Production of 3-geranyl-4-hydroxybenzoate acid in yeast, an important intermediate of shikonin biosynthesis pathway

Abstract

Shikonin and its derivatives are the main active components in the medicinal plant Arnebia euchroma and possess extensive pharmaceutical properties. In this study, we developed an optimized yeast system to obtain high-level production of 3-geranyl-4-hydroxybenzoate acid (GBA), an important intermediate involved in shikonin biosynthesis pathway. For host selection, recombinant expression of p-hydroxybenzoate:geranyltransferase (PGT) derived from A. euchroma was performed in Saccharomyces cerevisiae WAT11U strain and high yield monoterpene strain. In shake flask culture with 1 mM p-hydroxybenzoate acid (PHBA), they could yield GBA at 0.1567 and 20.8624 mg L⁻¹, respectively. Additionally, AePGT6 showed higher enzymatic activity than its homologs. Moreover, by combining improvement in the homologous mevalonate pathway with reconstruction in the heterologous shikimic pathway, a de novo GBA synthesis pathway was constructed in StHP6tHC with co-overexpressed SctHMG1, optimized EcUbiC and AePGT6. A high titer of 179.29 mg L⁻¹ GBA was achieved in StHP6tHC under shake flask fermentation with 1 mM PHBA. These results suggest that yeast could be engineered systematically to enable an efficient monoterpene–quinone or naphthoquinone production.

Abbreviations

Abbreviations
 
• GBA:

  3-geranyl-4-hydroxybenzoate acid

 
• HMG1:

  hydroxy-3-methylglutaryl coenzyme A reductase 1

 
• UbiC:

  chorismate pyruvate-lyase

 
• PHBA:

  p-hydroxybenzoate acid

 
• PGT:

  PHBA geranyltransferase

 
• PCR:

  Polymerase Chain Reaction

 
• RF:

  Restriction Endonuclease Free

 
• HPLC:

  High Performance Liquid Chromatography

 
• UPLC:

  Ultra Performance Liquid Chromatography

INTRODUCTION

Shikonin and its derivatives are a class of red naphthoquinone pigments that are only synthesized in Boraginaceous plants (Papageorgiou et al.1999). These naphthoquinone pigments are the active principles of the widely used traditional Chinese medicine Zicao, the root of medicinal plant Arnebia euchroma (Wang et al.2014). Shikonin has been shown to possess diverse pharmaceutical activities, including antioxidant, anti-inflammatory, antithrombotic, antimicrobial and wound healing effects (Andujar et al.2013). Besides functioning as natural red naphthoquinone pigments, shikonins were also observed on the food, cosmetics and pharmaceutical additives list approved by the United Nations. Although shikonins can be achieved through cell or tissue culture, at present the supply is still mainly dependent on extraction from wild plant resources. In recent years, plant cell culture technology has been widely applied to produce shikonin, and several strategies have been employed to improve its yield (Boehm et al.2000; Wang et al.2014). However, there is currently no metabolic engineering study on shikonin production in yeast via the technology of synthetic biology.

It is generally agreed that shikonins are synthesized via the ubiquinone and other terpenoid–quinone biosynthesis pathway (KEGG PATHWAY: rn00130) in plants (Fig. 1). In the first unique step of shikonin biosynthesis, the condensation of p-hydroxybenzoate acid (PHBA) derived from the shikimate pathway and geranylpyrophosphate (GPP) from the cytosolic mevalonate (MVA) pathway or plastid 2-C-methyl-D-erythritol-4-phosphate pathway leads to the generation of an important intermediate 3-geranyl-4-hydroxybenzoate acid (GBA), which is catalyzed by PHBA geranyltransferase (PGT) (Yazaki, Sasaki and Tsurumaru 2009). The following biosynthetic steps were ambiguous and were supposed to involve hydroxylation, decarboxylation and cyclization reactions, which have the potential to be catalyzed by cytochrome P450s (Yamamoto et al.2000).

A number of microorganisms have been engineered to produce ubiquinone such as CoQ6, CoQ8 and CoQ10, which, like shikonin, are formed by the conjugation of a benzoquinone ring with a hydrophobic isoprenoid chain (Meganathan 2001; Kawamukai 2009). The length of polyprenyl side chain was determined by polyprenyl diphosphate synthase genes of different microorganisms (Okada et al.1998). Generally, both Escherichia coli and yeast could not release GPP for the biosynthesis of monoterpene products in nature (Reiling et al.2004; Oswald et al.2007); there are many problems to be solved for the production of monoterpene–quinone and naphthoquinone such as GBA and shikonin.

