P1 enhances the severity of the HCTu-mediated serrated leaf phenotype and PTGS suppression
To further investigate the functions of P1 and HC-Pro in PTGS suppression, the P1/HC-Pro gene of TuMV was used in this study (Fig. 1). The P1/HCTu plants showed a severe serrated and curled leaf phenotype (Fig. 1A and B). The translated P1/HC-Pro protein contains an F362/S363 cleavage site (Fig. 1A), which can generate separated P1 and HC-Pro proteins through P1 cleavage (Fig. 1C). We generated two individual P1Tu and HCTu plants that expressed the P1 and the HC−Pro genes of TuMV individually (Fig. 1A and B). The P1Tu plant showed normal development similar to the Col-0 plants, whereas the HCTu plant showed mildly serrated leaves (Fig. 1B). In addition to the difference in the severity of the leaf phenotype, the size of HCTu plant was larger than that of the P1/HCTu plant (Fig. 1A, and B), suggesting that the released P1 enhances the HC-Pro-mediated serrated leaf phenotype.
In addition, an F362A substitution at the F362/S363-P1 cleavage site produced a P1HC-Pro fusion protein (P1HCTu − FA) (Fig. 1A, and C). This transgenic P1HCTu − FA plant showed a normal phenotype (Fig. 1B), suggesting that the separation of P1 from HC-Pro is necessary and sufficient to develop the serrated leaf phenotype. Furthermore, a kanamycin-resistant HCTu plant [HCTu (kan) plant] was generated for crossing with the P1Tu plant (Basta resistant) (Fig. 1A and B). Like the HCTu plant, the HCTu (kan) plant showed mildly serrated leaves (Fig. 1B). Interestingly, the P1Tu × HCTu (Kan) offspring showed severely serrated and curled leaves, but the size of P1Tu × HCTu (Kan) plant was larger than that of the P1/HCTu plant (Fig. 1B). In addition, only the P1/HCTu plant showed high levels of the P1 and HC-Pro proteins; while the other lines, even the P1Tu × HCTu (Kan) plant, showed low levels of P1 and HC-Pro (Fig. 1C), suggesting that P1HC-Pro fusion and the separation of P1 from HC-Pro must both occur in the transgenic cells to enhance the serrated leaf phenotype.
We compared 57 potyvirus amino acid sequences of P1/HC-Pro (Fig. 2A). The alignment results showed that the sequence and length of the P1 protein in different potyviruses are highly diverse (Fig. 2A). Only the C-terminal protease activity site (black boxes) is conserved (Fig. 2A). In contrast, several conserved domains of HC-Pro were found in different species (Fig. 2A). We evaluated the phenotype of the P1/HCZy plant and the P1/HCTe plant (Fig. 1A, and B). Both plants showed a severe serrated and curled leaf phenotype with high levels of P1 and HC-Pro (Fig. 1D). The results indicated that the P1/HC-Pro gene from various species can trigger a serrated leaf phenotype.
The next question was whether of the function of the HC-Pro from each virus requires the P1 from the same species. We generated 6 recombinant P1/HC-Pro plants in which HC-Pro was fused with a heterologous P1, namely, P1Zy/HCTu, P1Te/HCTu, P1Tu/HCZy, P1Te/HCZy, P1Tu/HCTe, and P1Zy/HCTe (Fig. 2B, and C). The P1Zy/HCTu and P1Te/HCTu plants, similar to the other recombinant transgenic plants, showed a severe serrated leaf phenotype (Fig. 2C). In addition, the P1Zy/HCTu and P1Te/HCTu plants showed detectable P1 and HC-Pro expression (Fig. 1D). These results suggest that multiple P1 genes have conserved functions in enhancing the HC-Pro-mediated serrated leaf phenotype.
