Hydrogen Production in Chlamydomonas: Photosystem II-Dependent and -Independent Pathways Differ in Their Requirement for Starch Metabolism  

Abstract

Under sulfur deprivation conditions, the green alga Chlamydomonas reinhardtii produces hydrogen in the light in a sustainable manner thanks to the contribution of two pathways, direct and indirect. In the direct pathway, photosystem II (PSII) supplies electrons to hydrogenase through the photosynthetic electron transport chain, while in the indirect pathway, hydrogen is produced in the absence of PSII through a photosystem I-dependent process. Starch metabolism has been proposed to contribute to both pathways by feeding respiration and maintaining anoxia during the direct pathway and by supplying reductants to the plastoquinone pool during the indirect pathway. At variance with this scheme, we report that a mutant lacking starch (defective for sta6) produces similar hydrogen amounts as the parental strain in conditions of sulfur deprivation. However, when PSII is inhibited by 3-(3,4-dichlorophenyl)-1,1-dimethylurea, conditions where hydrogen is produced by the indirect pathway, hydrogen production is strongly reduced in the starch-deficient mutant. We conclude that starch breakdown contributes to the indirect pathway by feeding electrons to the plastoquinone pool but is dispensable for operation of the direct pathway that prevails in the absence of DCMU. While hydrogenase induction was strongly impaired in the starch-deficient mutant under dark anaerobic conditions, wild-type-like induction was observed in the light. Because this light-driven hydrogenase induction is DCMU insensitive and strongly inhibited by carbonyl cyanide-p-trifluoromethoxyphenylhydrazone or 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone, we conclude that this process is regulated by the proton gradient generated by cyclic electron flow around PSI.

In the context of economical and environmental concerns around fossil fuel depletion and global warming, the interest in hydrogen as an energy carrier for the future has considerably grown. Because molecular hydrogen is scarce on our planet, the development of a hydrogen economy strongly depends on our ability to propose clean and sustainable technologies of hydrogen production. In this context, the ability of some photosynthetic microorganisms, and particularly cyanobacteria and microalgae, to convert solar energy into hydrogen has been considered as very promising (Ghirardi et al., 2000; Rupprecht et al., 2006). When cells of the unicellular green alga Chlamydomonas reinhardtii are illuminated after adaptation to anaerobic conditions, electrons originating from water splitting at PSII are driven by the photosynthetic electron transport chain to ferredoxin and to a reversible iron hydrogenase, thereby enabling the production of molecular hydrogen from water and solar energy. Because both hydrogenase activity and expression are highly sensitive to the presence of O₂ (Happe et al., 1994; Ghirardi et al., 1997; Happe and Kaminski, 2002) and because O₂ is produced at PSII, hydrogen photoproduction stops after a few minutes of illumination. Melis et al. (2000) proposed an experimental protocol based on sulfur (S) deprivation, allowing long-term hydrogen production. This protocol relies on a two-stage process: during a first stage, oxygenic photosynthesis drives production of biomass and carbohydrate stores, and during a second anaerobic stage, the hydrogenase is induced and hydrogen is produced. Sulfur starvation has two important effects regarding hydrogen production: (1) a massive accumulation of starch that defines a common response to nutrient starvation and (2) a gradual drop in PSII activity (Wykoff et al., 1998). Once the rate of photosynthetic O₂ evolution drops below the rate of respiration, anaerobic conditions are reached, enabling the induction of hydrogenase and the production of significant amounts of hydrogen for several days. In parallel to hydrogen production, starch is degraded (Melis et al., 2000; Melis, 2007).

The importance of starch fermentation in hydrogen production has been recognized early from the pioneering work of Gibbs and coworkers (Gfeller and Gibbs, 1984; Gibbs et al., 1986). Based on the observation that starchless C. reinhardtii mutants sta6 and sta7 are strongly affected in their ability to produce hydrogen, Posewitz et al. (2004) proposed that starch metabolism plays a central role in C. reinhardtii hydrogen production. Actually, two different pathways can supply reductants (i.e. reduced ferredoxin) for hydrogen production in the light, a direct pathway involving PSII and an indirect PSII-independent pathway that relies on a nonphotochemical reduction of plastoquinones (PQs; Fouchard et al., 2005; Melis, 2007). Starch catabolism was proposed to play a role in both pathways (Melis, 2007) by (1) sustaining mitochondrial respiration and allowing the maintenance of anaerobic conditions for the PSII-dependent direct pathway and (2) by supplying electrons to the chlororespiratory pathway and to the hydrogenase through a PSI-dependent process during the indirect pathway (Fouchard et al., 2005; Mus et al., 2005; Melis, 2007). Such a dual role of starch was first confirmed by the study of a Rubisco-deficient mutant (CC2653), unable to accumulate starch and to produce hydrogen in conditions of S deprivation (White and Melis, 2006), but was recently challenged by the study of another Rubisco-less mutant (CC2803), which was reported to produce significant amounts of hydrogen in S starvation conditions, although not accumulating starch (Hemschemeier et al., 2008). These conflicting results obtained on two different Rubisco-deficient mutants prompted us to reexamine the contribution of starch to both direct and indirect pathways of hydrogen production. For this purpose, we complemented the initial work of Posewitz et al. (2004) by revisiting the ability of C. reinhardtii mutants deficient in starch metabolism to produce hydrogen. We thus tested the ability to produce hydrogen in a starchless strain carrying defect in the structural gene encoding the small subunit of ADP-Glc pyrophosphorylase (AGPase; sta6; Zabawinski et al., 2001). We found that sta6 mutant produces significant hydrogen amounts in condition of S deprivation but shows a strongly reduced PSII-independent hydrogen production. We conclude that while the PSII-independent hydrogen production pathway strictly relies on starch catabolism, the PSII-dependent pathway may require either starch or acetate as a respiratory substrate to maintain anaerobiosis.

