Engineering a universal and efficient platform for terpenoid synthesis in yeast

Significance

Terpenoids are a class of high-value natural products with wide application in both industrial and human health spaces, but sourcing them from nature or deriving from petrochemicals is no longer sustainable. Microbial biosynthesis of terpenoids has emerged as the most commercially viable option for their large-scale production. However, economically viable titers and productivities are mostly hampered by the limited availability of the main precursors in their biosynthesis, prenol phosphates, in Saccharomyces cerevisiae. Here, we overcome this challenge by establishing the isopentenol utilization pathway–based precursor-forming routes in S. cerevisiae to augment the native mevalonate pathway, circumventing the competition from sterol biosynthesis. Our findings provide a universal and effective yeast platform for the production of diverse terpenoids.

Abstract

Engineering microbes for the production of valuable natural products is often hindered by the regulation of native competing metabolic networks in host. This is particularly evident in the case of terpenoid synthesis in yeast, where the canonical terpenoid precursors are tightly coupled to the biosynthesis of sterols essential for yeast viability. One way to circumvent this limitation is by engineering product pathways less connected to the host native metabolism. Here, we introduce a two-step isopentenol utilization pathway (IUP) in Saccharomyces cerevisiae to augment the native mevalonate pathway by providing a shortcut to the synthesis of the common terpenoid precursors, isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP). As such, the IUP was capable of elevating the IPP/DMAPP pool by 147-fold compared with the native pathway. We further demonstrate that cofeeding isoprenol and prenol enhances geranyl diphosphate (GPP) content for monoterpene biosynthesis. More importantly, we established a synthetic three-step route for efficient synthesis of di-and tetraterpene precursor geranylgeranyl diphosphate (GGPP), circumventing the competition with farnesyl diphosphate (FPP) for sterol biosynthesis and elevating the GGPP level by 374-fold. We combine these IUP-supported precursor-forming platforms with downstream terpene synthases to harness their potential and improve the production of industrially relevant terpenoids by several fold. Our exploration provides a universal and effective platform for supporting terpenoid synthesis in yeast.

Terpenoids (also known as isoprenoids) constitute the largest and most structurally diverse class of natural products, with over 75,000 members found to date (1). Their importance derives from numerous successful products belonging to this family and also because they have been exploited for numerous applications ranging from adhesive materials and pharmaceuticals to coloring agents, fragrances, and flavors (2–5). Due to their structural complexity, chemical synthesis of terpenoids has been challenging, and their supply still largely relies on extraction from plant sources, which suffers from low yields and different impurities caused by the complex chemical congener mixtures inherent to plant extracts (3). In light of these challenges, engineering microbes for the production of industrially important terpenoids provides a promising route for the cost-effective supply of these compounds.

All naturally produced terpenoids are derived from C₅-building blocks, isopentenyl pyrophosphate (IPP) and its isomer dimethylallyl pyrophosphate (DMAPP). Sequential addition of IPP units to DMAPP generates terpenoid precursors, like geranyl diphosphate (GPP, C₁₀), farnesyl diphosphate (FPP, C₁₅), and geranylgeranyl diphosphate (GGPP, C₂₀), which can be further converted by terpene synthases into typically cyclic terpenoid scaffolds. Such terpenoid scaffolds often undergo further stereo- and region-selective functionalization catalyzed by cytochromes P450 to produce monoterpene (C₁₀), sesquiterpene (C₁₅), diterpene (C₂₀), triterpene (C₃₀), and tetraterpene (C₄₀) compounds (6). Natively, both precursors IPP and DMAPP are generated through either the mevalonate (MVA) or 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway, which have been interesting targets for pathway engineering to produce copious quantities of precursors for a wide range of terpenoids in a variety of organisms (7, 8).

Saccharomyces cerevisiae, with a well-characterized MVA pathway, is an attractive host for de novo synthesis of plant-derived terpenoids, due to its robustness for genetic modulation and industrial bioprocessing, and additional engineering for further terpenoid scaffold decoration by functional expression of cytochrome P450 enzymes (9). In the past years, metabolic engineering of the MVA pathway in yeast has succeeded in increasing the flux of precursors, IPP and DMAPP, into terpenoid biosynthesis, such as overexpression of rate-limiting enzyme 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR) with N-terminal truncation (10–14) or even the whole MVA pathway (15), and heterologous introduction of the MEP pathway from Escherichia coli (16) and the Enterococcus faecalis genes EfmvaE and EfmvaS (17–19). However, these approaches have not succeeded in removing limitations owing to intrinsic regulatory mechanisms present in the native host and negative cross talk between product-forming and growth-sustaining reactions. In addition, due to the Crabtree effect, engineering S. cerevisiae for the overproduction of chemicals from glucose is often hampered by its inherent fermentation metabolism where most glucose is shunted toward ethanol (20), hindering the engineer’s goal of producing chemical products other than ethanol. A possible approach to overcoming these challenges would be to disconnect the heterologous product pathway from the host native metabolism.

