Global epigenetic and metabolic changes accompany the alterations in fruit size and shape of Cucurbita pepo L. intraspecies

Transcription

1 Global epigenetic and metabolic changes accompany the alterations in fruit size and shape of Cucurbita pepo L. intraspecies grafting Aliki Xanthopoulou 1, Aphrodite Tsaballa 2, Ioannis Ganopoulos 3, Aliki Kapazoglou 1, Evangellia Avramidou 4, Filippos Aravanopoulos 4, Theodoros Moysiadis 1, Maslin Osathanunkul 5, Athanasios Tsaftaris 6, Andreas Doulis 7, Apostolos Kalivas 3, Eirini Sarrou 3, Stefan Martens 8, Irini Nianiou- Obeidat 9, Panagiotis Madesis Corresp Institute of Applied Biosciences, Institute of Applied Biosciences/CERTH, Thermi Thessaloniki, Greece 2 School of Biological Sciences, University of East Anglia, Norwich, UK., School of Biological Sciences, University of East Anglia, Norwich, UK., Norwich, United Kingdom 3 Hellenic Agricultural Organization DEMETER (ex NAGREF), Institute of Plant Breeding and Genetic Resources, Thermi, Thessaloniki, Greece 4 Faculty of Forestry and Natural Environment, Laboratory of Forest Genetics and Tree Breeding, Aristotle University of Thessaloniki, Thessaloniki, Greece 5 Chiang Mai University, Department of Biology, Faculty of Science,, Chiang Mai, Thailand 6 American Farm School, Perrotis College, Thessaloniki, Greece 7 Hellenic Agricultural Organization DEMETER (ex NAGREF),, Institute of Olive Tree, Subtropical Crops and Viticulture, Heraklion, Greece 8 IASMA Research and Innovation Centre, Fondazione Edmund Mach (FEM),, Department of Food Quality and Nutrition Department, San Michele all' Adige, Italy 9 Department of Genetics and Plant Breeding, Aristotle University of Thessaloniki, Thessaloniki, Greece 10 Institute of Applied Biosciences, Institute of Applied Biosciences/CERTH Corresponding Author: Panagiotis Madesis address: pmadesis@certh.gr To further understand the impact of grafting on fruit characteristics and to comprehend the mechanisms involved in graft-induced changes we studied homo- and hetero- grafted Cucurbita pepo cultivars (cv.) that vary in fruit size and shape. C. pepo cv. Munchkin (small fruit) and cv. Big Moose (large fruit) as well as cv. Round green (round fruit) and cv. Princess (elongated fruit) were homografted and reciprocally heterografted. The results show significant changes in fruit size when Big Moose was grafted onto Munchkin rootstocks in comparison to homo-grafted controls. Moderate changes in fruit shape were observed when grafting of cv. Round green and cv. Princess were performed. This is the first report of such phenotypic changes after intra-species/inter-cultivar grafting in Cucurbitaceae. Additionally, we found significant changes in i) secondary metabolite profile, ii) global DNA methylation pattern and iii) mirna expression patterns in grafted scions and iv) DNA methylation on graft-induced phenotypic changes in grafted plants. Our results contribute to further understanding graft-induced effects on fruit morphology in intra-species grafting. Furthermore, our results pave the way for understanding the role of phenolic metabolites and epigenetic molecular mechanisms on the phenotypic changes recorded.

2 1 Global epigenetic and metabolic changes accompany the alterations in fruit size and shape 2 of Cucurbita pepo L. intra-species grafting 3 Aliki Xanthopoulou 1,2, Aphrodite Tsaballa 3, Ioannis Ganopoulos 4,, Aliki Kapazoglou 2, Evangelia 4 Avramidou 5, Filippos A. Aravanopoulos 5, Theodoros Moysiadis 2, Maslin Osathanunkul 9, 5 Athanasios Tsaftaris 1,2,8 Andreas G. Doulis 6, Apostolos Kalivas 4, Eirini Sarrou 4, Stefan Martens 7, 6 Irini Nianiou- Obeidat 1*, Panagiotis Madesis 2* 7 1 Aristotle University of Thessaloniki, Department of Genetics and Plant Breeding, Thessaloniki, 8 GR-54124, Greece 9 2 Institute of Applied Biosciences (INAB), CERTH, Thermi-Thessaloniki, GR-57001, Greece 10 3 School of Biological Sciences, University of East Anglia, Norwich, UK Institute of Plant Breeding and Genetic Resources, Hellenic Agricultural Organization 12 DEMETER (ex NAGREF), Thermi, Macedonia, GR-57001, Greece 13 5 Aristotle University of Thessaloniki, Faculty of Forestry and Natural Environment, Laboratory 14 of Forest Genetics and Tree Breeding, GR-54124, Greece 15 6 Hellenic Agricultural Organization DEMETER (ex NAGREF), Institute of Olive Tree, 16 Subtropical Crops and Viticulture, Heraklion, Crete, GR-71003, Greece Department of Food Quality and Nutrition Department, IASMA Research and Innovation 18 Centre, Fondazione Edmund Mach (FEM), Via E. Mach 1, San Michele all' Adige, (TN), 19 Italy 20 8 Perrotis College, American Farm School. Thessaloniki, GR Greece 21 9 Department of Biology, Faculty of Science, Chiang Mai University, Chiang Mai 50200, 22 Thailand Corresponding authors: pmadesis@certh.gr and nianiou@agro.auth.gr 25

3 27 Abstract 28 To further understand the impact of grafting on fruit characteristics and to comprehend the 29 mechanisms involved in graft-induced changes we studied homo- and hetero- grafted Cucurbita 30 pepo cultivars (cv.) that vary in fruit size and shape. C. pepo cv. Munchkin (small fruit) and cv. 31 Big Moose (large fruit) as well as cv. Round green (round fruit) and cv. Princess (elongated 32 fruit) were homografted and reciprocally heterografted. The results show significant changes in 33 fruit size when Big Moose was grafted onto Munchkin rootstocks in comparison to homo- 34 grafted controls. Moderate changes in fruit shape were observed when grafting of cv. Round 35 green and cv. Princess were performed. This is the first report of such phenotypic changes 36 after intra-species/inter-cultivar grafting in Cucurbitaceae. Additionally, we found significant 37 changes in i) secondary metabolite profile, ii) global DNA methylation pattern and iii) mirna 38 expression patterns in grafted scions and iv) DNA methylation on graft-induced phenotypic 39 changes in grafted plants. Our results contribute to further understanding graft-induced effects on 40 fruit morphology in intra-species grafting. Furthermore, our results pave the way for 41 understanding the role of phenolic metabolites and epigenetic molecular mechanisms on the 42 phenotypic changes recorded Keywords: Cucurbitaceae, MSAP methylation sensitive amplified polymorphisms markers, 45 epigenetic, phenolics, mirnas, fruit morphology

4 50 Introduction 51 Grafting is an important cultivation technique in agriculture, employed widely to improve 52 performance of important horticultural crops such as woody fruit trees, grapes and lately on 53 vegetable crops (Goldschmidt 2014; Mudge et al. 2009). It is an ancient technique used for over 54 2,000 years that was first applied in woody species for improving yield and disease resistance 55 (Kubota et al. 2008). Vegetable grafting which originated in the early 20 th century, involves 56 mainly Cucurbitaceae and Solanaceae species and has been utilized extensively for enhancing 57 fruit quality, resistance to diseases and tolerance to abiotic stresses (Goldschmidt 2014; Mudge et 58 al. 2009). Grafting was originally employed to avoid the detrimental effects of soil borne 59 pathogens infecting the roots of susceptible plants through the use of resistant rootstocks grafted 60 onto the plant of interest (King et al. 2008; Louws et al. 2010). Selected beneficial rootstocks can 61 provide resistance to bacteria, nematodes, fungi, as well as enhanced tolerance to abiotic stresses 62 such as extreme temperatures, increased salinity, increased calcium and heavy metals in the soil 63 (Colla et al. 2010; King et al. 2008). 64 Whereas horticulturists were trying to improve crops through grafting in order to provide 65 higher biomass performance and resistance in plants diseases and biotic stress, the quality 66 (nutritional/organoleptic properties) of product didn t attract much attention. Nowadays such 67 kind of improvement strategies seem to focus also on some specific compounds contributing to 68 the nutritional value of vegetables, such as polyphenols, carotenoids, volatiles etc. Polyphenols, 69 as products of secondary metabolism, have received much attention the last decades, due to their 70 antioxidant (Valenzuela et al. 2014), antimicrobial (Nascimento et al. 2014), antiviral (Katayama 71 et al. 2013) bioactivities among others. In the past decade, a great number of studies regarding 72 Cucurbitaceae grafting aimed at improving fruit quality and plant vigour of economically 73 important vegetables such as cucumber, melon and watermelon (Cohen et al. 2007). Grafting of 74 watermelon onto selected rootstock genotypes resulted in significant increase of lycopene 75 content in watermelon fruits in addition to enhanced resistance to soil-borne diseases (Mohamed 76 et al. 2012), whereas different rootstock-scion combinations led to alterations in the content of 77 plant nutrients in leaves and fruits of watermelon grafts (Yetisir et al. 2013). Moreover, 78 cucumber and melon grafted onto specific pumpkin rootstocks led to altered organoleptic 79 properties in addition to enhanced tolerance to high salinity (Huang et al. 2013; Orsini et al ; Rouphael et al. 2012). Numerous complex processes occur during grafting, including

