Light Driven CO2 Fixation by Using Cyanobacterial Photosystem I and NADPH-Dependent Formate Dehydrogenase

The ultimate goal of this research is to construct a new direct CO2 fixation system using photosystems in living algae. Here, we report light-driven formate production from CO2 by using cyanobacterial photosystem I (PS I). Formate, a chemical hydrogen carrier and important industrial material, can be produced from CO2 by using the reducing power and the catalytic function of formate dehydrogenase (FDH). We created a bacterial FDH mutant that experimentally switched the cofactor specificity from NADH to NADPH, and combined it with an in vitro-reconstituted cyanobacterial light-driven NADPH production system consisting of PS I, ferredoxin (Fd), and ferredoxin-NADP+-reductase (FNR). Consequently, light-dependent formate production under a CO2 atmosphere was successfully achieved. In addition, we introduced the NADPH-dependent FDH mutant into heterocysts of the cyanobacterium Anabaena sp. PCC 7120 and demonstrated an increased formate concentration in the cells. These results provide a new possibility for photo-biological CO2 fixation.


Introduction
Formate is an important energy carrier in the bacterial kingdom. Biological formate is produced by the degradation of pyruvate, amino acids, l-(+)-tartaric acid, oxalate, and hypoxanthine [1,2]. It can also be produced by the direct reduction of CO 2 catalyzed by formate dehydrogenase (FDH), and this approach is increasingly gaining attention [3].
FDHs consist of several groups of enzymes that vary significantly in their quaternary structure, occurrence, and type of prosthetic groups. One such group is the NADH-dependent FDH (EC1.2.1.2) (NADH-FDH), which catalyzes the oxidation of formate to CO 2 and the reduction of CO 2 to formate coupled with the oxidation-reduction of NAD + /NADH (HCOO -+ NAD + ↔ NADH + CO 2 ) [4]. NADH-FDH has previously been applied to light-driven formate production systems by using Zn-porphyrin or TiO 2 particles as a photosensitizer [5][6][7]. In these systems, electrons are supplied by photosensitizers and transported via mediators such as methyl viologen, to diaphorase reducing NAD + to NADH. The NADH-FDH then catalyzes the CO 2 reduction by using the resulting NADH to produce formate.
Based on the results of previous studies, we designed a new biological system ( Figure 1). In our system, the ability of cyanobacteria, algae, and plants to effectively produce NADPH using a photosynthetic light reaction [8,9] was taken as the basis for the design, in order to produce formate by using NADPH as a substrate for FDH. All the components are biologically producible molecules and can, therefore, be regenerated within the cell, enabling the creation of a sustainable formate production mechanism.
To create a light-driven formate production system in vivo, we focused on cyanobacterial heterocysts, which carry out a strictly anaerobic reaction, i.e., nitrogen fixation [10]. Heterocysts differentiate from vegetative cells, which carry out oxygen-evolving photosynthesis, and are deficient in PS II activity. Their internal environment is, therefore, micro-oxic and oxygen-sensitive nitrogenase is expressed in the cells. The electrons necessary for the nitrogen fixation are supplied from neighboring vegetative cells in the form of sucrose. Sucrose is degraded into CO 2 , protons, and electrons, and the electrons are photo-excited by PS I and transferred to nitrogenase to reduce nitrogen [10]. The micro-oxic and reductive environment makes heterocysts an ideal reaction vessel for light-driven formate production by FDH.
We first created NADPH-dependent FDH (NADPH-FDH) because all the known NADH-FDHs are specific for NADH. In a previous study, site-saturation mutagenesis at the amino acid residues determining the cofactor specificity of NADH-FDH were examined and some mutants, in which the cofactor preference in the degradation of formate was shifted from NAD + to NADP + , were found [11]. However, these studies did not report any data on the productivity of formate and its stability. Here, we prepared several mutants of FDHs from Pseudomonas sp. 101 (Pseudomonas), Candida boidinii (Candida), potato, Arabidopsis thaliana (Arabidopsis), and Thiobacillus sp. KNK65MA (Thiobacillus), and investigated their formate productivity from NADPH and CO 2 . We then verified our formate production system in vitro and in heterocysts by using the NADPH-FDH with the highest formate productivity.

