Multi-omic analysis of the extension of broccoli quality during storage by folic acid

Graphical abstract

A proposed model illustrating the response of broccoli to exogenous application of 5 mg/L folic acid based on an integrated analysis of DNA methylation, transcriptome, and metabolome data. Red word represents a positive effect while green word represents a negative effect.

Highlights

• •Application of FA on the postharvest physiology of fruits and vegetables during storage.
• •FA treatment changed the level of DNA methylation, transcription and metabolites in broccoli.
• •The main methylation type of broccoli is the CHH type.
• •FA treatment affected the methylation level and increased nutraceutical content.
• •FA treatment can delay in yellowing and inhibited off-odor biogenesis.

Abstract

Introduction

Folic acid (FA) is a critical metabolite in all living organisms and an important nutritional component of broccoli. Few studies have been conducted on the impact of an exogenous application of FA on the postharvest physiology of fruits and vegetables during storage. In this regard, the mechanism by which an exogenous application of FA extends the postharvest quality of broccoli is unclear.

Objective

This study utilized a multicomponent analysis to investigate how an exogenous application of FA effects the postharvest quality of broccoli.

Methods

Broccoli was soaked in 5 mg/L FA for 10 min and the effect of the treatment on the appearance and nutritional quality of broccoli was evaluated. These data were combined with transcriptomic, metabolomic, and DNA methylation data to provide insight into the potential mechanism by which FA delays senescence.

Results

The FA treatment inhibited the yellowing of broccoli during storage. CHH methylation was identified as the main type of methylation that occurs in broccoli and the FA treatment was found to inhibit DNA methylation, promote the accumulation of endogenous FA and chlorophyl, and inhibit ethylene biosynthesis in stored broccoli. The FA treatment also prevented the formation of off-odors by inhibiting the degradation of glucosinolate.

Conclusions

FA treatment inhibited the loss of nutrients during the storage of broccoli, delayed its yellowing, and inhibited the generation of off-odors. Our study provides deeper insight into the mechanism by which the postharvest application of FA delays postharvest senescence in broccoli and provides the foundation for further studies of postharvest metabolism in broccoli.

Introduction

Folic acid (FA), including tetrahydrofolate and its derivatives, is a water-soluble vitamin (B9) that is required by both animals and plants [1], [2], [3]. FA is an essential coenzyme involved in one carbon transfer reactions and is involved in the biosynthesis of purine, thymidine, DNA, amino acids, and proteins, as well as in the methyl cycle in animals and plants [4], [5], [6]. Normal FA levels in human blood are 2.7 to 17.0 ng mL⁻¹ [7]. FA-deficiency raises the potential of several human disorders, including cardiovascular disease, cancer, Alzheimer's, neonatal neural tube defects, and megaloblastic anemia [8]. Animals and humans cannot produce FA de novo, thus, they must acquire FA from food sources such as green vegetables [9]. Many fruits and vegetables, however, are low in FA and readily senesce after they are harvested, which rapidly decreases their nutritional value [10], [11], [12]. Newly identified functional roles have been identified for FA in plants. For example, FA has been linked to chlorophyll production, oxidative stress resistance, and the regulation of gene expression via riboswitch processes [13]. In short, the important functional role of FA in the physiology of plants is gradually emerging [14]. However, few studies have been conducted on the impact of an exogenous application of FA on the postharvest physiology of fruits and vegetables during storage.

Broccoli (Brassica oleracea var. Italica) is rich in vitamins, glucosinolates, flavonols, and phenolics to a greater extent than in many other green vegetables [15]. Broccoli, however, has a very short shelf life and once harvested its nutrient content declines rapidly and its florets quickly yellow [16], [17]. In particular, glucosinolates can decrease by >2.5 % during storage, resulting in the generation of offensive off-odors that seriously affect broccoli marketability [10]. The yellowing of broccoli florets that occurs during senescence is a complex process involving many biochemical and physiological pathways, including the degradation of chlorophyll [18], [19]. Several methods are presently used to slow the yellowing and senescence of broccoli, including the use of plant hormones, chemical treatments, and red light-emitting diode (LED) irradiation, [20], [21], [22], [23]. While partially effective, these treatments often do not address the need to maintain high nutritional value. Previous studies have demonstrated that compared to untreated broccoli, a postharvest application of 5 mg/L FA to broccoli delays the loss of chlorophyll, total soluble solids, vitamin C, total phenolics, flavonoids, and glucosinolates that occurs during storage [10]. FA also reduces the accumulation of malondialdehyde (MDA) and reactive oxygen species (ROS) and enhances antioxidant enzyme activity and corresponding gene expression [10]. However, further analysis is needed to better understand the underlying mechanism by which FA maintains the postharvest quality of broccoli.

The use of -omic technologies is an effective approach to comprehensively investigate the physiology, gene expression, and metabolism of plants at various developmental stages, under various natural environments, or in response to the application of different chemical compounds. Therefore, -omic analyses are very relevant to studying the mechanism(s) involved in the ability of FA to delay yellowing and senescence in broccoli during postharvest storage [24]. Transcriptomic and metabolomic analyses enable one to identify quality-metabolite networks associated with the physiological response of plants to various treatments or environmental variables [25]. DNA methylation is a significant epigenetic regulatory process that has been extensively studied in plants [26]. It is age-, species-, tissue- and organelle-specific in plants and plays a regulatory function in all DNA-related processes, including DNA repair, gene transposition, transcription, replication, and cell differentiation [27], [28]. Previous research has shown that changes in the level of DNA methylation in plants play a critical role in the vernalization and flowering of plants, fruit ripening, and environmental stress response [29], [30], [31]. The role of DNA methylation in plant response to abiotic stress has been investigated in many crops, including tomato [26], pineapple [32], wheat [33], and sweet orange [34]. However, few studies on DNA methylation, or other -omic analyses have been conducted on broccoli. Therefore, in the present study, a combination of DNA methylation, transcriptomic, and metabolomic analyses were conducted to identify the transcriptional and metabolomic events that occur in broccoli in response to the exogenous application of FA. Results of the study provide significant new insights on the response of broccoli to FA that can potentially be used to develop novel methods for extending the quality and nutritional value of broccoli during postharvest storage.

Materials and methods

Ethics statement

This study only utilized plants and did not involve animal and human subjects.

