Label-free quantitative proteomics analysis of jujube (Ziziphus jujuba Mill.) during different growth stages

Chinese jujube (Zizyphus jujuba Mill.), a member of the Rhamnaceae family with favorable nutritional and flavor quality, exhibited characteristic climacteric changes during its fruit growth stage. Therefore, fruit samples were harvested at four developmental stages on days 55 (young fruits), 76 (white-mature fruits), 96 (half-red fruits), and 116 (full-red fruits) after flowering (DAF). This study then investigated those four growth stage changes of the jujube proteome using label-free quantification proteomics. The results identified 4762 proteins in the samples, of which 3757 proteins were quantified. Compared with former stages, the stages examined were designated as “76 vs. 55 DAF” group, “96 vs. 76 DAF” group, and “116 vs. 96 DAF” group. Gene Ontology (GO) and KEGG annotation and enrichment analysis of the differentially expressed proteins (DEPs) showed that 76 vs. 55 DAF group pathways represented amino sugar, nucleotide sugar, ascorbate, and aldarate metabolic pathways. These pathways were associated with cell division and resistance. In the study, the jujube fruit puffing slowed down and attained a stable growth stage in the 76 vs. 55 DAF group. However, fatty acid biosynthesis and phenylalanine metabolism was mainly enriched in the 96 vs. 76 DAF group. Fatty acids are precursors of aromatic substances and fat-soluble pigments in fruit. The upregulation of differential proteins at this stage indicates that aromatic compounds were synthesized in large quantities at this stage and that fruit would enter the ripening stage. During the ripening stage, 55 DEPs were identified to be involved in photosynthesis and flavonoid biosynthesis in the 116 vs. 96 DAF group. Also, the fruit entered the mature stage, which showed that flavonoids were produced in large quantities. Furthermore, the color of jujube turned red, and photosynthesis was significantly reduced. Hence, a link was established between protein profiles and growth phenotypes, which will help improve our understanding of jujube fruit growth at the proteomic level.


Introduction
Chinese jujube (Zizyphus jujuba Mill.), a highly nutritious fruit, is one of the most popular fruits in China. About 6.25 million tons of jujube dry fruit are produced annually from a cultivated area of approximately 2 million ha. 1 Recent studies have revealed that its fruits contain various functional compounds such as vitamin C, amino acids, triterpene acids, polysaccharides, and polyphenols. [2][3][4] Based on bioactivity analyses, jujube fruit has been shown to lower blood pressure, reverse liver disease, treat anemia, and inhibit the growth of tumor cells. 5 Additionally, several reports have demonstrated the neuroprotective activities 6 and blood-nourishing function of jujube fruit. 7 Jujube exhibits characteristic climacteric changes during its fruit growth stages. During this stage, the jujube fruit cells divide quickly, and the number of cells increases rapidly; however, the cell volume increases slowly. Then, the volume of jujube fruit cells rapidly expands and continues to mature until the jujube reaches the white period. During the ripening process, owing to changes in the levels of avonoids, carotenoids, and chlorophyll, 8 jujube fruits pericarp color changes from green to white during early ripening and later turns red. Simultaneously, aromatics or characteristic components are produced in large numbers. During fruit development and ripening, the epigenetic characteristics of jujube fruit, such as color changes, sugar metabolism, cell wall restructuring, stress response, and aroma volatile synthesis, are involved in a series of physiological metabolic processes. 9 Recently, studies on jujube fruit focused on biochemical changes, including its metabolic and transcriptional regulation. Observing the physiological changes occurring in this fruit is a common method for studying the process of fruit growth and change. Li et al. reported how the skin color of jujube fruit during maturation was due to changes in the levels of avonoids, carotenoids, and anthocyanins. 8 Metabolomic proling also reected the genetic, epigenetic, and environmental factors that inuenced its cellular physiology. Similarly, Shi et al. reported that the ZjANS and ZjUGT79B1 promoters were activated by ZjMYB5, ZjTT8, and ZjWDR3 proteins, which regulated anthocyanin biosynthesis in jujube fruit skins. 10 Transcriptomic proling is another promising approach to analyze the entire genome, which provides details regarding the biological processes underlying jujube fruit development. Liu et al. conducted transcriptome analysis to investigate the changes in gene expression in surface-pitted jujube fruit. 11 Furthermore, Chen et al. found that 39 ZjWRKY genes were expressed during jujube fruit development and ripening. 12 These studies have provided an enormous amount of data that expand our knowledge of the molecular events associated with ripening. Proteomic proling bridges data-rich information regarding alterations of the proteome that occur because of the transcriptome, and metabolome has been used as well to analyze expressed proteins and protein function in a cellular context. Research on fruit proteomics, such as that for peaches, 13 tomatoes, 14 apples, 15,16 and muskmelons, 17 has also shown that proteomic analysis approaches can illustrate proteins involved in fruit ripening and senescence.
