|Home | About | Journals | Submit | Contact Us | Français|
Resveratrol synthesis from p-coumarate was analyzed in different Saccharomyces cerevisiae strains expressing the 4-coumaroyl-coenzyme A ligase (4CL1) from Arabidopsis thaliana and the stilbene synthase (STS) from Vitis vinifera and compared between yeast cultures growing in rich or synthetic medium. The use of rich medium considerably improved resveratrol production, and resveratrol yields of up to 391 mg/liter could be achieved with an industrial Brazilian sugar cane-fermenting yeast.
Resveratrol (3,5,4′-trihydroxy-trans-stilbene) is a polyphenolic compound produced by some plants in response to infections or environmental stresses. As an ingredient of grape juice, resveratrol might be responsible for the cardioprotective effect of red wine. Furthermore, resveratrol can prevent or delay the progression of cancer and it extends the life spans of various organisms by the activation of sirtuin deacetylases (reviewed in reference 1). Because of its beneficial properties, the biotechnological production of resveratrol in microorganisms has attracted increasing industrial interest.
Resveratrol is synthesized in the phenylpropanoid pathway from the precursor molecule p-coumarate, which is converted to p-coumaroyl-coenzyme A (CoA) by the enzyme 4-coumaroyl-CoA ligase (4CL1) (Fig. (Fig.1).1). In the following steps, the stilbene synthase (STS), a type III polyketide synthase, adds three units of acetate derived from malonyl-CoA to the molecule, resulting in resveratrol. In earlier studies, the production of resveratrol in Escherichia coli and Saccharomyces cerevisiae by the heterologous expression of STS and 4CL and feeding of p-coumarate was described (reviewed in references 4 and 6). Amounts from 1.5 mg/liter up to 6 mg/liter for S. cerevisiae (2, 3) and up to 171 mg/liter for E. coli (7, 10) were reported. In our study, we could increase the resveratrol yield up to 391 mg/liter with an industrial Brazilian S. cerevisiae strain expressing 4-coumaroyl-coenzyme A ligase from Arabidopsis thaliana and stilbene synthase from Vitis vinifera.
The 4CL1 gene (GenBank accession no. NM 104046) from A. thaliana was amplified by PCR from cDNA using oligonucleotides 5′-AACACAAAAACAAAAAGTTTTTTTAATTTTAATCAAAAAATGGCGCCACAAGAACAAGC and 5′-GAATGTAAGCGTGACATAACTAATTACATGACTCGAGTCACAATCCATTTGCTAGTT (4CL1 gene sequence is underlined). The STS gene (GenBank accession no. DQ 459351) from V. vinifera was amplified by PCR from cDNA using oligonucleotides 5′-AACACAAAAACAAAAAGTTTTTTTAATTTTAATCAAAAAATGGCTTCAGTCGAGGAAAT and 5′-GAATGTAAGCGTGACATAACTAATTACATGACTCGAGTTAATTTGTAACCATAGGAA (STS gene sequence is underlined). The DNA fragments were cloned into 2μ-based multicopy plasmids with the auxotrophic marker genes URA3 and LEU2, respectively, by homologous recombination (11), placing STS and 4CL under the control of a strong and constitutive HXT7 promoter fragment and the CYC1 terminator. The plasmids were transformed into S. cerevisiae CEN.PK2-1 cells by lithium acetate transformation (5), and positive transformants were selected on SD agar plates (6.7 g/liter yeast nitrogen base with amino acids, 20 g/liter glucose, 20 g/liter agar) lacking uracil and leucine.
A flask culture fermentation with the recombinant yeast strain was performed in SD medium supplemented with 5 mM p-coumarate (Sigma). Every 24 h, 1 ml of the culture was taken, mixed with 1 ml acetone, and spun down for 15 min at 13,000 rpm, and the supernatants were analyzed in a high-pressure liquid chromatography (HPLC) device (PDA-100 photodiode array detector and RF 2000 fluorescence detector [Dionex] and a 3.5-μm particle size, 0.6- by 150-mm SB-C8 column [Agilent] with 40% methanol-0.1% trifluoroacetic acid [TFA] used for the mobile phase). Quantification of p-coumarate and resveratrol was achieved by comparing the sample data with calibration curve data obtained by parallel HPLC analysis of dilution series of resveratrol and p-coumarate. Both substances were identified by their characteristic absorption spectra (resveratrol, 308 nm, and p-coumarate, 312 nm) (8). A resveratrol concentration of 6 mg/liter was detected after 144 h of incubation. Longer incubation resulted in a decrease in the resveratrol concentration, although there was still p-coumarate left in the culture.
