Strategies for improving biological hydrogen production

Bioresource Technology 110 (2012) 1–9

Contents lists available at SciVerse ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Review

Strategies for improving biological hydrogen production Patrick C. Hallenbeck ⇑, Mona Abo-Hashesh, Dipankar Ghosh Département de microbiologie et immunologie, Université de Montréal, CP 6128, Centre-ville, Montréal, Canada PQ H3C 3J7

a r t i c l e

i n f o

Article history: Received 2 December 2011 Received in revised form 15 January 2012 Accepted 19 January 2012 Available online 28 January 2012 Keywords: Biohydrogen Dark fermentation Metabolic engineering Hydrogen yields Biophotolysis

a b s t r a c t Biological hydrogen production presents a possible avenue for the large scale sustainable generation of hydrogen needed to fuel a future hydrogen economy. Amongst the possible approaches that are under active investigation and that will be briefly discussed; biophotolysis, photofermentation, microbial electrolysis, and dark fermentation, dark fermentation has the additional advantages of largely relying on already developed bioprocess technology and of potentially using various waste streams as feedstock. However, the major roadblock to developing a practical process has been the low yields, typically around 25%, well below those achievable for the production of other biofuels from the same feedstocks. Moreover, low yields also lead to the generation of side products whose large scale production would generate a waste disposal problem. Here recent attempts to overcome these barriers are reviewed and recent progress in efforts to increase hydrogen yields through physiological manipulation, metabolic engineering and the use of two-stage systems are described. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction The world is turning to a search for clean energy sources to mitigate coming climate change and the impending shortage of readily available fuel to provide the energy necessary for present and projected human activities. A variety of possible fuel sources are being examined at present. Among these, many have proposed using hydrogen as an energy carrier in a future hydrogen economy. However, a sustainable, renewable supply of hydrogen to power this economy is required. One option would be to use biological means to produce hydrogen. One approach would be to recruit the power of photosynthesis to capture sunlight and split water, a process that is called biophotolysis (Hallenbeck and Benemann, 2002; Hallenbeck, 2011). This approach, although attractive, suffers from major challenges that may require years of R&D to overcome. Photofermentation, use of bacterial photosynthesis to capture light energy and use it to drive hydrogen evolution from otherwise inaccessible substrates, can be used to show nearly complete substrate conversion to hydrogen (Hallenbeck, 2011; Keskin et al., 2011; Adessi and de Philippis, 2012). Nevertheless, it suffers from a number of limitations, including low light conversion efficiencies. These same substrates can be converted to hydrogen in microbial electrolysis cells, and although great advances have been made in both understanding the underlying mechanisms and in improving device performance, a number of challenges, as discussed below, remain. ⇑ Corresponding author. Tel.: +1 514 343 6278; fax: +1 514 343 5701. E-mail address: [email protected] (P.C. Hallenbeck). 0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2012.01.103

Finally, another approach, microbial fermentative production of hydrogen, draws upon already existing natural metabolic capacities. Moreover, it would use a variation of well-known technology that is already widely employed at industrial scale and hence is probably more likely to be realizable in the short term. A wide variety of potential substrates, wastes and biomass, are present in sufficient quantities to make a significant contribution to meeting energy needs if they were efficiently converted to hydrogen by fermentation. Nevertheless, there are significant challenges to the practical implementation of such a process (Hallenbeck and Ghosh, 2009; Hallenbeck, 2011). Here the current status of different strategies that are being used in attempts to improve the rates and yields of biohydrogen production are reviewed and strategies for improvement are discussed. Some of the advantages and disadvantages of each approach as well as some of the reported rates and yields are given in Table 1. Additional energy can be derived from the side products of fermentation (organic acids) by developing two stage systems (Hallenbeck, 2011; Adessi et al., 2012). Various second stages are under active investigation including; conversion to methane, photofermentation to hydrogen, and microbial electrolysis, where a small input of electricity drives conversion to hydrogen to completion. In addition, several metabolic engineering approaches are underway to develop a process capable of higher hydrogen yields in a single stage. These include; providing metabolic energy to overcome thermodynamic barriers, expression of heterologous proteins, including hydrogenases, and rerouting metabolism to achieve more complete substrate degradation and increased electron flux for proton reduction.

2

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

Table 1 Comparison of the Different Biohydrogen Processes. Process

Production rates (mls H2/ l/h)

Yields

Advantages

Disadvantages

Future prospects

Biophotolysis

2.5–13a

<0.1%b

Abundant, inexhaustible substrate (water)

Evolves oxygen, destroying the hydrogen evolving catalyst (hydrogenase) Low photosynthetic conversion efficiencies Potentially explosive gas mixtures formed Large surface areas required

Near term incremental improvements possible through creation of antenna mutants

Totally carbon independent pathway Simple products, hydrogen and oxygen

Photofermentation

12–83c

<1%d, 80%e

Uses readily available waste streams Nearly complete substrate conversion Can extract additional hydrogen from dark fermentation effluents

Dark fermentation

10–15  103

33%f

Can use a variety of waste streams Simple reactor technology, nonsterile conditions acceptable high rates achieved with immobilized mixed cultures

Need for inexpensive photobioreactors Low volumetric rates of production Low efficiency hydrogen production by nitrogenase Low photosynthetic conversion efficiencies Need for inexpensive photobioreactors Large surface areas required large amount of byproducts low COD removal

Immobilization might bring some improvement Creation of an oxygen resistant hydrogenase would be a breakthrough Materials science breakthrough Strain improvement through metabolic engineering replacement of N2ase with H2ase Near term improvement possible through creation of antenna mutants Materials science breakthrough

metabolic engineering could achieve breakthrough in metabolic limitations Two stage systems can extract additional energy, decrease COD

reactor to reactor variation

a

Sulfur-deprived green algae (Laurinavichene et al., 2006) and cyanobacteria (Tsygankov et al., 1998). Conversion of total incident light energy to hydrogen at full solar power. c (Eroglu et al., 1999; Kim et al., 2006). d At low (relative to full solar) light intensities (Abo-Hashesh et al., 2011b). e Conversion of substrate (organic acid) to hydrogen, does not account for light energy used. f 4 mol of hydrogen per mole of glucose equivalent, theoretically 12 mol are available. There appears to be an inverse relationship between hydrogen production rates and yields, so the high rate reactors giving the quoted high volumetric rates (Lee et al., 2006; Wu et al., 2007) have yields significantly lower than this. b

