Introduction

The non-conventional yeast Yarrowia lipolytica is an attractive microbial cell factory for the recombinant production of various chemicals. Due to its unique physiological characteristics including high tricarboxylic acid cycle (TCA) flux and abundant acetyl-CoA supply, it has been used to produce many valuable chemicals such as organic acids, lipids, terpenoids, polyketides, and flavonoids (Fickers et al. 2020; Chen et al. 2021; Zhang et al. 2022; Yu et al. 2018; Lv et al. 2019).

Engineering cell factories often requires genetic markers to select the cells with desired genetic modifications. Successive genetic manipulations are often limited by the number of available selection markers. The CRISPR/Cas system has been developed for genomic engineering in Y. lipolytica; however, Y. lipolytica has a preference for non-homologous end joining (NHEJ) and manipulation of key NHEJ or homologous recombination proteins is necessary to improve the efficiency of targeted genome integration (Schwartz et al. 2016; Wu et al. 2023). In contrast, marker-based random integration through NHEJ can be easily achieved. We previously demonstrated that random integration can be used to construct a gene expression library for high producer selection (Cui et al. 2019; Liu et al. 2022). Therefore, in metabolic engineering, genetic markers are required for gene manipulation. Currently, the most commonly used selectable markers in Y. lipolytica can be divided into auxotrophic and dominant markers (Larroude et al. 2018). The most common auxotrophic markers mainly include LEU2, URA3, and TRP1 (Barth and Gaillardin 1996), while the most frequently used dominant markers include hphMX (hygromycin resistance) and NAT1 (nourseothricin resistance) (Cordero Otero and Gaillardin 1996; Kretzschmar et al. 2013). In addition, bleoR (zeocin resistance), guaB (mycophenolic acid resistance), and CRF1 (Cu2+ resistance) were reported as dominant markers in Y. lipolytica (Tsakraklides et al. 2018; Wagner et al. 2018; Wang et al. 2011). Moreover, Hamilton et al. (2020) showed that YlAMD1 (encoding acetamidase) could be selected on acetamide media and counter-selected on fluoroacetamide media, while Vandermies et al. (2017) reported that EYK1 (encoding erythrulose kinase) can be used as an efficient catabolic selectable marker for genome editing. Loss of the MET25 gene (encoding O-acetyl homoserine sulfhydrylase) induces lead sulfide (PbS) aggregation in media supplemented with divalent lead, such that deficient cells become brown/black, providing a positive/negative color-associated selection marker in Y. lipolytica (Edwards et al. 2020). Furthermore, Larroude et al. (2020) did a phenotype screen by disrupting GSY1, MFE1, or LIP2 lipid metabolism genes to test CRISPR/Cas9 editing success rate. Despite these useful markers, other dominant selection markers are being sought to expand the genetic engineering toolbox for this species.

Marker recycling is necessary for iterative genetic engineering. The Cre/loxP system from bacteriophage P1 is a frequently used site-specific recombination system (Sauer 1987). The loxP site is a 34 bp sequence composed of two 13-bp inverted repeats separated by an asymmetric 8-bp core sequence (Hoess et al. 1982). When two loxP sites in the same orientation flank a marker gene, the expression of Cre recombinase will result in excision of the marker gene between the two loxP sites. Fickers et al. (2003) established the Cre/loxP system for marker recycling in Y. lipolytica. Using this system to rescue markers is highly efficient; however, recombinant loxP copies left in the genome may interfere with subsequent rounds of recombination (Suzuki et al. 2005a, b). This problem can be avoided by using mutant lox sites. One class of mutated loxP site is applied in the left element/right element (LE/RE) mutant strategy. The Cre recombinase can mediate recombination between the LE mutant lox site and the RE mutant lox site, generating a double-mutant lox site that remains in the genome and a wild-type loxP site in the excised fragment. After marker recycling, the double-mutant lox site in the genome is poorly recognized by Cre recombinase, which prevents further genome rearrangement (Albert et al. 1995). In addition to the Cre/loxP system, Wagner et al. (2018) developed the piggyBac transposon system to achieve footprint-free marker recycling.

