An advanced genome engineering platform for Komagataella phaffii enabling versatile in vivo DNA assembly and multiplex integration

Abstract

Background

The transition toward a sustainable bioeconomy necessitates the development of robust microbial platforms capable of C1 chemical valorization, particularly the conversion of methanol into high-value chemicals. Komagataella phaffii is a premier methylotrophic host for such applications due to its natural methanol utilization pathways and high-density fermentation capabilities. However, its potential as a C1-based synthetic biology chassis has been hampered by inefficient homologous recombination and the technical complexity of multigene pathway integration. To accelerate the design-build-test-learn (DBTL) cycle for methanol-based biomanufacturing, a more sophisticated and high-throughput genome engineering toolkit is required.

Results

We report the development of an HR-enhanced K. phaffii chassis (dKR) achieved by synergistically deleting Ku70 and overexpressing RAD52 at the chromosomal level. This engineering markedly improved the robustness and reproducibility of CRISPR/Cas9-mediated genome editing, increasing integration efficiency from 15.3% in the wild-type to 68.1% in the dKR chassis when using short 50-bp homology arms. The platform demonstrated exceptional capacity for vector-free multi-fragment assembly, achieving an integration efficiency of 86.5% for nine linear DNA fragments. Furthermore, the implementation of a tRNA-ribozyme CRISPR system enabled simultaneous multi-locus integration at three distinct genomic sites with 44.6% efficiency. Notably, the chassis supported direct in vivo plasmid assembly of ten fragments (9 fragments + 1 backbone) with 44.8% efficiency, effectively bypassing E. coli-based cloning. Functional validation via a carotenoid biosynthetic pathway resulted in β-carotene production while maintaining normal cellular growth and metabolic productivity.

Conclusion

This study established a versatile genome-engineering platform that advances K. phaffii from a protein-expression host to a broadly applicable biological engineering chassis. By enabling streamlined cellular design and reproducible multigene integration, the platform supports applications spanning synthetic biology, metabolic engineering, and bioproduction. More broadly, the approach illustrates how rational manipulation of DNA repair pathways can enhance genome engineering performance in non-conventional hosts, providing transferable design principles relevant to biological engineering across multiple domains.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13036-026-00731-z.

Introduction

The global imperative to transition toward a sustainable, carbon-neutral bioeconomy has accelerated the development of microbial platforms capable of converting non-food feedstocks into high-value chemicals and biomaterials [1, 2]. Conventional biomanufacturing, which primarily relies on sugar-based feedstocks, faces challenges regarding long-term economic viability and competition with food supply chains. Consequently, C1 valorization, the biological conversion of single-carbon compounds like methanol and CO₂, has emerged as a transformative strategy for the next generation of biorefineries [3, 4]. To realize this potential, there is an urgent need for robust, programmable microbial chassis that can be rapidly engineered for complex metabolic functions and industrial-scale bioprocessing [5, 6].

Among the available microbial hosts, the methylotrophic yeast Komagataella phaffii (formerly Pichia pastoris) occupies a strategic position in industrial biotechnology. Traditionally recognized for its exceptional capacity for heterologous protein secretion and high-density fermentation, K. phaffii offers several physiological advantages over the model yeast Saccharomyces cerevisiae [7–9]. Specifically, it is Crabtree-negative, avoiding the wasteful production of ethanol at high substrate concentrations, thereby ensuring efficient carbon flux toward target metabolites [10]. Its natural ability to utilize methanol as a sole carbon and energy source, supported by specialized peroxisomal metabolism, makes it an ideal candidate for C1-based precision fermentation and the production of diverse specialty chemicals, including organic acids, carotenoids, and functional lipids [11, 12].

Despite its industrial promise, the application of K. phaffii in complex synthetic biology projects is often hindered by its suboptimal genome-engineering efficiency. While CRISPR/Cas9 systems have been implemented, the predictability and throughput of genetic manipulation in K. phaffii remain significantly lower than in S. cerevisiae [13]. This bottleneck primarily originates from the intrinsic preference for double-strand break repair via non-homologous end joining (NHEJ) over homologous recombination (HR) [14, 15]. NHEJ, mediated by the Ku70-Ku80 heterodimer, directly ligates DNA ends without a homologous template and frequently introduces insertions or deletions, compromising targeting precision [16]. In contrast, HR enables accurate, site-specific integration using donor DNA templates. The dominance of NHEJ in K. phaffii limits precise gene insertion, multi-locus editing, and modular pathway assembly [15, 17]. To overcome these limitations, recent strategies have focused on shifting the DNA repair balance from NHEJ toward HR [18–22]. Specifically, combining Ku70 deletion with RAD52 and RAD59 overexpression has been demonstrated to substantially increase positive integration rates [22]. This shift establishes K. phaffii as a versatile engineering chassis with an HR-dominant repair machinery, supporting streamlined, biofoundry-compatible workflows [23, 24].

