Enhancing Bioethanol Production by Deleting Phosphoenolpyruvate Synthase and ADP-Glucose Pyrophosphorylase, and Shunting Tricarboxylic Acid Cycle In Synechocystis Sp. PCC 6803


 Background: The outstanding ability of directly assimilating carbon dioxide and sunlight to produce biofuels and chemicals impels photosynthetic cyanobacteria to become attractive organisms for the solution to the global warming crises and the world energy growth. The cyanobacteria-based method for ethanol production has been increasingly regarded as alternatives to food biomass-based fermentation and traditional petroleum-based production. Therefore, we engineered the model cyanobacterium Synechocystis sp. PCC 6803 to synthesize ethanol and optimized the biosynthetic pathways for improving ethanol production under photoautotrophic conditions.Results: In this study, we successfully achieved the photosynthetic production of ethanol from atmospheric carbon dioxide by an engineered mutant Synechocystis sp. PCC 6803 with over-expressing the heterologous genes encoding Zymomonas mobilis pyruvate decarboxylase (PDC) and Escherichia coli NADPH-dependent alcohol dehydrogenase (YqhD). The engineered strain was further optimized by an alternative engineering approach to improve cell growth, and increase the intracellular supply of the precursor pyruvate for ethanol production under photoautotrophic conditions. This approach includes blocking phosphoenolpyruvate synthetic pathway from pyruvate, removing glycogen storage, and shunting carbon metabolic flux of tricarboxylic acid cycle. Through redirecting and optimizing the metabolic carbon flux of Synechocystis, a high ethanol-producing efficiency was achieved (248 mg L-1 day-1) under photoautotrophic conditions with atmospheric CO2 as the sole carbon source. Conclusions: The engineered strain SYN009 (∆slr0301/pdc-yqhD, ∆slr1176/maeB) would become a valuable biosystem for photosynthetic production of ethanol and for expanding our knowledge of exploiting cyanobacteria to produce value chemicals directly from atmospheric CO2.


Background
Due to the high ability of directly integrating atmospheric CO 2 with sunlight into biomass, cyanobacteria have been increasingly proposed as one of the most promising biosystems for the solution to the global warming crises and the world energy growth [1]. Cyanobacteria are autotrophic prokaryotes that perform similar photosynthesis as higher plants [2]. However, compared to the traditional plants, cyanobacteria possess several advantages, including the simple inorganic nutrient requirement, the tolerant growth on non-arable land, and powerful genetic manipulation [3], which render photosynthetic cyanobacteria as attractive organisms for the direct production of important chemicals from atmospheric CO 2 through metabolic engineering methods [4]. Over the years, an enormous amount of the engineered cyanobacteria has been successfully developed to produce various industrial relevant chemicals, such as ethylene [5], isoprene [6], ethanol [7], isobutanol [8], acetone [9], and p-coumaric acid [10].
Ethanol is one of the major renewable biofuels that are a worldwide focus of main concern. Commercially, ethanol production is mostly based on the fermentation of starch or agricultural crops as feedstock [7]. The over-exploitation of this food-based raw material presents a signi cant bottleneck for expanding ethanol production. To avoid competition with the world food supply or agricultural land, the cyanobacteria-based method for ethanol production has been increasingly regarded as alternatives to biomass-based fermentation. In recent years, a lot of efforts have been made to realize ethanol production by engineered cyanobacteria. The rst production of ethanol was successfully achieved in Synechococcus elongatus PCC 7942 by expressing pyruvate decarboxylase (PDC) and alcohol dehydrogenase II (ADH II) from Z. mobilis [11]. Soon afterward, Synechocystis sp. PCC 6803 (hereafter Synechocystis 6803) was genetically engineered by expressing these two enzymes to produce ethanol, and obtained double the ethanol yield compared to S. elongatus PCC7942 strain [12]. To further improve ethanol production, many desirable strategies were adopted to optimize abiotic and biotic factors that have effects on cyanobacterial cell growth and metabolisms. For example, overexpressing the ethanolproducing steps and blocking the production of storage polymers (glycogen and PHB) were performed in Synechocystis 6803 to increase ethanol production [13]. Recently, another example has been signi cantly shown to improve the production of ethanol by engineered cyanobacteria with enhanced cell growth through overexpressing the Calvin-Benson-Bassham (CBB) cycle enzymes [14]. However, genetically engineered cyanobacteria-based biosystems still faced many challenges for ethanol production applications, such as redirecting carbon ux to the desired product instead of cellular biomass, promoting cell growth, and balancing the co-factor levels due to metabolic consumption of an introduced biosynthetic pathway. To enhance the ethanol production by cyanobacteria, more e cient and rational biosynthetic pathways need to be established through metabolic engineering.
