Numerous analyses of the virus-induced transcriptome in different host plants have been reported. These studies generally catalogued a set of induced changes to virus infection (reviewed in [1]). The most consistent observation is that viruses, in compatible host-virus interactions, trigger a range of stress responses, including defence response genes. Perturbations of different signalling pathways in the host are induced by change in gene expression involved in some specific interactions between virus and host proteins, as well as gene expression that are not directly involved in these interactions.

The Rice yellow mottle virus (RYMV, Sobemovirus) is the most damaging pathogen of rice, endemic to the Africa where it is widespread. It has a simple genomic organisation with a single-stranded RNA encoding four open reading frames (ORF). Its biology and structure are well known [2, 3]. The diversification as well as the phylogeography of this virus are well documented [4, 5]. Different levels of resistance against the RYMV have been described in the cultivated Asian and African rice species, O. sativa and O. glaberrima [6], and a partial resistance has been observed in some upland japonica rice cultivars [7]. A major recessive resistance gene located on chromosome 4, coding for an eIF(iso)4G protein, has been recently characterized [8]. Genetic basis of RYMV partial resistance has been studied using a doubled haploid population (IR64 (O. s. indica, highly susceptible to RYMV) × Azucena (O. s. japonica, partially resistant to RYMV) and showed seven quantitative trait loci involved in this partial resistance [7, 9]. This virus is thus a very attractive model since genomic studies on this pathogen give a well-known background, and open up horizon for other rice viruses.

The rice genome has been both well mapped genetically and physically [10, 11]. Genomes of the varieties Nipponbare (O. sativa japonica) and 93-11 (O. s. indica) have been integrally sequenced [12–20], and these 12 rice chromosomes are available online [21]. A large number of predicted genes correspond to putative proteins with unknown function and a portion of the genome corresponds to repetitive sequences [22, 23]. Nevertheless, it appears interesting to predict putative function of a gene based on homology searches, thus enabling the identification of candidate genes involved in a particular biochemical pathway, like virus resistance genes involved in the mechanisms developed by plant to resist virus infection.

We have already published a global protein profiling in rice cell suspension infected with RYMV, using 2-D gel electrophoresis. We also identified proteins (defence and stress related proteins, translation and protein turnover, and metabolic proteins) with altered accumulation in response to the virus in susceptible and partially resistant rice cultivars [24]. Moreover our first study on transcriptome of both susceptible and partially resistant rice cultivars infected with RYMV identified variations of gene expression in defence, metabolic and photosynthesis pathways [25]. These results, compared to others studies, indicate that similar events are happening at the protein and the mRNA levels. In this article, we proceed to a fine study of the transcriptome and report the location of 281 RYMV stress responsive sequences (RSR sequences identified by cDNA-AFLP). These RSR genes belong to different functional classes notably defence, photosynthesis pathway, and metabolism and have been mapped onto rice chromosomes using SIM4 subroutine that aligns cDNA fragment against genomic sequence. We present here the first compiled physical map for virus/Rice interaction with the model RYMV/Rice.

Viruses use a variety of strategies to promote their infection in plant. Large studies have already mentioned the various impacts of virus infection [27]. Well-documented modifications of host cells have already been reported [[1] for review]. Suppression of post-transcriptional gene silencing is one of the strategies to promote their infection [28]. Under viral infection, there are many concomitant plant gene variations and cellular changes [29]. The effects of virus infection remain complex and causal relationships are still difficult to establish. Moreover, major traits of host responses and the specificity of a susceptible compared to a tolerant response remain still unclear.

As the response to virus infection involves biological quantitative traits and has multi-factorial inheritance, a global analysis of the multiple genes that are affected during the RYMV infection appears to be a straightforward approach for the study of interactions between virus and host. Using two dimensional electrophoresis and ESTs analysis, we have already established a complex pattern of protein and gene regulations, and identified numerous changes in cell balance involved in host response ([25, 24]; data not shown). We identified at both transcriptional and post transcriptional levels some variations in the photosynthesis pathway and metabolism as potential clues of the susceptible response. Here we analyzed the transcript profiling using the cDNA-AFLP technique on the same two rice cultivars: IR64 (O. s. indica) highly susceptible to the RYMV, and Azucena (O. s. japonica) tolerant to the RYMV. This technique has been shown to be a powerful gel-based genome-scaled transcript profiling [30] and allowed us to identify a large number of genes involved in rice-RYMV interaction during the first week of infection.

