A large number of these genes were identified as differentially expressed over the course of fruit development, which is consistent with previous studies of transcriptome changes during fruit ripening in sweet orange . However, most genes showed similar temporal expression patterns among all rootstock genotypes. Furthermore, only ~15% of the genes were genotype-specific . Therefore, the remainder of this study focused on DEGs identified between these rootstock genotypes during fruit development. A total 684, 388, 361, 178, 395, and 885 genes were significantly differentially expressed between RL vs SO, CZ vs SO, TF vs SO, RL vs CZ, TF vs CZ and TF vs RL rootstocks respectively . The majority of the differentially expressed genes are observed in comparisons involving rough lemon rootstocks, especially compared to trifoliate orange. This is consistent with the observed differences in fruit quality traits, as fruit of trees grafted on rough lemon rootstock showed consistent significant differences from fruit of trees grafted on the other three rootstocks in many of the traits measured . These results suggest that rough lemon and trifoliate rootstocks show the greatest effects on the scion and are good candidates to identify graft-related genes playing a role in fruit quality. The largest and most significant changes in gene expression between rootstocks were observed at time points two and three . Among the DEGs were several genes with functions involved in fruit quality traits, such as those relating to starch and sucrose metabolism, fructose metabolism, and hormone signaling related genes. KEGG pathway analysis displayed plant hormone signal transduction, carotenoid biosynthesis, plastic pots for planting and fructose and mannose metabolism pathways to be significantly enriched. Several genes involved in various hormone-signaling pathways were DE, mainly genes in the abscisic acid and auxin-response pathways.

Several genes involved in these pathways were chosen to validate the RNA-seq data by qRT-PCR due to their potential biological significance regarding rootstock effects on fruit quality.ABA has been known to be a regulator of fruit ripening and response to abiotic stress in non-climacteric fruit. AHG1, a homolog of Arabidopsis PP2C family protein, was DE in this study. PP2C is a negative regulator of the ABA hormone-signaling pathway. This gene was slightly up-regulated when comparing fruit of trees grafted on trifoliate to fruit of trees grafted on rough lemon rootstock at time two and significantly down-regulated at time three . Upregulation of AHG1 is in accordance with previous studies showing this gene being induced by water stress, which may have occurred in September. The downregulation of this gene later in the season could be correlated with increased fruit maturation in fruit grown on trifoliate rootstocks. This is in agreement with a study in tomato where suppression of PP2C expression led to increased ABA accumulation and higher levels of ABA-signaling genes that increase the expression of ABA-mediated ripening-related genes.Auxin signal transduction is mediated by Aux/IAA and ARF genes. Aux/IAA proteins are negative regulators of the auxin signal transduction pathway. In this study, a gene encoding an Aux/IAA protein, IAA16, was up-regulated in fruit grown on trifoliate compared to rough lemon rootstocks at time two and three . A previous study revealed that a gain-of-function mutation in IAA16 displayed reduced response to auxin and ABA, which led to reduced plant growth. Silencing of related Aux/IAA genes increased fruit size in tomato due to auxin control of cell expansion and elongation. In addition to Aux/IAA, another early auxin-response gene, SAUR78, was DE in this study. This gene was down-regulated in fruit grown on trees grafted onto trifoliate compared to rough lemon rootstocks at time two and three . Small Auxin Up RNA genes are a group of auxin-inducible proteins.

Other SAUR genes have also been shown to promote cell expansion. Furthermore, a MYB77 gene encoding a transcription factor was DE in this study, displaying a slight increase in expression in fruit grown on trifoliate rootstock at time two, but a large decrease in expression at time three . This gene was previously described as a regulator of the auxin signal transduction pathway. This protein was shown to interact with ARFs to promote plant growth. Interestingly, the effects of MYB77 in Arabidopsis were found to be increased by endogenous exposure to ABA and further promote plant growth. While these two studies were performed in roots, this transcription factor was shown to be involved in citrus fruit ripening, where it was highly correlated with ABA and suggested to have a similar function in response to the hormone.Although there were not statistically significantly differences seen in other genes in the auxin- and ABA-signaling pathways, trends could be observed during hierarchical clustering of these genes. Many of the genes within a family shared common expression levels and generally follow the predicted regulatory patterns in their respective pathways . Taken together, the changes in ABA- and auxinresponsive genes suggest a potential mechanism for induced ripening by trifoliate rootstock and larger fruit produced when rough lemon is used as a rootstock.The expansion phase of citrus fruit development involves cell enlargement and water accumulation. Given the changes in hormone-signaling pathways that likely lead to changes in fruit size, other genes related to fruit growth, such as transporters and genes related to cell wall metabolism were investigated. This led to the identification of two DEGs that could be influencing fruit size. The first, a Plasma membrane Intrinsic Protein 2 gene encoding an aquaporin was down-regulated in fruit grown on trifoliate rootstock . Water import in plants is mediated by aquaporins and essential for cell expansion.

