Current crop breeding relies entirely on genetic diversity, because change in deoxyribonucleic acid (DNA) sequences (mutations, allelic diversity) that affects traits of agronomic importance, can be identified from population or individual using appropriate sequencing-based and genetic methods. However, only a tiny fraction of the heritability of plant traits can be associated with genetic variations (Brachi et al., 2011), suggesting that DNA polymorphisms among the parents are not sufficient to explain the wide range of phenotypic variation observed and limiting our ability to improve the quantity and quality of crop production. This hidden heritability is particularly critical considering the current and future challenges in agriculture due to a rapid increase in global population and climate change that threaten food security. Furthermore, genetic diversity has been progressively reduced in many crops through selection schemes, as is the case for tomatoes (Lin et al., 2014) or when clonal propagation is the main route for plant multiplication as in grapevine (Verhoeven and Preite, 2014). In this context, epigenetic variations may provide an additional source of phenotypic diversity independently of DNA sequence changes. Understanding the contribution of epigenetic variation to the inheritance of traits of agronomic importance can potentially transform our current understanding of quantitative (epi)genetics and allows for the development of new tools for crop improvement (Bourrat et al., 2017). So far, the importance of plant phenotypic traits under epigenetic control is underestimated in crops, and the ability to use them in breeding schemes still needs to be implemented.

Epigenetics encompasses the complement of genetic information carried in chromatin beyond the DNA sequence. It includes DNA methylation which occurs on the 5th carbon of cytosine (5mC, Zhang et al., 2018) and on the 6th carbon of Adenine (6mA, Boulias and Greer, 2022), histone post-translational modifications (HPTMs), histone variants and chromatin remodelling (Lauria and Rossi, 2011). It can influence gene expression resulting in specific gene expression patterns during development and in response to stresses (Pikaard and Mittelsten Scheid, 2014). In plants, DNA methylation occurs at cytosines in the symmetrical CG and CHG and in the non-symmetrical CHH (H being A, T or C) sequence contexts (Zhang et al., 2018). In addition to methylome dynamics that take place during plant development, for example, during tomato fruit ripening (Zhong et al., 2013), changes in methylation patterns can be generated in response to environmental stresses (Furci et al., 2019), in part memorised and eventually transmitted to the following generations (Gallusci et al., 2022). Similarly, HPTM remodelling was described both during plant development and in response to abiotic and biotic stresses (Zhao et al., 2019). In addition, heritable DNA methylation variations, known as epi-alleles, affect agronomical relevant traits including sex determination in melon (Martin et al., 2009), rice height (Miura et al., 2009), yield in oil palm (Ong-Abdullah et al., 2015) and ripening in tomato (Manning et al., 2006), have been recurrently observed in plants.

As far as the grapevine is concerned, epigenetic studies are in their early stages (Gallusci et al., 2017). Yet the evidence is accumulating that epigenetic mechanisms are important for the development and responses to environmental challenges in grapevine (Table 1) as in many other plants (Pikaard and Mittelsten Scheid, 2014). In this review article, after a general description of the threats that climate changes impose on grapevines, we present the current knowledge concerning epigenetic studies in grapevines. We describe the plant epigenetic memories of stresses and discuss how these can be used as a new lever to improve grapevine tolerance to stresses. We also discuss ways to generate epigenetic diversity by exploiting the specificity of grapevine, a grafted, perennial clonally propagated, woody plant. We present innovative approaches to develop epi-breeding strategies in grapevine, which would allow one to directly use elite varieties to generate phenotypic diversity independently of sequence variations, thereby providing innovative and more rapid ways for grapevine improvement.

Epigenetic mechanisms include genomic DNA methylation which can occur on the 5th carbon of cytosine (5mC; Law and Jacobsen, 2010; Zhang et al., 2018) and on the 6th carbon of Adenine (6mA; Boulias and Greer, 2022). The latter has been recently identified in multicellular eukaryotes, and little is known about the mechanisms underlying 6mA regulation and roles in plant cells. De novo DNA methylation at cytosine is mediated by the RdDM pathway which involves the Domain Rearranged Methyltransferases 1, 2 (DRM1/2), DRD1 and 24nt-long small RNAs, and by the chromomethylase 2 (CMT2) with Decrease in DNA Methylation (DDM1) for CHH localized in heterochromatic regions (Zemach et al., 2013). After cell division, newly produced DNA is hemimethylated at CG and CHG symmetrical sites. Maintenance of methylation in the CG context relies on the activity of MET1 together with VIM1, 2 and 3. The CHG methylation maintenance depends mainly on the activity of CMT3. In the asymmetrical context of CHH, re-methylation of the newly synthesized DNA strands depends on both the RdDM pathway and CMT2 activity. CMTs are dependent on histone methylation mediated by KYP and SUVH5 and 6 (for a complete review of DNA methylation, Zhang et al., 2018). The 5mC can be actively removed by DNA glycosylase/lyase, also called DNA demethylases (DMT). DNA demethylation can also occur passively in a replication-dependent manner when the methylation machinery is not or poorly active after cell division (Liu and Lang, 2020; Figure 2A).

