Engineering triterpene production in Saccharomyces cerevisiae–β-amyrin synthase from Artemisia annua

Abstract

Using a degenerate primer designed from triterpene synthase sequences, we have isolated a new gene from the medicinal plant Artemisia annua. The predicted protein is highly similar to β-amyrin synthases (EC 5.4.99.–), sharing amino acid sequence identities of up to 86%. Expression of the gene, designated AaBAS, in Saccharomyces cerevisiae, followed by GC/MS analysis, confirmed the encoded enzyme as a β-amyrin synthase. Through engineering the sterol pathway in S. cerevisiae, we explore strategies for increasing triterpene production, using AaBAS as a test case. By manipulation of two key enzymes in the pathway, 3-hydroxy-3-methylglutaryl-CoA reductase and lanosterol synthase, we have improved β-amyrin production by 50%, achieving levels of 6 mg·L⁻¹ culture. As we have observed a 12-fold increase in squalene levels, it appears that this strain has the capacity for even higher β-amyrin production. Options for further engineering efforts are explored.

Abbreviation

HMGR

HMG-CoA reductase

Triterpenes belong to the isoprenoid family of compounds and are recognized by their C₃₀ backbones. They are typically synthesized by the cyclization of the sterol precursor 2,3-oxidosqualene into a multi-ringed compound with a single alcohol group. Fungi and mammals convert 2,3-oxidosqualene to the triterpene compound lanosterol in the biosynthetic pathways to ergosterol and cholesterol, respectively. The equivalent step in plant primary metabolism is the cyclization of 2,3-oxidosqualene to cycloartenol for the production of membrane sterols. Cycloartenol is also the triterpene precursor of brassinosteroid phytohormones that regulate plant growth and development [1,2]. Based on chemical and genetic analyses performed to date, it appears that plants are more diverse than animals or fungi in the range of tritepene products synthesized [3]. However, despite the fact that a large variety of triterpene compounds have been isolated from plant sources [4], the majority of triterpene synthase genes isolated to date have encoded either lupeol or β-amyrin synthases (EC 5.4.99.–) [1]. β-amyrin in particular serves as the olefin precursor to a wide range of downstream products. The action of oxidative enzymes (typically cytochrome P450 monooxygenases) and glycosyltransferases convert β-amyrin to various triterpene saponins in different plant species [5–7]. These saponins may perform protective roles in the host plant, acting as antimicrobial [8] and insecticidal [9] agents, and many of these compounds are also of interest from a human health perspective. The effect of plant saponins on low-density lipoprotrein cholesterol absorption and arterial atherosclerosis has received much attention, leading to the development of several cholesterol-reducing dietary supplements [10]. Saponins may also find applications in ruminant nutrition [11], as anticancer agents [12,13], and as vaccine adjuvants [14].

Although triterpene synthases have been expressed in microbial hosts such as Saccharomyces cerevisiae, there has been little effort made so far to engineer the metabolism of a microbial host for enhanced production of triterpenes. By contrast, there have been many considerable efforts to engineer microbes for higher production of mono-, sesqi- and diterpenes [15]. These projects have mainly focused on the overexpression of enzymes involved in either of the two pathways (mevalonate or 1-deoxy-d-xylulose-5-phosphate) responsible for the biosynthesis of isoprenoids [16–18]. In S. cerevisiae, the mevalonate pathway is responsible for the biosynthesis of isoprenoids and sterols. A good deal is known about regulatory mechanisms within the pathway, although the majority of studies have focused on the upper part of the pathway, from acetyl-CoA to squalene. Our knowledge of how the lower half of the pathway, from squalene to ergosterol, is regulated remains somewhat limited. As the branch point for triterpene biosynthesis is located in this latter half of the pathway, the optimal steps to increase their production in yeast are not immediately apparent.

Artemisia annua, or sweet wormwood, has been used medicinally for centuries, predominantly in China [19]. A sesquiterpene constituent, artemisinin, is one of the most important drugs used in the treatment of malaria. In an effort to isolate and characterize new terpene synthases from A. annua, we have designed degenerate primers for use in RT-PCR. Here, we describe the isolation of a β-amyrin synthase gene from A. annua and its expression in S. cerevisiae. Our findings on engineering overproduction of β-amyrin in S. cerevisiae should be relevant to the production of any triterpene.

