Introduction

Epimedium L. is known as Barrenwort and Yin Yang Huo in Chinese, and it is also nicknamed “three branches–nine leaves grass” for having three petioles on its rhizome and three leaves on each petiole. Five Epimedium species, namely E. sagittatum Maxim, E. brevicornu Maxim, E. pubescens Maxim, E. wushanense T. S. Ying, and E. koreanum Nakai, are registered in the Pharmacopoeia of the People’s Republic of China (Committee 2005) due to their high contents of active compounds. So far, more than 260 compounds have been identified in the genus Epimedium with prenyl-flavonol derivatives being the major active components used as chemotaxonomic markers (Ma et al. 2011c). Especially, epimedin A, epimedin B, epimedin C, and icariin (hereafter designated as active components) are defined by the China Pharmacopoeia Committee as major standards for evaluating Epimedium quality (Committee 2005). Several lines of studies have reported that Epimedium flavonoid compounds can strengthen kidney, cure rheumatism and osteoporosis, benefit coronary heart patients, improve immunity, postpone caducity, and reduce tumor propagation (Ma et al. 2011c; Zhu et al. 2011; Tong et al. 2011). However, most previous studies have been focused on extracting and identifying novel phenolic metabolites. While an E. sagittatum 454 EST database has been established (Zeng et al. 2010), the molecular mechanism of the phenylpropanoid biosynthesis pathway in Epimedium L. still remains unclear (Fig. 1).

Fig. 1
figure 1

A simplified schematic pathway for phenylpropanoid biosynthesis in Epimedium. Enzyme abbreviations: PAL phenylalanine ammonia-lyase, CHS chalcone synthase, C4H cinnamate 4-hydroxylase, 4CL 4-coumaroyl: CoA-ligase, C3H p-coumarate 3-hydroxylase

Full size image

The phenylpropanoid pathway plays a pivotal role in the modulation of several secondary metabolites (caffeic acids, flavonoids, isoflavonoids, lignins, coumarins, salicylic acids (SAs), phytoalexins, chlorogenic acids, and stilbenes) responsible for biological functions such as plant development regulation, disease resistance, signal transduction, and pollination attraction. Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5), a rate-limiting enzyme in the phenylpropanoid pathway, concatenates primary and secondary metabolites by carrying out the first catalysis step in the biosynthesis of the phenylpropanoid pathway via catalyzing the non-oxidative deamination of l-phenylalanine to form trans-cinnamic acid and ammonia. Several PAL genes have been isolated from several plants, including Salvia miltiorrhiza (Song and Wang 2009), Angelica gigas (Park et al. 2010), and Solanum lycopersicum (Guo and Wang 2009). In short, PAL exists in plants with a small multiple-gene family (Guo and Wang 2009; Huang et al. 2010; Reichert et al. 2009).

Flavonoid biosynthesis is differentially modulated by different external (environmental) and internal (developmental) factors, such as the quality of light and stresses like wound (Xu et al. 2008), low temperature (Guo and Wang 2009), infection, nutrition, and phytohormones (Guo and Wang 2009; Jiang and Joyce 2003). Furthermore, flavonoids are synthesized in specific cells and/or organs at particular developmental stages and induced by certain stimuli. So far, to the best of our knowledge, no reports have been published concerning the improvement in the content of active components in Epimedium using an in vitro or in vivo method due to lack of mature biotechnologies, including in vitro regeneration (Ma et al. 2011a, b), in vitro hair root system, and transgenic system (Wang et al. 2012) in other species. Consequently, increasing the flavonoid content in Epimedium via induction by phytohormones has become an alternative.

In the present study, we isolated and characterized three EsPALs, given that the PAL enzyme channels the primary metabolite into phenylpropanoid enrichment in Epimedium. Gene expression and phytochemical assay were conducted to understand the roles of EsPALs in modulating different branch end-products biosynthesis in Epimedium phenylpropanoid pathway. Furthermore, the effects of phytohormones and sucrose on the expression of EsPALs and accumulation of flavonoids in Epimedium were evaluated.

