The effects of potato virus Y-derived virus small interfering RNAs of three biologically distinct strains on potato (Solanum tuberosum) transcriptome

Potato virus Y (PVY) continues to be an economically important pathogen of potato (Solanum tuberosum L.), worldwide, that causes significant yield losses and is detrimental to quality of tubers [1‐3]. PVY belongs to the family Potyviridae, and genus Potyvirus. It is a flexuous rod-shaped virus which exhibits a 9.7 kb + ss RNA genome [4]. The genome consists of two opening reading frames which encode 11 proteins. A single large open reading frame encodes a polyprotein cleaved into ten functional proteins while a second small open reading frame called PIPO (Pretty Interesting Potyiviridae ORF) encodes P3N-PIPO through RNA polymerase slippage mechanism in the P3-encoding region [2, 4‐7].

The genus Potyvirus is characterized as one of the largest genera of plant-infecting viruses with more than 200 approved and tentative species of pathogenic viruses [8]. The PVY-potato pathosystem is further complicated by the presence of a complex of different strains of viruses. The viral strains differ from one another in terms of symptoms they produce in same host and expectedly, there exists tremendous amount of genetic variation in their genomic sequences as well as capability for recombination [9‐11]. Various strains of PVY infecting potato include ordinary strain (PVY-O), stipple streak strain (PVY-C) and necrotic strains: tobacco veinal-necrotic strain (PVY-N), necrosis tuber-necrotic strain (PVY- NTN), necrotic-wilga (PVY- N:Wi) and a recombinant between N and O (PVY-N:O) [2, 12, 13]. Among the various strains, PVY-O is most prevalent strain in Europe and the USA, along with that, PVY-N and PVY- NTN are the most widely studied strains of PVY. However, recently, an increase in prevalence of recombinant strains, PVY-NTN and PVY-N:Wi over PVY-O has been reported [14]. Potato cultivars show differential host response when challenged with different strains of PVY. A deep sequencing study has revealed the differential accumulation of small RNAs derived from PVY-N, PVY-NTN and PVY-O upon infection of the same potato host cv. Russet Burbank [15].

Plant-pathogenic virus infections usually lead to production of virus-derived small interfering RNAs (vsiRNAs) in infected plant cells, as a result of host RNA silencing-mediated defense mechanism [16‐18]. Plant’s defense mechanism is established due to vsiRNAs generated as an outcome of virus infection targeting homologous viral transcripts. The process of formation of dsRNA sequences of virus genome is proposed to occur through various mechanisms like involvement of virus genome-encoded, RNA-dependent RNA polymerases (RdRp), inadvertent complementarity or base pairing between plus and minus strands of viral RNAs, formation of fold-back structures of viral genome sequences due to complementarity or due to the activity of host-derived RdRp etc. [19]. Despite the variation in mechanism, the trigger for RNA silencing is dsRNA which is processed into small interfering RNAs (siRNAs) by the activity of host RNAse III enzymes such as Dicer-like (DCLs). The resultant siRNAs are recruited on to RNA-induced silencing complex (RISC) mediated by family of proteins called argonautes (AGOs) that cleave target RNA in a sequence dependent manner [20‐22].

The RNA silencing process also comprises an amplification of signal strategy wherein host-derived RdRp are involved in generation of perfect dsRNA substrates for processing by DCL. This process leads to production of secondary siRNAs which further reinforces the activity of primary siRNAs [23‐25]. The role of secondary siRNAs pertains to the systemic spread of silencing signals throughout the plant system [26, 27]. Thus, the entire mechanism of RNA silencing mainly functions as a molecular antiviral defense system to resist the invasion of plant pathogenic viruses. However, the RNAi-mediated silencing of gene expression functions on the principle of sequence complementarity irrespective of the origin of transcripts. Hence it is conceivable and possible to study for the vsiRNAs to have unintended silencing effect on the host transcripts.

