Kenaf seed materials “Q303” and “L332,” which are cultivated in southern China, were provided by the Institute of Bast Fiber Crops, Chinese Academy of Agriculture Sciences. In this study, it was observed that the leaves of kenaf material “L332” belong to the lobed-leaf type and the leaves of kenaf material “Q303” belong to the round-like leaf type. Our experimental site was located in the Bairuopu Innovative Experimental Base at the Institute of Bast Fiber Crops, Chinese Academy of Agricultural Sciences in Changsha. The experimental area had a typical subtropical continental monsoon climate. A field with the same water and fertilizer conditions was selected as the experimental site. Two kenaf materials were sown on May 16, 2019 and harvested on October 20, 2019. The whole growth period lasted for 155 days. At the mature stage (September 17), all leaves within 50 cm below the top of the kenaf plant were collected and mixed for carrying out further analyses. The mixed samples were from the leaves of 20 kenaf plants. This work set 3 biological replicates for every kenaf material. For every kenaf material, this study gathered 30 g leaf samples from an individual plant. We froze sample leaves with liquid nitrogen at once, followed by preservation under −80°C until the application for RNA isolation, protein separation, and quantitative real-time PCR analyses.

This study determined kenaf leaf size using the graduated scale, while leaf weight using the electronic balance, and the total protein of kenaf leaves using the Kjeldahl method (Grzeszczuk et al., 2018). The leaves of kenaf were subjected to 0.5-h deactivation under 105°C, followed by drying under 65°C to constant weight. Subsequently, the leaves of kenaf were crushed to determine the protein content. We prepared 3 biological replicates of every kenaf material.

The leaves in the biological replicate for every sample were grinded to powder with liquid nitrogen. According to Chai et al. (2014), extraction of total RNA in two kenaf leaf samples was conducted. Total RNA extracted from kenaf leaves was dissolved in RNase-free water (TIANGEN, China). The A260/A280 ratio was measured using the Nanodrop ND-1000 system (Thermo Scientific, United States) and adopted for checking the RNA concentration in kenaf leaves, and 1.5% agarose gels were adopted for detecting the integrity of the RNA. After the RNA of kenaf samples was qualified, the library was prepared for the qualified samples. cDNA libraries were created with the use of the NEBNext^® Ultra™ RNA Library Prep Kit for Illumina^® (NEB, United States) in accordance with specific protocols. Kenaf mRNA purification was conducted using the extracted total RNA (4 μg) with oligo (dT) magnetic beads (Thermo Scientific) following the instructions of the manufacturer. Then, synthesis of first- and second-strand cDNA by reverse transcription was conducted, with mRNA being a template. After purification of double-stranded cDNA and then end repair, polyadenylation, and addition of adaptor sequences, DNA fragment sorting and PCR amplification were performed. Finally, a cDNA library was constructed. After purification of PCR products (AMPure XP system), we assessed cDNA library quality on an Agilent Bioanalyzer 2100 system. Finally, the cDNA library was constructed and sequenced after quality inspection.

The cDNA libraries for those 3 biological replicates of kenaf samples were constructed according to Cui et al. (2019). Subsequently, sequencing was carried out by an Illumina NovaSeq 6000 instrument by Shanghai Applied Protein Technology Co., Ltd. Sequencing was completed by an Illumina NovaSeq 6000 instrument from Shanghai Applied Protein Technology Co., Ltd. (Shanghai, China). Besides, through CASAVA base calling analysis, raw imaging data acquired by Illumina high-throughput sequencing were transformed into original sequenced reads. Besides, these findings were kept as the fastq file format. In the fastq file, four lines are regarded as a basic unit and correspond to the sequencing information of a sequence. The clean data (clean reads) were obtained by filter_fq software through filtering out low-quality and adaptor sequences as well as reads that contained the poly-N sequences. High-quality sequence data can be obtained by filtering out sequences that are too short or contain uncertain bases. We calculated the Q20, Q30, GC levels, as well as clean reads sequence duplicate degrees. Using Trinity software package¹ (Grabherr et al., 2011) to assemble and splice the clean reads, data assembly was later conducted. Trinity software workflow was used for de novo transcriptome assembly. It contains 3 separate software modules, respectively, Inchworm, Butterfly and Chrysalis. In addition, RNA-seq read data were processed by the three software modules in turn. Transcripts <200 bp were fully removed.

