Identifying new targets for improving terpenoid biosynthesis in Yarrowia lipolytica through random genomic cytosine base editing

Abstract

Background

Genome-scale mutagenesis integrated with high-throughput phenotypic screening and causal mutation mapping serves as a robust paradigm for systemic genetic dissection. Despite the application of non-homologous end joining (NHEJ)-mediated genome editing in Yarrowia lipolytica, the development of alternative genome-wide mutagenesis strategies remains unexplored in this industrially relevant oleaginous yeast.

Results

We developed the Helicase-Assisted (Helicase-CDA) system, a genome-wide mutagenesis platform integrating the helicase domain of Yarrowia MCM5 (Encoded by YALI1_A01766g) with cytidine deaminase (CDA). This system enables continuous C-to-T specific mutations at random genomic loci. Applied to an industrial β-carotene-producing Y. lipolytica strain, Helicase-CDA system generated a mutagenized library through 7-day subculturing. Through high-throughput screening, we successfully isolated the mutant strain CDA-14, which demonstrated a 25% enhancement in β-carotene production (448.1 mg/L) compared to the wild-type strain. Notably, its productivity of β-carotene reached 6.15 g/L in fed-batch fermentation. Whole-genome sequencing revealed a G1637A substitution in YALI1_B16239g, which encodes a membrane protein showing homology to sterol biosynthesis regulator MGA2. This mutation led to reduced ERG1 expression level and redirected central carbon flux toward carotenoid synthesis by perturbing isoprenoid precursor partitioning.

Conclusion

Helicase-CDA system circumvents the dependency on NHEJ-mediated whole-genome mutation approach, offering a robust tool for continuous genome evolution in pre-engineered industrial strains. This study not only enhances genome editing in Y. lipolytica but also identifies a practical target for optimizing terpenoid biosynthesis, demonstrating significant potential for applications in metabolic engineering and synthetic biology.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12934-025-02819-5.

Background

Metabolic engineering and synthetic biology have emerged as powerful tools for improving the production of high-value biochemicals in microbial hosts [1]. Among these, Y. lipolytica has gained prominence as a model organism due to its robust metabolism, high tolerance to stress, and ability to accumulate lipophilic compounds [2]. It has been widely applied in the synthesis of bulk chemicals, fatty acids, aromatic compounds, terpenoids, etc [3–7]. However, the complexity of cellular metabolic networks and incomplete understanding of context-dependent regulation continue to challenge rational engineering strategies for industrial trait improvement.

Genome-scale evolution has become an indispensable approach for developing robust cell factories, particularly when coupled with high-throughput screening methodologies [8]. While adaptive laboratory evolution (ALE) remains widely employed for phenotypic enhancement through spontaneous mutagenesis during serial passaging [9], recent advances in molecular biology tools have enabled more sophisticated genome editing strategies. Notable developments include RNAi-assisted genome evolution (RAGE) [10], CRISPR/Cas9-mediated homology-directed genome-scale engineering (CHAnGE), multifunctional genome-wide CRISPR systems (MAGIC) [11], and helicase-assisted cytidine deaminase fusion technologies [12]. A particularly promising approach involves the fusion of DNA replication-associated helicases with cytidine deaminases (CADs), exemplified by the Helicase-CDA base editing [12]. This technology leverages ssDNA-binding proteins fused with CADs to generate genome-wide C-to-T mutations, achieving remarkable success in enhancing industrial phenotypes across multiple microbial platforms - including β-carotene overproduction in Saccharomyces cerevisiae [12, 13], isobutanol tolerance in Escherichia coli [12], and stress resistance improvements in Corynebacterium glutamicum [14] and high concentrations of lovastatin production in Monascus purpureus [15]. Despite these advances, the application of such genome-wide mutagenesis tools remains unexplored in Y. lipolytica, representing a critical technological gap for this industrially important yeast.

In this study, we developed a novel genome editing platform for Y. lipolytica. We engineered a synthetic fusion protein combining a conserved helicase homolog YlMCM5, functionally analogous to S. cerevisiae MCM5, with cytosine deaminase (CDA) to enable programmable genome-wide C-to-T transitions via processive DNA unwinding-coupled base editing. Through continuous mutagenesis and fluorescence-activated cell sorting (FACS)-based high-throughput screening, we successfully isolated strains with significantly enhanced β-carotene production. Furthermore, through whole-genome resequencing and reverse genetic validation, we identified novel molecular targets contributing to carotenoid overproduction. This work not only establishes a powerful genome editing toolkit (Helicase-CDA system) for Y. lipolytica but also provides fundamental insights into the metabolic regulation of isoprenoid biosynthesis in oleaginous yeasts.

Materials and methods

Strains, media and growth conditions

Escherichia coli Trans1 T1 was used for plasmids construction and was cultivated at 37 °C in lysogeny broth (LB) medium supplemented with 100 µg/mL ampicillin or 50 µg/mL kanamycin. For flask fermentation, the Y. lipolytica strains were cultivated at 30 °C with shaking at 250 rpm in YPD medium with 30 g/L glucose (10 g/L yeast extract, 20 g/L peptone, and 30 g/L glucose) or YPDE medium (10 g/L yeast extract, 20 g/L peptone, 15 g/L glucose and 15 g/L erythritol) for 4 days. Synthetic complete medium (SC medium: 20 g/L glucose, 8 g/L SC (YGM003A; FUNGENOME; China), 18 g/L agar) without uracil was used to screen yeast transformants. Synthetic complete medium supplemented with 1 g/L 5-flfluoroorotic acid (5-FOA) was used to screen URA3 deficient strains.

Plasmids and strains construction

The plasmids used in this study are presented in table S2, and their construction is described in detail below and in Text S1. The primers are listed in table S1. The gRNA sequence used in this study are listed in table S3.

