Two Chinese chestnut cultivars, high-sugar cultivar ‘Yanlong’ (HS) and relatively low-sugar cultivar ‘Yanshanzaofeng’ (LS), were used as plant materials. The nuts of two Chinese chestnut cultivars were collected from the six 11-year-old trees at the Chinese chestnut base in Hebei, China (118°80′E, 40°13′N). Each tree received standard agronomic practices. ‘Yanshanzaofeng’ (LS) has female flowers in full bloom on June 5, and ‘Yanlong’ (HS) has female flowers in full bloom on June 25. Flowers that were at the anthesis stage simultaneously were marked, and fruits were collected at 60, 70, 80, 90, and 100 DAF (fruit maturity) in 2021. At each stage of development, nine nuts were mixed into a biological repetition from the selected three trees, with three repetitions per each stage, and then samples were quickly frozen with liquid nitrogen and stored in a refrigerator at -80°C.

The freeze-dried samples were ground into powder with a mixer mill (MM 400, Retsch) by filtration through a 0.5 mm filter. The sucrose content (about 50 mg of sample powder) was determined using the anthrone colorimetric method (Roggo et al., 2002). The contents of starch components (amylose and amylopectin) were determined through dual-wavelength spectroscopy, with amylopectin measured at 550 nm and 695 nm and amylose measured at 617 nm and 475 nm, using a spectrophotometer (UV-2102C) (Juliano et al., 1981). All the determinations were performed in triplicate.

Agilent 8890 gas chromatograph coupled to a 5977B mass spectrometer with a DB-5MS column (30 m length × 0.25 mm i.d. × 0.25 μm film thickness, J&W Scientific, USA) was employed for GC-MS analysis of sugars by MetWare (http://www.metware.cn/). Helium was used as carrier gas, at a flow rate of 1 mL/min. Injections were made in the split mode with a split ratio 5:1 and the intection volume was 1 μL. The oven temperature was held at 160°C for 1 min, and then raised to 200°C at 6°C/min, raised to 270°C at 10°C/min, raised to 300°C at 5°C/min, raised to 320°C at 20°C/min and held at the temperature for 5.5 min. All samples were analyzed in selective ion monitoring mode. The ion source and transfer line temperature were 230°C and 280°C, respectively (Lowell et al., 1989; Medeiros and Simoneit, 2007; Gomez-Gonzalez et al., 2010).

Total RNA was extracted from 0.5 g sample using RNAprep Pure Plant Kit (Tiengen, Beijing, China), and RNA purity was detected using a NanoPhotometer® spectrophotometer (IMPLEN, CA, USA). cDNA libraries were prepared using the NEBNext®Ultra™RNA Library Prep Kit (Illumina, San Diego, CA, USA). The NovaSeq 6000 sequencing system (Illumina) was used for sequencing with 150 bp paired-ends by Novogene Company (Beijing, China) (Modi et al., 2021). The raw data of each sample was more than 6 GB. RNA-seq data were uploaded to NCBI and can be accessed through BioProject accession number PRJNA883560.

The Chinese chestnut reference genome and gene model annotation files were downloaded from the website (https://github.com/yongshuai-sun/hhs-omei) (Sun et al., 2020). Clean reads were obtained by filtering raw reads using Perl scripts. Then, the clean reads were compared with the reference genome using Hisat2 v2.0.5 software, and the number of reads mapped to each gene was counted by features v1.5.0-p3 in the subread software to obtain the FPKM value. The differentially expressed genes (DEGs) were analyzed by OmicShare tool (www.omicshare.com/tools ) (|log2 (fold change)|≥1, FDR ≤ 0.05). KEGG enrichment analysis was performed using STRING database (http://www.string-db.org). Gene expression data were normalized and plotted using Tbtools V1.09876 software.

