Introduction

Triterpene synthases catalyze the cyclization of 2,3-oxidosqualene to triterpenes and sterols by diverse cationic intermediates. 2,3-Oxidosqualene is mainly cyclized to triterpenes based on the dammarenyl cation via the chair-chair-chair (C–C-C) conformation whereas sterols and minor triterpenes are derived from the protosteryl cation through the chair-boat–chair (C-B-C) conformation (Chin et al. 1973; Lou et al. 1991; Thimmappa et al. 2014). The dammarenyl cation could directly generate dammarane-type tetracyclic triterpenoids which undergo the ring expansion to yield the baccharenyl cation. Typical pentacyclic triterpenoids such as lupane, oleanane, ursane-type, and friedelane-type are derived from the baccharenyl cation while rare pentacyclic triterpenoids could be generated via the protosteryl cation (Fig. 1) (Xu et al. 2004). These cyclized skeletons could be further decorated by cytochrome P450 monooxygenases (P450s), cytochrome P450 reductases (CPRs), and UDP-dependent glycosyltransferases (UGTs), thereby generating numerous pentacyclic triterpenoids. These baccharenyl cation-derived pentacyclic triterpenoids showed various biological activities. For instance, lupeol, α-amyrin, and β-amyrin exhibit various biological activities including antitumor (Barros et al. 2011), anti-inflammatory (Lima et al. 2014), anxiolytic (Aragão et al. 2006), neuroprotective (Wu et al. 2020b), and hepatoprotective effects (Oliveira et al. 2004; Saleem 2009). Furthermore, friedelin was reported to be the precursor of antitumoral quinonemethide triterpenoids (Corsino et al. 2000).

Fig. 1
figure 1

Biosynthesis of triterpenes and sterols catalyzed by 2,3-oxidosqualene cyclases. 2,3-Oxidosqualene is the common precursor for the biosynthesis of triterpenes and sterols. Pentacyclic triterpenes are mainly derived from the baccharenyl cation in the “chair–chair–chair” conformation. The baccharenyl cation was generated by the dammarenyl cation via the ring expansion, which leads to the formation of a large diversity of pentacyclic triterpenoids. Sterols like cycloartenol and lanosterol are mainly generated by the protosteryl cation in the “chair–boat–chair” conformation

Full size image

Currently, the production of various triterpenoids was mainly achieved by direct extraction from plant tissues (Lima et al. 2015; Lu et al. 2018). However, the low abundance and unstable supply limited the sustainable production of triterpenoids (Sun et al. 2019). Microbial synthesis of triterpenoids is an attractive and promising alternative because of the short producing cycle and land-saving culture conditions (Li et al. 2019; Noushahi et al. 2022). Well-studied microbes such as Saccharomyces cerevisiae and Yarrowia lipolytica are good candidate hosts for microbial production of various triterpenoids as they provide the precursor 2,3-oxidosqualene by its endogenous mevalonate (MVA) pathway and membrane structure for the expression of corresponding enzymes P450s and CPRs involved in triterpenoid biosynthetic pathways (Cho et al. 2022; Guo et al. 2020; Zhao and Li 2018).

Metabolic engineering strategies have been employed to improve the production of triterpenoids in microbes (Li et al. 2016). By enhancing the mevalonate (MVA) pathway and triterpenoid biosynthetic pathway, the yield of dammarane-type triterpenoids in S. cerevisiae has reached the 10 g/L scale (Wang et al. 2019). Furthermore, several high 2,3-oxidosqualene-producing yeast strains have been successfully engineered to produce various triterpenoids by modifying the endogenous lanosterol synthase (Erg7p) (Guo et al. 2022a, b). Based on these strains, the production of various triterpenoids was increased to above 5 g/L in the bioreactor (Guo et al. 2022a). However, the production of typical pentacyclic triterpenoids is relatively lower than dammarane-type triterpenoids when similar strategies were applied to the same chassis (Guo et al. 2020; Huang et al. 2019; Wang et al. 2019). Hence, the major challenge for improving the production of typical pentacyclic triterpenoids is the poor catalytic activity of corresponding pentacyclic triterpene synthases (PTSs), and the strategy for improving the activity of typical PTSs is still lacking.

