Fine-tuning the expression of target genes using a DDI2 promoter gene switch in budding yeast

Tuned gene expression is crucial to the proper growth and response to the environmental changes of an organism. To enable tunable gene expression as designed is desirable in both scientific research and industrial application. Here, we introduce a novel promoter switching method based on the DDI2 promoter (PDDI2) that can fine tune the expression of target genes. We constructed a recyclable cassette (PDDI2-URA3-PDDI2) and integrated it upstream of yeast target genes to replace the native promoters by DDI2 promoter without introducing any junk sequence. We found that the presence or absence of cyanamide as an inducer could turn on or off the expression of target genes. In addition, we showed that PDDI2 could act as a gene switch to linearly regulate the expression levels of target genes in vivo. We switched the original promoters of RAD18, TUP1, and CDC6 with PDDI2 as a proof-of-concept.

located on different chromosomes. Protein sequence analysis and experimental verification revealed that DDI2/3 encode a cyanamide hydratase in S. cerevisiae [25][26][27] . Meanwhile, it has been reported that cyanamide can highly induce the expression of DDI2/3 as MMS 24,28 . Cyanamide is a clean fertilizer because in the air it can be converted naturally into ammonia and carbon dioxide 24 . Cyanamide is also used as an alcohol deterrent drug 29,30 and as a raw compound in the pharmaceutical industry to produce guanidine derivatives 31 . Hence, we propose to utilize the DDI2 promoter to develop a novel gene expression regulating system induced by cyanamide.
Here, we used the superfolder green fluorescent protein gene (sfGFP) as a reporter gene to determine the strength and regulation of P DDI2 compared with three other widely-used promoters (P ADH1 , P CUP1 and P GAL1 ) 32 . Quantitative analysis results demonstrated that the induction level of P DDI2 is strong and leaky level is low. As a proof-of-concept, we replaced the native promoters of three yeast genes (RAD18, TUP1, and CDC6) with P DDI2 , without redundant DNA sequence left. As anticipated, P DDI2 showed tunable control of the target genes, especially the regulation of the essential gene CDC6 as a yeast safeguard switch. We demonstrated a linear gene expression control on the target gene correlated with the inducing strength, which further facilitates the precise regulation of gene expression in budding yeast.

Results
optimal DDI2 promoter induction time and inducer concentration. P DDI2 is a novel inducible promoter that is efficiently induced by cyanamide 24,28 . We tested whether the strength of P DDI2 is affected by both the induction time and inducer concentration. To determine the optimal induction time and cyanamide concentration, we established a reporter gene system in which an sfGFP gene is under the control of DDI2 promoter (see Supplementary Fig. S1a). To avoid the potential effect from gene copy number variations a single-copy plasmid YCp (yeast centromere plasmid) was chosen as the backbone vector 33 . We measured the expression of the P DDI2 -sfGFP reporter gene at different induction time points (3-5 h) and cyanamide concentrations (0-8 mM) by flow cytometry. As shown in Fig. 1a, the mean fluorescence values of sfGFP increase with increasing cyanamide concentrations (0-8 mM) and stabilize when cyanamide concentrations are 5 mM and above. The highest sfGFP expression level is observed after 5-hour induction among 3-5 hour induction time points, which implies that P DDI2 activity is proportional to the induction time.
It has been reported that cyanamide has modest toxicity to humans 34 . To investigate the effect of cyanamide toxicity on the growth of yeast cells, we generated growth curves for wild-type S. cerevisiae cells grown in YPD medium supplemented with 0, 5 and 8 mM cyanamide. Figure 1b shows that the growth rate of yeast cells is hardly affected by 5 mM cyanamide, but with 8 mM cyanamide the cell concentration (OD600) is slightly affected and at longer induction time the cell growth rate is lower than at 0 and 5 mM cyanamide concentrations. These data suggest that P DDI2 is induced efficiently by cyanamide and has high promoter activity and that cyanamide has little effect on cell growth at concentrations equal to or lower than 5 mM. comparison of the DDI2 promoter with ADH1, CUP1 and GAL1 promoters. We compared the strength and induction process of P DDI2 with three classic promoters, P ADH1 , P CUP1 and P GAL1 . To facilitate the comparisons, we established the same reporter gene system as described above just by replacing DDI2 promoter with three other promoters. The four plasmids carrying P DDI2 -sfGFP, P ADH1 -sfGFP, P CUP1 -sfGFP and P GAL1 -sfGFP (see Supplementary Fig. S1a-d) were transferred to the S. cerevisiae cells. After the yeast cells were induced by corresponding inducer or water for 3 hours, we irradiated cells with the 488 nm laser to observe sfGFP signal by fluorescence microscopy. As shown in Fig. 2a, the fluorescence intensity of yFYV11 (P DDI2-sfGFP ) cells is much greater than that of yFYV9 (P ADH1-sfGFP ) cells. Comparing the three inducible promoters, the yFYV11 (P DDI2-sfGFP ) cells show higher fluorescence intensity than both yFYV10 (P GAL1-sfGFP ) and yFYV17 (P CUP1-sfGFP ) cells, (Fig. 2a). To get quantitive data to confirm these observations, we measured the expression of sfGFP using flow cytometry in different conditions. For yFYV17 (P CUP1 -sfGFP) cells, a final concentration of 0.5 mM CuSO 4 was applied for induction according to previous study 35 , then the mean fluorescence values were measured at different time points. yFYV10 (P GAL1 -sfGFP) cells were grown to log phase in glucose medium, then washed and incubated in raffinose medium for 3 hours, which does not repress or induce transcription of the GAL1 promoter 36 , prior to inducing in the galactose medium for 2-4 hours as described in the previous reports [37][38][39] . The yFYV11 (P DDI2 -sfGFP) cells were directly induced by 5 mM cyanamide for 2-4 hours.
