Integrated Metabolome–Transcriptome Profiling Identifies JrMYB8 as a Repressor of Polyphenol Biosynthesis in Walnut (Juglans regia L.)

TL;DR

This study identifies a gene, JrMYB8, that controls polyphenol levels in walnuts, which could help improve walnut flavor and quality through breeding.

Contribution

The study identifies JrMYB8 as a repressor of polyphenol biosynthesis in walnut using integrated metabolome–transcriptome profiling.

Findings

JrMYB8 overexpression reduces total phenol and flavonoid content in walnut tissues.

JrMYB8 directly represses the expression of JrC4H, a key gene in polyphenol biosynthesis.

Husk and pellicle tissues show distinct polyphenol composition profiles.

Abstract

Walnut is valued for being rich in nutrients and polyphenols, which are key bioactive metabolites; however, a comprehensive and dynamic assessment of metabolites in the husk and pellicle is still lacking. In this study, multi-omics approaches combining untargeted metabolomics and transcriptome analysis were conducted to systematically characterize the differential metabolite profile and regulatory networks in walnut husk and pellicle. Metabolomic profiling revealed a clear divergence in polyphenol compositions between the husk and the pellicle; the husk was predominantly enriched in nine phenolic acid compounds, whereas the pellicle accumulated eleven flavonoid compounds. Through co-expression network analysis, a transcription factor, JrMYB8, was identified and shown to act as a specific inhibitor and regulator of polyphenol biosynthesis. Functional characterization demonstrated that JrMYB8 overexpression significantly reduced the accumulation of the total phenol content (TPC) and the total flavonoid content (TFC) by directly repressing the expression of JrC4H. These findings not only provide a molecular target for manipulating polyphenol content in walnut tissues but also offer a target for improving flavor in walnut breeding.

Figures

Full text

1. Introduction

Polyphenols is the general term for secondary metabolites of polyhydroxyphenols, with phenol as their basic skeleton. Polyphenols play a decisive role in plant flavor and nutrient value [1] and have certain impacts on plant growth [2,3]. Plant polyphenols possess antioxidant, anti-aging, and anti-cancer properties. Walnut husks can be used as raw materials in the preparation of hair dyes, among other applications [4].

The polyphenol biosynthetic pathway and its transcriptional regulators have been extensively studied in plants [5,6]. Many types of transcription factors participate in the synthesis of plant polyphenols. Among them, the MYB-bHLH-WD40 (MBW) complex specifically functions as a key transcription factor of flavonoid biosynthesis, particularly anthocyanins and proanthocyanidins [7], while other transcription factor families (e.g., WRKY, R2R3-MYB repressors, and NAC) play predominant roles in regulating phenolic acid and lignin branches of the phenylpropanoid pathway. Some R2R3-MYB genes in subgroup 4 contain the conserved motifs C1 and C2; motif C2 includes the EAR repression sequence (LxLxL), which has been shown to inhibit PA biosynthesis [8,9,10,11]. However, transcription factors (TFs) in walnut have not been adequately studied, and the genes that control polyphenol accumulation and the underlying regulatory network remain unclear.

Juglans regia L., a deciduous tree of the genus Juglans in the family Juglandaceae, is mainly distributed in the northern hemisphere. Walnut is an economically important nut that is widely cultivated for its high contents of unsaturated fatty acids, melatonin, vitamins, and other bioactive compounds [12,13]. In addition, it is used for medicinal purposes because of its abundant natural products [14]. The husk and pellicle contain large amounts of polyphenols [15,16], but they are usually discarded as waste during production and processing. Recently, the integration of metabolomics and transcriptomics has provided an effective approach to gaining system-wide insights into underlying biological processes and to deciphering gene functions associated with metabolism and physiology [17]. To date, such a combined dynamic analysis is still lacking for walnut husk and pellicle.

In this study, non-targeted metabolomic and transcriptomic profiling was performed, and an R2R3-MYB factor, JrMYB8, was identified. JrMYB8 acts as a repressor that regulates polyphenol biosynthesis; moreover, it directly inhibits the expression of JrC4H1 and JrC4H2, leading to reduced polyphenol content in walnut. Our findings provide high-value information for elucidating the molecular mechanism of polyphenol formation in walnut and may contribute to improving walnut breeding and increasing the utilization of waste from walnut production.