With the increase in small RNAs datasets from deep sequencing studies, there is increased development of tools for prediction and identification of their prospective target genes in biological cells as way of functional characterization [28‐31]. The application of small RNA target prediction tools is paramount to the elucidation of cellular, physiological and ecological processes [28‐33]. This study exploits small RNA target prediction tools along with other bioinformatics tools to elucidate global virus-host interactions at the transcriptome level for the PVY-potato pathosystem.

In this study, we aimed to identify the derivation of vsiRNAs from the genome of three biologically distinct strains of PVY (PVY-N, PVY-NTN and PVY-O) and to reveal their propensity for potato transcripts for post-transcriptional gene silencing. This deciphered the role of PVY-derived vsiRNAs as potential pathogenicity determinants in interfering with host potato physiological processes, stress responses and subsequent symptom development.

The data for the study has not been obtained directly or indirectly from human or animal subjects nor has the study been conducted on human/animal subjects hence the research has been exempted from Washington State University Institutional Review Board’s (WSU-IRB) review.

We studied the role of highly frequent PVY-derived vsiRNAs of three biologically distinct strains of PVY (PVY-N, PVY-NTN and PVY-O) for their putative global effects on the host transcripts. The important resources of small RNA target prediction algorithms, psRobot and psRNATarget, to map vsiRNAs targets on the transcripts of potato and the potato genome annotations (S. tuberosum SolTub_3.0 cDNA sequences and S. tuberosum SolTub_3.0 non-coding RNAs) were employed effectively to identify genes that are potential targets for vsiRNA-induced silencing. It is vital to note that different small RNA target prediction algorithms predicted target transcripts at varying degrees, as such, intersections in outputs by different algorithms deliver precision and reduces false positives [28]. Large data sets of the potato coding transcripts were predicted to exhibit complementarity with PVY strain-specific vsiRNAs and hence the potential for their downregulation. The targeting vsiRNAs mapped to every genomic position of PVY. The genomic regions of 6 K1, P1 and Hc-Pro in PVY-N, P1, Hc-Pro and P3 in PVY-NTN and P1, 3′ UTR and NIa in PVY-O were dominant hotspots for the generation of vsiRNAs. The high frequency of vsiRNAs substantiates the plausibility of their off-target effects on the host transcripts.

The PVY strains (PVY-N, PVY-NTN and PVY-O) are known to elicit varied biological responses and differential expression of their genome-derived vsiRNAs [2, 15]. With that in mind, we anticipated varied interaction between PVY strain-specific vsiRNAs and the potato transcripts. As such, potato transcripts that were exclusively targeted by each strain may hold important clues about the biological variability and differential response of potato to each PVY strain. Further exploitation of these exclusively targeted transcripts may improve understanding of PVY-potato interactions. Considering off-target silencing, the differential accumulation of PVY-derived vsiRNAs of different PVY strains (Table 1) in the same potato host may also account for differential biological behavior.

The PVY-NTN strain is known to cause tuber necrosis in the potato tubers of susceptible plants and has been associated with manifestation of the potato tuber necrotic ringspot disease (PTNRD) [14, 52, 53]. The disease develops through protrusion of rings on the surface of the potato tuber, these subsequently become sunken and necrotic [52, 53]. While some PVY-N strains have been recorded to exhibit the same phenomena, it has not been reported for PVY-O [52]. It was therefore interesting to note that PVY-NTN, a recombinant strain, exhibited a higher repertoire of PVY-derived vsiRNAs as well as highest number of target coding transcripts. Also, the GO enrichment of PVY-NTN exclusively targeted transcripts which revealed among others, functions related to developmental process, anatomical structure development. It will be conceivable to investigate PVY-NTN exclusively targeted potato transcripts as well as those only common to PVY-N and PVY-NTN for involvement in PTNRD. Earlier studies have revealed that virus or viroid-derived siRNAs are involved in silencing of host genes which in turn leads to development of typical symptoms associated with respective diseases [54‐57]. Hence consideration and proof of vsiRNAs as major pathogenicity determinant of disease symptoms in virus-host relationship are not uncommon.

