Biofuels cause defects in cell growth of S. cerevisiae
Initially, under conditions of 0.8% (v/v) n-butanol, cell growth at the population level as measured by cell density (OD600) was decreased by 53.9% (Supplementary Fig. 1a). At the single-cell level, the fraction of viable cells (unstained by propidium iodide) was decreased by 44.4% (Supplementary Fig. 1b). Furthermore, it was found that: (1) the yeast cell became progressively longer and adopted a pseudohyphal morphology (Fig. 1a, c, and Supplementary Fig. 2); (2) the cell cycle was delayed at G1 phase (Supplementary Fig. 3). As budding formation is coordinated with cell cycle27, suggesting that the G1 phase delay caused by n-butanol may lead to budding process defects. Based on these, the budding capacity analysis showed that the budding pattern on the single-cell level was damaged (Fig. 1b) and the relative budding index was dropped by 19.8% after treating with 0.8% (v/v) n-butanol (Fig. 1d). The budding process determines the cell growth of S. cerevisiae28,29, indicating that budding index is one of the critical reasons for cell growth in the presence of n-butanol.
a The cell morphology change was detected by microscope and the cells were stained by calcofluor white (CFW). Additionally, field emission scanning electron microscopy (FESEM) was applied to confirm the morphology change under 0.8% (v/v) n-butanol treatment. The white arrows point to the elongated cells. b The time-lapse microscopy of a yeast cell that undergoing the budding process in the control and n-butanol treated group. The yeast cells were acquired at 20 min intervals. Time is indicated in minutes. c Cell length was detected by the microscope, and 120 cells were analyzed both in the control and n-butanol treated group. P values are from a Student’s two-sided t-test of the difference from the untreated control group. Values are shown as mean ± S.D. from 120 (n = 120) cells over three independent biological replicates. d The budding indexes of the control and n-butanol treated group was measured by FACS. The bud neck protein Cdc10 was used as an indicator that visualized the budding process of S. cerevisiae. Here, the fusion protein Cdc10-GFP was introduced into S. cerevisiae, and the GFP signal was just appearing at cells that were under budding process. The high GFP ratio indicates a higher budding index. The green arrows point to the bud necks. Values and error bars represent the mean values and standard deviations of biological repeats. Abbreviations: FACS Fluorescence-activated cell sorting; GFP green fluorescent protein; CFW calcofluor white; FESEM field emission scanning electron microscopy. a, b, d Three experiments (n = 3) were repeated independently with similar results. Source data are provided as a Source Data file.
Next, under growth conditions of 0.2 mM decanoic acid, the cell density (OD600) was decreased by 48.4% (Supplementary Fig. 4a), and the fraction of viable cells was decreased by 60.5% (Supplementary Fig. 4b). Further investigation on the decreased cell growth revealed that the presence of 0.2 mM decanoic acid delayed the formation of the endosome and vacuole, as revealed by an analysis with the lipophilic dye FM4-64 (Fig. 2a). Endocytosis was decreased by 47.3%, as revealed by an analysis using the fluorescent unit Cy5-conjugated α-factor (Fig. 2b). Intracellular pH (pHi) dropped from 7.0 to 5.4 (Fig. 2c, d and Supplementary Fig. 5). Because maintenance of pHi within physiological limits is vital for survival of S. cerevisiae cells, the results suggest that the presence of decanoic acid decreases cell growth via its effects on pHi.
