Photosynthetic Co-production of Succinate and Ethylene in a Fast-Growing Cyanobacterium, Synechococcus elongatus PCC 11801

Cyanobacteria are emerging as hosts for photoautotrophic production of chemicals. Recent studies have attempted to stretch the limits of photosynthetic production, typically focusing on one product at a time, possibly to minimise the additional burden of product separation. Here, we explore the simultaneous production of two products that can be easily separated: ethylene, a gaseous product, and succinate, an organic acid that accumulates in the culture medium. This was achieved by expressing a single copy of the ethylene forming enzyme (efe) under the control of PcpcB, the inducer-free super-strong promoter of phycocyanin β subunit. We chose the recently reported, fast-growing and robust cyanobacterium, Synechococcus elongatus PCC 11801, as the host strain. A stable recombinant strain was constructed using CRISPR-Cpf1 in a first report of markerless genome editing of this cyanobacterium. Under photoautotrophic conditions, the recombinant strain shows specific productivities of 338.26 and 1044.18 μmole/g dry cell weight/h for ethylene and succinate, respectively. These results compare favourably with the reported productivities for individual products in cyanobacteria that are highly engineered. Metabolome profiling and 13C labelling studies indicate carbon flux redistribution and suggest avenues for further improvement. Our results show that S. elongatus PCC 11801 is a promising candidate for metabolic engineering.


Introduction
Cyanobacteria are being explored as cell factories due to their ability to convert sunlight and atmospheric CO 2 into a variety of chemicals [1]. Cyanobacteria grow faster than other photosynthetic organisms, are tolerant to stresses and can be engineered genetically to express heterologous pathways. However, processes employing these photoautotrophic prokaryotes may not become commercially viable until the productivities are significantly improved. It has been proposed that the use of fast-growing and robust cyanobacteria as chassis may result in improved carbon capture efficiency and greater product formation rates [2][3][4]. Thus, the quest for the identification of fast-growing This study * slr0168::Sp r -PcpcB-efe/sll1981::Km r -PcpcB-efe/slr0370::Cm r -PcpcB-efe/phaAB::Gm r -pcpcB-kgtP; # slr0168::Sp r -PcpcBefe/sll1981::Km r -PcpcB-efe/slr0370::Cm r -PcpcB-efe/sll1423::Gm r -pcpcB-efe.
Heterologous expression of metabolic pathways is expected to re-distribute the carbon flux. Metabolic control analysis (MCA) has been used to show that the flux alteration is dependent on the flux control coefficient of the enzyme(s). Furthermore, MCA can be used to identify the flux controlling enzymes so that the metabolic engineering efforts can be focused on those enzymes [34,35]. Likewise, metabolomics and metabolic flux analysis can be employed to identify the potential bottlenecks and rationally engineer the strain to enhance productivity, as has been demonstrated for butyraldehyde production in S. elongatus PCC 7942 [36,37]. Similarly, Xiong et al. performed isotopic non-stationary 13 C metabolic flux analysis (INST-MFA) on an ethylene producing Synechocystis sp. PCC 6803 strain to obtain a better insight into the cyanobacterial metabolism. A greater carbon flux was reported to be directed towards the heterologous ethylene pathway leading to a larger than usual flux through the TCA cycle [30].
While the literature reports on heterologous production in cyanobacteria have primarily focused on single product formation, co-production of compounds might be more economical, provided the product separation cost can be minimised. The well-known EFE is a multi-product enzyme catalysing the formation of four products, ethylene, succinate, guanidine and (S)-1-Pyrroline-5-carboxylate (P5C). While ethylene is gaseous, the remaining three products remain in the liquid phase [38,39]. P5C is utilised for arginine biosynthesis, which acts as a substrate for the reaction [40]. Guanidine, on the other hand, was co-produced with ethylene showing a titre of 586.5 mg/L after seven days of cultivation [28]. However, no extracellular succinate accumulation was observed in the efe expressing Synechocystis sp. PCC 6803 strain [30]. Since S. elongatus PCC 11801 has been reported to exhibit high tolerance towards abiotic factors [6], we explored the potential of this strain to simultaneously produce succinate and ethylene.
S. elongatus PCC 11801 was engineered to co-produce ethylene and succinate by expressing the codon-optimised efe gene under the control of a strong truncated cpcB promoter of S. elongatus PCC 7942 (P cpcB7942 ) [41]. The CRISPR-Cpf1 based genome-editing strategy was employed for the knock-in of the promoter-gene-terminator cassette at the neutral site (NS) I of S. elongatus PCC 11801. Thus, this is the first report on the use of CRISPR-based genome editing of S. elongatus PCC 11801. The growth conditions for the production of succinate and ethylene were optimised. A detailed metabolome profiling and 13 C labelling studies were performed to obtain clues about the potential metabolic control points.

Results and Discussion
The co-production of succinate and ethylene by expressing efe gene has been reported to be thermodynamically feasible [42]. Further, both the products do not show toxic effect on cyanobacteria [19,25]. It is well known that the cyanobacterial TCA cycle is more active under low light or darkness than under high light [43,44]. Since the EFE enzyme utilises the TCA cycle intermediate, 2-oxoglutarate, we employed a promoter that would be highly active under low light. The characterisation of cpcB promoter from S. elongatus PCC 7942 (P cpcB7942 ) showed that it is one of the super-strong promoters exhibiting higher activity under low light and elevated CO 2 conditions [41]. The promoters have been observed to be highly portable among phylogenetic close neighbours [6,41]. A CRISPR-based, markerless genome editing approach was used to develop the efe-expressing strain of S. elongatus PCC 11801. Antibiotic marker-free recombinant strains are likely to be safer for use in large-scale, outdoor cultivations as they do not pose the risk of spread of antibiotic resistance [45]. Further, the CRISPR-based techniques speed up the chromosomal segregation in cyanobacteria that contain multiple copies of chromosome [3]. Each copy must carry the genetic modification of interest in order to obtain a stable recombinant strain. Moreover, additional genetic modifications can be performed once a stable, marker-free strain is obtained. We chose the CRISPR-Cpf1 technique over CRISPR-cas9 due to the various advantages that the former offers in cyanobacteria [12]. Chiefly, Cpf1, a type V-A nuclease, is significantly less toxic to cyanobacteria than cas9. Cpf1 carries out a 5-bp staggered double stranded break 17 nucleotides downstream from the PAM sequence, which can then be repaired with a repair template that comprises of the homology arms and the desired payload [12].
We used the Cpf1-harbouring, self-replicating plasmid pSL2680 that alleviates the need to integrate the nuclease into the chromosome, to construct the editing plasmid pSL3338 by incorporating the guide RNA and the repair template. The editing plasmid was cured after complete chromosomal segregation to obtain a truly markerless recombinant strain that is devoid of the Cpf1 nuclease and antibiotic resistance genes. This strategy allowed us to create the efe expressing strain, S. elongatus PCC 11801:NSI::P cpcB -efe-T rrnB strain (henceforth SL3338) ( Figure S1). NSI, locus tag DOP62_03525, a gene with~84% identity with NSI of S. elongatus PCC 7942, was chosen as the site for chromosomal integration in S. elongatus PCC 11801 based on previous reports [6,7]. Complete segregation of SL3338 strain was ascertained by gel electrophoresis ( Figure S2a). As a first step, the expression of efe gene was confirmed by isolating the RNA sample and by performing reverse transcriptase PCR (RT-PCR) and gel electrophoresis. Prominent bands corresponding to the efe gene, which was absent in the wild type strain and the negative control that lacked the reverse transcriptase step, confirmed the expression ( Figure S2b). Next, it was of interest to optimise the conditions for the co-production of ethylene and succinate. Note that there is significant difference in the methods to be used for the determination of the two products. Succinate accumulates in the culture medium and can be assayed either with an enzymatic reaction-based, spectrophotometric detection kit or via GC-MS upon appropriate derivatisation. For the determination of ethylene production rate, the culture first needs to be transferred to an airtight bottle followed by sampling of the headspace at regular intervals and assay via GC-FID. Thus, the growth conditions were first optimised for the production of succinate and then ethylene production rates were estimated for some of the conditions. This strategy was adopted primarily due to the relative ease of succinate detection.

