Epigenetics investigates heritable changes in gene function that occur without a change in DNA sequence (Holliday, 2006). Most of the epigenetic changes (Histone modification, DNA methylation, chromatin remodeling, and chemical modification) occur due to microRNAs (miRNAs). The epigenetic mechanisms are tightly connected to cellular development and differentiation (Barter et al., 2012). DNA methylation is a conserved epigenetic silencing process that is involved in a variety of biological functions, including the regulation of gene expression, prevention of the spread of transposons, and defense against these organisms. Additionally, changes in DNA methylation can result in abnormal development (Zhang et al., 2018). Mostly DNA methylation in plants occurs in the CG, CHG, and CHH sequences (where “H” can be either an A, C, or T), and it is especially prevalent over heterochromatic transposable elements (TEs) and repetitions, where it is crucial for transcriptional gene silencing (TGS) (Zhang et al., 2018). However, many plant species have DNA methylation over gene bodies but in this case precise role is still unknown (Bewick and Schmitz, 2017).

In plants, DNA methylation usually happens by RNA-directed DNA methylation (RdDM). RdDM is a biological mechanism in which DNA methylation is controlled by non-coding RNA molecules. RdDM-induced DNA methylation is usually linked to transcriptional suppression of the genomic regions targeted by the pathway (Erdmann and Picard, 2020). DNA methylation can change gene expression through direct and indirect mechanisms. Further, when it occurs in promoter region of a gene it inhibits transcription (Wierzbicki et al., 2008; Zhang et al., 2018). The key players of RdDM pathway are Pol IV (RNA Polymerase 4), RDR2 (RNA-dependent RNA polymerase 2), DCL3 (Dicer-like 3), and AGO4 (Argonaute 4) (Dinh et al., 2013). According to the canonical RdDM pathway, which is best explained in Arabidopsis, generation of 24-nt small interfering RNAs (siRNAs) depends on Pol IV. Pol IV transcribes single strand RNA (ssRNA), which is converted to double strand RNA (dsRNA) by RDR2. Then dsRNA is diced by protein DCL3 resulting in siRNAs (siRNAs 24nt). These siRNAs are loaded in AGO4 protein and AGO4 interacts with RISC (RNA-induced silencing complex) and methylates DNA (Matzke and Mosher, 2014). Apart from these proteins, some other proteins also play roles in RdDM pathways for example, SHH1 (SAWADEE HOMEO-DOMAIN HOMOLOG 1), CLSY (CLASSY), NRPD (NUCLEAR RNA POLYMERASE D), DRM (DOMAINS REARRANGED METHYLASE), RDM (RNA- DIRECTED DNA METHYLATION), RRP6L (RRP6-like), DRD (DEFECTIVE IN RNA-DIRECTED DNA METHYLATION), DDM (DECREASE IN DNA METHYLATION), IDN (INVOLVED IN DE NOVO), HEN (HUA ENHANCER). Moreover, SHH, NRPD, NRPE (DNA-DIRECIED RNA polymerase V subunit 1), DRM, RDM1, RDM2, RDM3, RMD4, and RDR have SAWADEE, RNA_pol Rpb2, RNA_pol Rpb1, DNA methylase, RdDM_RDM1, RNA_pol Rpb4, Spt5-NGN, lwr1, and RdRP conserved domain, respectively (Wassenegger and Krczal, 2006; Ream et al., 2009; He et al., 2009a; He et al., 2009b; Henderson et al., 2010; Hirtreiter et al., 2010; Zhang et al., 2013; Sasaki et al., 2014). CLSY, DRD, and DDM have two conserved domains; Helicase_C and SNF2-rel_dom (Smith et al., 2007; Hu et al., 2013). RRP6L also has two domains (DNA_pol_A_exo1 and HRDC) (Lange et al., 2008). IDNs (zf-XS, XS and XH), HEN (Hen1_Lam_C, Methyltransf_31, and dsRBD2), and AGO (n-side PAZs, C-side Piwi, and ArgoMid) each has three conserved domains (Ausin et al., 2009; Huang et al., 2009; Zhao et al., 2014). Further, DCL has six conserved domains (DEAD, Helicase-C, Dicer_dimer, PAZ, Ribonuclease_3, and dsrm) (Margis et al., 2006). In most of the crops, AGO, DCL, and RDR gene families have been identified and characterized.