RESULTS

The previously described starchless mutant sta6 was tested for its ability to produce hydrogen in response to S deficiency. The sta6 mutant, affected in the small subunit of the AGPase, is totally devoid of polysaccharides (starch or glycogen) (Zabawinski et al., 2001). This mutant being obtained by insertional mutagenesis of the cell wall less strain 330, the latter was used as our wild-type isogenic reference control. The hydrogen production capacity of the 330 strain was substantial but lower than that of wild-type 137C strain, representing from 30% to 40% of its maximal production in the same experimental setup (data not shown). Both sta6 mutant and wild-type 330 strains were placed in illuminated sealed flasks in conditions of S starvation. Gas concentrations and amyloglucosidase-sensitive carbohydrate contents were monitored during 7 d (Fig. 1  

Figure 1.

Hydrogen production in conditions of S deprivation in C. reinhardtii starchless mutants. Kinetics of H₂ production (A), maximal amounts of H₂ produced during the time course of the experiment (B), kinetics of O₂ evolution (C), and kinetics of intracellular carbohydrate content (D) were measured after placing cells in sealed flasks in conditions of S deprivation. A, C, and D, Measurements were performed four times in the starchless mutant sta6, in a complemented sta6 strain sta6-[C7], and in the control strain 330 with similar results. Due to the fact that anaerobiosis was reached at a few hours intervals depending on the experiment, error bars on kinetics of hydrogen production were relatively large (although similar amounts were produced at final). We therefore preferred to show a representative experiment. B, Average and standard deviation of maximal amounts of hydrogen produced during the time course of the experiment in the wild type (WT), in sta6 and sta7 starchless mutants, and in complemented sta6 strains (wild type, sta6, and sta7 are average ± sd of four independent experiments; sta6-C is the average ± sd of seven experiments, four for sta6-[C7], two for sta6-[C9], and one for sta6-[C13]). Cultures were inoculated at a cellular concentration of 4 × 10⁶ cells mL⁻¹.

). Surprisingly, although the sta6 mutant was previously reported to produce reduced hydrogen amounts (Posewitz et al., 2004), we observed similar hydrogen production patterns in sta6 and wild-type strain (Fig. 1A). Hydrogen production started after about 48 h of S deprivation in both wild-type 330 and sta6 strains (Fig. 1A). While the wild-type strain accumulated up to 60 μg 10⁻⁶ cells of starch during the first 48 h, we could not detect any storage polysaccharides in the sta6 mutant (Fig. 1C). Complemented strains were obtained by transformation of the sta6 mutant with a vector carrying a wild-type copy of the STA6 gene and paromomycin resistance cassette (Fig. 2A  

Figure 2.

Complementation of the sta6 starchless mutant. A, A genomic PCR fragment of the STA6 gene was cloned into pSL18 plasmid carrying a paromomycin resistance cassette (AphVIII). B, Characterization of transformants: control (330 and sta6) and three complemented strains (sta6-[C7], sta6-[C9], and sta6-[C13]) were plated on N-free TAP medium and stained after 5 d with iodine vapors to reveal the presence of starch (−N), on TAP medium (control), on TAP medium supplemented with Arg (+Arg), or on TAP medium supplemented with Arg and paromomycin (+ Arg + Paro). C, Starch contents were measured in control (330 and sta6) and complemented strains (sta6-[C7], sta6-[C9], and sta6-[C13]) in liquid cultures after 24 h of S deprivation under constant illumination (250 μmol photons m⁻² s⁻¹ PAR). Data are expressed as the average of eight independent experiments for the wild type (WT), three independent experiments for sta6 and sta6-[C7], and two independent experiments for sta6-[C9] and sta6-[C13]. Error bars represent ± sd.