Plasmids were assembled using the Seamless Cloning Kit (Beyotime, China) or Golden Gate method [16]. To construct the plasmid pURA-MCM5-CDA, the backbone was PCR-amplified from pURA-YL-IntC2 using the primer pair Cen1-R/AmpP-cyc1-F. The GPD promoter and YLMCM5(YALI1_A01766g) were amplified from the chromosomal DNA of Y. lipolytica using the primer pairs GPD-CEN-F/GPD-R and MCM5-GPD-F/MCM5-V40-R, respectively. The CDA fragment, which includes a large peptide linker (100 amino acids) and an SV40 nuclear localization signal (NLS) at the N-terminus, was amplified from pRS315e_pGal-dCas9-pmCDA1 (Addgene #79615) using the primer pair SV40-NLS-F/CDMA-CYC1-R. The CYC1 terminator was amplified from S. cerevisiae using the primer pair SCCyc1T-F/SCCyc1T-R. These five PCR products were digested with DpnI and ligated using the Seamless Cloning Kit to generate pURA-MCM5-CDA. The sequence of pURA-MCM5-CDA is provided in Table S4. The main primers and plasmids are listed in Table S1 and Table S2.

All yeast strains used in this study (Table 1) were derived from YL002 [17]. The methods for transformation and removal of the URA3 selection marker were described previously [18]. The YL002 strain was engineered with a rational metabolic engineering strategy to enhance β-carotene biosynthesis through the co-overexpression of key enzymes involved in the mevalonate pathway components ERG12, ERG20, IDI, HMG1 and GGS from Y. lipolytica, and the carotenoid biosynthesis pathway GGPPSxd from Phaffia rhodozyma, three copies of CarB and CarRP from Mucor circinelloides(SEQ ID N0.3) (Fig. 1A) were overexpressed to improve β-carotene production. A flowchart describing the construction of β-carotene-producing Y. lipolytica is shown in Fig. 1B. All Y. lipolytica strains constructed in this study are listed in Table 1. The genes were integrated into the yeast genome using CRISPR/cas9 [17]. The construction methods for plasmids containing donor DNA or gRNA are described in Text S1. The key genes abbreviations and identifiers used in this study were listed in table S5.

Table 1.

Strains used in this study

Strains	Relative characteristics	Sources
YL002	Po1g-Δku70::Cas9-Hyg, Δura3	[17]
YL045	Yl002-intE3::ERG12-ERG20	This work
YL046	Yl045-intE1::GGPPxd-IDI	This work
YL061	Yl049-intC2::CarRP-CarB	This work
YL062	Yl061-intF1::tHMG1-GGS	This work
YL063	Yl062-intF2::CarRP-CarB	This work
YL064	Yl063-intA8::tHMG1	This work
YL065	Yl064-intF17::CarRP-CarB	This work
CDA-14	Screened from library	This work
CDA-44	Screened from library	This work
YL065-MGA2	YL065-intC3::MGA2	This work
CDA-14-MGA2	CDA-14-intC3::MGA2	This work
YL065-△MGA2	YL065 with MGA2 deletion	This work

Fig. 1.

Reconstituted β-Carotene Biosynthetic Pathway and Strain Engineering Workflow in this study. A: Genes highlighted in red indicate overexpression, while genes marked in blue represent suppressed expression. Metabolites: G6P, Glucose-6-phosphate; F6P, Fructose-6-phosphate; FBP, Fructose-1,6-bisphosphate; GAP, Glyceraldehyde-3-phosphate; Acety-CoA, Acetyl-coenzyme A; MVA, Mevalonic acid; MVA-5-P, Mevalonate-5-phosphate; MVA-5-PP, Mevalonate-5-diphosphate; HMG-CoA, 3-hydroxy-3-methylglutaryl-CoA; IPP, isopentenyl diphosphate; DMAPP, dimethylallyl diphosphate; GPP, geranyl diphosphate; FPP, farnesyl diphosphate; GGPP, geranylgeranyl diphosphate; DMAPP, dimethylallyl diphosphate. Enzymes: ERG10, acetyl-CoA acetyltransferase; ERG13, HMG-CoA synthase; ERG12, mevalonate kinase; ERG8, phosphomevalonate kinase; ERG19, mevalonate diphosphate decarboxylase; IDI, IPP isomerase; ERG20, farnesyl pyrophosphate synthetase; ERG9, squalene synthase; ERG1, squalene epoxidase; GGPPxd, geranylgeranyl diphosphate synthase; CarRP, phytoene synthase/lycopene cyclase; CarB, phytoene dehydrogenase; tHMGR, truncated 3-Hydroxy-3-Methylglutaryl-CoA Reductase; PDH, Pyruvate Dehydrogenase; ALDO, Fructose-bisphosphate Aldolase; PFK, Phosphofructokinase; PGI, Phosphoglucose Isomerase; HK, Hexokinase. B: Strain Engineering Workflow in this study.

Random base editing procedure

The β-carotene-producing Y. lipolytica strains YL065 was used for chromosome random base editing. YL065 colonies with pURA-YLMCM5-CDA or without plasmid were first inoculated in a 2% glucose SD-URA⁻ medium and grown at 30 °C overnight, which was then inoculated into fresh medium at a 1% inoculation ratio and transferred every 24 h. After 5 and 7 consecutive transfers, strains exhibiting high fluorescence intensity within the 510–530 nm range were selected using flow cytometry due to the β-carotene’s characteristic fluorescence profile (excitation: 488 nm; emission: 535 ± 25 nm). The mutagenesis protocol and high-throughput screening schematic for β-carotene-producing strains are diagrammatically illustrated in Fig. 4A.

Fig. 4.

