Characterization of a Cytosolic Acyl-Activating Enzyme Catalyzing the Formation of 4-Methylvaleryl-CoA for Pogostone Biosynthesis in Pogostemon Cablin

Abstract

Pogostone, a compound with various pharmaceutical activities, is a major constituent of the essential oil preparation called Pogostemonis Herba, which is obtained from the plant Pogostemon cablin. The biosynthesis of pogostone has not been elucidated, but 4-methylvaleryl-CoA (4MVCoA) is a likely precursor. We analyzed the distribution of pogostone in P. cablin using gas chromatography-mass spectrometry (GC-MS) and found that pogostone accumulates at high levels in the main stems and leaves of young plants. A search for the acyl-activating enzyme (AAE) that catalyzes the formation of 4MVCoA from 4-methylvaleric acid was launched, using an RNAseq-based approach to identify 31 unigenes encoding putative AAEs including the PcAAE2, the transcript profile of which shows a strong positive correlation with the distribution pattern of pogostone. The protein encoded by PcAAE2 was biochemically characterized in vitro and shown to catalyze the formation of 4MVCoA from 4-methylvaleric acid. Phylogenetic analysis showed that PcAAE2 is closely related to other AAE proteins in P. cablin and other species that are localized to the peroxisomes. However, PcAAE2 lacks a peroxisome targeting sequence 1 (PTS1) and is localized in the cytosol.

Introduction

Pogostemonis Herba, commonly known as ‘Guang-Huo-Xiang’ in Chinese and patchouli in English, is produced from the dried aerial part of the plant Pogostemon cablin (Blanco) Benth. (Labiatae). As Pogostemonis Herba and its essential oil (patchouli oil) possess variety of pharmacological activities (Lu et al. 2011, He et al. 2013, Li et al. 2013, Wang et al. 2016), they have widely been used in many Asian countries as traditional Chinese medicine for the treatment of fatigue, summer heat, nausea, vomit, abdominal distension and other ailments since ancient time (Zhao et al. 2005, Chen et al. 2015, Hu et al. 2017). Patchouli oil has also been widely used for perfumes, cosmetics and food stuffs (Pattnaik et al. 1996, Lin et al. 2014, Yang et al. 2016). Pogostone (formula C₁₂H₁₆O₄, Fig. 1) is one of the major constituents of patchouli oil and is largely responsible for the intensive aromatic odor and, therefore, designated as a chemical marker for the quality control of patchouli oil (Swamy and Sinniah 2015, Yang et al. 2016). This compound has been demonstrated to exert various bioactive activities including anti-bacterial (Peng et al. 2014), anti-candida (Li et al. 2012), antifungal (Li et al. 2012, Luchesi et al. 2020), anticancer (Cao et al. 2017), insecticidal (Huang et al. 2014), anti-inflammatory (Li et al. 2014), immunosuppressive (Su et al. 2015) and gastro-protective (Chen et al. 2015) activities. The levels of pogostone in patchouli oil, mainly obtained from stems and leaves of Pogostemon plants, vary among samples collected from different geographic locations and at different periods (Luo et al. 2003, Li et al. 2004, Hu et al. 2006).

The biochemical steps involved in the biosynthesis of pogostone are currently unknown. It is likely that one or more polyketide synthases that use various starter molecules are involved (Fig. 1A). One obvious such starter is 4-methylvaleryl-CoA (4MVCoA), derived from the branched-chain fatty acid 4-methylvaleric acid (Fig. 1B). Acyl-CoA thioesters such as 4MVCoA are formed by members of the acyl-activating enzyme (AAE) superfamily that activate carboxylic acids through a two-step mechanism. First, the carboxylic acid group is adenylated to form an enzyme-bound acyl-AMP intermediate by ATP pyrophosphorylysis, and this step is followed by the displacement of AMP with a CoA group to form the relevant acyl-CoA thioester (Shockey and Browse 2011) (Fig. 1B). One of the unifying features of this superfamily is the well-conserved 12 amino acid residue AMP-binding motif (PROSITE PS00455), which is used to identify putative genes encoding AAEs (Shockey et al. 2003). Besides carboxylic acids, various plant hormones are also conjugated to amino acids by members of this superfamily (Nobuta et al. 2007, Okrent et al. 2009).

We have begun identifying and characterizing the enzymes involved in the biosynthesis of pogostone. The combined analysis of transcriptome and metabolome of different tissues at different developmental stages has been effectively used to discover enzymes involved in the specialized metabolic pathway in many species such as pyrethrum (Xu et al. 2018, 2019). To facilitate the elucidation of the pogostone biosynthetic pathway, a transcriptome assembly of P. cablin was generated in this study from RNAseq analysis of seedling, root, stem and leaf harvested at different developmental stages. GC-MS analysis established that pogostone is largely present in seedlings, the main stems and top leaves of main stem of P. cablin at earlier developmental stage and showed a developmentally specific accumulation pattern in top leaves of the main stem of P. cablin.

Here, we report the identification and characterization of the enzyme catalyzing the formation of 4MVCoA. Through transcriptome analysis of P. cablin RNAseq database, we identified 31 putative AAE genes including PcAAE2, the transcript profile of which shows a positive correlation with the accumulation pattern of pogostone. Through in vitro enzymatic assays, we identified PcAAE2 as encoding 4MVCoA synthetase. We further showed that PcAAE2 is a cytosolic enzyme that may have evolved from an ancestral peroxisome-targeted AAE.