Proteomics is therefore a powerful method for exploring differential proteins and their complex regulatory mechanisms. 18 Furthermore, the label-free quantitative (LFQ) proteomics approach is a mass-based quantitative technique for identifying and quantifying proteins. 19 Compared with proteome quantitation based on labeling techniques, LFQ can detect a large number of samples at a lower cost and ease. LFQ also has a simpler experimental workow. 20,21 However, quantitative analysis is comparatively time-consuming, as no multiplexed analysis is possible as in the case of label-free proteomics. Additionally, the instrumentation time necessary for LFQ analysis is signicantly longer than that of the labelbased quantitative method, and variations during sample preparation, including chromatography and MS acquisition, are increased. 22 Furthermore, the saturation of signals at high protein abundance levels can be due to limitations in the iontrapping capacity, duty cycle, or ionization efficiency of the mass spectrometer. 23 Therefore, Venable et al. suggested dataindependent analysis (DIA) as an alternative to datadependent analysis (DDA). 24 DIA increases the signal-to-noise effects during analysis by threefold to vefold and identies undetected peptides in the parent ion scan of a typical DDA experiment. 25 LFQ approaches are also suitable for comparing protein expression levels between different conditions and provide good approximations of absolute protein abundance, especially when used with carefully designed experiments and relevant statistical methods. 21 Interestingly, LFQ has recently been used in numerous plant proteomics studies, such as in apricot, 19 strawberry, 26 tomato, 14 apple, 27 and citrus. 28,29 In this research, fresh jujube were collected at four stages of development and were extracted to analyze its crude proteins using nanoow liquid chromatography-tandem mass spectrometry (nano LC-MS/MS) instruments. The four growth stage changes of jujube proteome were also studied using LFQ to investigate the differentially expressed proteins (DEPs) involved in signicant metabolic changes during development and ripening. Next, we used bioinformatics to investigate ripening-related pathways involved in signicant proteome changes during jujube ripening.

Material and treatments
Fresh jujube, "Xinxing," was obtained from a local farm in Binzhou, Shandong, China, in 2020. Young fruits (55 days aer owering [DAF]), white-mature fruits (76 DAF), half-red fruits (96 DAF), and full-red fruits (116 DAF) of Xinxing were collected separately from jujube trees (Fig. 1). 8,30 Pest-and disease-free jujube fruits with uniform size, having no mechanical damage, were used for the experiments. The jujube pulp was then frozen in liquid nitrogen and pulverized. Next, the powdered samples were stored in a refrigerator at À80 C. About 70-80 fruits were picked during each stage randomly and then divided into three biological replicates.

Proteomic analysis
2.2.1 Protein extraction. The method of protein extraction referenced previous studies and slightly altered. 19,31 Approximately 3 g of the powdered sample was thoroughly homogenized in 15 mL pre-cooling acetone containing a trichloroacetic acid mixed solvent (trichloroacetic acid/acetone, 1 : 9 [v/v], containing 0.07% b-mercaptoethanol) at 5000 rpm for 60 s. Then, the mixture was placed at À20 C for 4 h, and centrifuged at 12 000 Â g for 15 min at 4 C, aer which the supernatant was discarded. Next, pellets were resuspended in pre-cooling acetone and washed three times, aer which the precipitation was air-dried. Subsequently, crude proteins were divided by adding 15 volumes of urea extraction buffer (8 M urea, 0.1 M Tris-HCl, pH 8.5) and sonicated for 5 min (pulse 2 s on, 3 s off, 50% amplitude) in an ice-water bath (Sonics® Vibra-Cell VCX 130). Aer, centrifugation was conducted at 12 000 Â g for 15 min, and the supernatant was ltered with 0.45 mm lters. Bovine serum albumin was used as a reference substance, and Coomassie Brilliant Blue G-250 as the chromogenic agent. Also, visible spectrophotometry was used to determine the content of protein in the ltrate by UV at 595 nm.
