To identify P1-interacting host factors, we performed a Y2H screen of Arabidopsis thaliana cDNA library using turnip mosaic virus (TuMV)-encoded P1 as the bait. Yeast strain AH109 co-transformed with pGADT7 empty vector and pGBK-P1 did not survive on the selective media lacking tryptophan, leucine, histidine, and adenine (−WLHA), indicating that TuMV P1 has no self-activation activity (Figure 1A). The bait plasmid was then cotransformed with A. thaliana cDNA library in the pGADT7 vector. To avoid missing weak P1-interacting host factors, two rounds of screen were performed on the −WLHA and −WLH selective media, respectively. A total number of five and 18 colonies were recovered on the −WLHA and −WLH selective media, respectively. Plasmids were extracted from each of these colonies and Sanger sequenced. The results showed that the insertions of the 23 plasmids belong to eight genes: stress upregulated NOD19, VASCULAR ONE ZINC FINGER PROTEIN 2 (VOZ2), TEOSINTE BRANCHED1/CYCLOIDEA/PROLIFERATING CELL FACTOR 21 (TCP21), FORMS APLOID AND BINUCLEATE CELLS 1C (FAB1c), CYTOCHROME C OXIDASE 11 (COX11), NUCLEAR TRANSPORT FACTOR 2 (NTF2), LIGHT-HARVESTING CHLOROPHYLL-PROTEIN COMPLEX I SUBUNIT A4 (LHCA4), and the GIBBERELLIC ACID (GA)-STIMULATED IN ARABIDOPSIS 14 (GASA14; Table 1). We retransformed yeast cells with pGBKT7-P1 and each of the eight genes in pGADT7 vector and plated on −WLH or −WLHA selective media. The results showed that yeast cells transformed with all combinations of plasmids survived on −WLH media, but only those transformed with pGBKT7-P1 and either pGADT7-FAB1c or pGADT7-NTF2 grew on −WLHA media (Figure 1B), indicating that P1 may interact strongly with FAB1c and NTF2 and weakly with the other six proteins. In this study, we report the interaction between P1 and NOD19 and the biological function of the interaction in viral proliferation. The biological function of the interaction between P1 and other genes will be reported elsewhere.

Since only the N-terminal part of NOD19 was found in the three plasmids obtained from the Y2H screen, the full-length coding sequence of NOD19 was cloned from A. thaliana and was used to confirm the interaction with P1. Unexpectedly, full-length Arabidopsis NOD19 failed to interact with TuMV P1 in the Y2H assay, suggesting that the missing C-terminal part of NOD19 may influence its interaction with TuMV P1. As an alternative, bimolecular fluorescence complementation (BiFC) assay was performed to confirm the interaction. Leaves of four-week-old N. benthamiana seedlings were infiltrated with Agrobacterium harboring N-terminal part of YFP (YN) and C-terminal part of YFP (YC)-tagged NOD19 (NOD19-YC), YC and C-terminal YN-tagged P1 (P1-YN), P1-YN and NOD19-YC, or NIb-YC and YN-tagged NON-EXPRESSOR OF PATHOGENESIS-RELATED GENES1 (NPR1-YN). At 2 days post-infiltration (dpi), the infiltrated leave areas were analyzed by a laser scanning confocal microscope. The results showed that the epidermal cells of N. benthamiana leaves infiltrated with Agrobacterium harboring NPR1-YN and NIb-YC, which were included as a positive control (Martínez et al., 2023), displayed bright fluorescence in the nucleus, while no fluorescence was observed in those infiltrated with Agrobacterium harboring P1-YN plus YC or YN plus NOD19-YC (Figure 2A), confirming the specificity of the BiFC assay in our experimental conditions. Epidermal cells of N. benthamiana leaves infiltrated with P1-YN and NOD19-YC also showed bright fluorescence (Figure 2A), indicating that TuMV P1 interacts with full-length NOD19. Moreover, in contrast to the nuclear fluorescence from the interaction between NIb and NPR1, the fluorescence from the interaction between P1 and NOD19 did not localize the nucleus, but instead at the cell periphery, e.g., cytoplasm, on a membrane within or bordering the cytoplasm, or even in the extracellular space (Figure 2A). We thus analyzed the subcellular localization of P1 as N- or C-terminal YFP tagged recombinant protein in N. benthamiana epidermal cells. To avoid the release of P1 from the P1-GFP fusion, the last three amino acids of the cleavage site (Val-His-Phe) was deleted. Consistent with previous report of tobacco etch virus (TEV)-encoded P1 (Martínez and Daròs, 2014), TuMV P1 also distributed both in the cytoplasm and nucleus (Figure 2B). We further confirmed the interaction between TuMV P1 and a NOD19 mutant lacking the C-terminal 43 residues (NOD19ΔC) by BiFC. Results showed that the fluorescence from the interaction between P1-YN and NOD19ΔC-YC was observed in both the nucleus and cytosol. Together, these data suggest that full-length Arabidopsis NOD19 may only interact with TuMV P1 outside the nucleus, which explains the negative results of the Y2H assay.

