MicroRNA Identification and Integrated Network Analyses for Age-Dependent Flavonoid Biosynthesis in Ginkgo biloba

Abstract

Ginkgo biloba leaves contain abundant flavonoids, and flavonoid accumulation is affected by age. MicroRNAs (miRNAs) play an important role in the plant aging pathway. However, the miRNAs involved in flavonoid biosynthesis related to age in G. biloba have rarely been studied. In this study, we compared 1-, 4-, and 7-year-old ginkgo seedings and found a significant decrease in the content of quercetin, kaempferol, and total flavonol aglycones with age. We then profiled miRNAs in G. biloba through high-throughput sequencing on leaf samples of 1-, 4-, and 7-year-old ginkgo. GO and KEGG analyses suggest that photosynthesis and hormones may influence the flavonoid content. In particular, we identified 29 miRNAs related to the aging pathway according to their miRNA expression patterns. Correlation analysis of age-related miRNAs and major flavonoid compounds screened 17 vital miRNAs, including miRN79, miR535a, miR166a, miR171a, and miR396. Interactive miRNA-transcription factor network analysis suggested that the pivotal miRN79-DELLA and miR535a-SPL modules may be involved in flavonoid biosynthesis and aging pathways through post-transcriptional regulation. Our findings provide insights into the age-dependent regulatory roles of miRNAs in flavonoid biosynthesis.

1. Introduction

Flavonoids are important secondary metabolites, and they can be subdivided into diverse families, including flavonols, flavones, isoflavones, anthocyanidins, flavanones, flavanols, and chalcones, which are ubiquitous in a variety of plant organs and tissues [1]. To date, several key structural genes encoding multiple enzymes have been found to be involved in flavonoid biosynthesis. Early biosynthetic genes (EBGs), including chalcone synthase (CHS), chalcone isomerase (CHI), flavanone 3-hydroxylase (F3H), and flavonoid 3′-hydroxylase (F3′H), catalyze the synthesis of dihydroflavonols, which serve as common precursors for the production of downstream flavonoids. Subsequently, the precursors are enzymatically converted to a different class of flavonoids by the action of late biosynthetic genes. For instance, dihydroquercetin (DHQ) is converted by dihydroflavonol 4-reductase (DFR) to leucocyanidin [2,3]. Accumulating evidence suggests that the transcriptional levels of these structural genes are predominantly regulated by several transcription factors (TFs), including the WRKY, MYB, bHLH, WD40, and bZIP proteins [4,5]. TFs act independently or in combination with other TFs from the same or different families to regulate target gene expression [6]. Among these TFs, MYBs are the most comprehensively studied and are considered the most specific and prominent regulators of flavonoid biosynthesis [7].

There have been numerous studies conducted on flavonoid biosynthesis; most of these studies have focused on the response of flavonoid biosynthesis to external stress factors [8]. There have been relatively few studies carried out on the developmental regulation of flavonoid accumulation. However, plant age is also an important factor affecting the synthesis and accumulation of flavonoids. For example, the flavonoid content in young leaves of Cistus ladanifer was significantly higher than in old leaves [9]. Similar results were observed in birch leaves (Betula pendula, B. pubescens ssp. pubescens and B. pubescens ssp. czerepanovii) [10]. However, the underlying molecular mechanisms responsible for age-regulating flavonoid accumulation remain elusive.

Previous studies have shown that microRNAs (miRNAs) participate in various plant growth, development, and metabolic processes [11,12]. MiRNAs play key roles in the plant aging pathway, the core of which is made up of modules miR156–miR172 [13,14]. The abundance of miR156 in plants decreases with age, resulting in a gradual increase in miR156-targeted SQUAMOSA PROMOTER BINDING PROTEIN-LIKE genes (SPLs) [13,15]. SPL9 activates miR172, resulting in an expression pattern opposite to that of miR156 [14]. In addition, miRNAs also play an important role in regulating flavonoid synthesis. In Arabidopsis thaliana, the miR156-SPL module has been shown to inhibit anthocyanin production by influencing the MBW (MYB-BHLH-WD40) complex [8]. In Oryza sativa, the miR396-OsGRF module regulates flavonoid biosynthesis by directly regulating OsF3H [16]. In potato plants, miR858 directly represses MYB12 to negatively regulate flavonol biosynthesis [17]. MiRNA is also involved in flavonoid accumulation in gymnosperms. For example, the overexpression of miR166 in Larix kaempferi embryonic cells promotes the accumulation of compounds and genes related to flavonoid synthesis [18].