Results

Distribution of pogostone in P. cablin top leaves newly generated from the main stems shows a developmental pattern

The levels of pogostone in P. cablin seedlings, roots, main stems and top leaves from the main stems at different developmental stages (Fig. 2A) were determined by gas chromatography-mass spectrometry (GC-MS) analysis of methyl tert-butyl ether (MTBE)-extracted macerated tissues. This analysis showed that pogostone is largely produced in P. cablin seedlings and in the main stems and the top leaves of the main stem of plants younger than 7 weeks (Fig. 2B). Plants older than 7 weeks contained negligible amounts of pogostone compared to younger plants (Fig. 2B), which indicates that the distribution of pogostone in P. cablin newly generated leaves from the main stem is developmentally controlled (Fig. 2B). Besides pogostone, another compound related to pogostone (with one carbon shorter in the side chain and likely derived from isovaleryl-CoA), which we named pogostone B, was also detected in P. cablin plants (Supplementary Fig. S1). Pogostone B showed a very similar distribution pattern to that of pogostone in the tested tissues, albeit present at much lower levels (Fig. 2B).

Identification of candidate genes responsible for 4MVCoA biosynthesis

To identify the genes encoding the enzymes involved in the conversion of 4-methylvaleric acid to 4MVCoA, RNAseq libraries were constructed from six different P. cablin tissue samples including Seedling, 5W-Root, 5W-Stem, 5W-TLeaf, 8W-TLeaf and 8W-SLeaf from P. cablin plants (Fig. 2A). The corresponding transcriptome assemblies were then constructed and unigenes were annotated by comparison with genes on the UniProt database and using the BLASTX program. AAE genes of P. cablin were identified by using Arabidopsis representative AAE sequences of each of the seven clades of the AAE superfamily, which had been previously defined (Shockey and Browse 2011), to query our transcriptome assembly using TBLASTN software. In addition, we searched the functional annotations of the unigene with keywords such as ‘AMP-binding protein’. Since we sought to search for the gene responsible for 4MVCoA biosynthesis from 4-methylvaleric acid, the unigenes encoding putative AAEs catalyzing the conjugation of plant hormones to amino acids were not included in this search. A total of 31 unigenes encoding putative AAEs, containing the conserved AMP-binding motif (PROSITE PS00455) of the AAE family, were identified and further analyzed (Table 1, Supplementary Tables S1 and S2).

Previous phylogenetic analysis has defined seven subfamilies of AMP-dependent synthetases and ligases in Arabidopsis (Shockey and Browse 2011). Phylogenetic analysis indicated that all the 31 P. cablin AAE unigenes group into the previously established six clades of the AAE superfamily (Supplementary Fig. S2). Several members of clade VI from Arabidopsis thaliana and Humulus lupulus have been previously functionally analyzed by in vitro biochemical assays. HlCCL2 and HlCCL4 from H.lupulus utilize various short-chain fatty acids and branched short-chain fatty acids as substrates (Xu et al. 2013). Both of these CCLs prefer branched short-chain fatty acids as substrates, with isovaleric acid and isobutyric acid as the preferred substrates for HlCCL2 and HlCCL4, respectively (Xu et al. 2013). The Arabidopsis gene At2G17650 encodes an enzyme that also prefers branched short-chain fatty acids as the substrates, with isovaleric acid serving as the most preferred substrate in in vitro enzymatic assays (Xu et al. 2013). Recently, identified HcAAE1 from Hypericum calycinum exhibits promiscuous substrate preference, with benzoic acid and several straight/branched short-chain fatty acids such as butyric acid and isobutyric acid as preferred substrates (Singh et al. 2020). These data suggest that there might also be one or more members among the P. cablin clade VI genes that encodes an enzyme with a preference for branched-chain fatty acids, including 4-methylvaleric acid. Therefore, we concentrated on the eight genes in the P. cablin clade VI, which we designated as PcAAE1∼8, for further examination (Table 1 and Supplementary Table S1).

Tissue-specific/developmental expression patterns of PcAAE clade VI genes

Analysis of expression patterns of the eight genes in P. cablin RNAseq database indicated that PcAAE2 expression shows the strongest positive correlation with the distribution pattern of pogostone in these tissues of all these eight genes (Supplementary Table S1; Fig. 2B). In particular, PcAAE2 transcript reads are much lower in older tissues (8W-TLeaf and 8W-SLeaf) than in seedling and younger plants (5W-Stem and 5W-TLeaf). To more accurately test whether the transcript profiles of PcAAE1∼8 in the clade VI of AAE superfamily are correlated with the distribution pattern of pogostone in different tissues of P.cablin, their expression patterns were further confirmed by quantitative reverse transcription polymerase chain reaction (RT-qPCR) in the same tissue samples used for distribution analysis of pogostone (Fig. 3). The expression patterns of PcAAE1∼8 obtained by RT-qPCR analysis are mostly consistent with that in RNAseq database (Supplementary Table S1; Fig. 3). Again, the expression patterns of PcAAE2 in the tested samples show a strong positive correlation with the distribution pattern of pogostone: (i) PcAAE2 is highly expressed in the five tissue samples of seedling, 5W-TLeaf, 5W-Stem, 6W-TLeaf and 6W-Stem of P. cablin plants, which show large levels of pogostone; (ii) the expression levels of PcAAE2 in top leaves from the main stems of 7, 8 and 9-week-old P. cablin plants, including the tissue samples of 7W-TLeaf, 8W-TLeaf and 9W-TLeaf, dramatically decrease, and these tissues show negligible levels of pogostone compared to those observed in tissue samples of earlier developmental stages.

PcAAE2 encode a 4MVCoA synthetase

Sequence alignment showed that of the eight P. cablin clade VI proteins, PcAAE1 and PcAAE2 share the highest amino acid sequence identities with HlCCL2 (66.0% and 66.2%, respectively) and HcAAE1 (74.6% and 68.9%, respectively) (Supplementary Fig. S3) and, thus, are the closest homologs to HlCCL2 and HcAAE1, suggesting that they may have the similar substrate preference toward branched-chain fatty acids. The combination of the positive correlation of expression patterns of PcAAE2 with the distribution pattern of pogostone and the predicted substrate preference of the protein encoded by PcAAE2 toward branched-chain fatty acids made it the most likely candidate gene encoding the enzyme responsible for the formation of 4MVCoA precursor of pogostone compared with the other seven genes encoding P. cablin clade VI proteins.