5 81 formation of the graft union and the common vasculature, transport of information from 82 rootstock to scion and vice versa, as well as the differing effects of scion and rootstock genotype 83 combinations. Thus, it is reasonable to assume multiple molecular mechanisms operating 84 throughout the growth of a grafted plant. These mechanisms include transport of important 85 information molecules such as small RNAs as well as epigenetic effects, such as methylation of 86 DNA, both of which can, in turn, affect gene expression. Trafficking of macromolecules such as 87 proteins and of genetic material or hormones and nutrients, have been demonstrated in many 88 instances between grafting partners (Albacete et al. 2015). Regarding movement of genetic 89 material, an increasing number of reports implicate long distance transport of RNA molecules 90 such as messenger and small RNAs. Movement of RNA molecules over long distances, across 91 plant organs and through the phloem has been well documented (Kalantidis et al. 2008; Kehr & 92 Buhtz 2008; Lough & Lucas 2006; Lucas et al. 2001; Mermigka et al. 2015; Omid et al. 2007; 93 Spiegelman et al. 2013). Similarly, transport of small non-coding RNAs (sirnas and mirnas) 94 across graft segments of homo-grafted and hetero-grafted plants has been demonstrated in a 95 number of studies in Arabidopsis, Solanaceae and Cucurbitaceae (Haroldsen et al. 2012; Melnyk 96 et al. 2011; Palauqui et al. 1997).. In many occasions mobile RNA molecules have a 97 transcriptional or post-transcriptional gene silencing (Haroldsen et al. 2012; Melnyk et al. 2011; 98 Palauqui et al. 1997). It is known that mirnas play crucial roles in all plant biological 99 processes, including growth, vegetative and reproductive development, and response to abiotic 100 and biotic stresses (Barrera-Figueroa et al. 2013; Li et al. 2016; Zhang 2015). MiRNAs are post- 101 transcriptional gene regulators that target mrnas and either cleave the transcript or inhibit 102 translation (Dong et al. 2013). They target a wide range of transcription factors related to 103 developmental processes as well as genes that code for enzymes involved in metabolic pathways. 104 Several evolutionary conserved families of mirnas such as the mir156, mir159, mir165/166, 105 mirr171, mir172, control various aspects of development in plants. MiR156 has been 106 associated with changes in leaf biomass as its overexpression led to significant increase in total 107 leaf number and subsequent plant biomass in Arabidopsis (Schwab et al. 2005). 108 Global DNA methylation studies, using either methylation sensitive amplified 109 polymorphisms (MSAP) or locus-specific bisulfite sequencing analyses revealed significant 110 alterations in the DNA methylation pattern of grafted plants after reciprocal inter-species grafting 111 among three Solanaceae species (Wu et al. 2013). In particular, the global DNA methylation

6 112 pattern of tomato, eggplant and pepper scions was shown to be extensively altered in two DNA 113 methylation contexts (CG and CHG) upon grafting. Importantly, the graft-induced methylation 114 changes were inherited in the self-pollinated progeny, indicating that grafting can induce 115 epigenetic changes that can be stably transferred to subsequent generations (Wu et al. 2013). In 116 agreement with the study above, our group has recently demonstrated altered DNA methylation 117 patterns upon inter-species hetero-grafting of three Cucurbitaceae species 33. In particular, DNA 118 methylation levels were significantly increased in melon and cucumber scions grafted onto 119 pumpkin rootstocks (Avramidou et al. 2014). 120 Fruit morphology, which includes fruit shape and size, is another important trait of 121 agronomic importance in vegetable species of the Solanaceae and Cucurbitaceae families. 122 However, although studies during the past decades have focused on fruit yield and quality 123 (texture, aroma, flavor, nutrients), resistance to diseases and tolerance to abiotic stresses, little is 124 known about the effect of grafting on fruit morphology and the underlying molecular 125 mechanisms governing this process. A few reports have described an effect on fruit shape upon 126 intra-species grafting in pepper (Capsicum annuum) (Taller et al. 1999; Tsaballa et al. 2012; 127 Yagishita & Hirata 1987). Tsaballa et al. (2012) 35 have demonstrated remarkable effects on fruit 128 shape of the scion partner in a rootstock scion combination of two different pepper genotypes 129 with different fruit shape. Specifically, grafting of the round-fruit shaped cv. Round, onto a 130 long-fruit shaped cv. Long, resulted in substantially elongated fruits in the scion cv. Round 131 plants (Tsaballa et al. 2012). However, there are no reports, to date, concerning the effects of 132 grafting on fruit morphology in Cucurbitaceae. In order to further investigate grafting in 133 Cucurbitaceae and the effect of rootstock-scion combinations in fruit size and shape we have 134 applied reciprocal hetero-graftings of two different C. pepo cvs., Big Moose and Munchkin, 135 with different fruit sizes as well as two other C. pepo cvs., Princess and Round green, with 136 different fruit shapes. In addition, we have assessed changes in DNA methylation patterns 137 employing comparative MSAP methodology which allows for the documentation of epigenetic 138 alterations between different crosses and between plants bearing fruits of different sizes and 139 morphology. Further, we have analyzed the expression of several mirnas in order to probe 140 variations of these epigenetic modifiers linked to the grafting process and possibly with grafting- 141 induced fruit morphological changes. Additionally, we determined concentrations of a series of 142 selected secondary metabolites and compared thus produced metabolic profiles between different

7 143 grafting combinations and non-grafted controls. We report here graft-induced changes in fruit 144 size and moderate changes in fruit shape of C. pepo scions accompanied by significant changes 145 in scion genomic DNA methylation, patterns, abundance of specific mirna molecules and 146 secondary metabolites as compared to non-grafted control tissues Materials and Methods 149 Plant Material 150 Two independent grafting and MSAP experiments were conducted during two 151 consecutive summer periods (2014 and 2015). Samples from 2014 were used for MSAP analysis 152 in relation to fruit size, whereas samples from 2015 were used for MSAP analysis in relation to 153 fruit shape and metabolite profile. 154 Four Cucurbita pepo cultivars were used for grafting; (1) cv. Round green (round fruit 155 shape) (2) cv. Princess (long fruit shape) (3) cv. Big Moose (large fruit size) and (4) cv. 156 Munchkin (small fruit size). Seeds were initially sown in small pots that were kept in the 157 greenhouse at 25 C. Grafting was carried out by the time plants had 2 3 true leaves and a shoot 158 diameter of about mm. In total, four grafting combinations were produced representing 159 two reciprocal cv. combinations as following: (1) cv. Round green on cv. Princess, (2) cv. 160 Princess on cv. Round green, (3) cv. Big Moose on cv. Munchkin and (4) cv. Munchkin 161 on cv. Big Moose. Grafting was performed following the splice-grafting method employing 162 grafting clips (for a review of grafting methods see (Lee et al. 2010)). A total of 4 plants from 163 each grafting combination i.e. rootstock (RS) x scion (SC) were used for scoring for either fruit 164 size or fruit shape as well as for MSAP analysis. Controls included were (1) the homo-grafted 165 plants for example cv. Round green grafted onto cv. Round green for all four cvs. and (2) the 166 non-grafted s that served as reference state in MSAP comparisons as well (see below). Following 167 grafting, all plants were transferred in the growth chamber under constant temperature of 25 C, 168 RH 85 95%, light flux 2,000 lux, photoperiod of 16 h light/8 h dark and kept for 2 weeks. 169 Successfully grafted plants were transferred in the greenhouse for an acclimation period of one 170 month, transferred to 7 L pots filled with calcareous clay soil afterwards and allowed to grow in 171 the open while watered regularly with nutrient solution. The same grafting design was used both 172 years. 173

8 174 DNA extraction and ISSR molecular analysis 175 DNA extraction from leaves of controls and scions, from plants presenting statistically 176 different phenotypes were used, was performed using NucleoSpin Plant II Kit (Macherey Nagel, 177 Duren, Germany), according to manufacturer's instructions. 178 For the detection of putative intra-cultivar variation, we selected 6 genotypes of each 179 grafting cultivar combinations (72 individuals in total) and we used a set of six Inter Simple 180 Sequence Repeats (ISSR) primers as described previously by Xanthopoulou et al. (2015). PCR 181 reactions and gel electrophoresis were composed according to the authors above Methylation Sensitive Amplification Polymorphism (MSAP) analysis 184 For each sample, MSAP analyses were performed using both EcoRI/HpaII and 185 EcoRI/MspI digests. Digestion of 200-ng aliquots of genomic DNA with either of the two 186 isoschizomers (HpaII, MspI) was carried out in 20 μl containing 1Χ one for all Buffer, 4 U 187 EcoRI (New England Biolabs, Ipswich, MA, USA) and 3 U of either HpaII or MspI enzyme 188 (New England Biolabs, Ipswich, MA, USA) for 3 h at 37 C. Two different adapters, designed to 189 avoid reconstruction of restriction sites, one for the EcoRI sticky end and one for the HpaII/MspI 190 sticky end, were ligated to the DNA after digestion, by adding to each final digestion 5 μl of a 191 mix containing 5 pmol of EcoRI adapter, 50 pmol of HpaII/MspI adapter, 1 mm ATP, 1Χ one 192 for all Buffer, and 1 U of T4 DNA ligase (Invitrogen, CA, USA). The ligation was incubated for h at 25 C. The EcoRI adapter consisted of the combination of two primers: 5Ά- 194 CTCGTAGACTGCGTACC-3Ά and 3Ά-CTGACGCATGGTTAA-5Ά. EcoRI and HpaII/MspI 195 adapters are presented in Table S1. Digested and ligated DNA fragments were diluted 5-fold and 196 used as templates for the pre-amplification reaction. Two different pre-amplification reactions 197 were performed using EcoRI/MspI-A and EcoRI/HpaII-A primers in a total volume of 20 μl 198 containing 1X Kapa Taq Buffer, 0.2mM of each dntp, 2.5 mm MgCl 2, 30 ng of each primer 199 EcoRI+A, HpaII/MspI+A, 1U Taq DNA polymerase (Kapa Biosystems, Woburn, MA) and 5 μl 200 of diluted fragments. PCR amplifications were carried out in a BioRad thermocycler for cycles of 30 s at 94 C, 30 sec at 56 C, and 1 min at 72 C. Pre-amplified fragments were diluted fold and used as templates for the selective amplification. Οne labelled EcoRI and one 203 HpaII/MspI primer, with the same sequences as those used for the pre-amplification but with two 204 and three selective nucleotides respectively at the 3 end, were used in each analysis. PCR was

