Viral RNAs must disguise themselves to mimic eukaryotic mRNA, evade the host immune system, and hijack the host machinery for translation. Eukaryotic mRNA has a unique 5′ cap structure essential for translation initiation. The cap structure, which consists of a methylated guanosine residue, binds translation factors and ribosomes to initiate protein synthesis. Additionally, eukaryotic mRNA has a 3′ poly(A) tail that protects the mRNA from degradation by exonucleases [45]. Many plant RNA viruses lack either a 5′ cap, a 3′ poly(A) tail, or both, preventing recognition and translation by the host machinery. Moreover, many plant RNA viruses are multicistronic, containing several open reading frames (ORFs) within the same RNA strand or encoding a polypeptide that is processed by proteases into multiple peptides. To overcome the lack of a 5′ cap or a 3′ poly(A) tail, viral RNAs have evolved adaptations to recruit ribosomes. These adaptations include, e.g., 5′ internal ribosome entry sites (IRESs), 5′ viral genome-linked protein (VPg), 3′ cap-independent translation enhancers (CITEs), and cap snatching [46]. IRES elements are highly structured RNA sequences that allow ribosomes to initiate translation from a position internal to the RNA molecule, bypassing the requirement of a 5′ cap structure. VPg is a protein covalently linked to the 5′ end of some viral RNAs, and it is thought to enhance translation initiation [46]. CITEs are RNA elements located at the 3′ end of some viral RNAs that enhance translation initiation by interacting with translation initiation factors. Interestingly, both 5′ IRES and 3′ CITE elements are also present in some eukaryotic mRNAs and are thought to stimulate translation during stress and developmental circumstances when cap recognition is hindered [47]. These elements consist of cis-acting secondary structures that recruit host initiation factors or ribosomal subunits, enabling protein synthesis [48,49,50]. Another strategy involves VPg, which plays a crucial role in the translation of viral RNA by interacting with host translation factors, such as eIF4E and its isoform eIF4isoE. VPg competes with the cap structure of host mRNAs to bind these factors, thereby inhibiting host translation and redirecting the host machinery towards viral translation [51,52]. In contrast, some segmented negative-stranded RNA viruses have developed an alternative mechanism called cap snatching to initiate translation [53,54].

Different RNA virus families have developed diverse mechanisms to achieve efficient translation. Depending on the location of the encoded peptide sequence in the viral RNA, commonly used strategies include cap-independent translation, synthesis of subgenomic RNAs (sgRNAs), cap snatching, and translational recoding [55,56,57]. Cap-independent translation occurs at the first 5′ localized ORF and is regulated by IRES and 3′ CITE structures in the viral RNA. These structures interact with translation initiation factors, such as eIF4E and eIF4G, or directly with the ribosomal subunit through 18s rRNA, which stimulates host translation initiation complex assembly and translation of the viral RNA, mimicking cap-dependent translation of eukaryotic mRNA [48,49,50].

In positive-stranded RNA viruses, sgRNAs are synthesized from initial viral RNA by an RNA-dependent RNA polymerase (RdRp) encoded in the viral genome. RdRp recognizes subgenomic promoters in the viral RNA that give rise to different sgRNAs. These sgRNAs enable the host machinery to translate viral proteins located internally or at the 3′ end of the viral RNA [55]. Due to the dense coding of viral genomes, overlapping or adjacent ORFs often require translational recoding for translation. Translational recoding includes leaky ribosome scanning, non-AUG initiation, ribosomal codon read-through, ribosomal frameshifts, and translational bypassing. The mechanism used largely depends on the viral genus [58].