a Internalization of FM4-64 under 0 mM and 0.2 mM decanoic acid treatment. Cells were incubated with FM4-64 on ice, then the cells of different conditions were released and incubated at 30 oC. The fluorescence images were shown for timepoints of 0, 2, 4, 6 min at 30 oC. The red arrows point toward the endosome and the white one toward the vacuole. The red arrows point to the endosome and the white arrows point to the vacuolar. b The internalization of fluorescent unit Cy5 conjugated α-factor in 0 mM and 0.2 mM decanoic acid conditions. The cells were incubated with α-factor at 30 oC and then the ratio endocytosis process was dependent on the internalization of Cy5 conjugated α-factor. The data were analyzed by FACS. Significance (p-value) was evaluated by two-sided t-test. c The comparison of in vivo and in vitro pH values in 0 mM and 0.2 mM decanoic acid treatment. d The fluorescence microscope images of S. cerevisiae expression Sup35-GFP fusion protein (a protein that just aggregate under low pHi) under 0 mM and 0.2 mM decanoic acid treatment. Green arrows point toward Sup35 aggregations. Abbreviation: C10 decanoic acid. All the experiments were performed in three biological repeats. Values and error bars represent the mean values and standard deviations of biological repeats. Source data are provided as a Source Data file.
Engineering actin cytoskeleton to increase cell growth
The actin cytoskeleton, especially actin patches and actin cables, plays a fundamental role in cell division and endocytosis30. Thus, engineering the actin cytoskeleton has potential as a method to enhance the cell growth of S. cerevisiae in the presence of toxic biofuels. Based on our knowledge of the mechanism by which the actin cytoskeleton is formed, 12 genes, encoding components of the polarisome and formin, were selected to tune actin cables30, and 8 genes, encoding adapter proteins of patch components, were the potential targets to regulate actin patches31.
Actin cable tortuosity was used as a parameter to screen for budding process regulators because of its role in stabilizing the cell division process32. As shown in Fig. 3a, compared to the control strain, S. cerevisiae with the cap2, crn1, aip1, gic1, or spa2 gene deleted exhibited an increase in actin cable tortuosity of 7.1%, 10.7%, 11.1%, 6.7%, and −12.5%, respectively. Overexpression of tpm1, pfy1, bud6, bud14, cdc42, pea2, and bin1 resulted in increases in actin cable tortuosity of 7.7%, 14.8%, −5.7%, 6.1%, −7.0%, −1.3%, and 14.0%, respectively (Fig. 3a and Supplementary Fig. 6). Because reduced tortuosity resulted in reduced cell size and stabilized the budding process33, spa2, cdc42, and bud6 were selected to investigate the effect on the budding index (Supplementary Figs. 7–8), which was evaluated using two different approaches. First, the yeast cells were released from G1 phase arrest, and the cells were then plated on yeast nitrogen base (YNB) medium containing 0.8% (v/v) n-butanol. The number of colony-forming units (CFUs) was used to indicate the relative budding index. Compared with the control strain, the spa2 deletion strain, cdc42 overexpression strain, and bud6 overexpression strain exhibited an increase in CFUs by 18.4%, 15.4%, and 4.1%, respectively (Fig. 3b). Second, Cdc10-GFP (a fusion between a component of the septin ring and green fluorescent protein) was used as a marker to visualize the budding index of yeast cells. As shown in Fig. 3c, the relative budding indices of Cdc10-GFP in the SPA2 deletion strain and the cdc42 overexpression strain were 14.0% and 15.3% higher than that of the control strain in YNB medium with 0.8% (v/v) n-butanol, respectively. The bud6 overexpression strain showed a 6.1% decrease in budding index under the same conditions. Therefore, spa2 and cdc42 were selected as targets to improve the budding index. Furthermore, simultaneous deletion of spa2 and overexpression of cdc42 in strains W303-1A, W303-1B, BY4741, BY4742, Cen.PK2-1C, and thTAM in YNB medium with 0.8% (v/v) n-butanol was found to increase the OD600 by 88.3%, 79.6%, 53.9%, 104.0%, 106.7%, and 108.6%, respectively (Supplementary Fig. 9a), and increase the budding index by 18.9%, 21.3%, 19.0%, 16.5%, 14.9%, and 19.6%, respectively (Supplementary Fig. 9b).
