Cyanamide-inducible expression of homing nuclease ^I−SceI for selectable marker removal and promoter characterisation in Saccharomyces cerevisiae

Abstract

In synthetic biology, microbial chassis including yeast Saccharomyces cerevisiae are iteratively engineered with increasing complexity and scale. Wet-lab genetic engineering tools are developed and optimised to facilitate strain construction but are often incompatible with each other due to shared regulatory elements, such as the galactose-inducible (GAL) promoter in S. cerevisiae. Here, we prototyped the cyanamide-induced ^I−SceI expression, which triggered double-strand DNA breaks (DSBs) for selectable marker removal. We further combined cyanamide-induced ^I−SceI-mediated DSB and maltose-induced MazF-mediated negative selection for plasmid-free in situ promoter substitution, which simplified the molecular cloning procedure for promoter characterisation. We then characterised three tetracycline-inducible promoters showing differential strength, a non-leaky β-estradiol-inducible promoter, cyanamide-inducible DDI2 promoter, bidirectional MAL32/MAL31 promoters, and five pairs of bidirectional GAL1/GAL10 promoters. Overall, alternative regulatory controls for genome engineering tools can be developed to facilitate genomic engineering for synthetic biology and metabolic engineering applications.

1. Introduction

In synthetic biology and metabolic engineering, yeasts have been intensively engineered to produce and improve production of certain fuels, chemicals, and other bioproducts [[1], [2], [3], [4]]. It often requires genetic modification on multiple genes and at multiple genomic loci to generate ideal genotypes and phenotypes [5,6]. Such modification capacity is rapidly evolving as the result of developing various engineering tools, including computational design/learn tools, laboratory build/test tools, and high throughput procedures assisted by robotic automation [[7], [8], [9], [10], [11]]. Wet-lab strain engineering is essentially important in this process but often labour-intensive and resource-consuming [12]. To accelerate the process, strain engineering tools and strategies can be further explored.

Genomic integration of transgenes in the common yeast Saccharomyces cerevisiae was first reported in the 1970s [13]. Following this, genomic engineering methodologies and strategies, including plasmid vectors [14,15], positive/negative selectable markers [[16], [17], [18]], multi-gene introduction [[19], [20], [21]], and large-scale genome synthesis [22,23], have been continuously developed and improved. In rational yeast engineering, gene knock-out and gene knock-in are essential to deliver identical genotypes according to the research design (or hypotheses). Inherit advantages in S. cerevisiae include its highly efficient homologous recombination, whereas other non-conventional yeasts, like Komagataella phaffii (previously known as Pichia pastoris), may be engineered to reach the comparable efficiency as well [24,25]. Through homologous recombination, endogenous genes can be deleted through a knock-out strategy and heterologous genes can be integrated through a knock-in strategy. To facilitate knock-out/knock-in engineering processes, molecular biology tools are deployed, i.e., tyrosine recombinase-mediated site-specific recombination (like Cre and Flp) [26], CRISPR-associated nuclease-mediated genome double-strand breaking (like Cas9 and Cpf1) [27], meganuclease-mediated genome double-strand breaking (like the homing nuclease I-SceI) [28], and serine recombinase-mediated site-specific recombination (like Bxb1 and PhiC31) [29]. These tools may be further modified for alternative purposes. For example, Cre-loxP recombination is used for SCRaMbLE [30], synthetic chromosome rearrangement and modification by loxP-mediated evolution. CRISPR tools are used for CRISPRi (CRISPR interference) [31] and CRISPRa (CRISPR activation) [32]. Ideally, all these tools can be compatible with each other to realise their functions in a single host, allowing the engineering and integration of complexed biological systems.