Building on this repair-engineering concept, we aimed to expand the HR-enhanced K. phaffii chassis into a practical, self-sufficient genome engineering platform. By integrating this chassis with a tRNA–ribozyme CRISPR/Cas9 system and short-homology-based donor designs, we established a streamlined strategy that substantially reduces the cloning burden of conventional yeast engineering. Specifically, this platform demonstrates three core capabilities: (i) efficient multi-fragment in vivo genomic assembly using short homology arms, (ii) high-fidelity multi-locus chromosomal integration, and (iii) direct in vivo plasmid reconstruction that completely bypasses E. coli-based cloning. Ultimately, these integrated technologies accelerate iterative strain construction, providing a scalable foundation for sustainable precision fermentation.

Methods

Strains and culture conditions

The methylotrophic yeast Komagataella phaffii (ATCC 28485) was used as the host strain for all genetic manipulations and cultivation experiments. The strain was obtained from the Korean Collection for Type Cultures (KCTC, South Korea) and stored at -80 ℃ in 15% (v/v) glycerol stocks until use. For pre-culture, cells were grown in 5 mL of YPD20 medium consisting of 1% (w/v) yeast extract, 2% (w/v) peptone, and 20 g/L glucose at 30 ℃ with shaking at 250 rpm. For phenotype assay of transformants, the cells were cultured in synthetic complete (SC) medium (1.7 g/L yeast nitrogen base without amino acids and ammonium sulfate, 5 g/L ammonium sulfate, and 0.79 g/L complete supplement mixture) with 20 g/L glycerol. The medium was buffered with 50 mM potassium hydrogen phthalate (KHP) at pH 6.5. Cultivation was conducted in a 125 mL flask containing 25 mL working volume at 30 ℃ and 250 rpm, with inoculation at an initial OD₆₀₀ of 1.0.

Plasmid construction

All yeast strains, plasmids, primers, sgRNA for CRISPR/Cas9 engineering, synthetic DNA cassettes encoding gRNAs, and codon-optimized genes used in this study are summarized in Tables S1, S2, S3, S4, S5, and S6, respectively. For plasmid construction, Escherichia coli DH5α was used as the cloning host. Liquid cultures were grown in 5 mL LB medium (25 g/L) at 37 ℃ with shaking at 200 rpm. When required, the following antibiotics were supplemented: kanamycin (50 µg/mL), clonNAT (120 µg/mL), or ampicillin (100 µg/mL). LB agar plates (25 g/L LB with 1.8–2.0% agar) containing the respective antibiotics were used for colony isolation and maintenance.

Plasmids used in this study (Table S2) were constructed using a combination of Golden Gate assembly, Gibson assembly, and topoisomerase-based cloning, depending on the required architecture of each vector. To construct the β-carotene expression plasmids (pCrtYB-I-E) and the lactate dehydrogenase expression vector (pRoLDH), GoldenPiCS modular cloning (Addgene Kit #1000000133) was employed [25]. The resulting expression vectors served as the backbone for subsequent plasmid designs. To construct the pRAD52 plasmid, the RoLDH coding sequence in pRoLDH was replaced with the RAD52 ORF using Gibson assembly, while retaining the same promoter and terminator modules to maintain a consistent expression architecture.

For chromosomal integration at defined loci, plasmids pRoLDH_1000bp HA and pRAD52_1000bp HA were constructed by amplifying about 1000 bp left and right homologous arms (HAs) from K. phaffii genomic DNA via PCR. These HAs were inserted into the pRoLDH and pRAD52 backbones, respectively, using Gibson assembly to generate donor plasmids suitable for homology-directed repair (HDR)-mediated integration.

Construction of the Ku70 deletion donor plasmid (pKu70_DN) was performed using the TOPcloner Blunt Core Kit (Enzynomics, Republic of Korea), with cloning into the pTOP TA V2 vector included in the kit. This vector is maintained in E. coli using kanamycin resistance, and the resulting construct contained only the homologous donor DNA sequences required for targeted disruption of the Ku70 locus.

CRISPR/Cas9 vectors were constructed based on the pCast backbone (Addgene #200169) [26]. Plasmid assembly was carried out using the AarI restriction enzyme (Enzynomics, Republic of Korea) and T4 DNA ligase (New England Biolabs, USA). For bacterial propagation, ampicillin was used as the selection marker, while clonNAT was employed for yeast selection.

Gene fragments were obtained according to their genetic origin. Heterologous sequences, including RoLDH, crtYB, crtI, and crtE, were synthesized as gBlock gene fragments by Integrated DNA Technologies (IDT, USA). In contrast, endogenous sequences such as RAD52 and chromosomal homologous arms were amplified from K. phaffii genomic DNA using Phusion High-Fidelity DNA Polymerase (New England Biolabs, USA), depending on the amplification requirements.