Ginkgo biloba L. is an important economic tree species. Ginkgo biloba leaf extract (GBE) is a valuable herbal preparation used to treat cardiovascular diseases, dementia, and cancer [19,20,21]. The main active compounds in GBE are flavonoids, which commonly exist in nature as glycosides, including quercetin, isorhamnetin, and kaempferol [22,23]. However, flavonoid levels in ginkgo leaves decrease as the tree ages, which is why only ginkgo seedlings under 5 years old are typically used for medicinal purposes [24]. In this study, we investigated the miRNA profiles of G. biloba seedlings at three different ages through the construction of sRNA libraries. We then analyzed the relationships between miRNA expression patterns and flavonoid contents in leaves of different ages through a comprehensive metabolome, sRNA, and coexpression network. Our results provide insights into the regulatory roles of miRNAs in flavonoid biosynthesis in response to plant age.

2. Materials and Methods

2.1. Plant Materials

The leaves of 1-, 4-, and 7-year-old ginkgo trees used in this study were the same as the materials in our previous study [24]. Leaf samples were collected, immediately frozen in liquid nitrogen, and stored at −80 °C. Total RNA from these samples was isolated and quantified for sRNA sequencing.

2.2. Extraction and Quantification of Flavonol and Total Flavonoids

Dried ginkgo leaves were pulverized into powder to extract flavonol (quercetin, kaempferol, and isorhamnetin). A total of 0.1 g of the dry leaf powder samples were extracted with 1.5 mL of methyl alcohol. Extractions were performed for 30 min three times in an ultrasonic bath. After that, the combined supernatant was dried using nitrogen gas and redissolved in 1.0 mL of 25% (v/v) HCl-methanol. The solution was then transferred to a 10 mL chemical oxygen demand tube with a Teflon liner, followed by heating at 80 °C for 30 min and cooling at 4 °C for 10 min. The samples were mixed with 1.0 mL of methanol, and then filtered using an organic membrane with a pore diameter of 0.22 µm before performing high-performance liquid chromatography (HPLC) analyses. The total flavonol glycoside content was calculated by multiplying the sum of quercetin, kaempferol, and isorhamnetin by a factor of 2.51. Furthermore, the total flavonoid content was determined using a plant flavonoid kit (Suzhou Comin Biotechnology Co., Suzhou, China).

2.3. sRNA Library Construction and Sequencing

We used three biological replicates for each plant age for sRNA sequencing. Total RNA was extracted from leaves using the RNA Extraction Kit (TaKaRa, Shiga, Japan), and RNA quality and concentration were analyzed using a spectrophotometer (Implen, Munich, Germany) and an RNA Assay Kit (Life Technologies, Carlsbad, CA, USA), respectively. We constructed nine sRNA libraries from 1 µg of qualified total RNA from each sample. The libraries were sequenced and generated on an Illumina platform by Novogene (Beijing, China), following the manufacturer’s protocol.

2.4. Identification of Known and Novel miRNAs

Clean reads were generated by filtering low-quality reads and reads containing poly N, 5′ adapter contaminants, or poly A, T, G, or C, or without the 3′ adapter or insert tag. sRNA tags were mapped to the G. biloba genome without mismatches to analyze their expression and distribution. Then, the mapped sRNA tags were used to identify known miRNAs in G. biloba using the miRBase20.0 database (www.mirbase.org; accessed on 15 January 2020). MiREvo [25] and mirdeep2 [26] were integrated to predict novel miRNAs.

2.5. Differential miRNA Expression Analysis

To further explore functions, we normalized the miRNA levels of leaves from Y1, Y4, and Y7 ginkgo plants and screened differentially expressed miRNAs (DEMs). The expression levels of identified miRNAs (conserved and novel) were normalized by transcripts per million (TPM). DEMs between pairs of samples were identified using the DEGseq package in the R v1.8.3 software (R Core Team, Vienna, Austria).

