Litchi chinensis var. Xianjinfeng single plant buds were grafted onto healthy Heli litchi rootstocks in 2017 in Pingnan District, Guiping City. In 2019, a mutant single plant N7 was found in this batch of grafted seedlings. Two different types of leaves and fruits were found on N7. The leaves N7NL and fruits N7NF of one of them are consistent with the traits of Xianjinfeng. The leaves N7VL and fruits N7VF of the other one have different traits compared to those of Xianjinfeng. The skin of N7VF is turtle cracking, and the peak is raised. The flesh texture is crispier than that of Xianjingeng. The maturity period is 7–10 days less than that of Xianjinfeng. On the contrary, N7NF’s peel exhibits smooth cracking, whereas the mutant plants’ leaves (N7VL) are smooth compared to those of N7NL, which are curled (Figure 1). In 2022, the leaves N7NL and N7VL and fruits N7NF and N7VF of both types of grafts were collected and used for subsequent omics analyses.

Vacuum freeze-dried leaf and fruit samples of N7N and N7V were ground to powder using a grinder (MM 400, Retsch, Haan, Germany). Powder aliquots (100 mg) were extracted at 4°C with 0.6 mL of 70% aqueous methanol. The aliquots were vortexed during the extraction. The extracts were centrifuged (10,000 g) for 10 min and filtered through a microporous membrane (0.22 µm). After aspirating the supernatant, the sample was stored in a sample vial for UPLC-MS/MS analysis. Ultra-performance liquid chromatography (UPLC) (Shim-Pack UFLC SHIMADZU CBM30A, https://www.shimadzu.com.cn/) and tandem mass spectrometry (SHIMADZU Corp., Kyoto, Japan) (MS/MS) (Applied Biosystems 4500 QTRAP) were used for data acquisition. UPLC-MS/MS operating conditions were as previously reported (Yu et al., 2022).

The fruit pulp from three fruits were used for total RNA extraction using the Plant Total RNA Purification Kit (made) according to the manufacturer’s instructions. RNA purity testing, quantification, and integrity were performed using a NanoPhotometer (IMPLEN, Los Angeles, CA, United States) and the Qubit RNA Assay Kit in a Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, United States) and the RNA Nano 6000 Assay Kit on the Agilent Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA, United States). Sequencing of library preparation and quality control were performed as previously reported (Liu et al., 2019) and sequenced on the Illumina sequencing platform by Gene Denovo Biotechnology Co., Ltd. (Guangzhou, China).

The sequencing data were first processed for quality control using fastp (Chen et al., 2018) to remove reads with adapters, having an N ratio >10%, with all A bases, and those with Q ≤ 20 for more than 50% of the entire read. Moreover, the base composition and quality distribution analyses were performed. The high-quality reads were compared with the reference genome (JAHYJY000000000). According to the comparison results of HISAT2 (Kim et al., 2019), we used StringTie (Pertea et al., 2015) to reconstruct transcripts and used RSEM (Li and Dewey, 2011) to calculate the expression levels of the genes as Fragments Per Kilobase of transcript per Million mapped reads (FPKM) in each sample. The gene expression abundance distribution, PCA, and PCC were computed in R (www.r-project.org). Next, we computed differential gene expression [by using mean FPKM values (n = 3)] between N7NL vs N7VL and N7NF vs N7VF using DESeq2 (Love et al., 2014), and differentially expressed genes/transcripts (DEGs/DETs) were screened if the false discovery rate (FDR) was < 0.05 and log2 fold change was > 2. The DEG/DET data were visualized as volcano plots and clustered (hierarchical clustering) in R (www.r-project.org). The DEGs/DETs were mapped to the Gene Ontology (GO) database () and KEGG pathway database (https://www.genome.jp/kegg/pathway.html). A hypergeometric test was applied to find significant GO terms and KEGG pathways to which the DEGs/DETs were significantly enriched.