Yeast transformation

The transformation of Komagataella phaffii (ATCC 28485) was carried out via an optimized electroporation procedure as previously described with minor modifications [27–29]. Briefly, yeast cells were cultivated in 100 mL YPD20 medium in a 250 mL Erlenmeyer flask at 30 ℃ until mid-log phase (OD₆₀₀ ≈ 1.0–2.0; ~ 4 h). Cells were collected by centrifugation at 4000 rpm for 5 min, washed twice with sterile distilled water, and resuspended in 8 mL of transformation buffer containing 800 µL 1 M lithium acetate, 80 µL 1 M DTT, 480 µL 1 M sorbitol, 80 µL 1 M Tris-HCl (pH 7.5), and sterile distilled water to a final volume of 8 mL. The suspension was incubated at room temperature for 30 min, then centrifuged at 3000 rpm for 5 min. The cell pellet was resuspended in 1.5 mL ice-cold 1 M sorbitol and subsequently washed three times with ice-cold 1 M sorbitol (6000 rpm, 5 min each). Finally, cells were resuspended in 300 µL of ice-cold 1 M sorbitol to prepare electrocompetent cells at a final density of approximately 1 × 10¹⁰ cells/mL.

For transformation, 80 µL of competent cells were combined with the appropriate DNA constructs and incubated on ice for 5 min prior to electroporation. Electroporation was performed using a 2 mm gap cuvette at 1.5 kV, 25 µF, and 186 Ω with a Bio-Rad Gene Pulser system. The CRISPR/Cas9 plasmid was supplied at a fixed amount of 1.5 µg, while donor DNA quantities were adjusted based on assembly complexity: 1.5 µg for single-fragment integrations, 1.0 µg for assemblies consisting of two or more fragments, and 0.5 µg per fragment for highly multiplexed assemblies exceeding nine fragments. Immediately after pulsing, cells were recovered in 1 mL YPD20 medium and incubated at 30 ℃ for 2 h with shaking at 250 rpm. Transformants were selected by plating 100 µL of the recovered culture onto YPD20 agar plates containing 200 µg/mL clonNAT.

Molecular validation of yeast transformants

Putative transformants were subjected to colony PCR to verify correct genomic integration prior to downstream analyses. For rapid cell lysis, a small portion of each colony was resuspended in 20 µL of 0.2 M NaOH in a 1.5 mL microcentrifuge tube. The suspension was briefly heated using a microwave oven to disrupt the cells, followed by a short centrifugation (~ 30 s) to sediment cellular debris. The clarified supernatant was directly used as the template for PCR amplification. PCR was carried out using DreamTaq DNA polymerase (Thermo Fisher Scientific, USA) and primer pairs designed to span the homologous recombination junctions at the targeted genomic loci. The resulting amplicons were resolved by agarose gel electrophoresis using 0.8% (w/v) agarose gels prepared in 0.5× TAE buffer, and band sizes were compared with expected products to confirm correct integration.

Integration or in vivo assembly efficiency was evaluated as the fraction of colonies carrying the properly integrated construct. For RoLDH integration, transformants could not be visually distinguished based on colony phenotype. Therefore, when fewer than 20 colonies were obtained, all colonies were subjected to colony PCR to confirm correct integration. When more than 20 colonies were obtained, up to 20 randomly selected colonies were analyzed by colony PCR. For the integration of the β-carotene pathway, efficiency was evaluated by counting red-pigmented colonies among all transformants, as β-carotene-producing colonies were visually distinguished by their red/orange coloration. In all cases, integration efficiency was calculated as the percentage of positive colonies among the total number of tested colonies. All efficiency measurements were performed in biological triplicate.

Analytical methods

Growth was analyzed based on the optical density at 600 nm (OD₆₀₀) of cultures using a UV/Vis spectrophotometer (Nabi, MicroDigital Co., Ltd., Republic of Korea). Metabolites such as lactic acid and glycerol were analyzed using an Agilent 1260 Infinity series HPLC instrument (Agilent Technologies, USA) equipped with a refractive index detector (RID). Chromatographic separation was performed using an Aminex HPX-87 H column (Bio-Rad, USA) maintained at 50 ℃. The mobile phase consisted of 5.0 mM H₂SO₄ delivered at a flow rate of 0.5 mL/min.

For β-carotene analysis, 1 mL of culture broth was collected and normalized to an optical density at 600 nm (OD₆₀₀) of 10. Cells were harvested by centrifugation at 13,000 rpm for 5 min, washed once with sterile distilled water, and pelleted again under the same conditions. The pellet was resuspended in 600 µL of methanol: acetone (1:1, v/v). Cell disruption was performed using a FastPrep^®-24 Classic bead homogenizer (MP Biomedicals, USA) at 6 m/s for 40 s, with 9 cycles using Lysing Matrix C (1-mm silica beads, MP Biomedicals, USA). Samples were then centrifuged at 13,000 × g for 5 min, and the supernatant was collected for analysis. β-Carotene was quantified using an Agilent 1260 Infinity series HPLC instrument (Agilent Technologies, USA) equipped with a UV detector. Separation was performed using a Poroshell 120 EC-C18 column (4 μm, 4.6 × 150 mm; Agilent Technologies, USA) maintained at 40 ℃. The mobile phase consisted of 95% methanol and 5% acetonitrile (v/v) delivered at a flow rate of 1 mL/min. Prior to injection, samples were passed through a 0.20 μm PTFE syringe filter. The injection volume was 20 µL, and β-carotene was detected at 450 nm. Quantification was performed using standard calibration curves prepared with a β-carotene standard (Cat. No. C9750, Sigma-Aldrich, USA).