2. Results

2.1. Metabolic Profiling of Walnut Husk and Pellicle

Metabolites were determined using an untargeted metabolite method in the husk and pellicle. In total, 2621 and 1912 metabolites were detected and quantified throughout the growth and development period in the husk and pellicle, respectively. Principal component analysis (PCA) showed that principal components 1 and 2 explained 58.1% of the variation in metabolites among samples and revealed obvious clustering of metabolic profiles by tissue and developmental stage (Figure 1A). The metabolic profiles indicated distinct developmental differentiation between the husk and pellicle.

Based on the PCA, the differential metabolite analysis showed that the number of decreased metabolites in the husk and pellicle was more than that of increased metabolites in the growth and development stage, and the pellicle showed more up- and downregulated metabolites than the husk (Figure 1B). At the fruit-enlargement stage (FES), hard-stone stage (HSS), kernel-filling stage (KFS), and mature stage (MS), 92, 144, 155, and 170 metabolites were higher in the husk than in the pellicle, whereas 163, 114, 114, and 169 metabolites were lower, respectively (Table S1). KEGG enrichment of these stage-specific differential metabolites highlighted flavonoid and phenylpropanoid biosynthesis as the most significantly altered pathways (Figure 1D). A Venn diagram was used to visualize the metabolites that changed in common between the husk and pellicle; 155 metabolites were found (Figure 1C), and 87 differential metabolites were identified after further analysis. The heat map analysis revealed that 28 components were predominantly enriched in the husk, with the other 59 in the pellicle (Figure 1E; Table S2). Among them, 9 and 11 metabolites were highly corrected (with a fold change >2 or <1/2 and r > 0.8) and predominant in the husk and pellicle, respectively (Table S3).

2.2. RNA-Seq Profiling and Bioinformatic Analysis

RNA-Seq data were generated for five developmental stages. The filtered reads obtained from transcriptome sequencing were aligned to the walnut 2.0 assembly genome, and more than 94.1% of the reads were mapped to the genome. In total, 31,633 genes were expressed in the husk and pellicle. Correlation analysis and PCA were performed on all samples; as expected, husk transcriptomes clustered together and showed substantial differences across developmental stages, indicating obvious tissue-specific differences. The tissues showed higher correlation than the developmental stages, suggesting more similar transcriptomes and functions (Figure 2A).

To investigate the transcriptional differences that characterize different stages of fruit development in the husk and pellicle, specifically expressed genes were identified. Transcriptome analysis showed that 4271 and 3123 transcripts were expressed exclusively in the husk and pellicle, respectively. They were clustered into three branches according to their different tissues of expression and developmental stages, as shown in the heat map. The specific transcripts of branches 1 (2151), 2 (939), and 3 (2333) were analyzed through KEGG, and it was found that phenylpropanoid and flavonoid biosynthesis pathways existed in each branch (Figure 2B–D; Table S4). These transcripts were highly expressed at the early stage of husk and pellicle development, indicating that both husk and pellicle had an active polyphenol synthesis at that time.

2.3. Generation of Polyphenolic Metabolic Regulatory Networks

To further understand the phenylpropanoid and flavonoid biosynthetic pathways, we selected all expressed genes associated with these pathways, 247 in the husk and 197 in the pellicle. Low-expression (FPKM < 0.5) and redundant genes were removed, and genes with a fold change > 2 or <1/2 were screened; a total of 116 genes were related to the key metabolic pathways. Pearson correlation coefficient analysis (|PCC| ≥ 0.85) was performed between these genes and the main metabolites accumulated in the husk and pellicle, and gene–metabolite correlation networks were constructed in the husk and pellicle (Figure S1). A total of 22 hub genes were screened, including three UDP-glycosyltransferase (UDP), leucoanthocyanidin dioxygenase (LAR), three cinnamyl alcohol dehydrogenase (CAD), vinorine synthase (VS), two 4-coumarate-CoA ligase (4CL), spermidine hydroxycinnamoyl transferase (SHT), five phenylalanine ammonia-lyase (PAL), chalcone-flavonone isomerase (CHI), two trans-cinnamate 4-monooxygenase (C4H), caffeic acid 3-O-methyltransferase (COMT), caffeoyl-CoA O-methyltransferase (CCoAOMT), and anthocyanidin 3-O-glucosyltransferase (UA3GT) (Table S5). To link the metabolites to the underlying genes, we built a co-expression network. We constructed a proposed phenylpropanoid and flavonoid biosynthesis pathway for walnut and mapped the screened genes and metabolites onto it (Figure 2E). The majority of these 22 genes were highly expressed at the early stage, indicating that polyphenol synthesis is more active during early walnut fruit development.