DNA methylation performs diverse functions in plant growth and development including stress tolerance (Zhang et al., 2018). RdDM-related genes have been well studied in various crops for their critical roles in stress tolerance. For example, RNA Pol V (NRPE), RDR1/2/6, DCL2/4, CaDCLs, and RDR1 have been reported for their roles against different stresses in different crops i.e., Arabidopsis and pepper (Garcia-Ruiz et al., 2010; López et al., 2011; Qin et al., 2018). Further, SlAGO4 showed a negative role in salt and drought stress tolerance in tomato (Huang et al., 2016). DNA methylation may have potential roles in flowering, fruit setting, seed development, and fruit ripening. In Arabidopsis low levels of methylation caused delay in flowering (Jones and Sung, 2014). AtRRP6L1 and AtRRP6L2 regulated flowering in Arabidopsis via repressing the expression of FLC (Flowering Locus C) (Shin and Chekanova, 2014). On the other hand, high levels of DNA methylation are important for orange fruit development and ripening (Huang et al., 2019). In the abscission zone of the citrus sinensis fruit, key players of the RdDM pathway are down-regulated (Sabbione et al., 2019). Further, low DNA methylation levels in apples resulted in reduced fruit size (Daccord et al., 2017). In tomato a change in hypermethylation in CNR (COLORLESS NON-RIPENING) indicated abnormal fruit ripening (Manning et al., 2006). Cheng et al. (2018) has also reported the involvement of RdDM pathway in strawberry fruit ripening. During the ripening process in apples, DNA methylation of the MdMYB10 promoter affected gene expression and fruit color (El-Sharkawy et al., 2015).

The role of RdDM in seed development and germination has been studied (Xiao et al., 2006; Zhong et al., 2020). Furthermore, the data suggested that DNA methylation can play a critical role in seed dormancy (Kawakatsu et al., 2017). The association of DNA methylation with seed development has been reported in Arabidopsis thaliana and Brassica rapa (Grover et al., 2018). For example, the involvement of AtHEN1 in ovule development (Wei et al., 2020). Various phases of tissue have different degrees of DNA methylation. For example, the shoot apical meristem of early peach seedlings exhibited a higher amount of DNA methylation than adults (Bitonti et al., 2002). Most probably, the diverse DNA methylation levels in different tissues at different growth stages are due to different DNA methylation pathways (Bartels et al., 2018) and gene expression levels.

Grapevine (Vitis vinifera L.) is among the top fruit crops grown all over the world. The grapes are consumed in different ways including fresh fruit (table), wine, juices, and raisins (dried) (This et al., 2006). RdDM pathway genes have been identified in different plant species including Arabidopsis (Kurihara et al., 2008), Brassica rapa (Cao et al., 2016), Soybean (An et al., 2017), Citrus (Mosharaf et al., 2020), Apple (Li et al., 2019), Rice (Ahmad et al., 2019), Maize (Qian et al., 2011), Tomato (Bai et al., 2012), and Ginkgo (Gao et al., 2020). However, in most of the species scientists have reported only three RdDM (DCL, AGO, and RDR) gene families. There is a lack of information about RdDM genes in grapevine. The involvement of RdDM genes in different mechanisms of plant growth, development, and response to different stresses (biotic and abiotic) and antiviral defense, justifies the need of detailed bioinformatics studies of RdDM genes in grapevine. Here, we elaborated comprehensive genome-wide identification of RdDM grapevine genes, including phylogenetic analysis, chromosomal positions, intron-exon distribution, motif analysis, evolutionary history, selection pressure, and cis-elements analysis. Various transcriptome analyses under biotic stress, abiotic stress, hormone treatments, and at different stages of growth were performed. The expression of selected genes was further investigated during progressive phases of seed development in seedless and seeded cultivars. Moreover, the integration of all these studies can provide inklings about VvRdDM pathway genes function and their involvement in multiple biological processes. Our study provides basic information about RdDM grapevine genes and will provide new candidate genes for functional studies.

In plants, RdDM pathway genes control the epigenetic states via De novo methylation and play important roles in multiple biological processes (Matzke and Mosher, 2014; Zhao and Chen, 2014; Matzke et al., 2015). Due to recent advancements in whole-genome mapping and next-generation sequencing, plant breeders are paying more attention in epigenetic research and its use in breeding programmes. The RdDM pathway genes have been characterized in many plants i.e., Arabidopsis, Citrus, Tomato, etc (Kurihara et al., 2008; Qian et al., 2011; Bai et al., 2012; Cao et al., 2016; An et al., 2017; Ahmad et al., 2019; Li et al., 2019; Gao et al., 2020; Mosharaf et al., 2020). However, there have been limited reports for RdDM pathway genes in grapes. Previously, Zhao et al. (2014) identified only three important RdDM pathway-related gene families in grapes i.e., AGO, DCL, and RDR. Here by using the latest high-quality grape reference genome data, we identified 60 RdDM grapevine genes belonging to 14 different subfamilies. Moreover, integrated bioinformatics, transcriptomics, and expression profiling to identify their potential functional roles.

