Predicting the function of 4-coumarate:CoA ligase (LJ4CL1) in Lonicera japonica.

4-Coumarate:CoA ligases (4CLs) are a group of essential enzymes involved in the pathway of phenylpropanoid-derived compound metabolisms; however it is still difficult to identify orthologs and paralogs of these important enzymes just based on sequence similarity of the conserved domains. Using sequence data of 20 plant species from the public databases and sequences from Lonicera japonica, we define 1252 adenosine monophosphate (AMP)-dependent synthetase/ligase sequences and classify them into three phylogenetic clades. 4CLs are in one of the four subgroups, according to their partitioning, with known proteins characterized in A. thaliana and Oryza sativa. We also defined 184 non-redundant sequences that encode proteins containing the GEICIRG motif and the taxonomic distribution of these GEICIRG-containing proteins suggests unique catalytic activities in plants. We further analyzed their transcription levels in L. japonica and L. japonica. var. chinensis flowers and chose the highest expressed genes representing the subgroups for structure and binding site predictions. Coupled with liquid chromatography-mass spectrometry (LC-MS) analysis of the L. japonica flowers, the structural study on putative substrate binding amino acid residues, ferulate, and 4-coumaric acid of the conserved binding-site of LJ4CL1 leads to a conclusion that this highly expressed protein group in the flowers may process 4-coumarate that represents 90% of the known phenylpropanoid-derived compounds. The activity of purified crude LJ4CL1 protein was analyzed using 4-coumarate as template and high activity indicating that 4-coumarate is one of the substrates of LJ4CL1.

Introduction

4-Coumarate:CoA ligases (4CLs, EC 6.2.1.12) are a group of essential enzymes involved in the phenylpropanoid-derived compound (PDC) pathway, which converts hydroxylated cinnamic acids into their corresponding thioesters [1]. The PDC pathway as well as its branch pathways generates various classes of secondary compounds, including lignin, flavones, flavonols, anthocyanins, isoflavonoids, and furanocoumarins [2]. PDCs, as a group of the ubiquitous plant secondary metabolites, control flower color, pollination, and stress response [3]. In medicinal plants, certain PDCs have important functions, such as anti-inflammatory, anti-tumor, and anti-human immunodeficiency virus activity [4].

Due to the importance of phenylpropanoid-derived products in plants, 4CL, characterized as a member of a large AMP-binding protein family, has been studied extensively for nearly four decades [5]. However it is still difficult to identify them just based on sequence similarity of conserved domains. Although functionality can be deduced from the domain composition of proteins and enzymes [6], detailed domain analysis of 4CLs remain largely unknown. Their first signature domain (Box I) consists of a serine/threonine/glycine (STG)-rich domain followed by a proline/lysine/glycine (PKG) triplet [7], whereas the second signature domain contains a GEICIRG motif (Box II) [8]. The 4CL-catalyzed CoA ester formation takes place via a two-step reaction. In the first step, 4-coumarate and ATP form a coumaroyl-adenylate intermediate with simultaneous release of pyrophosphate. In the second step, the coumaroyl group is transferred to the sulfhydryl group of CoA, and AMP is subsequently released [8]. The mechanism of an adenylate intermediate formation is also common among a number of other enzymes with divergent functions, including luciferases, fatty acyl-CoA ligases, acetyl-CoA ligases, and the specialized domains of peptide synthetase multienzymes. Despite their low overall amino acid sequence identity, similar reaction mechanisms and the presence of conserved peptide motifs are used to classify 4CLs into a superfamily of adenylate-forming enzymes [9]. The relationship of 4CLs with other adenylate-forming enzymes is recently substantiated by a functional analysis of those key 4CL amino acid residues conserved in other adenylate-forming enzymes [10]. Phylogenetic analyses of the superfamily of adenylate-forming enzymes show that 4CL forms a monophyletic plant-specific group more closely related to luciferases rather than to the long-chain acyl-CoA ligases and acetyl-CoA ligases [11]. However, Souza et al. [12] reported that acyl-CoA synthetase is related to 4CL, although it encodes a novel fatty acyl-CoA synthetase. An in silico analysis revealed that the Arabidopsis genome has 14 genes annotated as putative 4-coumarate:CoA ligase isoforms or homologs. Of these genes, only four are catalytically active in vitro, with broad substrate specificities [13], and the functions of the others are yet to be characterized.