2.4. Identification of Transcription Factors Regulating Polyphenolic Metabolism

TFs may also participate in the regulation of phenylpropanoid and flavonoid biosynthesis. Based on previous studies of transcriptional regulatory mechanisms in phenylpropanoid and flavonoid biosynthesis, we focused on the MYB, bHLH, and WRKY classes. A total of 185 bHLHs, 325 MYBs, and 105 WRKYs were identified, and 225 of these transcription factors exhibited diverse transcriptional levels (FPKM > 0.5 and FC > 2 or <1/2). The main components in the husk and pellicle were used for correlation analysis with TFs to screen highly related transcription factors (|r| > 0.9). A total of 89 overlapping candidate transcription factors were detected in the husk and pellicle. Furthermore, a co-expression analysis of these transcription factors, the screened major metabolites, and structural genes was conducted using the gene–metabolite correlation-based network (Figure 2F,G). Finally, 33 hub candidate transcription factors were selected, comprising 15 MYB, 10 bHLH, and 8 WRKY (Table S6).

2.5. Phylogenetic Analysis and Amino Acid Sequence Analysis of MYB Transcription Factors

Among the screened TFs, although bHLH and WRKY hub TFs also scored highly in the co-expression analyses, MYB factors were initially selected based on existing research and their important roles in secondary metabolism; subsequent work will explore the remaining bHLH and WRKY candidates [18,19,20]. A phylogenetic analysis was performed to better analyze the function of the JrMYBs (Figure 3C). Among them, JrMYB8 was grouped together with the phenylpropanoid biosynthesis repressors AtMYBL2 [21], AtMYB7 [22], and AtMYB3 [20]; JrMYB20 was grouped with the phenylpropanoid regulators NtMYBGR1 [23] and AtMYB43 [24]; JrMYB59 was grouped with the flavonol regulator AtMYB48 [25]; and JrMYB108 was grouped with the anthocyanin regulator AtMYB112 [26]. Phylogeny and sequence analysis suggested that JrMYB8, JrMYB20, JrMYB59, and JrMYB108 are candidate phenylpropanoid and flavonoid regulators, and JrMYB8 might repress polyphenol biosynthesis; this hypothesis will be tested in subsequent experiments.

Sequence alignment showed that the four JrMYBs contain the R2 and R3 repeats, consistent with the canonical R2R3-MYB criteria (Figure 3D). The conserved protein C2 motif contains the EAR repressor domain (LxLxL) [9,27]. JrMYB8 carries an EAR-like motif in the same region, and this motif has been shown to function as a repressor in A. thaliana [21]. These results support the idea that JrMYB8 might act as a repressor of phenylpropanoid and flavonoid biosynthesis in walnut.

2.6. Analysis of Expression Pattern of MYB Transcription Factor

In qRT-PCR analysis, the expression of JrMYB8 was reduced from the fruit-bearing stage (FBS) to KFS and increased at MS in the husk, while it was the opposite in the pellicle; the expression trend of JrMYB20 was the opposite of that of JrMYB8; JrMYB59 was upregulated from FBS to KFS and then downregulated at MS in the husk but increased in the pellicle; and JrMYB108 was downregulated at FBS and upregulated from FES to MS, while it was increased in the pellicle (Figure 3A).

In order to further screen the transcription factors, the expression pattern of genes in the phenylpropanoid and flavonoid pathway was analyzed. JrC4H showed an upregulation trend with growth and development in the husk, while its expression decreased in the pellicle (Figure 3A), a pattern that was the opposite of that of JrMYB8 and consistent with the existing research results in other species [28,29]. That is to say, JrMYB8 was correlated with polyphenol biosynthesis in walnut.