a The tortuosity is defined as the ratio of the cable length (l) to the distance (d) between its two endpoints. A total of 12 candidates were observed by fluorescence microscope, and then the tortuosity was analyzed by Image J. Statistical significance were analyzed between control group and different actin cable engineered strains. b The stabilization effect of cdc42, spa2, and bud6 on the budding index of S. cerevisiae under 0.8% (v/v) n-butanol treatment. The budding index was determined by colony-forming unit (CFU). c The Cdc10-GFP was applied as a budding indicator to detect the budding process of engineered strains under n-butanol treatment. The GFP signal was just appearing at cells that were under budding process. The high GFP ratio indicates a higher budding index. The green arrows point to the bud necks. Statistical significance were analyzed between control group under 0.8% (v/v) n-butanol treatment and corresponding engineered strains with spa2Δ, cdc42, and bud6 overexpression. d The fluorescence microscope images of actin patch. The density of the patch determined the endocytosis process of S. cerevisiae. The effect of improving the density of the patch was analyzed by fluorescence microscope and the GFP dot indicated the actin patch. The patch was characterized by fusing the GFP with the genomic Abp1. The box plots showing the minima, maxima, center, bounds of box and whiskers and percentile. Statistical significance were analyzed between control group and different actin patch engineered strains. e The effect of relieving the decline of intracellular pH. Statistical significance were analyzed between control group under 0.2 mM decanoic acid treatment and corresponding engineered strains with clc1Δ, clc1Δsla2OE, and sla2 overexpression. f The fluorescence microscope images of S. cerevisiae expression Sup35-GFP fusion protein under 0.2 mM decanoic acid treatment in strain clc1Δ, sla2 overexpression strain, and syp1 overexpression strain. Green arrows point toward Sup35 aggregations. All experiments were performed in three biological repeats. Values and error bars represent the mean values and standard deviations of biological repeats. a, c–e, Significance (p-value) was evaluated by two-sided t-test. Source data are provided as a Source Data file.
The density of actin patches impacts endocytosis, which is important for maintaining the function of the vacuole30. Therefore, actin patch density was used to screen for pHi stabilizers. Eight potential genes were divided into two groups: the first group contained five overexpressed genes (pal1, syp1, sla2, ent1, and ent2), and the second group included three deleted genes (sla1, clc1, and ede1). As shown in Fig. 3d, overexpression of pal1, syp1, sla2, and ent2 led to increases in actin patch density of 57.1%, 110.7%, 152.9%, and 18.8%, respectively, whereas overexpression of ent1 led to a decrease of 8.1%. On the contrary, the deletion of sla1 and clc1 resulted in increases in actin patch density of 57.1% and 129.9%, respectively, whereas deletion of ede1 resulted in a decrease of 4.3% (Fig. 3d, Supplementary Fig. 10). Based on these data, clc1, sla2, and syp1 were selected to investigate their effects on pHi. The pHi values of the clc1δ, sla2 overexpression, and syp1 overexpression strains in the YNB medium containing 0.2 mM decanoic acid were 6.2, 6.0, and 5.5, respectively (Fig. 3e). This result was confirmed by Sup35-GFP aggregation (a translation termination factor and GFP fusion that forms aggregates under conditions of low intracellular pH34). The aggregation was found only in the syp1-overexpressing strain, in which pHi was decreased, but not the clc1Δ strain or the sla2-overexpressing strain (Fig. 3f). Therefore, clc1 and sla2 are identified as targets for stabilization of pHi. When sla2 was overexpressed along with clc1Δ in strains W303-1A, W303-1B, BY4741, BY4742, Cen.PK2-1C, and thTAM grown in YNB medium with 0.2 mM decanoic acid, the OD600 increased by 60.3%, 54.7%, 57.8%, 41.3%, 75.8%, and 59.1%, respectively, relative to control strains (Supplementary Fig. 11a), and the pHi was stabilized at 6.3, 6.1, 6.2, 6.2, 6.1, and 6.2, respectively (Supplementary Fig. 11b). However, in the YNB medium without n-butanol or decanoic acid, deleting and overexpressing genes related to actin cables (spa2Δcdc42OE) and patches (clc1Δsla2OE) resulted in 13.1% and 12.7% decreases in the OD600 values compared with the control strain (Supplementary Fig. 12). These data indicate that the static engineering strategies are not an optimal tool to tune the expression of the actin cytoskeleton.