In this study, we attempted to augment the native pathway in S. cerevisiae by introducing a synthetic route, the isopentenol utilization pathway (IUP), which can directly convert isopentenol isomers isoprenol and prenol into IPP and DMAPP, respectively, through two phosphorylation reactions (Fig. 1). The IUP was previously characterized in E. coli and demonstrated to be orthogonal to native pathways and central carbon metabolism (21–24). Here, upon activating the IUP in S. cerevisiae, the IPP/DMAPP pool was elevated by 147-fold relative to the native MVA pathway, illustrating the high efficiency of IUP at supplying ample amounts of critical precursors for terpene synthesis. We then optimized the combined IUP and downstream terpene precursor-forming pathway to establish a universal and efficient yeast platform for the biosynthesis of all kinds of terpenoid products. More specifically, we demonstrated that cofeeding isoprenol and prenol at a molar ratio of 7:3 enabled to further enhance the GPP content for monoterpene biosynthesis. More importantly, we established a three-step shortcut access to common di-and tetraterpene precursor GGPP synthesis, achieving a 374-fold increase compared with the native pathway. As a result, we successfully constructed the complete synthetic pathways that extend up to all kinds of terpenoids and achieved marked improvements in the production of several industrially important compounds.

Fig. 1.

Scheme of glucose-dependent controllable gene expression system in S. cerevisiae. In Gal regulatory system, Gal80p functions as a repressor to bind with Gal4p, thereby inhibiting Gal4-mediated transcription initiation of the Gal promoters in the absence of galactose. On the other hand, Gal3p interacts with Gal80p to eliminate its inhibition on Gal4p and thus allows Gal4p to bind the Gal10 and Gal1 promoters to activate the transcription of downstream genes. In the Snf1 network, Mig1p functions as a transcriptional repressor to inhibit the transcription of Gal4 in the presence of glucose. After Gal80p disruption, the expression of Gal4 can be controlled by the Snf1 signaling pathway only. Thus, the engineered pathway switch time could be controlled by adjusting the glucose supply.

Results

Construction of Diauxie-Inducible Biosynthetic Pathway in Yeast.

To evaluate the effect of IUP substrates isoprenol and prenol on cell growth, we cultured S. cerevisiae in media with varying concentrations of isoprenol or prenol, respectively. Cell growth was significantly inhibited when the medium was supplemented with increasing amounts of isoprenol or prenol (SI Appendix, Fig. S1). Considering this substrate toxicity, fine-tuning the timing of isoprenol or prenol feeding would be essential to maximizing the yields of desired chemicals. One tool that has been employed to this end is the use of inducible systems allowing control of metabolic enzyme expression (10, 25–27) such as to separate the growth phase from production in a bioreactor operation. Through the use of inducible expression systems, supply of inhibitory substrates can be coordinated with the initiation of the production phase such as to minimize accumulation of the inhibitory substrate.

The galactose regulatory network, in S. cerevisiae, is one of the best-characterized transcriptional induction systems, which is activated by galactose and inhibited by glucose (28, 29). This regulation is achieved via two signal pathways: the Gal4p–Gal80p regulatory axis responsible for galactose utilization and the Snf1 network involved in glucose repression (28, 30–32). In the core galactose regulon, in the absence of galactose, Gal80p binds the transcription activator Gal4p to inhibit Gal4p-mediated transcription initiation of the Gal promoters; in the opposite, the transcriptional factor Gal3p binds with Gal80p to relieve Gal80p repression on Gal4p transcription activation (Fig. 1). Therefore, galactose-independent activation of the Gal promoters can be achieved by disruption of Gal80p, leading to the Snf1 network to act as the sole controller, which allowed us to build the glucose-dependent controllable system (15, 31, 33). Consequently, pathways for desired chemical formation under the control of Gal promoters can be activated after glucose depletion at the diauxic phase in strains harboring Gal80p disruption. In this study, we employed the Gal1 and Gal10 promoters in combination with Gal80p disruption to control target pathways (Fig. 1), in which isoprenol or prenol was supplemented in the medium after glucose depletion at the diauxic phase when ScCK and AtIPK can be activated, thus avoiding substrate toxicity to cell growth.