The Use Cell PLo2.0 was used to analyze the amino acid sequence of JrMYB8, and the results showed a nuclear localization signal (NLS) within its subdomain. We then constructed the 35S::JrMYB8-GFP fusion vector to verify this prediction. Compared with 35S::GFP, JrMYB8-GFP was specifically localized to the nucleus (Figure 3B).

2.7. Persimmon Leaves Showed Reduced Polyphenol Contents After Overexpression of JrMYB8

Walnut is a perennial fruit tree, but the generation of transgenic plants by stable transformation technologies is difficult because of its high polyphenol content. The transformation of persimmon leaves by Agrobacterium tumefaciens is a fast and effective method for gene function analysis [27,30]. After the overexpression vector of JrMYB8 was transiently transformed into persimmon leaves, JrMYB8’s expression was significantly higher than in the control, whereas the contents of polyphenols and flavonoids decreased (Figure 4A,B). Analysis of phenylpropanoid and flavonoid pathway genes revealed that the expression of 4CL, anthocyanidin reductase (ANR), C4H, CHI, chalcone synthase (CHS), PAL, CoMT, hydroxycinnamoyl-CoA: shikimate hydroxycinnamoyl transferase (HCT), and dihydroflavonol 4-reductase (DFR) was significantly reduced, falling to 0.3–0.7-fold of the control level in persimmon leaves. The above results indicated that JrMYB8 acted as a negative regulatory transcription factor and reduced the accumulation of polyphenols by inhibiting the expression of polyphenol-related genes (Figure 4B). Consistently, a recent integrated omics survey in red walnut leaves also predicted JrMYB6a and JrMYB308e as the repressors of anthocyanin biosynthesis, supporting the existence of conserved MYB repressors in the genus Juglans [31].

2.8. Overexpression of JrMYB8 in Transgenic Tobacco Leads to Reduced Polyphenol Biosynthesis

JrMYB8 was introduced into tobacco to further identify JrMYB8’s potential function in regulating polyphenol biosynthesis (Figure 4C,E). In the JrMYB8 overexpression lines, no obvious phenotypic changes were observed in plant morphology; however, the polyphenol and flavonoid content was significantly reduced (Figure 4D), indicating that JrMYB8 could inhibit the biosynthesis of polyphenols in tobacco leaves. In addition, the expression of polyphenol biosynthesis genes showed a decrease, especially C4H, 4CL, CHI, COMT, and HCT (Figure 4E). These downregulated genes may be related to a function that has been reported for other transcription factors that co-cluster with JrMYB8, which can directly repress C4H activity. We speculate that JrMYB8 may have a similar function and lead to a reduction in other structural genes by suppressing JrC4H expression in walnut.

2.9. JrMYB8 Suppresses JrC4H Promoter Activity

According to the joint analysis, JrC4H1 and JrC4H2 were highly correlated with key components. After isolating the promoters of JrC4H1 and JrC4H2, we searched the plant cis-acting regulatory DNA elements (PLACE) and identified MYB-binding motifs, including H-Box, MYB, MYB-like, and Myb sequences, in the promoter regions (Figure 5A).

The β-glucuronidase (GUS) staining assay showed that the promoters of JrC4H1 and JrC4H2 had transcriptional activity (Figure 5B). In the yeast one-hybrid (Y1H) assay, the JrMYB8 bound to the promoters of JrC4H1 and JrC4H2 (Figure 5C). In the dual-luciferase assay, the co-expression of JrMYB8 markedly lowered the promoter activity of both JrC4H1 and JrC4H2 (Figure 5D,E). This indicates that JrMYB8 directly binds to the promoters of JrC4H1 and JrC4H2, which are related to polyphenol biosynthesis, and inhibits their expression.