We observed more regulations of host gene expression for the tolerant cultivar than for the susceptible one (Figure 3A). We noticed an up-regulation of expression of a gene encoding a photosystem II stability assembly factor (RSR sequences DQ884135 and DQ884136) for the cultivar IR64 during the kinetic of RYMV infection, and a down regulation of expression of gene encoding a photosystem II phosphoprotein for the cultivar Azucena under RYMV infection (DQ884058). These results were consistent with our previous study on ESTs which mentioned down-regulations in photosynthesis pathway for the partially resistant cultivar and an up-regulation for the susceptible cultivar [25]. Nevertheless, these two approaches (ESTs analysis and cDNA-AFLP technique) were too different for a complete overlapping of the results, and our different experiments did not underline same aspects of the processes. The ESTs approach was based on large data sets analysis and revealed global patterns, whereas cDNA-AFLP technique allowed fine analysis of genes, even with low expression, and variations of expression can be validated by RT-PCR. Moreover, the experiments were not carried out at the same time, and variations in the conditions of experiments could have thus occurred. A general property of viral infections might cause increasing accumulation of host gene products as part of a stress response, especially HSPs genes. [1]. We observed an up regulation of the expression of HSP70 protein (gb|DQ883891) in wounded leaves and virus infected Azucena leaves harvested at 3 dpi. Many viruses elicit the expression of HSP70 genes (and other heat shock genes) and in some case HSPs have been shown to facilitate viral infection [31]. It seems that most viruses trigger these generic stresslike or defence responses which occur in the absence of typical gene-for-gene or resistance gene-avirulence gene interactions. RYMV infection also induced the expression of WRKY DNA binding protein (gb|DQ884128) in virus infected IR64 leaves at 7dpi. WRKY6, a member of the WRKY family, has been identified by Whitham as elicited by viruses and other pathogen infections [1]. The RYMV caused both specific and general changes in host gene expression and such examples show that this virus elicits expression of common genes induced by other viruses.

This study is the first compiled map on RYMV/Rice interaction. We physically mapped 281 RSR genes using Sim4 [32] on the rice physical map, with respect to intron/exon structure. The map is available, (see Availability and requirements section). We also mapped the ESTs and proteins previously identified [24, 25] and the position of the 7 QTLs involved in the partial resistance [7]. The ESTs or cDNA-AFLP analysis revealed changes in host gene expression as a consequence of the infection, whereas the QTLs are involved in a direct response of the tolerance of the host. We did not expect to find any adding up, as the ESTs analysis revealed largely expressed genes, and cDNA-AFLP technique showed up some genes with weak expression level, moreover we do not know if the QTLs involved in partially resistance or tolerance are induced or repressed.

The statistical analysis of the distribution of the RSR sequences showed non uniform distribution of RSR sequences on chromosomes 4 and 10 and thus might underline the hypothesis of RSR sequences clusters of deregulation. We did not observe any cluster regrouping RSR sequences with a similar function ("functional cluster"). But we did not identify all the genes involved in the response to RYMV infection and a deeper study could identify more clusters (especially functional ones) at a microlevel.

Some sequences from IR64 and Azucena colocalized (Table 2). The pattern of gene expression under RYMV infection has been confirmed by quantitative RT-PCR for some RSR genes and showed some differences between the two cultivars. Such RSR genes might be involved in the specificity of the host response. One of these genes encodes an ABC transporter (DQ883962 and DQ884118 RSR sequences). This gene is highly repressed at 5 dpi in the susceptible cultivar and over expressed in wounded leaves of the partially resistant cultivar (Figure 4, A2). ABC transporter genes were the most predominant genes identified by cDNA-AFLP (6 RSR sequences for Azucena and 4 RSR sequences for IR64) and thus might play an important role in plant response to RYMV infection. Such transporters are involved in the membrane transport of a wide range of structurally and functionally unrelated compounds, such as glutathione conjugates, lipids, inorganic acids, peptides, secondary metabolism, toxin and drugs [33]. It has been shown that some ABC transporter, like the AtPDR8, could be a key factor controlling the extent of cell death in the defence response [34], and conferring heavy metal resistance [35]. We also observed a variation of gene expression of a sphingosine-1-phosphate lyase in the two cultivars (Table 2, DQ883945 and DQ884107 RSR sequences). We observed an important decrease of gene expression in RYMV infected leaves of the susceptible cultivar as soon as 2 dpi (Figure 4, A14). Lipids influence pathogenesis and resistance mechanisms associated with plant-microbe interactions, and sphingolipids act as elicitors of defence response [36]. Such families of genes with a down regulation of expression during the kinetic might be targets of small RNAs and deeper study should be undertaken to analyze the potential effects of small RNAs on their regulation of expression. Finally, we noticed the down regulation of 5 different RSR sequences encoding retroelements (DQ884156, DQ883989, DQ883835, DQ883911, and DQ883916). This negative regulation occurred during the whole kinetic in Azucena infected leaves. A deeper transcriptomic analysis is now carried on to determine the relation between the virus infection and such down regulation of retroelement.

This study presents for the first time a compiled physical map on rice/virus interaction. It gives a strong basis for future studies on RYMV/Rice interaction and for other rice pathogens, as CMAP tool allows the combination of maps. We identified variations of host responses to RYMV infection in susceptible and partially resistant cultivars and we showed that the phenomenon of tolerance to the RYMV involves regulations at different cellular levels. Some genes appeared to be potentially important during the response of partially resistant and susceptible cultivars and it would be interesting to study the variation of their expression in a near isogenic line with only the major QTLs of resistance, this would allow avoiding any bias due to the difference of the genetic background of the 2 cultivars. Moreover, it could be interesting to analyze the variation of expression (using quantitative RT-PCR) of such genes in O. glaberrima species with different levels of resistance assuming that the sequencing is available. This would free further studies from inter-species variations of response to virus infection.

http://cmap.bioinfo.mpl.ird.fr/cmap/: Physical map of 281 RSR sequences. On the left of each chromosome, start positions of RSR sequences were in bp *100000, and the ESTs and proteins identified as candidates involved in host response [24, 25]. On the left, we mentioned the position of the 7 QTLs involved in the partial resistance [7].