As it is expected, the mean fluorescence values for yFYV9 (P ADH1 -sfGFP) cells are relatively low and stable at different time points (Fig. 2b). At the 4-hour time point, compared the mean fluorescence values of yFYV11 (P DDI2 -sfGFP) cells induced with 5 mM cyanamide to that of yFYV9 (P ADH1 -sfGFP) cells and yFYV17 (P CUP1 -sfGFP) induced by 0.5 mM CuSO 4 , P DDI2 exhibited upwards of an 8.72-fold and a 2.03-fold increase in mean fluorescence over P ADH1 and P CUP1 . After 4-hour induction, the mean fluorescence of yFYV10 (P GAL1 -sfGFP) reaches a rather high level. However, the mean fluorescence of P DDI2 is still 1.23-fold higher than that of P GAL1 at this time point (Fig. 2b). These results suggest that P DDI2 has significantly higher activity than P ADH1 , P CUP1 , and P GAL1 at the same induction time. Hence, P DDI2 can be used as a novel promoter with high inducibility.
Switching the native promoters of target genes RAD18 and TUP1 with DDI2 promoter. P DDI2 can be induced by cyanamide, therefore, it is reasonable that the native promoter of a target gene on its genomic locus could be replaced with P DDI2 to regulate its expression. As a proof-of-concept, we first replaced the promoters of two target genes (RAD18 and TUP1) with P DDI2 . RAD18 plays an important role in DNA damage repair caused by UV or methyl methanesulfonate 23,40 . TUP1 is a general repressor of transcription in yeast, and a tup1Δ strain exhibits an obvious clumpy cell morphology phenotype 41,42 that can be observed directly.
We used the recyclable cassettes method reported previously 37,43 to replace the native promoter in the yeast genome by one-step transformation followed by two-step selection. First, the P DDI2 -URA3-P DDI2 (DUD) cassettes were amplified by PCR to replace the promoter of the target gene by homologous recombination. Secondly, the cassette-integrated strain was obtained using auxotrophic plates (SD-URA). The most important step is that one copy of the P DDI2 and URA3 deleted cells are selected on 5-FOA plates so that only one P DDI2 regulating the target gene is obtained without redundant DNA. For more efficient promoter shuffling, the several hundred base pairs flanking the homology arms were generated by overlapping PCR.
To confirm that P DDI2 could switch on/off gene expression on the yeast genome, we labeled the two target genes (RAD18 and TUP1) with 5 × Flag tag in both the promoter-shuffled and corresponding wild-type strains. In order to determine the relationship between protein abundance and cyanamide concentrations, yFYV12 (P DDI2 -RAD18) cells were induced with 0-9 mM cyanamide (Fig. 3a). It can be seen that the abundance of the Rad18-5 × FLAG increased linearly after the 0-9 mM cyanamide induction and almost no signal is detected in the absence of cyanamide. The subsequent quantitative analyses show that protein abundance is linearly related to the cyanamide concentration (Fig. 3a). Hence, the protein abundance can be quantitatively regulated by the P DDI2 .
Because rad18Δ strain is radiation sensitive, we measured the survival rate of cells after irradiation with different UV doses. The yFYV12 (0/8 mM), yFYV15 (rad18Δ) and wild-type cells were respectively spread on YPD plates containing cyanamide, then exposed with different doses of UV. The yFYV12 (P DDI2 -RAD18) cells that were treated with 0 and 8 mM cyanamide show similar survival rates as yFYV15 (rad18Δ) and wild-type strains, respectively (Fig. 3b).