The key amino acid residue motifs such as DCTAE and MXXXXR motifs have been characterized in 2,3-oxidosqualene cyclases (Chen et al. 2021; Wang et al. 2022). The MXXXXR motif is generally characterized as an essential characteristic of typical PTSs, and the active sites of the MXXXXR motif are considered highly conserved (Fig. 2). In the case of β-amyrin synthase from Euphorbia tirucalli (EtbAS), the active-site residues Trp257 and Tyr259 involved in the M256WCYCR261 motif were reported to stabilize the oleanyl cation and the baccharenyl cation via cation-π interaction, respectively (Ito et al. 2016). For lupeol synthases from Olea europaea (OEW), a previous study demonstrated that the active-site residues Leu256 and Tyr258 involved in the M255LCYCR260 might stabilize the lupenyl cation (Kushiro et al. 2000). Modification of these active sites could affect the cation-π interaction, thereby leading to altered end products. In addition, the Trp260 residue in the α-amyrin synthase of Camellia sasanqua was proposed to stabilize the ursanyl cation through cation − π interactions and CH − π interactions (Huang et al. 2022). In contrast, the non-active-site residues such as Met255, Cys257, and Cys259 in the MXXXXR motif of EtbAS are not highly conserved in typical PTSs. Engineering the residue such as Cys262 in the M258WCYCR263 motif of the β-amyrin synthase from Panax ginseng did not alter the product selectivity (Kushiro et al. 2000). However, the influence of these residues on the catalytic activity has not been characterized to date. Recently, two PTSs in Iris tectorum ItOSC2 and ItOSC6 have been characterized to possess non-canonical motifs F255LALAR260 and L255MVLAR260, respectively (Wu et al. 2020a). This study also demonstrated that these non-active-site residues might function as second shell residues in the non-canonical motif and mutation of these non-active sites could influence the catalytic activity (Wu et al. 2020a). As the biosynthesis of the typical pentacyclic triterpenoids is derived from the baccharenyl cation, engineering the non-active-site residues in the MXXXXR motif of the PTSs might be a general strategy for improving the catalytic activities.

Fig. 2
figure 2

Phylogenetic analysis combined with the MXXXXR motif of characterized pentacyclic triterpene synthases (PTSs). Sequences were selected from GenBank based on their authentication in the literature (unless otherwise indicated) which was built using the neighbor-joining method. Distances were computed using the Poisson correction method. The phylogenetic tree was constructed using 1000 bootstrap replicates. The percentage of trees in which the associated taxa clustered together was shown next to the branches. PTSs could be functionally grouped into lupeol synthases (red), β-amyrin synthases (purple), multifunctional α-amyrin synthases (orange), taraxerol synthase (pink), and friedelin synthase (yellow). Evolutionary analyses were conducted in the software MEGA11. The scale bar indicates 0.2 amino acid substitutions per site

Full size image

In this study, we employed the above strategy to improve the activity of typical PTSs. Taking lupeol synthase, β-amyrin synthase, and α-amyrin synthase as examples, we first investigated the function of the non-active-sites residues in the MXXXXR motif by the computational model. We further investigated the influence of these residues on the catalytic activity of typical PTSs by site-directed saturation mutagenesis, which was characterized by the production of the corresponding products. Finally, we investigated whether the combinatorial mutants based on the beneficial mutations could influence the catalytic activity of typical PTSs. This study provides a general and novel target that could improve the activity of typical PTSs.

Materials and methods

Chemicals, strains, and medium

All chemicals and reagents used were obtained from the manufacturer Sinopharm Chemical Reagent Co., Ltd. 601 (Beijing, China). Standards of lupeol (purity ≥ 98%), β-amyrin (purity ≥ 95%), and α-amyrin (purity ≥ 95%) were supplied from Shanghai yuanye, Bio-Technology Co., Ltd. (China). Primers were synthesized by GENEWIZ Bio Inc. (Suzhou, China). Strain SP1 derived from S. cerevisiae BY4742-trp was used as the initial strain for the expression of all PTSs in this study. Gene cloning was carried out in Escherichia coli XL10-gold. All plasmids and strains used in this study are listed in Table S1 and S2, respectively. S. cerevisiae strains were cultured at 30 °C whereas E. coli strains were grown at 37 °C in an incubator shaker at 200 rpm. YPD medium consists of 10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose (20 g/L agar, only for solid medium). Synthetic complete (SC) medium consists of 6.7 g/L yeast nitrogen base without amino acids and with ammonium sulfate, 20 g/L glucose, yeast synthetic drop-out medium supplements without uracil, histidine, leucine, and tryptophan (20 g/L agar, only for solid medium). LB medium consists of 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl (20 g/L agar, only for solid medium). For E. coli, 50 µg/mL of ampicillin was routinely supplemented in broth or solid media, respectively.