Unlike the soluble prenyltransferases from fungi and bacteria, aromatic prenyltransferases from plants are all membrane-bound proteins thus far (Yazaki, Sasaki and Tsurumaru 2009). The enzymes involved in shikonin biosynthesis pathway are mostly membrane-bound proteins, which are often difficult to express in bacteria. In contrast, yeast (Saccharomyces cerevisiae) is well suited as a cell factory platform for aromatic terpenoids production. On the other hand, CYP450s are widely involved in the process of plant secondary metabolism. With chromosomal integration of NADPH-cytochrome P450 reductase gene of Arabidopsis thaliana, the yeast strains WAT11U (Pompon et al.1996; Urban et al.1997) and high yield monoterpene strain (HMT1, a kind gift from Werck-Reichhart Daniele’ Lab) were both engineered to obtain plant secondary metabolisms.

Recently, we cloned three PHBA geranyltransferase genes (AePGTs) in RNA sequence database constructed from callus of A. euchroma (unpublished data). Unlike mitochondrial polyprenyltransferase for ubiquinone biosynthesis in yeast, AePGTs, like LePGTs (Yazaki et al.2002), are ER membrane-bound enzymes and could specifically recognize GPP (C-10) as the prenyl donor (unpublished data). In a previous study, GBA was enzymatically synthesized from PHBA and GPP using polyprenyldiphosphate: 4-hydroxybenzoate polyprenyltransferase (UbiA) from E. coli as GBA is non-commercial (Melzer and Heide 1994). In this study, we established a yeast strain for the production of GBA by recombining expression of PGT derived from A. euchroma and optimized the platform by combining improvement in the homologous mevalonate pathway with reconstruction in the heterologous shikimic pathway.

A major regulatory control point of the MVA pathway is the formation of MVA, which is catalyzed by HMG-CoA reductase (HMG) 1 and 2. In S. cerevisiae, HMG1 contributes at least 83% of the activity (Basson et al.1986). Early studies showed that additional overexpression of the catalytic domain of HMG1 (tHMG1) led to a 2.6-fold increase of miltiradiene content (Zhou et al.2012). PHBA is directly formed from chorismate in bacteria and catalyzed by chorismate pyruvate-lyase (UbiC), while it can be formed from either chorismate or tyrosine in yeast. But the homolog of UbiC in yeast is unknown, nor the enzymes responsible for PHBA biosynthesis via the coumarate pathway (Marbois et al.2010). A de novo synthesis pathway of GBA was constructed in the selected host strain HMT1 with co-overexpressed SctHMG1 from S. cerevisiae, optimized EcUbiC from E. coli and AePGT6 from A. euchroma. Also, we developed an optimized high-efficient yeast system to obtain GBA through the biotransformation of PHBA, an industrial chemical estimated with a cheap price of around US $3000/t (Kromer et al.2013). A high-level titer of 179.29 mg L⁻¹ GBA was achieved through the biotransformation of PHBA in shake flask fermentation. These results suggested that yeast could be engineered systematically to enable an efficient production of monoterpene–quinone or naphthoquinone and engineering geranyltransferase and the precursor supplying enzymes in both mevalonate pathway and shikimic pathway are essential for an efficient heterologous production of monoterpene–quinone in S. cerevisiae.

MATERIALS AND METHODS

Strains, plasmids and growth conditions

Strains and plasmids used in this study are listed in Table 1.

Cloning was carried out using Escherichia coli strain DH5α. Escherichia coli was grown at 37°C under shaking conditions. When necessary, growth media were supplemented with 100 μg mL⁻¹ ampicillin for E. coli culture.

All the fragments used for overexpression of genes were amplified by polymerase chain reaction (PCR) using primers and templates as described in Table 2.

The fragment encoding AePGT (GenBank accession number: DQ397513.2), AePGT4 (GenBank accession number: KT991522) and AePGT6 (GenBank accession number: KT991524) were amplified from cDNA of shikonin-sufficient suspension cell of Arnebia euchroma. The EcUbiC gene from E. coli was codon-optimized for high expression in Saccharomyces cerevisiae with GenScript OptimumGene technology and synthesized to make it possible to improve gene expression in yeast. SctHMG1, the catalytic domain of HMG1, was cloned from gDNA of WAT11U strain. The amplified products involved were cloned into the corresponding parental expression plasmids as described in Table 1 by restriction endonuclease free (RF) cloning method (Unger et al.2010). The clones with correct inserts were confirmed by sequencing. The list of the constructed vectors can be found in Table 1 and the details on the cloning are listed in Table 2.

