Differential gene expression and physiological changes during acute or persistent plant virus interactions may contribute to viral symptom differences

Viruses have different strategies for infecting their hosts. Fast and acute infections result in the development of severe symptoms and may cause the death of the plant. By contrast, in a persistent interaction, the virus can survive within its host for a long time, inducing only mild symptoms. In this study, we investigated the gene expression changes induced in CymRSV-, crTMV-, and TCV-infected Nicotiana benthamiana and in PVX- and TMV-U1-infected Solanum lycopersicum plants after the systemic spread of the virus by two different high-throughput methods: microarray hybridization or RNA sequencing. Using these techniques, we were able to clearly differentiate between acute and persistent infections. We validated the gene expression changes of selected genes by Northern blot hybridization or by qRT-PCR. We show that, in contrast to persistent infections, the drastic shut-off of housekeeping genes, downregulation of photosynthesis-related transcripts and induction of stress genes are specific outcomes with acute infections. We also show that these changes are not a consequence of host necrosis or the presence of a viral silencing suppressor. Thermal imaging data and chlorophyll fluorescence measurements correlated very well with the molecular changes. We believe that the molecular and physiological changes detected during acute infections mostly contribute to virus symptom development. The observed characteristic physiological changes associated with economically more dangerous acute infections could serve as a basis for the elaboration of remote monitoring systems suitable for detecting developing virus infections in crops. Moreover, as molecular and physiological changes are characteristics of different types of virus lifestyles, this knowledge can support risk assessments of recently described novel viruses.

Introduction and Northern blot validation to follow the transcriptome changes. Employing diverse bioinformatics tools, we unravelled commonalities and specific features of acute and persistent infections. We have also shown that specific changes in physiological parameters are characteristic in acute infections. Based on these data, we propose that the molecular and physiological alterations may contribute to viral symptom development during acute infection.

Plant materials and virus infection
Wild-type Nicotiana benthamiana [23] and Solanum lycopersicum ('Kecskeméti jubileum' cultivar of ZKI (www.zki.hu)) plants were used as systemic hosts. Plants were grown under a 16 h light/8 h dark regime at 21˚C in soil. The N. benthamiana plants were infected with CymRSV, CymRSV19S (a p19 silencing suppressor mutant CymRSV), carnation Italian ringspot virus (CIRV), CIRV19S (a p19 silencing suppressor mutant CIRV), crTMV and TCV. The S. lycopersicum plants were infected with PVX and TMV-U1 strain. For infection, we used in vitro transcripts of the viruses (CIRV, CIRV19S, crTMV, PVX and TMV-U1) or purified virions (CymRSV, CymRSV19S, and TCV). The efficiency of the virus infection was tested either by EtBr gels or by Northern blot analysis of extracted RNA using radioactively labelled virus-specific probes. Cloned partial viral pieces were the templates for radioactive probe preparation. S19 Table contains a list of the primers used for cloning these viral sequences.

Sample collection for high-throughput analysis
Infection experiments were performed three times, resulting in three biological replicates. In the case of N. benthamiana, we collected systemically infected leaves at 5 days post inoculation (dpi) (CymRSV-, crTMV-and mock-infected) and 11 dpi (TCV-infected) for microarray analysis and further validation. RNA was extracted from 10 plants/infection in three biological replicates (S1 Fig). In the case of S. lycopersicum, mock-and virus-infected 'Kecskeméti jubileum' variety at 14 dpi was used for RNA-seq and further validation (S2 Fig). Leaf samples were collected from the second leaf level from the bottom from four plants/infection in three biological replicates. RNA was immediately extracted from the samples, or the leaves were stored at -70˚C.

Total nucleic acid and RNA isolation
Total RNA was extracted [23] from the mock-or virus-infected (systemically infected) leaves. Briefly, frozen plant material was homogenized in an ice-cold mortar, suspended in 650 μl of extraction buffer (100 mM glycine, pH 9.0, 100 mM NaCl, 10 mM EDTA, 2% SDS and 1% sodium lauroylsarcosine) and mixed with an equal volume of phenol, centrifuged for 5 minutes. The aqueous phase was treated by equal volumes of phenol, chloroform, and isoamylalcohol (25:24:1), and after subsequent treatment with chloroform:isoamyl-alcohol (24:1), it was precipitated with ethanol and then resuspended in sterile water. Sample quality and viral presence were examined on a 1.2% agarose gel and by Northern blot analysis. This total nucleic acid was used for the Northern blot experiments. For microarray analysis and RNA-seq, the samples were further purified to eliminate DNA and improve the sample quality. Total nucleic acid extracts obtained were stored at −70˚C until used.

