Introduction

With the advantages of mild reaction conditions and high reaction specificity, a vast number of microbial cell factories have been constructed in recent years to synthesize a variety of bulk chemicals through synthetic biology and metabolic engineering. Yeast, as a typical eukaryotic microorganism, plays an important role in the microbial cell factory; for example, Liu et al. (2020) achieved the compartmentalized synthesis of squalene in the peroxisome organelle of Saccharomyces cerevisiae. Among various yeasts, K. phaffii is a well-established GRAS (generally recognized as safe) eukaryotic host that is typically used for the expression and production of heterologous proteins due to its excellent posttranslational modification ability and tight methanol regulation mechanism (Ahmad et al. 2014; Yang and Zhang 2018). In contrast to S. cerevisiae, K. phaffii is a methylotrophic Crabtree-negative yeast and accordingly produces almost no ethanol as a byproduct during glucose fermentation. Additionally, K. phaffii could achieve high-density fermentation in basic media with methanol as the only carbon source to maintain cell growth and synthesize products. And the peroxisome organelles of K. phaffii could proliferate and expand under methanol-induced conditions (Bernauer et al. 2021; Ohsawa et al. 2022). Therefore, an increasing number of researchers have focused on yeast engineering to construct cellular factories of K. phaffii (Guo et al. S5).

Fig. 6
figure 6

Design and validation of a yeast combinatorial library. a Schematic illustration of a yeast combinatorial library. The library consists of three promoters, four genes, and four terminators. The scissor means the cleavage site of CRISPR/Cas9. UHA and DHA are the upstream and downstream homologous arms of ADH900 loci, respectively, and their lengths were both approximately 1000 bp. b The screening process for yeast libraries. c Fermentation results of multiple clones picked from yeast libraries. Results of fermentation show that the maximum yield was 25 times higher than the minimum yield. All samples were run in triplicate

Full size image

Next, 288 clones were transferred into 96 deep-well plates and cultured for 2–3 days (Fig. 6b). Then, 14 clones with different color grades were picked for sequencing and fermentation, with zw107 as a control. As expected, the results of both sequencing and HPLC analysis demonstrated that different combinations were generated in this yeast library and exhibited various phenotypes, and the lycopene yields varied from 7.28 to 182.73 mg/L (Fig. 6c). The maximum production was 25 times higher than the minimum. The sequencing results showed that strain TX-L3, in which crtI, crtYBW61R, tHMG1, and crtE were controlled by PGAP, PTEF1, PPGI1, and PTEF1, respectively, achieved maximum yield. The strain TX-L7, in which crtI, crtYBW61R, tHMG1, and crtE were controlled by PPGI1, PGAP, PPGI1, and PGAP, respectively, achieved minimum production. Moreover, we found that in higher-yielding strains, crtE (encoding GGPP synthases) and crtI (encoding phytoene desaturase), which catalyze the synthesis of GGPP from FPP and lycopene from phytoene, respectively, were under the control of the stronger promoter PGAP or PTEF1, suggesting that GGPP synthases and phytoene desaturase are two of the key enzymes in lycopene synthesis, which is consistent with previous reports (Zhang et al. 2020). Thus, it is practical and rapid to coordinate the expression levels of individual genes in the metabolic pathway to achieve a better combination.

Discussion

The methylotrophic yeast, K. phaffii, has been increasingly engineered into cell factories for the synthesis of various secondary metabolites in recent years (Cai et al. 2022; Guo et al. 2021; Zuo et al. 2022). To enable the construction of cell factories, several CRISPR-based gene editing systems have been established in K. phaffii (Cai et al. 2021; Liu et al. 2019; Nishi et al. 2022; Weninger et al. 2015, 2016; Zhang et al. 2021); however, compared to S. cerevisiae, K. phaffii still has potential for improvement in marker recycling, multifragment integration, etc. In this research, Kp6 (CBS7435ΔKu70) and Kp9 (CBS7435ΔKu70ΔDNL4) were constructed, and the KpHis4 locus was selected as the target to evaluate the homologous recombination efficiency of the two strains. The result showed better homologous recombination efficiency in Kp9, which is consistent with previously reported research results (Ito et al. 2018; Nishi et al. 2022). In addition, our previous experiments showed that the growth of the K. phaffii ΔURA3 strain was impaired, which is consistent with previously reported research results (Lin Cereghino et al. 2001); therefore, rapid marker recycling could not be performed similarly to that in S. cerevisiae on screening plates supplemented with 5-fluoroorotic acid (Kotaka et al. 2009; Moon et al. 2022). To simplify the marker recycling process, we redesigned the gRNA plasmid to determine whether plasmid elimination had been achieved through the visualization of green fluorescent protein. Compared to plasmid elimination methods in other systems (Liao et al. 2021; Liu et al. 2019; Weninger et al. 2018; Yang et al. 2020), the method we designed takes less time and may be more conducive to iterative gene editing. In addition, the GFP in the plasmid could assist in screening positive transformants.