To test this hypothesis, we performed in vitro biochemical assays with all eight AAE proteins in the P. cablin clade VI of the family (Table 1). All proteins were produced in Escherichia coli with a fused N-terminal His6 tag and purified as soluble proteins. All the eight purified recombinant proteins were initially tested in the coupled enzyme assay with a variety of fatty acid substrates including straight short/medium-chain fatty acids and branched short/medium-chain fatty acids and benzoic acid at a 0.5 mM concentration of the acid substrate and 0.5 mM of CoA. Except for the three recombinant proteins PcAAE1∼3, the other five recombinant proteins showed no detected enzymatic activities with any of these selected substrates.

Recombinant PcAAE1 utilizes isovaleric acid as the best substrate among the selected substrates and shows lower activities toward butyric acid, valeric acid and 4-methylvaleric acid (Table 2). Recombinant PcAAE2 shows high activities toward 4-methylvaleric acid, valeric acid and butyric acid and utilizes 4-methylvaleric acid as the most preferred substrate (Table 2). Recombinant PcAAE2 also accepts 5-methylhexanoic acid, heptanoic acid, hexanoic acid and isovaleric acid as substrates but at a lower rate (Table 2). Purified PcAAE3 accepts very short-chain fatty acids as substrates, including acetic acid, propanoic acid and butyric acid, with propionic acid as the preferred substrate (Table 2). Therefore, PcAAE1 and PcAAE2 are assigned as the probable isovaleryl-CoA synthetase and 4MVCoA synthetase, respectively, and may supply CoA ester precursors for pogostone and its analogue pogostone B biosynthesis. Thus, in-depth studies of the dependency of their enzymatic activities on incubation conditions and components were further performed. Highest activities of PcAAE1 and PcAAE2 with their respective preferred substrates were observed at pH 7.5 and 30°C. The activities of PcAAE1 and PcAAE2 were dependent on divalent cations with Mg²⁺ being preferred and increased by 1.7-fold and 0.8-fold, respectively, when the concentration of Mg²⁺ rose from 1 to 10 mM. Additional supplementation with dithiothreitol or a univalent cation at 2.5 mM including K⁺ does not obviously increase their activities. These two enzymes were not stable at 4, −20 and −80°C and lost more activities at 4, −20°C than at −80°C for 24 h (Supplementary Fig. S4).

Since we aim to identify the AAEs responsible for the formation of 4MVCoA and isovaleryl-CoA, the predicted CoA ester precursors of pogostone and its analogue pogostone B, respectively, the enzymatic products of PcAAE1 with isovaleric acid as substrate and PcAAE2 with 4-methylvaleric acid and isovaleric acid as substrates were further verified with LC-QTOF-MS system (Fig. 4) and the kinetic properties of PcAAE1 with isovaleric acid as substrate and PcAAE2 with 4-methylvaleric acid and isovaleric acid as substrates were further determined (Table 3). The kinetic analysis revealed that PcAAE1 has a K_m value of 12.9 ± 0.8 µM and a catalytic efficiency of 3.33 × 10⁵ s⁻¹M⁻¹ for isovaleric acid, and PcAAE2 has a K_m value of 43.4 ± 5.1 µM and catalytic efficiency of 5.19 × 10⁴ s⁻¹M⁻¹ for 4-methylvaleric acid, while the K_m value and catalytic efficiency for isovaleric acid are 216.9 ± 13.1 µM and 0.78 × 10⁴ s⁻¹M⁻¹, respectively (Table 3).

Considering the previously reported substrate promiscuity of AAEs such as CsAAE1, Ph4CL1 and HcAAE1 (Stout et al. 2012, Klempien et al. 2012, Singh et al. 2020), we further determined the kinetic properties of PcAAE1 and PcAAE2 with substrates toward which PcAAE1 and PcAAE2 show relative activities higher than 40% in Table 2. The kinetic analysis further revealed that PcAAE2 also has low K_m values with the substrates valeric acid, hexanoic acid and heptanoic acid besides 4-methylvaleric acid (Table 3). Although the K_m value of PcAAE2 with hexanoic acid is about the same as that of PcAAE2 with 4-methylvaleric acid, the catalytic efficiency of PcAAE2 with hexanoic acid is only 44% of that of PcAAE2 with 4-methylvaleric acid (Table 3).

Subcellular localization of PcAAE1 and PcAAE2

Subcellular localization prediction software TargetP 2.0 and PTS1 predictor were utilized for signal peptide prediction of PcAAE1∼8 (Table 1). PcAAE1 is predicted to harbor a mitochondrial transit peptide with a high score, while no signal peptide was found for PcAAE2, suggesting that PcAAE1 may target mitochondria while PcAAE2 is probably localized in cytoplasm. Since 4-methylvaleric acid and isovaleric acid are the most preferred substrates of PcAAE2 and PcAAE1, and the corresponding acyl-CoA ester products are precursors of pogostone and pogostone B, respectively, the subcellular localizations of these two proteins were further analyzed experimentally. This was done by transiently expressing fluorescent protein fusions of the first 30 amino acids (predicted mitochondrial transit peptide) of PcAAE1, whole PcAAE1 and PcAAE2, respectively, in Arabidopsis leaf protoplasts. When the first 30 amino acids of PcAAE1 and whole PcAAE2 were expressed as C-terminal green fluorescent protein (GFP) fusions (SP_PcAAE1-GFP and PcAAE2-GFP, respectively), complete co-localization of SP_PcAAE1-GFP signal with MitoTracker Red, a marker dye specific for mitochondria, was observed, indicating that PcAAE1 is a mitochondria-targeted protein (Fig. 5A). The GFP signal of PcAAE2-GFP was detected exclusively in the cytoplasm and no GFP signal could be found to overlay with the red chloroplast autofluorescence (Fig. 5B). When PcAAE1 and PcAAE2 were expressed as N-terminal GFP fusions (GFP-PcAAE1 and GFP-PcAAE2, respectively), partial co-localization of GFP-PcAAE1 signal with MitoTracker Red was observed (Fig. 5A), suggesting that N-terminal GFP partially blocks the mitochondria signal of PcAAE1. The GFP signal of GFP-PcAAE2 was the same as that observed for The GFP signal of PcAAE2-GFP, indicating that PcAAE2 is a cytosol protein (Fig. 5B).

