Enhancing the light-driven production of d-lactate by engineering cyanobacterium using a combinational strategy

It is increasingly attractive to engineer cyanobacteria for bulk production of chemicals from CO2. However, cofactor bias of cyanobacteria is different from bacteria that prefer NADH, which hampers cyanobacterial strain engineering. In this study, the key enzyme d-lactate dehydrogenase (LdhD) from Lactobacillus bulgaricus ATCC11842 was engineered to reverse its favored cofactor from NADH to NADPH. Then, the engineered enzyme was introduced into Synechococcus elongatus PCC7942 to construct an efficient light-driven system that produces d-lactic acid from CO2. Mutation of LdhD drove a fundamental shift in cofactor preference towards NADPH, and increased d-lactate productivity by over 3.6-fold. We further demonstrated that introduction of a lactic acid transporter and bubbling CO2-enriched air also enhanced d-lactate productivity. Using this combinational strategy, increased d-lactate concentration and productivity were achieved. The present strategy may also be used to engineer cyanobacteria for producing other useful chemicals.

. Because of these concerns, biomass resources such as sugars are considered as the major substitutes for fossil resources 3 . However, using biomass as fossil substitutes leads to direct competition for resources between energy and food supplies. Therefore, it is necessary to develop biosynthetic processes which don't need to use edible biomass as feedstocks. Direct conversion of CO 2 to biofuels and carbohydrates using photoautotrophic organisms such as cyanobacteria can resolve the issues regarding both CO 2 emission and resources shortage simultaneously 1 . d-Lactate, an isomeric form of lactate, is used as basic feedstock of biodegradable polylactide, a well-known sustainable bioplastic material with lots of commercial applications 4 . d-Lactate is a chiral chemical, which is also used in the pharmaceutical industry and as a precursor for industrial chemicals such as cosmetics 5 . Moreover, its ester derivatives can be used to produce perfumes, coatings, adhesives, and printing ink, and have applications in the electronics industry [5][6][7] . Generally, d-lactate is produced by lactic acid bacteria from foods containing hexoses and pentoses or from sugar-containing raw materials [7][8][9] . To reduce utilization of food-related biomass resources, it is necessary to design cell factories that can directly use CO 2 as the carbon source. More importantly, the cell factories consume CO 2 , thereby relieving material shortage and retarding climate change 1,10 . As lactate biosynthesis in cyanobacteria is attractive, several studies have investigated lactate production in cyanobacteria [11][12][13][14][15][16][17][18] . However, engineering of cyanobacteria for d-lactate production is limited by relatively low productivity, although efforts have been made to enhance its biosynthesis through both genetic engineering and optimization of culture conditions 14,16,17 . Low productivity may be attributed to cofactor imbalance (insufficient supply of NADH), because cyanobacteria produce NADPH as the major carrier of reducing equivalents 19 . This hypothesis was confirmed by previous studies where soluble transhydrogenase (sth) was introduced into cyanobacteria during d-lactate synthesis 11,16,17 . Other studies using native NADPH-dependent enzymes were successful with high titers of target products 3,[20][21][22][23] . However, considering that bacterial NADH-dependent oxidoreductases are more abundant than NADPH-dependent ones, it would be interesting to reverse coenzyme specificity using protein engineering [24][25][26][27][28][29] . Recently, Angermayr et al. 13 reported that genetically engineered l-lactate dehydrogenase (coenzyme specificity changed from NADH to NADPH) resulted in increased l-lactate production in Synechocystis sp. PCC6803. Therefore, rational creation of efficient d-lactate dehydrogenase features a high preference for NADPH, and it is promising to utilize NADPH in cyanobacteria for d-lactate production.