2.2.2 Filter-aided sample preparation digestion. Protein extracts were digested with trypsin following the lter-aided sample preparation procedure. 31,32 Briey, total-protein samples (200 mg) were loaded into ltration devices (10 kD Microcon, Millipore) and centrifuged at 14 000Âg for 15 min. Then, detergent, DTT, and other low-molecular-weight components were removed using urea extraction buffer by repeated ultraltration. Aer centrifugation, concentrates were reduced with 100 mL of 100 mM DTT, containing 50 mM NH 4 HCO 3 (50 C, 30 min). Next, 100 mL of 100 mM iodoacetamide in a urea extraction buffer was added to the block-reduced cysteine residues. The samples were incubated for 40 min in darkness. Subsequently, the lters were washed with 100 mL of urea extraction buffer three times and then with 100 mL of 25 mM NH 4 HCO 3 buffer twice. Finally, the concentrate was digested at 37 C for 16 h with 200 mL trypsin (Promega), aer which the resulting peptides were collected as ltrates. Eluted peptides were subsequently acidied with TFA to a nal concentration of 0.1% and cleaned up using PepClean™ C-18 Spin Columns (Pierce) according to the manufacturer's instructions.
2.2.3 Mass spectrometry analysis and protein identication. The method of mass spectrometry analysis and protein identication referenced previous studies and slightly altered. 33 The digested peptides were analyzed on a Q-Extractive HF mass spectrometer (Thermo Scientic, Waltham, MA, USA) coupled with an online nano LC-MS/MS on an Ultimate 3000 system (Dionex). LC was then used to separate collected fractions. Briey, the peptide mixture was loaded on a reversed-phase trap column (Thermo Scientic Acclaim PepMap 100, 100 mm Â 2 cm, nano-Viper C 18 ) connected to a C-18 reversed-phase analytical column (Thermo Scientic Acclaim PepMap 100, 15 cm long, 75 mm inner diameter, 3 mm resin). Then, buffer A (0.1% formic acid) was used in the separation, with a linear gradient of buffer B (80% acetonitrile and 0.1% formic acid) at a ow rate of 300 nL min À1 . The linear gradient was processed as follows: 4-40% solution B for 98 min, 40-99% solution B for 5 min, and 99% solution B for 5 min. Furthermore, MS data were acquired using a data-dependent top 20 method, by dynamically choosing the most abundant precursor ions from the survey scan (m/z 150-2000) for higher-energy collisional dissociation (HCD) fragmentation. The instrument parameters were set as follows: automatic gain control target, 3e; 6 dynamic exclusion duration, 30 s; resolution for survey scans, 120 000 at m/z 200; resolution for HCD spectra, 30 000 at m/z 200; and isolation width, m/z 2. The normalized collision energy was 27, and the underll ratio was 0.1%. Dynamic exclusion was employed again for 30 s to prevent repetitive selection of peptides. Each sample was analyzed in three technical triplicates.
2.2.4 Protein identication. Raw data from mass spectrometric analysis were analyzed using Proteome Discoverer soware (ver. 2.2, Thermo Scientic, Bremen, Germany). Search parameters were precursor error tolerance (10 ppm) and fragment ion tolerance (0.5 Da). The peptide section was greater than six amino acids in length. Tryptic cleavage was then selected, and up to two missed cleavages were allowed. Carbamido methylation on cysteine was also set as a xed modication, and oxidation on methionine was assigned as a variable modication parameter. For protein identication, searches were conducted against the UniProtKB/Swiss-Prot ziziphus jujuba database (release 2020_12, 35 660 total entries, downloaded 12/26/20) using the Sequest HT search engine. The false discovery rate of less than 0.01 and relaxed criteria of 0.05 was used. Razor and unique peptides were also used for protein quantication.
2.2.5 Data analysis and statistical analysis. Data were analyzed using Perseus (ver.1.6.5.0) soware and the "Wu Kong" platform powered by R language (https://www.omicsolution.com/ wkomics/main/). 34 Groups with values greater than or equal to 2/ 3 were included. Missing values le were imputed using the KNN method. The t-test was then performed to analyze the differences between two groups, whereas a one-way analysis of variance test was used among more than three groups. To analyze signicant differences between the quantitative results, at least two non-null data from triplicated experimental data in the same sample group were statistically analyzed. log2 intensity values that were missing in one sample out of the groups were also inputted using a downshied normal distribution with width 0.3 and downshi 1.8 for each sample. During the screening of DEPs, we used p-value < 0.01 and jlog2 (fold change)j > 1 as the threshold criteria to determine the signicance of differences between protein expression levels.
Principal component analysis (PCA) is a common method for analyzing data dimension reduction. This method can reduce the dimension of the data and keep the feature that contributes most to the variance. It is equivalent to keeping the low-order principal components and ignoring the higher-order ones. Hierarchical cluster analysis (HCA) was set as the distance method for rows and columns was Euclidean, and the clustering method was complete.