In addition to cis elements that regulate translation, viruses have evolved various strategies to achieve maximum translation efficiency. For example, a commonly observed phenomenon is the induction of a ‘host shut-off’ mechanism, where the translation of endogenous mRNA is suppressed [59]. This can be accomplished through interference with the cap-dependent translation of host mRNAs, resulting in a decrease in, among other things, host antiviral responses and an increase in viral RNA translation. Moreover, viruses have optimized translation by facilitating RNA cyclization, interfering with host translation initiation, and compartmentalizing translation in virus factories (VFs) [60,61]. Viral RNA cyclization is promoted through a specific type of 3′ CITE structure that binds to eIF4F, leading to the interaction of eIF4F with a hairpin structure at the 5′ end of the RNA. In turn, this results in the cyclization of translation [62,63]. This cis-element-stimulated cyclization has been demonstrated for viruses in the Tombusviridae family [62,63]. Multiple viruses interfere with host translation by targeting eIF4E, altering its phosphorylation status [64]. In plants, phosphorylated eIF4isoE shows an increased binding affinity for VPg and enhanced mRNA translation [65,66], perhaps favoring the translation of viral VPg-containing RNAs over host mRNA. VFs, viroplasms, inclusion bodies (IBs), or viral replication complexes (VRCs) are intracellular structures induced by viruses. These membrane-bound inclusion-like bodies or spherules concentrate viral RNA and proteins, potentially enhancing viral replication. The formation of VFs is often facilitated by viral proteins that manipulate membranes of the endoplasmic reticulum, mitochondria, peroxisomes, and/or chloroplast membranes [67,68,69,70,71]. Although VFs are primarily associated with RNA replication, they have also been suggested to play a role in viral translation and cell-to-cell movement in certain cases [72,73].

Symptom recovery is observed in certain plant–virus interactions, where asymptotic leaves emerge or symptomatic leaves recover after systemic infection with a virus. This phenomenon has been reported for numerous unrelated viruses and can also be induced through mutation of VSRs [34,82,83,84,85]. Symptom recovery is associated with systemic sequence-specific resistance, providing protection against reinfections and cross-protection against related viruses [84]. Symptom recovery from RNA viruses can result from either degradation or translational repression of viral transcripts [42,43,86,87,88]. Symptom recovery is the only known phenomenon in plants where PTGS leads to translational repression [34,82].

In Nicotiana benthamiana, symptom recovery from tomato ringspot virus (ToRSV) is associated with PTGS but not accompanied by viral clearance [86]. Instead, ToRSV recovery is accompanied by an accumulation of the ToRSV RNA2 viral segment and a reduction in its encoded CP and MP through translational repression [42,87]. Whether translation of RNA1 of ToRSV is also repressed remains elusive. Similarly, symptomless recovery from an engineered tobacco rattle virus clone (TRV-GFP) infection in Arabidopsis thaliana is associated with a decrease in GFP fluorescence, followed by a drop in the viral titer [43,88]. During this stage, the TRV RNA is less associated with ribosomes, which results in reduced levels of GFP. This observation suggests that translational repression is involved in TRV recovery [43]. In contrast, in N. benthamiana, TRV RNA is known to be targeted for slicing [89]. In A. thaliana undergoing infection by oil-seed rape mosaic virus (ORMV), symptom recovery occurs in newly developed leaves in the presence of infection-competent replicating viral RNAs. It has been proposed that 21–22 nt siRNAs in concert with PTGS and TGS machinery are responsible for this recovery in a non-cell-autonomous manner, generating a source-to-sink siRNA gradient, where VSR function becomes saturated, perhaps inhibiting viral protein translation. Accordingly, A. thaliana mutants that result in a dysfunctional PTGS and/or TGS fail to undergo symptom recovery, while impairing RNA decay machinery enhances such a response, probably by allowing the accumulation of dsRNA, which can be processed into vsRNAs and exported to new leaves [90]. These data suggest that PTGS participates in translational repression during viral recovery. Moreover, under certain interactions, translational repression and slicing can occur in parallel [87].

During NLR-mediated TA, the translation of viral transcripts of the virus triggering the immune response is arrested, as well as that of transcripts from other viruses present in the cell [25,39,91]. The resulting TA appears to be cell-autonomous and only occurs in cells where the NLR is activated [25,39]. Two structurally divergent NLRs, a CNL (Rx1) and a TNL (N), are known to induce TA in N. benthamiana [25,28].