SUS was assayed in both the synthetic and cleavage directions. One gram of the frozen powder was resuspended at 4°C in 5 ml of 100 mM HEPES (pH 7.5), 2 mM EDTA and 5 mM dithiothreitol. The medium for SUS synthesis contained in a 0.5 mL volume: extract, 80 mM Hepes (pH 8.5), 5 mm KCN, 5 mm NaF, 100 mM fructose, and 15 mM UDPG. The medium for SUS cleavage in a similar volume consisted of extract, 80 mM MES (pH 5.5), 5 mM NaF, 100 mM sucrose, and 5 mM UDP. Reactions proceeded for 15 min at 30°C and were terminated by boiling for 1 min. UDP production was quantified by measurement of pyruvate kinase-specific loss of NADH in the presence of lactic dehydrogenase, and UDPG production was quantified by measuring UDPG dehydrogenase-specific synthesis of NADH (Baroja-Fernández et al., 2009).

Weighted gene co-expression network analysis (WGCNA) (V1.69) in R software package was used to construct the gene co-expression network, using the signed-hybrid network type. The co-expression network was mapped using Cytoscape V3.7.1 (https://cytoscape.org/) software. The description of gene function comes from the STRING database (Szklarczyk et al., 2019).

Real-time quantitative PCR experiments were performed on ABI 7500 Real-Time PCR system (Applied Biosystems Inc., Foster City, CA, USA) with TB Green Premix Ex Taq (Takara). The instrument settings were: 95°C for 300 s; 40 PCR cycles, with each cycle set at 95°C for 10 s and 60°C for 30 s. The specific primer information was shown in Table S1 , in which the 18S gene of Chinese chestnut was used as the reference gene. The relative expression levels were calculated using the 2^-ΔΔCt method. Three biological replicates were performed.

We have sampled five nut development stages of the two cultivars under the same site conditions. The mature stage of LS was 20 days earlier than HS, and the development temperature of HS is lower than that of LS ( Figure 1A ). The accumulation of nutrients in two cultivars was mainly carried out from 60 to 100 DAF. At 100 DAF, the dry weight of HS was 6.77g, which was 0.49g heavier than in LS ( Figure 1B ). The changes of amylose (Figure 1C), amylopectin (Figure 1D), ratio of amylopectin to amylose (Figure 1E) and total starch (Figure 1F) contents in the two cultivars were similar. It is worth noting that the soluble sugar of HS is 9%, which was 1.5 times higher than LS at the mature stage ( Figure 1G ). The soluble sugar in HS increased rapidly from 6.46% to 6.84% at 90-100 DAF, and the amylopectin decreased from 56.03% to 44.77%. The increase of soluble sugar in HS from 90 to 100 days was due to the decomposition of amylopectin. The difference of temperature and daylength during nut development may be one of the reasons for the difference in soluble sugar content in two cultivars.

We tested a total of 32 kinds of sugar metabolites, of which 30 kinds of metabolites were detected, and the two metabolites (D-Arabinitol and L-Rhamnose) were not detected ( Table 1 ). The content of the total sugar metabolites was related to soluble sugar content ( Figure 2B ), indicating that the experimental data was reliable. The results showed that the soluble sugar composition changed obviously with embryo development. The soluble sugar was mainly sucrose, glucose and fructose at 60-70 DAF ( Figure 2A ). In addition, the content of glucose ( Figure 2D ) and fructose ( Figure 2E ) was high (> 87.01 mg/g) at 60 DAF but low (<3.68 mg/g) at 80-100 DAF. In mature chestnuts, the soluble sugar was mainly sucrose (> 38.67 mg/g), and the content of other sugar metabolites was very low (<3.30 mg/g). At 90-100 DAF, the sucrose content in HS was 1.5 times higher than that in LS (Figure 2C). In addition, the content of raffinose (Figure 2F), 1,5-Anhydroglucitol (Figure 2H) and maltose (Figure 2I) was higher in HS than that in LS at mature stage. The content of inositol (Figure 2G) was higher in LS than that in HS at mature stage. During ripening, the increased content of these soluble sugar metabolites will make the chestnuts sweeter.

Thirty libraries from Chinese chestnut embryos were sequenced. We obtained a total of 1.41 billion base pairs, with an average of 47,036,059 raw reads and 45,312,612 clean reads per sample. The average ratio of clean reads to raw reads was 96.33% ( Table S2 ). The clean reads are made freely available in the NCBI (accession number: PRJNA883560).