Results

Isolation and verification of a β-amyrin synthase

In order to isolate new triterpene synthase genes from A. annua, several degenerate primers were designed from an alignment of plant triterpene synthase protein sequences. 3′ RACE reactions were carried out on RNA isolated from A. annua leaf tissue, and a product of the expected size was obtained with the primer TriF1 (Table 1). The fragment was cloned and the sequence was found to be homologous to triterpene synthase genes. Gene-specific primers were designed for 5′ RACE, and a product was obtained that contained the likely start codon, based on protein sequence alignments, with 175 nt of the upstream 5′ UTR sequence. The predicted full-length gene encodes a 762 amino acid protein that shares over 70% identity with plant β-amyrin synthases. The most closely related protein (AAX14716), sharing 86% sequence identity, is the β-amyrin synthase from Aster sedifolius, a plant which belongs to the asteroideae, the same sub-family as A. annua (Fig. 1).

Table 1.   Oligonucleotides used. The restriction sites used in cloning are underlined.

Oligonucleotide	Sequence (5′- to 3′)
TriF1	ATGYTNGCNTGYTGGRTNGARGAYCC
PolyT-anchor	GAGCTCGAGATCTAAGCTTGCTTTTTTTTTTTTTTTTTTTT
Anchor	GAGCTCGAGATCTAAGCTTGC
TriRaceR1	GATCCTGCTGGTTCCCATGCGGCTA
TricdsF	AACGAATTCAACAATGTGGAGATTGAAAATAGCAGAAGGGCGCAATG
TricdsR	AACGAGCTCCTAGGTGCCTTTGAGCTGTGGCAGCACCTGCTTG
ERG7F	TACCCATGGCAGAATTTTATTCTGACACAATCGGTC
ERG7R	CCATCGATCCATCAACCGGATGTGCTGTATTGACG

Figure 1.  Pairwise sequence alignment of AaBAS from A. annua (upper line) with its closest known relative, β-amyrin synthase from A. sedifolius (AAX14716, lower line). The position of the primer TriF1 is indicated by the arrow.

Published detection methods for triterpenes such as β-amyrin are generally laborious, and are inconvenient when processing a large number of samples [5,20]. Therefore, we attempted to streamline the process by eliminating the sample clean-up steps normally performed after cell extraction, or the need for derivatization. Cell disruption and saponification was performed as previously described, using a mixture of EtOH and KOH [20]. We found that using the nonpolar solvent dodecane for extraction allowed us to follow this directly with separation by GC/MS, using the highest possible temperature settings for the mass spectrometer ion source and quadrupole. This proved to be a sensitive and robust method for the detection of β-amyrin and the cell sterol components squalene and ergosterol.

The coding sequence of the gene, designated AaBAS, was cloned into the high-copy yeast expression vector pESC-URA, under control of the GAL10 promoter, and transformed into S. cerevisiae to create the strain βamy1. After induction of AaBAS expression with galactose, sterols were extracted from cells and analyzed by GC/MS. A single chromatographic peak was found in extracts from βamy1 cells that was absent in cells containing an empty vector. The retention time of this compound was identical to that of a β-amyrin standard, and the corresponding mass spectra were found to match (Fig. 2). An in vitro assay using a βamy1 cell extract and the triterpene substrate 2,3-oxidosqualene was also found to result in production of the β-amyrin product. The product was not observed in the absence of either the substrate or the AaBAS gene (data not shown).

Figure 2.  Confirmation of β-amyrin production in S. cerevisiae by expression of AaBAS. (A) Overlaid GC/MS chromatographs of extracts from strains BY4742 and βamy1 with an authentic β-amyrin standard (with β-amyrin structure shown) (B) Mass spectra extracted from the peaks shown in (A).