Constructed yeast expression plasmids were transformed into the yeast strain WAT11U (Urban et al.1997) and HMT1, respectively, as listed in Table 1. Transformants were selected on relevant synthetic complete (SC) medium as well as drop-out media (SC-His, SC-Ura-His, SC-Ura-His-Leu) containing 20 g L⁻¹ glucose. The obtained recombinant strain was initially grown in drop-out liquid medium with 20 g L⁻¹ glucose at 30°C for about 48 h. Cells were centrifuged followed by washing with sterile water to remove any residual glucose. Cells were then resuspended in drop-out liquid medium with 20 g L⁻¹ galactose (PHBA was added according to situation) and grown for 24 h at 30°C to induce recombinant protein expression and 48 h to produce GBA. The culture broth (medium and cells together) was analyzed for GBA concentration using ultra performance liquid chromatography (UPLC).

In vitro AePGTs enzymatic activity assay

The culture broths grown in drop-out liquid medium with 20 g L⁻¹ galactose for 24 h at 30°C to induce recombinant protein expression were spun down and cells were used for the preparation of microsomes. All subsequent steps were carried out at 4°C. Cells were manually broken with glass beads (0.45 mm diameter) in a breaking buffer (50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 600 mM sorbitol). Microsomal fractions were prepared from yeast cells using a glycerol–NaCl precipitation procedure (Pompon et al.1996). Microsomal membranes were suspended in storage buffer containing 50 mM Tris HCl (pH 7.5), 1 mM EDTA and 20% (vol/vol) glycerol. In vitro AePGTs enzymatic activity assays were conducted according to previous study with slight modifications (Melzer and Heide 1994; Yazaki et al.2002). Briefly, the assay was performed in a total volume of 250 μL mixture, which consists of 50 mM Tris-HCl (pH 7.8), 200 nmol PHBA, 100 nmol GPP, 10 μmol MgCl₂ and 250 μg of microsomal protein. Reactions were incubated with shaking for 1 h at 30°C and then were terminated by extraction with an equal volume of ethyl acetate. The substrate (PHBA) and reaction product (GBA) was dried with N₂, dissolved in methanol and detected by Waters ACQUITY-UPLC-PDA system equipped with a Waters C18 1.8 μm 2.1 × 100 mm T3 HHS column at 254 nm absorbance. Samples were analyzed using an isocratic elution method with two solvents: 50% A-acetonitrile and 50% B-water (formic acid 0.1%) at 0.5 mL/min.

Quantification analysis of GBA in yeast

For sample preparation of GBA quantification, 5 mL culture broth (with medium and cells together uniformly) processed by galactose was extracted with equal volume ethyl acetate by sonication for 60 min for three times and dried with N₂. The residue was dissolved in methanol then subjected to Waters ACQUITY-UPLC-PDA system equipped with a Waters C18 1.8 μm 2.1 × 100 mm T3 HHS column at 254 nm absorbance. The mobile phase comprised A-acetonitrile and B-water (formic acid 0.1%) at 0.5 mL/min with a gradient program (0 min, 10% A; 4 min, 35% A; 4.3 min, 60% A; 8 min, 72% A; 8.5 min, 98% A; 10.5 min, 98% A; 11 min, 10% A; 0 min, 13% A). The substance GBA was extracted, isolated and purified in our laboratory. Their structures were confirmed by mass spectrometry and ¹H NMR, ¹³C NMR (Fig. S1–S3, Supporting Information). The purity was better than 98%, based on the percentage of total peak area determined by high performance liquid chromatography (HPLC) analysis. For all the strains, at least three biological replicates were analyzed. The bioconversion ratio of PHBA is calculated according to the ratio of yield of product (GBA, mol L⁻¹) and the amount of the substrate (PHBA, mol L⁻¹).

RESULT AND DISCUSSION

Yeast strain selection for production of GBA

The activity of PHBA geranyltransferase was examined by in vitro assays using microsomal fraction of recombinant yeast. AePGT was overexpressed in the Saccharomyces cerevisiae WAT11U strain. Microsomal fraction from the recombinant yeasts StWP and StWHis (vector control) were then assayed in the presence of PHBA, GPP and MgCl₂. UPLC–MS/MS analyses indicated that AePGT was able to transfer a GPP to PHBA C-3 and produce GBA in vitro (Fig. 2A).