3. Discussion

Polyphenolic biosynthesis is an intricate process orchestrated by the gene and metabolite network. Researchers have profiled walnut compounds using liquid chromatography–mass spectrometry (LC–MS) [32,33,34], but the metabolome—especially the husk and pellicle—remains uncharted [35,36,37]. In this study, metabolomic analysis was carried out at different developmental stages of walnut husk and pellicle. By comparing differential metabolites, we found that phenylpropanoid and flavonoid biosynthesis were the main enrichment pathways, and phenylpropanoid compounds were predominant in the husk, whereas flavonoids were more abundant in the pellicle. The results indicate that metabolites may play a key role in characterizing the formation of different tissues in the walnut fruit. Similar patterns observed in the inner and outer seed coats of pomegranate also support this conclusion [38]. Polyphenols in the husk peak during FBS and then decline, whereas in the pellicle, they rise steadily as the kernel fills. This accumulation pattern mirrors tissue function: high polyphenol levels in the husk provide allelopathic and antioxidant defenses against insects and UV, safeguarding early growth. After HSS, the seed coat becomes the metabolic hub; proanthocyanidins and condensed tannins accumulate, inhibit lipoxygenase, and chemically protect the accumulating endosperm lipids. Therefore, screening the differential metabolites accumulated during husk and pellicle formation provides insights into the metabolic factors that influence fruit differentiation. This study offers a valuable reference for research on fruit developmental biology and walnut genetic improvement.

A single metabolite change cannot explain the molecular–phenotype transition. Multi-omics links metabolites and genes to the pathway [39]. In walnut, JrPAL, JrC4H, Jr4CL, JrCHI, JrLAR, JrCAD, JrCOMT, and JrUDP are involved in phenylpropanoid and flavonoid biosynthesis, possibly influencing husk/pellicle formation. R2R3-MYBs are the key transcription factors that control flavonoid and anthocyanin biosynthesis in plants [40,41,42]. However, few MYBs have been reported to regulate walnut polyphenol biosynthesis [43]. Here, we identified JrMYB8, an R2R3-MYB transcription factor regulating walnut polyphenol metabolism. Stable transformation and expression of JrMYB8 in tobacco suppressed polyphenol biosynthesis (Figure 4D,E). Like other inhibitory MYBs, JrMYB8 contains a glycine (Gly) residue at position 50 of the R2 domain and the Asp-Asn-Glu-Ile motif within the R3 domain [44].

Phylogenetic analysis grouped JrMYB8 with AtMYB3, AtMYB4, and PtMYB221—members of the MYB4 subfamily known to suppress phenylpropanoid/lignin accumulation [20,45,46]. The EAR motif serves as the primary inhibitory domain [47]. In Arabidopsis, chimeric AtMYB23 (containing the EAR motif) with EIN3 or PAP1 significantly inhibited target gene expression [48]. The C-terminal 35 amino acids of NtERF3 (191–225) are essential for transcriptional repression in tobacco [49]. The amphiphilic LxLxL motif in EAR domains mediates transcriptional repression [9,27]; therefore, JrMYB8’s C2 LxLxL motif functions similarly. AtMYBL2’s TLLLFR motif (EAR-like but non-amphiphilic) also acts as an inhibitor when fused to heterologous DNA-binding domains [48]. JrMYB8 contains a similar TLLLF motif, though its distinction from EAR requires further study. These findings demonstrate JrMYB8’s negative regulatory role in walnut polyphenol biosynthesis. In dicotyledons, R2R3-MYB transcription factors often repress flavonoid biosynthesis. In Arabidopsis, AtMYB7 deletion upregulates C4H, 4CL1, flavonoid 3’-hydroxylase (F3’H), DFR, and UDP-glycosyltransferase (UGT) [22]. FaMYB1 overexpression alters anthocyanin and flavonol pathway genes [10]. In grapevine, VvMYB4-like and VvMYB4A suppress anthocyanin accumulation by repressing DFR, UDP-glucose flavonoid 3-O-glucosyltransferase (UFGT), and anthocyanidin synthase (ANS) [50,51]. AtMYB4 overexpression downregulates upstream flavonoid genes, such as C4H, CHS, and 4CL. UV-B exposure reduces AtMYB4 transcripts, thereby elevating AtC4H expression and hydroxycinnamate production [28]. Similarly, PhMYB4 inhibits PhC4H in petunia, altering phenylpropanoid flux [29]. In this study, JrMYB8 overexpression in tobacco downregulated phenylpropanoid and flavonoid genes (4CL, CHS, CHI, ANR, C4H, CoMT, and HCT) (Figure 4C–E). JrMYB8 directly bound and repressed JrC4H1 and JrC4H2 promoters (Figure 5), indicating that JrMYB8 plays an inhibitory role in walnut polyphenol biosynthesis. However, the expression pattern of JrMYB8 does not consistently align with polyphenol accumulation across different developmental stages [16]. This discrepancy likely arises because the transcription factor itself is subject to multiple layers of regulation; its transcript abundance, therefore, does not directly reflect its final repressive strength. In tea (Camellia sinensis), CsMYB75- and CsMYB86-directed MBW complexes preferentially activate anthocyanin and catechin biosynthesis, respectively, whereas the homologous proteins CsMYBL2a and CsMYBL2b suppress light- and temperature-induced anthocyanin and catechin production in both Arabidopsis and tea; thus, different MYB factors exert distinct regulatory effects on flavonoid pathways [52]. In walnut, polyphenol accumulation is governed by a balance between positive activators and negative repressors; within this multi-factor framework, the inhibitory effect of JrMYB8 can be masked. The initiation of polyphenol biosynthesis in plants is controlled by diverse developmental and environmental cues. In poplar, the lignin-biosynthetic transcription factor LTF1 binds the 4CL promoter and represses lignin formation when unphosphorylated; upon phosphorylation by PdMPK6, LTF1 is degraded via the proteasome pathway, thereby activating lignification [53]. Similarly, the function of JrMYB8 in walnut is likely modulated by upstream kinase activities, thereby influencing polyphenol accumulation.