Epigenetics mechanisms also include histone variants and histone-post translational modifications (HPTMs) that mainly occur on lysine, threonine and serine residues within the histone amino-terminal tail. HPTMs include acetylation, methylation phosphorylation, ubiquitination, sumoylation and ADP ribosylation. Nucleosome composition and HPTMs can affect chromatin structure and directly determine the formation of a transcriptionally active or inactive state by modifying the interaction between DNA and nucleosomes (Lauria and Rossi, 2011). While histone acetylation localised within coding regions is associated with active transcription, histone methylation has been correlated to different types of effects on gene expression, depending both on the position of the lysine and on the number of methyls added (mono, di, tri-) (Figure 2B). Similar to DNA methylation, the addition and removal of histone marks are regulated by specific proteins respectively called writers (for example, Histone Methyltransferases/HMT, Histone Acetyltransferases (HATs) and erasers (Histone Demethylases/HDM, Histone Deacetylases/HDACs) (Figure 2B). The activity of enzymes involved in epigenome regulation relies on the availability of metabolic precursors and co-factors, that can be altered by environmental stresses (Leung and Gaudin, 2020).

Table

(A) de novo DNA methylation relies on the RNA-directed DNA methylation pathway (RdDM), which involves DRM1/2, small interfering RNAs (siRNAs) and scaffold RNAs in addition to an array of proteins. DNA methylation maintenance in both CG and CHG contexts are performed by MET1 and CMT3, CMT2, respectively. The maintenance of CHH methylation is carried out by the RdDM pathway and CMT2. DNA methylation can be passively lost after DNA replication in a non-targeted or actively removed by the DMTs enzymes that are targeted to specific loci. DNA methylation can impact gene expression by modulating chromatin accessibility to the transcriptional machinery when found in the promoter (coloured in red) and to a lesser extent within genes (coloured in blue). (B) Histone H3 methylation and acetylation and associated chromatin states as examples of Histone H3 post-translational modifications. Methylated histones can either be found in transcription-restrictive or permissive chromatin, depending on the modified lysine residue position on the histone tail (Kx) and the number of methyl groups carried (me1, me2, me3). Acetylated histones are correlated with active gene expression. Only the most studied acetylation marks in plants are listed here. Different protein families are involved in the addition (writers, in blue) or the removal (erasers, in orange) of histone marks. Amino acids that are modified and their position are indicated. DRM1/2 (Domain Rearranged Methyltransferase ½), CMT2/3 (chromomethylase 2/3), MET1 (DNA methyltransferase 1), VIM1-3 (variant in methylation 1-3), KYP/SUVH4 [KYP/Su-(var)3-9 homolog 4], SUVH5/6 [Su-(var)3-9 homolog 5/6], DRD1 (Defective in RNA-directed DNA methylation), DDM1 (Decrease in DNA methylation) and 24nt siRNA (24 nucleotide small interfering RNAs), DML (DEMETER(DMT)-LIKE demethylase), E(Z) (Enhancer of Zeste proteins), KYP (KRYPTONITE), LSD (Lysine Specific Demethylases), JMJ (Jumonji-type histone demethylases), CBP (p300/CREB binding protein), GCN5 (General Control Non-derepressible protein 5), MYST (MOZ, Ybf2/Sas3, Sas2 and Tip60), TAF (TATA-binding protein (TBP)-associated factor, RPD3 (Reduced Potassium Deficiency 3).

Works in Arabidopsis thaliana have shown the stable transmission of epigenetic marks over generations (Figure 4). Whether the memory of stresses can be transmitted to the next generation(s) of plants is however still a matter of intense research (Quadrana and Colot, 2016). Evidence of an intergenerational transmission (one generation) of epigenetic memory of stress has however been demonstrated in model plants and seems preferentially transmitted through the female gamete (Wibowo et al., 2016). In contrast, there is no clear demonstration of a transgenerational (several generations) inheritance of epigenetic information generated by stress. For example, inter- but not trans-generational memory of stress was described in the case of Arabidopsis thaliana after exposure to the mild drought stress of a first generation followed by four generations without applying any stress (Van Dooren et al., 2020). Mechanisms have already been described that control the balance between resetting and maintenance of epigenetic marks during gametophyte formation, fertilisation and embryo development (Ono and Kinoshita, 2021).