This work annotated de novo assembled unigenes with the National Center for Biotechnology Information (NCBI) non-redundant protein sequence database (NR), the database for manual annotation and protein sequence reviewing (Swiss-Prot), a protein family database (Pfam), an integrated database on the basis of Gene Ontology (GO), as well as a genome database on Kyoto Encyclopedia of Genes and Genomes (KEGG), and threshold value was set as an E-value < 10^–5. After gene GO annotation, the annotated genes were categorized in accordance with GO (Biological processes BPs, Cellular Components CCs, Molecular Functions MFs). KEGG database contributes to studying genes and expression information as a whole network. KEGG pathway analysis was accomplished through KOBAS 2.0 test statistical enrichment (Chen et al., 2011).

In this experiment, the Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) value was used for transcription and quantification of gene expression, and DEG-seq was applied to explore the differential expression of samples. We controlled false discovery rate using the p-value adjustment methods of Benjamini and Hochberg (1995). The screening criteria for obvious differences were a p < 0.05 and | log2 (FC)| ≥ 1. The expression was deemed to be different between the two samples. Cluster Profiler R software was employed to realize GO as well as KEGG analysis on DEGs. Ap, the PlantTFDB² was used to analyze transcription factors (TFs) in DEGs.

Protein was extracted in kenaf leaves of every sample based on previous description, with three biological replicates for each treatment (An et al., 2018). BCA Protein Assay Kit (Bio-Rad, United States) was adopted for determining the protein concentration.

Protein was digested based on Liu et al.’s method (Liu et al., 2018), with some modifications. In filter-aided sample preparation (FASP Digestion), we poured all samples in the buffer [consisting of 150 mM Tris-HCl (pH 8.0), 4% SDS, 100 mM DTT], followed by 7-min boiling of mixed solution and cooling immediately. The samples were treated following specific protocols of TMT Kit (Thermo Scientific, United States). This work thereafter utilized TMT 6-plex to categorize trypsin (that contained 100 μg protein) for 126-tag (HPC-1), 127-tag (HPC-2), 128-tag (HPC-3), 129-tag (LPC-1) 130-tag (LPC-2) and 131-tag (LPC-3) labeling. The label assimilation was checked, and all the above 6 labeled samples in diverse set were later combined.

Thereafter, this work loaded the mixed peptide sample to the reversed-phase trap column (nanoViper C18, 100 μm*2 cm; Thermo Scientific Acclaim PepMap100) equipped with the C18 reversed-phase analytical column (length, 10-cm; 3-μm resin; inner diameter, 75-μm; Thermo Scientific Easy Column) within buffer A consisting of 0.1% formic acid, followed by separation using the gradient buffer B (0.1% formic acid and 84% acetonitrile) at a 300 nl/min flow rate under the control by IntelliFlow Technology. Afterward, peptides were detected using Q-Exactive High-Resolution Mass Spectrometer (Thermo Scientific). This study obtained survey scans at the 70,000 at m/z 200 resolution, whereas HCD spectra were obtained at 17,500 and 35,000 at m/z 200 (TMT 6plex and 10plex, respectively), with isolation width being set at 2 m/z. Meanwhile, this work set the underfill ratio (specifying the minimal target value percentage possibly reached at the maximal fill time) and normalized collision energy as 0.1% and 30 eV separately.

This work obtained original MS data in RAW files. This work utilized Proteome Discoverer 1.4 software as well as MASCOT engine (Matrix Science, London, United Kingdom; version 2.2) for searching MS/MS spectra against kenaf database with the settings below, enzyme trypsin and two as the maximal missed cleavage allowed; variable oxidation modification (M), fixed carbamidomethyl modification (C), TMT-6plex (K) and TMT-6plex (N-term); and mass tolerance for peptide ions and fragment ions were 20 ppa and 0.1 Da, separately, at both protein and peptide levels of false discovery rate (FDR) < 0.01. Proteins were detected and quantified using specific protein peptides. Differentially abundant proteins (DAPs) were detected by the thresholds of FC ≤ 0.83 or ≥ 1.2 and p < 0.05.