Bacterial strains and Culture conditions
Escherichia coli (E. coli) DH5α (RBC Bioscience, New Taipei City, Taiwan) and E. coli BL21 (DE3) (Merck, Rahway, New Jersey) were used for the genetic manipulation and expression of recombinant FDH, respectively. For protein expression, 0.5 mM (as a final concentration) of isopropyl-l-dthiogalactopyranoside (IPTG; Nacalai tesque, Kyoto, Japan) was added to each Luria-Bertani (LB) medium. The filamentous heterocystous cyanobacterium Nostoc sp. strain PCC 7120 (also called Anabaena sp. strain PCC 7120) (hereinafter, Anabaena) was purchased from the Pasteur Culture Collection of Cyanobacteria (France). Anabaena and the transformed cells were grown at 30 • C under a light intensity of 30 µE•m -2 •s -1 in the BG11 medium as previously described [12]. Liquid cultures were bubbled with air. For nitrogen deprivation experiments, cells grown in the BG11 medium until they reached an OD 600 of 0.4-0.5 were washed with the nitrogenfree medium (BG11 0 ) and then re-suspended in the BG11 0 medium. Neomycin was added to the medium at a final concentration of 30 µg/mL, when required.

Plasmids
The synthesized genes coding for FDHs from Pseudomonas, Candida, potato, Arabidopsis, and Thiobacillus were subcloned using NdeI and NotI sites in pET22b (Merck) as C-terminal Histagged proteins. According to previous studies [11], the amino acid residues determining the cofactor specificity of NADH-FDH were suggested to be at 222 and 224 in Pseudomonas, 197 and 199 in potato, 227 and 229 in Arabidopsis, and 222 and 224 in Thiobacillus. The nucleotides corresponding to the above pairs of amino acids were substituted with those corresponding to Glu/Ser/Ala/Gln and Asn/His, respectively, by means of PCR site-directed mutagenesis using mixed primers, 5′-GTTCATCTGCACTATACC(C or G or T)(A or C)GCGT(A or C)AC CGCCTGCCGGAAAGCGTG-3′ for Pseudomonas, 5′- CTGCAATCTGCTGTACCAC(C or G or T)(A or C)GCGT(A or C)ACAAAATGGATTCTGAACTG-3′ for potato, 5′-GTTGCAACCTGCTGTACCAT(C or G or T)(A or C)GCGT(A or C)ACCAGATGGCACCGGAACTG -3′ for Arabidopsis, and 5′-GTTAAACTGCATTATACC(C or G or T)(A or C)GCGT(A or C)ACCGTCTGCCGGAAGCAGTG-3′for Thiobacillus, and their complementary primers (underlines indicate codons corresponding to amino acid residues determining the cofactor specificity). The introduction of mutations into Candida gene was performed using non-mixed primers, 5′-CGAAAGAACTGCTGTATTACCAGCGTAATGCCCTGCCGAA AGAAGCAG-3′ and its complementary primer (underlines indicate codons corresponding to mutation sites (Asp195 → Gln, Tyr196→Arg and Gln197→Asn)).
For the expression of NADPH-FDH in Anabaena heterocysts, we used a hetR promoter and a shuttle vector, pAM505, which contained replicons of E. coli and Anabaena [12,13]. A fragment coding for the NADPH-FDH gene was amplified from pET22b harboring a gene coding for a Pseudomonas FDH mutant, PsFDH[QN] (please see Results and Discussion), using primers, 5′-AAACTGGAGACTCATAATGGCGAAAGTGCTGTG-3′, and 5′-AATAATTGAGCTCTCAGTGGTGGTGGTGGTGGTGTTCGAG TGCGGCCGC-3′. The hetR promoter (approximately 1000 bp upstream of hetR coding region) was amplified from Anabaena genome according to Wang et al. [14]. The fragments were then digested and inserted into pAM505 using SalI and SacI sites, resulting in pAM505-fdh. Furthermore, the FNR gene was inserted into the downstream of the FDH, generating a vector for co-expression of FDH and FNR, pAM505-fdhfnr.