In this study, we developed dsdA from Escherichia coli, which encodes a D-serine deaminase as a dominant marker. In Y. lipolytica, it also confers the capability to use D-serine as the sole nitrogen source. In addition, we also determined the selection conditions of different dominant markers including kanMX, bleoR, and guaB (Table 1). The markers documented here expand the available genetic tools for Y. lipolytica and may facilitate the engineering of Y. lipolytica strains to produce various chemicals.

Table 1 Summary of the markers used in our study and the selection conditions for the PO1f strain
Full size table

Materials and methods

Strains and media

The yeast strains used were Y. lipolytica PO1f (ATCC MYA-2613, matA leu2-270 ura3-302 xpr2-322 axp-2), W29 (ATCC 20460, CLIB89 matA wild type), Y1444, and Y31248 (from the China Center of Industrial Culture Collection, CICC). A YPD medium (10 g/L yeast extract, 20 g/L tryptone, and 20 g/L glucose) was used to cultivate yeast cells under non-selective conditions. The medium that consisted of SC-URA is 1.7 g/L YNB (Yeast Nitrogen Base without amino acids and ammonium sulfate) (BBI Life Science Corporation, Shanghai, China) (see Supplemental Table S3 for the ingredients of YNB), 5 g/L (NH4)2SO4, and 0.77 g/L CSM-URA (Sunrise Science Products, Knoxville, USA) (see Supplemental Table S4 for the ingredients of CSM-URA). The DH5α (TransGen Biotech Co., LTD, Beijing, China) of E. coli was used for plasmid construction and propagation. E. coli was cultivated in a Luria–Bertani (LB) medium (5 g/L yeast extract, 10 g/L tryptone, and 10 g/L NaCl) supplemented with 0.1 to 0.2 mg/mL ampicillin at 37 °C (Liu et al. 2021). For solid media, 20 g/L of agar was added to the media.

For drug testing, the Y. lipolytica strains were revived in YPD medium at 30 °C with shaking at 200 rpm for 12 h. After overnight incubation, the cells were plated on YPD agar plates with different concentrations of antibiotics at 30 °C for 3 days. G418 was added at the concentrations of 1 mg/mL, 2 mg/mL, 4 mg/mL, and 7.2 mg/mL, respectively. Zeocin was added at the concentrations of 0.2 mg/mL, 0.4 mg/mL, 0.6 mg/mL, and 0.8 mg/mL, respectively. Mycophenolic acid was added at the concentrations of 0.1 mg/mL, 0.3 mg/mL, and 0.5 mg/mL, respectively. CuSO4·5H2O was added at the concentrations of 1.6 mg/mL, 3.2 mg/mL, 4.39 mg/mL, 4.55 mg/mL, 4.7 mg/mL, 5 mg/mL, and 6.3 mg/mL, respectively. The lowest concentration without any colony arising (minimal inhibition concentration) was taken as the required level of antibiotics to select transformants. The expression of dsdA was chosen in a D-serine selection medium containing 20 g/L glucose and 1.7 g/L YNB. Concentrations of 100 μg/mL, 125 μg/mL, 250 μg/mL, and 500 μg/mL D-serine were added, respectively. Uracil and leucine were added if necessary.

Plasmid construction and strain construction

The dsdA (from E. coli), kanMX (from bacteria), bleoR (from Streptoalloteichus hindustanus), and guaB (from E. coli) genes were codon-optimized (Supplemental Material) (GenScript Co., LTD, Nanjing, China). The CRF1 gene was PCR amplified from W29 genomic DNA. The marker genes were assembled under the control of the EXP1 promoter and the ICL1 terminator with the YLEP-ET2 plasmid (Cui et al. 2019) to generate the episomal plasmids YLEP-ET2-dsdA, YLEP-ET2-kanMX, YLEP-ET2-bleoR, YLEP-ET2-guaB, and YLEP-ET2-CRF1 (Supplemental Fig. S1). These plasmids were transformed into the E. coli strain DH5α for propagation. Finally, the plasmids were transformed into the yeast strain using a lithium acetate method (Chen et al. 1997).