). Eighty transformed paromomycin-resistant strains were screened on petri dishes for their ability to accumulate starch in response to nitrogen (N) deprivation using iodine staining. Among 24 colonies showing a dark staining, 12 were selected and tested for their starch content in liquid culture under conditions of N deprivation. Total rescue of the starch accumulation was obtained in eight strains (between 20 and 30 μg starch 10⁻⁶ cells) and partial rescue in four strains (below 10 μg starch 10⁻⁶ cells). Five strains (four showing total rescue, sta6-[C2], sta6-[C6], sta6-[C7], and sta6-[C9], and one showing partial rescue under N deprivation, sta6-[C13]) were tested for their ability to accumulate starch in experimental conditions corresponding to the H₂ production protocol (S deficiency). In our experimental conditions, a higher starch accumulation (about 50 μg starch 10⁻⁶ cells) was measured in the control strain, and only partial restoration (maximum 50%) was observed in the five strains tested (three of them are shown in Fig. 2C). Such a difference might be due to differences in experimental conditions (such as illumination) between N deprivation (using a standard laboratory protocol) and S deprivation experiments (carried out using the Melis protocol in sealed flasks) and to the fact that for complementation the sta6 gene was not placed under the control of its own promoter but under the control of the light-dependent psaD promoter. None of the paromomycin-resistant strains obtained by transforming the starchless sta6 mutant with a plasmid carrying only the resistance cassette were able to produce cell patches with an iodine stain nor assayable starch, thus confirming the link between storage polysaccharide synthesis and the presence of a functional ADP-Glc pyrophosphorylase small subunit expression (Supplemental Fig. S1). In conditions of S depletion, the complemented strain sta6-[C7] produced as much hydrogen as other wild-type and mutant strains, although reaching anoxia and starting to produce hydrogen a little later (Fig. 1). Similar hydrogen production rates were observed in sta7 (Fig. 1B; Supplemental Fig. S2), a mutant deficient in isoamylase, a debranching activity required to synthesize crystalline starch (Mouille et al., 1996). This mutant, which does not synthesize starch, produces small amounts (from 5% to about 20% of the normal starch amount depending on physiological conditions and genetic background) in the form of a soluble polysaccharide sharing structural and biochemical similarities with glycogen and thus named phytoglycogen (Mouille et al., 1996). Note that both anaerobiosis and hydrogen production started 24 h earlier in sta7 (Supplemental Fig. S2, A and B) when compared to other strains (Fig. 1, A and B).

Posewitz et al. (2004) reported that the induction of hydrogenase, followed at the transcript level, declined gradually under dark anaerobic conditions in the sta7 mutant, explaining the decrease in hydrogen production. At variance with the experimental protocol used by Posewitz et al. (2004), S deficiency experiments were performed in continuous light. We postulated that conditions of illumination may affect hydrogenase induction and therefore used H/D isotope exchange measurements by mass spectrometry to monitor the hydrogenase activity (Fig. 3A  

Figure 3.

Hydrogenase activity measured by H/D exchange in starch-deficient mutants following anaerobic induction in the dark or in the light and effect of inhibitors on the light induction. A, After induction of hydrogenase for 45 min under anaerobic conditions (obtained by addition of Glc and Glc oxidase), deuterium (D₂) was bubbled to saturation in the algal sample. Concentrations of D₂ (m/e = 4), HD (m/e = 3), and H₂ (m/e = 2) were monitored by mass spectrometry. A similar experiment was carried out in mutant (sta6) and complemented (sta6-[C7]) strains. B, Hydrogenase activities were measured by mass spectrometry in the same type of experiment as described in Figure 3A. Calculation of hydrogenase activity from H/D exchange kinetics was performed as previously described (Cournac et al., 2004) in mutant (sta6) and complemented (sta6-[C7]) strains following a 45-min induction in darkness (black boxes), in the light (100 μmol photons m⁻² s⁻¹ PAR; dark gray boxes), or in the light in the presence of 20 μ  m (final) DCMU (gray boxes), of 20 μ  m (final) DCMU and 10 μ  m (final) DBMIB (light-gray boxes), or 2 μ  m (final) FCCP (white boxes). Data are expressed as the average of seven independent experiments for the sta6 mutant and three independent experiments for both the wild type (WT) and complemented strain sta6-[C7]. Error bars represent ± sd.

). The rationale of this measurement is that the catalytic site of the hydrogenase, by splitting D₂ into D⁻ and D⁺, allows a random exchange between D⁺ and H⁺ ions originating from the medium. Subsequent recombination of H⁺ with D⁻ forms HD, whose concentration increases in the medium in the presence of an active enzyme (Cournac et al., 2004). Consistent with the lack of hydrogen production observed in these conditions and in agreement with the data of Posewitz et al. (2004), the sta6 mutant showed a very low H/D exchange activity after dark anaerobic induction (Fig. 3, A and B). However, when the anaerobic induction was carried out in the light, significant hydrogenase activity was measured in the sta6 mutant, the activity measured in these experimental conditions being close to that measured in the complemented strain. In contrast, the light treatment had no major effect on hydrogenase activity in both complemented and wild-type strains (Fig. 3B). Note that for unknown reasons, a significantly higher hydrogenase activity was measured in the latter. Posewitz et al. (2004) proposed that hydrogenase induction may be triggered by reduction of the PQ pool. In order to investigate this possibility, we studied the effect of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), a PSII inhibitor, on the light activation process. Surprisingly, we found that addition of DCMU had no effect on the light activation of the hydrogenase (Fig. 3B), clearly showing that this process is not regulated by the PSII-mediated reduction of the PQ pool. We then tested whether cyclic electron flow around PSI, which is a DCMU-insensitive process, may be involved in hydrogenase activation. For this purpose, we studied the effects of the cytochrome b  ₆  /f inhibitor 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone (DBMIB) and of the uncoupler carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP). Both compounds were found to strongly inhibit the light induction of hydrogenase (Fig. 3B), therefore supporting a role of the proton gradient generated by cyclic electron flow around PSI in the activation process.