2. Light-driven biohydrogen production At first glance, the use of biological systems to capture solar energy and convert it to energy in the form of hydrogen is inherently appealing. In fact, two different types of microbial photosynthesis are available for the task; biophotolysis using plant-type photosynthesis and photofermentation using bacterial photosynthesis. Some of the mechanistic details behind both biophotolytic hydrogen production and photofermentation have recently been reviewed elsewhere (Hallenbeck, 2011, 2012; Eroglu and Melis, 2011) and will not be discussed in great detail here. A large compilation of rates for both types of photon driven hydrogen production, in terms of mls of hydrogen per liter of reactor, has recently been published (Eroglu and Melis, 2011). These vary more than one hundred fold, from about 1 ml l 1 h 1 to more than 100 ml l 1 h 1, with production rates of biophotolytic systems generally being significantly lower than those of photofermentation systems. Unfortunately, an areal measure of production, permitting the calculation of the efficiency of solar energy conversion, would be a much more useful measure. This is because ultimately what matters is the surface area that intercepts the incoming radiation that must be covered with photobioreactors to produce a given quantity of hydrogen (moles of hydrogen produced per Watts m 2 or lE m 2 s 1). 2.1. Water-splitting systems (Biophotolysis) Light-driven biophotolysis has many attributes that make it a promising concept. The natural capacity of microbial photosynthesis, either microalgal or cyanobacterial, is used to capture solar energy and split water. Highly reducing electrons produced by

photosystem I can then be used to reduce a ferredoxin that can drive hydrogen evolution by a hydrogenase enzyme (Hallenbeck, 2011). Since the overall machinery is a living cell, there would be an inherent capacity for stability and self-renewal. The substrate, water, is naturally abundant, especially since clean water is not required, and in fact some types of wastewater would be preferred as they would contain the necessary growth nutrients. However, despite this promise, and several decades of research, serious technical barriers to a practical application of this technology persist (Table 2) (Beer et al., 2009; Ghirardi and Mohanty, 2010; Hallenbeck, 2011, 2012; Tsygankov, 2012). For one thing, light utilization, in particular at full solar power, is inefficient due to a number of factors. The efficiency of light utilization by an individual cell is compromised by the fact that photosynthetic antenna sizes are usually optimized for sub-maximal light intensity leading to an appreciable amount of energy dissipation after superfluous light capture at high light intensities. Coupled with self-shading, limiting effective light penetration into cultures, overall photosynthetic efficiencies are no more than 1% and often closer to 0.1%. This problem could possibly be somewhat circumvented by the creation of antenna mutants which reduce antenna sizes to those most efficient at high light intensities. This is discussed in more detail in the next section. Various studies have estimated that efficiencies of the order of 10% must be achieved to have an economically viable process. Of course, low efficiencies translate into a requirement for a larger surface area to obtain the same amount of energy. This is critical since fabrication and installation of low-cost transparent hydrogen impermeable photobioreactors is already very problematic. Sturdy, long-lived materials that meet these specifications are simply not available at present. Development of such materials would require a breakthrough in materials sciences and

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

3

Table 2 Strategies for improving light-driven biohydrogen production. Challenges for implementation Biophotolysis Low efficiencies of light conversion Oxygen inhibition of hydrogen production High cost of photobioreactors Photofermentation Low efficiencies of light conversion High cost of photobioreactors

Possible workarounds Reduce antenna size Create oxygen tolerant hydrogenase Decouple oxygen and hydrogen evolution through two stage or two phase process Develop high-tech hydrogen impermeable plastics Reduce antenna size Replace nitrogenase with hydrogenase Develop high-tech hydrogen impermeable plastics

engineering, but is essential if biophotolysis is to be implemented on a practical level. Finally, simultaneous evolution of oxygen and hydrogen requires a proton activating catalyst that can function at saturating, or even super saturating oxygen levels. Unfortunately, the hydrogenases with the highest activities, [FeFe] hydrogenases, are highly oxygen sensitive and irreversibly destroyed by short term exposure to low concentrations of oxygen. This represents the major fundamental obstacle to this approach, and efforts are underway to attempt to engineer hydrogenases that are less sensitive to oxygen, as detailed in the following section. Some years ago this field received a great deal of interest through the demonstration of prolonged hydrogen production with sulfur-deprived cells of the green alga, Chlamydomonas reinhardtii (Melis et al., 2000), and much further research along these lines has been performed. However, despite much enthusiasm, and even the occasional establishment of companies to exploit this process commercially (http://www.solarvest.ca/), it seems that this approach, although a nice demonstration of principles, is destined to remain a laboratory curiosity. This is because lack of oxygen inhibition is achieved by reducing photosynthesis 90%, at which point the culture is anaerobic since the small amount of oxygen produced is consumed by respiration. The net effect is to reduce efficiencies, already low, ten-fold, and since normal efficiencies are ten-fold too low for a practical process, a difference of 100 fold in efficiencies must now be overcome. Recent work has shown that somewhat higher efficiencies (0.9% PAR) can be obtained with alginate entrapped cells (Kosourov and Seibert, 2009) but this is only at low light intensities, not readily applicable to a practical system. In general, the sulfur deprivation regime appears more applicable to green algae than to cyanobacteria, for which one can in some cases show low level hydrogen production without it. For example, the filamentous non-heterocystous cyanobacterium, Lyngbya perelegans, produced hydrogen (maximum reported: 19.4 mmol g 1 (dry wt) when incubated under an argon atmosphere. Highest production rates were noted at mid-exponential phase, probably due to the accumulation of appreciable amounts of glycogen at this point in cultures incubated under optimal conditions (Kaushik and Anjana, 2011). On the other hand, hydrogen production by Spirulina platensis has been noted after induction of severe chlorosis mediated reduction in photosynthetic oxygen evolution brought about by sulfur deprivation (Morsy, 2011). In the green alga Chlamydomonas, the vast majority of studies have been carried out under sulfur deprivation conditions, usually required for the reduction of photosynthetic oxygen evolution that occurs as PSII complexes are degraded. However, one recent study (Wang et al., 2011) showed that hydrogen evolution can occur without sulfur deprivation in Chlorella pyrenoidosa. However, this only occurs under very low light intensities (<30 lE m 2 s 1), well below the compensation point (light intensity where photosynthetic oxygen evolution meets respiratory demand). As well, at these very low light intensities, the corresponding hydrogen production rate was extremely low, 6 lmol H2 l 1 h 1. Indeed, the challenges to be met with the much studied sulfur-deprived C. reinhardtii system are formidable. For example, in a recent study