Bioinformatic analysis
The altered abundance of proteins was further analyzed using bioinformatic methods such as Gene Ontology (GO) functional annotation, KEGG pathways, and protein-protein interaction (PPI) network predictions as described previously. 32,35 Briey, GO annotation, KEGG pathway, and PPI analyses were carried out using the online soware OmicsBean (www.omicsbean.cn). 36 The generated data were mapped to Arabidopsis thaliana. PPI information of the studied proteins were also retrieved from the IntAct molecular interaction database (http://www.ebi.ac.uk/intact/) using their gene symbols. The PPI prediction network was also analyzed using the online soware STRING version 11.0 (http://string-db.org/) and revised using Cytoscape version 3.7.2. 37 Additionally, the Molecular Complex Detection (MCODE) plugin in Cytoscape version 3.7.2, was used to screen modules of the PPI network (degree cutoff ¼ 2, node score cutoff ¼ 0.2, k-core ¼ 2, and max. depth ¼ 100).

Data analysis
3.1.1 Protein identication and quantication. As shown in Fig. 2, 4762 master proteins were identied, and 3757 were quantied. Also, 4225, 4205, 4119, and 4017 proteins were identied in 55, 76, 96, and 116 DAF, respectively. Additionally, 3757 proteins were identied in all growth stages, accounting for about 79% of all identied proteins (Fig. 2). Furthermore, the results showed that the physiological and biochemical processes were similar in the four developmental and ripening stages of jujube.
3.1.2 PCA and hierarchical cluster analysis of proteins. As a dimensionality reduction method, unsupervised PCA effectively represented the original data using fewer dimensions, retained the distribution characteristics, and maximized the variance of data in each dimension. 38 The results of the PCA and HCA of proteins are shown in Fig. 3.
The score plot of PCA showed that the rst two principal components explained 81.9% of the variability in the data, which each accounted for 56.7% and 25.2% of the total variance. Cluster analysis of the different samples also revealed that 116 and 96 DAF were aggregated into a cluster, and 76 and 55 DAF into another cluster. Overall, the results of the statistical analysis of the heat map and PCA were highly consistent, indicating that the objects to be analyzed can be divided into two clusters. Results showed that the individuals within the clusters had high similarity and that the individuals between clusters had great differences. It was therefore concluded that the samples can be divided to distinguish among these jujube samples. Therefore, samples belonging to 116 and 96 DAF were placed at the ripening stage, and those under 76 and 55 DAF were placed at the growth stage.  The proteins identied and sorted by p-value were then classied into three types, namely, CC, MF, and BP ( Fig. 5 and ESI Table 1 †). From the results, proteins of the 76 vs. 55 DAF group in the BP category were mainly related to single-organism metabolic processes (56%), catabolic processes (6%), response to chemicals (8%), single-organism cellular processes (5%), and cell wall organization or biogenesis groups (2%). Cellular components that were over-represented among proteins with increased abundance were the cell periphery (31%), apoplast (3%), and plasmodesma (2%). Furthermore, oxidoreductase   (20%), hydrolase (32%), lyase (3%), and lipid-binding (2%) activities were signicantly enriched in the GO-MF class.
In the 116 vs. 96 DAF group, proteins mainly identied in the BP category were related to responses to abiotic stimulus (24%), response to chemicals (14%), single-organism metabolic processes (22%), and single-organism cellular processes (9%). Similarly, cellular components that were signicantly enriched were related to intracellular parts (73%) and the cell periphery (7%). Meanwhile, two sub-categories were mainly related to the GO-MF class, such as protein-binding (24%) and oxidoreductase (19%) activities.
The three most enriched GO terms for DEPs in jujube are listed in Table 1. GO annotation showed that cell-wall-related categories, such as the "plasma membrane," "apoplast," and "oxidoreductase activity," were signicantly enriched GO terms in the 76 vs. 55 DAF group. It was also suggested that the formation of cell walls played a vital role in this stage. Moreover, in the 96 vs. 76 DAF group, the "intracellular" and "protein binding" enriched terms from the CC and MF ontology, respectively, inferred that identied proteins were those found mainly in the cell and referred to the ability of proteins to bind to other substances. Additionally, the "response to the abiotic stimulus" pathway was enriched in BP, which shows that jujube proteins were stimulated by a large number of external environmental factors from 76 to 96 days. In the 116 vs. 96 DAF group, the "intracellular organelle part," "protein-binding" and "response to abiotic stimulus" pathways were enriched. This result proposes that most of the metabolic activity in this group was concentrated in the organelle and that proteins such as carrier proteins were involved in the transport of substances.
