Self-grafting-induced epigenetic changes leading to drought stress tolerance in tomato plants

Abstract Grafting is widely used as a method to increase stress tolerance in good fruiting lines of Solanaceae plants. However, little is known about how grafting, affects epigenetic modifications and leads to stress tolerance, especially within the same line. Here, we studied the effects of self-grafting in tomato plants on histone and DNA modifications and changes in gene expression related to drought stress. We found that at the three-leaf stage, 1 week after self-grafting, histone H3 K4 trimethylation and K27 trimethylation changes were observed in more than 500 genes each, and DNA methylation changes in more than 5,000 gene regions at the shoot apex compared to the non-grafted control. In addition, two weeks after the epigenomic changes, global expression changes continued to be observed at the shoot apex in several genes related to the metabolic process of nitrogen compounds, responses to stimulus, chromosome organization, cell cycle-related genes, and regulation of hormone levels. Finally, these grafted seedlings acquired remarkable drought tolerance, suggesting that epigenomic modifications during the wound-healing process mitigate stress tolerance in tomato plants.


Introduction
Plant grafting is a technique of primarily connecting the scion and the root 'rootstock'. 1 This technique is commonly used to increase stress tolerance, such as soil-borne disease and abiotic stress like drought, and increase yield by developing a vigorous rootstock in Solanaceae and Cucurbitaceae crops. 2 To increase production and reduce losses due to environmental stress, graft vegetables with vigorous rootstock, especially from different species or hetero-grafting, to reduce the stress in the shoot. 3,4 Tomato (Solanum lycopersicum) is commonly used as a model for vegetable plants because its genomic information is readily available, is easy to cultivate and graft, and is high-yielding. 5, 6 We previously found that Momotaro (Momo) homo-grafting, using the same species as a vigorous rootstock, and self-grafted, increased drought tolerance after 12 days of drought stress treatment. 7 While it was expected in homo-grafting, we observed that selfgrafting also increased drought tolerance in Momo shoots. Several differentially expressed genes were found in the shoot apical meristem; however, the mechanism behind its activation is still unclear.
Some recent studies have focused on long-distance communication of grafted junctions during the wound-graft-healing process, such as the movement of several hormones, metabolites, macromolecules, proteins, and RNA signals. 1,8,9 The latter is highly important because mobile small RNAs associated with gene silencing are exchanged in the graft junctions and cause genetic modifications to the DNA. 9 Hence, further analysis is required to determine whether these epigenetic changes are involved in essential traits such as stress tolerance in grafted plants. Plants must adapt to environmental stress through mechanisms that include stress acclimation by priming, usually referred to as stress memory. This phenomenon involves priming a seed or a plant to mild stress, causing the plants to store the information. 10 However, this information enhances adaptation to subsequent stresses and accelerates the stress response. Additionally, several studies have revealed epigenetic processes in plant adaptation to abiotic and biotic stresses, 11 including memory mechanisms involved in epigenetic changes, such as histone modification, DNA methylation, chromatin dynamics, and microRNA. 12 This study assessed whether self-grafted tomato plants could improve drought tolerance and alter transcriptional and epigenetic changes, such as histone H3K4me3, H3K27me3, and DNA cytosine methylation, compared to non-grafted Momo. This tomato variety is mainly cultivated in Japan and is known for its sweetness, hardiness, and flavour. 13 Our data further support the hypothesis that grafting alters epigenetic marks in the apical tissues of the scion, including the meristems, resulting in increased drought stress tolerance through stress memory within specific genes.

Plant materials, growth conditions, and grafting
Momotaro (Momo) tomato seeds were grown in a large growth cabinet (Espec Ltd., Osaka, Japan) at 28/22°C (day/ night), with a 12-h photoperiod (300 μmol/m 2 /s) and 85/75% relative humidity. Seeds were obtained from Takii Seed Co., Ltd., Kyoto, Japan. The soil was kept moist by daily irrigation to avoid water logging. When Momo plants had three true leaves (three-leaf stage; 10-d old seedlings), scion and rootstock stems were cut diagonally and attached to the same or different individuals using a joint graft holder (Seem, Kita-Kyushu, Japan) for self-grafting. The plants were maintained under high humidity (85-90%) and low light for 3 days until grafting was completed. The grafted plants were then acclimated to the large growth cabinet, and healthy grafted plants at 7 days post-grafting (still at the three-leaf stage) were used for epigenetic modification analysis. In addition, Momo 10-dold plants (three-leaf stage) were used as non-grafted controls. Shoot apexes (approximately 10 mm long) containing apical meristematic tissue were cut from each plant with a surgical scalpel and applied for chromatin immunoprecipitation sequencing (ChIP-seq) and whole-genome bisulfite sequencing (BS-seq) below.