Results and discussion

Enhancement of CRISPR/Cas9 efficiency through ΔKu70/RAD52 overexpression engineering

Efficient and precise genome engineering is essential for the construction of advanced metabolic pathways in K. phaffii. However, unlike S. cerevisiae, K. phaffii preferentially repairs DNA double-strand breaks via NHEJ, resulting in relatively low HR efficiency [30]. To overcome this limitation, we engineered strains by deleting Ku70, a central NHEJ component, and overexpressing RAD52, a mediator of HR pathway activity [21].

To support efficient CRISPR/Cas9-mediated genome engineering, we employed a tRNA-gRNA architecture consisting of a tRNA_Gly sequence, a 20 bp gRNA targeting region, the standard gRNA scaffold, and a hepatitis delta virus (HDV) ribozyme (Fig. 1A) [31, 32]. The precursor RNA undergoes sequential processing by endogenous RNase P and RNase Z, which cleave the tRNA sequence at the 5′ and 3′ ends, respectively, and the HDV ribozyme mediates self-cleavage at the 3′ terminus of the gRNA, generating a precisely trimmed mature gRNA [31, 33]. This processing strategy ensures accurate gRNA maturation independent of promoter-encoded processing signals and avoids the incorporation of extraneous nucleotides that can impair Cas9-gRNA complex formation [31, 33].

Fig. 1.

Enhancement of CRISPR/Cas9 genome editing efficiency in K. phaffii via Ku70 deletion and RAD52 overexpression. (A) tRNA-based gRNA expression design enabling precise gRNA maturation through RNase P/Z processing and HDV ribozyme self-cleavage. (B) Schematic of donor constructs with 1000 bp, 500 bp, and 50 bp homology arms with pGAP(P)-RoLDH-RPS25Att(T) cassette. (C) Comparison of HDR integration efficiencies across WT, dK, ROE, and dKR strains. Data are presented as mean ± standard deviation (SD) from three independent biological replicates. For each replicate, integration efficiency was calculated as the percentage of PCR-positive colonies among the colonies analyzed. Significant differences compared with the WT control with the same homology arm length, as determined by Student’s t-test, are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001). (D) Lactic acid production performance of L1 and dKRL1 strains showing both productivity (g/L/h) and yield (g/g glycerol). Data are presented as mean ± SD from two independent biological replicates

Using this tRNA-based CRISPR/Cas9 platform, three engineered strains were constructed in the K. phaffii background: (i) a Ku70-deleted strain (dK), (ii) a RAD52 overexpression strain (ROE), and (iii) a strain combining Ku70 deletion and RAD52 overexpression (dKR). These strains, together with the wild-type (WT) control, were evaluated to determine whether simultaneous suppression of NHEJ and enhancement of HR could synergistically improve CRISPR/Cas9-mediated chromosomal integration. To quantitatively assess HR efficiency, chromosomal integration of a Rhizopus oryzae lactate dehydrogenase (RoLDH) cassette was evaluated using donor DNA fragments carrying homology arms (HAs) of 1,000 bp, 500 bp, or 50 bp (Fig. 1B and C).

In the WT strain, integration efficiencies were uniformly low and insensitive primarily to HA length, reaching only 15.3%, 13.1%, and 13.7% with 50 bp, 500 bp, and 1,000 bp HAs, respectively, confirming that NHEJ predominates in the native K. phaffii background even when long homology regions are provided. Deletion of Ku70 dramatically enhanced integration efficiency across all HA lengths, yielding efficiencies of 75.0% (50 bp), 69.3% (500 bp), and 60.5% (1,000 bp). In contrast, RAD52 overexpression alone resulted in only modest and inconsistent improvements relative to WT (8.3%, 20.0%, and 11.7% for 50 bp, 500 bp, and 1,000 bp HAs, respectively), indicating that reinforcement of the HR machinery is insufficient to overcome the intrinsic dominance of NHEJ when competing repair pathways remain active. The combined dKR strain consistently exhibited high integration efficiencies across all HA lengths, reaching 68.1% with 50 bp HAs and remaining above 50% with longer homology arms (58.8% for 500 bp and 52.4% for 1,000 bp).

Although dKR did not exhibit the highest integration efficiency under the tested conditions, we selected this strain as a minimal dual-repair-engineered chassis rather than the strain with the highest overall efficiency. Ku70 deletion suppresses NHEJ and improves targeted integration in K. phaffii [20, 34], but permanent impairment of Ku70-dependent NHEJ may affect DNA damage responses and cellular robustness, including potential effects on chromosome stability, growth, or production [30, 34]. Mechanistically, RAD52 acts as a critical recombination mediator that binds to replication protein A-coated single-stranded DNA and facilitates the loading of RAD51, thereby catalyzing strand invasion and homologous pairing [35, 36]. In an NHEJ-deficient (ΔKu70) background, unresolved or slowly processed double-strand breaks pose a severe threat to genome integrity, often inducing lethal DNA damage or gross chromosomal rearrangements [30, 37]. Overexpressing RAD52 ensures that the kinetic flux of double-strand break repair is actively and safely channeled into the high-fidelity HR pathway [22, 30]. While the precise impacts of RAD52 overexpression on genomic stability and recombination kinetics remain to be experimentally validated, this dual-engineered strategy was adopted as a proactive safeguard against intensive engineering stress. Consequently, our selection of the dKR strain was guided by a holistic consideration of the balance of repair machinery, rather than by absolute integration efficiency alone. This approach is consistent with previous studies showing that combined suppression of NHEJ and overexpression of HR factors can improve precise genome editing [22, 38, 39].