Plant material and FA treatment

‘Youxiu’ broccoli heads were harvested in the Shunyi District, Beijing on their optimum harvest date (100 d after planting) and immediately brought to the laboratory. Broccoli heads that were uniform in size and color and showed no signs of mechanical or insect damage were selected for use in the study. Before initiating the experimental application of FA, nine broccoli heads were randomly selected to serve as an initial control group. The remaining broccoli heads were randomly partitioned into two groups, comprising 54 broccoli heads in each group. One group was immersed in 5 mg/L FA for 10 min (FA treatment), and one group was soaked in distilled water for 10 min as a control, check group (CK treatment). The treated broccoli heads were air-dried and then put in a 0.04 mm thick polyethylene film bags, with three broccoli heads in each bag, and placed at 20 °C and 90 % relative humidity (RH) for 4 d. Approximately 5 g (FW) of florets and a small section of subtending stalk tissue were collected from nine broccoli heads in each experimental group quickly frozen in liquid nitrogen and stored at −80 °C until subsequent analysis. Three heads of broccoli were placed in each bag. The heads in each bag were pooled and represented a single biological replicate. Thus, three bags were collected at each time point (0–4 d) for each group. Separate bags were collected for the broccoli heads used in the transcriptome, methylation, and metabolic analyses. Three technical replicates were used at each sampled time point. Data from each of the three technical replicates were pooled to obtain a single value for each biological replicate. Our previous studies indicated that broccoli begins to yellow on the third day of storage [10]. Therefore, samples were collected at 3 d and used for the subsequent -omic analyses. Three biological replicates were collected just before treatment (Initial 1, Initial 2, Initial 3), and then three biological replicates of the control group (CK 3d 1, CK 3d 2, CK 3d 3) or the FA group (FA 3d 1, FA 3d 2, FA 3d 3) of broccoli were collected after 3 d of storage.

Color

A digital colorimeter was used to gauge the color of the broccoli (CR-400, Konica Minolta, Japan). Daily measurements of L*, a*, and b* were made in a fixed, marker-outlined region of each broccoli head.

Transcriptome analysis

RNA-seq library preparation and sequencing

RNA-seq library preparation and sequencing followed the protocol described by Zuo et al. [35]. A Nanodrop 2000 was used to measure the amount of RNA. Following the manufacturer's instructions, sequencing libraries were created using a NEBNext UltraTM RNA Library Prep Kit for Illumina, and index codes were added to assign sequences to particular samples. The acquired sequence dataset was further filtered to provide clean reads after adaptor sequences and poor-quality sequence reads were eliminated. HISAT2 software was used to map the clean reads to the reference genome (https://www.genoscope.cns.fr/externe/plants/chromosomes.html). StringTie software was used to estimate transcript levels using a maximum flow algorithm, and Fragments Per Kilobase of transcript per Million fragments mapped (FPKM) was used to express transcript expression levels.

Identification of differentially expressed genes (DEG)

The R program Deseq was used to determine DEGs (1.10.1). DEGs were calculated using a |log2 (foldchange)| ≥ 2. And a False Discovery Rate (FDR) < 0.01 [36]. The ratio of gene expression levels in two groups is represented as log2 (foldchange) and the FDR represents the false discovery rate, corrected p-value. The p-value was modified using the Benjamini-Hochberg technique.

DNA methylation analysis

Library generation for bisulfite sequencing

The preparation of samples for bisulfite sequencing followed the standard procedure provided by Illumina. Genomic DNA was extracted, and the quantity of the extracted DNA was assessed. Fragment size was determined using agarose gel electrophoresis, and the chosen fragments underwent bisulfite treatment and PCR amplification to create a sequencing library. The quality of the created libraries was then evaluated and high-quality libraries were subsequently commercially sequenced on an Illumina Hiseq Xten platform. The quality of the generated reads (paired-end sequences) was evaluated, and the reads filtered to obtain clean reads for use in the subsequent bioinformatic analyses. Clean reads were mapped to the broccoli genome in Bismark (version 0.12.5) using default parameters after low-quality reads were removed from the raw data using Trimmomatic (version 0.36) (https://www.genoscope.cns.fr/externe/plants/chromosomes.html) [37].

Differentially methylated regions (DMR) and differentially methylated gene identification

A comparison of base information of cytosine (C) sites for the entire genome was extracted and used to obtain the coverage statistics of 5mC sites utilizing Bismark software, which uses the binomial distribution test principle to detect 5mC at each C locus. This was done in accordance with the best comparison results of clean reads in the reference genome. Coverage > = 4x, FDR < 0.05, and the site of methylation MoAbs were employed to screen the DMR region, according to the screening parameters [38], [39]. The coverage depth was required to be at least 10x and to have three separate methylation sites. The methylation level difference was just 0.2 at the most (CG type was 0.3). The p-value of <0.05 was employed along with Fisher's exact test. Based on the position of DMR on the genome and the annotation data for the genome, we defined a 3000 bp sequence upstream of the gene as the promoter region to annotate DMR [40].

Gene functional annotation

Genes were annotated using the following databases: The National Center for Biotechnology Information (NCBI) non-redundant (NR) protein database, (ftp://ftp.ncbi.nih.gov/blast/db/FASTA/); Protein family (Pfam, https://pfam.xfam.org/); Kyoto Encyclopedia of Genes and Genomes (KEGG, (https://www.genome.jp/kegg/); and Gene Ontology Consortium (GO, https://www.geneontology.org/). Using the Wallenius, non-central hypergeometric distribution, the GOseq R package was used to conduct the GO enrichment analysis of DEGs. Statistically significantly enriched DEGs in the KEGG pathways was determined using the KOBAS program [35]. Beijing Biomarker Technologies, Beijing, China, provided these analyses.

Metabolomic analysis

A total of 50 mg of sample was added to a centrifuge tube with a 1000 µL extract solution (acetonitrile: methanol: water = 2:2:1) and an internal standard (L-2-Chlorophenylalanine, 2 µg mL⁻¹). After a 30-seconds of vortexing to homogenize the samples, the tubes were sonicated at a frequency of 35 Hz for five minutes in a cold bath. The samples were then centrifuged after being incubated at −40 °C for 1 h. A fresh centrifuge tube was then filled with 250 µL of supernatant, which was then dried in a vacuum concentrator. The dried materials were then sonicated for 10 min over ice in 300 µL of 50 % acetonitrile to reconstitute them. One microliter was injected into each of the positive (POS) and negative (NEG) ion detection modes.

The metabolomic analysis was conducted by Beijing Biomarker Technologies, Beijing, China. The resulting data was subjected to a principal component analysis (PCA). The data model was validated using orthogonal projections to latent structures-discriminant analysis (OPLS-DA), which was utilized to extract information from the data set based on the variety and good group separation in PCA. Using projection analysis, the differential metabolites were sorted according to variable relevance. Differential metabolites (DEM) in the analyzed samples were determined based on a |log2 (foldchange)| ≥ 2, and a FDR < 0.01 [36] and an adjusted p-value. The ratio of gene expression levels in two groups is represented as log2 (foldchange) and the FDR represents the corrected p-value on log2 (foldchange) and a p-value < 0.05.

