Comparative transcriptomic analysis of two Cucumis melo var. saccharinus germplasms differing in fruit physical and chemical characteristics

To compare the fruits of the two assessed Hami melon varieties, we examined the physicochemical characteristics of the fruits collected at different stages of development (Fig. 1). The color of the fruit peel of the two varieties was similar at 5 and 10 days after pollination (DAP). However, the peel of the Yaolong fruit turned darker green than that of Guimi from 15 to 20 DAP. With continued growth, Guimi melons developed into medium-sized oval fruits with yellow skin. In both varieties, juicy flesh underwent a change in color from white to yellow during ripening.

Fruits of the Guimi and Yaolong varieties of Hami melon at different stages of development. For each sampling time image, Guimi and Yaolong fruits are depicted on the left- and right-hand sides, respectively

Trends in changes in the physicochemical characteristics of the fruits, including weight, size, and soluble sugar content, are shown in Fig. 2. Similar trends were observed in the two varieties with respect to the weight and size of the fruit during ripening. Interestingly, fruit weight increased rapidly from 35 to 40 DAP, whereas there was only a slight increase in fruit size. At 40 DAP, the soluble sugar content of Guimi fruit reached a value of 12, which was higher than that obtained for Yaolong, although differences in the Brix values of the two varieties were non-significant.

Differences in the weight, size, and soluble sugar contents of fruits of the Guimi and Yaolong varieties of Hami melon. Trends in weight (A) and size (length, diameter) (B) at 5, 10, 15, 20, 25, 30, 35, and 40 days after pollination (DAP). (C) Total soluble sugar contents at 40 DAP. Data represent the means of three individual replicates. The bars denote standard error values (n = 3)

To determine the gene expression patterns, we performed RNA-seq analysis using the C. melo reference genome. After filtering out the rRNAs and low-quality reads, 111 million reads were mapped to the reference genome (Table S1). For these clean reads, we obtained an average mapped read per sample of greater than 90%. One exception was the low (74.44%) alignment ratio obtained for Yaolong at 40 DAP. In total, we detected 21,172 expressed genes in the melon fruit samples.

To determine the differences in the expression of genes between two close sampling times, we identified the DEGs in Guimi and Yaolong based on the threshold criteria of a log2 fold change ≥ 1 and FDR ≤ 0.05. For this purpose, we defined the earliest time point (10 DAP) of the paired groups as the control sample for subsequent measurements. The number of DEGs in Guimi was found to be markedly higher than that in Yaolong at 20 and 30 DAP, whereas fewer DEGs were identified in Guimi than Yaolong (223 vs. 95) at 40 DAP (Fig. 3A, Table S2). These findings indicate that during fruit development, the constituents of Guimi fruit undergo more pronounced changes than those of Yaolong fruit. However, over time, there was a gradual reduction in the number of DEGs in both melon varieties during fruit development, indicating a corresponding reduction in the speed of fruit growth and that the fruit was fully mature at 40 DAP. For both Guimi and Yaolong, we detected perturbations in gene expression at the same time points. The number of genes differentially expressed between Guimi and Yaolong initially increased, reaching a peak at 20 DAP, after which the number declined to 458 at 40 DAP (Fig. 3B). The majority of the top 30 significantly altered genes were downregulated in the comparisons between Yaolong and Guimi at 10, 20, and 40 DAP (Table S3, S4, S5, S6). We observed that caffeoyl-CoA O-methyltransferase-like (At1g67980), which is involved in the reinforcement of the plant cell wall, was more highly expressed in Yaolong during fruit development compared with the expression in Guimi.

Distribution of differentially expressed genes (DEGs) at different time points after pollination during melon fruit development and ripening. (A) The number of DEGs in different samples. Light coral and medium turquoise colors denote DEGs in the Guimi and Yaolong varieties, respectively. (B) Distribution of DEGs between Guimi and Yaolong at specific sampling times