In this study, considerable attention is paid to achieve the ethanol production in the model cyanobacterium Synechocystis 6803 by genetically integrating the genes encoding Z. mobilis PDC and E. coli YqhD into the Synechocystis 6803 chromosome. To investigate the potential of optimizing metabolic pathways for improving ethanol productivity, the engineered cyanobacterial strains were genetically modi ed in a stepwise approach via inhibiting phosphoenolpyruvate pathway from pyruvate, removing glycogen storage, and shunting carbon metabolic ux of the tricarboxylic acid cycle. These approaches led to the high-e cient ethanol production directly from solar energy and atmospheric CO 2 under photoautotrophic conditions, and signi cantly contribute to enhancing the biological synthesis of the desired carbon-based biofuels.

Results
Determining the metabolic pathways for the ethanol production To achieve photosynthetic production of ethanol in Synechocystis 6803, it is essential to rationally construct and optimize an exogenous biosynthetic pathway for redirecting carbon uxes towards ethanol. Thus, the precursor intracellular level and the catalytic pathway e ciency are generally considered as two regulating factors for high-e cient production of ethanol. In cyanobacteria, the pyruvate and acetyl-CoA are two important intermediate metabolites, which are naturally accumulated in cells, and successfully exploited by several studies as the starting precursors to be converted into ethanol and other products. To evaluate which intermediate precursor is more e cient for ethanol production, we measured the intracellular content of the primary metabolites in the wild-type Synechocystis 6803 strain.
As shown in Fig. S1, the intracellular concentration of pyruvate reached up to an average of 1.05 µmol g -1 (dry cell weight), which was approximately 3 times that of acetyl-CoA. This result indicated that Synechocystis 6803 possesses a high potential ability to produce ethanol by using pyruvate as the starting precursor rather than acetyl-CoA.
The major pathway of ethanol production was constructed by expressing two enzymes, Z. mobilis PDC and E. coli YqhD. As shown in Fig. 1, the pyruvate was directly converted to acetaldehyde and CO 2 by PDC, and subsequently, acetaldehyde was reduced by YqhD to generate ethanol and NADP + . The Cu 2+ inducible promoter PpetE was initially selected to drive the expression of the exogenous genes. The promoter PpetE and the codon-optimized genes were integrated into the neutral site (slr0168) of Synechocystis 6803 genome to generate the initial strain SYN001 (∆slr0168/pdc-yqhD, PpetE) ( Fig. 2A,  Fig.2B). The resulting transformants were maintained in liquid BG-11 medium with the addition of 50mg/L spectinomycin. HPLC analysis showed that ethanol accumulated up to a yield of 230 mg L -1 (OD 730 ≈0.64) in the strain SYN001 after 7 days of photoautotrophic growth (Fig. 3A, Fig. 3B). According to the published studies [15,16], the strength of the promoter PpetE was categorized as "medium", which led to relatively low expression levels of the targeted genes. This enabled us to hypothesize that the mRNA expression levels of the pdc and yqhD gene were potentially correlated with ethanol production yield. To overcome the potential effects of possible low expression of the ethanol biosynthesis genes, we substituted the 'medium' promoter PpetE with the light-sensitive and strong promoter PpsbA2s to construct the second producer SYN002 (∆slr0168/pdc-yqhD, PpsbA2s) ( Fig. 2A, Fig.2B). After 7 days of cultivation, HPLC analysis demonstrated that 474 mg L -1 (OD 730 ≈0.64) of ethanol was accumulated in the culture medium, which was 2-fold higher than that of strain SYN001. These results demonstrated that enhancing the gene expression was required in Synechocystis for the high-e cient production of ethanol. This is also consistent with previous studies where the relatively high expression levels of biosynthetic genes were required to improve the production of the desired chemicals.
