Enhancing beta-carotene production in Saccharomyces cerevisiae by metabolic engineering


  • Qian Li,

    1. CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Zhiqiang Sun,

    1. CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Jing Li,

    1. CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China
    2. University of Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Yansheng Zhang

    Corresponding author
    • CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, China
    Search for more papers by this author

Correspondence: Yansheng Zhang, CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, 430074, China. Tel./fax: +86 27 8761 7026; e-mail: zhangys@wbgcas.cn


Beta-carotene is known to exhibit a number of pharmacological and nutraceutical benefits to human health. Metabolic engineering of beta-carotene biosynthesis in Saccharomyces cerevisiae has been attracting the interest of many researchers. A previous work has shown that S. cerevisiae successfully integrated with phytoene synthase (crtYB) and phytoene desaturase (crtI) from Xanthophyllomyces dendrorhous could produce beta-carotene. In the present study, we achieved around 200% improvement in beta-carotene production in S. cerevisiae through specific site optimization of crtI and crtYB, in which five codons of crtI and eight codons of crtYB were rationally mutated. Furthermore, the effects of the truncated HMG-CoA reductase (tHMG1) from S. cerevisiae and HMG-CoA reductase (mva) from Staphylococcus aureus on the production of beta-carotene in S. cerevisiae were also evaluated. Our results indicated that mva from a prokaryotic organism might be more effective than tHMG1 for beta-carotene production in S. cerevisiae.


Carotenoids are health-promoting metabolites that are widely biosynthesized in plants, microorganisms, algae, etc. Beta-carotene, which humans cannot synthesize de novo, is the precursor of Vitamin A and astaxanthin. It has been reported that beta-carotene functions as an antioxidant preventing angiocardiopathy, strengthens the immune system, and decreases the risk of cancer (Bracco et al., 1981; Buring & Hennekens, 1995; Kritchevsky, 1999). Nowadays, beta-carotene has also gained increasing attention in food and feed additives (Nelis & De Leenheer, 1991), cosmetics (Anunciato & da Rocha Filho, 2012), and health food and pharmaceutical industries (Bauernfeind et al., 1970; Johnson, 2002).

Due to its importance, biosynthesis of carotenoid has aroused many researchers' interests, with most studies focusing on the metabolic engineering of carotenoid biosynthesis (Misawa & Shimada, 1997; Lee & Schmidt-Dannert, 2002). Saccharomyces cerevisiae is an edible microorganism which is widely used in fermentation industries. Compared with other heterologous beta-carotene producing microorganisms such as Escherichia coli (Kim et al., 2006) and Candida utilis (Miura et al., 1998), S. cerevisiae is considered to be a safe yeast and has the advantage of easy genetic manipulation with established host–vector systems. As a conventional yeast, S. cerevisiae can produce geranylgeranyl diphosphate, which is the precursor of carotenoid biosynthesis. Previous studies have shown that S. cerevisiae integrated with the two main carotenogenic genes, phytoene synthase (crtYB) and phytoene desaturase (crtI), from Xanthophyllomyces dendrorhous could produce carotenoids (Verwaal et al., 2007).

Experimental investigation demonstrated that different species favor distinctive codons in translating nucleotides into amino acids throughout evolution (Hershberg & Petrov, 2008; Plotkin & Kudla, 2010). Due to the difference in the codon usage preference among species, expressions of foreign enzymes often result in suboptimal performance. Therefore, codon optimizations have been applied to improve the productions of many natural products. For example, the codon-optimized taxadiene synthase cDNA directed a 40-fold increase in the production of taxadiene in yeast (Engels et al., 2008). The original cDNA encoding tyrosine ammonia lyase (TAL) from Rhodobacter sphaeroides was not utilized in a yeast translation process (Zhang et al., 2006). However, the TAL enzyme was highly expressed in S. cerevisiae when the TAL cDNA was full-length codon optimized, which led to a higher level of resveratrol production in the yeast cells (Wang et al., 2011). In the present study, we report an improvement in the production of carotenoids in S. cerevisiae by specific site optimizations of crtI and crtYB cDNAs from X. dendrorhous.