Flow Cytometry-Based Screening (FACS)and Fermentation Optimization of β-Carotene-Producing Microbial Strains. A: FACS Workflow for Helicase-CDA System-Generated Mutant Libraries; B: Growth and β-carotene yield of High-Fluorescence Strains Enriched by Flow Cytometry; C: Growth and β-carotene yield of screened high β-carotene-producing strain from Mutant Libraries. Error bar represents standard deviation of triplicate experiments

Preliminary screening of high-β-carotene-producing mutants

The cytometric analysis and sorting of the chromosome random mutagenesis library were performed following the protocol described in a previous report [19]. Cytometric analysis and sorting of YL065 chromosome random mutagenesis library were executed with a flow cytometer (BD FACS Aria Fusion) using a 50-mW argon ion laser emitting at 488 nm. The cells were washed twice with 10 mM potassium phosphate buffer (pH 7.4), filtered through a 40-mm nylon mesh (assembled in a 5 mL polystyrene tube; cell-strainer cap; Nippon Becton, Dickinson) to remove accumulated cells, and then diluted with the same buffer to an optimum concentration. The fluorescence emission was measured at 510 nm to 560 nm. The self-defined high fluorescein area cells were sorted (around 15,000 cells). Fluorescence data were analyzed by Accuri CFlow software (Accuri Cytometers, Inc., Ann Arbor, MI, USA) and Flowjo software (FlowJo, LLC, Ashland, OR, USA) for statistical analysis.

The mutagenized libraries from days 5 and 7 were sorted using flow cytometry, and the top 0.05% of cells with the highest fluorescence intensity were collected as high-yield strains. For each sample, 2,000 cells were pooled and mixed, while an additional 100 colonies were streaked onto plates for further analysis.

Whole genome sequencing of evolved strains

Whole genome sequencing of CDA14 and CDA44 was performed together with YL065. Genomic DNA was extracted from the strains using a cetyltrimethylammonium bromide (CTAB)-based method and sequencing libraries were prepared using the standard library building process of the Illumina TruSeq DNAPCR-Free Prep Kit reagent. Libraries with a 350 insert were constructed for each sample and were subjected to paired-end sequencing using next-generation sequencing (NGS) based on the Illumina/BGI sequencing platform. There were, in total, 7,842,520, 8,072,614, and 7,588,142 reads for YL065, YL065-CDA-14, and YL065-CDA-44, respectively. The GATK software was used to analyze SNP and InDels. and were filtered with the following parameters: QD < 2.0 || MQ < 40.0 || FS > 60.0 || QUAL < 30.0 || MQrankSum < −12.5 || ReadPosRankSum < −8.0 -clusterSize 2 -clusterWindowSize 5. The SNPs identified by GATK were further filtered. Finally, High-quality SNPs were obtained. Exonic SNPs identified in all evolved strains as compared with Y. lipolytica (Yarrowia lipolytica genome assembly ASM176148v1 - NCBI - NLM) are provided in Supplementary Excel S1.

Fed-batch fermentation of the strains YL065 and CDA-14

Strains YL065 and CDA-14 were utilized for β-carotene production through fed-batch fermentation. Fed-batch fermentation was performed as described [20, 21] with the following modification. A two-stage system was used for preparing seed culture. In the first stage, fresh colonies were inoculated into a 100 mL flask containing 10 mL of medium, incubated at 30 °C with shaking at 250 rpm until reaching an OD₆₀₀ of 1.0–4.0. Subsequently, 1 mL from this initial seed culture was transferred to a 1 L flask containing 200 mL of medium and further incubated under the same conditions until the OD₆₀₀ reached 15. This second-stage seed solution was transferred to a 5 L bioreactor containing 2 L substrate medium. The fermentation substrate medium contained (per liter): 100 g glucose, 50 g yeast extract, 100 g peptone.

During the early fermentation phase, dissolved oxygen (DO) concentration was maintained at 30% by cascading with agitation speed and ventilation, the temperature was controlled at 30℃, and pH was maintained at 5.5 by automatic addition of 13% ammonia. Glucose content was measured every 4–6 h. Once glucose concentration dropped below 10 g/L, feed solution containing 20 g/L of yeast extract, 50 g/L of (NH₄)₂SO₄ and 500 g/L of glucose was added to restore the glucose concentration within the bioreactor to 80 g/L. The airflow rate and agitation speed were kept at the maximum value. Then the concentration of DO was settled at 30%.

Determination of squalene levels

Squalene content in YL065, CDA-14, and YL065-ΔMGA2 strains was measured according to the method reported previous [22] with some modification. YL065, CDA-14, and YL065-ΔMGA2 were cultured in YPD medium, and squalene levels were determined in samples collected at 48 h and 72 h of fermentation. Briefly, a 200 µL aliquot of the culture broth was centrifuged (12,000×g, 5 min), mixed with 3–4 mm glass beads, and resuspended in 1 mL acetone. Cells were disrupted by vortexing at 3,000 rpm for 10 min, followed by sonication for 5 min in an ultrasonic cleaner. The lysate was then incubated at 55 °C for 15 min in the dark. Afterward, the samples were centrifuged at 12,000×g for 10 min to obtain the supernatant containing carotenoids, which was used for HPLC analysis as described before [22].

RNA isolation and qRT-PCR

Flask cultures were inoculated into YPD liquid medium containing 3% (w/v) glucose and cultured for 24 h. After cultivation, 1 mL of cultures were collected by centrifugation for 5 min at 12,000 g at 4℃. Total RNA was extracted using the RNeasy Mini Kit (QIAGEN, Germany) according to the manufacturer’s protocol, with gDNA elimination performed using DNase I (Thermo Scientific, USA). cDNA was synthesized using the EasyScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, China). RT-qPCR analysis was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme, China) on an Eco Real-Time PCR System (Illumina, USA).