Integrated DNA methylation and transcriptome analysis

The weight technique was used to determine the methylation level of a specific region of the genome using the CGmap Tools program [40]. The methylation levels within each gene, 2kbp upstream, and 2kbp downstream were determined and sorted into three categories; CG, CHG, and CHH. The correlation coefficient (CC) was used to determine the degree of correlation between variables. The relationship between the level of transcript expression and gene methylation was utilized to identify differentially methylated genes. A |CC| > 0.80 and a p-value ＜ 0.05 were used to identify substantially different levels of methylation.

Integrated transcriptome and metabolome analysis

The correlation between all genes and metabolites was determined by calculating the Pearson correlation coefficient. The data were preprocessed using the Z-value transformation method before calculating the correlation. The correlation between transcript levels and metabolite levels was determined using CC and CCP. The threshold used to define a significant correlation was |CC| > 0.80 and CCP < 0.05.

Ethylene generation

Gas chromatography was used to assess the rate of ethylene production. Broccoli heads from each treatment were placed in a sealed container (14 L) with a gas extraction port at (20 ± 0.5) ℃. After 1 h, 1.0 mL the headspace gas was injected into a gas chromatograph (7820A, Agilent Technologies, Inc., USA). The level of ethylene was calculated using a standard curve based on the linear relationship between peak area (y) and the concentration of ethylene (x). The obtained equation representing this relationship was y = 1395.8x − 4781.6 (R² = 0.9995). The rate of ethylene generation is expressed as µL kg⁻¹h⁻¹.

Chlorophyll, FA, and glucosinolate content

The method provided by Xu et al. [10] was used to determine the amount of chlorophyll. Samples of frozen broccoli powder were homogenized in an acetone/ethanol (2:1) solution. The homogenate was then centrifuged at 12,000 g for 10 min at 4 °C. The supernatants were then collected. Sample absorbance at 645 nm and 663 nm was determined spectrophotometrically (UV-1800, Shimadzu Corporation) and the values used to calculate total chlorophyll content.

FA levels were determined using high-performance liquid chromatography (Agilent 1260, Agilent Technologies, Inc., USA) as described in Xu et al. [10]. Each sample (0.5 g) was mixed with 300 μL ethanol-aqueous solution (90:10, v/v), then homogenized for 10 min. The extracts were centrifuged at 12000 × g at 4 °C for 15 min. The supernatant was then vacuum freeze-dried. Methanol (300 μL) was used to redissolve the extract. FA levels in the samples were then determined using high performance liquid chromatography using a folic acid standard solution in 1 % methanol-H₂O (v/v).

UV spectrophotometry was used to calculate the level of glucosinolates as described by Xu et al. [10]. The reaction system consisted of 0.05 g frozen tissue and 3 mL methanol-acetic acid solution (40 % methanol and 0.5 % acetic acid). In the blank group, 0.05 g sample was added to 3 mL distilled water. The sample solutions were incubated at 37 ℃ for 5 min. Then, 2 mL methanol and 3 mg activated carbon were added to terminate the reaction. Samples were then centrifuged at 12,000 g for 20 min at 4 ℃. One mL of supernatant was then added to four mL standard glucose solution and incubated at 37 ℃ for 30 min. Four mL H₂SO₄ was then added to terminate the reaction. The absorbance of the reaction mixture was then measured at 240 nm. The amount of glucose present in the reaction solution was used to determine total glucosinolate content.

Quantitative reverse transcription-quantitative PCR (RT-qPCR) of DEGs

RT-qPCR was used to verify the relative gene expression level of DEGs as described in Yan et al. [15]. The selected primers were designed using Primer 3 software. Expression of the broccoli actin gene was used for normalization. The IQ5 system (Bio Rad, Hercules, CA, USA) was used to conduct the RT-qPCR. The analysis of each of the selected genes utilized three biological replicates and three technical replicates of each biological replicate. Each biological replicate was obtained from a different set of cDNAs. The relative expression level of candidate genes was determined using the 2^−ΔΔCT tower method and expression levels between treated and untreated samples were compared.

Analysis of volatile sulfur compounds

A total of 3.5 g of frozen cauliflower florets were placed in a 20 mL sealed vial with a PTFE silicone septum (Supelco, Bellefonte, Pennsylvania, USA). The vials were incubated for 1 h at room temperature and then placed in an automated sampling device. HS-GC-MS is comprised of a Headspace Sampler-10 (HS-10) and a GC-MSQP 2010 Plus system (Shimadzu, Kyoto, Japan). Volatile compounds were identified by comparing the mass spectra in a sample to the mass spectra in the NIS11 mass spectra database. The relative content of each identified volatile chemical was determined based on peak area percentage. The HS-GC-MS settings that were used are the same as reported in Lv et al. [41].

Statistical analysis

Statistical analyses were conducted using SPSS 22 (SPSS, Chicago, Illinois, USA) software. The data were subjected to a one-way analysis of variance, and the LSD test was used to compare the mean values between treatments. Statistical differences between samples was determined at p ＜ 0.05 and p ＜ 0.01. The presented data represent the mean ± standard deviation (SD).

Results

FA-treatment maintained the appearance of broccoli and modifies gene expression

Visual observations indicated that the CK treatment group of broccoli florets began to generally turn yellow within 3 days of storage at 20 °C, while those in the FA treatment group began to turn yellow on day 4 but only in small areas (Fig. 1A). The a* value represents a range from green to red, and the b* value represents a range from yellow to blue. The a* value of the CK and FA treatment groups of broccolis exhibited a gradual decreasing trend, which rapidly decreased in the CK treatment after 3 days of storage, while the b* values exhibited an increasing trend (Fig. 1B and C). Notably, the a* values in the FA treatment group were significantly higher (p < 0.05), that is more negative, than those in the CK treatment group on all four days of storage while b* values were significantly lower than the CK treatment on the third and fourth days of storage.

Fig. 1.

(A) Representative photographs comparing broccoli immersed in 5 mg/L folic acid for 10 min with untreated broccoli stored for four days at 20 ± 1 ℃. (B) Effect of 5 mg/L folic acid on a* values. (C) Effect of folic acid treatment on b* values in storage. (D) Upset diagram of DEGs identified in Initial / CK 3d, Initial / FA 3d, and CK 3d / FA 3d comparisons. (E) KEGG pathway enrichment analysis of DEGs in the CK 3d / FA 3d comparison and presentation of the the top 20 of the most significant (i.e., lowest q-value) pathways. Data represent the mean ± SD (n = 3). Asterisk (*) indicates a significant difference between the control and treatment groups at p < 0.05 while a double asterisk (**) indicates a significant difference at p < 0.01 based on an LSD test.

