Argonaute 1 and 5 proteins play crucial roles in the defence against cucumber green mottle mosaic virus in watermelon

Abstract

RNA silencing, a core part of plants' antiviral defence, requires the ARGONAUTE, DICER‐like, and RNA‐dependent RNA polymerase proteins. However, how these proteins contribute to watermelon's RNA interference (RNAi) pathway response to cucumber green mottle mosaic virus (CGMMV) has not been characterized. Here, we identify seven ClAGO, four ClDCL, and 11 ClRDR genes in watermelon and analyse their expression profiles when infected with CGMMV. ClAGO1 and ClAGO5 expression levels were highly induced by CGMMV infection. The results of ClAGO1 and ClAGO5 overexpression and silencing experiments suggest that these genes play central roles in watermelon's antiviral defence. Furthermore, co‐immunoprecipitation and bimolecular fluorescence complementation experiments showed that ClAGO1 interacts with ClAGO5 in vivo, suggesting that ClAGO1 and ClAGO5 co‐regulate watermelon defence against CGMMV infection. We also identified the ethylene response factor (ERF) binding site in the promoters of the ClAGO1 and ClAGO5 genes, and ethylene (ETH) treatment significantly increased ClAGO5 expression. Two ERF genes (Cla97C08G147180 and Cla97C06G122830) closely related to ClAGO5 expression were identified using co‐expression analysis. Subcellular localization revealed that two ERFs and ClAGO5 predominantly localize at the nucleus, suggesting that enhancement of resistance to CGMMV by ETH is probably achieved through ClAGO5 but not ClAGO1. Our findings reveal aspects of the mechanisms underlying RNA silencing in watermelon against CGMMV.

ClAGO1 and ClAGO5 interaction in vivo inhibits CGMMV infection and there is probably a crosstalk between the ethylene signalling pathway and RNA silencing.

1. INTRODUCTION

Watermelon (Citrullus lanatus) is a herbaceous vine in the family Cucurbitaceae (Guo, Zhao, et al., 2019), an important horticultural crop ranking fifth in global production after apple, citrus, banana, and grape (http://faostat.fao.org/). However, watermelon production and, consequently, food sustainability are under constant threat from emerging and re‐emerging diverse viral diseases, such as cucumber green mottle mosaic virus (CGMMV), zucchini yellow mosaic virus (ZYMV), cucumber mosaic virus (CMV), and watermelon mosaic virus (WMV). Cucumber green mottle mosaic virus is a well‐studied species in the genus Tobamovirus, family Virgaviridae. CGMMV causes disease in Cucurbitaceae plants (e.g., watermelon, melon, cucumber, and gourds; Dombrovsky et al., 2017). Plants infected with CGMMV show leaf mosaic (irregular, mottled patches of discolouration on the leaves), plant dwarfing, and fruit distortion, and their yields and qualities are remarkedly reduced (Dombrovsky et al., 2017). The CGMMV resistance genes in watermelon are poorly investigated, and no resistant cultivars are available commercially (Cai et al., 2022). It is therefore of particular interest to study the defence response mechanisms of cucurbit plants to CGMMV infection.

Plants have evolved multiple defence mechanisms against viruses, including RNA silencing, a conserved mechanism for plant antiviral immunity (Guo, Li, et al., 2019; Yang & Li, 2018). Argonaute (AGO), Dicer‐like (DCL), and RNA‐dependent RNA polymerase (RDR) proteins are essential components of RNA silencing machinery (Baulcombe, 2004). On virus infection, RNA silencing relies on recognizing viral double‐stranded RNA (dsRNA) by DCL enzymes, which cleave the dsRNA into 21‐ to 24‐nucleotide small RNAs called virus‐derived small interfering RNAs (vsiRNAs). These vsiRNAs are recruited by an AGO protein and form RNA‐induced silencing complexes (RISCs), which target complementary viral RNAs for inactivation through slicing or translational inhibition (Ma & Zhang, 2018; Machado et al., 2017; Szittya & Burgyán, 2013; Yang & Li, 2018).

The Arabidopsis genome encodes 10 AGO proteins (Vaucheret, 2008). Many AGO proteins show direct antiviral activity and multiple viral suppressor of RNA silencing proteins (VSRs) have been shown to target AGO proteins (Pumplin & Voinnet, 2013; Schuck et al., 2013). AGO1, 2, 4, 5, 7, and 10 are virus‐specific in RNA virus defence (Huang et al., 2019). For example, AGO1 has been implicated in resistance to turnip crinkle virus (TCV), brome mosaic virus (BMV), bamboo mosaic virus (BaMV), and CMV (Huang et al., 2019; Morel et al., 2002; Qu et al., 2008; Wang et al., 2011). The expression of AGO2 and AGO3 genes is regulated by abscisic acid (ABA), which mediates resistance against BaMV (Alazem et al., 2017). In rice, OsAGO18 can promote rice antiviral defence by sequestering miR168 to alleviate repression of rice AGO1, which is essential for antiviral RNA interference (RNAi) (Wu et al., 2015), and jasmonic (JA)‐responsive transcription factor JAMYB regulates AGO18 expression to promote rice antiviral defence (Yang et al., 2020). Despite the relative wealth of studies on the importance of AGOs in antiviral mechanisms in the plants mentioned above, the role of AGOs in resistance against viral infections in watermelon remains largely unknown.