Establishing Efficient IPP/DMAPP Biosynthesis in Yeast Cells.

To explore IUP functionality in S. cerevisiae, this two-step phosphorylation pathway consisting of two kinase enzymes, choline kinase (CK) and isopentenyl phosphate kinase (IPK), was implemented. ScCK from S. cerevisiae and AtIPK from Arabidopsis thaliana were expressed in Gal80p disruption strain SCMA00, and the intracellular isopentenyl monophosphate/dimethylallyl monophosphate (IP/DMAP) and IPP/DMAPP levels were measured using LC-MS/MS. Thirty millimolar isoprenol or prenol was determined as the optimal concentration for supplementation based on growth inhibition considerations (SI Appendix, Fig. S1). Upon activation of the IUP in S. cerevisiae, the pool size of IP/DMAP and IPP/DMAPP significantly increased by 9.1-fold and 147-fold, respectively, compared with the base strain in the presence of isoprenol in the medium (Fig. 2 A and B). Time courses of intracellular intermediate levels showed that the engineered strain SCMA01 harboring the IUP exhibited consistently elevated fluxes to IP/DMAP and IPP/DMAPP compared with the base strain of no IUP expression (Fig. 2 C and D). Similar results were obtained when cultures were supplemented with prenol, and the levels of IP/DMAP and IPP/DMAPP were elevated by 3.3-fold and 36.2-fold, respectively, in the engineered strains with ScCK and AtIPK overexpression (Fig. 2 E and F). The lower levels observed with prenol suggest that isoprenol is a preferred substrate for ScCK as observed before (21). Taken together, our observations demonstrated that the IUP in S. cerevisiae can serve as an efficient augmentation to the native MVA pathway for the supply of terpenoid precursors similar to previously reported (34). Since S. cerevisiae already harbors a native copy of ScCK, we also tested whether the truncated IUP could function by only expressing AtIPK. Although this single-enzyme-variant of the IUP is sufficient to enhance IPP/DMAPP synthesis, achieving a 26.3-fold and 7.5-fold increase when supplemented with isoprenol and prenol, respectively (SI Appendix, Fig. S2), and a decrease in performance compared with the full IUP was also evident, necessitating the use of both enzymes in subsequent engineering.

Fig. 2.

The IUP enhances the synthesis of IPP/DMAPP in S. cerevisiae. (A and B) Comparison of the intracellular IP/DMAP (A) and IPP/DMAPP (B) in strains with and without ScCK/AtIPK overexpression. Thirty millimolar isoprenol was supplemented in the YPD medium after glucose depletion. Intracellular intermediates were measured at 12 h after isoprenol addition. (C and D) Time courses of intracellular intermediate level suggest that the engineered strain SCMA01 harboring ScCK and AtIPK can constantly promote IP/DMAP (C) and IPP/DMAPP (D) biosynthesis in the presence of isoprenol compared with base strain without the IUP. (E and F) Comparison of the intracellular IP/DMAP (E) and IPP/DMAPP (F) in strains with and without ScCK/AtIPK overexpression when 30 mM prenol was supplemented in the YPD medium. Data are represented as mean ± SD (n = 3).

We also tested other CK and IPK from different species. SmDAGK (equivalent to ScCK), diacylglycerol kinase from Streptococcus mutans, and MvIPK, IPK from Methanococcus vannielii, were reported to be the optimal combination for lycopene synthesis in E. coli by comparing enzyme variants from different species (35). However, when SmDAGK and MvIPK were expressed in S. cerevisiae, the IPP/DMAPP level was significantly lower than that of ScCK and AtIPK (SI Appendix, Fig. S3). To reduce the connection with the native cytosolic sterol synthesis, we attempted to compartmentalize the IUP into suborganelles. Unexpectedly, either the endoplasmic reticulum (ER)–or peroxisome-assembled IUP significantly decreased IP/DMAP and IPP/DMAPP contents in cells relative to their cytosolic counterparts (SI Appendix, Fig. S4), possibly due to the instability of ScCK or AtIPK in compartments. Therefore, we opted to use the cytosolic IUP consisting of ScCK and AtIPK for further experiments.