Primers for qRT-PCR are listed in Supplementary Table 1.The ACT1 gene (actin gene) served as the internal control for data normalization. Gene relative expression was calculated via the 2^−ΔΔCT method [23]. All experiments included biological triplicates.

Measurement of carotenoid production

Carotenoid extraction was performed as described before with the following modification [24]. Carotenoids were extracted using acetone following the same method as for squalene extraction, and the supernatant containing carotenoids was obtained and used for HPLC analysis.

The analytes were separated on a Symmetry C18 column (250 mm×4.6 mm, 5 μm, Waters, Ireland), which was kept at 30 °C, and were detected at 476 nm using a Shimadzu UV-2550 spectrophotometer (Shimadzu, Kyoto, Japan). The mobile phase and flow rate in measurement of carotenoids production were the same as that described previously [25]. The results represent the means ± SD of three independent experiments. Dry cell weight (DCW) was calculated from the optical density at 600 nm (OD₆₀₀) using the empirical formula 1 OD₆₀₀ = 0.39 g DCW L⁻¹.

Results

Constructing the Helicase-CDA molecular device in Y. lipolytica

To establish a CDA-mediated targeted mutagenesis system in Y. lipolytica, we employed the previously developed Target-AID (activation-induced cytidine deaminase) base editor [26]. For this system, we selected a helicase as the docking platform to fuse with the cytidine deaminase. By attaching the cytidine deaminase to the helicase, we aimed to exploit this unwinding process, enabling the enzyme to act directly on the exposed single-stranded DNA [12]. Mcm2-7 complex is the replicative helicase in eukaryotes, which is constituted of 6 different subunits [27]. One of the subunits, MCM5, was selected and fused to CDA to construct MCM5-AID complex expressed by plasmid pMCM5-AID [12].

Genomic alignment revealed that the protein encoded by YALI1_A01766g in Y. lipolytica shares 53% sequence similarity with the MCM5 helicase subunit of S. cerevisiae, suggesting evolutionary conservation of DNA replication fork unwinding functionality. Based on this phylogenetic evidence, we designated this protein as YLMCM5 and engineered a fusion construct with cytidine deaminase (CDA) from Petromyzon marinus using flexible peptide linkers, resulting in the pMCM5-CDA plasmid (Fig. 2A) for genome-wide C→T mutagenesis (Fig. 2B). The molecular mechanism involves YLMCM5-mediated recruitment of CDA to single-stranded DNA regions at replication forks during chromosomal DNA synthesis. CDA catalyzes deamination of cytosine to uracil (C→U) on nascent DNA strands, which is subsequently fixed as thymine (T) through replication, thereby achieving directional C/G→T/A base substitutions (Fig. 2C).

Fig. 2.

Mechanistic Basis of Helicase-CDA Genome Editing. A: Diagram of the plasmid of pURA-MCM5-CDA and Helicase-CDA Fusion Protein, The SV40 nuclear localization signal (NLS) directs efficient nuclear targeting of the fusion protein, while the helicase simultaneously unwinds duplex DNA and catalyzes CDA1-mediated deamination of cytosine (C) to uracil (U) during DNA replication. B: Schematic Diagram of Cytosine Deamination to Uracil. C: The mechanism of Helicase-CDA system mediated random base editing. The schematic is hypothesis, and may not reflect the actual mechanism

To validate the mutagenic capacity of this system while simultaneously uncovering novel targets for improving carotenoid biosynthesis, pMCM5-CDA would be transformed into a β-carotene-producing chassis strain. Then a genome-wide saturation mutagenesis library was constructed and subjected to high-throughput screening via fluorescence-activated cell sorting (FACS), leveraging the intrinsic autofluorescence of carotenoids to establish a phenotype-genotype correlation platform.

Strain design for high β-carotene production

To achieve dual objectives of functional validation for the MCM5-guided CDA mutagenesis system and discovery of novel β-carotene biosynthesis enhancers, a high-titer β-carotene-producing Y. lipolytica chassis strain was engineered. The previously engineered Y. lipolytica strain YL002, in which KU70 was deleted through the Cas9 expression cassette insertion and URA3 was deleted using CRISPR/Cas9 system, was used to construct β-carotene-producing strain [17]. The β-carotene synthetic pathway in yeast can be decomposed to the endogenous mevalonate (MVA) pathway section generating geranylgeranyl pyrophosphate (GGPP) as the precursor for carotenoids. In previous studies, overexpression of the genes ERG12, ERG20, IDI, HMG1, and GGS have been reported to enhance the supply of IPP and DMAPP, thereby increasing carotenoid production [7, 28]. Moreover, the GGPPxd derived from Xanthophyllomyces dendrorhous exhibits superior catalytic activity [20]. Additionally, the CarRP and CarB genes from Mucor circinelloides, encoding phytoene synthase/lycopene cyclase and phytoene dehydrogenase were used to catalyze the synthesis of β-carotene from GGPP. Notably, we employed an engineered CarRP variant (Y27R) that completely abolishes substrate inhibition while preserving catalytic efficiency, as previously characterized through structural enzymology studies [20]. Therefore, ERG12, ERG20, IDI, HMG1, GGS and the codon optimized GGPPxd, CarB, CarRP were overexpressed in Y lipolytica for improve carotenoid production.

Specifically, the ERG12 and ERG20 genes are inserted at the intE3 locus, the GGPPxd and IDI genes at the intE1 locus, and the HMG1 and GGS genes at the intF1 locus. The gene clusters (CarRP and CarB) were integrated into the genome of Y. lipolytica at the intC2, intF2 and intF17 loci to enhance β-carotene production (Figs. 1B and 3A). The engineered strain YL061 with co-expression of CarRP, CarB, ERG12, ERG20, GGPPxd and IDI successfully produced 31.81 mg/L, 2.84 mg/g DCW of β-carotene (Fig. 3B).