Phylogenic relationships of PcAAE2 homologs

To gain further insight into the evolution of PcAAE2, we expanded the phylogenetic analysis of AAE clade VI to include 77 homologs from chlorophytes including Chlamydomonas reinhardtii and Volvox carteri, Embryophyte, Tracheophyte and angiosperms. The phylogenetic tree was constructed using a maximum likelihood algorithm (Fig. 6). The proteins in the expanded clade VI were mainly grouped into four distinct subclades (Fig. 6). We designated the group that contains PcAAE1, PcAAE2 and PcAAE7 as subclade VIa, the group that contained PcAAE3 as subclade VIb, the group that included PcAAE5, PcAAE6 and PcAAE8 as subclade VIc, and the group that included PcAAE4 as subclade VId (Fig. 6). Six homologs of the AAE clade VI from Chlamydomonas reinhardtii, Volvox carteri, Physcomitrella patens, Selaginella moellendorffii grouped out of the designated four subclades, suggesting an earlier evolutionary split.

It is interesting that most of clade VI AAEs possess canonical PTS1 signals and are predicted to be targeted to the peroxisome by PTS1 predictor (Fig. 6). We also further investigated subcellular localization prediction of other selected genes using Targetp 2.0. 3 out of the 77 clade VI AAEs contain mitochondrial transit peptides at the N-terminus and are predicted to be located in mitochondria, while 15 out of these 77 AAEs contain no obvious signal peptides and are predicted to be located in cytoplasm (Fig. 6). In addition, all the six outgroups contain prototypical PTS1 signals with high prediction score, suggesting likely peroxisomal localization. PcAAE1 and PcAAE2 are grouped closely with HcAAE1, HlCCL2 and At2G17650 in the subclade via (Fig. 6), which are consistent with the fact that they share very similar substrate preference toward branched/straight short-chain fatty acids.

DISCUSSION

PcAAE2 catalyzes the formation of 4MVCoA precursor of pogostone in P. cablin

The medicinal plant P. cablin is an important commercial source of patchouli oil (Yang et al. 2016, van Beek and Joulain 2018). Patchouli oil has the ability to blend with other essential oils and give a strong base, lasting character, fixative properties and aids in preventing evaporation, thus promoting tenacity; hence, it is widely used in the manufacturing of perfume, cosmetics, soaps, scents body lotions and detergents (Gubareva 2004, Kiso et al. 2004, Swamy and Sinniah 2015, Hu et al. 2017). Previous studies have indicated that pogostone, one of the major chemical constituent of patchouli oil, is largely responsible for its intense aromatic odor, and more recently, this compound has been proved to exert many pharmaceutical activities (Li et al. 2012, 2014, Yi et al. 2013, Su et al. 2015, 2017, Chen et al. 2016, Wang et al. 2016, Cao et al. 2017). At present, pogostone is mainly isolated from patchouli oil, which is usually produced by steam distillation of the shade dried leaves and stems (Swamy and Sinniah 2015).

In patchouli oil, besides pogostone, a previous study also detected its different analogues including pogostone B in much smaller concentrations (Nakahara et al. 1975, Rijke et al. 1978, van Beek and Joulain 2018), which is similar to that was observed in the present study. Pogostone B, albeit at a much lower level, showed very similar accumulation pattern to that of pogostone in tested P. cablin tissues (Fig. 2B). The distribution of both pogostone and pogostone B in P. cablin top leaves newly generated from main stems shows a developmental pattern (Fig. 2). To begin elucidating the biosynthetic pathway of pogostone, we searched for the enzyme that catalyzes the formation of 4MVCoA, one of the precursors of pogostone. Transcriptome profiling and searches of RNAseq databases obtained from three different stages of developing P. cablin plants identified 31 AAE unigenes (Supplementary Table S1), with one unigene, named PcAAE2, showing the strongest positive correlation with the distribution pattern of pogostone (Supplementary Table S1; Figs. 2B and 3).

The purified recombinant PcAAE2 protein is able to effectively catalyze the formation of 4MVCoA in vitro from 4-methylvaleric acid, with a K_m value of 43.4 µM and a catalytic efficiency of 5.19 × 10⁴ s⁻¹M⁻¹ for 4-methylvaleric acid (Table 3; Fig. 4). In addition to 4-methylvaleric acid, PcAAE2 is also able to effectively catalyze the formation of CoA esters in vitro from valeric acid and hexanoic acid, with K_m values of 53.9 and 41.5 µM, respectively (Table 3), thus indicating substrate promiscuity, which has been commonly observed for AAEs (Stout et al. 2012, Klempien et al. 2012, Xu et al. 2013, Singh et al. 2020). Despite the observation of substrate promiscuity, PcAAE2 does show the highest catalytic efficiency toward 4-methylvaleric acid among the tested fatty acid substrates (Table 3), suggesting it as the 4MVCoA synthetase in vivo. In addition, PcAAE2 is also able to catalyze the formation of isovaleryl-CoA in vitro from isovaleric acid with a K_m value of 216.9 µM and a catalytic efficiency of 0.78 × 10⁴ s⁻¹M⁻¹, which accounts for about 15% of that with 4-methylvaleric acid (Table 3). Therefore, given the much lower level of pogostone B compared with that of pogostone in tested P. cablin tissues (Fig. 2), PcAAE2 may also supply isovaleryl-CoA precursor for pogostone B biosynthesis in cytoplasm, where the polyketide pogostone and pogostone B are believed to be synthesized.