Optimising Growth Condition for Improved Succinate Production in the Recombinant Strain
We characterised the growth of the recombinant S. elongatus PCC 11801 (SL3338) strain under low and high light (LL and HL), ambient and 1% CO 2 (LC and HC) and in shake flask (SF) or multi-cultivator (MC). In SF, while SL3338 showed growth defect with~20% reduction in biomass accumulation under LC-HL (Figure 1a), the extent of growth retardation was negligible under HC-HL (Figure 1a). On the contrary, under HC-LL, SL3338 showed lower growth for the initial 24 h or so but eventually accumulated biomass that was comparable to that of HL-HC ( Figure S3). Under both high and low light, the WT and SL3338 strains accumulated~6-7 g DCW in 72-120 h under HC. Thus, similar to earlier reports, the expression of efe did not impose a severe metabolic burden under HC and showed no significant growth retardation. Moreover, the photosynthetic efficiency, as a function of the Fv/Fm was observed to be not compromised between the strains (Table S3). Unlike SF, growth in MC 1000-OD under 20 µE·m −2 ·s −1 with bubbled 1% CO 2 showed significantly retarded growth for both strains (Figure 1b). However, high light intensity ensured higher growth and biomass accumulation in MC 1000-OD with minimum growth defect between wild type and recombinant strains (Figure 1b). The best growth conditions might not be the best production conditions as product formation is also dependent on the transcriptional and translational rates of protein production.
MS upon appropriate derivatisation. For the determination of ethylene production rate, the culture first needs to be transferred to an airtight bottle followed by sampling of the headspace at regular intervals and assay via GC-FID. Thus, the growth conditions were first optimised for the production of succinate and then ethylene production rates were estimated for some of the conditions. This strategy was adopted primarily due to the relative ease of succinate detection.

Optimising Growth Condition for Improved Succinate Production in the Recombinant Strain
We characterised the growth of the recombinant S. elongatus PCC 11801 (SL3338) strain under low and high light (LL and HL), ambient and 1% CO2 (LC and HC) and in shake flask (SF) or multicultivator (MC). In SF, while SL3338 showed growth defect with ~20% reduction in biomass accumulation under LC-HL (Figure 1a), the extent of growth retardation was negligible under HC-HL (Figure 1a). On the contrary, under HC-LL, SL3338 showed lower growth for the initial 24 h or so but eventually accumulated biomass that was comparable to that of HL-HC ( Figure S3). Under both high and low light, the WT and SL3338 strains accumulated ~6-7 g DCW in 72-120 h under HC. Thus, similar to earlier reports, the expression of efe did not impose a severe metabolic burden under HC and showed no significant growth retardation. Moreover, the photosynthetic efficiency, as a function of the Fv/Fm was observed to be not compromised between the strains (Table S3). Unlike SF, growth in MC 1000-OD under 20 μE·m −2 ·s −1 with bubbled 1% CO2 showed significantly retarded growth for both strains (Figure 1b). However, high light intensity ensured higher growth and biomass accumulation in MC 1000-OD with minimum growth defect between wild type and recombinant strains (Figure 1b). The best growth conditions might not be the best production conditions as product formation is also dependent on the transcriptional and translational rates of protein production. Succinate was estimated to optimise the growth conditions as succinate is excreted into the culture medium and easily detected. While no experimental reports are available on extracellular succinate accumulation by efe-expressing cyanobacteria, the engineered strain of this study (SL3338) showed ~7-fold higher extracellular succinate level in shake flask as compared to wild type after 120 h of SF cultivation under LC-HL. The succinate productivity may have been limited by the availability of the carbon substrate as daily addition of 50 mM sodium bicarbonate to the culture or growth under 1% CO2 have improved the succinate productivity by 40% (Figure 2a). Likewise, an improved productivity was observed under HC compared to LC (data not shown). Thus, all the Succinate was estimated to optimise the growth conditions as succinate is excreted into the culture medium and easily detected. While no experimental reports are available on extracellular succinate accumulation by efe-expressing cyanobacteria, the engineered strain of this study (SL3338) showed 7-fold higher extracellular succinate level in shake flask as compared to wild type after 120 h of SF cultivation under LC-HL. The succinate productivity may have been limited by the availability of the carbon substrate as daily addition of 50 mM sodium bicarbonate to the culture or growth under 1% CO 2 have improved the succinate productivity by 40% (Figure 2a). Likewise, an improved productivity was observed under HC compared to LC (data not shown). Thus, all the subsequent trials for succinate productivity were carried out under 1% CO 2 (or HC), both in SF and MC1000-OD. Next, it was of interest to identify the optimal light intensity for product formation. In going from the light intensity of 300 to 20 µE·m −2 ·s −1 , the biomass levels decreased but the product titre increased yielding the highest specific productivity at 20 µE·m −2 ·s −1 (Figure 2b). Therefore, the increased productivity under LL-HC may be a combined effect of the high promoter activity of P cpcB7942 [41] and a greater TCA cycle activity in at low light [43,44]. Note that α-ketoglutarate, a key precursor of the EFE enzyme, is an intermediate of the TCA cycle and its availability is expected to affect the product formation. fumarate in SL3338 compared to 5-10 mg/L for the WT cultures under identical conditions (data not shown). We hypothesise that the levels of fumarate can be minimised by knocking out the enzyme succinate dehydrogenase [21]. However, for ethylene production, the use of fumarate as an endogenous substrate has been predicted to be thermodynamically advantageous to enhance the driving force towards ethylene [39]. Therefore, we propose the Sdh knockdown as the future work for improved succinate yield and reduced fumarate side product.  The availability of oxygen is essential for EFE catalysed reaction, therefore different cultivation flasks were tested to investigate the effect of oxygen under the production condition. Although the biomass accumulation in the bottle was lower than in shake flasks, no significant improvement in succinate titre was observed in baffled flasks as compared to unbaffled flasks or bottles ( Figure S4). Cyanobacterial cultures exhibit super-saturated levels of dissolved oxygen under photoautotrophic conditions [46,47] and this may explain the lack of effect of baffles on succinate production.
Fumarate is a product of both the oxidative and reductive arm of the TCA cycle. Since excess succinate can potentially get converted into fumarate via the oxidative arm [30], fumarate can potentially accumulate as an undesirable by-product. We observed accumulation of 30-40 mg/L fumarate in SL3338 compared to 5-10 mg/L for the WT cultures under identical conditions (data not shown). We hypothesise that the levels of fumarate can be minimised by knocking out the enzyme succinate dehydrogenase [21]. However, for ethylene production, the use of fumarate as an endogenous substrate has been predicted to be thermodynamically advantageous to enhance the driving force towards ethylene [42]. Therefore, we propose the Sdh knockdown as the future work for improved succinate yield and reduced fumarate side product.

Nitrate Limitation Positively Affected Succinate Production
Nitrogen deprivation is known to induce overflow metabolism in cyanobacteria apart from causing an increase in the intracellular 2-oxoglutarate pool [48,49]. It was of interest to test the effect of nitrate on succinate productivity as 2-oxoglutarate is a key precursor for the EFE enzyme. Previous reports have shown that growth in either BG11 0 or 1 mM NaNO 3 supplemented BG11 was highly retarded with insignificant ethylene production [25,33]. Therefore, the effect of nitrate on product formation was investigated in the recombinant and wild type strains under different nitrate (NaNO 3 ) concentrations, viz., 0.25×, 0.5× and 1× (17 mM) NaNO 3 . Though there was no change in the succinate accumulation in recombinant strain across the conditions after 72 h, the succinate level in 0.5× nitrate concentration was higher than in 1× nitrate after 120 h ( Figure 3a). Further reduction in nitrate concentration led to reduced succinate titres (Figure 3a). While growth was retarded under nitrate limited state, the recombinant strain did not show bleached phenotype unlike the wild type strain (Figure 3b). Generally, glycogen deficient strains exhibit a non-bleaching phenotype during nitrogen starvation as a result of an unknown mechanism [48,50]. Recently, a correlation between glycogen deficiency and alternation in the adenylate energy charge has been shown which in turn leads to higher accumulation of organic acids under nitrogen deplete condition [51]. The energy charge level in the cell has been reported to be associated with cyanobacterial growth and partitioning of carbon. We hypothesised that non-bleaching phenotypes is triggered as a result of difference in energy charge distribution between wild type and mutant cells [52]. Since the recombinant strain showed lower glycogen accumulation than wild type under both high and low light intensities (Figure 3c and Figure S5), similar metabolic re-arrangement might have triggered non-bleaching phenotype in the SL3338 strain.

