Screening and evaluation of the strong endogenous promoters in Pichia pastoris

Background Due to its ability to perform fast and high-density fermentation, Pichia pastoris is not only used as an excellent host for heterologous protein expression but also exhibits good potential for efficient biosynthesis of small-molecule compounds. However, basic research on P. pastoris lags far behind Saccharomyces cerevisiae, resulting in a lack of available biological elements. Especially, fewer strong endogenous promoter elements available for foreign protein expression or construction of biosynthetic pathways were carefully evaluated in P. pastoris. Thus, it will be necessary to identify more available endogenous promoters from P. pastoris. Results Based on RNA-seq and LacZ reporter system, eight strong endogenous promoters contributing to higher transcriptional expression levels and β-galactosidase activities in three frequently-used media were screened out. Among them, the transcriptional expression level contributed by P0019, P0107, P0230, P0392, or P0785 was basically unchanged during the logarithmic phase and stationary phase of growth. And the transcriptional level contributed by P0208 or P0627 exhibited a growth-dependent characteristic (a lower expression level during the logarithmic phase and a higher expression level during the stationary phase). After 60 h growth, the β-galactosidase activity contributed by P0208, P0627, P0019, P0407, P0392, P0230, P0785, or P0107 was relatively lower than PGAP but higher than PACT1. To evaluate the availability of these promoters, several of them were randomly applied to a heterogenous β-carotene biosynthetic pathway in P. pastoris, and the highest yield of β-carotene from these mutants was up to 1.07 mg/g. In addition, simultaneously using the same promoter multiple times could result in a notable competitive effect, which might significantly lower the transcriptional expression level of the target gene. Conclusions The novel strong endogenous promoter identified in this study adds to the number of promoter elements available in P. pastoris. And the competitive effect observed here suggests that a careful pre-evaluation is needed when simultaneously and multiply using the same promoter in one yeast strain. This work also provides an effective strategy to identify more novel biological elements for engineering applications in P. pastoris. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-021-01648-6.


Background
As one of the typical methylotrophic yeasts, Pichia pastoris (also known as Komagataella phaffii) is not only widely used as an excellent host for heterologous protein expression [1] but also exhibits good potential for efficient biosynthesis of economic compounds such as nootkatone [2], carotenoids [3,4], xanthophylls [5], ricinoleic acid [6], hyaluronic acid [7] and dammarenediol II [8]. Although a lot of work had been carried out in P. pastoris, compared with Saccharomyces cerevisiae, fewer strong endogenous promoter elements available for foreign protein expression or construction of biosynthetic pathways were carefully investigated in P. pastoris [9].
Using methanol as a carbon source is one of the most favorable properties of P. pastoris. Several promoters from the methanol utilization (MUT) pathway were identified as the methanol-dependent strong promoters [10], such as P AOX1 (the promoter of alcohol oxidase 1) [11], P FLD1 (the promoter of formaldehyde dehydrogenase 1) [12], and P DAS (the promoter of dihydroxyacetone synthase) [13]. On the contrary, P PEX8 (the promoter of peroxisomal matrix protein) [14] or P AOX2 (the promoter of alcohol oxidase 2) [15] was classified as a weak promoter. Considering that these promoters are tightly regulated by methanol, making them useful for the efficient production of foreign proteins when using methanol as a carbon source and inducer [16]. P GAP is the most frequently used constitutive promoter that drives the expression of glyceraldehyde-3-phosphate dehydrogenase at a high level in the medium with glucose or glycerol but at a moderate level in the medium with methanol as a carbon source. Although in many cases, P AOX1 works more efficiently than P GAP for heterologous protein production [17,18]. Some cases also gave the opposite results. Waterham HR et al. proved the expression level of β-lactamase contributed by P GAP in the medium containing glucose was even higher than P AOX1 in the medium containing methanol [19]. Similar result was obtained with the P GAP regulated mammalian peptide transporters rPEPT2 by Doring F et al. [20]. A promoter library of mutated P GAP was generated, and it spanned an activity range between 0.6% and 19.6-fold of the wild-type P GAP activity [21]. As another well-known constitutive and strong promoter, the transcriptional activity of P TEF1 , which controls the expression of translation elongation factor 1 alpha [22], was also found to be similar to or higher than that of P GAP in batch cultures or fed-batch cultures (glucose or glycerol was used as carbon source). P GCW14 is a strong constitutive promoter discovered recently [23]. It exhibited stronger promoter activity than P TEF1 or P GAP under different carbon sources (glucose, glycerol, or methanol) when an enhanced green fluorescent protein (EGFP) was used as the reporter. Moreover, some other constitutive promoters, P ENO1 , P GPM1 , P HSP82 , P KAR2 , P PGK1 , P SSA4 , and P TPI1 [24], were also reported with little attention paid. The expression levels of EGFP contributed by these promoters were showed to be 10-80% of that of P GAP .
Although most studies have focused on strong promoters, it is worth noting that strong promoters are not suitable in some applications. For example, promoters with moderate expression levels are sometimes more desirable for transporter or chaperone expression. Thus, the studies on various promoters with different transcriptional activities should also be emphasized. In this study, based on the RNA-seq data and LacZ reporter system, eight novel endogenous promoters that exhibited strong transcriptional activities in three frequently-used media were screened out. A heterogenous β-carotene biosynthetic pathway was chosen to further evaluate the availability of these promoters. It was also found that simultaneously using the same promoter multiple times could result in a notable competitive effect, which might significantly lower the transcriptional expression level of the target gene.

