SUMMARY
SUMMARY This review summarizes recent aspects of (di)nitrogen fixation and (di)hydrogen metabolism, with emphasis on cyanobacteria. These organisms possess several types of the enzyme complexes catalyzing N2 fixation and/or H2 formation or oxidation, namely, two Mo nitrogenases, a V nitrogenase, and two hydrogenases. The two cyanobacterial Ni hydrogenases are differentiated as either uptake or bidirectional hydrogenases. The different forms of both the nitrogenases and hydrogenases are encoded by different sets of genes, and their organization on the chromosome can vary from one cyanobacterium to another. Factors regulating the expression of these genes are emerging from recent studies. New ideas on the potential physiological and ecological roles of nitrogenases and hydrogenases are presented. There is a renewed interest in exploiting cyanobacteria in solar energy conversion programs to generate H2 as a source of combustible energy. To enhance the rates of H2 production, the emphasis perhaps needs not to be on more efficient hydrogenases and nitrogenases or on the transfer of foreign enzymes into cyanobacteria. A likely better strategy is to exploit the use of radiant solar energy by the photosynthetic electron transport system to enhance the rates of H2 formation and so improve the chances of utilizing cyanobacteria as a source for the generation of clean energy.
INTRODUCTION
Biological (di)nitrogen fixation is catalyzed by the enzyme complex nitrogenase, where the formation of molecular hydrogen accompanies ammonia production according to equation 1: $$mathtex$$\[8\mathrm{H}^{{+}}{+}8e^{{-}}{+}\mathrm{N}_{2}{+}16\mathrm{MgATP}\ {\rightarrow}\ 2\mathrm{NH}_{3}{+}\mathrm{H}_{2}{+}16\mathrm{MgADP}{+}16\mathrm{P}_{\mathrm{i}}\]$$mathtex$$(1) Whereas H2 formation by nitrogenases is unidirectional, H2 production by some hydrogenases is reversible, as shown in equation 2: $$mathtex$$\[2\mathrm{H}^{{+}}{+}2e^{{-}}\ {\leftrightarrow}\ \mathrm{H}_{2}\]$$mathtex$$(2) N2 fixation and H2 formation are closely linked processes, as has been known at least since a publication by Phelps and Wilson in 1941 (39). Hydrogenase recycles the H2 produced in N2 fixation, thereby minimizing the loss of energy during nitrogenase catalysis. A rather simple scheme showing the relationship between pyruvate degradation, N2 fixation, and production and uptake of H2, as occur in strict anaerobes such as Clostridium pasteurianum or in the facultative anaerobe Klebsiella pneumoniae, is shown in Fig. 1. However, H2 can also be produced independently of N2 fixation, e.g., as an end product of fermentation, which can also take place in N2-fixing organisms.
A simple scheme showing the relationship between pyruvate degradation, ammonium and hydrogen formation by nitrogenase, and hydrogen uptake by hydrogenase. This pathway is typical in strict or facultative anaerobes but also proceeds in cyanobacteria.
As described in detail below, nitrogenases (Mo, V, and homocitrate) and hydrogenases (Ni, CO, and CN−) contain unusual components in their prosthetic groups (Fig. 2 and 3) that are not or only rarely employed elsewhere in nature. Their roles and their biosyntheses pose fascinating questions that are as yet only partly resolved. Most cyanobacteria are aerobic organisms producing O2 photosynthetically. They are generally not exposed to environmental molecular H2. Despite this, and paradoxical at first glance, the capability to metabolize H2 is constitutively expressed in many aerobic cyanobacteria. N2 fixation and H2 metabolism have been key research areas in microbiology over the years. Cyanobacteria are the best suited organisms for studies on the subject, because several of them, both unicellular and heterocystous forms, can be easily genetically modified by molecular techniques. Moreover, cyanobacterial H2 production offers perspectives for potential applications.
The structure of the 2:1 Fe protein-MoFe protein complex of the Azotobacter vinelandii nitrogenase stabilized by MgADP plus AlF4−. Each Fe protein molecule (shown at the top left and bottom right of the complex in brown) docks directly over the interface between an α/β subunit pair of the MoFe protein (in black and gray), which occupies the center of the structure, to juxtapose its [4Fe-4S] cluster (in yellow) with a P cluster (in red) at this interface. One FeMo cofactor (in pale blue) is accommodated within each α subunit. The two β subunits (in gray) provide the interactions among the two α/β subunit pairs (183) (Protein Data Bank [PDB] code 1N2C). (Adapted from reference 183 with permission from Macmillan Publishers Ltd.)