In this study, a combinational strategy was used for construction of a cyanobacterial strain for d-lactate production. Firstly, the comparison of cyanobacterial genomes was performed. Then Synechococcus elongatus PCC7942 was selected as the host strain because it lacks l-and d-lactate dehydrogenases, ethanol dehydrogenase, and formate-lyase genes (GenBank ID, CP000100), and could grow to high density within enclosed bioreactors 30 . Secondly, the coenzyme specificity of the key d-lactate producing enzyme, LdhD, was switched from NADH to NADPH by protein engineering. Thirdly, codon usage of LdhD was optimized. Furthermore, as photoautotrophs, cyanobacteria generally lack transporters to move hydrophilic organic molecules across cell membranes 31 . Therefore, a lactate transporter was integrated into S. elongatus PCC7942. Finally, CO 2 bubbling was used to enhance d-lactate production by the constructed S. elongatus strain.

Results
Switching the coenzyme specificity of LdhD. Bacteria are important gene sources for cyanobacterial engineering. For example, bacteria possess a large number of NADH-dependent oxidoreductases, which are crucial enzymes in metabolic pathways. Unfortunately, the concentration of NADPH in cyanobacteria is much higher than that of NADH 19 ; this limits the application of NADH-dependent oxidoreductases in cyanobacteria. A previous study has shown that it is possible to reverse the coenzyme specificity of xylitol dehydrogenase (XDH) from NADH to NADPH using site-directed mutagenesis 25 . To design NADPH-dependent enzymes for cyanobacterial engineering, several key enzymes in bacterial pathways were analyzed. As shown in Table S1, there are sequence gaps around the putative coenzyme binding regions of the first four enzymes, but Asp 176 -Asn 180 in LdhD and its corresoponding regions in XDH and other enzymes are obviously homologous, indicating that aspartate, asparagine, and the hydrophobic residues are conserved. Therefore, these enzymes may be engineered to utilize NADPH as the preferred cofactor.
D-Lactate is a bio-based chemical that can be produced by fermentation. The key enzyme for d-lactate production in Lactobacillus bulgaricus ATCC11842 32 has the same discriminatory sites between NADH and NADPH (Asp 176 , Ile 177 , Phe 178 , and Asn 180 ) as XDH (Table 1). Thus, this enzyme was selected to investigate the applications of LdhD in cyanobacterial engineering. To evaluate the effect of single substitution mutation on cofactor specificity, four single mutants, LdhDn A (D176A), LdhDn R (I177R), LdhDn S (F178S), and LdhDn R2 (N180R), were constructed and were expressed in recombinant Escherichia coli BL21(DE3) ( Table S2). All four single mutations produced positive effects on NADPH kinetics, compared with wild-type LdhD, suggesting that these single substitutions might contribute independently to cofactor reversal in LdhD (Table 2). However, these single mutants still preferred NADH to NADPH.
To further increase the cofactor specificity of LdhD towards NADPH, a quadruple mutant LdhDn ARSdR (D176A/I177R/F178S/N180R) was generated and was expressed in E. coli (Table S2). As shown in Construction of d-lactate-producing S. elongates strains. The biosynthetic pathway of d-lactate uses pyruvate, a central metabolic intermediate that can be reduced to lactate. To engineer S. elongatus PCC7942 for d-lactate production, LdhDn ARSdR was then introduced into the strain to facilitate direct utilization of the NADPH pool (Fig. 1). The original enzyme, LdhD, was used to construct a control strain (Fig. 1). To enhance expression in Synechococcus, codon-optimized versions of the above two genes, termed as ldhDc and ldhD ARSdR , were also synthesized (Table S2) Table S3). To further enhance d-lactate production, LldP, a lactate transporter, was expressed under the same promoter ( Fig. 2e; Table  S3). Neutral site I of S. elongates PCC7942 chromosome was used to integrate the cassettes that contained in the constructed plasmids pYLW11, pYLW12, pYLW13, pYLW14, and pYLW24. The resulting strains were named as YLW01, YLW02, YLW03, YLW04, and YLW05, respectively (Table S3). Integration of the inserted genes into the chromosome was verified with PCR and DNA sequencing (Fig. 2f).

d-Lactate production from CO 2 by S. elongates.