9 205 performed in a total volume of 10 μl including 1X Kapa Taq Buffer 2.5 mm MgCl 2, 0.08 mm of 206 each dntp, 5 ng of labeledecori primer, 30 ng of HpaII/MspI primer, 1 U of Taq DNA 207 polymerase (Kapa Biosystems), and 3 μl of diluted pre-amplified DNA. The selective 208 amplification was carried out using classical AFLP cycling parameters (Vos et al. 1995), 209 employing eight different primer combinations (Table S1) MSAP data collection and processing 212 MSAP product mixtures were denatured in formamide at 94 C for 2 min and 213 electrophoretically fractionated on an ABI Prism 3730xl (Applied Biosystems, Carlsbad, CA, 214 USA). In order to reduce the impact of potential size homoplasy only fragments ranging from to 500 bases in size were counted (Vekemans et al. 2002). Fragment size was determined 216 using Genemapper v4.0 (Applied Biosystems, Carlsbad, CA, USA) employing an internal 217 standard (GS 500 LIZ, Applied Biosystems) and an initial data matrix was produced with allele 218 presence scored as 1 and absence as Our scoring initially entailed determining marker presence / absence between a seedling 220 (reference state) and grafted plants of the same cv. even if a single band in one of the two 221 isoschizomer profiles of a single individual was detected. Scions from homo- or hetero- 222 grafted plants were scored independently and were considered as two different treatment states. 223 The same individual control seedlings were considered as reference states in comparisons 224 involving scions of the same cv. irrespective of rootstock cv. (i.e. irrespective of homo- or 225 hetero- treatment states). Each methylation state was described with two binary digits (duplets) 226 corresponding to the two isoschizomer profiles. Quadruplets (a series of four binary digits) were 227 subsequently, produced by pooling together duplets from the reference and from the treatment 228 state (two different treatment states in total; homo- and hetero- ) and were assigned to state 229 change events between reference and treatment states. Sixteen alternate quadruplets could thus 230 be produced each grouped according to (Haoa et al. 2004) as de novo methylation (three events), 231 demethylation (three events), other (ambiguous or too complex pattern to be explained; four 232 events) or no change (six events; not specifically enumerated) (Haoa et al. 2004). All data 233 management steps, from initial matrix formation till event change classification and tallying 234 were performed employing an in-house script. Only polymorphic fragments (more than 5% 235 present or absent) were retained for subsequent analysis (MSAP polymorphic loci). Once the

10 236 quadruplets of each sample (grafting combination) have been derived (by comparison to 237 respective non-grafted controls) they were categorized according to deduced methylation change 238 event (de-methylation, de novo methylation, no change, other-non classifiable). The number of 239 each one of the four change events -per sample- was converted to percentages against the 240 number of total events of each sample separately. Four different percentages are produced per 241 each sample corresponding to the four different change events. Comparisons and statistical 242 analyses of such percentage across samples could thus be made (Figures 3 and 4). Change event 243 other is not portrayed in Figures 3 and 4). Subsequently, statistically significant differences 244 between the two treatment states i.e. homo- vs. hetero- grafted plants were assessed for the 245 MSAP polymorphic loci by means of a t-test employing SPSS (ver. 22). Separate tests were 246 performed for each cv. serving as rootstock (four independent experiments / statistical analyses 247 in total) and the two combinations employing the self cv. as scion mirna expression profile via stem-loop RT-PCR 250 Total RNA was extracted from leaves (from plants presenting statistically different 251 phenotypes) using the TRIzol Reagent (Invitrogen, USA) and DNase I was then used to 252 remove DNA. The integrity and purity of the total RNA were evaluated using 2% denaturing gel 253 electrophoresis and NanoDrop DU8000 spectrophotometry (A260/A280 and A260/A230). The 254 expression profiles of five mirnas were assayed by stem loop reverse transcription-pcr (RT 255 PCR). A 200 ng aliquot of total RNA was used for the initiation of the reverse transcription 256 reaction. The stem loop reverse transcription primers were designed following the method 257 described by (Varkonyi-Gasic et al. 2007). The reverse transcription product is amplified using a 258 mirna-specific forward primer and a universal reverse primer. The stem loop reverse 259 transcription reactions were performed by Superscript III reverse transcriptase (Life 260 Technologies, Carlsbad, CA, USA) transcriptase according to the supplier's manual. PCR 261 primers were then added to perform the PCR. Relative quantification was performed using 262 Elongation Factor-1 (EF-1) as the endogenous control reference gene (Obrero et al. 2011). The 263 data were analyzed using the REST software (Pfaffl et al. 2002) Phenolic metabolite analysis 266 Fresh collected samples (each sample obtained from pooling fruit material from three

11 267 individual plants) were freeze dried (Freeze-dryer Alpha 1-2 LD plus, Christ, Germany; at C), and then pulverized to fine powder. The extraction was performed by mixing 100 mg of 269 freeze-dried fruit sample with 5 ml 80% methanol into 15 ml falcon tube. The samples and 270 solvent were mixed by orbital shaker for 3 h at room temperature and the extraction proceeded 271 overnight at 4 C in the dark. The resulting solutions were filtered on a 0.22 µm PFTE membrane 272 into a glass vial and analyzed as described below. Three replicates for each sample were done. 273 The analysis of phenolic compounds was performed using the method described previously 274 by (Vrhovsek et al. 2012). Samples were directly injected after extraction. Targeted Ultra 275 Performance Liquid Chromatography was performed on a Waters Acquity UPLC system 276 (Milford, MA, USA) consisting of a binary pump, an online vacuum degasser, an autosampler, 277 and a column compartment. Separation of the phenolic compounds was achieved on a Waters 278 Acquity HSS T3 column 1.8 μm, 100 mm 2.1 mm (Milford), kept at 40 C. 279 Mass spectrometry detection was performed on a Waters Xevo TQMS (Milford) 280 instrument equipped with an electrospray (ESI) source. Data processing was performed using the 281 Mass Lynx Target Lynx Application Manager (Waters) Results Graft-induced changes in fruit size and shape 286 Two C. pepo cultivars differing in fruit size, cv. Munchkin, a small fruited cultivar, and 287 cv. Big Moose, a large fruited cultivar (Figure 1A) and two cultivars differing in fruit shape, 288 cv. Round green, a round fruited cultivar and cv. Princess, an elongated fruited cultivar 289 (Figure 2A), were used to assess possible graft-induced alterations in fruit morphology in 290 reciprocal hetero-grafting combinations. Fruit morphology of grafted plants was assessed using 291 morphological traits according to UPOV descriptors such as weight, polar diameter and 292 equatorial diameter for the scion fruits from each grafting combination as well as from control 293 (non-grafted). The majority of the fruits produced by the Munchkin and Big Moose grafting 294 combinations although they had the expected phenotype i.e. they were similar to the fruit 295 phenotype of the non-grafted scion in every case (Figure 1B), they presented statistically 296 significant phenotypic differences with the fruits from the non-grafted or homografted plants in 297 terms of weigh, equatorial and polar diameter. In addition a single grafted Munchkin (RS) x

12 298 Big Moose (SC) plant (termed with altered fruit size ; Table 1) produced fruits with 299 pronounced morphological differences when compared to the non-grafted Big Moose plants 300 (Figure 1C). A Distinct reduction in fruit weight and fruit size in this plant could be observed 301 when compared to the rest of the fruits produced by this grafting combination (Figure 1D). 302 Further, the reduction in fruit weight was approximately 2.6-fold compared to Big Moose non- 303 grafted control, more than 3-fold compared to Big Moose homografts and 2.4-fold compared to 304 the majority of the Munchkin (RS) x Big Moose (SC). In addition, a significant decrease in 305 both equatorial and polar fruit diameter of about 2-fold was evidenced in this heterograft as 306 compared to the control non-grafted Big Moose and Munchkin (RS) x Big Moose (SC) 307 plants (Table 1) On the other hand the reciprocal combination, Big Moose (RS) x Munchkin (SC) did 310 not produce any discrete evident morphological changes (i.e. in fruit size) like the one before 311 (Figure 1B-bottom) (Table 1). However, moderate changes statistically significant were observed 312 in fruit shape in the reciprocal combinations of Round green and Princess. Princess (RS) x 313 Round green (SC) resulted in less round fruits and the reverse combination, Round green 314 (RS) x Princess (SC) led to fruits that were smaller as compared to the control seedlings 315 (Figure 2B). Grafting combinations of Princess and Round green did not produce extreme 316 differences in terms of morphological traits yet there were statistically significant especially for 317 equatorial and polar diameter when compared to non-grafted or homografted plants (Table 1). 318 To exclude the possibility that fruit morphology changes rise from genetic variation 319 within cultivars or cross-pollination between them since C. pepo is an openly pollinated species, 320 molecular analysis with six polymorphic Inter simple sequence repeat (ISSR) markers was 321 performed according to (Xanthopoulou et al. 2015). No intra-cultivar variation was found 322 among the 72 examined plants suggesting that epigenetic (rather than genetic) changes 323 associated with grafting should be sought as the underlying mechanism inducing morphological 324 alterations Graft-induced changes in scion global DNA methylation pattern 327 The effect of grafting on the pattern of scion DNA methylation pattern was assessed 328 utilizing a comparative MSAP analysis (Supplementary Figure 1). Initially, the MSAP method