For the yFYV13 (P DDI2 -TUP1) cells, we first observed phenotypic changes at different cyanamide concentrations. The tup1Δ strain (yFYV16) was used as a negative control. As shown in Fig. 3c, the yFYV13 cells without cyanamide settle on the bottom of the tube, similar to the yFYV16 strain. Meanwhile, yFYV13 cells induced www.nature.com/scientificreports www.nature.com/scientificreports/ with 5 mM cyanamide were suspended in the same medium as the wild-type cells. Next, we examined the Tup1-5 × FLAG protein abundance by Western blot after treatment with 0, 2, and 4 mM cyanamide, respectively. We detected an obvious increase in Tup1-5 × FLAG protein level after cyanamide treatment (2 or 4 mM), and no Tup1-5 × FLAG protein expression was detected without inducer cyanamide (Fig. 3d). The results indicate that, by using different concentrations of cyanamide, P DDI2 is capable of boosting the expression of the target gene to a level that compares to the tup1Δ and wild-type strains. Together, these data suggest that the P DDI2 gene switch can fine-tune the expression of the selected target genes.
Using DDI2 promoter controls the expression of essential gene CDC6. To investigate whether the P DDI2 can linearly and tightly regulate the expression of essential genes as well. Here, we chose CDC6 encoding the protein required in the pre-replicative complex formation in DNA replication 44 as a gene of interest. Compared to other essential gene manipulating method which must transfer the strain with a wild copy of the essential gene to ensure the survival during the modification, we replaced the promoter of CDC6 with the P DDI2 in a simple and efficient method. We designed a P DDI2 -down fragment (Fig. 4a) which partially drove the expression of the essential gene under the induction of cyanamide. Thus, only adding cyanamide can ensure the cell viability during the promoter replacing process, and an extra wild-type copy of the target gene is unnecessary. Based on the recyclable cassette method 37 , we replaced the CDC6 native promoter with P DDI2 (Fig. 4a).
To confirm that the P DDI2 can tightly control the essential gene (CDC6) expression as a gene switch. Overnight cultured yFYV14 (P DDI2 -CDC6) cells (with 5 mM cyanamide) were respectively transferred into 5 mM cyanamide or cyanamide-free medium for 4 h. Finally, 10000, 1000, and 500 cells were spread on YPD plates containing 0 and 5 mM cyanamide. As shown in Fig. 4c, no cells survive on YPD plates without cyanamide, which clearly indicates that the P DDI2 is a tight promoter with no detectable expression leakage. The expression of the Cdc6-5 × FLAG protein was also examined by western blot analysis using an anti-Flag antibody, and a corresponding experimental result is shown in Fig. 4b. It clearly indicates that the protein production corresponds well with the cyanamide concentration (0-5 mM). A small amount of Cdc6-5 × FLAG protein was detected in 0 mM cyanamide sample probably due to intracellular residual cyanamide from the overnight culture. The western blot results show that a www.nature.com/scientificreports www.nature.com/scientificreports/ linear equation fitted the changes in protein abundance. Together, these results implied that the P DDI2 can act as a gene switch to control the expression of essential genes.

Discussion
In this study, we demonstrate that P DDI2 is a strong inducible promoter. We establish a simple promoter switching method to control the expression of target genes on its genomic locus in budding yeast and quantitatively analyzed the expression of target genes under the control of P DDI2 . Our study suggests the potential of P DDI2 for linear control of gene expression and as a tight on-off switch.
Four plasmids, containing constitutive promoter ADH1 or inducible promoters GAL1, CUP1 and DDI2 linked to the sfGFP reporter gene were constructed to compare promoter strengths. The experimental results show that the strength of P DDI2 is stronger than all P ADH1 , P CUP1 , and P GAL1 (Fig. 2). It has been reported that inducible promoter P CUP1 is leaky 45,46 . We also observed a rather high basal expressing level of sfGFP even in absence of CuSO 4 induction (data not shown). Both GAL1 and DDI2 are stringent promoters. In our experiments, the sfGFP expression under control of these two promoters was undetectable in the absence of corresponding inducers, which is important in cases where background expression is unacceptable, such as the expression of proteins is toxic to yeast cell. Unlike P GAL1 , P DDI2 did not require replacement of the culture medium during the process www.nature.com/scientificreports www.nature.com/scientificreports/ of induction, thus total culture time was shorter. Moreover, the induced strength of P DDI2 was higher than P GAL1 under the corresponding conditions, and the cost of the cyanamide inducer was less than that of P GAL1 (galactose) 8 . Accordingly, using P DDI2 to precisely control gene expression can save a lot of effort and be economical for researchers. More importantly, due to the low price of cyanamide, it has good application prospects in the industrial production that requires high-level induction of target gene with low cost.