(Tamburic et al., 2011) in which a tubular flow photobioreactor with a large surface to volume ratio was used and key parameters, including, pH, dissolved oxygen, light intensity, temperature, and optical density were all optimized, a cumulative yield of only 3 ml H2 per liter of culture (over 120 h) was achieved, representing a dismal light conversion efficiency of <0.1%, and an extremely low hydrogen production rate, 1.04 lmol H2 l 1 h 1. In a novel approach to improving overall hydrogen production by sulfur-deprived Chlamydomonas, codon optimized ferrochelatase (hemH) and leghemoglobin (iba) were expressed in the chloroplast (Wu et al., 2011). In theory, this could improve performance and stability as these proteins would help insure a very low partial pressure of oxygen within the cells while at the same time possibly permitting oxygen delivery to the respiratory complexes in the membrane. The mutant strain showed four times the hydrogen production rates as the wild-type, 19.1 lmol H2 l 1 h 1 but at fairly low (and thus favorable) light intensities 50 lE m 2 s 1 (PAR). Thus, systems producing hydrogen through biophotolysis (water splitting photosynthesis) fall far short of what is needed to develop a practical system capable of converting captured solar energy into hydrogen. As was presented above, there are a number of barriers to achieving substantial system improvement. As well, there are a number of strategies that are guiding current efforts to overcome these barriers and they are discussed below. 2.2. Strategies for improvement 2.2.1. General strategies for improvement of photosynthetic efficiencies Strategies for improving hydrogen yields from biophotolytic systems fall into two categories; those that are general and aimed at increasing photosynthetic efficiencies, and those that specifically seek to increase hydrogen production. Some of what follows has also been discussed in recent reviews (Kruse and Hankamer, 2010; Srirangan et al., 2011). Photosynthetic efficiency, the quantum requirement for production of the desired product, is of concern no matter what the specific biofuel end product. There are two main areas that present targets for potential improvement; increasing the total spectrum that is captured, and increasing the quantity of light that is captured and productively used at high light intensities. The efficiency of conversion of total incident light energy to useful chemical energy is a direct reflection of how effective the light reactions of photosynthesis are. Many different factors reduce the amount of energy that can be obtained from incoming light, the basic efficiency train of photosynthesis, and some of these are seemingly immutable (Table 3). For one thing, substantial energy in incident solar radiation cannot be used because it falls outside the absorption abilities of the photosynthetic pigments (chlorophyll a) that make up photosystem II (PSII) and photosystem I (PSI). One radical proposition is to make a new hybrid photosynthetic organism where additional optical bandwidth is accessed by combining a normal PSII (chlorophyll a) with a bacterialchlorophyll containing PSI (Blankenship et al., 2011). Since bacteriochlorophyll a absorbs light at wavelengths inaccessible to chlorophyll a an additional fraction of the solar spectrum becomes

4

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

Table 3 Efficiency train for photosynthesis. Factor Total incoming solar flux Absorbtion abilities of photosynthetic pigments Quantum requirement for product formation Inefficiency at high light intensity

Percent of total incident remaining (%) 100 45 12.2 3

available for conversion to chemical energy, thus potentially increasing the energy efficiency of this primary step. It remains a formidable task to put this into practice. However, energy absorption is also inefficient because the size of the photosynthetic unit may not be appropriate for the incident light intensities. This is especially true at high light intensities where many more photons are captured than can be fruitfully converted to chemical energy and the excess, which can be as high as 80%, are dissipated as fluorescence or thermal energy. Thus, the thinking has been that creating strains with a smaller sized antenna would increase photosynthetic efficiencies, and indeed this seems to be the case (Polle et al., 2002). However, this has not necessarily translated into increased efficiency of hydrogen production, at least with the sulfur-deprived green algal system, because, as discussed above, hydrogen production by this system depends upon decreasing photosynthetic capacity, harder to do in a more photosynthetically efficient strain. Nevertheless, cell entrapment permitted the demonstration of increased efficiencies (although still quite low at 0.08%) at moderately high light intensities (285 lE m 2 s 1) (Kosourov et al., 2011). This question has also been addressed in another manner. Hydrogen production was examined in a D1 mutant with reduced total chlorophyll (dry weight basis) and a higher photosynthetic capacity (Torzillo et al., 2009). Indeed, the mutant strain showed interesting properties with regard to hydrogen production, with a substantially higher hydrogen production rate, 240 lmol H2 l 1 h 1 versus 88 lmol H2 l 1 h 1, with a total accumulation of H2 by the mutant that was much greater than wild-type, 504 versus 80 ml, due to a combination of higher hydrogen production rate and a much longer period of production, 285 h versus 53 h (Torzillo et al., 2009). Another factor that might impinge on the overall light-dependent reactions is a limitation in re-reduction of the primary electron donor of photosystem I (PSI). During linear, non-cyclic photosynthesis this is accomplished by an electron that is ultimately derived from the water-splitting reaction of PSII. However, this electron must reach PSI by traveling through a connecting electron transfer chain that normally serves to couple electron transfer to proton translocation, thereby creating a proton gradient which can subsequently be used for ATP synthesis. One thought has been that if the proton gradient is not dissipated quickly enough through ATP synthesis, overall electron transfer will be slowed, reducing the rate at which PSI can accept further photons. Thus, it has been suggested that collapsing this proton gradient, perhaps through introduction of a mutation in ATP synthase, could accomplish this, but this hypothesis has yet to be put to the test (Ghirardi and Mohanty, 2010). Finally, production of a desired product, in this case hydrogen, but equally applicable to other biofuels, depends upon how effectively electrons are transferred from the primary acceptor of PSI to the metabolic pathway leading to product formation. In practical terms, all electrons from PSI are carried by ferredoxin, so this translates to what fraction of the reduced ferredoxin donates to the particular product. Therefore, to increase product formation, oxidation of reduced ferredoxin by competing pathways should be reduced or eliminated. Thus, reducing cyclic flow around PSI carried out by reduced ferredoxin by locking it

into so-called state 1 where cyclic flow is minimal, increases hydrogen production fivefold (Kruse et al., 2005). Another strategy would be to eliminate the major sink for reduced ferredoxin, formation of NADPH for use in carbon dioxide fixation, which should also increase hydrogen production, but this has yet to be tested. Of course, the ultimate would be to do away with ferredoxin altogether, and hardwire hydrogenase directly to PSI. Initial attempts to do this have met with some success, although present rates are rather low (Lubner et al., 2010). 2.2.2. Strategies for increased hydrogen production in the presence of oxygen Finally, the major limiting factor in biophotolysis is the fact that the hydrogenase present in green algae, like all other known [FeFe] hydrogenases, is extremely sensitive to oxygen, undergoing irreversible inactivation in the present of even small amounts of oxygen (Lambertz et al., 2011). Creating an oxygen tolerant hydrogenase would go a long way to making biophotolysis a practical method of hydrogen production, but this remains a formidable and perhaps even unreachable goal. In fact, how one might even engineer this into a hydrogenase is not clear. One line of thinking has been that a degree of oxygen tolerance could be gained by altering the properties of the protein channel through which oxygen (and hydrogen) are thought to diffuse, specifically by adding bulky side chains that could restrict the passage of gases. Molecular dynamics and X-ray crystallographic structures indicate gas diffusion channels in the protein leading from the surface to the active site buried in the core of the protein, suggesting that mutations that narrow this channel could favor hydrogen diffusion over oxygen diffusion, leading to an improved protein. However, at least in some cases, gas discrimination may be driven by features at or near the active site, rather than filtering effects of the bulk protein (Stripp et al., 2009). Interestingly, a recent study determined that in one FeFe hydrogenase, from Desulfovibrio fructosovorans, reaction at the active site is the rate-limiting step and not oxygen diffusion per se (Liebgott et al., 2010). Thus, although mutations in the putative channel can indeed have large effects on diffusion, reaction with oxygen is unaffected. On the other hand, the FeFe hydrogenase from Clostridium acetobutylicum reacts much more slowly with oxygen, to the extent that diffusion along the channel is the rate limiting step (Liebgott et al., 2010). This suggests that in fact, oxygen tolerance may, depending on the particular hydrogenase, be more likely introduced by either altering the channel, or by altering the environment around the active site. How the local environment of the active site might affect oxygen sensitivity has been illuminated by recent studies of the oxygen tolerant NiFe membrane-bound hydrogenase of Ralstonia, where it was shown that the iron sulfur cluster proximal to the active site cluster is formed by additional protein cysteine residues and has altered redox properties (Goris et al., 2011; Pandelia et al., 2011). Further information about the structure/function relationship of FeFe hydrogenases and oxygen sensitivity may allow the rational engineering of a reasonably O2 tolerant hydrogenase in the near future. The fact that different factors play a major role in setting oxygen sensitivity depending upon the specific hydrogenase (Liebgott et al., 2010), leaves open the possibility for creating in the future a hydrogenase that is much more resistant to oxygen but that has retained its high activity for proton reduction and diffusion of hydrogen to the surface. This is the real challenge for biophotolytic hydrogen production. 2.3. Bacterial photosynthesis (Photofermentation) Purple non-sulfur photosynthetic bacteria contain a single photosystem and are thus unable to carry out water splitting