Screening the potential strong promoters independent of carbon sources based on RNA-seq
In order to screen out strong promoter candidates from P. pastoris, the assessment of cell growth status was firstly carried out. As shown in Fig. 1A, the yeast cells gradually entered the stationary phase after 36 h growth in YPD, YPG, or YPM broth. Considering that the accumulation of protein expression or small-molecule compound has occurred chiefly during the stationary phase of yeast growth, total RNA isolation and subsequent RNA-seq were performed after 36 h growth.
As shown in Fig. 1B, the distribution of FPKM (referred to fragments per kilobase of transcript per million base pairs sequenced) values of genes was similar when the GS115 WT strain was cultured in three broths, and the FPKM values of most genes fall in the range of 10-50. The FPKM value of ACT1 (PAS_ chr3_1169), which is usually used as an endogenous reference gene (housekeeping gene) in yeast [25,26], was 624 in YPD, 858 in YPG, and 315 in YPM, respectively. It ranked 245th out of 4764 (YPD), 185th out of 4897 (YPG), and 473th out of 4410 (YPM) expressed genes (FPKM < 10 was defined as not expressed in the corresponding medium), indicating that P ACT1 could be considered as a strong promoter. To explore the potential strong promoters in all the above media, the genes with FPKM values five times higher than that of ACT1 were selected as the candidate genes. Finally, the promoters of PAS_chr1-4_0586 (GCW14), PAS_chr4_0627, PAS_chr2-2_0019, PAS_chr2-1_0437 (GAP), PAS_chr1-1_0407, PAS_chr2-2_0392, PAS_chr3_0230, PAS_chr2-2_0208, PAS_chr4_0785, and PAS_chr1-1_0107 were screened out (Additional file 1). For the subsequent description's convenience, the above genes were represented by GCW14, 0627, 0019, GAP, 0407, 0392, 0230, 0208, 0785, and 0107, respectively. In addition, the expression levels of these candidate genes were also confirmed by real-time quantitative PCR (RT-qPCR). As shown in Fig. 1C, all the RT-qPCR results were consistent with the RNA-seq data, suggesting these promoters could be used for further evaluation.