Some of these molecular biology tools are not compatible with each other for parallel applications in the host. For example, galactose-inducible promoters are used to control the induction of Cre-LoxP-mediated SCraMbLE [33], Cre-LoxP-mediated marker removal [34], ^I−SceI-mediated marker excision [28], ^I−SceI-mediated on-genome assembly of multiple genes [19], and mostly the induction of heterologous metabolic pathways in metabolic engineering [[35], [36], [37]]. To solve the compatibility problem, orthogonal genetic regulatory systems can be used to control each of these engineering machineries. In S. cerevisiae, both native and synthetic regulatory systems have been widely investigated in synthetic biology and yeast engineering studies [3,38,39]. The cyanamide-inducible promoter (DDI2) and maltose-inducible promoter (MAL32) are the options of native systems, and do not require the introduction of additional synthetic regulatory factors [[40], [41], [42]]. Therefore, we aimed to explore their usage for the orthogonal control on genome engineering tools.

In this study, we developed a cyanamide-inducible ^I−SceI system for selectable marker removal. Incorporating the maltose-inducible expression of the bacterial toxin gene MazF [18], we further developed an in-situ promoter substitution system to facilitate promoter characterisation. Using this tool, we characterised five categories of inducible promoters.

2. Materials and methods

2.1. Plasmid and strain construction

Yeast strains used in this study are listed in Supplementary Table 1. Plasmids used in this study are listed in Supplementary Table 2. Primers used in yeast colony PCR are listed in Supplementary Table 3. The CYC1core[4×Z268 + 1×LmrO] promoter sequence is listed in Supplementary Information file.

2.2. Yeast cultivation and transformation

Yeast extract Peptone media (YP media; 10 g L⁻¹ yeast extract, and 20 g L⁻¹ peptone), supplemented with 20 g L⁻¹ glucose (YP-glucose media), were used in general strain maintenance and development. Antibiotic G418 sulphate (300 μg mL⁻¹) or hygromycin B (100 μg mL^-1) was used in the selection of the strain harbouring the KanMX4 or HphMX selectable marker [34]. Yeast nitrogen base media (YNB media; 6.9 g L⁻¹ yeast nitrogen base without ammonia sulphate and ammino acids, pH 6.0), were used to grow the strain harbouring the URA3 selectable marker in yeast transformation and characterisation. When maltose and/or galactose were used as the carbon source(s), each of them was supplemented at the concentration of 20 g L⁻¹. For characterisation of genetic induction, 2 mM cyanamide (diluted from 4 M stock in H₂O), 125 μΜ tetracycline (diluted from 125 mM stock in dimethylsulfoxide:ethanol 1:1 mixture; stored at −80 °C), or 1 μM β-estrodiol (diluted from 1 mM stock in dimethylsulfoxide) was used. Agar (20 g L⁻¹) was added to prepare the solid media.

S. cerevisiae transformation was performed using the LiAc/ssDNA/PEG method with modifications [43]. Yeast cells from the recovery plate were inoculated into YP-glucose media at OD₆₀₀ = 0.001 for routine transformation or 0.0004 for the transformation with 2-h cyanamide induction and grown overnight. The LiAc/SScarrierDNA/PEG method [43] was followed to prepare the transformation mixture with heat-shock treatment. For antibiotic selection, cells collected from transformation mixture were precultured for at least 3 h in YP-glucose media and then spread onto the selection agar plates for further cultivation. For the yeast promoter knock-in with the negative selection through maltose-induced MazF toxin expression, cells were precultured in 1 mL YP-glucose media and incubated at room temperature overnight, and 10 μL of overnight culture was spread onto each YP-maltose agar plate supplemented with the required inducer like galactose, tetracycline, cyanamide, or β-estrodiol. After 48-h incubation at 30 °C, the plates were imaged using a BioRad ChemiDoc Imaging System.