Plasmid construction

Gibson assembly technique was used for the construction of all plasmids. All plasmids and strains used in this study are described in Table S1 and S2, respectively. Primers used in this study are shown in Table S3. GgbAS (encoding β-amyrin synthase from Glycyrrhiza glabra, GenBank: AB037203), OEW (encoding lupeol synthase from Olea europaea, GenBank no. AB025343), and MdOSC1 genes (encoding α-amyrin synthase from Malus × domestica, GenBank no. FJ032006.1) were generated by PCR-based gene synthesis, and the resulting fragments were ligated into the pCfB255, resulting in the plasmids p056-LOXP-GgbAS1-URA, p056-LOXP-OEW-URA, and p056-LOXP-MdOSC1-URA. The GgbAS, OEW, and MdOSC1 genes were expressed under the control of a constitutive PPGK1 promoter and the TADH1 terminator. To mutate the GgbAS, OEW, and MdOSC1 genes, primers containing corresponding mutations were designed for PCR amplification. Plasmids p056-LOXP-GgbAS1-URA, p056-LOXP-OEW-URA, and p056-LOXP-MdOSC1-URA were used as the template for PCR. To achieve multi-site-directed mutagenesis, plasmids p056-LOXP-GgbAS1-URA, p056-LOXP-OEW-URA, and p056-LOXP-MdOSC1-URA containing the single-site mutants were used as a template for PCR amplification. PrimerSTAR DNA polymerase was employed to amplify genes and mutants. The resulting linear fragments were chemically transformed into E. coli XL10-gold for construction of target plasmids. All mutant genes were verified by DNA sequencing. The genes of GgbAS, OEW, and MdOSC1 were synthesized by approximately 60 bp oligonucleotides with 20 bp overlaps.

Strain construction

The corresponding plasmids were transformed into strain SP1 for generating the desired strains. The standard lithium acetate method was applied to transform yeast cells (Daniel Gietz and Woods 2002). The desired transformants of S. cerevisiae were selected on synthetic complete (SC) media with appropriate amino acids or uracil. Before the transformation, the expression cassette of genes in each plasmid was achieved by PCR amplification and then purified from agarose gel. The desired transformants were verified by colony PCR and sequencing. The PCR mixtures contained 0.1 mM each of the four deoxynucleoside triphosphates (dNTPs), 20 ng of the plasmid template, 2 × buffer, 0.1 µM of each primer, 2 U of Phanta MAX Super-Fidelity DNA Polymerase, and double-distilled water to a final volume of 50 µL. The PCR conditions were 95 °C for 3 min, 30 cycles of 95 °C for 10 s, 56 °C for 30 s, 72 °C for 3 min, and a final extension at 72 °C for 30 min.

Shake flask cultivation

Pre-cultures were prepared by inoculating a single colony from the fresh agar plate. To prepare the shake flask cultivation, pre-cultures were transferred into the main culture to an initial OD600 of 0.4. All the main culture were grown in 250-mL shake flasks containing 25 mL (10% (v/v) of the flask). The cultures were incubated at 30 °C and 220 rpm in a rotary shaker for 72 h. Samples were regularly taken for analysis of various triterpenes and measurement of optical densities. All optical densities at 600 nm (OD600) were monitored by using the spectrophotometer. Cell biomass was calculated with an OD600/dry cell weight conversion factor of 0.21 g DCW/L.

Homology modeling and bioinformatics analysis

The protein structure models of the GgbAS, OEW, and MdOSC1 were predicted by using the online bioinformatics tool, iterative threading assembly refinement (I-TASSER). Each model with the highest C-score was chosen for molecular docking. The molecular docking was performed using AutoDock 1.5.6 with the corresponding substrate (lupeol, PubChem CID: 259,846; β-amyrin, PubChem CID: 73,145; α-amyrin, PubChem CID: 73,170) as the ligand. The grid maps were calculated by the AutoGrid program. The box size was set to enveloping ligands. The active sites in the motifs MXXXXR and DCTAE of GgbAS, OEW, and MdOSC1 were kept in the center of the grid box with an 80 × 80 × 80-point. The grid spacing was set to 0.375 Å. The Lamarck’s genetic algorithm (LGA) was used to optimize the conformation of lupeol, α-amyrin, or β-amyrin in the binding pocket. The PyMOL viewer was utilized to check the interaction between key amino acid residues and ligands in the models.