For functional annotation, the DEGs identified in the two melon varieties were assigned to GO terms and KEGG pathways. We found that for Yaolong melons, the majority of enriched GO terms between 10 and 20 DAP were involved in biological processes, including ‘histone lysine methylation,’ ‘peptidyl-lysine methylation,’ and ‘DNA alkylation,’ whereas for Guimi melons, the most significantly enriched GO terms were ‘polysaccharide metabolism,’ ‘cell wall biogenesis,’ and ‘external encapsulating structure organization’ (Fig. S1 and S2). For the second comparison (20 DAP vs. 30 DAP), the DEGs in Guimi melons were found to be significantly enriched in the processes of ‘response to acid chemical,’ ‘plant-type cell wall biogenesis,’ and ‘response to chemical,’ whereas for Yaolong, the three most enriched biological processes were ‘hemicellulose metabolism,’ xylan metabolism,’ and ‘cell wall polysaccharide metabolism.’ Comparisons between 30 and 40 DAP samples revealed that the DEGs of Guimi melons were notably enriched for ‘amine metabolism,’ ‘cellular amine metabolism,’ and ‘jasmonic acid metabolism,’ whereas those of Yaolong melons were enriched in the cellular component categories ‘external encapsulating structure,’ ‘cell periphery,’ and ‘vacuole.’ With regards to KEGG pathway annotation, we found that in the 10 to 20 DAP comparison, DEGs in Yaolong were enriched for ‘DNA replication,’ ‘ABC transporters,’ and ‘flavone and flavonol biosynthesis,’ whereas ‘metabolic pathways,’ ‘biosynthesis of secondary metabolites,’ and ‘phenylpropanoid biosynthesis’ were enriched by Guimi DEGs (Fig. S3 and S4). Interestingly, for the 20 to 30 DAP comparison, we observed that for both melon varieties, the three most significantly DEG-enriched KEGG pathways were ‘biosynthesis of secondary metabolites,’ ‘metabolic pathways,’ and ‘phenylpropanoid biosynthesis.’ For the third comparison between 30 and 40 DAP, Guimi DEGs were found to be enriched in pathways such as ‘valine, leucine, and isoleucine biosynthesis,’ ‘biosynthesis of secondary metabolites,’ and ‘cyanoamino acid metabolism,’ whereas pathways enriched with Yaolong DEGs included ‘biosynthesis of secondary metabolites’ and ‘metabolic pathways.’

To gain further insight into the changes in gene expression during fruit development, we clustered 4,731 DEGs from Guimi melons and 3,198 DEGs from Yaolong melons into 38 profiles using the STEM algorithm. Among these, 2,897 Guimi DEGs were significantly clustered into the following six profiles: two upregulated profiles (profiles 17 and 12), three downregulated profiles (profiles 0, 7, and 2), and one biphasic expression pattern profile (profile 18) (Fig. 4A). Similarly, 2,217 Yaolong DEGs were classified into the following six profiles based on P-values ≤ 0.05: two upregulated patterns, one biphasic expression pattern, and three downregulated patterns (Fig. 4B).

Enriched profiles of differentially expressed genes (DEGs) during fruit development. Profiles of Guimi (A) and Yaolong (B) were clustered into three groups, namely Up (upregulated), Bi (biphasic expression pattern), and Down (downregulated). Profile numbers are indicated in the top left-hand corner, and the corresponding P-values for each profile are shown in the bottom left-hand corner. The number of DEGs within each profile is shown in brackets

To systematically investigate the biological functions of candidate genes, we extracted DEGs from the up- and downregulated cluster groups for further GO term and KEGG pathway analyses. GO analysis revealed 18 biological processes significantly enriched by the Yaolong DEGs assigned to profile 17 (Fig. S5A), and the DEGs in profile 19 were strongly categorized into two molecular processes (Fig. S5B), whereas the Guimi DEGs in profile 12 were enriched with respect to 110 major functions in the biological process, cellular component, and molecular function categories (Fig. S5C). However, we detected no significant enrichment of the Guimi DEGs clustered in profile 17 based on the adjusted P-values (Fig. S5D). The GO terms with the highest representation for the downregulated cluster groups are shown in Fig. S6. Among biological functions, ‘gibberellin metabolic process’ (GO:0,009,685), ‘cytoskeleton’ (GO:0,005,856), and ‘cell wall organization or biogenesis’ (GO:0,071,554) were the most significantly enriched functions in Guimi profiles 7, 2, and 0, respectively. For profile 0, 2, and 7 of Yaolong DEGs, ‘cell wall’ (GO:0,005,618), ‘histone lysine methylation’ (GO:0,034,968), and ‘phenylpropanoid biosynthetic process’ (GO:0,009,699), respectively, were the most enriched biological functions.