The importance of starch on the indirect pathway of hydrogen production was then investigated. In contrast to the direct pathway, which is fed by electrons originating from PSII, the indirect pathway is supplied by nonphotochemical reduction of the PQ pool. In order to determine whether nonphotochemical reduction of the PQ pool was affected in starch-deficient mutants, we first monitored changes in chlorophyll fluorescence during a transition from aerobic to anaerobic conditions (Supplemental Fig. S4). A decline of the PSII quantum yield (Φ_PSII) was observed when reaching anoxia both in control and starch less strains, indicating a reduction of the PQ pool by stromal reductants. However, the decline in Φ_PSII was less pronounced in the starchless mutant, indicating that reductants are supplied less efficiently to the PQ pool in the absence of starch. We then measured the capacity of the indirect pathway by following hydrogen production in a S-deprived medium supplemented with DCMU (Fig. 4  

Figure 4.

Hydrogen production and intracellular starch content in conditions of hydrogen production by the indirect pathway. After 24 h of S deficiency, DCMU (20 μ  m final concentration) was added to the culture medium and the cell suspension was bubbled by N₂ to remove O₂. A, Hydrogen production measured in the starchless mutant sta6 mutant (sta6), in the complemented strain (sta6-[C7]), and in the control strain (330; labeled as WT). B, Intracellular carbohydrate content. Data are expressed as the average of three independent experiments for each strain. Cultures were inoculated at a cellular concentration of 4 × 10⁶ cells mL⁻¹. Error bars represent ± sd.

). In these conditions, the direct pathway, which is PSII dependent, is totally switched off (Fouchard et al., 2005; Melis, 2007), and hydrogen production essentially depends on electrons donated to the intersystem electron transport chain by the NAD(P)H-PQ oxidoreductase (Jans et al., 2008; Desplats et al., 2009). As previously reported (Fouchard et al., 2005; Melis, 2007), hydrogen production rates measured in these conditions are much lower (at least 10 times lower) than in the absence of DCMU (compare Fig. 1B and Fig. 4A). In contrast to hydrogen production measured in the absence of DCMU, hydrogen production by the indirect pathway was severely affected in the sta6 mutant. Similar hydrogen production rates were observed in the wild type and in the complemented strain (sta6-[C7]), despite a 50% decrease in the starch pool in the particular complemented strain used by comparison to the wild type (Fig. 4). In the sta7 mutant, hydrogen production by the indirect pathway was significantly reduced compared to the wild type (Supplemental Fig. S3), although to a lesser extent than in sta6. We conclude from these experiments that although nonphotochemical reduction of the PQ pool does not rely on starch in short-term experiments, hydrogen production by the indirect pathway requires the presence of starch. We also conclude that hydrogen produced in starch-deficient mutants in the absence of DCMU should mainly depend on the direct pathway.

The nature of the substrate supplying electrons to mitochondrial respiration to scavenge O₂ from PSII and maintain anaerobiosis in the absence of starch was then questioned. Indeed, previous studies reported that acetate was not consumed during hydrogen production (Kosourov et al., 2003). We followed acetate consumption during hydrogen production experiments in wild-type and mutant strains (Fig. 5  

Figure 5.

Mass balances of hydrogen, intracellular starch, and extracellular acetate during anaerobic hydrogen production in C. reinhardtii cells in response to S deprivation. Cultures were inoculated at a cellular concentration of 4 × 10⁶ cells mL⁻¹. After 72 h of S deficiency, cells were transferred to sealed flasks and bubbled with N₂ to reach anoxia simultaneously. Evolutions of hydrogen, starch, and acetate concentrations were monitored during 96 h. Black bars, Starch-deficient mutant (sta6); gray bars, complemented strain (sta6-[C7]); white bars, control strain (330). Data are expressed as the average of three independent experiments for each strain. Error bars represent ± sd.

). In agreement with Kosourov et al. (2003), no significant acetate consumption was observed during the first 96 h of hydrogen production in the wild-type strain. However, significant acetate consumption was measured in the starch-deficient mutant, the complemented strain showing an intermediate behavior. We conclude from this experiment that in the absence of cellular starch, acetate may be used as substrate to maintain anaerobiosis during the process of hydrogen production by the direct pathway.

