Overproduction of the Flv3B flavodiiron, enhances the photobiological hydrogen production by the nitrogen-fixing cyanobacterium Nostoc PCC 7120

Background The ability of some photosynthetic microorganisms, particularly cyanobacteria and microalgae, to produce hydrogen (H2) is a promising alternative for renewable, clean-energy production. However, the most recent, related studies point out that much improvement is needed for sustainable cyanobacterial-based H2 production to become economically viable. In this study, we investigated the impact of induced O2-consumption on H2 photoproduction yields in the heterocyte-forming, N2-fixing cyanobacterium Nostoc PCC7120. Results The flv3B gene, encoding a flavodiiron protein naturally expressed in Nostoc heterocytes, was overexpressed. Under aerobic and phototrophic growth conditions, the recombinant strain displayed a significantly higher H2 production than the wild type. Nitrogenase activity assays indicated that flv3B overexpression did not enhance the nitrogen fixation rates. Interestingly, the transcription of the hox genes, encoding the NiFe Hox hydrogenase, was significantly elevated, as shown by the quantitative RT-PCR analyses. Conclusion We conclude that the overproduced Flv3B protein might have enhanced O2-consumption, thus creating conditions inducing hox genes and facilitating H2 production. The present study clearly demonstrates the potential to use metabolic engineered cyanobacteria for photosynthesis driven H2 production.


Background
Development of renewable fuel as a clean alternative to fossil fuels is nowadays strongly needed. Besides solar energy, which represents the most abundant renewable energy, hydrogen (H 2 ) is regarded as an attractive option for its high energy content and null ecological impact: its combustion only releases water as a byproduct. In this regard, growing autotrophic, photosynthetic organisms (cyanobacteria and algae) to yield H 2 with minimized energy supply is a very promising alternative to fossil fuels.
In cyanobacteria, H 2 is produced by two different enzymes: hydrogenase and nitrogenase. In diazotrophic strains, H 2 is formed as a by-product of N 2 fixation activity performed by the nitrogenase. However, the nitrogenase is often associated to an uptake hydrogenase, encoded by the hup genes that catalyze the oxidation of H 2 into protons; the amount of H 2 produced during nitrogen fixation is thus rather limited [1]. The second type of enzymes producing H 2 are hydrogenases (H 2 ases). Bidirectional NiFe H 2 ases (called Hox), which catalyze both H 2 oxidation and proton reduction, are largely distributed across the

Open Access
Microbial Cell Factories *Correspondence: latifi@imm.cnrs.fr 1 Aix Marseille Univ, CNRS, LCB, Laboratoire de Chimie Bactérienne, Marseille, France Full list of author information is available at the end of the article cyanobacterial phylum [2,3]. They form a heteropentamer with a H 2 ase part (HoxYH) and a diaphorase part (HoxEFU). The physiological function of Hox hydrogenases in cyanobacteria is not well understood but they may serve as electron valve during photosynthesis in the unicellular cyanobacterium Synechocystis sp. PCC 6803 [4]. The expression of hox genes is induced in dark and/or anaerobic conditions [5] and is under the control of the regulators LexA and two members of the AbrB family (antibiotic resistance protein B) [6][7][8]. The sensitivity of cyanobacterial bidirectional H 2 ases to oxygen (O 2 ) and the fact that their activity occurs in the dark or under anaerobic conditions are the major obstacles to obtaining efficient solar driven production of H 2 in cyanobacteria. Several strategies have so far been adopted to overcome the limits of the natural H 2 -evolving mechanisms in cyanobacteria (for a review see [9]).
During photosynthesis, O 2 can be reduced to water through an enzymatic process involving flavodiiron proteins (Flvs) [10]. In cyanobacteria, Flvs catalyze the reduction of O 2 into water using NADPH as an electron donor [11] and play a critical role during growth under fluctuating light regimes [12]. The filamentous heterocyte-forming cyanobacterium Anabaena/Nostoc PCC7120 (hereafter Nostoc) produces four Flvs proteins in the vegetative cells (Flv1A, Flv2, Flv3A, and Flv4) and two Flvs (Flv1B and Flv3B) specific to the heterocyte [13]. The Flv3B protein mediates lightinduced O 2 -uptake in the heterocyte, which benefits nitrogenase activity by providing a protection mechanism against oxidation [14]. In addition, the ∆flv3B mutant displayed a broad effect on gene expression, which indicates that a regulation process links gene transcription to O 2 level in the heterocyte [14].
We recently reported that decreasing the O 2 level inside the heterocyte by producing the cyanoglobin GlbN allowed it to host an active FeFe H 2 ase from Clostridium acetobutylicum. The recombinant strain displayed a significant H 2 -production yield under phototrophic conditions [15]. These data suggest that engineering approaches increasing the anaerobiosis inside the heterocyte can be highly profitable for the activity of O 2 -sensitive enzymes. To go further, we investigate here the impact of an overproduction of the flavodiiron Flv3B protein on the production of H 2 in Nostoc. We demonstrate that the recombinant strain produces on average tenfold more H 2 than the parental strain and that the expression of the hox genes is induced in this genetic background.