ChIP-seq analyses with antibodies against histone H3 K4 trimethylation and K27 trimethylation
ChIP assay for three biological replicates per line was performed as described 14,15 with anti-trimethyl-Histone H3-Lys4 (EDM Millipore Co., Temecula, CA, USA) and antitrimethyl-Histone H3-Lys27 (EDM Millipore) antibodies. In addition, Dynabeads Protein A (Thermo Fisher Scientific), NucleoSpin® Gel and PCR Clean-up (Macherey-Nagel, Düren, Germany), NEBNext® Ultra™ II DNA Library Prep Kit, and NEBNext® Multiplex Oligos (New England Biolabs, Ipswich, MA, USA) were used for ChIP-DNA sequencing by MiSeq® (Illumina). Enriched peaks with significant grafting effects (P values of < 0.001 as calculated fold enrichment between the average of the three replicates) were analysed by the ChIP-Seq program. 16 A significant increase in methylation by grafting was termed hypermethylation, whereas the loss was named hypomethylation.

Whole-genome BS-seq analyses
For cytosine-5 DNA methylation analysis, whole-genome DNA of three biological replicates per line was extracted as described, 17 end-repaired, and ligated to the enzymatic methyl-seq (EM-seq) ligators of NEBNext® UltraTM II reagents (New England Biolabs). Libraries were prepared by NEBNext® EM-seq Kit (New England Biolabs) and sequenced using MiSeq® (Illumina). We identified the sequence variants 18 and extracted the single nucleotide polymorphisms (SNPs). 19 SNPs with cytosine or guanine in the control samples and thymine or adenine in the grafting samples were identified as unmethylated sites by grafting (hypomethylation), and vice versa was recognized as methylated sites (hypermethylation) with Python 3 in-house scripts and defined these as differentially methylated positions (DMPs). CG and CH (CHG, CHH) methylation contexts in which H is any base other than G, were included in the results.

Gene ontology categorization
The differentially methylated genes (DMGs) were analysed for gene ontology (GO) enrichment using the PANTHER 14.0 tool 20 with a P value of 0.05. Only the GOs categories of interest were chosen for the family of biological processes.

Drought stress treatment and sample collection
In the drought stress treatment, each plant was cultured in a large growth cabinet to about 10 cm plant height and similar leaf area (Fig. 2B). The non-grafted Momo controls were at the four-leaf stage at 14-d old, the self-grafted plants were at the five-leaf stage 14 d after grafting, and the predrought-experienced plants (deprived water for 4 d or 7 d from the three-leaf stage and then 4 d water supply) were at the four-leaf stage. For drought treatments, each plant was not watered for 12 d under a large growth cabinet. The plants were then irrigated for 3 d, and those with new leaves emerging from the apical meristem, or from axillary buds due to loss of apical dominance, were counted as surviving plants. 7 Meanwhile, no surviving plants showed wilted and no growth. Shoot apexes (approximately 15 mm long) containing apical meristematic tissue were collected before stress (D0) and on day 3 (D3) during drought stress treatment in three biological replications in each group.
2.6. RNA extraction, sequencing, data analysis, and RT-PCR TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA) and RNeasy Minikit (Qiagen, Hilden, Germany) were used to extract and purify total RNA. RNA pools of the three biological replicates of control and grafted (Day 0 and Day 3 of drought stress treatment) were used and sequenced by Ribo-Zero rRNA Removal Kit [Plant Leaf (Illumina, San Diego, CA, USA)], SureSelect Strand-Specific RNA Library Prep (Agilent Technologies), and NextSeq 500 (Illumina). All data analysis was done following refs [21][22][23][24] with ITAG4.0 tomato annotation information 5 and featureCounts v.2.01. 25 In addition, PrimeScript RT Reagent Kit (Takara Bio Inc., Shiga, Japan) and KAPA SYBR FAST universal (Kapa Biosystems Inc., Wilmington, MA, USA) were used to perform real-time quantitative RT-PCR. SAMs samples of control, grafted, and pre-drought stress-treated plants (on Day 0 and Day 3) were run in the CFX Connect Real-Time System (Bio-Rad Laboratories, Hercules, CA, USA) with the following cycling conditions: 3 min at 95°C, then followed by 40 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 1 min. The RT-qPCR primers are listed in Supplementary Table S1. The internal control was set to the housekeeping gene 18S rRNA. The 2 −∆C´T method 26 was used to calculate each gene's relative expression.