Furthermore, the DDI2 promoter can act as a linear gene switch to regulate the gene expression. As a proof-of-concept, the parental promoters of RAD18, TUP1, CDC6 genes were successfully replaced with DDI2 promoter in the genomic locus, so that the expression of the above three genes are under the control of P DDI2 as desired. Especially, for regulating an essential gene on its genomic locus, there are several advantages for the P DDI2 -shuffling method over other gene regulated expression systems that function in yeast. Because of the special character of DDI2 promoter, an extra wild-type copy of the target gene is not required to transfer into the host cell during the promoter switching process, which is simply not possible with other systems. Thus, the counter-selection step for losing the extra copy of the target gene can be omitted to save time and the cost. The linear correlation between cyanamide concentration and expression level is a great benefit for regulating an essential gene, the target essential gene would be expected to produce the designed amount of product to carry out its function.
Recently, the booming development of synthetic biology has made genetically engineered budding yeast strains more widely used in large-scale cultivation and open environments. To prevent unconscious diffusion of genetically engineered strains into natural ecosystems or protect the intellectual property, an important strategy is utilizing on-off switch that strictly controls the expression of essential genes 47 . According to our data, we anticipate that P DDI2 can be used as such a switch. In our study, the parental promoter of the essential gene CDC6 was successfully replaced by P DDI2 in the genomic locus of S. cerevisiae. P DDI2 tight regulated CDC6 expression and the yeast cells did not survive without the induction of cyanamide. In the future, we anticipate P DDI2 system could be used as a safeguard switch in the recombinant yeast to achieve the integral biosafety.
We envision that DDI2 promoter is used for fine-tuning the expression of the interested gene in both the experimental studies and industrial applications. The main concern for the widespread use of DDI2 promoter could be the low toxicity of inducer cyanamide. The cyanamide can highly induce the expression of DDI2/3 as MMS which is the DNA-damaging agent. It has been reported that cyanamide has modest toxicity to human 34 , however, no direct evidence has been proved that the toxicity is related to DNA damage. Little is known about the toxicity of cyanamide to yeast, although it has been reported that budding yeast cells lacking the cyanamide hydratase activity displayed an enhanced sensitivity to cyanamide 24 . With the development of chemical synthesis technology, it is expected to design the analogs of cyanamide. Hopefully, the analogous compounds can achieve high-level induction as cyanamide, but with non-toxic to cells.
To obtain the P DDI2 -URA3-P DDI2 fragments, we inserted the upstream copy of P DDI2 into the BamHI and EcoRI sites of pBluescript-URA3 37,48 , then the corresponding downstream P DDI2 fragment was cloned into the HindIII and SalI sites to generate the DUD plasmid (see Supplementary Fig. S1e). To facilitate PCR amplification of the entire DUD cassette, the 3′ end of the upstream promoter and the 5′ end of the downstream copy were truncated to serve as optimized templates.