It must be stated that, while some target coding transcripts were exclusive to each strain of PVY, majority of the predicted transcripts were common amongst them (Fig. 3). The same applies for the predicted ncRNAs (Fig. 4). This may be accounted for by the considerable identity between the PVY strains at genomic level [36]. Thus, there could be shared identity among the vsiRNAs of the PVY strains which subsequently had similar targets in the potato transcripts. This suggests a similar mechanism of action of strain-specific PVY-derived vsiRNAs on the potato transcripts. Moreover, this phenomenon is further corroborated by shared KEGG pathways (Fig. 5a-c) and GO terms enriched (Fig. 6a-d) in target potato coding transcripts amongst the PVY strains.

Functional annotation indicated that the target potato transcripts encompassed broad functional properties. The ncRNAs MIR821, 28S rRNA,18S rRNA, snoR71, tRNA-Met and U5 were identified amongst top target candidates for vsiRNAs of PVY. The MIR821 has been shown to be involved in metabolic processes and stress responses [58‐61]. The possible involvement of tRNA-Met in translational control in stress has been highlighted [62]. The top target pathways for the PVY strains were plant hormone signaling, genetic information processing pathways (RNA-related processes), plant-pathogen interactions, phenylpropanoid biosynthesis as well as starch-sucrose metabolism amongst others. These pathways represent high level molecular functions that are related to the GO terms enriched for the target transcripts. For example, the high targeting of the GO term related to RNA processing, adenyl ribonucleotide binding (Fig. 6d) relates to KEGG pathways involved in genetic information processing pathways, spliceosome, RNA transport and ribosome (Fig. 5a -c). Enrichment of GO terms related to membrane part, single cell signaling, defense response, protein kinase activity, cell communication and protein phosphorylation (Fig. 6d) suggests targeting of cellular signaling pathways which may be related to the enriched KEGG pathway in plant hormone signaling (Fig. 5a -c). Interference with plant hormone signaling is further revealed by enrichment of GO terms in system development and reproductive process, physiological processes that are most likely to be under the regulation of phytohormones. These phenomena suggest counter-defense strategy by the PVY-derived vsiRNAs to host RNA silencing to perturb plant defense responses and developmental processes, which may ultimately lead to development of symptoms associated with PVY infection. The vsiRNAs could therefore be important determinants in plant-pathogen interactions by mediating post-transcriptional gene regulation of host genes. Furthermore, the off-target silencing of host transcripts is conceivable and potentially enormous since vsiRNAs and target host mRNAs are present in same cellular environment. Nevertheless, at a given point of time not all the transcripts that are potential targets for vsiRNA based repression are present, hence the presence of transcripts and abundance of vsiRNAs, sequence characteristic features of vsiRNAs like 5′ nucleotide etc. are the determinants that could play a major role in off-target silencing activity of vsiRNAs.

In our study, the HSP-90 was identified to be significantly downregulated upon PVY-NTN infection in potato (Fig. 9). The HSP-90 has been demonstrated to be vital in the function of various proteins. It has been shown to be pertinent to stabilize auxin response phenotypes, to influence the loading of small RNAs into argonautes (AGOs) and subsequently the RNA-induced silencing complex and in plant responses to stresses [63‐65]. This relates to the GO and KEGG pathway annotations identified in this study. The silencing of HSP-90 suppressed the Pvr9 resistance gene-mediated hypersensitive response in Nicotiana benthamiana to Pepper mottle virus, a Potyvirus [66]. The demonstration of the requirement of HSP-90 for efficient plant response to stress [51] and its importance in host RNA silencing processes [64] highlights the importance of downregulation of potato HSP-90 gene by PVY. In silico predictions have previously demonstrated the HPS-90 as a validated target of vsiRNAs of Cucumber mosaic virus, a + ss RNA virus, in tomato [67]. This substantiates the suggestion that the PVY-derived vsiRNAs are a counter-strategy to thwart plant defense processes. This not only validates the in silico prediction, but highlights its importance in elucidating and understanding virus-host interactions.