Genetic marker testing

The plasmids with a resistance gene or linear DNA fragments (obtained by PCR) containing a resistance gene were transformed into the yeast, and cells were plated onto YPD agar supplemented with working concentrations of an antibiotic. After incubation for 3 days at 30 °C, the genomic DNA of transformants was extracted to amplify the target gene. The resulting PCR products were then sequenced to confirm the presence of a marker gene. The primers used in this study are listed in Supplemental Table S1. Furthermore, real-time quantitative reverse transcription PCR (RT-qPCR) was used to confirm the expression of dsdA, kanMX, guaB, and bleoR. For the dsdA selection, the nitrogen source was replaced with D-serine in the selection medium.

RT-qPCR

For RT-qPCR, the cells with a marker gene were revived in a YPD liquid medium supplemented with the working concentrations of an antibiotic (30 °C, 200 rpm for 8 to 12 h). The RNA was extracted and purified using a UNlQ-10 column TRIzol Total RNA Isolation Kit (Sangon Biotech, Shanghai, China). Next, cDNA was synthesized with the PrimeScript RT-PCR Kit (TaKaRa, Shiga, Japan). RT-qPCR was performed by using an Applied Biosystems Quant-Studio 3 Real-Time PCR system (Applied Biosystems, Waltham, MA, USA) with the SYBR Green Real-Time PCR Master Mix (Toyobo, Osaka, Japan). The ACT1 gene (encoding yeast actin, GenBank accession number YALI0_D08272g) was used as the reference gene to quantify the relative expression levels of marker genes.

Confirmation of the lox66-URA3-lox71 cassette

The EXP1p-URA3-ICL1t fragment was amplified from the YLEP-ET2-URA3 plasmid, and the lox66 and lox71 sequences (Suzuki et al. 2005a) were synthesized by Tsingke Biotechnology Co., Ltd (Qingdao, China). The two mutant lox sequences were inserted on the alternate sides of the EXP1p-URA3-ICL1t fragment, and the resulting fragment was named the lox66-URA3-lox71 cassette. The lox66-URA3-lox71 cassette was then transformed into PO1f (PO1f-URA3), and SC-URA plates were used to screen positive transformants. The cassette was further verified by PCR and sequencing. Next, the YLEP-Leu-Cre plasmid (Cui et al. 2017) was transformed into the positive transformants to produce Cre recombinase. After transformation, the colonies were picked and incubated on SC-URA and YPD agar plates, respectively, to confirm the deletion of the URA3 marker.

Nucleotide sequence accession numbers

The nucleotide sequences of dsdA, kanMX, bleoR, and guaB were deposited in the NCBI Sequence Read Archive under the BioProject PRJNA961308. The CRF1 GenBank accession number of Y. lipolytica is XM_500631.1. Their sequences are also provided at the end of the Supplemental Material.