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

photosynthesis. Nevertheless, they have been extensively studied for use in biological hydrogen production due to their ability to capture solar energy and carry out the conversion of substrates to hydrogen for which this would not be possible without an additional energy input (acetate, lactate, etc.). In addition, they are being investigated as part of two-stage systems for deriving additional hydrogen from the effluents of dark, hydrogen producing bioreactors (Gadhamshetty et al., 2011; Hallenbeck, 2011; Keskin et al., 2011; Oh et al., 2011). In this case, hydrogen evolution is catalyzed by the ATP-requiring nitrogenase enzyme, capable of reducing protons to hydrogen in the absence of other substrates. This process has been called photofermentation and allows in some cases the stoichiometric conversion of substrate to hydrogen and carbon dioxide. As nitrogenase is repressed by fixed nitrogen, suitable substrates must have a high C/N ratio, but this is the case of many different wastes that might serve as feedstocks for photofermentative production of hydrogen (Keskin et al., 2011). Although this process has been under study for decades, there are a number of limiting factors that prevent practical application at present (Table 2). These include the overall low light conversion efficiencies, the inability to use high (full solar) light effectively, low volumetric rates of hydrogen production, and the low yields observed with some substrates. 2.3.1. Strategies for improving photofermentation As another light driven process, some of the same considerations as already discussed for biophotolysis apply to photofermentation. Thus, for example, higher efficiencies at high light intensities are likely to be obtained with strains carrying truncated photosynthetic antennas. This is corroborated by one report, where a 50% decrease in LHII content translated into a 50% increase in hydrogen production rate at 300 W/m2 (Kondo et al., 2002). Several other studies have also examined this question. A reduced pigment mutant of Rhodobacter sphaeroides was reported to give higher hydrogen production, but only at low (10 W/m2) light intensities, contrary to what one might expect (Kim et al., 2006a,b). The difference at a higher light intensity (100 W/m2) was quite small, difficult to explain since mutants with less antenna pigment would be expected to greatly outperform the wild-type under these conditions. A more recent study has reexamined this question as well and suggested that mutants unable to synthesize LHII antenna pigments gave greater hydrogen production rates at high light intensities, but the results were compromised by the fact that correct regulation of the photosynthetic apparatus was not obtained since the operons for the photosynthetic clusters were carried on plasmids (Eltsova et al., 2010). Moreover, for reasons that are not clear, the two strains that were compared did not possess the same levels of nitrogenase activity. Clearly further research on this question is required. While alterations in light absorption by manipulating antenna complexes might be expected to affect overall rates of hydrogen production, other metabolic changes can be proposed that should directly affect substrate conversion yields. Similarly to the strategies proposed above for biophotolytic systems, where reactions draining the reduced ferredoxin pool might increase yields, in principle, suppressing metabolic reactions that divert electron flow away from nitrogenase should increase hydrogen yields in photofermentation. Several different pathways can be targeted. For one, under excess carbon conditions, which is the case at high C/ N necessary for nitrogenase expression, synthesis of the reduced carbon storage compound, polyhydroxybutyrate, is activated. Thus, in theory phb-mutants should show increased yields. In practice this has been somewhat difficult to demonstrate clearly. A Phbmutant of a R. sphaeroides strain when grown on malate gave a 34% increase in specific hydrogen production, but only a 21% increase in volumetric hydrogen production (Kim et al., 2006a,b).

5

Moreover, when grown on acetate, the substrate that might be expected to show the largest difference, the mutant strain, for reasons that are not clear, grew poorly. The effect of a mutation in the uptake hydrogenase appeared additive, and a double mutant strain had a specific hydrogen production on malate that was 2.5-fold higher than that of the wild-type. In fact, mutating the uptake hydrogenase is another strategy that has been widely applied in efforts to increase photofermentative hydrogen production. In general, Hup-strains show a small but significant increase in total hydrogen accumulation (11–40%) (Liu et al., 2010). Finally, as noted above, photofermentative hydrogen production is carried out by nitrogenase which inflicts a significant energy penalty due to the requirement for ATP for hydrogen evolution by this enzyme (4ATP/H2). Thus, in theory replacing nitrogenase with a FeFe hydrogenase would accomplish two things. First, it would replace an enzyme with a slow turnover rate (6.4 s 1) with one with a much higher turnover rate (2000–6000 s 1), possibly leading to increased volumetric production rates. Secondly, this would eliminate the requirement to use a large fraction of the captured photons for ATP production, potentially leading to higher light conversion efficiencies. Thus strains with FeFe hydrogenase would be postulated to show higher photosynthetic efficiencies of hydrogen production. Some results have been presented that purport to show this (Kim et al., 2008a,b), but these studies are difficult to understand in light of what is presently known about FeFe protein maturation, necessary to achieve activity with heterologously expressed hydrogenases. In all cases that have been examined in detail, synthesis of active FeFe hydrogenase requires the accessory factors, HydE, F, and G (Posewitz et al., 2004). In one case, moderately increased hydrogen production was said to be due to the presence of plasmid-borne hydA (encoding FeFe hydrogenase) that was introduced into a strain for which the hydEHG are absent in the genome (Kim et al., 2008a). In another case, mildly increased hydrogen production was said to be due to plasmid-borne hydA and hydrogenase maturation genes (Kim et al., 2008b), but the maturation genes carried by this plasmid are those for a NiFe hydrogenase, thought not to be capable of maturing a FeFe hydrogenase. These claims need further supporting evidence, and in any case the strains used contained active nitrogenase in addition to the introduced hydrogenase. Thus, whether or not replacing nitrogenase with an active FeFe hydrogenase would lead to greatly increased hydrogen production efficiencies remains an open question. Obviously, a fair amount of improvement can be sought in photofermentative hydrogen production through extensive metabolic remodeling. Successful application of strategies using multiple mutations and the introduction of novel pathways and enzymes probably requires an overall consideration of the metabolic fluxes in the different pathways and their possible interaction. This can be guided by metabolic models, several of which are now available for the photosynthetic bacteria (Golomysova et al., 2010; Imam et al., 2011). In addition, several recent studies have determined the metabolic flux through different pathways during photofermentative hydrogen production from a variety of organic acids (McKinlay and Harwood, 2011), or from glucose (Tao et al., 2012) in both wild-type and some mutant strains. With these tools and information at hand, the near future should bring increased yields through extensive metabolic engineering of these organisms.