The structure of the FeMo cofactor of the Azotobacter vinelandii nitrogenase MoFe protein with its α subunit-based ligating amino acid residues (αCys-275 and αHis-442) and homocitrate. The Mo (red), Fe (gray), and S (pale green) atoms are individually colored. The identity of the central atom (blue) remains unassigned (PDB code 1M1N). (Reprinted from reference 61 with permission from AAAS.)
Both N2 fixation (153, 177) and H2 metabolism (226, 228) have been reviewed. Excellent accounts on cyanobacterial hydrogenases (82, 212, 214) are available, and those articles should be consulted for primary references. The aim of this review is not to reiterate these subjects but to highlight facts and ideas, particularly on the physiology, that have not received much attention in the past. This review also emphasizes the more recent developments and focuses on the fact that nitrogenases and hydrogenases are common players in H2 metabolism. The restriction to cyanobacteria as the best candidates for applications appears to be timely.
MOLYBDENUM NITROGENASE
The longest-known and best-studied nitrogenase is the Mo nitrogenase, which occurs in all N2-fixing organisms with the exception of some CO-oxidizing bacteria (178). The Mo nitrogenase is encoded by the structural genes nifHDK. It consists of two component proteins. Figure 2 shows the structure of a 2:1 complex of the two components, which might approximate an electron transfer transition state, with the larger component in the center and one molecule of the smaller component at each end (see the legend to Fig. 2 for more information). The nifH gene codes for the smaller, homodimeric (γ2) protein, which has a molecular mass of about 64 kDa and is termed Fe protein, (di)nitrogenase reductase, or protein 2. Its prosthetic group is a [4Fe-4S] cluster that bridges the subunit interface and is ligated by two cysteinyl residues from each subunit. This cluster accepts reducing equivalents from electron carriers which are either ferredoxin or flavodoxin, depending on the organism. Each subunit possesses a MgATP/MgADP binding site. When provided with MgATP and reductant, the Fe protein undergoes a conformation change combined with a change of its redox potential of ca. −200 mV. Docking to the larger component protein (Fig. 2) lowers the redox potential further to about −600 mV and is accompanied by an additional conformation change. All these changes are prerequisites for the transfer of one electron from the Fe protein to the larger component protein with concurrent MgATP hydrolysis. Multiple electron transfers prepare the larger component for substrate binding and reduction. The Fe protein has the most conserved amino acid sequence among all nitrogenase proteins. Therefore, the nifH gene is best suited for DNA probing when searches for the occurrence of nitrogenase in organisms or different environments are undertaken (181).
The larger component protein (MoFe protein, dinitrogenase, or protein 1) is a tetrameric (α2β2) protein of about 240 kDa. It contains two unique prosthetic groups, the P cluster and the MoFe cofactor (Fig. 3). Each αβ dimer of the larger nitrogenase protein binds one FeMo cofactor and one P cluster. The P cluster is composed of both a [4Fe-4S] subcluster and a [4Fe-3S] subcluster, which share one S2−. It sits at the interface of the α and β subunits and is usually depicted as an intermediate in electron transfer from the Fe protein to the FeMo cofactor. However, there is no direct evidence to support this supposition. The P cluster may have an N2 fixation-specific role through which it provides the impetus to commit the reversibly bound N2 to the irreversible reduction pathway (70). The MoFe cofactor consists of 1 Mo atom, 7 Fe atoms, 9 S atoms, and homocitrate, plus an as-yet-unidentified light atom (or ion) at its center (Fig. 3). Although an educated first guess might be that it is N based, this suggestion remains unproven (see, for example, reference 234). The FeMo cofactor is the site of substrate binding and reduction. This cluster can again be subdivided into two subclusters, one [Mo-3Fe-3S] and one [4Fe-3S]. These are bridged by 3 S2− ligands and the light atom. Homocitrate, which is bound to Mo by two O ligands, is required for full catalytic activity, but its specific role remains unclear.
The substrate binding and reduction sites have not yet been identified definitively. The N2 molecule may be bound at a central 4Fe-4S face, possibly with participation of the light atom. The Mo-homocitrate entity would then not be directly involved in catalysis but could determine the redox potential of the cofactor. Alternatively, N2 may be bound directly to and be reduced at the Mo-homocitrate part of the FeMo cofactor. It is somewhat surprising that this issue has not yet been resolved despite extensive research for many years. However, neither the FeMo cofactor nor any other part of the nitrogenase complex binds a substrate on its own. Substrate binding and reduction commence only when both nitrogenase component proteins plus MgATP and reductant are available.