Results

Identification of dsdA as a selectable marker in Y. lipolytica

It is reported that the dsdA gene from E. coli encodes a D-serine deaminase that converts D-serine to pyruvate and ammonia, which confers host resistance to D-serine and the ability to use D-serine as the nitrogen source (Cosloy and McFall 1973; Maas et al. 1995) (Fig. 1A). To determine whether dsdA can work in Y. lipolytica, we first tested whether Y. lipolytica can grow in defined medium containing D-serine as the sole nitrogen source. We found that PO1f could not grow in the YNB medium (without (NH4)2SO4) at different D-serine concentrations. Then, we transformed the centromere (CEN)/autonomously replicating sequence (ARS) plasmid (YLEP-ET2-dsdA) expressing dsdA as a selectable marker (under the control of the EXP1 promoter and the ICL1 terminator) into Y. lipolytica. As shown in Fig. 1B, when the concentration of D-serine was lower than 250 μg/mL, the transformants showed poor growth. However, transformants could grow normally when the D-serine concentration exceeded 250 μg/mL. The linear DNA fragment containing the dsdA expression cassette was then transformed into Y. lipolytica. The transformants with an integrated dsdA expression cassette in the genome could grow at 500 μg/mL D-serine in plate media (Fig. 1C). To confirm the dsdA expression in PO1f, 5 transformants were randomly selected from plasmid containing transformants or genomically integrated transformants. Genomic DNA was then isolated. The presence of the dsdA gene in Y. lipolytica was verified by PCR (Fig. 1D). Furthermore, we also extracted the RNA and analyzed the transcription of dsdA (Supplemental Table S2). The results confirmed that dsdA was expressed in the transformants. We also tested if dsdA can be used for other strains. As shown in Supplemental Fig. S3, when dsdA containing plasmid was transformed to Y1444 and Y31248 (the strain from the CICC), they can also grow in the minimal medium containing 500 μg/mL D-serine (Supplemental Fig. S3). Our results showed that dsdA was effectively expressed and could be used as a genetic marker in Y. lipolytica.

Fig. 1
figure 1

dsdA confers host resistance to D-serine and the ability to use D-serine as a nitrogen source in the Y. lipolytica strain PO1f. A dsdA encodes a D-serine deaminase that converts D-serine to pyruvate and ammonia. B Determining the concentration of D-serine to maintain the growth of PO1f after the transformation of dsdA. C The growth of the dsdA expression strains on 500 µg/mL D-serine. The transformants were plated on 500 µg/mL D-serine and incubated at 30 °C for 2 days. D Agarose gel confirmation of the presence of dsdA. After the transformation of the dsdA gene, genomic DNA was isolated, and the dsdA gene was amplified by PCR

Full size image

Determination of the selection condition of G418, zeocin, and mycophenolic acid

In addition to dsdA, we also tested the applicability of several other resistance markers, which have been used in different yeast species. G418 is widely used for genetic manipulation in Saccharomyces cerevisiae (Webster and Dickson 1983; Wang et al. 1996). It is reported that many prokaryotes and eukaryotes are sensitive to G418 (Colbère-Garapin et al. 1981). In S. cerevisiae, the kanMX gene coding for aminoglycoside 3-phospho-transferase (APH) confers resistance to G418 (Jimenez and Davies 1980). Here, we tested whether the drug can be used as a selectable marker in Y. lipolytica. The growth of PO1f was completely inhibited at a concentration of 7.2 mg/mL, which was much higher than the effective concentration in S. cerevisiae (Qiu et al. 2022) (Fig. 2A). We then evaluated the resistance of kanMX as a selectable marker by transforming the plasmids (YLEP-ET2-kanMX) that contain kanMX-transformed into PO1f. The kanMX-transformed cells gave rise to well-defined colonies on G418 plates. At the same time, control transformations (PO1f without kanMX gene) failed to form colonies (Fig. 2B). Both PCR and RT-qPCR results confirmed that the kanMX gene was expressed in PO1f (Fig. 2C, Supplemental Table S2, and Supplemental Fig. S2). Moreover, we demonstrated that the 7.2 mg/mL concentration of G418 also applies to screen W29 (Supplemental Fig. S4). However, the required concentration is much lower for Y1444 and Y31248 (Supplemental Fig. S5 and Supplemental Fig. S6).