H3K4me3, H3K27me3, and DNA methylation preferentially mark genes in grafted compared with control plants
To investigate whether self-grafting induces epigenetic changes in the Momo scion, ChIP-and BS-seq analyses were performed on shoot apex containing apical meristem at the three-leaf stage of ungrafted control and grafted plants. Selfgrafting was done at the three-leaf stage, after which the grafted plants took a week to recover, but no new fourth leaf emerged during this wound-healing process (Fig. 1A). The grafted plants grew transiently slower than the ungrafted controls. The result of ChIP analyses showed that the selected histone H3 modifications were consistent with normal and common patterns, with H3K4me3 showing a peak at the beginning of the 5' end of the gene and H3K27me3 indicating a uniform distribution throughout the gene body ( Supplementary  Fig. S1). Next, we found 630 hypermethylated and 43 hypomethylated DMGs in H3K4 and 527 hypermethylated and 32 hypomethylated DMG in H3K27 that resulted from grafting (Supplementary Fig. 1B and Tables S2 and S3). All DMGs of H3 modifications were respectively listed with fold enrichments calculated with pile-up values in each grafted vs. control (positive: fold increase, negative: fold decrease in grafted samples), in which values were normally distributed in either grafted or control samples. In addition, DNA hypermethylation and hypomethylation (at least one DMP) were observed in 3,334 and 3,002 DMGs, respectively, by grafting (Fig. 1C, Supplementary Table S4). Furthermore, Venn diagrams dividing active and inactive modifications showed that histone H3 modifications and DNA methylation did not intersect much, indicating that either epigenetic modification occurred independently in each DMG (Fig. 1C).
Gene ontology (GO) enrichment analyses of these DMGs were classified into 'nitrogen compound metabolic process', 'cellular component organization or biogenesis', 'biosynthetic process', 'cell cycle', 'chromosome organization', 'response to stimulus', 'biological regulation', 'regulation of gene expression', and 'regulation of hormone levels' (Fig. 1D). The enrichment was notable in higher GO categories such as biological regulation and middle categories such as response to stimulus.

Momo scion increases survival rate to drought stress treatment through self-grafting
To study the effect of self-grafting and pre-drought stresstreated Momo in drought tolerance, we subjected tomato plants to 12 days of drought and 3 days of recovery ( Fig. 2A). Each drought treatment was applied to plants approximately 10 cm in height and with similar leaf area, but the grafted plants had five true leaves (five-leaf stage), one leaf more than the ungrafted control due to slow stem growth but increased leaf growth (Fig. 2B). Expression levels of 13 genes involved in lowering, such as single flower truss (SFT, Solycg063100) and falsiflora (FA, Solyc03g118160), 27 were quite low at the shoot apex of both non-grafted control and grafted plants and were not significant difference between them (Supplementary Fig. S2 and Table S5). Their low expression indicates that the floral bud transition has not yet occurred. In general, tomatoes such as Momo develop their first inflorescence on the primary stem after eight leaves have developed. 28 Thus, most of the shoot apical tissue at the four-to five-leaf stage is in the vegetative growth stage. Control plants survival after 12 days of drought stress treatment was severely reduced to 20%, which recovered from the apical meristem area ( Fig. 2C and D). The remaining 80% of the plants remained brown and never emerged with new leaves. In contrast, the self-grafted Momo was significantly more resistant to drought stress (56% recovery rate). Surviving grafted plants developed new leaves from the apical meristem (about 43%) and axillary bud (about 13%), which were thought to have reduced apical dominance. However, pre-drought treatment (4 and 7 days) did not increase tolerance but conversely decreased it (no survival individuals). These results show that drought tolerance was acquired by self-grafting without using a vigorous rootstock, while this effect was significantly different from the pre-drought stress exposure.