A total of 42,740 unigenes were identified from transcriptome data. Principal components analysis (PCA) of all 30 samples was conducted based on RNAseq FPKM, and two principal components were found to explain 39.9% of the overall variance (27.2% and 12.7% for PC1 and PC2, respectively). The three samples at the same stage were close to each other, indicating that there was a high consistency between the three biological replicates ( Figure 3A ). A total of 14,430 DEGs were identified by pair comparison between two cultivars at each stage ( Figure 3B ). The cluster analysis of DEGs was shown in Figure 3C . The results showed that the 30 samples could be divided into three groups: LS60 and HS60 constituted the first group, LS70, LS80, LS90, HS70 and HS80 constituted the second group, and LS100, HS90 and HS100 formed the third group.

In order to further understand the mechanism of sugar synthesis in Chinese chestnut embryos, we focused on the expression trend of DEGs. The 14,430 DEGs were divided into five modules using WGCNA ( Figure 4A ), and the number of DEGs and KEGG pathways of each module were listed in each module. The largest module (turquoise) contained 6489 DEGs whose expression was highest in the LS60 sample, which included genes related to glycolysis/gluconeogenesis, brassinosteroid biosynthesis, and flavonoid biosynthesis. The second largest module (blue) contained 3484 DEGs whose expression was the highest expression in the LS100 sample, which included genes related to glycolysis/gluconeogenesis, glycerolipid metabolism, limonene and pinene degradation, and flavone and flavonol biosynthesis. The third largest module (brown) contained 1985 DEGs whose expression was the highest expression in the HS60 sample, which included genes related to photosynthesis, oxidative phosphorylation, nitrogen metabolism, tyrosine metabolism.

In order to understand the relationship between the gene expression patterns and metabolites, association analysis was performed using WGCNA ( Figure 4B ). Sugar content was highly correlated with turquoise and brown modules. Combined with KEGG pathway and module-trait correlation analysis, turquoise, blue and brown modules were associated with sugar biosynthesis.

The sucrose produced by photosynthesis will be stored in the chestnut as starch. As shown in Table S3, a total of 66 unigenes related to starch synthesis were identified, of which 15 belong to the turquoise module, 9 belong to the blue module, and 4 belongs to the other module. Most of the genes related to starch synthesis were highly expressed at 70 DAF. Some genes such as SUS, GPT and GBSS had FPKM values >1000 in our transcriptome data. These highly expressed genes may play an important role in starch synthesis in Chinese chestnut embryos. The variation trend of these genes related to starch synthesis was similar to that reported in previous articles (Zhang et al., 2015; Shi et al., 2021), indicating that the regulation of these genes in chestnut cultivars was similar.

At 70-80 DAF, the expression level of genes related to starch synthesis (i.e., SUS, PGI, UGP, PGM and AGP) was higher in LS than that in HS ( Figure 5 ). At 100 DAF, the expression level of genes related to starch synthesis (i.e., HXK, PGI, PGM, AGP, SS, SBE, ISA) was higher in HS than that in LS ( Figure 6 ). This indicates that the sugar metabolism of HS was more active than that of LS at mature stage.

Starch degradation during ripening is a key additional process for sugar accumulation and sweetness in fruit (Liu et al., 2021). As shown in Table S4 , a total of 33 unigenes related to starch degradation were identified, of which 6 belong to the turquoise module, 6 belong to the blue module, and 3 belongs to the other module. Some genes such as PWD, BAM, PHS1, GLT, PHS2 had FPKM values >100 in our transcriptome data. These highly expressed genes may play an important role in starch degradation in Chinese chestnut embryos. At 100 DAF, the expression level of genes related to starch degradation (i.e., GWD, PWD, LDA, ISA, PHS1, MEX1) was higher in HS than that in LS ( Figure 6 ). This indicated that starch degradation of HS was more intense than that of LS at mature stage, which was consistent with soluble sugar content.

It is worth noting, the expression level of GWD genes was higher in HS than that in LS at 70-100 DAF. GWD catalyzes the phosphorylation of amylopectin to form phosphoglucan, which is a key enzyme in the degradation of plant starch (You et al., 2020). In rice research, GWD can improve the yield and quality characteristics, which has great application potential (Wang et al., 2021). Therefore, GWD may play an important role in starch degradation of chestnut at mature stage.