The helper plasmid pRL623, which harbors a mob gene and methylase genes [15] and a self-transmissible plasmid RP4 [16] were provided from Prof. Wolk (Michigan State University).
Introduction of the shuttle vector into Anabaena. Plasmid pAM505-fdh was first introduced into E. coli HB101 carrying a helper plasmid pRL623 [15]. The resultant transformed cells were mixed with Anabaena and E. coli J53 carrying a self-transmissible plasmid RP4 [16][17][18], and the exoconjugant Anabaena carrying pAM505-fdh (Anabaenafdh) was selected on a BG11 agar plate containing neomycin. For co-expression of PsFDH(QN) and FNR, we introduced pAM505-fdhfnr into Anabaena and isolated the exoconjugant Anabaena-fdhfnr similarly to that described above.

Purification of proteins
The FDH proteins, Pc, Fd and FNR were purified under aerobic conditions using a His-Accept Ni-chelating resin (Nacalai tesque). The 50 mM Tris buffer (pH 7.6) containing 500 mM NaCl was used as binding and wash buffer. The binding buffer containing 500 mM imidazole was used for the elution of His-tagged proteins.
Purification of PS I was conducted essentially as previously described by Nakamoto and Hasegawa [19] except for the solubilization condition. To solubilize PS I from thylakoid membrane, the membrane was suspended in the buffer containing 1.6% dodecyl-β-d-maltoside, 50 mM Tris-HCl (pH8.0), 10 mM NaCl and 10% sucrose.

Assays
The formate degradation activities of the FDH proteins were measured spectroscopically in 2 mL of phosphate-buffered saline (PBS) containing 1 mM NADP + or NAD + , 100 mM formate, 0.05% Triton X-100, and 5 µg/mL FDH. The reaction was initialized by the addition of a FDH sample and the change in the absorption of NADPH (or NADH) was monitored at 340 nm (the extinction coefficient is 6.22 mM -1 cm -1 ). The analysis was performed in triplicate. The Michaelis-Menten parameters were determined from the NAD(P) + dependence of initial rates of NAD(P) H production in PBS containing 0.05-2 mM NAD(P) D + , 500 mM formate, 0.05% Triton X-100, and 5 µg/mL FDH, and from the formate dependence obtained in the same conditions, except for the NAD(P) D + and formate concentrations being constant (2 mM) and variable (10-500 mM), respectively.
The formate formation reaction was carried out under strictly anaerobic conditions. A tightly sealed branched flask was used with FDH placed in one arm and PBS, NADPH, and Triton X-100 in the other. The 2 solutions were degassed by repeated cycles of evacuation and argon gas purging, followed by charging with 100% CO 2 for 1 min at the final cycle. The reaction was initiated by mixing the 2 solutions at 25° C. The final concentrations of FDH, NAD(P) H, and Triton X-100 were 1.7 mg/ml (38 µM), 1 mM, and 0.05%, respectively. The concentration of the dissolved CO 2 in the buffer was measured using a CO 2 sensor, CGP-31 (TOA DKK, Tokyo, Japan), and it was observed to be between 1.0 and 1.3 mg/ml. The concentration of formate was determined by withdrawing 100-µl aliquots from the solution without exposure to O 2 at intervals of 60 min, and subsequently analyzing them with ion chromatography by using a HPLC system equipped with a Shim-pack SCR-102H column (300 mm length × 8.0 mm φ; mobile phase, 4 mM p-toluenesulfonic acid; column temperature, 40° C) and a conductivity detector, CDD-10A VP (Shimadzu, Kyoto, Japan). The analysis was performed in triplicate.