The β-1,3 galactosyltransferase 2 was also significantly downregulated in PVY-NTN infected potato. Galactosyltransferases are proteins exhibiting membrane-spanning domains, and through glycosylation of glycoconjugates (glycoproteins, glycolipids, and proteoglycans), are involved in a diversity of functions that include cell wall structure, cell–cell interactions and signaling and host–pathogen interactions [68‐71]. A β-1,3 galactosyltransferase has been shown to be paramount for pollen development and viability in Arabidopsis [72]. This validates the interference of PVY infection with plant signaling pathways in potato as predicted in functional annotation. The aquaporin NIP1–1 was shown to be downregulated by log₂(FC) = −1.15 (Fig. 9) in PVY-NTN infected potato plants. Aquaporin proteins are pertinent for water and nutrient transport through living membranes i.e. transmembrane transport, recently, they have also been demonstrated to be involved in carbon dioxide delivery for photosynthesis as well as confer responses to abiotic and biotic stresses [73‐76]. This is interesting because it also validates the functional annotation of PVY strain-specific commonly targeted potato transcripts in the cellular component to the membrane part and in the molecular function category to transmembrane transporter activity protein (Fig. 6d). Moreover, the functional annotation of target coding transcripts of PVY-NTN vsiRNAs revealed enrichment of transmembrane transport in the biological process category, membrane in the cellular component category and active transmembrane transporter activity in the molecular function category (Fig. 6b). It has been shown that aquaporin proteins (tonoplast-intrinsic aquaporins) are upregulated in response to biotic stress, furthermore, tomato (Solanum lycopersicum) aquaporin transcripts are high in lines resistant to Tomato yellow leaf curl virus (TYLCV) as opposed to susceptible lines [77]. This suggests that downregulation of aquaporins and other transmembrane proteins such as the β-1,3 galactosyltransferases may be involved in conditioning the potato plants for susceptibility to PVY.

The findings presented here are in accordance with the earlier works on virus-derived siRNAs and off-target silencing of host transcripts [54, 55, 57, 78, 79]. Similarly, siRNAs derived from TMV (TMV-Cg: Crucifer infecting isolate of TMV) were revealed to target host mRNAs involved in RNA processing and defense response of the host [80]. Off-targeting potential of Sugarcane mosaic virus (SCMV)-derived siRNAs with maize mRNAs have been proven and it was revealed that host mRNA involved in ribosome biogenesis, other biotic and abiotic stress-related pathways were the targets [81]. Similarly, tomato target genes (SolWD40-repeat) were identified for small RNAs derived from Pospiviroid infecting tomato and downregulation of host mRNAs was demonstrated [79].

Evidence of vsiRNAs interfering with and silencing host plant genes has also been shown [82]. In addition, the credibility for the hypothesis comes from the successful deployment of 21 nt virus genome-derived sequences as effector molecules of silencing in amiRNA-mediated antiviral resistance [83, 84]. It implies that sequence complementarity of 21 nt length with target mRNA is sufficient to induce RNA silencing of cognate transcripts. The predicted interactome scenario is a first report on the interaction between one of the most important Potyvirus genome-derived siRNAs and potato transcripts. It highlights the significance of deeper understanding of the role of vsiRNAs on viral replication, pathogenicity and host machinery. These clues on virus-host interactions can be applied in developing novel strategies for disease management.

The differential accumulation of PVY strain-specific vsiRNAs in potato cv. Russet Burbank was demonstrated. Most of the vsiRNAs populations were derived with high frequency from 6 K1, P1 and Hc-Pro for PVY-N, P1, Hc-Pro and P3 for PVY-NTN and P1, 3′ UTR and NIa for PVY-O genomic regions. The host genes targeted by the PVY-derived vsiRNAs were found to be involved in plant hormone signaling, genetic information processing and defense/ stress responses suggesting a counter-defense strategy by the virus to the host RNA-silencing machinery. The broad range of host genes targeted by PVY-derived vsiRNAs in infected potato suggests a diverse role for vsiRNAs. Future work could focus on the confirmation of the vast majority of the predicted transcripts, more so, on the transcripts exclusively targeted by the PVY strains.