DISCUSSION

We report here that a C. reinhardtii mutant (sta6) strongly affected in starch biosynthesis, produces as much hydrogen as the wild type in conditions of S deficiency. The sta6 strain, mutated in the AGPase, does not produce any starch (Zabawinski et al., 2001). Similar results were obtained in the sta7 mutant, which is affected in the isoamylase and produces about 20% of phytoglycogen, a noncrystalline, water-soluble, polysaccharide (Mouille et al., 1996). Our results contrast with the generally accepted role of starch in hydrogen photoproduction (Melis, 2007). Indeed, based on the study of the same two starch-deficient mutants, Posewitz et al. (2004) concluded that starch metabolism plays an important role in C. reinhardtii photoproduction. The striking difference in the conclusions of these two studies may be attributed to differences in experimental conditions used for hydrogen photoproduction measurements. Posewitz et al. (2004) measured hydrogen photoproduction following dark anaerobic induction, whereas in this study hydrogen photoproduction was measured in conditions of S deficiency and light anaerobic induction. Hydrogen production may be affected at two different levels: electron supply for hydrogen production and hydrogenase enzyme induction.

Hydrogen photoproduction results from the contribution of two electron transfer pathways: a direct and an indirect pathway. In the direct pathway, PSII is active and supplies electrons to PSI and to the hydrogenase. This pathway requires O₂-consuming processes to maintain anaerobic conditions at the level of hydrogenase. Starch was proposed to be involved in the two pathways of hydrogen production (Melis, 2007). In the direct pathway, starch metabolism would contribute to maintain low O₂ concentration by feeding mitochondrial respiration (Zhang et al., 2002). In the indirect pathway, starch breakdown supplies electrons to the PQ pool (Gibbs et al., 1986), through Nda2, a plastidial type II NAD(P)H dehydrogenase (Mus et al., 2005; Jans et al., 2008; Desplats et al., 2009). Clearly, the existence of two starch-requiring pathways of hydrogen production is inconsistent with our data. When DCMU was added after the starch accumulation phase, conditions where hydrogen photoproduction essentially results from the indirect pathway, hydrogen production was strongly decreased in both starch-deficient mutants, thus showing that the PSII-independent indirect pathway activity depends on starch metabolism. Since hydrogen photoproduction was not affected in conditions of S deficiency in the absence of DCMU, we conclude that the PSII-independent pathway does not operate in these conditions, or at least is not necessary for sustaining the production rates that are observed. We also conclude that in our experimental conditions, the direct pathway does not rely on starch metabolism and that the direct pathway prevails in conditions of S deficiency.

The major differences in hydrogen photoproduction rates observed in starch-deficient mutants upon dark anaerobic induction (Posewitz et al., 2004) and upon light anaerobic induction (in conditions of S deficiency) can be explained by differences in hydrogenase induction. In agreement with Posewitz et al. (2004), we observed that induction of hydrogenase activity is severely decreased in a starch-deficient mutant upon dark anaerobic adaptation. However, when anaerobic adaptation was performed in the light, no major difference was observed between the wild type and the starch-deficient mutant sta6. This clearly shows that the induction of hydrogenase depends on two parameters: the establishment of anaerobic conditions and a cellular metabolic and/or bioenergetic signal that is triggered either in the dark in the presence of starch or in the light. It has been shown that the PQ pool redox state may be involved in target gene expression (Escoubas et al., 1995; Bellafiore et al., 2005), and Posewitz et al. (2004) proposed, based on transcript analysis, that hydrogenase expression may be triggered by the redox state of the PQ pool. The light induction of the hydrogenase activity observed in our study was insensitive to DCMU but strongly inhibited by the cytochrome b  ₆  /f inhibitor DBMIB or by the uncoupler FCCP. Based on these data, we propose that the proton gradient (or ATP) generated by cyclic electron flow around PSI is involved in the induction process. Two cyclic electron pathways have been identified in Chlamydomonas based on their sensitivity to antimycin A (Ravenel et al., 1994). The antimycin A-insensitive pathway likely involves the newly discovered Nda2, which catalyzes nonphotochemical PQ reduction in Chlamydomonas chloroplasts (Jans et al., 2008; Desplats et al., 2009), while the antimycin A-sensitive pathway may involve, as described for Arabidopsis (Arabidopsis thaliana), PGR5 and PGRL1 (Munekage et al., 2002; DalCorso et al., 2008). Which pathway prevails under anaerobic conditions and at which level (transcription, translation, enzyme stability, or activity) the H₂ase is induced by light will need further investigations to be elucidated.