2.6. miRNA Target Prediction and Functional Enrichment Analysis

MiRNA target genes were predicted by the psRobot_tar function in psRobot [27]. Gene Ontology (GO) enrichment analysis of target gene candidates was conducted using the clusterProfiler R package to evaluate enrichment in biological processes (BP), molecular functions (MF), and cellular components (CC). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses of target gene candidates were conducted through the clusterProfiler package in R and the KEGG database (http://www.genome.jp/kegg/; accessed on 15 January 2020).

2.7. Coexpression Analyses of miRNAs and Flavonoid Compounds

To evaluate correlations between key flavonoid compounds and age-related miRNAs, we constructed a coexpression network. Spearman’s correlation coefficients were calculated using the stats package in R v3.6.3. Clustering correlation results were expressed as a heatmap using the OmicStudio platform (https://www.omicstudio.cn; accessed on 25 April 2023), and Pearson’s correlation coefficients were used to evaluate significance at correlation ≤−0.8 or correlation ≥0.8.

2.8. Statistical Analysis

Statistical data were presented as means ± standard deviation (SD) of three biological replicates and analyzed using one-way analyses of variance (ANOVA). Significant differences were calculated at p-values < 0.05 using GraphPad 7.0 software (GraphPad Software Inc., La Jolla, CA, USA).

3. Results

3.1. Flavonoid Analyses of Ginkgo Leaves of Different Ages

To profile changes in flavonol content at different age stages, we collected samples of 1-, 4-, and 7-year-old G. biloba leaves (referred to as Y1, Y4, and Y7, respectively) and measured the flavonoid content. The results revealed a decrease in total flavonoid content with increasing age (Figure 1A). Additionally, we utilized high-performance liquid chromatography (HPLC) to detect the flavonol contents in G. biloba leaves at three age stages (Figure 1B). The quercetin content significantly declined with age (Figure 1C). Compared to 1-year-old ginkgo leaves, the quercetin content in 4- and 7-year-old ginkgo leaves was 13.2% and 52.6% lower, respectively (Figure 1C). Furthermore, the kaempferol content decreased by 24.4% and 48.7%, respectively, in 4- and 7-year-old ginkgo leaves compared to 1-year-old ginkgo leaves (Figure 1D). The total flavonol aglycone content displayed a similar trend (Figure 1F). These results suggest that the flavonoid content decreases with age in 1-, 4-, and 7-year-old G. biloba leaves.

3.2. sRNA Sequencing and Identification Results for Conserved and Novel miRNAs

Our study has found that the accumulation of flavonoids in ginkgo had an obvious age effect, and the flavonoid content and flavonoid-related genes decreased with the increase in age at 1, 4, and 7 years. To better understand the regulatory role of miRNAs in age-dependent flavonoid synthesis, the miRNA profiles of G. biloba leaves at different ages were investigated. sRNA samples from the leaves of Y1, Y4, and Y7 G. biloba plants were used for sRNA sequencing, and nine sRNA libraries were constructed, which generated a total of 154,154,295 raw reads. After the removal of unqualified reads, we obtained 144,284,274 clean reads for further analysis (Table S1). The size distribution of the clean sRNA reads was summarized, and the most abundant classes consisted of 20–25 nt sRNAs, with 21, 24, and 25 nt sRNAs showing the highest abundance among all samples (Figure 2 and Table S2). To identify conserved and novel miRNAs in G. biloba, the valid unique reads were aligned to the G. biloba genome (http://gigadb.org/dataset/100613; accessed on 15 January 2020). In total, 236 expressed miRNAs, including 209 conserved and 27 novel miRNAs, were identified in the sequenced libraries. The 209 highly conserved miRNAs were grouped into 150 families, among which miR169 was the largest family (10 members), followed by miR171 (9), miR482 (9), and miR396 (7). More than half of all families had only one member. Of the rest, most families consisted of two to five members (Table S3).

3.3. Differential Analysis of Candidate miRNAs According to Plant Age

All miRNA expression levels were calculated through the TPM algorithm. MiRNA expression levels were compared in pairs of samples to identify DEMs according to plant age. To determine the relationships between nine ginkgo leaf samples at different ages, a Pearson correlation matrix was established to assess the degree of similarity. Based on standardized expression levels, high correlations were observed among Y1, Y4, and Y7 samples (Figure S1). Specifically, all of the correlation coefficients of Y1, Y4, and Y7 samples were above 0.859 (Figure S1). Furthermore, totals of 36 (Y4 vs. Y1), 41 (Y7 vs. Y1), and 19 (Y7 vs. Y4) significant DEMs (padj < 0.05 and fold change > 2) were detected (Figure S2). Hierarchical clustering analysis was performed on the standardized expression levels of conserved and novel miRNAs; the results suggested strongly differential expression among plant ages for most miRNAs (Figure S3). Furthermore, expression was highly variable within miRNA families (Table S3). For example, the miR394 and miR4414 families had low expression in the leaves of Y1, Y4, and Y7, while the miR482 family had high expression in all ages.