To determine the optimal IPTG concentration required for d-lactate production, all the engineered Synechococcus strains were grown in the presence of 0.1, 0.5, 1, and 2 mM of IPTG. d-Lactate yields were highest at 1 mM IPTG for strains YLW01, YLW02, YLW03, YLW04, and YLW05 (titers of d-lactate were 101 ± 5.3, 104 ± 5.7, 362 ± 17.1, 452 ± 18.7, and 798 ± 30.3 mg/L, respectively; Figure S1a). d-Lactate synthesis was reduced when the concentration of IPTG was above 1 mM. Therefore, 1 mM IPTG was chosen as the optimal concentration for all subsequent experiments. In addition, reverse transcription (RT)-PCR was performed to investigate the expression of ldhD, ldhDc, ldhDn ARSdR , ldhD ARSdR , and lldP ( Figure S1b). The result revealed that the transcription of all genes was detectable at this IPTG concentration. As controls, the five mutant strains were cultured in the BG-11 medium 20 in the absence of IPTG. A small amount of d-lactate was detected in all the strains, indicating slight leaky expression of the LdhDs (data not shown). It is notable that the difficulty in obtaining the transformants (YLW03, YLW04, and YLW05) harboring ldhDn ARSdR and ldh-D ARSdR proved to be quite difficult, suggesting that leaky expression of LdhDn ARSdR might have resulted in low growth rate and transformation efficiency.  YLW01, YLW02, YLW03, YLW04, and YLW05 were cultured for d-lactate production under constant light exposure (100 μE·s −1 ·m −2 ); the wild type strain S. elongates 7942 was used as the control. After induction for 10 days, d-lactate was not detected in the wild type strain (Fig. 3a). Upon introduction of NADPH-utilizing LdhDn ARSdR , d-lactate production increased by 3.6-fold and 4.2-fold in YLW03 (37.9 mg/L per day) and YLW04 (46.1 mg/L per day), compared with YLW01 and YLW02 (which harbor native LdhD), respectively (Fig. 3a). The enzymatic activities of the LdhDs in crude S. elongatus cell lysate were estimated to confirm the expression of the introduced lactate dehydrogenase genes. As shown in Table 3, although high LdhD activity for NADH was detected in both YLW01 and YLW02, d-lactate production remained low (Fig. 3). This may be attributed to the insufficient concentration of intracellular NADH in cyanobacteria. Conversely, although the activity of LdhDn ARSdR in YLW03 and YLW04 was significantly low, d-lactate production was enhanced in these strains, attributable to the abundant NADPH pool for LdhDn ARSdR in these strains. These results are consistent with a recent report that l-lactate productivity was enhanced by introducing a mutated l-lactate dehydrogenase that could co-utilize NADPH 13 . In addition, to determine the effect of codon optimization on d-lactate production, the relative protein expression profiles of LdhDs were measured. Higher protein levels were observed in YLW02 and YLW04 (~0.80 and ~0.38 × 10 −1 mg/mg in YLW02 and YLW04, respectively; compared with ~0.56 and ~0.13 × 10 −1 mg/mg in YLW01 and YLW03, respectively). However, d-lactate production increased only by 1.04-and 1.21-fold in YLW02 and YLW04, compared with YLW01 and YLW03, respectively ( Table 2; Table 3). This indicated that codon optimization of ldhD and ldhDn ARSdR increased d-lactate production only marginally. Overall, our results reinforce the importance of using cofactor-altered LdhDn ARSdR for d-lactate production in S. elongatus PCC7942.
It is notable that the growth rate of strains containing LdhDn ARSdR was different from that of other strains. There was no significant difference in cell growth among the wild type, YLW01, and YLW02 strains, which had not reached stationary phase at the tenth day and seemed to be able to continue (Fig. 3b). On the other hand, the cell growth rate of strains YLW03, YLW04, and YLW05 was impaired (maximum OD 730 values of 0.95, 0.99, and 1.07, respectively). A similar phenomenon was also observed in previous reports in which a soluble transhydrogenase or NADPH co-utilizing l-lactate dehydrogenase was introduced 11,13 (Fig. 3b).
