Kiwifruit development
The ‘Ganmi 6’ kiwifruit samples used as the test materials were selected at eleven developmental stages from initial fruit appearance at 20 DAF to its full maturity at 170 DAF (Fig. 2a). S1-S6 is the first rapid growth stage of fruit (Fig. 2b) and the seeds have been brown (Fig. 2a), S8-S10 is the second rapid growth stage of fruit (Fig. 2b). AsA content was also determined, it was found that AsA began to accumulate and its content increased at the early stage of fruit development, reached the first peak of AsA accumulation at S2 (7.56 mg·g− 1). Subsequently, AsA content continued to decline until S7, the decline was interrupted and reached the second peak (7.63 mg·g− 1). AsA content resumed its downward trend at S8 and has a steep decrease at S9, then decreased until it’s harvested (Fig. 2c). The dynamic changes of T-AsA content and AsA/DHA (Fig. 2d) was basically consistent with that of AsA content, the DHA content were lower throughout development (Fig. 2c).
Activity of AsA related metabolic enzymes
The activity of dehydroascorbate reductase (DHAR) was higher than that of other enzymes in the whole development period, between 35.38–70.04 U·g− 1 FW, and the first and second peak of DHAR activity coincided with the peak of AsA (Fig. 3c). D-galacturonic acid reductase (GalUR) activity was between 11.04–36.05 U·g− 1 FW, reached the maximum value (36.05 U·g− 1 FW) at S8, and then declined rapidly until GalUR activity reached the minimum value (11.04 U·g− 1 FW) at S10 (Fig. 3b). The lowest activity of ascorbic peroxidase (APX) corresponds to the first peak of AsA, and with the increase of ascorbic acid content, the activity of APX decreases during S2-S5, with the ripening of fruit, AsA content decreased and APX activity increased (Fig. 3d). The activity trend of L-galactose dehydrogenase (GalDH, Fig. 3a) and L-galactono-1,4-lactone dehydrogenase (GalLDH, Fig. 3a) was similar to that of AsA content, the range was 3.76–8.21 and 1.65–3.81, respectively. The changes of monodehydroascorbate reductase (MDHAR, Fig. 3c) and L-ascorbate oxidase (AAO, Fig. 3d) activities were relatively small during fruit development, were 3.75–7.73 and 0.26–1.39 respectively. In addition, there was a similar trend between the MDHAR and AsA.
Sequencing, basic transcriptome assembly, and functional annotation
We obtained 98,565 unigenes with an average length of 932 bp, the longest unigenes sequence length is 16,709 bp, the shortest unigenes sequence length is 201 bp, and N50 of 1609 bp (supplementary Table 2). The length distribution of unigenes (supplementary Figure 1) showed that the number of genes decreased gradually with the increase of gene length.
The 98,656 unigenes obtained by the assembly were aligned to Nr, SwissProt, KOG and KEGG databases. A total of 50,184 unigenes were annotated, with an annotation rate of 50.87% (supplementary Figure 2).
Of the 98,656 unigenes, 88,867 (90.08%) were annotated with GO terms that were offered by the GO-based annotation (an internationally standardized gene functional classification system) as a strictly defined conceptualization for comprehensively describing the properties of genes and their products within any organism. The 88,867 unigenes were classified into three functional categories: biological process, cellular component, and molecular function (supplementary Figure 3).
A total of 28,919 unigenes were annotated in the KOG database, and these unigenes were divided into 25 categories. Among them, general function prediction only (R), posttranslational modification (O) and signal transduction mechanisms (T) were the dominant categories (supplementary Figure 4).
KEGG pathway analysis
A total of 31,069 unigenes were annotated into the KEGG database and assigned to the following five KEGG biochemical pathways (supplementary Figure 5). We pay more attention to AsA and aldehydic acid pathway, which enriched 133 unigenes, these unigenes were involved in ASA metabolism.