In this study, Y1H assays demonstrated that JrMYB8 can directly recognize the JrC4H promoter (Figure 5C), whereas the empty-vector control produced no signal. Numerous previous studies have shown that the three families form an MBW ternary complex that regulates anthocyanin and proanthocyanidin biosynthesis in plants. However, whether JrMYB8 acts indirectly by interfering with or forming part of the MBW complex was not experimentally tested here. Some reports have demonstrated that AtMYB4 indirectly restricts MBW-mediated activation of anthocyanin synthesis by directly binding MYB elements in target-gene promoters and repressing transcription, whereas R2R3-MYB repressors such as CsMYBL2b, which carry an EAR motif, inhibit anthocyanin/catechin synthesis by competing with activating MYBs for bHLH cofactors, thereby blocking the formation of the MBW complex [52,54]. Sequence analysis revealed that JrMYB8 possesses only an EAR repression motif (LxLxL) downstream of its R2R3 domain and lacks the conserved signature [DE]Lx_2_[RK]x_3_Lx_6_Lx_3_R required for stable interaction with R/B-like bHLH proteins [55]. Taken together with the Y1H results, we propose that JrMYB8 exerts negative regulation by directly binding the JrC4H promoter, possibly limiting MBW-mediated activation of polyphenol synthesis indirectly; this hypothesis awaits confirmation by interaction and in vitro competition assays. In walnut, the low expression of JrMYB8 allows phenolic acid synthesis in the husk in early development, whereas its later upregulation during HSS restrains excessive tannin buildup in the seed coat, ensuring nutrient flow to the endosperm. The husk is rich in feruloylquinic, neochlorogenic, and p-coumaroylquinic acids and syringic acid hexosides—bioactives linked to anti-atherogenic, hypolipidemic, and anti-inflammatory effects that underpin walnut’s “functional food” status. The husk is currently discarded, creating waste and pollution; extracting these phenolics for nutraceuticals or natural antioxidants could add commercial value across the supply chain. Conversely, the pellicle accumulates ellagic and syringic acids and juglone, which scavenge radicals during storage, delaying lipid oxidation and thus determining shelf life and flavor stability. The polyphenol network is therefore an adaptive system coupling development, stress responses, and nutritional quality; deciphering this biology will guide breeding for cultivars high in antioxidants and enable whole-fruit valorization.