4CLs often present in multiple isoforms that exhibit distinct substrate specificity and coincide with specific metabolic functions (Figure S1). Substrate of 4CLs include sinapic acid, 5-hydroxyferulate, ferulic acid, caffeic acid, 4-coumarate, and trans-cinnamic acid. Three 4CL isozymes from Sorbus aucuparia L. prefer 4-coumaric acid over cinnamic acid in the spectrophotometric assays, but fail to utilize benzoic acid in radioisotopic assays [14]. Allina et al. [15] confirmed that multiple 4CL isoforms present in poplar tissues. However, there has been no evidence to support the differences in substrate-utilization profiles of the partially purified native 4CL isoforms or of the two isoforms expressed in the recombinant forms. Three of the 4CLs from the bryophyte Physcomitrella patens display similar substrate utilization profiles with high catalytic efficiency towards 4-coumarate, but similar efficiency with cinnamate as the substrate to those with caffeate and ferulate [16]. All are efficiently activated by 4CLs from various sources except sinapate [17]. Recombinant Ocimum sanctum 4CL showed the highest activity with p-coumaric acid, followed by ferulic, caffeic, and trans-cinnamic acids [18]. One of the Petunia 4CLs has broad substrate specificity and represents a bona fide 4CL, whereas the other is a cinnamate:CoA ligase [19]. The crystal structures of 4CLs from Arabidopsis thaliana [20] and Populus tomentosa [21] have already been reported. Information regarding 4CL specificity may facilitate predicting substrate preference for the characterization of 4CL-like proteins [22]. Divergent substrate preference also affects the expression of 4CL genes [1]. A differential transcription pattern of each 4CL, in various organs and tissues, as well as distinct temporal patterns of expression, has been observed during flower and fruit development of raspberry [23]. The controlled silencing of At4CL1 and At4CL2 alter the lignocellulose composition of Arabidopsis without affecting its stem growth [24]. Likewise, severe suppression of 4CLs in the coniferous Pinus radiata substantially affects plant phenotype and results in dwarfed plants [25].

The major active PDCs in Lonicera japonica are flavones and flavonols, including chlorogenic acid (CGAs) and luteoloside [26-28]. In this study, we aim to determine the characteristics and function of 4CLs in L. japonica. Recently, a number of 4CLs have been characterized in A. thaliana and Oryza sativa based on transcriptomic studies [29]. Here we report the identification and characterization of LJ4CLs and propose the relationship of LJ4CL function and the related active compounds in L. japonica, based on expression data [30], protein structure analysis, and substrate characterization.

Results and Discussion

Global Phylogeny and Duplication of AMP-Binding Proteins

Using Pfam (AMP-binding enzyme PF00501) and Interpro (IPR000873 and IPR020845), as well as information from public genome databases and our own transcriptome databases of L. japonica, we gathered 1252 non-redundant sequences that encode AMP-binding proteins from 20 different species, representing a diverse taxonomic background (Table S1). The result shows that AMP-binding proteins are widely distributed among bacteria, fungi, animals, and plants. Among 1252 sequences, 146 putative AMP-binding protein sequences are identified in A. lyrata as compared to 46 and 10 AMP-binding proteins in Culex quinquefasciatus and Escherichia coli, respectively (Table 1).

Table 1.

Copy number of AMP-binding domain in 20 species.