We then determined the expression profile of NOD19 in A. thaliana by reverse transcription-polymerase chain reaction (RT-PCR). The results showed that transcript of NOD19 was detected in shoot meristem, flower, leaf, and stem, but not in the root or seed (Figure 3A). Bioinformatic analyses showed that the N-terminal 26 amino acid of NOD19 may function as a signal peptide, the middle part is the stress upregulated Nod 19 super family domain (SURNod19), while its C-terminal contains a hydrophobic helix (Figure 3B). We also analyzed the alphafold2-predicted structure of NOD19 and found that the middle SURNod19 domain of NOD19 is folded into a compact structure with high confidence score (Figure 3C). The predicted signal peptide at the N-terminal end and the putative transmembrane motif at the C-terminus formed into an α-helix structure with relative lower confidence score (Figure 3C). We further analyzed the subcellular localization of NOD19 using C-terminal YFP-tagged construct (NOD19-YFP) in N. benthamiana epidermal cells. We found that the fluorescence of NOD19-YFP was located at the cell periphery (Figure 3D). Detailed inspection showed that hollow fluorescent vesicles was always found near the cell periphery (Figure 3D). Together, these data suggest that NOD19 maybe a membrane-associated protein that may interact with endomembrane via a hydrophobic α-helix at the C-terminus.

To understand the role of NOD19 in viral proliferation, we obtained a T-DNA insertion mutant of NOD19 (nod19-1; SALK_034861C) from the Arabidopsis Biological Resource Center (ABRC). Genotyping, Sanger sequencing, and RT-PCR showed that the T-DNA was inserted in the first exon of NOD19, which caused the loss of full-length NOD19 transcript (Figures 4A,B). Three-week-old seedlings of wild-type A. thaliana ecotype Col-0 (WT) or nod19-1 were inoculated with TuMV-GFP, an infectious cDNA clone expressing a free GFP between P1 and HcPro cistrons (Cotton et al., 2009), by agroinfiltration. At 14 dpi, we found that about 11.4% leaf area of WT seedlings showed GFP signal from TuMV-GFP, while only about 5.8% leaf area of nod19-1 seedlings showed GFP signal (Figures 4C,D). Statistical analyses showed that virus-infected leaf area on nod19-1 was significantly lower than that on WT plants (Figure 4D). We further analyzed the amount of viral RNA in WT and nod19-1 seedlings by amplifying a fragment of the CP gene through reverse transcription-quantitative PCR (RT–qPCR). Consistently, seedlings of nod19-1 also accumulated less viral RNAs as compared with that in WT seedlings (Figures 4E,F), indicating that knockout NOD19 attenuates TuMV proliferation.