3. Dark fermentative biohydrogen production 3.1. Principles and present status While oxygen sensitivity of hydrogenase, as well as the other factors described above, may or may not prove to be intractable

6

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

problems, it seems fairly certain that practical application biophotolysis will require a fairly long-term R&D commitment. Another possibility, and one that at least on the face of it seems to perhaps be more realizable in the short term, is to use (and manipulate) the natural ability of some micro-organisms to produce hydrogen during their fermentative metabolism. Some other fermentations are already carried out on the industrial scale, so reactor technology and bioprocess control are already well in hand. These are anaerobic processes, so hydrogenase sensitivity to oxygen is not an issue. As well, no direct capture of solar energy is required, substrates, plant derived carbohydrates, are produced from normal agricultural or forestry operations which, at least in developed countries, are fairly well optimized. Despite these clear advantages, serious challenges also have to be overcome to make dark fermentative hydrogen production a practical reality (Table 4). The major barrier can be found in the limitations of the natural metabolic process available (Fig. 1). Even though there is some diversity in pathways, and there are various hydrogen evolving hydrogenases available, the major problem with existing pathways is that only one-third of the substrate can be used for hydrogen production, with the remaining twothirds (acetyl-CoA) forming another fermentation product; acetate, butyrate, butanol, acetone, etc. (Hallenbeck, 2009). In terms of the growth and survival of the organism this makes sense because formation of some of these products, e.g. acetate, allows ATP formation while formation of other, reduced products allows the oxidation of NADH, necessary to maintain redox balance in the fermentation. Basically, sugars are broken down to pyruvate which then gives acetyl-CoA and either reduced ferredoxin (PFOR (pyruvate:ferredoxin oxidoreductase) pathway, or formate (PFL (pyruvate:formate lyase) pathway). Organisms which only have the PFL pathway cannot access NADH for hydrogen production and thus are limited to 2 mol of hydrogen per mole of glucose, whereas organisms that use the POR pathway can potentially derive some hydrogen from NADH oxidation using one or more of several different [FeFe] hydrogenases. Thus depending upon the hydrogenase used; NADH-dependent, Fd-dependent with ferredoxin reduced via NFOR (NADH:ferredoxin oxidoreductase), or the bifurcating NADH-Fd dependent hydrogenase, 2–4 mol of hydrogen per mole of glucose can be obtained. Efforts to maximize yields through metabolic engineering have involved almost exclusively Escherichia coli, which however, only has the PFL pathway, so is limited to 2 mol of hydrogen per mole of glucose. Indeed, yields approaching this have been obtained with suitably modified strains (Abo-Hashesh et al., 2011a; Hallenbeck, 2009; Hallenbeck et al., 2011). In theory of course, Clostridium which have the PFOR pathway are better targets for metabolic engineering since they have the potential capability of generating 4 mol of hydrogen per mole of glucose. However, until recently many clostridial species were relatively intractable genetically. With the recent development of better genetic tools, creating strains through rational engineering with increased hydrogen production is now possible (Abo-Hashesh et al., 2011a,b; Hallenbeck et al., 2011; Heap et al., 2010) and interesting results should

Fig. 1. Hydrogen producing fermentation pathways In fermentations with hydrogen as one of the products, as in many other fermentations, glucose is broken down to pyruvate, generating ATP and NADH. Pyruvate is then converted to acetyl-CoA, and depending upon the organism, either formate, through the PFL pathway, or reduced ferredoxin and CO2, through the PFOR pathway. Formate can be converted to hydrogen and CO2, by either the formate hydrogen lyase pathway which contains a [NiFe] hydrogenase (the Ech hydrogenase), or possibly in some other organisms another pathway which contains a formate dependent [FeFe] hydrogenase. NADH, generated during glycolysis, is oxidized through the production of various reduced carbon compounds, typically ethanol. A variety of [FeFe] hydrogenases can be used to reoxidize ferredoxin and produce hydrogen, including; a ferredoxin-dependent H2ase (Fd-[FeFe]). In some cases, NADH can also be used in hydrogen production, either by reducing ferredoxin (NFOR), by directly reducing H2ase (NADH-[FeFe]), or as a co-substrate with reduced ferredoxin (Fd-NADH-[FeFe]). Excess NADH is used to produce other reduced fermentation products. In both cases, acetyl-CoA can also be used to produce ATP.

be forthcoming. Nevertheless, even achieving 4 mol of hydrogen per mole of glucose, already demonstrated with a hyperthermophilic bacterium (Zeidan and van Niel, 2010), only represents a 33% yield since 12 mol of hydrogen are theoretically available in glucose. As discussed elsewhere, this is unacceptable for several of reasons; other biofuels could be made from the same substrates at substantially higher yields with present technologies, and the excess carbon that is not converted to a biofuels represents a serious waste treatment challenge (Hallenbeck and Ghosh, 2009;

Table 4 Strategies for improving dark fermentative biohydrogen production. Challenges for implementation Yields limited by metabolic pathways Large quantities of side products (organic acids) produced Most abundant substrate, lignocellulose, difficult to ferment. Need to ferment variety of hexoses and pentoses

Possible workarounds Metabolic engineering to introduce novel pathways/enzymes Use one of several possible second stages to convert these to hydrogen (or methane). Develop pretreatment (solublization and hydrolysis) Use natural cellulose degraders or endow strong fermenters with this capacity Introduce substrate flexibility into fermenter Increase hydrogen production by omnivorous organism

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9 Table 5 Strategies for second stage technologies. Challenges for implementation Methane digestion Different media requirements Different growth rate Photofermentation Low light conversion efficiencies High cost of photobioreactors Excess energy demand by nitrogenase (ATP) Microbial electrolysis Expensive precious metal cathodes required Current densities need to be increased Excess voltages required for high yields

Possible workarounds

Find low cost alkalinisation method Use larger reactor, greater HRT

Reduce antenna size Develop high-tech H2 impermeable plastics Replace N2ase with H2ase

Develop inexpensive cathodes (Ni, stainless steel) Better anode geometry, eliminate ‘‘short circuit’’ metabolic reactions (H2 cycling) Engineer cells with lower internal resistance