Light-dependent formate production was performed in 3 mL PBS by using isolated PS I at a concentration of 100 µg chlorophyll/mL, 10 mM ascorbate, 10 µM PC, 2 µM Fd, 0.5 µM FNR, 1 mM NADP + , 4 mM NADPH, 2.5 mg/mL (56 µM) FDH mutant (PsFDH[QN]), 0.05% Triton X-100, and 10% sucrose. The solution containing all the components was degassed by repeating cycles of evacuation and argon gas purging, followed by charging with 100% CO 2 for 1 min during the final cycle. The reaction was initiated by illuminating with visible light (<420 nm) at an intensity of 1000 µmol (photon) m -2 s -1 and monitored by sampling at regular intervals without exposure to O 2 . The analysis was performed in triplicate.
Intracellular formate levels were estimated using the extracts from Anabaena-WT, Anabaena-fdh and Anabaena-fdhfnr cell pellets. Anabaena-WT, Anabaena-fdh and Anabaena-fdhfnr were initially grown in BG11 medium by bubbling with air under a fluorescent lamp at an intensity of 15 µmol photon m -2 s -1 , and the cells in the exponential growth phase (OD 700 ~3) were washed and transferred into BG11 0 medium and cultured for a further 36 h. The formation of heterocysts was confirmed microscopically. The media were then degassed and bubbled with air, air + 10% CO 2 , N 2 + 10% CO 2 , or argon + 10% CO 2 , for 12 h under relatively strong light (150 µmol photon m -2 s -1 ). The cultures were transferred anaerobically to the centrifuge tubes in a globe box, centrifuged at 20,000 × g and lysed by grinding with glass beads in 0.1N HCl. The lysates were adjusted to pH 3, centrifuged, and subjected to HPLC analysis. The cultivation and analysis were performed in triplicate.

Conversion of the cofactor specificities of FDH mutants
To engineer NADH-FDHs, the FDH genes from Pseudomonas, Candida, potato, Arabidopsis, and Thiobacillus were subcloned into pET22b and mutated at the positions corresponding to the amino acid residues determining the cofactor specificity (i.e., 222 and 224 in Pseudomonas, [195][196][197]197 and 199 in potato, 227 and 229 in Arabidopsis, and 222 and 224 in Thiobacillus). The resulting plasmids were transformed into E. coli BL21 (DE3). Cells were grown in LB medium at 37° C and induced with 0.5 mM IPTG. FDH mutants were expressed as a soluble and apo form and purified under aerobic conditions. SDS-PAGE analyses indicated that purified FDH samples from Arabidopsis and Thiobacillus contained additional bands at a slightly lower molecular weight besides an expected band, indicating the partial digestion during the purification; however, those from Pseudomonas, Candida and potato were purified to homogeneity.
Cofactor specificities of FDH mutants were first evaluated by measuring the NAD + or NADP + reduction rate. FDH mutants from Arabidopsis and Thiobacillus tended to aggregate and become inactivated during the reaction; however, those from Pseudomonas, Candida, and potato were relatively stable. Several mutants switched the cofactor specificity from NADH to NADPH (Table 1) Cofactor specificities of PsFDH(QN), CbFDH(QRN), and PoFDH(QH) for formate production were determined for the following reaction: NAD(P)H + CO 2 → HCOO -+ NAD(P) + . Because this reaction is inhibited by gaseous oxygen, it was performed under strict anaerobic conditions (100% CO 2 atmosphere). Consequently, we observed that the reactivity of PsFDH(QN) with NADPH was 5 times that with NADH, and produced formate at 32 µM/h, corresponding to a specific activity of 19 nmol (mg protein) -1 h -1 , which was twice the amount produced in the reaction of the wild-type protein with NADPH ( Figure 2). The reaction of PsFDH(QN) was also carried out in the presence of a high concentration of NADPH (10 mM), but it resulted in only a 1.3-fold activity compared to when 1 mM NADPH was used (closed diamonds in Figure 3). This suggested that the cofactor binding sites were almost saturated with NADPH at 1 mM concentration.
