Introduction

Fuels derived from biological renewable energy sources (e.g., bioethanol or biodiesel) are carbon neutral, since fuel production through photosynthesis captures CO2 from the atmosphere and fuel utilization re-emits CO2 back into the atmosphere. The ultimate carbon-free fuel is hydrogen, when produced directly from inorganic sources such as water. A number of photosynthetic microbes are known to uncouple H2 photoproduction from photosynthetic carbon fixation, using reductant generated directly from photosynthetic water oxidation to reduce protons (Weaver et al. 1980; Brand et al. 1989; Boichenko et al. 2004; Skjanes et al. 2008). If these organisms were to be utilized as photocatalysts, they would produce H2 fuel in a non-polluting, carbon-independent manner, and thus bypass the issue of CO2 emissions that plagues most of the biofuel industry (Fargione et al. 2008; Searchinger et al. 2008). Theoretically, production of H2 linked to photosynthetic water oxidation has a maximum solar to hydrogen (STH) conversion efficiency of about 10–13 % (Ghirardi et al. 2009; Walker 2009; Blankenship et al. 2011), which is lower than the 24 % that more expensive photovoltaic/electrolysis approaches can be expected to achieve (Blankenship et al. 2011; Hanna and Nozik 2006). However, in practice, photosynthetic microbes are only capable of photoproducing H2 from water for short periods of time, following prolonged (hours) anaerobic incubation. In order to achieve sustained H2 photoproduction, numerous challenges related to the organism’s metabolism need to be addressed (Dubini and Ghirardi 2014; Posewitz et al. 2008; Rupprecht 2009), the major being the O2 sensitivity of the hydrogenase catalysts. Extensive research has been done in the last 50 years to further understand, solve, or bypass this challenge. We will describe the past research and current state of the art in this review, and discuss perspectives for the future.

Hydrogenases

The photoproduction of H2 gas by microalgae was first discovered in the 1940s (Gaffron and Rubin 1942) and was the focus of extensive research starting in the 1970–1980s (Healey 1970; Kessler 1974; Godde and Trebst 1980; Maione and Gibbs 1986a, b; Ben-Amotz and Gibbs 1975; Roessler and Lien 1984). At that time, it was recognized that the various capabilities required for H2 production were distributed among photoautotrophic and non-photoautotrophic microbes in communities (Adams and Mortenson 1984; Adams et al. 1980; Vignais et al. 1985). These include: (a) hydrogenase-containing photoautotrophic microbes (green algae and cyanobacteria) and anaerobic hetero- or lithotrophic bacteria, in which the H2 production reaction is easily reversible, due to the isopotential or slightly negative ∆G 0 between hydrogenases and their electron donors; and (b) nitrogenase-containing cyanobacteria and other N2-fixing microbes, using a reaction that requires ATP as a co-factor and is essentially irreversible. Although promising, due to its irreversible nature, the ATP requirement lowers the potential solar to hydrogen (STH) conversion efficiency and renders nitrogenase-containing organisms inadequate as commercial H2 producers (Ghirardi et al. 2009). Indeed, STH values are directly correlated with the overall cost of the process, and technoeconomic analyses have been used by the U.S. Department of Energy to limit funding only to the most promising technologies (James et al. 2009).

Algal and cyanobacterial hydrogenases belong to different classes of enzymes. The first contain a [2Fe2S] catalytic cluster linked by a cysteine residue to a [4Fe4S] center (H-cluster), while the latter operate through a catalytic center comprised a [NiFe]-cluster attached to a conventional [4Fe4S] unit through 3 cysteine residues (Adams and Mortenson 1984; Adams et al. 1980; Vignais et al. 2001). [FeFe]-hydrogenases, in general, favor the H2 production reaction, while [NiFe]-hydrogenases are mostly involved in H2 uptake, the reversible reaction. This advantage of the green algal enzyme is counteracted by the fact that [FeFe]-hydrogenases are irreversibly inhibited by O2, while the O2-inhibition of [NiFe]-hydrogenases has a reversible character (English et al. 2009). The prompt inactivation of hydrogenases by O2 is considered to be a major limiting factor preventing sustained and efficient H2 production by photosynthetic microbes (Ghirardi et al. 2007; Goldet et al. 2009; Lambertz et al. 2011).