The transcriptomes of the three sample groups (Initial, CK 3d, and FA 3d) were analyzed to provide information on the potential mechanism underlying the ability of FA to delay the senescence of broccoli. Analysis of gene expression indicated major differences among the three sample groups, providing evidence that the postharvest application of FA altered gene expression during storage, and 46,764 genes were identified in this study. (Fig. S1A, Table S1). A total of 10,642 (4401 up-regulated, 6241 down-regulated), 10,715 (4780 up-regulated, 5935 down-regulated), and 13,888 (5777 up-regulated, 8111 down-regulated) DEGs were identified in CK 3d/FA 3d comparison, initial/CK 3d comparison, and initial/FA 3d comparison, respectively (Fig. 1D, Table S1). An upset diagram was constructed to illustrate how DEGs were distributed among the three treatment groups, (Fig. 1D). KEGG pathway enrichment analysis was conducted on the CK 3d/FA 3 d DEGs to gain insight into the potential function of the identified DEGs. Interestingly, KEGG enrichment analysis indicated that numerous of the DEGs identified in the CK 3d/FA 3 d group comparison is involved in the metabolism of cysteine, methionine, and sulfur (Fig. 1E).

FA-treatment methylation and metabolite levels in broccoli

Bisulfite sequencing was conducted on initial, CK 3d, and FA 3d samples to determine the DNA methylation response of broccoli to the exogenous application of FA. After data filtering, 42.36, 43.61, and 47.94 million clean reads from the Initial, CK 3d, and FA 3d samples were mapped to the broccoli reference genome, respectively (Table S2 and Table S3). The quantity of mCG, mCHG, and mCHH methylations was calculated, reflecting the amounts of 5mCs methylation in the three samples. The proportion of the different types of methylation sites in the whole genome differed, with CG and CHH types exhibiting the largest proportion, indicating that CG and CHH types of methylation sites account for the majority of methylation sites in broccoli (Fig. S1A). A total of 5,180, 9,979, and 28,934 DMRs for CG, CHG, and CHH methylation types were identified in the CK 3d / FA 3d comparisons, respectively (Fig. 2A; Table S4). The promoter region of a gene was defined as the region 3,000 bp upstream of the start site of a gene and the identified DMRs were mostly located in intergenic and promoter regions of genes [40]. In our results, a greater number of DMRs were also found in these two regions than were found outside of these regions (Fig. 2B). The FA treatment predominantly changed CHH and CHG DNA methylation types in the promoter and intergenic regions of the broccoli genome. The role of DMRs in the CK 3d / FA 3d was evaluated based on KEGG annotation. Results indicated that many of the DMRs were associated with processes, such as homologous recombination, nucleotide excision, and the transduction of plant hormone signals, closely connected to fruit ripening (Fig. S1C-S1E and Table S5).

Fig. 2.

(A) Several differentially methylated regions (DMRs) were identified in the Initial / CK 3d, Initial / FA 3d, and CK 3d / FA 3d comparisons. (B) Annotation of DMRs and their distribution in the CK 3d / FA 3d comparison group. Heat map of differentially abundant metabolites in the Initial, CK 3d, and FA 3d sample groups identified in the POS ion mode (C) and in the NEG ion mode (D). Volcano plot of differentially abundant metabolites identified in the CK 3d / FA 3d sample group comparison in a POS ion mode (E) and in the NEG ion mode (F).

In order to evaluate changes in the amount and/or presence of metabolites in broccoli in response to the FA treatment, an untargeted metabolomic analysis was performed (n = 3). A total of 3,116 metabolites with POS ions and 2,801 metabolites with NEG ions were separately identified (Fig. 2C and D). Notably, 487 DEMs (290 up-regulated, 197 down-regulated) with POS ions and 670 DEMs (158 up-regulated, 512 down-regulated) with NEG ions were identified in the CK 3d / FA 3d comparison (Fig. 2D and E). KEGG enrichment analysis indicated that metabolites in several KEGG pathways were differentially abundant in the CK 3d / FA 3d comparison, including 47 pathways in the POS ion mode and 71 pathways in the NEG ion mode. These pathways included FA biosynthesis, glutathione metabolism, glucosinolate biosynthesis, and others (Table S6).

Integrated analysis of multi-omic data from broccoli treated with FA

Combined analysis of DNA methylation and transcriptome data can provide information on the effect of methylation on gene expression, and provide insight into the regulation of gene expression. The integrated analysis of DMRs and DEGs revealed significant differences in the methylation level of 1,540 DEGs between the CK 3d and FA 3d sample groups (Table S7). Among the identified DEGs, 815 were positively correlated and 725 were negatively correlated. KEGG enrichment analysis indicated that both types of DEGs are related to glucosinolate biosynthesis pathways, chlorophyll metabolism, and glutathione metabolism (Figure S2 and Table S8).

An integrated analysis of the transcriptomic and metabolomic data was also conducted. Results revealed a total of 71 DEMs in the NEG ion mode and 2,252 DEGs between the CK 3d and FA 3d sample groups that were enriched in 69 KEGG pathways. Additionally, a total of 43 DEMs in the POS ion mode and 2,252 DEGs were identified that were enriched in 46 KEGG pathways (Figure S3). These pathways included FA biosynthesis, chlorophyll metabolism, sulfur metabolism, and flavonoid biosynthesis. These pathways were also found to be enriched in the integrated analysis of DNA methylation and transcriptome data.

FA-treatment maintained FA levels in stored broccoli

A total of 36 DEGs and 3 DEMs identified in our study are involved in folate metabolism pathways. After three days of storage, the FA treatment significantly upregulated the expression of genes encoding dihydropteroate synthase (DHPS) and aminodeoxychorismate lyase (ADCL) compared to the CK treatment group, and downregulated the expression of genes encoding GTP cyclohydrolase 1 (GCH1) and folylpolyglutamate synthase (FPGS) (Fig. 3A). The FA treatment also significantly inhibited the expression of genes related to FA metabolism, including serine hydroxymethyltransferase (SHM), methylenetetrahydrofolate reductase (MTHFR), SAH hydrolase (SAHH), S-adenosylmethionine synthase (MAT), etc. (Fig. 3A), thus, increasing the content of homocysteine (Hcy) and decreasing the content of methionine (Met) and S-adenosylhomocysteine (SAH) metabolites (Fig. 3B). FA levels in FA-treated broccoli increased continuously throughout the storage period, while levels in the CK treatment group decreased (Fig. 3C). After the second day of storage, FA content in FA-treated broccoli was significantly higher than it was in the control group (p < 0.01). After four days of storage, FA content in broccoli in the FA treatment group (225 g kg⁻¹) had increased by about 20 %, while FA content in the control group (184 g kg⁻¹) had decreased by about 7.6 % (Fig. 3C).