Microarray and bioinformatics analysis
For microarray analysis, total nucleic acid extracts of the pooled N. benthamiana samples (10 samples/infection in 3 biological replicates) were further purified by Tri Reagent (Sigma) and treated with DNase (Thermo Scientific). The concentration of the purified DNA-free RNAs was determined by NanoDrop, and their quality was further checked by Agilent scanner, FE SW 9.5. Samples with adequate quality were used for microarray hybridization using the Agilent Tobacco Gene Expression Microarray 4x44K array. Expression profiles of the virusinfected samples were compared to those of the control plants. Raw data were deposited into NCBI-GEO: GSE113774/ GPL21056. Mean normalized expression values for each probe were calculated (3 biological replicates for 3 virus and mock infections), and these normalized results for all samples were also deposited into NCBI-GEO. Genes showing statistically significant changes in response to infection by crTMV, CymRSV and TCV were identified using corrected p value assignments <0.05 that were derived from unpaired T-test variance analysis and followed by multiple testing correction by the Benjamini-Hochberg procedure. Changes in gene expression were calculated as normalized fold changes, and changes greater than twofold were analysed (S1 Table).

RNA-sequencing and bioinformatics analysis
For RNA-seq, total nucleic acid extracts of pools derived from S. lycopersicum plants (4 samples/infection in 3 biological replicates) were purified with RNAzol. Their quality was checked on a Bioanalyser. Libraries were prepared with Illumina TruSeq Stranded mRNA Library prep kit, and 50 base-pair single-end reads were sequenced on an Illumina HiScanSQ by the UD-GenoMed company. The third biological replicates were sequenced twice as a technical replicate; therefore, four fastq files containing results of the sequencing were deposited into NCBI-GEO: GSE113774/GPL24943. For quality control, the reads were checked with fastqc, ribosomal contaminations were removed with SortMeRNA [24], and adapter trimming and quality filtering were performed by Trimmomatic [25]. After quality control, the reads were aligned to the genome (Sly2.5 with ITAG2.4 annotation) and quantified, and a significant expressional difference was calculated using the Tuxedo protocol [26]. Only genes showing significantly different expression (FDR-adjusted p value less than 0.05) and twofold changes were considered (S2 Table). Heat maps were created to show the gene expression of several probes or genes using red colour for up-and green for downregulated genes. A detailed version of each heat map is provided as a supplementary Excel table containing log2-fold change values with their corresponding p values.

Northern blot analysis of host endogenous genes
N. benthamiana and S. lycopersicum total RNA extracts were used for first-strand cDNA synthesis (RevertAid First Strand cDNA Synthesis Kit, Thermo Scientific) according to the manufacturer's recommendation and using oligo dT. Target gene fragments were amplified from the produced cDNA by PCR using Taq polymerase (Thermo Scientific). The sequences of the oligonucleotides used as primers are listed in S11 and S12 Tables. The products of the reverse transcription PCR (RT-PCR) were purified by GenJet (Thermo Scientific). The gel-purified PCR products were ligated into pBlueScript plasmid (Promega). The sequences of the clones were confirmed by Sanger sequencing. For Northern blot analysis, 1-5 μg of total RNA were separated on formaldehyde-containing 1.2% agarose gels and blotted to Hybond-N membranes, as described previously [27]. Cloned fragments of the viruses and endogenous genes being studied were used as templates for probe preparation. Before probe preparation, the dNTPs were eliminated from the PCR product by gel filtration of the product on homemade G-50 Sephadex micro columns. Radioactively labelled random DNA probes were generated from these purified PCR products with a DecaLabel DNA Labeling Kit (Thermo Scientific). Northern blots were hybridized with these probes either in Church buffer (1% BSA, 1 mM EDTA, 0.25 M Na2HPO4, 7% SDS, pH 7.2), DIG Easy Hyb(Roche) or PerfectHyb Plus Hybridization buffer (Sigma-Aldrich), at 65˚C, washed according to the manufacturer's instructions and exposed to X-ray films. Signal intensity was calculated using the rRNA band of the EtBr-stained gel before blotting as a loading control.