Besides PcAAE2, PcAAE1, another member of clade VI AAEs from P. cablin, also shows substrate preference toward branched short-chain fatty acid (Table 2). The purified recombinant PcAAE1 protein accepts isovaleric acid as the best-preferred substrate and is able to effectively catalyze the formation of isovaleryl-CoA in vitro from isovaleric acid, with a K_m value of 12.9 µM and a catalytic efficiency of 3.33 × 10⁵ s⁻¹M⁻¹ for isovaleric acid (Table 3; Fig. 4). Although the expression pattern of PcAAE1 did not show an obvious positive correlation with the accumulation pattern of pogostone B in the tested P. cablin tissues (Supplementary Table S1; Figs. 2 and 3), the possibility of involvement of PcAAE1 in pogostone B biosynthesis cannot be completely ruled out. Subcellular localization assays indicated that PcAAE1 is a mitochondria-targeted protein (Fig. 5A). Therefore, it is mostly likely that PcAAE1 maintains the isovaleryl-CoA pool in mitochondria under certain conditions and plays a role in a dynamic balance between leucine degradation and the electron transport chain of mitochondria (Araujo et al. 2010).

Cytosolic PcAAE2 may have evolved from an ancestral peroxisomal AAE

The AAE gene family expanded and diversified as land plants evolved from their smaller, simpler algal ancestors (Shockey and Browse 2011). Each of the AAE gene family of Chlamydomonas reinhardtii and Volvox carteri, two unicellular green algae, contains one gene (Cr03g211745 for Chlamydomonas reinhardtii and Vc0013s0189 for Volvox carteri) only distally related to PcAAE2, and this gene was grouped out of the AAE clade VI subfamily in our phylogenetic tree (Fig. 6), suggesting that this clade VI is specific to multicellular land plants as previously reported (Shockey and Browse 2011).

Interestingly, most of clade VI AAEs in this phylogenetic analysis (46 out of 71) possess a predicted peroxisomal targeting signal 1 (PTS1) and are therefore predicted to be targeted to the peroxisomes (Fig. 6), while three clade VI AAEs possess predicted mitochondrial transit peptides and are predicted to be located in the mitochondria (Fig. 6). Fifteen clade VI AAEs possess no obvious predicted signal peptides and are predicted to be targeted to cytoplasm (Fig. 6). Each of PcAAE2’s closest homologs (Cr03g211745, Vc0013s0189, Pp3c25280, Pp3c96840, Sm409095, Sm440236, Sm236556 and Sm81559) from Chlamydomonas reinhardtii (a unicellular green algae), Volvox carteri (unicellular green algae), Physcomitrella patens (a moss) and Selaginella moellendorffii (an ancient vascular plant) contains a prototypical PTS1 sequence at the C-terminus with high prediction score (Fig. 6) and is therefore highly likely to be located in peroxisome where the catabolic process of β-oxidation occurs. These ancestor peroxisomal PcAAE2 homologs may play important roles in β-oxidation. PcAAE2 lacks a PTS1 and locates in cytoplasm where the polyketide pogostone is believed to be synthesized (Fig. 5B).

Since the possession of a PTS1 sequence is the ancestral trait in the AAE VI clade throughout the plant kingdom, and a PTS1 sequence is also present in the P. cablin VIa clade (in PcAAE7, a protein that has 65% amino acid identity to PcAAE2), the recruitment of PcAAE2 for pogostone biosynthesis appears to have arisen through a duplication event of an ancestral peroxisomal protein and subsequent divergence in both subcellular localization and substrate specificity. Similar cases for a cytoplasm localized hexanoyl-CoA synthetase (CsAAE1) recruited to cannabinoid biosynthesis in Cannabis sativa trichomes and for two cytoplasm localized branched short-chain CoA ligases (HlCCL2 and HlCCL4) recruited to bitter acid biosynthesis in H.lupulus trichomes and for recently identified cytoplasm and peroxisome dually localized benzoyl-CoA ligase (HcAAE1) recruited to xanthone biosynthesis in H.calycinum have been already reported (Stout et al. 2012, Xu et al. 2013, Singh et al. 2020). A similar case may also have occurred for PcAAE1, which encodes a mitochondrial isovaleryl-CoA synthetase (Table 3; Fig. 5A), probably playing a role in a dynamic balance between leucine degradation and the electron transport chain of mitochondria through maintaining the pool of isovaleryl-CoA in mitochondria (Araujo et al. 2010). This conclusion could be confirmed by further experimental subcellular localization analysis of more clade VI genes from different plant species.

Assessment of P. cablin PcAAE1 and PcAAE2 sequence properties

AAEs belonging to the acyl-CoA synthetase subfamily are members of the so-called ANL superfamily, in which some recognized conserved signature motifs (A1-A10), common to all adenylation domains have been identified (Marahiel et al. 1997, Gulick 2009) and are also present in P. cablin PcAAE1 and PcAAE2 (Supplementary Fig. S3). AAEs exhibit similar overall folding patterns with large N-terminal and small C-terminal domains. Biochemical and structural studies demonstrate that AAE enzymes undergo large-scale domain movement to facilitate the two-step catalytic reaction, an adenylate-forming conformation during the first half-reaction and a thioester-forming conformation during the second step (Bar-Tana and Rose 1968, Gulick et al. 2003, Reger et al. 2008, Gulick 2009, Li and Nair 2015). The hinge residue that facilitates this movement is usually aspartate in the A8 motif (Gulick et al. 2003), which is conserved in P. cablin PcAAE1 and PcAAE2 (Supplementary Fig. S3). As in all adenylate forming enzyme, two conserved AMP-binding motifs (AMP1 and AMP2) (Weimar et al. 2002, Morgan-Kiss and Cronan 2004) and the catalytically relevant amino acid lysine (Reger et al. 2007, Hu et al. 2010, Li and Nair 2015) are also present in P. cablin PcAAE1 and PcAAE2 (Supplementary Fig. S3).