Rx1 has been introgressed into commercial potato cultivars from the wild potato species Solanum andigena [31,32,92]. The Rx1 gene confers extreme resistance to PVX. In transgenic N. benthamiana, Rx1 is activated in the cytoplasm upon recognition of the PVX CP [32,93,94,95,96]. Recently, it was shown that Ran GTPase Activating Proteins 1/2 (RanGAP1/2) are targets of PVX CP [97]. Since inactive Rx1 and RanGAP2 interact through their CC and WPP domains, respectively, Rx1 activation likely occurs by indirect recognition through RanGAP2 [97,98,99]. Once activated, Rx1 is translocated to the nucleus, which is essential to the mounting of a full immune response [28,95,96,100]. Rx1 shuttling is mediated by RanGAP2 and the co-chaperone SUPRESSOR OF G2 ALLELE OF SKP1 (SGT1) [96,99,100]. In the nucleus, the activated form of Rx1 directly binds and distorts double-stranded DNA through its NB-ARC domain [93]. The DNA-binding capacity of Rx1 could enable it to function as a transcriptional regulator by facilitating access to DNA for transcription machinery. The binding specificity of Rx1 is suggested to be provided by the Golden2-like transcription factor (TF) NbGlk1 [101]. In the absence of PVX, the CC domain of Rx1 binds to NbGlk1 and a bromodomain (BD)-containing protein (NbBDCP), preventing chromatin interaction [23]. Upon PVX infection, NbGlk1 mediates Rx1 binding to Golden2-like consensus DNA sequences, which may lead to TA by regulating unknown target genes.

N mediates resistance to TMV through recognition of the p50 helicase domain present in the TMV replicase [102,103]. In N. tabacum, activation of this NLR results in HR [33]. However, when introduced in N. benthamiana, N activation leads to ER through TA [25,33]. Recognition of p50 is mediated by N-receptor-interacting protein 1 (NRIP1), which is normally localized in chloroplasts [104]. However, p50 recruits NRIP1 to the cytoplasm and nucleus to form p50-NRIP1 complexes [104]. Afterward, NRIP1 binds to the TIR domain of N, triggering activation [104,105]. The N-mediated defense response encompasses multiple pathways. However, the specific mechanism by which it induces TA remains unknown. Currently, the only protein suggested to be involved in TA is the helper NLR N REQUIREMENT GENE 1 (NRG1) via an unknown mechanism [39]. Future studies aimed at unveiling the proteins involved in NLR-mediated transcriptional activation could help determine whether both NLRs require the same partners for signaling or utilize distinct pathways. Differences in sensitivity to the VSR p38 of turnip crinkle virus (TCV) suggest that the mechanisms that lead to TA are different for N and Rx1 [28].

Alternatively, virus-derived nucleic acids from begomovirus, a DNA virus, can trigger a global TA through NIK1 [40,41]. NIK1 is part of the same LRR-RLK subfamily as BRI1-ASSOCIATED RECEPTOR KINASE-1 (BAK1), which is well known for its role in plant defense against bacteria, fungi, and oomycetes [106,107,108]. The global TA following NIK1 activation is thought to be induced through the downregulation of translational-machinery-related genes [109]. This downregulation is indirectly facilitated by the ribosomal protein L10A (RPL10A), which translocates to the nucleus upon phosphorylation by NIK1, where it interacts with the L10-interacting MYB domain-containing (LIMYB) TF [41,41]. However, the mechanism of NIK1 activation and the signaling pathway leading to global TA remain obscure, though transcriptional reprogramming might be common during TA in response to plant viruses.

To mount a TA of viral RNAs, the plant must distinguish viral RNAs from endogenous RNAs. Recognition of a viral transcript can occur through the sequence specificity brought by vsRNA in an RISC complex or by the binding of RNA-binding proteins (RBPs) to common viral motifs present in the RNA. Some of these common motifs include dsRNA, the 5′-ppp region of the RNA, the 5′ UTR region, and hairpin structures [110,111,112,113].