Production of β-amyrin in engineered strains of S. cerevisiae

We attempted to increase production of β-amyrin by modifying flux through the sterol biosynthetic pathway of S. cerevisiae. The enzyme 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR, represented by two isozymes: HMG1 and HMG2; Fig. 3) is known to act as a control point in the sterol pathway, with one of the primary control mechanisms being degradation of the HMGR protein in response to accumulation of the squalene precursor farnesyl pyrophosphate [21,22]. Expression of a truncated form of the HMG1 (tHMG1) protein circumvents this feedback, which occurs via the N-terminal transmembrane domain of the enzyme [23]. Furthermore, expression of tHMG1 from an independently-regulated promoter will bypass any transcriptional control of expression. Strain βamy2 contains an integrated copy of tHMG1 under control of the GAL1 promoter in addition to pESC-AaBAS. In agreement with previous studies [23,24], βamy2 accumulated significantly higher levels of squalene compared to βamy1 (Fig. 4A). However, this eight-fold increase in squalene did not translate into increased yields of β-amyrin; rather the β-amyrin levels from βamy2 were one-third of that produced by βamy1 (Fig. 4C). The ergosterol content of the cells remained essentially unchanged in strain βamy2 (Fig. 4B), which is consistent with the view that there is a feedback control mechanism in the pathway between squalene and ergosterol [23]. βamy2 grew extremely slowly at first (Fig. 4D), as observed previously [23], where it was attributed to accumulation of toxic intermediates such as farnesyl pyrophosphate.

Figure 3.  The yeast sterol pathway with the branch point for β-amyrin synthesis. Multiple steps are indicated by dashed lines.

Figure 4.  Determination of (A) squalene, (B) ergosterol, (C) β-amyrin, and (D) D₆₀₀ for strains βamy1 (BY4742, pESC-AaBAS), βamy2 (BY4742, pESC-AaBAS, P_GAL1-tHMG1), βamy3 (BY4742, pESC-AaBAS, P_MET3-ERG7), and βamy4 (BY4742, pESC-AaBAS, P_GAL1-tHMG1, P_MET3-ERG7 ). Error bars represent the SD for three independent cultures per strain.

In a separate approach, we tested whether downregulation of lanosterol synthase (ERG7; Fig. 3) would provide more 2,3-oxidosqualene substrate for AaBAS. Other studies have shown that triterpene production can be enhanced by deleting ERG7 completely [20]. However, as erg7 strains require feeding with ergosterol, this approach is economically limited for industrial purposes. To enable the downregulation of ERG7, we modified strain βamy1 by replacing the native ERG7 promoter with the methionine-repressible promoter of the MET3 gene [22]. Thus, the strain βamy3 contains a P_MET3-ERG7 replacement of the native ERG7 gene in addition to pESC-AaBAS. Following downregulation of ERG7, strain βamy3 was found to accumulate similar levels of squalene to strain βamy1, whereas β-amyrin levels were slightly higher (Fig. 4). It is interesting to note that ergosterol levels in the cell were not reduced in response to ERG7 limitation. Veen et al. [25] have shown that, when squalene epoxidase (ERG1) is overexpressed, there is no accumulation of 2,3-oxidosqualene, but rather of lanosterol, indicating that ERG7 is not a flux-limiting enzyme. In addition, it is likely that regulation of the pathway was adjusted in response to the reduced ERG7 transcript levels in order to maintain ergosterol production. Indeed, it was necessary to optimize the relative timing of AaBAS induction and ERG7 repression to even maintain the same β-amyrin production levels as strain βamy1. When induction and repression were simultaneous, β-amyrin production levels were actually lower in strain βamy3 and, thus, it was necessary to delay repression of ERG7 until 24–48 h after AaBAS induction in order to allow AaBAS to first accumulate (data not shown). The data shown in the present study were generated by inducing AaBAS at inoculation and repressing ERG7 43 h later by the addition of 1 mm methionine.

We next decided to combine the two strategies in order to test whether the feedback regulation that appears to take place in the tHMG1 strain βamy2 may be overcome by downregulating ERG7. Strain βamy4 was therefore constructed from strain βamy2 by replacing the native ERG7 promoter with the MET3 promoter. Again, it was found that optimal results were obtained when ERG7 was repressed by the addition of methionine 24–48 h after induction of AaBAS expression. βamy4 did not exhibit the same growth lag phase observed in βamy2, and grew at approximately the same rate as βamy1 (Fig. 4D). Interestingly, squalene accumulated in βamy4 to even greater levels than those found in βamy2, corresponding to a 12-fold increase over those in βamy1 (Fig. 4A). In this case, however, β-amyrin production levels were also significantly enhanced, resulting in a 50% increase in yield over βamy1 (Fig. 4C).