In addition, PcAAE1∼3 from P. cablin and some other medium/short-chain acyl-CoA synthetases from A.thaliana, C.sativa, H.lupulus, H.calycinum, P.cablin, Taxus media have been shown to be active with a range of medium/short-chain fatty acids (Stout et al. 2012, Xu et al. 2013, Srividya et al. 2020, Singh et al. 2020). Phylogenetic analysis indicated that they fall into two clades (Fig. 7A). Clade I, which contains acyl-CoA synthetases preferring branched/straight short-chain fatty acids, is clearly distinct from clade II, which contains CsAAE3, TmAAE13, TmAAE15 and TmAAE16, all of which show high activities toward straight medium-chain fatty acids. Previous analysis of bacterial AAEs indicated an fatty acyl-CoA synthetase (FACS) signature motif appearing to contribute to the fatty acid-binding pocket and thus affects the preference for fatty acids of different chain lengths (Black et al. 1997, 2000). Sequence divergence across their FACS signature motifs indicated that they fall into two groups that are consistent with the result of phylogenetic analysis (Fig. 7B), suggesting that this motif is an important determining factor for their functional differentiation and evolutionary split. Phe⁴, Val¹¹ and Lys²¹ in the 25 amino acids of the FACS signature motif are specifically conserved in the first group, while Tyr¹¹, Asp¹³ and Val²¹ are specifically conserved in the latter group (Fig. 7B), suggesting that these amino acids may play a vital role in fatty acid chain length specificity. To verify this hypothesis, in silico homolog model based on the available crystal structures of AAEs such as Pt4CL1 and Nt4CL2 (Hu et al. 2010, Li and Nair 2015) should be required. Further extensive comparisons on the structures and gain-of-function mutations in this motif are needed for a deeper understanding of the substrate preferences for branch/straight chain and chain length of these enzymes.

Metabolic engineering of pogostone biosynthesis

P.cablin (Lamiaceae) is a commercially important plant for its essential oil (patchouli oil). It is native to southeast Asia and is now cultivated widely in many tropical and subtropical regions (Wu et al. 2010, Xie et al. 2017), including China, Indonesia, the Philippines and Thailand for the extraction of essential oils from its leaves and stems, and especially as a natural source of patchouli alcohol and pogostone (Singh and Rao 2009, He et al. 2016, Swamy and Sinniah 2016). Different P. cablin cultivars exhibit significant differences in quality and bioactive components, as these factors are influenced by climate, soil nutrients and water in the different locations (Swamy et al. 2015). Based on the main components of essential oils, P. cablin in China can be divided into two chemotypes: pogostone type, with a high content of pogostone and low content of patchouli alcohol, and patchoulol type, which has a high content of patchouli alcohol but a low content of pogostone (Liu et al. 2002, Luo et al. 2003, Huang et al. 2016). Traditionally, the patchoulol-type cultivars are mainly used in the perfume industry, whereas the pogostone-type cultivars are considered as medicinal plants in China. However, the production of pogostone, likely the main effective compound in medicinal use, is extremely low due to the limitation of cultivation area (Hu et al. 2006, Wu et al. 2010). Due to the discovery of more and more pharmacological activities of pogostone, the pogostone-type cultivars are highly popular (Chen et al. 2021). However, their natural supplies cannot sufficiently satisfy the economic demand (Zhang et al. 2019).

Our discovery of PcAAE2, the enzyme that catalyzes the biosynthesis of 4MVCoA, will serve as a step forward in the construction of high-yield pogostone P. cablin varieties. Increasing PcAAE2 transcript levels or changing its developmental expression pattern could lead to further increase in the production of pogostone. Pogostone has received increased attention along with the discovery of its additional potential bioactivities and pharmacological properties such as antibacterial, anti-inflammatory, anticancer activities and others (Chen et al. 2015, 2016, Sun et al. 2016, Wang et al. 2016, Cao et al. 2017). The discovery of 4-MVCoA synthetase could also facilitate the functional characterization of other enzymes such as polyketide synthases involved in the pogostone biosynthesis and, thus, will allow for the reconstruction of this pathway in the heterologous system using a synthetic biology approach to produce pogostone in large amounts.

Materials and Methods

Plant materials and chemicals

P.cablin seeds were germinated in soil, and plants were grown in a greenhouse with a 16 h light/8 h darkness photoperiod at 30℃. The various tissues samples including seedling, roots, main stems and leaves were collected from P.cablin plants. For RT-qPCR analysis, the harvested tissues were flash frozen in liquid nitrogen and stored at −80°C until use. Arabidopsis plants were grown in soil in a growth chamber with the photoperiod (12 h light/12 h dark at 23°C) under 60% humidity. The well-expanded leaves from 4-week-old Arabidopsis plant were selected for protoplast isolation for subcellular localization.

Pogostone and pogostone B standard were synthesized as described previously (Yu et al. 2018). Since standard 4MVCoA is not available commercially, the CoA esters produced by PcAAE2 from 4-methylvaleric acid were identified as 4MVCoA based on the exact mass and fragment pattern by LC-QTOF-MS. MitoTracker Red used for mitochondria staining was purchased from ThermoFisher, and other chemicals were purchased from Sigma-Aldrich.

GC-MS analysis of pogostone and pogostone B from P. cablin seedling, roots, main stems and leaves from the main stem at different stages of development

P. cablin seedlings, roots, main stems and leaf tissues at different stages of development were collected and cut into small pieces, 100 mg of which were transferred into a tube containing 500 µL MTBE with 0.05 ng/µl tetradecane as internal standard. The tube was vortexed for 3 min at maximum speed, incubated at room temperature for 1 h and then centrifuged for 3 min at the speed of 10,000 g/min. The MTBE phase was collected and analyzed by GC-MS. For GC-MS analysis, 1 µl aliquot of the sample was injected into Agilent GC-MSD (Agilent 7890B-5977B) system equipped with the DB-5MS column (30 m × 0.25 mm × 0.25 µm film thickness, Agilent, USA). The oven temperature was programmed as follows: initial temperature, 50℃ for 3 min, 50 to 320℃ at a rate of 10℃ min⁻¹ and hold for 5 min. The identification of compounds was assigned by comparison of their retention times and mass fragmentations with corresponding authentic standards. The measurement of pogostone and pogostone analogue was performed by comparison to the corresponding authentic standard curves.