We further analyzed the function of NOD19 in the proliferation of other potyviruses on their respective hosts, i.e., SMV on its host soybean [Glycine max (L.) Merr.]. Soybean is an ancient tetraploid, which underwent two whole genome duplications. Most of soybean genes are paralogous genes with multiple copies. Blast search using the amino acid sequence of NOD19 as the query showed that there are at least 11 genes in G. max var. Willioms 82 with the of SURNod19-like domain in the genome of soybean. Phylogenetic analyses showed that six of the 11 soybean homologs were clustered with NOD19 of Arabidopsis in same cluster with GmNOD19a1 and GmNOD19a2 as the closest ones (Figure 5A). Multiple sequence alignment showed that the nucleotide sequences of GmNOD19a1 and GmNOD19a2 are also highly similar (Supplementary Figure 1). Thus, we tried to simultaneously silence GmNOD19a1 and GmNOD19a2 in soybean cultivar Zhonghuang13 using an ALSV-based RNA silencing vector. At 20 dpi, soybean plants infected by ALSV inserted with the fragment of soybean PHYTOENE DESATURASE (GmPDS) gene (ALSV::GmPDS) showed the photobleaching phenotype, while those infected by ALSV harboring GmNOD19a fragment (ALSV::GmNOD19a) displayed no obvious symptoms (Figure 5B). RT–qPCR showed that the expression level of GmNOD19a was reduced more than 60% in the systemic leaves infected by ALSV::GmNOD19a as compared with that in soybean plants infected by ALSV::GmPDS (Figure 5C). GmNOD19a-silenced soybean leaves were then mechanically inoculated by SMV strain SC7. At 2 dpi, the accumulation of SMV genomic RNA on the inoculated leaves were evaluated by amplifying a fragment of the CP gene of SMV through RT–qPCR. The results showed that seedlings inoculated by ALSV::GmNOD19a accumulated significantly fewer viral genome RNA than that on WT seedlings (Figures 5D,E), indicating that SMV infection is negatively influenced by GmNOD19a knockout.

A. thaliana ecotype Col-0 and N. benthamiana seeds were maintained by our lab and Soybean (G. max) cultivar Zhonghuang-13 seeds were kindly provided by the Heilongjiang Academy of Agricultural Science. Seedlings of A. thaliana and N. benthamiana were grown in a plant growth chamber under a 16-/8-h photoperiod at 23°С. Soybean seedlings were grown under a 16-/8-h photoperiod at 26°С in a greenhouse. Agroinfiltration was performed as described earlier with seedlings of the same age and similar sizes (Wu et al., 2022). In brief, Agrobacterium strain GV3101 harboring the proper plasmid were washed with infiltration buffer (10 mM MES, pH 5.6, 10 mM MgCl; 100 μM acetosyringone) twice by low-speed centrifuge at room temperature and infiltrated into N. benthamiana leaves at the indicated OD values. Mechanical inoculation was performed as described earlier (Li et al., 2022). The whole aerial part was taken for further analyses, e.g., RNA extraction and Western blot.

Full-length coding sequences of turnip mosaic virus (TuMV) P1 and Arabidopsis NOD19 (AT5G61820) were amplified using the primers listed in Supplementary Table 1 with the Phanta DNA polymerase (Vazyme, Nanjing, China) and inserted into a modified pDONR207m vector by a ClonExpress II One Step Cloning Kit (Vazyme). The gateway-compatible pEarleyGate101 and pEarleyGate104 plant binary expression vectors were used to produce C- or N-terminal YFP-tagged constructs (Earley et al., 2006). The pGADT7 and pGADT7 or gateway-compatible version pGBKT7-DEST and pGADT7-DEST were used to produce plasmids for the Y2H assay. The pEarleyGate201-YN, pEarleyGate202-YC, YC-pEarleyGate100 and YN-pEarleyGat100 were used to generate constructs for the bimolecular fluorescence complementation (BiFC) assay (Lu et al., 2010). All plasmids were verified by Sanger sequencing.