The enzyme 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGR) has been shown to be a major rate-limiting enzyme in the mevalonate (MVA) pathway in S. cerevisiae (Britton et al., 1998). Overexpression of the catalytic domain of the HMGR (namely, truncated HMG-CoA reductase gene or tHMG1) in yeast cells has been observed to boost isoprenoid biosynthesis (Donald et al., 1997; Polakowski et al., 1998). Through in vitro and in vivo assays of various HMGRs, the HMGR from Staphylococcus aureus, named mva, has been noted to exhibit a higher activity than tHMG1 from E. coli (Ma et al., 2011). In the present study, the effects of tHMG1 and mva on carotenoid production in S. cerevisiae were investigated to compare their overall performance in S. cerevisiae.

Materials and methods

Cloning and site-directed mutagenesis of crtI and crtYB

The cDNAs corresponding to crtI (GenBank accession no. AY177424.1) and crtYB (GenBank accession no. AY177204.1) were isolated from X. dendrorhous by RT-PCR, and cloned into the pMD18-T vector to obtain the constructs pMD18-T-crtI and pMD18-T-crtYB, respectively, for DNA sequencing. To determine the nucleotide sites whose utilization frequency is < 15% in S. cerevisiae, the full-length cDNA sequences of crtI and crtYB were analyzed using a graphical codon usage analyzer (GCUA; Fuhrmann et al., 2004) with the codon usage table of S. cerevisiae. Subsequently, five nucleotides of crtI and eight nucleotides of crtYB were identified to be poor sites (usage frequency of < 15%) with regard to the translation efficiency in S. cerevisiae. Based on these results, the poor codon sites of crtI and crtYB were subjected to mutation to improve their codon usage frequency. All of the site-directed mutagenesis processes were accomplished by overlapping extension PCR with primers 5–30, using the plasmid pMD18-T-crtI or pMD18-T-crtYB as the template (Kanoksilapatham et al., 2007). The mutated crtI and crtYB were designated as McrtI and McrtYB, respectively. All the primers used for the PCR in this study are listed in Table 1.

Table 1. Primers used in this study
Primer no.DescriptionSequence (5′ to 3′)

Plasmid constructions

A DNA cassette for yeast expression was prepared in the vector PUC19 to obtain the construct pLQ01 (Fig. 1). The plasmid pLQ01 comprised 680-bp TDH3 gene promoter and 250-bp CYC1 gene terminator sequences, which were amplified from the genomic DNA of S. cerevisiae using primers 31–34. The open reading frames (ORFs) of crtI, crtYB, McrtI and McrtYB were linked to the TDH3 promoter and CYC1 terminator by inserting the ORFs into the plasmid pLQ01 at the BamHI/NotI site. The expression cassette ‘TDH3 promoter-gene-CYC1 terminator' was then amplified using primers 35–38. The DNA cassettes ‘TDH3 promoter-crtI-CYC1 terminator' and ‘TDH3 promoter-McrtI-CYC1 terminator' were digested with SpeI and SalI restriction enzymes and introduced into an integrative yeast expression vector pRS406 at SpeI/SalI sites, resulting in the constructs pRS406-crtI and pRS406-McrtI, respectively. The SalI/XhoI fragment of ‘TDH3 promoter-crtYB-CYC1 terminator' was subsequently ligated into pRS406-crtI at the SalI/XhoI site to obtain the construct pRS406W (Fig. 1). The same approach was applied to construct pRS406M by excising the SalI/XhoI fragment of ‘TDH3 promoter-McrtYB-CYC1 terminator' and introducing into pRS406-McrtI.

Figure 1.

Structure of the vectors. (a) pLQ01 containing a TDH3 promoter and a CYC1 terminator. (b) pRS406. (c) pRS406W carrying the wild-type genes responsible for carotenoids expression. (d) pRS406M carrying the optimized carotenoids expression genes. (e) pESC-TDH3-tHMG1 holding the catalytic domain of tHMG1 gene in Saccharomyces cerevisiae. (f) pESC-TDH3-tHMG1 holding the mva gene derived from Staphylococcus aureus.