Analysis of the triterpenes

UHPLC system was applied for the quantification of various triterpenes. A C18 column was used for chromatographic separation. The chromatographic conditions were applied as described previously (Czarnotta et al. 2017). The chromatographic conditions were as follows: 10 µL of the sample was used for measurement; the column oven temperature was set to 40 ℃. Acetonitrile (ACN), methanol, and milli-Q water with 0.2% (v/v) formic acid were used as the mobile phase at a flow rate of 1 mL/min. The ratio of acetonitrile, methanol, and milli-Q water with 0.2% (v/v) formic acid was set to 90:9:1 for 1 min after injection. The total run time was 23 min. For calibration, standard curves of lupeol, β-amyrin, and α-amyrin were prepared with concentrations ranging from 1 to 500 mg/L. The samples were prepared as previous procedures with some modifications. A 500 µL cell suspension aliquots were taken from the shake flask culture and transferred into a 2-mL screwed cap tube. Glass beads (250 µL, 0.5 mm), 1 M HCl (50 µL), and hexane (500 µL) were added and homogenized for 5 min in a Mini-BeadBeater. The initial homogenate was clarified by centrifugation at 13,000 rpm for 15 min at room temperature. The upper hexane layer (300 µL) was evaporated and resuspended in 100 µL methanol. The sample was injected directly into the UHPLC system (Figure S6).

Results

Engineering lupeol synthase for increasing lupeol production

To rapidly demonstrate the changes in the catalytic activity of various PTSs, a squalene-producing strain SP1, previously constructed based on S. cerevisiae BY4742-Trp, was used as the chassis to investigate the activity of typical PTSs. One copy of the gene encoding the target PTS was integrated into the genome of the chassis to ensure constitutive expression throughout the cultivation. To assess the validity of our approach, saturation mutagenesis was applied to three non-active sites in the MXXXXR motif of each PTS. The catalytic activity of the mutant and wild-type PTSs was characterized by the titer and specific titer of the corresponding pentacyclic triterpene in the strain.

Lupeol is derived from the lupenyl cation which is first generated through the ring expansion of the baccharenyl cation. Lupeol synthase has been identified from numerous plants such as Arabidopsis thaliana, Betula platyphylla, Bruguiera gymnorrhiza, and Olea europaea (Fig. 2). The previous report demonstrated that strain harboring lupeol synthase from Olea europaea (OEW) could generate higher titer of lupeol than that of strain harboring most of other lupeol synthases (Guo et al. 2022a). Here, we first investigated whether engineering the non-active-site residues in the MXXXXR motif could influence the activity of OEW. Molecular docking between the receptor OEW and the ligand lupeol was employed to predict the role of the non-active-site residues Met255, Cys257, and Cys259 of the MXXXXR motif in the OEW. The best-docked model was visualized with the PyMOL viewer. As shown in Fig. 3a and Fig. S1, the non-active-site and active-site amino acid residues were labeled in pink and green, respectively. The lupeol was labeled in yellow. The residue Tyr258 is located near the lupeol. The residues Met255, Cys257, and Cys259 are spatially proximal to the residues Leu256 and Tyr258 which might function as the shell residues. The results of the prediction model suggested that the non-active-site residues might function as the shell residues surrounding the active sites of Leu256 and Tyr258.