Materials and methods

Epimedium samples and phytohormone treatment

All the diploid Epimedium plants tested were grown in the shadow in Wuhan Botanical Garden, China Academy of Sciences. E. sagittatum plants were used to isolate PAL genes. The inner sepals and petals of five Epimedium species, namely E. acuminatum, E. mikinorii, E. leptorrhizum, E. franchetii, and E. sagittatum, were chosen to establish the relationship between EsPAL expression and anthocyanin biosynthesis. The experimental results showed that the E. sagittatum H population contains a significantly higher level of active components than the L population. Subsequently, leaves in the H and L populations were harvested at different developmental stages to identify which EsPAL is in charge of the biosynthesis of the active components in Epimedium. Samples that were designated as S1–S6 and used as temporal samples were cropped at completely curly leaves (S1), partially curly leaves (S2), completely expanding leaves (S3), one-fourth size of mature leaves (S4), one-half size of mature leaves (S5), and full size of mature leaves (S6). The content of lignin in leaves gained a steady increase from S1 to S6. The H2 and L6 lines, representing respectively the H and L populations, were selected for the characterization of the EsPAL expression and the phytochemical assay. In addition, roots, petioles, leaves, flowers, and siliques at the florescence stage were harvested and used as spatial samples. All samples were replicated three times and prepared for the assessment of the amount of EsPAL transcripts.

Epimedium pubescens plants were treated under light stress at 63 μE m−2 s−1 light intensity on day 1 and day 4 to investigate the effects of phytohormones and sucrose on the accumulation of flavonoids and to screen the optimum factors inducing flavonoids in Epimedium plants. Six-year-old E. pubescens plants with new growth roots were soaked in water mixed with 100 μM phytohormones [SA, abscisic acid (ABA), indole-3-acetic acid (IAA), or gibberellic acid (GA)] or 5 % sucrose, respectively. Subsequently, such treated plants were grown at 25 °C in a light bin (SPX-300B-G; Nanjing Ascent Technology Development Co., Ltd). At the same time, a control, designated as CK, was treated with pure water and grown in the same light bin. Triple specimens from each treatment were collected for the characterization of EsPALs in response to plant hormones or sucrose, and the ultimate quantitative assessment of transcript levels using real-time Polymerase Chain Reaction (RT-PCR).

Nucleotide extraction and isolation of EsPALs

DNA was extracted from young leaves of E. sagittatum as previously described (Porebski et al. 1997) and used as the template with which to amplify EsPAL genomic sequences. Total RNAs were extracted from E. sagittatum leaves using TRIzol Kit (Invitrogen, USA) according to the manufacturer’s instructions and prepared for the amplification of EsPAL full-length CDS. An Epimedium 454-EST database (Zeng et al. 2010) was used for the isolation of full-length EsPAL ORF (Open Reading Frame) through the rapid amplification of cDNA ends (RACE) and the digital cloning. For EsPAL1, degenerate primers PAL-DF and PAL-DR were designed and a band of approximately 1,800 bp was amplified and sequenced. Based on the known sequence, the gene-specific primers 5-3EsPAL1 and 3-5EsPAL1 were designed and combined with SMARTIV and CDSIII primer for 5′RACE and 3′RACE, respectively. Finally, the gene-specific primers EsPAL1-FL-F and EsPAL1-FL-R were designed to confirm the full-length EsPAL1 transcript. An approximate 2,200 bp fragment was obtained, resequenced, and checked. In addition, EsPAL2 and EsPAL3 were retrieved from the Epimedium 454-EST database. The EsPAL2-FL-F/EsPAL2-FL-R and EsPAL3-FL-F/EsPAL3-FL-R primer pairs were designed for confirming the full-length sequences of EsPAL2 and EsPAL3, respectively. The detailed primer information of EsPALs was listed in Table 1. The EsPAL sequences were submitted to NCBI under accession numbers HQ331118, JQ319805, and JQ319806.

Table 1 List of primers for EsPALs used in this study
Full size table

Sequence and phylogenetic tree analysis

PALs from the Phytozome database (http://www.phytozome.net/) were retrieved with default parameters for a comprehensive analysis of the PAL phylogeny in the Viridiplantae family. Detailed information for PALs is listed in Table S1. The genomic structure of PAL was predicted based on the Phytozome V7.0 database. For sequence analysis, ClustalX1.83 (Thompson et al. 1997) was used for sequence alignment with manual correction. A PAL phylogenetic tree was constructed via the Neighbor-Joining method using MEGA4.0 (Tamura et al. 2007). The isoelectric points and molecular weights of EsPALs were predicted by Compute pI/Mw (http://expasy.org/tools/pi_tool.html). Sequence identity and similarity between EsPALs were analyzed using EMBOSS Needle (http://www.ebi.ac.uk/Tools/psa/).