In this study, we identified and analysed the expression profiles of all ClAGO, ClDCL and ClRDR genes in response to CGMMV infection. Subsequently, we investigated the roles of ClAGO1 and ClAGO5. We report that ClAGO1 and ClAGO5 might co‐regulate resistance against CGMMV in vivo. Furthermore, ethylene (ETH) treatment induced the expression of the ClAGO5 gene. To our knowledge, this is the first comprehensive study of the AGOs in response to virus infections in C. lanatus, demonstrating the unique contributions of the ClAGO1 and ClAGO5 proteins against CGMMV.

2. RESULTS

2.1. Identification of the AGO , DCL, and RDR genes in watermelon

Although the genome and protein sequences information of watermelon C. lanatus subsp. vulgaris ‘97103’ is available (CuGenDB; http://cucurbitgenomics.org), the genes encoding AGOs, DCLs, and RDRs have not been thoroughly identified. To provide a comprehensive understanding of the AGOs, DCLs, and RDRs involved in the defence responses against viruses, we used BlastP to search the C. lanatus genome and eliminated redundant sequences by evaluating the structural integrity of conserved domains (Figure S1). A total of seven AGO, four DCL, and 11 RDR genes were identified (Table 1). To illustrate the evolutionary relationship of AGO, DCL, and RDR proteins in watermelon with those of known model plants, we performed multiple sequence alignments using ClustalW followed by phylogenetic analysis of the identified ClAGO, ClDCL, and ClRDR proteins and representatives from Arabidopsis thaliana (AtAGOs, AtDCLs, and AtRDRs), using the general time reversible substitution model and the maximum‐likelihood method for phylogenetic tree reconstruction in MEGA 7. Following previous designations to A. thaliana AGO, DCL, and RDR proteins (Rodríguez‐Leal et al., 2016), the seven ClAGO proteins were separated into four distinct groups and named ClAGO1, ClAGO4, ClAGO5, ClAGO6, ClAGO7, ClAGO10a, and ClAGO10b (Figure 1). The four ClDCL proteins were named ClDCL1, ClDCL2, ClDCL3, and ClDCL4 (Figure S2a). The 11 ClRDR proteins were divided into four clades. Among the four groups, group I contained seven members, named ClRDR1a, ClRDR1b, ClRDR1c, ClRDR1d, ClRDR1e, ClRDR1f, and ClRDR1g (Figure S2b), which is consistent with the number of RDR1 proteins found in Triticum aestivum (Akond et al., 2022). However, four RDR1 proteins were identified in strawberry, which were also diploid (Jing et al., 2023), we speculate that this difference may be due to variation among different species. In addition, the identification of pseudogenes among the identified ClRDR1 genes requires further experimentation. Groups II and III contained one member each, ClRDR2 and ClRDR6, respectively. There were two ClRDR proteins in group VI, which were named ClRDR3 and ClRDR5 (Figure S2b).

TABLE 1.

Basic information of ClAGO, ClDCL and ClRDR genes in watermelon.

Gene	Accession number	Location		Coding sequence (bp)	PI	Mol. wt. (kDa)
ClAGO1	Cla97C04G079510	Chr4	26688374…26694303 (+)	3180	4.84	262.4
ClAGO10a	Cla97C05G085880	Chr5	4414438…4420960 (+)	2955	4.87	242.1
ClAGO10b	Cla97C06G119800	Chr6	21261033…21269560 (+)	2849	4.87	236.2
ClAGO7	Cla97C05G105980	Chr5	33389625…33392984 (−)	3084	4.84	257.3
ClAGO6	Cla97C09G173890	Chr9	10554665…10560152 (+)	2718	4.90	223.9
ClAGO5	Cla97C01G012980	Chr1	26778114…26785345 (−)	2910	4.88	239.3
ClAGO4	Cla97C02G050470	Chr2	37627289…37633077 (−)	2769	4.88	229.9
ClDCL1	Cla97C05G079940	Chr5	63206…89286 (−)	5970	4.74	486.7
ClDCL2	Cla97C03G061160	Chr3	17896468…17919838 (+)	2624	4.92	218.5
ClDCL3	Cla97C08G152540	Chr8	20916213…20937301 (−)	4932	4.78	407.4
ClDCL4	Cla97C06G121290	Chr6	23677330…23692362 (+)	4926	4.77	407.4
ClRDR1a	Cla97C07G134280	Chr7	10599513…10604135 (−)	3393	4.86	281.9
ClRDR1b	Cla97C00G000380	Scf031	6535…8499 (−)	1491	5.03	124.1
ClRDR1c	Cla97C07G133680	Chr7	7013734…7016075 (−)	1494	5.03	125.2
ClRDR1d	Cla97C07G133670	Chr7	7011452…7013724 (−)	1812	5.00	149.6
ClRDR2	Cla97C06G126490	Chr6	28211339…28215849 (+)	3354	4.86	277.5
ClRDR5	Cla97C03G066490	Chr3	29913496…29922075 (−)	1752	5.01	144.1
ClRDR3	Cla97C03G066500	Chr3	29922328…29928188 (−)	1218	5.06	101.4
ClRDR6	Cla97C08G156700	Chr8	24428769…24432636 (+)	3591	4.85	295.3
ClRDR1f	Cla97C01G006910	Chr1	6968097…6972390 (+)	3363	4.87	276.8
ClRDR1g	Cla97C01G006770	Chr1	6830896…6837155 (+)	3135	4.90	260.3
ClRDR1e	Cla97C07G133700	Chr7	7096033…7101045 (+)	3351	4.86	278.5