Comparing the transcriptional activities of promoter candidates
In order to compare the transcriptional activities of above-selected promoters, the β-galactosidase activities of transformants harboring vector pZeocin/ P XXX -lacZ-T ADH1 (P XXX indicates the name of candidate promoter) were measured after 60 h growth. The transcriptional levels of lacZ in all transformants were also tested by RT-qPCR. All experimental data was obtained from three independently screened colonies. As shown in Fig. 2, the β-galactosidase activity contributed by each selected promoter was higher than that by P ACT1 when the transformant was cultured in three mentioned broths, except P 0107 which exhibited a lower β-galactosidase activity compared to P ACT1 in YPM broth. In YPD or YPG broth, the activity of P GAP or P GCW14 was significantly higher than those of the rest promoters. In YPM broth, P GCW14 contributed the highest β-galactosidase activity among the allselected promoters, and P GAP , P 0208 , P 0019 , P 0392 , P 0230 , or P 0627 contributed a similar β-galactosidase activity.
In summary, among the ten selected promoters, the previously reported P GAP and P GCW14 showed strong transcriptional activities. The transcriptional activity of P 0019 , P 0107 , P 0208 , P 0230 , P 0392 , P 0407 , P 0627 , or P 0785 was between that of strong promoters P GAP and P ACT1 . It indicates that these promoter candidates could be recognized as strong endogenous promoters in P. pastoris.
Although the transcriptional activities of most selected promoters corresponded with the β-galactosidase activities at 60 h, there were still few exceptions, such as P 0407 and P 0230 . For example, the transcriptional level of P 0407 -lacZ was much higher than that of P GAP -lacZ when the relative transformants were cultured in YPG or YPM broth. However, the β-galactosidase activities controlled by P 0407 were less than half of that by P GAP . So was P 0230 -lacZ. We evaluated the copy number of lacZ controlled by each promoter, the results proved all clones harboring only one lacZ, eliminating the influence of transcriptional level by copy number (data not shown). It indicated that the transcriptional and protein level were not coordinated in some genes. The reason was not clear. To evaluate whether the above promoters are constitutive, the β-galactosidase activities of different transformants at 16 h (exponential phase) and 60 h (stationary phase) were measured. In view of P GAP being identified as a recognized constitutive promoter, it was chosen as a control in this study. As shown in Table 1, most promoter candidates, except P 0407 , P 0627 , and P 0208 , could be defined as constitutive promoters due to their relatively stable β-galactosidase activities at 16 and 60 h. The relative β-galactosidase activity driven by P 0208 , P 0407 , or P 0627 at 16 h was at least 16.5, 2.5 or 7.5 times lower than that at 60 h in three media, reminding us that these promoters might be associated with cell growth.
To test this possibility, the transcriptional levels of 0407, 0208, and 0627 genes in the GS115 WT strain were detected at the different growth stages. As shown in Additional file 2, the transcriptional levels of 0407, 0627 and 0208 genes increased along with the growth time of GS115 WT strain in YPD and YPG  media, indicating P 0407 , P 0208 and P 0627 could be considered as growth-dependent promoters. In YPM broth, they behaved not growth-dependent at 60 h. We suspect that it may be related to the depletion of carbon source (methanol). In our study, YPD (2%), YPG (2%) or YPM (1%) has 32.58, 33.36 and 12.26 mmol carbon atoms, respectively. At 60 h, the methanol had been depleted which may affect the cell metabolism, so the relative transcriptional level of 0627, 0407 and 0208 at 60 h was lower than that at 36 h, showing non-growthdependent characteristic. From 16 to 36 h, the relative transcriptional level was increased. Throughout the performance of 0208 and 0627 in three media for three time points, it demonstrates that they are growthdependent. In addition, the transcriptional level of 0407 was higher than that of ACT1 at 16 h in YPD, YPG and YPM (Additional file 2), indicating P 0407 was not a strictly growth-dependent promoter.
To further investigate the transcriptional properties of P 0208 and P 0627 , the GS115/P 0208 -lacZ-T ADH1 and GS115/P 0627 -lacZ-T ADH1 strains were cultured in YPD broth and sampled every 12 h from 12 h to 60 h. It was found that the relative transcriptional level of P 0208 was increased from 0.002 to 16 h to 26 at 60 h, while that of P 0627 was increased from 0.005 to 16 h to 40 at 60 h (Fig. 3A). Although the growth status of GS115/ P 0208 -lacZ-T ADH1 or GS115/P 0627 -lacZ-T ADH1 strain was similar, the β-galactosidase activity of GS115/P 0208 -lacZ-T ADH1 strain was lower than that of GS115/P 0627 -lacZ-T ADH1 strain at each sampling point (Fig. 3B), suggesting that P 0627 has good potential to be used in expressing the toxic proteins independent of methanol induction.