Additionally, our results demonstrate that 50-bp HAs are fully sufficient to support efficient CRISPR/Cas9-mediated chromosomal integration, eliminating the strict requirement for long homology regions in this optimized system. This feature offers a major practical advantage, as short HAs can be readily introduced via PCR primers, thereby simplifying fragment preparation and reducing the cloning burden. By leveraging this short-homology-based strategy, we established various downstream synthetic biology workflows, including biosynthetic pathway assembly, multi-fragment or multiplex genomic integration, and direct in vivo plasmid reconstruction. Thus, this platform extends beyond repair-pathway engineering itself by streamlining the construction of complex genetic designs and facilitating future iterative strain engineering.

To ensure the practical utility of this platform for such downstream applications, we next verified that the genetic modifications did not compromise the physiological integrity. Because disruption of Ku70 can potentially impair cellular robustness, we evaluated growth and metabolic productivity by comparing RoLDH-expressing strains constructed in the WT and dKR backgrounds, designated L1 and dKRL1, respectively (Table S1). When cultivated on glycerol, the WT-derived L1 strain exhibited a productivity of 0.085 g/L/h and a yield of 0.445 g/g, whereas dKRL1 achieved 0.105 g/L/h and 0.555 g/g, respectively (Fig. 1D), confirming that the engineered background maintains robust cellular performance.

Taken together, these results indicate that NHEJ suppression enhanced homologous recombination in K. phaffii. At the same time, additional reinforcement of the HR machinery through RAD52 overexpression may contribute to more robust and reproducible CRISPR/Cas9-mediated genome editing. This combined strategy enables efficient short-homology-based integration with reduced variability. Based on these findings, we established the final platform strain KpKR, incorporating Ku70 deletion and chromosomal RAD52 overexpression, for subsequent multi-locus pathway assembly and strain development.

Development of a vector-free multi-fragment in vivo assembly system

With the enhanced CRISPR/Cas9-driven homologous recombination capacity of the engineered chassis (KpKR), we next sought to develop a vector-free in vivo DNA assembly and genomic integration system. As an initial proof-of-concept, we first validated the feasibility and fidelity of in vivo multi-fragment assembly using a fluorescent reporter system. A pGAP-mCherry-RPS25Att expression cassette was assembled directly in vivo from multiple linear DNA fragments comprising the promoter, coding sequence, terminator, and short homology arms targeting the int1 locus (Fig. S1A). Correct chromosomal integration was confirmed by diagnostic PCR (Fig. S1B). Functional expression of the assembled cassette was further verified by fluorescence quantification normalized to OD₆₀₀, which showed robust mCherry expression in representative assembled strains compared with the wild-type control (Fig. S1C). Flow cytometry analysis revealed a distinct mCherry-positive population, confirming homogeneous reporter expression at the single-cell level (Fig. S1D). These results demonstrate that the HR-enhanced KpKR chassis supports accurate and functional in vivo multi-fragment assembly prior to application to more complex biosynthetic pathways.

Following this initial validation, we extended the in vivo multi-fragment assembly strategy to a more complex biosynthetic pathway by selecting the β-carotene biosynthetic genes crtYB, crtI, and crtE from Xanthophyllomyces dendrorhous (Fig. 2A). Each gene was codon-optimized for expression in K. phaffii. Successful pathway assembly was readily visualized by the appearance of orange-red pigmentation, providing a convenient phenotypic readout for efficient screening [40]. Three configurations were tested to assess the flexibility and efficiency of the KpKR chassis: (i) a single β-carotene cassette (crtYB-crtI-crtE, Kp1F), (ii) three promoter-gene-terminator cassettes (Kp3F), and (iii) nine individual promoter, gene, and terminator fragments (Kp9F) (Fig. 2B-D). For all genome integration experiments in this section, the terminal homology arms flanking the donor construct were fixed at 50 bp. For multi-fragment assembly, adjacent DNA fragments were designed with about 40 bp of overlap to facilitate in vivo recombination before or during chromosomal integration.

Fig. 2.