13 329 can detect the DNA methylation status of 5 -CCGG-3 sites, employing two pairs of 330 isoschizomers EcoRI/HpaIIand EcoRI/MspI that have different sensitivity to methylation at the 331 inner or outer cytosines. For each sample the DNA methylation pattern at 5 -CCGG-3 sites is 332 determined by scoring the presence or absence of marker bands of EcoRI/HpaII and EcoRI /MspI 333 digestion of genomic DNA followed by PCR amplification (Supplementary Figure 1). 334 Differences in PCR products obtained from two different samples reflect different methylation 335 states at the cytosines restriction site. Per sample, methylation patterns are deduced by 336 determining amplification differences between the two pairs of isoschizomers while methylation 337 pattern changes are determined following comparisons between treatments (different grafting 338 combinations) and controls (non-grafted seedlings). We consider the case with four different 339 methylation change events (1. de novo methylation, 2. demethylation, 3. no change, 4. other 340 variation). 341 As a first step, methylation change (MSAP) data were averaged and plot by event and by 342 grafting combination (Figures 3 and 4). In the Munchkin (RS) x Big Moose (SC) combination, 343 which resulted in the significant fruit size reduction, an increase of approximately 5-fold in DNA 344 de novo methylation was observed in the scion leaves compared to the control Big Moose non- 345 grafted seedlings (Figure 3A). In addition, an increase of about 2-fold in demethylation was 346 observed in the scions in comparison to the control Big Moose non-grafted, whereas 40% of 347 cytosine sites remained unchanged (Figure 3A). In the Big Moose (RS) x Munchkin (SC) 348 combination an increase in DNA methylation of approximately 2-fold was evidenced in the 349 leaves of the scion plants as compared to the Munchkin control seedlings, whereas a change of 350 approximately 8-fold in DNA demethylation levels was detected (Figure 3B). In the Princess 351 (RS) x Round green (SC) graft which resulted to a flat-shaped-fruit, a remarkable increase in de 352 novo DNA methylation of 20-fold is evidenced in the scions as compared to the control Round 353 green non-grafted (Figure 4A). A marked increase in DNA demethylation of 12-fold was also 354 observed in the Round green scions, as compared to the control Round green (Figure 4A). 355 Likewise, in the Round green (RS) x Princess (SC) graft, which resulted in a smaller fruit 356 compared to the non-grafted seedling, a significant increase in DNA methylation of 357 approximately 16-fold and an increase in DNA demethylation of about 12- fold was observed in 358 Princess scions, relative to the control Princess non-grafted plants (Figure 4B). 359 As a second step, methylation change data underwent a detailed statistical analysis. The

14 360 chi-square test was applied to test the null hypothesis that homografting vs heterografting (four 361 independent comparisons-please see a)-d) below) are homogeneous regarding the methylation 362 event, i.e. the respective theoretical distributions of percentages regarding the categories of the 363 methylation event do not differ statistically significantly. The significance level was set to We considered the case with four different methylation event categories (1. de novo methylation, demethylation, 3. no change, 4. other variation). A significant result would indicate significant 366 overall difference in the methylation categories percentages between homografting and 367 heterografting, and therefore statistically significant difference in how methylation is expressed 368 in these two cases. We performed independently four comparisons, i.e. a) Munchkin homo 369 versus Big Moose x Munchkin, b) Big Moose homo versus Munchkin x Big Moose, c) 370 Princess homo versus Round green x Princess and d) Round green homo versus Princess 371 x Round green. The results were a) , p-value<0.001, b) , p-value<0.001, c) , p-value=0.001 and d) , p-value< Therefore, the null hypothesis of 373 homogeneous methylation behaviour was rejected for all comparisons, i.e. all differences 374 considered were statistically significant. In addition to the above comparisons, for each one of 375 the four cultivars, the goodness of fit test was applied to test the null hypothesis that the 376 theoretical percentages regarding the categories of the methylation event are equal, and by that to 377 actually test if the type of methylation event appears at random within each cultivar. The results 378 were i) , p-value<0.001, ii) , p-value<0.001, iii) , p-value=0.001 and iv) , p-value< Therefore, the null hypothesis of equal methylation categories 380 percentages was rejected for all comparisons, i.e. the type of change did not appear at random. 381 The Cochran s criterion for the validity of all the above chi-square based tests has been checked 382 and was satisfied in all cases. 383 Graft induced changes in mirna expression 384 As depicted in Figure 5A, significant changes were observed in expression levels of a 385 number of mirnas in the Munchkin and Big Moose grafted plants, when compared to the 386 non-grafted controls. In the Munchkin (RS) x Big Moose (SC) that was the combination in 387 the grafted plant with the reduced fruit size, the expression of mir159 and mir164 was 388 significantly decreased by approximately 4-fold, whereas mir171 was decreased by 2-fold and 389 mir166 by 1.6-fold. On the contrary, non-significant expression changes were recorded in the 390 scions of the Big Moose (RS) x Munchkin (SC) heterograft, except for mir171, which was

15 391 induced by about 5-fold and mir164, which was 1.5 times more abundant in comparison to the 392 control (Figure 5A). A significant induction of about 2-fold, for mir171 was also observed in the 393 Round green (RS) x Princess (SC) heterograft, which had a mild effect in fruit shape (Figure 394 5B). A reduction of 1.6-fold was evidenced for mir159 in the Round green (RS) x Princess 395 (SC) heterograft (Figure 5B) Phenolic metabolite analysis 398 Fruit samples from all grafting combinations were subjected to targeted LC-MS/MS 399 analysis in order to identify the effect and possible changes on the secondary metabolite level in 400 different grafting combinations. In total, seventeen compounds were identified in C. pepo fruit 401 samples from the size differing cultivars and their grafting combinations while eighteen 402 compounds were identified in the shape differing cultivars and their grafting combinations. 403 Compounds detected varied in presence and concentration, as shown in the heatmaps produced 404 (Figures 6A and B). In the majority of samples, kaempferol-3-o-rutinoside (Km3rut), rutin and 405 isorhamnetin-3-o-rutinoside (Iso3rut) were the most abundant compounds, whereas in the 406 cultivars differing in fruit size significant amounts were detected for p-hydroxybenzoic (p-hba), 407 cinnamic, ferulic and vanilic acid in scion fruits. In the different cultivars, nine ( Big Moose, 408 Princess ), ten ( Munchkin ) and fourteen ( Round green ) phenolic metabolites were identified 409 (Tables 2 and 3), while several metabolites seem to be influenced by grafting combination. 410 Concerning the fruit cultivars differing in shape, some of the identified metabolites such as 411 Km3rut, quercetin-3-o-glucoside, rutin and Iso3rut, which are present in the control non grafted 412 and the Big Moose homograft as long as in the heterograft Munchkin (RS) x Big Moose 413 (SC), were absent in all the rest of grafting samples. On the other hand, the compounds p-hba, 414 sinapoyl alcohol and arbutin were present only in the fruits obtained from the Munchkin non- 415 grafted plants, its homografts, and the heterografts where Munchkin cultivar was used as the 416 scion. Furthermore, it became apparent in the case of the heterograft Princess (RS) x Round 417 (SC) that there was a stimulation in most of the identified metabolites in the cultivars with green 418 fruit colour, especially in the category of rutinosides. Both homografts of Princess and Round 419 green exhibited a limited level of metabolites Discussion

16 422 In the current study we investigated the effect of C. pepo intra-species grafting on scion s 423 fruit size and shape and its association with global epigenetic and metabolic profiles. The 424 heterograft of two C. pepo cultivars with extremely different fruit size, Munchkin (RS) x Big 425 Moose (SC) displayed remarkable alterations in fruit size as it resulted in a dramatic decrease in 426 fruit mass and a reduction in both equatorial and polar fruit diameter. This is the first time that 427 intra-species grafting in a Cucurbita species results in changes in fruit morphology. However, 428 grafting has been associated with changes in a number of fruit quality characteristics in the past. 429 Pepper is one of the plants where fruit shape changes attributed to grafting were early recorded 430 and actually found to be inherited to the grafting progenies introducing the whole concept of 431 grafting-induced phenotypic variation) (Taller et al. 1999; Tsaballa et al. 2012; Yagishita & 432 Hirata 1987). (Taller et al. 1999) managed to produce pepper varieties with stable but different 433 fruit characteristics such as fruit shape and flavour from the graft-induced variants of an intra- 434 species grafted plant. In our previous work we have found that grafting-induced fruit shape 435 changes were stably inherited to the grafting progenies when a round-fruited pepper cultivar was 436 grafted on a long-fruited pepper cultivar; scion s fruits in this case resembled more the fruit 437 shape of the rootstock than that of the scion (Tsaballa et al. 2012). These findings support the 438 already formed idea of graft hybridization that involves the movement of genetic material from 439 the rootstock to the scion, a process also called graft transformation and results in changes that 440 are heritable to the scion s progenies (Goldschmidt 2014; Mudge et al. 2009). It was recently 441 found that whole nuclear genomes can be transferred between plant cells during grafting creating 442 new allopolyploid species (Fuentes et al. 2014). Our results support the notion that grafting is 443 associated with genetic changes in the phenotype of the grafted plants in a way that it could also 444 be used for the introduction of new characteristics in crops. 445 Apart from genetic changes leading to phenotypic changes, agronomical traits such as 446 yield, plant architecture, fruit morphology, fruit quality, as well as the response to abiotic stress 447 or resistance to diseases can be affected substantially by changes in DNA methylation, histone 448 post-translational modifications and mirnas, all leading to profound changes in gene expression 449 (Dong et al. 2013; Fujimoto et al. 2012; Tsaftaris et al. 2012; Zhong et al. 2013). Likewise, the 450 same traits can be affected by grafting, while increasing evidence has indicated that grafting may 451 induce DNA methylation-pattern alterations. Global DNA methylation patterns were found 452 significantly altered in Solanaceae and Cucurbitaceae inter-species grafting (Avramidou et al.