Besides the study of starch-deficient mutants, the role of carbohydrate stores in the process of hydrogen photoproduction has been investigated in C. reinhardtii by the study of Rubisco-deficient mutants. Such mutants are unable to accumulate starch in the light and require acetate for growth. However, contrasting results were reported in two different Rubisco deficient strains, CC-2653 and CC-2803 (White and Melis, 2006; Hemschemeier et al., 2008). White and Melis (2006) first reported the absence of significant hydrogen production in the mutant strain CC-2653. In sharp contrast, Hemschemeier et al. (2008) reported that the Rubisco-deficient mutant CC-2803 produces significant hydrogen amounts in the absence but also in the presence of S. From the effect of DCMU, these authors concluded that hydrogen was essentially produced by the PSII-dependent direct pathway in CC-2803. Our results, which conclude that starch is not involved in the direct hydrogen production pathway, are in line with this view. Since both mutants do not accumulate starch, the contrasting results were proposed to result from Rubisco turnover in conditions of S deficiency, one strain (CC-2653) producing a truncated Rubisco, while the other (CC-2803) is totally devoid of the CO₂-fixating enzyme (Hemschemeier et al., 2008). It should be emphasized here that since Rubisco-deficient mutants do not produce oxygen in the light, they rapidly reach anaerobic conditions upon closure of the reaction vessel, therefore favoring a rapid induction of hydrogenase. These differences might alternatively result from differences in mitochondrial respiration in the two strains, CC-2653 showing a reduced activity of respiration (White and Melis, 2006).

A question that remains to be answered is to elucidate which mechanisms are involved in the consumption of oxygen produced at PSII when hydrogen is produced in starch-deficient or in the Rubisco-deficient mutant CC-2803. In the absence of starch, other storage compounds or acetate could be involved. The role of acetate during the hydrogen photoproduction process has been a matter of debate. Acetate was recognized as the main substrate for respiration during the establishment of anaerobic conditions, but based on the observation that acetate consumption stopped during the hydrogen production phase, it was concluded that acetate does not play a role in the hydrogen production process by itself (Ghirardi et al., 2000; Melis and Happe, 2001; Kosourov et al., 2003). However, acetate measurements performed in starch-deficient mutants showed a significant consumption of this compound during the hydrogen production phase, and some acetate consumption could also be observed in starch-containing strains at the beginning of the hydrogen production phase. We conclude from these experiments that acetate metabolism, by supplying electrons to mitochondrial respiration, might in the absence of starch contribute to maintain micro-aerobic conditions required for hydrogen production by the direct pathway. Moreover, we conclude from measurements of starch, hydrogen, and acetate balances (Fig. 5) that variations in intracellular starch content in the wild type and in extracellular acetate concentration in the starch-deficient strain are sufficient to quench O₂ produced during the hydrogen production phase. Indeed, one molecule of hydrogen requires two reducing equivalents corresponding to half O₂ produced at PSII. Since acetate contains eight reducing equivalents per molecule, acetate consumption by the sta6 mutant (about 300 μmol 10⁻⁹ cells during the time course of the experiment) would be theoretically sufficient to quench the 600 μmol O₂ 10⁻⁹ cells corresponding to 1200 μmol H₂ 10⁻⁹ cells. In the same way, Glc molecules containing 24 reducing equivalents, starch consumption in the wild-type strain (about 300 μmol Glc equivalent 10⁻⁹ cells during the time course of the experiment) is largely sufficient to quench O₂ produced at PSII.

Experimental conditions (differences between strains, conditions of preculture, light intensity, etc.) may affect the contribution of both direct and indirect pathways and may partly explain some discrepancies in the literature concerning the relative importance of these pathways. Different experimental conditions, by affecting, for instance, kinetics of PSII inhibition in conditions of S depletion, activity, or expression of Nda2 (the enzyme involved in nonphotochemical reduction of PQ), or starch mobilization (which supplies electrons to the indirect pathway), could in fine lead to different ratios between direct and indirect pathways. Also, the fact that suppression of a pathway does not lead to significant changes in hydrogen production rates does not necessarily mean that this pathway is not operating. Indeed, it seems highly probable that these two pathways, which are somehow competing for electron carriers, may complement each other to some extent. There could also be some subtle synergetic modes of interactions between pathways. This is well illustrated by Hemschemeier et al. (2008) who observed that, although hydrogen production by the indirect pathway (measured on the long term after DCMU addition) did not exceed 30% (most often around 10%) of hydrogen production in standard S-deprived conditions, it could contribute to >80% when measured in the short term (by adding DCMU at different periods of the hydrogen production process). This suggests the existence of different limitations in short-term and long-term experiments.