Hallenbeck, 2011). Thus, ways to extract more energy, preferably hydrogen, must be found. As detailed below, a number of options are under active study at present. 3.2. Metabolic engineering for increased yields One approach is to increase yields above the present metabolic limits by metabolic engineering. Adding entire pathways from a heterologous organism and altering metabolic flux to channel flow of metabolic intermediates through the added pathway could lead to higher hydrogen yields. Computational modeling has been used to predict a number of ways in which this might work (Jones, 2008). It was suggested to redirect glucose catabolism through the pentose phosphate pathway, thus, at least theoretically, allowing the complete oxidation of glucose. In fact, obtaining higher hydrogen yields requires that more complete oxidation than is possible with glycolysis alone be achieved. Several studies have examined the effect of increasing flux through the pentose phosphate pathway on hydrogen production (Veit et al., 2008; Kim et al., 2011). This also requires the introduction of an artificial pathway from the NADPH which is generated through the pentose phosphate pathway to an introduced hydrogenase. Although both studies demonstrated increased hydrogen production, yields were still low, even when gluconeogenic flux was boosted by overexpression of glpX (Kim et al., 2011). One problem is of course related to the adverse thermodynamics of reducing ferredoxin from NADPH (Veit et al., 2008). A number of other ways in which increased glucose oxidation and hence hydrogen production might be possible have also been suggested (Hallenbeck et al., 2011). The citric acid cycle, normally operational only under aerobic conditions, allows the complete dissimilation of glucose to CO2 and NADH and FADH. If this cycle could be made to operate under anaerobic conditions, large amounts of NADH would become available. In common with the approach of sending most of the carbon through the pentose phosphate pathway however, is how the pool of reduced NAD(P)H could be converted quantitatively to hydrogen since the redox potential of NAD(P)H is higher than that of H2. It seems likely that this would only be possible if some type of reduced electron flow system were used, which would require the input of some amount of energy. A number of these systems are known, and it has been suggested that the requisite energy could be provided by allowing a limited amount of oxygen-dependent respiration, although this has yet to be demonstrated.

7

Other synthetic biology approaches might be applied to increase flux through hydrogenase, thereby at least increasing hydrogen productivities if not yields. Naturally highly efficient metabolic pathways are often physically organized in supramolecular complexes. Some well known examples include the NRPS (Non-ribosomal peptide synthetase) and PKS (polyketide synthases) (Fischbach and Walsh 2006). These systems carry out substrate channeling and have a number of advantages including; operation independent of diffusion, lower buildup of potentially toxic intermediates, and overall coordination of enzymatic activity. Recently, it was shown that artificially constructed protein complexes offer many of the same advantages, giving a new method for efficient de novo metabolic pathway construction (Dueber et al., 2009). RNAs have been used to build protein binding scaffolds through their aptamer domains and when coupled with a heterologous NADPH dependent hydrogenase pathway in E. coli gave eleven-fold higher hydrogen production (Delebecque et al., 2011). 3.3. Hybrid systems for increased yields Another possibility is to use a second stage system to extract additional energy from the side products of a first stage hydrogen producing fermentation operating under present metabolic constraints. At present there are three different proposals (Table 5) (Hallenbeck and Ghosh, 2009; Hallenbeck, 2011). In one approach, already demonstrated at the pilot scale, the organic acids produced in the first stage are used to produce methane. This involves the process of anaerobic digestion and the bioprocess parameters for the successful functioning of an anaerobic digester are well known. Although less desirable than a pure hydrogen stream, hydrogen/ methane mixtures are of some utility since they burn considerably cleaner in an internal combustion engine than methane alone. Two different approaches are under study which would lead to additional hydrogen production from the organic acids produced in the first stage. In both cases, some additional external energy input is required to drive the thermodynamically unfavorable conversion of the organic acids to hydrogen. In one scenario, photosynthetic bacteria, which can capture solar energy and convert it to chemical energy, would stoichiometrically convert the organic acids to hydrogen and carbon dioxide by a process that is called photofermentation (Hallenbeck and Benemann, 2002; Hallenbeck, 2011; Keskin et al., 2011). Although this process has been well studied, there are still a number of technical barriers including; sensitivity to fixed nitrogen, low light conversion efficiencies, inability to use high light intensities, and the need for low cost, transparent, hydrogen impermeable photobioreactors. In another approach, under study for only the past five years, second stage MECs (microbial electrolysis cells) are used in conjunction with a small electrical current to quantitatively convert the organic acids produced in the first stage to hydrogen. Although tremendous advances in rates and current densities have been achieved in relatively little time, more R&D is required before MECs can be deployed on a practical level. MECs are a variation on microbial fuels cells and suffer from some of the same technical challenges as those devices (Logan, 2010). 4. Conclusions A variety of microbial paths to renewable hydrogen production are available and are under active study. Although a number of advances have been made recently, there are a number of technical challenges in each area that must be overcome before these technologies can be adopted on a practical large scale. Extensive R&D in this area is underway worldwide, but practical development of