Fig. 3.

(A) DEGs involved in folate metabolism as determined by KEGG pathway analysis [5]. Log2-based FPKM values were used to generate the heat map. GTP, guanosine triphosphate; DHN-P3,dihydroneopterin triphosphate pyrophosphatase; HMDHP-P2, 6-hydroxymethyldihydropterin pyrophosphate; DHP, dihydropteroate; ADC, aminodeoxychorismate; pABA, para-aminobenzoic acid; THF, tetrahydrofolate; THF-Glu(n), tetrahydrofolate polyglutamate; 5,10-CH2-THF, 5,10-methylene-THF; Met, methionine; SAM, S-adenosylmethionine; SAH, S-adenosylhomocysteine; Hcy, Homocysteine; ACC, aminocyclopropane-1- carboxylic acid; GCH1, GTP cyclohydrolase I; DHPS, dihydropteroate synthase; ADCL, aminodeoxychorismate lyase; FPGS, folylpolyglutamate synthetase; SHM, serine Hydroxymethyltransferase; MTHFR, methylenetetrahydrofolate reductase; MS, methionine synthase; MAT, S-adenosylmethionine synthase; SAHH, SAH hydrolase; DNMET, DNA methyltransferase; ACS, ACC synthase; ACO, ACC oxidase. (B) Relative abundance of the DEMs Hcy and Met in Initial, CK 3 d, and FA 3d samples. ** indicates a significant difference (p < 0.01) between the designated groups. (C) Effect of folic acid treatment on folic acid content in CK and FA-treated broccoli during storage. ** indicates a significant difference (p < 0.01) between the designated groups. (D) Effect of folic acid treatment on the rate of ethylene generation in CK and FA-treated broccoli in storage. * indicates a significant difference (p < 0.05) between the designated groups. (E) RT-qPCR analysis of DEGs involved in folic acid metabolism in initial, FA 3d, and CK 3d samples of broccoli. Different lowercase letters above the bars indicate a significant difference (p < 0.05) between sample groups. (F) Annotation of DEGs and DMRs involved in folic acid metabolism.

Compared to the CK treatment group, the FA treatment significantly decreased the expression of DNA methyltransferase (DNMET), which is associated with DNA methylation levels, and suppressed the expression of ACC synthase (ACS) and ACC oxidase (ACO), which are associated with ethylene biosynthesis (Fig. 3A). Ethylene production in both groups of broccolis exhibited a fluctuating upward trend during storage, however, on the third- and fourth-day ethylene generation in the FA treatment group was significantly lower (p < 0.05) than it was in the CK treatment group (Fig. 3D and Table S9). RT-qPCR was used to determine the reliability of the transcriptome data. Based on significant increases in FPKM, five DEGs related to FA metabolism were selected for RT-qPCR analysis. RT-qPCR results confirmed that the results of the RNA-seq were reliable. The combined analysis of transcriptome and DNA methylation data indicated that there were 29 DMRs in GCH1, MTHFR, SHM, methionine synthase (MS), MAT, SAHH, ACS, ACO, and DNMET DEGs in the CK 3d/FA 3d group, including five DMRs for the CG type, nine DMRs for the CHG type, and 15 DMRs for the CHH type (Fig. 3F). Notably, 12 DMRs were located in the promoter regions of genes, of which seven DMRs exhibited an increased level of methylation and a decreased level of transcription in the CK 3d / FA 3d group, including GCH1, MAT, ACS, and DNMET. The methylation level of five DMRs decreased as did the transcription level, including in SHM, MS, SAHH, ACS, and ACO.

FA-treatment maintained chlorophyll levels in stored broccoli

Integrated analysis of the transcriptome and metabolomic data identified 67 DEGs and one differentially abundant metabolite (bilirubin) in the POS ion mode associated with chlorophyll metabolism (Figure S4, Fig. 4A, and B). Among the DEGs, 14 DEGs are involved in the synthesis of chlorophyll precursors and the formation of chlorophyll a/b, including glutamyl-tRNA (EARS), glutamyl-tRNA reductase 1 (Hema1), coproporphyrinogen-III oxidase 1 (CPOX), protoporphyrinogen oxidase 1(PPOX), magnesium chelatase subunit D (CHLD), magnesium chelatase subunit I (CHLI), magnesium protoporphyrin IX methyltransferase (CHLM), protochlorophyllide reductase C (POR), divinyl chlorophyllide an 8-vinyl-reductase (DVR), chlorophyll synthase (CHLG), and chlorophyll b reductase (NYC1) (Fig. 4A). In addition, the FA treatment significantly delayed the decline in the expression level of EARS, CPOX, PPOX, CHLD, CHL, CHLM, POR, DVR, and CHLG, and inhibited an increase in the expression level of Hema1 and NYC1 (Fig. 4A). While the chlorophyll content in both the CK and FA treatment groups declined, chlorophyll content in CK treatment group decreased more rapidly and chlorophyll content in the FA treatment group was significantly higher (p ＜ 0.01) than in the CK treatment group on the third and fourth days of storage (Fig. 4C). RT-qPCR analysis of the expression of five DEGs involved in chlorophyll metabolism validated the accuracy of the RNA-seq data (Fig. 4D). Combined analysis of the transcriptome and DNA methylation data identified three CHG types of DMR located at distal intergenic regions in NYC1 and revealed that the methylation and transcription levels in NYC1 were reduced after FA treatment (Fig. 4E). Moreover, four DMRs were located at the distal intergenic region of Hema1, among which one was a CG type and one was a CHG type. The FA treatment increased the methylation level in Hema1 but decreased its transcription level. The other two DMRs were CHH type, and the methylation and transcription levels of both of the associated genes were reduced after FA treatment.

Fig. 4.

(A)DEGs involved in chlorophyll metabolism as determined by KEGG pathway analysis [23]. Log2-based FPKM values were used to generate the heat map. EARS: glutamyl-tRNA synthetase; Hema1: glutamyl-tRNA reductase 1; CPOX: coproporphyrinogen III oxidase; PPOX: protoporphyrinogen oxidase; CHLD: magnesium chelatase subunit D; CHl1: magnesium chelatase subunit I; ChlM: magnesium-protoporphyrin O-methyltransferase; DVR: divinyl chlorophyllide an 8-vinyl-reductase; POR: protochlorophyllide reductase; CHlG: chlorophyll/bacteriochlorophyll a synthase; NYC1: chlorophyll(ide) b reductase. (B) Relative abundance of DEMs involved in chlorophyll metabolism in Initial, CK 3d, and FA 3d samples. ** indicates a significant difference (p < 0.01) between the designated groups. (C) Effect of folic acid treatment on chlorophyll content in CK and FA-treated broccoli samples during storage. * indicates a significant difference (p < 0.05) and ** (p < 0.01) between the designated groups. (D) RT-qPCR analysis of DEGs involved in chlorophyll metabolism in Initial, CK 3 d, and FA 3 d samples of broccoli. Different lowercase letters above the bars indicate a significant difference (p < 0.05) between sample groups. (E) Annotation of DEGs and DMRs involved in chlorophyll metabolism.