Evaluation of the promoter candidates through the heterogenous β-carotene biosynthetic pathway
According to the previous report, the optimized β-carotene biosynthetic pathway always includes four genes, crtE/tHMG1/crtYB/crtI [27]. In this study, six promoter candidates (P GAP , P GCW14 , P 0208, P 0019 , P 0230 , and P 0107 ) with different transcriptional activities were randomly selected and applied to the β-carotene biosynthetic pathway to further evaluate the availability of promoters. As shown in Fig. 4, a promoter library was first constructed by random linkage of two reverse-aligned promoters (Step 1). Then two vectors (pZeocin/P XXX -crtE-P YYY -tHMG1 and pG418/P XXX -crtYB-P YYY -crtI, P XXX and P YYY indicate the name of candidate promoters) containing the above four genes were constructed (Step 2) and transformed into the GS115 WT strain in turn (Step 3). The promoter activities were qualitatively observed by the color of the transformants on plates (Step 4) and then quantitively determined by liquid fermentation (Step 5).
Finally, fourteen transformants with significant color differences were screened out for further liquid fermentation (Fig. 5A). The result indicates that the darker the color of transformant, the higher the yield of β-carotene (Fig. 5B). And the yield of β-carotene from these yeast mutants was from 0.0123 mg/g DCW (Caro-1 strain) up to 1.07 mg/g DCW (Caro-14 strain). Fig. 3 The properties of P 0208 and P 0627 . A The relative transcriptional level of P 0208 -lacZ or P 0627 -lacZ after 12-60 h growth in YPD broth. The ACT1 gene was used as the reference gene. B The growth curve and β-galactosidase activity of GS115/P 0208 -lacZ-T ADH1 or GS115/P 0627 -lacZ-T ADH1 strain cultured in YPD broth. Error bars indicate the SD for samples tested in triplicate The promoters of four genes (crtE, tHMG1, crtYB, and crtI) in each strain were verified by PCR and the primers were shown in Additional file 3. The relationship between the transcriptional activity of each gene and the β-carotene production of the engineering strain was then analyzed. As shown in Fig. 5C, when both crtYB and crtI had a stronger promoter (similar to P GAP ), the β-carotene production of engineered strain was generally higher, and when crtYB had a relatively weaker promoter compared to P GAP , the β-carotene production of engineered strain was decreased. This result was consistent with the previous report that CrtYBp was the rate-limiting enzyme in β-carotene production [28]. Compared to the promoter combinations in the β-carotene biosynthetic pathway of Caro-13 and Caro-14 strain, all corresponding genes used the same promoter except crtE. The promoter of crtE was P GAP in the Caro-13 strain and P 0230 in the Caro-14 strain. Although the transcriptional activity of P 0230 was lower than that of P GAP at 60 h, the β-carotene production of the Caro-14 strain was much higher than that of the Caro-13 strain. This result suggested that not all genes should have stronger transcriptional levels to achieve higher β-carotene production. Balancing the expression level of each gene in the biosynthetic pathway could be a better choice. The existence of a significant competitive effect when one promoter was used multiple times in a yeast strain In the Caro-9 strain, the combination of exogenous genes and promoters was P GCW14 -tHMG1/P GAP -crtE/P GCW14 -crtYB/P GAP -crtI. In the Caro-13 strain, that was P GCW14 -tHMG1/P GAP -crtE/P GCW14 -crtI/P GAP -crtYB. Although both P GAP and P GCW14 were strong promoters, and their related β-galactosidase activities were similar (Table 1), the β-carotene production of Caro-9 strain was over 3-fold lower than that of Caro-13 strain (Fig. 6A, B). The RT-qPCR result at 16 h also showed that the transcriptional level of crtYB in the Caro-13 strain was over 3-fold higher than that in the Caro-9 strain (Fig. 6C). To figure out the reason, the transcriptional levels of four genes in the Caro-9 or Caro-13 strain were measured, respectively. Accidentally, Fig. 5 Evaluation of the promoter candidates through the β-carotene biosynthetic pathway. A The fourteen engineering strains with significant color differences grown on the YPD agar plate. B The production of β-carotene of the selected engineering strains. Error bars indicate the SD for samples tested in triplicate. C Heat map of the transcriptional levels (blue) and the β-carotene productions (orange) the transcriptional levels of two exogenous genes controlled by the same promoter behaved differently. As shown in Fig. 6C, although an endogenous gene (GAP) and two exogenous genes (crtE and crtI) were all driven by P GAP in the Caro-9 strain, the transcriptional levels of both exogenous genes were much lower than that of GAP, presenting a significant "competitive effect". In addition, the transcriptional levels of crtE and crtI were also different. So were GCW14, crtYB, and tHMG-1controlled by P GCW14 . A similar phenomenon was also observed in the Caro-13 strain (Fig. 6D). Such a "competitive effect" when the same promoter was used multiple times in one yeast strain resulted in a lower expression level of the related gene than expected. It might be an important factor limiting the β-carotene yields of engineering strains constructed in this study.