For yeast promoter characterisation, yeast cells were pre-cultured in 200 μL YNB-glucose media in 5 ml sterile falcon tubes for ∼24 h to the stationary phase. The cultures were grown in a 200 rpm 30 °C incubator. For characterisation in YNB-glucose media, the precultures were diluted by 1000 times in 200 μL YNB-glucose media with or without inducer supplement. For the characterisation in YNB-galactose media, the precultures were diluted by 100 times in 200 μL YNB-galactose media. For the characterisation in YP media, the precultures were diluted by 10,000 times in 200 μL fresh YP-glucose or YP-maltose media. The diluted cultures were grown overnight for fluorescence measurement through flow cytometer.

2.3. Yeast colony PCR

Yeast colony PCR was performed using the protocol we published previously [5]. Yeast cells were resuspended in 5 μL of yeast cell digestion solution, the 10:1 mixture of 1 × phosphate-buffered saline (PBS) buffer and Zymolyase-20 T (nacalai tesque, Japan) stock solution (1U per μL in 1 × PBS buffer containing 100 mM DTT and 50 % v/v glycerol). Yeast cells were digested at 37 °C for 30 min, denatured at 95 °C for 5 min, and diluted with 100 μL of water. The mixture (1 μL) was used as the template in 10 μL of PrimeSTAR GXL DNA Polymerase (Takara Bio Inc., Japan) reaction. The typical PCR thermal-cycle conditions are (1) 98 °C for 30 s, (98 °C for 15 s, 50 °C for 10 s, 68 °C for 5 min)*3, (98 °C for 15 s, 55 °C for 10 s, 68 °C for 5 min)*27, 68 °C for 5 min. The extension duration at 68 °C was adjusted according to the amplicon length.

2.4. Flow cytometry

A BD FACSCanto II flow cytometry (Firmware version 1.49) was used to measure yEGFP [44] and E2Crimson [45] fluorescence in single cells. Yeast cells in overnight cultures were used directly for fluorescence analysis. yEGFP fluorescence was monitored at the FITC-A channel with a 488 nm laser. E2Crimson fluorescence was monitored at the APC-A channel with a 633 nm laser. Mean values of FSC-A (PMT voltage = 200; gating threshold = 3000), SSC-A (PMT voltage = 200), FITC-A (PMT voltage = 450), and APC-A (PMT voltage = 535) for all detected events were extracted using BD FACSDiva Software (version 8.0.1; CST Version 3.0.1; PLA version 2.0). yEGFP and E2Crimson fluorescence levels were expressed as the percentage of the average background autofluorescence from the exponential-phase cells of FP-negative reference strain CEN.PK113-7D [46].

3. Results

3.1. Cyanamide-inducible expression of ^I−SceI for selectable marker removal

In S. cerevisiae, ^I−SceI expression causes the double-strand break (DSB) at the 18 bp ^I−SceI recognition site (5′-AGTTACGCTAGGGATAACAGGGTAATATAG-3′) [19,28]. DSBs trigger DNA repair, prevalently through homologous recombination in S. cerevisiae when homologous sequences surround ^I−SceI recognition site. This mechanism was applied to facilitate selectable marker removal and the assembly of multiple genes into genome [19,28]. Previously, galactose-inducible GAL1 promoter was used to control ^I−SceI expression [19,28]. However, GAL promoters are broadly used to control the expression of heterologous genes in metabolic engineering, i.e., the genes in terpene synthetic pathways [5,14,47]. Induction of GAL promoters-controlled terpene pathway genes may cause metabolic burden, challenging the strain development and therefore not compatible with GAL promoter-mediated induction of ^I−SceI expression. We aimed to use an alternative regulatory promoter to control ^I−SceI expression to facilitate strain engineering. The DDI2 promoter was considered because of its binary ‘OFF-to-ON’ induction in the presence of cyanamide [40,41].