Proteins sequences of one monocot (rice) and three dicots (Arabidopsis, grapes, and tomato) species were used to study the phylogenetic relationships of different RdDM families. Based on phylogenetic studies, VvAGO, VvDCL, and VvRDR proteins were more closely related with SlAGO, SlDCL, and SlRDR proteins, respectively. However, VvRDM proteins were closely related to AtRDM proteins. These results suggest that AGO, DCL, and RDR protein families of grapes and tomato have similarities in evolution and ancestral history. Likewise, RDM proteins of grapes and Arabidopsis might have common ancestors. The pattern of exon-intron distribution within a gene family can provide inklings about evolutionary history. Some of the VvRdDM genes families (AGO, DCL, and IDN) showed similar exon-intron numbers within the same sub-families, proposing that these subfamilies are functionally conserved. Other families showed differences in exon-intron number and distribution patterns even within the subfamilies. These results suggest that during the evolutionary process these subfamilies have either gained or lost exon/introns. Changes in exon number during the evolutionary process play important functions in the evolution of new gene families (Xu et al., 2012). Further, the newly evolved genes are initially redundant and gain new functions with the passage of time (Dias et al., 2003). Like intron-exon arrangement, most of the families (DRD, DCL, CLSY, and DDM) showed similar motif distribution patterns within the subfamilies suggesting these subfamilies are conserved during the evolutionary process. More interestingly, DRD, DCL, CLSY, and DDM all have a common domain (Helicase_C), which consists of ATP-dependent proteins which participate in epigenetic mechanisms via chromatin remodeling by using the energy provided by ATP (Flaus et al., 2006). This indicates this domain has some key roles in the RdDM pathways. However, RDM members showed differences in motif arrangements.

Gene duplications and exon-intron patterns have played a key role in the expansion of gene families during the evolutionary process (Cannon et al., 2004). In case of duplicated gene pairs previous trend was observed i.e., members of four duplicated gene pairs (VvAGO10a-VvAGO10b, VvIDN2a-VvIDN2b, VvAGO5a-VvAGO5b, and VvSHH1a-VvSHH1b) showed a similar trend of exon-intron and motif distribution patterns. Moreover, some duplicated gene pairs (VvAGO10a-VvAGO10b, VvAGO3a-VvAGO3b) showed similar expression patterns under high temperature treatment ( Figure 11 ). Only thirteen genes (8 segmental and 5 tandem) belonging to 4 families (AGO, IDN, SHH, and RDR) are the result of duplication events. All duplicated gene pairs have a purifying mode of selection. These results suggest that these four families are diversified and all others are conserved. Further, the AGO family is more diversified because 50% of members of this family have undergone duplication events. Overall, in the expansion of VvRdDM gene families, we noticed the previous trend that segmental duplication occurs more often than tandem duplication (Cannon et al., 2004). Synteny analysis can provide inklings about the gene functions. We identified 19 syntenic pairs between grape and Arabidopsis RdDM genes. These findings propose that these homologous genes have common ancestors before the speciation. However, relying solely on syntenic linkages to explain evolutionary relationships is difficult. Interestingly, two genes VvCLSY2 and AT5G59390 paired with more than one gene, implying that these genes have gained more changes during the evolutionary process. Based on the stated functions of orthologous genes in one species (Arabidopsis) functions can be predicted in other species (grapes).

According to promoter analysis, most of the VvRdDM genes have hormone-responsive cis-elements. However, ABA response element (ABRE) was a more abundant hormone-responsive cis-element, and genes VvDCL2, VvRDR1a (contained ABRE elements) significantly responded ABA treatment, moreover, the DEGs after MeJA treatment (VvRDR1b, VvAGO7-CGTCA-motif, TGACG-motif) and GA₃ treatment (VvDCL2, VvAGO6a, VvAGO7-P-box) also contained related elements, suggesting the involvement of VvRdDM genes in hormone signaling pathways. Previously, Mosharaf et al. (2020) has also reported the presence of more hormone-responsive cis-elements in CsRdDM genes. Apart from hormone-responsive elements, stress and growth-related elements were also identified, signifying the role of RdDM genes in disease resistance and grapevine growth signaling pathways. These findings predict that regulation of RdDM genes is a complex process and rationalizes the need for further research.