Nitrogenase catalyzes the reduction of many substrates other than N2, nearly all of which have a complete or partial triple bond in common, e.g., HC≡N (hydrocyanic acid), R,C≡N (nitriles), RN≡C (isonitriles), N2O (nitrous oxide), N≡N—N− (azide), and HC≡CH (acetylene); the main exceptions are H+ and NO2−. Of particular interest is the reduction of C2H2 to C2H4. In contrast to carbon fixation research, where an easily manageable isotope (14C) is available, N2 fixation research suffers from the absence of a similar isotope of N. 13N is highly radioactive and very unstable, and because 15N is nonradioactive, its reduction can be determined only by the somewhat laborious technique of mass spectrometry. In contrast, the gases C2H2 and C2H4 can be easily and quickly separated and quantified with high accuracy by gas chromatography. Unless special questions (e.g., the determination of the ratio between C2H2 and N2 reduction) are to be resolved, nitrogenase activity is routinely assayed by the C2H2 reduction method despite the fact that the ratio between N2 fixation and C2H2 reduction is not always 3:1. The reduction of all nitrogenase substrates is inhibited by CO, with the exception of H+ conversion to H2 (see below).
The reduction of N2 but not that of all other nitrogenase substrates is accompanied by the evolution of one H2 molecule for each N2 molecule that is reduced (203) (see equation 1). This formation of H2 could represent an activation step that is uniquely required for N2 binding (196). In the absence of any other substrate, nitrogenase catalyzes an ATP-dependent reduction of H+. The relationship(s) between the binding of N2, the other substrates, and inhibitors such as CO is apparently very complex and at best only partly understood. The complexity of the situation is evidenced by the fact that N2 is a competitive inhibitor of C2H2 reduction but C2H2 is a noncompetitive inhibitor of N2 reduction (179).
In addition to the three structural genes nifHDK, nitrogenase expression requires altogether 20 genes in the enterobacterium Klebsiella pneumoniae, all of which are contiguously located on the chromosome. In other bacteria, these genes are interspersed throughout the genome, and other fix genes may be necessary for nitrogenase synthesis and catalysis.
ALTERNATIVE NITROGENASES
Mo nitrogenase is now known to have two close relatives, the V nitrogenase and the Fe nitrogenase, but the distribution of these two enzymes appears to be haphazard (see below). The discovery of the alternative nitrogenases without molybdenum in their prosthetic groups can be regarded as a milestone in nitrogenase research. Reviews on this subject are available (18, 59, 167, 242). The aerobe Azotobacter vinelandii possesses gene sets for all three different types of nitrogenases (Fig. 4). Under conditions of Mo sufficiency in the culture medium, A. vinelandii expresses nifHDK, encoding Mo nitrogenase. When Mo is limiting but V is sufficiently available, A. vinelandii synthesizes a V nitrogenase with a VFe cofactor in the N2 binding and reducing site through expression of the alternative structural genes vnfHDGK. The occurrence of V in the prosthetic group of an enzyme complex is remarkable because, other than in V nitrogenase, the element V has only rarely been found to have a biological function, e.g., in some uncommon peroxidases (95). When the concentrations of both Mo and V are growth limiting, A. vinelandii synthesizes a third nitrogenase with an FeFe cofactor in the active site and encoded by the structural genes anfHDGK.
Genes coding for nitrogenases in two cyanobacteria and two other microorganisms. (Courtesy of Teresa Thiel, University of Missouri—St. Louis.)
All three nitrogenases are rather similar. They require both a larger and a smaller component protein for catalytic activity and possess the P cluster, with identical spectroscopic properties, and a special cofactor for the substrate binding and reducing site. All three nitrogenases show extensive but not identical amino acid sequence homologies. Most importantly, both alternative nitrogenases possess the additional G gene located between the D and K genes and the resulting component proteins are, therefore, α2β2δ2 heterohexamers. The δ subunit has no counterpart with similar sequence homologies elsewhere. Its function has not been finally resolved, but it is apparently required for processing the apoprotein of the alternative nitrogenases to the functional enzyme complex by assisting in the insertion of the cofactor, as has been specifically shown for the V nitrogenase (45, 46). Remarkably, although the proteins VnfG and AnfG are required for N2 fixation by A. vinelandii, they are not required for C2H2 reduction (45, 46, 228).
