All chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) unless otherwise specified. Mogrol standards were purchased from Sigma-Aldrich (St. Louis, MO, United States ). PrimeSTAR HS DNA polymerase used for target genes amplification was purchased from Takara (Japan). The GeneJET polymerase chain reaction (PCR) purification kit was purchased from Thermo Fisher Scientific (United States ), and SanPrep Spin Column & Collection Tube used for plasmid extraction was purchased from Sangon (Shanghai, China). The primers used in this study were synthesized by Genewiz (Suzhou, China). 5-Fluoroorotic acid (5-FOA), tyrosine, histidine, leucine, and uracil were purchased from Solarbio (Beijing, China).

The strains and plasmids used in the study are listed in Table 1. Escherichia coli JM109 was used for constructing recombinant plasmids. S. cerevisiae Y4 was employed as the original strain, a high-yield strain of squalene constructed previously. All strains constructed in this study are listed in Table 1. E. coli was cultured in Luria–Bertani medium (10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl) containing 100 mg/L ampicillin at 37°C and 220 rpm, and yeast strains were cultured in YPD medium (10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose) at 30°C and 220 rpm. Yeast transformants were selected on synthetic dextrose (SD-Ura) agar plates containing 6.9 g/L yeast nitrogen base without amino acids (YNB medium), 20 g/L glucose, 0.1 g/L amino acids (l-leucine, l-tryptophan, and l-histidine), and 20 g/L agar. Dimethyl sulfoxide with 1 g/ml 5-FOA was used for the counterselection of recombinants containing the URA3 marker. G418 sulfate (200 mg/L; Sigma) was supplemented into the medium as required.

SgCDS, SgEPH3, and CYP87D18 involved in mogrol synthesis were synthesized and codon-optimized by Genewiz. The CPRs coding AtCPR1 (GenBank ID: 828554) and AtCPR2 (GenBank ID: 829144) from Arabidopsis thaliana and SgCPR2 (GenBank ID: AYE89265.1) from S. grosvenorii (Zhao et al., 2018) were synthesized and codon-optimized by Genewiz. SrCPR1 from Stevia (Gold et al., 2018) was synthesized and codon-optimized by Genscript (Nanjing, China). CPRs coding CsCPR from Cucumber were provided by Prof. Sanwen Huang (Chinese Academy of Agricultural Sciences). All genetic modifications used the CRISPR-Cas9-mediated genome editing method and were performed as previously described (Gilbert et al., 2013). Genome integration loci (Reider Apel et al., 2017) and the corresponding single guide RNA (sgRNA) plasmids are listed in Supplementary Table S1. The target fragments were amplified using PCR, and the corresponding plasmid or S. cerevisiae genome was used as the template. The primers used in this experiment were synthesized by Genewiz. Overlaps (40–50 bp) of each adjacent fragment were used to implement homologous recombination in yeast. Yeast transformation was performed by the LiAc/SS carrier DNA/PEG method previously described (Gietz and Schiestl, 2007).

A green fluorescent protein (GFP) was fused with SgCDS, SgEPH3, CYP87D18, and AtCPR1, generating plasmids pY15-SgCDS-GFP, pY15-SgEPH3-GFP, pY15-CYP87D18-GFP, and pY15-AtCPR1-GFP, respectively, to determine the localization of heterologous enzymes involved in the biosynthesis pathway of mogrol. Most plant-derived P450 enzymes are located in the endoplasmic reticulum (ER) in S. cerevisiae (Arendt et al., 2017). An expression cassette P_GAP-mCherry-Linker-SCE12-TADH1 was inserted into pY15-CYP87D18-GFP and pY15-AtCPR1-GFP, resulting in plasmids pY15-CYP87D18-GFP-mCherry-SEC12 and pY15-AtCPR1-GFP-mCherry-SEC12, respectively.

CYP and CPR genes, promoter P _GAL1,10 , and terminators T _TDH3 and T _CYC1 were amplified using PCR to construct pESC-G418 series vectors, and the corresponding plasmid or S. cerevisiae genome was used as the template. These fragments with a 40-bp homology arm were inserted into multiple cloning sites of pESC-G418 using the Gibson Assembly^® Cloning Kit (NEB). Potential sgRNA targets were predicted at http://www.rgenome.net/cas-designer/ to design specific sgRNAs for corresponding targeted genes. All plasmids used in the study are listed in Table 2, and the corresponding primers are listed in Supplementary Table S2.