To explore the miRNAs (conserved and novel) expression profiles, K-means clustering was performed. The miRNAs were divided into 20 subclusters (Figure S4 and Table S4), among which we screened 9 age-related miRNA (ARM) groups. Expression levels increased with plant age among most miRNAs (15/21) in subclusters 1, 5, 9, 10, and 14, such as miRN79, miR398, miR396i, and miR535a. Furthermore, in the remaining five subclusters (3, 7, 8, 13, and 19), the expression levels of most miRNAs (14/23), such as miR11534c, miR11534h, miR164a, miR166a, and miR535a, decreased with plant age (Figure S5). Notably, miR171a and miR171b, members of the miR171 family, showed opposite expression trends in three age groups. Specifically, the expression of miR171a increased with age, while the expression of miR171b decreased with age. Then, we analyzed the correlation of the expression levels of 29 miRNAs with age. As shown in Figure S6, the expression levels of most miRNAs were significantly correlated with age (correlation > 0.8; p < 0.05). These results imply that miRNAs with age-dependent expression play key roles in the G. biloba aging pathway. In addition, different members of the same miRNA family may have different regulatory mechanisms in the G. biloba aging pathway.

Next, we analyzed the miR156 and miR172 expression levels, which are widely recognized as critical regulators of aging pathways in angiosperms [13,14]. Interestingly, we did not find members of the miR172 family in our data. Additionally, the expression patterns of the miR156 family, including miR156a, miR156b, and miR156c, were different (Table S3). Specifically, miR156a was undetectable in the leaves of Y1 and Y4 and was only observed in Y7. In addition, miR156b was highly expressed in leaves of different ages, but its expression pattern was not age-dependent. Notably, the expression of miR156c was age-dependent and increased with age.

3.4. Identification and Analysis of miRNA Target Genes

In total, 17,791 target genes were predicted. Among these, 2278 target genes were targeted by both known and novel miRNAs (Table S5). Based on the annotation results, the regulatory relationship between miRNAs and their targets is complex. Most miRNAs target multiple unigenes, and multiple miRNAs also regulate a particular target gene. Furthermore, the target genes of the same miRNA family are different. For example, miR171d can regulate 195 target genes, but miR171d can only regulate 54 target genes. In particular, among all miRNAs, miRN107 has the largest number of target genes (624), suggesting that it may have a wide range of functions in G. biloba. In addition, among the 17,791 target genes, 13,132 genes were annotated, and more than 25% of the genes (4659) were unknown (Table S6).

We further analyzed GO function enrichment among the targets regulated by miRNAs, and found that most targets were involved in biological processes (BPs) and molecular functions (MFs), whereas fewer targets were involved in cellular components (CCs). Specifically, BPs, including the primary metabolic process (GO:0044238), macromolecule metabolic process (GO:0043170), and cellular macromolecule metabolic process (GO:0044260), were significantly enriched between Y1 and Y4, whereas between Y4 and Y7, the metabolic process (GO:0008152) and cellular macromolecule metabolic process (GO:0044260) were significantly enriched. Furthermore, among MFs, they were mainly involved in binding (GO:0005488), ion binding (GO:0043167), heterocyclic compound binding (GO:1901363), organic cyclic compound binding (GO:0097159), and organic cyclic compound binding (GO:0097159) (Figure 3). We further analyzed biological functions using the KEGG pathway database and found that the flavonoid biosynthesis pathway (ko00941) was significantly (q value < 0.05) enriched (Figure 4 and Table S7).