Nitrate Limitation Positively Affected Succinate Production
Nitrogen deprivation is known to induce overflow metabolism in cyanobacteria apart from causing an increase in the intracellular 2-oxoglutarate pool [45,46]. It was of interest to test the effect of nitrate on succinate productivity as 2-oxoglutarate is a key precursor for the EFE enzyme. Previous reports have shown that growth in either BG110 or 1 mM NaNO3 supplemented BG11 was highly retarded with insignificant ethylene production [25,47]. Therefore, the effect of nitrate on product formation was investigated in the recombinant and wild type strains under different nitrate (NaNO3) concentrations, viz., 0.25, 0.5 and 1 (17 mM) NaNO3. Though there was no change in the succinate accumulation in recombinant strain across the conditions after 72 h, the succinate level in 0.5 nitrate concentration was higher than in 1 nitrate after 120 h ( Figure 3a). Further reduction in nitrate concentration led to reduced succinate titres ( Figure 3a). While growth was retarded under nitrate limited state, the recombinant strain did not show bleached phenotype unlike the wild type strain ( Figure 3b). Generally, glycogen deficient strains exhibit a non-bleaching phenotype during nitrogen starvation as a result of an unknown mechanism [45,48]. Recently, a correlation between glycogen deficiency and alternation in the adenylate energy charge has been shown which in turn leads to higher accumulation of organic acids under nitrogen deplete condition [49]. The energy charge level in the cell has been reported to be associated with cyanobacterial growth and partitioning of carbon. We hypothesised that non-bleaching phenotypes is triggered as a result of difference in energy charge distribution between wild type and mutant cells [50]. Since the recombinant strain showed lower glycogen accumulation than wild type under both high and low light intensities (Figure 3c and Figure  S5), similar metabolic re-arrangement might have triggered non-bleaching phenotype in the SL3338 strain. Figure 3. The effect of nitrate limitation on succinate production: (a) The OD730nm and succinate titres were measured for the SL3338 strain at 72 and 120 h under the LL-HC in 25 mL shake flasks with 7 mL BG-11 media and varying NaNO3 concentrations. The 1 nitrate is equivalent to 1.5 g/L. The error bars correspond to SEM from the biological replicates (n = 3). (b) The non-bleaching phenotype exhibited by the recombinant strain as compared to wild type under nitrate limitation. Refer to section 4.5 for experimental details. (c) Glycogen content measured as per the protocol reported in [6]. For glycogen estimation, the cultures were grown at 38 °C under low light intensity (20 μE·m −2 ·s −1 ) in shaker incubator aerated with 1% CO2 to different cell densities (OD730nm). The error bars for the glycogen data correspond to SEM from three biological replicates (and two technical replicates for each biological replicate) (n = 3) and the letters a and b denote statistically different values for each category (p < 0.05) using t-test. The effect of nitrate limitation on succinate production: (a) The OD 730nm and succinate titres were measured for the SL3338 strain at 72 and 120 h under the LL-HC in 25 mL shake flasks with 7 mL BG-11 media and varying NaNO 3 concentrations. The 1× nitrate is equivalent to 1.5 g/L. The error bars correspond to SEM from the biological replicates (n = 3). (b) The non-bleaching phenotype exhibited by the recombinant strain as compared to wild type under nitrate limitation. Refer to Section 3.5 for experimental details. (c) Glycogen content measured as per the protocol reported in [6]. For glycogen estimation, the cultures were grown at 38 • C under low light intensity (20 µE·m −2 ·s −1 ) in shaker incubator aerated with 1% CO 2 to different cell densities (OD 730nm ). The error bars for the glycogen data correspond to SEM from three biological replicates (and two technical replicates for each biological replicate) (n = 3) and the letters a and b denote statistically different values for each category (p < 0.05) using t-test.

Efficient Carbon Partitioning Between Product Formation and Biomass
The recombinant and wild type strains were cultivated to measure the levels of succinate under more controlled conditions in MC 1000-OD with bubbling 1% CO 2 . While cultivation under high light intensity of 300 µE·m −2 ·s. −1 exhibited higher growth, the production was restricted to 250 mg/L (200 mg/gDCW) of succinate in 120 h ( Figure S6). On the contrary, the combination of low light and 1% CO 2 (LL-HC) highly impaired growth of both wild type and recombinant, but showed significantly higher succinate titres ( Figure 2c). The peak specific production rate of 121.1 mg/gDCW/h was observed, which is much higher than the reported rates (Table 1).
Optimal carbon partitioning would ensure a perfect balance between biomass accumulation and product formation [23]. Several of the metabolic engineering studies in cyanobacteria suffer from low specific productivity as the recombinant strain exhibits a substantial increase in biomass [19,22,25]. There have been efforts to balance the carbon flow between biomass and product accumulation by downregulating the citrate synthase gene, which arrests growth and diverts greater amounts of carbon towards product formation [23]. This strategy of targeting citrate synthase might not be applicable for our study as we focus on increasing flux through TCA cycle. Besides genetic modifications, external factors such light, temperature and pH have also been observed to modulate total intracellular carbon flux in organism [53]. In this study, a similar carbon partitioning was facilitated in the recombinant strain when cultivated in MC 1000-OD under 20 µE·m −2 ·s −1 light intensity and 1% CO 2 . The combination of low light and high CO 2 arrested growth, ensuring carbon flow towards the product of interest, thereby enhancing the specific productivity and carbon partitioning toward the product (Figures 2c and 4). The carbon partitioning was calculated using the equations described below (Equations (1) and (2)): where the molecular wt. of succinate is 118.09 g/M while that of biomass (with chemical composition of CH 1.8 O 0.5 N 0.15 ) is 23.9 g/M.