The GO annotation based on three categories, including BPs, MFs, and CCs, together with KEGG analysis, was carried out by Fisher’s exact test (p < 0.05). This work also adopted STRING 9.0 software for constructing the PPI network³. Later, Java Treeview software⁴ and Cluster 3.0⁵ were utilized for hierarchical clustering.

By adopting SYBR Green PCR master mix (Aidlab Biotechnologies, Co., Ltd.), qRT-PCR was conducted to analyze 20 potential DEGs levels according to the manufacturers’ instructions on Bio-Rad CFX96 Touch detection system (Bio-Rad, Richmond, CA, United States). The total RNA isolation method and cDNA preparation method were the same as previous description. Genomic DNA was removed using gDNA Eraser (Tsingke), later, the isolated total RNA was prepared into cDNA with Goldenstar™ RT6 cDNA Synthesis Kit Ver. 2 through reverse transcription. Primers were designed with Primer 5.0 software (Supplementary Table 5) for qPCR experiment, followed by blasting of these gene sequences to NCBI database. This study utilized around 1 μg total RNA in every reaction. For removing gDNA, we cultured sample for a 2-min period under 42°C and for another 5-min period under 60°C. Reaction buffer (10 μl) in step 1 (20-μl system) was utilized in every reverse transcription. After mixing, the mixture was subject to 30-min incubation under 50°C and 5-min heat shock under 85°C. The PCR conditions were shown below, 30-s under 95°C; 5-s under 95°C and 55-s under 60°C for altogether 40 cycles. Meanwhile, this work utilized gene-specific primers (F/R) of 20 candidate DEGs and the kenaf internal control gene GhACTIN (GenBank accession no. AY305733) (Lu et al., 2018), with action being the control. Further, 2^–ΔΔCt approach was utilized for determining 20 gene levels. qRT–PCR procedure was conducted in triplicate.

For verifying the TMT-measured protein levels, a PRM experiment was adopted for further quantifying the expression level of the selected proteins. For every sample, we added the PRTC stable isotope peptide as control. After protein extraction, sample (200 μg) was subject to trypsin hydrolysis. Meanwhile, the C18 cartridge was utilized to desalt peptides; 40 μl of freeze-dried peptide was added to 0.1% formic acid solution, followed by peptide quantification. Peptides of the same mass were mixed and separated by Easy nLC-1200 system (Thermo Fisher Scientific, MA, United States). Meanwhile, this work utilized acetonitrile (ACN) within 40 min with a 1-h liquid chromatographic gradient of 5–30%, and PRM mass spectrometry was carried out using the Q Exactive HF mass spectrometer (Thermo Scientific) in the positive-ion mode, with a full MS1 scanning resolution ratio of 60,000 (200 m/z). This work also set the target values of maximal ion injection time and automatic gain control (AGC) at 200 ms and 3e6, separately; after full MS scanning, 25 PRMs (MS2 scans) were performed, at the resolution and maximal injection time of 120 ms and 30,000 (m/z 200), and agc3e6. In addition, a second window was utilized to isolate target peptides; whereas the normalized collision energy (Grzeszczuk et al., 2018) was adopted for ion activation/dissociation within the high energy dissociation (HCD) collision pool. The 20 target proteins quantified by PRM were imported into Skyline software⁶, the peptide settings were selected, the background proteome database file was added, the protein quantitative polypeptide was selected according to the ion signal in the spectral library, the list of related peptides with retention time was derived from Skyline, and the quantitative results of each peptide of target protein were manually checked. Finally, each sample took 1 μg with 20 fmol heavy isotope-labeled peptide section PRTC as the internal label, and the target peptides and proteins were quantitatively analyzed by the same chromatographic conditions and mass spectrometry methods as the previous PRM method. The signal strength of the single peptide sequence of each significantly changed protein was measured for all samples relative to control (Gillet et al., 2012).