Gene expression analysis of EsPALs

To investigate the expression of EsPAL isoforms in another four Epimedium species, we isolated, sequenced, and confirmed the fragments representing each EsPAL isoform of the four Epimedium species. Total RNAs were isolated following the TRIzol protocol (Invitrogen, USA) for a quantitative test of the expression profile of EsPALs. The total RNAs were then reverse-transcribed using PrimeScript RT Reagent Kit with gDNA Eraser (DDR047; TaKaRa), which digested the residual DNA and reverse-transcribed RNA in one step. The EsPAL transcripts were amplified using SYBR Premix ExTaqTMII (DDR081S; TaKaRa) and detected by an ABI 7500 Real-Time PCR system. The PCR program was set as follows: stage 1: 95 °C for 30 s, stage 2: 95 °C for 5 s followed by 60 °C for 34 s repeated for 40 cycles, and stage 3: 95 °C for 15 s, 60 °C for 1 min, and 95 °C for 15 s. Quantitative real-time PCR (qRT-PCR) experiments were performed in triplicate. The expression level of the Epimedium actin (EsActin) was used to standardize the RNA sample. The expression levels of EsPALs relative to EsActin were presented. In order to evaluate the levels of EsPAL1, EsPAL2, and EsPAL3 transcripts, primer pairs were designed for qRT-PCR (Table 1).

Phytochemical determination

For the analysis of active components, approximately 50 mg of ground sample was soaked in 5 ml of 70 % ethanol and ultrasonicated for 30 min before being filtrated through 0.45 μm polyfilters for HPLC analysis. HPLC under 272 nm wavelength was carried out at a flow rate of 1.0 ml/min on Agilent Technologies Series 1100 (Agilent Technologies, Palo Alto, CA, USA). Zorbax SB-C18 chromatographic columns (250 mm × 4.6 mm I.D., 5 μm; Agilent Technologies) were used at 25 °C. The mobile phase consisted of acetonitrile and 36 % acetic acid. Data analysis was performed using Agilent ChemStation (VA.10.02). Epimedin A, epimedin B, epimedin C, and icariin standards were purchased from ChromaDex (Santa Ana, USA).

For the anthocyanin assay, separated sepals and petals of five different Epimedium species were harvested and powdered using liquid nitrogen for anthocyanin extraction using a previously described method with modifications (Mancinelli 1984). Approximately 0.2 g of powders was soaked in 10 ml of acidic methanol (0.1 % HCl, v/v) buffer at 4 °C for 24 h. Subsequently, the mixture was centrifuged at 3,000 rpm for 10 min, and the supernatant was collected to measure the A530 and A657 values using a U-3900 spectrophotometer (Hitachi High-Tech, Japan). Finally, the relative content of anthocyanin in each specimen was calculated as follows: (A530 − 0.25 * A657)/Fresh Weight (g). The data reported for sepals and petals of the Epimedium species represented the mean values of three replicates. For the total flavonoid assessment, leaf samples from phytohormone- and sucrose-treated E. pubescens plants were harvested and dried by silica gel before they were ground using liquid nitrogen. A sample of approximately 20 mg was used to extract the total flavonoid following a previously reported method with minor modifications (Zhang et al. 2008). The ground sample was soaked in 30 ml of ethanol (50 %) and ultrasonicated at 50 °C for 90 min. Subsequently, the filtrate was collected through a 0.22 μm microporous membrane. Finally, the total flavonoid content was quantified using a U-3900 spectrophotometer at 270 nm.