Eliminating phosphoenolpyruvate pathway from pyruvate Due to continuous production and consumption by several metabolic pathways, the intermediate pyruvate undergoes dynamic uctuation at the intracellular level. This rendered us attempt to lessen the competitive consumption of the precursor pyruvate by inhibiting the catalytic activity of phosphoenolpyruvate synthase encoded by the slr0301 gene of Synechocystis 6803 genome. Theoretically, disruption of the slr0301 gene would metabolically block the catalytic conversion of the pyruvate to phosphoenolpyruvate, and then could increase the endogenous carbon ux from pyruvate to ethanol. To test if the catalytic inhibition of phosphoenolpyruvate (PEP) synthase activity could contribute to ethanol formation, the engineered strain SYN003 (∆slr0301/pdc-yqhD) was constructed by transforming the plasmid pBE03 inculcating the slr0301 knockout cassette into the wide-type Synechocystis strain. The ethanol-producing assay was performed to compare the yield of the newly engineered strain SYN003 with the strain SYN001 and SYN002. The ethanol yield was improved to 600 mg L -1 (OD 730 ≈0.64) after 7 days of photoautotrophic growth. The yield of the strain SYN003 was approximately 2.6-fold and 1.3-fold higher than that of SYN001 and SYN002, respectively. This result gave an implication that catalytic inactivation of phosphoenolpyruvate synthase led to the level increase of the precursor pyruvate, and then boosted the ethanol formation. To determine whether disruption of the slr0301 gene has negative effects on the photoautotrophic growth of the engineered Synechocystis strain SYN003, the time-courses of OD 730 were performed to compare the growth pattern of the strain SYN003 and SYN004 (∆slr0301) with the wild-type strains. Obviously, no differences were observed between the engineered strains and the wild-type strain (Fig. 4), indicating that the deletion of the slr0301 gene has no signi cant effects on the physiological activities of Synechocystis 6803. Based on these results, rational optimization of the pyruvate consumption pathway might provide a useful strategy that can contribute to effectively increase the ethanol biosynthesis in the engineered strain of Synechocystis 6803.

Blocking glycogen synthesis pathway
To enhance photosynthetic carbon ux towards chemical production by cyanobacteria, removing the glycogen biosynthetic pathway would become a helpful strategy for the generation of biofuels and chemicals. In this study, to test if inactivation of glycogen synthesis contributes to the ethanol production in the phosphoenolpyruvate synthase-de cient strain (∆slr0301/pdc-yqhD), we also performed complete inhibition of the AGPase activity by disrupting the slr1176 gene of Synechocystis 6803 genome. The engineered strain SYN007 (∆slr0301/pdc-yqhD, ∆slr1176) was constructed by transforming the plasmid pMD-slr1176-Ω inculcating the slr1176 gene knockout cassette into the strain SYN003. The ethanol yield in this double-knockout strain reached up to 689 mg L -1 (OD 730 ≈0.57) under the photoautotrophic conditions. The strain SYN007 showed about 1.5-fold ethanol yield compared with the starting strain SYN002. Similar to the strain SYN003 and SYN004, no remarkable differences were shown between the engineered SYN007 and the wild-type strain (Fig. 4). These data con rmed that complete inactivation of glycogen synthesis can be bene cial for the direct conversion of the xed carbon in photosynthesis to the desired ethanol. This is also consistent with the previous studies that blocking the glycogen synthesis pathway can shift excess carbon from the glycolytic pathway and pentose phosphate pathway to a favorable biosynthetic pathway.