Kenaf leaves are applied as livestock feed because of their high protein concentration and large biomass. The crude protein in leaves of two kenaf materials were determined. As a result, mature “Q303” leaves had markedly increased protein levels compared with mature “L332” (Figure 1). The average percentage of crude protein content in leaves of kenaf material “Q303” reached 28.43%. Compared with kenaf material “Q303,” the percentage of crude protein content in leaves of material “L332” decreased by 6.13%.

For summarizing the transcriptomes of the two kenaf materials in the mature period, this work built six cDNA libraries (i.e., L332 and Q303, three repeats). Altogether 49.33 and 47.74 million raw sequences were generated in cleft Q303 and L332 libraries, separately. Later, adaptor sequences, low-quality reads and those with uncertain base ratios >10% were eliminated, leaving 46.32 and 47.78 million clean reads with Q30 base percentages and GC contents of 94.23-94.38 and 47.07-48.00%, respectively (Supplementary Table 1). A total of 222,979 transcripts (range, 201–2,000 bp), with altogether 101,679 unigenes (>200 bp) were discovered (Table 1). The size distributions of unigenes and transcripts are presented in Supplementary Figure 1.

For determining those identified transcripts’ candidate functions, this study utilized Basic Local Alignment Search Tool (BLAST) to annotate unigenes on the basis of 5 databases, like National Center for Biotechnology Information non-redundant protein sequences (NR) (60,171, 59.18% of all the identified unigenes), SwissProt (40,606, 39.94%), Protein families (Pfam) (31,431, 30.91%), Gene Ontology (GO) (34,448, 33.88%), and KEGG (10,456, 10.28%) databases. Based on the above results, NR database had the most functional annotations, which suggested that all the 61141 unigenes corresponded to sequences from one or more public databases. Finally, there were 6879 functionally annotated unigenes from different databases (Supplementary Table 2).

Of those, 3,778 DEGs were found by p < 0.05 and log2-FC > 1 thresholds in leaves of two kenaf materials. Q303 group showed 1,350 and 2,428 genes with up-regulation and down-regulation, respectively, relative to L332 group. Supplementary Figure 2A displays a volcano plot of the above findings. Table 2 summarized data regarding the identified DEGs. Many of the above genes obtained from annotation had not been characterized before.

For better understanding DEGs’ function, this work carried out bioinformatics analysis through hierarchical clustering and GO analysis. As revealed by GO functional annotation, many DEGs were enriched into serine family amino acid catabolic process and glycine catabolic process with regard to BP terms; oxidoreductase activity and glycine dehydrogenase (decarboxylating) activity with regard to MF terms; and ribosome and intracellular non-membrane-bounded organelle with regard to CC terms (Figure 2A).

Additionally, this work conducted KEGG analysis for evaluating DEGs. As a result, most DEGs were associated with secondary metabolite biosynthesis and antibiotic metabolic pathway biosynthesis in the leaves of the two kenaf materials (Figure 2C). Besides, this study conducted hierarchical clustering for improving the understanding on alterations of protein synthesis-associated DEG levels (Figure 3A). In all, 141 protein synthesis-related pathways, including amino acid biosynthesis and metabolism, energy metabolism, carbohydrate metabolism, genetic information processing, and carbon metabolism, were clustered closely.

For exploring the mechanism underlying high protein content in kenaf leaves, this work carried out quantitative proteomics analysis by LC-MS/MS analysis and TMT platform in mature stage, for the sake of complementing transcriptomic analysis. Proteomic analysis identified altogether 66,347 spectra, 26,972 unique peptides and 32,558 identified peptides, among which, we discovered 7,196 proteins (Table 1 and Supplementary Table 3). As for distribution of protein mass, proteins whose molecular weights (MWs) were >9 kDa showed a broad range as well as favorable coverage, and the maximal distribution area reached 10–40 kDa (Supplementary Figure 3A). According to protein peptide quantification, protein number declined as matching peptide increased (Supplementary Figure 3B).