Fig. 3.

β-carotene production in engineered strains of Y. lipolytica A: Schematic representation of high-expression genetic elements, including gene, promoter, terminator, and insertion loci. B: Cell growth (OD₆₀₀), β-carotene production (mg/L), and content (mg/g DCW) in β-carotene-producing strains after 4-day cultivation in YPD medium. Error bar represents standard deviation of triplicate experiments

Previous research demonstrated that overexpression of 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase (HMGR1) significantly enhances terpenoid production [29, 30]. Furthermore, a truncated form of HMG1 (tHMG1), lacking the N-terminal 500 amino acids, has been shown to markedly increase β-carotene production and other terpenoids in Y. lipolytica [31]. In addition to tHmgR, expression of geranylgeranyl diphosphate synthase (GGS1) has also been reported to elevate the synthesis of carotenoid precursors [30, 31]. Based on these findings, the construct P_EXP-tHmgR-P_GPD-ggs1 was integrated into the YL061 genome at the intF1 locus, generating strain YL062 (Fig. 3A). As anticipated, strain YL062 demonstrated β-carotene production levels of 119.94 mg/L and 10.05 mg/g DCW, corresponding to 3.77-fold and 3.54-fold increases, respectively, compared to YL061 (Fig. 3B).

In addition, CarB, CarRP, and tHMG1 have been identified as bottlenecks in β-carotene synthesis in Y. lipolytica. To further enhance β-carotene production, additional copies of CarRP and CarB were expressed along with tHMG1 in strain YL062, resulting in the development of strains YL063, YL064, and YL065 (Fig. 1B). After the introduction of a third copy of CarRP and CarB into YL062, stain YL063 produced 236.71 mg/L and 18.69 mg/g DCW of β-carotene, representing improvements of 97.4% and 86.1%, respectively, compared to the parent strain YL062. The incorporation of tHMG1 into the YL063 strain further increased β-carotene production and yield to 247.27 mg/L and 20.66 mg/g DCW, reflecting improvements of 4.5% and 10.5% over the parent strain. Subsequently, the integration of a third copy of CarRP and CarB further enhanced β-carotene production and yield to 352.79 mg/L and 27.92 mg/g DCW, corresponding to increases of 42.7% and 35.1% relative to the parent strain. When compared to YL061, the production and yield in YL065 exhibited 11.1-fold and 9.8-fold enhancements, respectively (Fig. 3B).

YL065 contained three copies of CarRP and CarB, providing a sufficient pathway for β-carotene synthesis. Mutations that enhance precursor supply can be reflected in increased β-carotene production. Therefore, starting from this strain, whole-genome mutagenesis was performed to identify new targets for improving carotenoid synthesis in Y. lipolytica.

Constructing and sorting the chromosome random mutagenesis library

The pMCM5-CDA plasmid was transformed into the YL065 strain, and the transformants were cultured on SD-URA⁻ medium to induce whole-genome random mutagenesis. A fluorescence-activated cell sorting (FACS) strategy was previously established for efficient high-throughput screening of β-carotene-producing strains [19]. This approach utilizes the green fluorescence characteristics of β-carotene, which exhibits an excitation peak at 488 nm and an emission wavelength of 535 ± 25 nm. In this study, flow cytometry was employed to isolate high β-carotene-producing strains from mutagenized libraries generated via genome-wide mutation introduced by pMCM5-CDA.

First, cells mix of 2000 colonies sorted by flow cytometry from the random chromosomal mutagenesis library after 5/7 days continuous cultivation were cultured, and their β-carotene production was quantified to evaluate the impact of mutagenesis duration on yield. Second, a defined number of colonies with intense pigmentation from plate were selected and subjected to shake-flask fermentation to identify high-yield β-carotene-producing strains.

The results of fermentation are shown in Fig. 4B. Strains obtained after 5 days of continuous culture, the β-carotene production reached 398 mg/L, representing a 13% increase compared to the control strain. For strains obtained for 7 days, β-carotene production reached 438.43 mg/L, a 24% increase over the control strain. These results demonstrate that flow cytometry can effectively sort and enrich strains with high β-carotene production.

Fifty visually yellowish colonies were selected from the plates and fermented in shake flasks to measure β-carotene production. The production levels ranged from 214.84 to 447 mg/L, corresponding to 0.60–1.24 times the production of the original YL065 strain (Fig. S1). High-yield strains, including CDA-37, CDA-15, CDA-45, CDA-44, CDA-9, CDA-12, and CDA-14, were selected for repeated fermentation to obtain accurate data on cell growth and β-carotene production.

The results are shown in Fig. 4C. The selected strains exhibited similar cell growth, with OD₆₀₀ values ranging from 28 to 34.8, and their β-carotene production ranged from 88 to 125% of the parental strain. Among these, the β-carotene production of CDA-12, CDA-14, and CDA-44 exceeded that of the control strain, reaching 412.23 mg/L, 448.1 mg/L, and 416.95 mg/L, representing increases of 15%, 25%, and 17%, respectively.

The CDA-44 and CDA-14 were functionally validated to investigate if Helicase-CDA system could mediate genome-wide C→T mutation, and identify new target for improving β-carotene biosynthesis.

Performance validation of the Helicase-CDA system by whole-genome sequencing

Whole-genome analysis was conducted on CDA-14 and CDA-44, which exhibited significant and moderate improvements in β-carotene production, respectively, using YL065 as the control strain (Excel 1 in supplementary). Compared to the reference strain W29, after excluding the initial mutations introduced in YL065 (i.e., the three strains share identical bases but differ from the reference sequence), base deletions and insertions (indels), 90, 77, and 4 base substitutions were accumulated in CDA-14, CDA-44, and YL065, respectively(Fig. 5 and Excel 2 in supplementary).

