Reconstitution of the vindoline biosynthetic pathway in yeast
As the biosynthetic pathway from tabersonine has been fully elucidated with several CYPs involved (i.e., T16H2 and T3O, Fig. 1a), VSY002, in which CrCPR and CrCYB5 had been integrated, was firstly constructed for the introduction of vindoline biosynthetic pathway. Unfortunately, the production of vindoline in strain VSY006, containing a single copy of each gene of the whole vindoline biosynthetic pathway, was nearly undetectable with MRM mode or scan mode (as low as ~4.2 ng/L). On the contrary, the by-product vindorosine, which could be synthesized from tabersonine by the downstream pathway (T3O/T3R, NMT, D4H, and DAT), was accumulated to high levels (Fig. 2). These results indicated that the T16H2 and 16OMT activities might be the bottlenecks for vindoline biosynthesis, which was consistent with the previous study17. To verify the hypothesis, the plasmid pESC-LEU2d-T16H2-16OMT was transformed into the VSY006 strain. Compared with VSY006, the introduction of T16H2 and 16OMT on a multi-copy plasmid (strain VSY007) significantly increased the vindoline production (13.8 μg/L). Meanwhile, the accumulation of vindorosine was dramatically decreased and the production level was ~2.34 fold lower than that of vindoline (Fig. 2a). The LC-MS SIM mode profiles clearly indicated the metabolic shift from vindorosine (tR = 34.02 min, m/z = 427.05) accumulation in VSY006 to vindoline (tR = 33.16 min, m/z = 457.05) biosynthesis in VSY007 (Fig. 2b). These results revealed that increasing the copy numbers of T16H2 and 16OMT was beneficial for converting tabersonine to vindoline.
a Productivity changes of vindoline and vindorosine in strain VSY006 and VSY007, with the production level of vindorosine in strain VSY006 set as the reference value. b MS spectra of vindoline and vindorosine produced by strain VSY006 and VSY007, respectively. Error bars represented SD of biological triplicates (n = 3).
Improving vindoline production via biosynthetic pathway optimization
Considering that higher copy numbers of T16H2 and 16OMT were required for vindoline production and the plasmid system was not suitable for practical applications, multiplex and multi-copy genome integration of the biosynthetic pathway genes was employed using the CRISPR/Cas9-mediated genome editing tool (Supplementary Fig. S1). As shown in Fig. 3, the integration of another copy of T16H2 and 16OMT expression cassettes into strain VSY006 increased the production of vindoline to 130.3 μg/L (VSY008), which was also significantly higher than that of the plasmid system (strain VSY007), indicating the advantage of genome integration for cell factory development. Further integration of another copy of the other pathway genes into VSY008 resulted in the construction of VSY009, which harbored two copies for each gene of the vindoline biosynthesis pathway. The production of vindoline in strain VSY009 was increased ~1.71-fold to 221.9 μg/L, when compared with strain VSY008. However, vindorosine was still accumulated to high levels, indicating that T16H2 and 16OMT remained the bottlenecks for high-yield conversion of tabersonine to vindoline.
With VSY006 as the parent strain, additional copies of T16H2–16OMT expression cassettes and other pathway gene expression cassettes were introduced by CRISPR/Cas9-mediated multi-copy genome integration technology. Error bars represent SD of biological triplicates (n = 3).
To further enhance 16-hydroxytabersonine and 16-methyoxytabersonine synthesis and accordingly direct the metabolic flow toward vindoline production, additional copies of T16H2 and 16OMT were further integrated into VSY009, resulting in the construction of VSY014 (three copies of T16H2 and 16OMT), VSY015 (four copies of T16H2 and 16OMT), and VSY016 (five copies of T16H2 and 16OMT). VSY015 harboring four copies of T16H2 and 16OMT and two copies of the remaining pathway genes resulted in the highest production of vindoline, which was 3.2- and 1.88-fold higher than that of VSY008 and VSY009, respectively. Integration of extra copies of T16H2 and 16OMT failed to further increase the conversion of tabersonine into vindoline (Fig. 3). Therefore, the biosynthetic pathway with four copies of T16H2 and 16OMT and two copies of T3O, T3R, NMT, D4H, and DAT were optimal for vindoline conversion, and VSY015 was chosen for subsequent optimization.
Engineering CYPs microenvironment and increasing co-substrate availability for increased vindoline production
As mentioned above, two plant CYPs (T16H2 and T3O) were involved in vindoline biosynthesis. Considering the pivotal role of CPRs in NADPH-mediated electron transfer and the requirement of endoplasmic reticulum (ER) membrane localization for CYPs activities, the following strategies were employed to improve the microenvironment of CYPs for optimal performance: (1) CYP-CPR pairing; (2) ER expansion; (3) NADPH supply enhancement.