Gene localization of the mogrol biosynthesis pathway was observed in strain CEN-PK2-1C. Strain CEN-PK2-1C was transformed by plasmids pY15-SgCDS-GFP, pY15-SgEPH3-GFP, pY15-CYP87D18-GFP-mCherry-SEC12, and pY15-AtCPR1-GFP-mCherry-SEC12, thereby yielding strains LMOR001, LMOR002, LMOR003 and LMOR004. These strains were streaked onto an SD-Leu plate and cultured at 30°C for 2 days. Later, a single colony was picked into a 2 ml SD-Leu medium, cultured at 30°C and 220 rpm for 24 h, and washed twice with 1× phosphate-buffered saline (0.8% NaCl, 0.02% KCl, 0.144% Na₂HPO₄, and 0.024% KH₂PO₄, pH 7.4). Subsequently, 2-μL preparations were directly plated on slides. Strain LMOR001 culture was stained with Nile red solution in acetone (1:10, v/v; 1 mg/ml; Solarbio) and incubated for 60 min in the dark at 25°C to further confirm the specific distribution of SgCDS.

Fluorescence imaging was performed on a Nikon C-HGF microscope with a ×100 oil immersion objective. GFP fluorescence (excitation, 490 nm; emission, 530 nm), mCherry fluorescence (excitation, 580 nm; emission, 615 nm), and Nile red fluorescence (excitation, 561 nm; emission, 615 nm) were detected by microscopy. Image analysis was carried out on the Leica TCS SP8 software package and ImageJ (NIH).

The endogenous promoter of ERG7 and the gfp gene were inserted into plasmid pY13 to construct plasmid pY13-P_ERG7-GFP. Strain CEN-PK2-1C was transformed by plasmid pY13-P_ERG7-GFP to yield strain FMOR001. The mCherry protein shared the bidirectional promoter P _GAL1,10 with the dcas9 protein on plasmid pML104-dCas9. The corresponding sgRNA was integrated into plasmid pML104-dCas9 to construct plasmid pML104-dCas9-mCherry-1/2/3. Strain CEN-PK2-1C was transformed by plasmids pY13-P_ERG7-GFP and pML104-dCas9, thereby yielding strains FMOR002, FMOR003, and FMOR004, respectively. These strains were streaked onto an SD-Ura-His (FMOR002, FMOR003 and FMOR004) or SD-His (FMOR001) plate and cultured at 30°C for 2 days. Later, a single colony was picked into a 2 ml SD-Ura-His or SD-His medium and cultured at 30°C and 220 rpm for 24 h. The 2% (v/v) seed cultures were inoculated into a 200 μL SD-Ura-His or SD-His medium in a 96-well fluorescent plate and cultured at 30°C for 72 h. Strains with plasmids pML104-dCas9 and pY13-P_ERG7-GFP were cultivated in the SD-Ura-His medium, and the strain with pY13-P_ERG7-GFP was cultivated in the SD-His medium. All strains containing plasmid pML104-dcas9 needed 2%, 4%, 6%, 8%, 10%, and 20% galactose at 0 h. GFP fluorescence (excitation, 490 nm; emission, 530 nm), mCherry fluorescence (excitation, 580 nm; emission, 615 nm), and OD600 of strains were detected using a Cytation microplate reader (BioTek). Eq. 1 was used to calculate the repression fold of GFP fluorescence (θ) by different sgRNAs. θ = [ G F P / O D ] 1 − [ G F P / O D ] 0 [ G F P / O D ] x − [ G F P / O D ] 0 − 1 (1) where (GFP/OD)_x is the relative fluorescence intensity of GFP in strain FMOR002, FMOR003 or FMOR004; (GFP/OD)₀ is the relative fluorescence intensity of GFP in strain MOR001 (without GFP); and (GFP/OD)₁ is the relative fluorescence intensity of GFP in strain FMOR001.

The strains were streaked onto a YPD solid plate and cultured at 30°C for 2 days. A single colony was picked into a 2 ml YPD medium at 30°C and 220 rpm for 24 h. The 1% (v/v) cultures were inoculated into a 250 ml flask containing 50 ml YPD medium. Strains containing plasmid pESC-G418 needed 200 mg/L G418 to add to the YPD medium. All flask fermentation results represented the mean ± standard deviation (SD) of three independent experiments.