Discussion
The strong promoter is an indispensable element in the engineering retrofit and application of yeast. Due to the lack of basic research compared with S. cerevisiae, the availability of strong promoter is scarce in P. pastoris, which seriously restricts the subsequent application.
In this study, by analyzing the RNA-seq data of GS115 WT strain in the stationary phase under different media (YPD, YPG, and YPM), ten strong endogenous promoters (P GCW14 , P 0627 , P 0019 , P GAP , P 0407 , P 0392 , P 0230 , P 0208 , P 0785 , and P 0107 ) which were not affected by carbon sources were screened out. Except for P GCW14 and P GAP , eight of them were first reported here. Five (P 0019 , P 0392 , P 0230 , P 0785 , and P 0107 ) were identified as strong constitutive promoters, and two (P 0627 and P 0208 ) as growthdependent strong promoters. The growth-dependent transcriptional behavior of P 0627 or P 0208 makes it a good selection for the expression of toxic foreign proteins in P. pastoris. P 0407 showed a similar trend as P 0627 and P 0208 in β-galactosidase activity. However, the transcriptional level of P 0407 -lacZ was higher than that of ACT1 at 16 h, indicating this promoter was not a strictly growthdependent promoter. In addition, in our study, the activity of P GCW14 was lower than that of P GAP in YPD broth and similar to that of P GAP in YPG broth when measured with the β-galactosidase activity assay. These results were not consistent with the previous report that P GCW14 activity was much higher than P GAP by EGFP reporter in YPD or YPD broth [23]. However, the results of RT-qPCR and the β-galactosidase activity did not agree in all cases. For example, the transcriptional level of P 0407 -lacZ was much higher than that of P GAP -lacZ when the relative transformants were cultured in YPG or YPM broth. However, its related β-galactosidase activity was much lower than P GAP . P 0107 presented the lowest transcriptional level at 60 h, but its corresponding β-galactosidase activity was not the lowest (Fig. 2). These results agree with the suggestion of previous reports that it was not that the higher of transcriptional level, the higher of protein expression level [29,30].
In this study, a heterogenous biosynthetic pathway of β-carotene containing four genes was used to further evaluate the properties of six strong endogenous promoters (P GAP , P GCW14 , P 0208 , P 0019 , P 0230 , and P 0107 ). The Caro-14 strain contributed the highest yield of β-carotene (1.07 mg/g DCW). It is higher than the yield in S. cerevisiae containing integrated carotenogenic gene cassettes crtYB/I/E/tHMG1 (501 µg/g DCW) [31], suggesting that P. pastoris might have the advantages to be engineered as a carotene producer.
In addition, the competitive effect was first evaluated in this study. It was found that using the same promoter multiple times in one strain could lead to a significant decrease of the expression level of the target gene, reminding a careful evaluation is required before using a promoter in this way.
Some reports considered a 500 bp upstream sequence covering the whole promoter region of most genes in S. cerevisiae [32,33]. As the most known strong constitutive promoter P GAP in P. pastoris, its' sequence length is 477 bp. It should be noted that the intergenic region between GAP and its neighbor gene was 1576 bp. On the contrary, the sequence length of another wellknown strong inducible promoter, P AOX1 , is 939 bp in the commercial vector pPICZαC, mainly covering the intergenic region between AOX1 and its neighbor gene (998 bp). The study of P AOX1 showed that the region − 1055 to − 809 and − 653 to − 514 significantly affects the promoter activity [34]. In our study, the 500 bp upstream sequence of candidate gene was amplified to act as a promoter. However, the lengths of the intergenic region of some genes are more than 500 bp, which might cause incomplete promoter elements like P AOX1 , thereby affecting their normal functions. The corresponding intergenic regions of six selected promoters are over 500 bp except P 0208 , reminding the possibility that several key regulatory elements might not be included in the 500 bp sequence (Additional file 4). The promoter length would be appropriately extended in the follow-up study to explore whether it will significantly impact promoter activity.