Algal hydrogenases were purified in the late 1980s (Roessler and Lien 1984; Happe and Naber 1993). It was found that Chlamydomonas harbors two separate hydrogenase genes that encode for two proteins, HYDA1 and HYDA2 (Forestier et al. 2003). The two proteins have molecular weights of about 47 kDa and interact physiologically with Ferredoxin (Winkler et al. 2009), which shuttles electrons to hydrogenases from either the photosynthetic electron transport chain or from the fermentative pyruvate-ferredoxin oxidoreductase (PFOR) enzyme (Hemschemeier and Happe 2011). To date, the differences in the physiological roles of the two hydrogenases are not known, since both are involved in fermentative and photoproduction of H2, as shown by Meuser et al. (2012) in their seminal paper describing the functions of single and double knock-out mutants of Chlamydomonas reinhardtii hydrogenases. Although the algal enzymes have not been crystallized in their mature form [the apoprotein’s structure was solved by Mulder et al. (2010)], homology models were constructed (Cohen et al. 2005a, b), based on the available crystal structures of the Clostridium pasteurianum (Peters et al. 1998) and Desulfovibrio desulfuricans (Nicolet et al. 1999) hydrogenases. These models show that the major differences between these two classes of [FeFe]-hydrogenases are (a) the lack of additional FeS centers in the algal enzymes; and (b) the presence of additional amino acid loops of unknown function in the algal hydrogenases. Interestingly, non-algal hydrogenases are about two orders of magnitude more tolerant to O2 inactivation (Cohen et al. 2005a); the hypothesis that their higher O2 tolerance was due to the lack of the additional FeS clusters was proved not to be true (King et al. 2006).

In contrast with algal [FeFe]-hydrogenases, cyanobacterial [NiFe]-hydrogenases are pentameric complexes, consisting of a large and small hydrogenase subunits (HoxH and HoxY) and three diaphorase subunits (HoxE, HoxF, and HoxU), the first of which is a membrane-bound protein. All three diaphorase subunits harbor additional FeS clusters, and they are responsible for creating an interface for interaction of the hydrogenase with NAD(P)H (Boison et al. 1998; Aubert-Jousset et al. 2011). Cyanobacterial hydrogenases, thus, are linked to both photosynthetic and respiratory electron transport chains (Cournac et al. 2004; Schmitz and Bothe 1996; Appel and Schulz 1996). Moreover, recent evident suggests that both ferredoxins and flavodoxins could act as electron donors to the cyanobacterial hydrogenase (Gutekunst et al. 2014). Besides the major enzymatic differences between [FeFe] and [NiFe]-enzymes described above, it must be noted that [NiFe]-hydrogenases are constitutively expressed, while the expression of [FeFe]-hydrogenases requires anaerobic induction. Both enzymes have similar turnover numbers when measured on an electrode surface (Jones et al. 2002; Armstrong 2004), although the redox potential of NAD(P)H is more positive than that of ferredoxin (−0.32 vs. −0.42 eV), which favors H2 production by [FeFe]-hydrogenases and H2-uptake by [NiFe]-hydrogenases at pH7 and standard conditions. Moreover, recent work (Maness et al. 2002) has shown that it is possible to purify truncated [NiFe]-hydrogenases containing only two subunits that are active in H2 production and are able to interact with ferredoxin. However, the expression of heterologous O2-tolerant hydrogenases in cyanobacteria has been extremely challenging, due to the fact that the maturation of [NiFe]-hydrogenases requires a large number of species-specific maturation enzymes (Ghirardi et al. 2014).

Research effort to understand the factors responsible for the high O2 sensitivity of hydrogenases is taking place in many parts of the world. The use of cyclic voltammetry to study hydrogenases made a major impact in the field (Armstrong 2004; Armstrong and Albracht 2005; Lamle et al. 2005; Leger et al. 2002; Vincent and Armstrong 2005; Vincent et al. 2005a, b), and provided a tool to measure the performance of hydrogenases under a variety of redox conditions, following exposure to H2, O2, or CO. These measurements confirmed previously observed major differences between [FeFe]- and [NiFe]-hydrogenases, which included the reversibility of inactivation by the latter (Vincent et al. 2007), and the existence of [NiFe] enzymes with lower sensitivity to O2 (Cracknell et al. 2008). Additionally, these studies, combined with more refined EPR and FTIR techniques, are still unveiling a series of redox states of [FeFe]-hydrogenases that have been shown to affect its catalytic activity and O2 sensitivity (Mulder et al. 2013; Stripp and Happe 2009), such as the fact that reduced hydrogenases and CO-treated [FeFe]-hydrogenases are less sensitive to O2 inactivation (Vincent et al. 2005b; Stripp et al. 2009), probably because of competition between H2, CO, and O2 for binding at the catalytic site.