FA-treatment inhibited a decrease in glucosinolate and quality loss in broccoli in storage

Seven DEGs and one DEM (Indole-3-acetonitrile) were found to be associated with the synthesis and metabolism of sulfur glycosides in the CK 3d/FA 3d group comparison, including Cytochrome P450 79B1 (CYP79B1), Cytochrome P450 83B1 (CYP83B1), S-alkyl-thiohydroximate lyase (SUR), UDP-glycosyltransferase 74B1 (UGT74B1), Cytosolic sulfotransferase 16 (SOT16), and β-thioglucoside glycopyrrolate 4 (TGG4) (Fig. 5A and B). In addition, 13 DEGs identified in the CK 3d/FA 3d group comparison are associated with primary S metabolism, including adenylylsulfate reductase (APR), adenylyl-sulfate kinase (APK), ATP amylase (APS), and Methionine gamma-Lyase (MGL). These 20 genes exhibited lower levels of expression in the FA 3d group, relative to the CK 3d group (Fig. 5A). RT-qPCR analysis of the expression of the four DEGs associated with glucosinolate metabolism and primary S metabolism corroborated the validity of the RNA seq results (Fig. 5E). Combined transcriptomic and DNA methylation analysis indicated that seven DEGs possessed a total of 14 DMRs, including five CHG-type DMRs and nine CHH-type DMRs (Fig. 5F). Among the DMRs identified in the CK 3d/FA 3d comparison, the methylation level of 12 DMRs decreased, while transcriptome levels also decreased. The methylation level of two of the DMRs increased but the transcription level decreased. Notably, four DEGs (APS1, SOT16, UGT74B1, and SUR1) possessed one CHG-type DMR and three CHH-type DMRs located in the promoter region, and DNA methylation and transcription levels both decreased in the CK 3d/FA 3d comparison. Notably, on the third and fourth days of storage, the concentration of glucosinolates was significantly greater (p ＜ 0.05) in the FA treatment group of broccoli than it was in the CK treatment group. HS-GC-MS analysis identified 23 volatile, sulfur-containing compounds in broccoli, including three sulfur-containing free amino acid metabolites, six glucosinolate decomposition products (Fig. 5C), and 14 other sulfur-containing compounds (Table S10). A lower number of volatile sulfur-containing compounds were found in the FA treatment group than in the CK treatment group. The metabolites of sulfur-containing amino acids identified in the present study were mainly dimethyl sulfide, dimethyl trisulfide, and methyl (methylthio) methyl disulfide. The FA treatment group had a lower relative level than the CK treatment group for three of the sulfur compounds. Six glucosinolate breakdown products, including thiocyanate, nitrile, thioether, and other substances, were found to be present at lower levels in the FA treatment group than in the CK treatment group.

Fig. 5.

(A) DEGs involved in glucosinolate and primary S metabolism as determined by KEGG pathway analysis [23], [75]. Log2-based FPKM values were used to generate the heat map. CYP79B1: Cytochrome P450 79B1; CYP83B1: Cytochrome P450 83B1; SUR1: S-alkyl-thiohydroximate lyase SUR1; UGT74B1: UDP-glycosyltransferase 74B1; SOT16: Cytosolic sulfotransferase 16; TGG4: β-thioglucoside glycopyrrolate 4; APS: ATP sulfurylase 1; APK: Adenylyl-sulfate kinase; APR: adenylylsulfate reductase 2; MGL: methionine-gamma-lyase. (B) Relative abundance of DEM (indole-3-acetonitrile) in Initial, CK 3d, and FA 3d samples. ** indicates a significant difference (p < 0.01) between the designated groups. (C) Relative content (%) of volatile sulfur compounds in CK 3d and FA 3d samples of broccoli. ** indicates a significant difference (p < 0.01) between the CK and FA sample groups. (D) Effect of folic acid treatment on glucosinolate content in CK and FA-treated broccoli in storage. * indicates a significant difference (p < 0.05) and ** (p < 0.01) between the CK and FA sample groups. (E) RT-qPCR analysis of DEGs involved in glucosinolate and primary S metabolism in Initial, CK 3 d, and FA 3 d samples of broccoli. Different lowercase letters above the bars indicate a significant difference (p < 0.05) between sample groups. (F) Annotation of DEGs and DMRs involved in glucosinolate and primary S metabolism.

Discussion

Fresh broccoli is a popular, nutrient-rich vegetable that begins to rapidly senesce after it is harvested. Senescence is accompanied by changes in respiration, as well as nutrient and hormone levels, with symptoms such as yellowing, off odors, and decay [15]. The current research was conducted to determine the effect of postharvest immersion of broccoli in 5 mg/L FA for 10 min followed by 4 d of storage on its physiology, gene expression, metabolite levels, and degree of methylation, relative to untreated broccoli. Results indicated that the FA treatment significantly inhibited chlorophyll degradation, reduced the rate of ethylene generation, and helped to maintain higher levels of FA and glucosinolate.

DNA methylation is a dynamic process that regulates gene expression during plant growth and in reaction to abiotic stress factors [42], [43], [44], [45]. The amount or pattern of DNA methylation has been shown to significantly alter in response to environmental stimuli [46], [47], [48]. In the present study, methylcytosine was primarily found at CHH sites (36.07–36.36) and CG sites (35.98–36.21 %), while the peak frequency of the CHG sequence ranged from 27.6 to 27.8 % (Fig. S1B). These findings indicate that CHH and CG methylation sites dominate in broccoli. Similar results were reported in unfertilized ears of corn and pineapple fruit but contrast with results reported for corn leaves and tomato fruit [32], [35], [49]. We speculate that differences in methylation patterns can be attributed to differences between species and varieties and/or developmental stages. In the present study, CHH methylation types in all DMRs were found to be significantly higher than CG and CHG methylation types (Fig. 2A; Table S2). This may be because CHH methylation types predominate in the plant genome, relative to CG and CHG types. Notably, CHH sites are associated with stress response, developmental regulation, and DNA methylation regulation [50]. Four classes of DNA methyltransferases and four functional DNA demethylases, including DNA methyltransferase 1, domains rearranged methyltransferase 1 and 2, and chromomethylase chromo methylase 3; ROS1, and DEMETER, DEMETER LIKE 2 and 3, control DNA methylation in Arabidopsis thaliana, and can form either a symmetrical (CG or CHG) or non-symmetrical (CHH) structure at cytosine [51], [52], [53]. In our study, we found that the overall methylation level in broccoli decreased during storage and that the expression level of DNMET exhibited a declining tendency. This downward trend may be attributed to the fact that, in the absence of methyltransferase activity, all organisms passively lose cytosine methylation during DNA replication, leading to decreased methylation at locations essential to plant growth and development [54]. In general, the decrease in methylation levels may represent a mechanism that regulates plant development and cell fate reprogramming. Our findings suggest that the expression of some genes is negatively correlated with methylation while the expression of others is positively correlated. (Fig. 3F, 4F, and 5F; Table S8). Rishi et al. [55] indicated that methylation may occasionally be necessary for the activation of transcription. It has also been reported that DNA methylation at sites within the promoter region of genes and transcription termination region can suppress gene expression [56], [57], [58]. Notably, however, Medvedeva et al. [59] stated that it is not possible to observe the direct and selective methylation of certain transcription factor binding sites, which prevents transcription factor binding being recognized as a general transcriptional control mechanism since it only applies in certain situations. This latter point of view is supported by our findings in broccoli, where methylation events within a promoter region or a transcription termination site were both negatively and positively correlated with the expression of specific genes.