Quantitative RT-PCR analysis
When Northern blot analysis was not successful in detecting endogenous target transcripts, we used quantitative RT-PCR for validation of gene expression changes. For qRT-PCR, the total RNA isolated by the phenol-chloroform method was treated with DNase (Thermo Scientific), followed by purification using RNAzole (Sigma). Then, 3 μg RNA was used to prepare cDNA using RevertAid Reverse Transcriptase and a random oligo primer (Thermo Scientific). cDNA was used as a template for PCR, which used gene-specific primers (S11 and S12 Tables) and Power SYBR Green PCR Master Mix (Applied Biosystems) and was performed in a Rotor Gene 3000 real-time DNA detection system (Qiagen). Three biological replicates and three technical replicates were used per sample. For relative gene expression results, the delta-delta Ct was calculated. Ubiquitin showing stable expression changes in every infection in both hosts was used as an internal control.

IR thermography
For thermography, we selected the third, fully developed young leaves from the apical shoots of Nicotiana benthamiana (CymRSV and crTMV on 4 dpi; TCV on 8 dpi) or Solanum lycopersicum (13 dpi for both viruses) plants. Thermographic images were taken using a Stirlingcooled infrared scanning camera (varioSCAN 3021 ST, Jenoptik, Jena, Germany). All plants under treatment were assessed by thermography, with 3 plant replicates per treatment. Leaf temperatures were analysed independently from each other [28,29]. The mean leaf temperature per plant was used for statistical analysis (n = 3 to 5).

Chlorophyll a fluorescence and Chl fluorescence imaging
Leaves of the third fully developed branch from the apical shoot were used for the study in both hosts. Plants were analysed at 7 dpi (in the case of CymRSV-and crTMV-infected N. benthamiana), at 11 dpi (in the case of TCV-infected N. benthamiana) and at 8 dpi (in the case of TMV-and PVX-infected S. lycopersicum) together with their appropriate controls. OJIP chlorophyll a fluorescence transients were measured by a Plant Efficiency Analyzer (Pocket Pea, Hansatech, UK). Prior to the OJIP measurements, plants were dark-adapted for 20 min under greenhouse conditions. Nine leaves from 3 independent plants/treatment were analysed to obtain data. Quenching analysis of Chl fluorescence was detected by the maxi-head of a PAM Chl fluorescence imaging system (Walz Effeltrich, Germany).
The chlorophyll fluorescence parameters (that determine photosynthetic performance) were measured or calculated as follows: the initial fluorescence yield was obtained in a darkadapted state, when the reaction centres are open (Fo); maximal fluorescence yield, in a darkadapted state, when the reaction centres are closed (Fm); maximal quantum yield of photosystem II photochemistry (Fv/Fm); electron transport rate at donor side of PSII (Area) reflecting the size of the plastoquinone pool; the amount of active reaction centres per absorption (RC/ ABS); efficiency of water splitting complex on the donor side of PSII (Fv/Fo); probability of electron transport out of QA ((1-Vj)/ Vj, where Vj = (F2ms-Fo)/Fv); and performance index (potential) for energy conservation from photons absorbed by PSII to reduce the intersystem electron acceptors (PI) [30].

Results and discussion
Based on viral symptoms and infection lifestyles, we employed a range of single-stranded RNA viruses to study whole transcriptome changes induced during acute or persistent infections. N. benthamiana plants were infected with CymRSV (Tombusviruses, acute), crTMV (Tobamoviruses, acute) and TCV (Tombusviruses, persistent), whereas S. lycopersicum plants were infected with PVX (Potexviruses, acute) and TMV-U1 (Tobamoviruses, persistent). CymRSV and crTMV infections on N. benthamiana plants resulted in a high level of virus accumulation in the systemic leaves at 5 dpi, which is associated with drastic downregulation of housekeeping gene mRNAs, as demonstrated by monitoring virus and GAPDH mRNA levels [22]. This acute phase of infection turns rapidly into necrosis and subsequent death of plants at 7-8 dpi. The acute PVX infection on S. lycopersicum is slower, with the virus reaching the systemic leaves at approximately 14 dpi, coinciding with the downregulation of housekeeping genes. Although severe disease symptoms appear, the plant can survive while exhibiting prolonged downregulation of housekeeping genes. TCV infection on N. benthamiana is slower than CymRSV and crTMV infections, with the virus reaching the systemic leaves at 11 dpi. In this case, the downregulation of housekeeping genes never occurs, and the infection never turns into the acute phase. TMV infection on S. lycopersicum shows the same characteristic pattern, with the virus reaching the systemic leaves at 14 dpi but never inducing downregulation of the housekeeping genes [22]. According to these results, we identified sampling timepoints in this current work to represent the earliest timepoints when viruses systemically infect their host plants and induce symptoms such as stunting and chlorosis, but only when necrotic reactions are not yet initiated (S1 and S2 Figs).