Results

Isolation and sequence analysis of EsPALs

We isolated EsPAL1, EsPAL2, and EsPAL3 mRNA sequence, encoding respectively 709, 716, and 739 amino acids, with the theoretical isoelectric point/molecular weight values being 6.07/77,353.6 Da for EsPAL1, 5.83/77,705.7 Da for EsPAL2, and 6.23/80,315.3 Da for EsPAL3. The protein identities/similarities of EsPAL1/EsPAL2, EsPAL1/EsPAL3, and EsPAL2/EsPAL3 were calculated as 84.9/92.9 %, 60.8/74.1 %, and 62.0/73.8 %, respectively, suggesting that EsPAL1 is closer to EsPAL2 than to EsPAL3, which was confirmed by sequence alignment (Fig. S1), genomic structure (Fig. 2a), and phylogenetic tree (Fig. 2b). As shown in Fig. S1, the predicted active residues, MIO group (Ala–Ser–Gly triad), and phosphorylation residues are conserved in EsPALs and ArthPAL1 (Allwood et al. 1999). The predicted substrate specificities of F127 in EsPAL1, F135 in EsPAL2, and F157 in EsPAL3 are also highly conserved to F134 in BaolPAL2/3 that has PAL activity (Hsieh et al. 2010), suggesting that EsPALs possess PAL activity.

Fig. 2
figure 2

Genomic structure (a) and Neighbor-Joining phylogenetic tree (b) of EsPALs. The numbers labeled in the exon box or on the intron line represents the sizes of the exon and intron, respectively. The codons in red, interrupted by an intron, correspond to amino acids on top of the exons. Percentages indicate the similarity of fragments between EsPALs. Detailed information of the model species PALs is listed in Table S1. Sequence alignment was performed by ClustalX1.83 and corrected manually. An NJ tree was constructed using MEGA4.0 with 1000 bootstrap replicates. Light green, light purple, blue, and red branches represent the dicot PAL, monocot PAL, SemoPAL, and PhpaPAL subgroups, respectively. Furthermore, the closed red circle, light purple rectangle, light green square, and light blue oval near the PALs in the phylogenetic tree represent the numbers of introns present: zero, one, two, and three, respectively. (Color figure online)

Full size image

Tissue expression profiles of EsPALs

To dissect the spatiotemporal expression profiles of EsPALs, roots, petioles, leaves, flowers, and siliques of E. sagittatum were harvested at florescence stage S4 for qRT-PCR. As shown in Fig. 3, the expression patterns of EsPAL1 and EsPAL2 were conserved and abundant in roots. EsPAL3 transcripts were remarkably enriched in leaves, flowers, and siliques, where flavonoid, anthocyanin, and proanthocyanin are generally considered to be enriched.

Fig. 3
figure 3

Spatial expression pattern of EsPALs in E. sagittatum. Vertical bars indicate the standard deviation of three replicates. Mean ± SD determined from three independent samples are shown. Different letters indicate significant differences in gene expression of each EsPAL isoform in different tissues as calculated by Duncan statistical analysis (P < 0.01)

Full size image

Profiles of EsPAL expression and active components in leaves of E. sagittatum

In order to identify the EsPAL isoform modulating the biosynthesis of active compounds in Epimedium, we investigated the profile of EsPALs expression and accumulation of active components in S1–S6 leaf samples in H2 and L6 line, which represent Epimedium populations with a high and low level of active components, respectively. As presented in Fig. 4a, c, the EsPAL3 expression pattern is in perfect agreement with the accumulation trend of active components in the H2 line, except that the content of the active components peaks at S5 while the EsPAL3 transcript peaks at S4. In contrast, EsPAL1 and EsPAL2 were highly expressed in S4–S6 (Fig. 4a), during which Epimedium leaves begin to mature and turn leathery, along with increased lignification as observed in line L6 (Fig. 4b, d). As shown in Fig. 4, the metabolite accumulation lags behind the gene expression of EsPALs. The high level of EsPAL2 transcript accounts partially, if not completely, for the peak of active components at S5 in the H2 and L6 lines (Fig. 4). In addition, the comparative analysis reveals that the EsPAL3 transcript level is lower in the L6 line than in the H2 line at S3 and S4, which might account for the lower content of active components in the L6 line than in the H2 line (Fig. 4).

Fig. 4
figure 4

Differential expression profiles of EsPALs (a, b) and the accumulation patterns of active components (c, d) in leaves of E. sagittatum from H and L populations at five developmental stages. Active components represent icariin, epimedin A, epimedin B, and epimedin C. Mean ± SD determined from three independent samples are shown. Different letters and double stars respectively indicate significant differences in gene expression of each EsPAL isoform in different stages (a, b) and the content of active components (c, d) in leaves at different developmental stages as calculated by Duncan statistical analysis (P < 0.01)