These genes were highly expressed in expanding green grapes and one was identified as a candidate gene under the QTL for berry weight. PIP genes were also associated with an increase in volume of fruit in apple and strawberry. The second DEG, an expansin , was also down-regulated in fruit grown on trifoliate rootstock . Expansins play various roles in fruit development, including cell elongation and cell wall softening. A homolog of EXP1 in tomato was expressed during green fruit cell division and expansion with maximum accumulation of EXP1 during the late phase of green fruit expansion and early maturation. The increase in expression of these two genes in fruit grown on rough lemon rootstock could contribute to the larger fruit size observed. In addition to cell division and cell expansion, during fruit development, fruit softening is also an important feature that relies on cell wall metabolism. The Trichome Birefringence-Like gene, which encodes a protein required for cellulose biosynthesis, was identified in our study as DE. Mutations in this gene caused a reduction in the amount of pectins and an increase in pectin methylesterase activity. PME catalyses the demethylesterification of pectin, which may undergo depolymerisation by glycosidases. TBL23 was up-regulated in fruit grown on trifoliate rootstock compared to rough lemon , suggesting a potential role in fruit softening during citrus ripening. Transcription factors also play an important role in plant development and fruit ripening. Several transcription factors were differentially expressed in this study. GO enrichment showed the molecular function GO term ‘DNA-binding transcription factor activity’ was significantly enriched. In addition to the MYB77 transcription factor gene described earlier, a GRAS transcription factor gene, HAM3, was DE in this study. GRAS transcription factors were previously found to play a role in berry development and ripening in grapes, tomato, and citrus. This transcription factor showed increased expression later in the season when fruit were grown on trifoliate rootstock, drainage for plants in pots suggesting the rootstock influences its role in improved citrus fruit quality.The largest phenotypic differences seen in mature fruit grown on trifoliate compared to rough lemon rootstock were in the levels of total soluble sugar and titratable acid in ripe fruit. The levels of sugars and acids and their ratio in fleshy fruits is one of the most important determinants of sensory traits such as taste and flavor. Two genes were identified as differentially expressed that could play a role in the accumulation of these compounds. Firstly, a P-type ATPase was DE in fruit growing on trees grafted onto trifoliate versus rough lemon. This gene was down-regulated at time two, but upregulated at time three . Studies have proposed a number of ATPases as proton pumps that are responsible for organic acid accumulation in citrus fruit.

The reduced expression of this ATPase gene later in the season in fruit grown on rough lemon rootstocks could contribute to the lower accumulation of titratable acid levels in these fruits. This ATPase gene identified in this study was not identified in the previous citrus studies, but the regulation of acid accumulation is a complex, as can be seen in other fruits, such as papaya and apple. It is possible this is a graftinduced effect observed with these specific rootstocks, which were not examined in the previous studies. Secondly, a homolog of Arabidopsis BETAFRUCT4 was down-regulated in fruit of trees grown on trifoliate rootstock compared to rough lemon at time three . This gene encodes a vacuolar invertase. Decreased expression of vacuolar invertases has been associated with increased sucrose content and accelerated ripening. Interestingly, by using an antisense acid invertase gene in transgenictomato to reduce acid invertase activity, fruit displayed higher levels of sucrose, as well as smaller fruit. We see similar trends in sugar accumulation and alterations in fruit size in this study. Klann et al. suggested that the water influx that drives fruit expansion is closely related to the concentration of osmotically active soluble sugars and therefore, all genotypes accumulate water until they reach a similar threshold of soluble sugar concentration. This could also contribute to the increased size of fruit grown on rough lemon fruit compared to trifoliate rootstocks.This study did not identify any statistically significant differentially expressed miRNAs from our fruit small RNA seq data. Therefore, potential miRNAs that target DEGs were predicted. An in-house R-script was used to select for miRNA-mRNA interaction pairs with an expected negative correlation in gene expression. These pairs were identified for the ten genes described above. All ten miRNA genes and their target mRNAs were detected by qRT-PCR. Pearson correlation coefficient value between the relative expression level detected by qRT-PCR and by RNA-sequencing was highly significant with r = 0.94. Of the ten interaction pairs, eight followed expected fold changes between timepoints . Therefore, it is likely that these eight target mRNAs are being regulated to some extent by their respective miRNA.Citrus is now grown in more than 140 countries in tropical, subtropical and Mediterranean regions. It is one of the most economically important crops in the world. Citrus are rarely grown from seed and virtually all commercial citrus is propagated by grafting. This reduces the juvenile phase, allowing for the trees to produce fruit many years earlier than would trees grown from seed1 . Due to the large variation in growing conditions and climate in the regions where citrus is grown, different citrus rootstocks are required to improve yield and fruit quality in numerous diverse climates, as well as resist various pests and diseases. Rootstocks impart certain traits to the scion and the effects of rootstocks can be large. The most significant impacts are on growth, vigor and yield, tree nutrition, stress resistance, and fruit quality. The rootstock effects on various aspects of tree growth and fruit development are well documented, but the molecular mechanisms underlying most of these differences are unknown. Previous studies have shown changes in the transcriptome of various rootstock genotypes, especially in response to biotic and abiotic stressors. These types of changes have been seen in Arabidopsis, corn, mulberry, tomato, and poplar. In citrus, gene expression profiling has been used to understand rootstock effects and responses to biotic and abiotic factors. In another study, expression studies of leaves from mandarin grafted onto various rootstocks were analyzed in order to explain rootstock effects on the growth of scions. There is extremely limited tissue-specific transcriptome knowledge in citrus, especially for root tissue. A small number of studies have evaluated trifoliate, trifoliate hybrid, and mandarin root transcriptomes in response to citrus diseases, but these studies each assessed only one genotype. Only recently has an RNA-seq based approach been used to establish a reference transcriptome for citrus and of the 28 samples used in the study, only two were obtained from roots.