During viral recovery, transcript recognition is assumed to occur through sequence specificity provided by siRNAs [38]. This hypothesis is supported by the observation that plants have sequence-specific resistance to reinfection after viral recovery [84]. However, transcript recognition has not been reported for ToRSV or TRV recovery, where translational repression is involved. In NLR-mediated TA, viral transcripts present in the cell are halted [25,39,91]. Most likely, a common element present in viral RNA is recognized by an RBP, resulting in the translational repression of viral transcripts. Evidence suggests that a structural motif present in the CP-encoding part of the PVX RNA is essential for TA of these transcripts [25]. In correspondence with this, the translation of PVX transcripts lacking the CP region was no longer halted when N-mediated TA was triggered. The PVX-induced TA could be reverted when the CP ORF was replaced with another viral CP with low sequence similarity, suggesting that the recognition and subsequent targeting for repression is likely caused by a structural element in this region. Moreover, activation of the NLR Tm-2a from a wild relative of tomato (Solanum lycopersicum), Solanum peruvianum, inhibited the accumulation of PVX but not that of PVXΔCP in tobacco [25]. This suggests that the structural element present in the CP region might be a common target for NLRs. However, the possible binding partner and RNA structure both remain unidentified. Other elements of the viral RNAs, such as their 5′ cap structure and their poly-A tail at the 3′ region, are unlikely to be recognized during NLR-mediated TA, since TCV sgRNA lacks these structures and is still halted during N-mediated TA [39,114].

After dsRNA sensors recognize the viral transcripts, the translation of transcripts can either be repressed before translation initiation or later, during translation elongation. For example, in miRNA-mediated translational repression, translation initiation is known to be inhibited through the physical blocking of mRNAs by AGO1-RISC binding at the 5′ UTR to prevent recruitment of the 48S ribosomal subunit and ribosome assembly [115]. In addition, depending on the position of the RISC target site, the binding of RISC to the RNA can physically block translation elongation [115]. The stage during translation at which the transcripts are halted can be determined through polysome profiling, as the association of ribosomes is prevented by translational inhibition prior to translation initiation [116]. During ToRSV recovery, monosome and polysome profiling revealed that translation inhibition occurred after translation initiation. However, during TRV recovery, translation inhibition appeared to occur before translation initiation [43,87]. In NLR-mediated TA, polysome profiles revealed that PVX RNAs were no longer associated with polysomes [39]. In N-mediated TA, PVX RNA was prohibited from associating with monosomes, which suggests that repression occurs before translation initiation [25]. However, the mechanism behind translation initiation inhibition during NLR-mediated TA remains obscure.

Liquid–liquid phase separation (LLPS) is a physical phenomenon where a homogeneous solution separates into two distinct liquid phases, driven by changes in concentration, temperature, or other factors [117,118]. RNA granules, which comprise all intracellular aggregations of ribonucleoprotein (RNP) complexes large enough to be microscopically visible, can undergo LLPS. Cells undergoing a viral infection often accumulate RNA granules, which dynamically influence each other by exchanging components, such as transcripts and proteins [119].

In plants, RNA granules can be classified into PBs, stress granules (SGs), and siRNA bodies. In PBs and SGs, LLPS behavior is thought to be driven by multivalent interactions between RBPs, RNA, and other proteins that enable the formation of weak reversible bonds, which, in turn, can dynamically change the properties of the granules [120,121,122,123]. PBs are essential in regulating gene expression, mRNA decay, and translation [124]. These bodies reside within cells to process aberrant mRNA transcripts. During mRNA decay, endogenous aberrant mRNA or viral RNAs are degraded through decapping by the mRNA decapping protein 2 (DCP2), followed by de-adenylation and 5′ to 3′ decay by the exoribonuclease XRN4 [124]. Decapping is assisted by the co-activator DCP1 and scaffold varicose (VCS) [125,126]. Besides mRNA decay, PBs are sites for storing translationally repressed RNAs [121]. Once the abundance of repressed RNA increases due to RNAi, virus infection, or UV irradiation, the number and size of PBs changes consequently [36,39,43,124,127]. Plant viruses have been proposed to utilize components of PBs for their benefit, and PBs have been suggested to store repressed viral RNA during immune responses and viral recovery [39,43,128,129].

