All the nucleotide primers for amplifying genes in this study are listed in Supplementary Table 1. The BTS1 (S. cerevisiae geranylgeranyl pyrophosphate synthase gene, GenBank accession no. NM_001183883) expression cassette, ‘pTEF-BTS1-CYC1t,’ was amplified by linking the BTS1 open reading frame (ORF) with the TEF promoter (GenBank accession no. EF210199) and CYC1 terminator (GenBank accession no. DQ232604) using primers 1–6 (Supplementary Table 1). The DNA cassette ‘pTEF-BTS1-CYC1t’ was then introduced into an integrative yeast expression vector YIplac128 at BamHI/SalI sites, resulting in the constructs YIplac128-BTS1. YIplac128-BTS1 was linearized with an EcoRV digestion and integrated into the LEU2 locus of the WAT11/pRS406W strain, which is a carotene-producing yeast previously prepared by Li et al. (2013). The resultant yeast strain bearing the BTS1 expression cassette was designated WYIB in this study. Yeast cells were grown on appropriate dropout medium supplemented with 2% glucose or 2% galactose as the carbon source. For monitoring the growth property of the recombinant yeast strains, single colonies were picked from plates, inoculated into appropriate liquid medium containing 2% glucose, and grown overnight at 30°C. The yeast cells were then washed with sterile water and sub-cultured into a fresh induction medium at an OD₆₀₀ of 0.1, where 2% glucose was replaced with 2% galactose. Starting from the sub-culture, the OD₆₀₀ of the cell cultures were measured every 24 h and the cell biomass was recorded by weighing lyophilized cell weight.

The λTRP-S. cerevisiae cDNA library, which was generated by cre-lox recombination from λTRP (Elledge et al., 1991), was ordered from the American Type Culture Collection (ATCC no. 87277). In this library, individual yeast cDNAs were inserted between GAL1 promoter and CYC1 terminator in the pTRP vector with TRP1 as a selection marker. To screen the gene targets that potentially improve β-carotene biosynthesis, 10 μg of the yeast cDNAs was transformed into the WYIB strain and grown on the selective plates (dropout medium lacking uracil, leucine, and tryptophan, and containing 2% galactose and 2% agar) for colony color visual screening. The WYIB strain transformed with the empty vector pESC-TRP (Stratagene, La Jolla, CA, United States) served as a control. After being grown for 7 days, the colony color phenotype was suitable for visibly validation. Three rounds of the colony color screening were applied: First, the colonies harboring the S. cerevisiae cDNA library were compared against the control one on same plates and the colonies with increased orange color were selected; second, the selected colonies were re-spotted onto plates and the truly colonies with increased orange color relative to the control were used for the next screening; third, the plasmids containing the S. cerevisiae cDNAs were isolated from the selected colonies and re-transferred into the WYIB strain to confirm the color phenotype compared with the control strain. After the three rounds of selections, colonies with increased orange color were eventually considered as the positive clones and their contained cDNA inserts were discovered through plasmid isolation followed by sequencing.

To examine whether the above positive cDNA inserts truly enhance β-carotene production in S. cerevisiae, the ORFs of the cDNA candidates were amplified using primers 7–18 and individually inserted into the yeast expression vector pESC-TRP under the control of a galactose inducible promoter, yielding the constructs of pESC-TRP-BMH1, pESC-TRP-CAR1, pESC-TRP-DID2, pESC-TRP-PDC5, pESC-TRP-TIF5, and pESC-TRP-VOA1. The resultant constructs were then separately transformed into the WYIB strain to give the strains of WYIB-BMH1, WYIB-CAR1, WYIB-DID2, WYIB-PDC5, WYIB-TIF5, and WYIB-VOA1, respectively. The WYIB strain transformed with the empty vector pESC-TRP served as the control and designated WYIB-TRP. Single colonies of each transformant were cultured in dropout liquid medium as described above, and the 72 h-grown cells at stationary phase were harvested for measuring β-carotene yield. Two-tailed t-test was performed for the statistical analysis. To further confirm the increment in β-carotene production by the best target (DiD2), the strain WYIB-DID2 was compared with the control strain WYIB-TRP in a time course with respect to the ability of producing β-carotene.