Host transcriptomes are strongly affected during acute virus infections
To unravel changes caused by CymRSV-, crTMV-, TCV (and mock) infection in N. benthamiana plants a microarray analysis was carried out (see Materials and Methods). To find differentially expressed genes, we collected all probes showing 2-fold changes with p values smaller than 0.05 in any of the virus-infected plants (listed in S1 Table). In CymRSV-and crTMVinfected plants, the number of differentially expressed probes (DEPs) was very high (6639 and 6318), while in TCV-infected samples, the number of DEPs was lower by approximately an order of magnitude (515 DEPs) (Fig 1A and 1B). Among these DEPs, 2807, 2033 and 275 probes were upregulated, whereas 3832, 4285 and 240 were downregulated ( Fig 1B). Importantly, most of up-(1473) and downregulated (3219) probes were similar in the CymRSV-and crTMV-infected plants, suggesting common regulation ( Fig 1C). These results show that acute infections (CymRSV and crTMV) severely alter the transcriptome of the host, whereas the persistent infection (TCV) has a limited impact.
An RNA-seq analysis was carried out to study transcriptome changes during PVX and TMV-U1 infection in S. lycopersicum plants (see Materials and Methods). The differentially expressed genes (DEGs, 2-fold changes, p <0.05) observed during virus infection are listed in S2 Table. A total of 5711 DEGs were identified in the PVX-infected samples (vs mock), whereas in the TMV-infected samples, only 1672 DEGs were found. Among these, 2736 and 1008 DEGs were upregulated and 2975 and 664 DEGs were downregulated in PVX-and TMV-infected plants, respectively (Fig 2A and 2B). Only a small fraction of the DEGs (542 and 288) overlapped between the two infections (the majority of DEGs were present only in the PVX-infected plants) (Fig 2C).
Comparison of Volcano plots shows that the number of downregulated genes was much higher in N. benthamiana than that of upregulated genes, whereas in S. lycopersicum, this ratio was approximately 1:1 (Figs 1A and 2A). However, probes on the N. benthamiana chip could not be directly correlated to changes in the whole genome because some transcripts could be overrepresented, whereas probes for others could be absent. An imbalance of the genes represented on the chip could lead to an imbalance in the identified DEGs and not actually show their real ratio.  Table. https://doi.org/10.1371/journal.pone.0216618.g001  Table. https://doi.org/10.1371/journal.pone.0216618.g002 Therefore, in summary, acute infections cause severe transcriptome changes, whereas persistent infections have a moderate impact. The changes caused by acute and persistent infections in N. benthamiana and S. lycopersicum plants are consistent with each other.

Photosynthesis-related transcript downregulation and stress-related transcript up-regulation occur specifically during acute infection
To reveal the molecular basis of acute and/or persistent infections, we analysed the functional distribution of the DEGs (annotated according to their Bin codes). The most differentially regulated genes common for acute and persistent infections (for all viruses tested) were the ones that participate in protein metabolism, followed by the DEGs of RNA regulation, signal processes and transport (S3 Fig). These changes were observed in both hosts. When we searched for DEGs, we found that transcripts that play role in photosynthesis were downregulated, whereas stress response DEGs were upregulated almost exclusively during acute (CymRSV, crTMV and PVX) infections and were typically absent in persistent (TCV and TMV-U1) infections (S4 Fig and S3-S6 Tables). Investigating the top 20 down-and upregulated genes in both hosts, we found the same situation: changes were more severe during acute infections (S7-S10 Tables). The most downregulated genes in N. benthamiana plants were almost the same during acute infections (CymRSV and crTMV), and the expression of these probes was unaltered during persistent (TCV) infection. The most upregulated genes in virus-infected N. benthamiana plants were stress-related genes and were very similar during CymRSV-and crTMV-infected tobacco but were unchanged in the TCV-infected tobacco (S8 Table).
The most downregulated DEGs in S. lycopersicum have diverse functions, except the cytochrome P450s, whose levels decreased in both PVX and TMV infections. Although some of the top 20 downregulated genes were similar, the extent of their downregulation was more severe during acute infection (PVX-infected tomato) (S9 Table). Pathogenesis-related proteins (PRs) and various transcription factors that play a role in hormone metabolism were among the most upregulated genes in the virus-infected tomato. In addition, in this case, the PVX-induced changes were even much higher compared to TMV-U1 (S10 Table). In summary, photosynthesis-related and stress-related transcripts are specifically affected during acute infection.