Fig. 3
figure 3

Modification of the non-active-site residues of lupeol synthase from Olea europaea (OEW). a Molecular docking model of OEW and lupeol using PyMOL software. b Influence on the titers of lupeol, specific titer, and cell biomass when saturation mutagenesis was applied on the sites Met255 (b), Cys257 (c), and Cys259 (d) of OEW

Full size image

Furthermore, site-directed saturation mutagenesis was employed to engineer the aforementioned sites in the MXXXXR motif. As shown in Fig. 3b, strain SP1 harboring OEWM255L, OEWM255V, OEWM255I, OEWM255Y, and OEWM255C displayed higher titer and specific titer of lupeol compared with these of the wild-type OEW after 72 h cultivation. To be noted, the strain harboring OEWM255L produced approximately 50 mg/L and 14 mg/g biomass, which are 2.7-fold and 2.3-fold higher than these of the strain harboring the wild-type OEW (Fig. 3b). For the strain SP1 carrying OEWM255A, OEWM255W, OEWM255S, OEWM255Q, OEWM255G, and OEWM255K, the titer and specific titer of lupeol were significantly decreased. The production of lupeol was not detected in the strain SP1 harboring OEWM255H, OEWM255P, OEWM255E, OEWM255R, OEWM255D, and OEWM255N. The OD600 of strain SP1 harboring the mutant OEW did not show a significant change compared with that of the wild-type OEW. These results demonstrated that modification of the Met255 could affect the catalytic activity of OEW. Moreover, the five mutants showed higher activity than that of the wild-type OEW.

We then investigated whether the modification of Cys257 and Cys259 in the MXXXXR motif could influence the activity of OEW. To this end, single-site saturation mutagenesis was applied to Cys257 and Cys259. For the Cys257, strain SP1 harboring OEWC257X did not show an increase in the titer and specific titer of lupeol despite that the OD600 of these mutants variably increased (Fig. 3c). Moreover, the production of lupeol in the strain SP1 harboring OEWC257H, OEWC257K, OEWC257Y, OEWC257V, OEWC257D, and OEWC257I was not detected. For the Cys259, most mutants lowered the production of lupeol compared with that of the strain SP1 harboring wild-type OEW, and the strain SP1 harboring OEWC257X showed similar cell biomass. However, strain SP1 harboring OEWC259A and OEWC259G accumulated 65 mg/L and 37.7 mg/L lupeol with the specific titer of 15.5 mg/g biomass and 9.4 mg/g biomass, respectively, which displayed a 3.5-fold and twofold increase in the production of lupeol compared with that of the strain SP1 harboring the wild-type OEW (Fig. 3d). These results demonstrated that modification of Cys257 and Cys259 could affect the activity of OEW, and substitution of Cys259 with A and G could significantly improve the activity of OEW.

Engineering β-amyrin synthase for increasing β-amyrin production

To demonstrate its applicability, such a strategy was applied to β-amyrin synthase. β-Amyrin is the direct deprotonation product of the oleanyl cation which is formed by further ring expansion of the lupenyl cation via the proton transfer of the intermediate germanicyl cation. β-Amyrin synthase is one of the most common PTSs, widely distributed in plants (Fig. 2). β-Amyrin synthase from Glycyrrhiza glabra (GgbAS) was reported to exhibit better catalytic efficiency than that of other β-amyrin synthases from A. annua, P. sativum, and P. ginseng (Dai et al. 2014). Here, GgbAS was used as the representative β-amyrin synthase to verify our strategy. The non-active-site residues Met255, Cys257, and Cys259 in the MXXXXR motif of OEW correspond to Met256, Cys258, and Cys260 in GgbAS. First, the β-amyrin was docked with GgbAS for predicting the role of Met256, Cys258, and Cys260 in the GgbAS. As shown in Fig. 4a and Fig. S2, the residues Trp257 and Tyr259 were spatially proximal to β-amyrin. The residues Met256, Cys258, and Cys260 surrounded the active-site residues Trp257 and Tyr259. This result suggested that the non-active-site residues Met256, Cys258, and Cys260 might function as the shell residues of the active-site residues Trp257 and Tyr259 in the GgbAS.

Fig. 4
figure 4

Modification of the non-active-site residues of β-amyrin synthase from Glycyrrhiza glabra (GgbAS). a Molecular docking model of GgbAS and β-amyrin using PyMOL software. b Influence on the titers and specific titer of β-amyrin as well as cell biomass, when saturation mutagenesis was applied on the sites Met256 (b), Cys258 (c), and Cys260 (d) of GgbAS