Efficient Carbon Partitioning Between Product Formation and Biomass
The recombinant and wild type strains were cultivated to measure the levels of succinate under more controlled conditions in MC 1000-OD with bubbling 1% CO2. While cultivation under high light intensity of 300 μE·m −2 ·s. −1 exhibited higher growth, the production was restricted to 250 mg/L (200 mg/gDCW) of succinate in 120 h ( Figure S6). On the contrary, the combination of low light and 1% CO2 (LL-HC) highly impaired growth of both wild type and recombinant, but showed significantly higher succinate titres (Figure 2c). The peak specific production rate of 121.1 mg/gDCW/h was observed, which is much higher than the reported rates (Table 1).
Optimal carbon partitioning would ensure a perfect balance between biomass accumulation and product formation [23]. Several of the metabolic engineering studies in cyanobacteria suffer from low specific productivity as the recombinant strain exhibits a substantial increase in biomass [19,22,25]. There have been efforts to balance the carbon flow between biomass and product accumulation by downregulating the citrate synthase gene, which arrests growth and diverts greater amounts of carbon towards product formation [23]. This strategy of targeting citrate synthase might not be applicable for our study as we focus on increasing flux through TCA cycle. Besides genetic modifications, external factors such light, temperature and pH have also been observed to modulate total intracellular carbon flux in organism [51]. In this study, a similar carbon partitioning was facilitated in the recombinant strain when cultivated in MC 1000-OD under 20 μE·m −2 ·s −1 light intensity and 1% CO2. The combination of low light and high CO2 arrested growth, ensuring carbon flow towards the product of interest, thereby enhancing the specific productivity and carbon partitioning toward the product (Figure 2c and Figure 4). The carbon partitioning was calculated using the equations described below (Equations (1) and (2)): where the molecular wt. of succinate is 118.09 g/M while that of biomass (with chemical composition of CH1.8O0.5N0.15) is 23.9 g/M. Unlike MC, the shake flask showed much greater carbon diverted towards biomass formation under conditions of LL-HC maintained in a shaker incubator (Figure 2d). MC represents a more carbon saturated state because 1% CO2 is bubbled into the culture than shake flask where mixing is facilitated by surface aeration [38]. Since bicarbonate has been reported to be used as a buffering agent, 1% CO2 bubbling in MC might provide an improved buffering system than 1% CO2 in chamber. Moreover S. elongatus PCC 11801 exhibits better growth rate Unlike MC, the shake flask showed much greater carbon diverted towards biomass formation under conditions of LL-HC maintained in a shaker incubator (Figure 2d). MC represents a more carbon saturated state because 1% CO 2 is bubbled into the culture than shake flask where mixing is facilitated by surface aeration [41]. Since bicarbonate has been reported to be used as a buffering agent, 1% CO 2 bubbling in MC might provide an improved buffering system than 1% CO 2 in chamber. Moreover S. elongatus PCC 11801 exhibits better growth rate under unbuffered media ( Figure S7) in shake flask than in MC under 1% CO 2 [6]. Apart from the differences in the pH, the difference in the light penetration in the cultures due to the different arrangement of the LED lights in a shaker versus the MC might also affect the growth pattern. Thus, the reduction in biomass accumulation observed in MC 1000-OD might be a combinatorial effect of changed external growth conditions.

Co-Production of Ethylene in S. elongatus PCC 11801-efe Strain
Ethylene is the primary product of the EFE mediated reaction and stoichiometrically two moles of ethylene are produced from three moles of 2-oxoglutarate [31]. Since it is a gaseous product, airtight bottles were used for ethylene production. The recombinant and the wild type strains were grown in a shake flask or MC 1000-OD under LL-HC, the conditions that were optimised for succinate production and were inoculated with different biomass in bottles under~20 µE·m −2 ·s −1 light intensity. Ethylene accumulated in the headspace was estimated every hour using GC-FID. While the wild type strain showed no ethylene accumulation, the recombinant strain showed ethylene production which increased when incubated with higher biomass (Figure 5a). Although the accumulation was observed to be saturated within 8 h (Figure 5a), the same biomass could be utilised multiple times for ethylene production without compromising the potential of the strain ( Figure S8). The early saturation in ethylene accumulation was overcome with very low inoculum (OD 730nm 0.05) culture incubated in a bottle under low light and 50 mM sodium bicarbonate. The low inoculum ensured better light penetration as compared to higher cell density cultures and further supplementation with bicarbonate might have drastically changed the growth condition thereby facilitating steady increase in ethylene accumulation. Although a steady increase in ethylene accumulation was observed over a period of 12 days with negligible increment in biomass, the rate of increase was slower in comparison to higher biomass ( Figure 5b). As observed with respect to succinate, production in the bottle showed steady maintenance of biomass and diverting more carbon towards the product of interest, facilitating better partitioning of carbon than shake flasks ( Figure S9). Therefore, under optimal production conditions, the SL3338 strain was able to co-produce 338.26 µmole/gDCW/h of ethylene using a single copy of efe gene, which is by far significant specific productivity of ethylene with minimal modifications (Table 1). A detailed comparison of the product levels along with the reported conditions is provided in Table S2.
Metabolites 2020, 10, x FOR PEER REVIEW 9 of 22 minimal modifications (Table 1). A detailed comparison of the product levels along with the reported conditions is provided in Table S2.

Metabolic Profiling and 13 C Labelling
Metabolomics and 13 C isotopic carbon labelling can provide some key insights into a strain engineering workflow. Metabolite pool sizes provide a static snapshot of the metabolism, which in turn may be dependent on the net rate of synthesis of each metabolism. Mapping the changes in metabolite levels can potentially provide unique insights in the differential metabolic activity between the wild type and the engineered strain. 13 C isotope labelling studies, on the other hand, can reveal additional information on the changes in the reaction rates in metabolic pathways [22]. To that end, we utilised orthogonal metabolomics techniques, viz., measurement of relative pool sizes and the dynamic 13

Metabolic Profiling and 13 C Labelling
Metabolomics and 13 C isotopic carbon labelling can provide some key insights into a strain engineering workflow. Metabolite pool sizes provide a static snapshot of the metabolism, which in turn may be dependent on the net rate of synthesis of each metabolism. Mapping the changes in metabolite levels can potentially provide unique insights in the differential metabolic activity between the wild type and the engineered strain. 13 C isotope labelling studies, on the other hand, can reveal additional information on the changes in the reaction rates in metabolic pathways [22]. To that end, we utilised orthogonal metabolomics techniques, viz., measurement of relative pool sizes and the dynamic 13 C isotopic enrichment of the various intermediate metabolites of the central carbon metabolism, amino acids, sugar nucleotides and cofactors. Since low light intensity ensured the highest productivity of succinate, the metabolomics experiments were performed under similar conditions in shake flask for recombinant and wild type strains. The results of metabolite profiling and 13 C enrichment were analysed together to identify possible redistribution of the carbon. For example, a slower 13 C enrichment can result either from a slower rate of reaction or from a larger pool size. However, a slower reaction rate can be unequivocally concluded from a smaller pool size together with slower labelling rate.
Transient 13 C-labelling experiments were performed on both strains by adding 13 C-labelled sodium bicarbonate to the culture medium during the exponential growth phase [54]. Culture samples were collected at predetermined intervals, quenched, cellular metabolites extracted and 13 C enrichment quantified via LC/MS/MS [14,[54][55][56]. We monitored labelling in 34 metabolites but only those that show satisfactory labelling pattern and appreciable accumulation of 13 C carbon are shown in Figures S10 and S11. Within the 30-min labelling experiment, a number of intermediate metabolites accumulated substantial amounts of the isotopic 13 C carbon [57]. On the other hand, the metabolite pool sizes were estimated for both the strains during the early (exponential phase) and late phase of growth. Late phase metabolite profiling was performed to further investigate the possible causes of reduced rate of product formation. As the mass detectors of LC/MS are known to be non-quantitative, relative quantitation of metabolites was obtained via the isotopic ratio method [16]. The metabolite extract from fully 13 C-labelled biomass of S. elongatus PCC 11801 was used as an internal standard, which was mixed with every sample in equal proportion [7]. The monoisotopic 13 C peak of each metabolite serves as an internal standard thereby accounting for the matrix effects and possible drifts in the MS detector. The ratio of monoisotopic 12 C and 13 C peaks for each metabolite thus aids the relative quantitation. Overall, 50 metabolites were monitored (Table S4) but only those that showed fold change of >1.4 or <0.71 at a p-value of 0.05 are shown in Figure 6. While the area ratios ( 12 C/ 13 C) for all the metabolites in wild type and SL3338 during early and late phase are provided in File S2 (Tables S5 and S6), the MID and the mean enrichment values of the metabolites are provided in Files S3 and S4.