Validation of the IUP by ¹³C-Labeling Assay.

We were next interested in assessing the flux generated by the IUP using ¹³C metabolic tracer assays (Fig. 3A). The engineered strain SCMA01 harboring ScCK and AtIPK was grown in the YNB medium containing [U-¹³C₆]glucose as the sole carbon source and then supplied with natural abundance isoprenol after glucose had been depleted. Immediately prior to the addition of unlabeled isoprenol, we detected fully labeled IP generated from mevalonate-5-phosphate (MVAP, Fig. 3A) and fully labeled IPP/DMAPP (Fig. 3 B and C). After 2 h of isoprenol addition, unlabeled IP and IPP/DMAPP were observed and predominately accumulated (Fig. 3 B and C), suggesting a strong flux toward IPP/DMAPP synthesis provided by the IUP. In addition, we also observed that the labeling patterns of the glycolytic intermediates fructose-1,6-disphosphate (FBP) and phosphoenolpyruvate (PEP) remained unchanged after supplementation of natural abundance isoprenol (Fig. 3 D and E), indicating total uncoupling of the IUP from native glycolysis in S. cerevisiae. We concluded from these results that the IUP is functional for synthesizing IPP/DMAPP in S. cerevisiae from isopentenol and that the function of IUP was independent from the native pathway.

Fig. 3.

Validation of functioning of the IUP using ¹³C-labeling and metabolic tracer analyses. (A) Schematic diagram of carbon flux through the mevalonate pathway and IUP. (B and C) The fraction of ¹³C-labeled IP (B) and IPP/DMAPP (C). The engineered strain SCMA01 harboring the IUP was grown in the YNB medium containing [U-¹³C₆]glucose as the sole carbon source. Within 2 h of natural abundance isoprenol feeding, significant amounts of unlabeled IP (B) and IPP/DMAPP (C) were observed. (D and E) Labeling patterns for fructose-1,6-disphosphate (FBP) (D) and phosphoenolpyruvate (PEP) (E), respectively, in the IUP strain over the 8-h incubation. Data are represented as mean ± SD (n = 3).

Specifically Redirecting Flux to Precursors of Different Terpene Pathways.

To evaluate the effect of IUP expression on the native downstream pathways of terpenoid synthesis, especially the synthesis of prenol phosphates, we expanded the intracellular intermediate measurements to include GPP, FPP, and GGPP (Fig. 4A). Although increases in the quantities of all three intermediates were observed in the IUP engineered strain SCMA01, the improvement of GPP, FPP, and GGPP was limited, especially when compared with those of IPP/DMAPP, leading to substantial accumulation of the latter in cells harboring the IUP (Fig. 4B). The highly accumulated IPP is toxic to cells, resulting in growth retardation and a decrease in glucose consumption (36). Therefore, we next aimed to relieve the IPP/DMAPP accumulation by redirecting the flux toward biosynthesis of downstream terpene precursor.

Fig. 4.

Redirecting increased IPP/DMAPP supply, achieved by flux enhancement through the IUP, to terpene precursor biosynthesis. (A) Schematic diagram of terpene precursor–synthesizing pathways combining native MVA enzymes and synthetic IUP. Ac-CoA, acetyl-CoA. (B) Intracellular intermediate levels of IPP/DMAPP, GPP, FPP, and GGPP in IUP strains with engineered downstream pathways. Introduction of ERG20M (ERG20^F96W-N127W), ERG20, and SaGGPPs, along with IDI overexpression, enabled to specifically elevate the pool sizes of terpene precursors GPP, FPP, and GGPP, respectively, thus creating a platform for biosynthesis of diverse terpenes in S. cerevisiae. The YPD medium containing 30 mM isoprenol was used in these experiments. Intracellular precursor levels were determined after 12-h incubation with isoprenol. Data are represented as mean ± SD (n = 3).