Both alternative nitrogenases can support growth of A. vinelandii, albeit with lower rates than Mo nitrogenase. Both N2 and C2H2 are poorer substrates for the alternative nitrogenases than for the Mo enzyme. Whereas with Mo nitrogenase the stoichiometry between ammonia production and H2 formation is about 2:1, as shown in equation 1, the reaction via the V nitrogenase proceeds optimally as shown in equation 3: $$mathtex$$\[12\mathrm{H}^{{+}}{+}12e^{{-}}{+}\mathrm{N}_{2}{+}24\mathrm{MgATP}\ {\rightarrow}\ 2\mathrm{NH}_{3}{+}3\mathrm{H}_{2}{+}24\mathrm{MgADP}{+}24\mathrm{P}_{\mathrm{i}}\]$$mathtex$$(3) With Mo nitrogenase, virtually all electrons are allocated to C2H2 when it is the only substrate available. In contrast, C2H4 formation by V nitrogenase is accompanied by a significant production of H2. This H2 formation in the presence of either N2 or C2H2 seems to be even higher with the Fe nitrogenase, although these reactions have not been examined in comparable detail. These differences between the three nitrogenases are not due to differences in the apparent Km values for N2 and C2H2 and are also not caused by restricted electron transfer within or between the nitrogenase proteins (59). The differences may lay in the rate-limiting step in the nitrogenase catalytic cycle (220), which is the final dissociation of the oxidized Fe protein-MgADP from the electron transfer complex.
The production of NH3 from N2 by the V nitrogenase is accompanied by the release of the presumptive reduction intermediate N2H4 (57). In addition, both the V and Fe nitrogenases reduce C2H2 beyond C2H4 to produce some C2H6. Although this ethane formation amounts to only about 3% of the total C2H2-reducing capacity, it can easily be assessed by gas chromatography and is therefore indicative for the expression of an alternative nitrogenase in an organism (56). Mo nitrogenase does not catalyze the reduction of ethene, but some H2-consuming methanogenic enrichment cultures have been reported to produce ethane from ethene apparently independently of nitrogenases (118).
The apparently haphazard distribution of nitrogenases results in some organisms having all three, some possessing only Mo nitrogenase, and others having the Mo and V but not the Fe nitrogenase or the Mo and Fe nitrogenases without the V nitrogenase. Azotobacter vinelandii (19), Azotobacter paspali (129), Rhodopseudomonas palustris (155), and the archaeon Methanosarcina acetivorans (76) are the only organisms so far identified that possess gene sets for all three nitrogenases. The combination of a Mo and a V nitrogenase is found in Azotobacter chroococcum, Azotobacter salinestris, and the archaeon Methanosarcina barkeri 27 (129) and in several cyanobacteria (see below). The Mo and Fe nitrogenases but not the V enzyme occur in Clostridium pasteurianum, Azomonas macrocytogeneses, and Azospirillum brasilense Cd (44) and in the phototrophs Rhodospirillum rubrum, Rhodobacter capsulatus, and Heliobacterium gestii (17).
Probes have been developed from vnfG and anfG to specifically amplify gene segments by PCR and to detect the alternative nitrogenases in organisms. By this technique, Loveless et al. (130) were able to isolate seven diazotrophs from aquatic environments that possess an alternative nitrogenase(s) and belong to the fluorescent pseudomonads and azotobacteria of the gammaproteobacteria. Recently, 24 bacteria of the same group, one closely related to Enterobacter and another with sequences almost identical to those of Paenibacillus, were isolated from diverse habitats, all with an alternative nitrogenase(s) (17). Besides in pseudomonads and azotobacteria, alternative nitrogenases occur only occasionally and in prokaryotes of totally unrelated taxonomic affinities. The rather close sequence similarities of the nitrogenase genes suggest that they may have arisen by gene duplication in the azotobacter-fluorescent pseudomonad group (17). In other organisms, however, there is little correlation between vnfG and anfG sequences on the one hand and the phylogeny inferred from the 16S rRNA gene sequence data on the other. This could mean that alternative nitrogenase genes may have been interspersed by lateral gene transfer among nonmembers of the azotobacter-pseudomonad group.
An indicator of this situation seems to occur in Methanosarcina barkeri 227 (47). This archaeon possesses a D gene and a G gene with close sequence homologies to vnfDG from other organisms, particularly Anabaena variabilis. The vnfH gene is separated from the vnfDGK cluster by two open reading frames (ORFs). Phylogenetic analysis indicates that this H gene is a member of a separate cluster comprising anfH genes of several bacteria and is closely related to anfH from Rhodobacter capsulatus and Clostridium pasteurianum. This cluster might also include vnfH from A. vinelandii. In another methanogen, Methanococcus maripaludis, with only a single nitrogenase, nifD and nifK cluster with the other genes for the Mo nitrogenase, whereas the H gene is an amalgam of both Mo and V nitrogenase H genes (113). Thus, vnfH and vnfDGK may have been acquired from other organisms by two independent gene transfers.