We constructed a two-vector system to introduce the ^I−SceI expression cassette under the control of the DDI2 promoter (Fig. 1A: pITESceIcmd & pITEdv1). These two plasmids were co-transformed into CEN.PK113-32D [46] genome with the replacement of the HO gene, the gene required for mating-type switch in natural S. cerevisiae isolates (Fig. 1A: strain oQIE0 with genotypes confirmed by yeast colony PCR). As the results, the 174-bp repeated sequences were at the both ends of the fragment ^I−SceI_Site-HphMX (the hygromycin resistant selectable marker)-^I-SceI_site. We grew oQIE0 in YPD broth supplemented with 2 mM cyanamide from a low OD₆₀₀ ∼ 0.001 for 20 h and then isolated the single colonies by spreading the diluted culture on a YPD agar plate supplemented with 2 mM cyanamide. We tested 24 colonies through yeast colony PCR, and the Hph gene was successfully removed in ∼75 % of the colonies, shown by the loss of the amplification of PCR fragment #3 (Fig. 1A: generating strain oQIE1).

Fig. 1.

Cyanamide-inducible expression of ^I−SceI facilitating selectable marker removal for iterative strain engineering. (A) Integration of cyanamide-inducible ^I−SceI cassette at ho locus and selectable marker removal. (B) Integration of tetracycline-inducible GAL80 module for tetracycline-mediated repression on galactose-inducible promoters and selectable marker removal. (C) Integration of the mevalonate pathway modules and selectable marker removal. Negative, the KanMX4 marker not being removed; Positive, the KanMX4 marker being removed. P_XXX, the promoter of gene XXX; T_XXX, the terminator of gene XXX. T_synth3, a synthetic minimal terminator. NLS, nucleus-localising sequence. TetR-HPV16L^NLS-TUP1, encoding a tetracycline-depressible repressor. TetO, TetR-binding sequence. Metabolic pathway genes for terpene precursor synthesis: ERG12, ERG8, MVD1, IDI1, EfmvaS, ACS2, and EfmvaE. GAL, the genes involving in galactose metabolism and regulation. M, Thermo Scientific GeneRuler DNA Ladder Mix (Thermo Scientific#SM0331). #1, yeast clones replicated on the YPD agar supplemented with 300 μg ml⁻¹ G418. The yeast colony PCR results are shown in DNA agarose gel images.

To further investigate the applications of the marker-recycling mechanism, we re-constructed the background platform strain for improved synthesis of heterologous terpenoids. Firstly, we transformed the construct that encoded the genetic circuit for tetracycline-mediated repression of GAL promoters (Fig. 1B: pIRTetR-GAL80) [38]. This construct is targeted to replace the promoter of the GAL repressor gene GAL80 with the TEF1 promoter inserted with 4 TetO elements, the G418 selectable marker KanMX4, and the expression cassette of an artificial transcription factor TetR-Tup1p. In this construct, we have previously cloned two 40-bp repeated sequences at the both ends of KanMX4-^I-SceI_site. Here, the 18-mer ^I−SceI_site (5′-TAGGGATAACAGGGTAAT-3′) was used. This construct was transformed into strain oQIE1 to generate strain oQIE2 (Fig. 1B: with genotype confirmed by yeast colony PCR). After cyanamide-mediated induction of ^I−SceI, 27 colonies were characterised through yeast colony PCR. None of these colonies showed the successful removal of the KanMX4 marker. Instead, each colony likely contained the mixture of two genotypes: KanMX4 unremoved (Fig. 1B: PCR fragment #5) and KanMX removed (Fig. 1B: PCR fragment #7). We chose first four colonies (replicated on the plate without G418 supplemented) to perform the second round of cyanamide induction and tested 27 colonies for the G418 resistance. Five colonies had the KanMX marker removed (Fig. 1B: strain oQIE3). This case indicates that 40-bp repeats and a single 18-mer SceI_site do not work efficiently for marker removal.