Discussion

We have shown that it is possible to engineer increased production of tritepenes in S. cerevisiae without the need for feeding with exogenous sterols. The 50% increase in β-amyrin levels demonstrated in the present study should be considered in the light of the fact that triterpene production may not be as amenable to engineering efforts as the volatile sesquiterpenes and monoterpenes that readily diffuse out of the cell. However, it is apparent that further progress can be made and there are some clues as to what these next steps may comprise. We have achieved a 12-fold increase in squalene levels over the initial βamy1 strain, and a logical course of action would be to find a way to convert this squalene into β-amyrin.

M’baya et al. [26] demonstrated that ERG1 activity is reduced in the presence of excess sterols through a mechanism other than enzyme inhibition, most likely transcriptional repression. Not a great deal is known about how the latter half of the sterol pathway is regulated and exactly what role ERG1 plays in this process. However, the fact that we observe a further increase in squalene upon downregulation of ERG7 in strain βamy4 would indicate that 2,3-oxidosqualene can act as a repressor of ERG1. Tight regulation at ERG1 would make sense as it marks the beginning of the oxygen-dependent reactions in the pathway. If the feedback regulation is transcriptional in nature, then it should be possible to circumvent it by overexpressing ERG1 under the control of an independent promoter. Veen et al. [25] overexpressed ERG1 and tHMG1 together and observed a 50% increase in sterol concentrations.

We have subsquently tested this hypothesis by transforming the strain βamy4 with a high-copy expression vector harboring ERG1 under control of the GAL1 promoter (to create strain βamy5). A comparison between these two strains showed that β-amyrin production levels were essentially unchanged whereas squalene levels actually increased slightly in strain βamy5 (data not shown). This would indicate that flux from squalene to β-amyrin is not limited by ERG1 transcription levels, but the possibility remains that there is regulation of ERG1 at the protein level. It also is likely that there are other factors contributing to the lack of flux from squalene to β-amyrin. In particular, the availability of squalene for conversion by ERG1 may be limited by its biochemical state. In cases where tHMG1 has been overexpressed in yeast, squalene accumulates predominantly in an insoluble form that is not immediately available to the sterol pathway [25,27]. The sterol-acyl transferases ARE1 and ARE2 are responsible for esterification of excess squalene for storage in insoluble lipid particles [27]. Thus, it appears that attenuation of this process would be the next logical step for further engineering triterpene production in S. cerevisiae. Additional studies into the possible regulation of ERG1 by 2,3-oxidosqualene at either post-translational or enzyme kinetic levels may also be warranted.

Experimental procedures

Isolation of a triterpene synthase gene from A. annua

Leaf tissue from A. annua was collected predominantly from new growth at branch tips and immediately frozen in liquid nitrogen. The tissue was ground to a fine powder using a cooled mortar and pestle, and RNA was purified by the Qiagen Plant RNeasy extraction method using RLC buffer as supplied (Qiagen, Valencia, CA, USA). RNA was quantified and checked for integrity using the Bioanalyzer 2100 (Agilent, Foster City, CA, USA). A single triterpene-specific primer (TriF1; Table 1) was designed from an alignment of various plant triterpene synthase protein sequences. For 3′ RACE, two primers were used in conjunction with TriF1: a poly(dT) primer with a 5′‘anchor’ sequence that has a melting temperature matching that of TriF1; and a primer composed of only the anchor sequence (Table 1). cDNA was synthesized from 3 μg of leaf total RNA using the polyT-anchor primer and Superscript II (Invitrogen, Carlsbad, CA, USA) with reverse transcription at 50 °C, followed by RNase H treatment. Touchdown PCR was carried out on the cDNA using the TriF1 and anchor primers by dropping the annealing temperature by 0.4 °C per cycle from 61 °C to 56 °C for the first eight cycles, followed by 30 cycles of amplification with an annealing temperature of 56 °C. A product within the expected size range (1.4 kb, based on the average length of a triterpene synthase gene and a plant 3′ UTR) was cloned into the TOPO TA vector (Invitrogen) and sequenced to reveal an ORF that appeared to encode a triterpene-synthase-like protein. The 5′ end of the cDNA was recovered using the GeneRacer kit (Invitrogen) according to the manufacturer’s guidelines, with the gene-specific primer TriRaceR1. The complete coding sequence of the candidate triterpene synthase gene was confirmed by cloning and sequencing three PCR products, amplified using PFU Turbo (Stratagene, La Jolla, CA, USA) from three independent cDNA preparations. The sequence has been submitted to Genbank and is available under accession number EU330197.