A 1 ml fermentation culture was collected and broken by a high-pressure homogenizer (FastPrep-24 5G, United States ) and extracted twice with ethyl acetate to measure the mogrol and squalene concentrations. For squalene analysis (Paramasivan and Mutturi, 2017), the samples were filtered using a 0.45 μm organic phase filtration membrane and injected into a high-performance liquid chromatography instrument (Agilent 1260 series). The column was a C18 ODS column (5 μm, 250 × 4.6 mm; Thermo Fisher Scientific), and the column temperature was 40°C. The pump flow rate was 1.6 ml/min, and the mobile phase was 100% acetonitrile. The injection volume was fixed to 10 μL. For mogrol analysis, the samples extracted with ethyl acetate were resuspended in 100 μL methanol after evaporation. Mogrol was quantified using liquid chromatography mass spectrometry (LC–MS) according to Itkin et al. (2016).

According to the mogrol synthetic pathway in plants, the initial step in mogrol biosynthesis is the epoxidation of 2, 3-oxidosqualene to form 2, 3; 22, 23-diepoxysqualene (the triterpenoid skeleton of mogrol). Then, 2, 3; 22, 23-diepoxysqualene is cycled into 24, 25-epoxycucurbiadienol by SgCDS. SgEPH3 catalyzes the unique hydroxylation of 24, 25-epoxycucurbiadienol to form trans-24, 25-dihydroxycucurbitadienol. Finally, CYP87D18 adds an oxygen atom to position C11 of 24, 25-dihydroxycucurbitadienol to form mogrol (Figure 1).

Therefore, to construct the de novo biosynthetic pathway of mogrol in S. cerevisiae, SgCDS, SgEPH3, and CYP87D18 from S. grosvenorii and the CPR gene AtCPR1 from A. thaliana were integrated into the chromosome of strain Y4 (a high-yield strain of squalene constructed previously), yielding strain MOR001. The expression of these genes was under the control of the galactose-inducible promoter P _GAL1,10 . Synthetic genes were first employed as templates and then the genes were amplified to investigate the expression and distribution of SgCDS, SgEPH3, CPY87D18, and AtCPR1. Genes were integrated into plasmids, and the gfp gene was fused downstream of them (Figure 2A). These plasmids were transformed into strain MOR001, thereby yielding strains LMOR001, LMOR002, LMOR003 and LMOR004. Comparing the fluorescent images of SgCDS to the bright-field images, the green fluorescent spots showed a point-like distribution. These apparent traits showed that these spots are like lipid drops (LDs), the primary storage sites for neutral lipids in cells (Hammer and Avalos, 2017). Then, cells were stained with Nile red, a dye that can directly stain LDs (Liu et al., 2020). The spots in Nile red images exactly overlapped with the green fluorescent spots. It was concluded that SgCDS was specifically localized in LDs (Figure 2B), which was similar to the cycloartenol synthase from Arabidopsis thaliana (Milla et al., 2002) and α-amyrin synthase from Malus × domestica (Yu et al., 2020). According to the merged results of the colocalization of AtCPR1 and ER, the localization of GFP fluorescence was consistent with sec12p, suggesting that AtCPR1 was an ER-localized protein in S. cerevisiae (Figure 2C). Besides, laser scanning confocal microscopy (LSCM) analysis revealed that the GFP fluorescence of SgEPH3 filled the entire cytoplasm of S. cerevisiae, showing that SgEPH3 was located in the cytoplasm (Figure 2D). Furthermore, the GFP fluorescence of CYP87D18 was weak, indicating that the enzyme activity of CYP87D18 was a bottleneck in the biosynthesis pathway of mogrol (Figure 2E). These results proved that heterologous SgCDS, SgEPH3, CPY87D18 and AtCPR1 genes were successfully expressed in strain MOR001.

After strain MOR001 cultivation, a peak with a mass-to-charge ratio (m/z) value of 521.389 was detected using LC–MS in strain MOR001, in accordance with the m/z values of mogrol standard (Figures 3A,B). Mogrol could only be detected when the inducer galactose was added to the medium. These results demonstrated that the de novo biosynthetic pathway of mogrol was successfully constructed in strain MOR001. However, the mogrol titer in strain MOR001 was only 19.1 ng/L. The metabolic network was divided into three modules, including the squalene, 2, 3; 22, 23-diepoxysqualene, and mogrol synthesis modules, to synergistically regulate the metabolic network of mogrol (Figure 1). For the squalene synthesis module, the precursor acetyl-CoA supply was enhanced to facilitate squalene synthesis. In the cytoplasm, CIT2 and MLS1 involved in the glyoxylate pathway consumed intracellular acetyl-CoA. Therefore, CIT2 and MLS1 were knocked out to reduce the consumption of the precursor acetyl-CoA using the CRISPR–Cas9 system, generating strain MOR002. However, the mogrol titer in strain MOR002 only reached 25.9 ng/L, which was 1.3-fold higher than that of MOR001 (Figure 3D). These results indicated that the intracellular content of acetyl-CoA and squalene was sufficient to supply mogrol synthesis, which may be due to the high content of precursor squalene in the original strain Y4.