In clonally propagated plants such as grapevine, the progeny is generated by cutting. In this case, the maintenance of epigenetic marks is mediated through mitosis in the stem cells located in meristems. For example, in white clover (González et al., 2017) there is a clear demonstration of the maintenance of stress-induced epigenetic imprints in clonally generated progeny when still attached to the mother plant. A similar observation was performed in the progeny of dandelion plants, which is characterised by an apomictic type of reproduction (Verhoeven et al., 2010). The functional consequences of such an inheritance on the adaptation capacities of the progeny require additional investigations. Of course, we cannot rule out that when cuttings are generated, as in grapevine, re-juvenilization may occur and reset some or most of the parental epigenetic imprints. However, recent work has shown that even when going through plant regeneration which implies a major developmental reprogramming, the regenerated plant maintains part of the epigenetic imprints of the organ of origin and this impacts its phenotype (Wibowo et al., 2018).

As far as grapevine is concerned there is little work performed to investigate how the growing conditions of mother plants will impact the phenotypes of the progeny generated by cuttings, either for scions or rootstocks. Studies in poplar, a woody perennial, have however shown that ramets originating from the same plants but grown in contrasted environments displayed different methylation landscapes consistent with methylation imprints reflecting the environment in which plants are grown, rather than the parental origin (Guarino et al., 2015). To what extent these differences in methylation profiles result in a better adaptation to a specific environment has however not been investigated. Similarly in grapevine, the study of methylation variations of grapevine across regions in Australia (Xie et al., 2017) showed that methylation patterns varied with geographical location and vineyard management, suggesting that the plant environment, in a broad sense, was shaping the plant methylome. Consistently methylation profiles of Merlot and Pinot Noir (PN) were more affected by their location than by their clonal origin (Baránková et al., 2021). These results would suggest that the environment of the plant is more important than its origin in defining the methylome of plants and eventually plant phenotypes.

In contrast, cuttings collected from 60 adult Lombardy poplars growing in different environments, were gathered in a single place and compared for phenology and epigenetic parameters. Results indicate that epigenetic marks could be transferred from parents to asexually multiplied offspring likely reflecting the adaptation to the environment of the parents (Vanden Broeck et al., 2018). Similarly, the analysis of DNA methylation patterns of grapevine grown in various Argentinian vineyards indicates that they are more dependent on clonal origin than location (Varela et al., 2021). Hence, in these studies the clonal origin of plants seems to significantly impact their methylation landscape, suggesting that the progeny may have inherited the parental adaptation to the environment. However, different approaches between these different studies were used which makes it difficult to compare them and conclude about the relative contribution of parental origin versus the place of growth of the progeny on epigenetic information and phenotypes. In addition, various parameters may affect the memory of parental imprints through non-sexual reproduction such as growth conditions of the progeny, propagation methods, or simply the age of the progeny at the date of analysis because of an eventual epigenetic drift of the progeny, among others.

In this context, more studies are necessary to better understand to what extent and for how long parental epigenetic imprints can be transmitted to asexually reproduced grapevine plants. Important questions to investigate may include: 1) Are growing conditions generating an epigenetic drift that may lead to a better adaptation of plants to their environment? This question could be addressed by analysing the evolution of the epigenetic landscape of individual plants deriving from the same parental plant and grown in the contrasted environment; 2) to what extent environmentally induced changes in the epigenetic landscape are heritable after clonal propagation and 3) confer better adaptation of the progeny to the environment. Indeed, transgenerational priming of plants generated by clonal propagation (considering both rootstocks and scions and their epigenetic interactions) would provide innovative strategies for grapevine adaptation to climate changes, using already cultivated cultivars.

A grapevine is a grafted and clonally propagated plant. This makes it unlikely that the phenotypic diversity observed in this plant is solely due to genetic variations because in most cases genetic variants may remain hidden as heterozygous recessives. Epigenetic variants which may affect similarly both alleles in the genome could therefore be critical in shaping phenotypic variations in this plant. Consistently, clonal diversity within V. vinifera varieties was recently associated with methylation changes illustrating the importance of epigenetic markers for intra-varietal diversity (Varela et al., 2021).

