistance to necrotrophic fungal pathogens Botrytis cinerea and Alternaria brassicicola (Zheng et al. 2006). In this study, at least two WRKY family transcription factors homologous to At2g38470 and At5g13080 were identified. These two genes were co-localized with the positions of previously mapped QTL. The expression levels of these two genes were verified by Northern-blot analysis, and for At2g38470, also confirmed by quantitative RT-PCR. These two genes were also highly induced after S. sclerotiorum infection in another partially resistant B. napus line with a lineage similar to RV289 (D. H. Hegedus, Saskatoon Research Center, Canada, personal communication). WRKY transcription factors regulate down-stream gene expression via combining the W box of the promoter regions of the target genes. How the transcription factors identified in this study regulate the response to S. sclerotiorum infection remains to be elucidated by the identification and characterization of the target genes and its regulatory sequences.

Two types of transporters have been reported to function in the defense response, the monosaccharide transporter gene, STP4 (Truernit et al. 1996; Fotopoulos et al. 2003), and the Arabidopsis ATP transporter, AtPDR12 (Campbell et al. 2003; Jasinski et al. 2003). These reports suggested a model for increased demand for carbohydrate and energy caused by environmental stress. We identified several genes encoding putative transporters including a monosaccharide transporter, an ABC transporter, amino acid transporters and sugar transporters that were highly up-regulated in response to this infection. This is consistent with the previous observations on the defense functions of monosaccharide and ATP transporters, and may also suggest that besides monosaccharide transporter and ABC transporter, other types of transporters are demanded in the infection point.

Gene expression profile studies using microarray usually result in hundreds of genes showing up- or down-regulation. Verification of hundreds of genes is not practical. Thus, which and how many genes to choose has been a debate for researchers. In this study, we used a comparative mapping method to integrate the differentially expressed genes with the previously mapped QTL to narrow down the number of genes. Four genomic regions on N2, N12, N5, and N14 from B. napus containing resistance alleles from the partially resistant parent RV289 correspond to two regions in Arabidopsis due to a transposition of the N2 region containing the QTL onto N12 and homeology between N5 and N14 in the regions of the QTLs (Udall et al. 2005; Zhao et al. 2006). Conserved blocks between B. napus and Arabidopsis were found between N2/N12 and At3/At5, and N5/N14 and At2 by blast analysis. This result agrees with the detailed alignment reported by Parkin et al. (2005). Candidate genes were chosen based on two criteria: (1) location in these collinear regions of Arabidopsis, and (2) a significant expression change in RV289 at 24 hpi and a consistent expression change at 48 hpi. Twelve candidate genes were selected based on these criteria from the integration. Northern-blot analysis verified seven of the nine genes which were amplified as a single and distinct band from B. napus. A quantitative RT-PCR was also used to verify the expression level of these seven genes. In general, the quantitative PCR gave higher expression values than the microarray data. However, only three of the seven genes were verified, including a homolog of At2g38470. The main reason for this is that the PCR primers were designed based on Arabidopsis sequence and in B. napus genome, each homologous single gene was assumed to have at least six copies. Thus, the PCR amplification is not specific enough. Further work is needed to design and test primers for gene specificity. Cross-amplification should be considered in verification of homologous genes or members of a gene family. Products of these seven genes that were confirmed by Northern-blot analysis include hairpin-induced family protein (YLS9), WRKY and zinc finger proteins, plant cell wall related proteins, and ET-responsive proteins. Although functions of these genes need further characterization, our data provide new insights into the mechanism of defense response to S. sclerotiorum and suggest that this defense might involve various proteins and signaling pathways. Early induction of WRKY and zinc finger proteins and plant cell wall related proteins may play important roles in enhancing resistance to S. sclerotiorum disease.