The ORF of tHMG1 cDNA (NCBI Reference Sequence NM_001182434.1) was isolated by RT-PCR with primers 39–40 from S. cerevisiae WAT11 strain, and the cDNA corresponding to mva (NCBI Reference Sequence NC_017342.1) was amplified using primers 41–42 from the genomic DNA of S. aureus (ATCC25923). The S. aureus strain was kindly provided by Dr. Qiang Gao (Biological Control of Arborvirus Vectors, Wuhan Institute of Virology, Chinese Academy of Sciences). Through overlapping extension PCR with primers 43–44, the ORF of tHMG1 was linked to the TDH3 promoter to obtain the fragment ‘TDH3-tHMG1’. The DNA fragment ‘TDH3-tHMG1’ was then digested and ligated into a yeast expression vector pESC-HIS (Stratagene) at the SpeI/BamHI site to obtain the construct pESC-HIS-TDH3-tHMG1. Similarly, the ORF of mva was cloned at the TDH3 promoter by overlapping PCR using primers 45–46, and then ligated into pESC-HIS at the EcoRI/BamHI site to obtain the pESC-HIS-TDH3-mva vector. In the constructs pESC-HIS-TDH3-tHMG1 and pESC-HIS-TDH3-mva, the endogenous galactose inducing promoters (Gal1 and Gal10) in pESC-HIS were removed (Fig. 1).

Yeast strain construction and cultivation

The integration vectors pRS406, pRS406W, and pRS406M were linearized with StuI and integrated into the ura3-52 locus of S. cerevisiae WAT11 strain to create the yeast strains WAT11/pRS406, WAT11/pRS406W, and WAT11/pRS406M, using the PEG/LiAc method (Gietz & Woods, 2002). The episomal vector pESC-HIS-TDH3-tHMG1 was transformed into the yeast strain WAT11/pRS406M to obtain WAT11/pRS406M-tHMG1. Similarly, the strain WAT11/pRS406M-mva was prepared by transforming the construct pESC-HIS-TDH3-mva into the yeast strain WAT11/pRS406M. As the control, the empty vector pESC-HIS was transferred into the yeast strain WAT11/pRS406M to construct WAT11/pRS406M-HIS. Detailed information regarding the constructed yeast strains is presented in Table 2.

Table 2. Strains and plasmids used in this study
Strain or plasmid Relevant features
Yeast strainsWAT11MATα (leu2-3,112 trp1-1 can1-100 ura3-1 ade2-1 his3-11,15)
Integrative vector transformants
WAT11/pRS406 WAT11+ pRS406
WAT11/pRS406W WAT11+ pRS406W
WAT11/pRS406M WAT11+ pRS406M
Episomal vector transformants
WAT11/pRS406M-tHMG1 WAT11+ pRS406M + pESC-TDH3-tHMG1
WAT11/pRS406M-mva WAT11+ pRS406M + pESC-TDH3-mva
pRS406W pRS406 TDH3-WCrtYB-CYC1, TDH3-WCrtI-CYC1
pRS406M pRS406 TDH3-MCrtYB-CYC1, TDH3-MCrtI-CYC1
pESC-TDH3-mva pESC-TDH3-mva-CYC1

The yeast strains were cultured at 250 rpm and 30 °C in an appropriate yeast liquid medium. The strains WAT11/pRS406, WAT11/pRS406W, and WAT11/pRS406M were grown in SD-URA medium (amino acid dropout medium), and the strains WAT11/pRS406M-tHMG1, WAT11/pRS406M-mva, and WAT11/pRS406M-HIS were grown in SD-HIS-URA medium (amino acid dropout medium). To monitor the yeast growth, a single clone from each of the yeast strains was initially cultured in 5 mL of yeast medium at 30 °C under constant shaking to an optical density (measured at 600 nm; OD600) of 0.6. The yeast cultures were then inoculated into fresh SD medium at a ratio of 1 : 40. Subsequently, 200 μL of the samples were collected at 0, 6, 12, 24, 30, 36, 48, 60, 72, 96, and 120 h to measure OD600.