Full size image

Single-site saturation mutagenesis was applied to Met256, Cys258, and Cys260. For Met256, strain SP1 harboring GgbASM256Q, GgbASM256A, GgbASM256H, GgbASM256E, GgbASM256N, GgbASM256P, and GgbASM256F showed higher titer and specific titer of β-amyrin than these of strain SP1 harboring the wild-type GgbAS after 72 h cultivation. Among these beneficial mutants, strain SP1 harboring GgbASM256Q could produce 45.2 mg/L β-amyrin with the specific titer of 10 mg/g biomass, which are 6.5-fold and 5.9-fold increase compared with that of strain SP1 harboring the wild-type GgbAS. Strain SP1 harboring GgbASM256N showed a slight decrease in the cell biomass. Furthermore, strain SP1 harboring GgbASM256W, GgbASM256R, GgbASM256K, and GgbASM256G displayed a significantly decreased or complete loss in the production of β-amyrin (Fig. 4b). These results demonstrated that the activity of β-amyrin synthase could be improved by the modification of Met256.

In addition, we investigated whether the modification of Cys258 and Cys260 could improve the activity of GgbAS. As shown in Fig. 4c, most of the substitutions at the Cys258 did not lead to an increase in the titer and specific titer of β-amyrin when single-site saturation mutagenesis was applied. However, strain SP1 harboring GgbASC258V, GgbASC258A, and GgbASC258S produced 51.6 mg/L, and 48.2 mg/L, 33.3 mg/L β-amyrin which were 7.3-fold, 6.8-fold, and 4.7-fold higher titer and specific titer of β-amyrin than these of strain SP1 harboring the wild-type GgbAS. The specific titer of β-amyrin in strain SP1 harboring GgbASC258V, GgbASC258A, and GgbASC258S reached 10.6, 10.9, and 6.6 mg/g biomass, respectively. Compared with Cys258, more mutants at the site of Cys260 could result in a higher titer and specific titer of β-amyrin. Among these mutants, strain SP1 harboring GgbASC260W produced 30.6 mg/L β-amyrin with 5 mg/g biomass, which was 4.4-fold and 2.8-fold higher than these of strain harboring wild-type GgbAS. Strain SP1 harboring GgbASC260R and GgbASC260S resulted in a significant loss in the production of β-amyrin (Fig. 4d). In addition, the cell growth of most mutants at the site of Cys260 could be dramatically increased. These results demonstrated that mutations on the sites of Cys258 and Cys260 could improve the activity of GgbAS. Hence, our strategy could be applied to β-amyrin synthase for improving the activity.

Engineering α-amyrin synthase for increasing α-amyrin production

As many PTSs have been characterized as multifunctional enzymes in plants, we wished to investigate whether our strategy could be applied to multi-function PTSs. The monofunctional α-amyrin synthase has not been identified in plants to date, which could be utilized as the target PTS for verification of our strategy. α-Amyrin is generated by the ursanyl cation which is derived from the germanicyl cation via the proton transfer (Fig. 2). The previous study reported that the activity of α-amyrin synthase from Malus × domestica (MdOSC1) was higher than that of other characterized α-amyrin synthases (Yu et al. 2018). Here, we used MdOSC1 as the target PTS for verification of our strategy. The non-active-site residues in the MFCYCR motif of MdOSC1 correspond to Met256, Cys258, and Cys260. First, the role of non-active-site residues Met256, Cys258, and Cys260 was predicted by molecular docking using the MdOSC1 and the α-amyrin. As shown in Fig. 5a and Fig. S3, the active-site residues Phe257 and Tyr259 were adjacent to the ligand α-amyrin whereas the residues Met256, Cys258, and Cys260 surrounding the active-site residues were distant from the ligand α-amyrin. This result suggested that the residues Met256, Cys258, and Cys260 might be the shell residues around the active-site residues Phe257 and Tyr259 in the MdOSC1.

Fig. 5
figure 5

Modification of the non-active-site residues of α-amyrin synthase from Malus × domestica (MdOSC1). a Molecular docking model of MdOSC1 and α-amyrin using PyMOL software. b Influence on the titers and specific titer of α-amyrin and β-amyrin as well as cell biomass when saturation mutagenesis was applied on the sites Met256 (b), Cys258 (c), and Cys260 (d) of MdOSC1