During TRV recovery, A. thaliana shows an increase in PBs potentially containing repressed TRV RNA aggregates. Eventually, these RNAs are degraded by decapping enzymes present in these PBs. However, degradation of the TRV RNA is not necessary for recovery, as was demonstrated by the normal TRV recovery observed in the DCP2 mutant its1 [43]. Although DCP1 was used as a PB marker, the exact composition of the PBs that form during TRV recovery remains unknown.

When cells underwent NLR-mediated TA, PBs were shown to accumulate as well [39]. These PBs contained DCP1 but were depleted of DCP2, resulting in transcript accumulation caused by the lack of RNA degradation [39]. The protein composition within these types of PBs remains unknown so far. After NLR activation, the cellular DCP1 levels remain unchanged, which suggests that a pre-existing pool of DCP1 acts as a nucleation point for PB formation [39]. While Meteignier et al. (2016) suggested that the lack of DCP2 in PBs could be due to limited levels of cellular DCP2, previous studies have shown that DCP1 and -2 are both recruited into PBs depending on the stress perceived and that this is not necessarily reflective of their cellular concentrations [130]. Therefore, it is likely that the depletion of DCP2 in NLR-mediated PB formation is an active exclusion rather than a passive reflection of PB protein component concentrations in the cytoplasm. However, the mechanisms that stimulate PB formation during NLR-mediated immunity remain to be elucidated.

The formation of PBs in response to NLR activation occurs through a different mechanism than that triggered by UV irradiation or RNAi [36,39]. For example, UV irradiation triggers phosphorylation of eIF2a, resulting in a global TA [131], which does not occur after N activation [39]. Additionally, when PTGS is repressed by VSR P19, PB formation is reduced, while P19 has no influence on PB formation following N activation [39]. Although the mechanisms that induce PB formation after NLR activation are distinct, whether the formed PBs are qualitatively similar or distinct remains under debate and requires further research. Whether PBs form during ToRSV recovery is unknown. However, it is unlikely, since the inhibition of translation initiation is known to result in PB formation, whilst inhibiting translation elongation leads to a decrease in PBs [36]. As discussed, translational repression during ToRSV recovery likely happens during elongation, whilst TRV recovery and NLR-mediated TA occurs before translation initiation. However, the exact mechanisms underlying PB formation have not been investigated for these viruses.

AGOs form an integral part of the RNAi pathway, where they act as the catalytic subunit of the RISC complex. Multiple AGOs (AGO1, AGO2, AGO4, and AGO7) are known to be involved in and are essential for antiviral immune responses [25,43,80,132,133,134,135,136]. All these AGO proteins have RNA slicing activity, whilst only AGO1 has been proven to be directly involved in translational repression [38,80,114,115,137,138]. AGOs can directly perform RNA slicing through their endonucleolytic PIWI domain [139]. Although the composition of the AGO protein interactome remains elusive, translational repression by AGO is thought to operate through association with WG/GW motif-containing proteins [140,141,142,143]. In A. thaliana, genetic studies identified a gene encoding the GW-containing protein SUO involved in miRNA-mediated translational repression [138]. However, whether SUO directly interacts with AGO through its GW motifs to perform its function has not been investigated, and how AGOs determine whether to perform slicing or repression is poorly understood. Possibly, AGO function is regulated through post-translational modifications, such as phosphorylation or changes in subcellular localization [42].

