Suppression of a dsRNA-induced plant immunity pathway by viral movement protein

The virome of plants is dominated by RNA viruses 1 and several of these cause devastating diseases in cultivated plants leading to global crop losses 2. To infect plants, RNA viruses engage in complex interactions with compatible plant hosts. In cells at the spreading infection front, RNA viruses replicate their genome through double-stranded RNA (dsRNA) intermediates and interact with cellular transport processes to achieve cell-to-cell movement of replicated genome copies through cell wall channels called plasmodesmata (PD) 3. In order to propagate, viruses also must overcome host defense responses. In addition to triggering the antiviral RNA silencing response, RNA virus infection also elicits pattern-triggered immunity (PTI) 4 whereby dsRNA, a hallmark of virus replication, acts as an important elicitor 5. This innate antiviral immune response is also triggered when dsRNA is applied externally and does not require sequence homology to the virus 5. However, the mechanism by which PTI restricts virus infection is not known. Here, we show that dsRNA inhibits the progression of virus infection by triggering callose deposition at plasmodesmata and the inhibition of transport through these cell-to-cell communication channels. The dsRNA-induced signaling pathway leading to callose deposition is independent of ROS production and thus distinguished from pathways triggered by bacterial and fungal elicitors. The dsRNA-induced host response at plasmodesmata is suppressed by the Tobacco mosaic virus movement protein (MP). Thus, the virus uses MP to inhibit innate dsRNA-induced immunity at plasmodesmata, which could be a general strategy of phytoviruses to overcome plant defenses and spread infection.

(flg22) than in control plants treated with buffer ( Fig. 1). The treatments did not cause a significant change in GFP fluorescence ( Fig. 1) and had no bulk effect on viral RNA accumulation in the infected cells (Fig. S1). The smaller infection sites in treated leaves suggested that the immunity is linked to reduced cell-to-cell movement of the virus.
Virus movement occurs through plasmodesmata (PD). We thus hypothesized that dsRNA inhibits virus movement by causing PD closure. Treatment of N. benthamiana plants with poly(I:C), flg22, or water, and quantification of PD-associated callose by in vivo aniline blue staining revealed that both poly(I:C) and flg22 trigger an increase in PD-associated callose levels in a concentration-dependent manner ( Fig. 2A and 2B). In agreement with this observation, treated tissues showed reduced PD permeability as determined by a GFP mobility assay ( Fig. 2C and 2D). As previously noted in Arabidopsis 5 , poly(I:C) triggered MPK activation but the level of activation is significantly weaker than the activation observed with flg22 (Fig. 2E). Treated  and ENHANCED DISEASE SUSCEPTIBILITY 1 (EDS1), but not the gene for BRASSINOSTEROID INSENSITIVE 1 (BRI1) (Fig. S2).
To determine how dsRNA elicits the deposition of callose at PD, we turned our attention to Arabidopsis.
Treatment of A. thaliana Col-0 plants with poly(I:C) led to increases in PD-associated callose levels similar to those in N. benthamiana (Fig. 3A). These treatments confirmed the ability of poly(I:C) to activate MPK and to inhibit seedling growth (Fig. S3, A-C), as previously reported 5 . Interestingly, dsRNA-induced callose deposition was strongly inhibited in bik1 pbl1 plants (Fig. 3B), which are deficient in the receptor-like cytoplasmic kinase (RLCK) BIK1 and its homolog PBS1-LIKE KINASE1 (PBL1). BIK1 is a central component of PTI signaling that integrates signals from multiple pathogenrecognition receptors (PRRs), as shown by its direct interaction with FLS2, EFR, PEPRs, and CERK1 7-9 and its ability to phosphorylate and activate downstream targets, such as the NADPH Oxidase RBOHD 10 . BIK1 and PBL1 have additive effects; the bik1 pbl1 double mutant was shown to strongly inhibit PAMP-induced defense responses 8 . In addition to the defect in dsRNA-induced callose deposition, bik1 pbl1 plants are also deficient in poly(I:C)-induced MPK activation and growth inhibition (Fig. S3, D-F). These findings establish a role of BIK1/PBL1 in the innate immunity triggered by dsRNA that leads to callose deposition at PD.
To further determine the signaling pathway downstream of BIK1, we examined whether the treatment of plants with poly(I:C) leads to the production of reactive oxygen species (ROS). To our surprise, neither the treatment of Arabidopsis Col-0 plants (Fig. 3C) nor the treatment of N. benthamiana plants ( Fig. 3D) with poly(I:C) led to the production of ROS. By contrast, strong responses were recorded in both plant species upon treatment with the flg22 elicitor. These observations indicate that dsRNAinduced callose deposition and the dsRNA-based perception of viruses are independent of ROS.