Full size image

The single-site saturation mutagenesis was applied to Met256, Cys258, and Cys260. Most of the substitutions for Met256 displayed a significant loss in the titer and specific titer of α-amyrin and β-amyrin. Strain SP1 harboring MdOSC1M256K, MdOSC1M256H, MdOSC1M256W, MdOSC1M256D, and MdOSC1M256F completely abolished the production of α-amyrin and β-amyrin. Nevertheless, strain SP1 harboring MdOSC1M256I produced about 14 mg/L α-amyrin and 4.2 mg/L β-amyrin with the specific titer of 3.7 mg/g biomass and 1.1 mg/g biomass, which showed a 2.4-fold and 1.8-fold increase in the titer of α-amyrin and β-amyrin, compared with these of the strain SP1 harboring the wild-type MdOSC1 (Fig. 5b). Furthermore, strain SP1 harboring MdOSC1M256S, MdOSC1M256C, MdOSC1M256A, and MdOSC1M256Y produced more β-amyrin than α-amyrin. The α-amyrin to β-amyrin ratios in these mutant strains were compared with the strain harboring wild-type MdOSC1. Strain SP1 harboring MdOSC1M256P produces exclusively β-amyrin instead of α-amyrin and β-amyrin. These results demonstrated that engineering Met256 could change its activity and the ratio of α/β-amyrin.

For Cys258, we did not observe the overall increase in the titer and specific titer of α-amyrin and β-amyrin when single-site saturation mutagenesis was applied. Nevertheless, most mutants such as MdOSC1C258G, MdOSC1C258Q, MdOSC1C258M, MdOSC1C258H, and MdOSC1C258T showed the change in the ratio of α/β-amyrin (Fig. 5c). To be noted, strain SP1 harboring MdOSC1C258P exclusively produced α-amyrin after 72 h cultivation. For Cys260, the substitution at the Cys260 showed the changed ratio of α-amyrin/β-amyrin and the overall decrease in the titer and specific titer of α-amyrin. Strain SP1 harboring MdOSC1C260E, MdOSC1C260N, and MdOSC1C260Y produced β-amyrin instead of α-amyrin and β-amyrin (Fig. 5d). These results demonstrated that our strategy could be applied to multifunction PTSs for improving their activity, and the ratio of mixed products could also be changed when the non-active-site residues of the MXXXXR motif were engineered.

Combinatorial mutation of beneficial mutations for increasing the production of various PTSs

Based on the aforementioned beneficial sites of various PTSs, we then investigated whether combinatorial mutagenesis could improve the activity of various PTSs. For OEW, double mutants were constructed based on the significant beneficial single-site mutants of the sites Met255 and Cys259 including M255V, M255L, M255C, M255I, M255Y, C259A, and C259G (Fig. 6a). All strain SP1 carrying these double mutants except OEWM255YC259G accumulated the lower titer and specific titer of lupeol, compared with strain SP1 carrying the wild-type OEW. Strain SP1 carrying OEWM255YC259G achieved an approximately 1.9-fold higher titer of lupeol than SP1 carrying the wild-type OEW. The titer and specific titer of lupeol reached 35 mg/L and 8.27 mg/g biomass, respectively.

Fig. 6
figure 6

Influence of double and triple mutants on the activity of PTSs. a Influence on the titers and specific titer of lupeol as well as cell biomass when double mutants were applied on the sites Met255 and Cys259 of OEW. b Influence on the titers and specific titer of β-amyrin as well as cell biomass when double mutants were applied on the sites Met256, Cys258, and Cys260 of GgbAS. c Influence on the titers and specific titer of β-amyrin as well as cell biomass, when triple mutants were applied on the sites Met256, Cys258, and Cys260 of GgbAS

Full size image

For the GgbAS, combinatorial mutagenesis was employed based on the top two beneficial mutants of Met256, Cys258, and Cys260. To be specific, M256A, M256Q, C258A, C258V, C260W, and C260H were chosen to construct twelve double mutants and eight triple mutants, respectively. As shown in Fig. 6b, the strain SP1 carrying GgbASM256AC258V, GgbASC258VC260H, and GgbASC258AC260W showed a significant increase in the titer and specific titer of β-amyrin where other double mutants lowered the production of β-amyrin. Among these double mutants, the strain SP1 carrying GgbASC258VC260H produced 20.8 mg/L β-amyrin with a specific titer of 4.7 mg/g biomass. For triple mutants, the strain SP1 carrying GgbASM256AC258VC260W and GgbASM256AC258AC260W showed a 1.55-fold and 1.23-fold increase in the production of β-amyrin compared with that of strain SP1 harboring the wild-type GgbAS. The production of β-amyrin in the strain SP1 carrying GgbASM256AC258VC260W and GgbASM256AC258AC260W reached 12.8 mg/L and 10.1 mg/L with a specific titer of 2.9 mg/g biomass and 2.6 mg/g biomass (Fig. 6c). For MdOSC1, triple mutants were constructed based on the beneficial single-site mutants in the MdOSC1 and GgbAS. As shown in Figure S4, all the strain SP1 carrying triple mutants accumulated the lower titer and specific titer of α-amyrin, compared with strain SP1 carrying the wild-type MdOSC1. These results indicated that the combinatorial mutagenesis of the beneficial mutant could also influence the activity of PTSs despite that these recombinant mutants showed relatively lower efficiency than that of the single beneficial mutants of PTSs.