Partitioning of Carbon in the Recombinant Strain
The Calvin Benson Bassham (CBB) cycle acts as the entry point for the inorganic carbon into the cyanobacterial carbon metabolism. There were no significant differences between the two strains in the pool sizes of the CBB cycle intermediates at early phase ( Figure 6). However, the 13 C labelling studies showed higher 13 C enrichment at later time points (pseudo isotopic steady state) in the majority of the CBB cycle intermediates with similar initial labelling rates as wild type (Figure 7 and Figure S12). Therefore, the increased enrichment with no significant change in the total pool and rates indicate that the carbon stored as inactive pool of these metabolites may have been utilised for active metabolism in the engineered strain. This is more clearly evident in the plots of the mass isotopologue distribution (MID) of metabolites ( Figure S10). The level of M0, the unlabelled state, is expected to reach zero at steady state, which is expected to be seen at 3× the doubling time of the culture. However, in a very short period of 10-30 min, the MIDs of the fast-labelling metabolites reach a pseudo-steady state. To exemplify, the M0 of 3-phosphogyceric acid (3-PGA) reached a pseudo-steady state level of~35% in 10 min ( Figure S11). This suggests that there may be an inactive pool of 3-PGA inside the cell, approximately 35% in WT, that does not participate in the active metabolism. In the recombinant strain, the inactive pool of 3-PGA reduces to~28%. Likewise, we observed a reduction in the inactive pools of ribulose-1,5-bisphosphate (RuBP) and phosphoenolpyruvate (PEP) in the recombinant strain.
Interestingly, despite the absence of obvious differences in the pools of central metabolites, other peripheral metabolites such as sugar nucleotides, which act as activated precursors in the synthesis of storage molecules and cell wall components, were found to be depleted in the engineered strain ( Figure 6). This suggests that the cyanobacterial metabolism responds to the stress induced by heterologous pathways by re-partitioning the carbon stored as inactive pools and by lowering the flux towards the synthesis of storage compounds [16,58]. The hypothesis is very well complemented by lower rate of 13 C enrichment in ADP-glucose, a precursor for glycogen synthesis, and a lower glycogen content (major storage compound in cyanobacteria) in the engineered strain (Figures 7  and 8). On the other hand, the lower abundance of other of metabolites responsible for peptidoglycan biosynthesis ( Figure 6) might lead to altered cell wall architecture. Although the cell size was not affected (Figure S13), the cell wall composition can potentially undergo a change in response to the additional metabolic burden. Sucrose acts as an osmolyte and a relatively greater abundance of this metabolite was observed in the engineered strain with a reduced pool of UDP-glucose (UDPG) during the early phase ( Figure 6). Moreover, no significant difference in the mean enrichment of UDPG was observed for the first 10 min (Figure 7 and Figure S12). Thus, this increase in the intracellular sucrose pool might be to compensate for the changed osmolarity of the medium due to the presence of extracellular succinate.
Metabolites 2020, 10, x FOR PEER REVIEW 11 of 22 control point and can be overexpressed to improve productivity as observed in earlier reports [7,59,60].
Metabolite profile in late phase shows interesting results. The metabolites that showed significant changes are presented in Figure 6. In contrast to the results for early phase, there was no significant difference in the levels of ADP-Glucose, UDP-Glucose and various other sugar nucleotides ( Figure 6) between the engineered and wild type strains. This possibly suggests that wild type and the engineered strain allocates approximately equal carbon towards storage molecules during the late phase metabolism. This is supported by the fact that there is no significant change observed for both glycogen and sucrose between the two strains (Figure 3c,6b). Further, the glycogen content was in general higher in late phase under LL-HC and HL-HC for both the wild type and the engineered strains (Figure 3c and Figure S5). Since the recombinant strain showed increased glycogen content similar to wild type with no change in the relative abundance of ADPG (Figure 3c and 6), knockout of genes for glycogen synthesis might divert more carbon towards the pathway of interest increasing the productivity during the later phase of growth [47].

Figure 6.
Fold change in the metabolite levels between SL3338 and WT strains at early phase and late phase of growth. While the early phase corresponds to the exponential phase and it was achieved in 18 h, the late phase of growth corresponds to the 96 h when the OD730nm was on average 11. Only those metabolites are included in the panel where the statistically significant fold change was > 1.4 or < 0.71 at p -value of < 0.05 in at least one of the conditions. Samples were drawn, quenched and extracted as reported previously [55]. The extracts were mixed with the extract of fully 13 C-labelled biomass of S. elongatus PCC 11801 WT that was used as the internal standard [7]. Fifty metabolites were monitored via LC/MS and quantified as isotopic area ratio. The ratio of the metabolite pool of recombinant to wild type was calculated as ratio of isotopic ratios of recombinant to WT strains and a symbol * represent the metabolites having a fold change of > 1.4 or < 0.71 and a p-value of < 0.05 using t-test. The metabolites were identified at MS2 level using the MS-DIAL [61] and MetDIA [62] tool. Error bars shown correspond to SEM from three biological replicates (and three technical replicates for each biological replicate) (n = 3) and are calculated as mentioned in the statistical analysis section. The full list of metabolites with abbreviations is provided in Table S4. More details are presented in Section 4.8. Figure 6. Fold change in the metabolite levels between SL3338 and WT strains at early phase and late phase of growth. While the early phase corresponds to the exponential phase and it was achieved in 18 h, the late phase of growth corresponds to the 96 h when the OD 730nm was on average 11. Only those metabolites are included in the panel where the statistically significant fold change was > 1.4 or < 0.71 at p-value of < 0.05 in at least one of the conditions. Samples were drawn, quenched and extracted as reported previously [55]. The extracts were mixed with the extract of fully 13 C-labelled biomass of S. elongatus PCC 11801 WT that was used as the internal standard [7]. Fifty metabolites were monitored via LC/MS and quantified as isotopic area ratio. The ratio of the metabolite pool of recombinant to wild type was calculated as ratio of isotopic ratios of recombinant to WT strains and a symbol * represent the metabolites having a fold change of > 1.4 or < 0.71 and a p-value of < 0.05 using t-test. The metabolites were identified at MS2 level using the MS-DIAL [61] and MetDIA [62] tool. Error bars shown correspond to SEM from three biological replicates (and three technical replicates for each biological replicate) (n = 3) and are calculated as mentioned in the statistical analysis section. The full list of metabolites with abbreviations is provided in Table S4. More details are presented in Section 3.8.

Lower Part of Glycolysis and the TCA Cycle
The recombinant SL3338 strain showed lower pool of PEP ( Figure 6) during the early phase and higher 13 C enrichment than wild type (Figure 7), suggesting greater carbon flux in the lower branch of glycolysis. However, the lack of difference in pool size and lower rates of 13 C enrichment of acetyl CoA (Figures 6 and 8) combined with the lower expression of pyruvate dehydrogenase (PDH) ( Figure  S14) suggest a lower flux through PDH in the recombinant strain. It is likely that a greater proportion of PEP may be diverted to the TCA cycle via PEP carboxylase (ppc) in order to replenish the TCA cycle intermediates that get depleted by the action of efe. Thus, we suggest simultaneous overexpression of ppc and PDH enzymes to improve flux through the TCA cycle. The late phase metabolism showed a higher abundance of PEP and Aspartate (ASP) which may result from slower  Figure 8 and Section 3.9 for experimental details. The values represent the average of two biological replicates (n = 2), each having two technical replicates. Error bars correspond to the SEM of the averaged biological replicates (n = 2). The abbreviations used for metabolites are: Acetyl CoA, acetyl coenzyme A; AKG, alpha-ketoglutaric acid; ARG, arginine; ASP, aspartic acid; CIT, citric acid; DHAP, dihydroxyacetone phosphate; FBP, fructose 1,6 bisphosphate; FUM, dumaric acid; F6P, fructose-6phosphate; GAP, glyceraldehyde-3-phosphate; GLU, glutamate; G1P, glucose-1-phosphate; G6P, glucose-6-phosphate; ISO-CIT, isocitric acid; MAL, malic acid; OAA, oxaloacetic acid, PEP, phosphoenolpyruvate; PYR, pyruvic acid; P5C, (S)-1-pyrroline-5-carboxylate; RuBP, ribulose 1,5 bisphosphate; Ru5P, ribulose-5-phosphate; R5P, ribulose-5-phosphate; SBP, sedoheptulose 1,7 bisphosphate; SUC, sucrose; SUCC, succinate; S7P, sedoheptulose-7-bisphosphate; UDPG, UDP-glucose; 2PGA, 2-phosphoglyceric acid; and 3PGA, 3-phosphoglyceric acid. The gene efe (ethylene forming enzyme) (red) was inserted into the engineered SL3338 strain. towards succinate completing the bifurcated TCA cycle due to the expression of efe gene (Figure 7). Since the majority of succinate is exported out of the recombinant cell (Figure 2), a basal level of intracellular succinate content was observed similar to the wild type ( Figure 6). On the contrary, the low 13 C mean enrichment pattern observed for succinate in recombinant (Figure 7) might indicate dilution of the intracellular succinate pool due to the rapid exchange of succinate across the leaky cell membrane.