Housekeeping-, stress-and some metabolism-related transcripts are differentially expressed during acute but not persistent infection
In our previous work, we showed that, in systemic leaves of CymRSV-infected N. benthamiana, the level of GAPDH, tubulin, Rubisco, chlorophyll a/b binding protein (CP) 29 (CP29) and histone (H1E) transcripts were drastically downregulated, and the elongation factor (EF) 2 transcript was slightly decreased, whereas the glutathione S-transferase gene (GST) and the heat shock protein (HSP) 90 gene (HSP90) transcripts were increased [22]. We checked the expression levels of these genes in our microarray and RNA-seq data. Several Rubisco-, GAPDH-and CP-specific probes were present on the microarray chip. Their expression showed severe changes only during acute (CymRSV and crTMV) infections, and their expression was not altered during persistent (TCV) infection (Fig 3A upper panel). We validated the changes of the Rubisco, GAPDH and CP29 transcripts by Northern blot analysis (Fig 3A bottom panel). The RNA-seq results from S. lycopersicum plants showed that Rubisco, GAPDH and CP RNA levels decreased during the PVX infection but were not altered or only slightly so during TMV-U1 infection (Fig 3B upper panel). Validation of their levels by Northern blot analysis showed the same trend (Fig 3B bottom panel).
According  (Fig 4A upper panel). According to the RNA-seq results, the GST genes of S. lycopersicum were usually upregulated during virus infection, and these changes were more severe in PVX-infected plants (Fig 4B upper panel). These changes were validated by Northern blot analysis (Fig 4A and 4B lower panels).
We checked the relative expression of PR, SAR and HSP transcripts in our microarray and RNA-seq data (S6 Fig); all were upregulated only during acute (CymRSV, crTMV and PVX) infections, whereas their induction was absent or mild during persistent (TCV or TMV-U1) infections. Induction of the pathogen-related protein Q gene (PR-Q), pathogen-related protein 1 gene (PR1) and SAR in N. benthamiana and HSP20 in S. lycopersicum was also validated by Northern blot analysis (S6A and S6B Fig).
Protodermal factor 1 (PDF1) and lactate dehydrogenase (LDH) were among the DEPs whose levels were severely changed in acute infection in N. benthamiana plants but stayed unchanged during persistent infection (S7A and S7B Fig). LDH was shown to play a role in proline metabolism [31]. During stress conditions, fast degradation of proline happens as a result of the activation of proline dehydrogenase (ProDH) [32]. Although lactate is a substrate of LDH, it is also a competitive inhibitor of ProDH (S7A Fig, lower panel at right). The increase in LDH expression during virus infection enhances ProDH activity, allowing a fast response to stress conditions. PDF1 encodes a putative extracellular proline-rich protein exclusively expressed in the L1 layer of meristems and the protoderm of organ primordia. Its level was strongly downregulated in acute infection compared to mock-and TCV-infected leaves (S7B Fig right panel). In S. lycopersicum, changes in PDF1 and LDH expression showed the same trend. PDF1 was markedly upregulated during PVX and only slightly in TMV-U1 infection. LDH was downregulated during PVX infection and slightly upregulated during TMV-U1 infection (S7A and S7B Fig right panels).
The gene expression patterns analysed above show that specific transcriptome changes are induced during acute infections. These changes are not characteristic of persistent infections. Acute transcriptome changes, however, are consistently similar in the two investigated hosts.