Engineering O2-tolerant [FeFe]-hydrogenases

The O2 inactivation of hydrogenases has been the major factor precluding their commercial application (Dubini and Ghirardi 2014; Rupprecht 2009; Lee et al. 2010). Earlier research efforts demonstrated that the Chlamydomonas reinhardtii hydrogenases had an I50 of 0.3–0.4 % O2 following a 2-min incubation in vivo, and that using chemical mutagenesis one could select for strains with O2 I50s up to 9 times higher than the wild-type strain (Ghirardi et al. 1997; Seibert et al. 1998, 2001a). At the time, the lack of inexpensive genome sequencing technologies and of high-throughput screening approaches precluded the investigation of the genotype responsible for these results. However, these observations provided the evidence that it was possible to genetically alter the overall sensitivity of algal H2 production to O2.

The subsequent search for O2-tolerant [FeFe]-hydrogenases focused on two approaches: (a) the use of site-directed mutagenesis to address the origin of O2 sensitivity; and (b) the generation of random hydrogenase mutants followed by high-throughput screening for H2-producers in the presence of or following exposure to O2. These studies were only possible due to the discovery of the maturation proteins required for assembly of the [FeFe]-hydrogenases catalytic site (Posewitz et al. 2004a) and the demonstration that they were the only requirement for expression of the structural genes in E. coli (King et al. 2006) or in Clostridium acetobutylicum (Girbal et al. 2005; von Abendroth et al. 2008). This capability opened up the door for studies requiring large amounts of these enzymes. Indeed, by co-expressing the maturation genes with the [FeFe] structural hydrogenase genes, it has been possible to detect heterologous expression of the Clostridium acetobutylicum hydrogenases in Synechococcus elongatus (Ducat et al. 2011) and of the Shewanella MR-1 hydrogenase in the heterocystis of Anabaena PCC 7120 (Gartner et al. 2012). Similarly, the development of high-throughput screening methods to detect H2 production by individual colonies proved essential for progress in this area. Among the various techniques devised, the following have been successfully used by various research groups to screen large populations of recombinant hydrogenases of H2-producers:

  1. (a)

    Chemochromic films consisting of layers tungsten oxide (WO3) and platinum, which turn blue as the WO3 becomes reduced (Seibert et al. 2001b); the response time of the films is fast and they are able to sense down to 1 nmol H2 emitted from a 3-mm-diameter colony. However, they are not available commercially and their signal is transient. The films have been successfully used to screen chemical mutagenesis-based libraries of Chlamydomonas for mutants unable to produce H2 (Seibert et al. 1998, 2001b), and they identified novel factors required for expression of an active hydrogenase from random insertional mutagenesis libraries (Posewitz et al. 2004a, b).

  2. (b)

    Sulfonated Wilkinson’s catalyst coupled with a tetrazolium indicator. This method is less sensitive than the WO3 films, detecting down to 20 nmol H2 in microwell plates. It has been used to detect H2 production by cyanobacterial libraries (Katsuda et al. 2006; Schrader et al. 2008) and for identifying algal strains with lowered Photosynthesis/Respiration capacity ratios (Ruhle et al. 2008; Hemschemeier et al. 2008)—see “By-passing the O2 sensitivity of [FeFe]-hydrogenases” section for context.

  3. (c)

    Reduction of resazurin to resorufin by H2 (Stapleton and Swartz 2010a) or production of H2 by reduced ferredoxin or methyl viologen (Bingham et al. 2012a; Stapleton and Swartz 2010b). Since resazurin is not specific for H2 and can be reduced by other metabolic intermediates, the assay’s specificity was optimized by absorption of the dyes to beads displaying recombinant hydrogenases. This method was used for detecting O2-tolerant, H2 uptake activity by libraries of recombinant hydrogenases.

  4. (d)

    Biosensor based on the H2-sensing promoter of Rhodobacter uptake hydrogenases coupled to the GFP marker. This method relies on an overlay of GFP-expressing Rhodobacter that fluorescences when in contact with H2-producing single colonies of various organisms (Wecker et al. 2011). The assay is extremely sensitive to H2, and can detect fermentatively produced H2 gas as well. It has been successfully used to identify Chlamydomonas strains that photoproduce H2 upon exposure to high light, conditions that normally inhibit the wild-type activity under the assay conditions (Wecker and Ghirardi 2014).