Measurement of the yeast growth curves in different carbon source broths
The GS115 WT strain was pre-cultured in 5 mL YPD broth at 30 °C, 220 rpm for 24 h and followed by inoculating to an initial OD 600 of 0.1 in 250 mL shaking flasks containing 50 mL YPD (2% glucose), YPG (2% glycerol), or YPM (1% methanol) for 72 h. The optical absorbance (OD 600 ) of the culture was detected by an Evolution ™ 220 spectrophotometer (Thermo Fisher Scientific Inc., USA), and each medium had three replicates.

RNA-seq analysis
Three independent clones of the GS115 WT strain were incubated in 5 mL YPD broth for 24 h. The pre-cultured pellet was washed with presterilized water and transferred into YPD, YPG, or YPM broth for incubating. The yeast cells (total OD 600 of 16) were harvested by centrifugation (5000 g for 5 min, 4 °C) at 36 h when they entered the stationary phase. Total RNA was isolated using a Yeast RNA kit (Omega Bio-Tek Inc., USA) through the operation manual. The residuary genomic DNA was eliminated using a DNase I treatment (Takara Bio Inc., Japan). The qualified RNA samples determined with Bioanalyzer 2100 (Agilent Technologies, USA) were used for library construction with the sequencing platform BGISEQ-500 (BGI Shenzhen, China). Compared to the genome, the average comparison rate of the sample is 97.13%, and the average comparison rate of the compared gene set is 82.58%. All the generated raw sequencing reads were filtered to remove reads with adaptors, reads with more than 10% unknown bases, and low-quality reads. Clean reads were then obtained and stored in FASTQ format. The clean reads were mapped to the genome of GS115 by HISAT [35]. The software package RSEM [36] was used to quantify the gene expression level. The ratio of FPKM for candidate genes to that for ACT1 was used as the relative fold change for RNA-seq data treatment.

Real-time quantitative PCR (RT-qPCR) analysis
The methods for obtaining pure total RNA were mentioned above. After DNase I treatment, the absence of DNA contamination was confirmed by PCR using a pair of genomic primers, and the genome of the GS115 WT strain was used as a positive control. 800 ng of total RNA was used to reverse transcription to generate cDNA with HiScript II Reverse Transcriptase (Vazyme Biotech Co., Ltd., China). RT-qPCR was carried out on a Mastercycler ® ep gradient S instrument (Eppendorf, Hamburg, Germany) with 2 × ChemQ SYBR qPCR Mester Mix (Vazyme Biotech Co., Ltd., China). The ACT1 gene was chosen as an internal control to normalize the relative expression levels of target genes. All reactions were performed in three replicates. A ΔΔCt method was used for the relative fold change analysis [37].

Construction of the vectors containing P XXX -lacZ and yeast transformation
The genomic DNA of E. coli BL21 (DE3) was used as a template to amplify the open reading frame (ORF) of lacZ. DNA fragments of the integration site (AOX1 promoter, which helped the plasmid inserted into the chromosome) were amplified from GS115. ADH1 terminator (T ADH1 ) was amplified from the genomic DNA of S. cerevisiae BY4741. The fragment of ori and zeocin-resistant marker came from the commercial plasmid pPICZαC (Invitrogen Corporation, USA). All the fragments were amplified with Hieff Canace ™ high-fidelity DNA Polymerase (Yeasen, China) and then be linked by the homologous arm with pEASY ® -Basic Seamless Cloning and Assembly Kit (TransGen Biotech Co., Ltd, China) to construct a vector pZeocin/lacZ-T ADH1 .
The promoter fragments of candidates were amplified from the genomic DNA of P. pastoris GS115 by PCR and then inserted into the HindIII site of pZeocin/lacZ-T ADH1 with seamless cloning to yield vector pZeocin/P XXX -lacZ-T ADH1 (P XXX indicates the name of candidate promoters).
The linearized vector pZeocin/P XXX -lacZ-T ADH1 with PmeI (located in the middle of AOX1 promoter locus) digested were transferred into the GS115 cells by electro-transformation and integrated into the genome by homologous recombination [38]. The strains containing P XXX -lacZ were selected from the agar plates containing zeocin (0.1 mg/mL) and verified by PCR. The template came from colony or genomic DNA. The colony was boiled in water for 5 min before PCR and the genomic DNA was extracted as the previous report [39].