Metabolic interventions of the TCA cycle
It has been reported that the major portion of the xed carbon in Synechocystis 6803 was metabolically consumed by the aerobic TCA cycle, which is one of the central carbon metabolisms in cyanobacteria [17]. Those carbon reserve metabolites produced from the TCA cycle are of great interest because they are e cient substrates for the production of ethanol or other biofuels. Thus, in the next optimization step, we attempted to improve ethanol production by shunting the metabolic route of Synechocystis TCA cycle metabolism. The metabolic intervention of the TCA cycle was conducted by over-expressing NADPdependent malic enzyme (maeB) from E. coli. The maeB gene was introduced into the slr1176 site of the strain SYN003, establishing SYN009 (∆slr0301/pdc-yqhD, ∆slr1176/maeB). After 7 days of photoautotrophic growth, the strain SYN009 was able to generate ethanol, which the production yield reached up to 1.09 g L -1 (OD 730 ≈0.7) (Fig. 3B), and showed more than 1.6-fold improved ethanol yield compared with the strain SYN007. To our surprise, under photoautotrophic conditions, the strain SYN009 displayed normal growth and propagated steadily over culture time, and showed 30% higher cell density than the wide-type strains (Fig. 5A).
To indicate the hypothesis that the engineered strains produce higher yield of ethanol over long-term growth, the strain SYN009 was cultivated for up to 14 days. As expected, the ethanol accumulation was increased by 30% and improved to 1.3 g L -1 (OD 730 ≈1.37) at 14 days of photoautotrophic culture.
Interestingly, the ethanol yield in strain SYN009 was increased dramatically during the growth stage of days 0-4. During each of these days, the ethanol productivity was greater than 200 mg L -1 day -1 (Supplementary materials Fig. S2). The highest ethanol productivity reached 248 mg L -1 day -1 (OD 730 ≈0.70) at the 4th day. After day 4, the ethanol yield showed a slight increase, but the ethanol productivity signi cantly decreased to less than 100 mg/L/day. This may be that the insu cient supply of CO 2 and light limits ethanol production of the strain SYN009 under photoautotrophic conditions. Taken together, these data indicated that a basic photo-biosystem for ethanol production was developed in our experiment through metabolic engineering optimization, which contributes to redirecting the metabolic carbon ux towards ethanol production.

Discussion
It has been reported that the metabolic imbalance of endogenous metabolism and the biosynthetic pathway often limits the desired chemical productivity and yield by microbial systems [18]. To enhance the carbon ux towards ethanol production, it is necessary to counterbalance the contradictory relationship between endogenous metabolism and the biosynthetic pathways. Therefore, we focused on the genetically engineering of a cyanobacterial strain for enhancing the capacity of metabolic ux toward the pyruvate, an important intermediate metabolite for ethanol production. For this purpose, we developed several modi ed metabolic bypasses to boost the intracellular supply of pyruvate in the engineered Synechocysits harboring the ethanol-producing pathway. Our integrative approach led to successful improvement of the ethanol yield and productivity via stepwise metabolic engineering, e.g., disruption of consuming pyruvate pathways and reorientation of the carbon ux of the TCA cycle towards pyruvate.
In the glycolytic pathway, phosphoenolpyruvate (PEP) is catalyzed by pyruvate kinase into pyruvate, and reversibly, the pyruvate is converted into PEP through the activity of PEP synthase (PpsA). This means that PpsA could be an attractive target for pyruvate forming metabolic bypass from PEP. In Synechocystis 6803, PpsA is encoded by the slr0301gene. To the best of our knowledge, no experimental evidence has to date been reported on the potential effects of the disruption of Synechocystis PpsA on the ethanol production. To test if this could be the case in Synechocystis, we constructed the strain SYN003 by deleting the PEP synthase coding locus: slr0301. As expected, with the inhibition of PEP synthase activity, the ethanol production yield in the strain SYN003 showed signi cant increase when compared to the strain SYN001 and SYN002. Moreover, the cell growth of the PEP synthase-de cient Synechocystis is normal and similar to that of the wide-type strains. These results suggested that the optimized metabolic bypass can regulate the strain SYN003 intracellular metabolism to improve the pyruvate supply for ethanol production. It can be suggested that deletion of the PEP synthase contributed to the yield of ethanol production by repression of PEP synthesis.