We further aimed to transform two plasmids pIMVAd39T (Addgene #98303) and pIMVAu1 (Addgene #98301) to overexpress the genes from the mevalonate pathway [47]. In these two plasmids, autotrophic selectable markers were used for yeast transformation, but were not available in oQIE3. We further used the KanMX4 marker containing two ^I−SceI sites and the 80-mer AB repeats (EuroScarf#P20769: pDS1) [28], to investigate the efficiency of cyanamide-induced marker removal. We replaced the selectable marker in pIMVAd39T and pIMVAu1 with AB-^I-SceI-KanMX4-^I-SceI-AB fragment, and sequentially transformed them into oQIE3 with cyanamide-induced marker removal (Fig. 1C). The success rates for selectable marker removal were ∼37 % and 33 %. These are lower than the efficiency (>50 %) reported previously when galactose-inducible promoter was used for ^I−SceI induction [47].

3.2. In situ gene replacement design for plasmid-free cloning of heterologous promoters into yeast genome

The fundamental principles in metabolic engineering and synthetic biology rely on the expression of heterologous genes under the control of different promoters [48]. The promoters can be either constitutive or regulatory and show a range of transcription strengths [49]. Nevertheless, it is important to characterise these features for genetic design in metabolic engineering and synthetic biology. In S. cerevisiae, promoters can be characterised using an episomal plasmid system or a genome-integrating system. Genome-integrating systems may deliver a genetically homogeneous cell population, independent from specific selection pressure for plasmid maintenance. However, the molecular cloning procedure for genome-integrating systems is very lengthy, time-consuming, and labour-intensive. We, applying cyanamide-inducible ^I−SceI expression, designed an in-situ promoter substitution system, which can significantly simplify the molecular cloning process (Fig. 2).

Fig. 2.

Diagrams of in situ gene replacement design for plasmid-free cloning of heterologous promoters into yeast genome. MazF, encoding an E. coli interferase toxin.

We constructed two plasmids, pILGFPEasy6R and pILGFPEasy8R. Both plasmids can be integrated into the yeast genome at the ura3 locus, containing a KlURA3 selectable marker, yEGFP gene fused with the PGK1 terminator, an ^I−SceI site, a G418-resistant selectable marker KanMX4, a MAL32 promoter-controlled toxin gene MazF inserted with the RPL28 intron. In plasmid pILGFPEasy8R, a E2Crimson reporter [50,51] fused with the PDC1 terminator was inserted at the downstream of MazF and in the opposite direction of yEGFP gene. For in situ promoter substitution, the plasmid pILGFPEasy6R or pILGFPEasy8R can be transformed into a background strain harbouring a cyanamide-inducible ^I−SceI expression cassette. Prior to yeast transformation, cyanamide is added to induce ^I−SceI expression, and the transformed promoter fragment can repair ^I−SceI-mediated DSBs through homologous recombination. The cell mixtures can be selected on YP-maltose agar. In theory, toxin MazF gene is induced by maltose, and only the cells with MazF replaced can form colonies. Furthermore, the inducer chemicals can be supplemented to activate the heterologous promoter for yEGFP fluorescence-assisted selection.

We further constructed another plasmid (pILGFPEasy9R; Supplementary Fig. 2; Addgene# 218291) by inserting the P_DDI2-^I-SceI-T_synth3 expression cassette into the downstream of the PDC1 terminator in plasmid pILGFPEasy8R. This plasmid can be used in the background strain that does not harbour a pre-transformed ^I−SceI expression cassette.

3.3. Characterisation of regulatory promoters in S. cerevisiae

To test efficiency of in situ promoter substitution, we characterised a set of regulatory promoters, which were regulated by either endogenous regulatory mechanisms or synthetic regulatory mechanisms. We constructed a yeast strain (strain oQR36o). In strain oQR36o, four heterologous expression cassettes were introduced at the HO locus: (1) an artificial transcription factor (TF) Zif268-hER-VP16^A (the fusion of Zif268 zinc finger DNA binding domain, estrogen sensing domain, and VP16 transactivating domain) was expressed under the control of the YPK2 promoter, a weak promoter; (2) an artificial TF LmrA-SV40^NSL-Mig1^C (the fusion of a bacterial TetR-like TF LmrA, SV40 nucleus-localising sequence, and Mig1p C-terminal trans-repression domain) was expressed under the control of the ADR1 promoter; (3) an artificial TF TetR-HPV16L^NLS-Tup1 (the fusion of tetracycline repressor TetR-HPV16L nucleus-localising sequence, and trans-repressing TF Tup1; TetR-Tup1) was expressed under the control of the HAC1 promoter; and (4) ^I−SceI was expressed under the control of cyanamide-inducible DDI2 promoter.