Fig. 5.

Analysis and Validation of Nucleotide Mutations in Screened Microbial Strains A: Proportional Distribution of SNP Counts in CDA-14, CDA-44, and YL065; Pre-analysis Filtering of Shared Mutations in Strains CDA-14, CDA-44, and YL065 Relative to the Reference Genome; B: Mutation type of SNP in CDA-14, CDA-44, and YL065; C: Cell Growth and β-Carotene Production of Strain with Reverse Genetics Validation of YALI1_B16239g; D: β-Carotene yield of Strain with Reverse Genetics Validation of YALI1_B16239g; Error bar represents standard deviation of triplicate experiments.

Among the substitutions identified in strains CDA-14, CDA-44, and YL065, C: G → T: A transitions accounted for 61, 55, and 0 base changes, representing 68%, 71%, and 0% of their total substitutions, respectively. This pattern aligns with the known catalytic activity of cytidine deaminase (CDA), which specifically promotes C-to-T conversions, confirming the functional efficacy of the Helicase-CDA fusion in Y. lipolytica for inducing genome-wide C-to-T mutations. Notably, C:G → G: C transversions were undetectable in both CDA-14 and CDA-44, while the frequency of T: A → A: T transversions remained similar to that observed in the absence of Helicase-CDA. These results indicate that Helicase-CDA selectively suppresses C: G → G: C transversions without significantly affecting the rate of T: A → A: T transversions (Fig. 5B).

Analysis of mutations in high β-carotene-producing strain CDA-14

Given that the β-carotene yield of CDA-14 significantly exceeded that of CDA-44, we prioritized mutation analysis in CDA-14. Among all identified SNPs (Table 2), 27 and 25 mutations were located in upstream and downstream regions, respectively, while 17 were synonymous mutations. Intergenic, intronic, and stop-lost mutations accounted for 2, 1, and 1 instances, respectively, all of which are unlikely to substantially impact strain phenotype. Notably, 15 non-synonymous coding mutations and 3 nonsense mutations were identified, some of which may influence β-carotene production by altering transcriptional/translational efficiency or enzymatic activity. These 18 mutations mapped to 17 genes, categorized as follows: 1 storage function, 2 DNA/RNA-related, 1 flavin-linked, 4 membrane-associated, 1 inorganic ion transport, 2 molecular chaperones, 5 functionally uncharacterized, and 1 involved in metabolic pathways (Table 3). All the nonsynonymous, upstream region of genes and termination codon base conversions are listed in Table S3, along with their kegg suggested functions (https://www.kegg.jp/). Given the lipophilic nature of β-carotene, we prioritized functional analysis of four membrane-associated proteins. Bioinformatics annotation identified YALI1_B16239g as a putative MGA2 homolog.

Table 2.

Distribution of SNP variant types and their counts in CDA-14

Mutation Category	Numbers
DOWNSTREAM	25
NON_SYNONYMOUS_CODING	15
stop	3
INTERGENIC	2
intron	1
SYNONYMOUS_CODING	17
UPSTREAM	27

Table 3.

Gene functional classification and annotations of the NON_SYNONYMOUS_CODING and stop mutation of CDA-14

Functionl Category	Base Mutation	Gene ID	Related Annotation
Storage Function	Ctc/Atc	Gene-YALI1_C14738g	Storage vacuole, lytic vacuole
	Ctc/Atc	gene-YALI1_C14738g	Storage vacuole, lytic vacuole
DNA and RNA Related	Gaa/Aaa	gene-YALI1_C13302g	DNA replication initiation
	Ctt/Ttt	gene-YALI1_D00577g	RNA processing
flavin-linked	Gat/Aat	gene-YALI1_D34258g	flavin-linked sulfhydryl oxidase activity
Membrane-related	Gtt/Att	gene-YALI1_F25309g	transmembrane transport
	Ttt/Ctt	gene-YALI1_D16990g	integral component of membrane
	cGa/cAa	gene-YALI1_B16239g	integral component of membrane
	cGc/cAc	gene-YALI1_E27273g	protein-cysteine S-palmitoyltransferase activity
Inorganic ion transport	Ggt/Agt	gene-YALI1_B28659g	Inorganic ion transport and metabolism
Molecular Chaperone	Ggc/Agc	gene-YALI1_D00231g	Posttranslational modification, protein turnover, chaperones
	Cag/Tag	gene-YALI1_C29052g	Posttranslational modification, protein turnover, chaperones
Function Unknown	cGa/cAa	gene-YALI1_D27547g	Uncharacterized protein
	gCa/gTa	gene-YALI1_F25212g	Uncharacterized protein
	Tcc/Gcc	gene-YALI1_D00040g	Uncharacterized protein
	Cca/Tca	gene-YALI1_B00711g	Uncharacterized protein
	Caa/Taa	gene-YALI1_F25212g	Uncharacterized protein
	Cag/Tag	gene-YALI1_D01796g	Uncharacterized protein

Given the lipophilic nature of β-carotene, we prioritized functional analysis of four membrane-associated proteins. Bioinformatics annotation identified YALI1_B16239g as a putative MGA2 homolog. Sequence alignment revealed weak homology between YALI1_B16239g and S. cerevisiae S288c MGA2, a membrane-anchored transcriptional regulator that coordinates unsaturated fatty acid biosynthesis through dual modulation of OLE1 (encoding Δ9 fatty acid desaturase) transcription and mRNA stability [32, 33]. Notably, S. cerevisiae mga2Δ mutants exhibited a 5.5-fold increase in squalene accumulation compared to wild-type strains. Previous studies have demonstrated that MGA2 regulates sterol biosynthesis by activating ERG1 (squalene epoxidase) expression [34], a pathway competing with carotenoid synthesis for shared precursors (IPP and DMAPP). To investigate whether YALI1_B16239g serves as a novel engineering target for carotenoid overproduction, we performed reverse genetic manipulation of this gene to assess its impact on β-carotene yield, with particular focus on its potential role in redirecting terpenoid precursor flux.