8

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9

biohydrogen production is a long term prospect, commensurate with the time frame required to adopt hydrogen as a major fuel source. Acknowledgements Biohydrogen research in the laboratory of PCH is supported by grants from NSERC (Natural Sciences and Engineering Research Council of Canada) and FQRNT (Le Fonds québécois de la recherche sur la nature et les technologies). M. A.-H. is supported by the Ministry of Higher Education (Egypt). D.G. is supported by a scholarship from PBEEE/ FQRNT (Le Fonds Québécois de la recherché sur la nature et les technologies). References Abo-Hashesh, M., Wang, R., Hallenbeck, P.C., 2011a. Metabolic engineering in dark fermentative hydrogen production; theory and practice. Bioresour. Technol. 102 (18), 8414–8422. Abo-Hashesh, M., Ghosh, D., Tourigny, A., Taous, A., Hallenbeck, P.C., 2011b. Single stage photofermentative hydrogen production from glucose: An attractive alternative to two stage photofermentation or co-culture approaches. Int. J. Hydrogen Energy 36 (21), 13889–13895. Adessi, A., De Philippis, R., 2012. Hydrogen Production: Photofermentation. In: Hallenbeck, P.C. (Ed.), Microbial Technologies In Advanced Biofuels Production. Springer, New York, pp. 53–75. Adessi, A., De Philippis, R., Hallenbeck, P.C., 2012. Combined systems for maximum substrate conversion. In: Hallenbeck, P.C. (Ed.), Microbial Technologies In Advanced Biofuels Production. Springer, New York, pp. 107–126. Beer, L.L., Boyd, E.S., Peters, J.W., Posewitz, M.C., 2009. Engineering algae for biohydrogen and biofuel production. Curr. Opin. Biotech. 20, 264–271. Blankenship, R.E., Tiede, D.M., Barber, J., Brudvig, G.W., Fleming, G., Ghirardi, M., Gunner, M.R., Junge, W., Kramer, D.M., Melis, A., Moore, T.A., Moser, C.C., Nocera, D.G., Nozik, A.J., Ort, D.R., Parson, W.W., Prince, R.C., Sayre, R.T., 2011. Comparing photosynthetic and photovoltaic efficiencies and recognizing the potential for improvement. Science 332 (6031), 805–809. Delebecque, C.J., Lindner, A.B., Silver, P.A., Aldaye, F.A., 2011. Organization of intracellular reactions with rationally designed RNA assemblies. Science 333 (6041), 470–474. Dueber, J.E., Wu, G.C., Malmirchegini, G.R., Moon, T.S., Petzold, C.J., Ullal, A.V., Prather, K.L.J., Keasling, J.D., 2009. Synthetic protein scaffolds provide modular control over metabolic flux. Nat. Biotechnol. 27 (8), 753–759. Eltsova, Z., Vasilieva, L., Tsygankov, A., 2010. Hydrogen production by recombinant strains of Rhodobacter sphaeroides using a modified photosynthetic apparatus. Appl. Biochem. Microbiol. 46 (5), 487–491. Eroglu, I., Aslan, K., Gunduz, U., Yucel, M., Turker, L., 1999. Substrate consumption rates for hydrogen production by Rhodobacter sphaeroides in a column photobioreactor. J. Biotechnol. 70, 103–113. Eroglu, E., Melis, A., 2011. Photobiological hydrogen production: recent advances and state of the art. Bioresour. Technol. 102, 8403–8413. Fischbach, M.A., Walsh, C.T., 2006. Assembly-Line Enzymology for Polyketideand Nonribosomal Peptide Antibiotics: Logic, Machinery, and Mechanisms. Chem. Rev. 106, 3468–3496. Ghirardi, M.L., Mohanty, P., 2010. Oxygenic hydrogen photoproduction – current status of the technology. Curr. Sci. 98, 499–507. Gadhamshetty, V., Sukumaran, A., Nirmalakhandan, N., 2011. Photoparameters in photofermentative biohydrogen production. Crit. Rev. Env. Sci. Technol. 41 (1), 1–51. Golomysova, A., Gomelsky, M., Ivanov, P.S., 2010. Flux balance analysis of photoheterotrophic growth of purple nonsulfur bacteria relevant to biohydrogen production. Int. J. Hydrogen Energy 35, 12751–12760. Goris, T., Wait, A.F., Saggu, M., Fritsch, J., Heidary, N., Stein, M., Zebger, I., Lendzian, F., Armstrong, F.A., Friedrich, B., Lenz, O., 2011. A unique iron-sulfur cluster is crucial for oxygen tolerance of a NiFe-hydrogenase. Nat. Chem. Biol. 7 (5), 310– 318. Hallenbeck, P.C., 2009. Fermentative hydrogen production: principles, progress, and prognosis. Int. J. Hydrogen Energy 34, 7379–7389. Hallenbeck, P.C., 2011. Microbial paths to renewable hydrogen production. Biofuels 2, 285–302. Hallenbeck, P.C., 2012. Hydrogen production by cyanobacteria. In: Hallenbeck, P.C. (Ed.), Microbial Technologies In Advanced Biofuels Production. Springer, New York, pp. 15–28. Hallenbeck, P.C., Benemann, J.R., 2002. Biological hydrogen production; fundamentals and limiting processes. Int. J. Hydrogen Energy 27, 1185–1193. Hallenbeck, P.C., Ghosh, D., 2009. Advances in fermentative biohydrogen production: the way forward? Trends Biotechnol. 27 (5), 287–297. Hallenbeck, P.C., Ghosh, D., Abo-Hashesh, M., Wang, R., 2011. Metabolic engineering for enhanced biofuels production with emphasis on the biological production of hydrogen. Adv. Chem. Res., 125–154, J. C. Taylor ed., vol. 6. Heap, J.T., Kuehne, S.A., Ehsaan, M., et al., 2010. The ClosTron: Mutagenesis in Clostridium refined and streamlined. J. Microbiol. Meth. 80, 49–55.