Construction of an autonomous bidirectional signal conditioner for manipulating the actin cytoskeleton
To achieve autonomous, temporal, and dual control of the actin cytoskeleton when biofuels accumulate, an autonomous bidirectional signal conditioner (ABSC) containing an activation unit, a repression unit, and a biofuel signal converter was constructed (Fig. 4a, b and Supplementary Fig. 13).
a, b The schematic diagram of the autonomous bidirectional signal conditioner (ABSC). c The activation unit of ABSC in biofuels conditions was characterized by GFP. d The repression unit of ABSC in biofuels conditions was characterized by mKate2. c, d Values and error bars represent the mean values and standard deviations of three biological repeats, respectively. Source data are provided as a Source Data file.
The activation unit consisted of a synthetic transcriptional activator (LexA-Ste12PRD), a hybrid LexO-cup1 promoter comprising a LexA operator (LexO), a cup1 core promoter sequence, and a GFP reporter gene. LexA-Ste12PRD can be recruited to LexO-cup1 to allow for transcriptional activation of the GFP reporter. In the absence of α-factor, the yeast mating pheromone response pathway is inactivated, which prevents the activation of LexA-Ste12PRD and the expression of GFP. Conversely, in the presence of the α-factor, LexA-Ste12PRD is activated, which induces the cup1 promoter and finally activates GFP expression (Fig. 4a). The repression unit was constructed based on the activation unit, the yeast galactose (GAL) regulon composed of transcriptional regulators (Gal80 and Gal4), and the gal1 promoter was introduced as a “NOT” gate that can turn off gene expression when required. Here, mKate2 is the reporter gene, and Gal80 is expressed under the control of the LexO-cup1 promoter, which is activated by LexA-Ste12PRD. In the absence of α-factor, Gal80 is repressed, so Gal4 can activate mKate2 expression from the gal1 promoter. Conversely, in the presence of α-factor, Gal80 is expressed, binds to Gal4, and prevents mKate2 expression from the gal1 promoter (Fig. 4a).
Furthermore, to regulate the activation and repression units simultaneously and autonomously, a biofuel signal converter was constructed. At first, available literature about growth in the presence of n-butanol and MCFAs was mined for transcriptome data35,36. From this analysis, potential biofuels responsive promoters were screened. In n-butanol growth conditions, the 15 most differentially expressed genes identified in the transcriptome data were selected, and their promoters were identified as potential n-butanol responsive promoters (Supplementary Fig. 14a). Similarly, 15 potential MCFA responsive promoters were identified from relevant transcriptome data (Supplementary Fig. 14a). Thereafter, to obtain sensors that were tunable, stable, and had a low background of activity, the potential responsive promoters were characterized in a series of different levels of n-butanol or decanoic acid, as appropriate. As a result, nrp1 and pdr12 were selected as the n-butanol- and MCFA responsive promoters (Supplementary Fig. 14b), respectively, which showed good specificity for the corresponding biofuel (Supplementary Fig. 14c). Subsequently, the n-butanol-responsive promoter nrp1 was used to regulate the expression of mfα2, the gene encoding α-factor (Fig. 4b and Supplementary Fig. 15). In the presence of n-butanol, mfa2 is expressed and α-factor is generated, inducing the regulation of the activation and repression units, as described above. Similarly, the MCFA responsive promoter pdr12 was then used to induce the generation of α-factor and autonomously drive the activation and repression units (Fig. 4b and Supplementary Fig. 15). Based on these results, the corresponding biofuel signal converter, activation unit, and repression unit could thus be assembled to create ABSC-butanol and ABSC-MCFA (Fig. 4b), respectively.