3.5. Prediction of miRNAs Related to the Flavonoid Synthesis

We constructed an integration network of miRNAs and 19 targets related to the flavonoid synthesis pathway. In this network, 26 miRNAs were identified as possibly being involved in flavonoid synthesis regulation, including 25 known and 1 novel miRNA (Figure 5A and Table S8). We constructed a clustering correlation heatmap of the miRNA data using R v3.6.3 to further elucidate the relationships among 21 miRNAs with TPM > 1 in Y1, Y4, and Y7 plants and 7 main compounds involved in flavonoid biosynthesis (catechin, quercitrin, kaempferin, dihydrokaempferol, naringenin chalcone, L-epicatechin, and L-phenylalanine) [24]. The levels of most miRNAs were strongly correlated with flavonoid biosynthesis compounds (correlation ≤ −0.8 or correlation ≥ 0.8). The contents of these compounds were significantly positively correlated with the miR166a, miR11534c, and miR11534h expression levels and negatively correlated with the miR396f and miR171a expression levels. The miR397h, miR408, miRN97, and miR396g expression levels were negatively correlated only with catechin content. In addition, the expression levels of the remaining five miRNAs (miRN112, miR399a, miRN94a, miRN84, and novel_25) were not significantly correlated with the contents of these compounds (Figure 4B). These results suggest that 17 miRNAs have different regulatory effects on the synthesis of flavonoids in G. biloba.

Because flavonoid content declines with plant age, we performed an overlap analysis of ARMs and flavonoid-related miRNAs to evaluate the key miRNAs regulating flavonoid biosynthesis in the G. biloba age pathway. Six key miRNAs were identified: miRN79, miR535a, miR171a, miR11534c, miR11534h, and miR166a (Figure 5C). The target genes of these miRNAs include key enzymes for flavonoid biosynthesis, such as F3′H, F3H, and C4H. miRN79, miR535a, and miR171a target three key structural genes of the flavonoid pathway, GbF3′H1 (Gb_19800), GbF3′H2 (Gb_13074), and GbF3′H3 (Gb_33145), respectively. GbF3′H1 is a vital enzyme gene in the flavonoid synthesis pathway, and overexpression of GbF3’H1 can promote the accumulation of flavonoids [28]. miR11534c and miR11534h together target GbF3H1 (Gb_05056) and GbF3H2 (Gb_05058). miR166a targets GbC4H1 (Gb_23185) and GbC4H2 (Gb_38916) (Figure 5A). These key miRNAs exhibit different expression trends; the expression levels of miRN79, miR535a, and miR171a increase with plant age, whereas those of miR11534c, miR11534h, and miR166a decrease (Figure 5D).

MiRNAs that regulate flavonoid synthesis through TF regulation have been widely reported in different species. We further analyzed the targets of the six key miRNAs, and identified important TF families, including SBP, AP2, GARS, MYB, and NB-ARC (Figure 6). An interactive miRNA-TF network analysis was carried out to elucidate the biological functions of the miRNAs. As shown in Figure 6, miR535a targets SBP and AP2 genes, which are key regulators of aging pathways [13,29]. MYB TFs promote or inhibit flavonoid synthesis by directly regulating flavonoid synthesis genes. The miRNA-TF network analysis also showed that miR166a targets MYB genes. Thus, these miRNAs may be involved in the synthesis of flavonoids and G. biloba aging pathways through post-transcriptional regulation.

3.6. miRNAs Associated with Photosynthesis

In our previous study, high stomatal density and photosynthetic capacity were observed in the annual seedling leaves [24]. Here, GO enrichment analysis revealed that the primary metabolic process (GO:0044238) was significantly enriched (Figure 3). In addition, carbon fixation in photosynthetic organisms (ko00710) was significantly enriched based on the KEGG analysis (Figure 4). A total of seven miRNAs (miRN38a, miRN8, miRN529d, miRN11, novel_29, novel_25, and novel_34) were identified that can medicate targets involved in photosynthesis (Figure S7). The target genes of these miRNAs encode different genes, such as phosphoglycerate kinase (PGK, Gb_38763), Phosphoenolpyruvate carboxykinase (PCKA, Gb_33546), Phosphoenolpyruvate carboxylase 4 (PPC4, Gb_40755), and glutamate glyoxylate aminotransferase 1 (GGAT1, Gb_00661). Furthermore, the miRNA expression levels in the annual seedling leaves were relatively low. Interestingly, among the seven miRNAs, we found that miRN38a and novel_29 expression levels were age-dependent and increased with age (Figure S7).