Transcriptome data from SRA can provide some information for further research. In this study, previously published transcriptome data was used to evaluate the response and expression of VvRdDM genes against different treatments. Following different stress treatments or in different organs, the AGO family showed high expression or significant expression change as compared to control, suggesting it’s involvement in multiple biological processes ( Figure 11 ). Previously, AGO family members have displayed diverse functions in different species. For example, in A. thaliana AGO mutant showed enhanced susceptibility to fungi (Ellendorff et al., 2009) and in another study AtAGO4 mutants were associated with resistance to bacterial pathogens (Le et al., 2014). SlAGO4A has been reported for its role in salt and drought stress tolerance in tomato (Huang et al., 2016) and OsAGO17 regulated seed size in rice (Zhong et al., 2020). AGO proteins are core components of the RNA-induced silencing complex (RISC) (Hutvagner and Simard, 2008), and this may be the potential explanation to its diverse roles. VvRDM2b and VvAGO1 ( Figure 7 – 9 ) gene showed a broad spectrum of high expression in different organs at various stages even under different treatment. VvRdDM genes not only showed diverse expression but also showed spatio-temporal expression patterns. For example, VvRDR1a, VvIDN2a, and VvNRPD2b, were significantly more expressed at the hard green stage than at the ripened stage. VvAGO5a was significantly differentially expressed at all stages of fruit ripening. Most of the VvRdDM genes showed significant down-regulation following different treatments ( Figure 11 ). Surprisingly, VvDCL2 displayed same expression trend under the conditions of downy mildew, low temperature, GA₃ and ABA treatment. Further, the expression level of VvDCL2 was significantly down-regulated following low temperature treatment (108 to 27) suggesting some role in the adaptation of grape to low temperatures. Our results are also supported by the earlier findings i.e., the activity of DCL2 in A. thaliana was related with temperature (Zhang et al., 2012). Prupe DCL under drought treatment was differentially expressed in peach (Belal et al., 2022). These findings suggest, that the functions of DCL2 are diverse, and further research is needed for exploring the exact functions of VvDCL2. The VvRdDM genes showed differential expressions in different organs, indicating their involvement in the growth and development process. Our results are in line with the previous findings, as AtRRP6L1 and AtRRP6L2 (homologous genes of VvRRP6L1a) regulated flowering (Shin and Chekanova, 2014), DDM1 (homolog VvDDM1) mutant effected seed development process (Xiao et al., 2006) in A. thaliana. Silencing of FvAGO4 (homolog VvAGO4a) during fruit development resulted in early ripening of strawberry fruit (Cheng et al., 2018). Based on these findings, we speculate that VvRdDM genes can play critical role in the growth and development of grapes.

Expression patterns of fifteen genes during seed development were compared among four different (2 seeded and 2 seedless) cultivars of grapes. Interestingly, three genes (VvSHH1a, VvDRD1a, and VvIDN2a) showed higher expression in seeded cultivars especially at 40 and 50 DAF, signifying their key roles in grapevine seed development pathways. Previous results support our findings; for example, in Hypericum perforatum L, a gene with high similarity to AtIDN2 demonstrated differential expression in the ovule of apomictic H. perforatum. Furthermore, in A. thaliana, silencing the IDN2 gene led to decreased seed set and changes in seed size (Basso et al., 2019). However, in seedless cultivars, VvRDR1a and VvRDR6 showed significantly higher expression throughout all phases of seed formation. Similar trend of higher expression was noted during transcriptomic analysis. Relying on the findings, we can predict that these two genes might have roles in ovule abortion in grapes. Our results are in concordant with the previous findings of Grover et al. (2018) who noticed the abortion of brassica rapa seeds due to maternal alterations in the Pol IV-dependent small RNA pathway. They reported that RdDM pathway genes have critical roles in brassica rapa seed development. These findings rationalize the need for a future study about the functional characterization of these (VvRDR1a and VvRDR6) genes.

A total of 60 RdDM genes in grapevine were identified, and comprehensive bioinformatic and expression analysis were performed. VvRdDM proteins were evaluated in terms of chromosomal positions, exon-intron distribution, phylogenetic, and evolutionary history. Expression analysis including transcriptomic analysis of VvRdDM genes following different stresses and hormonal treatments and qRT-PCR analysis of selected VvRdDM genes during different stages of seed development suggested their roles in multiple biological processes. VvRDR1a and VvRDR6 gene were proposed as candidate gene resources for further functional characterization. Our study provides new resources for grape molecular breeding and inklings about the role of epigenetics in grapevine growth and development.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

LW and RX designed the research. LW and GD supervised the experiments. RX performed most of the experiments. BA, LY and CL provided technical assistance. RX and XS analyzed the data. BA, RX and LW wrote the article with contributions of all the authors. All authors contributed to the article and approved the submitted version.