To increase the resveratrol productivity, a fermentation in rich YEPD medium (10 g/liter yeast extract, 20 g/liter bacterial peptone, 20 g/liter glucose) was tested. Therefore, the STS and 4CL1 genes were recloned into the 2μ-based multicopy drug resistance marker plasmids pRS42k (G418 resistance) (9) and p426iRAD (hygromycin resistance) (Wiedemann and Boles, unpublished data), still under the control of the same HXT7 promoter fragment and the CYC1 terminator. The plasmids were transformed into CEN.PK2-1 cells, and positive transformants were selected on YEPD agar supplemented with G418 and hygromycin (200 mg/liter each). A shaking flask fermentation of this strain in YEPD medium with G418 and hygromycin supplemented with 10 mM p-coumarate was performed; an additional amount of 5 mM p-coumarate was added after 120 h for an optimal precursor supply. Under these conditions, the CEN.PK2-1 strain produced 262 mg/liter resveratrol within 144 h. This result shows that a fermentation in rich medium yields much higher resveratrol levels than a fermentation in synthetic medium.
In the next step, the plasmids were transformed into four S. cerevisiae strains from industrial sources to test their efficiency in resveratrol production. Three strains produced resveratrol in various concentrations, and interestingly, one strain metabolized p-coumarate without resveratrol synthesis. The highest resveratrol yield, 391 mg/liter, was produced by an S. cerevisiae strain isolated from a Brazilian sugar cane plantation (Barra Grande) (Fig. (Fig.2).2). The increase in the resveratrol productivity of this strain in comparison to that of CEN.PK2-1 may be due to its different genetic background.
To ensure that the greatly increased resveratrol productivity was indeed an effect of the rich medium and not an effect of the drug resistance marker plasmids, fermentations with the strains in SD medium with G418 and hygromycin selection were performed. As expected, this resulted in much lower resveratrol yields that were comparable to those obtained with the auxotrophic selection marker plasmids (not shown).
As in SD medium, incubation for longer than 144 h did not increase the resveratrol concentration. One possible reason could be feedback inhibition effects of resveratrol because of its intracellular accumulation (in fact, 43% ± 9% [mean ± standard deviation] of the total resveratrol amount was detected in the cell pellet fraction of a centrifuged flask culture sample after 144 h of incubation). Therefore, fermentations were performed after the addition of 500 mg/liter resveratrol to the medium. In these experiments, the resveratrol levels still increased by 370 mg/liter, up to 870 mg/liter, which shows that the high resveratrol concentration does not have an inhibitory effect. Furthermore, different fermentation conditions, i.e., high glucose concentrations, anaerobic conditions, and different carbon sources, were tested. All of these approaches resulted in lower resveratrol yields (not shown).
In this study, we showed that the resveratrol productivity of recombinant S. cerevisiae strains starting from p-coumarate can be increased by fermentation in YEPD instead of SD medium. Furthermore, the increase in the resveratrol concentrations achieved with a Brazilian wild-type strain (391 mg/liter) compared to the concentrations produced by the laboratory CEN.PK2-1 strain (262 mg/liter) show that unknown strain-specific features seem to be important for resveratrol productivity. Moreover, it might also be that the source of the 4CL gene, A. thaliana, might have contributed to the high resveratrol productivity, as in previous studies the 4CL gene had been selected from a hybrid poplar (2) or from Nicotiana tabacum (3). In general, this study shows that S. cerevisiae is well suited as a host for industrial resveratrol production. By further optimization of the fermentation conditions and metabolic engineering strategies, the productivity might be further enhanced.
We thank Volker Müller for the kind provision of the HPLC equipment and Stefan Saum for his expert technical support. We thank Claudia Stamme and Beate Wiedemann for the kind provision of vectors pRS42k and p426iRAD.
This work was funded by Evonik Industries, financially supported by the State of North-Rhine Westfalia, and cofinanced by the European Union.
Published ahead of print on 26 March 2010.