Enzymatic function of recombinant AePGT in vivo was indicated by the GBA concentrations of yeast culture broth, which were fermented in SC-His medium with 20 g L⁻¹ galactose for 48 h (Fig. 2B). Results showed that no GBA was observed in the StWP and StWHis cultured in the medium without PHBA addition. Although 1 mM PHBA exists in the medium, only trace amounts of GBA (0.1567 ± 0.0491 mg L⁻¹) can be detected in StWP. This can be attributed to the fact that WAT11U yeast system is incapable of producing GBA under the condition lacking PHBA and GPP, even though PGT is overexpressed in the recombinant strain.

In contrast to the low price (around US $3000 t⁻¹) and toxicity of PHBA (the minimal inhibitory concentration of PHBA in yeast is 38.3 g L⁻¹ (Kromer et al.2013), the application of AePGT in yeast has been largely limited by the relatively expensive prenyl donor GPP. In order to produce GBA without feeding of GPP, AePGT was overexpressed in different yeast strains for the biotransformation of PHBA.

As wild-type yeast only carries a famesyl pyrophosphate synthase (FPPS, Erg20) rather than a specific GPP synthase (GPPS) as plants, FPP (C15) was synthesized through the intermediate GPP (C10) continuously. Consequently, GPP was not released from the catalytic site and was entirely converted to FPP (Oswald et al.2007). But the endogenous FPPS (Erg20) of the HMT1 was engineered into a GPPS by site mutation of K197G (Fischer et al.2011). At the same time, the HMT1 also have expressed the cytochrome P450 reductase gene from Arabidopsis thaliana as WAT11U. Thus, HMT1 was selected as the host for AePGT with the corresponding transgenic yeasts named StHP and StHHis.

As expected, the corresponding transgenic yeast StHP gives a dramatic increase in the production of GBA (20.86 ± 1.12 mg L⁻¹) in the medium with 1mM PHBA (Fig. 2B). Thus, we selected the HMT1 strain as the host of AePGT for the production of GBA.

Screening for PGT with higher enzymatic activity

The prenylation of PHBA, a key step linking the shikimate and mevalonate pathways, plays a critical role in the regulation of shikonin biosynthesis. Yazaki et al. (2002) have isolated two prenyltransferases in Lithospermum erythrorhizon designated LePGT-1 and LePGT-2, which were indicated by different geranyltransferase activity. In our study, we cloned six AePGT genes from callus of A. euchroma, among them, AePGT, AePGT4 and AePGT6 showed specific geranyltransferase activity in vitro (unpublished). Therefore, the three geranyltransferases were overexpressed in HMT1, separately. After 48 h of shake-flask fermentations in galactose induced medium with 1 mM PHBA, the corresponding strains StHP, StHP4 and StHP6 produced GBA at a titer of 20.86, 45.08 and 70.96 mg L⁻¹, respectively (Fig. 3). As AePGT6 showed the highest geranyltransferase activity in yeast, StHP6 was selected as parent strain for further engineering.

Coexpression of SctHMG1 and optimized EcUbiC led to higher GBA production in yeast

In order to produce GBA without feeding of PHBA in StHP6, we tried to construct a de novo GBA synthetic pathway in yeast. The modular pathway engineering strategy was applied for rapid assembling synthetic GBA pathway in yeast by combining improvement in the homologous mevalonate pathway with reconstruction in the heterologous shikimic pathway. To improve metabolic flux of MVA pathway, tHMG1 derived from S. cerevisiae was additionally overexpressed in the StHP6 background (StHP6tH). In order to overproduce PHBA in yeast, we overexpressed the UbiC gene from Escherichia coli constitutively (Kromer et al.2013). In addition, the UbiC gene from E. coli was codon optimized for S. cerevisiae with GenScript OptimumGene technology to make it suitable for expression in yeast and promising in production enhancement (Hoekema et al.1987; Trotta 2013).

As expected, overexpression of the optimized EcUbiC had a positive effect on the GBA production in the medium without PHBA as well as overexpression of SctHMG1. The strain individually overexpressing EcUbiC (StHP6C) and SctHMG1 (StHP6tH) produced 0.56 mg L⁻¹ and 0.54 mg L⁻¹ GBA, respectively; and the simultaneous overexpression of SctHMG1 and EcUbiC (StHP6tHC) increased the production of GBA to 0.98 mg L⁻¹ (Fig. 4A). The strategy here achieved an 11.27-fold increase in GBA yield from the 0.086 mg L⁻¹ observed with the base strain StHP6.