The hydrophobic cell membrane is the main barrier for the production and secretion of hydrophilic products such as lactate by genetically engineered cyanobacteria 33 . An l-lactate transporter, LldP, has been described as a nonspecific d-lactate transporter that efficiently transports d-lactate by using proton motive force in E. coli and cyanobacteria 16,34 . With the expression of the additional lldP gene, YLW05 secreted 829 mg/L of d-lactate in 10 days with an average productivity of 82.9 mg/L per day (Fig. 3a). Although the activity of LdhDn ARSdR was lower in YLW05 (compared with that in YLW04), probably because of downregulation of the two genes upon co-expression (Table 3), d-lactate titer in YLW05 was approximately 1.8-fold higher than that in YLW04. This indicates that the transporter efficiently translocated lactate in YLW05, thereby improving d-lactate productivity. Effect of aerating CO 2 on d-lactate production. In order to test if lactate production by the Synechococcus mutant strain could be enhanced, strain YLW05 was grown in a bubble column photobioreactor by continuously aerating CO 2 -enriched air (5%, v/v; Fig. 4). As expected, aeration with CO 2 increased d-lactate production in YLW05 (~1.6-fold), reaching a titer of 1.31 g/L in 10 days with a maximum productivity of 221 mg/L per day. Moreover, d-lactate production did not cease after the ninth day, although productivity decreased slightly. As for cell growth, the YLW05 culture with CO 2 -enriched air exhibited slight increase in cell density ( Fig. 3; Fig. 4). To further simulate natural production conditions, S. elongatus mutant YLW05 was maintained at alternating dark and light periods (at an interval of 12 h) instead of constant light exposure. Under the conditions employed, cell growth was limited to the light period, and cell density decreased slightly in the dark period (Fig. 4b). This result suggests that d-lactate production in Synechococcus strains might be promoted by light and inhibited in the dark. This hypothesis is consistent with a previous report stating that cyanobacterial cultures accumulate polysaccharides when they are exposed to light, and they mobilize these intracellular reserve materials in the dark 35 . Finally, strain YLW05 produced a mere 563 mg/L of d-lactate (maximum production rate is 75.1 mg/L per day) after 10 days (Fig. 4b). Nevertheless, these results (productivity of 56.3 mg/L per day) under the day-night cycle conditions suggest that Synechococcus strains may be applied to d-lactate production.

Discussion
Cofactor preference of enzymes is important for microbial organisms to produce metabolites 36 . In this study, to directly use the abundant NADPH pool in cyanobacteria for d-lactate production, a cyanobacterial cell factory was designed by introducing an NADPH-utilizing enzyme, LdhDn ARSdR . Significant changes in the kinetic constants of LdhDn ARSdR suggested that the increased d-lactate productivity in YLW03 and YLW04 might stem from the increased activity with NADPH (Table 2). Multiple strategies were tested to optimize lactate production. Altering the cofactor preference of LdhD resulted in over 3.6-fold increase while introducing the transporter, LldP, resulted in approximately 1.8-fold increase, and bubbling CO 2 resulted in approximately 1.6-fold increase, in d-lactate production. These results show that altering cofactor specificity contributes mostly in enhancing the d-lactate production, which    indicates the feasibility of altering the cofactor specificity. In addition, altering the cofactor preference of an existing enzyme has the following possible advantages. First, the cofactor-altered enzyme could directly utilize NADPH, which might be more efficient for product synthesis. Second, compared to the use of a transhydrogenase, it simplifies the metabolic pathways using just one enzyme. Third, altering cofactor specificity might be faster than the process of identifying NADPH-dependent enzymes. Use of the cofactor-altered LdhDn ARSdR resulted in impaired cell growth. This might be attributed to the high rate of d-lactate production, resulting in decreased NADPH level and activation of the oxidative pentose phosphate cycle. This cycle is the major route of carbon metabolism in cyanobacteria 37 . To confirm this hypothesis, the intracellular levels of NADPH/NADH in both wild type S. elongatus PCC7942 and mutant strains were determined during the cultivation process. The ratio of NADPH/NADH slightly decreased in strain YLW04, compared with that in S. elongatus PCC7942 ( Figure S2). This suggested that the NADPH/NADH ratio altered in the mutant strains, which might have affected the cell growth of strains YLW03, YLW04, and YLW05 (Fig. 3b), respectively. Another possible reason is that redirection of carbon flux from cellular biomass toward synthesis of d-lactate disrupts cell growth. Therefore, the intracellular pyruvate concentration in the wild type and mutant strains was measured. Pyruvate concentration was slightly higher in the wild type strain than that in YLW04 ( Figure S3). To overcome this problem of attenuated cell growth, it is necessary to maintain the balance between growth and lactate production by precisely controlling the expression of mutated ldhD. This result is also consistent with the above IPTG concentration optimization.