To this end, we first overexpressed FPP synthase (ERG20), which sequentially catalyzes the condensation of IPP and DMAPP to form GPP and the conversion of GPP to FPP, along with an additional copy of isopentenyl diphosphate isomerase (IDI) in the engineered strain harboring IUP (Fig. 4A). We found that, in the resulting strain SCMA03, the levels of GPP and FPP significantly increased by 1.67- and 2.13-fold, respectively, compared with base strain SCMA01, and the IPP accumulation was significantly reduced as well (Fig. 4B). To be noted, although ERG20 has both GPP and FPP synthase activities, only a small fraction of GPP was accumulated, and most of it was used by ERG20 to synthesize FPP, allowing to build a platform for FPP-derived terpene synthesis, yet hindering monoterpene production (Fig. 4B). Therefore, aiming to decrease the competition for GPP by ERG20 and establish a platform specially for monoterpene biosynthesis, we alternatively introduced the ERG20^F96W-N127W (ERG20M; Fig. 4A), a dominant-negative ERG20 variant that inhibits the FPP synthesis step of the endogenous wild-type enzyme, and increased the GPP pool without abolishing sterol synthesis (17, 37, 38) to substitute overexpression of wild-type ERG20 (resulting in SCMA04 strain). Consistent with our expectation, this exploration allowed strain SCMA04 to achieve a 2.4-fold increase in the GPP pool along with a significant decrease in the FPP level compared with SCMA03 with ERG20 overexpression (Fig. 4B), indicating that the introduction of ERG20^F96W-N127W redirects the flux of substrates away from FPP synthesis to GPP formation.

In parallel with these studies, we also noticed that in all engineered strains harboring a native copy of the GGPP synthase, the GGPP content was very low even in strain SCMA03 optimized for FPP production, indicating that conversion of FPP to GGPP was a rate-limiting step for GGPP-derived compound biosynthesis. To enhance GGPP accumulation for diterpene and tetraterpene biosynthesis, a GGPP synthase from Sulfolobus acidocaldarius (SaGGPPs) was introduced into the IUP engineered strain SCMA01 along with overexpression of IDI (Fig. 4A). The resulting strain SCMA05 exhibited a 142-fold increase in the GGPP level, while other intracellular intermediates, such as IPP/DMAPP, GPP, and FPP, also significantly decreased relatively to base strain SCMA01 levels (Fig. 4B). Our results thus demonstrated that the accumulated IPP/DMAPP provided by the IUP can be effectively converted by SaGGPPs into the GGPP pool in the strain SCMA05. To be noted, compared with strain SCMA05, engineered for GGPP accumulation, the fold change in GPP and FPP contents in their corresponding overproduction strains showed relatively modest improvements, possibly due to an amount of flux channeled into sterol synthesis. This is supported by the observation that the levels of squalene, the precursor for sterol formation, in GPP- or FPP-overproducing strains were significantly higher than those in others (SI Appendix, Fig. S5). This result also strongly suggests that the increased GPP and FPP flux redirected from the IUP was efficiently channeled toward sterol synthesis. Overall, by optimizing the downstream terpene precursor-forming pathways, we have constructed three strains that can selectively accumulate GPP, FPP, and GGPP. These strains can potentially serve as platforms for the production of GPP-derived (monoterpenoid), FPP-derived (sesquiterpene and triterpene), and GGPP-derived (diterpene and tetraterpene) terpenoid products.

Maximizing IUP Performance through Cofeeding Isoprenol and Prenol.

Naturally, GPP is derived from IPP and its isomer DMAPP in a head-to-tail manner. Further condensation of GPP with one or two additional IPP units yields FPP and GGPP, respectively (Fig. 4A). Since the substrate isoprenol or prenol can be directly converted into IPP or DMAPP, respectively, through the IUP, supplying both substrates in the medium enables the possibility of enhancing the performance of IUP, thus achieving further improvement of terpenoid precursors (GPP, FPP, and GGPP). Furthermore, we also confirmed that, in S. cerevisiae, the growth properties of prenol-supplementing yeast cells were very similar to those of isoprenol-only-containing media (SI Appendix, Fig. S1), unlike what was observed with other organisms (E. coli and Yarrowia lipolytica) in which prenol was observed to be more toxic to cell growth than isoprenol.