Kingdom Group		Class	Clusters *	Number of copies
			Species	1	1-1	1-2	1-3	1-4	2	2-1	2-2	2-3	2-4	3	Total
Animal			Culex quinquefasciatus	39	5	5	27	2	6	1	4	1	0	1	46
Bacteria			Escherichia coli	5	0	3	2	0	3	0	2	1	0	2	10
Fungus			Penicillium marneffei	20	3	3	12	2	25	3	7	15	0	5	50
			Aspergillus nidulans	32	2	8	17	5	28	3	5	19	1	1	61
Plant	Algae		Chlamydomonas reinhardtii	18	11	2	0	5	9	0	8	0	1	0	27
	Bryophyte		Physcomitrella patens	43	8	16	14	5	2	0	0	1	1	0	45
	Pteridophyta		Selaginella moellendorffii	70	15	26	26	3	14	3	9	2	0	0	84
	Gymnospermae		Pinus taeda	4	0	0	4	0	0	0	0	0	0	0	4
			Larix gmelinii	18	0	0	18	0	0	0	0	0	0	0	18
			Pseudotsuga menziesii	17	0	0	17	0	0	0	0	0	0	0	17
	Angiospermae	Dicotyledoneae	Glycine max	99	30	29	35	5	12	5	6	1	0	0	111
			Populus trichocarpa	71	21	24	23	3	8	3	4	1	0	0	79
			Arabidopsis thaliana	48	13	17	16	2	6	0	5	1	0	0	54
			Arabidopsis lyrata	146	24	0	122	0	25	8	15	2	0	1	172
			Vitis vinifera	34	8	9	14	3	6	1	4	1	0	0	40
			Lonicera japonica	103	43	36	17	7	14	0	12	2	0	0	117
			Lonicera japonica var. chinensis	73	33	20	16	4	12	0	10	2	0	1	86
		Monocotyledoneae	Zea mays	82	23	21	35	3	15	3	9	3	0	0	97
			Sorghum bicolor	48	16	13	17	2	10	3	6	1	0	0	58
			Oryza sativa	60	31	12	16	1	16	3	9	4	0	0	76
Total															1252

Clusters were showed in Figure S2.

We classified all AMP-binding protein sequences into three clusters, where 82% of them are in Cluster 1. The gymnospermae species are all found in cluster 1, whereas pteridophyta, algae, monocotyledoneae, and dicotyledoneae are divided into both Clusters 1 and 2. A few sequences from A. lyrata and L. japonica Thunb var. chinensis (Wats.) are in Cluster 3. Cluster 1 has four subgroups. According to the known function of the proteins in A. thaliana and O. sativa, we speculate that long chain acyl-CoA synthase (ACS) belongs to Subgroup 1 and that acyl-acting enzyme (AAE), o-succinylbenzoate-CoA ligase, and benzoate-CoA ligase are in Subgroup 2. 4CL is expected to be in Subgroup 3, and Subgroup 4 includes acyl-CoA and malonyl-CoA synthases. The predicted 4CL group is the same as true 4CL enzymes from genome-wide analysis of a land plant-specific acyl:coenzymeA synthetase (ACS) gene family in Arabidopsis, poplar, rice, and Physcomitrella [31] and 4CL-like. From the Neighbor-Joining trees, we also found that the gymnospermae sequences are only clustered in Subgroup 3, but the algal sequences are not (Figure S2).

Global Phylogeny and Duplication of GEICIRG-Containing Proteins

In the species of which the genome is completely known, there are 184 non-redundant sequences that encode GEICIRG-containing proteins, which are unique to Plantae, including gymnospermae, algae, bryophyte, pteridophyta, and angiospermae (Table 2). The GEICIRG motif is absolutely conserved in all 4CLs, and its central cysteine residue is suggested to be directly involved in catalysis [8], and the participation of a cysteine residue in catalysis has also been observed for other adenylate-forming enzymes [32].

Table 2.

Copy number of containing-GEICIRG protein.

		Clusters *	Number of copies				Total	Genome size	Reference
			ACS	AAE	4CL	ACS
Species			1	2	3	4		M
Gymnospermae		Pinus pinaster	0	0	1	0	1
Algae		Chlamydomonas reinhardtii	2	0	0	1	3
Bryophyte		Physcomitrella patens	0	3	12	0	15
Pteridophyta		Selaginella moellendorffii	3	2	5	6	16
Angiospermae	Dicotyledoneae	Glycine max	3	7	10	4	24	1100	[33]
		Populus trichocarpa	5	1	6	2	14	485	[34]
		Arabidopsis thaliana	2	1	7	5	15	135	[35]
		Vitis vinifera	2	2	3	2	9	505	[36]
		Lonicera japonica	5	4	1	5	15
		Lonicera japonica var. chinensis	4	2	2	3	11	~800	Our group
	Monocotyledoneae	Zea mays	3	0	12	8	23	2300	[37]
		Sorghum bicolor	2	1	8	4	15	730	[38]
		Oryza sativa	6	0	7	10	23	467	[39]
Total			37	23	74	50	184

Clusters were showed in Figure S3. ACS, long chain acyl-CoA synthase; AAE, Acyl-acting enzyme/o-succinylbenzoate-CoA ligase/benzoate-CoA ligase; 4CL, 4-coumarate:CoA ligase.