FA are cofactors in one carbon transfer reactions that play a role in DNA methylation, the synthesis of amino acids (methionine, glycine, and serine), nucleic acids, and S-adenosylmethionine (SAM), among other metabolic processes in plants [8], [60], [61]. Folic acid also plays an extremely important role in maintaining genomic stability. The enzyme GTP cyclohydrolase (GCH) is thought to play a crucial role in regulating FA biosynthesis because it activates a branch of the FA biosynthesis pathway [62], [63], [64]. The conversion of aminodeoxychorismate to para-aminobenzoic acid (pABA) is mediated by the enzyme ADCL [65]. In our study, the expression of GCH1 decreased during the storage of broccoli, while the expression levels of ADCL and DHPS increased. Therefore, we speculate that the increase in FA levels in broccoli in response to the FA treatment may be due to the increased expression of these two genes (Fig. 3A). The last step in the synthesis of folates is the attachment of a glutamate tail to a tetrahydrofuran (THF) molecule in a reaction catalyzed by the enzyme folylpolyglutamate synthetase (FPGS) [66]. FPGS activity is affected by FA availability in the cell, with a high abundance of FA inhibiting FPGS activity in plants [67], [68], [69]. Therefore, we speculate that the accumulation of FA in broccoli brought about by soaking broccoli in a solution of FA may be responsible for the decrease in FPGS expression. Importantly, MS synthesizes methionine and regenerates the methyl group of SAM after methylation events by using 5-methyl-THF (5-CH₃-THF) as a methyl donor. After providing methyl groups, SAM is converted into sadenosyl homocysteine (SAH). Previous studies have shown that SAH is an effective inhibitor of methyltransferase [70]. In our present study, exogenous folic acid treatment maintained SAH levels and promoted MET levels. However, the overall methylation level in broccoli and the expression level of DNMET decreased, suggesting that the inhibitory effect of SAH on methylation may be influenced by other factors. Thus, further research is needed. ACC synthase (ACS) converts SAM into aminocyclopropane-1-carboxylic acid (ACC), which is then further oxidized by ACC oxidase (ACO) to create ethylene [69]. Our study indicated that the FA treatment of broccoli significantly inhibited the expression of SHM, MTHFR, MAT, ACS, and ACO, relative to untreated broccoli, which may be the main reason for the reduced rate of ethylene production. Collectively, our data indicate that exogenous application of FA may affect the methylation level of GCH1 and other genes, as well as the transcription level of these genes, which would directly impact the concentration of Met, Hcy, and SAH, thus, promoting the accumulation of FA and inhibiting ethylene generation.

Broccoli kept at room temperature after harvest undergoes rapid chlorophyll degradation, turns yellow, and has a shelf life of fewer than three days [69]. Hence, reducing chlorophyll degradation and loss would aid in preserving the sensory quality of broccoli. Webb et al. [72] reported that a small change in the content and distribution of FA derivatives in plants could affect chlorophyll synthesis in an entire plant through its effect on the methyl cycle. FA metabolism can transfer the methyl group provided by SAM to porphyrin IX, which is facilitated by the enzyme CHLM, to generate Mg-protoporphyrin IX methyl ester, and this compound can then enter the downstream pathway of chlorophyll synthesis [73]. Chlorophyll synthesis involves two processes (Fig. 4A), the synthesis of chlorophyll precursors and the production of chlorophyll a/b [74]. Enzymes, including EARS, Hema, CPOX, and PPOX are crucial to the production of chlorophyll [75], [76], [77]. In our study, the FA treatment (FA 3d) increased the expression of chlorophyll synthesis-related genes, relative to the control group (CK 3d). These findings support earlier findings showing that FA treatments stimulate chlorophyll production [71]. Our present data indicate that the FA treatment differentially impacts the expression of CAO and NYC, both of which are involved in chlorophyll biosynthesis. In response to the FA treatment, the expression of CAO did not change much, however the expression of NYC1 drastically decreased. Previous studies have found that the early yellowing of broccoli can be attributed to the accumulation of chlorophyll b and the lack of chlorophyll a [78]. We suggest that the FA treatment alters the ratio of chlorophyll a and b content in broccoli and inhibits the yellowing of broccoli during storage by promoting chlorophyll synthesis and affecting the synthesis of chlorophyll a and chlorophyll b.