Extraction and quantification of beta-carotene

The yeast cells grown at 30 °C for 72 h in 50 mL of yeast medium were pelleted by centrifugation at 3500 g for 5 min, washed twice with 0.9% (w/v) NaCl, and lyophilized (Lange & Steinbuchel, 2011). Approximately 100 mg of the freeze-dried cells were suspended in 2 mL of acetone/0.2% pyrogallol in methanol (w/v; 80 : 20, v/v), and 1 g of glass beads (diameter, 425–600 μm) was added. The mixture was shaken for 3 min in a high-throughput tissue grinder, and the acetone–methanol fraction was collected by centrifugation at 6000 g for 5 min. The extraction was repeated four to five times. The organic extractions were pooled, evaporated to dryness, and re-dissolved in 1 mL of acetone for HPLC analysis. To avoid photo-oxidation, all the extraction procedures were carried out in darkness.

The HPLC analysis was performed on an LC-20AT instrument equipped with a binary pump, an autosampler, and a photodiode array detector (Shimadzu, Kyoto, Japan). An Agilent HC-C18 (2) reversed-phase column (4.6 × 250 mm, 5 μm) was used with acetonitrile/methanol (50 : 50 v/v) as the mobile phase at a flow rate of 1 mL min−1. The column temperature was set at 25 °C and the detection wavelength was 450 nm. Beta-carotene generated from the yeast cultures was quantified based on a standard calibration curve made with authentic beta-carotene (Sigma Aldrich GmbH, China) of various concentrations.

Results and discussion

Codon-optimized crtI and crtYB highly improved carotenoid production in S. cerevisiae

Saccharomyces cerevisiae has been routinely used as a heterologous expression system because it is safe to use, involves mature fermentation technology, etc. However, the bias of codon usage is an important restriction factor for the expression of foreign enzymes in S. cerevisiae. In the present study, S. cerevisiae integrated with crtI and crtYB genes from X. dendrorhous was capable of expressing those genes and producing carotenoids (Fig. 2). To investigate the possibility of improving carotenoid production in S. cerevisiae, the codon usage of crtI and crtYB cDNA sequences in S. cerevisiae were examined using GCUA. As shown in Fig. 3, five codons of crtI and eight codons of crtYB were found to be poor sites with less than 15% of usage frequency. Using overlapping extension PCR, the poor codons were mutated to favorable ones of S. cerevisiae (Fig. 3). The codon-optimized genes, McrtI and McrtYB, were successfully integrated into the genome of S. cerevisiae. The yeast strain integrated with both wild-type crtI and crtYB genes served as the control. For the negative control, the empty integrative vector pRS406 was integrated into the genome of the yeast strain. A representative clone from these transgenic yeast strains was streaked onto an SD-URA plate and grown at 30 °C for 5 days. The color of the negative control strain was white, whereas the strain containing wild-type crtI and crtYB genes was yellow. Apparently, a yellow color with more intensity was observed from the strain integrated with McrtI and McrtYB (Fig. 4a), suggesting an evident improvement in beta-carotene production in this strain. The growth properties of these yeast strains grown in liquid media were monitored by measuring OD600 through a time course. None of the three strains showed significant differences in their growth curves and the cell densities were almost saturated at 72 h, starting from the cells inoculated into the fresh medium (cf. yeast strains cultivation in 'Materials and methods'). Therefore, the cells grown for 72 h were harvested for extraction and quantification of beta-carotene. HPLC analysis showed around 200% improvement in the production of beta-carotene in the yeast strain integrated with mutated genes when compared with that in the control yeast bearing wild-type genes. The concentration of beta-carotene produced by the control strain was about 85.6 μg g−1 dw, whereas the yeast strain containing mutated genes produced beta-carotene up to a concentration of 251.8 μg g−1 dw (Fig. 4b). The negative control yeast strain integrated with empty vector pRS406 did not produce any beta-carotene (data not shown). Surprisingly, only site optimizations resulted in such a distinct improvement in beta-carotene production, leading to the expectation that more beta-carotene could be generated in S. cerevisiae by optimizing the full-length crtI and crtYB cDNAs. The present study provides another example of the effectiveness of codon optimization for heterologous expressions of foreign metabolic pathways.

Figure 2.