Distribution of H3K4me3, H3K27me3, and DNA methylation causes different gene expression intensity levels in grafted
In a comparison of grafted and control at D0 [Gr/Co (D0)], grafted transcriptionally upregulated 809 genes and downregulated 77 genes (Fig. 3A). Several hormone-related genes include ABA receptors, auxin, cytokinin, ethyleneresponsive transcription factors, gibberellins (GA), and jasmonic acid were differentially expressed by grafting (Supplementary Table S6). In addition, DNA replicationrelated genes, stress-related genes, chaperone/heat shock protein genes, and reactive oxygen species-related genes were also altered (Fig. 3B). These indicate that grafting changed these gene expressions throughout the wound-healing process.
3.4. Graft-induced changes in gene expression with epigenetic modifications do not change similarly in response to pre-drought stress To study whether gene expression differs between grafted (drought tolerance) and pre-drought stress treatment (drought susceptible), RT-PCR analyses were performed on a subset of expression-altered genes with an epigenetic modification in the grafted samples (selected from Table 1). An ABAinsensitive 5-like protein 5 or ABI5 (Solyc01g104650) and two heat shock protein (HSP) genes, LeHSP110/ClpB (Solyc02g088610) and HSP70 kDa (Solyc07g043560), were hypermethylated with H3K4me3 and were significantly upregulated in grafted samples under D0 conditions (Fig. 4A-C). In contrast, pre-drought stress treatment was not induced or slightly reduced compared to control. The germin-like protein (GLP) (Solyc04g041720), an H3K27me3 hypermethylated gene, showed a significant reduction in gene expression in the pre-drought-treated samples at D0 (Fig.  4D). During D3 drought stress, grafted samples showed less reduction in gene expression. GLPs are involved in various abiotic and biotic stress responses in plants. 29 A similar pattern was observed in the H/ACA ribonucleoprotein complex non-core subunit NAF1 (Solyc07g054460), which increased 11 DMPs in the grafted samples (Fig. 4E). NAF1 is essential for several RNA processing and ribosome biogenesis. [30][31][32] Several genes showed more than one epigenetic modification, such as ABA PYL10 (Solyc01g095700) and ERD or earlyresponsive to dehydration stress (Solyc07g048110), were modified through hyper and hypomethylation of DNA ( Fig.  4F and G). Both genes showed significantly increased gene expression in grafted samples under D0 conditions. These results clearly show that the expression of the molecular chaperon HSPs, drought-responsive genes like ABI5, PYL10, and ERD, and growth-critical genes, GLPs and NAF1 is altered by grafting but not our pre-drought stress treatments. These differences would be directly related to the acquisition of drought tolerance in self-grafted tomatoes.