Multiple VSRs interact with AGOs to impair RNAi. For instance, AGO1 expression, stability, and activity can be targeted by VSRs [8,14,87,137,144,145]. In line with this, VSRs, including the carmovirus p38 and ipomovirus P1, contain the AGO-interacting motif WG/GW and compete with host proteins for AGO binding [144,146]. Some VSRs have multiple strategies to interfere with plant signaling. For example, p38 of TCV prevents the processing of dsRNA into siRNA and impairs siRNA loading into AGO1 and AGO2 in A. thaliana [144,147,148].

During symptom recovery of ToRSV in N. benthamiana, translational repression is AGO1-dependent [42]. It is likely that AGO1 is (in)directly involved in the PTGS mechanism leading to translational repression of ToRSV RNA2. This is supported by the finding that ToRSV CP hinders AGO1 function through interaction with the WG/GW motif of AGO1, triggering AGO1 degradation and probably competing with cellular WG/GW proteins involved in translational repression [13,87]. During TRV infection in A. thaliana, AGO2 and AGO4 are involved in initial susceptibility to TRV, whilst other unidentified proteins are required during recovery [43]. Multiple VSRs are known to affect symptom recovery and the expression of strong VSRs, such as HC-Pro from potyvirus, and p25 from PVX can eliminate symptom recovery during ToRSV infection [149]. Similarly, viral recovery of TRV was abolished by p38 of TCV [43]. Moreover, the inactivation of VSRs can lead to symptom recovery of virus isolates that typically do not display recovery. This has been shown for 2b of cucumovirus, HC-Pro of potyvirus, and P19 of tombusvirus [150,151,152,153]. Recovery in ToRSV and TRV is not dependent on a lack of VSRs in the virus isolates but rather on the relative strength of the VSRs present. As mentioned earlier, ToRSV CP acts as an AGO-hook triggering AGO1 degradation, and TRV 16K acts as a VSR by preventing AGO4-RISC assembly [87,154]. Additionally, the loss-of-function mutants for AGO1 and other genes involved in siRNA signal amplification, such as RDR6, SGS3, and DCL4, fail to alleviate infection symptoms [90]. These observations together suggest that AGO protein(s), in complex with partners containing WG/GW motifs, could participate in the TA of RNA viruses and that VSRs target this function.

Rx1-mediated and N-mediated TA are AGO1-independent and AGO4-dependent [25]. AGO4 is involved in resistance to many viruses through the RNA-directed DNA methylation (RdDM) pathway, yet it is also involved in antiviral mechanisms unrelated to RdDM in the cytoplasm [135,155,156]. During plantago asiatica mosaic virus (PIAMV) infection, AGO4 is suggested to re-localize to the cytoplasm and directly target the PIAMV RNA [157]. The exact mechanisms behind this interaction remain unknown. It is likely that AGO4 acts independently of the RdDM pathway as well during NLR-mediated TA, since the viral replication cycle is restricted to the cytoplasm. Involvement of the AGO proteins in NLR-mediated TA is further proven by the expression of VSR p0 from beet western yellows virus (BWYV), p38 of TCV, and p19 of cymbidium ringspot virus (CymRSV) [25,28,158]. p0 inhibits N-mediated TA and is known to induce the degradation of AGO proteins in A. thaliana [25,146]. However, its effect on Rx1-mediated TA has not been tested. Possibly, N-mediated TA is repressed by p0 promoting the degradation of AGO4. Besides p0, p38 of TCV inhibits N-mediated TA [25]. However, p38 does not physically interact with AtAGO4, suggesting that N-mediated TA requires additional AGO proteins or that p38 inhibits TA at the dsRNA processing level [144]. The latter is unlikely since NLR-mediated transcript recognition is supposed to occur through common motif recognition rather than through siRNA complementarity, suggesting that the process might be siRNA-independent. Rx1-mediated TA is not inhibited by p38, suggesting a distinct mechanism leading to Rx1-mediated TA [28]. p19 of CymRSV is known to repress systemic PTGS by binding dsRNAs in order to prevent incorporation into RISC complexes. However, TA induced by both Rx1 and N is not influenced by p19, further supporting a distinction between the TA mechanisms involved in NLR and viral recovery [28,133,158].