A major hypothesis regarding the factors responsible for the high O2 sensitivity of [FeFe]-hydrogenases was derived from comparative studies of hydrogenases from different organisms, which revealed the presence of putative hydrophobic gas channels connecting the surface of the protein to the catalytic site. Such channels have been proposed to be pathways by which O2 and/or H2 diffuse in and out of a protein (Cohen et al. 2005a; Montet et al. 1997). These features suggested that O2 sensitivity could be altered by interfering with the gas diffusion pathway, as long as the rate of diffusion is limiting the overall H2 production reaction, compared to the rate of catalysis (Fontecilla-Camps et al. 2007). In order to further investigate this hypothesis and apply it to [FeFe]-hydrogenases, molecular dynamic simulations (MD), volumetric solvent accessibility maps (VOAM), and potential energy (PMF) plots of the diffusion of O2- and H2-sized particles from the surface to the catalytic site of clostridial and algal [FeFe]-hydrogenases were performed (Cohen et al. 2005a, b).

The studies indicated that there were two major H2 diffusion pathways converging at a large central cavity located near the catalytic site of [FeFe]-hydrogenases, while a small percentage of the H2 molecules diffused through other less-defined pathways, suggesting that the Clostridium pasteurianum CpI hydrogenase is fairly porous to H2 gas. In contrast, O2 molecules diffuse at much slower rates through the same pathways and often stay within the channels, without equilibrating with the solvent. The O2 diffusion motion suggested that it depends on the opening/closing of transient channels. Indeed, when the enzyme was probed for transient cavities using the VOAM method, it was found that the two major pathways seen by the MD simulations correspond to major trajectories predicted by VOAM. These structures seem to correspond to transient protein conformations that are formed spontaneously, as a result of the equilibrium dynamics of the protein structure. Moreover, VOAMs were used to identify specific residues that might be responsible for transiently blocking gas diffusion along each pathway. The relative relevance of each residue was assessed through PMF maps. It was found that the major barrier to O2 diffusion along either pathway was at the intersection of both pathways, while minor barriers (or wells) were observed along both pathways. Specific residues in CpI were selected for mutagenesis, aimed at replacing them with bulkier residues. However, only minor improvements in O2 tolerance were detected with the L283 W mutant, accompanied by decreased activity and increased O2 sensitivity in all other single or double mutants (unpublished). A few years later, an explanation for the lack of success in generating high H2-producing, O2-tolerant mutants at those specific positions appeared, when Mulder et al. (Mulder et al. 2010) showed that the residues lining the hydrophobic cavity in the algal hydrogenases were required for the proper folding of the apoprotein; mutation of any of the residues would be expected to prevent maturation of the protein and result in loss of activity, improper protein folding, and higher permeability to O2. Interestingly, the narrowing of gas channels to slow down O2 inactivation has been successful when implemented with [NiFe]-hydrogenases (Dementin et al. 2009), as described in more detail in “Engineering O2-tolerant [NiFe]-hydrogenases ” section.

Swartz’ group at Stanford took advantage of in vitro cell-free translation systems and used directed evolution to generate hydrogenase mutants, followed by the methyl viologen assay [see (c) above] to screen for H2 uptake activity after exposure of the enzyme to O2 (Stapleton and Swartz 2010a, b). After a series of rounds of mutagenesis using error-prone PCR to amplify the CpI gene, the authors detected 20 mutants with H2 uptake activity levels above the CpI control but only one mutant with decreased O2 sensitivity. Among the 13 mutations present in this strain, 3 were found to be responsible for the higher O2 tolerance phenotype. Site-directed saturation mutagenesis was then used to optimize the phenotype. However, neither of the mutants showed increased O2 tolerance during active H2 production (Bingham et al. 2012b).

All these discouraging results, however, have been accompanied by progress in the understanding of the factors that confer O2 tolerance to [NiFe]-hydrogenases (see “Engineering O2-tolerant [NiFe]-hydrogenases” section), suggesting that cyanobacteria may be a more appropriate organism for expressing heterologous O2-tolerant hydrogenases.