Compared to the direct pathway, the indirect pathway of hydrogen production presents some advantages but also suffers from limitations for biotechnological applications. First, as evidenced by this study, and in contrast to the direct pathway, the indirect pathway does not require acetate to maintain anaerobic conditions. Second, and probably the main advantage of the indirect pathway is related to its lower quantum requirement, starch being converted to hydrogen by the photochemical activity of PSI, in the absence of PSII activity. Hydrogen production processes in which aerobic photosynthesis would be driven at low cost in open ponds and the subsequent conversion of biomass into hydrogen, based on the indirect pathway, performed in closed photobioreactors have been proposed (Benemann, 1997). In such a process, the second step would require a costly closed photobioreactor optimized for hydrogen production. Even if the total quantum yield of the process would be increased since it would require six photons instead of four per molecular hydrogen produced, the lower quantum yield of the anaerobic phase would be an advantage since it would directly impact the light efficiency of the conversion and the dimensions and cost of the photobioreactor (Benemann, 1997). On the other hand, the control of hydrogen production processes based on the indirect pathway requires a tight control of PSII activity. This can be achieved by S deficiency, but the use of nutrient starvation has a negative long-term impact on production yields. This could be alternatively achieved by controlling PSII activity using inducible promoters to switch on/off the activity of PSII (Surzycki et al., 2007). The main disadvantage of the indirect pathway is that its maximal rate of hydrogen production is much lower (at least 10 times lower) than by the direct pathway (compare Fig. 1B and Fig. 4A). It has been proposed that metabolic steps involved in starch breakdown and/or reduction of the PQ pool from stroma donors likely limit the process (Cournac et al., 2002). The starch level by itself does not appear to be limiting, since hydrogen production rates measured in the complemented sta6 strains, which accumulated about half less starch amounts as the wild type, were similar to that measured in the wild type. Although this decrease in the starch content compared to the wild type is not explained at the moment, this could be due to subtle differences in transcriptional regulation of the complementing transgene by comparison to the wild-type gene. Indeed, transcription of STA6 is known to be under circadian clock control and exerts a tight control in the carbon flux to starch (Zabawinski et al., 2001; Ral et al., 2006). Limitations in the indirect pathway more likely rely on enzymatic steps involved from starch breakdown to nonphotochemical reduction of PQs. Recently, a type II NAD(P)H dehydrogenase activity has been evidenced in C. reinhardtii chloroplasts and shown to be involved in nonphotochemical reduction of PQs and hydrogen production (Jans et al., 2008; Desplats et al., 2009). This enzyme, but also enzymes involved in starch breakdown, which remain to be identified for most of them, may represent good targets for future biotechnological improvements.

MATERIALS AND METHODS

Strains and Culture Conditions

Wild-type Chlamydomonas reinhardtii strains used in this study are 137C (mt– nit1 nit2) and 330 (mt+ arg7-7 cw15 nit1 nit2). Starch-deficient mutants sta6  (mt+ cw15 nit1 nit2 sta6-1∷ARG7) and sta7 (mt+ nit1 nit2 cw15 arg7-7 sta7-2∷ARG7) were obtained from strain 330 by random integration of pARG7 into the nuclear genome (Mouille et al., 1996; Zabawinski et al., 2001). Complemented strains sta6-[C7], sta6-[C9], and sta6-[C13] were obtained by transformation of the sta6 mutant by a vector carrying a genomic copy of wild-type STA6 gene. Transformation controls (strains called sta6-[TC1], sta6-[TC2], and sta6-[TC3]) were obtained by transformation of the sta6 mutant using a plasmid only carrying the paromomycin resistance cassette. Algal cells were maintained on agar Tris-acetate phosphate (TAP; Harris, 1989) plates under constant illumination (40 μmol photons m⁻² s⁻¹) at 25°C and replated every 3 weeks. Unless indicated otherwise, Arg (100 μg mL⁻¹) was added to the growth medium of the 330 strain, both on solid or liquid medium. Liquid cultures were performed in TAP liquid medium with constant stirring at 25°C and continuously illuminated with a mix of Cool-white and Grolux fluorescent OSRAM tubes (140 μmol photons m⁻² s⁻¹ photosynthetically active radiation [PAR]).

Plasmid Constructs and Transformation of Chlamydomonas Strains

Complementation of sta6 mutants was carried out by transformation with the plasmid pSL-STA6 carrying a genomic copy of the STA6 gene, which encodes the small subunit of AGPase and the paromomycin resistance cassette AphVIII. The genomic fragment was obtained by PCR using the protocol provided for long PCR with the Ext DNA polymerase (Finnzyme). The following primers were used to amplify the complete genomic sequence and introduce EcoRI sites: R1STA6for, 5′-GGAATTCATGGCCCTGAAGATGCGGGTG-3′; R1STA6rev, 5′-GGAATTCTTAGATGATGGTGCCGGC-3′. The PCR-amplified fragments were then subcloned using the EcoRI restriction sites introduced by PCR in the pSL18 plasmid previously used for functional complementation of Chlamydomonas mutant strains (Dauvillée et al., 2006). Transformation was achieved using the glass beads method (Kindle, 1990). One hundred milliliters of a 2 × 10⁶ cells mL⁻¹ culture was concentrated 100 times, and 300 μL of concentrated cells were poured into a glass tube containing 300 mg of sterilized glass beads (0.45–0.52 mm diameter) and 1 μg of plasmid DNA. After vortexing, 600 μL of TAP medium were added and cells were spread on TAP plates supplemented with paromomycin (10 μg mL⁻¹). After drying, plates were sealed with parafilm and incubated in a 25°C growth chamber under continuous illumination (40 μmol photons m⁻² s⁻¹ PAR). The transformants were purified on paromomycin containing TAP plates and then transferred on N-free TAP medium (TAP-N) plates and further stained with iodine to detect the presence of starch.