Full size image

EsPAL3 is predominantly involved in anthocyanin biosynthesis in Epimedium

To determine which EsPAL isoforms being responsible for the anthocyanin biosynthesis, we investigated the EsPAL expression patterns in the sepals and petals of five Epimedium species with different colors (Fig. 5a). Under our detection system, EsPAL1 transcripts were not detected in the petals and sepals of all five Epimedium species (data not shown). The EsPAL3 expression profile was perfectly coincident with the color phenotypes in E. acuminatum and E. leptorrhizum (Fig. 5a, c), where tissue(s) with a high level of anthocynain express a high level of EsPAL3 transcripts (Fig. 5b, c). However, the number of EsPAL2 transcripts was smaller than that of EsPAL3 transcripts (Fig. 5c). Low anthocyanin levels were detected in E. franchetii and E. sagittatum (Fig. 5b), although EsPAL3 was highly expressed in petals of both species (Fig. 5c), suggesting that anthocyanin biosynthesis in the two species was blocked in the downstream of the PAL committed step. Surprisingly, both EsPAL2 and EsPAL3 exhibited a slightly higher expression in the white inner sepal than in the orchid petal of E. mikinorii (Fig. 5c), and they accumulated at a lower level than their counterparts in E. acuminatum and E. leptorrhizum petals (Fig. 5c), suggesting that another Epimedium PAL member may control the anthocyanin biosynthesis in E. mikinorii. Alternatively, there may exist certain trans-regulators, such as transcription factors like MYB and bHLH, which may negatively regulate anthocyanin genes in sepals, resulting in less anthocyanin accumulation in sepals than in petals. For instance, FaMYB1 (Aharoni et al. 2001) and MdMYB6 (Gao et al. 2011) have been reported to repress the anthocyanin in transgenic plants.

Fig. 5
figure 5

The color phenotype (A) anthocyanin content (B) and EsPAL expression patterns (C) in separated sepals and petals of five Epimedium species. Mean ± SD determined from three independent samples are shown. Different letters indicate significant differences in the anthocyanin content in the petal or sepal of five different species (B) and the gene expression of EsPAL2 or EsPAL3 isoform in petal and sepal of certain species (C) as calculated by Duncan statistical analysis (P < 0.01)

Full size image

Effects of phytohormones and sucrose on EsPAL expression

As a result of lack of mature biotechnologies, including in vitro regeneration, in vitro hair root system, and transgenic system, phytohormones and sucrose induction has become an alternative to improve the content of active components in Epimedium. In the SA sample, EsPAL1 transcripts were downregulated, but EsPAL2 and EsPAL3 were enhanced on the fourth day as compared with the CK sample (Fig. 6a–c). For the GA3 sample, EsPAL1 and EsPAL2 were remarkably induced by GA3 on the first day and decreased thereafter, whereas EsPAL3 was continuously increased (Fig. 6a–c). For the ABA treatment, EsPAL2 and EsPAL3 transcript levels were reduced on the first day and increased to the level of CK sample. On the contrary, EsPAL1 was constant on the first day and upregulated on the fourth day as compared with the CK sample (Fig. 6a–c). For IAA, there was no statistic difference of EsPALs transcript level compared to CK sample, except that EsPAL2 transcripts increased on the first day and EsPAL1 was downregulated on the fourth day (Fig. 6a–c). In the sucrose sample, a continuous increase was detected in the transcript levels of EsPAL2 and EsPAL3 in response to sucrose (Fig. 6a, b). In contrast, EsPAL1 transcripts increased significantly initially and then sharply decreased (Fig. 6c).

Fig. 6
figure 6

Effects of phytohormones and sucrose on the expressions of EsPAL1 (a), EsPAL2 (b) and EsPAL3 (c), and the accumulation pattern of total flavonoid (d) in E. pubescens. Mean ± SD determined from three independent samples are presented. Different letters and double stars respectively indicate significant differences in the gene expression of each EsPAL isoform in different treatments (a–c) and the flavonoid content (d) in samples treated on day one and day four as calculated by Duncan statistical analysis (P < 0.01)

Full size image

Effects of phytohormones and sucrose on flavonoid accumulation in Epimedium

Assessing the trend of flavonoids in response to phytohormones and sucrose is a prerequisite in drafting strategies for improving the content of active components in Epimedium to increase production for human consumption while minimizing the damage to wild resources of Epimedium caused by commercial overexploitation. As shown in Fig. 6d, the total flavonoid content in the CK sample decreased under light stress. In the SA and GA3 samples, the flavonoid content was continuously enhanced when compared with the CK sample. On contrary, the total flavonoid level increased on the first day and decreased on the fourth day in IAA, Sucrose, and ABA samples, although the total flavonoid content was enhanced compared with that in the CK sample (Fig. 6d).