we overexpressed YALI1_B16239g in parental strain YL065 and high-producing strain CDA-14 and the result stains named YL065-MGA2 and CDA-14-MGA2. As shown in Fig. 5C, overexpression strains YL065-MGA2 and CDA-14-MGA2 produced 269.14 mg/L and 275.18 mg/L β-carotene, representing 75% and 61% of their respective parental strains’ β-carotene production (Fig. 5C). Conversely, MGA2 knockout in YL065 (YL065-ΔMGA2) reduced total β-carotene production to 85% of wild-type levels, while demonstrating a 27% increase in β-carotene yield per dry cell weight (36.18 mg/g DCW) (Fig. 5D).

The levels of squalene accumulation and ERG1 transcription in YL065, YL065-ΔMGA2 and CDA-14

To investigate the relationship between ERG1 transcription levels, squalene accumulation and MGA2 alteration, squalene levels were measured in YL065, YL065-ΔMGA2, and CDA-14 strains at 48 and 72 h of fermentation, while ERG1 transcription was analyzed by qRT-PCR at 24 h.

The results showed that, regardless of the fermentation time, the squalene production in YL065-ΔMGA2 and CDA-14 strains was higher than that of the YL065 strain. Specifically, the squalene accumulation in CDA-14 at 48 h and 72 h was 5.6% and 8.0% higher, respectively, compared to the control strain. The squalene accumulation in YL065-ΔMGA2 increased by 65.4% and 24.6% at these two time points, respectively(Fig. 6AB). Squalene accumulation followed the order: YL065 < CDA-14 < YL065-ΔMGA2.

Fig. 6.

The levels of squalene accumulation and ERG1 transcription in YL065, YL065-ΔMGA2 and CDA-14. A: Squalene yield after 48 h fermentation; B: Squalene yield after 72 h fermentation; C: Relative expression levels of ERG1 after 24 h fermentation

The qRT-PCR analysis revealed that after 24-h fermentation, ERG1 transcription in CDA-14 and YL065-ΔMGA2 was reduced by 15% and 68%(Fig. 6 C), respectively, relative to YL065, indicating that MGA2 deletion represses ERG1 transcription, thus likely enhancing β-carotene synthesis.

Fed-batch fermentation of strains YL065 and CDA-14 for β-carotene production

To validate enhanced β-carotene production by strain CDA-14 under fed-batch fermentation, YL065 and CDA-14 were cultivated in 5-L bioreactors at pH 5.5. Prior to conducting fed-batch fermentation, the LEU and URA3 genes were supplemented into the YL065 and CDA-14 strains to eliminate the need for amino acid supplementation during fermentation. YL065 produced 4.26 g/L and 28.44 mg/g DCW of β-carotene within 110 h (Fig. 7AC). CDA-14 produced 6.15 g/L and 39.28 mg/g DCW of β-carotene within 110 h (Fig. 7BC), representing 31% and 28% increases over YL065, respectively.

Fig. 7.

Fed-batch fermentation results for β-carotene production by strains YL065 and CDA-14. A: Strain YL065; B: Strain CDA-14; C: The β-carotene yield of YL065 and CDA-14; A: β-carotene cultures displayed a deepred-orange color after 110 h cultivation. Error bar represents standard deviation of triplicate experiments

Discussion

Current research on genome-scale mutagenesis technologies in Y. lipolytica remains limited, with existing studies primarily exploiting the yeast’s robust non-homologous end joining (NHEJ) repair mechanism to construct genome-wide mutant libraries through NHEJ-mediated integration [35]. Unlike other mutagenesis approaches, NHEJ-mediated insertional mutagenesis facilitates straightforward library generation while eliminating the need for heterologous transposons or CRISPR systems, and enables rapid mutation tracking. Leveraging this NHEJ-driven insertional mutagenesis platform, researchers have efficiently identified novel targets for enhancing β-carotene biosynthesis and acetic acid tolerance, demonstrating its utility for functional genomics in this industrially relevant yeast. However, a critical technological paradox emerges when applying this strategy to pre-optimized industrial strains. The prevailing paradigm for constructing Y. lipolytica cell factories requires suppression of NHEJ activity (typically through KU70 deletion) to enhance homologous recombination efficiency for precision genome editing. This essential optimization directly conflicts with the fundamental requirement of NHEJ-mediated mutagenesis approaches, creating incompatibility between random insertion library strategies and continuous strain improvement in engineered production platforms.

To overcome this technological barrier, we successfully developed a Helicase-Assisted mutation (Helicase-CDA) system, a genome-wide mutagenesis platform based on a helicase-deaminase complex [12]. By functionally fusing the MCM5 helicase of Y. lipolytica with cytidine deaminase (CDA), this innovative system enables the establishment of a controllable, continuous mutagenesis mechanism in engineered strains through a single transformation of a plasmid carrying the MCM5-CDA fusion module. MCM5 represents an optimal fusion subunit within the MCM2-7 helicase complex, attributed to its dual structural roles in replication fork anchoring and gating regulation [27]. The molecular basis for its precise replication fork localization involves two domain-specific mechanisms: (1) electrostatic docking of its C-terminal domain (C5) onto Orc3 of the ORC complex for origin positioning [36] and (2) interlocking of its N-terminal zinc finger domain with adjacent hexamers to stabilize the double hexamer (DH) interface encircling dsDNA. Helicase activation is initiated by ORC-Cdc6 ATP hydrolysis, which triggers closure of the MCM2-MCM5 gating gap to form a topologically closed ring entrapping dsDNA [37]. This structural transition releases Cdt1 and converts the inactive DH into a processive helicase stably anchored at the fork. The spatial precision afforded by MCM5-mediated localization (via ORC interaction and DH stabilization) ensures targeted delivery of fused enzymatic payloads [37]. This mechanistic rationale is functionally validated in S. cerevisiae, where MCM5-AID fusion increased editing efficiency by (2.1 ± 0.4) × 10³-fold over genomic background rates, demonstrating its capacity for replication-coupled enzymatic action [34]. Following a 7-day continuous subculturing regimen (once-daily passages), the system achieves C-to-T site-specific base substitutions across the entire genome via helicase-mediated localized DNA unwinding and deaminase targeting.