Of the above proteins, we discovered 1,320 differentially abundant protein (DAPs) upon the upregulated and downregulated thresholds being FC > 1.2 as well as p < 0.05, separately, including 636 with higher abundance levels, whereas 684 with lower abundance levels between Q303/L332 groups, as displayed by the volcano plot Supplementary Figure 2B.

A higher number of DEGs (1,350 with up-regulation and 2,428 with down-regulation) was obtained than DAPs (636 with up-regulation and 684 with down-regulation). For evaluating the relation of proteomic and transcriptomic alterations during protein synthesis, correlation analysis was performed using the quantification of DAPs and DEGs. As a result, 134 DAPs and their corresponding DEGs were identified. Of these, 122 DAP abundances increased and 12 DAP abundances decreased (Figure 5A). The results showed that there were more DEGs than DAPs, and there were significant differences in transcription level and protein abundance.

Moreover, Pearson’s correlation test suggested that FCs of DAPs showed moderate correlation with specific DEGs (r = 0.3379, p < 0.01), which indicated that transcript contents were more closely related to protein levels (Figure 5B). Meanwhile, FCs in DAPs exhibited positive relation to DEGs (r = 0.7756, p < 0.01) (Figure 5C).

For better understanding those co-expressed DEG-DAP genes for their biological roles, this work performed enrichment analysis on the basis of GO as well as KEGG analysis. Consequently, the markedly enriched BP terms were cellular processes and metabolic processes, whereas cellular parts and binding and catalytic activities were the obviously enriched CC and MF terms, separately (Supplementary Figure 4A). According to the association analysis, 134 DAPs were enriched in 160 metabolic pathways, 10 significantly enriched pathways in the transcriptome, and 20 significantly enriched pathways in the proteome. There were fifty-seven common pathways between proteomic and transcriptomic data. The comparison demonstrated 2 common pathways, namely, phenylpropanoid biosynthesis, and starch/sucrose metabolism pathways, in proteomic and transcriptomic data (Supplementary Figure 4B). Therefore, these shared metabolic pathways might exert possible functions in the protein synthesis of kenaf leaves.

For validating whether our TMT and RNA-Seq data were reliable, this work conducted correlation and expression analyses by qPCR, and later acquired fragments per kilobase per million reads mapped (FPKM) values based on proteomic and transcriptomic data. Those 20 most significant DEGs were enriched into starch and sucrose metabolism, carbon fixation in photosynthetic organisms, glutathione metabolism, aminoacyl-tRNA biosynthesis, nitrogen metabolism, glycine/serine/threonine metabolism, and other protein synthesis of kenaf leaves were tested by qPCR experiments, and the results are shown in Figure 6. Specifically, most of these DEGs were significantly upregulated, including the starch and sucrose metabolism pathway related genes-sucrose-phosphate synthase (SPS), beta-fructofuranosidase (INV1 and INV2), beta-amylase 3 (BAM3-1 and BAM3-2) and sucrose synthase 6 (SUS6); glycolysis/gluconeogenesis metabolism related gene-aldehyde dehydrogenase (ALDH); carbon fixation in photosynthetic organisms related gene-fructose 1,6-bisphosphate 1-phosphatase activity (FBP); nitrogen metabolism related genes-nitrate reductase (NR), nitrate transporter (NRT), nitrite reductase (NirA); serine and threonine metabolism related gene-glycine dehydrogenase (GLDC),Glycine dehydrogenase (decarboxylating) A (GDCSPA) and aminoacyl-tRNA biosynthesis related gene-proline-tRNA ligase (PARS); and cysteine/methionine metabolism associated gene-S-adenosylmethionine decarboxylase (SAMDC), alanine/aspartate/glutamate metabolism associated gene-glutamate decarboxylase (GAD). Translation initiation factor 4E (EIF4E), transaldolase-like (TAL), glucan endo-1,3-beta-glucosidase (EGLC), and L-3-cyanoalanine synthase D-1 (CYSD1) were downregulated. Additionally, 14 potential genes, like FBP (r = 0.9349, p < 0.01), SPS (r = 0.8301, p < 0.05), INV1 (r = 0.9106, p < 0.05), INV2 (r = 0.7154, p < 0.05), NRT (r = 0.6840, p < 0.05), NR (r = 0.6553, p < 0.05), NirA (r = 0.8855, p < 0.05), PARS (r = 0.8863, p < 0.01), GLDC (r = 0.8451, p < 0.01), GDCSPA (r = 0.7583, p < 0.01), SAMDC (r = 0.7693, p < 0.05), ALDH (r = 0.7815, p < 0.05), EIF4E (r = 0.8543, p < 0.01), and CYSD1 (r = 0.7442, p < 0.05), were strongly related to RNA-Seq results, whereas 6 genes showed low relation to related protein levels (Figure 6 and Table 3). Generally speaking, qPCR analysis verified gene expression profiles acquired based on proteomic and transcriptomic data, indicating the reliability of our observations.