Conclusion
Metabolic engineering of cyanobacteria for the co-production of two essential products in two different phases might be economically advantageous. To that end, the EFE enzyme was expressed under a strong PcpcB7942 promoter in a fast-growing cyanobacterium, S. elongatus PCC 11801, to simultaneously produce ethylene and succinate. The best production condition was observed to be low light, high CO2, and this might be due to the enhanced activity of the promoter in use [38] or the activated TCA cycle [40]. The low light production strategy in MC 1000-OD ensured the accumulation of lower biomass with higher cumulative product titre. On the contrary, cultivation in shake flask limited the carbon partitioning towards the product of interest. Under optimal conditions, SL3338 strain co-produced succinate and ethylene at peak specific production rates of 338.26 and 1044.18 μmole/gDCW/h, respectively. These results obtained by expressing a single copy of the efe gene in S. elongatus PCC 11801 compare favourably with the reported productivities for strains that have been engineered either with multiple copies of efe or have additional genetic modifications. Further, metabolite profiling was performed to identify the potential bottlenecks under low light and 1% CO2 in the shake flask. The metabolome analysis showed that the recombinant SL3338 strain during the early phase did not alter the relative pool sizes of CBB cycle intermediates but enhanced 13 C enrichment at later time points (pseudo isotopic steady state) towards the CBB metabolites. This Unlike the early phase, the late phase metabolism of the recombinant strain manifested significant accumulation of ribulose bisphosphate (RuBP), 3PGA, fructose-1,6-bisphosphate (FBP) and sedoheptulose-1,7-bisphosphate (SBP) in the recombinant strain ( Figure 6). It is noteworthy that the conversion of RuBP, FBP and SBP to respective products is catalysed by the potential flux controlling enzymes, viz., RuBisCO and sedoheptulose/fructose bisphosphatases (SBPase/FBPase) that catalyse irreversible reactions. The accumulation of FBP and SBP suggests that SBPase/FBPase may become limiting during the late phase of metabolism in the engineered strain, as the cells have additional metabolic burden of making products apart from carrying out normal cellular activities. RT-PCR to estimate the mRNA expression levels of SBPase/FBPase showed qualitatively lower expression in recombinant strain ( Figure S14). These results suggest that this enzyme may be a major control point and can be overexpressed to improve productivity as observed in earlier reports [7,59,60].
Metabolite profile in late phase shows interesting results. The metabolites that showed significant changes are presented in Figure 6. In contrast to the results for early phase, there was no significant difference in the levels of ADP-Glucose, UDP-Glucose and various other sugar nucleotides ( Figure 6) between the engineered and wild type strains. This possibly suggests that wild type and the engineered strain allocates approximately equal carbon towards storage molecules during the late phase metabolism. This is supported by the fact that there is no significant change observed for both glycogen and sucrose between the two strains ( Figure 3c, Figure 6). Further, the glycogen content was in general higher in late phase under LL-HC and HL-HC for both the wild type and the engineered strains (Figure 3c and Figure S5). Since the recombinant strain showed increased glycogen content similar to wild type with no change in the relative abundance of ADPG (Figures 3c and 6), knockout of genes for glycogen synthesis might divert more carbon towards the pathway of interest increasing the productivity during the later phase of growth [33].

Lower Part of Glycolysis and the TCA Cycle
The recombinant SL3338 strain showed lower pool of PEP ( Figure 6) during the early phase and higher 13 C enrichment than wild type (Figure 7), suggesting greater carbon flux in the lower branch of glycolysis. However, the lack of difference in pool size and lower rates of 13 C enrichment of acetyl CoA (Figures 6 and 8) combined with the lower expression of pyruvate dehydrogenase (PDH) ( Figure S14) suggest a lower flux through PDH in the recombinant strain. It is likely that a greater proportion of PEP may be diverted to the TCA cycle via PEP carboxylase (ppc) in order to replenish the TCA cycle intermediates that get depleted by the action of efe. Thus, we suggest simultaneous overexpression of ppc and PDH enzymes to improve flux through the TCA cycle. The late phase metabolism showed a higher abundance of PEP and Aspartate (ASP) which may result from slower rates of succinate and ethylene production and higher biomass accumulation ( Figure 6). Therefore, overexpression of ppc might improve and sustain the productivity by optimising the carbon flow towards TCA cycle [19,22].
Certain metabolites that are directly derived from the intermediates of TCA cycle were observed to be depleted in the early phase suggesting significant product formation ( Figure 6). The lower glutamine (GLN) level ( Figure 6) depicts more carbon directed to the EFE-catalysed reaction, which utilises 2-oxoglutarate. Since arginine (ARG) is a substrate for ethylene formation and N-acetyl glutamate (N-acetyl GLU) is a precursor for ARG synthesis via the ornithine pathway, the depletion of N-acetyl GLU was not surprising. On the other hand, the abundance of glutamate (GLU) pool was maintained due to a cyclic reaction involving P5C and guanidine which are important for arginine biosynthesis replenishing the GLU pool [40]. Subsequently, the significantly lower rate of 13 C enrichment for GLU in SL3338 strain than wild type confirmed the efficient partitioning of carbon towards succinate completing the bifurcated TCA cycle due to the expression of efe gene (Figure 7). Since the majority of succinate is exported out of the recombinant cell (Figure 2), a basal level of intracellular succinate content was observed similar to the wild type ( Figure 6). On the contrary, the low 13 C mean enrichment pattern observed for succinate in recombinant (Figure 7) might indicate dilution of the intracellular succinate pool due to the rapid exchange of succinate across the leaky cell membrane.

Chemicals and Reagents
All enzymes were purchased from New England Biolabs (NEB, Ipswich, MA, USA) and ThermoFischer Scientific (Waltham, MA, USA). The isolation and purification kits were obtained from Qiagen (Hilden, Germany). All chemicals, reagents and antibiotics used were of analytical/HPLC grade, whereas reagents for LCMS analysis were of LCMS grade and were procured from Sigma-Aldrich (St. Louis, MO, USA). The efe gene [25] was codon optimised for S. elongatus PCC 7942 and synthesised from ThermoFischer Scientific (Waltham, MA, USA).

Strain Development Strategy
The markerless, efe expressing strain of S. elongatus PCC 11801 was developed using the CRISPR-Cpf1 technique as described previously [63]. The protocols in brief is given below.

Plasmid Construction
The pSL2680 vector formed the base plasmid for markerless genome editing [12]. The gRNA sequence was restriction digested and cloned at the AarI site of the pSL2680 ( Figure S1b) to generate pSL2680-gRNA, which targeted the neutral site I (NSI) sequence of S. elongatus PCC 11801. The resultant plasmid pSL2680-gRNA was transformed in E. coli XL1-Blue strain and was selected by blue-white screening. Next, the DNA elements comprising of the NSIa, 300 bp cpcB promoter, efe gene-T rrnB and NSIb were joined by Gibson Assembly [64] and cloned into the KpnI site of the pSL2680-gRNA vector and transformed into E. coli XL1-Blue. The resultant plasmid now contains all the elements for genome editing and was named as the editing plasmid pSL3338 ( Figure S1b). Due to the small size of the rrnB terminator (T rrnB ), the efe gene and the terminator (T rrnB ) were joined separately by overlap PCR before performing Gibson assembly. The NSIa and NSIb correspond to the 750 bp sequences upstream and downstream of the targeting site of gRNA. These plasmids were sequenced from Genewiz ® (South Plainfield, NJ). The plasmids were maintained in E. coli strain XL1-Blue.