In the 2, 3; 22, 23-diepoxysqualene synthesis module, Erg7p (lanosterol synthase) catalyzed the cyclization of (3S)-2, 3-oxidosqualene to lanosterol and further flowed into the synthesis pathway of ergosterol, one of the main components of the plasma membrane (Nes, 2011; Jordá and Puig, 2020). In general, 2,3-oxidosqualene prefers to be cyclized by ERG7 rather than oxidized into 2, 3; 22, 23-diepoxysqualene (Corey and Gross, 1967; Shan et al., 2005). Therefore, knocking out or inhibiting the activity of ERG7 is crucial for 2, 3; 22, 23-diepoxysqualene synthesis. However, directly blocking the metabolic flux to ergosterol would affect cell growth and require additional supplementation of ergosterol and Tween 80 in the medium to maintain cell growth, which was not cost-saving and conducive to industrial development (Itkin et al., 2016; Qiao et al., 2019).

The CRISPRi system is widely used to inhibit competing pathways in metabolic networks of various products (Gilbert et al., 2013; Jensen et al., 2017; Wu et al., 2018). Therefore, the CRISPRi system was used to inhibit ERG7 expression, thereby redirecting the metabolic flux toward mogrol biosynthesis, enabling the decoupling of cell growth and product synthesis. First, the inhibitory targets of the CRISPRi system on the promoter of ERG7 (P _ERG7 ) and their corresponding inhibitory effects were screened using the gfp reporter gene (Figure 4A). Control plasmid pY13-P_ERG7-GFP containing the reporter gene gfp was constructed to characterize the transcription strength of promoter P _ERG7 . Three sgRNAs targeting different positions of the promoter P _ERG7 were inserted into plasmid pML104-dCas9-mCherry, which contained a tandem expression cassette dCas9-mCherry under the control of galactose-inducible promoter P _GAL1,10 , thereby obtaining plasmids pML104-dCas9-mCherry-1/2/3 (Figure 4B). The dCas9 expression level in plasmid pML104-dCas9-mCherry-1/2/3 was characterized according to the fluorescence intensity of mCherry. The control plasmid pY13-P_ERG7-GFP was transformed into the wild-type strain CEN-PK2-1C, yielding strain FMOR001. Plasmids pY13-P_ERG7-GFP and pML104-dCas9-mCherry-1/2/3 were cotransformed into the wild-type strain CEN-PK2-1C, thereby yielding strain FMOR002, FMOR003, and FMOR004 (Figure 4B). Finally, the fluorescence intensity of GFP and mCherry in strains FMOR001, FMOR002, FMOR003, and FMOR004 was measured. The results showed that when the inducer galactose concentration was 6 g/L, the repression fold of GFP fluorescence (θ) in strains FMOR002, FMOR003 and FMOR004 were 6.1-, 7.4-, and 3.5-fold, respectively (Figure 4C, Supplementary Figure S1). When the galactose concentration was >6 g/L, the repression fold of GFP fluorescence (θ) by different sgRNAs was gradually decreased (Figure 4C). In addition, the expression level of dCas9 in strains FMOR004 was higher than FMOR002 and FMOR003, but the repression fold of GFP fluorescence (θ) in strains FMOR004 were lowest (Figure 4D). This result indicating that sgRNA rather than the dCas9 protein was the most essential for the efficiency of the CRISPRi system.

Subsequently, dcas9-sgRNA expression cassettes in plasmids pML104-dCas9-mCherry-1/2/3 were cloned and integrated into the genome of strain MOR002 to repress ERG7 expression, thereby yielding strains MOR003, MOR004, and MOR005, respectively. The growth of strains MOR003, MOR004, and MOR005 was similar to strain MOR002 (Figure 4E), demonstrating that the CRISPRi system could inhibit ERG7 expression without affecting cell growth. Meanwhile, the mogrol titer in strains MOR003, MOR004, and MOR005 reached 43, 66, and 30 ng/L, increased by 53%, 135%, and 7%, respectively, compared to that of MOR002 (Figure 4F), consistent with the GFP fluorescence repression fold of different sgRNAs.