To further improve the ethanol yield in our study, an alternative strategy is to perform the complete inhibition of glycogen synthesis pathway. Glycogen is generated from CBB cycle, and considered as one of major storage components for carbon resources in cyanobacteria. Signi cantly, if Synechocystis 6803 cells lack the ability of glycogen synthesis, they will show an over ow of carbon metabolism leading to the excretion of pyruvate [19]. This effect may be hijacked for product formation by introducing a pyruvate-utilizing reaction such as ethanol production [20]. Several studies have been conducted to increase the ethanol production in cyanobacteria by deleting the glycogen synthesis pathway [13,21]. To check if this could also be the case for Synechocystis, we optimized the strain SYN003 by knocking out the glgC: slr1176 gene encoding AGPase to obtain the glycogen-de cient strain. As expected, the strain SYN007 produced more ethanol than the strain SYN003. The marked increase of ethanol yield in the strain SYN007 strongly con rmed that complete inhibition of glycogen synthesis could contribute to the ethanol production. Also, this is the rst investigation of the effect of combinatorial inhibition of PEP synthase and glycogen synthesis on ethanol production. It can be hypothesized that the production of the increased level of ethanol in the stain SYN007 could be caused by the repression of the cell growth and glycogen storage. Although the enhancement of ethanol production was observed in the strain SYN007, it would be necessary to preform metabolomics and proteomics analysis for further understanding the regulating mechanisms or metabolic network.
In this study, considerable research attention has been made to remove the potential bottlenecks in ethanol biosynthetic pathway. As the intercellular concentration of pyruvate, the major intermediate precursor, is metabolically controlled by several endogenous pathways. Traditionally, it holds the view that the pyruvate was mainly generated from the PEP through pyruvate kinase [7]. However, recent studies have shown that a carbon ux is signi cantly channeled via the TCA cycle through the malic enzyme to pyruvate, rather than generating pyruvate directly from the ATP-generating reaction catalyzed by pyruvate kinase [22,23]. Thus, the metabolic interference of the TCA cycle is suggested as an alternative way to boost ethanol production [24]. For this purpose, a previous study has been performed in an attempt to investigate the effects of an endogenous gene slr0721 encoding malic enzyme (me) in Synechocysitis 6803 on the ethanol production [25], suggesting that the ethanol production was associated with the optimal level of malic enzyme activity in Synechocysitis 6803. To test whether the potential effects of exogenous malic enzyme on the ethanol production, we rst integrated E. coli maeB encoding NADP-dependent malic enzyme into the slr1176 site of the strain SYN007. Our results provided strong evidence that E. coli maeB was indeed shown to function as malic enzyme, catalyzing malate conversion to pyruvate. The improved ethanol production in the SYN009 strain may result from an increase in the intracellular level of the precursor pyruvate, which is subsequently metabolized for ethanol production. Equally important, the overexpression of this malic enzyme maeB led to not only the reversible oxidative decarboxylation of malate to pyruvate and CO 2 , but also accompanied with reduction of NADP + to NADPH [26]. Impressively, NADPH is an important reduced co-factor in cyanobacterial cells, and provides more favorable advantages for the NADPH-dependent metabolic pathways. Increasing NADPH production in cyanobacteria can further improve the production of the desired chemicals [27]. With respect to the nal step of ethanol synthesis, the reduced co-factor NADPH may contribute to enhancing the catalytic activity of NADPH-dependent enzyme YqhD, which catalyze acetaldehyde to ethanol. The utilization of NADPH-dependent enzymes linking an exogenous biosynthetic pathway to the modulation of the cellular metabolism are of particular interest because they likely contribute to exploiting cyanobacterial NADPH pool in the biological processes of the ethanol production. Therefore, this synergy between NADPH-producing pathway and NADPH-consuming pathway may effectively improve the activities of NADP-dependent maeB and NADPH-dependent yqhD, which cause more metabolic carbon ux towards ethanol production.