HC-Pro-mediated PTGS suppression
Previous studies demonstrated that an abnormal accumulation of miRNA and miRNA* occurs in several transgenic viral suppressor plants because suppressors interfere miRNA biogenesis (Kasschau et al. 2003; Kung et al. 2014; Wu et al. 2010). In addition to miRNA/miRNA* accumulation, miRNA targets were also upregulated in the transgenic plants because of miRNA misregulation (Kasschau et al. 2003; Kung et al. 2014; Wu et al. 2010). Therefore, phenomena of abnormal miRNA/miRNA* and target RNA accumulation are the molecular phenotypes of PTGS suppression. Except for the P1HCTu − FA plant, all transgenic lines that contained HCTu showed abnormal miRNA and miRNA* accumulations (Fig. 1E), confirming that HCTu is the dominant player in PTGS suppression. Surprisingly, the P1Tu plant also showed miRNA and miRNA* accumulation through an unknown mechanism (Fig. 1E). The P1/HCZy, P1/HCTe, and 6 recombinant P1/HC plants also showed identical miRNA/miRNA* accumulations (Fig. 2D). Transcriptome profiles also indicated that miRNA targets were upregulated in HCTu, HCTu (kan), P1Tu × HCTu, and P1/HCTu plants (Fig. 1F), suggesting that miRNA regulation was blocked by HC-Pro. However, DICER-LIKE 1 (DCL1; miR162 target) and two translation inhibition genes, APETALA 2 (AP2; miR172target) and SHORT VEGETATIVE PHASE (SVP; miR396 target), showed no change in their transcript levels (Fig. 1F). Except for the DCL1, AP2, and SVP genes, the P1/HCTu plant suppressed most of the miRNA-target regulations (Fig. 1F). We conclude that the P1/HCTu plant has a stronger suppressive effect than the HCTu plants. In addition, the heterologous P1s have conserved function(s) in enhancing the HC-Pro-mediated PTGS suppression.
Identification of host P1-interacting proteins
We hypothesize that various P1 proteins have (a) conserved interacting protein(s) in Arabidopsis that enhance HC-Pro-mediated PTGS suppression. To identify the host P1-interacting proteins, the P1/HCTu, P1/HCZy, and P1/HCTe plants were used for IP with a-P1Tu, a-P1Zy, and a-P1Te antibodies, respectively. These IP eluates were analyzed by LC-MS/MS. We identified 101 cytoplasmic P1 of TuMV (P1Tu)-interacting proteins (Supplementary Data). Furthermore, we identified 56 cytoplasmic P1 of ZYMV (P1Zy)-interacting proteins and 20 cytoplasmic P1 of TEV (P1Te)-interacting proteins (Supplementary Data). Importantly, only one consensus cytoplasmic protein, VERNALIZATION INDEPENDENCE 3/ SUPERKILLER8 (VIP3/SKI8; AT4G29830), was found in the IP profiles of 3 viral P1s (Table 1). VIP3/SKI8 is a subunit of the RNA exosome complex that is required for degradation of the RISC 5'-cleavage fragment (Branscheid et al. 2015; Orban and Izaurralde 2005). In contrast, 12 consensus cytoplasmic proteins were identified in the P1Tu and P1Zy IP profiles, whereas 10 consensus proteins were identified in the P1Tu and P1Te IP profiles (Table 1). Moreover, 5 consensus cytoplasmic proteins were found in the P1Zy and P1Te IP profiles (Table 1).