Construction and characterization of a Nostoc recombinant strain overexpressing the flv3B gene
In a transcriptomic study using an RNAseq approach, the transcription of flv3B (all0178) gene was induced 12 h after nitrogen starvation [16]. In order to specifically overexpress the flv3B gene in the heterocytes without competing with the natural promoter of this gene, we decided to place it under the control of a heterocyte-specific promoter whose transcription is induced at the same time than flv3B. For this, we analyzed the transcription of flv3B throughout the differentiation process by quantitative RT-PCR. We also concomitantly monitored the transcription of the patB gene, known to be expressed after the initiation of heterocytes development [17]. flv3B and patB genes showed very similar transcription profile ( Fig. 1). Both genes were induced 18 h after nitrogen stepdown and their transcription increased through the development program (compare Fig. 1a, b). The patB promoter was therefore chosen to drive flv3B overexpression in Nostoc, and the resultant recombinant strain was named WT/patB-flv3B. As a first step in the characterization of this strain, we checked the overexpression of flv3B in response to nitrogen starvation. We first carried out quantitative RT-PCR analyses and expressed the amount of flv3B transcripts in the recombinant strain relatively to their amount in the wild type. Results reveal a more than tenfold increase in flv3B gene expression in the recombinant strain, also starting much sooner after nitrate depletion, indicating that flv3B gene was strongly overexpressed (Fig. 1c). Because Flv3B from Nostoc and FlvB from Chlamydomonas reinhardtii amino acid sequences present 51% identity (Additional file 1: Figure  S1), we hypothesized that antibodies produced against FlvB from C. reinhardtii [18] could cross-react with Flv3B and hence could be used to analyze the amount of Flv3B protein in Nostoc. Since Flv1B from Nostoc displays 30% identity with FlvB from C. reinhardtii, the anti-FlvB antibodies could also cross-react with this protein. However, as only the flv3B gene was overexpressed, we assumed that FlvB antibodies could help assessing Flv3B overproduction. In the western blot analyses, the amount of RbcL protein served to check that equal amounts of proteins were loaded in each condition [19]. Data on Fig. 1d show that a protein of the expected size of Flv3B (64 kDa) was detected only in BG11 0 medium (without nitrate), which is in agreement with flv3B gene being specific to the heterocyte [13]. Moreover, this protein accumulated at a higher level in the WT/patB-flv3B strain. Altogether, these results indicate that the flv3B gene was overexpressed in the recombinant strain. The WT/patB-flv3B strain showed similar growth efficiency than the wild type under both nitrogen replete and deplete conditions ( Fig. 2a, Table 1), and both strains differentiated heterocytes equally well (Fig. 2b). The frequency of heterocytes along the filament was similar between the two strains, with 12 vegetative cells on average in between two heterocytes (Fig. 2c). Given that the overexpression of flv3B did not impair the growth ability of the strain, we proceeded with an analysis of its impact on H 2 -production.