Engineering O2-tolerant [NiFe]-hydrogenases

In contrast with [FeFe]-hydrogenases, the O2 sensitivity of [NiFe]-hydrogenases has been found to be amenable to molecular engineering, and O2 tolerance was conferred to the Desulfovibrio fructosovorans hydrogenase through two major strategies: a reduction in O2 access to the active site (as described in “Engineering O2-tolerant [FeFe]-hydrogenases ” section with respect to [FeFe]-hydrogenases), and scavenging O2 from the active site by methionine residues located close to it (Dementin et al. 2009; Leroux et al. 2008). These modifications were achieved by exchanging a valine residue within the enzyme’s hydrophobic channel by a methionine, and by substituting another residue near the active site by a second methionine. Both modifications were shown to additively contribute to spontaneous enzyme reactivation after O2 exposure. These genetic modifications were shown to be further associated with changes in the gas diffusion characteristics and they did affect the directionality of the enzyme (Abou Hamdan et al. 2012). Recent reports have shown that a similar strategy could be developed on the hydrogenase of Synechocystis, resulting in the expression of an engineered hydrogenase exhibiting a sustained activity in the presence of O2 with a bias toward production (Cano et al. 2014). Other engineering strategies involving different targets can be considered and might even be required as part of a larger scale combination of unique redesigning in order to obtain an efficient O2-tolerant hydrogenase. Indeed, O2 tolerance has also been tightly linked to electron transfer reactions involving the proximal Fe-S cluster (Fargione et al. 2008; Wu et al. 2011; Lukey et al. 2011) which can be altered through its coordinating cysteine residues (Fargione et al. 2008; Pandelia et al. 2011). Mechanisms involving water molecules and protons transport pathways have also been suggested to play a role in [NiFe]-hydrogenase O2 tolerance (Cornish et al. 2011; Sumner and Both 2012).

Other approaches toward efficient biohydrogen photoproduction include the heterologous expression of [NiFe]-hydrogenases more tolerant toward O2 in cyanobacteria, which has proved to be particularly difficult, since their maturation proteins seem to be extremely species specific (English et al. 2009) and, to date, very few successful efforts have been reported (Weyman et al. 2011). Strategies aiming at sequestrating the intracellular O2 have also been contemplated (Shestakov and Mikheeva 2006), but no reports of such approaches are currently available.

Bypassing the O2 sensitivity of [FeFe]-hydrogenases

A physiological method to bypass the O2 inactivation of algal hydrogenases was first demonstrated in 2000 by Melis et al. (2000). These studies relied on the effect of sulfur deprivation on algal physiology. When sulfur is depleted from the growth medium, Chlamydomonas cultures gradually lose their O2-evolution capacity, over-accumulate starch, and eventually become anaerobic. Anaerobiosis induces both the expression of the hydrogenase enzyme and starch degradation. As a result, the culture produces H2 in the light using reductant from the residual water oxidation activity and from the indirect, starch degradation-linked photosynthetic pathway. Recent papers showed a similar effect by depriving the cultures of other nutrients, such as nitrogen (Philipps et al. 2012) which was shown to also induce H2 production but not always as promptly or as specifically as the sulfur-deprivation method.

The sulfur-deprivation process consists of two phases: a photosynthetic phase, during which O2 evolution is gradually inhibited and starch overaccumulates; and a H2 production anaerobic phase that is characterized by three major metabolic activities—residual O2 evolution and H2 photoproduction from water oxidation, respiration of organic substrates, and starch fermentation, yielding additional H2 gas and other fermentative products such as formate, acetate, and ethanol (Kosourov et al. 2003). The effects of sulfur deprivation on algal physiology have been extensively studied; the loss of O2 evolution was attributed to the inhibition of D1 (a major protein forming the Photosystem II reaction center) turnover, which requires sulfurylated amino acid residues, followed by loss of Rubisco, the most abundant protein in Chlamydomonas. It was hypothesized that the latter was a result of an attempt by the organism to recycle its sulfurylated residues from abundant proteins in order to preserve its minimum metabolic functions and survive the deprivation period ((Zhang et al. 2002). Various studies aimed at deconvoluting the complexities associated with the process and optimizing the H2 production yield of sulfur-deprived cultures, either by engineering additional properties into Chlamydomonas or by further manipulating the process followed. It was observed that, although the original method was implemented under photoheterotrophic conditions (acetate was present in the growth medium), the process could be reproduced under photoautotrophic conditions (using CO2 as the only carbon source), as long as optimum conditions were maintained for starch accumulation and degradation (Tolstygina et al. 2009).