3.7. miRNAs Associated with Plant Hormones

Hormones play key regulatory roles in the biosynthesis of flavonoids and the development of plants [30,31]. We also found that the hormone pathway (ko04075) was also significantly enriched between Y1 and Y7 (p-value < 0.05) (Figure S8B). In total, 70 miRNAs (66 conserved and 4 novel) were involved in the regulation of different hormone pathways (Table S9). Among the 70 hormone-related miRNAs, 20 showed age-related expression patterns through overlap analysis (Figure S8A). In particular, we focused on the JA, GA, and ETH pathways, which are involved in plant age signaling and flavonoid synthesis [32,33,34,35]. The results showed that two miRNAs targeted the JAZ protein in the JA pathway. The expression of the two miRNAs, miR396i and miRN96, increased with age (Figure S8B). In particular, in the GA pathway, we found that miRN79 can simultaneously target two DELLA genes, Gb_34637 and Gb_28283 (Figure S8C,E). In addition, we predicted four miRNAs involved in the ETH pathway, including miR169g, novel_24, miR11534c, and miR11411. Except for miR169g, the other three miRNA levels decreased with age (Figure S8D). Interestingly, we also found that novel_24 and miR396i are involved in two hormonal pathways, respectively. miR396i was involved in the JA and GA pathways, while novel_24 was involved in the ETH and GA pathways (Figure S8). These results suggest that these miRNAs may be involved in flavonoid synthesis and aging pathways by regulating target genes related to hormone signaling.

4. Discussion

G. biloba is an important economic and medicinal plant and has been subjected to extensive chemical investigation because of the wide-ranging pharmacological value of its leaves. However, ginkgo leaves older than 5 years old cannot be used for medicinal purposes due to their low flavonoid content [36,37]. In this study, we found that the contents of flavonoid compounds, including quercetin, kaempferol, and total flavonol aglycones, decreased significantly with age. Consistent with our results, it has been reported that the total flavonol glycoside content was significantly related to the age of the ginkgo tree and decreased with increasing tree age [38,39]. These results indicate that age signaling is an important factor affecting flavonoid accumulation in G. biloba.

So far, considerable progress has been made in understanding the aging pathway in angiosperms. MiR156 is among the most evolutionally conserved miRNAs and acts as an age biomarker in angiosperms. It was highly expressed in juvenile plants and gradually decreased with age in A. thaliana, Arabis alpina, maize, rice (Oryza sativa), and soybean (Glycine max L.) plants [13,40,41,42,43]. The highly conserved miR156 family was also found in gymnosperms, including Picea abies and loblolly pine [44,45]. In this study, we also identified three miRNA156s: miR156a, miR156b, and miR156c. The expression pattern of miR156b was not correlated with plant age in G. biloba. miR156a expression levels were not detected in Y1 and Y4 samples but were suddenly upregulated in 7-year-old G. biloba leaves. Additionally, miR156c expression was age-dependent and increased with age. Interestingly, our previous study also found that miR156 expression in the vascular cambium of G. biloba did not change with age [46]. Similar results have been reported in conifers, and the miR156 expression and targets in Pinus tabulaeformis are not age-dependent [47,48]. Furthermore, we also did not find members of the miR172 family, which also plays a critical role in the aging pathways of angiosperms [14]. These results suggest major differences in the aging pathway regulatory mechanisms of gymnosperms and angiosperms [49,50]. Here, we found 29 miRNAs that were correlated with aging, including miRN79, miR535a, miR166a, miR171a, and miR396, which showed different expression patterns. In Chinese fir (Cunninghamia lanceolata) trees, the abundance patterns of miR166, miR171, and miR396 were also age-related [51]. Similarly, the expression of miR166 was age-dependent and increased with age in 1-, 2-, 5-, and 10-year-old Larix Kaempferi [52]. Thus, these vital miRNAs may be involved in the age-dependent regulatory network of G. biloba.