Moreover, to further validate the proteomics data, we quantified the protein abundance of 40 randomly selected genes by PRM and analyzed the protein abundance of 20 specific genes associated with protein synthesis, 19 of which were successfully quantified. Fifteen of the 19 (78.9%) proteins exhibited close abundance trends in PRM compared with TMT analysis, including ferredoxin-nitrite reductase (NirA), glutathione S-transferase DHAR2 (DHAR2), prolyl-tRNA synthetase (PARS), 3-phosphoshikimate 1-carboxy vinyl transferase (EPSPS), fructose-bisphosphate aldolase (ALD), peroxidase (PRX), ferredoxin-NADP+ reductase (petH), catalase isozyme 2 (CAT2), divinyl chlorophyllide an 8-vinyl-reductase (DVR), 1,4-alpha-glucan-branching enzyme 1 (GBE1), glycine dehydrogenase (GLDC), beta-fructofuranosidase INV1 (INV1), fructose 1,6-bisphosphate 1-phosphatase activity (FBP), cysteine synthase (Csase), and glycine-tRNA ligase (GlyRS). Additionally, four genes [glycerate dehydrogenase (HPR), glyceraldehyde-3-phosphate dehydrogenase GAPA2 (GAPA2), fructokinase-2 (FRK2), glutamine synthetase cytosolic isozyme 2 (GS1-2)] exhibited different levels relative to TMT-measured protein levels (Table 4). Thus, the results showed that PRM results were well correlated with proteomics analysis results, besides, some genes related to protein synthesis-associated metabolic pathways were consistently up-regulated and down-regulated within proteome and transcriptome (Figure 7).

The high protein content in leaves is the key factor affecting kenaf’s multi-functional breeding. Illustrating the molecular mechanism of protein synthesis in leaves is helpful for the development of forage products using kenaf leaves. However, the underlying mechanism of kenaf protein synthesis is greatly unknown. According to TMT and RNA-seq data, the current work explored the different proteomic and transcriptomic data of two kenaf materials at the same stage. Since the transcriptome database is applied in identifying proteins, transcriptomic data assembly and sequencing quality is important for the following study. Totally, this work assembled 101,679 unigenes (>200 bp) into kenaf transcriptome, and the total was much higher than that formerly reported for this species as well as other Malvaceae Juss. plants, for example, [55,623 unigenes by Tang et al. (2021); 71,318 unigenes by Li et al. (2016)], and Hibiscus rosa-sinensis (30988 unigenes) (Trivellini et al., 2016). The obtained percentages of the Q20 bases (98.08-98.15%), Q30 bases (94.23-94.38%), with close GC concentrations (47.07-48.00%) to other researches on kenaf transcriptome, e.g., (98.19–98.31%), (94.45-94.71%), and (46.15-46.47%) (Tang et al., 2021), respectively. Li et al. (2016) only studied Q20 bases (97.70-97.77%) and GC contents (45.42-46.05%). In total, 60171 (59.18%) and 40606 (39.94%) of all unigenes detected in the present work matched those in SwissProt and NR databases, separately. The findings supported the results obtained by Hibiscus L. (20,547 and 13,184; 66.31 and 42.55%) (Trivellini et al., 2016). Therefore, our results provide a wide range of sequence and unigene resources for Malvaceae Juss.