The RNAs were extracted, quantified, and tested for quality from triplicate samples, as mentioned in Section 2.3. The qRT-PCRs for 13 selected genes were carried out as reported earlier (Li et al., 2013). Primers were designed using Primer Premier 5 (Lalitha, 2000). The qRT-PCR for each reaction mixture was performed using 5 µL 2x qPCR mix, 10 pmol/μL each of forward and reverse primers, 2 µL cDNA, and 2.5 µL ddH₂O. The reactions were carried out on the QuantStudio™ 5 Real-Time PCR System (Applied Biosystems, Waltham, Massachusetts, United States). The data quality control analysis was performed by using QuantStudio™ Design & Analysis Software (Applied Biosystems, Waltham, Massachusetts, United States). The 2^−ΔΔCT method was used for measuring relative gene expression (Livak and Schmittgen, 2001).

Metabolomic analysis of 12 litchi samples (six fruit and six leaf samples) revealed 1,060 metabolites. These metabolites belonged to 20 compound classes. Amino acids and derivatives constituted the highest number of metabolites (180), followed by flavonoids (176), lipids (101), organic acids and their derivatives (92), and organoheterocyclic compounds (66) (Figure 2A). Leaves had a higher content of alcohols and polyphenols, flavonoids, lipids, nucleotides and derivatives, organooxygen compounds, phenolic acids, and vitamins than fruits, whereas the content of alkaloids and derivatives, amines, amino acids and derivatives, benzene and substituted derivatives, carbohydrates and their derivatives, organic acids and their derivatives, organoheterocyclic compounds, organosulfur compounds, phenols and their derivatives, phenylpropanoids and polyketides, phytohormones, polyamines, and terpenoids was higher in fruits (Figures 2B, C).

In general, N7VL had a higher content of alcohols and polyphenols, alkaloids and derivatives, amines, carbohydrates and derivatives, nucleotides and derivatives, organosulfur compounds, phenolic acids, and terpenoids than N7NL, while the content of the other compounds was higher in N7NL than in N7NV. In fruits, the content of alcohols and polyphenols, alkaloids and derivatives, amino acids and derivatives, lipids, nucleotides and derivatives, organosulfur compounds, phenols and their derivatives, and phenylpropanoids and polyketides was higher in N7VF than in N7NF, while the content of the other compounds was higher in N7NF than in N7NF. These changes indicate that both leaves and fruits differ in their metabolic content. Furthermore, the organs of mutant and normal plants also differ in terms of metabolite content.

Principal component analysis showed that replicates of each sample were grouped together (Figure 2D). Similarly, hierarchical clustering showed that leaf samples from both N7N and N7V were clustered together, while fruit samples were clustered together (Figure 2E). The PCA further confirmed these observations, where replicates of the same tissue had a relatively higher correlation compared to the different tissues and plants (Figure 2F).

The screening of metabolites based on VIP ≥ 1 and t-test p < 0.05 in the OPLS-DA model resulted in 106 DAMs between N7VL and N7NL, out of which 37 and 69 metabolites showed increased and decreased accumulation in N7VL compared to N7NL, respectively (Figure 3A). The top 15 metabolites based on the VIP of OPLS-DA are shown in Figure 3B. Notably, epicatechin, shikimic acid, procyanidin A3, adenosine, eicosadienoic acid, and trilobatin contents were higher in N7VL than in N7NL. The DAM with highest log2FC was 2-amino-4-(methylthio)butyric acid (5.99), followed by 4-hydroxyquinazoline (2.82), keracyanin chloride (2.31), 3,5-dihydroxybenzoic acid (1.67), and eicosadienoic acid (1.63).,while those having significantly higher contents in N7NL than in N7VL were sattabacin (−3.29), manglieside E (−3.05), (9Z,11E)-octadecadienoic acid (−3.0), chrysoeriol 7-O-rutinoside (−2.56), and 14,15-dehydrocrepenynic acid (−2.3). The DAMs in N7NL compared to N7VL were enriched in biosynthesis of secondary metabolites; starch and sucrose metabolism; citrate cycle; valine, leucine, and isoleucine degradation; glycerolipid metabolism; and others (Figure 3C).