FIGURE 1.

Phylogenetic relationship of the Argonaute (AGO) family proteins of watermelon (Citrullus lanatus). The amino acid sequences of AGO proteins identified in watermelon, with those from Arabidopsis thaliana as references, were aligned using the ClustalW software and subjected to phylogenetic analysis using the maximum‐likelihood method in MEGA 7 with 1000 bootstrapping replicates using default parameters. The bootstrap values are shown next to the branches. Major clades are boxed in different colours. Scale bar, 0.1 substitutions per site.

2.2. ClAGO , ClDCL , and ClRDR gene expression profiles following CGMMV infection

To examine the roles of the ClAGOs, ClDCLs, and ClRDRs in the antiviral defence responses, we first performed RNA sequencing (RNA‐Seq) for watermelon leaves at 20 days after CGMMV inoculation. Then, we analysed gene expression profiles through RNA‐Seq data at 10 and 20 days after CGMMV inoculation (data not shown). We found that the transcript accumulation levels of ClAGO1, ClAGO5, ClDCL2, ClDCL4, and ClRDR1e genes increased remarkedly in CGMMV‐infected watermelon (Figure 2a). The expression of those genes by reverse transcription‐quantitative PCR (RT‐qPCR) was consistent with transcriptome data, confirming the induction of those genes in CGMMV‐infected watermelon (Figure 2b).

FIGURE 2.

The gene expression profiles of ClAGO, ClDCL, and ClRDR in virus‐infected leaves and accumulation of viral coat proteins (CP) at intervals following infections by CGMMV in watermelon leaves. (a) Heatmap showing the gene expression patterns of ClAGO, ClDCL, and ClRDR in watermelon leaves under CGMMV infection 10 and 20 days postinoculation (10d, 20d). The value in each box represents the gene expression level, obtained after averaging the three biological replicates and the gene data were standardized by log₁₀ using TBtools (Chen et al., 2020). (b) The expression levels of ClAGO, ClDCL, and ClRDR genes were examined by reverse transcription‐quantitative PCR. The leaves were collected at 10 and 20 days postinoculation for total RNA extraction. Values are means ± SD of three biological replicates. Norm. Exp, normalized expression level; *p < 0.05, **p < 0.01, and ***p < 0.001 significant difference determined by Student's t test. (c) Detection of CGMMV genomic (gRNA) and subgenomic RNA (sgRNA) 1 and 2 in the mock‐inoculated (CK) and CGMMV‐infected (CG) plants by northern blotting. rRNA was stained with ethidium bromide and served as a loading control.

In addition, we measured the accumulation levels of viruses by northern blotting. Interestingly, the expression levels of ClAGO1, ClAGO5, ClDCL2, and ClDCL4 genes were directly proportional to the levels of CGMMV accumulation (Figure 2c). To determine the expression patterns of the candidate genes in the various organs of watermelon, RT‐qPCR was also performed to analyse the transcript levels of ClAGO1, ClAGO5, ClDCL2, ClDCL4, and ClRDR1e. We found that these genes had relatively higher expression levels in roots; in leaves, ClAGO1 showed the highest expression level among all examined genes, followed by ClDCL4 and ClAGO5 (Figure S3). In addition, ClAGO1 and ClAGO5 are expressed at much higher levels than ClDCL and ClRDR genes in mock‐inoculated and CGMMV‐infected watermelon plants (Figure 2a); we therefore presumed that ClAGO1 and ClAGO5 are more important for watermelon antiviral defence than the other genes tested. Accordingly, ClAGO1 and ClAGO5 were characterized further in subsequent studies.