RNAseq analysis

Total RNA was extracted from seedling, root, main stem and leaf parts of P. cablin plants at different developmental stages using Total RNA Isolation Kit from Omega. The RNAseq library construction and transcriptome assembling were performed as described previously (Xu et al. 2018). The expression levels of all transcripts were estimated using the Trinity software, which estimated transcript abundance using the RSEM function (Li and Dewey 2011).

Identifying AAE genes in the P. cablin transcriptome

To find candidate unigenes encoding AAEs in our P. cablin transcriptome database, we constructed local Nucleotide Database using the software (BioEdit Sequence Alignment Editor, https://bioedit.software.informer.com/7.2/) and queried it with Arabidopsis representative AAEs from each of seven clades of AAE superfamily using the TBLASTN function. We also screened the entire annotated database for the functional annotations ‘AMP binding protein’ and ‘CoA ligase’ and added transcripts identified in this way to our list of candidates.

Quantitative reverse transcription polymerase chain reaction analysis of PcAAE1∼8 transcript Levels

For quantitative real-time PCR analysis of transcripts in different tissues, total RNA was isolated using total RNA Isolation Kit from Omega with a DNA digestion step. RNA concentration and quality (260/280 and 260/230 ratios) were determined using the Thermo Scientific™ NanoDrop™ Lite Spectrophotometer. Reverse transcription polymerase chain reaction (RT-PCR) was prepared using High Capacity cDNA Reverse Transcription Kit from Thermo Fisher Scientific following the manufacturer’s instructions with minor modification. 3 μg total RNA was used for making cDNAs in a 20 μl RT-PCR reaction system containing 5 μM oligo(dT), which was obtained from Thermo Fisher Scientific (cat number; #SO132). RT-qPCR analysis was performed using Power SYBR Green PCR master mix (Thermo Fisher) using Bio-Rad CFX Connect Real-Time PCR Detection System. Each reaction consisted of 10 μl Power SYBR Green PCR master mix, 2 μl cDNA and 1 μl of 5 μM forward and reverse primers in a total volume of 20 μl made up with sterile water. The thermal profile included, after a 2 min step at 95°C, 40 cycles of denaturation at 95°C for 5 s and annealing at 55°C for 30 s and amplification for 30 s. Melting curves were recorded by melting the amplified amplicon at 65–95°C in an increment increase of 0.5°C per cycle. P. cablin glyceraldehyde 3-phosphate dehydrogenase (PcGAPDH), PcActin7 and PcTubulin3 were used as internal reference genes. For PcAAE1∼8 and the three reference genes, single melting curve peaks were obtained. Primer specificities were evaluated by melt curve analysis and agarose gel electrophoresis. cDNAs of four biological replicates were analyzed. No‐template (NTC) and no‐reverse transcriptase controls (NRC) were used to rule out primer dimer formation and genomic DNA contamination, respectively.

The RT-qPCR data were expressed as relative expressions of PcAAE1∼8 relative to the three reference genes in each tissue. The relative expressions of PcAAE1∼8 in each tissue were calculated using 2^dCq with the dCq calculated by subtracting Cq values of PcAAE genes from the geometric mean Cq values of the three reference genes in every tissue and the amplification efficiency (E) set as 2. The equation used in this process is defined by just one group of Cq values containing Cq for the gene of interest and geometric mean Cq for the three reference genes in tested tissues (Vandesompele et al. 2002). Stable expression of the three reference genes among all selected samples was confirmed before integrating their Cq values in normalization. Primers used in this study are designed using Primer 5.0 software and listed in Supplementary Table S4. Assay of unbiased RNA coverage of oligo (dT) and random primer for PcAAE1∼8 in the top leaf and main stem tissues of 6-week-old P. cablin plants was performed and the relevant results were supplied as Supplementary Fig. S5.

Isolation of His-tagged recombinant proteins of PcAAE1∼8

The open reading frames of PcAAE2∼8 were obtained by RT-PCR from prepared cDNA of P. cablin seedlings and were introduced into the expression vector pET-28a(+) between the restriction sites of NdeI and BamHI using NEBuilder® HiFi DNA Assembly Cloning Kit (NEB, Catalog number: E5520S), in each case generating a fusion gene that encoded a ‘tag’ of His6 residues at the N-terminus for expression in Escherichia coli. To obtain soluble proteins for expression in E. coli, a truncated open reading frame of PcAAE1, missing the first 30 codons, was obtained by RT-PCR from prepared cDNA of P. cablin seedlings and was introduced into the expression vector pET-28a(+) to generate a fusion gene that encoded a ‘tag’ of His6 residues at the N-terminus for expression in E. coli. Recombinant proteins of PcAAE1∼8 were expressed in E. coli strain BL21(BE3) and purified using Ni-NTA agarose chromatography (Qiagen) as previously described (Xu et al. 2018).