In addition, 66,347 match spectra, 26,972 unique peptides, and 7,196 proteins were detected in accordance with the transcriptome data from kenaf leaves. RNA-seq and protein sequencing methods have found and annotated many genes and proteins, providing the foundation for a more correct and thorough description of molecular processes and analysis of genetic regulation mechanisms and complex physiological processes (Vanessa et al., 2012). The data presented here are sufficient and accurate and will provide useful guidance tools for the study of protein synthesis in kenaf leaves and other mallow plants. In addition, there were a lot of “predicted,” “putative,” or “uncharacterized” transcripts and proteins in the annotation results, suggesting that the nature of the analysis was restricted because of the lack of detailed genomic information (Niu et al., 2020). As a result, the functions of these unknown or unidentified genes and proteins in the protein synthesis of kenaf leaves should be explored in further researches.

The regulation of storage protein gene expression is complex and includes transcription regulation, processing regulation for transcription products, translation regulation, hormone regulation, and so on (Kroj et al., 2003). As a result, the differences in the expression of genes and proteins in a single period alone cannot fully and correctly reflect the cause of protein accumulation in kenaf leaves. In the current work, the comparative analysis of gene expression and protein levels suggested that there existed more DEGs than DAPs. This may be due to protein post-translational modifications that manage the degradation or secretion of the protein, the technical limitations of MS-based proteomics, and the availability of a few samples. These reasons limit the ability to identify proteins (Zhou et al., 2015). Investigating the association between proteins and the network formed by their interaction is significant to show the roles of proteins (Xing et al., 2016). The PPI analysis showed that the interaction between proteins was obvious, and there was a complex network between interacting proteins. The interactions between proteins constitute a major part of the cellular biochemical reaction network, which is of great significance in regulating cells and their signals (Liu et al., 2021). In addition, correlation analysis demonstrated that there existed a weak correlation between the proteome and transcriptome, suggesting that there existed low correlation between transcription level and protein abundance. The findings were similar to those found in former studies, indicating that post-transcriptional and post-translational regulation, reversible phosphorylation, cell splicing events and translation efficiency exert a vital function in the regulation of protein accumulation in leaves (Chen et al., 2019). Therefore, gene translation and post-translational modification could be important methods of protein accumulation in leaves.

The GO and KEGG functional enrichment provides the definition and description of gene and protein functions and integrates genomic, chemical and systemic functional information. GO term analysis showed that in addition to the co-enrichment of glycine dehydrogenase (decarboxylation) activity and serine family amino acid catabolic processes related to amino acid metabolism, oxidoreductase activity was also co-enriched. Similar to former researches, the target genes can be classified into different categories based on their roles, such as amino acid anabolism and peroxidase activity (Skadsen and Cherry, 1983). Peroxidase participates in the photorespiration of plant cells and redox equilibrium reactions (Jin et al., 2013). Glycine is mainly used as a glycine residue in leaf protein synthesis (Guo et al., 2021). These modifications improved the efficiency of photosynthesis and provided raw materials for protein synthesis. KEGG analysis showed that starch and sucrose metabolism is a shared pathway of DEGs and DAPs and other pathways involved in carbon metabolism, which provide basic substances for protein synthesis. Starch and sucrose metabolism can generate a vital function in the process of protein accumulation in kenaf leaves. In addition to the 21 amino acid synthesis pathways, other metabolic pathways are indirectly involved in amino acid synthesis (Iqbal et al., 2020). These processes include gluconeogenesis, the pentose phosphate pathway, the citrate cycle, carbon fixation in photosynthetic organisms as well as pyruvate metabolism. These metabolic pathways are closely related to the synthesis of amino acids and provide raw materials for protein synthesis. Furthermore, the findings showed that the proteomic data and transcriptome data were complementary and that the proteome could verify the transcriptome data; besides, genes exert the same role at the transcriptome and proteome levels (Muers, 2011). Besides, the functional classification of the transcriptome and proteome is conducive to improving the comprehension of protein synthesis physiology and molecular biology.