Sucrose synthase (SUS) is the key enzyme of sucrose metabolism, which has the function of decomposition and synthetic sucrose (Stein and Granot, 2019). In this study, we identified a total of 8 SUSs in Chinese chestnut, including three SUS2, one SUS3, two SUS4, and two SUS6 genes. Among these 8 SUSs, 2, 4, and 2 genes belong to the SUSI subfamily, SUSII subfamily, SUSIII subfamily, respectively ( Figure 7A ). The SUS2 (bl_022590, bl_022593 and bl_022594), SUS3 (bl_024378), and SUS4 (bl_040354) genes were highly expressed (FPKM> 100) ( Table S3 ), which may play an important role in sucrose content during Chinese chestnut embryo development.

The SUS6 gene were lowly expressed, but the SUS2 were highly expressed in developing Chinese chestnut embryos. The expression heat map revealed that three SUS2 genes (bl_022590, bl_022593 and bl_022594) could be grouped into a category, with the high expression levels observed at 60-90 DAF ( Figure 7B ). It is noteworthy that the expression level of SUS3 (bl_024378) was higher in HS than that in LS at 90-100 DAF.

In addition, we measure the enzyme activity of SUS. The enzyme activity of SUS-cleavage continues to decrease during embryo development, and there is no significant difference between the two cultivars ( Figure 7C ). However, the enzyme activity of SUS-synthetic was significantly different between the two cultivars. The enzyme activity of SUS-synthetic was higher in LS than that in HS at 70-80 DAF, while was higher in HS than that in LS at 90-100 DAF. The enzyme activity of SUS changes were consistent with changes of sucrose content in the two cultivars.

One study showed that after 48 h cold stress treatment of Arabidopsis thaliana (Baslam et al., 2017), the expression levels of AtSus2, AtSus4, AtSus5 and AtSus6 were not affected, while the expression levels of AtSus1 and AtSus3 were 24.5 times and 5.5 times of those in the untreated group, respectively. The changes of these genes were consistent with our results, and cold stress may affect SUS3 (bl_024378) gene expression, and then change the enzyme activity of SUS-synthetic in Chinese chestnut.

Some studies have shown that cold stress can promote the increase of soluble sugar in plants (Zhang et al., 2011; Ding et al., 2019) ( Figure 8A ). In addition, COR is a key gene in response to cold stress, which encoded hydrophilic peptide can protect the cells from freezing damage. Our data showed that the expression level of COR gene was higher in HS than that in LS at 60-100 DAF. Notably, COR (bl_030433) and GWD (bl_017818) were co-expressed, which may reflect the correlation between starch degradation and low temperature. Then, we analyzed the 412 genes co-expressed with COR (bl_030433) by string analysis ( Figure 8B ). There were 9 genes associated with ABA signaling pathway and 8 genes associated with peroxisome. ABA and peroxide may play an important role in starch decomposition during Chinese chestnut ripening.

Therefore, we also analyzed ABA content in two Chinese chestnut cultivars during embryo development ( Figure 8C ). At 90 DAF, ABA content in HS was higher than that in LS. Correspondingly, the expression level of genes related to ABA (i.e., PYL, PP2C and SnRK2) was also higher in HS than in LS ( Figure 8A ). In addition, the expression of CSLA gene related to polysaccharide synthesis was higher in HS than in LS at 80-100 DAF. Some studies have also shown that ABA plays an important role in promoting starch degradation and sucrose synthesis in fruits. Exogenous ABA treatment could promote mango ripening and increase soluble sugar (Wu et al., 2022).

In order to verify the relative expression pattern of unigenes in RNA-seq analysis, twelve key genes related to sugar metabolism were analyzed by qPCR (Panels A-L) ( Figure 9 ). Panel M shows that RNA-seq is highly correlated with qPCR data (R² = 0.74, p <0.01), indicating that the expression data obtained by RNA-seq is reliable. Among all 12 genes, the expression level of 10 genes (HXK, PGI, PGM, GPT, GBSS, SS, SBE, ISA, GWD and CLSC20) was higher in HS than that in LS at 100 DAF, indicating that the active sugar metabolism was related to the increase in the content of soluble sugar in LS at the mature stage.