Photosynthesis-related, cell wall-remodelling and hormone-responsive transcripts are severely altered during acute infection
Although extensively studied, the molecular background of viral symptom development is still elusive. Symptom development in virus-infected plants reflects the sum of various molecular and physiological changes [5,33]. Chlorosis, yellowing, leaf size reduction, malformation or senescence are 'general' symptoms caused by several plant viruses. These changes usually coincide with alterations in photosynthetic activity or molecular structure of the host chloroplast [34][35][36][37][38][39], but insights into molecular changes are still elusive.
Chlorosis and yellowing can be a result of chlorophyll (Chl) degradation, an early mark of leaf senescence. The key regulator of Chl degradation is pheophorbide a oxygenase (PAO), which catalyses the cleavage of the porphyrin ring of pheophorbide, resulting in a red Chl catabolite intermediate (RCC), which later degrades in the vacuole (Fig 5A). The level of PAO increases during biotic stresses [40,41]. As on the microarray chip, we did not find a specific probe for N. benthamiana PAO, we showed by Northern blot analysis that its level was induced during acute (CymRSV and crTMV) infections (Fig 5B). By contrast to this observation, our RNA-seq results in S. lycopersicum showed three different PAO-like DEGs, whose expression slightly decreased in PVX and slightly altered in TMV-U1 infection (Fig 5C, S13 Table).
During acute infection, in addition to chlorosis and yellowing, slight stunting of both hosts was observed. When cell wall synthesis is impaired, the leaf size decreases, malformation of the leaf shape occurs, and the total growth of the plant is reduced. These changes can be observed in cell wall synthase (CESA) mutants or CESA-silenced plants [42]. By contrast, cell wall invertase (CWINV), which converts sucrose into fructose, is usually induced early during any defence [43] (Fig 6A left panel). With tomato yellow leaf curl virus, infected S. lycopersicum plants, especially when co-infected with tomato chlorosis virus, stunting was directly correlated to CESA8 down-and CWINV2 upregulation [34]. Investigation of the DEPs in N. benthamiana and the DEGs in S. lycopersicum that play a role in cell wall metabolism showed the same trend (Fig 6, S13 and S14 Tables); that is, the levels of the CESA8-type synthases were downregulated, whereas the invertases were upregulated. The findings were validated by qRT-PCR (Fig 6B and 6C).
According to our results, PAO, CESA and CWINV expression levels correlated very well with the type of infection, being more severe during acute infections in both hosts. These changes probably have a role in the development of the stunted phenotype during acute infections.
Different morphological abnormalities which are usually present in acute infection can also be the result of changes in hormone metabolism (reviewed recently in [5]). Auxins, cytokinins, gibberellic acid and brassinosteroids that regulate plant growth, development and elongation can have ambiguous effects during virus infection (reviewed by [44]). Although improper annotation did not allow us to map DEPs and DEGs of these pathways completely, after compiling gene expression changes of all probes and genes annotated to these pathways, these pathways were observed to be differentially regulated in acute infection, whereas changes were almost absent in persistent infection in both hosts (S15 and S16 Tables). Transcription factors regulated by other hormones, such as ethylene, abscisic acid, jasmonic acid and salicylic acid, can differentially regulate senescence-associated genes (SAGs), which finally lead to chlorophyll breakdown and leaf senescence [45,46]. To summarize our transcriptomics results obtained by microarray and RNA-seq experiments, all hormone metabolism annotated changes were collected (S17 and S18 Tables) and visualized as heat maps of these pathways (S8 Fig). The overall or combined effects of hormone biosynthesis and regulation would finally lead to the development of leaf senescence. Changes at each level were more pronounced in the acute infections, which could explain the development of more severe symptoms during this type of infection.

Physiological changes in virus-infected N. benthamiana and S. lycopersicum plants
To check whether the molecular changes caused by the acute infection (vs persistent ones) are able to cause physiological alterations, we characterized the photosynthetic response of the host in the above host-virus interactions by the application of various physiological measurements (chlorophyll fluorescence, leaf temperature, area photosynthetic activity and calculated performance index (PI)) (Fig 7 and Fig 8).
Evaporative cooling of healthy and infected plants could be distinguished by using thermal imaging approaches. With better stomatal regulation, leaves of healthy plants transpire more water than is transpired by infected plants, which can be the effect of stomatal closure induced by the pathogen. Thermal images of acute and persistent virus infections were quantified by thresholding the evaporative cooled area relative to the temperature of the surrounding air (Fig 7). We could observe similar transpiration rates similar to those of the mock-infected plants during persistent (TCV-and TMV-U1 -infected N. benthamiana and S. lycopersicum, respectively) infections. By contrast, during acute (CymRSV, crTMV and PVX) infections the leaf temperature was found to be significantly higher (almost similar to that of ambient air), suggesting that acute viruses have a drastic negative impact on the evapotranspiration of their hosts. The chlorophyll fluorescence fast kinetics approach also proved to be very effective in measuring physiological responses. Chlorophyll fluorescence parameters characterizing the electron transfer on both the acceptor and the donor sides of photosystem II drastically affected the performance index parameter, the maximal quantum yield for primary photochemistry and the quantum yield for electron transport [47,48] in CymRSV-and crTMV-infected N. benthamiana and in PVX-infected S. lycopersicum plants. In TCV-infected N. benthamiana and TMV-U1-infected S. lycopersicum plants, the chlorophyll fluorescence parameters were the same or a little higher than in the control (Fig 8A and 8B).
Physiological data are consistent with molecular changes, and the mild alterations of gene expression during persistent infection do not alter the physiology of the leaves. During acute