Recombinant Strain Construction via Conjugation
The editing plasmid pSL3338 was transferred in S. elongatus PCC 11801 by conjugation as described previously [12]. Briefly, 6 mL of the exponential phase culture of S. elongatus PCC 11801 grown in BG11 medium [13], 4 mL culture of the conjugal strain HB101 pRL443 grown in LB with 100 µg/mL ampicillin and 4 mL culture of E. coli XL1-Blue strain containing pSL3338 cultivated in LB with 50 µg/mL kanamycin were centrifuged and washed gently. The washed E. coli and cyanobacterial cell pellets were mixed together gently in 200 µL of BG-11 medium and evenly spread on to a HATF nitrocellulose filter paper (MilliporeSigma, St. Louis, MO, USA) that was placed on a BG-11 5% LB agar plate without antibiotic, prior to the experiment. The spread plates were incubated for 24 h under 100 µE·m −2 ·s −1 light at 38 • C to obtain mat growth of cyanobacterial cells. The filters were then carefully transferred to BG-11 agar plate with 50 µg/mL kanamycin and incubated under same condition until the lawn growth disappeared and individual colonies appeared. The colonies were initially patched on 10 µg/mL kanamycin plates and then further re-streaked on 50 µg/mL kanamycin plate several times (7 streak) to facilitate chromosomal integration and segregation. Segregation was checked by PCR using confirmation primers NS1-CF-F/NSI-CF-R. Post segregation, the recombinant S. elongatus PCC11801 strain was cured of the editing plasmid pSL3338 to generate S. elongatus PCC 11801:NSI::P cpcB -efe-T rrnB (referred to as the SL3338 strain) strain by growing in BG11 media without antibiotic pressure, resulting in a clean markerless recombinant strain [63]. All the primers used in the study are listed in Table S1.

Strain Cultivation and Determination of Cell Biomass
The wild type S. elongatus PCC 11801 [6] and SL3338 strains were cultivated and maintained in 50 mL BG-11 media at 38 • C under 20 or 300 µE·m −2 ·s −1 of light intensity (LL and HL) and ambient air or 1% CO 2 (LC or HC) in a shaker incubator (Kuhner AG, LT-X, Switzerland). For cultures grown in Multi-Cultivator MC 1000-OD (MC 1000-OD, Photon Systems Instrument, Drásov, Czech Republic), ambient air or air with 1% CO 2 was bubbled at a rate of 100 sccm per tube. The light intensity varied from 20 to 600 µE·m −2 ·s −1 and was specified for each case. The SL3338 as well as wild type cells were grown under different experimental conditions to determine the dry cell weight. Cells were harvested at different time points, washed, dried in flat aluminium cups at 85 • C overnight and their dry weight was measured.

Optimisation of Conditions for Succinate Production
The SL3338 strain and the wild type S. elongatus PCC 11801 were grown under different conditions in shaker incubator and MC 1000-OD to optimise the succinate titre. The effect of carbon in sufficient and limited states on succinate production was tested in shaker incubator under 200 µE·m −2 ·s −1 light intensity with or without 50 mM bicarbonate. The light characterisation was performed under four different light intensities of 20, 50, 100 and 300 µE·m −2 ·s −1 in MC 1000-OD bubbled with 1% CO 2 . The effect of O 2 was investigated by cultivating the strains in different growth vessels (baffled flask, unbaffled flask and bottle) under optimal light and CO 2 levels for succinate production. The starting inoculum was maintained around 0.1-0.2 OD 730nm . For estimating succinate levels in the extracellular media, 1 m of culture was collected every day from each experimental setup, centrifuged and the supernatant was used for measuring the succinate titre using the succinate estimation kit from Megazyme (Bray, Ireland). The assay was performed according to the manufacturers' instructions. Samples were temporarily stored at −20 • C until ready for analysis. All experiments were performed in triplicates.

Effect of Nitrate Limitation on Succinate Production
The wild type and the recombinant cells were inoculated in 25 mL shake flask with an OD 730nm of 0.2 in 7 mL BG-11 0 media supplemented with three different concentration of NaNO 3 , 1.5 (termed as 1× or as present in the BG-11 medium), 0.75 (0.5×) and 0.375 g/L (0.25×) in shaker incubator at 120, under 20 µE·m −2 ·s −1 light and 1% CO 2 to study the effect of nitrate concentration on succinate production.

Estimation of Ethylene Production Using GC-FID
The recombinant and wild type S. elongatus PCC 11801 strains were grown under optimal growth conditions in shake flask or multicultivator and transferred to 120 mL air-tight rubber cock Hungate glass bottles to measure ethylene formation rates. The initial OD 730nm was adjusted to 0.1, 0.2 or 2.15 as specified and the total culture volume was adjusted to 20 mL using BG-11 media. The bottles were incubated in shaker incubator under 20 µE·m −2 ·s −1 light intensity. Ethylene accumulated in the head space of the culturing bottle was estimated using GC-FID (6890N Gas Chromatograph-Flame Ionization Detector, Agilent). Four hundred microlitres of gas from the head space of the bottle was sampled every hour using an air-tight syringe and was injected in GC-FID. An ethylene standard curve was prepared prior the experimentation for quantitative estimation of ethylene. The gas samples were separated in a Poropak N column having inner dimensions of 5 × 1/8" and detected by flame ionisation detector using argon as the carrier gas at a flow rate of 65 ml/min. The oven temperature was maintained at 100 • C, whereas the injector and detector temperatures were kept at 150 and 200 • C, respectively [65].

Isolation of Total RNA and Reverse Transcriptase (RT)-PCR of Selected Genes
RNA isolation kit from Sigma-Aldrich (St. Louis, MO, USA) was used for RNA extraction from cell pellets. The isolated total RNA was cleaned to remove genomic DNA contamination and then proceeded to convert the total RNA to cDNA by iScript Advanced cDNA synthesis kit from Bio-Rad Laboratories (Hercules, CA, USA) by following the manufacturer's instructions. Subsequently, the comparative transcript levels of the genes were determined by qualitative RT-PCR with 30 PCR cycles. The respective genes were amplified from the cDNA using respective set of primers (Table S1) and visualised on a 0.8% agarose gel.

Metabolite Pool Size Estimation of SL3338 Strain
S. elongatus PCC 11801 WT and SL3338 were grown under 1% CO 2 and 20 µE·m −2 ·s −1 in CO 2 incubator shaker at 38 • C. Metabolomic analysis was performed with cells that were at exponential growth phase (OD 730nm of 0.6) and stationary phase (4th day, OD 730nm of~11). Metabolite extraction was performed as described previously [55]. Briefly, the sample volume was adjusted to use 3 mg DCW. The culture was filtered rapidly through nylon membrane filters (Whatman, 0.8 µ) in the presence of light. To quench the metabolism, the biomass together with the filter was transferred quickly to 80:20 methanol:water, which was precooled to −80 • C. The metabolites were extracted as described [14,55] and the extracts were lyophilised and stored at −80 • C until ready for LCMS analysis. The metabolite extracts were reconstituted in 100 µL of water:methanol mixture (50/50, v/v) and filtered. Then 20 µL of the reconstituted metabolite extract was mixed with equal volume (20 µL) of metabolite extract fully 13 C-labelled biomass of S. elongatus PCC 11801, which served as the internal standard for every metabolite. Eight microlitres of the sample was injected on LCMS. The data were acquired using information dependent acquisition (IDA) method on a Q-TOF instrument (TripleTOF 5600, SCIEX, Framingham, MA) interfaced with Shimadzu Ultra Performance-Liquid Chromatography (UPLC) system (Shimadzu, Nexera LC −30 AD, Singapore). The instrument was operated under negative ion mode. Reverse phase ion-pairing chromatography was performed using C18 Synergi 4 µm Hydro-RP LC column 150 × 2 mm (Phenomenex Inc, Torrance, CA, USA). The gradient program and other instrument parameters were set as reported previously [14]. The relative quantification of metabolites was performed by normalising area under the peak for monoisotopic m/z of a particular metabolite by its respective 13 C-monoisotopic isotopologue present in the internal standard giving isotopic area ratio. The areas were estimated using using Multiquant 3.0.2 (Sciex, Framingham, MA, USA). The fold change of metabolite pools between the SL3338 and WT strains was calculated by dividing the isotopic area ratios (i.e., 12 C/ 13 C) of a metabolite in SL3338 by that of WT. These experiments were performed in three independent biological replicates with three technical replicates for each experiment (n = 3).