3.2.2 Distribution of enriched KEGG pathway. KEGG enrichment showed that many pathways were involved in the developmental and ripening stages of jujube fruit. Results from the DEPs in the 76 vs. 55 DAF group were mainly enriched in "amino and nucleotide sugar metabolism," including "ascorbate and aldarate metabolism" (Fig. 6A), which shows that primary metabolic proteins in this group were involved in cell wall formation, consistent with the GO enrichment analysis results. Furthermore, 11 DEPs were identied in the amino and nucleotide sugar metabolism, 10 were downregulated, and only 1 was upregulated (Table 1). However, most were signicantly enriched in fatty acid biosynthesis and phenylalanine metabolism in the 96 vs. 76 DAF group (Fig. 6B). Additionally, 11 DEPs were identied to be related to fatty acid biosynthesis, of which 10 were upregulated and only 1 was downregulated. In phenylalanine metabolism, ve DEPs were identied as well, of which one was downregulated and others were upregulated (Table 1). Two biosynthetic pathways, called photosynthesis and avonoid biosynthesis-related pathways (Fig. 6C), were enriched in the 116 vs. 96 DAF group, which indicated that they were involved with changes in fruit color. Similarly, ve DEPs were also identied in photosynthesis, all of which were downregulated. The same is true for avonoid biosynthesis (Table 1).

Protein-protein interaction.
The functional signicance of a protein can be predicted on the basis of its neighboring interacting partners. 39 PPIs have led to the understanding of protein complexes and cellular protein functions through transient or stable interactions. 40 Therefore, to predict assumed functional relationships between these DEPs, a PPI of all signicantly different proteins identied at three groups was constructed and evaluated (Fig. 7).  group. However, the network contains 379 edges and 1457 nodes in the 116 vs. 96 DAF group.

Discussion
Jujube fruit undergoes a rapid developmental and ripening process that involves several differential expression proteins. In this study, proteins identied were related to fruit enlargement, aroma formation, and color change. These DEPs were also enriched in many pathways such as amino and nucleotide sugar metabolism, including ascorbate and aldarate metabolism, fatty acid biosynthesis, phenylalanine metabolism, photosynthesis, and avonoid biosynthesis.

Proteins associated with cell wall metabolism
In early developmental stages, fruit growth results from cell division, aer which the fruit produced continues growing through cell expansion. The enlargement of the jujube fruit is normally completed before its color transition. During plant growth, the increase in cell number and volume is accompanied by changes in xyloglucan content and cell wall remodeling. However, the plant cell wall is a complex network structure containing cellulose, hemicellulose, pectin, and a small number of structural proteins. 41 Therefore, many kinds of monosaccharides are present in cell wall polysaccharides, such as arabinose (Ara), rhamnose (Rha), mannose (Man), and xylose (Xyl), which can be linked in multiple positions to make them multipurpose cell wall building materials. Of these sugar units, ascorbate is among the most effective reducing agents in organisms, playing key roles in mitosis and cell growth. 42,43 Ascorbate also acts as a reducing and chelating agent and scavenges free radicals. It is an important ingredient of the antioxidative defense mechanism in cells and tissues as well. 44 Research shows that antioxidant activity gradually reduces throughout fruit development. 8 The amino and nucleotide sugar metabolic pathways (ESI Fig. 1 †), including ascorbate and aldarate metabolic pathways (ESI Fig. 2 †), were associated with fruit enlargement and were enriched in the 76 vs. 55 DAF group. Results from the KEGG annotation and enrichment analysis showed increased expression of alpha-L-arabinofuranosidase in 76 DAF. Furthermore, it is the lowly expression of a UDP-arabinopyranose mutase that catalyzes the formation of UDP-L-arabinofuranose from UDP-L-arabinopyranose. So, the content of Ara should be increased. Furthermore, the activities of UDP-glucose 4,6-dehydratase and 3,5-epimerase/4reductase were decreased, thus reducing Rha. Owing to the downregulation in hexokinase and xylan 1,4-beta-xylosidase expression, Man and Xyl were reduced in the meantime. Furthermore, UDP-glucuronate 4-epimerase contributes to the dimerization of pectin rhamnogalacturonan-II, 45 and the lower expression results in low pectin content in the cell wall. In the 76 vs. 55 DAF group, Ara was accumulated, whereas Rha, Man, and Xyl were decreased. Meanwhile, the expression of UDP-glucuronate 4epimerase was decreased as well. These results propose that cell wall synthesis is slower and cell division was slowed down. Therefore, with the fruit at the 76 DAF stage, the jujube fruit puffing slowed down and attained a stable growth stage.