Expression of different regulators active in the nucleus showed the same pattern during different infections
Molecular and physiological changes in N. benthamiana and S. lycopersicum plants followed the same pattern in the different viral lifestyles. Severe changes occurred during acute infection but were absent or mild during persistent infection in both hosts, suggesting the presence of a universal regulatory pathway. In vitro transcription assays have proven that shut-off happens at the transcription level in the nucleus [22], where transcription factors regulate their targets. During acute infection, these types of DEPs in N. benthamiana were seriously altered. The expression of six of them was confirmed by Northern blot analysis, and their possible role is discussed in the Supplementary Material (S1 File and S9 Fig). In the nucleus of virus-infected plants, where transcription factors regulate their targets, the transcription factors themselves are often under the control of RNAi-based regulation executed by microRNAs (miRNA) and small interfering RNAs (siRNA) (reviewed by [5,49,50]). During transcriptional gene silencing (TGS) 24-nt-long small RNAs are loaded into Argonaute 4 protein (AGO4) to direct changes in DNA methylation and histone modification. The TGS may be modulated during pathogen attack [51]. When analysed, AGO4 was found to be differently regulated; its level was downregulated only in acute infection. AGO4 mRNA changes were also validated by Northern blot analysis in N. benthamiana (S10A Fig). Methyltransferases changed in a similar manner (S10B Fig).

Persistent downregulation of housekeeping genes during acute infection is not a consequence of necrosis
As we have shown, acute infections have a profound effect on the gene expression pattern and physiology of the hosts. However, whether the molecular and physiological changes are the cause or the consequence of the early necrosis is not clear. Tombusviruses, such as CymRSV and CIRV, systemically infect N. benthamiana plants, and the infection ultimately culminates in the death of the host. These viruses encode a very efficient viral suppressor protein of RNA silencing (VSR), p19. p19 binds virus-derived small interfering RNA (vsiRNA) duplexes in a size-specific manner [52], consequently blocking their loading into RNA-induced silencing complexes and thus inhibiting the antiviral RNA silencing defence [53]. During infection with suppressor-deficient mutant viruses (CymRSV19S and CIRV19S), however, the antiviral system works efficiently, and the plants are able to recover from infection [54]. After a recovery period, an almost healthy phenotype develops (recovery phenotype). In recovery plants, due to the efficient antiviral silencing, the virus is present only at a low level. The efficiency of RNA silencing decreases at lower temperatures [55]. Under low-temperature conditions, suppressor-deficient viruses can also accumulate to a high level. It was also shown that CymRSV19S or CIRV19S viruses do not induce necrosis even at low temperatures [56]. To exclude the possibility that downregulation of housekeeping genes is a consequence of early necrosis, we analysed housekeeping gene expression patterns in response to CymRSV19S-and CIRV19Sinfection at 15˚C. Samples from newly developed, systemically infected leaves were collected at 12 dpi, and the expression levels of two housekeeping genes (Rubisco, CP29) and a stress gene (PR-Q) were investigated by Northern blot analysis (Fig 9).
The Rubisco and CP29 levels were downregulated and the expression of PR-Q was induced irrespective of the presence of the VSR. Similarly, as we showed earlier two other housekeeping transcript (GAPDH and tubulin) levels were reduced during CymRSV19S infection [22]. These findings suggest that the shut-off of housekeeping genes is not a consequence of the upcoming necrosis or the presence of VSR.