Expression of the candidate triterpene-synthase gene in S. cerevisiae

The full-length coding sequence of the candidate triterpene synthase gene was amplified using the primers TricdsF and TricdsR, which contain the 5′ restriction sites EcoRI and SacI, respectively. The sequence AACA was included immediately 5′ to the start codon in TricdsF to provide a favorable translation start context [28]. The PCR product was cloned into the EcoRI and SacI sites of the yeast expression vector pESC-URA (Stratagene) under control of the GAL10 promoter and ADH1 terminator to create pESC-AaBAS. Following sequence verification, the plasmid was transformed into S. cerevisiae BY4742 by the lithium acetate method [29], and transformants were selected on synthetic complete minus uracil agar (SC-URA). Synthetic complete media were made by adding 6.7 g·L⁻¹ Difco yeast nitrogen base (Becton, Dickinson & Co., Sparks, MI, USA) to a complete supplemental mixture (MB Biomedicals, Solon, OH, USA) of vitamins, minerals and amino acids, with the appropriate amino acid dropped out. Transformed yeast strains were maintained on SC medium with 2%d-glucose as carbon source, and induced by inoculating into SC + 2%d-galactose at D₆₀₀ of 0.02–0.03.

Modification of the sterol biosynthesis pathway in S. cerevisiae

A strain containing a soluble, truncated form of HMG-CoA reductase 1 (HMG1) under control of the GAL1 promoter was constructed by transformation of BY4742 with the integrating plasmid pδ-tHMGR and selection for loss of the URA3 selection marker using 5-fluoroorotic acid, as described previously [17].

The native promoter for the S. cerevisiae lanosterol synthase gene, ERG7, was replaced with the methionine-suppressible promoter P_MET3 to create the strain P_MET3-ERG7. The strategy used essentially follows that described by Gardner and Hampton [21]. Genomic DNA from strain BY4742 served as a template for amplification of the first 422 bp of the ERG7 cds using the primers ERG7F and ERG7R (Table 1). The amplified fragment was cloned into the NcoI and ClaI restriction sites of the vector pRS-ERG9 [17], thus replacing the ERG9 cds fragment with the ERG7 cds fragment, 3′ to the MET3 promoter. For integration into S. cerevisiae, the vector was digested at the unique BbvCI site in the ERG7 cds to facilitate homologous recombination with the native ERG7 gene. Upon transformation into S. cerevisiae, the successful promoter replacement was confirmed by PCR using a P_MET3 forward primer and an ERG7 reverse primer.

Extraction, identification, and quantitation of β-amyrin and sterols

A single method was developed to extract and quantify β-amyrin and the native yeast sterols squalene and ergosterol. Yeast culture (1 mL) in a microfuge tube was centrifuged for 1 min at 17 900 g to pellet cells. The cells were resuspended in 0.6 mL of a fresh solution of 20% (w/v) KOH in 50% ethanol containing 30 μg·mL⁻¹ cholesterol as an internal standard. The cells were boiled for 5 min in this solution in 2 mL screw-cap tubes. After cooling, the sterols and β-amyrin were extracted by vortexing with 0.6 mL dodecane (Sigma, St Louis, MO, USA) for 5 min at room temperature. The dodecane phase was transferred to a glass vial and directly subjected to GC/MS (GC model 6890, MS model 5973 inert, Agilent). An aliquot of the sample (1 μL) was injected into a DB5-MS column (Agilent) operating at a helium flow rate of 1 mL·min⁻¹. The oven temperature was held at 80 °C for 1 min after injection, and was then ramped to 280 °C at 20 °C·min⁻¹, held at 280 °C for 20 min, ramped to 300 °C at 20 °C·min⁻¹ and finally held at 300 °C for 2 min. The MS ion source was held at 300 °C throughout, with the quadrupole at 200 °C and the GC/MS transfer line at 280 °C. Full mass spectra were generated for metabolite identification by scanning within the m/z range of 40–440. For quantification of metabolites, samples were run in selected ion mode, detecting ions 203, 218, and 426. Standard curves for β-amyrin, squalene and ergosterol were run at the start and end of each batch of samples.