Table 1
The P1 interacting proteins
AGI | Protein Name | Description | Interacting with |
P1Tu | P1Zy | P1Te |
AT4G29830 | VIP3/SKI8 | WD repeat-containing protein | +a | + | + |
AT5G61780 | TSN2 | Ribonuclease TUDOR 2 | + | | |
AT5G07350 | TSN1 | Ribonuclease TUDOR 1 | + | | |
AT3G13300 | VSC | VARICOSE | + | | |
AT2G15430 | DdRp | DNA-directed RNA polymerases subunit 3 | + | | |
AT3G18165 | MOS4 | Modifier of SNC1,4 | + | | |
AT1G79280 | NUA | Nuclear-pore anchor (NUA) | + | | |
AT5G53480 | Importin | Importin subunit beta-1 | + | | |
AT4G16143 | Importin | Importin subunit alpha-2 | + | | |
AT3G43300 | BIG5 | Brefeldin A-inhibited guanine nucleotide-exchange protein 5 | + | | |
AT3G47810 | VSP29 | Vacuolar protein sorting-associated protein 29 | + | | |
AT5G24780 | VSP1 | Vegetative storage protein 1 | + | + | |
AT1G52400 | | Beta-D-glucopyranosyl abscisate beta-glucosidase | + | + | |
AT1G53310 | | Phosphoenolpyruvate carboxylase 1 | + | + | |
AT1G16460 | | Thiosulfate/3-mercaptopyruvate sulfurtransferase 2 | + | + | |
AT3G27300 | | Glucose-6-phosphate 1-dehydrogenase, cytoplasmic isoform 1 | + | + | |
AT2G23930 | | Probable small nuclear ribonucleoprotein G | + | + | |
AT2G31390 | | Probable fructokinase-1 | + | + | |
AT3G62830 | | UDP-glucuronic acid decarboxylase 2 | + | + | |
AT5G03630 | | Monodehydroascorbate reductase 2 | + | + | |
AT3G52560 | | Ubiquitin-conjugating enzyme E2 variant 1D | + | + | |
AT1G64520 | | 26S proteasome non-ATPase regulatory subunit 8 homolog A | + | + | |
AT1G08830 | SOD1 | Superoxide dismutase [Cu-Zn] 1 | + | | + |
AT3G55620 | | Eukaryotic translation initiation factor 6 − 2 | + | | + |
AT5G41220 | GST | Glutathione S-transferase T3 | + | | + |
AT4G24190 | | Endoplasmin homolog | + | | + |
AT5G42980 | | Thioredoxin H3 | + | | + |
AT1G72730 | | Eukaryotic initiation factor 4A-3 | + | | + |
AT1G77760 | | Nitrate reductase [NADH] 1 | + | | + |
AT1G78370 | GST | Glutathione S-transferase U20 | + | | + |
AT3G61220 | | (+)-neomenthol dehydrogenase | + | | + |
AT3G06650 | | ATP-citrate synthase beta chain protein 1 | | + | + |
AT5G49460 | | ATP-citrate synthase beta chain protein 2 | | + | + |
AT5G44316 | | Putative UPF0051 protein ABCI9 | | + | + |
AT3G44300 | | Nitrilase 2 | | + | + |
aThe protein was identified in the relevant P1 transgenic plants and marked as “+”. |
Next, we focused on P1Tu-interacting proteins because the P1/HCTu plant was the model used in this study. In the P1Tu IP profile, two TUDOR-SN ribonucleases [(TSN1 (AT5G07350) and TSN2 (AT5G61780)] were uniquely identified 5 to 6 times in a total of 6 IP experiments with P1/HCTu plants (Table 1, and Supplementary Table 1). TSN1 and TSN2 have been suggested to be involved in the regulation of uncapping mRNA and localize to processing bodies (P-bodies) and stress granules (Yan et al. 2014). These data suggested that P1Tu might alter the function of TSN1 and TSN2. Moreover, VARICOSE (VSC; AT3G13300) and MODIFIER OF SNC1,4 (MOS4; AT3G18165), which are involved in RNA regulation, were identified in the P1Tu IP profile (Table 1, and Supplementary Table 1). We also identified the NUCLEAR-PORE ANCHOR (NUA; AT1G79280), two IMPORTIN subunits (AT5G53480 and AT4G16143), and BREFELDIN A-INHIBITED GUANINE NUCLEOTIDE-EXCHANGE PROTEIN 5 (BIG5; AT3G43300), which are involved in nuclear and cytosolic transport (Table 1, and Supplementary Table 1) (Xue et al. 2019). Moreover, VACUOLAR PROTEIN SORTING-ASSOCIATED PROTEIN 29 (VSP29; AT3G47810) was identified, which participates in vacuolar protein trafficking and vacuolar sorting receptor recycling (Table 1, and Supplementary Table 1) (Kang et al. 2012).