Vector-free multiplex genome integration of the β-carotene pathway in the HR-enhanced K. phaffii chassis. (A) Schematic overview of β-carotene biosynthesis from the mevalonate (MVA) pathway. (B-D) Donor DNA architectures for Kp1F (B), Kp3F (C), and Kp9F (D) configurations and corresponding colony morphology following CRISPR-mediated insertion. All designs utilized 50-bp homology arms with > 40-bp overlaps between adjacent fragments. P1, P2, and P3 represent the promoters of GAP, MDH3, and POR, respectively, whereas T1, T2, and T3 represent the terminators of RPS25A, TDH3, and RPS2, respectively. Highly pigmented colony counts highlight the capacity of the chassis to support multi-fragment in vivo DNA assembly. (E) Colony PCR confirmation of correct integration at the int1 locus for 1 F (Kp1F), 3 F (Kp3F), and 9 F (Kp9F) assemblies, showing the expected integration-specific band (8.7 kb). Non-integrated transformants yielded a shorter negative-control amplicon (0.4 kb), allowing a clear distinction between successful and unsuccessful recombination events. (F) Integration efficiencies of one-fragment (Kp1F), three-fragment (Kp3F), and nine-fragment (Kp9F) donor designs in the ΔKu70/RAD52 OE strain (KpKR). Data are presented as mean ± SD from three independent biological replicates. For each replicate, efficiency was calculated based on a total-plate phenotypic screening of all emerging colonies based on pigmentation

Colony PCR analysis verified the correct genomic integration of crtYB, crtI, and crtE at the int1 locus across all configurations (Fig. 2E). Amplified fragments corresponding to the expected band sizes confirmed accurate assembly and locus-specific insertion. Remarkably, integration efficiencies reached 27.6%, 92.1%, and 86.5% for the 1-, 3-, and 9-fragment configurations, respectively (Fig. 2F). The lower efficiency observed in the 1 F design is likely attributed to the extended DNA length (8–9 kb) of the single-fragment donor, which reduces transformation efficiency and increases susceptibility to shearing and incomplete homologous pairing during repair [41, 42]. In contrast, dividing the cassette into smaller modular fragments (3 F and 9 F) facilitated more efficient recombination, as shorter DNA fragments are more readily imported and processed by the homologous recombination machinery. These results suggest that fragment size is a primary determinant of recombination efficiency under the tested conditions.

Functional expression was confirmed by cultivating a representative β-carotene strain (Kp1F). Growth and glycerol-consumption profiles remained comparable to the control strain (Fig. S2A and B). The engineered strain achieved a β-carotene titer of 123.9 mg/L in glycerol medium (Fig. S2C) and exhibited stronger pigmentation than the parental strain (Fig. S2D). To verify the robustness of this approach, β-carotene production was assessed across different donor configurations (Kp1F, Kp3F, and Kp9F) using HPLC (Fig. S3). Although pathway functionality was successfully reconstructed in all cases, colony-level variations in titers were observed. This is unlikely to stem from assembly errors or overlapping-sequence issues, because correct integration and junction fidelity were verified by colony PCR and sequence analysis of representative constructs. Instead, these differences are more likely attributable to potential CRISPR/Cas9-associated off-target effects or subtle clonal variations that affect host fitness and metabolic flux [43, 44]. To systematically address these issues, further genome-wide analyses will be useful in future studies for a more comprehensive evaluation of construct stability, genome integrity, and platform robustness.

Collectively, these findings establish the ΔKu70/RAD52 OE chassis (KpKR) in K. phaffii as an efficient platform for direct in vivo multi-fragment genomic assembly. By enabling pathway integration without plasmid cloning, this system greatly simplifies strain construction workflows and provides a scalable foundation for rapid pathway prototyping and genome-scale metabolic engineering in K. phaffii [17, 21, 45].

Multiplex genome editing enabled by the tRNA–Ribozyme CRISPR system in the ΔKu70/RAD52 overexpression platform

To further exploit the HR-enhanced strain background for simultaneous multi-locus editing, we implemented a modular tRNA-guided CRISPR cassette designed for multiplex gRNA expression (Fig. 3A). In this system, repeated tRNA_gly-gRNA-HDV ribozyme units are transcribed as a single polycistronic RNA, which is subsequently segmented into individual mature gRNAs through endogenous tRNA processing pathways [31]. This architecture enables the coordinated generation of multiple gRNAs without requiring multiple promoters [46], ensuring balanced expression levels and synchronous Cas9 loading across targets [33]. By integrating three gRNAs targeting distinct genomic loci, we evaluated the efficiency and scalability of multiplex editing in K. phaffii and determined whether the enhanced HR chassis can support parallel chromosomal integration events.

Fig. 3.

Multiplex genome editing using a tRNA-gRNA-ribozyme expression system in K. phaffii. (A) Schematic of the polycistronic tRNA-gRNA-HDV ribozyme array, which enables endogenous RNase P/Z processing and ribozyme self-cleavage to release individual mature gRNAs for multiplex CRISPR targeting. (B) Single-site integration of a crtYB-crtI-crtE cassette at int21 and corresponding colony phenotype. (C) Dual-site integration of crtYB and crtI at int21, and crtE at int1. (D) Triple-site integration of crtYB, crtI, and crtE across int21, int1, and int6, demonstrating successful three-locus editing in a single transformation. (E) Colony PCR confirmation of single-, dual-, and triple-locus integrations (1 S, 2 S, 3 S). Negative-control (non-integrated) bands appeared at 0.3 kb for int21, 0.4 kb for int1, and 0.6 kb for int6. Correctly edited strains produced locus-specific integration bands of 8.6 kb (1 S-int21), 6.3 kb (2 S-int21), and 2.7 kb (2 S-int1) for dual-site editing, and 3.3 kb (3 S-int21), 3.4 kb (3 S-int1), and 2.9 kb (3 S-int6) for triple-site editing. These band patterns confirm precise CRISPR-mediated insertion at all designated loci. (F) Editing efficiencies for single-, dual-, and triple-locus integration events in the ΔKu70/RAD52 OE chassis (KpKR). Data are presented as mean ± SD from three independent biological replicates. For each replicate, efficiency was calculated based on a total-plate phenotypic screening of all emerging colonies based on pigmentation