Protein synthesis is a complex physiological, biochemical, and molecular biological process. It is not directly determined by a single gene but is jointly controlled by a lot of genes with different functions (Skadsen and Cherry, 1983). Fructose-1,6-bisphosphate aldolase (ALD) is a critical enzyme for the conversion from 3C to 6C compounds after CO₂ fixation. The upregulated ALD catalyzes the alcohol aldehyde condensation of dihydroxyacetone phosphate (DHAP) and glyceraldehyde 3-phosphate (GAP) to produce more fructose-1,6-diphosphate (FDP). Fructose-1,6-bisphosphatase (FBP) catalyzes fructose-1,6-diphosphate (FDP) and water with the purpose of generating fructose-6-phosphate and inorganic phosphorus, providing raw material for protein synthesis and has been verified in other crops (Li et al., 2022). Upregulated sucrose phosphate synthase (SPS) catalyzes UDP glucose and fructose 6-phosphate in order to yield more sucrose 6-phosphate (S6P), and sucrose 6-phosphate is hydrolyzed to produce sucrose. The main enzymes of sucrose catabolism are sucrose invertase (INV) and sucrose synthase (SUS), which degrade sucrose to glucose and fructose (Duan et al., 2021). The sucrose synthesized by photosynthesis is converted into reducing sugar by INV to supply the growth of young tissues and provide raw materials for protein synthesis (Deng et al., 2019). Nitrate transporters (NRTs) are a key factor in plant perception, absorption, and transport of nitrate (Zhang et al., 2021). The first step of nitrate degradation occurs in the cytoplasm. Nitrate is lowered to nitrite by nitrate reductase (NR). Nitrite enters chloroplasts or plastids and is degraded to ammonium by nitrite reductase (NirA) to provide nitrogen for protein synthesis (Eady et al., 2016). Aminoacylation of tRNA is the first step in protein synthesis. In this process, aminoacyl tRNA synthetases (aaRSs) connect specific amino acids to homologous tRNA to complete protein synthesis (Smirnova et al., 2012). Some examples include Val tRNA synthetase and threonyl tRNA synthetases (Wang et al., 2016; Zhang et al., 2017). Amino acids play a vital function in plant central metabolism. Amino acids can also play the role of intermediates of final metabolites in some metabolic pathways and be engaged in regulating various metabolic pathways and other physiological and biochemical pathways, thereby influencing protein synthesis. The Aspartic acid (Asp) participates in protein synthesis and provides raw materials for protein synthesis (Yang et al., 2020).

In the current work, 14 DEGs and 15 DAPs which were engaged in protein synthesis presented strong associations with RNA-seq data and protein expression levels, and these correlations are consistent with the results that the metabolic pathway that is directly or indirectly involved in amino acid synthesis may play a key role in protein synthesis (Iqbal et al., 2020). Obviously, the expression levels of four proteins (FBP, NirA, PARS and GLDC) related to carbon fixation in photosynthetic organisms, nitrogen metabolism, aminoacyl-tRNA biosynthesis and glycine, serine and threonine metabolic pathways demonstrated higher protein and gene expression in kenaf leaves with high protein content, indicating FBP, NirA, PARS and GLDC may exert an essential function in protein synthesis of kenaf leaves.

The protein and gene expression levels of ALD, SPS, INV, NR, and NRT in kenaf leaves with a high protein content were higher than those of kenaf leaves with a low protein content. The expression of metabolic proteins related to carbon transport and nitrogen transport increased significantly, indicating that the physiological synthesis of raw amino acids and the transcription and translation of storage protein genes constitute a network in kenaf leaves with high protein content. In future work, transgenic plants are required to overexpress candidate genes to verify the functions of these genes.