Differentially expressed host proteins in transgenic plants
We performed label-free proteomics to identify the differentially expressed host proteins between Col-0 and other transgenic plants. We identified 2,757 Arabidopsis proteins in Col-0, P1Tu, HCTu, and P1/HCTu plants (Supplementary Data). We found that ADP-GLUCOSE PYROPHOSPHORYLASE (APL3; AT4G39210), 6-PHOSPHOGLUCONOLACTONASE (PGL5; AT5G24420), and TONSOKU (TSK)-ASSOCIATING PROTEIN 1 (TSA1; AT1G52410) were decreased in P1Tu and P1/HCTu plants but were not decreased in HCTu plants compared to Col-0 plants (Fig. 3A-C, panel i). However, the transcript of APL3 showed no significant difference among the various transgenic plants, whereas PGL5 and TSA1 were upregulated in HCTu and P1/HCTu plants compared with Col-0 plants (Fig. 3A-C, panel ii). APL3 is a starch biosynthesis enzyme, whereas PGL5 is a catalyzed enzyme in the oxidative pentose-phosphate pathway (OPPP) (Lansing et al. 2020; Liu et al. 2019). TSA1 was induced by methyl jasmonate (MeJA) and triggers endoplasmic reticulum (ER) body formation (Geem et al. 2019; Suzuki et al. 2005). These data indicated that P1Tu might trigger APL3, PGL5, and TSA1 protein degradation through an unknown type of posttranslational regulation.
Next, we identified differentially expressed proteins between the HCTu and P1/HCTu plants. Nine proteins, including 2 superoxide dismutase [SOD1 (AT1G08830), and SOD2 (AT2G28190)], COPPER CHAPERONE FOR SOD1 (CCS1; AT1G12520), and ENHANCED MIRNA ACTIVITY 1/SUPER SENSITIVE TO ABA AND DROUGHT 2 (EMA1/SAD2; AT2G31660), were increased in P1/HCTu plants compared with HCTu plants (Fig. 3D-G, panel i). The EMA1/SAD2 contains an importin-beta domain and negatively regulates in miRNA activity and also involved in abscisic acid (ABA) signaling (Cui et al. 2016; Panda et al. 2020; Wang et al. 2011).
In contrast, 8 photosystem proteins (ATCG00340, AT1G55670, AT1G31330, AT4G12800, AT1G52230, AT1G44575, ATCG00350, and AT2G20260) were decreased in the P1Tu, HCTu, and P1/HCTu plants compared with Col-0 (Fig. 3H-O, panel i). However, their transcript levels were not significantly different (Fig. 3H-O, panel ii). We also found that that PATHOGENESIS-RELATED GENE 5 (PR5; AT1G75040) was decreased in P1/HCTu plants (Fig. 3P, panel i), whereas JASMONATE RESISTANT 1 (JAR1; AT2G46370) was decreased in HCTu and P1/HCTu plants compared with Col-0 (Fig. 3Q, panel i). Similarity, the transcript levels of PR5 and JAR1 were not significantly different between the plants (Fig. 3P and Q, panel ii). In summary, many instances of posttranslational regulations were occurred in the P1Tu, HCTu, and P1/HCTu plants.
The posttranscriptional and posttranslational regulation of miRNA targets in P1/HCTu plants
CCS1 is involved in copper delivery, and SOD1 and SOD2 participate in Cu/Zn superoxide dismutase activities. The transcripts of these three genes are regulated by miR398 (Bouché 2010; Sunkar et al. 2006). However, there were a high level of CCS1, SOD1, and SOD2 accumulation in the HCTu and P1/HCTu plants, which corresponded to their transcript levels, indicating the P1/HC-Pro-mediated PTGS suppression (Fig. 3D-F, panel ii). Indeed, the transcript level of miR168-regulated AGO1 (AT1G48410) was high in HCTu and P1/HCTu plants compared with Col-0 (Fig. 3R, panel ii). Surprisingly, the level of AGO1 protein was decreased via an unknown mechanism in HCTu and P1/HCTu plants (Fig. 3R, panel i). The western blot data also indicated that the level of AGO1 was low in P1/HCTu plants, but was normal expression similar to Col-0 in P1/HCZy and P1/HCTe plants (Fig. 1G). These data suggested that the P1/HC-Pro of TuMV has a specific ability to trigger the posttranslational degradation of AGO1.