Discussion
A previous study found that DNA methylation changes by grafting were related to rootstock-scion interactions in heterografting tomato plants. 33 In addition, there are increasing reports of similar DNA methylation and non-coding RNA changes by grafting in some hetero-grafting plants. 34 We here found that even self-grafted tomato plants that do not use vigorous rootstocks increase drought tolerance (Fig. 2D). In addition to changes in gene expression, epigenetic changes due to H3K4me3, H3K27me3, and DNA methylation occur in at least a total of 20% of genes of grafted plants ( Fig. 3A and C). DNA methylation in plant genomes shows the highest in the CG context but also exists in non-CG methylation, where is highly stable between tissues. 35,36 In tomato leaf, the percent of CG, CHG, and CHH is 84.1, 54.8, and 8.4, respectively. 36 In our analysis, we englobe the methylations type in control and grafted, only comparing the DMGs in any context due to the difference of Momotaro and reference tomato background.
Selected histone modifications have been reported to play a role in drought stress memory in several genes in Arabidopsis thaliana. 37,38 DNA methylation has also been reported to function in drought adaptation of the rice genome. 39,40 We observed that the H3K4me3, H3K27me3, and DNA methylation DMGs belong to several pathways (Fig. 1D). The GO analysis of these DMGs includes 'response to stimulus', 'cell cycle', and 'regulation of hormone levels'. Several phytohormones, such as abscisic acid, ethylene, auxin, cytokinin, and gibberellin, have been essential for stress response and graft union. 41 Many of these hormone-related genes were found in DMGs and DEGs in the grafted plants in this study (Supplementary  Tables S2-S4, S6, and S7). Abscisic acid is not required for the graft reconnection process, and its activation occurs in early response to wounding. [42][43][44] During water deficit stress, ABA is important for plant adaptation through the induction of stomatal closure as a measurement of internal water control. 45 A study found a correlation between histone modification and the ABA pathway, affecting the gene expression of dehydration-responsive genes. 46 In our data, we observed that genes related to the ABA pathway, ABI5 and PYL10, and ERD were activated by grafting (Fig. 4A, F, and G) under normal conditions. Those genes may be induced by H3K4me3 and DNA hypomethylation. ABI5 is a bZIP transcription factor that functions in the centre of the ABA signalling where the overexpression of Arabidopsis orthologue ABI5 increases sensitivity to ABA and sugar levels. 47,48 The overexpression of ABA PYL9 in Arabidopsis (orthologue with PYL10) has been demonstrated to confer drought resistance; however, it also induces leaf senescence. 49 Auxin and cytokinin are essential for graft reconnection. 9,50 In addition, gibberellin is activated during grafting and plays a role in expanding the cortex cells of the vascular tissue formed during the graft-healing process. 42 Like the ERF family, several transcription factors, such as apetala 2/ethylene response factor (AP2/ERF) family, play an essential role in growth and development, stress tolerance, and hormonal signal transduction. 51,52 These findings suggest that hormone signalling contributes to stress tolerance through the epigenetic modification caused by grafting. While several of these hormones have an essential role in growth and development, they can also aid in coordinating the plant to the stress response. It is well known that hormones and epigenetic regulation have a crosstalk mechanism where the hormone can directly affect the epigenetic modifier activity, or the epigenetic modification targets genes related to the hormone pathway. 53 In addition, the apical meristem reprograms stem cells to differentiate into new tissues. 54,55 Some of the expressed genes whose effect was less affected by grafting during D3 belonged to the GO term 'cell cycle'. The inhibition of the cell cycle reduces plant growth, which is a stress response mechanism. 56,57 Cell cycle regulation involves cell division and expansion regulated by chromatin modifiers. 58,59 Several DMGs classified as 'cell cycle' were marked by H3K4me3 and DNA hypomethylation. Cell cycle-related genes are important during the wound and regeneration process in the grafting area. Several studies have demonstrated that cellular reprogramming is regulated by several histone modifications and DNA methylation, causing gene expression changes after wounding. 44,60 Grafting is assumed to cause priming drought stress due to the lack of water supply caused by xylem disconnection. On the other hand, 4-and 7-day pre-drought stress treatments in this study were expected to have similar drought tolerance effects, but the results were the opposite and became more sensitive. It may be due to different root damage between grafting and pre-drought treatment. In Arabidopsis, drought stress has been reported to cause epigenetic changes within 5 h, 37 and multiple air-dryings for 2 h also acquire a memory of drought stress with epigenetic modifications. 46 In contrast, we found here that even self-grafting is a simple and reproducible procedure for acquiring stress memory with epigenetic modifications. Recently, similar gene expression changes were found in a homo-grafted tomato with vigorous rootstock. 7 However, self-grafting through wounding could have strengthened the root system. One mechanism of wounding or mechanical stress that induces changes in plants is thigmomorphogenesis, where the signalling pathways of phytohormones are implicated and thus may trigger potential responses to stress. 61 Combined with this study, there are at least two grafting effects, stress memory due to the wound-healing process and stress tolerance acquired by vigorous rootstock combination. We are now attempting to consider other epigenetic mechanisms like small RNAs and histone modifications such as acetylation involved in grafting and stress memory. In conclusion, self-grafting generated several changes in the epigenomics of the tomato scion, where some DMG influenced some DEGs. In addition, we observed that H3K4me3, H3K27me3, and Table 1. Classification of differentially methylated genes by H3K4me3, H3K27me3, and cytosine-5 DNA methylation in grafted and gene expression on control (Co) and grafted (Gr) on before (D0) and during drought stress (D3) treatment. Histone modification presented as fold enrichment (P < 0.001) and DNA methylation as differentially methylated position (DMPs > |1|  . Asterisks show significant differences between the control and each treated plant according to the t-test (Mann-Whitney test), where * denotes P < 0.05, ** P < 0.01, *** P < 0.001, and P < 0.0001. The error bard displays the three biological replicates' standard error.
DNA methylation DMGs in several pathways could significantly regulate drought stress tolerance.