The redox process of ascorbate is closely related to the growth and development of plants and has regulatory effects on the cell cycle. Studies have also shown that ascorbate promotes the cell cycle, whereas oxidative ascorbic acid (DHA) hinders normal cell cycle progression. 46 Additionally, ascorbate is involved in regulating cell division and elongation and participates in the synthesis and elongation of the cell wall. 47 Formation of GDP-L-galactose from GDP-D-mannose is catalyzed by GDP-man-nose-3 0 ,5 0 -epimerase (GME), which is the rst step in plant ascorbate biosynthesis at the glucoside level. This step is also the key link in ascorbate feedback regulation. Different redox states of cells affect this regulation, including internal and external environmental stress conditions. GME is one of the enzymes that convert from GDP-D-mannose and GDP-L-gulose to Lgalactose. L-Galactose, the last precursor in ascorbic acid synthesis, promotes ascorbic acid synthesis in tobacco cells while improving cell mitotic index. 48 Ascorbate oxidase (AO) and ascorbate peroxidase (APX) are two main oxidases that catalyze the decomposition of ascorbate. Ascorbate can also be oxidized to form monodehydroascorbate (MDA), which produces ascorbate and dehydrogenated ascorbic acid (DHA) by disproportionation without enzyme catalysis. Alternatively, MDA reacts with NADH to produce ascorbate as catalyzed by monodehydroascorbate reductase (MDHAR). Studies have shown that increased levels of MDA outside the cytoplasm promoted cell elongation, which was related to high expression levels of AO. 48 During the fruit-puffing period, fruit suffer stronger oxidation stress; however, during stable growth periods, oxidation stress is mitigated. In this study, GME, AO, and APX were downregulated, whereas MDHAR was upregulated, which indicated that ascorbate production decreases gradually when jujube fruit attains a stable growth stage.
As shown in Table 1 and Fig. 7B, C, MDAR1, RHM1, and NRS/ ER are identied. In Table 1, MDAR1 is involved in ascorbate and aldarate metabolism, and RHM1 and NRS/ER are involved in amino and nucleotide sugar metabolism. As illustrated in Fig. 7B, GPX1, FSD2, and APX3 are connected to MDAR1 directly, whereas BCCP2, At1g24360, EMB3147, and At5g64250 are connected to MDAR1 indirectly. A possible fact is that GPX1, FSD2, and APX3 are closely related to ascorbate and aldarate metabolism and act on jujube. GLC, GME, and TT4 are also connected to RHM1 and NRS/ER directly. Similarly, TT2 and TT10 are connected to TT4 directly as well. All results propose that GLC, GME, TT4, RHM1, and NRS/ER are associated with amino and nucleotide sugar metabolism and that TT2 and TT10 are associated with other KEGG pathways in the 76 vs. 55 DAF group.

Proteins associated with energy metabolism and production of phenolic compound
Jujube avor is a combination of volatile organic compounds, which are synthesized during ripening and characterize a specic fruit or variety. 49 As is the case with many other fruits, volatiles are derived from nutritionally important metabolites such as essential amino acids, fatty acids, and carotenoids. 50 In most plant tissues, the biosynthesis of fatty acids, in which acetyl-CoA is converted to C 16 fatty acids, occurs in the cytosol. Also, a heterogeneous group of molecules with saturated and unsaturated, straight-chain, branched-chain, and cyclic structures bearing various functional groups such as alcohols, aldehydes, ketones, and esters, is found as well in fresh jujube fruits. 51 In higher plants, the amino acid phenylalanine (Phe) is a substrate of phenylpropanoid metabolism, a network of secondary metabolic pathways in the cytosol. 52,53 Plants use Phe as a precursor to synthesize various phenolic natural products, including diverse and abundant phenylpropanoids such as anthocyanins, avonoids, isoavonoids, tannins, and volatiles. 54,55 Most phenylpropanoids are also related to fruit coloring. Phe is an aromatic amino acid required for protein synthesis in plants. 53 Yet in many cases, more phenylalanine is allocated to the phenylpropanoid metabolism than to protein biosynthesis. 52 Fatty acid biosynthesis (ESI Fig. 3 †) and phenylalanine metabolism (ESI Fig. 4 †) were enriched in the 96 vs. 76 DAF group. The most important process in fatty acid biosynthesis is the formation of malonyl-CoA under the action of acetyl-CoA carboxylase (ACCase). Here malonyl-CoA, in turn, acts as a building block for fatty acid elongation. ACCase is composed of a biotin carboxyl carrier protein (BCCP). Also, three subunits, namely, biotin carboxylase (BC), BCCP, and carboxyl transferase (CT), were linked with BCCP. 56 In our study, these enzymes, BC, BCCP, and CT, were signicantly increased, suggesting that malonyl-CoA synthesis was promoted.