Discussion

Phenylalanine ammonia-lyase concatenates primary and secondary metabolites by carrying out the first catalysis step in the biosynthesis of the phenylpropanoid pathway. The active components in Epimedium are a major subset of flavonoid products derived from phenylpropanoid pathway. Although many PAL genes, consisting of small gene family, were isolated and characterized in many species, the characteristics of Epimedium PAL are still unclear. In this study, three PAL genes were isolated. EsPAL1, EsPAL2, and EsPAL3 encode 709, 716, and 739 amino acids, respectively. Our results show that EsPAL1 is closer to EsPAL2 than to EsPAL3, which has been confirmed by sequence alignment (Fig. S1), genomic structure (Fig. 2a), and phylogenetic tree (Fig. 2b).

Based on the results shown in Figs. 3, 4, and 5, we can draw two tentative conclusions: (1) EsPAL3 and, to a lesser extent, EsPAL2 are in charge of funneling metabolic flux into flavonoid secondary metabolism toward flower pigments (anthocyanin), UV-protective flavonol derivatives and proanthocyanins; (2) EsPAL1 and EsPAL2 are involved in the lignin biosynthesis. The two conclusions are supported by the following observations: (a) EsPAL3 transcripts are abundant in flavonoid-enriched tissues, including leaves, flowers, and siliques, in which EsPAL2 and EsPAL1 transcripts are low or undetectable (Fig. 3). (b) The expression pattern of EsPAL3, but not EsPAL1, perfectly matches the accumulation profile of active components in different developmental stages of leaves (Fig. 4). EsPAL2 contributes at a lesser extent to flavonoid biosynthesis at S4–S6 (Fig. 4). (c) The expression patterns of EsPAL1 and EsPAL2 are in accordance with the extent of increasingly elevated lignification in mature leaves from S4 to S6 (Fig. 4a, b). (d) EsPAL3 and EsPAL2 account for the anthocyanin biosynthesis in the sepals and petals of Epimedium (Fig. 5). EsPAL3 transcripts highly accumulated in the sepals and petals of the five Epimedium species with different colors, whereas EsPAL2 was expressed at a low level and EsPAL1 transcripts were undetectable (Fig. 5). Altogether, EsPALs have experienced functional diversity in their evolutionary history, resulting in the branch-preference expressions of EsPAL1 for the lignin pathway, EsPAL3 for the flavonoid and anthocyanin pathways, and EsPAL2 for the lignin, flavonoid, and anthocyanin pathways (Figs. 2, 3, 4). We previously found that Es4CL1 and Es4CL2 channel the metabolic flux into the lignin and flavonoid pathways in Epimedium, respectively (Huang et al., unpublished data). It could be concluded that EsPAL1/2 paralleling with Es4CL1 and EsPAL2/3 paralleling with Es4CL2 are responsible for lignin and flavonoid biosynthesis, respectively. In P. tremuloides, the gene expressions of PtPAL1 and PtPAL2, paralleling with Pt4CL2 and Pt4CL1, respectively, are associated with proanthocyanidin and lignin biosynthesis (Kao et al. 2002). Similarly, Arabidopsis 4CL3 and 4CL1/2, which differ in their substrate preferences and expression patterns, have been reported to be responsible for the biosynthesis of flavonoids and lignins and other phenolics, respectively (Ehlting et al. 1999; Humphreys and Chapple 2002). In summary, the differential expressions among the gene family members involved in the phenylpropanoid pathway encode distinct enzyme isoforms, distinguish substrate selectivity, and form distinct membrane-associated metabolons, which further dedicate to different end-products of phenylpropanoid branches (Winkel 2004). Epimedium likely adopts these strategies to function divergently in both EsPALs and Es4CLs and produce various phenylpropanoid end-products.