Previous studies demonstrated that the source of CPRs could affect the activities of CYPs and the native CPRs were not necessarily the best partner for CYPs in a heterologous host18,19,20. Therefore, CYPs should be appropriately paired with CPRs for optimal function and accordingly vindoline biosynthesis. In yeast strain VSY015, the native CPR from C. roseus (CrCPR) was integrated. To investigate the effect of CYPs-CPRs pairing on vindoline biosynthesis, four different CPR-encoding genes including AtCPR1 (Arabidopsis thaliana), GuCPR1 (Glycyrrhiza uralensis), GlCPR (Ganoderma lucidum), and MTR2 (Medicago) were chosen to replace CrCPR in VSY015, resulting in the construction of VSY017, VSY018, VSY019, and VSY020, respectively. The replacement of CrCPR with AtCPR1 resulted in significant improvements in vindoline production (∼1.8-fold, 662.4 μg/L), while a comparable titer was observed for all the remaining CPR replacement (Fig. 4). Interestingly, the integration of an additional copy of CrCPR into VSY015 and VSY017 resulted in decreased vindoline production, 253.4 μg/L in VSY021 and 455.8 μg/L in VSY022, respectively.
Five CPRs with different origins (CrCPR from C. roseus, AtCPR1 from A. thaliana, GuCPR1 from G. uralensis, GlCPR from G. lucidum, and MTR2 from Medicago) were integrated into the yeast genome and their effects on vindoline production were investigated. All strains were cultured in SC with 2% galactose in the presence of 50 mg/L tabersonine. Error bars represent SD of biological triplicates (n = 3).
As AtCPR1 demonstrated the best compatibility with the two C. roseus CYPs (T16H2 and T3O), an additional copy of AtCPR1 was integrated into VSY017. Unfortunately, the production of vindoline was decreased in VSY017-2 (Supplementary Fig. S2), indicating the benefits of overexpression of AtCPR1 was counteracted by the endogenous CPR (ScCPR), encoded by NCP121. ScCPR has been found to possess low compatibility with plant CYPs and might interfere with the electron transfer between AtCPR1 and C. roseus CYPs (T16H2 and T3O). Therefore, to further explore the CYP-CPR interaction for optimal catalytic activity of CYPs, NCP1 was deleted and the effect on vindoline production was investigated in the present study. VSY017 was further engineered by inserting CPRs and CrCYB5 expression cassettes into the NCP1 locus, resulting in the construction of VSY017-3, VSY017-4, VSY017-5, VSY017-6, and VSY017-7, respectively. As expected, the integration of an additional copy of CPRs into the NCP1 locus of VSY017 increased vindoline production, with the highest production achieved in VSY017-3 (1264.2 μg/L), representing a two-fold increase over the parental strain VSY017 (Supplementary Fig. S2). In other words, the deletion of NCP1 in the yeast strain harboring two copies of AtCPR1-CrCYB5 increased vindoline production by about 2.6-fold. Based on these results, two genomic copies of AtCPR1 together with NCP1 deletion contributed to the best electron transfer compatibility with CYPs of the vindoline biosynthetic pathway.
As plant CYPs and CPRs are generally ER-localized membrane proteins22, the expansion of ER should enable higher enzymatic activities by providing more space for the folding of CYPs and CPRs23. Previous investigations have demonstrated that ER expansion could be achieved by the deletion of PAH123, the deletion of OPI1, and the overexpression of INO224. Although attempted multiple times, the deletion of PAH1 was not successful in the vindoline-producing strain. The discrepancy with the previous study might result from the use of yeast strains with different genetic background. Therefore, the overexpression of INO2 and the deletion of OPI1 were employed in the present study. In addition, NADPH functions as the cofactor for CPRs mediated electron transfer and the increased NADPH availability has been shown to enhance CYPs activities25. The introduction of GAPN26 (NADP+-dependent glyceraldehyde-3-phosphate dehydrogenase from Streptococcus mutans) and overexpression of ZWF1 were commonly employed strategies to enhance NADPH supply. Enhancing the supply of S-Adenosyl-Methionine (SAM) via the overexpression of SAM2 has been demonstrated to improve the methyltransferase activity and accordingly the biosynthesis of secondary metabolites in yeast as well5. Therefore, VSY017 was further engineered by inserting the GAPN and ZWF1 expression cassettes into the OPI1 locus (VSY023), followed by the integration of SAM2 and INO2 expression cassettes (VSY024). The production of vindoline in VSY023 and VSY024 was increased to 1339.0 μg/L and 1662.7 μg/L, respectively (Fig. 5). Overall, the production of vindoline was increased for about fourfold (from ~417.1 μg/L in VSY015 to ~1662.7 μg/L in VSY024) via engineering of the CYP microenvironment, including CYPs-CPRs pairing, ER expansion, and NADPH supply enhancement.