4. Materials and Methods

4.1. Plant Materials

Fruits of “Wen 185” were sampled from the orchard in Tarim University (44°55′ N, 81°28′ E) during five developmental stages, namely, at the FBS, FES, HSS, KFS, and MS. A photograph of the five developmental stages collected in the orchard is provided in Supplementary Figure S2; the exact tissues dissected for multi-omics (husk at FBS; husk and pellicle at FES, HSS, KFS, and MS) are indicated by red lines in the figure. At each stage of development, fifteen fruits of three different trees were mixed into a biological repetition, with three repetitions per stage. For each stage, the husk and pellicle were immediately separated from the walnut on ice, flash-frozen in liquid N_2_ within 5 min, and stored at –80 °C until use. Tobacco (Nicotiana tabacum L.) was used in transient expression and stable transformation. Seeds were surface-sterilized with 75% ethanol (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) and 2% NaClO (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China), washed 3–5 times, and cultured on MS medium (Lanboled, Wuhan, China). Healthy and sterile tobacco leaves were selected and transformed via the leaf disc method; three independent transgenic lines were retained for subsequent experiments.

4.2. Metabolite Extraction and Untargeted Determination of Key Metabolites

The metabolites were extracted with 50% methanol (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) as described by Sheng et al. (2021) [16]. After filtration, the data was imported into Simca software (http://umetrics.com/) for unsupervised PCA and orthogonal partial least squares discriminant analysis (OPLS-D). Hierarchical clustering was performed with TBtools (v2.0x) (https://www.tbtools.com/). Metabolites meeting |fold change| ≥ 2, VIP ≥ 1, were designated as significant differences and visualized in Venn diagrams.

4.3. RNA Sequencing and Data Analysis

Total RNA was isolated from the husk and pellicle. RNA-Seq was performed by Biomarker Technologies Co., Ltd. (Beijing, China). Libraries were prepared with the NEBNext^®^ Ultra^TM^ RNA Library Prep Kit for Illumina^®^ (New England Biolabs, USA (Ipswich, MA, USA)) and sequenced on an Illumina HiSeq X Ten platform (Illumina, Inc., San Diego, CA, USA). Clean reads were aligned to the Juglans regia genome (https://www.ncbi.nlm.nih.gov/genome/17683 (accessed on 7 January 2021)) with HISAT2 (https://ccb.jhu.edu/software/hisat2/index.shtml (accessed on 7 January 2021)). Differentially expressed genes were identified with the DESeq R package (1.10.1) (|FC| ≥ 2, FDR < 0.01). A metabolite–gene co-expression network (Pearson |r| > 0.9) was built in Cytoscape 3.8.2.

4.4. Quantitative Real-Time PCR (qRT-PCR) Analysis

Total RNA was isolated from the husk and pellicle with the Spin Column Plant Total RNA Purification Kit (Sangon Biotech, Shanghai, China). RNA quality and integrity were assessed using a NanoDrop 2000 spectrophotometer (Thermo Scientific, https://www.thermofisher.com) and gel electrophoresis. First-strand cDNA was synthesized using the PrimeScript^TM^ RT Kit with a gDNA Eraser (TaKaRa) according to the manufacturer’s instructions. TB Green Premix Ex Taq^TM^ (TaKaRa) was used to perform qRT-PCR on a real-time PCR instrument (QuantStudio 7 Flex Real-Time PCR system, Applied Biosystems, https://www.thermofisher.com). JrActin (LOC109008570) was used as an internal reference for walnut. The primers for qRT-PCR are listed in Table S7.

4.5. Full-Length Cloning and Bioinformation Analysis of MYB Genes

The full-length ORF of JrMYB8 was amplified with gene-specific primers designed from the walnut genome database, and its molecular weight, isoelectric point (pI), and subcellular localization were predicted. Transcription factors were predicted with PlantTFD. Phylogenetic tree analysis of JrMYB and its homologs was carried out by MEGA 6 [56] with bootstrap values calculated from 1000 replicate analyses. JrMYB and orthologous amino acid sequences from other species were aligned with DNAMAN 6.0 (Lynnon Corporation, San Ramon, CA, USA).

4.6. Dual-Luciferase Assay

The JrMYB8 coding sequence and the promoters of JrC4H1 and JrC4H2 were cloned into pGreenII 62-SK and pGreenII 0800-LUC vectors, respectively. After transformation into A. tumefaciens (GV3101), effector and reporter strains were co-transformed into tobacco leaves. All primers are listed in Table S7. Dual-luciferase activities (LUC and REN) were quantified with the Promega dual-luciferase assay kit on a Tecan Infinite M200 microplate reader.