When constructing cell factories, the number of genes integrated in one step is one of the factors that affect the efficiency of the construction. Although Nishi et al. (2022) realized the integration of four heterologous gene expression cassettes totaling 7.8 kb, which were divided into 10 fragments, in a single transformation in K. phaffii, there is still a gap compared to S. cerevisiae in which up to 15 fragments can be integrated in a one-step transformation at a single locus (Jakociunas et al. 2015). In this study, the one-step integration of 11 expression cassettes for a complete lycopene synthesis pathway with a total length of 26.5 kb at a single locus was achieved in K. phaffii for the first time. To the best of our knowledge, both the number of fragments and the total length of the fragments are by far the highest integrated in K. phaffii to date. Previously, NHEJ-based knockout of large fragments was performed by introducing DSBs via CRISPR/Cas9 in K. phaffii, and a positive rate of ~ 40% was achieved for a 4000-bp DNA fragment, but only ~ 2% of transformants were cleanly eliminated (Schusterbauer et al. 2022). Additionally, although Schusterbauer et al. (2022) achieved a “nonclean knockout” of a 100-kb DNA fragment in K. phaffii, the transformants lost the complete 3′ end of a chromosome rather than showing precise knockout of a large fragment inside a chromosome, as in our work. In addition to CRISPR/Cas9-based large fragment knockout, Zhang et al. (2021) realized CRISPR/Cpf1-based large fragment knockout with 11% efficiency for 20 kb. In this study, we first precisely knocked out a 27-kb DNA fragment inside a chromosome by introducing two DSBs at both ends of the fragment to be knocked out, thereby completely eliminating the lycopene synthesis pathway with 50% efficiency, which represented superior efficiency and length to those of the knockout achieved in the CRISPR/Cpf1 system (Zhang et al. 2021). Subsequently, three neutral targets, II-4, II-5, and II-6, were selected for testing larger DNA fragment knockouts in this study. However, in several repetitions of the experiment, no transformants successfully grew. The results of sequence alignment of proteins in the NCBI database indicated that the genes encoding the ribosomal biosynthesis protein RLP24 and the large ribosomal subunit biogenesis protein JIP5 were included in the region between II-4 and II-5, both of which are homologous analogs that are essential in S. cerevisiae for biogenesis of the large ribosomal subunit. Therefore, we speculated that the lack of transformants might be due to gene loss preventing the growth of cells. Further studies may be needed to determine the deletion of larger fragments. These results show that our tool could be used to efficiently and rapidly edit genomes, accelerating metabolic network rearrangements and facilitating the construction of cell factories.

We first constructed yeast combinatorial libraries in K. phaffii for the fast balancing of metabolic pathways, which has been demonstrated previously in S. cerevisiae (EauClaire et al. 2016). Through the yeast combinatorial library, we successfully obtained different lycopene synthesis strains with yields up to 182.73 mg/L in shake flask fermentation. This result suggests that in constructing metabolic pathways, not all genes need to be under the control of a strong promoter. Instead, the metabolic pathway should be optimized by combining the demands of the metabolic flux and the catalytic efficiency (specific activity) of the enzyme itself. In previous reports, the ratio of multiple enzymes associated with product synthesis was determined by individually overexpressing each protein and measuring the resulting catalytic capacity in vitro, which in turn informed the optimization of the metabolic pathway in vivo (Liu et al. 2017). This is an effective strategy to improve product synthesis. However, in many cases, the catalytic capacity of each individual enzyme is not easily determined. For example, some enzymes may be difficult to heterologously express or purify, or their substrates/products may not be easily detected, which makes the implementation of this strategy difficult. Other researchers optimized the copy number of each gene to determine the rate-limiting step and enhance the yield of the product (Lv et al. 2019), which involves an extensive and time-consuming workload when the metabolic pathway involves a large number of genes. In contrast, yeast combinatorial libraries could effectively facilitate the regulation and assembly of metabolic pathways. Additionally, this strategy could be extended to the free assembly of other elements, such as different genes, terminators, and transcriptional regulators, making the construction of cell factories more efficient and more time-saving.

In conclusion, this work further expands the application of CRISPR/Cas9 in K. phaffii; provides newly developed rapid, efficient, and continuous genome editing tools; and establishes a flexible and concise genome editing method. In this research, marker recycling, multifragment integration, and large fragment knockout were optimized, and yeast combinatorial libraries were introduced for the first time. In summary, the CRISPR/Cas9 system described in this study provides a more efficient tool for the construction of K. phaffii cell factories.