17 ; Wu et al. 2013). In Solanaceae species grafting, scion genomic loci with altered DNA 454 methylation patterns contained genes associated to diverse cellular functions while a series of 455 genes encoding DNA methylation-related enzymes were found to have altered gene expression 456 patterns (Wu et al. 2013). In addition, a portion of DNA methylation changes were found to be 457 inherited in the self-pollinated progeny (Wu et al. 2013). Changes in DNA methylation patterns 458 were also shown in grafts between Brassica juncea and B. oleraceae and it was suggested that 459 they are possibly linked to the phenotypic variability observed although in these species it was 460 found that the changes can be both heritable and reversible in the subsequent generations (Cao et 461 al. 2016). In our study, we have noticed significant changes in the DNA methylation pattern of 462 the grafted plants. In the Munchkin (RS) x Big Moose (SC) combination which resulted in a 463 dramatic fruit size reduction, de-novo DNA methylation levels were markedly increased, 464 suggesting that graft-associated DNA methylation alterations in certain loci may be responsible 465 for altering fruit shape in the scion. On the other hand, in the reciprocal combination, Big 466 Moose (RS) x Munchkin (SC), a significant increase in DNA demethylation levels was also 467 detected. Although fruit morphology changes were not evidenced in this combination it is 468 possible that increased DNA demethylation has affected other loci associated with fruit quality, 469 and/or environmental responses. Further experiments are needed in order to elucidate the effect 470 of enhanced DNA demethylation in these grafts. Fruit shape changes on grafted plants was also 471 found to be accompanied by alterations in the DNA methylation patterns of scion leaves as 472 shown by the results of Princess (RS) x Round green (SC) and the reciprocal graft 473 methylation patterns. 474 Changes in DNA methylation of the graft partners could result in profound changes in 475 gene expression in both scion and rootstock tissues and consequently affect a multitude of 476 cellular pathways. The latter can be associated with distinct phenotypes including shoot 477 architecture, total yield, fruit morphology and quality, and responses to environmental stress. Our 478 data showed that intra-species Cucurbita heterografting altered fruit morphology. In the past we 479 have reported similar results when intra-species grafting in pepper using two cultivars of 480 different shape, cv. Long (RS) x cv. Round green (SC) resulted in a change in scion s fruit 481 morphology as the fruits had acquired a more elongated shape compared to the control s and the 482 homografted plants, resembling more the fruit phenotype of the rootstock rather than the scion. 483 Taking into account all the above considerations, it is possible that fruit morphology changes in

18 484 inter-species heterografting of Cucurbitaceae could be associated with DNA methylation 485 alterations in scion tissues that in turn affect gene expression programs linked to fruit 486 morphology traits. 487 However, epigenetic changes in gene expression that may result in phenotypic variation 488 are not only produced by changes in DNA methylation but by other epigenetic mechanisms 489 linked to small RNAs. The 21nt small RNA class of mirnas has been shown to play crucial 490 roles in plant growth and development controlling seed, vegetative, flowering, reproductive and 491 developmental processes and the responses to abiotic and biotic stress (Li et al. 2016; Zhang ). In addition, changes in mirna expression have been demonstrated recently to take place 493 during grafting. MiRNA transcriptomic analyses in Cucurbitaceae grafted plants have revealed 494 that groups of conserved mirnas such as mir156, mir159, mir164, mir165/166, mir171, 495 mir172, were either induced or repressed in inter-species hetero-grafted scions implying a 496 regulatory role of mirnas in the grafting process (Li et al. 2013). In our experiments, mir159, 497 mir164, mir166 and mir171 showed significant downregulation in the Munchkin (RS) x Big 498 Moose (SC) heterograft that resulted in the changed, smaller fruit. These results point out that 499 there is a graft-induced mirna controlled epigenetic mechanism that possibly underlies fruit 500 development in grafted scions. MiR156 targets plant specific transcription factors of the 501 SQUAMOSA-promoter binding protein (SBP) type, or SBP-like (SPL) type, as well as 502 APETALA2 (AP2) which have been associated with flowering time and organ size (Kim et al ; Wang et al. 2009; Wang et al. 2008; Zhang et al. 2011). MiRNA 159 has been implicated 504 in floral meristem development by targeting SPL transcription factors which in turn regulate 505 SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), LFY and AP1 genes 506 involved in SAM cell differentiation and floral meristem initiation (Achard et al. 2004). In 507 addition, mir159 mediates floral meristem formation by regulating MYB family transcription 508 factors and in particular gibberillin (GA)-specific MYBs. Loss of function 509 of mir159a and mir159b results in the abnormal expression of MYB33 and MYB65 which 510 influences plant development and decreases cell proliferation in leaves (Allen et al. 2010; Allen 511 et al. 2007). Mir164 controls proper formation and separation of adjacent plant organs by 512 targeting genes coding for the NON APICAL MERISTEM (NAM), ATAF and NAC family of 513 transcription factors. These genes regulate the establishment and maintenance of the shoot apical 514 and axillary meristem during floral development and control the formation of floral organs and

19 515 organ boundaries (Aida & Tasaka 2006; Baker et al. 2005; Kusumanjali et al. 2012; Mallory et 516 al. 2004). MiRNA172 targets the transcription factor gene AP2 and regulates floral meristem 517 initiation, flower patterning and floral organ identity (Chen 2004; Li & Zhang 2015). In addition, 518 mir172 which is regulated by FRUITFULL and AUXIN RESPONSE FACTOR positively 519 controls carpel size via repression of AP2 (Ripoll et al. 2015). MiRNA171 targets the GRAS 520 family transcription factors that play a key role in meristem maintenance, shoot development and 521 flowering time (Wang et al. 2010). Overexpression of mir171 in Arabidopsis led to altered plant 522 height, short branching, root elongation, leaf shape and flower structure (Wang et al. 2010). 523 Overexpression of mir171 in rice and barley had an effect on phase transitions and floral 524 meristem determinacy (Curaba et al. 2013; Fan et al. 2015), whereas overexpression of mir in tomato resulted in taller plants and earlier flowering (Huang et al. 2013). In conclusion all the 526 mirnas that were found to be downregulated in the grafted plants we used have been shown to 527 be implicated directly or indirectly in processes linked to cell, meristem, flower and fruit 528 development. Changes in expression of such important mirna molecules may have huge 529 implications in fruit morphology and may be responsible for the altered phenotypes of the 530 grafted plants. It is possible that some of these mirna expression changes control key-genes in 531 the fruit growth process that result in grafted plant s fruit phenotypic change. 532 Apart from the transcriptomic level was studied the grafting effect on the qualitative and 533 quantitative profile of secondary metabolites, mainly polyphenols, of C. pepo fruit samples. A 534 fast, efficient and sensitive state of the art targeted LC-MS/MS analysis was used for this 535 purpose. In the samples used, seventeen phenolic metabolites were identified in comparison to 536 (Iswaldi et al. 2013) who reported twenty-five phenolic compounds obtained from two green 537 zucchini varieties. To a certain extend qualitative differences may arise due to different technical 538 issues: extraction methods, plant material used (genotype, tissue), environmental factors 539 (altitude, climate, rainfall), abiotic stress, drought, UV-radiation, salt stress and nutrient 540 availability, harvesting season, and/or postharvest processing of fruits fruits (Boeing et al. 2014; 541 Cheynier et al. 2013; Nayak et al. 2015; Tiwari & Cummins 2013). Five additional metabolites 542 were identified in this study; elculetin, coniferyl and sinapoyl alcohol, arbutin, carnosic acid and 543 carnosol, even as traces. In fact, to the best of our knowledge this is the first report that the last 544 two diterpenes (carnosic acid and carnosol) are detected in C. pepo, except for Rosmarinus and 545 Salvia species (Birtić et al. 2015; Sarrou et al. 2016). The promotive effect of grafting on