Fig. 2
figure 2

Determination of the selection conditions for the dominant markers in PO1f. A Exploration of the drug concentration that inhibits the growth of Y. lipolytica strain PO1f. B The growth of the strain after transforming plasmids with selection-marker genes into PO1f. The transformants were plated on YPD plates (adding 7.2 mg/mL G418, 0.8 mg/mL zeocin, 0.5 mg/mL mycophenolic acid, and 6.3 mg/mL or 5 mg/mL CuSO4·5H2O, respectively) and incubated at 30 °C for 3 days. C Agarose gel confirmation of the presence of selection marker genes. After the transformation of a marker gene, genomic DNA was isolated, and the marker genes were amplified by PCR

Full size image

Zeocin is a copper-chelated glycopeptide antibiotic that causes cell death by intercalating into DNA and cleaving it (Utashima et al. 2017). The bleoR gene from S. hindustanus encodes a small acidic protein (MW 13665) that binds to zeocin at a specific ratio, making it unable to bind DNA and inhibiting its DNA cleavage activity (Drocourt et al. 1990; Leiting and Noegel 1991). The resistance gene against zeocin is an excellent marker for transforming eukaryotes and prokaryotes. Tsakraklides et al. (2018) used 1 mg/mL zeocin to select transformants Y. lipolytica strain YB-392, while Wang et al. (2022) used 50 to 250 μg/mL bleomycin to inhibit or delay the growth of strain PO1g. In our case, we also showed that the bleoR conferred PO1f resistance to zeocin when the selection condition of zeocin was 0.8 mg/mL for PO1f (Fig. 2, Supplemental Table S2, and Supplemental Fig. S2). Differently, W29, Y1444, and Y31248 were completely inhibited at the concentrations of 1.2 mg/mL, 2 mg/mL, and 2 mg/mL, higher than those needed for PO1f (Supplemental Fig. S4, Supplemental Fig. S5, and Supplemental Fig. S6).

It is reported that the guaB-encoding inosine monophosphate dehydrogenase from E. coli confers mycophenolic acid resistance (Farazi et al. 1997). The first report of the dominant selection marker in Y. lipolytica was published by Wagner et al. (2018). The concentration of mycophenolic acid they used for PO1fΔTRP1 was 100 mg/L. In our experiment, we tested the different concentrations of mycophenolic acid in PO1f and found that 500 mg/L mycophenolic acid could achieve the selection of guaB-containing transformants of PO1f (Fig. 2, Supplemental Table S2, and Supplemental Fig. S2). Furthermore, 1700 mg/L, 2000 mg/L, and 2000 mg/L of mycophenolic acid were required to inhibit the growth of W29, Y1444, and Y31248, respectively. These concentrations were higher than those necessary to inhibit PO1f (Supplemental Fig. S4, Supplemental Fig. S5, and Supplemental Fig. S6).

Copper confers toxicity in Y. lipolytica. It is reported that Y. lipolytica cannot grow when the concentration of CuSO4 is higher than 22 mM (Wang et al. 2011). Wang et al. (2011) overexpressed the endogenous copper-resistance gene CRF1 to resist higher copper concentrations (31 mM CuSO4) and successfully selected seven transformants into the strain W29. Here, we added different concentrations of CuSO4·5H2O into the YPD medium and found that CuSO4 could inhibit the growth of PO1f at 25 mM (6.3 mg/mL CuSO4·5H2O). However, when the CRF1 gene–containing plasmids (YLEP-ET2-CRF1) were transformed into PO1f, no transformant was obtained at 25 mM CuSO4 (Fig. 2A, B). Thus, we tried to reduce the concentration of CuSO4 to determine whether the CRF1 gene confers resistance to Cu2+. We found that reactions with 5 mg/mL CuSO4·5H2O did not inhibit cell growth completely, as there were very few colonies. Therefore, we used 20 mM (5 mg/mL CuSO4·5H2O) for the selection of transformants of the CRF1 gene. However, no transformants were obtained (Fig. 2B).