Using this system, three editing configurations were tested to evaluate the performance of multiplex integration: (i) single-site insertion of the crtYB-crtI-crtE cassette at int21 (Kp1S), (ii) dual-site editing inserting crtYB and crtI at int21 and crtE at int1 (Kp2S), and (iii) triple-site editing placing crtYB, crtI, and crtE at int21, int1, and int6, respectively (Kp3S) (Fig. 3B-D). Based on the HA-length comparison, 50-bp homology arms were used as the final donor design for all multiplex genome integration experiments. Accordingly, all integration donors were flanked by 50-bp locus-specific homology arms and delivered as linear DNA fragments. Successful insertion events were confirmed by colony PCR and β-carotene pigmentation. Colony PCR analysis verified correct integration at the intended loci for all editing modes (Fig. 3E). Editing efficiencies were 55.4% for single-site integration, 48.0% for dual-site editing, and 44.6% for triple-site editing (Fig. 3F). Although a gradual decline in efficiency was observed with increasing editing complexity, the decrease remained modest, indicating that the tRNA–ribozyme CRISPR system and the ΔKu70/RAD52 OE background effectively supported parallel processing of multiple gRNAs and corresponding recombination events. β-carotene production in the three transformants was quantitatively confirmed by HPLC analysis (Fig. S3). Furthermore, for the most complex triple-site editing case (Kp3S), Sanger sequencing of all three targeted loci verified accurate, precise chromosomal integration without detectable rearrangements (Fig. S5).

Overall, the ability to simultaneously modify up to three genomic loci using short (50 bp) homology arms and linear donor DNA demonstrates the robustness of the ΔKu70/RAD52 OE chassis for multiplex genome engineering in K. phaffii. This capability facilitates efficient assembly of multigene constructs or combinatorial variants within a single transformation step. Collectively, these results demonstrate that the tRNA–ribozyme CRISPR system, when combined with the HR-enhanced chassis, provides a practical and scalable framework for multiplex genome editing in K. phaffii [21, 45, 47].

In vivo plasmid assembly in the HR-enhanced K. phaffii chassis

To further extend the utility of the HR-enhanced chassis, we evaluated whether K. phaffii could directly assemble plasmid DNA in vivo, thereby eliminating the need for the conventional E. coli cloning workflow (Fig. 4A). Instead of constructing vectors in E. coli, linear fragments and a plasmid backbone were co-transformed into the KpKR strain to test whether the yeast could autonomously reconstitute a functional circular plasmid through homologous recombination.

Fig. 4.

In vivo plasmid assembly in K. phaffii eliminates the need for E. coli cloning. (A) Comparison of traditional E. coli-based plasmid construction vs. direct in vivo assembly in K. phaffii. (B) Donor DNA designs for 3F1B and 9F1B assembly and representative colonies expressing β-carotene. All fragments featured 50-bp homology arms with > 40-bp overlaps, demonstrating efficient in vivo plasmid reconstruction via homologous recombination. (C) Gel electrophoresis confirms accurate in vivo assembly of the 9-fragment plasmid (Kp9F1B). Vector-verification PCR (V-PCR) yielded the expected 8.8 kb product. Individual fragment PCRs showed bands corresponding to their predicted sizes: P1 (GAP promoter; 0.5 kb), crtYB (2.0 kb), T1 (RPS25A terminator; 0.5 kb), P2 (MDH3 promoter; 1.0 kb), crtI (1.7 kb), T2 (TDH3 terminator; 0.2 kb), P3 (POR promoter; 0.7 kb), crtE (1.1 kb), T3 (RPS2 terminator; 0.5 kb), and BB (backbone; 2.9 kb), confirming precise reconstitution of all components. (D) Assembly efficiencies for the 3 fragments + 1 backbone construct (Kp3F1B) and 9 fragments + 1 backbone construct (Kp9F1B). Data are presented as mean ± SD from three independent biological replicates. For each replicate, efficiency was calculated based on a total-plate phenotypic screening of all emerging colonies based on pigmentation