It is noteworthy that in the medium with 1 mM PHBA added, the improved strains leading to a high-level production of GBA (Fig. 4B). The strain StHP6tH with single overexpression of tHMG1 produced 171.36 mg L⁻¹ of GBA (Fig. 4B), which represents a 2.41-fold improvement in comparison to StHP6. And the strain StHP6C produced 171.14 mg L⁻¹ of GBA with the similar level of StHP6tH. The highest production (179.29 mg L⁻¹) was obtained in the strain with coexpression of SctHMG1 and EcUbiC. In this condition, the bioconversion ratio of PHBA reached 65.40% (Fig. 4C), most of PHBA in the medium transferred to GBA.

For GBA production through biotransformation of PHBA, the improvements described in this study (Fig. 5) achieve a 1093-fold increase in GBA yield from the 0.16 mg L⁻¹ observed with the strain StWP to 179.29 mg L⁻¹ with the strain StHP6tHC and an 8.22-fold increase from the 20.86 mg L⁻¹ observed with StHP. It is likely that we can expect to obtain a better yield of GBA and higher bioconversion ratio of PHBA in controlled fed-batch fermentation. To the best of our knowledge, this is already the highest titer of a compound derived from GPP in yeast. There are studies of monoterpene production in yeast achieved approximately 5 mg L⁻¹ geraniol from a haploid strain that contains a mutant form of Erg20p (K197G) (Fischer et al.2011) and 17.5 mg L⁻¹ sabinene by engineering monoterpene production in yeast using a synthetic dominant negative geranyl diphosphate synthase (Ignea et al.2014).

In order to enrich the precursor GBA for the production of shikonin in yeast through synthetic biology, it is necessary to optimize shikimate pathway as well as MVA pathway to ensure sufficient supplement of PHBA and GPP at the same time (Fig. 5). The engineered strain StHP6tH have shown sufficient ability for GPP supplement in medium with 1 mM PHBA. PHBA has ever been produced biotechnologically in recombinant strains of S. cerevisiae, by knocking out chorismate mutase and overexpressing UbiC gene from E. coli (Kromer et al.2013). But the accumulation of PHBA was still restricted by the shortage of endogenous chorismate, an important intermediate in shikimate pathway in yeast. Rodriguez et al. (2015) established a yeast platform strain for production of p-coumaric acid, which was derived from chorismate as well as PHBA. The highest titer of p-coumaric acid of 1.93 g L⁻¹ was obtained, by eliminating feedback inhibition of key enzymes, reducing by-product formation and overexpressing enzymes of flux-controlling steps. These works implied that there a great potential and prospect for PHBA production improvement in such a yeast-based process. Furthermore, these strategies of eliminating feedback-resistant enzymes, overexpressing the controlling enzymes of pathway flux and elimination of by-products accumulation may be beneficial to assemble a high efficient synthetic monoterpene–quinone pathway in yeast.

Prenylation is an important derivatization of plant aromatics, contributing to the chemical diversification of phenolic secondary metabolites in plants due to differences in prenylation positions, prenyl chain lengths and further modifications of prenyl chains. Recently, geranyl diphosphate-specific aromatic prenyltransferase was only identified in plants such as olivetolic acid geranyltransferase in Cannabis sativa (Gagne et al.2012), coumarin 8-geranyltransferase in Citrus limon (Munakata et al.2014) and PHBA geranyltransferase in Boraginaceous plants. The strategies in this will boost the research for the production of other geranylated compounds in yeast especially for the geranylated aromatic natural product in plant.

In our study, the engineered yeast StHP6tHC produced GBA up to 179 mg L⁻¹ through biotransformation of PHBA under flask-shake conditions. The engineered strain provided a platform for the functional identification of further steps involved in the shikonin biosynthesis pathway (Li et al.2015). Recently, several P450s genes have been cloned from the transcriptome database of Arnebia euchroma in our laboratory. Also, GBA, the proposed precursor for these reactions, has been synthesized in our engineered yeast. Taken together, all these progresses hold promise for the completion of shikonin biosynthesis pathway.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSYR online.

Acknowledgements

We thank Prof. Oliver Yu for the gift of yeast expression strain WAT11U and the expression vector, Prof. Werck-Reichhart Daniele for the HMT1 strain. We also thank AP Juan Guo, AP Zhubo Dai and Dr Xiaohui Ma for their kind assistance.

FUNDING

This work was supported by the National Natural Science Foundation of China (No. 81473307), Finance Ministry Projects (No. 2060302) and the Important National Science & Technology Specific Projects (2016YFF0202802).

Conflict of interest. None declared.

REFERENCES