Comparative gene-to-gene network and transcriptome analysis
In the transcriptome analysis, we constructed a gene-to-gene correlation network to study PTGS suppression from a different perspective. First, we constructed a network for Col-0 vs. P1/HCTu plants in the ContigViews system. A list of 2-fold DGEs between Col-0 and P1/HCTu plants was used to generate a Pearson correlation network (Fig. 4). A group of positive correlations (red lines) and a group of negative correlations (green lines) are highlighted in the network (Fig. 4). Importantly, AGO1, AGO2 (AT1G31280), and AGO3 (AT1G31290) were present in the group of negative correlations (Fig. 4). AGO2 and AGO3 were positively correlated (red line) but had an indirect correlation with AGO1 through XYLOGLUCAN ENDOTRANSGLUCOSYLASE/HYDROLASE 7 (XTH7; AT4G37800) (Fig. 4). Notably, the transcripts of AGO1, AGO2, and AGO3 were upregulated, but the XTH7 transcripts were downregulated in the HCTu and P1/HCTu plants, suggesting that the AGOs and XTH7 might have opposite functions in PTGS (Fig. 3R, panel ii; Fig. 6A-C).
Next, we made two comparative networks, which were generated by a list of 2-fold DEGs between Col-0 and HCTu plants or between Col-0 and P1Tu plants (Fig. 5A and B). The gene positions in the comparative networks were followed with the Col-0 vs. P1/HCTu network for comparison (Fig. 4 and Fig. 5). The gene numbers in the Col-0 vs. HCTu network were much less than in the Col-0 vs. P1/HCTu network (Fig. 4 and Fig. 5A). However, the main genes involved in PTGS, such as AGO1, AGO2, AGO3, and XTH7 etc., were remained in the Col-0 vs. HCTu network (Fig. 5A). This suggested the presence of a basic network backbone in the HCTu-mediated PTGS suppression that occurs without the effects of P1Tu. In contrast, the Col-0 vs. P1Tu network only had 7 genes in 2 small groups that also presented in parts of the Col-0 vs. HCTu or Col-0 vs. P1/HCTu networks (Fig. 4 and Fig. 5). Moreover, XTH7 had less than 49 connected genes in the Col-0 vs. HCTu network, whereas XTH7 had 61 connections in the Col-0 vs. P1/HCTu network (Fig. 4 and Fig. 5A). These data indicated that the XTH7 connection is variable in different networks, and it might play a significant role in PTGS suppression. Overall, the comparative network analysis, it highlights the effecrs of P1Tu on HCTu-mediated PTGS suppression. It also explains why the P1/HCTu plant has a severe phenotype because of how many pathways were interfered with.
Critical genes in the Col-0 vs. P1/HCTu network
The importance of XTH7 is not only in number of gene connections it has or that it is connected with AGO1 and AGO2; XTH7 also had a negative-correlation with several miRNA targets in the Col-0 vs. P1/HCTu network, such as 2 auxin response transcription factor genes [ARF3 (AT2G33860), and ARF8 (AT5G37020)], PHOSPHATE 2 (PHO2; AT2G33770), GROWTH-REGULATING FACTOR 1 (GRF1; AT2G22840), CCS1, SOD1, and SOD2 (Fig. 5). However, ARF3, ARF8, PHO2, GRF1, CCS1, SOD1, and SOD2 formed a positive correlation in the network (Fig. 4). These miRNA target transcripts were upregulated in HCTu and P1/HCTu plants because of PTGS suppression (Fig. 3E-F; panel ii; Fig. 6D-G). Moreover, SEP3 (AT1G24260) showed negative correlations with XTH7, ARF3, ARF8, and SOD1 (Fig. 5). In addition, SEP3 transcript levels were lower in P1Tu, HCTu, and P1/HCTu plants (Fig. 6H). Notably, SOD1 was shown to have a physical interaction with P1Tu and P1Te (Table 1) and was also highlighted in the network, suggesting the importance of SOD1 in PTGS suppression.
Four miRNA targets, including TARGET OF EARLY ACTIVATION TAGGED 2 (TOE2; AT5G60120), and 2 squamosa promoter-binding protein-like genes [SPL13A (AT5G50570), and SPL13B (AT5G50670)], were also found in group of negative correlation area, whereas ARABIDOPSIS THALIANA HOMEOBOX PROTEIN 15 (ATHB-15; AT1G52150) and PHABULOSA (PHB; AT2G34710) were found in the boundary between the positive and negative correlations groups (Fig. 4). Notably, miR172b-regulated TOE2 modules in regulating plant innate immunity (Zou et al. 2018). In Col-0 vs. P1/HCTu network, CYCLING DOF FACTOR 2 (CDF2; AT5G39660), and 2 carbon catabolite repressor 4 (CCR4)-associated factor genes [CAF1A (AT3G44260), and CAF1B (AT5G22250)] that are involved in RNA regulation were identified in the Col-0 vs. P1/HCTu network (Fig. 4). CAF1A and CAF1B catalyze mRNA deadenylation, whereas CDF2 interacts with DICER-LIKE 1 (DCL1) for miRNA biogenesis (Liang et al. 2009; Sun et al. 2015; Walley et al. 2010). These genes were also upregulated in HCTu and P1/HCTu plants (Fig. 6Q, R, and Y).