To confirm the photosensitivity of PsFDH(QN), the reaction was conducted under visible light (<420 nm) at an intensity of 1000 µmol photon m -2 s -1 . No significant change was observed, demonstrating an insensitivity to light (closed squares in Figure  3). When instead of charging with CO 2 , NaHCO 3 was added to the phosphate buffer (final concentration 50 mM) with NADPH, and Triton X-100, followed by degassing and mixing with the FDH solution, the rate of formate production was significantly lower (crosses in Figure 3). The observed concentration of dissolved CO 2 in the buffer (0.040-0.060 mg/mL) was significantly lower than that in the solution charged with CO 2 (1.1-1.3 mg/mL), which explains why no significant formate production from NaHCO 3 was observed. The significant low CO 2 concentration in the buffer added with NaHCO 3 was mainly attributed to the degassing process.

In vitro light-driven formate production
Using PsFDH(QN), a light-driven formate production system was constructed in vitro. The system consisted of PS I (100 µg chlorophyll/mL) as photosensitizer; 10 mM ascorbate as a sacrificial electron donor; 10 µM PC as an electron carrier at Table 1. Cofactor specificities of the WT and mutant FDH proteins.   Figure 4), formate production was observed at a rate of 5 µM/h, which was significantly lower than that under illumination. In the reaction under a 10% O 2 and 90% CO 2 gas atmosphere (open triangles in Figure 4), formate was not detected, demonstrating the O 2 -sensitivity of the system. The reactions were carried out in PBS containing 0.05-2 mM NAD(P) + 500 mM formate, 0.05% Triton X-100, and 5 µg/mL FDH or in the same conditions, except for the NAD(P) D + and formate concentrations being constant (2 mM) and variable (10-500 mM), respectively. Michaelis-Menten parameters were determined by Lineweaver-Burk plots. ND indicates the case in which the data could not be fitted to Lineweaver-Burk plots.   The rate-determining factor was then assessed using the total activity of PsFDH(QN) and the rate of supply of NADPH. The total activity of 7.5 mg of PsFDH(QN) in 3 mL of reaction solution was estimated to produce 47 µM formate/h using its specific activity of 19 nmol (mg protein) -1 h -1 as described above. This rate was in the same range as the maximum rate of formate production at 30-60 min after initiation. On the other hand, the total activity of the 0.3 mg chlorophyll PS I in 3 mL solution was estimated to be 15 mM NADPH/h using its specific activity of 150 µmol (mg chlorophyll) -1 h -1 which was determined in the same conditions except for no FDH protein and no NADP + being added and the concentration of PS I being 5 µg chlorophyll/mL. However, because the total NADP(H) concentration was 5 mM, almost all NADP + was expected to be reduced to NADPH. In fact, the NADPH concentration in the control reaction solution without PsFDH(QN) was analyzed spectrophotometrically and estimated to reach 4.5 mM (NADPH/NADP + ~ 9), which is high enough to push the equilibrium towards formate production, within 60 min after the initiation of the reaction under anaerobic conditions. The time required for all NADP + to be converted to NADPH was consistent with the lag time observed in Figure 4. Therefore, the rate-determining factor was identified in the accumulation of NADPH from 0 min to 30 min, and in the activity of FDH from 30 min onward.

In vivo light-driven formate production
To recreate the light-driven formate production system in vivo, we attempted to express PsFDH(QN) in Anabaena heterocysts, which differentiate in order to carry out nitrogen fixation during nitrogen starvation [10]. Heterocysts are deficient in PS II, and their internal environment is micro-oxic. The cells receive carbohydrates from neighboring vegetative cells and use them as electron donors to PS I. The photoexcited electrons derived from PS I are used for the reduction of N 2 gas to ammonium by nitrogenase [20]. The micro-oxic and reductive environment makes heterocysts an ideal reaction vessel for light-driven formate production by PsFDH(QN). For the expression of PsFDH(QN) in heterocysts, we constructed two shuttle vectors, pAM505-fdh and pAM505-fdhfnr, and introduced them into Anabaena (Anabaena-fdh and Anabaenafdhfnr).