Glucosinolate is the main bioactive compound in broccoli. Glucosinolate is decomposed into isothiocyanates, indoles, nitriles, and other substances when plants are subjected to biological stress and mechanical injury. In our investigation, five DEGs in the glucosinolate biosynthesis pathway (CYP79B1, CYP83B1, SUR1, UGT74B1, SOT16) exhibited lower expression in the FA 3d group than in the CK 3d group (Fig. 5A). During storage, however, glucosinolate levels in the FA treatment group were greater than they were in the control group. (Fig. 5D). The major enzyme that catalyzes the decomposition of glucosinolate is β-thioglucoside glycopyrrolate (TGG)[79], and the expression level of TGG4 in FA 3d samples was much lower than it was in CK 3d samples. (Fig. 5A). Therefore, we suggest that one of the main reasons the FA treatment maintains glucosinolate content in broccoli is due to its ability to inhibit the decomposition of glucosinolate. Glucosinolate is also a secondary metabolite of sulfur. Mugford et al. [80] found that Arabidopsis mutants lacking two subtypes of APS kinase, thus, limiting 3′-phospho-adenosine-5′-phosphosullfate (PAPS) synthesis, had only 10–15 % of the glucosinolate present in wild-type plants. The biosynthesis of the sulfur-containing amino acid cysteine is promoted by the enzymes APR, APK, and APS, which also aid in the accumulation and homeostasis of sulfur [79]. Di Pentima et al. [81] stated that large concentrations of free sulfur amino acids can stimulate the synthesis of volatile sulfur compounds. In addition to yellowing, the most common storage problem of broccoli is the production of off-odors [82]. The main components of broccoli odors are volatile sulfur compounds, including methyl mercaptan, isothiocyanate, nitriles, oxazolidine thione, and various indoles [41]. These volatile sulfur compounds are mainly products of sulfur metabolism, including the enzymatic hydrolysis of sulfur-containing amino acids and glucosinolates [29], [30]. The enzyme MGL catalyzes the production of methyl mercaptan from methionine [83]. FA-treated broccoli produced a lower level of volatile sulfur compounds, relative to the control group of broccolis, which may be attributed to the ability of FA to inhibit the expression of APR, APK, and APS, and prevent the excess accumulation of sulfur products. We speculate that the reduced synthesis of sulfur-containing amino acids is due to the lower expression level of these genes, which in turn reduces the amount of MGL expression and, thus, decreases the conversion of sulfur-containing amino acids into volatile sulfur compounds. Glucosinolates can also be hydrolyzed by TGG into thiocyanate, nitrile, and thioether [84]. We found that the FA treatment dramatically reduced TGG4 expression. The HS-GC–MS data obtained in our study indicate that FA-treated broccoli contains fewer glucosinolate decomposition products and sulfur amino acid metabolites than untreated broccoli. Therefore, we suggest that the FA treatment inhibits the production of off-odors in broccoli, due to the ability of FA to decrease the accumulation of sulfur-containing amino acids and glucosinolates.

Conclusions

The effect of an exogeneous folic acid treatment on the postharvest quality of broccoli was evaluated using a multi-omic approach. A combination of transcriptome, DNA methylation, metabolomic, physiological, and biochemical analyses, provided insight into how folic acid maintains the nutritional quality, color, and aroma of stored broccoli (Fig. 6). Results indicated that FA induces the expression of genes and alters the level of metabolites associated with folic acid biosynthesis, chlorophyll metabolism, and sulfur metabolism. The upregulation of the identified genes is associated with a delay in yellowing, inhibition of off-odor development, and the maintenance of nutritional quality in stored broccoli. In general, changes in the pattern of gene methylation, gene expression, and metabolite levels induced by FA, all play a role in regulating postharvest aging in broccoli. In summary, exogenous folic acid treatment represents a promising method for maintaining the quality and nutritional status of broccoli after harvest. However, it’s use and effect on other vegetables remains to be explored.

Fig. 6.

A proposed model illustrating the response of broccoli to exogenous application of 5 mg/L folic acid based on an integrated analysis of DNA methylation, transcriptome, and metabolome data. Gene name abbreviations are presented in circles and red type indicates up-regulation while green type represents down-regulation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Funding

This work was supported by the China Agriculture Research System Project (CARS-23), the Collaborative innovation center of Beijing Academy of Agricultural and Forestry Sciences (KJCX201915), Beijing Municipal Science and Technology Commission (Z191100008619004, Z191100004019010, and Z181100009618033), the National Natural Science Foundation of China (31772022), the Natural Science Foundation of Beijing (6182016), Special innovation ability construction fund of Beijing Academy of Agricultural and Forestry Sciences (20180705 and 20200427), Special innovation ability construction fund of Beijing Vegetable Research Center, Beijing Academy of Agriculture and Forestry Sciences (KYCX2020112). We thank Dr. Michael Wisniewski (Adjunct Professor, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA) for his critical reading of the manuscript.

Compliance with Ethics Requirements:

All Institutional and National Guidelines for the care and use of animals (fisheries) were followed.

CRediT authorship contribution statement

Yaqi Zhao: Data curation, Methodology, Investigation, Software, Writing – original draft, Writing – review & editing. Junyan Shi: Data curation, Methodology, Investigation, Software. Bihong Feng: Writing – review & editing, Resources. Shuzhi Yuan: Data curation, Writing – original draft. Xiaozhen Yue: Data curation, Writing – original draft. Wenlin Shi: Software, Writing – original draft. Zhicheng Yan: Conceptualization, Formal analysis. Dongying Xu: Software, Writing – original draft. Jinhua Zuo: Conceptualization, Formal analysis, Resources, Supervision. Qing Wang: Conceptualization, Data curation, Formal analysis, Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary materials

The following are the Supplementary data to this article:

Supplementary data 1

Table S1. All the genetic information obtained by transcriptome sequencing

Supplementary data 2

Table S2. DNA methylation sequencing statistics. Table S3. DNA methylation sequencing statistics for the three sample groups (Initial, CK 3d, and FA 3d). Table S4. Quantitative statistics of genome-wide, differentially methylated regions (DMRs).

Supplementary data 3

Table S5. KEGG analysis of DMRs in the CK 3d / FA 3d sample group comparison.

Supplementary data 4

Table S6. KEGG analysis of DEMs in the CK 3d / FA 3d sample group comparison identified in the POS and NEG ion mode.

Supplementary data 5

Table S7. Correlation between methylation levels and the expression level of DEGs (|coefficient|>0.80 and p value <0.05).

Supplementary data 6

Table S8. List of all of the identified DEGs and DMRs in the three comparative groups.

Supplementary data 7

Table S9. Effect of folic acid treatment on ethylene production rate in CK and FA-treated broccoli in storage.

Supplementary data 8

Table S10. Composition and content of volatile sulfur compounds in broccoli.

Supplementary data 9

Figure S1. (A) PCA analysis of transcriptome data for each sample group. (B) Statistics on the percentages of mCs. (C) KEGG pathway analysis of DMRs for CG type in the CK 3d/FA 3d sample group comparison. (D) KEGG pathway analysis of DMRs for CHG type in the CK 3d/FA 3d sample group comparison. (E) KEGG pathway analysis of DMRs for CHH type in the CK 3d/FA 3d sample group comparison.

Supplementary data 10

Figure S2. KEGG analysis of the integrated analysis of transcriptomic and methylation data in the comparisons between different sample groups.

Supplementary data 11

Figure S3. KEGG analysis of the integrated analysis of transcriptomic and metabolomic data in the comparisons between different sample groups.

Supplementary data 12

Figure S4. (A) Heat map of DEGs involved in chlorophyll metabolism. (B) Correlation between DEGs and differentially abundant metabolites involved in chlorophyll metabolism. A circle represents a metabolite. A box represents a gene. Values presented on the branched lines are the correlation coefficient. Red represents a positive correlation and green represents a negative correlation. Increasing line width and color intensity indicates a larger coefficient value.