De novo biosynthesis of a vinblastine precursor in Pichia pastoris

Plant natural products have been widely used in medicine, food, and cosmetics. Of a particular note, many natural products from Catharanthus roseus have important biological activities and medicinal value [1], with vinblastine as one of the most effective drugs for cancer treatment [2]. Due to difficulties in chemical synthesis of such complex structures, plant extraction is still the major method for large-scale preparation of plant natural products. In recent years, with the development of synthetic biology, there is a growing interest to synthesize plant natural products using microbial cell factories [3,4], demonstrating benefits in low cost, green manufacturing, and sustainability. However, grand challenges in the manipulation of multi-gene biosynthetic pathways and improvement of expression levels and activities of plant proteins remain to be addressed for wide applications.

The methylotrophic yeast strain Pichia pastoris possesses one of the most popular protein expression systems and has been employed for the expression, preparation, and characterization of thousands of recombinant proteins [5] such as insulin, human serum albumin, and epidermal growth factors. Despite the poor expression and low activity of key enzymes of the plant natural product biosynthetic pathways, such as polyketide synthase (PKS) and cytochrome P450 enzyme (CYP), P. pastoris is still a promising chassis for producing plant natural products due to its ability to express eukaryotic proteins at high levels. Luckily, several groups have established CRISPR/Cas9 based genome editing system in P. pastoris, enabling multiplex genome integration of heterologous polygene biosynthetic pathways with high efficiency [[6], [7], [8], [9]]. Accordingly, several value-added compounds, such as carotenoids [6], fatty acid derivatives [10], and α-santalene [11], have been effectively synthesized in P. pastoris. The baicalein and oroxylin A, which are the well-known plant derived products, are also successfully synthesized in P. pastoris by introducing the complicated pathway involving several CYP enzymes using CRISPR/Cas toolkits [12]. This is the first instance of heterologous production of a plant medicine in P. pastoris. The final yield of these products is higher than that was previously reported in model yeast strains [13], indicating P. pastoris is an excellent host for plant-derived metabolites. However, the reconstitution of biosynthetic pathways as complex as vinblastine is yet to be explored in P. pastoris.

In a recent work published in Nature Synthesis [14], Jiazhang Lian and co-workers from Zhejiang University engineered P. pastoris for de novo biosynthesis of catharanthine, a vinblastine precursor. Considering the complexity of catharanthine biosynthesis, the authors divided the biosynthetic pathway into three functional modules for construction and optimization (Fig. 1). With strictosidine and nepetalactol as the branching points, the pathway was divided into the CAN module (from strictosidine to catharanthine), the STR module (from nepetalactol to strictosidine), and the NPT module (from carbon sources to nepetalactol). Using such a modular pathway construction and optimization strategy, bottlenecks in catharanthine biosynthesis were identified and subsequently addressed by gene amplification of the rate-limiting enzyme encoding genes, selection of appropriate plant enzymes with high activity and/or specificity, and expression of fusion protein to facilitate substrate trafficking. With the recombination of host metabolic engineering and bioprocess engineering, the authors accomplished the full biosynthesis of catharanthine from simple carbon sources for the first time, achieving the highest titer of 2.57 mg/L in a fed-batch bioreactor.

Fig. 1.

Complete biosynthesis of catharanthine in P. pastoris. The authors divided the biosynthetic pathway into three functional modules for construction and optimization. With strictosidine and nepetalactol as the branching points, the pathway was divided into the NPT module (from carbon sources to nepetalactol), the STR module (from nepetalactol to strictosidine), and the CAN module (from strictosidine to catharanthine) [15], as represented in blue, salmon and gray, respectively.

The development of synthetic biology is always accompanied with the discussion on the optimal chassis for biotechnology applications. Compared with prokaryotic cell factories (e.g., Escherichia coli), the ability of post-translational modifications and the presence of inner membrane systems for functional expression of CYPs make yeasts preferred hosts for the biosynthesis of plant natural products. Noteworthy, the model yeast Saccharomyces cerevisiae is currently a favorite chassis for plant natural products, with the success in the production of artemisininic acid [16], opioids [17], noscapine [18], and scopolamine [19]. More importantly, part of the vinblastine biosynthetic pathway, such as the biosynthesis of strictosidine [20] from carbon sources and the conversion of tabersonine to vindoline, have been reconstituted in S. cerevisiae, which might be the reason for Prof. Keasling to choose this chassis for the reconstitution of the catharanthine and vindoline biosynthetic pathways [21]. Considering the advantages in high levels expression of plant proteins, Lian and co-workers chose P. pastoris as the chassis for the reconstitution of catharanthine biosynthetic pathway. Using fed-batch fermentation, the yield of catharanthine in P. pastoris (∼2.57 mg/L) is much higher than that in S. cerevisiae (∼0.09 mg/L), highlighting the advantage and potential of P. pastoris in large-scale production of vinblastine precursors as well as other plant natural products.

Overall, the authors constructed a highly complex pathway for the biosynthesis of catharanthine in a non-model yeast for the first time. The CRISPR/Cas9 based genome integration toolkit established in this study can be employed for rapid construction of multi-gene biosynthetic pathways for metabolic engineering applications. Nevertheless, it should be acknowledged that the synthetic biology tools are still not as well developed as those in S. cerevisiae. For example, while the traditional pathway optimization method significantly enhanced the performance of CAN module and NPT module, its contribution to the STR module was rather limited, indicating that more delicate pathway optimization tools should be established in near future. With the rapid growing of the P. pastoris community, it will be established as one of the most promising chasses to produce plant natural products.