A large number of studies have reported the involvement of miRNAs in the synthesis of flavonoid compounds in various species [8,53]. In this study, we found that some miRNAs were involved in flavonoid biosynthesis. For example, the expression of miR171a was significantly negatively correlated with flavonoid compound contents, and targets F3′H to participate in flavonoid synthesis. Previous studies have also found that miR171 can target multiple flavonoid synthesis genes, such as DFR, F3′5′H, and F3′H, involved in flavonoid biosynthesis in G. biloba [54]. Similarly, a recent study found that miR171f regulates flavonoid synthesis genes in rice, such as F3′H and F3H [55]. We also found that miR396 was negatively correlated with flavonoid compound levels, consistent with similar results for soybean plants [56]. Interestingly, miR396 has also been reported to be involved in the synthesis of terpene trilactones (TTLs), the important medicinal components of G. biloba leaves, suggesting that miR396 plays a vital role in the synthesis of secondary metabolites of G. biloba [57]. In Larix kaempferi, miR166 overexpression promoted the accumulation of flavonoid compounds and flavonoid-related genes [18]. We also found that miR166 expression was significantly positively correlated with flavonoid levels. Thus, these miRNAs may be involved in flavonoid synthesis.

In plants, photosynthesis affects the production of flavonoid compounds. Flavonoid content was positively correlated with photosynthetic rate, and a higher photosynthetic rate can promote the accumulation of flavonoids [24]. Our previous studies showed that photosynthesis was involved in the synthesis of flavonoid compounds in ginkgo leaves at different ages [24]. In addition, studies on yellow-leaf ginkgo mutants have also found that the expression levels of photosynthesis-related genes were positively correlated with the content of epicatechin [58]. In this study, we also found several vital miRNAs involved in the photosynthesis pathway. Notably, we identified two key miRNAs, miRN38a and novel 29, whose expression was age-dependent. We speculate that miRN38a and novel 29 mediated their target genes to weaken the photosynthetic activity of leaves and reduce the production of flavonoids as age increases.

Besides photosynthesis, plant hormones have also been widely reported to participate in the synthesis of flavonoids [59]. For example, JA promoted flavonoid content in pigeon pea (Cajanus cajan (L.) Millsp.) [60]. The accumulation of flavonoids significantly increased in ginkgo after exogenous GA treatments [37]. In this study, we found several age-related miRNAs, including miR396i and miRN96, which can target the key protein JAZ in the JA pathway. In addition, we also found a vital age-related miRNA, miRN79, that simultaneously targets two DELLA proteins, which acts as a repressor of the GA signaling pathway [61]. These results suggest that these miRNAs may be involved in flavonoid synthesis by regulating target genes related to hormone signaling.

miRNAs that regulate flavonoid synthesis through TF regulation have been widely reported in different species, such as the miR156-SPL module in Arabidopsis [8], the miR858-MYB and miR828-MYB modules in grapes [62], and the miR396-GRF-F3H module in rice [16]. In this study, we found that miR166a targets MYB genes, which are the most comprehensively studied and are considered the most specific and prominent regulators of flavonoid biosynthesis [7]. Additionally, miRN79 targets DELLA proteins, which have been confirmed to be involved in aging pathways and flavonoid synthesis by interacting with other proteins [34,63]. In particular, our results showed that a vital miRNA (miR535a) targets F3′H and SPL genes in miRNA-mRNA and miRNA-TF networks. Previous studies have revealed that SPLs are involved in the aging pathway and regulate flavonoid synthesis by targeting the F3′H gene [8,13]. Overall, we conclude that miRN79 and miR535a may be involved in G. biloba aging and flavonoid biosynthesis pathways via two “miRN79-DELLA” and “miR535a-SPL” modules. To date, the roles of miR535a and miRN79 have not been widely reported in plants. Therefore, the molecular mechanisms of miR535a and miRN79 in age-mediated flavonoid biosynthesis should be investigated in detail.

5. Conclusions

Here, we found that the contents of flavonoid compounds decreased significantly with age. Furthermore, sRNA sequencing was performed to explore vital miRNAs involved in flavonoid synthesis and aging pathways in G. biloba. We identified 29 miRNAs related to the aging pathway according to their miRNA expression patterns. Based on the miRNA-target network and correlation analysis of miRNAs and flavonoids, we identified 17 vital miRNAs, including miRN79, miR535a, miR166a, miR171a, and miR396, which were involved in the synthesis of flavonoid compounds. In particular, we speculate that the “miRN79-DELLA” and “miR535a-SPL” modules may be involved in age-mediated flavonoid synthesis through post-transcriptional regulation. Our results provide insights into the regulatory roles of miRNAs in flavonoid biosynthesis in response to plant age.