This observation provided a foundation for optimizing the substrate ratio to augment the IUP in S. cerevisiae. To test this notion, a preliminary experiment was conducted by feeding both isoprenol and prenol with varying molar ratios in terpene precursor–overproducing strains. We found that, in the GPP-overproducing strain SCMA04, the optimal condition for high-level GPP accumulation was a 7:3 of substrate ratio, yielding an increase of 2.1-fold compared with using isoprenol addition only (Fig. 5A). Considering the effect of the enzyme IDI on the distribution of IPP and DMAPP fluxes, we also investigated the effectiveness of supplying both isoprenol and prenol in the GPP production media for strain SCMA06 without IDI overexpression. Compared with IDI-overexpressed strain SCMA04, the GPP content displayed a significant decrease in the strain SCMA06 harboring native copy of IDI only, regardless of the substrate ratio (Fig. 5A). A similar strategy was applied next to FPP- and GGPP-accumulating strains. We found that the highest level of FPP was achieved in the IDI overexpression strain SCMA03 supplemented with isoprenol exclusively (Fig. 5B), which was likely due to the increased stoichiometric need for IPP over DMAPP. As for GGPP production, although the highest level in the IDI overexpression strain SCMA05 was achieved by feeding isoprenol only, interestingly, the engineered strain SCMA08 without IDI overexpression showed further GGPP improvement when feeding prenol exclusively and increased by 2.6-fold compared with the best performance of IDI-overexpressed strain (Fig. 5C). This investigation allowed us to utilize the full capacity of the IUP-augmented strains and provided a foundation for high-yield terpenoid biosynthesis in S. cerevisiae.

Fig. 5.

An optimal level exists for the ratio of isoprenol/prenol supplementation in the media. (A) A maximum GPP production was obtained in strain SCMA04 with IDI overexpression and cofeeding isoprenol and prenol at the ratio of 7:3, resulting in a 2.1-fold increase compared with isoprenol addition only. (B) The level of FPP was maximized in the IDI-overexpressed strain SCMA03 when supplemented with isoprenol exclusively. The maximum was much higher than that of strain SCMA07 without IDI overexpression. (C) The engineered strain SCMA08 without IDI overexpression supplemented with prenol exclusively achieved the highest GGPP production and increased by 2.6-fold compared with IDI-overexpressing strain SCMA05 with isoprenol supplementation. The YPD medium containing 30 mM isoprenol and prenol varying with their ratio from 10:0 to 0:10 was used in these experiments. Intracellular precursor levels were determined after 12-h incubation with a substrate. The optimal substrate ratio with a high level of terpene precursors was indicated by an arrow. Data are represented as mean ± SD (n = 3).

A Three-Step Pathway Shortcut for GGPP Synthesis.

Since the engineered strain SCMA08 harboring ScCK, AtIPK, and SaGGPPs without an additional copy of IDI showed the best performance when supplemented with prenol only, achieving a 374-fold increase in the GGPP level compared with the strain utilizing the native pathway (Fig. 6 A and B), we hypothesized that SaGGPPs may directly utilize the DMAPP to synthesize GGPP. To test this hypothesis, we examined the functionality of bacterially expressed SaGGPP enzyme in an in vitro assay (SI Appendix, Fig. S6) and found that SaGGPPs can accept DMAPP as a substrate for GGPP synthesis (Fig. 6C), consistent with a previous report (39). This discovery allowed us to establish a pathway for the synthesis of the universal precursor GGPP of di-and tetraterpenes in combination with the IUP using prenol as a substrate (Fig. 6A). This alternative three-step route is completely decoupled from native metabolism and could efficiently and directly convert the increased DMAPP pool, which results from the IUP-enhanced flux, toward GGPP, while avoiding competition for FPP for sterol biosynthesis. To validate this pathway activity, an in vitro multienzyme sequence was carried out using enzymes expressed in and purified from E. coli (SI Appendix, Fig. S6). By supplying prenol as a substrate, we monitored the expected product from the in vitro biosynthetic sequence. As shown in Fig. 6D, the three-step enzyme sequence was capable of converting prenol to GGPP. Thus, through a combination of in vivo and in vitro experiments, we concluded that it is possible to achieve conversion of prenol to GGPP through a simple three-step pathway with the potential to exceed the native MVA pathway in producing di-and tetraterpene products.

Fig. 6.

A three-step shortcut pathway for efficient GGPP synthesis. (A) Schematic of the mevalonate (22 steps) and 3-step shortcut pathways for GGPP synthesis in yeast. (B) Comparison of the GGPP levels in strain SCMA08 harboring the synthetic 3-step pathway and a strain with the native pathway only. Data are represented as mean ± SD (n = 3). (C and D) In vitro confirmation of GGPP biosynthesis from DMAPP (C) and prenol (D) through the shortcut pathway comprising enzymes expressed in and purified from E. coli. Samples in all assays were analyzed by LC-MS/MS. Samples without enzyme were considered negative control.