As photoautotrophs, cyanobacteria lack many of the transporters found in E. coli or yeast 16 . In two previous studies in which a transporter was introduced into a cyanobacterium for the secretion of lactate, significant improvement in lactate production was observed 15,16 . Here, strain YLW05 expressing the ldhD ARSdR and lldP genes secreted relatively high levels of d-lactate into the medium, suggesting that integration of the lactate transporter aids lactate secretion. Moreover, the substrate transport process was mediated by proton translocation, resulting in the accumulation of H + -a necessary material for the synthesis of NADPH 34 . Therefore, introduction of LldP might contribute to the high yield of d-lactate in strain YLW05 by promoting NADPH production, which can then be used by LdhDn ARSdR .
Biosynthesis of d-lactate from CO 2 has been achieved and is characterized in cyanobacteria, such as Synechocystis sp. PCC6803, through genetic engineering [11][12][13][14][15][16][17][18]32 . As shown in Table S4, both Hollinshead et al. 14 and Varman et al. 17 have reported the enhanced d-lactate production using Synechocystis sp. PCC6803 by adding acetate as an organic carbon source. Herein, the concentration and average productivity of d-lactate increased by approximately 90% using S. elongatus strain YLW05, compared with strain AV10 14,17 , within 10 days. It should be noted that the production in this case was purely photosynthetic. Apparently, the titer values and average productivity of YLW05 were considerably higher than those of other reported strains (10 days), without the addition of an additional carbon source. Based on the above results, it is reasonable to conclude that our combinational strategy for the production of d-lactate might be more effective. Furthermore, to examine whether S. elongates PCC7942 was superior, the actual partitioning of carbon between cellular biomass and d-lactate production was evaluated at the late log phase of growth (6 to 8 days for YLW05; 18 to 21 days for AV10). The results revealed that the values for strains YLW05 and AV10 were approximately 80.7 mg/L/OD 730 and 5.9 mg/L/OD 730 per day, respectively. This result suggests that strain YLW05 might be more efficient than strain AV10 17 .
In summary, LdhD, a key enzyme in the d-lactate production pathway, was successfully engineered for cofactor reversal, and was used in engineered cyanobacteria for efficient production of d-lactate. Other methods, including introducing a lactate transporter and optimizing codon usage were also adopted in the construction. Under conditions of constant light exposure and bubbling CO 2 -enriched air, the resulting strain (YLW05) achieved the highest lactate concentration and productivity reported for engineered cyanobacteria within 10 days (Table S4). This indicates that the systematic combination of different methods is promising in cyanobacteria engineering. This method of cyanobacterial engineering might have applications in the efficient biosynthesis of other chemicals as well.

Methods
Chemicals and reagents. The d-lactate standard, NADH, NADPH, and isopropyl-β-d-thiogalactoside (IPTG) were obtained from Sigma-Aldrich (St. Louis, MO). Oligonucleotides and gene synthesis were carried out by Sangon Biotech Co., Ltd. (Shanghai, China). All other chemicals and reagents were of at least analytical grade and were available commercially.
Strains and growth conditions. Lactobacillus bulgaricus ATCC11842 and Escherichia coli K-12 strain MG1655 were used as the sources of ldhD (GenBank no. 103422405) and the l-lactate transporter gene (lldP) (GenBank no. 1790031), respectively. E. coli strain DH5α was used as the host for vector construction. S. elongates PCC7942 (ATCC33912) was from ATCC (American Type Culture Collection). The S. elongates strain was cultured in the BG-11 medium 20 unless otherwise stated, and cells were incubated statically, at 30 °C and at an illumination intensity of 100 μE·s −1 ·m −2 , as described elsewhere 14 . Cell growth was monitored by measuring the optical density at 730 nm (OD 730 ).