Metabolic pathway of endogenous ergosterol biosynthesis and exogenous beta-carotene biosynthesis. The solid arrows show the one-step conversions of biosynthesis, and the dashed arrows show the several-step conversions. crtYB and crtI are the endogenous beta-carotene synthesis genes.

Figure 3.

The low usage codons of (a) crtI and (b) crtYB gene from Xanthophyllomyces dendrorhous, and (c and d) codon usage frequency of these sites after optimization.

Figure 4.

(a) Colors of different carotenoid-producing Saccharomyces cerevisiae strains on a – URA plate for (a) WAT11/pRS406, (b) WAT11/pRS406W, and (c) WAT11/pRS406M; on a – URA-HIS plate for (d) WAT11/pRS406M-HIS, (e) WAT11/pRS406M-tHMG1, and (f) WAT11/pRS406M-mva. (b) The concentrations of beta-carotene produced in the different S. cerevisiae transformants. Values are the mean ± SE of three experiments.

Higher beta-carotene production in S. cerevisiae directed by mva than tHMG1

To compare the overall performance of mva and tHMG1 in S. cerevisiae in vivo, the mva and tHMG1 cDNAs were cloned at the TDH3 constitutive promoter and separately overexpressed in the constructed carotenoid-producing S. cerevisiae. The carotenoid-producing yeast transformed with the empty vector pESC-HIS served as the control strain. When the representative clones from these transgenic strains were streaked onto an SD-HIS-URA plate, a clear difference in yellow-color intensity was observed (Fig. 4A). The strain harboring either tHMG1 or mva was much yellower than the control strain, suggesting a higher content of beta-carotene. To investigate whether the expression of tHMG1 or mva inhibits the yeast growth, the cell densities of the three transgenic strains were evaluated by measuring the OD600 values during a time course. The yeast growth curves were very similar, indicating that the expression of tHMG1 or mva had no or negligible effects on the yeast growth. In addition, the biomass yields of those strains were also almost saturated at 72 h in culture conditions (data not shown). After 72 h of growth, the yeast cells were harvested for the quantification of beta-carotene using HPLC analysis. The beta-carotene content of the control strain was found to be 237 μg g−1 dw. When compared with the control strain, the strain overexpressing tHMG1 exhibited a 30% improvement in the production of beta-carotene (310.8 μg g−1 dw), whereas mva expression directed around 60% increase in the beta-carotene levels (390 μg g−1 dw; Fig. 4B). This result indicates that mva was more efficient than tHMG1 in promoting isoprenoid biosynthesis in S. cerevisiae. This higher efficacy of mva than tHMG1 in S. cerevisiae could possibly be related to the higher catalytic efficiency of mva (Ma et al., 2011). When the codon usage preference analysis was applied to mva and tHMG1 cDNA sequences using S. cerevisiae as the host organism, we found that the overall codon usage frequency of mva was lower than that of tHMG1 (data not shown). Thus, codon-optimized mva can be expected to further increase isoprenoid biosynthesis in S. cerevisiae.

In summary, our results strongly suggested that the codon optimizations of CrtI and CrtYB led to a higher concentration of beta-carotene produced in S. cerevisiae strain. Moreover, compared with the tHMG1 expressed, mva seemed to be better to direct more carbon fluxes into isoprenoid pathway, which is consistent with a previous biochemical study (Ma et al., 2011). In comparisons with other reports, the overall beta-carotene yield of the S. cerevisiae strains in this study was comparable to that of S. cerevisiae strains engineered by Yamano et al. (1994) but was more than 10-fold lower relative to the beta-carotene yield engineered in S. cerevisiae by Verwaal et al. (2007). This most likely resulted from the different number of genes engineered in yeast cells or genotype differences between S. cerevisiae stains, e.g. S. cerevisiae CEN. PK contained an unusually high content of ergosterol and fatty acids compared with other S. cerevisiae strains (Daum et al., 1999).


This project was supported by the Grant for One Hundred Talents Program of the Chinese Academy of Sciences, China (Project No. Y129441R01).

Authors' contribution

Q.L., Z.S. and J.L. contributed equally to this work.