Viroids and viruses depend on host components to complete their life cycle and thus interfere with host processes, such as host gene expression during infection. The identification of genes with altered expression in response to a certain virus/viroid infection can provide clues about the viral life cycle and requirements which need to be met by the host cell. Moreover, a better understanding of host responses to infection facilitates the development of methods to control pathogen infection. To date, the effects of viroid infection on the host transcriptome have been reported for four viroids. These studies used PCR-based cDNA library subtraction ; differential display RT-PCR and quantitative real-time RT-PCR [31, 32] or cDNA microarray [17, 33] to detect host gene expression changes. Currently, a number of studies explore gene expression changes in virus-infected plants using DNA microarrays. All these studies involve a given virus and a model host plant, usually A. thaliana or N. benthamiana. In the present work, we collected peach fruits, a natural host of both PNRSV and PLMVd, at development stage (S4) and studied changes in their transcriptome upon infection with the two pathogens. Today, no array including the whole peach fruit genome is available. Thus, we used a cDNA microarray containing probes for 4,261 genes expressed during chilling injury (CI) development .
During infection, the concentration of PNRSV and PLMVd in fruits is significantly higher (100:1) than in leaves [35, 36]. In addition, the occurrence of peach trees doubly infected with both, PNRSV and PLMVd in the field is considerable (e.g. Fiore et al., unpublished data); . Here, we compared transcript changes in asymptomatic singly and doubly PNRSV or PLMVd infected peach fruits. Overall, our gene expression analysis revealed a relatively low number of differentially expressed genes in single PNRSV infections compared to the healthy control plants (82 genes). This is consistent with the results obtained by Dardick  in which only 89 genes were significantly differentially expressed in PNRSV-infected N. benthamiana plants compared to mock controls. Peach fruits infected with PLMVd also mounted only a moderate response at transcript level (16 genes). The PLMVd variants detected in our study are all considered as latent. In future studies it would be interesting to compare the host transcripts changes caused by variants causing the “calico”  syndrome with those caused by milder variants. PNRSV infected samples were also asymptomatic and all the sequence variants recovered belonged to the PV32 group . Remarkably, in the doubly infected peach fruits a significant synergistic effect on the host transcriptome was observed. Considering the three different infections, the total number of genes with significantly altered expression (at least 1.5 fold change in expression level) was 783, which represents 18% of the whole array. From these 783 genes and eliminating those with significant expression changes occurring in common upon the three scenarios, 627 have orthologs in Arabidopsis (Additional file 3: Table S2). Among those genes 211 exhibited induced and 416 reduced expression. Thus, the total number of genes with reduced expression exceeded that of genes with induced expression by two-fold. This is surprising as comparative analysis of the Arabidopsis transcriptome during compatible interaction with plant viruses  revealed that there was a greater number of up-regulated than repressed genes in the course of viral pathogenesis. However, the authors demonstrated that each virus-host interaction is unique in terms of the genes with altered expression levels and to find a common pattern among different viruses is difficult . In addition, in one of the few studies in which a temporal analysis after virus/viroid infection was carried out, Rizza et al.,  showed that the pattern of up-regulated vs down regulated genes can change in pre-symptomatic when compared to post-symptomatic stages of Etrog citron infected with Citrus exocortis viroid (CEVd).
The overlap in significantly altered gene expression among the two single and the double infections was low. Common genes that were differentially expressed during each infection scenario are shown in Additional file 4: Table S3. This is not surprising considering the small number of genes with significant changes in both single infections. Consistently, Postnikova and Nemchinov  have recently shown that the number of host genes commonly affected during infection of Arabidopsis with either of twelve different viruses studied is very limited. Due to the nature of this specialized database it is difficult to establish a relationship between genes involved in peach ripening and those affected by the pathogens. Nevertheless in our study we observed some genes previously described as related with ripening with altered expression. Most of them did it in the, a priori, unexpected wayside slowing down the ripening (see Additional file 5: Table S4 and references [38–41]). Although we did not follow with great detail whether or not mixed infections had modified the ripening it is noteworthy to emphasize that alterations in ripening date have been described in several virus-host interactions regarding fruit trees (see  for review).
The higher number of genes significantly altered exclusively upon double infection with PLMVd and PNRSV led us to investigate whether some specific functional categories were over- or under- represented among those genes. Interestingly, among the 198 significantly induced genes unique to double infection (Figure 4) none of the functional categories appeared to be enriched in comparison with the normal distribution of genes in the whole microarray. As the array used in this study is enriched with sequences of genes that are implicated in chilling injury development, the lack of overrepresentation of functional categories among the induced genes upon infection compared to the genes represented on the array may indicate a similarity between the functional classes induced upon the two types of stresses, chilling injury and virus/viroid infection. By contrast, we identified over-represented functional categories and subcategories among the set of repressed genes belonging to each of the main GO domains (Table 2, threshold > 1.5 fold). We found genes involved in “response to external stimulus”, “defense response”, “catabolic processes” and “binding function or cofactor requirement”. Downregulation of these functions is presumably due to the virus counterattack against host defense-related pathways. Moreover, these host functions are frequently found in gene expression studies in response to virus infection .