2.3. ClAGO1 and ClAGO5 reduce CGMMV accumulation in watermelon

To verify the anti‐CGMMV function of ClAGO1 and ClAGO5, we used the pV190 virus‐induced gene silencing (VIGS) system to generate ClAGO1‐ and ClAGO5‐knockdown plants. The 300‐bp fragments from the coding sequences of ClAGO1 and ClAGO5 were cloned into a pV190 vector to generate pV190‐ClAGO1 and pV190‐ClAGO5, which was introduced into watermelon via Agrobacterium‐mediated infiltration. At 20 days postinfiltration (dpi), pV190‐PDS‐treated plants showed visible photobleaching symptoms, while mock‐ and pV190‐treated plants showed no mosaic symptoms; however, pV190‐GUS, pV190‐ClAGO1, and pV190‐ClAGO5 showed mosaic symptoms on leaves, and more severe mosaic symptoms occurred in ClAGO1‐ and ClAGO5‐silenced plants than in pV190‐GUS‐treated plants (Figure 3a). In addition, the height of pV190‐ClAGO1 and pV190‐ClAGO5‐infected plants was significantly shorter than that of pV190‐PDS and pV190‐GUS‐infected plants (Figure 3b).

FIGURE 3.

Effects of knockdown of ClAGO1 or ClAGO5 on symptoms and CGMMV accumulation. (a) Symptoms in CGMMV‐infected watermelon leaves from various gene‐silenced plants or controls at 20 days postinoculation (dpi). (b) Height of watermelon plants treated with different pV190 vectors. (c) Silencing efficiency of ClAGO1 and ClAGO5 in watermelon leaves determined by reverse transcription‐quantitative PCR. (d) Accumulation of CGMMV genomic RNA and subgenomic RNA (sgRNA) in gene‐silenced watermelon leaves determined by northern blotting.

ClAGO1 expression in ClAGO1‐knockdown plants was decreased to 47% of that of control plants (pV190‐PDS) at 20 dpi (Figure 3c), and ClAGO5 expression remained stable. ClAGO5 expression in ClAGO5‐knockdown plants was decreased to 52% of that of control plants at 20 dpi (Figure 3c), and ClAGO1 expression was not changed. Next, to investigate whether down‐regulation of ClAGO1 and ClAGO5 affected CGMMV accumulation, systemic leaves from mock‐, PDS‐, ClAGO1‐silenced, and ClAGO5‐silenced watermelon plants were analysed by northern blotting. We found that CGMMV genomic RNA (gRNA) increased by 120% and 170% in ClAGO1‐knockdown and ClAGO5‐knockdown plants compared to PDS‐silenced plants (Figure 3d), indicating the critical role of ClAGO1 and ClAGO5 in anti‐CGMMV defence.

To further analyse the functions of ClAGO5 in anti‐CGMMV defence, we also generated ClAGO5 overexpression lines in Nicotiana benthamiana. Transgenic plants overexpressing ClAGO5 grew faster (Figure 4a), and the ClAGO5 transcript abundance was significantly increased (Figure 4b). Then, we examined the role of ClAGO5 in response to CGMMV. Milder mottle and mosaic symptoms of systemically infected leaves were shown in ClAGO5 overexpression lines inoculated with CGMMV than in wild‐type (WT) plants (Figure 4c). These phenotypes were correlated with CGMMV accumulation, which was significantly lower in ClAGO5 overexpression lines than in WT plants (Figure 4d).

FIGURE 4.

Effects of overexpression of ClAGO5 on CGMMV accumulation. (a) Representative growth phenotypes of wild‐type (WT) and 35S‐ClAGO5 lines. (b) The transcript levels of ClAGO5. (c) Symptoms of systemic leaves on the 35S‐ClAGO5 and WT Nicotiana benthamiana plants inoculated with CGMMV at 10 days postinoculation. (d) Accumulation of CGMMV coat protein (CP) in ClAGO5 overexpression N. benthamiana leaves determined by reverse transcription‐quantitative PCR.

2.4. ClAGO1 and ClAGO5 interact in vivo

Because ClAGO1 and ClAGO5 both significantly reduced CGMMV accumulation, we suspected that there might be an interaction between these two proteins. We first performed a co‐immunoprecipitation (Co‐IP) assay. We found that the Co‐IP of ClAGO5‐HA with ClAGO1‐FLAG revealed strong signals, while a sample of a single expression of ClAGO5‐HA or ClAGO1‐FLAG had no detectable signals, indicating the interaction of ClAGO5 with ClAGO1 in vivo (Figure 5b). This was further confirmed by bimolecular fluorescence complementation (BiFC) assays. Co‐expression of ClAGO1‐cYFP with ClAGO5‐nYFP resulted in strong yellow fluorescent protein (YFP) fluorescence in the cells in N. benthamiana leaves. This strong YFP fluorescence appeared in the cytoplasm and nucleus (Figure 5a), indicating the specific interactions of ClAGO1 with ClAGO5 in vivo. Furthermore, a yeast two‐hybrid (Y2H) assay showed ClAGO1 could not interact with ClAGO5 protein (Figure 5c), indicating that the interaction between ClAGO1 and ClAGO5 is indirect. To explore the proteins that might interact with ClAGO1 and ClAGO5, their amino acid sequences were entered into the functional protein interaction network database (STRING 11.0, https://string‐db.org/). The results showed that other RNA silencing components, such as DCL and HYL1, might interact with ClAGO1 and ClAGO5 (Figure S4).