To verify the universality of the improving role of DID2 on β-carotene biosynthesis, another yeast strain with a genetic background distinct from the WAT11 strain used above, namely CEN.PK2-1C, was used as a host to test the DID2 effect. The carotenoid-producing vectors, pRS406W and plac128-BTS1, were linearized with StuI and EcoRV, respectively, and simultaneously integrated into the genome of CEN.PK2-1C under the URA3 and LEU2 locus, yielding the carotene-producing strain of CYIB. To test the DID2 effect under the CYIB background, the constructs of pESC-TRP-DID2 and pESC-TRP were separately transformed into the CYIB strain to give the strains of CYIB-DID2 and CYIB-TRP, then, the β-carotene yields by both strains were compared in a time course.

The β-carotene was extracted from the yeast cultures with acetone/0.2% pyrogallol in methanol (w/v) (80:20, v/v) as described previously (Li et al., 2013), and its yields were quantified by high performance liquid chromatography (HPLC) analysis. The HPLC analysis was performed on an LC-20AT instrument (Shimadzu, Kyoto, Japan) with an inertsil ODS-SP reversed-phase column (4.6 mm × 250 mm, 5 μm). Metabolites were separated at 2 mL/min and 25°C in acetonitrile/methanol (v/v = 50:50), and monitored at a wave length of 450 nm. The β-carotene production was quantified using a standard calibration curve which was obtained by authentic β-carotene from Sigma Chemical Co. (Cat. no. 7235407) in five gradient concentrations.

Total RNA was isolated from the yeast cells using an EASYspin RNA kit (Aidlab, Beijing, China). RNA was treated with DNase I (Thermo Scientific) to remove genomic DNA contaminations, and reversed to cDNA using Oligo-dT18 primer (Qsingke, Wuhan, China) and M-MLV reverse transcriptase (Thermo Scientific) following the provided protocols. To analyze the transcript levels of genes encoding 3-hydroxy-3-methylglutaryl coenzyme-A reductase (HMG1, GenBank accession no. NM_001182434), mevalonate kinase (ERG12, GenBank accession no. NM_001182715), phosphomevalonate kinase (ERG8, GenBank accession no. NM_001182727), farnesyl pyrophosphate synthase (ERG20, GenBank accession no. NM_001181600), geranylgeranyl pyrophosphate synthase (BTS1, GenBank accession no. NM_001183883), phytoene synthase and cyclase (crtYB, GenBank accession no. AY177204) and phytoene desaturase (crtI, GenBank accession no. AY177424), specific primers 19–32 (Supplementary Table 1) were used to amplify their transcripts. To normalize the variation between the cDNA preparations, the S. cerevisiae actin gene (GenBank accession no. NM_001179927) was used as an internal standard and amplified with the primers 33/34. Real-time RT-PCRs were performed in three biological replicates with three technical repeats, using FastStart Universal SYBR Green Master mix (Rox) (Roche, Mannheim, Germany) on an ABI 7500 FAST Real Time PCR Systems (Life Technologies, United States). The relative expression was estimated using the comparative threshold cycle method (Schmittgen and Livak, 2008). Two-tailed t-test was performed to evaluate the gene expression difference between the control strain WYIB-TRP and the strain WYIB-DID2.

To identify the genes that potentially improve β-carotene production, the carotenoid-producing strain WYIB was screened with the collection of the yeast cDNAs (ATCC no. 87277). Transformants with a higher production of β-carotene were visually sieved by the increased orange color. Previously, this approach was successful in quickly identifying gene targets that affect carotenoid biosynthesis (Ozaydin et al., 2013). In the first round of the screening, 500 transformants among those showing the increased orange color out of about 80,000 colonies were selected. The colonies selected from the primary screening were then subjected to additional two rounds of the color validations. Eventually, there were twenty-two colonies with the deepest orange color identified from the color screening (Figure 2) The cDNA inserts of these 22 colonies were then discovered by sequencing. Each of the colonies was found to bear only one cDNA insert. Among these, there were five gene targets that encode ribonucleoproteins (RNPs) and nine genes that were present in an anti-sense form. By a blast search, the remaining eight cDNA inserts were found to encode enzymes of glyceraldehyde-3-phosphate dehydrogenase (Tdh1), pyruvate kinase (Cdc19), arginase (Car1), class E protein of the vacuolar protein-sorting pathway (Did2), pyruvate decarboxylase (Pdc5), 14-3-3 protein (Bmh1), translation initiation factor (Tif 5), and vacuolar H(+)-ATPase subunit 1 (Voa1), respectively. Tdh1 and Cdc19 are the enzymes required for glycolysis process which is upstream of the isoprenoid pathway and which also generate acetyl-CoA, the precursor of the carotenoid biosynthesis. Because this study focused on the genes outside the isoprenoid pathway, the Tdh1 and Cdc19 were not further investigated and the remaining six candidates were in the further validations.