Plasmid pILGFPEasy6R (linearised by PmeI) was transformed into strain oQR36o for characterisation of mono-directional promoters, and plasmid pILGFPEasy8R (linearised by PmeI) was transformed for characterisation of bidirectional promoters. Noted is that pILGFPEasy8R covers pILGFPEasy6R function. Each promoter fragment contains 40 or 60-bp homologous recombination arms for genomic integration. For yeast transformation, cells were spread to YP-maltose agar plates. At this step, it is recommended to spread 10 μl or less cells per plate, otherwise it can be difficult to isolate single colonies. Although brighter colonies were only a small proportion of those plated, they were found to be correctly transformed (Supplementary Fig. 1). Yeast colony PCR showed the successful promoter substitution. However, in GAL10/1 promoter cases, it was difficult to isolate the single bright colonies from other non-fluorescent colonies on yeast transformation plates, which led to the amplification of two DNA fragments (Supplementary Fig. 1). Therefore, spreading less cells per plates is important for some cases. In the cases that we could not isolate the pure colonies from the transformation plates, the pure fluorescent colonies were isolated through streaking on fresh plates for further characterisation.

To test the use of pILGFPEasy9R, pILGFPEasy9R (linearised by PmeI) was transformed into strain CEN.PK113-5D. Using the resulting strain, we transformed S. kudriavzevii GAL1 promoter [52]. Consistently, brighter colonies, although forming a smaller population, were verified using yeast colony PCR (Supplementary Fig. 2).

In total, we characterised five groups of inducible promoters. They are (1) tetracycline-inducible TEF1[4×TetO] promoters, (2) cyanamide-inducible DDI2 promoter, (3) synthetic β-estradiol-inducible promoters, (4) maltose-inducible MAL32/31 bidirectional promoters, and (5) five galactose-inducible GAL10/1 promoters.

We previously synthesised a TEF1[4×TetO] promoter by inserting four TetO motifs between the core promoter and upstream activation sequence (UAS) of the TEF1 promoter [38]. The TEF1[4×TetO] promoter is repressed in the strain expressing TetR-Tup1 and is inducible by tetracycline addition. The originally tested TEF1[4×TetO] promoter (Fig. 3A: UAS-579), containing a full-length UAS, demonstrates a moderate transcription strength. The tetracycline-inducible promoters with weaker strength might be useful to control the expression of certain genes, like the TF-encoding gene. We hypothesized that truncating UAS sequences might decrease promoter strength. We then tested two truncated TEF1[4×TetO] promoters, UAS-382 (the promoter starting at −382 position of the TEF1 gene) and UAS-288 (the promoter starting at −288 position). Consistent with the hypothesis, truncated TEF1[4×TetO] promoters were weaker than the full-length one (Fig. 3A) by up to 62 %.

Fig. 3.

Characterisation of regulatory promoters in S. cerevisiae. (A) Tetracycline-inducible TEF1[4×TetO] promoters: truncating the upstream activation sequence (UAS) resulted in decreased expression outputs. TetR-HPV16L^NLS-Tup1, an artificial trans-repressor binds to TetO sequences. Tetracycline binding to TetR disassociates the TetR-TetO interaction. (B) Cyanamide-inducible DDI2 promoter. (3) Estrogen (β-estrodiol)-inducible promoters. Zif268-hER-VP16, an artificial trans-activator targeting to Z268 sequences. β-Estrodiol binding results in nuclear translocation of hER (human estrogen receptor). LmrA-SV40^NSL-Mig1^C, an artificial trans-repressor binding to the LmrO sequence. (D) Maltose-inducible promoters. (E) Galactose-inducible promoters. (D & E) The bold lines show the fluorescence level of yEGFP under the control of the TEF1 promoter, and dashed lines were of E2Crimson. Cells were grown in YNB-glucose media (A and C), YP-glucose media (B & D), YP-maltose media (D), or YNB-galactose media (E). Mean values ± standard deviation were shown (n = 3 biological replicates).