Engineering the Cre/lox system for marker recycling

The Cre/loxP system is used for Cre-mediated site-specific fragment excision when two copies of the loxP sequence flank the fragment DNA in tandem. To avoid repeated recombination of loxP sites, the mutant lox71 and lox66 sites were introduced into constructs to transform Corynebacterium glutamicum (Suzuki et al. 2005a, 2007; Hu et al. 2013). The mutant lox71 site contains mutations in the left inverted repeats of loxP, while lox66 contains mutations in the right inverted repeats (Suzuki et al. 2005a). The Cre-catalyzed recombination between lox66 and lox71 will produce a mutant lox site (loxLR) at the place of fragment excision that is poorly recognized by Cre recombinase (Fig. 3A). In our study, in constructs for the transformation of Y. lipolytica (PO1f-URA3), lox66 and lox71 were introduced at the left and the right sides of the URA3 marker, respectively (Supplemental Fig. S1). Cre recombinase was expressed by the transformation of the Cre-containing plasmid (YLEP-Leu-Cre). The transformants were randomly selected from PO1f-URA3 after Cre recombinase expression and streaked on YPD and SC-URA plates. As seen in Fig. 3B, no transformant could grow on a SC-URA plates. This result confirms that the Cre/mutant lox system can achieve marker deletion with a deletion efficiency of almost 100% (Fig. 3B).

Fig. 3
figure 3

The Cre/mutant lox-mediated marker recycling in the Y. lipolytica strain PO1f-URA3. A Recombination between a lox66 site and a lox71 site produces a wild-type loxP site within the excised DNA and a double-mutant site left in the host genome (Suzuki et al. 2005a). The double-mutant site cannot be recognized by Cre. B After the expression of Cre recombinase, the URA3 marker was efficiently removed from the genome of YLEP-Leu-Cre transformants. The left plate is a SC-URA plate while the right plate is a YPD plate. The plates were incubated at 30 °C for 3 days

Full size image

Discussion

In this study, we developed a dsdA selection marker for Y. lipolytica and determined the selection conditions of several infrequently used markers in Y. lipolytica. The dsdA gene has previously been developed for the genetic manipulation of S. cerevisiae (Vorachek-Warren and McCusker 2004). When using dsdA as the selection marker in S. cerevisiae, it was reported that it allows this yeast to grow in a medium with 500 µg/mL D-serine only when expressed in a CEN/ARS plasmid. However, when dsdA was integrated into the genome, an additional nitrogen source (5 mg/mL L-proline and 2 mg/mL D-serine) was required. In contrast to the situation in S. cerevisiae, the Y. lipolytica strains in our study could all grow well in 500 µg/mL D-serine with either a CEN plasmid containing the dsdA cassette or an integrated dsdA cassette in the genome.

Apart from dsdA, we also demonstrated that the kanMX gene can be used as a backup selection marker in Y. lipolytica. However, the selection concentration of G418 was higher than in other yeasts such as S. cerevisiae (Qiu et al. 2022). For the antibiotic markers we tested, we found that the selection concentrations of the antibiotics for different strains (including PO1f, W29, and Y. lipolytica strains from CICC) were different. For example, the inhibition concentrations of zeocin and mycophenolic acid for PO1f were lower than those needed for W29 and the strains from the CICC. However, the inhibition concentrations of G418 for PO1f are higher than those required by strains from the CICC. Compared with the auxotrophic strain of PO1f, the prototrophic strains seem to have a higher tolerance to zeocin and mycophenolic acid. These findings demonstrate that the concentrations of these drugs are not the same under different conditions. Ultimately, it is necessary to explore the appropriate concentrations for each specific strain.

In contrast to the previous study, in our study, the CRF1 gene did not confer PO1f the resistance to CuSO4 (Wang et al. 2011). These authors used CRF1 for gene disruption in W29, and they selected the recombinant strain in the presence of 31 mM CuSO4. However, we did not obtain positive transformants in the presence of 20 mM CuSO4. This may be because the expression level of the CRF1 gene was not high enough. Wang et al. (2011) did not provide details about the promoter of the CRF1 gene. Therefore, if the CRF1 gene is used as the selection marker, it is necessary to test the proper expression level in the following study.

Unlike the auxotrophic markers, the dominant markers developed here do not require any modification to the host organism. Therefore, they can be easily applied to different prototrophic strains. The markers developed and optimized here expand the genetic tools available for engineering in Y. lipolytica and may be helpful for cell factory construction.