Based on KEGG analysis, we identified that nine KEGG pathways including ‘metabolic pathways’ (ko01100), ‘biosynthesis of secondary metabolites’ (ko01110), ‘phenylpropanoid biosynthesis’ (ko00940), and ‘biosynthesis of various secondary metabolites—part 2’ (ko00998) were enriched with Guimi DEGs assigned to downregulated profile 0, 2, and 7 (Fig. S7A). In contrast, only two pathways were enriched with Guimi DEGs within profiles 12 and 17, namely ‘carbon fixation in photosynthetic organisms’ (ko00710) and ‘galactose metabolism’ (ko00052) (Fig. S8A). In total, we identified nine pathways including ‘phenylpropanoid biosynthesis’ (ko0094), ‘metabolic pathways’ (ko01100), ‘biosynthesis of secondary metabolites’ (ko01110), ‘phagosome’ (ko04145), ‘purine metabolism’ (ko00230), ‘starch and sucrose metabolism’ (ko00500), ‘DNA replication’ (ko03030), ‘phenylalanine, tyrosine, and tryptophan biosynthesis’ (ko00400), and ‘biosynthesis of amino acids’ (ko01230) with significant enrichment of downregulated Yaolong DEGs (Fig. S7B). Protein processing in the endoplasmic reticulum (ko04141) was highly enriched by Yaolong DEGs in profiles 19 and 17 (Fig. S8B). These results indicate that most DEGs regulated during fruit development appear to be associated with the functioning of metabolic pathways.

Fifty-one DEGs in Guimi, clustered in profiles 0 (n = 28) and 2 (n = 23), were significantly associated with cell wall biogenesis, showing downregulated gene expression patterns; 48 DEGs showing downregulated trends, clustered into profiles 0, 2, and 7, showed similar associations. Figure 5 shows the differences in the expression trends of these DEGs between Guimi and Yaolong. We observed 33 common DEGs in the cell wall biogenesis-related profiles of both melon varieties, the majority of which showed similar patterns of expression regulation at different stages of fruit development and ripening (Fig. 5A). For example, gradual reductions in GUX3 expression levels were observed during fruit development in both germplasms (Fig. 5B and C). A similar reduction in expression was detected for ODO1, which peaked during the early stages of fruit development, after which there was a slight reduction that became more pronounced prior to maturity. Although melon type-specific DEGs involved in cell wall biogenesis showed an overall downregulated pattern, changes in the direction of gene expression were still different between consecutive stages of development. For example, in Guimi melons, UAM1 was downregulated from stages 1 to 2, and then constantly expressed at a stable level from stages 2 to 3, prior to undergoing a decline in the mature stages. These findings indicated differences in the expression patterns of melon cultivar-specific DEGs associated with cell wall synthesis during fruit development.

Trends in the expression changes of key genes associated with cell wall biogenesis in two Hami melon varieties. (A) Common differentially expressed genes (DEGs) in the cell wall biogenesis pathway of the two varieties. DEGs for Guimi (B) and Yaolong (C)

Some of the genes involved in sugar metabolism were found to be differentially expressed during melon fruit development, including those associated with ‘galactose metabolism,’ ‘starch and sucrose metabolism,’ ‘fructose and mannose metabolism,’ and ‘amino sugar and nucleotide sugar metabolism’ (Fig. 6A). However, in the case of Yaolong, only profile 0 DEGs were significantly associated with starch and sucrose metabolism pathways. In contrast, in Guimi, ‘amino sugar and nucleotide sugar metabolism’ were enriched with profile 2 DEGs. For the majority of DEGs associated with the ‘amino sugar and nucleotide sugar metabolism’ pathways, including MUR4, UGD1, and UPTG2, the lowest levels of expression were detected in ripe fruit (Fig. 6B). Differences in gene expression were also detected during different stages of fruit development. For example, there was a marked reduction in the expression of GAUT6 during the middle phase of fruit development, followed by a slight increase in pre-mature fruit, before declining to the lowest levels in mature fruit. Although the trend in UGD1 expression in young fruit was similar to that observed for GAUT6, it differed in that it was expressed at a constant level in the pre-mature fruit. In Yaolong, a total of 11 profile 0 DEGs were significantly associated with starch and sucrose metabolism pathways. In contrast to trends in the expression of sugar metabolism-related genes in Guimi, nine of these DEGs were characterized by gradually downregulated expression during fruit development and ripening (Fig. 6C). For example, the lowest level of At4g02290 expression in Yaolong was detected at stage T3. Conversely, whereas INV1 was characterized by a downregulated expression trend in young fruit, high expression levels were detected during the expansion stage, prior to a subsequent decline in expression, reaching the lowest values in ripe fruit.

Trends in the expression changes of key genes associated with sugar metabolism pathways in the two Hami melon varieties. (A) The number of differentially expressed genes (DEGs) in sugar metabolism pathways in the two varieties. Expression trends of DEGs in Guimi (B) and Yaolong (C)