Conclusions
In our work, we investigated genome-wide gene expression changes and physiological alterations during acute and persistent infections in two different hosts. Persistent shut-off of housekeeping-and photosynthesis-related transcripts, induction of stress-related transcripts and alterations of cell wall remodelling enzyme transcripts were exclusively caused by viruses that had an acute lifestyle strategy, but not in the case of persistent viruses. Gene expression changes on the global scale (microarray and RNA-seq data) were validated by an alternative method (Northern blot analysis or qRT-PCR) for representative transcripts of each class. Acute virus-specific changes were very similar in the two host plants (N. benthamiana and S. lycopersicum) and in the case of different viruses (Tombusviruses, Tobamoviruses and Potexviruses), thereby suggesting that general mechanisms are involved. Two methods were used to serve as a cross-reference in our study, validating the reliability of the results and strengthening our conclusions.
During infection, viruses harness the host machinery to replicate themselves. During this process, the host endomembrane system, RNA transport, transcription and translation machinery are hijacked by the virus (deployed from the host). This shortage of host enzymatic processes may be directly or indirectly responsible for acute virus-caused gene expression changes and subsequent physiological alterations. However, the extremely high virus level during persistent infection in the investigated host-virus systems (TCV-infected N. benthamiana and TMV-U1-infected S. lycopersicum plants) contradict this hypothesis and suggest that extreme overuse of the host transcriptional and translational system is probably not the primary cause of severe host gene alteration during acute infections.
We show that the housekeeping gene shut-off is not the consequence of the upcoming necrosis or the presence of RNA silencing suppressor protein. Previously it was proposed that interference of the silencing suppressor with endogenous silencing pathways might contribute to viral symptom development [57,58]. This possibility has since been questioned [59] because the p19 suppressor is not able to bind endogenous small RNAs in a virus-infected environment. We propose, therefore, that the severe transcriptome and physiological changes caused by the acute viruses but not alterations of the miRNA-guided developmental pathway are the primary contributors of viral symptoms. Our finding that key regulators of TGS (AGO4) were altered during acute infection implies that the altered TGS-mediated pathway could cause broad-range changes in host gene regulation. Mapping 24-nt small RNA-mediated methylation changes of the host in this type of infection could answer this question in the future. Moreover, as has been shown for AGO1, translational regulation (which cannot be detected by either microarray or RNA-seq experiments) can further influence this complex pattern.
While investigating the mechanism of shut-off in virus-infected mammalian cells, Gilbertson and colleagues recently showed that cytoplasmic degradation of mRNAs can negatively regulate their initial transcription rate in the nucleus [60]. During mRNA degradation, RNAbinding proteins are released and shuttle back to the nucleus, where they block early stages of transcription. During virus infection, the translation machinery is redirected to produce viral proteins, which may affect both host RNA stability and protein translation, leading to the induction of gene expression changes as observed during acute infections. Beside cell-autonomous impacts, viruses may alter plant metabolism on tissue or organism level (water and nutrient transport), which would affect cell-autonomous regulatory pathways. Testing how these regulatory mechanisms operate in virus-infected plants could further contribute to our understanding of host-virus interactions and development of viral symptoms.
We used two different high-throughput methods in our experiments: microarray hybridization and RNA sequencing to detect changes in the gene expression pattern of the host in compatible virus infections. Changes in expression levels were higher in the case of DEPs compared to DEGs, which could be a result of different detection methods. The ratio of downregulated genes in microarray analysis was higher compared to that in upregulated ones, which could be a result of the presence of overrepresented probes. This trend is a warning signal that interpretation of the most extensively changed gene ontology categories cannot be precisely defined based on the microarray data.
Severe downregulation of genes that play a role in photosynthetic and other housekeeping processes has a marked effect on plant physiology. Thermal imaging and variable chlorophyll fluorescence transients to follow leaf temperature and photosynthetic activity, respectively, could measure these effects. According to our results, physiological changes associated with the extensive shut-off of host genes in acute infections make it possible to differentiate the latter from persistent infections, even before visible symptoms appear. Remote sensing methods are able to monitor these physiological changes [61][62][63]; however, the development and use of these new technologies must be based on detailed knowledge of both the changes in gene expression and the resulting physiological changes in virus-infected plants. Consequently, our data could be utilized in the future by remote sensing techniques in precision agriculture, helping crop management by predicting possible infection risks.
High-throughput sequencing methods lead to the discovery of more and more viruses each day, but determining the actual importance (disease-causing potential) of these newly discovered viruses is very difficult. Characterizing gene expression patterns of housekeeping genes and identifying the type of viral infection based on their expression levels could be used in risk assessments ( Table 1).
Coexistence of a virus and its host in a persistent virus infection without visible symptoms could be beneficial for the virus and the host at an evolutional scale. Interestingly, the persistently infecting TCV caused improved physiological attributes (e.g., Performance index increased by 1.4-fold, Fig 8A), suggesting that TCV may be a mutualist rather than an agent of disease. Although we gained deep insight into the regulation of gene expression in various viral infections, how viruses can avoid or minimize the host reaction during persistent infections to reach a 'peaceful' coexistence with their host is a very exciting question that needs to be further investigated.  indicated on the heat map, along with their characteristic parameters, is detailed in the S3 and S4 Tables (for photosynthesis) and in the S5 and S6 Tables (for stress responses Green shows downregulation, whereas red shows upregulation. The list of probes and genes whose levels are indicated on the heat map is detailed in the S17 Table (N. benthamiana) and S18 Table (