For d-lactate production, S. elongatus cells in the exponential phase were diluted to 0.05 (OD 730 ) in 100 mL BG-11 medium containing 20 mg/L spectinomycin in 300 mL flasks. Cultures were induced with a suitable concentration of IPTG after growing to an OD 730 of 0.4-0.6. Daily, 1 mL of each sample was collected for analysis, and equivalent BG11 was supplemented.
Site-directed mutagenesis and plasmid construction. All primers used for plasmid construction are listed in Table S5. The constructed plasmids are listed in Table S3. A neutral site I (NSI) in S. elongates PCC7942 chromosome was used for inserting an expression cassette.
lldP was obtained via PCR amplification from E. coli MG1655 using two primers, lldP_3 F and lldP_2 R, it was cloned into the XhoI/BamHI site of pYLW14, resulting in plasmid pYLW24. The Shine-Dalgarno (SD) sequence of lldP was obtained from E. coli MG1655.
Transformation of Synechococcus. Transformation of Synechococcus host cells was carried out by using a double homologous-recombination procedure as previously described 16 . Integration of vectors into neutral site I was verified by PCR using gene-specific primers (Table S5) to demonstrate the presence of appropriate novel chromosome-transgene junctions and the absence of uninserted sites. The genetic stability of the mutant strains was evaluated by the serial subcultivation. Table S3 lists the strains that were constructed and used in this study. Briefly, mutant strains were obtained via integrating the aimed DNA fragments harbored by plasmids pYLW11, pYLW12, pYLW13, pYLW14, and pYLW24 to Synechococcus chromosome, respectively.

Culture conditions of Synechococcus cells.
To investigate the effect of the initial concentration of IPTG on d-lactate production, IPTG concentration was adjusted in the culture media to 0.1, 0.5, 1, and 2 mM. To determine the effect of aeration on d-lactate production, a bubble column photobioreactor equipped with a glass column was used. S. elongatus strains were separately suspended in BG11 medium by aerating CO 2 -enriched air under constant light exposure as described in a previous study 10 . To study the effect of the day-night cycle on d-lactate production, S. elongatus strain was grown under the aerating condition with day and night periods that alternated every 12 h.
Enzyme assays. S. elongatus cells were harvested via centrifugation (6,000 × g, 5 min) 72 h after induction, washed twice with 50 mM Tris-HCl buffer (pH 7.0), and resuspended in the same buffer containing 2 mM dithiothreitol. Crude extracts were prepared via bead beating 22 . Total protein concentration was determined according to the method of Bradford 10 using bovine serum albumin as the standard. The standard reaction mixture (1 mL) contained 50 mM Tris-HCl buffer (pH 7.0), 0.2 mM NAD(P)H, and 0.05 mM pyruvate. One unit of protein activity was calculated as micromoles of pyruvate consumed per minute per milligram of the total protein at 30 °C.
To characterize the kinetic constants of LdhDn ARSdR after the reversal of coenzyme specificity, the enzyme was expressed in E. coli BL21(DE3), with the wild-type LdhD as a control. The purification of the two enzymes were performed using the method of Wang et al. 40 The reduction activity of purified LdhD and LdhDn ARSdR on pyruvate were assayed at 30 °C. Oxidation of NADPH/NADH (340 = 6220 m −1 cm −1 ) was monitored by the decrease in absorbance at 340 nm 41 . One unit of protein activity was defined as the amount enzyme that catalyzed the consumption of 1 μmol pyruvate per minute. The reaction mixture (1 mL) contained 50 mM Tris-HCl buffer (pH 7.0), 0.2 mM NAD(P)H, and different concentrations of substrate. The Michaelis-Menten equation was used for determination of the kinetic parameters. To determine the stability of LdhDn ARSdR , the purified LdhDn ARSdR were incubated at 30 °C for 24 h.

Reverse transcription PCR (RT-PCR).
RT-PCR was performed as previously described 21 . Total RNA of the various cyanobacteria strains was extracted using an RNAprep Pure Cell/Bacteria Kit (TIANGEN