20 546 biochemical and nutritional traits of some vegetables such as tomato, watermelon and eggplant 547 has been previously reported (Krumbein & Schwarz 2013; Kyriacou et al. 2015; Nicoletto et al. 2013; 548 Soteriou et al. 2014). In the majority of these reports the observed changes were also indirectly 549 connected to the belated harvest point (due to rootstock effect), the seasonal change and the 550 postharvest storage too. According to the data obtained, the cultivar that is used as the scion to 551 the heterografts plays a crucial role for the biosynthesis of the phenolic metabolites. That became 552 apparent, as compounds that were not present in the control s or the homograft of Big Moose, 553 were detected in the heterograft of Big Moose x Munchkin, while Munchkin contained such 554 phenolic compounds in relative concentrations as well. Thus, the changes observed in the 555 phenolic profile of Big Moose (RS) x Munchkin (SC) and Munchkin (RS) x Big Moose 556 (SC) are more likely scion-dependent than rootstock-dependent. These observations are in 557 agreement with previous studies on Solanaceae species that also reported the main scion effect 558 on final yield and fruit quality traits in grafted plants, while rootstock effects can alter such 559 characteristics possibly due to changes in vigor and concentration of plant growth regulators 560 induced by the rootstocks (Gisbert et al. 2012; Gisbert et al. 2011; Moncada et al. 2013; Muñoz- 561 Falcón et al. 2008; Passam et al. 2005). Furthermore, the trend in the accumulation of phenolics 562 seems to be also genetically controlled as in green fruit C. pepo cultivars and more complex than 563 yellow ones. 564 Previous investigations in Cucurbita inter-species heterografting involving significant 565 DNA methylation changes in melon and cucumber scions grafted onto pumpkin rootstocks has 566 been expanded by the current study (Avramidou et al. 2014) Conclusion 569 Herein, for the first time, has been demonstrated that intra-species grafting among C. pepo 570 cultivars induced dramatic changes in scion fruit size and that these changes are accompanied by 571 significant alterations in global DNA methylation patterns of scion genomes, mirna expression 572 changes, and phenolic profile changes. Moreover, it is shown that intra-species C. pepo grafting 573 induced moderate alterations in fruit shape with concomitant substantial changes in scion DNA 574 methylation and mirna expression patterns. These investigations contribute to a deeper 575 understanding of grafting in Cucurbitaceae and will shed light to the poorly understood 576 molecular mechanisms underlying in the effect of grafting on fruit morphology. Moreover, they

21 577 will help in understanding the association between graft-induced DNA methylation, mirna 578 expression and altered phenolic profiles in fruits of grafted plants. Taken together, the 579 differential epigenetic marks and phenolic patterns induced by grafting may lead to the 580 development of epi-molecular markers and metabolic predictors for rapid and efficient selection 581 of grafted vegetables with desired traits of agronomical relevance. 582 References 583 Achard P, Herr A, Baulcombe DC, and Harberd NP Modulation of floral development by a 584 gibberellin-regulated microrna. Development 131: Aida M, and Tasaka M Genetic control of shoot organ boundaries. Current opinion in plant 586 biology 9: Albacete A, Martinez-Anddujar C, Martinez-Perez An, Thompson AJ, Dodd IC, and Perez-Alfocea F Unravelling rootstock scion interactions to improve food security. Journal of 589 Experimental Botany /jxb/erv Allen RS, Li J, Alonso-Peral MM, White RG, Gubler F, and Millar AA MicroR159 regulation of 591 most conserved targets in Arabidopsis has negligible phenotypic effects. Silence 1: Allen RS, Li J, Stahle MI, Dubroué A, Gubler F, and Millar AA Genetic analysis reveals 593 functional redundancy and the major target genes of the Arabidopsis mir159 family. Proceedings 594 of the National Academy of Sciences 104: Avramidou E, Kapazoglou A, Aravanopoulos FA, Xanthopoulou A, Ganopoulos I, Tsaballa A, Madesis 596 P, Doulis AG, and Tsaftaris A Global DNA methylation changes in Cucurbitaceae inter- 597 species grafting. Crop Breeding and Applied Biotechnology 15: Baker CC, Sieber P, Wellmer F, and Meyerowitz EM The early extra petals1 mutant uncovers a 599 role for microrna mir164c in regulating petal number in Arabidopsis. Current Biology 15: Barrera-Figueroa BE, Wu Z, and Liu R Abiotic stress-associated micrornas in plants: discovery, 602 expression analysis, and evolution. Frontiers in biology 8: Birtić S, Dussort P, Pierre F-X, Bily AC, and Roller M Carnosic acid. Phytochemistry 115: Boeing JS, Barizão ÉO, e Silva BC, Montanher PF, de Cinque Almeida V, and Visentainer JV Evaluation of solvent effect on the extraction of phenolic compounds and antioxidant capacities 606 from the berries: application of principal component analysis. Chemistry Central Journal 8: Cao L, Yu N, Li J, Qi Z, Wang D, and Chen L Heritability and Reversibility of DNA Methylation 608 Induced by in vitro Grafting between Brassica juncea and B. oleracea. Scientific reports 6.

22 609 Chen X A microrna as a translational repressor of APETALA2 in Arabidopsis flower 610 development. Science 303: Cheynier V, Comte G, Davies KM, Lattanzio V, and Martens S Plant phenolics: recent advances 612 on their biosynthesis, genetics, and ecophysiology. Plant Physiology and Biochemistry 72: Cohen R, Burger Y, Horev C, Koren A, and Edelstein M Introducing grafted cucurbits to modern 614 agriculture: the Israeli experience Plant dis 91: Colla G, Rouphael Y, Leonardi C, and Bie Z Role of grafting in vegetable crops grown under 616 saline conditions. Scientia Horticulturae 127: Curaba J, Talbot M, Li Z, and Helliwell C Over-expression of microrna171 affects phase 619 transitions and floral meristem determinancy in barley. BMC plant biology 13: Dong H, Lei J, Ding L, Wen Y, Ju H, and Zhang X MicroRNA: function, detection, and 621 bioanalysis. Chemical reviews 113: Fan T, Li X, Yang W, Xia K, Ouyang J, and Zhang M Rice osa-mir171c mediates phase change 623 from vegetative to reproductive development and shoot apical meristem maintenance by 624 repressing four OsHAM transcription factors. PloS one 10:e Fuentes I, Stegemann S, Golczyk H, Karcher D, and Bock R Horizontal genome transfer as an 626 asexual path to the formation of new species. Nature 511: Fujimoto R, Sasaki T, Ishikawa R, Osabe K, Kawanabe T, and Dennis ES Molecular Mechanisms 628 of Epigenetic Variation in Plants. International Journal of Molecular Sciences 13: Gisbert C, Prohens J, and Nuez F Performance of eggplant grafted onto cultivated, wild, and 630 hybrid materials of eggplant and tomato. International Journal of Plant Production 5: Gisbert C, Prohens J, Raigón MD, Stommel JR, and Nuez F Eggplant relatives as sources of 632 variation for developing new rootstocks: Effects of grafting on eggplant yield and fruit apparent 633 quality and composition. Scientia Horticulturae 128: Goldschmidt EE Plant grafting: new mechanisms, evolutionary implications. Frontiers in plant 635 science Haroldsen V, Szczerba MW, Aktas H, Lopez J, Odias MJ, Chi-Ham CL, Labavitch J, Bennett AB, and 637 Powell ALT Mobility of transgenic nucleic acids and proteins within grafted rootstocks for 638 agricultural improvement. Frontiers in plant science 3: Huang Y, Bie Z, Liu P, Niu M, Zhen A, Liu Z, Lei B, Gu D, Lu C, and Wang B Reciprocal 640 grafting between cucumber and pumpkin demonstrates the roles of the rootstock in the 641 determination of cucumber salt tolerance and sodium accumulation. Scientia Horticulturae :

23 643 Iswaldi I, Gómez-Caravaca AM, Lozano-Sánchez J, Arráez-Román D, Segura-Carretero A, and 644 Fernández-Gutiérrez A Profiling of phenolic and other polar compounds in zucchini 645 (Cucurbita pepo L.) by reverse-phase high-performance liquid chromatography coupled to 646 quadrupole time-of-flight mass spectrometry. Food Research International 50: Kalantidis K, Schumacher HT, Alexiadis T, and Helm JM RNA silencing movement in plants. 648 Biology of the Cell 100: Katayama S, Ohno F, Yamauchi Y, Kato M, Makabe H, and Nakamura S Enzymatic synthesis of 650 novel phenol acid rutinosides using rutinase and their antiviral activity in vitro. Journal of 651 agricultural and food chemistry 61: Kehr J, and Buhtz A Long distance transport and movement of RNA through the phloem. Journal 653 of experimental botany 59: Kim JJ, Lee JH, Kim W, Jung HS, Huijser P, and Ahn JH The microrna156-squamosa 655 PROMOTER BINDING PROTEIN-LIKE3 module regulates ambient temperature-responsive 656 flowering via FLOWERING LOCUS T in Arabidopsis. Plant Physiology 159: King SR, Davis AR, Liu W, and Levi A Grafting for Disease Resistance. HortScience 43: Krumbein A, and Schwarz D Grafting: A possibility to enhance health-promoting and flavour 660 compounds in tomato fruits of shaded plants? Scientia Horticulturae 149: Kubota C, McClure MA, Kokalis-Burelle N, Bausher MG, and Rosskopf EN Vegetable Grafting: 662 History, Use, and Current Technology Status in North America. HortScience 43: Kusumanjali K, Kumari G, Srivastava PS, and Das S Sequence conservation and divergence in 664 mir164c1 and its target, CUC1, in Brassica species. Plant biotechnology reports 6: Kyriacou MC, Soteriou GA, Rouphael Y, Siomos AS, and Gerasopoulos D Configuration of 666 watermelon fruit quality in response to rootstock mediated harvest maturity and postharvest 667 storage. Journal of the Science of Food and Agriculture. 668 Lee J-M, Kubota C, Tsao SJ, Bie Z, Echevarria PH, Morra L, and Oda M Current status of 669 vegetable grafting: Diffusion, grafting techniques, automation. Scientia Horticulturae 127: Li C, and Zhang B MicroRNAs in Control of Plant Development. Journal of Cellular Physiology : /jcp Li S, Castillo Gonzalez C, Yu B, and Zhang X The functions of plant small RNAs in development 674 and in stress responses. The Plant Journal.