In higher plants, the acyl-carrier-protein (ACP) is an important cofactor in fatty acid biosynthesis, 57 which is closely related to the content, composition, and proportion of unsaturated fatty acid storage by plants. During the synthesis of long-chain fatty acids, the starting unit is accepted to form the ACP complex, which is then transferred to the long chain of macromolecules that need to be extended. The expression of [ACP] S-malonyltransferase is thus increased, which then catalyzes the synthesis of the malonyl-[ACP] by the ACP and malonyl-CoA. Subsequently, 3-oxoacyl-[ACP] synthase II and 3oxoacyl-[ACP] reductase, which are associated with the synthesis of long-chain fatty acid, was signicantly increased in this group. The fatty acyl-ACP thioesterase B is also involved in the elongation of fatty acid chains. Alternatively, long-chain acyl-CoA synthetase participates in the beta-oxidation process of fatty acids. Furthermore, numerous fatty acid synthase subunits, which catalyze the carbon chain extension, were upregulated in the 96 DAF group, showing that fatty acids were rapidly synthesized around 96 DAF. This result correlates with the high demand for lipid biosynthesis during fruit ripening, which participates in the formation of jujube aroma as a precursor and provides a storage matrix for the accumulating carotenoids. 14 Additionally, compared with all hub genes, in the 96 vs. 76 DAF group (Fig. 7e and Table 1), identied common genes were FaTA, At1g24360, EMB3147, CAC3, and CAC2. Furthermore, P5CS1, ACC1, FAB1, and E1-BETA-1 were closely related to those common genes. This result indicates that this hub module is mainly involved in fatty acid biosynthesis and proposes that P5CS1, ACC1, FAB1, and E1-BETA-1 are also associated with fatty acid biosynthesis.
Phenylalanine ammonia lyase was downregulated, illustrating that the content of phenylalanine was increased. Therefore, more phenylalanine owed into the phenylpropanoid biosynthesis, preparing jujube for color changes. According to a recent report, the content of phenylalanine increased gradually with the coloring of the pericarp, 58 and aspartate aminotransferase (AAT) can interact with the amino acid Phe and phenylpyruvate. 59 AAT were up-expressed in "96 vs. 76 DAF" group, which indicated that metabolism was active between phenylalanine and phenylpyruvate. Additionally, a previous study showed that phenylacetaldehyde (PHA) and phenylethylalcohol are aromatic aroma compounds directly catalyzed by phenylalanine lyase. These are important scent compounds in numerous fruits. L-Tryptophan decarboxylase catalyzes the synthesis of phenylethylamine, and then PHA is synthesized by primary amine oxidase. The expression of the primary amine oxidase was increased, in 96 DAF, which proposes that aromatic component-synthesis was accelerated. Furthermore, 3-hydroxybutyryl-CoA dehydrogenase was upregulated, resulting in the increased synthesis of acetyl-CoA and succinyl-CoA, and an increased entry into the tricarboxylic acid cycle. This result proposes that the growth of jujube fruit needed more energy, which was related to the biosynthesis of secondary metabolites as identied in 96 DAF.