With VSY017 as the parent strain, OPI1 was deleted and INO2 was overexpressed for ER expansion, ZWF1 and GAPN were overexpressed to enhance NADPH supply, and SAM2 was overexpressed to increase SAM availability. All strains were cultured in SC with 2% galactose in the presence of 50 mg/L tabersonine. Error bars represent SD of biological triplicates (n = 3).
Optimization of fermentation conditions for the biosynthesis of vindoline
To further increase vindoline production, the fermentation conditions were briefly optimized in shaker flasks. First, SC and YP medium were investigated with different concentrations of tabersonine (10, 25, 50, 75, 100, 125 mg/L). Generally, more vindoline could be produced if a higher concentration of tabersonine was supplemented. As shown in Fig. 6a, compared with the synthetic medium, the rich medium YP led to a pronounced increase in vindoline titers (as high as ~5.8 mg/L), particularly under low tabersonine concentration. The better performance might result from improved enzyme expression levels and better cell growth in the rich medium. Nevertheless, in the YP medium, the conversion yield of tabersonine to vindoline was dramatically decreased with higher substrate supplementation (Fig. 6a).
a The strain (VSY024) was cultured in SC (shown in blue) or YP (shown in red) medium with 2% galactose in the presence of various concentrations of tabersonine at 30 °C. Conversion yield was calculated by the conversion of tabersonine to vindoline, with the pink line representing the conversion yield in SC medium and the black line for that in YP medium. Error bars represent SD of biological triplicates (n = 3). b The accumulation of vindoline and intermediate metabolites in strain VSY024 when different amounts of tabersonine were supplemented into YP medium. Data are average of biological triplicates (n = 3).
To explore the possible reasons, the accumulation of the intermediate metabolites was analyzed when different concentrations of tabersonine were supplemented into YP medium. As shown in Fig. 6b, the percentage of 16-OH-tabersonine (light blue box) and 3-OH-16-MOH-2,3-2H-tabersonine (light green box) were increased with higher substrate supplementation, indicating that vindoline production was limited by the methyltransferases (16OMT and NMT). In addition, the percentage of 3-OH-2,3-2H-tabersonine (light orange box), an intermediate metabolite of the vindorosine pathway, was increased with higher substrate supplementation, implying that the promiscuity of T3O/T3R was another reason for low yield vindoline biosynthesis.
To address the dilemma between titer and yield, a low concentration of tabersonine (15 mg/L) was fed into the fermentation broth every 24 h after galactose induction in VSY024 and VSY025. With a similar amount of tabersonine supplemented (roughly 100 mg/L), the intermittent supply of the substrate at a lower concentration resulted in improved both titer and yield of vindoline, whose maximal titer reached 11.7 mg/L and 16.5 mg/L, respectively (Fig. 7a). In addition, by keeping tabersonine at a low concentration, the accumulation of vindorosine and its biosynthetic intermediates were largely decreased, while the vindoline biosynthetic intermediates (such as 3-OH-16-MOH-2,3-2H-tabersonine and desacetoxyvindoline) were still accumulated to high levels (Fig. 7b). In other words, the concerns with the promiscuity of T3O/T3R could be addressed to some extent by lowering tabersonine concentration, while the biosynthetic pathway should be further optimized to minimize intermediate accumulation.
a Enhancing vindoline production in YP medium via feeding low concentration of tabersonine (~15 mg/L) every 24 h after galactose induction. A total of ~100 mg/L tabersonine was fed into the fermentation broth. Error bars represent SD of biological triplicates. b The accumulation of vindoline and intermediate metabolites in strain VSY025 by intermittent feeding of low concentrations of tabersonine and high cell density yeast biotransformation (BioT). Data are average of biological triplicates (n = 3).
Finally, to compare the performance of VSY025 with the previously constructed vindoline producing strain by Qu et al.17, which harbored the biosynthetic pathway genes on multi-copy plasmids, yeast biotransformation assay was carried out under the same conditions. With the supplementation of ~75 mg/L, while the strain constructed by Qu et al. could produce ~1.1 mg/L vindoline, VSY025 was able to produce 17.7 mg/L and 29.4 mg/L vindoline in SC medium and YP medium, respectively (Supplementary Fig. S3), indicating the advantages of the genome-integrated strains in the biosynthesis of natural products. More importantly, tabersonine biotransformation using high concentration of yeast cells not only decreased by-product formation, but also minimized intermediate accumulation, particularly in YP medium (Fig. 7b).