To obtain a global view of the phylogenetic relationships among GEICIRG-containing proteins, we first constructed an NJ tree, based on their AMP-dependent synthetase/ligase domain sequences (Figure S3). These sequences clustered together with a strong bootstrap support and four clades are clearly distinguishable, including 4CLs, two ASCs, and one AAE protein as subgroups, based on the known functions of the proteins in A. thaliana and O. sativa.

The phylogenetic reconstructions also revealed that subsequent duplication of proteins containing the GEICIRG motif occurred in different lineages. Analysis on these clades—4CLs, ACSs, and AAE in dicotyledoneae, monocotyledoneae, pteridophyta, bryophyte, algae, and gymnospermae—suggests that gene duplication occurred among the GEICIRG-containing proteins prior to the divergence of Angiospermae and Pteridophyta. In the NJ tree, representatives of 4CL and AAE are classified into two ASC clades. The major sequences in the AAE clade (Cluster 2) are from Pteridophyta and Dicotyledoneae, where only one copy is found in Sorghum bicolor.

The phylogeny based on AMP-dependent synthetase/ligase domain sequences (Figures S4 and S5) revealed that ACS proteins are classified into two clusters (1 and 4). In dicots, the lineage-specific duplication events in Cluster 1 are similar to those in Cluster 4; however, Cluster 4 has more duplication events than Cluster 1 in monocots. In cluster 1, two ACS copies are paralogous in Chlamydomonas reinhardtii, indicating that duplication occurred early in angiosperm evolution. Three ACS copies are also present in Selaginella moellendorffii and are separated by angiosperm sequences, which are related to ancient gene duplication.

The most abundant group, containing 4CLs (Cluster 3), shares the GEICIRG-motif. The phylogeny based on AMP-dependent synthetase/ligase domain sequences (Figure 1) revealed distinct and highly supported clades that group the 4CL proteins into: (1) Dicotyledoneae; (2) Mixed group; and (3) Monocotyledoneae. Apart from the early duplication of 4CLs during the evolution of higher plants, successive duplications must have occurred among eudicots and monocots as additional copies are present in Arabidopsis (7 copies) and rice (7 copies). Ten and 12 copies are also evident in Glycine max and in Zea mays, respectively (Figure 2). The high occurrence of duplication events are attributable to frequent genome duplications in angiosperm evolution.

Figure 1.

Phylogeny and expression of 4CL sequences in Cluster 3. A neighbor-joining tree containing 62 sequences was generated based on the AMP-dependent synthetase/ligase domain sequences. A bootstrap value of 1000 replications was applied. The RPKM value of sequences in flowers of Lonicera japonica is shown.

Figure 2.

PDCs in Lonicera japonica.

Expression of GEICIRG-Containing Proteins in L. japonica Flowers

We analyzed the transcription of the GEICIRG-containing proteins based on our own data. In Cluster 1, the GEICIRG-containing proteins are clustered into two groups and the first group contains two pairs of orthologs. In the other group, the paralogs of L. japonica are found to be expressed at low levels, and their average Reads Per Kilo bases per Million reads (RPKM) is 22.91, which is lower than that of the first group. Although there are a similar number of orthologs in L. japonica and L. japonica var. chinensis, the collective average RPKM value of the GEICIRG-containing paralogs is 1.87-fold higher in L. japonica than that in L. japonica var. chinensis. A similar trend is found in Cluster 2 and other results are summarized in Table 3.

Table 3.

Copy and RPKM of contained GEICIRG protein in Lonicera japonica.

Cluster *	Varieties	Lonicera japonica	Lonicera japonica var. chinensis	Total
	Copy and RPKM
1	Number of copies	5	4	9
	RPKM	320.4257	171.5426	491.9683
2	Number of copies	4	2	6
	RPKM	250.9583	306.7299	557.6882
3	Number of copies	1	2	3
	RPKM	50.3633	234.6021	284.9654
4	Number of copies	5	3	8
	RPKM	292.3302	236.8353	529.1655

Clusters were shown in Figure S3.