flv3B overexpression in the heterocyte powers H 2 -production
The sensitivity of H 2 ases and nitrogenase to O 2 is an important limitation to H 2 -photoproduction. By promoting O 2 consumption in the heterocyte, the Flv3B protein is ought to protect enzymes evolving H 2 . To test this hypothesis, the wild type and the WT/patB-flv3B strains were first grown exponentially under aerobic conditions in nitrate replete medium. H 2 -production yield was then measured and compared after cells were transferred to nitrate-depleted medium. The recombinant strain produced 10 to 30-fold more H 2 than the wild type under the same conditions (Fig. 3a). H 2 production increased with the experienced light irradiance, with the highest yield obtained under 60 µE m −2 . Flv3B overproduction is thus an efficient way to enhance H 2 photoproduction in Nostoc.

The presence of the uptake H 2 ase is required for a maximal H 2 production
Since the uptake H 2 ase consumes the H 2 produced by the nitrogenase in the heterocyte and since its deletion enhanced H 2 production [20], we investigated whether a deletion of hupL gene, encoding the large subunit of the uptake H 2 ase would show a cumulative effect with Flv3B overproduction. For this purpose, a deletion of hupL was constructed and the resultant strain transformed with the patB-flv3B containing plasmid. The deletion of hupL gene in an otherwise wild type background increased the H 2 production level, which is in agreement with data  (a, b) or the WT/patB-flv3B (c) strain at four different times (7, 18, 24 and 48 h) after the onset of nitrogen depletion. Each sample was measured in triplicate and the standard deviation is indicated by error bars. Values were normalized to the rnpB transcript, relatively to the value obtained for the wild type strain, which was set to 1. d Immunoblot analysis of the amount of Flv3B protein (upper panel) in the wild type and WT/patB-flv3B strains, carried out using antibodies produced against FlvB from Chlamydomonas reinhardtii [18]. Immunoanalysis of RbcL protein amount was carried out as a loading control (lower panel). The condition (+ Nitrate) stands for cultures performed in nitrate-containing medium, and the condition (− Nitrate) indicates cultures grown in nitrate-free medium published previously [20] (Fig. 3b). However, the absence of a further enhanced H 2 production following the overproduction of Flv3B in the ∆hupL strain was unexpected.  Intriguingly, the ∆hupL/patB-flv3B strain produced 3.5fold less H 2 than the WT/patB-flv3B strain (Fig. 3b).

Flv3B overproduction does not stimulate nitrogenase activity
The deletion of the flv3B gene was shown to result in a decrease in both the amount of nitrogenase subunits and nitrogenase activity [14]. Therefore, the increased H 2 production in the flv3B overproducing strain could be a consequence of an increase in the activity of the nitrogenase.
To test this hypothesis, we monitored nitrogenase activity in exponentially growing cultures after their transfer to a medium devoid of combined nitrogen. Results demonstrated that the overproduction of Flv3B protein did not enhance nitrogenase activity (Table 1). Therefore, the effect of Flv3B on H 2 production is unlikely to result from nitrogenase activity.

Flv3B overproduction induces the expression of the bidirectional H 2 ase encoding genes
Since the only other enzyme able to produce H 2 in cyanobacteria is the bidirectional Hox H 2 ase, we analyzed whether an induced expression of hox genes then results from the overproduction of Flv3B. The hoxH and hoxY genes encoding the H 2 ase subunits as well as the hoxE,F,U genes encoding the diaphorase subunits belong to two separate operons [21]. To evaluate the expression of these operons, the transcription of two genes from each operon (hoxH,Y and hoxE,F) was comparatively monitored in the wild type and the recombinant strains. Quantitative RT-PCR analysis was used to evaluate the transcription of these four genes after transfer of the strains into nitrogen deplete conditions to induce flv3B expression. The transcription of the four hox genes was weak in the wild type strain (Figs. 4a, b; 5a, b), which is in agreement with the fact that the hox genes are not expressed under aerobic conditions [21]. However, in the WT/patB-flv3B strain, 18 h after nitrogen step down, the hoxE,F, H and Y transcripts level were on average tenfold higher than in the wild type (Figs. 4c, d and 5c, d). The expression of the two hox operons encoding the H 2 ase and diaphorase proteins is therefore induced in the strain overexpressing the flv3B gene under the heterocyte specific promoter patB. Consequently, the effect of flv3B overexpression on H 2 production may be mediated by the induction of hox genes.