FIGURE 5.

Interaction of ClAGO1 with ClAGO5 in vivo. (a) Confirmation of the interaction between ClAGO1 and ClAGO5 by bimolecular fluorescence complementation. ClAGO1 and ClAGO5 were fused to the C‐terminal (cYFP) and N‐terminal (nYFP) half of yellow fluorescent protein (YFP). Confocal imaging was performed at 2–3 days postinoculation. White arrowheads indicated nuclear signals. Scale bar, 20 μm. (b) Confirmation of the interaction between ClAGO1 and ClAGO5 in vivo by co‐immunoprecipitation assay. Nicotiana benthamiana leaves were co‐infiltrated with Agrobacterium cultures harbouring expression vectors to express ClAGO1‐FLAG and ClAGO5‐HA. Samples before (Input) and after (IP) immunopurification were verified using anti‐FLAG and anti‐HA antibodies. (c) No direct interaction between ClAGO1 and ClAGO5 shown by yeast two‐hybrid assay.

2.5. The expression of ClAGO5 is affected by exogenous ETH

Previous studies have reported that salicylic acid (SA), JA, and ABA can regulate the expression of the AGO, DCL, and RDR families (Alamillo et al., 2006; Alazem et al., 2017; Hunter et al., 2013; Jaubert et al., 2011; Lee et al., 2016; Li et al., 2012; Yang et al., 2020). In our previous study, SA and ETH could increase watermelon resistance to CGMMV infection (authors' unpublished data). We therefore suspected that SA and ETH contribute to watermelon antiviral defences by regulating the expression of the ClAGO1 and ClAGO5 genes. First, we identified the cis‐elements in the ClAGO1 and ClAGO5 genes promoter sequences (c.2 kb) that were involved in plant hormone responses. The ABA‐responsive element, methyl JA/GA‐responsive element, and ERF (ETH‐related transcription factor) binding sites were found in the promoter of ClAGO1 (Figure 6a). The binding sites of these elements were also found in the promoter of the ClAGO5 gene, in addition to SA‐ and auxin‐responsive elements (Table 2). To confirm that ETH affects the expression of the ClAGO1 and ClAGO5 genes, we used RT‐qPCR to measure the ClAGO1 and ClAGO5 transcripts in watermelon leaves treated with these hormones. We found that ETH treatment significantly increased the expression of ClAGO5, while ClAGO1 was unaffected (Figure 6b,c).

FIGURE 6.

Identification of ethylene‐responsive elements (EREs) in promoters of ClAGO1 and ClAGO5 genes and the effect of ethylene (ETH) on the expression of ClAGO1 and ClAGO5. (a) EREs in the ClAGO1 and ClAGO5 promoters. (b) and (c) Relative expression of ClAGO1 (b) and ClAGO5 (c) with or without ETH treatments. CK, control.

TABLE 2.

Cis‐acting regulatory elements were predicted in the promoter regions of ClAGO1/5 related to plant hormone responses in watermelon.

Gene	Cis‐element	Sequence	Probable function
ClAGO1	ABRE	ACGTG	Abscisic acid‐responsive element
		CACGTG
		CACGTA
		TACGTG
	TGACG‐motif	TGACG	Methyl jasmonate (MeJA)‐responsive element
	CGTCA‐motif	CGTCA	MeJA‐responsive element
	DRE	ACCGAGA	ERF (ethylene response factor) binding site
	ERE	ATTTTAAA	ERF binding site
	GARE‐motif	TCTGTTG	Gibberellin‐responsive element
ClAGO5	ABRE	ACGTG	Abscisic acid‐responsive element
		CACGTG
		GACACGTGGC
	ERE	ATTTTAAA	ERF binding site
	P‐box	CCTTTTG	Gibberellin‐responsive element
	TCA‐element	CCATCTTTT	Salicylic acid‐responsive element
	TGA‐element	AACGAC	Auxin‐responsive element

In addition, we found two ERFs (Cla97C08G147180 and Cla97C06G122830) that have a co‐expression relationship with ClAGO5 (Table S2); moreover, both ERFs have conserved ERF domains, and their expression is affected by ETH treatment (Figure S5). We also performed localization of ClAGO5 and the two ERFs. We found that the two ERFs and ClAGO5 were mainly localized in the nucleus (Figure 7a,b). These results indicate that ETH might regulate the expression of ClAGO5 by ERF and then induce antiviral RNA silencing in watermelon.

FIGURE 7.