Hydrogen Production in Response to Sulfur Deficiency

We used an experimental protocol adapted from the initial S deprivation protocol described by Melis et al. (2000). Cells were grown to late exponential phase (6 × 10⁶ cells mL⁻¹), washed twice in a S-deprived medium (TAP-S), and then resuspended in 280 mL of TAP-S medium in glass flasks (Schott). The flasks were sealed with rubber septa. Sealed flasks were placed under continuous light (50% Cool-white, 50% Grolux fluorescent tubes; Osram) at 200 μmol photons m⁻² s⁻¹ PAR with constant stirring. Two methods were used to transfer cultures to anoxic conditions. The first one consisted of incubating algae in S-deprived medium, S deprivation leading to the inhibition of O₂ production by PSII (Melis et al., 2000). In the second method, PSII was blocked at t = 24 h as described by Fouchard et al. (2005) by supplying the medium with DCMU (20 μ  m final concentration). Nitrogen gas was bubbled for 6 min to eliminate dissolved oxygen in the medium and reach anoxic conditions more rapidly. Every day, 0.5 mL of gas sample was taken out of the flask with a tight syringe and introduced through an argon-flushed line into a mass spectrometer (Prisma QMS 200; Pfeiffer Vacuum) to measure the gas phase composition (H₂, O₂, and CO₂).

Starch and Acetate Measurements

Starch extraction was performed using a method slightly modified from that of Klein and Betz (1978). One milliliter of culture was sampled, centrifuged at 18,000g for 2 min, suspended in 1 mL of methyl alcohol for chlorophyll extraction, and centrifuged again. The pellets were rinsed with 0.5 mL of Na-acetate buffer (100 mm, pH 4.5), resuspended in 300 μL of Na-acetate buffer, and heated in an oven for 15 min at 120°C for starch solubilization. Starch was then degraded to Glc by a commercial amyloglucosidase solution (Starch Assay Reagent; Sigma-Aldrich), and Glc was subsequently assayed using an automated sugar analyzer (YSI model 2700 select; YSI Life Sciences). Acetate assay was performed using an enzymatic kit based on the monitoring of NADH production at 340 nm using acetyl-CoA synthetase, citrate synthase, and l-malate dehydrogenase (BioSentec).

Hydrogenase Activity Measurement

Hydrogenase activity was determined using a water-jacketed chamber (1.5 mL at 25°C) coupled to a mass spectrometer (model MM 8-80; VG Instruments) through a membrane inlet system. Cultures were inoculated at about 0.5 × 10⁶ cells mL⁻¹ and grown to 5 × 10⁶ cells mL⁻¹. A cell sample containing 30 μg of chlorophyll was harvested, centrifuged, resuspended in 1.5 mL of fresh TAP medium, and poured in the water-jacketed chamber. Glc (20 mm final concentration), 0.5 mg of Glc oxidase, and 5,000 units catalase were added to reach and maintain anoxia. After 45 min of induction of the hydrogenase under anaerobic conditions in the dark or in the light (100 μmol photons m⁻² s⁻¹ PAR), and in the presence of different inhibitors (20 μ  m DCMU, 2 μ  m FCCP, or 10 μ  m DBMIB final concentrations), deuterium (D₂) was bubbled to saturation in the algal sample. Concentrations of D₂ (m/e = 4), HD (m/e = 3), and H₂ (m/e = 2) were monitored by mass spectrometry. Hydrogenase activation resulted in H⁺/D⁺ scrambling, which resulted in the progressive replacement of dissolved D₂ by HD and ultimately by H₂. Calculation of hydrogenase activity from H/D exchange kinetics was performed as described by Cournac et al. (2004).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Characterization of transformation controls of the sta6 starchless mutant.

• Supplemental Figure S2. Hydrogen production in conditions of S deprivation in the C. reinhardtii starchless mutant sta7.

• Supplemental Figure S3. Hydrogen production and intracellular starch content during hydrogen production by the indirect pathway in the starchless mutant sta7.

• Supplemental Figure S4. Chlorophyll fluorescence measurements during a transition from aerobic to anaerobic conditions in C. reinhardtii and in the starchless mutant sta6.

ACKNOWLEDGMENTS

We thank Patrick Carrier (Commissariat à l'Energie Atomique Cadarache) and Thierry Duchêne (Université des Sciences et Technologies de Lille) for excellent technical help.

Author notes

1

This work was supported by the French “Agence Nationale pour la Recherche” (PHOTOBIOH2 project) and by the European FP7-Energy-RTD program (SOLAR-H2 project 212508). V.C. was a recipient of a Ph.D. thesis grant cofinanced by Commissariat à l'Energie Atomique and the “Région Provence Alpes Côte d'Azur.”

^*

Corresponding author; e-mail .

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Gilles Peltier ().

[W]

The online version of this article contains Web-only data.

www.plantphysiol.org/cgi/doi/10.1104/pp.109.144576