Discussion
In this work we show that overexpression of flv3B gene from a promoter specific to the heterocyte enhanced the production of H 2 in aerobic cultures of Nostoc. So far, the only conditions in which H 2 -evolution had been recorded in aerobically grown Nostoc were the use of mutants lacking the HupL subunit of the uptake H 2 ase or the last step of the maturation system of this H 2 ase [20,22]. H 2 evolution mediated by Flv3B overproduction presents the advantage of sustaining the protective effect of the uptake H 2 ase on the nitrogenase. By studying the phenotype of a ∆flv3B mutant of Nostoc, Ermakova et al. [14] showed that Flv3B protected nitrogenase through light-induced O 2 consumption inside the heterocytes. The effect of Flv3B overproduction evidenced in our work could therefore result from a stimulated nitrogenase activity. But the recombinant strain displayed similar nitrogenase activity as the wild type (Table 1), proof that another mechanism operates to enhance H 2 production.
In C. reinhardtii, the existence of intracellular microoxic niches in the chloroplast preserve FeFe-hydrogenase activity and support continuous H 2 production during growth in aerobic cultures [23]. The same authors suggested that Flvs proteins were involved in this process [23]. A similar mechanism may be proposed to explain the effect of the Flv3B protein overproduction on H 2 evolution, in which the decrease in O 2 concentration in the heterocyte would reinforce the anaerobiosis in this cell type, thus promoting H 2 ase synthesis and/or activity. We studied the transcription of hox genes encoding the bidirectional H 2 ase as their induction is known to be concomitant to high H 2 ase activity [21]. Data in Figs. 4, 5 indicate that flv3B overproduction led to a substantial induction of hoxE,F,H,Y genes expression that can explain the H 2 production measured in this strain. The LexA transcriptional factor was proposed to regulate hox genes transcription in Nostoc [21]. In the unicellular cyanobacterium Synechocystis PCC6803, LexA was suggested to act as a transducer of the intracellular redox state, rather than of the SOS response as in E. coli [24]. Based on this information, we suggest that an increased O 2 -uptake driven by Flv3B overproduction can modify the redox state in the heterocytes, resulting in the observed induction of hox genes transcription.
Surprisingly, and contrary to what happens in the wild type background, the lack of the uptake H 2 ase in the WT/patB-flv3B strain led to a decrease in H 2 production (Fig. 3b). As the H 2 ases are bidirectional enzymes, a possible interpretation of this result is that the Hup enzyme is responsible of the H 2 production observed in this recombinant strain. However, this is rather unlikely since it was demonstrated that the Hup H 2 ase is not able to produce H 2 at any significant rate, and is considered to react only in the uptake direction [1,25]. Through the oxidation of H 2 , the Hup H 2 ase provides electrons to the photosynthesis and respiratory processes [1] (Fig. 6). Since the Hox H 2 ase was suggested to use ferredoxin as reducing partner rather than NAD(P)H as previously admitted (reviewed in [9]), this enzyme may benefit from the electrons generated by the Hup H 2 ase through regeneration of the reduced ferredoxin pool (Fig. 6). This could explain the negative impact of the hupL deletion on the H 2 -production yield in the WT/patB-flv3B strain (Fig. 6). Our data show that metabolic engineering approaches are particularly relevant in the use of photosynthetic bacteria for biofuel production.