Enzymatic assays of recombinant PcAAE1∼8

Acyl-CoA synthetase activities of recombinant PcAAE1∼8 proteins were measured using the coupled enzymes assay (Koo et al. 2006, Chen et al. 2011). For substrate specificity assays, reaction mixtures (250 µl) contain 0.1 M Tris-HCl, pH 7.5, 2 mM DTT, 5 mM ATP, 10 mM MgCl₂, 0.5 mM CoA, 0.4 mM NADH, 0.5 mM fatty acid substrate, 1 mM phosphoenolpyruvate and four units each of myokinase, pyruvate kinase, and lactate dehydrogenase. The reaction was initiated by adding 1 μg of purified enzyme. Oxidation of NADH was monitored at 30°C for 15 mins at 340 nm (ε_{340 nm} = 6220 M^–1 cm^–1) with a Beckman DU530 spectrophotometer. The enzyme concentration (1 μg) and incubation time (15 min) were selected to attain a linear reaction velocity during the assay period. To determine the kinetic parameters of recombinant PcAAE2, a similar protocol was followed. The K_m value for fatty acid substrate was determined by fixing the concentration of CoA at 1 mM and ATP at 5 mM, whereas the K_m value for CoA was determined by fixing the concentration of fatty acid substrate at 1 mM and ATP at 5 mM, and the K_m value for ATP was determined by fixing the concentration of fatty acid substrate at 1 mM and CoA at 1 mM. K_m and k_cat values were calculated from initial rate data by using the hyperbolic regression analysis method in Hyper32 software (version 1.0.0, http://hyper32.software.informer.com/).

In addition, Acyl-CoA synthetase activity of recombinant PcAAE1 and PcAAE2 proteins was further measured using an LC-QTOF-MS-based method. Briefly, the reaction mixtures (250 µl) contain 0.1 M Tris-HCl, pH 7.5, 2 mM DTT, 5 mM ATP, 10 mM MgCl₂, 0.5 mM CoA, 0.5 mM fatty acid substrate. The reaction was initiated by adding 10 μg of purified PcAAE1 or PcAAE2 protein. Assays were stopped after 30 min by the addition of 250 µl methanol. The solution was filtered through a 0.45 μm filter prior to analysis by LC-QTOF-MS system (Agilent, 6545 LC/QTOF-MS) coupled with a C₁₈ column (ZORBAX RRHD Plus C₁₈, Φ2.1 × 50 mm, 1.8 μm) at 35°C. The gradient (solvent A, water + 0.1% formic acid; solvent B, methanol) program was set as follows at a flow of 0.5 ml min^–1: 0–0.5 min, a linear gradient from 5% to 40% of B; 0.5–2.5 min, 40% of B; 2.5–3.5 min, a linear gradient to 100% of B; 3.5–4.5 min, 100% of B; 4.5–4.6 min, a linear gradient to 100% of A and equilibration for 2 min before starting the next sample. The operating parameters were set as follows: capillary voltage, 3,500 V; nebulizer pressure, 35 pounds per square inch gauge; drying gas flow rate, 8 L min⁻¹; gas temperature, 320°C; sheath gas temperature, 350°C; sheath gas flow, 11 L min⁻¹ and the fragmentor, 175 V. Mass spectrum were acquired in negative mode, and lock mass correction was performed using standard hexakis (2,2,3,3-tetrafluoropropoxy) phosphazine and purine. The compounds were analyzed with the parameters as follows: 766.1079 (m/z) for CoA, 850.1654 (m/z) for isovaleryl-CoA and 864.1811(m/z) for 4MVCoA.

Subcellular Localization of PcAAE1 and PcAAE2

The vector used for subcellular localization of analysis was pTF486. For the construction of these vectors expressing fusion proteins with GFP fused with C-terminus of the first 30 amino acids of PcAAE1 and whole PcAAE2, the open reading frames of the first 30 amino acids of PcAAE1 and whole PcAAE2 were amplified from P. cablin seedling cDNA, respectively, and inserted into pTF486 vector between restriction sites of SalI and BamHI using In-Fusion® HD Cloning Kit (TaKaRa, Catalog number: 638910) following the protocol. For the construction of these vectors expressing fusion proteins with GFP fused with N-terminus of PcAAE1 and PcAAE2, respectively, the open reading frames of PcAAE1 and PcAAE2 and gene encoding GFP were amplified from P. cablin seedling cDNA and pTF486 plasmid, respectively, the purified PCR product of either PcAAE1or PcAAE2 and gene encoding GFP and purified double enzyme digestion product of pTF486 plasmid between restriction sites of SalI and BamHI were assembled together using In-Fusion® HD Cloning Kit (TaKaRa) following the protocol. Protoplasts were prepared from Arabidopsis leaves, and transformation and confocal microscopy were performed as described previously (Yoo et al. 2007). Primers used in the construction of these vectors are listed in Supplementary Table S4 and the overlaps (15–30 bp) in the primer sequences that were used for DNA fragment assembly were underlined.

Supplementary Data

Supplementary data are available at PCP online.

Data Availability

Data supporting the finding of this study are available within the article and its supplementary files. The sequences of the genes used in this study are available under GenBank accessions as followsPcAAE1, MW413950; PcAAE2, MW413951; PcAAE3, MW413952; PcAAE4, MW413953; PcAAE5, MW413954; PcAAE6, MW413955; PcAAE7, MW413956; PcAAE8,MW413957; PcGAPDH,MW735976; PcActin7, MW735977; PcTubulin3, MW735978. The bioproject accession number for the RNAseq data is PRJNA713906.

Funding

This work was supported by the National Natural Sciences Foundation of China (Grant Nos. 31970319 to H.X.); Fundamental Research Funds for the Central Universities (Grant Nos. 2020CDJQY-A077 to H.X.); the National Key Research and Development Program (Grant Nos. 2018YFD1000800 to H.X.); 100 Talent Program (Chongqing University, Grant Nos. 0304001104433 to H.X.

Acknowledgments

We would like to thank Dr. Guocan Zhen at Analytical and Testing Center of Chongqing University for their assistance with LC-ESI/MS analysis.

Author Contributions

H.X. designed the experiments; J.C., L.L., Y.W. and Z.L. conducted the experiments, analyzed data or provided material; J.C., G.W., E.P. and H.X wrote the article; all authors edited the article.

Disclosures

The authors have no conflicts of interest to declare.

References