Such processes are difficult to understand because there is no apparent selective pressure to acquire and maintain alternative nitrogenases. Conditions in nature where Mo is growth limiting in soils or aqueous habitats are unknown, and microorganisms have high-affinity transport systems that effectively mobilize Mo from habitats (168). These mobilizations may result in microzones of Mo depletion around microorganisms where bacteria that can express an alternative nitrogenase(s) have a selective advantage (142). However, the isolation of diazotrophs with alternative nitrogenases from habitats with sufficient Mo concentrations (17) may indicate that these enzymes could have other, but so far totally unresolved, functions in nature. Otherwise, why would these genes, if redundant, be retained in organisms during evolution?
NITROGENASES IN CYANOBACTERIA
Occurrence of Nitrogenases in HeterocystsCell-free preparations of nitrogenases from all organisms are irreversibly damaged by O2, and different groups of microorganisms have been versatile in developing various means to protect their nitrogenases against the O2 of the air (153). In cyanobacteria, the O2 problem is enhanced by the photosynthetic production of this gas. Many filamentous cyanobacteria solve the issue by cell differentiation. Under aerobic growth conditions, their vegetative cells perform photosynthetic O2 evolution and CO2 fixation, whereas nitrogenase resides in specialized cells, the heterocysts (66). These differentiate from vegetative cells by cell division and extensive metabolic changes (133, 162). Photosystem II (PSII) is largely degraded in heterocysts so that they cannot perform the photosynthetic water-splitting reaction. They are also unable to fix CO2 photosynthetically. Vegetative cells provide photosynthetically fixed carbon, which may be exported as sucrose to the heterocysts (52). In turn, heterocysts provide nitrogen, likely as glutamine formed via ammonia generated by N2 fixation and both glutamine synthetase and glutamate synthase (219). Alternatively, glutamine may be converted to arginine which is then incorporated into the cyanophycin granule. This may be degraded by cyanophycinase in a dynamic way depending on the N demand of heterocysts and vegetative cells (86).
Heterocysts possess a thick cell envelope composed of long-chain, densely packed glycolipids providing a barrier to gas exchange (9). The main diffusion pathway for O2 and N2 might be through the terminal pores (“microplasmodesmata”) (83) that connect heterocysts with vegetative cells. Walsby (230) suggested that transmembrane proteins make the narrow pores permeable enough and might provide a means of regulating gas exchange. Residual O2 reaching the inside of the heterocysts might be immediately consumed by their high respiratory activity and also other reactions in these cells. In this way, heterocysts provide an anaerobic environment which allows nitrogenase to function.
The occurrence of nonspecific intercellular channels between heterocysts and vegetative cells has recently been confirmed (149). Any analogy to the plasmodesmata of higher plants is misleading, however, because cyanobacteria do not possess an endoplasmic reticulum. However, the export of metabolites might follow the source-sink gradient along the intercellular channels of both plants and cyanobacteria. Alternatively, the periplasmic space between the peptidoglycan layer and the outer membrane could constitute a communication conduit for the transfer of compounds, since this space is continuous between heterocysts and vegetative cells (72).
Heterocyst formation from vegetative cells of Anabaena species takes about 24 h after the cells have suffered N deprivation. More than 500 proteins are differentially expressed in heterocysts during cellular transformation from vegetative cells (162), showing that this complex process is under the control of many genes. Master regulators are HetR, a serine-type protease, and NtcA, a nitrogen control transcription factor in cyanobacteria (115, 152, 160, 200). Expression of hetR is upregulated by nitrogen deprivation, and this upregulation depends on NtcA (62). Heterocyst formation is also controlled by the availability of 2-oxoglutarate, which provides the carbon skeleton for the incorporation of inorganic nitrogen and which also serves as a signal molecule of the organic carbon content in the developing heterocysts (122, 161, 223). NtcA is the main 2-oxoglutarate sensor for the initiation of heterocyst differentiation (239). The otherwise important signal protein PII, which is involved in regulation of nitrogen metabolism in bacteria and plants, is apparently not required for heterocyst formation (240). Nitrogenase synthesis has a high demand for Fe. The uptake of this element is controlled by furA, whose expression is also modulated by NtcA and HetR (128). The reader is referred to review articles on this complex regulatory cascade (84, 94).