In order to ensure their positive effects of the above six candidates (CAR1, DID2, PDC5, BMH1, TIF5, and VOA1) on the β-carotene production in S. cerevisiae, their ORFs were individually re-amplified, cloned into the yeast expression vector pESC-TRP, and transformed into the WYIB strain. As a control, the WYIB strain was transformed with the empty vector pESC-TRP. The β-carotene yield accumulated by these transformants was measured by HPLC analysis. Compared with the control, the expressions of DID2, VOA1, TIF5, and BMH1 increased the β-carotene production to 205, 152, 148, and 133%, respectively, whereas no significant increment was observed by over-expressing either PDC5 or CAR1 (Figure 3). Among these targets, the DID2 over-expression led to the largest increase in β-carotene levels. The enhanced β-carotene production in the DID2 over-expressing strain (WYIB-DID2) was further assessed in a time course compared to the control strain (WYIB-TRP). The strain WYIB-DID2 produced significantly higher levels of β-carotene than the control strain at all the time points (Figure 4A). The β-carotene level accumulated by the strain WYIB-DID2 at the 120 h-time point was 5.9 ± 0.1 mg g^-1 dry cell weight, which was 2.1-fold of that of the control strain (2.8 ± 0.2 mg g^-1 DW). The growth property of the strain WYIB-DID2 was measured as well. As shown in Figure 4B, except for a little bit reduction, the growth profile of WYIB-DID2 was comparable to that of the control strain, indicating that the up-regulation of DID2 exerted no serious inhibition on yeast growth.

To test the universality of the DID2 effect, we integrated the carotenoid pathway into another yeast strain CEN.PK2-1C and expressed the DID2 gene under this yeast background. The production of β-carotene was compared in a time-course manner between the strain containing the DID2 (CYIB-DID2) and the control strain harboring the empty vector pESC-TRP (CYIB-TRP). As shown in Supplementary Figure 1A, in comparison with the control, the DID2 expression increased the β-carotene yield at all the time points, and the levels of β-carotene produced by CYIB-DID2 were 1.8- to 2.4-fold of those produced by the control strain CYIB-TRP. Moreover, once again, the DID2 expression did not significantly inhibit the yeast growth of the CEN.PK2-1C strain (Supplementary Figure 1B).

To explain why the DID2 over-expression improve the production of β-carotene in S. cerevisiae, we tested the effects of the DID2 over-expression on the expression levels of the seven β-carotene biosynthetic genes (HMG1, ERG12, ERG20, ERG8, BTS1, crtYB, and crtI) (Figure 1). Transcript abundances of these genes were measured at either the exponential (24 h-culture) or stationary phase (72 h-culture). Compared to those in the control yeast cells (WYIB-TRP), the expression levels of these genes in the DID2-overexpressed yeast cells (WYIB-DID2) generally increased to different extents at both growth stages (Figure 5). For instance, at the exponential phase, the transcript levels of HMG1, ERG20, and ERG8 in the upstream pathway were up-regulated by 0.9-, 1.3-, and 1.8-fold, respectively (Figure 5A); at the stationary stage, the increment of the crtYB expression levels reached up to 2.0-fold while the crtI expression increased up to 2.5-fold (Figure 5B). Thus, it suggested that the improvement in the β-carotene production induced by DID2 over-expression was most likely caused by the elevated transcript levels of β-carotene biosynthetic genes.