Among the DAMs, most of the compounds classified as carbohydrates and derivatives, nucleotides and derivatives, and phenolic acids were accumulated in higher amounts in N7VL than in N7NL, where the higher carbohydrate content, i.e., D-fructose 6-phosphate, D-glucose 6-phosphate, D-glucarate, D-galactaric acid, D-mannose 6-phosphate, pteroside A, and D-3-phosphoglyceric acid in N7VL indicates higher resources for downstream processes. In addition, the higher levels of hydroxycinnamate, p-coumaric acid 4-O-glucoside, p-coumaric acid 4-O-hexoside, 1-O-gentisoyl-D-glucoside, shikimic acid, and caffeic acid O-glucoside in N7VL suggest that the mutant leaves are better in these traits. On the other hand, the content of most of the amino acids and derivatives, flavonoids, lipids and organic acids and derivatives, organoheterocyclic and organooxygen compounds, phenylpropanoids, phytohormone (N6-isopentenyladenine-7-glucoside, iP7G), and vitamins was lower in N7VL than in N7NL (Figure 3D). This change may be related to the higher cytokinin (iP7G) content (Wang et al., 2016).

The screening of metabolites yielded 101 DAMs, of which 45 and 56 showed increased and reduced accumulation in N7VF compared to N7NF, respectively (Figure 4A). The top 15 metabolites based on the VIP of OPLS-DA are shown in Figure 4B. N7VF had higher content of maleic acid (3.69), proline (3.56), aspartic acid (3.18), D-proline (2.84), L-saccharopine (2.47), L-methionine (1.82), typhaneoside (1.77), and others. On the contrary, N7NF had higher amounts of scopoletin, benzamidine, 7-methylguanine, L-phenylalanine, D-phenylalanine, HBOA, delphinidin-3-O-rutinoside, citric acid, and others. DAMs in N7NF compared to N7VF were enriched in butanoate metabolism, amino acid biosynthesis, 2-oxocarboxylic acid metabolism, glycerolipid metabolism, and others (Figure 4C).

N7VF fruits contained higher levels of amino acids and derivatives, carbohydrates and derivatives, and a limited number of compounds classified as lipids (oleamide, 6-aminocarboxylic acid, sn-glycero-3-phosphocholine, and gamma,gamma-dimethylallyl pyrophosphate) and organic acids and derivatives (L-malic acid, fumaric acid, succinic anhydride, and maleic acid). N7NF fruits contained higher levels of flavonoids, scopoletin, amines, some amino acids and derivatives (L-phenylalanine, D-phenylalanine, L-glutamic acid, 2-aminoisobutyric acid, etc.), benzamidine, carbohydrates and carbohydrate derivatives, (D-glucose-1-phosphate, turanose, alpha, alpha-trehalose, and galactinol), uridine 5′-monophosphate, some organic acids and derivatives (citric acid, furanoic acid, isocitrate, D-threo-isocitric acid, and quinic acid), organoheterocyclic compounds, shikimic acid, scopoletin, iP7G, and vitamin U (Figure 4D).

The sequencing of 12 libraries generated 513.84 million raw reads, with an average of 42.82 million raw reads per library. After filtering, 510.64 million clean reads (>99.1% of the raw reads) were obtained. On average, 81.56% of the clean reads could be mapped to the reference genome (Supplementary Table S1). In general, the FPKM distribution density was higher for N7NF and N7NV than for N7VF and N7VL (Figure 5A). The PCA showed a consistent grouping with that of the metabolomic samples, i.e., replicates were grouped according to samples. In particular, fruit samples showed less variation in PC2 compared to leaf samples. However, there was greater variability between the fruit and leaf (Figure 5B). This was also observed for PCA, i.e., there was a higher correlation between the expression in fruits of both samples compared to the fruit and leaf within a litchi species (Figure 5C). The qRT-PCR analysis results also complied with the RNA sequencing results (Figure 5D).