Confirming this result, rbohd and rbohf mutants deficient for the major ROS producing NADPH oxidases RBOHD and RBOHF 11 responded like wild-type Col-0 plants to the presence of dsRNA, showing induced callose deposition at PD (Fig. 3E).
To determine the signaling pathway induced by dsRNA, additional mutants were tested. We started with Arabidopsis mutants deficient in the PD-localized proteins (PDLPs), which are a family of eight proteins that dynamically regulate PD. For example, PDLP5 mediates salicylic acid (SA)-induced callose deposition and PD closure, which is required for plant resistance to bacterial pathogens 12,13 . pdlp5 mutant plants showed strong callose deposition at PD upon poly(I:C) treatment (Fig. 3F), indicating that dsRNA-induced callose deposition is independent of PDLP5 and, thus, of a potential SA response mediated through PDLP5. pdlp1 pdlp2 double mutant plants also showed normal callose deposition. In contrast, pdlp1 pdlp3 double mutant and pdlp1 pdlp2 pdlp3 triple mutant plants were unable to significantly increase PD-associated callose levels in response to poly(I:C) (Fig. 3G). The involvement of PDLP1 and/or PDLP3 is consistent with their redundant roles in callose deposition at PD 14 and also in callose deposition within haustoria formed in response to infection by mildew fungus 15 .
We found that dsRNA-induced callose deposition at PD also depends on the Ca 2+ -binding, PD-localized CALMODULIN-LIKE protein 41 (CML41). This protein was shown to mediate rapid callose deposition at PD associated with decreased PD permeability following flg22 treatment 16 . Plants of a CML41 overexpressing transgenic line (CML41-OEX-2) showed increases in PD-associated callose upon poly(I:C) treatment, like non-transgenic Col-0 controls. By contrast, plants in which CML41 is downregulated by an artificial miRNA (CML41-amiRNA-1) showed a decreased response (Fig. 3H).
Consistent with the role of CML41 in the callose deposition response to dsRNA, the permeability of PD was previously shown to be sensitive to cytosolic Ca 2+ concentrations 17,18 . To test the role of Ca 2+ in dsRNA-triggered innate immunity, we treated plants with poly(I:C) together with EGTA, a Ca 2+chelating molecule. EGTA reduced the level of callose induced at PD after dsRNA treatment (Fig. 3I).
This effect was EGTA concentration-dependent and indicates a role of Ca 2+ in dsRNA-triggered PD regulation. Together, these results suggest a role for CML41 and Ca 2+ in the dsRNA-induced defense response at PD.
We previously showed that poly(I:C) treatment protects Arabidopsis plants against infection by Oilseed rape mosaic virus (ORMV) 5 . If BIK1 and CML41 mediate dsRNA-induced PD closure, we hypothesized that this antiviral defense should be weakened in bik1 pbl1 mutant plants and in the CML41 amiRNA-1 plants. Whereas poly(I:C) treatment prevented symptoms at 28 dpi and resulted in a strongly reduced virus titer in wild-type Col-0 plants (as previously reported) 5 , both the bik1 pbl1 and CML41 amiRNA-1 plants showed symptoms and accumulated high virus levels in poly(I:C)-treated plants, like in ORMV-infected but untreated controls ( Fig. 3J and 3K). Ablation of virus-inoculated leaves from plants at different times after inoculation showed that the time required for the virus to exit the inoculated leaf and to cause systemic infection was 3 days in Col-0 plants. By contrast, this time was reduced to 24 hours in bik pbl1 mutants and CMLl41-amiRNA-1 plants (Fig S4, A and B). Taken together, these findings show that dsRNA-induced callose deposition at PD correlates with antiviral plant defense at the level of virus movement. Furthermore, the experiments reveal a signaling pathway that requires BIK1/PBL1, CML41 and Ca 2+ , as well as PDLP1 and PDLP3 for callose deposition. This dsRNAinduced callose deposition is independent of ROS and thereby distinguished from innate immunity pathways that close PD in the presence of fungal and bacterial elicitors 10,19 .
The plant-pathogen arms race causes pathogens to evolve effectors that overcome host defenses. which is the center of the infection site where MP is no longer expressed. Zones II-IV continuously accumulate dsRNA in distinct replication complexes that also produce MP (Fig. S5). Aniline blue staining demonstrates high PD-associated callose levels within and around the infection site (Fig. 4B).
However, cells in zone II and zone III, where virus cell-to-cell movement is associated with a transient activity of MP in increasing the PD size exclusion limit 20 , exhibit a marked reduction in PD-associated callose levels as compared to cells in zone I (ahead of infection) and zone IV (center of infection) ( Fig.   4B and C). The low level of PD-associated callose in cells at the virus front (zone II) is consistent with the ability of MP to transiently suppress dsRNA-induced immunity at PD.