In the present study, all the engineered strains were cultivated under autotrophic conditions without optimizing growth medium, and faced many challenges in the ethanol yield. Traditionally, appropriate nutritional conditions such as certain amount of CO 2 or the reduced cofactor (NADPH) are needed to boost ethanol production. As shown in Fig. 5B, the synthetic ethanol of the nal strain SY009 showed a dramatic increase within the rst 4 days of cultivation, which almost linearly increased with the time, and then its production slightly increased. When the time consumption was counted for the whole process, the strain SYN009 showed relatively lower ethanol productivity of 93 mg L -1 day -1 after 14 days compared to other studies (Supplementary materials Fig S2). This may be the exhausted nutrient in the medium and the insu cient CO 2 in the air (less than 0.03% vol/vol) during the later stage of cultivation, which limits the cell growth rate. Thus, the requirement of high cell density became the critical factor controlling the ethanol yield in cyanobacteria. By pumping 5% CO 2 -air (vol/vol) into the photo-bioreactor, the ethanol yield of 5.5 g L -1 was achieved with a high cell density (OD 730 ≈15) after 26 days of fermentation [7]. In comparison, without using photo-bioreactor and pumping CO 2 , the strain SYN009 produced only 1.3 g L -1 of ethanol when the cell density was much lower, which OD 730 value was 1.37 after 14 days of culture. However, if the cell density were considered, the ethanol-producing e ciency of the strain SYN009 reached up to 68 mg OD 730 unit -1 L -1 day -1 , which was signi cantly better than that of the previous reports [7,12,14]. Thus, it will be one of important challenges for the engineered cyanobacteria to improve the ethanol productivity at high cell density over a long period. If overcame this problem, the utilization of the nal strain SYN009 as photosynthetic biosystem would be expected to make the ethanol production more competitive in the future.

Conclusions
In this study, we genetically engineered a cyanobacterial strain that is entitled to converting atmospheric CO 2 into ethanol at high e ciency via stepwise genetic engineering. Combined with biosynthetic pathway bottleneck disruption and genetic interventions, the engineered strain showed a higher ethanol-producing e ciency (248 mg L -1 day -1 ) compared to previous studies under photoautotrophic conditions with nutrient limitations. Our ndings indicated that the SYN009 strain (∆slr0301/pdc-yqhD, ∆slr1176/maeB) would become a useful biosystem for photosynthetic production of ethanol through optimizing the fermentation conditions, and for expanding our knowledge of exploiting cyanobacteria to produce value chemicals directly from atmospheric CO 2 .

Strains and growth conditions
The plasmids were constructed by using E. coli DH5α. The E. coli strains carrying the different plasmids were grown in the liquid LB medium or on agar plate containing corresponding antimicrobial agents, such as 50μl ml -1 spectinomycin (Sp R ), 50μl/ml kanamycin (Km R ), and 25μl ml -1 chloramphenicol (Cm R ). The wild-type and engineered Synechocystis 6803 strains were grown in the liquid BG-11 medium or on agar plates containing 1.5% agar, and kept at 30℃ under continuous illumination with an intensity of 50 μmol photons m -2 s -1 , unless otherwise noted. Engineered Synechocystis 6803 strains were grown in BG-11 medium supplemented with the corresponding antimicrobial agents according to details of each cyanobacterial strains.

Plasmid Constructions
The pMD18-T simple vector (Sangon Biotech) was used as a foundation to construct the plasmids, which were listed in Supplementary materials Table S1. Using PCR to amplify the fragments, the fragments and the vectors are double-digested by recombinase (NEW ENGLAND BioLabs Beijing, China) and ligated by T4 ligase (NEW ENGLAND BioLabs Beijing, China). All the primers were listed in Supplementary materials Table S2.