Indeed, one of the major questions regarding the sulfur-deprivation method is the role of starch. Many reports showed that DCMU, an inhibitor of O2 evolution, partially inhibited H2 production when added at the beginning of the anaerobic phase, and the relative level of DCMU inhibition decreased during the 3–4-day duration period (Laurinavichene et al. 2004). This suggested that the contribution from the direct (reductant originated from PSII-catalyzed water oxidation) and indirect (reductant originated from the initial steps of starch degradation and transferred to the photosynthetic plastoquinone pool by the NAD(P)H/plastoquinone oxidoreductase) pathways varies during the process. The H2 photoproduction capability of starch-less mutants sta6 (Zabawinski et al. 2001) and sta7 (Posewitz et al. 2004) following dark anaerobic induction showed transient and low levels of H2 photoproduction; however, wild-type rates of H2 photoproduction by both starch-less mutants were detected under sulfur deprivation (Chochois et al. 2009). The latter was explained by the observation that significant acetate consumption occurs in starch-deficient mutants during sulfur deprivation, and acetate consumption is responsible for maintaining anaerobiosis and providing reductant through the indirect H2 photoproduction pathway during sulfur deprivation. Interestingly, though, the low H2 production of starch-less mutants under dark anaerobic conditions correlates with low hydrogenase genes transcription (Posewitz et al. 2004). It was hypothesized that, since electrons generated from starch degradation reduce the PQ, this triggers a regulatory process that affects the expression of various photosynthetic components (Escoubas et al. 1995) (Jans et al. 2008), including hydrogenases (Hemschemeier et al. 2008; White and Melis 2006). Alternatively, based on inhibitor studies under sulfur-deprivation conditions, Chochois et al. (2009) proposed that the regulation of hydrogenase activity occurred instead by the proton gradient generated by cyclic electron flow during the process, thus confirming the ATP requirement for hydrogenase induction reported previously by others (Lien and Pietro 1981).

Curiously, two Rubisco mutants that are unable to fix CO2 have been reported to exhibit either low (White and Melis 2006) or high H2 photoproduction activity under sulfur deprivation (Hemschemeier et al. 2008). Since neither mutant accumulates starch, it was finally agreed that the two different phenotypes were due to the production of an unstable truncated Rubisco, while the other completely lacks starch (Hemschemeier et al. 2008).

Although not directly related to the O2 sensitivity limitation, the non-dissipation of the proton gradient during H2 photoproduction and the induction of state transitions (which direct light energy preferentially toward PSI and CEF) during anaerobiosis are major issues responsible for the low rates of the process (Dubini and Ghirardi 2014; Hankamer et al. 2007). A recent manuscript (Tolleter et al. 2011) describes the effects of genetically uncoupling the proton gradient from photosynthetic electron transport, which is accompanied by an increase in the rates of H2 photoproduction, either under anaerobically induced or sulfur-deprived conditions. Similarly, efforts to eliminate state transitions while increasing the rates of respiration have successfully led to higher rates of H2 photoproduction ((Kruse et al. 2005).

The success of partial inactivation of Photosystem II activity to achieve culture anaerobiosis led to the design of alternative processes to inhibit D1 turnover. Initial experiments targeted the sulfate permease enzyme, which translocates sulfate from the cytosol into the chloroplast. Sulfate permease mutants had a phenotype similar to that of sulfur-deprived wild-type algae and produced H2 at similar levels (Chen et al. 2005). A different approach focused on the feasibility of using D1 mutants with reported lower O2 evolution capability as H2-producers in closed photobioreactors (Makarova et al. 2007). Unfortunately, the specific mutants used in this study were unable to over-accumulate starch, a sine-qua-non for efficient H2 photoproduction during anaerobiosis. Markov et al. (Markov et al. 2006) demonstrated sustained H2 photoproduction for a total of about 90 min by algal cells previously photoinhibited to decrease their O2 evolution capability. More recently, Surzycki et al. (2007) devised a cyclic process based on the properties of cytochrome c promoter sequence, which is responsive to anaerobiosis and low Cu levels, as a regulator expression of the nucleus-encoded Nac2 chloroplast protein. The latter is required for stable accumulation of the psaD RNA (encoding the reaction center D2 protein) and, if it is not expressed, Photosystem II activity is inhibited. The cycle operates by manipulating the copper concentration in the growth medium; in the presence of copper, PSII is inhibited, the culture gradually becomes anaerobic, the hydrogenase is induced, and H2 production is observed. Hydrogen accumulates for a total of about 1.5 h, at which point anaerobiosis induces the synthesis of the Nac2 protein and PSII activity is restored. The cycle can be repeated upon re-addition of Cu2+ to the medium, and the authors demonstrated the occurrence of two successive cycles during a total period of 50 h. These results compare well with the cyclic production of H2 using sulfur deprivation (Ghirardi et al. 2000), which showed three cycles over 350 h of operation.