The performance of ABSC-butanol and ABSC-MCFA was evaluated in a series of experiments. With ABSC-butanol, the addition of 0.8% (v/v) n-butanol triggered a 6.2-fold increase in GFP fluorescence intensity driven by the activation unit and a 4.8-fold decrease in mKate2 fluorescence intensity driven by the repression unit compared to the control (Fig. 4c, d). The GFP and mKate2 fluorescence intensities did not change in strains devoid of the nrp1-LexO or nrp1-GAL regulons, respectively. In ABSC-MCFA, the addition of 0.2 mM decanoic acid led to a 7.6-fold increase in GFP fluorescence intensity driven by the activation unit and a 6.1-fold decrease in mKate2 fluorescence intensity driven by the repression unit compared to the control (Fig. 4c, d). The GFP and mKate2 fluorescence intensities did not change in strains devoid of the pdr12-LexO or pdr12-GAL regulons, respectively. However, this narrow dynamic range was deemed insufficient to regulate the expression of the actin cytoskeleton. Therefore, to optimize the activation unit, the number of lexO-binding sites (1× to 8× LexO) and the core promoters of LexO-cup1 were changed. As a result, GFP fluorescence intensity increased 12.7-fold and 16.3-fold following treatment with 0.8% (v/v) n-butanol and 0.2 mM decanoic acid, respectively (Supplementary Fig. 16a). In addition, optimization of the number of LexO-binding sites and the constitutive promoter of Gal4 allowed the mKate2 fluorescence intensity to decrease by 12.0-fold and 16.8-fold following treatment with 0.8% (v/v) n-butanol and 0.2 mM decanoic acid, respectively (Supplementary Fig. 16b). Moreover, functioning of the ABSC system was confirmed at the 5-L bioreactor level (Supplementary Fig. 17). A computational model was constructed to better understand the behavior of ABSC-butanol and ABSC-MCFA, respectively (Supplementary Fig. 18).
To confirm that ABSC could efficiently regulate the expression of endogenous genes, ABSC-butanol and ABSC-MCFA were introduced into strain BY4741 to generate strains LH001-B and LH001-M, respectively, and the fluorescent proteins were replaced by the corresponding genes encoding components of the actin cytoskeleton. Following 0.8% (v/v) n-butanol treatment, compared to the control strain, the budding index and cell density (OD600) in strain LH001-B were increased by 1.2- and 1.7-fold, respectively (Supplementary Fig. 19a). Additionally, following 0.2 mM decanoic acid treatment, the pHi of strain LH001-M was maintained at 6.1 and the cell density (OD600) was 56.2% higher than that of the control strain (Supplementary Fig. 19b). Furthermore, the universality of ABSC to regulate the expression of the actin cytoskeleton was confirmed in S. cerevisiae strains, including W303-1A and Cen.PK2-1C (Supplementary Fig. 20). Overall, these results suggest that ABSC has broad applicability for tuning the actin cytoskeleton.
n-Butanol production is enhanced by improving the cell growth
Saccharomyces cerevisiae LH002-B1 was constructed by inserting genes encoding enzymes related to the n-butanol production pathway from Clostridium acetobutylicum (Fig. 5a). Strain LH002-B1 produced 634.3 mg L−1 n-butanol (Fig. 5b). Compared with the corresponding values in the control strain LH002-B0, the accumulation of n-butanol in strain LH002-B1 led to (1) elongated cells (Figs. 5c), (2) a 23.5% decrease in the budding index (Figs. 5d), (3) a 44.1% lower of cell density (OD600) (Fig. 5e), and (4) a 28.9% decrease in the fraction of viable cells of strain LH002-B1 (Supplementary Fig. 21b). These results indicated that a decreased budding index causes a lower cell growth, thus leading to cell growth defects and potentially limiting the potential for n-butanol production. Therefore, to increase the cell growth, the n-butanol fermentation process was divided into two phases by introducing ABSC-butanol to improve the budding index (Supplementary Fig. 22) as follows: (1) the n-butanol accumulation phase, in which the expression of spa2 is activated by the gal1 promoter and the expression of cdc42 is repressed; (2) the repair phase, in which ABSC-butanol is self-induced by n-butanol, leading to the repression of spa2 expression by Gal80 and the activation of cdc42 expression by the tdh3 promoter.