Imam, S., Yilmaz, S., Sohmen, U., Gorzalski, A.S., Reed, J.L., Noguera, D.R., Donohue, T.J., 2011. IRsp1095: a genome-scale reconstruction of the Rhodobacter sphaeroides metabolic network. BMC Syst. Biol. 21 (5), 116. Kim, M.-S., Baek, J.S., Lee, J.K., 2006b. Comparison of H2 accumulation by Rhodobacter sphaeroides KD131 and its uptake hydrogenase and PHB synthase deficient mutant. Int. J. Hydrogen Energy 31, 121–127. Kim, Y.M., Cho, H.-S., Jung, G.Y., Park, J.M., 2011. Engineering the pentose phosphate pathway to improve hydrogen yield in recombinant Escherichia coli. Biotech. Bioeng. 108 (12), 2941–2946. Jones, P.R., 2008. Improving fermentative biomass-derived H-2-production by engineering microbial metabolism. Int. J. Hydrogen Energy 33, 5122–5130. Kaushik, A., Anjana, K., 2011. Biohydrogen production by Lyngbya perelegans: Influence of physico-chemical environment. Biomass & Bioenergy 35, 1041– 1045. Keskin, T., Abo-Hashesh, M., Hallenbeck, P.C., 2011. Photofermentative hydrogen production from wastes. Bioresour. Technol. 102, 8557–8568. Kim, E.-J., Kim, J.-S., Kim, M.-S., Lee, J.K., 2006a. Effect of changes in the level of light harvesting complexes of Rhodobacter sphaeroides on the photoheterotrophic production of hydrogen. Int. J. Hydrogen Energy 31 (4), 531–538. Kim, E.-J., Kim, M.-S., Lee, J.K., 2008a. Hydrogen evolution under photoheterotrophic and dark fermentative conditions by recombinant Rhodobacter sphaeroides containing the genes for fermentative pyruvate metabolism of Rhodospirillum rubrum. Int. J. Hydrogen Energy 33 (19), 5131–5136. Kim, E.-J., Lee, M.-K., Kim, M.-S., Lee, J.K., 2008b. Molecular hydrogen production by nitrogenase of Rhodobacter sphaeroides and by Fe-only hydrogenase of Rhodospirillum rubrum. Int. J. Hydrogen Energy 33 (5), 1516–1521. Kondo, T., Arakawa, M., Hirai, T., Wakayama, T., Hara, M., Miyake, J., 2002. Enhancement of hydrogen production by a photosynthetic bacterium mutant with reduced pigment. J. Biosci. Bioeng. 93 (2), 145–150. Kosourov, S.N., Seibert, M., 2009. Hydrogen photoproduction by nutrient-deprived Chlamydomonas reinhardtii cells immobilized within thin alginate films under aerobic and anaerobic conditions. Biotech. Bioeng. 102, 50–58. Kosourov, S.N., Ghirardi, M.L., Seibert, M., 2011. A truncated antenna mutant of Chlamydomonas reinhardtii can produce more hydrogen than the parental strain. Int. J. Hydrogen Energy 36, 2044–2048. Kruse, O., Hankamer, B., 2010. Microalgal hydrogen production. Curr. Opin. Biotechnol. 21 (3), 238–243. Kruse, O., Rupprecht, J., Bader, K.P., Thomas-Hall, S., Schenk, P.M., Finazzi, G., Hankamer, B., 2005. Improved photo biological H2 production in engineered green algal cells. J. Biol. Chem. 280, 34170–34177. Lambertz, C., Leidel, N., Havelius, K.G.V., Noth, J., Chernev, P., Winkler, M., Happe, T., Haumann, M., 2011. O2 reactions at the six-iron active site (H-cluster) in [FeFe]hydrogenase. J. Biol. Chem. 286 (47), 40614–40623. Laurinavichene, V., Fedorov, A.S., Ghirardi, M.L., Seibert, M., Tsygankov, A.A., 2006. Demonstration of sustained hydrogen photoproduction by immobilized, sulfur-deprived Chlamydomonas reinhardtii cells. Int. J. Hydrogen Energy 31, 659–667. Lee, K.-S. et al., 2006. Improving biohydrogen production in a carrier-induced granular sludge bed by altering physical configuration and agitation pattern of the bioreactor. Int. J. Hydrogen Energy 31, 1648–1657. Liebgott, P.P., Leroux, F., Burlat, B., et al., 2010. Relating diffusion along the substrate tunnel and oxygen sensitivity in hydrogenase. Nat. Chem. Biol. 6, 63–70. Liu, T., Li, X., Zhou, Z., 2010. Improvement of hydrogen yield by hupR gene knock-out and nifA gene overexpression in Rhodobacter sphaeroides 6016. Int. J. Hydrogen Energy 35 (18), 9603–9610. Logan, B.E., 2010. Scaling up microbial fuel cells and other bioelectrochemical systems. Appl. Microbiol. Biotechnol. 85, 1665–1671. Lubner, C.E., Knorzer, P., Silva, P.J.N., Vincent, K.A., Happe, T., Bryant, D.A., Golbeck, J.H., 2010. Wiring an FeFe-Hydrogenase with Photosystem I for Light-Induced Hydrogen Production. Biochemistry 49 (48), 10264–10266. McKinlay JB, Harwood CS.2011. Calvin cycle flux, pathway constraints, and substrate oxidation state together determine the H2 biofuel yield in photoheterotrophic bacteria. mBio 2 (2):e00323–10. doi:10.1128/mBio.0032310. Melis, A., Zhang, L.P., Forestier, M., Ghirardi, M.L., Seibert, M., 2000. Sustained photobiological hydrogen gas production upon reversible inactivation of oxygen evolution in the green alga Chlamydomonas reinhardtii. Plant Physiol. 122, 127–135. Morsy, F.M., 2011. Acetate versus sulfur deprivation role in creating anaerobiosis in light for hydrogen production by Chlamydomonas reinhardtii and Spirulina platensis: two different organisms and two different mechanisms. Photochem. Photobiol. 87, 137–142. Oh, Y.-K., Raj, S.M., Jung, G.Y., Park, S., 2011. Current status of the metabolic engineering of microorganisms for biohydrogen production. Bioresour. Technol. 102 (18), 8357–8367. Pandelia, M.E., Nitschke, W., Infossi, P., Giudici-Orticoni, M.T., Bill, E., Lubitz, W., 2011. Characterization of a unique FeS cluster in the electron transfer chain of the oxygen tolerant NiFe hydrogenase from Aquifex aeolicus. Proc. Nat. Acad. Sci. USA 108 (15), 6097–6102. Polle, J.E.W., Kanakagiri, S., Jin, E., Masuda, T., Melis, A., 2002. Truncated chlorophyll antenna size of the photosystems, a practical method to improve microalgal productivity and hydrogen production in mass culture. Int. J. Hydrogen Energy 27, 1257–1264. Posewitz, M.C., King, P.W., Smolinski, S.L., Zhang, L., Seibert, M., Ghirardi, M.L., 2004. Discovery of two novel radical S-adenosylmethionine proteins required for the assembly of an active [Fe] hydrogenase. J. Biol. Chem. 279, 25711–25720.

P.C. Hallenbeck et al. / Bioresource Technology 110 (2012) 1–9 Srirangan, K., Pyne, M.E., PerryChou, C., 2011. Biochemical and genetic engineering strategies to enhance hydrogen production in photosynthetic algae and cyanobacteria. Bioresour. Technol. 102 (18), 8589–8604. Stripp, S.T., Goldet, G., Brandmayr, C., et al., 2009. How oxygen attacks FeFe hydrogenases from photosynthetic organisms. Proc. Nat. Acad. Sci. USA 106, 17331–17336. Tamburic, B., Zemichael, F.W., Maitland, G.C., Hellgardt, K., 2011. Parameters affecting the growth and hydrogen production of the green alga Chlamydomonas reinhardtii. Int. J. Hydrogen Energy 36, 7872–7876. Tao, Y., Liu, D., Yan, Z., Zhou, Z., Lee, J.K., Yang, C., 2012. Network identification and flux quantification of glucose metabolism in Rhodobacter sphaeroides under photoheterotrophic H2-producing conditions. J. Bacteriol. 194, 274–283. Torzillo, G., Scoma, A., Faraloni, C., Ena, A., Johanningmeier, U., 2009. Increased hydrogen photoproduction by means of a sulfur-deprived Chlamydomonas reinhardtii D1 protein mutant. Int. J. Hydrogen Energy 34, 4529–4536. Tsygankov, A.A., Hall, D.O., Liu, J., Rao, K.K., 1998. An automated helical photobioreactor incorporating cyanobacteria for continuous hydrogen production. In: Zaborsky, O.R. (Ed.), Biohydrogen. Plenum Press, London, pp. 431–440.

9

Tsygankov, A.A., 2012. Hydrogen Production: Light-Driven Processes – Green Algae. In: Hallenbeck, P.C. (Ed.), Microbial Technologies In Advanced Biofuels Production. Springer, New York, pp. 29–51. Veit, V., Akhtar, M.K., Mizutani, T., Jones, P.R., 2008. Constructing and testing the thermodynamic limits of synthetic NAD(P)H:H2 pathways. Microbial Biotechnol. 1, 382–394. Wang, H.Y., Fan, X.L., Zhang, Y.T., Yang, D.W., Guo, R.B., 2011. Sustained photohydrogen production by Chlorella pyrenoidosa without sulfur depletion. Biotechnol. Lett. 33, 1345–1350. Wu, K.-J. et al., 2007. Simultaneous produc tion of biohydrogen and bioethanol with fluidized-bed and packed-bed bi oreactors cont aining immobilized anaerobic sludge. Process Biochem. 42, 1165–1171. Wu, S.X., Xu, L.L., Huang, R., Wang, Q.X., 2011. Improved biohydrogen production with an expression of codon-optimized hemH and lba genes in the chloroplast of Chlamydomonas reinhardtii. Bioresour. Technol. 102, 2610–2616. Zeidan, A.A., van Niel, E.W.J., 2010. A quantitative analysis of hydrogen production efficiency of the extreme thermophile Caldicellulosiruptor owensensis OLT. Int. J. Hydrogen Energy 35, 1128–1137.