Grapevine provides a unique system to investigate the consequences of epigenetic variations in clones, their potential role in shaping the phenotypes of plants and opportunities to develop epi-breeding strategies. The generation of Epigenetic Recombinant Inbred Lines (EpiRILs), which takes advantage of the segregation and recombination between heritable differentially methylated regions (DMRs) to perform (epi)QTL mapping (Quadrana and Colot, 2016), has proven to be an efficient strategy to investigate the contribution of heritable epigenetic variation to agronomic traits. This strategy is based on sexual reproduction where parental lines having the same genotype but with different epigenomes are crossed. After inbreeding over seven generations to fix the epigenetic variations, lines with stable epigenetic differences can be compared in a common genetic background. As examples, Arabidopsis EpiRILs have been generated by crossing a wild-type plant with plants with the same genetic background, but mutated in the gene encoding the DNA METHYLTRANSFERASE-1 (MET1) or DECREASE IN DNA METHYLATION (DDM-1), proteins that are both involved in the maintenance of DNA methylation (reviewed in Catoni and Cortijo, 2018).

EPIRIL populations cannot be developed in crops such as Grapevine that require long generation times and are essentially clonally propagated. In this case, other strategies need to be implemented to investigate the contribution of epigenetic variation to phenotypic diversity and traits of agronomic importance in the absence of DNA sequence polymorphisms. As a first strategy, we have now developed a unique population of isogenic plants displaying epigenomic diversity. This Grapevine “EpiClonal” population was initially created from cuttings of a single Pinot Noir plant that were in vitro cultured to generate a population of plants with identical genomes and is currently being extended to other cultivars. Stems were subjected to demethylating agents, which cause variable degrees of genome-wide hypomethylation and lead to diverse DNA methylation, eventually to diverse phenotypes. The “EpiClonal” populations are currently under vegetative propagation and will be used for methylome and phenotypic analyses. Alternatively, stress conditions may be used to generate an epigenetic drift. Using plants produced from a limited number of mother plants, it is expected to impact the epigenome of plants in different ways if they are grown in contrasted conditions with recurrent stresses, thereby generating epigenetic and phenotypic diversity.

This epigenetic diversity could then be used in different ways: (1) evaluate the resilience of treated versus non-treated plants when subjected to similar stresses; (2) measure the epigenetic drift and identify epi-allelic regions that are critical for the better adaptation of plants to stresses; (3) investigate the stability of epigenetic imprints that have been created over seasons and (4) during plant propagation. Finally, parental priming of rootstocks and scions could provide new ways to develop grapevine propagation strategies better adapted to climate changes building on the epigenetic exchanges existing between the graft partners (Rubio et al., 2022). Such populations could also be used for epigenome-wide association studies (EWAS), an approach that has emerged as a powerful way to identify DNA methylation variants associated with phenotypic changes as shown following the identification of an epiallele underlying the metastable mantled somaclonal variant of oil palm (Ong-Abdullah et al., 2015).

There is evidence of various types of epigenetic memories in plants (Gallusci et al., 2022) and this may provide innovative ways to generate lines with new adaptive potential, yet without affecting their genetic information. Figure 4 summarises the different strategies that can be developed in the case of grapevine to use epigenetic regulations and memories as a new lever to develop cultivars and/or rootstocks better adapted to their environment.

Table

(A) Untargeted epigenetic variation can occur naturally or be generated experimentally. Epigenetic diversity can be found in natural populations or occur spontaneously in cultivated populations. Environmental stresses may also result in epigenetic variations that are memorized by plants and generate epigenomic variations. Epigenetic variations can be created when two distantly related species are crossed (genomic choc). An epigenetic dialogue occurs during grafting that impacts the epigenome of both partners, which may lead to new epiallelic combinations. In vitro propagation of plants in the presence of methylation inhibitors generates randomly hypomethylated genomes that result in clones with modified epigenomes and identical genomes. (B) Targeted epigenetic changes can be created using epigenome editing for example using CRISPR-ca9. The use of specific small RNAs (24nt) can allow addressing the novo methylation by the RdDM at specific loci.

This work was supported by the EPIDEP program (Plant National du Dépérissement, PNDV). BR received a grant from EPIDEP. MB received a PhD grant from the CNIV (Comité National des Interprofessions des Vins) and a Post-doctoral grant in the TEPIMEPORY program (Plant Science Project, University of Bordeaux).