We characterised the DDI2 promoter which was used to control ^I−SceI expression in this study. In previous studies [40,41], the DDI2 promoter was not characterised with an external reference. The DDI2 promoter showed a strength lower than the TEF1 promoter, causing ∼6.4-fold decrease in yEGFP fluorescence (Fig. 3B).

Synthetic estrogen-inducible promoters were developed previously [53]. However, these promoters were leaky, i.e., for the CYC1 core promoter fused with Zif268-binding elements (Fig. 3C) [38] and the GAL1 promoter inserted with Zif268-binding elements [39]. To solve this problem, we hypothesized that an artificial trans-repressor could be used to repress its basal expression. Here, we tested a synthetic CYC1core promoter fused with four Z268 (Zif268-binding) elements and one LmrA-binding LmrO sequence (CYC1core[4×Z268 + 1×LmrO]). The CYC1core[4×Z268 + 1×LmrO] showed a very low basal expression and was inducible by estrogen β-estradiol. But the induction level was lower than that of the CYC1core[4×Z268] promoter, shown by ∼4.7-fold decrease in yEGFP fluorescence (Fig. 3C).

For bidirectional promoters, we characterised maltose-inducible MAL32/31 promoters and the galactose-inducible GAL1/10 promoters from five Saccharomyces species: S. cerevisiae (ScGAL10/1), S. arboricola (SaGAL1/10), S. kudriavzevii (SkGAL1/10), S. mikatae (SmGAL1/10), and S. paradoxus (SpdGAL1/10). The MAL32/31 and ScGAL1/10 promoters have been characterised for bidirectional transcriptional activation and strength previously [52]. For the SaGAL1/10, SkGAL1/10, SmGAL1/10, and SpdGAL1/10 promoters, we previously measured the strength at the GAL1 direction, and the GAL10 promoters were not tested. We therefore used our system to re-characterise their activities. As the external reference, the strength of the TEF1 promoter was characterised on glucose using E2Crimson, which was under the control of the PGK1 terminator (Fig. 3D & E: dashed line). Consistent with the previous study, both sides of the MAL32/31 promoter were induced by maltose. The MAL32 promoter was induced to the similar level of the TEF1 promoter, and the MAL31 promoter was at the halved level the TEF1 promoter.

The GAL1/GAL10 promoters were characterised on galactose in a GAL80 wildtype background. Using E2Crimson as the reporter, the output from the ScGAL1 promoter was ∼2.6-fold of that from the TEF1 promoter. This fold-change was in the same range with the ∼2.5 fold in our previous measurement when yEGFP was used as the reporter [49]. For the GAL1 promoters, the strength was sequenced as: SkGAL1 > SmGAL1 > SaGAL1/SpdGAL1 > ScGAL1. All five GAL10 promoters were stronger than the TEF1 promoter. The strongest GAL10 promoter was the SmGAL10 promoter. The SpdGAL10 promoter was the weakest GAL10 promoter with the activity higher than the TEF1 promoter by 50 %. These suggest that these five bidirectional GAL1/10 promoters are good options for gene overexpression in metabolic engineering.

4. Discussion

In this study, we tested two yeast engineering methods for selectable marker removal and the plasmid-free promoter characterisation. Although not showing superior engineering efficiency, both methods can practically facilitate routine yeast engineering. Applying cyanamide-mediated gene induction allows its compatibility with pathway engineering using the galactose-inducible promoters. The in-situ promoter substitution system minimises the operational steps for characterisation of yeast promoters through genome integration. Both systems can be further improved and may be replicable in other non-conventional yeast platforms.