Demonstration of the Established Yeast Platform for Terpenoid Biosynthesis.

To fully exploit the potential of IUP-driven precursor-forming platforms, we set out to engineer terpenoid production in our yeast strains, with the goal of translating elevated intracellular intermediate concentrations to increased terpenoid synthesis. We first targeted limonene as a model monoterpene compound. The Citrus limon limonene synthase (LS) (9) lacking plastid-targeting peptide was introduced to GPP-accumulating strain (resulting in SCMA09). As control, we also transformed the strain lacking the IUP but harboring IDI and ERG20^F96W-N127W (resulting in SCMA10). In this strain, IPP/DMAPP production relies on the native MVA pathway. We obtained a 20-fold increase in the limonene titer of strain SCMA09, up to 20.2 mg/L, compared with control (Fig. 7A). We similarly investigated sesquiterpene, diterpene, triterpene, and tetraterpene synthesis with amorphadiene, taxadiene, β-amyrin, and lycopene as a representative compound of each group of terpenoids, respectively. To this end, we introduced yeast codon-optimized versions of four different terpene synthases, including Artemisia annua amorphadiene synthase (ADS) (15), Taxus brevifolia taxadiene synthase (TDS) (40), Glycyrrhiza glabra β-amyrin synthase (AS) (41), and Mucor circinelloides lycopene synthase (CarB and CarRP^E78K) (42), into the corresponding terpene precursor–overproducing strains, respectively (Fig. 7 B–E). Product titers in all cases were significantly increased in the engineered IUP-expressing chassis strains compared with those in strains utilizing the native MVA pathway only, ranging from 1.74- to 4.3-fold improvement (Fig. 7 B–E). In addition, experiments using ¹³C-labeled glucose confirmed that the carbon flux generated by the IUP from isoprenol or prenol as a substrate was channeled for increased terpenoid synthesis (Fig. 7F). We also overexpressed tHMGR, a rate-limiting enzyme involved in the MVA pathway, in strains harboring the IUP, and found a further increased terpene titer (Fig. 7), suggesting that the IUP contributes to upregulation of the MVA pathway, and utilization of both the MVA pathway and IUP would further enhance terpenoid production. Taken together, these findings support the strategy that terpenoid synthesis can be significantly enhanced by the increasing availability of the intracellular precursors of each product. This can be accomplished by combining expression of the IUP with specific genes directing the enhanced IUP flux toward the synthesis of the key precursor of each terpenoid product. The strategy produced overall higher yields, thus establishing an efficient universal platform for productive terpenoid synthesis in yeast.

Fig. 7.

Demonstration of enhanced terpenoid synthesis by IUP-engineered strains harboring pathways for the conversion of isoprenol or prenol to various products. Limonene (A), amorphadiene (B), taxadiene (C), β-amyrin (D), and lycopene (E) as representatives of monoterpene, sesquiterpene, diterpene, triterpene, and tetraterpene were targeted, respectively. The YPD medium containing corresponding substrate was used in these experiments. Limonene-producing strains were grown in the YPD medium with cofeeding of 30 mM isoprenol and prenol at a ratio of 7:3. Amorphadiene- and β-amyrin-producing strains were cultured in medium containing 30 mM isoprenol. Taxadiene- and lycopene-producing strains were grown in medium supplemented with 30 mM prenol. tHMGR was overexpressed in strains harboring the IUP for further increase in terpenoid production. Fold changes were shown above columns. Samples were collected and measured after 5-d cultivation. Data are represented as mean ± SD (n = 3). (F) Mass spectra of ¹³C₆ labeled (in red) and unlabeled (in black) compounds.

Discussion

Overexpression of endogenous genes and heterologous pathways are often undermined by native regulatory networks and competition with native metabolism. This drawback is especially evident in the synthesis of compounds requiring long and complex pathways for their biosynthesis (e.g., natural products such as terpenoids). In this regard, utilization of orthogonal pathways decoupled from native metabolism has shown promise for improved pathway control and overall performance (9). This is especially important when the biosynthetic pathway of interest competes with endogenous pathways that are essential for cell growth.