By increasing the threshold to at least 2 fold changes we identified enriched functional categories only belonging to two of the main GO domains: Biological Process and Cellular Component (Table 2). Using this more stringent analysis we found groups of genes related with “response to stress”, to “external stimulus”, “amino acid metabolism” and “vacuole”. Importantly, the most overrepresented functional category was “response to water deprivation”. Among the genes assigned to this functional group genes related with “plant hormones” were found. Plant hormones play important roles in regulating developmental processes and signalling networks involved in plant responses to a wide range of biotic and abiotic stresses. For instance we identified a tonoplast resident H+-pump (At1g15690 H+-PPase AVP1) that contributes to vacuolar acidification, regulation of apoplastic pH and auxin transport . Auxin acts as an important component of hormone signaling network involved in the regulation of defense responses against various biotrophic and necrotrophic pathogens. Additionally, this hormone regulates the expression of genes associated with the biosynthesis, catabolism and signaling pathways of other hormones . Viral pathogens manipulate auxin signaling components to promote virulence and cause disease. One example is the interaction of Tobacco mosaic virus (TMV) replicase with Aux/IAA proteins. Development of symptoms promoted by this interaction has been described in Arabidopsis and tomato [45–47] and in addition disrupts the nuclear localization of several Arabidopsis Aux/IAA proteins. A second example is the protein phosphatase AtPP2CA (At3g11410) which acts as strong negative regulator of ABA signal transduction during seed germination and the regulation of stomatal closure . Tobacco plants infected with TMV showed increased ABA levels and treatment with ABA enhanced TMV resistance in tobacco . Some pathogens might have evolved the ability to produce ABA or ABA-mimicking substances to interfere with host defence. In any case, the role of ABA during plant-pathogen interactions depends on the individual plant-pathogen combination .
As members of the functional group “response to water deprivation” we also found proteins directly related with protection against oxidative stress like the Aldehyde dehydrogenase (At1g54100) and the Late embryogenesis abundant (LEA)-like protein (AtLEA5) (At4g02380). Aldehyde dehydrogenases (ALDHs) regulate the level of aldehydes by metabolizing excessive amounts of aldehyde molecules, which accumulate as a result of perturbed environmental conditions. Aldehydes are involved in different cellular metabolic processes but in excess they can have toxic effects on the cells. It is important then to maintain an appropriate balance of these molecules to avoid cellular damages. It has been described that over-expression of different ALDHs in A. thaliana confers tolerance to abiotic stress and protects plants against lipid peroxidation and oxidative stress . (LEA)-like proteins protect other proteins from aggregation from desiccation. AtLEA5 is the unique which is induced specifically by reactive oxygen species (ROS) as well as by ABA but it is unlikely to act as an antioxidant enzyme. It has been suggested that LEA5 may cooperate with other factors to protect cellular components against ROS-induced damage or indeed to enhance the turnover of specific proteins during stress to enable rapid acclimation to the prevailing conditions .
In the few cases in which changes in the host transcriptome have been investigated upon double virus infections (e.g. the PVX-PVY interaction in N. benthamiana) a severe oxidative stress was inferred to be induced in infected plant leaves, as increased transcript levels of genes encoding proteins important for lipid peroxidation for the generation of ROS were observed. We did not observe any significant over- or underrepresentation of genes involved in the antioxidative system in the doubly infected peach fruits. This difference between PLMVd-PNRSV-infected peach fruits and PVX-PVY infected N. benthamiana plants may be explained by the nature of the interaction between the two pathogens and the pathogens and the host. While co-infection of N. benthamiana with PVX-PVY can be considered as a true biological synergistic interaction leading to enhanced disease phenotypes compared to single infection with either pathogen, the doubly-infected PNRSV-PLMVd peach fruits did not exhibit an enhanced disease phenotype compared to single infection with either PNRSV or PLMVd. Interestingly, PNRSV was able to induce significant oxidative stress and an imbalance in the antioxidant systems in infected apricot-seeds  resulting in a decrease in seed germination. This observation indicates that host responses are not only specific for the host plant and the pathogen, but also tissue and/or organ-specific.
In the functional category “response to dehydration” which was strongly overrepresented upon double infection with PNRSV and PLMVd, we also found other important players with known functions in plant immunity. Among those was Cysteine protease RD21 (At1g47128), a Papain-like cysteine protease (PLCP). These proteins participate in immune responses and are targeted by pathogen- derived inhibitors. PLCPs are also required to trigger the hypersensitive response (HR) [54, 55]. Consistently, PLCP RD19 is required in Arabidopsis for RRS1-R-mediated resistance against the bacterial pathogen Ralstonia solanacearum producing effector PopP2 . A crucial role of PLCPs in disease immunity is also indicated by the observation that many pathogens produce effectors that manipulate these proteases [56–61].