When applied the Helicase-CDA system to an industrial β-carotene-producing Y. lipolytica strain, a 7-day continuous culture and mutagenesis process successfully yielded the superior mutant CDA-14 with 25% increased β-carotene production. This system, which enables continuous mutagenesis in engineered strains without the need to restore high NHEJ efficiency, provides a novel platform for rapid directed evolution of metabolically engineered strains.

Whole-genome resequencing of the CDA-14 strain which screened from library that subjected to 7-day continuous mutagenesis revealed only 96 accumulated SNPs, indicating suboptimal mutagenesis efficiency. We hypothesize this limitation arises from two molecular mechanisms: (1) The absence of codon optimization for the CDA gene in Y. lipolytica likely reduced translational efficiency, resulting in insufficient MCM5-CDA fusion protein expression [38]. (2) Lack of uracil glycosylase inhibitor (UGI) co-expression may permit repair of CDA-induced C-to-U mutations through base excision repair (BER) pathways [39]. Strategic improvements including codon usage adaptation and UGI overexpression could therefore enhance mutation rates. Crucially, mutagenesis intensity requires systematic optimization to balance genetic burden and phenotypic screening efficiency – a principle critical for evolutionary engineering success. The optimal mutation efficiency can be achieved through the molecular optimization of Helicase-CDA system components, including modulating the expression of the Helicase-CDA fusion protein to regulate editing activity, or by precisely controlling the duration of exposure to active mutagenesis phases.

The G-to-A mutation at position 1637 of the YALI1_B16239g gene (encoding MGA2) results in a codon change from CGA (arginine) to CAA (glutamine). This mutation not only alters the amino acid but also replaces a high-frequency codon (41.4‰) with a low-frequency codon (0.7‰). In strains with G-to-A mutations or deletion of YALI1_B16239g, ERG1 transcription levels decreased, while squalene accumulation and β-carotene yield increased. We hypothesized that the introduction of this rare codon reduces the expression of the YALI1_B16239g-encoded protein, leading to lower ERG1 activity. This decrease in ERG1 expression reduces the metabolic flux toward sterols and potentially redirects it toward β-carotene. The increase in squalene accumulation due to ERG1 downregulation is partially offset by the diversion of metabolic flux toward β-carotene. As a result, the overall increase in squalene production in this study is only 65% higher than in the wild-type, which is much lower than the 5.5-fold increase observed upon MGA2 deletion in S. cerevisiae [34]. Furthermore, squalene accumulation varied significantly between different fermentation time points, with lower accumulation at 72 h compared to 48 h. This suggests that while the MGA2 gene regulates squalene accumulation, other factors may also influence its levels during fermentation.

Although the YL065-ΔMGA2 strain exhibits a 15% reduction in total β-carotene production due to a significant biomass decrease (49% lower OD₆₀₀), it paradoxically achieves a 27% higher β-carotene yield per unit biomass. This yield increase stems from enhanced carbon flux efficiency toward β-carotene biosynthesis. Knocking out MGA2 suppresses ERG1 expression and sterol biosynthesis. While sterol depletion likely limits growth and biomass [40], it simultaneously redirects metabolic flux away from the competing sterol pathway and into β-carotene production, boosting its accumulation per cell mass.

Notably, YALI1_B16239g represents a novel target, and there are no reports yet regarding its influence on β-carotene production in Y. lipolytica or any other species. Future research can further explore this hypothesis.

Conclusion

In this study, we developed the Helicase-CDA base editing system to enhance C: G-to-T: A mutation efficiency in Y. lipolytica. Functional validation demonstrated that the Helicase-CDA system significantly increased the occurrence of C: G-to-T: A substitutions compared to the parental strain. Through targeted mutagenesis, screening and whole genome sequencing, we identified a novel genetic target, YALI1_B16239g (a putative membrane protein involved in OLE1 transcriptional regulation), where a specific C-to-T mutation correlated with a modest increase in β-carotene production. This discovery not only validates the utility of the Helicase-CDA system for uncovering functional genetic elements but also reveals a previously uncharacterized regulatory node in carotenoid biosynthesis.

CDA-14 exhibited 25% and 31% increases in β-carotene production under shake-flask and 5-L bioreactor conditions, respectively, demonstrating the Helicase-CDA system’s robust efficacy as a high-performance tool for microbial strain engineering. The system’s unique capability to enable continuous mutagenesis without restoring NHEJ pathway activity provides a critical advantage for iterative strain optimization in genetically engineered industrial microbes. The Helicase-CDA system establishes a versatile platform for advancing microbial metabolic engineering and accelerating the development of high-performance cell factories.

Author contributions

Zhenxia Li and Bo Liu: Data curation, Formal analysis, Writing - original draft. Rongtao Lv: Data curation, Methodology. Sun Zhe: Conceptualization, Formal analysis，Writing - review & editing. QingYan Li: Supervision, Writing - review & editing. XueLi Zhang: Supervision, Funding acquisition.

Funding

This research was supported by grants from the National Science Fund for Distinguished Young Scholar (No.32225031), the Basic Science Center Project of the National Natural Science Foundation of China (No. 32488301).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.