Before nitrogenase can be expressed in Anabaena sp. strain 7120, a gene rearrangement has to occur within nifD. An 11-kb DNA element is excised by a specific enzyme (XisA), and the two fragments of nifD are ligated to allow nitrogenase transcript formation to proceed. The excisase gene xisA is located on the excised DNA element. This gene rearrangement occurs in heterocystous cyanobacteria, such as the best-studied species Anabaena variabilis (37) and Anabaena (Nostoc) PCC 7120 (43), but not in nonheterocystous, N2-fixing forms (93). Similar rearrangements happen during the late stages of heterocyst development of some cyanobacteria. These include excision within a special ferredoxin (fdxN) of a 55-kb element by XisF and excision of a 10.5-kb element within the large subunit of uptake hydrogenase (hupL) (see below) mediated by XisC. These genetic elements may represent ancient viruses that have come under the control of the host and are excised as required. Similar gene rearrangements were detected during spore formation in bacteria. The subject has been reviewed (93), and newer publications on this subject are available (43, 96, 198).
Electron Transport to Nitrogenase in CyanobacteriaElectron transport to nitrogenase has been studied extensively in heterocystous cyanobacteria. Heterocysts have a very active ferredoxin- and photosystem I-dependent cyclic photophosphorylation (28) which generates the ATP for N2 fixation. These cells possess several ferredoxin-like Fe-S proteins. Of these, a special FdxH is expressed only in heterocysts and was proposed to serve as the electron carrier to nitrogenase. However, mutants with mutations in FdxH can still perform N2 fixation at a high rate (138), indicating that this protein can be replaced by others. Another ferredoxin-like protein, FdxB (PatB), is specifically expressed in heterocysts (107). Neither ferredoxin was identified in a quantitative proteomic investigation of heterocysts (163).
Reducing equivalents for the reduction of ferredoxins can be generated by several pathways (Fig. 5). In heterocysts, in the light, ferredoxin can be reduced via photosystem I. Alternatively, either NAD(P)H and a dehydrogenase or H2 and uptake hydrogenase (see below) can feed in electrons at the plastoquinone site (or close to it). In darkness, ferredoxin can be reduced by NAD(P)H and NAD(P)H:ferredoxin oxidoreductase (FNR) present in heterocysts and vegetative cells. The reduction of ferredoxin can also be achieved by the pyruvate phosphoroclastic reaction. Here, pyruvate and coenzyme A are cleaved to acetyl coenzyme A and CO2, and the remaining two electrons are transferred to ferredoxin. The enzyme involved, the pyruvate:ferredoxin oxidoreductase (PFO) is typically distributed among anaerobes, either strict (Clostridium) or facultative (Escherichia coli).
Generation of reductant for N2 fixation in cyanobacteria. The details are explained in the text.
A somewhat controversial issue arose regarding the occurrence of PFO in cyanobacteria. The enzyme was originally observed in extracts from Anabaena variabilis (120) and was then characterized in much greater detail from Anabaena cylindrica (151). Extracts from the latter cyanobacterium catalyzed the pyruvate-dependent reduction of methyl viologen (as an artificial substitute of ferredoxin) with formation of CO2 and the synthesis of acetohydroxamate from the acetyl coenzyme A produced. The reverse reaction, the synthesis of pyruvate from acetyl coenzyme A, CO2, and reduced ferredoxin, was also demonstrated. This reaction is even more indicative for the occurrence of the pyruvate:ferredoxin oxidoreductase because the pyruvate dehydrogenase complex is thermodynamically unable to catalyze this reaction.
Despite all this work, the occurrence of the phosphoroclastic reaction in cyanobacteria was not readily accepted in the literature until 1993, when two groups independently published sequences of the nifJ gene, encoding PFO. The enzyme from Anabaena sp. PCC 7120 was expressed only under Fe deficiency in the growth medium (12), whereas it was constitutive and independent of the Fe content in A. variabilis (192). The sequenced parts of the two nifJ genes showed only a low similarity of ca. 75%, in contrast to the sequences of other genes from the two organisms, which did not differ by more than 5%. The genome sequencing project for Anabaena 7120 then revealed that this cyanobacterium contained two nifJ genes and that the two above-mentioned groups had each sequenced a different nifJ copy. All cyanobacterial PFO sequences cluster with those from strict anaerobes, such as Clostridium or Desulfovibrio (191). However, as shown by the lux reporter system, PFO is expressed both under aerobic growth conditions and in Fe-replete medium in the unicellular, non-N2-fixing Synechococcus sp. PCC 7942. This cyanobacterium and other completely sequenced unicellular cyanobacteria contain only one PFO. Their genomes also contain sequences for phosphotransacetylase and acetate kinase. Acetyl coenzyme A could, therefore, be converted to acetyl-phosphate and then to ATP as a fermentative generation of additional energy. Such ATP generation has, however, never been verified experimentally in cyanobacteria.