The cDNA library of Arabidopsis mRNA in the pGADT7 vector was constructed as described earlier (Xu et al., 2020). Y2H assays were performed using the yeast strain Golden (Clontech, Beijing, China) as described previously (Cheng and Wang, 2017). All yeast selective media were purchased from Coolaber Technology (Beijing) Co., Ltd.

The fluorescence was visualized with a TCS SP8 LIGHTNING Confocal Microscope (Leica, Wetzlar, Germany) as described previously (Chai et al., 2020). Sequential mode was used when necessary.

Plant phenotype images were quantified by Fiji, a distribution of ImageJ for biologists and computer scientists (Schindelin et al., 2012). In brief, images were firstly split into red (chloroplast fluorescence), green (GFP fluorescence), and blue channels. The analyze particles function in Fiji was used to analyze the total leaf area using the red channel pixels, the virus-infected area using the green channel pixels with size 100-infinity and circularity 0.0–1.0. The ROI manager function was used to produce the pseudocolor image.

Apple latent spherical virus (ALSV)-based gene silencing was performed as described earlier (Don et al., 2022). In brief, a 309 base-pair cDNA fragment of GmNOD19 was amplified from soybean cultivar Zhonghuang-13 and inserted into the pALSV2 vector to construct pALSV2-GmNOD19. The pALSV1 were co-infiltrated with pALSV2-GmNOD19 or pALSV2-GmPDS into N. benthamiana seedlings. At 20 days post infiltration (dpi), viral particles were purified from systemic leaves of N. benthamiana by PEG method. The purified viral particles in 0.01 mol/L phosphate buffered saline (pH 7.0) were directly used to inoculate the first-round true leaves of Zhonghuang-13 seedlings that were pre-dusted with carborundum. At about 16 dpi, the photobleaching phenotype was observed on soybean plants inoculated with pALSV1 and pALSV2-GmPDS. Reverse transcription and quantitative polymerase chain reaction (RT–qPCR) was then used to check the silencing efficiency of GmNOD19 in the systemic leaves of soybean plants inoculated by pALSV1 and pALSV2-GmNOD19. The GmNOD19-silenced or photobleaching leaves were then mechanically inoculated by soybean mosaic virus (SMV) N1 isolate and the virus replication was evaluated by RT–qPCR at 2 dpi using primers listed in Supplementary Table 1.

Total RNA was extracted using the Eastep® Super Total RNA Extraction Kit (Promega, Beijing, China) and reverse transcribed into complementary DNA (cDNA) by a HiScript III 1st Strand cDNA Synthesis Kit with gDNA wiper (Vazyme) with Oligo (dT)₂₀ and random hexamers. RT–qPCR was performed on a CFX Connect qPCR System in a 20 μL volume system containing 4 μL of 5 μM of each primer, 10-fold-diluted cDNA, and 10 μL 2 × AceQ® Universal SYBR qPCR Master Mix (Vazyme). Arabidopsis ACTIN II (AT3G18780) and GLYCERALDEHYDE 3-PHOSPHATE DEHYDROGENASE (GAPC2; At1g13440) and soybean ACTIN 7 (GmACT7; LOC100806695) and CONSTITUTIVE 4 (GmCONS4; LOC100783869) were used as the internal controls. All primers were listed in Supplementary Table 1.

Two-sided Student’s T-test was performed with Office Excel. The data are represented as the mean ± SD as indicated.

DeepTMHMM (Hallgren et al., 2022), SignalP 6.0 (Teufel et al., 2022), and Phobius predictors (Käll et al., 2007) are used to infer signal peptides and transmembrane regions. Conserved domain was searched with the CDD/SPARCLE (Lu et al., 2020). Alphafold2-predicted structure of NOD19 was downloaded from the AlphaFold Protein Structure Database.¹ The structure was visualized and rendered by iCn3D structure viewer.²