24 675 Li X, Bian H, Song D, Ma S, Han N, Wang J, and Zhu M Flowering time control in ornamental 676 gloxinia (Sinningia speciosa) by manipulation of mir159 expression. Annals of botany 111: Lough TJ, and Lucas WJ Integrative plant biology: role of phloem long-distance macromolecular 679 trafficking. Annu Rev Plant Biol 57: Louws FJ, Rivard CL, and Kubota C Grafting fruiting vegetables to manage soilborne pathogens, 681 foliar pathogens, arthropods and weeds. Scientia horticulturae 127: Lucas WJ, Yoo B-C, and Kragler F RNA as a long-distance information macromolecule in plants. 683 Nature Reviews Molecular Cell Biology 2: Mallory AC, Dugas DV, Bartel DP, and Bartel B MicroRNA regulation of NAC-domain targets is 685 required for proper formation and separation of adjacent embryonic, vegetative, and floral organs. 686 Current Biology 14: Melnyk CW, Molnar A, and Baulcombe DC Intercellular and systemic movement of RNA 688 silencing signals. The EMBO journal 30: Mermigka G, Verret F, and Kalantidis K RNA silencing movement in plants. Journal of 690 integrative plant biology. 691 Mohamed F, El-Hamed K, Elwan M, and Hussien M Impact of Grafting on Watermelon Growth, 692 Fruit Yield and Quality. Vegetable Crops Research Bulletin Moncada A, Miceli A, Vetrano F, Mineo V, Planeta D, and D Anna F Effect of grafting on yield 694 and quality of eggplant (Solanum melongena L.). Scientia Horticulturae 149: Mudge K, Janick J, Scofield S, and Goldschmidt EE A History of Grafting. Horticultural reviews: 696 John Wiley & Sons, Inc., Muñoz-Falcón JE, Prohens J, Rodríguez-Burruezo A, and Nuez F Potential of local varieties and 698 their hybrids for the improvement of eggplant production in the open field and greenhouse 699 cultivation. Journal of Food Agriculture and Environment 6: Nascimento PLA, Nascimento TCES, Ramos NSM, Silva GR, Gomes JEG, Falcão REA, Moreira KA, 701 Porto ALF, and Silva T Quantification, antioxidant and antimicrobial activity of phenolics 702 isolated from different extracts of Capsicum frutescens (Pimenta Malagueta). Molecules 19: Nayak B, Liu RH, and Tang J Effect of processing on phenolic antioxidants of fruits, vegetables, 705 and grains A review. Critical reviews in food science and nutrition 55: Nicoletto C, Tosini F, and Sambo P Effect of grafting on biochemical and nutritional traits of 707 Cuore di Bue tomatoes harvested at different ripening stages. Acta Agriculturae Scandinavica, 708 Section B Soil & Plant Science 63:

25 709 Obrero An, Die JV, Román Bn, Gómez P, Nadal S, and González-Verdejo CI Selection of 710 reference genes for gene expression studies in zucchini (Cucurbita pepo) using qpcr. Journal of 711 agricultural and food chemistry 59: Omid A, Keilin T, Glass A, Leshkowitz D, and Wolf S Characterization of phloem-sap 713 transcription profile in melon plants. Journal of experimental botany 58: Orsini F, Sanoubar R, Oztekin GB, Kappel N, Tepecik M, Quacquarelli C, Tuzel Y, Bona S, and 715 Gianquinto G Improved stomatal regulation and ion partitioning boosts salt tolerance in 716 grafted melon. Functional Plant Biology 40: Palauqui JC, Elmayan T, Pollien JM, and Vaucheret H Systemic acquired silencing: 718 transgene specific post transcriptional silencing is transmitted by grafting from silenced stocks to 719 non silenced scions. The EMBO Journal 16: Passam HC, Stylianou M, and Kotsiras A Performance of eggplant grafted on tomato and eggplant 721 rootstocks. European Journal of Horticultural Science: Pfaffl MW, Horgan GW, and Dempfle L Relative expression software tool (REST ) for group- 723 wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic 724 acids research 30:e36-e Ripoll JJ, Bailey LJ, Mai Q-A, Wu SL, Hon CT, Chapman EJ, Ditta GS, Estelle M, and Yanofsky MF microrna regulation of fruit growth. Nature plants Rouphael Y, Cardarelli M, Bassal A, Leonardi C, Giuffrida F, and Colla G Vegetable quality as 728 affected by genetic, agronomic and environmental factors. Journal of Food, Agriculture & 729 Environment 10: Sarrou E, Martens S, and Chatzopoulou P Metabolite profiling and antioxidative activity of Sage 731 (Salvia fruticosa Mill.) under the influence of genotype and harvesting period. Industrial Crops 732 and Products 94: Schwab R, Palatnik JF, Riester M, Schommer C, Schmid M, and Weigel D Specific effects of 734 micrornas on the plant transcriptome. Developmental cell 8: Soteriou GA, Kyriacou MC, Siomos AS, and Gerasopoulos D Evolution of watermelon fruit 736 physicochemical and phytochemical composition during ripening as affected by grafting. Food 737 chemistry 165: Spiegelman Z, Golan G, and Wolf S Don t kill the messenger: long-distance trafficking of mrna 739 molecules. Plant science 213: Taller J, Yagishita N, and Hirata Y Graft-induced variants as a source of novel characteristics in 741 the breeding of pepper (Capsicum annuum L.). Euphytica 108:73-78.

26 742 Tiwari U, and Cummins E Factors influencing levels of phytochemicals in selected fruit and 743 vegetables during pre-and post-harvest food processing operations. Food Research International : Tsaballa A, Athanasiadis C, Pasentsis K, Ganopoulos I, Nianiou-Obeidat I, and Tsaftaris A Molecular studies of inheritable grafting induced changes in pepper (Capsicum annuum) fruit 747 shape. Scientia Horticulturae 149: Tsaftaris AS, Kapazoglou A, and Darzentas N Epigenetics, Epigenomics, and Implications in Plant 749 Breeding "Plant biotechnology and agriculture: Prospects for the 21st century" eds A. 750 Altman and P.M. Haegawa Elsevier Press. 751 Valenzuela GM, Soro AS, Tauguinas AL, Gruszycki MR, Cravzov AL, Giménez MC, and Wirth A Evaluation polyphenol content and antioxidant activity in extracts of Cucurbita spp. Open 753 Access Library Journal 1: Varkonyi-Gasic E, Wu R, Wood M, Walton EF, and Hellens RP Protocol: a highly sensitive RT- 755 PCR method for detection and quantification of micrornas. Plant methods 3: Vekemans X, Beauwens T, Lemaire M, and Roldán Ruiz I Data from amplified fragment length 757 polymorphism (AFLP) markers show indication of size homoplasy and of a relationship between 758 degree of homoplasy and fragment size. Molecular ecology 11: Vos P, Hogers R, Bleeker M, Reijans M, Van de Lee T, Hornes M, Friters A, Pot J, Paleman J, and 760 Kuiper M AFLP: a new technique for DNA fingerprinting. Nucleic acids research : Vrhovsek U, Masuero D, Gasperotti M, Franceschi P, Caputi L, Viola R, and Mattivi F A versatile 763 targeted metabolomics method for the rapid quantification of multiple classes of phenolics in 764 fruits and beverages. Journal of agricultural and food chemistry 60: Wang J-W, Czech B, and Weigel D mir156-regulated SPL transcription factors define an 766 endogenous flowering pathway in Arabidopsis thaliana. cell 138: Wang J-W, Schwab R, Czech B, Mica E, and Weigel D Dual effects of mir156-targeted SPL 768 genes and CYP78A5/KLUH on plastochron length and organ size in Arabidopsis thaliana. The 769 plant cell 20: Wang L, Mai Y-X, Zhang Y-C, Luo Q, and Yang H-Q MicroRNA171c-targeted SCL6-II, SCL6-771 III, and SCL6-IV genes regulate shoot branching in Arabidopsis. Molecular plant 3: Wu R, Wang X, Lin Y, Ma Y, Liu G, Yu X, Zhong S, and Liu B Inter-Species Grafting Caused 773 Extensive and Heritable Alterations of DNA Methylation in <italic>solanaceae</italic> Plants. 774 PloS one 8:e61995.