Transcript screening revealed 2,144 DEGs/DETs, of which 895 and 1,249 had higher and lower expressions in N7VL than in N7NL, respectively (Figure 6A). The number of upregulated and downregulated genes is consistent with the observed number of DAMs. GO enrichment showed that the DEGs/DETs were mostly associated with biological processes (chitin response, small-molecule catabolism, organic acid catabolism, carboxylic acid catabolism, and chitin metabolism), molecular function (hydrolase activity, oxidoreductase activity, and tetrapyrrole binding), oxidoreductase activity, tetrapyrrole binding, heme binding, catalytic activity, chitin binding, aconite hydratase activity, 3-isopropylmalate dehydratase activity, and cellular component (cell periphery, plasma membrane, 3-isopropylmalate dehydratase complex, and the intrinsic component of the plasma membrane) (Supplementary Figure S1A). The DEGs/DETs were enriched in the GO process associated with polysaccharide metabolism processes, followed by hormone-mediated signaling pathways, protein secretion, protein localization, and glucan-mediated processes (Figure 6B). The DEGs/DETs were enriched in KEGG pathways associated with carbon metabolism; phenylpropanoid biosynthesis; biosynthesis of amino acids; terpenoid backbone biosynthesis; pentose and glucuronate interconversions; cutin, suberin, and wax biosynthesis; and carotenoid biosynthesis (Figure 6C). These observations are consistent with those of metabolome profiles of the two fruit types.

As the metabolite profiles indicated differences in carbohydrate-related compounds, we filtered the genes enriched in related pathways. Two genes enriched in the carbon fixation pathway in photosynthetic organisms, i.e., LITCHI027517, sedoheptulose bisphosphatase, and LITCHI014769, photosystem I subunit II, showed lower expressions in N7VL than in N7NL. In addition, 31 transcripts annotated as 21 genes enriched in carbohydrate biosynthesis-related pathways showed differential expression in the leaf samples of the two cultivars. In particular, we found that genes associated with the biosynthesis of D-glucose-6-P, D-fructose, trehalose, D-glucose, ADP-glucose, starch, and dextrin had higher expressions in N7VL than in N7NL. In the case of N7NL, transcripts associated with the biosynthesis of D-glucose, cellobiose-6-P, sucrose-6-P, dextrin, and maltose had higher expressions. Taken together, these observations indicate that both leaves had different expressions for carbohydrate biosynthesis-related genes, and the observed content of carbohydrate-related metabolites in N7VL may be related to these expressions (Supplementary Table S2).

In N7VL, we observed higher expressions of 5-O-(4-coumaroyl)-D-quinate 3′-monooxygenase (C3′H, LITCHI008559), consistent with higher shikimic acid and caffeic acid O-glucoside contents. In contrast, the increased accumulation of DAMs such as flavonoids and anthocyanins could be due to higher expressions of caffeoyl-CoA O-methyltransferase (LITCHI017676, LITCHI017674, and LITCHI017673) and isoflavone/4′-methoxyisoflavone 2′-hydroxylase (CYP81E, LITCHI006751) in N7NL. Similarly, consistent with the lower amino acid content in N7VL, we found that S-sulfo-L-cysteine synthase (LITCHI003294) and glycine hydroxymethyltransferase (glyA, LITCHI007175) had higher expressions in N7NL. However, several other transcripts associated with the GO term “vitamin” had higher expressions in N7VL, suggesting that more in-depth studies will be required to define the role of these genes and leaf-specific traits. Next, we searched for transcripts/genes enriched in amino acid-related pathways, as in general, amino acids and derivatives were highly accumulated in N7NL compared to N7VL. Consistent with the metabolomic observations, 19 out of 28 transcripts were highly expressed in N7NL compared to N7VL. In the case of transcripts enriched in lipid and related pathways, 7 out of 13 genes were highly expressed in N7NL compared to N7VL, confirming the metabolomic observations. Finally, two of the five genes associated with organic acid-related GO terms, i.e., PYR/PYL (LITCHI018118) and ATP-binding cassette, subfamily B (MDR/TAP), and member 1 (LITCHI026684), had higher expression in N7NL. In contrast, those highly expressed in N7VL were associated with a carbon pool through the folate pathway. These observations confirmed the metabolome-based results (Supplementary Table S2).