Eight calcium signaling genes were identified in the group of positive correlation in the network and were significantly upregulated in P1/HCTu plants (Fig. 4, and Fig. 6I-P). In the network, CALMODULIN-LIKE 24 (CML24; AT5G37770), and CAM-BINDING PROTWIN 60-LIKE G (CBP60G; AT5G26920) have significantly functions in the regulation of autophagy and innate immunity, respectively (Qin et al. 2018; Tsai et al. 2013). In addition, jasmonic acid (JA) signaling and defense genes were highlighted in the positive correlation and their transcripts were upregulated in HCTu and P1/HCTu plants (Fig. 4, and Fig. 6Q-V). In addition, the network indicated that FUMARASE 2 (FUM2; AT5G50950) and BARELY ANY MERISTEM 2 (BAM2; AT3G49670) were present in the boundary region between groups of positive and negative correlation (Fig. 4). BAM2 is a CLAVATA1-related receptor kinase and promotes the differentiation of stem cells on the meristem flank (DeYoung et al. 2006). FUM2 transcripts were upregulated in HCTu and P1/HCTu plants, whereas BAM2 transcripts were decreased (Fig. 6W and X). We also showed that CDF2, which is involved in miRNA biogenesis (Sun et al. 2015), is in the negative correlation group and is indirectly connected (negative correlation) with AGO1 through HB2 (AT3G10520) and XTH7 (Fig. 4). The CDF2 transcripts were upregulated in HCTu and P1/HCTu plants (Fig. 6Y). The functions of critical genes in the Col-0 vs. P1/HCTu network are listed in Supplementary Table 2.
The auxin, ethylene, and ABA signaling pathway in PTGS suppression
The auxin response can induced ethylene and concomitantly trigger ABA biosynthesis (Hansen and Grossmann 2000). Importantly, the auxin, ethylene, and ABA signaling genes could be found in the Col-0 vs. P1/HCTu network (Fig. 4). MiRNA-regulated ARF3 and ARF8 targets are also auxin response genes, which were highly expressed in HCTu and P1/HCTu plants (Fig. 6D, and E). In contrast, BAM2 expressed is antagonistic with auxin transporters (Cecchetti et al. 2015) and its transcripts were downregulated in HCTu and P1/HCTu plants (Fig. 6X). Ethylene signaling genes, 1-AMINOCYCLOPROPANE-1-CARBOXYLIC ACID (ACC) SYNTHASE 6 (ACS6; AT4G11280), SCARECROW-LIKE 13 (SCL13; AT4G17230), and 8 ethylene responsive element binding factors [ERF1 (AT4G17500), ERF4 (AT3G15210), ERF5 (AT5G47230), ERF6 (AT4G17490), ERF105 (AT5G51190), ERF104 (AT5G61600), ERF12 (AT1G28360), ERF8 (AT1G53170)] were present in group of positive correlation and their transcripts were upregulated in HCTu and P1/HCTu plants (Fig. 4; Fig. 6Z-AI; and Supplementary Table 2). Moreover, results from endogenous ethylene emission experiments showed higher levels over a time courses in P1/HCTu plants compared with Col-0 (Fig. 1H). ABA signaling genes, SULFATE TRANSPORTER 3;1 (SULTR3;1; AT3G51895), and 2 DIVARICATA genes [DIV1 (AT5G58900), and DIV2 (AT5G04760)] were also highlighted in the Col-0 vs. P1/HCTu network (Fig. 4) (Chen et al. 2019; Fang et al. 2018). These data suggested that P1/HC-Pro-mediated PTGS suppression has even interfere with plant hormone signaling pathways.