To test this hypothesis, we examined whether MP suppresses the poly(I:C)-induced callose deposition at PD in the absence of viral infection. Transgenic N. benthamiana plants that stably express MP:RFP have the ability to complement a MP-deficient TMV mutant for movement, thus indicating that the MP:RFP in these plants is at least partially active (Fig. S6). Treatment of such plants with poly(I:C) showed a 50% lower induction of callose deposition at PD as compared to wild type plants ( Fig. 4D and   E). The induction of callose by poly(I:C) was also reduced upon transient expression of MP:GFP ( Fig.   4F and G). Importantly, the same effect was observed with MP C55 :GFP. This mutant MP lacks 55 amino acids from the C-terminus but still accumulates at PD and is functional in TMV movement 21 . By contrast, dysfunctional MP P81S carrying a P to S substitution at amino acid position 81, fails to target PD 22 and does not interfere with poly(I:C)-induced callose deposition. These experiments show that the TMV MP has the capacity to significantly reduce the dsRNA-induced callose deposition at PD, and that this inhibition of callose deposition requires a MP that can facilitate virus movement.
Replication of an MP-deficient TMV replicon was previously shown to induce callose deposition at PD 23 , but the viral molecules inducing this deposition remained obscure. Consistent with findings that viruses induce innate immunity 4 and that dsRNA is a potent PAMP in plants 5 , we show here that dsRNA-induced immunity leads to PD callose deposition and closure. The required signaling pathway involves BIK1/PBL1, CML41, Ca 2+ , and PDLP1,2,3 but not ROS production or PDLP5. Thus, although RBOHD and ROS production are a hallmark of PTI 24 and both ROS and PDLP5-mediated SA signaling play a role in PD regulation 25 , other pathways also exist. The lack of ROS production distinguishes virus/dsRNA-induced signaling from the ROS-associated responses induced by other pathogens. The PD-localized CML41 protein was previously shown to participate in flg22-triggered, but not chitintriggered PD callose deposition 16 . Thus, although differing in upstream components, dsRNA and flg22induced signaling act on shared PD-associated regulatory components. The absence of ROS signaling in dsRNA-induced PD regulation could reflect this specific elicitor type or its location of perception. ORMV replicate in association with punctate sites at the cortical endoplasmic reticulum (ER) that may represent ER:PM contact sites that also occur at PD and may allow interaction of the viral replication complexes with PM-localized signaling proteins. Importantly, dsRNA-induced innate immunity is unaffected by mutations in dsRNA binding DICER-LIKE (DCL) proteins, which excludes these proteins as the dsRNA receptors for PTI and shows that dsRNA silencing and dsRNA-induced innate immunity require different protein machineries 5 .
TMV and its MP represent the paradigm model for virus movement 27 . The ability of MP to reduce poly(I:C)-induced callose deposition at PD is consistent with previous findings suggesting that viral MPs operate callose-degrading enzymes or PD structural components to increase the PD size exclusion limit 28-30 . Our results widen this model by suggesting that viral MPs may act through interaction with components of the dsRNA-induced immunity pathway leading to PD closure. Identifying the PTI dsRNA receptor and also the components through which MP reduces the dsRNA-induced PTI response at PD will be the challenge for future studies.  Immediately after treatment, remaining elicitors, buffers and virions were washed off the leaf surface.
Symptoms were analysed at 28 dpi. At the same time young, systemic leaves were sampled for analysis of virus accumulation by quantitative Taqman RT-PCR using previously described methods 5,38 .
Analysis of differential gene expression measurement by RT-qPCR N. Benthamiana leaf discs were excised with a cork borer and incubated overnight in 12-well plates containing 600 μl deionized, ultra-pure water. The leaf disks were washed several times with water and then incubated with elicitor (1 μM flg22, 0.5 μg/μl poly(I:C), or water as control) for 3 hours. After washing the discs with deionized, ultra-pure water three times, samples were ground to a fine powder in liquid nitrogen and total RNA was extracted by TRIzol™ reagent according to the protocol of the manufacturer. 2 ug of RNA were reverse transcribed using a reverse transcription kit (GoScript™ Reverse Transcription System, Promega). The abundance of specific transcript was measured by probing 1 μL cDNA by quantitative real-time PCR in a total volume of 10 µl containg 5 μL SYBRgreen master mix (Roche), 0.5 μM forward and reverse primer and water. PCR was performed in a  Table S1.

Competing Interests
Authors declare that they have no competing interests

Data and materials availability
All data are available in the main text or the supplementary materials