Under Fe deficiency conditions, some cyanobacteria synthesize flavodoxin (formerly termed phytoflavin) instead of ferredoxin (221). Despite statements to the contrary (12), flavodoxin effectively transfers electrons to nitrogenase when properly reduced (32). Flavodoxin exists in three redox states, the oxidized, semiquinone, and fully reduced (hydroquinone) forms. Only the hydroquinone/semiquinone couple, with an E0′ of about −500 mV, can transfer electrons to nitrogenase in cyanobacteria (32) and in Azotobacter (235). Reduction of flavodoxin to the fully reduced state does not occur effectively using NAD(P)H [E0′ of NA(P)H/NAD(P)+ = −320 mV], but it can proceed via photosystem I or from pyruvate (E0′ for the pyruvate cleavage ∼ −500 mV). Flavodoxin is constitutive in the nonphotosynthetic aerobe Azotobacter vinelandii (225). It remains to be elucidated under what conditions flavodoxin has a physiological role in cyanobacteria. Fe deficiency is generally not a constraint in nature that demands the expression of flavodoxin. The demonstration of flavodoxin, other flavoproteins, and other ferredoxin-like electron transferring proteins in heterocysts of Nostoc sp. PCC 7120 (162) in non-Fe-limited cultures may indicate that other, still unresolved electron transfer pathways operate in these specialized cells. Similar evidence may be derived from work with Nostoc punctiforme ATCC 29133, where two ferredoxin-like electron transport proteins show a markedly increased abundance together with FNR in heterocysts (163). Flavodoxin was reported to enhance cyclic electron flow around photosystem I in salt-stressed cells (89), which may also occur in N2-fixing heterocysts.
Alternative Nitrogenases in CyanobacteriaThe occurrence of the V nitrogenase in cyanobacteria was first inferred from physiological evidence with A. variabilis (111). Under Mo deficiency and with V in the culture medium, this cyanobacterium reduced significant amounts of C2H2 to C2H6 and also produced much more H2 than Mo-grown cells. Subsequently, Thiel and coworkers performed the molecular characterization in great detail (216). In A. variabilis, the vnfDGKEN genes occur as a cluster, whereas four other H genes, in addition to nifH, are interspersed on the chromosome (Fig. 4). A vnfH gene is located 23 bp from vnfDGK. Either NifH or VnfH can act to complement either Mo or V nitrogenase. Two copies of the H gene exist in Nostoc punctiforme, which does not possess any other genes encoding an alternative nitrogenase (Fig. 4).
Among cyanobacteria, the V nitrogenase has been found only in A. variabilis, in an Anabaena isolate from the fern Azolla (154), in the southern Chinese rice field isolates Anabaena CH1 and Anabaena azotica (26), and recently in one Nostoc strain and two Anabaena strains (141). Anabaena azotica thrives at high temperatures at which Azolla dies. A different expression pattern for the two cyanobacterial nitrogenases, possibly dependent on growth temperature, was suspected (26). In support of this idea, the V but not the Mo nitrogenase of A. vinelandii has been found to be active at lower temperatures (167). However, the specific activities of C2H2 reduction for both Mo and V nitrogenase of A. azotica were found to be the same over a range of temperature and light regimens (26). Thus, the V nitrogenase is unlikely to provide a selective advantage for A. azotica at higher temperatures. Other conditions, such as Mo-deficient microzones around microbial colonies, unusually high W concentrations (which block Mo nitrogenase synthesis), or high alkalinity (pH of ∼10), have been suggested, but not proven, to favor V nitrogenase gene expression (222).
The close sequence similarity of the cyanobacterial vnfDG genes to those of Methanosarcina spp. could indicate an archaeal origin for the alternative nitrogenase similar to that for the Mo enzyme (176). Alternatively, these two groups of organisms with totally unrelated taxonomic affinities may have retained these genes in evolution by chance.
Some physiological evidence has been presented for the existence of the Fe nitrogenase in A. variabilis (112). However, the completely sequenced chromosome of this organism and of more than 30 other cyanobacteria did not reveal genes coding for the Fe nitrogenase, and a nifH vnfH double mutant of Anabaena variabilis did not grow diazotrophically (172). Thus, the evidence, particularly the positive results after hybridization with an anfH probe from Azotobacter vinelandii (112), must indicate the presence of some other sequence-related entity (possibly two other nifH copies [Fig. 4]). In the past, searches for nitrogenases were often based on probing with the nifH gene. However, sequences of anfH are significantly divergent from those of nifH and vnfH (59), and thus possibly a cyanobacterial Fe nitrogenase, say, occurring on a plasmid, may have been missed by probing with the nifH gene.