Strain construction. The yeast strains used in this study are listed in Table 1. Four plasmids with promoter-sfGFP fragments were transformed into yeast strain W303 using a highly efficient LiAc transformation method 49 . The strains were selected on SD-Leu medium (synthetic dextrose medium without leucine) and named yFYV9 (P ADH1 -sfGFP), yFYV10 (P GAL1 -sfGFP), yFYV11 (P DDI2 -sfGFP) and yFYV17 (P CUP1 -sfGFP) ( Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ The reported recyclable cassettes method 37 was used to replace the native promoters. The promoters of TUP1, RAD18, and CDC6 yeast genes were scarlessly replaced by P DDI2 , and the resultant strains were named yFYV12, yFYV13, and yFYV14, respectively (Table 1). DUD recyclable cassettes were PCR amplified using target gene promoter-specific primers linked to the cassette-specific primer sequences 50 and DUD plasmid as template. The primers used in this study are listed in Supplementary Table S1. A two-step method was used to replace the genomic promoter, and the transformed strain was selected on SD-Ura medium. Subsequently, the URA3 marker pop-out strain was achieved by using 5-FOA based method 37 . The resulting strains were confirmed by genomic PCR followed by Sanger sequencing, and the 5 × FLAG tag was added to the C-terminus of the target genes (TUP1, RAD18, and CDC6) for western blot detection. The RAD18 and TUP1 genes were knocked out using the NAT1 gene, respectively, and the resultant strain was named yFYV15 and yFYV16. fluorescence microscopy. yFYV9 (P ADH1 -sfGFP), yFYV10 (P GAL1 -sfGFP), yFYV11 (P DDI2 -sfGFP) and yFYV17 (P CUP1 -sfGFP) cells were cultured in liquid SD-Leu medium overnight at 30 °C. Next morning, 500 µl of the cultures were transferred to tubes containing 5 ml fresh SD-Leu medium (OD600 of 0.4) and incubated for 3 h to an OD600 of 1.0 to 1.2. Next, the yFYV9 (P ADH1 -sfGFP) cells were cultured in SD-Leu medium for 3 h, and the yFYV10 (P GAL1 -sfGFP) cells were starved in raffinose medium for 3 h, then induced in galactose medium for 3 h. The yFYV11 (P DDI2 -sfGFP) cells were cultured in SD-Leu medium with 5 mM cyanamide (C87908-100G, Aldrich) for 3 h and the yFYV17 (P CUP1 -sfGFP)cells were cultured in SD-Leu medium with 0.5 mM CuSO 4 (10008218, ACR) for 3 h. At the end of incubation, 1 OD yeast cells were collected by centrifugation at 2400 g for 2 min, washed once with 1 ml PBS, and resuspended in PBS to final OD600 = 40. Fluorescence microscopy of sfGFP expression was performed using an Olympus FV1200 optical microscope (IX81) with a 60× objective lens. flow cytometry. Cells were incubated and treated as described above. The yFYV9 (P ADH1 -sfGFP) cells were cultured for different time points (2, 3, and 4 h), the yFYV10 (P GAL1 -sfGFP) cells were induced with galactose for different induction time points (2, 3, and 4 h), the yFYV11 (P DDI2 -sfGFP) cells were induced with different concentrations of cyanamide (0-8 mM) for 3-5 h, and the yFYV17 (P CUP1 -sfGFP) cells were induced with 0.5 mM Cu 2+ for different induction time points (2, 3, and 4 h). After cell culture, 1 OD cells were collected and resuspended in 1 ml PBS. The fluorescence expression of sfGFP in each of the strains was measured using a MoFlo XDP flow cytometry (Beckman). For each strain, approximate 50,000 events were collected to measure fluorescence intensity. Summit 5.2 software was used to analyze the data, and mean fluorescence values were calculated in biological triplicates.
Irradiation, clumping assay, and cyanamide induces the expression of the essential gene. Cultures of yFYV12, yFYV15, and BY4741 were grown overnight in YPD liquid medium, then transferred to fresh medium for another 4 h. Next, 0/8 mM cyanamide concentrations were added to the yFYV12 cultures. After 4 h, the yeast cells were harvested and diluted to 10 4 cells/ml. Then, approximately 50 µl of the least diluted cultures were spread on YPD or cyanamide-containing (8 mM) YPD plates in triplicate. The plates were treated with different doses of UV irradiation (0, 10, 20, and 30 J/m 2 ) using a UV cross-linking instrument (Analytik Jena's Model CL-1000), then incubated at 30 °C for at least 4 days.
The yFYV16, BY4741, and yFYV13 cells supplemented with 0 or 5 mM cyanamide were incubated overnight before imaging.
The yFYV14 cells were cultured in liquid YPD with 5 mM cyanamide at 30 °C for 12-16 h. Then, the cells were washed and resuspended in YPD or 5 mM cyanamide-containing YPD medium for 4-6 h. After the cultivation, the yFYV14 cells were harvested and diluted to 10 5 cells/ml. Approximately 10 4 , 10 3 , or 500 cells were spread on YPD plates (with 0 or 5 mM cyanamide) for 3 days.
Total protein was extracted from yFYV12, yFYV13, and yFYV14 cells induced with different cyanamide concentrations for 4 h, separated by SDS-PAGE, then gels were transferred to a polyvinylidene fluoride membrane and analyzed using immunoblotting. The antibodies used in this study were FLAG tag antibody (M20008M, Abmart) and Pgk1 antibody (ab199438, Abcam). Protein quantification was compared under mean gray value and processed by Image J software (1.50i).
Statistical analysis. Data were presented graphically using GraphPad Prism software. Mean values and standard deviations were obtained for data analyzed in the study. The T test was applied to assess whether the among groups was significantly different. The Significance of statistical analysis was established at α = 0.01.

Data Availability
All data generated or analyzed in this study are included in this published article and its Supplementary Information files.