In this study, the effects of phytohormone and sucrose on EsPALs expression and flavonoid accumulation in Epimedium were also investigated. As presented in Fig. 6b–d, the expression patterns of EsPAL2 and EsPAL3 were in accordance with the flavonoid profile in the SA sample and thus EsPAL2 and EsPAL3, especially EsPAL3, might account for the increase of the flavonoid content in the SA sample. One recent study has documented that the phenylpropanoid pathway may be responsible for the production of a regulatory molecule for SA biosynthesis or accumulation (Huang et al. 2010). Consequently, it can be deduced that exogenous SA and light stress-induced endogenous SA may regulate phenylpropanoid biosynthesis in a feedback manner. When applied, the exogenous SA initially blocks SA biosynthesis and then further channels the metabolic flux into the flavonoid pathway, playing a larger role over the effect of SA-induced depression of EsPALs, until a tradeoff is achieved, which determines the flavonoid content on the first day (Fig. 6d). In contrast, the enhanced EsPAL2 and EsPAL3 transcripts contributed to the high flavonoid content on the fourth day (Fig. 6b, c). For the GA3 sample, the sharp enhancement of EsPAL transcripts resulted in a high flavonoid content (Fig. 6d). EsPALs expression patterns suggested that EsPAL1 and EsPAL3 might be responsible for the high content of flavonoids on the first day, whereas EsPAL3 might dominate the flavonoid accumulation on the fourth day (Fig. 6). The transcripts of genes, including PAP1, F3’H, LDOX, and UF3GT, were highly accumulated in Arabidopsis under low phosphate conditions, and GA treatment rapidly reduced the amount of these transcripts (Jiang et al. 2007). In the present study, GA3 treatment possibly blocked anthocyanin biosynthesis in E. pubescens cultured with water, resulting in the accumulation of an anthocyanin precursor, dihydroflavonol (Fig. 6d). In addition, GA3 application increased CHS and CHI mRNA levels, as well as the protein level in Petunia hybrida (Weiss et al. 1990). Taken together, the integrative signals of hormone (GA) and abiotic stress (light stress and phosphate deficiency) promote the expression of upstream genes in the flavonoid pathway and decrease the expression of downstream genes in the anthocyanin pathway, which might account for the accumulation of higher levels of flavonoids in the GA3 sample (Fig. 6d). Previous studies demonstrated that ABA lessens the activity of PAL at molecular and biochemical levels (Ward et al. 1989). At the same time, a previous study reported that ABA enhances the mRNA level and enzymatic activity of PAL in grape (Jeong et al. 2004). Furthermore, ABA treatment promotes anthocyanin and phenolic accumulation in strawberry (Jiang and Joyce 2003). In Epimedium, the flavonoid content in the ABA sample was higher than that in the CK sample and approached the level in the RT sample, suggesting that the effect of ABA-enhanced EsPAL enzymatic activity might overwhelm the effect of ABA-suppressed EsPAL expression (Fig. 6). For IAA, EsPAL2 transcripts enhanced significantly on the first day, which might account for the increase of the flavonoid content in the IAA sample (Fig. 6). Recently, IAA-induced signals were found to upregulate flavonoid-related genes, including AtCHS, AtCHI, AtFLS, and AtMYB12 (Lewis et al. 2011). Perhaps, IAA signals could also activate the flavonoid structural genes and promote the flavonoid biosynthesis in Epimedium, although only EsPAL2 transcripts were enhanced on the first day (Fig. 6a–c). In the sucrose sample, the expression pattern of EsPAL1 was perfect in accordance with the accumulation pattern of flavonoid, suggesting that EsPAL1 might play a larger role than EsPAL2 and EsPAL3 in the flavonoid biosynthesis under sucrose treatment (Fig. 6).

In conclusion, three EsPALs were isolated and characterized in the present study. Sequence analysis indicated that EsPALs possess PAL activity as in model species. Sequence alignment, phylogeny, and genome structure showed that EsPAL1 shares a higher level of identity with EsPAL2 than with EsPAL3. Gene expression profiles and flavonoid accumulation patterns revealed that EsPAL1 and EsPAL2 may be in charge of lignin biosynthesis, whereas EsPAL3 and, to a lesser extent, EsPAL2 may account for flavonoid and anthocyanin biosynthesis. Our findings suggest that EsPALs play a pivotal role in modulating the metabolic flux into different branches of the phenylpropanoid pathway, resulting in abundant phenolic compounds in Epimedium. Furthermore, under light stress, SA, GA, IAA, ABA, and sucrose can promote flavonoid accumulation in Epimedium via activating different EsPAL isoform(s).