In the specific case of terpenoid synthesis, the MVA pathway for IPP/DMAPP synthesis starts from acetyl-CoA derived from central carbon metabolism, thereby competing with other cellular processes for resources, which can complicate attempts to increase terpenoid pathway flux. Furthermore, the MVA pathway requires 2 molecules of NADPH and 3 molecules of ATP per molecule of IPP synthesized. This nontrivial demand for cofactor sharpens competition for cellular resources and often limits the flux achievable by overexpressing endogenous genes (43). Besides carbon and cofactor limitations, current microbial production of terpenoids also faces challenges associated with highly regulated enzymes. Pathway intermediates or downstream products have been reported to inhibit enzymes in the MVA pathway: acetoacetyl-CoA thiolase can be inhibited by free CoA (44), HMG-CoA synthase can be inhibited by acetylacetyl-CoA, HMG-CoA, and CoA (45), HMG-CoA reductase is capable of being inhibited by HMG, free CoA, and NAD(P)⁺/NADPH (46, 47), and IPP, DMAPP, GPP, and FPP have been shown to inhibit mevalonate kinase (48). Those complex regulations often hinder attempts to up-regulate the MVA pathway for efficient terpenoid synthesis.

To overcome these limitations, we here introduced the two-step IUP, which can directly convert isoprenol and prenol to IPP and DMAPP, respectively, into S. cerevisiae. Our results demonstrated that instead of extensively engineering the native pathways, the IUP can serve as a powerful augmentation to the MVA pathway, strengthening the flux toward IPP formation and downstream pathways. As a metabolic pathway decoupled from central carbon metabolism, the IUP not only allowed shortcut access to IPP, circumventing the bottlenecks encountered in the native pathways, but also achieved minimal cofactor requirements. Since the IUP consists of two phosphorylation steps, only two molecules of ATP are required per molecule of IPP synthesized, reducing the competition for NADPH with the decoration of terpene scaffold by cytochrome P450 enzymes. These benefits underline the potential of IUP in yeast metabolic engineering for terpenoid synthesis.

In this study, we first demonstrated the utility of IUP in the model industrial microbial host S. cerevisiae. The introduction of IUP led to a significant increase (147-fold) in intracellular IPP/DMAPP, which subsequently translated to increased terpenoid precursors by overexpressing ERG20 and introducing ERG20^F96W-N127W and SaGGPPs, respectively. Further optimization was achieved by designing a cofeeding strategy with both isoprenol and prenol, making use of the desirable characteristics of S. cerevisiae that, in yeast, have similar tolerance to isoprenol and prenol. Its stronger tolerance than that of the other two common hosts namely E. coli and Y. lipolytica also suggests that IUP utilization in S. cerevisiae is more competitive (SI Appendix, Fig. S1). This is also supported by the best performance in S. cerevisiae in terms of content and productivity of IPP/DMAPP augmented by the IUP than the other two organisms harboring ScCK and AtIPK only (SI Appendix, Table S4). In addition, engineering efficient terpene scaffold production in S. cerevisiae is of particular interest because yeast cells have good capability for the functional expression of cytochrome P450 enzymes, which are required for the synthesis of complex terpenoids. Last but not least, the established three-step pathway based on the IUP enabled our strain to achieve a 374-fold increase in the GGPP level compared with the native pathway. As such, this three-step pathway has a promising potential for diterpene and tetraterpene biosynthesis. Overall, our exploration demonstrated a universal and effective platform for supporting plant-derived terpenoid synthesis in S. cerevisiae.

Materials and Methods

A full Materials and Methods section is provided in SI Appendix.

Strain Engineering and Metabolite Analysis.

Strains used in this study were constructed in the wild-type S. cerevisiae strain, CEN.PK2-1C, through genomic integration employing a sequential overexpression strategy. When activating the IUP, the substrate isoprenol or prenol was added in the YPD medium after glucose depletion. The intracellular intermediates IP/DMAP, IPP/DMAPP, GPP, FPP, and GGPP in strains were extracted by ice-cold buffer (40% (v/v) methanol + 40% (v/v) acetonitrile + 20% (v/v) water) following 12-h incubation with isoprenol or prenol and then were quantified via LC-MS/MS. For terpene quantification, the organic layer in the medium culturing strains for volatile compounds (limonene, amorphadiene, and taxadiene) was harvested and assessed by GC–MS. β-amyrin was extracted by hexane following the cell that was broken using glass beads in methanol and qualified by GC–MS. Lycopene extraction was performed in dimethyl sulfoxide (DMSO) at 50 °C for 1 h and qualified by HPLC.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.