Subcellular localization of ClAGO5 and two ClERFs (Cla97C08G147180 and Cla97C06G122830). (a) Subcellular localization of ClAGO5. Bars, 20 μm. (b) Subcellular localization of the two ClERFs. Bars, 50 μm. Green fluorescent protein (GFP) control vector (35S:GFP), ClAGO5‐GFP, Cla97C08G147180‐GFP, or Cla97C06G122830‐GFP fusion protein (35S:ClAGO5‐GFP, 35S:Cla97C08G147180‐GFP, or 35S:Cla97C06G122830‐GFP) were co‐transformed with a nuclear marker gene H₂B fused to mCherry (red fluorescent protein, 35S:H₂B‐mCherry) in Nicotiana benthamiana leaves. Confocal microscopic images of epidermal cells were taken under green (for GFP), red (for nuclear marker) fluorescence, and bright field. The images on the right are overlapped by the three images on the left.

3. DISCUSSION

RNA silencing is a basic antiviral mechanism for plant resistance to viral infection. AGO, DCL, and RDR are critical components of RNA silencing and participate in the synthesis of small interfering (si)RNA (Qin et al., 2018). Watermelon is an important horticultural crop but is highly susceptible to various viral diseases. CGMMV outbreaks in grafted watermelon‐producing areas seriously hindered the sustainable development of the watermelon industry (Dombrovsky et al., 2017). Therefore, it is important to identify the core elements of the RNA silencing machinery, analyse their expression patterns, and explore their role in watermelon defence against CGMMV infection. In addition, the availability of the watermelon genome sequence has enabled genome‐wide gene expression analysis in watermelon (Guo, Zhao, et al., 2019). This study identified seven ClAGO, four ClDCL, and 11 ClRDR encoding genes in the watermelon genome and analysed their accumulation levels following CGMMV infection. We further verified the antiviral functions of ClAGO1 and ClAGO5 by VIGS and genetic transformation approaches. We report that ClAGO1 and ClAGO5 play key roles in antiviral response machineries.

3.1. ClAGO1 and ClAGO5 play central roles in watermelon defence against CGMMV

Many AGOs have been reported to be essential for plant defence against virus infection. However, the functions of AGO proteins vary with the virus–host pathosystem. For example, in A. thaliana, AGO1 and/or AGO7 were reported to have activity against BMV, CMV, and TCV (Azevedo et al., 2010; Dzianott et al., 2012; Morel et al., 2002; Qu et al., 2008). However, these two proteins appear dispensable for defence against TRV, and AGO2 is more important than AGO1 in defence against WT TRV (Ma et al., 2015). In this study, we found that the ClAGO1 and ClAGO5 genes are induced by CGMMV infection, and the knockdown of their expression could significantly increase viral accumulation. In addition, we also measured the expression of NbAGO1a, NbAGO1b, and NbAGO5 after CGMMV infection and found that their expression patterns in tobacco are opposite to those in watermelon (Figure S6). This indicates that the same virus could have a different effect on the AGO proteins of different hosts and may be why silencing efficiency differs between watermelon and tobacco, as described in our previous study (Liu, Liang, et al., 2020).

AGO1 is up‐regulated during virus infections and can interact with virus‐derived siRNAs to posttranscriptionally promote viral RNA degradation (Carbonell & Carrington, 2015; Guo, Zhao, et al., 2019; Yang & Li, 2018). However, the involvement of AGO5 might vary with viruses (Brosseau & Moffett, 2015). In our study, ClAGO1 and ClAGO5 can co‐enhance watermelon resistance to CGMMV infection. In addition, Co‐IP and BiFC experiments detected a physical interaction between ClAGO1 and ClAGO5. Similarly, both AGO2 and AGO5 act nonredundantly to restrict potato virus X (PVX) accumulation in A. thaliana (Brosseau & Moffett, 2015). It remains unknown whether AGO1 and AGO5 have an additive or nonredundant effect on restricting CGMMV accumulation in watermelon.

3.2. ETH signalling promotes watermelon anti‐CGMMV defence, probably through inducing ClAGO5 expression

Several plant hormones, such as SA, ABA, and JA, have been shown to promote antiviral defence through enhancing RNA silencing (Alamillo et al., 2006; Alazem et al., 2017; Hunter et al., 2013; Jaubert et al., 2011; Lee et al., 2016; Li et al., 2012). JA signalling promotes rice antiviral defence by inducing AGO18 expression (Yang et al., 2020). Our previous study found that CGMMV infection could increase ETH content in watermelon, and ETH treatment could enhance watermelon resistance to CGMMV (authors' unpublished data). We tested whether the ETH signalling promotes watermelon antiviral defence through enhancing RNA silencing. We identified several ERFs motifs in the promoter regions of the ClAGO1 and ClAGO5 genes, and ETH treatment could increase ClAGO5 gene expression levels (Figure 6). We also identified a positive correlation between the expression of two ERFs and ClAGO5, but we need to further examine whether both ERFs could regulate the expression of ClAGO5. These findings reveal that there is probably a crosstalk between the ETH signalling pathway and RNA silencing. AGO18 not only sequesters miR168 to derepress the expression of AGO1 (Wu et al., 2015) but also sequesters miR528 to derepress the expression of AO (Wu et al., 2017) for the enhancement of plant resistance to viruses. Whether ClAGO5 binds to miRNA requires further study.