a The schematic diagram of the n-butanol pathway. The red cross symbols indicate that the corresponding gene was knocked out and the red arrows symbols indicate the corresponding gene was overexpressed. But-P means n-butanol pathway. b n-Butanol production of different engineered strains in shake flask culture. The histogram graphs were the titer of different engineered strains, whereas the triangle symbols were the yield of the corresponding engineered strain. c The population distribution of normal yeast morphology in LH002-B0, LH002-B1, and LH002-B5. The morphology was characterized by SSC v.s. FSC density plot by cell cytometry and 20,000 cells were collected. Each dot or point on the plot represents an individual cell that has passed through the laser. LH002-B0 strain was the control strain without the n-butanol pathway. LH002-B1 strain was only harbored the n-butanol pathway, whereas the LH002-B5 strain was harbored the n-butanol pathway and ABSC-butanol with actin cable genes. d The relative cell budding index were analyzed in LH002-B0, LH002-B1, and LH002-B5 strain, respectively. e The cell density was characterized by OD600. Additionally, methylene blue was applied to stain the dead cells. Three experiments (n = 3) were repeated independently with similar results. b, e Values and error bars represent the mean values and standard deviations of three biological repeats, respectively. Source data are provided as a Source Data file.
To autonomously and temporally tune these two phases, ABSC-butanol was introduced into strain LH002-B1, and the DNA sequences that encoded GFP and mKate2 proteins were replaced by cdc42 and spa2, respectively. As shown in Fig. 5b, these manipulations generated a series of strains, namely LH002-B2, LH002-B3, LH002-B4, and LH002-B5. In strain LH002-B5, the expression of spa2 was activated during the first 50 h of n-butanol fermentation but was repressed thereafter, whereas cdc42 was increasingly expressed between 50 and 120 h (Supplementary Fig. 23). Based on these manipulations, it was found that: (1) the population distribution of normal yeast morphology was 81.5%, which was 32.8% higher than that of LH002-B1 and 5.2% lower than LH002-B0 (Fig. 5c and Supplementary Fig. 24); (2) the budding index was 14.0% higher than that of strain LH002-B1 and reached 90.6% higher than that of LH002-B0 (Fig. 5d); and (3) the cell density (OD600) and the fraction of viable cells was 59.6% and 26.8% higher than that of strain LH002-B1, respectively (Fig. 5e and Supplementary Fig. 21). As a result, the titer, yield, and titer per dry cell weight (DCW) of n-butanol achieved by strain LH002-B5 were 1674.3 mg L−1,120.3 mg g−1 glucose, and 1511.6 mg g−1, respectively (Fig. 5b); these values were substantially higher than those of other strains (titer, yield, and titer per DCW): LH002-B1 (164.0%, 118.2%, and 92.9%, respectively), LH002-B2 (110.0%, 87.0%, and 65.6%, respectively), LH002-B3 (108.3%, 78.8%, and 64.3%, respectively), and LH002-B4 (46.9%, 21.0%, and 21.4%, respectively) (Fig. 5b and Table 1).