Proteins associated with color changes
High-quality fruit typically have deep red coloration, high sugar content, and a balanced sugar-to-acid ratio. 60,61 The color change of jujube fruit depends on the content and proportion of chlorophyll, carotenoids, avonoid, and anthocyanins in the fruit peel. 62 The type and content in the pigment also determines the color of jujube fruit. The green and yellow colors of the fruit epidermis are derived from chlorophyll and carotenoids as well, whereas the red and reddish-purple coloration is mainly determined by the content and proportion of phenolic acid, avonoids, avanols, and anthocyanins, which are associated with the antioxidant activity of jujube fruit. 63 Photosynthesis (ESI Fig. 5 †) and avonoid biosynthesis (ESI Fig. 6 †) pathways were associated with fruit color changes and were enriched in the 116 DAF vs. 96 DAF group. Converting chloroplast into chromoplast is an important part of the ripening process and is typically associated with dismantling the photosynthetic machinery and accumulating carotenoids in chromoplasts. 64 Also, chlorophyll is a key photosynthetic pigment in plant chloroplasts. 65 During the light reaction, the thylakoid membrane of chloroplasts absorbs light energy, which decomposes water and releases oxygen. This membrane then stores light energy in the form of ATP and NADPH through the photosynthetic electron transport chain. The photosynthetic electron transport chain is composed of four complexes on the thylakoid membrane: the photosynthetic system II(PSII), cytochrome b6/f complex (Cyt b6/f), photosynthetic system (PSI), and ATP synthase complex (ATPase). The Cyt b6/f, located between PSII and PSI in the photosynthetic electron transport chain, is a plastoquinone (PQH2)/plastoblue (PC) oxidoreductase. In contrast, Cyt b6f plays a key role in electron transfer and energy conversion during photosynthesis. First, it acts as an electron carrier to mediate electron transfer between PSII and PSI. Furthermore, it is used as an energy converter to transfer protons from the outer side of the thylakoid membrane to its inner side using the electron-free energy released in the process of electron transport. This step results in the formation of a transmembrane proton electrochemical gradient, which provides power for ATP synthesis by ATP synthase. From the results, 10 DEPs mainly related to the photosynthesis pathway, that is, four proteins related to photosystem I (PSI), four related to photosystem II (PSII), and two related to the cytochrome b6/f complex, were identied. It means that jujube decreased photosynthesis by decreasing their enzyme activity in the 116 vs. 96 DAF group. The decrease in photosynthesis intensity should be related to the decrease in chlorophyll content in the process of jujube peel color progression from semi-red to full red. In this stage, the energy required for jujube metabolism is proposed to be reduced by photosynthesis. Additionally, as shown in Table 1 and Fig. 7f, PSBR and DRT112 were common genes. Studies have shown that PSBR and DRT112 are involved in photosynthesis. According to the edge color, most nodes were closely related to ACP4, PSBQ-2, PSAE-2, PETC, SBPASE, PSAH2, NPQ4, GAPB, PSAK, AOR, ENH1, and PSBP-1. This result indicates that these hub genes led to the fruit changes observed during this period (116 vs. 96 DAF group).
Flavonoids are a series of natural compounds that are widely distributed in plants and play important roles in diverse BPs such as imparting various colors, as they are major pigmentation factors. 35 Various avonoid substances have been reported in jujube fruit, including catechins, epicatechins, naringenin, and other avanols. 66 In the 116 vs. 96 DAF group, it was observed that the content of avanols or avonols increased by promoting the process of phenylpropanoid metabolism in the early stage of ripening. The expression of these key enzymes such as chalcone synthase, cinnamate-4-hydroxylase (C4H), avanone 4-reductase, anthocyanidin reductase, and chalcone isomerase, were, however, decreased, which indicated that the content of avonoids in jujube fruit decreased from 96 DAF to 116 DAF. Previous studies by X. Li et al. also had the same conclusion. 66

Conclusion
To better understand the change of melatonin in jujube fruit developmental and ripening, a LFQ proteomic tool was used to investigate the expression proteins present during the developmental and ripening stages (55,76,96,and 116 DAF). Results showed that physiological traits of the seedless jujube fruits varied signicantly at four growth stages, with a remarkable correlation with fruit quality changes. From the study, 262, 315, and 532 DEPs were identied in the 76 vs. 55 DAF, 96 vs. 76 DAF, and 116 vs. 96 DAF groups, respectively. Hierarchical clustering of stage-responsive proteins also revealed the metabolic changes during fruit growth. Similarly, GO and KEGG annotation and enrichment analysis of DEPs showed that the 76 vs. 55 DAF group pathways was amino and nucleotide sugar metabolic pathways, including those related to ascorbate and aldarate metabolism. These pathways were associated with cell division and resistance. Fatty acid biosynthesis and phenylalanine metabolism was mainly enriched in the 96 vs. 76 DAF group. Also, aromatic compounds were synthesized in large quantities at this stage. During the ripening stage, 55 DEPs were involved in photosynthesis and avonoid biosynthesis. In the 116 vs. 96 DAF group as well, the color of jujube turned red, avonoids were produced in large quantities, and photosynthesis was signicantly reduced. Hence, our results established a link between protein proles and growth phenotypes, which will help improve our understanding of seedless jujube fruit growth at the proteomic level. Our results also added information on proteome changes during ripening.

Conflicts of interest
The authors declare that there are no competing nancial interests.