β-Galactosidase activity assay
The β-galactosidase activity was assayed according to the previous report with slight modification [40]. Briefly, the yeast cells (total OD 600 of 3) were collected. The supernatant was discarded by centrifuge, and the pellet was resuspended in 1 mL Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, and 1 mM MgSO 4 ) and diluted to make the final OD 600 of 0.2-0.6. Take suitable volumes of the above cells into a 2 mL Eppendorf tube, and make up to 1 mL with Z Buffer. Add 25 µL 0.1% SDS and 50 µL chloroform, vortex for 30 s, and incubate at 30 °C for 15 min. The reaction was started through the addition of 200 µL of 4 mg/mL onitrophenyl-β-D-galactopyranoside (ONPG) solution at 30 °C and stopped by the addition of 500 µL of 1 M NaCO 3 . After removal of cell debris by centrifugation, the OD 420 and OD 550 of the reaction solution were detected by Evolution ™ 220 spectrophotometer, and the relative activity of β-galactosidase was calculated using the formula below: T is the reaction time (min) and V is the volume (mL) of culture used for the assay.

Construction of promoter library in the β-carotene biosynthetic pathway
The vector pMRI-34-crtE-tHMG1 and vector pMRI-35-crtYB-crtI were the kind gifts from Hongwei Yu's lab [41]. The two gene cassettes with opposite directions of tHMG1/crtE or crtI/crtYB containing the promoter, ORF, and terminator were amplified from the above vectors. The AOX1 promoter (located in GS115 chromosome 4) and ENO2 promoter (located in GS115 chromosome 3) were chosen as the integration site. The ble (coding zeocin-resistant protein in E. coli and yeast) and kan (coding kanamycin-resistant protein in E. coli and G418-resistant protein in yeast, respectively) were chosen as screening markers. All those fragments were linked with the homologous arm by seamless cloning to yield vector pZeocin/crtE-tHMG1 and vector pG418/crtYB-crtI.
The candidate promoters (P 0107 , P GCW14 , P 0019 , P GAP , P 0208 , and P 0230 ) were amplified from the genomic DNA of GS115 by PCR. Two promoters in opposite directions were co-amplified by fusion PCR. Then a pool of combinations with two reversed promoters was obtained. Those promoters were inserted into BamHI/NotI sites of vector pZeocin/crtE-tHMG1 and vector pG418/crtYB-crtI with seamless cloning to yield vector pZeocin/P XXX -crtE-P YYY -tHMG1 and vector pG418/P XXX -crtYB-P YYY -crtI (P XXX and P YYY indicate the name of candidate promoters). The vector pZeocin/P XXX -crtE-P YYY -tHMG1 was linearized with PmeI and then transferred into the GS115 cell by electro-transformation. All the transformants grown on the agar plates containing zeocin were selected and co-cultured for the next round of electrotransformation with linearized pG418/P XXX -crtYB-P YYY -crtI by ScaI in ENO2 promoter. The transformants with different colors on the agar plates containing G418 (0.25 mg/mL) were picked out. The promoter combinations were verified by PCR.

Measurement of the β-carotene production
The extraction and analysis of β-carotene were performed according to a previous report [41]. Briefly, the β-carotene engineering strains were pre-cultured in 5 mL YPD broth at 30 °C, 220 rpm for 24 h. Precultures were inoculated to an initial OD 600 of 0.05 in 20 mL of YPD broth with 100 mL flasks and grown under the same condition for 120 h. Samples were taken every 24 h to determine the DCW and β-carotene yield. The carotenoids were co-extracted using hot HCl-acetone. The analyses of β-carotene were performed on an HPLC system (LC 20AT) equipped with a Supersil ODS2 C18 column (4.6mm × 250mm). The mobile phase is composed of acetonitrile, methanol, and isopropanol with a volume ratio of 5:3:2, followed by a flow rate of 1 mL/min at 40 °C. The UV/VIS signals were detected at 450 nm, and the analysis time of each sample is 40 min. The standard product of β-carotene was purchased from Solarbio (Beijing Solarbio Science & Technology Co., Ltd., China). The standard curve of concentration with the peak area was verified by triplicate.