Substrate-Binding Diversity in the Expressed GEICIRG-Containing Proteins in L. japonica

We examined the structure of four GEICIRG-containing proteins in L. japonica based on their expressions—the most highly expressed genes in each cluster (Figure S6; Table S2). Only six conserved residues (L-243, H-247, S-316, Y-342, T-345, and E-346, in LJ4CL1) were found, and the T-345 residue has the highest frequency over all others. Another high-frequency residue, M-332, belonging to the B-subdomain (adenylation domain), is relatively conserved [40] but is not found in LJACS2. Thus, T-345 may be the conserved residue responsible for the function of adenylation domain in L. japonica.

Highly expressed LJ4CL1 is found in L. japonica bud [30], and it has the following putative substrate binding residues: I-196, Y-197, G-317, G-343, P-349, V-350, and L-351. The substrate-binding residues Y-236, G-306, G-331, P-337, and V-338 in 4CL1 of P. tomentosa are identified, based on their crystal structures, as well as mutagenesis and enzymatic activity studies. The Y residues may relate to Pt4CL1 activity against caffeic and ferulic acids, but not 4-coumaric acid. Similarly, the G residue may relate to Pt4CL1 activity against 4-coumaric and caffeic acids [21]. However, Schneider et al. [22] also reported that 12 amino acid residues, I-252, Y-253, N-256, M-293, K-320, G-322, A-323, G-346, G-348, P-354, V-355, and L-356, form the substrate specificity code of At4CL. The N-256 residue is located at a distance of 3.1 Å from the hydrogen atom of the 4-hydroxyl group of caffeic acid. The residues M-293 and K-320 formed a clamp structure around the 3-hydroxyl group of caffeate. However, the residues N-256, M-293, and K-320 do not seem to have the corresponding residues in LJ4CL1.

Our results suggest that ferulic acid and 4-coumaric acid are candidate substrates of LJ4CL1. To verify our prediction, we analyzed the PDCs of L. japonica flowers using LC-MS. Ten compounds, namely, chlorogenic acid, ferulic acid, rutin, hyperoside, isoquercitrin, luteoloside, quercitrin, luteolin, quercetin, and apigenin, are identified. Considering the transferase reaction that couples quinic acid to cinnamic acid derivatives is reversible, the CGAs are a storage form of cinnamic acid derivatives, and are considered as intermediates in the lignin biosynthetic pathway [41,42]. Based on their chemical structure, the intermediates are classified into two groups. The putative substrates of these two groups are 4-coumarate and ferulic acid, consistent with the putative function of LJ4CL1.

I-196, Y-197, G-317, and G-343 are also observed in LJACS1 and LJACS2, and their frequencies are lower than those in LJ4CL. P-349, V-350, and L-351 are not found in LJACSs and LJAAE, and they may be conserved residues related to 4-coumarate. The size of the binding pocket is most important in determining the substrate specificities of P. tomentosa 4CLs [21]. Our results suggest that the diversity of the residues of the binding-site controls the enzymatic function, although they have the same conserved catalytic motif. Ninety percent of the substrates for the known PDCs in L. japonica are 4-coumarate, and the substrate specificity may be related to the conserved residues in LJ4CLs.

To confirm our suspect, LJ4CL1 protein was expressed in E. coli. The activity of crude LJ4CL1 protein was analyzed using 4-coumarate as template and high activity (19.36 U μg−1 protein−1) was observed, indicating that 4-coumarate is one of the substrates of LJ4CL1.

Experimental Section

Classification of AMP-Dependent Synthetase/Ligase Sequences and GEICIRG-Motif Proteins

We searched for the AMP-dependent synthetase/ligase sequences of twenty-one species (Table S1) using the pfam [43] and InterPro databases. These species include one animal, one bacterium, two fungi, two algae, three gymnospermae, two pteridophyta, seven dicotyledoneae, and three monocotyledoneae. We compared the sequences against the sequence “GEICIRG” with an e-value cut-off below e−20 using BlastP (protein-protein BLAST) [44] to determine the GEICIRG-containing proteins from the best reciprocal hits.