MCFAs production is enhanced by improving the cell growth
Saccharomyces cerevisiae LH003-M1 was constructed by reinforcing the endogenous fatty acid pathway and engineering fatty acid synthases (Fig. 6a). Strain LH003-M1 could produce 446.6 mg L−1 MCFAs (Fig. 6b). The accumulation of MCFAs in strain LH003-M1 led to a decline in pHi from 7.1 to 5.4, which was 23.9% lower than that of the strain LH003-M0 (Fig. 6c, d) and a 43.0% lower cell density (OD600) and a 42.4% decrease in the fraction of viable cells, compared to the corresponding values of the control strain LH003-M0 (Fig. 6e and Supplementary Fig. 25b). These results demonstrate that a lower pHi leads to cell growth defect and ultimately affects MCFAs production in strain LH003-M1. To solve these limitations, the MCFAs fermentation process was divided into two phases by introducing the ABSC-MCFA to stabilize the pHi (Supplementary Fig. 26) as follows: (1) the MCFAs accumulation phase, in which the ABSC-MCFA is inactivated and the genes that encode the actin patch are not manipulated; (2) the repair phase in which the ABSC-MCFA is self-induced by the accumulation of MCFAs, leading to the repression of clc1 expression by Gal80 and the activation of sla2 expression by the tdh3 promoter.
a The schematic diagram of medium-chain fatty acid production. The red arrows indicate the enhancement of carbon flux into medium-chain fatty acids. The thioesterase AcTesA was inserted into fatty acid synthase, and the location of AcTesA was adjacent to the acyl carrier protein. Meanwhile, two mutations (G1250S and M1251W) were performed in Fas2. MCFA-P means MCFAs pathway. b MCFA production of different engineered strains in shake flask culture. The histogram graphs were the titer (C6-C12) of different engineered strains, whereas the round symbols were the yield of the corresponding engineered strains. c The intracellular pH of strain LH003-M0, LH003-M1, and LH003-M5 during the MCFAs production. LH003-M0 strain was the control strain without the MCFAs pathway and two-phase damage buffer. LH003-M1 strain was only harbored the MCFAs pathway, whereas the LH003-M5 strain was both harbored the MCFAs pathway and the autonomous two-phase damage buffer. d The microscopy images of strain LH003-M0, LH003-M1, and LH003-M5 during the MCFAs production. e The cell density (OD600) was analyzed in LH003-M0, LH003-M1, and LH003-M5 strain, respectively. Additionally, methylene blue was applied to stain the dead cells. b, c, e Values and error bars represent the mean values and standard deviations of three biological repeats, respectively. d, e Three experiments (n = 3) were repeated independently with similar results. Source data are provided as a Source Data file.
To autonomously and temporally tune these two phases, the ABSC-MCFA was introduced into strain LH003-M1, and the DNA sequences that encoded GFP and mKate2 proteins were replaced by sla2 and clc1, respectively. As shown in Fig. 6b, these manipulations generated a series of strains, namely, LH003-M2, LH003-M3, LH003-M4, and LH003-M5. In strain LH003-M5, the expression of clc1 under the gal1 promoter occurred during the first 24 h, whereas the expression of sla2 was activated between 24 and 60 h (Supplementary Fig. 27). Based on these manipulations, the following phenomena were observed: (1) the pHi was restored to 6.4, which was 18.5% higher than that of strain LH003-M1 and reached 90.1% of the value in LH003-M0 (Fig. 6c, d), and (2) the cell density (OD600) and the fraction of viable cells of strain LH003-M5 were 36.0% and 43.8% higher compared to those of strains LH003-M1 and LH003-M0, respectively (Fig. 6e and Supplementary Fig. 25). Consequently, the titer, yield, and titer per DCW of MCFAs produced by strain LH003-M5 were 692.3 mg L−1, and 19.3 mg g−1 glucose, and 280.2 mg g−1, respectively (Fig. 6b). These values were substantially higher compared to those of other strains (titer, yield, and titer per DCW): LH002-M1 (55.1%, 37.3%, and 12.3%, respectively), LH002-M2 (28.1%, 18.7%, and 2.7%, respectively), LH002-M3 (27.2%, 21.5%, and 4.2%, respectively), and LH002-M4 (19.4%, 12.0%, and 1.9%, respectively) (Fig. 6b and Table 2).