To further improve mogrol production, the metabolic flux of the mogrol synthesis module was engineered. First, the precursor 2, 3-oxidosqualene flow was “pulled” into the mogrol synthesis module rather than cell growth by overexpressing ERG1 (squalene epoxidase), which catalyzed squalene to 2, 3-oxidosqualene and 2, 3; 22, 23-diepoxysqualene. Another copy of ERG1 was integrated into the genome of strain MOR002, and GAL80 (a transcriptional factor) (Pilauri et al., 2005) was simultaneously knocked out, yielding strain MOR006. LC–MS results showed that the mogrol titer in strain MOR006 reached 48 ng/L, which showed a further 240% increase than that of MOR002 (Figure 5A). When another copy of ERG1 was integrated into the genome of strains MOR003, MOR004 and MOR005, the mogrol titer in the resulting strains MOR007, MOR008 and MOR009 were 13.9-, 33-, and 4.8-fold as high as that of strain MOR006, reached 670, 1600, and 230 ng/L, respectively (Figure 5A). These results indicated that only inhibiting the synthetic pathway of lanosterol and synergistically “pulling” the metabolic flux flow into the 2, 3; 22, 23-diepoxysqualene could maximize mogrol production. Subsequently, the copy number of ERG1 in strain MOR008 was further increased to improve mogrol production. When the copy number of ERG1 increased to three copies (MOR010), the mogrol titer was further increased to 5.67 μg/L. When the copy number of ERG1 reached four copies (MOR011), the mogrol titer decreased to 0.68 μg/L (Figure 5C). Finally, the titer of the intermediate metabolite squalene was also measured in strains MOR004, MOR008, MOR010, and MOR011. The results showed that the squalene titer in strains MOR008, MOR010 and MOR011 at 120 h reached 617, 1002.9, and 971.4 mg/L, which was 1.1-, 1.86- and 1.81-fold higher respectively, compared to that of the strain MOR004 (Figure 5D). The biomass of MOR008, MOR010 and MOR011 was significantly improved compared to MOR004, which may be due to squalene being an essential precursor for cells to synthesize the plasma membrane (Figure 5B). These results demonstrated that ERG1 increased mogrol production by “pulling” the metabolic flux toward the biosynthesis pathway of mogrol.

Most plant-derived P450 enzymes (CYP) are located in the ER and are often considered as the rate-limiting enzymes in the biosynthetic pathway (Hausjell et al., 2018). They rely on CPRs to obtain electrons from NAD(P)H for subsequent catalytic reactions (Sadeghi and Gilardi, 2013). Many studies showed that different sources of CPRs may affect CYP activity, and the compatibility of native CPRs with CYPs in heterologous hosts is often not optimal (Theron et al., 2019). Hence, screening CPRs from different sources is an effective strategy to increase the activity of P450 enzymes.

Therefore, CYP87D18 was first inserted into a high-copy plasmid pESC-G418 to test whether increasing its expression would improve mogrol production, yielding plasmid pESC-G418-CYP. Plasmid pESC-G418-CYP was transformed to MOR004 and MOR010 to construct strains MOR012 and MOR013, respectively. Compared to MOR004, CYP87D18 expression on a multicopy plasmid significantly increased mogrol production (1.05 μg/L; Figure 6A). This result revealed that increasing P450 enzyme expression was beneficial to mogrol synthesis, whereas it decreased in strain MOR013 (1.64 μg/L) compared to strain MOR010 (5.67 μg/L).

Subsequently, CYP87D18 was coexpressed in combination with five different CPRs in plasmid pESC-G418, including AtCPR1 and AtCPR2 from A. thaliana, SgCPR2 from S. grosvenorii (Zhao et al., 2018), CsCPR from Cucumber, and SrCPR1 from Stevia (Gold et al., 2018). The resulting plasmids were transformed into strain MOR010, thereby obtaining strains MOR014, MOR015, MOR016, MOR017, and MOR018, respectively. Fermentation data showed that the mogrol titer in strain MOR015 reached 9.1 μg/L, which was 1.6-fold higher than that of the strain MOR010. In contrast, the mogrol titer in strains MOR014 MOR016, MOR017 and MOR018 was lower than that of MOR010 (Figure 6B). These results demonstrated that the selection of appropriate CPR is crucial for the catalytic efficiency of CYPs in heterologous hosts. Additionally, the biomass of these strains was only 70% of MOR010 (Figure 6C), which may be due to the additional burden on the organism caused by that the overexpression of the second recombinant protein (Hausjell et al., 2018).