13 C Enrichment Analysis of SL3338 Strain
The 13 C labelling experiment was conducted as described previously [54] under ambient air and 20 µE·m −2 ·s −1 light intensity in an incubator shaker at 38 • C in a shake flask. Briefly, the labelling experiment was performed when the culture reached exponential growth phase (0.6-0.7 OD 730nm ). Twenty millilitre sample corresponding to the time (t = 0) (unlabelled) was collected. Then, the cultures were cultivated for 30 min to avoid dilution by unlabelled CO 2 into the system by restricting the exchange of air. Post incubation, 1g/L 13 C-Sodium bicarbonate was added to the system and 20 mL of culture were rapidly sampled at various timepoints (0, 0.5, 1, 1.5, 2, 3, 5, 10 and 30 min). The culture samples were filtered rapidly through a 0.8 µm Whatman nylon membrane filter (Catalogue No. 7408-004) procured from Merck Millipore (MilliporeSigma, St. Louis, MO, USA) in the presence of light and then quenched quickly by submerging the biomass together with the filter paper in chilled 80:20 methanol:water (precooled at −80 • C).
The extraction of metabolite and LCMS analysis was done as described in Section 3.8. The extracted ion chromatograms (XIC) of metabolites were visualised with PeakView 2.2 software and MasterView1.0 (Sciex, Framingham, MA, USA). The areas under the curve for the XICs corresponding to each isotopologue for each metabolite were estimated using Multiquant 3.0.2 (Sciex, Framingham, MA, USA) and then converted to mass isotopologue distribution (MID). All the experiments were performed in two independent biological replicates and two technical replicates for each set, n = 2. MIDs for each technical replicate were then corrected for natural abundance of the 13 C isotope using the IsoCor software, as described by Millard et al. [66,67]. Average values obtained and statistical analysis performed on IsoCor corrected data.

Determination of Photosynthetic Efficiency and Capacity
The maximal PSII quantum yield (F V /F M ) and the PSII photosynthetic capacity (ETRmax) was determined for the wild type and recombinant SL3338 strains in vivo by using a pulse amplitude modulated fluorimeter (Phyto-PAM, Heinz Walz, Germany). Briefly, 3 mL of exponentially grown culture (under optimal production condition in shake flask) was dark-adapted for 15-20 min before measurement under the same experimental growth temperature. The F 0 , Fm, Fv/Fm and ETRmax were determined with the saturation pulse method and Rapid Light Curve as per the instruction manual and previous reports [68,69].

Statistical Analysis
Statistical significance of the product titres and the metabolite levels between the engineered and wild type strains were examined by t-test. Metabolites with a fold change of ≥1.4 or ≤0.71 between the engineered and WT strains and p-value < 0.05 were considered as statistically significant.
The standard error of mean (SEM) of ratio of metabolite levels between the engineered and WT strains, denoted as X and Y, respectively, was calculated as follows. The variance of the ratio X/Y was first calculated using Taylor expansion method according to Equation (3) [70]: The fold change between two conditions is represented as: where σ denotes the standard deviation and σ 2 is the variance amd X and Y represents two conditions. X is engineered strain, whereas Y is wild type strain. X and Y are the respective mean of the metabolite levels in the X and Y conditions and cov(X,Y) implies the sample covariance. The number of replicates is denoted by n.

Conclusions
Metabolic engineering of cyanobacteria for the co-production of two essential products in two different phases might be economically advantageous. To that end, the EFE enzyme was expressed under a strong P cpcB7942 promoter in a fast-growing cyanobacterium, S. elongatus PCC 11801, to simultaneously produce ethylene and succinate. The best production condition was observed to be low light, high CO 2 , and this might be due to the enhanced activity of the promoter in use [41] or the activated TCA cycle [43]. The low light production strategy in MC 1000-OD ensured the accumulation of lower biomass with higher cumulative product titre. On the contrary, cultivation in shake flask limited the carbon partitioning towards the product of interest. Under optimal conditions, SL3338 strain co-produced succinate and ethylene at peak specific production rates of 338.26 and 1044.18 µmole/gDCW/h, respectively. These results obtained by expressing a single copy of the efe gene in S. elongatus PCC 11801 compare favourably with the reported productivities for strains that have been engineered either with multiple copies of efe or have additional genetic modifications. Further, metabolite profiling was performed to identify the potential bottlenecks under low light and 1% CO 2 in the shake flask. The metabolome analysis showed that the recombinant SL3338 strain during the early phase did not alter the relative pool sizes of CBB cycle intermediates but enhanced 13 C enrichment at later time points (pseudo isotopic steady state) towards the CBB metabolites. This suggests better utilisation of the inactive pools of CBB cycle intermediate, in turn improving the carbon capture efficiency of the cycle in recombinant strain. Since the glycogen accumulation was lower in engineered strain than the wild type, the increased carbon uptake was diverted towards the product of interest. A similar result was observed for the efe-expressing Synechocystis sp. Strain, which also showed improved flux diverted from the carbon fixation cycle towards ethylene production [30]. Though the recombinant strain increased the efficiency of carbon sequestration, SBPase/FBPase enzyme remains to be a major control point in the pathway. Beside SBPase, PDH and ppc enzymes have also been identified as the bottlenecks. As future work, tuning the potential metabolic control knobs identified in this study might help enhance the photosynthetic co-production of ethylene and succinate.
Supplementary Materials: The following are available online at http://www.mdpi.com/2218-1989/10/6/250/s1, 1. Supplementary Information containing figures and tables: Figure S1: The plasmid map for markerless genome editing of cyanobacterial genome using CRISPR-Cpf1: (a) the schematic showing the construct assembled using Gibson Assembly along with the respective primers; and (b) pSL3338; Figure S2: The Genome and RNA level confirmation; Figure S3: The biomass accumulation of the wild type and recombinant strains; Figure S4: The effect aeration on succinate levels was studied by growing the strain in different cultivation vessels: air-tight bottle, baffled shake flask and unbaffled shake flask; Figure S5: Estimation of glycogen accumulated in wild type and recombinant SL3338 strains under high light and 1% CO 2 in shaker incubator; Figure S6: Estimation of succinate levels and growth profile of recombinant SL3338 strain grown in MC 1000-OD at 1% CO 2 bubbled under high light intensity; Figure S7: The comparison of the growth profile (representative) of S. elongatus PCC 11801 when the culture was cultivated in BG11 media with or without buffered pH; Figure S8: Re-using biomass for ethylene production over a period of three days; Figure S9: The carbon partitioning between biomass and product of interest when cultivated in bottle; Figure S10: The mass isotopologoue distribution of the metabolites under study in wild type and the recombinant strains; Figure S11: The comparison of mass isotopologue of metabolites under study quantified from S. elongatus PCC 11801 and recombinant strain; Figure S12: Comparison of mean enrichment of metabolites between the wild type and recombinant strains after: (a) 10 min; and (b) 30 min; Figure S13: The microscopic image of the wild type and recombinant strains; Figure S14: Qualitative RT-PCR was performed for both the wild type and recombinant strains; Table S1: The list of primers employed in the study; Table S2: Detailed comparison of succinate and ethylene productivity reported in the literature for different cyanobacteria; Table S3: The Quantum efficiency (Fv/Fm) of the wild type and recombinant strain; Table S4: List of metabolites analysed in the study. 2. File S2: Table S5: The area ratios ( 12 C/ 13 C) for all the metabolites in the wild type and SL3338 strain of S. elongatus PCC 11801 under early phase (EP) (XLSX); Table S6: The area ratios ( 12 C/ 13 C) for all the metabolites in the wild type and SL3338 strain of S. elongatus PCC 11801 under late phase (EP) (XLSX).