Phylogeny of AMP-Dependent Synthetase/Ligase Sequences and GEICIRG-Motif Proteins

We used the AMP-dependent synthetase/ligase domain of AMP-dependent synthetase/ligase sequences and GEICIRG-motif proteins to construct neighbor-joining trees using Mega 5.0 [45] and ClustalW2 [46], respectively, with a bootstrap value of 1000 replicates. In addition, we reconciled preliminary trees by setting the bootstrap value greater than 50% to yield a more credible consensus tree.

Indentification of Orthologs and Paralogs

To identify orthologs, we performed an all-against-all sequence comparison using BLAST with an e-value cut-off below e−20. The orthologs were then determined based on the best reciprocal hits [47]. We implemented a more stringent criterion that the alignment length percentage against the longer protein must be above 80%.

Gene Expression Analyses

The gene expression profiling of L. japonica flowers was performed in a previous work [30]. The expression level was normalized with total mapped reads and the contig length, similar to the reads per kilobase of exon model per million mapped reads (RPKM) method [48]. The RPKM value for each transcript was calculated as the number of reads per kilobase of the transcript sequence per million mapped reads [49].

Protein Structure and Binding Site Prediction

The three-dimensional protein structures were predicted from the amino acid sequences using the online version of I-TASSER [50]. Based on the C-score and TM-score, the top ten models were predicted and the structural analogs with similar binding sites were identified. All of the predicted binding site residues in the model were summarized. The diversity of the predicted binding site residues of four proteins was analyzed.

LC-MS Analysis of L. japonica Flowers

Dried L. japonica flowers (medicinal materials) were separately comminuted with a miller. Each solid sample (40 mesh, 0.50 g) was accurately weighed, and extracted with 50 mL of 70% aqueous ethanol with ultrasonication for 30 min. The extract was cooled to 25 ºC, diluted to 50 mL with 70% aqueous ethanol, and filtered with a 0.45 μm Millipore filter membrane. Then, 10 μL of the filtrate was injected into the liquid chromatography–mass spectrometry (LC-MS) system (Agilent RRLC/Agilent ion trap 6320, Agilent, Santa Clara, CA, USA) for analysis (Figure S7). The LC-MS/MS systems were set to a 1.0 mL/min flow rate and performed in an Agilent TC-C18 reserved-phase column (5 μm, 250 mm × 4.6 mm). The mobile phases consisted of deionized water-formic acid (99:1, v/v) and methanol. The elution conditions were same as the high-performance liquid chromatography (HPLC, Agilent, Santa Clara, CA, USA) conditions used in a previous work [30]. The detection wavelength was set to 242 nm, and the column temperature was maintained at 25 ºC. All standard compounds were purchased from the National Institutes for Food and Drug Control, Beijing, China.

Expression of 4CL Protein in E.coli and Enzyme Activity Assay

The open reading frame (ORF) of LJ4CL was cloned into the expression vector pGEX-4T-1 and transformed into Transetta (DE3) chemically competent cells (Beijing TransGen Biotech Co., Ltd., Beijing, China), respectively. The vector pGEX-4T-1 (+) allows inframe cloning of PCR products resulting in a GST-tag attached at the N-terminal end of the recombinant protein. Expression of the recombinant protein was induced by adding isopropyl-β-d-1-thiogalactopyranoside (IPTG) and cells were harvested at 9 h. The activity of 4CL was analyzed according to Voo et al. [51]. The 1 mL reaction mixture contained 50 μL crude enzyme, 0.2 mM 4-coumarate, 0.8 mM ATP, 7.5 mM MgCl2, and 38 M CoA in 100 mM Tris-HCl buffer (pH 7.5). One unit of 4CL was defined as the amount of enzyme that causes a decrease in A333 of 0.01 units min−1. Protein concentration in the extracts was determined using the Lowry method [52].

Conclusions

4CLs form an important enzyme family for the phenylpropanoid-derived pathway in plants. Our analysis on AMP-binding and GEICIRG-containing proteins from the genome and transcript sequences of 19 species; including an in-house generated dataset containing 40,000 transcript scaffolds of L. japonica; allowed us to further exploit 4CL structural (domain and motif) features and validate structural predictions based on chemical assays. We also propose the putative substrate-binding residues of LJ4CLs and defined the major substrate of the PDC pathway in L. japonica. Our study paves a way for further studies on 4CLs and their related metabolic pathways in medicinal plants.