Transcript screening revealed 2,699 DEGs/DETs, out of which 1,547 and 1,152 had lower and higher expressions, respectively, in N7VF than in N7NL (Figure 6D). The number of upregulated and downregulated genes is consistent with the observed number of DAMs. GO enrichment showed that the DEGs/DETs were mostly associated with biological processes (cellular, metabolic, developmental, signaling, reproductive, growth, and pigmentation-related processes), molecular function (catalytic, binding, transporter, transcriptional regulation, ATP-dependent, antioxidant, and nutrient reservoir activities), and cellular component (cellular anatomical unit and protein-containing complex) (Figure 6E; Supplementary Figure S1B). DEGs/DETs were enriched in secondary metabolite, phenylpropanoid and flavonoid biosynthesis, and the phytohormone and MAPK signaling pathway (Figure 6F).

As fruit development is influenced by hormone levels and signaling, we searched for DEGs/DETs enriched in related pathways. Eleven out of 37 transcripts enriched in plant hormone signaling and plant MAPK pathways had higher expressions in N7VF than in N7NF. These transcripts were related to auxin (AUX1, AUX/IAA, and GH3), cytokinin (CRE1), GA (PIF3/PIF4), ABA (PYR/PYL), and JA (MYC2) signaling. Interestingly, those with higher expressions in N7NF than in N7VF were also associated with auxin (TIR1, AUX/IAA, ARF, and GH3), cytokinin (B-ARR), GA (DELLA), ABA (PYR/PYL), ethylene (ETR and ERF1/2), BR (BIN2 and TCH4), JA (MYC2), and SA (PR-1) signaling. These observations suggest that hormones may play a role in the differences in leaf and fruit phenotypes (Supplementary Table S2). Considering the observation that N7VF had a higher total carbohydrate content, we searched for genes associated with pathways related to carbohydrate biosynthesis. The transcripts were annotated as trehalose-6-phosphate synthase (ostA, LITCHI020469), glucose-1-phosphate adenylyltransferase (glgC, LITCHI014739), endoglucanase (EG, LITCHI001925 and LITCHI022015), beta-glucosidase (BGLU, LITCHI024092), and glucan endo-1, 3-beta-D-glucosidase (EGLC, LITCHI017822). In contrast, those involved in the biosynthesis of D-glucose-6P, D-fructose, trehalose, D-glucose, and cellobiose had higher expression in N7NF, indicating that a complex network of genes is involved in the differential carbohydrate biosynthesis in the two fruit types. There were also DEGs/DETs enriched in galactose metabolism, fructose and mannose metabolism, gluconeogenesis, galactose metabolism, and pentose and glucuronate interconversion. Thus, the observed differences in carbohydrate content between the two fruits also involve these pathways.

Twenty-four out of the 45 DETs enriched in pathways related to amino acid biosynthesis had higher expressions in N7NF than in N7VF. The others had higher expressions in N7VF than in N7NF. These pathways include metabolism pathways such as glutathione metabolism, cyanoamino acid metabolism, amino acid biosynthesis, valine, leucine, and isoleucine biosynthesis, and phenylalanine, tyrosine, and tryptophan metabolism. The higher content of lipid-related metabolites is consistent with the higher expression of 12 out of 21 DETs, which are enriched in glycerolipid metabolism, ether lipid metabolism, and glycerophospholipid metabolism. As most organic acid-related metabolites had higher levels in N7NF than in N7VF, we observed consistent gene expression. Sixty-three of the 117 DEGs/DETs associated with the GO term “organic acid-related functions,” showed higher expressions in N7NF than in N7VF (Supplementary Table S2).

Since N7NF had a higher flavonoid content, we looked for DETs enriched in related pathways. In particular, caffeoyl-CoA O-methyltransferase (CCOMT, LITCHI017676, and LITCHI001586) and C3′H (LITCHI008559) had higher expression in N7NF. On the other hand, N7VF had higher levels of several compounds such as flavonoids including luteolin 7-O-glucoside, syringetin 7-O-hexoside, keracyanin chloride, and typhaneoside. These higher levels may be due to the relatively higher expression of chalcone isomerase (CHI, LITCHI027595), chalcone synthase (CHS, LITCHI020852 and LITCHI015108), CCOMT (LITCHI017672), anthocyanin reductase (ANR, LITCHI029356), naringenin 3-dioxygenase (F3H, LITCHI006477), and others. Overall, the transcriptome data are consistent with the metabolome profiles of the two fruits (Supplementary Table S2).