Conclusion
In the present study, the flv3B gene was specifically overexpressed in the heterocyte of Nostoc under the control of the patB promoter. The overproduction of the Flv3B flavodiiron enhanced the H 2 production yield by a factor of ten on average, which is not to be attributed to the nitrogenase since no increase in the nitrogenase activity was observed. The transcription of the hox genes was induced in the recombinant strain expressing the flv3B gene, suggesting that the additional H 2 produced relates to the activity of the bidirectional H 2 ase. Facilitating the consumption of O 2 inside the heterocyte thus appears as a relevant step towards the design of an optimized Nostoc strain for H 2 production. This paves the way to further improvement to achieve sustainable production of H 2 by air-grown cyanobacteria.

Growth conditions and heterocytes induction
Cyanobacterial strains were grown in BG11 medium (nitrate replete) at 30 ℃ under continuous illumination (30 µE m −2 s −1 ). Cultures of recombinant strains were supplemented with neomycin (50 μg mL −1 ). Heterocyte formation was induced by transferring the exponentially growing cultures (OD 750 = 0.8) to BG11 0 (BG11 devoid of sodium nitrate) by filtration (0.2 µm pore size filters, Sigma) and resuspension of cells into the nitratefree medium. The growth was maintained for 4 days. The presence of heterocytes was confirmed by light microscopy and their distribution within filaments was rated visually by counting the number of vegetative cells between two heterocytes. At least 400 total vegetative cells were counted for each strain.
In the H 2 production experiments, the strains were grown under continuous illumination of 20 µE m −2 s −1 or 60 µE m −2 s −1.

Construction of plasmids and strains
To construct the Flv3B overproducing strain, the promoter region of patB (all2512, 500 bp upstream the start codon) was amplified by PCR from Nostoc sp. PCC 7120 genomic DNA using the ppatB forward and ppatB reverse primers ( Table 2). The ppatB reverse primer contained a multiple cloning site (ApaI, ClaI, BamHI, SalI, ScaI, EcoRI). The amplified promoter was cloned into BglII and EcoRI restriction sites of the pRL25T plasmid [26], yielding the pRL25T-patB plasmid. The open reading frame of flv3B gene was amplified using the flv3B forward and reverse primers (Table 2), and cloned into the ApaI and ScaI restriction sites of the pRLpatB. The recombinant plasmid (pRL25T-patB-flv3B) was analyzed by sequencing (Millegen). Conjugation of Nostoc was performed as described in Ref. [27]. Briefly, E. coli strains (bearing the replicative pRL25T-patB-flv3B and the RP-4 conjugative plasmid) grown to exponential growth phase, were mixed to an exponentially grown Nostoc culture. The mixture was plated on BG11 plates and Neomycin was added 24 h later for plasmid selection. Plasmid extraction was used to analyze the obtained recombinant clones.
Deletion of the hupL gene, yielding the ∆hetL strain, was obtained by homologous recombination replacing the hupL3′ gene (all0687C) with the gene encoding the spectinomycin/streptomycin resistance (Sp/ Sm cassette hereafter). For this purpose, the upstream and downstream 1500 bp flanking the hupL3′ gene were amplified form Nostoc genomic DNA using the all0678 forward/all0678 reverse and the Strp-all0678 forward/Strp-all0678 forward, respectively; The Sp/ Sm cassette was amplified using the Strp forward/Strp reverse primers (Table 2), using the pBAD42 plasmid (Addgen) as template. Gibson's assembly technique (New-England Biolabs) was applied to insert the three resulting fragments into the suicide pRL271 vector linearized by SpeI. The resulting recombinant plasmid was conjugated into Nostoc as described above. The initial conjugants were selected by screening for resistance to The strains and plasmids used in this study are listed in Table 3.

RNA preparation and reverse transcription
RNAs were prepared using the Qiagen RNA extraction kit (Qiagen) following the manufacturer instructions. An extra TURBO DNase (Invitrogen) digestion step was undergone to eliminate the contaminating DNA. The RNA quality was assessed by tape station system (Agilent). RNAs were quantified spectrophotometrically at 260 nm (NanoDrop 1000; Thermo Fisher Scientific).    [26,30] pRL25T-patB-flv3B pRL25T harboring the flv3B gene under the control of the patB promoter (Neo R ) This study