To confirm the expression of PsFDH(QN), western blot analysis was carried out. Anabaena-fdh and Anabaena-fdhfnr were cultivated in nitrogen-deficient BG11 medium (BG11 0 ), in which the differentiation to heterocysts and the expression of PsFDH(QN) was to be induced. Cell extracts were separated on 10% polyacrylamide gels and then blotted onto polyvinylidene difluoride membranes. Chemiluminescence images were obtained after blocking, incubation with anti-His tag antibody-biotin conjugate and streptavidin-Horseradish Peroxidase conjugate and reaction with chemiluminescence reagent. The bands were successfully detected at the expected molecular weight of PsFDH(QN) in the samples from both exoconjugants, but the intensity of the sample from BG11 0 was approximately 2-fold that from BG11 ( Figure 5). Although it is suggested that the expression of the PsFDH(QN) was enhanced in response to nitrogen deprivation, it remains to be solved whether the increase occurred mainly in the heterocysts. Assuming the heterocyst-specific response, taken together with the fact that approximately 1 heterocyst develops for every 10-20 vegetative cells; it is speculated that the PsFDH(QN) concentration in the heterocysts was 10-20 times that in the vegetative cells.
Anabaena-WT, Anabaena-fdh and Anabaena-fdhfnr were initially grown in BG11 0 medium by bubbling with air, air + 10% CO 2 , N 2 + 10% CO 2 , or argon + 10% CO 2 . We observed that only when Anabaena-fdh was incubated under argon + 10% CO 2 , a peak was detected at the same retention time (12.6 min) as formate with an intensity corresponding to a concentration of 28 µM ( Figure 6). To confirm whether the peak could be assigned to formate, the lysate was neutralized and treated with 10 µM Candida boidinii FDH (Sigma-Aldrich, St. Louis, Missouri) and 10 mM NAD + to advance the reaction: HCOO -+ NAD + → CO 2 + NADH. Thereupon, only the peak at 12.6 min among the observed peaks disappeared ( Figure 5), demonstrating that it was derived from formate. Therefore, we concluded that introducing PsFDH(QN) into heterocysts accelerated in vivo formate production.
One of the possible interpretations of this result is that under argon + 10% CO 2 , the electron flux toward N 2 fixation in heterocysts is decreased and the NADPH concentration is increased, facilitating the reaction: CO 2 + NADPH → HCOO -+ NADP + by PsFDH(QN). Another interpretation is that nitrogenase can accept electrons in the absence of N 2 and produce H 2 [20], which reacts with endogenous hydrogenase, producing NAD(P) H, which is used as a substrate for PsFDH(QN) to produce formate. In this study, no significant amounts of formate were detected in the lysates of Anabaenafdhfnr. Although the level of expression of PsFDH(QN) in Anabaena-fdhfnr was almost the same as that in Anabaenafdh, the growth rate of Anabaena-fdhfnr was lower than Anabaena-WT and Anabaena-fdh (see Figure S1), and a large quantities of cell debris appeared in the BG11 0 medium, suggesting the occurrence of cell lysis and the outflow of cell contents. Notably, by over repeated subculture, the ability of Anabaena-fdh to produce formate sometime disappeared because of the appearance of plasmids lacking FDH gene. In addition to this problem, the analysis of these phenomena at molecular level becomes difficult because of the low frequency of heterocysts. The development of stable expression system and a quantitative understanding of the process in heterocysts are currently under investigation.

Summary
Here we have shown a novel example of direct CO 2 reduction using the photo-reducing power of PS I. The system proposed here consisted of proteins (PS I, Fd, FNR, and PsFDH[QN]) and a natural electron carrier (NADP[H]). We have demonstrated the feasibility of creating a CO 2 reduction system in vivo by introducing PsFDH(QN) into heterocysts and have successfully enhanced the level of formate production. Further work investigating the detailed electron flux in the cells needs to be carried out, in addition to the optimization of the electron supply to the CO 2 reduction system, the suppression of the N 2 fixation and the design of the system for formate secretion. Figure S1. Growth curves of Anabaena wild-type (open circles), Anabaena-fdh (closed squares) and Anabaenafdhfnr (closed triangles). xmlns:xlink="http://www.w3.org/ 1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/ MathML">Culture conditions were: at 30 • C, light intensity of 30 mE•m -2 • s -1 and bubbling with air in BG11 medium. (TIF)