Screening of DEGs
The focus of the present study is that the decreasing trend of AsA content was interrupted at S7. Three group comparisons consisting of S6 vs S7, S6 vs S8, S7 vs S8 were determined to identify the DEGs. As can be seen from supplementary Figure 6, the down-regulated genes were more than the up-regulated genes in the three periods of S6, S7 and S8. AsA content in fruit depends on its biosynthesis ability and degradation recycling level. AsA biosynthesis and degradation are catalyzed by a series of enzymes. The level of gene expression that encodes these enzymes directly determines the amount of AsA synthesis or degradation. Finally, according to the DEGs among S6 vs S7, S6 vs S8 and S7 vs S8, the 23 unigenes related to AsA anabolism and with high expression were screened out (supplementary Table 3). These unigenes are mainly distributed in the AsA biosynthetic pathway including L-galactose pathway, D-galacturonic acid pathway and inositol pathway as well as the AsA circulation pathway.
Expression profiles of 23 DEGs, cluster analysis and correlation analysis
The qRT-PCR data of 23 candidate genes were compared with transcriptome data, we found that the trends were basically the same. The gene expression involved in the AsA synthesis pathway was shown in Fig. 4, in the L-galactose pathway, PGI1 expression was higher in early and late fruit development (supplementary Figure 7 & Fig. 4a); the relative expression level of PMI1 was higher than that of PMI2 throughout development (Fig. 4b and c); the relative expression level of PMM and GPP1 was similar and generally low throughout development (Fig. 4d and h); the relative expression level of GME was higher in the early stage of fruit development and showed a downward trend as a whole (Fig. 4e); the relative expression level of GGP2 is higher than that of GGP1, and the relative expression level of GGP1 generally shows a downward trend (Fig. 4f and g); the expression trends of GalDH and GalLDH were basically similar, generally showing a descending - then ascending - and then descending trend, and the expression levels of both were higher in early fruit development (Fig. 4i and j). As for the D-galacturonic acid pathway, the relative expression level of GalUR1 was higher than that of GalUR2 in the whole development stage, the relative expression level of GalUR1 is the highest at S7, and the relative expression level of GalUR2 was almost undetectable after S3 (Fig. 4k and l). L-gulonolactone oxidase gene (GuLO) is the structural gene in the last step of AsA synthesis in L-gullosugar pathway, the expression of GuLO6 shows a downward trend during the whole development period, and the expression level of GuLO6 was significantly decreased after S7 (Fig 4m). Finally, the relative expression levels of MIOX1 and MIOX2 in the inositol pathway were low (Fig. 4n and o).
In AsA cycle pathway, the expression of MDHAR fluctuated in early fruit development and decreased in late fruit maturity (supplementary Figure 8 & Fig. 5a); the relative expression of DHAR2 showed an overall downward trend, and increased at S7(Fig. 5b); the relative expression of DHAR3 was lower during the whole fruit development stage (Fig. 5c); the relative expression levels of APX1 and APX5 were consistent, lower in the early and late-stage and higher in the middle stage of fruit development (Fig. 5d and g); the relative expression levels of APX2, APX3 and AAO showed a downward trend, in which the relative expression levels of AAO could hardly be detected in the later stage of fruit development (Fig. 5e, f and h).
Correlation analysis of metabolism components, related enzymes and genes of AsA
It can be seen from supplementary Table 4, AsA content in fruit was significantly negatively correlated with DHA content (R2 = − 0.65), and was significantly positively correlated with T-ASA (R2 = 1.00) and AsA/DHA (R2 = 0.89), respectively; DHA content was significantly negatively correlated with T-ASA (R2 = − 0.59) and AsA/DHA (R2 = − 0.85), respectively; and T-ASA and AsA/DHA (R2 = 0.85) were significantly positively correlated.
GalDH and GalLDH in the synthetic pathway and MDHAR and DHAR in the regenerative cycle pathway were significantly correlated with AsA content, and the correlation coefficients were 0.73, 0.73, 0.68 and 0.89, respectively. DHA was negatively correlated with the activity of MDHAR (R2 = − 0.61) and DHAR (R2 = − 0.60). The activity of T-ASA was significantly positively correlated with GalDH (R2 = 0.74), GalLDH (R2 = 0.74) and DHAR (R2 = 0.88), while the activity of MDHAR (R2 = 0.66) was significantly positively correlated with T-ASA.
GME, GGP1 and GalLDH in L-galactose pathway were significantly correlated with the expression levels of GalUR1 in D-galacturaldehyde pathway and MDHAR5, DHAR2 and AAO in AsA regeneration cycle pathway, with correlation coefficients of 0.67, 0.61, 0.58, 0.87, 0.63, 0.60 and 0.59, respectively.