Two assembly designs were examined: (i) Kp3F1B, consisting of three β-carotene pathway cassettes (crtYB, crtI, crtE) plus a backbone, and (ii) Kp9F1B, consisting of nine promoter–gene–terminator fragments plus a backbone (Fig. 4B). In these in vivo plasmid reconstruction experiments, plasmid assembly relied on overlaps between adjacent fragments. Therefore, each insert and backbone fragment was designed to contain about 40 bp of overlap with neighboring fragments, consistent with the multi-fragment assembly strategy. The backbone was derived from the GoldenPiCS kit (Addgene Kit #1000000133) [25] using the BB3rN_AD plasmid, which carries CEN/ARS for plasmid replication and a clonNAT resistance marker (NrsR) for selection. Successful in vivo plasmid reconstruction was indicated by β-carotene pigmentation and nourseothricin resistance. Colony PCR further validated precise reconstruction in both Kp3F1B and Kp9F1B strains, with amplicons matching expected sizes (Fig. 4C). Assembly efficiencies were 55.56% for Kp3F1B and 39.13% for Kp9F1B (Fig. 4D), demonstrating that the HR-optimized chassis can reliably construct multi-fragment plasmids without E. coli-based cloning. This yeast-based cloning capability effectively transforms K. phaffii from a downstream expression host into an autonomous genetic engineering platform capable of assembling complex plasmids directly within the cell [48]. By bypassing E. coli propagation, restriction enzyme cloning, and plasmid purification, this approach enables rapid, scarless, and marker-free vector construction, thereby accelerating the design-build-test cycle for pathway prototyping [48, 49].

Together, this study establishes a homologous recombination-enhanced K. phaffii chassis that substantially expands the genetic engineering capabilities of this industrially relevant yeast. By combining NHEJ suppression with HR reinforcement, the platform enables efficient multi-fragment genomic assembly, robust multiplex genome editing, and reliable reconstruction of functional biosynthetic pathways using short homology arms and linear DNA donors. These integrated capabilities streamline strain development workflows and improve the flexibility and reproducibility of pathway engineering in K. phaffii.

Despite these advantages, a current limitation of this platform is that its validation was restricted to a three-gene biosynthetic model, representing a relatively simple pathway architecture. Furthermore, while the platform demonstrates high efficiency and reproducibility in initial construct assembly and integration, the long-term genetic and production stability of the engineered pathways over extended cultivation or multiple passages remains to be fully evaluated. Future high-value applications, such as complex pathway construction, dynamic genetic circuits, or biocontainment systems, will inevitably require the orchestration of much larger and more complex genetic programs. Because eukaryotic yeasts require distinct regulatory elements for each gene rather than utilizing simple operon structures, scaling up to multi-gene pathways exponentially increases the genetic design burden. Therefore, future studies must systematically evaluate both the scalability of this vector-free platform with larger gene clusters and its long-term operational stability under prolonged cultivation conditions. Additionally, integrating this system into automated, biofoundry-compatible ‘design-build-test-learn’ workflows will be critical to achieving high-throughput, iterative strain optimization. Ultimately, this high-performance engineering platform provides a critical technical foundation for advanced biomanufacturing and C1 valorization strategies, enabling the rapid and scalable development of microbial cell factories for the sustainable production of diverse high-value chemicals and carbon-neutral biomaterials [24, 50].

Conclusions

We developed a homologous recombination-enhanced K. phaffii chassis by combining Ku70 deletion with chromosomal RAD52 overexpression, resulting in robust and reproducible CRISPR/Cas9-mediated genome engineering. This strategy enabled efficient short-homology-based integration while maintaining normal cellular growth and metabolic performance. Using this chassis, we established a vector-free in vivo DNA assembly framework that supports multi-fragment genomic integration, multiplex genome editing, and direct plasmid reconstruction from linear DNA. These capabilities substantially simplify strain construction and accelerate pathway prototyping by eliminating plasmid cloning and long homology requirements. Overall, this platform advances K. phaffii into a versatile biological engineering chassis and provides a scalable foundation for synthetic biology, metabolic engineering, and C1-based biomanufacturing requiring rapid and reliable pathway construction.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

crtE

Geranylgeranyl pyrophosphate synthase

crtI

Phytoene desaturase

crtYB

Phytoene synthase/lycopene cyclase

GGPP

Geranylgeranyl pyrophosphate

gRNA

Guide RNA

HA

Homology arm

HDV

Hepatitis delta virus

LHA/RHA

Left/right homology arm

MVA

Mevalonate

PScTDH3

Saccharomyces cerevisiae TDH3 promoter

RNase P/Z

tRNA-processing endoribonucleases

RoLDH

Rhizopus oryzae lactate dehydrogenase

TScCYC1

Saccharomyces cerevisiae CYC1 terminator

tRNAGly

Glycine transfer RNA

V-PCR

Vector-verification PCR

WT

Wild type

Author contributions

Y.C. contributed to the writing-original draft preparation, conceptualization, visualization, data curation, formal analysis, investigation, and methodology. A.K. contributed to the visualization, formal analysis, investigation, and methodology. N.K.K contributed to the writing of the review and editing, conceptualization, supervision, and funding acquisition.

Funding

This work was supported by the Bio&Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2025-02214232, RS-2026-25471273, RS-2025-23442968, and RS-2026-25523532), and the R&D Program of MOTIE/KEIT (20025698).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.