27 775 Xanthopoulou A, Ganopoulos I, Kalivas A, Nianiou-Obeidat I, Ralli P, Moysiadis T, Tsaftaris A, and 776 Madesis P Comparative analysis of genetic diversity in Greek Genebank collection of 777 summer squash (Cucurbita pepo) landraces using start codon targeted (SCoT) polymorphism and 778 ISSR markers. Australian Journal of Crop Science 9: Yagishita N, and Hirata Y Graft-induced change in fruit shape in Capsicum annuum LI genetic 780 analysis by crossing. Euphytica 36: Yetisir H, Özdemir E, Aras V, Candır E, and Aslan H Rootstocks effect on plant nutrition 782 concentration in different organ of grafted watermelon Agricultural Sciences Zhang B MicroRNA: a new target for improving plant tolerance to abiotic stress. Journal of 784 experimental botany 66: Zhang X, Zou Z, Zhang J, Zhang Y, Han Q, Hu T, Xu X, Liu H, Li H, and Ye Z Over-expression 786 of sly-mir156a in tomato results in multiple vegetative and reproductive trait alterations and 787 partial phenocopy of the sft mutant. FEBS letters 585: Zhong S, Fei Z, Chen Y-R, Zheng Y, Huang M, Vrebalov J, McQuinn R, Gapper N, Liu B, Xiang J, Shao 789 Y, and Giovannoni JJ Single-base resolution methylomes of tomato fruit development 790 reveal epigenome modifications associated with ripening. Nat Biotech 31:

28 793 Figure Legends 794 Figure 1: Fruits of cv. Munchkin and cv. Big Moose grafted plants and their controls 795 (A) Control non-grafted Munchkin (left) and Big Moose (right) 796 (B) Grafted Big Moose (SC) on Munchkin (RS) (top) and reciprocal combination (bottom) 797 with no change in fruit morphology that is the same as the fruit morphology of the non-grafted 798 scions. 799 (C) Grafted Big Moose (SC) on cv. Munchkin (RS) with the altered fruit morphology 800 growing on the grafted plant. 801 (D) Grafted Big Moose (SC) on cv. Munchkin (RS) with the altered fruit morphology (left) 802 next to grafted Big Moose (SC) on cv. Munchkin (RS) with no alteration (right) Figure 2: Fruits, flowers and leaves of grafted plants cv. Round green, cv. Princess and 805 their controls. 806 (A) Control non-grafted Round green 807 (B) Grafted Round green (SC) on Princess (RS) 808 (C) Control non-grafted Princess 809 (D) Grafted Princess (SC) on Round green (RS) Figure 3: Relative (by comparison to non-grafted seedlings) DNA methylation and 812 demethylation levels (percent of total within each sample; averaged) of grafted plants based 813 on MSAP analysis. Relative levels at randomly sampled 5 - CCGG -3 sites with the MSAP 814 marker (A) Munchkin x Big Moose grafted plants versus Big Moose s. B) Big Moose x 815 Munchkin plants versus Munchkin s. C) Munchkin x Big Moose grafted plant with fruit 816 size changes versus Big Moose seedlings. Asterisks indicate values significantly lower than the 817 values of the corresponding seedling controls (P<0.05).y-axis: relative (percent) methylation 818 level Figure 4: Relative (by comparison to non-grafted seedlings) levels at randomly sampled CCGG -3 sites with MSAP markers (A) Round green x Princess grafted plants versus 822 Princess seedlings. B) Princess x Round green plants versus Round green seedlings 823 (percent of total within each sample; averaged). Asterisks indicate values significantly lower

29 824 than the values of the corresponding seedling controls (P<0.05).y-axis: relative (percent) 825 methylation level. 826 Figure 5: Quantitative qrt-pcr analyses of selected mirnas in grafted plants and their 827 non-grafted controls 828 (A) mirna expression in heterografts of Big moose and Munchkin. B= Big Mouse, M= 829 Munchkin. 830 (B) mirna expression levels in heterografts of Princess and Round green. P= Princess, 831 R= Round green 832 Expression values were normalized to those of the reference gene CpEF-1a. Relative expression 833 ratio of each sample was compared to the control group which was assigned arbitrarily the value 834 of 1. Data represent mean values from two independent experiments with standard deviations. 835 Values significantly different (P < 0.05) from the control sample ( Big Moose ) are marked with 836 an asterisk (*) Figure 6: Overlay heat map of metabolite profiles in grafted plants in comparison with 839 non-grafted controls. 840 (A) For the cultivars that differ in fruit size and their grafted plants. On the left the part of the 841 heatmap that corresponds to the graftings and the control that produced small fruits in terms of 842 size (red color), on the right the part of the heatmap that corresponds to the ones that produced 843 large fruit (pink color). 844 (B) For the cultivars that differ in fruit shape and their grafted plants. On the left the part of the 845 heatmap that corresponds to the graftings and the control that produced round fruits (red color), 846 on the right the part of the heatmap that corresponds to the ones that produced elongated fruits 847 (pink color). The metabolites in each square represent the effect of plant grafting on the amount 848 of every metabolite using a false-color scale. 849

30 Figure 1(on next page) Fruits of cv. Munchkin and cv. Big Moose grafted plants and their controls. Figure 1: Fruits of cv. Munchkin and cv. Big Moose grafted plants and their controls (A) Control non-grafted Munchkin (left) and Big Moose (right) (B) Grafted Big Moose (SC) on Munchkin (RS) (top) and reciprocal combination (bottom) with no change in fruit morphology that is the same as the fruit morphology of the non-grafted scions. (C) Grafted Big Moose (SC) on cv. Munchkin (RS) with the altered fruit morphology growing on the grafted plant. (D) Grafted Big Moose (SC) on cv. Munchkin (RS) with the altered fruit morphology (left) next to grafted Big Moose (SC) on cv. Munchkin (RS) with no alteration (right).

31 A) B) Munchkin x Big Moose Munchkin Big Moose C) D) Big Moose x Munchkin Munchkin x Big Moose fruit changed plant Munchkin x Big Moose

32 Figure 2(on next page) : Fruits, flowers and leaves of grafted plants cv. Round green, cv. Princess and their controls Figure 2: Fruits, flowers and leaves of grafted plants cv. Round green, cv. Princess and their controls. (A) Control non-grafted Round green (B) Grafted Round green (SC) on Princess (RS) (C) Control non-grafted Princess (D) Grafted Princess (SC) on Round green (RS)

33 A) B) Round green C) D) Princess x Round green Princess Round green x Princess

34 Figure 3(on next page) Relative (by comparison to non-grafted seedlings) DNA methylation and demethylation levels (percent of total within each sample; averaged) of grafted plants based on MSAP analysis. Figure 3: Relative (by comparison to non-grafted seedlings) DNA methylation and demethylation levels (percent of total within each sample; averaged) of grafted plants based on MSAP analysis. Relative levels at randomly sampled 5 - CCGG -3 sites with the MSAP marker (A) Munchkin x Big Moose grafted plants versus Big Moose s. B) Big Moose x Munchkin plants versus Munchkin s. C) Munchkin x Big Moose grafted plant with fruit size changes versus Big Moose seedlings. Asterisks indicate values significantly lower than the values of the corresponding seedling controls (P<0.05).y-axis: relative (percent) methylation level.

35 A) Munchkin x Big Moose gra@ed plants versus Big Moose seed-plants No change Demethyla5on De-novo methyla5on B) Big Moose x Munchkin gra?ed plants versus Munchkin seed-plants No change Demethyla4on De-novo methyla4on C) Munchkin x Big Moose gra?ed plant with fruit size changes versus Big Moose seed-plant No change Demethyla4on De-novo methyla4on

36 Figure 4(on next page) Relative (by comparison to non-grafted seedlings) levels at randomly sampled 5 - CCGG -3 sites with MSAP markers Figure 4: Relative (by comparison to non-grafted seedlings) levels at randomly sampled 5 - CCGG -3 sites with MSAP markers (A) Round green x Princess grafted plants versus Princess seedlings. B) Princess x Round green plants versus Round green seedlings (percent of total within each sample; averaged). Asterisks indicate values significantly lower than the values of the corresponding seedling controls (P<0.05).y-axis: relative (percent) methylation level.

37 A) Round green x Princess gra@ed plants versus Princess seed-plants B) No change Demethyla5on De-novo methyla5on Princess x Round green gra?ed plants versus Round green seed-plants No change Demethyla4on De-novo methyla4on

38 Figure 5(on next page) Quantitative qrt-pcr analyses of selected mirnas in grafted plants and their nongrafted controls Figure 5: Quantitative qrt-pcr analyses of selected mirnas in grafted plants and their non-grafted controls (A) mirna expression in heterografts of Big moose and Munchkin. B= Big Mouse, M= Munchkin. (B) mirna expression levels in heterografts of Princess and Round green. P= Princess, R= Round green Expression values were normalized to those of the reference gene CpEF-1a. Relative expression ratio of each sample was compared to the control group which was assigned arbitrarily the value of 1. Data represent mean values from two independent experiments with standard deviations. Values significantly different (P < 0.05) from the control sample ( Big Moose ) are marked with an asterisk (*).

39 A) mir156 mir159 Rela%vre expression ra%o 1,2 1 0,8 0,6 0,4 0,2 0 B MXB M BXM 1,2 1 0,8 0,6 0,4 0,2 0 B MXB M BXM mir164 mir166 1,4 1,2 1,2 1 0,8 0,6 0,4 0,2 0 B MXB M BXM 1 0,8 0,6 0,4 0,2 0 B MXB M BXM mir B MXB M BXM B)

40 Figure 6(on next page) Overlay heat map of metabolite profiles in grafted plants in comparison with nongrafted controls. Figure 6: Overlay heat map of metabolite profiles in grafted plants in comparison with non-grafted controls. (A) For the cultivars that differ in fruit size and their grafted plants. On the left the part of the heatmap that corresponds to the graftings and the control that produced small fruits in terms of size (red color), on the right the part of the heatmap that corresponds to the ones that produced large fruit (pink color). (B) For the cultivars that differ in fruit shape and their grafted plants. On the left the part of the heatmap that corresponds to the graftings and the control that produced round fruits (red color), on the right the part of the heatmap that corresponds to the ones that produced elongated fruits (pink color). The metabolites in each square represent the effect of plant grafting on the amount of every metabolite using a false-color scale.

41 A) B)