Discussion

In this study, we first reported that engineering the non-active-site residues of the MXXXXR motif could improve their activity, which could be applied to typical PTSs. As a proof of concept, the computational models were used to predict the function of these non-active sites. Several single mutants of OEW, GgbAS, and MdOSC1 exhibiting the higher production of corresponding pentacyclic triterpenes were achieved when site-directed saturation mutagenesis was applied on the non-active-site residues. In addition, double and triple mutants of GgbAS and OEW could also improve the production of corresponding triterpenes. This strategy has the potential to be applied to other non-active-site residues close to the active sites of typical PTSs for improving the activity.

Recently, many strategies have been applied to improve the activity of typical PTSs. Genome mining is a traditional strategy to identify a PTS with high activity in nature (Dale et al. 2020; Wang et al. 2021). Various isoenzymes could be characterized based on the genome and transcriptome data of different plant species. The typical PTSs with high activity could be screened out based on the titer and specific titer of corresponding triterpenes (Dale et al. 2020). However, it is time-consuming to dig up the PTSs with beneficial mutations. In addition, a semi-rational design has been recently applied to improve the activity of a specific PTS. Based on the prediction of the computational model, a mutant MdOSC1N11T/P250H/P373A with high activity was achieved by the single and combinatorial site-direct saturation mutagenesis (Yu et al. 2020). We also investigated whether the mutant N11T/P250H/P373A combined with M255T could further enhance the production of α-amyrin. Nevertheless, the quadruple mutant N11T/P250H/P373A/M255T did not show higher production of α-amyrin than that of the wild-type MdOSC1 (Figure S5).

The active sites of cyclases function as the stabilization of cationic intermediates, which are critical for the catalytic activities and specificities (Hoshino 2017). Most mutations in these active sites would disrupt the cation-π interaction and affect the stability of cationic intermediates, which could decrease the activity of cyclases (Srisawat et al. 2019). For instance, most mutations in the Phe413, Phe728, Trp257, and Tyr259 of EtbAS could decrease its catalytic activity (Ito et al. 2013, 2016). Nevertheless, a previous study suggested that increasing the stability of the cationic intermediates could improve the activity of squalene-hopene cyclase by enhancing the cation-π interaction (Dang and Prestwich 2000). In addition to cyclases, many studies demonstrated that engineering the residues surrounding the active site of enzymes could generate mutated enzymes with improved catalytic activity (Krah et al. 2021). Taken together, engineering the non-active-site’s proximity to active sites of PTSs might influence the cation-π interaction and then led to a change in the catalytic efficiency or specificity of PTSs. For the MdOSC1, we did not observe that engineering Cys258 and Cys260 in the MXXXXR motif could improve the enzyme activity, which might be due to that MdOSC1 is a multi-function PTS. Furthermore, engineering the non-active-site residues in the MXXXXR motif could switch the favor of the ursanyl cation to the oleanyl cation as the non-active-site residues might stabilize the ursanyl cation through cation − π interactions and CH − π interactions (Huang et al. 2022; Wu et al. 2019).

In summary, we developed a novel strategy for improving the activity of PTSs in S. cerevisiae. Results demonstrated that the titer and specific titer of lupeol, β-amyrin, and α-amyrin were significantly increased up to 7.3-fold when the PTSs harbored corresponding mutations. As a proof of concept, we demonstrated that the activity of typical PTSs could be influenced by engineering the non-active-site residues of the MXXXXR motif, and the activity of three PTSs was successfully improved. This work also provided a new approach for improving the production of pentacyclic triterpenoids.