Recent reports show that the engineered D1 double mutant in which L159 was replaced by isoleucine and N230 by tyrosine exhibits higher rates of photosynthesis and respiration, lower Chl and higher D1 content (but only slightly higher Chl a/b ratio, resembling high light-grown cells), and higher rates of H2 photoproduction under sulfur deprivation than the common cc124 wild-type strain (Scoma et al. 2012; Torzillo et al. 2009). The two mutated amino acid residues are located, respectively, near Tyrosine YZ (the primary electron donor to Photosystem II), and in the loop between helixes IV–V. The double mutant accumulates more biomass during photomixotrophic growth and has higher levels of zeaxanthin during the H2 production phase, suggesting higher photoprotection by the violoxanthin/zeaxanthin cycle, resulting in a longer H2 production phase.

Finally, a novel method to bypass the O2-sensitivity property of hydrogenases was reported by Wu et al. (2011). The authors expressed two codon-optimized O2-binding proteins, leghemoglobin (from Glycine max), and ferrochelatase (from Bradyrhizobium japonicum) in the chloroplast of Chlamydomonas. Both genes were shown to be transcribed and translated, although biochemical activity measurements were unsuccessful. Nevertheless, faster O2 consumption under sulfur deprivation and higher H2 yields were reported, although the strain chosen for transformation is a known low H2-producer and the levels of the transgenes were still too low.

It is difficult to compare literature data on the application of sulfur deprivation to H2 photoproduction, since different researchers use different cell densities, report either maximum or average rates, and are able to sustain the process for different amounts of time. Table 1 summarizes the highest rates that have been currently achieved using a combination of sulfur deprivation with other genetic or cultivation approaches to increase productivity. It is important to point out that neither of the reported processes has achieved economic parity with the current cost of gasoline. More details about the genotypes of each mutant can be found in the original reference (4th column).

Table 1 Rates, yield, and long-term duration of H2 photoproduction by different mutants of Chlamydomonas reinhardtii under sulfur-deprivation conditions
Full size table

Future directions

Although most of the research to date has focused on Chlamydomonas reinhardtii and Synechocystis sp. PCC 6803 as model organisms, H2 photoproduction has been observed in other algal species (Brand et al. 1989; Boichenko et al. 2004; Skjanes et al. 2008; Timmins et al. 2009; Wang et al. 2011). It is interesting to notice that, instead of using sulfur deprivation, a large number of researchers cultivate algae under low light intensity in a sealed reactor, as an alternative method to induce anaerobiosis (Wang et al. 2011; Maneeruttanarungroj et al. 2010). Although very useful from a research point of view, such systems are incompatible with commercial applications, where decreases in light intensity require larger areas of cultivation which would increase capital and operational costs (James et al. 2009; Hankamer et al. 2007).

The limited success in generating organisms that photoproduce H2 from water at high STH conversion efficiency by genetically engineering hydrogenases or through physiological approaches to prevent O2 accumulation is forcing the community to re-evaluate the relevance of O2 sensitivity in limiting H2 photoproduction vis a vis other known barriers. Those have been mentioned above and were discussed in detail by Dubini and Ghirardi (2014). Suffice it to say that success will involve engineering multiple traits into a single organism, and that various research groups are currently involved in approaches to achieve this goal, as shown in Table 1. A recent report on the transformation of the algal [FeFe]-hydrogenase in the Chlamydomonas chloroplast (Reifschneider-Wegner et al. 2014) of a double hydrogenase knock-out strain (Meuser et al. 2012) may represent a potential useful method to increase the availability of the enzyme, which may be a rate-limiting step in H2 production. However, current expression levels and rates are still too low and need to be substantially increased to demonstrate its benefit over the wild-type strain.