4. EXPERIMENTAL PROCEDURES

4.1. Plant materials, growing conditions, and virus inoculation

Watermelon plant (cv. Zhengkang no. 2, ZK) and N. benthamiana plants were used in this study. These plants were grown in pots in a growth room maintained at 28°C and a 16‐h light/8‐h dark photoperiod. For rub‐inoculation, cotyledons of 7‐day‐old watermelon seedlings and N. benthamiana leaves at the five‐ to six‐leaf stages were inoculated with crude extracts from CGMMV‐infected plant leaves. For agro‐infiltration, Agrobacterium tumefaciens GV3101 carrying individual infectious viral clones was suspended in an infiltration buffer (10 mM MgCl₂, 10 mM MES, 100 μM acetosyringone, pH 5.6) until OD₆₀₀ = 1, incubated at room temperature for 2–4 h, and then infiltrated into leaves of N. benthamiana and watermelon using 1‐mL needleless syringes.

4.2. RNA sequencing and analysis

Leaves of the assayed ZK plants were collected at 20 days after CGMMV or phosphate buffer inoculation. These samples were labelled as ZK‐20d‐CK and ZK‐20d‐CG. Each leaf sample had three biological replicates, and each biological replicate had leaf tissues from at least nine plants. RNA sequencing was performed by the MetWare Biological Science and Technology Co. Ltd on an Illumina HiSeq 2000 platform. The resulting reads were aligned to the watermelon reference genome (http://cucurbitgenomics.org/organism/21), and the differentially expressed genes (DEGs) were identified using the DEseq2 software as previously described (Love et al., 2014; Varet et al., 2016), using a false discovery rate (FDR) threshold at <0.05. RNA‐Seq data on ZK plant leaves at 10 days after CGMMV or phosphate buffer inoculation were obtained previously (authors' unpublished data). Heatmaps about expression patterns of ClAGO, ClDCL, and ClRDR genes were performed.

4.3. Western blot assays

The total proteins of leaf tissues were extracted using a plant protein extraction kit (Solarbio). Then, the proteins were detected by specific anti‐HA (Abcam) or anti‐FLAG (Sigma) antibodies and were visualized using enhanced chemiluminescence reaction, ECL (Transgene Biotech).

4.4. VIGS

In our previous study, a VIGS system for cucurbits was developed (Liu, Liang, et al., 2020), which has the potential for functional analysis of screened AGOs. In this study, a 300‐bp fragment of AGO1 or AGO5 was obtained by PCR amplification using 2 × Fast Pfu Master Mix (Quick Load; Novoprotein) with gene‐specific primers pV190‐ClAGO1‐F and pV190‐ClAGO1‐R, or pV190‐ClAGO5‐F and pV190‐ClAGO5‐R (Table S1), then ligated into the pV190 vector. An empty pV190 vector, a pV190 vector carrying a 300‐bp fragment of the β‐glucuronidase (GUS) gene, and a pV190 with a 300‐bp fragment of the phytoene desaturase (PDS) gene were used for control treatments. All constructs were transformed into A. tumefaciens GV3101 and infiltrated into watermelon plants as previously described (Liu, Liang, et al., 2020).

4.5. Stable transformation of N. benthamiana

The full‐length open reading frame of ClAGO5 was amplified from watermelon leaves by RT‐PCR and cloned into the pBWA(V)HS‐ccdB‐GFP expression vector (Biorun). The resultant construct was introduced into A. tumefaciens GV3101, which was then used to stably transform N. benthamiana plants. Positive T₂ transgenic lines were selected and used in further experiments.

4.6. RNA analysis

Total RNA from the leaf tissues of nine plants was extracted by using TRIzol (Tiangen) according to the manufacturer's instructions. For northern blotting analysis, total RNA was separated in a 1.2% denaturing agarose gel containing 2% formaldehyde by electrophoresis, and then RNA bands were transferred onto a membrane as previously described (Liu, Liu, et al., 2020). The nylon membrane was hybridized with the probe, a digoxigenin‐labelled probe specific for the CGMMV sequence. Finally, the nylon membrane was incubated with CDP‐Star (Roche), and the results were visualized by a Tanon 5200 chemical luminous imaging system (Tanon). For RT‐qPCR assays, first‐strand cDNA was synthesized using HiScript III RT SuperMix (+gDNA wiper; Vazyme). Briefly, 2 μL of five‐fold‐diluted cDNA as a template, specific primers (Table S1), and the SYBR Green master mix (Vazyme) were used for qPCR analyses. The watermelon ClCAC gene and N. benthamiana GAPDH were selected as the reference genes, and the relative transcription level of each assayed gene was calculated using the 2^−ΔΔCt method (Livak & Schmittgen, 2001).