The cyanamide-inducible DDI2 promoter was used in the aim to deliver orthogonal control on different genetic engineering tools. However, in comparison to the marker removal system applying the GAL1 promoter to control ^I−SceI expression [47], the system applying the DDI2 promoter showed a lower marker removal efficiency. This might be due to that the strength of the DDI2 promoter is lower than the strength of the GAL1 promoter by ∼17 folds (Fig. 3). In further development, alternative small-molecule-inducible genetic circuits can be considered, i.e., the ligand/factor pairs of 2,4-diacetylphloroglucinol/PhlF [54], camphor/CamR [54], 1,2-bis(4-hydroxyphenyl)ethane-1,2-dione/hER mutant [55], steroids/mammalian type I nuclear receptors [56]. These inducible circuits may not only be useful to control alternative meganucleases or genome-engineering tools, but also alternative killing toxins (like MazF; Fig. 3) for negative selections.

Nevertheless, it might be necessary to optimise the regulatory promoters for a tight ON/OFF response and minimise the promoter lengths [57] to facilitate PCR-based cloning and assembly. We have previously identified that the fusion of 4 × Z268-containing DNA sequence to the CYC1 core promoter contributed to the basal expression [38], possibly due to endogenous trans-activators activating the transcription. Here, we solved the leakiness problem in a β-estradiol-inducible genetic circuit by introduction of a synthetic trans-repression mechanism in the regulatory promoter (Fig. 3C). The high affinity of a synthetic trans-activator to its target binding sites might also result in the leaky trans-activation [58]. To solve this problem, using a low-affinity DNA-binding domain eliminated the leakiness at the expense of the induction strength. To increase the induction output strength in this case, a molecular clamp was used to stabilise the interaction of multiple synthetic trans-activators and target DNA motifs at the target promoter. In the tetracycline-inducible circuit case, we showed that the length of upstream activation sequence (UAS) determined the induction strength (Fig. 3A). Consistently, tandem assembly of multi-UASs has been used to engineer super strong yeast promoters [59]. These emphasise that there are alternative solutions in synthetic biology to achieve similar biological effects and genetic circuit design principles shall be discovered for precise genetic design.

In summary, we prototyped the cyanamide-induced meganuclease tools to facilitate in situ genome engineering in S. cerevisiae. These tools may facilitate strain engineering at multiple genome loci and characterisation of yeast promoters. Availability of multiple fine small-molecule-inducible genetic circuits are important to develop more complexed networks for precise control on more genome engineering tools in a single cell [60,61]. In this scenario, it will be also important to develop computer software so that individual genome-engineering tools may work synergistically in a biofoundry setting to power the engineering biology capacity for synthetic biology and metabolic engineering applications [62].

CRediT authorship contribution statement

Liam McDonnell: Investigation, Methodology, Validation, Writing – review & editing, Formal analysis. Samuel Evans: Formal analysis, Investigation, Methodology, Validation, Writing – review & editing. Zeyu Lu: Methodology, Resources, Supervision. Mitch Suchoronczak: Investigation. Jonah Leighton: Investigation. Eugene Ordeniza: Investigation. Blake Ritchie: Investigation. Nik Valado: Investigation. Niamh Walsh: Investigation. James Antoney: Investigation, Resources. Chengqiang Wang: Conceptualization, Formal analysis. Carlos Horacio Luna-Flores: Conceptualization, Formal analysis, Resources. Colin Scott: Conceptualization, Resources. Robert Speight: Conceptualization, Funding acquisition, Project administration, Resources, Supervision. Claudia E. Vickers: Conceptualization, Funding acquisition, Resources, Supervision. Bingyin Peng: Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Validation, Visualization, Writing – original draft.

Declaration of competing interest

Authors declare no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article.