For construction of gene deletion vector, plasmid pBE406 was constructed by inserting the slr0168 gene knockout cassette into the pMD18-T vector. The slr0168 gene knockout cassette was constructed by integrating 600 bp sequence located immediately upstream slr0168 (0168 up), spectinomycin resistance gene (Sp R ), and 600 bp sequence located immediately downstream slr0168 (0168 down). The upstream and downstream of the slr0168 gene were ampli ed by PCR from Synechocystis 6803 genome. The Sp R sequence was optimized and synthesized by Sangon Biotech as the template. Similar to the method of constructing pBE406, pMD-slr0301-Ω and pMD-slr1176-Ω were constructed and the speci c method was shown in Supplementary materials Table S1. The Cm R sequence was also optimized and synthesized by Sangon Biotech.
For construction of ethanol production vector, plasmid pBE01 was constructed by inserting PpetE-pdc-yqhD expression cassette and TrbcL terminator into the plasmid pBE406. The fragment sequence of PpetE and TrbcL were both PCR-ampli ed from Synechocystis 6803 genome. The pdc and yqhD gene was PCR-ampli ed from the nucleotide sequence that has been optimized and synthesized by Sangon Biotech. Plasmid pBE02 was constructed by replacing the promoter PpetE in pBE01 with PpsbA2s. Promoter PpsbA2s was also PCR-ampli ed from Synechocystis 6803 genome. Plasmid pBE03 was constructed by inserting PpsbA2s-pdc-yqhD expression cassette and TrbcL terminator into the plasmid pMD-slr0301-Ω .
For construction of genetic intervention vector, plasmid pBE09 was constructed by inserting PpsbA2s-maeB expression cassette and TrbcL terminator into the pMD-slr1176-Ω vector. The maeB gene was ampli ed by PCR from E. coli genome.
Construction of engineered strains.
The constructed plasmids were independently transformed into Synechocystis 6803 according to the approach performed by previous studies [7]. Brie y, exponentially growing Synechocystis cultures were collected, washed with fresh BG11 medium three times, and then mixed with the corresponding plasmids. The mixture was incubated at the temperature of 30 °C for 5 hours under constant illumination and shaken periodically. The mixture was then streaked on a sterile lter membrane placed on BG11 solid medium and maintained at 30°C under continuous illumination. After 24 hours, the lter membrane was transferred to solid BG11 medium with the corresponding antibiotic. Single colonies appeared after two weeks of cultivation. A single transformant was sub-cultured stepwise on new BG11 plates containing increasing concentration of antibiotic for the occurrence of segregation and cultivated in a liquid medium for analysis. All the strains referred in this study were listed in Table 1.

Analytical methods
Synechocystis growth status was recorded by measuring transmittance at wavelength λ = 730 nm (OD730) using a spectrometer (YOKE INSTRUMENT L6 UV-Vis, Shanghai, China). The biomass is determined according to the published procedure [28]. OD730 was converted to biomass (dry cell weight g/L) by multiplying it by 0.177, which had been calculated from a calibration curve.

HPLC analysis
For the ethanol production assay, all the mutants were cultured in a fresh BG11 medium with initial OD730 = 0.1 and cultivated photoautotrophically in a ask. The BG11 medium of SYN001 contains 500 nM copper ions to induce the expression of ethanol-producing genes [29]. To get the samples, a certain amount of mutant cultures was regularly obtained and centrifuged at 10000×g for 2 minutes (MRK MG1450). Special lter membrane with 0.22-micron (Sigma -Aldrich) was used to lter the supernatant obtained by centrifugation. The ltered solution was used for ethanol analysis by the method of highperformance liquid chromatography (HPLC) [30]. According to the detection of HPLC, no ethanol was detected in the lysed precipitate.

RT-qPCR
The wild-type and ethanol-producing Synechocystis cultures (30 mL, OD 730 ≈0.6) was collected by centrifugation at 3500×g for 15min at 4℃ (BIORIDGE TGL-16M, Shanghai, China). Three biological replicates were carried out for each sample. RNA extraction and RT-qPCR analysis were performed according to the methods described previously [7]. The relative expression level of the targeted mRNA could be estimated using the calculation method of 2 -△△CT ; the higher the △CT value is, the less expression level of the targeted mRNA [31]. The endogenous gene 16S was selected as an internal reference gene.