4.7. Co‐immunoprecipitation assays

For transient expression of ClAGO1‐FLAG or ClAGO5‐HA in planta, pEarleyGate201‐ClAGO1‐YC‐FLAG or pEarleyGate201‐ClAGO5‐YN‐HA was introduced into A. tumefaciens GV3101, followed by induction and infiltration into the youngest fully expanded leaves of 3‐week‐old N. benthamiana plants. Total proteins from leaves were extracted at 48–72 h postinfiltration in lysis buffer and 2 mM dithiothreitol and incubated with anti‐HA‐tag magnetic beads for 3 h at 4°C. The beads were collected by brief centrifugation, washed at least four times in wash buffer, and immunoblotted with anti‐HA or anti‐FLAG antibodies.

4.8. BiFC and subcellular localization assays

For BiFC, proteins were expressed in young leaves of 3‐ to 4‐week‐old N. benthamiana plants by infiltrating A. tumefaciens GV3101 harbouring the pEarleyGate201‐ClAGO1‐YC‐FLAG or pEarleyGate201‐ClAGO5‐YN‐HA plasmid. During 48–72 h after agroinfiltration, leaves were observed for corresponding fluorescence by confocal microscopy. For subcellular localization experiments, pBWA(V)HS‐ClAGO5‐GFP, pCNF‐Cla97C08G147180‐GFP, and pCNF‐Cla97C06G122830‐GFP vectors were transiently transfected into N. benthamiana leaves, together with plasmids expressing nuclear markers. Subcellular localization of the target protein was observed under a confocal laser scanning microscope (TCS SP8; Leica).

4.9. Yeast two‐hybrid assay

The coding sequences of ClAGO1 and ClAGO5 were introduced into pGADT7 and pGBKT7. Yeast constructs were co‐introduced into Y2HGold by lithium acetate‐mediated transformation, as previously described (Yang et al., 2020). Yeast cells were initially plated on a double dropout medium (SD−Trp/−Leu) to test if a good transformation efficiency had been achieved and were then transferred to SD/−Leu/−Trp/−Ade/−His/+X‐α‐Gal medium to analyse interactions between the expressed proteins. Plasmids pGADT7‐T were co‐transformed with pGBKT7‐lam as a negative control and with pGBKT7‐53 as a positive control.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Figure S1. Structural analysis of ClAGO (a), ClDCL (b), and ClRDR (c) proteins in watermelon.

Figure S2. Phylogenetic relationship of the Dicer‐like protein (DCL) and RNA‐dependent RNA polymerase (RDR) family proteins of watermelon. The amino acid sequences of DCL and RDR proteins identified in watermelon, with those from Arabidopsis thaliana as references, were aligned using the ClustalW software and subjected to phylogenetic analysis using the maximum‐likelihood method in MEGA 7 with 1000 bootstrapping replicates using default parameters. The bootstrap values are shown next to the branches. Major clades are boxed in different colours. Scale bar, 0.2 substitutions per site.

Figure S3. Heatmap showing the expression patterns of the ClAGO, ClDCL, and ClRDR genes in various organs. Relative expression levels of the ClAGO, ClDCL, and ClRDR genes in watermelon were determined by reverse transcription‐quantitative PCR (RT‐qPCR) at corresponding organs, including roots, leaves, flowers, stems, and fruit.

Figure S4. The prediction results of functional interaction protein.

Figure S5. The prediction results of ERF conserved domain and the effect of ETH on the expression of two ERFs. (a, b) The ERF conserved domain prediction results in Cla97C08G147180 (a) and Cla97C06G122830 (b). (c, d) Relative expressions of Cla97C08G147180 (c) and Cla97C06G122830 (d) with or without ETH treatments.

Figure S6. The relative transcript accumulation levels of NbAGO1a, NbAGO1b, and NbAGO5, genes in CGMMV‐infected Nicotiana benthamiana systemic leaves at 9, 12, 15 and 18 days postin.

Table S1. Primers used in this study.

Table S2. Correlation analysis between common differentially expressed transcription factor and ClAGO5 gene of two sampling stages (10‐and 20‐days postinoculation [dpi]) in CGMMV.

Liu, M. , Wu, H. , Hong, N. , Kang, B. , Peng, B. , Liu, L. et al. (2023) Argonaute 1 and 5 proteins play crucial roles in the defence against cucumber green mottle mosaic virus in watermelon. Molecular Plant Pathology, 24, 961–972. Available from: 10.1111/mpp.13344

DATA AVAILABILITY STATEMENT

The data supporting the findings of this study are available from the corresponding author upon request.