4.7. Yeast One-Hybrid Assay

The CDS of JrMYB8 was amplified and subcloned into the pGADT7 vector. The promoter fragments of JrC4H1 (1062 bp upstream of the start codon, Chr 5744791-5745853 on the Walnut 2.0 genome assembly) and JrC4H2 (1434 bp upstream of the start codon, Chr 14: 5732593-5734027 on the Walnut 2.0 genome assembly) were PCR-amplified from walnut genomic DNA with gene-specific primers (Supplementary Table S7), verified by sequencing, and cloned into the pAbAi vector (Clontech). Pairs of plasmids were co-introduced into yeast strain Y1H Gold and cultured on SD/-Ura/-Leu medium containing 0–200 ng/mL AbA at 30 °C for 72 h.

4.8. GUS Staining

The promoter of JrC4H was inserted into the DX2181 vector, then transferred into tobacco leaves and cultured in a light incubator for 2–3 days. GUS staining in transgenic plants was performed as described previously [57].

4.9. Stable Overexpression of Transcription Factor Genes in Tobacco

To generate binary vectors for JrMYB8 overexpression, the cDNA sequences of JrMYB8 were cloned into pDONR207 and pK7WG2D using Gateway BP (Invitrogen, Carlsbad, CA, USA, no. 11789-020) and LR (Invitrogen, Carlsbad, CA, USA, no. 11791-020). ClonaseTM II Enzyme Mix was used in the recombination reaction to generate JrMYB8-pK7WG2D binary vectors. All constructs were transformed into A. tumefaciens strain GV3101 cells. Tobacco was stably transformed via the leaf disc method as previously described [58]. Regenerated shoots were selected on MS medium containing 300 mg/L cefotaxime and 50 mg/L kanamycin. PCR was performed to confirm the transgenic tobacco plants. The transgenic plants were moved to soil for their normal growth.

4.10. Subcellular Localization of JrMYB8

The JrMYB8 CDS (minus the stop codon) was cloned into pAN137 to create a 35S-driven JrMYB8-GFP fusion, which was then introduced into A. tumefaciens strain GV3101. A. tumefaciens containing the JrMYB8 or GFP construct was resuspended in infiltration buffer and infiltrated into tobacco leaves. After infiltration, plants were placed at 24 °C for 48 h before GFP observation. The nucleus and GFP fluorescence were observed under a confocal laser scanning microscope (Leica TCS SP8, https://www.leica-microsystems.com).

4.11. Statistical Analysis

Pearson correlation coefficients were calculated with IBM SPSS version 28.0 (IBM Corp., Armonk, NY, USA) statistical software (https://www.ibm.com/analytics/spss-statistics-software (accessed on 22 February 2021)), and two-tailed p-values < 0.001 were considered significant. Three biological replicates per condition were employed in differential expression analyses. Read counts were analyzed with DESeq2 (v1.34.0) under the Wald test, and genes with |log2 fold-change| ≥ 1 and FDR < 0.05 were deemed differentially expressed. All analyses were expressed as the mean ± standard error (SE), with significance levels denoted as p < 0.05 (*) or p < 0.01 (**). For comparisons involving more than two sets of data, one-way analysis of variance (ANOVA) was performed, followed by Fisher’s least significant difference post hoc test, and a p-value < 0.05 was considered to be significant by different lowercase letters above the columns.

5. Conclusions

In this study, we investigated the regulatory network connected to the phenylpropanoid and flavonoid pathways using a co-expression analysis of the metabolome and transcriptome in five different stages of walnut husk and pellicle. Correlation analyses identified the hub TFs, among which JrMYB8 was functionally characterized as a repressor that reduces polyphenol levels by repressing the expression of JrC4H. Moreover, as illustrated by our studies on polyphenols, a correlation network of differential metabolites and genes was obtained. The exploration and utilization of this joint analysis could contribute to understanding the biochemical regulation mechanisms of metabolic pathways in walnut.