In waters, cyanobacteria thrive under oxygenic conditions where Fe is generally limiting but Mo or V is abundantly available (63). Those authors suggest that these conditions may favor the expression of Mo or V nitrogenase, whereas the concentration of Fe is possibly too low to allow synthesis of the Fe nitrogenase.
In 1995, two groups independently reported the existence of a second Mo nitrogenase in A. variabilis (193, 217). The “classical” Mo nitrogenase occurs only in heterocysts of this organism. The second Mo nitrogenase is encoded by a separate set of nifHDK genes and is expressed in vegetative cells under anaerobic or, more precisely, low-O2-tension conditions because these cells produce O2 photosynthetically. It resembles, by its expression under anaerobic conditions, the enzyme from the filamentous, nonheterocystous Plectonema (Leptolyngbya) boryanum (209). Its physiological and biochemical properties in A. variabilis have not been studied extensively. The distribution of this enzyme has recently been screened in several cyanobacteria (141).
Nitrogen Fixation in Nonheterocystous CyanobacteriaThe literature on nitrogen fixation in nonheterocystous cyanobacteria up to the mid-1990s was extensively reviewed (14). Therefore, this section concentrates on more recent results.
Many nonheterocystous cyanobacteria can fix N2, but almost all of them do so under anaerobic conditions, or, rather, under conditions of decreased O2 tension. Several of them were shown to separate these two incompatible reactions, with photosynthetic CO2 fixation being performed in the light and N2 fixation in darkness. Thus, at night, nitrogenase is not exposed to the photosynthetically produced O2 and respiration might then utilize most of the O2 of the air to provide anaerobic conditions, especially in dense cultures or in biofilms. However, not all nonheterocystous cyanobacteria show this circadian rhythm. Gloeothece and Synechococcus (Cyanothece) spp. also fix N2 during the day and can grow slowly under continuous illumination. In the oceans, the filamentous Trichodesmium may show a division of labor in which some cells perform photosynthesis whereas others fix N2 (14). However, a recent immunological study (156) revealed that more than 77% of all cells were nitrogenase immunopositive, indicating that Trichodesmium does not develop heterocyst-equivalent cells. Immunological studies indicated that nitrogenase in Plectonema, Gloeothece, and others is also uniformly distributed throughout all cells, thus showing no preferential association with a cell structure (14). Cyanobacteria did not develop O2 protection devices, such as changes in the enzyme's conformation upon exposure to excess O2 as in azotobacteria, production of leghemoglobin as in the rhizobia, or reversible modification of the Fe protein by ADP-ribosylation controlled by the DRAT/DRAG enzymes as in photosynthetic purple bacteria or azospirilla. Their respiratory activity does not seem to be extraordinarily high as in Azotobacter sp. or in heterocysts, to consume all O2 reaching within the cells (66). Thus, N2 fixation in light by these few aerobic cyanobacteria remains an enigma.
Cyanobacterial N2 fixation in the oceans contributes significantly to the global N budget (15, 55, 202). In temperate areas, heterocystous species can form blooms in summer, but they are somewhat unpredictable in time and location, as exemplified for the fresh- and brackish-water species Aphanizomenon flos-aquae and the toxin-producing Nodularia spumigena (143). The major organisms in oceanic N2 fixation in areas of the warmer tropical and subtropical regions of the Pacific Ocean are Trichodesmium sp. and the heterocystous Richelia intracellularis, which lives inside diatoms (74, 206). In other areas of the Pacific Ocean, N2-fixing cyanobacteria, such as Crocosphaera watsonii, and the non-N2-fixing Prochlorococcus marinus thrive in abundance (238). Other nanoplanktonic organisms may be even more important there. Small uncultured cyanobacteria that fix N2 but are unable to perform photosynthetic CO2 fixation and thus O2 evolution have now been recognized (237), and they are particularly active during winter in areas of the Pacific Ocean (117). They have not yet been characterized properly, but their nitrogenase DNA sequences resemble those of the “spheroid bodies” that occur in the fresh water diatoms Rhopalodia gibba and Epithemia sp. (80). These diatoms grow very slowly on agar plates. During the time before the use of molecular biology techniques, physiological experiments demonstrated light-dependent C2H2 reduction by R. gibba even with the rather small amounts of cell material then available (71). More recently, DNA sequencing showed that the spheroid bodies of R. gibba indeed possess the structural nitrogenase genes (173). The spheroid bodies and uncultured marine cyanobacteria either could perform cyclic phosphorylation or may be completely dependent on a supply of both ATP and reductant from organic carbon in the environment. These spheroid bodies, being N2-fixing entities within eukaryotic cells, might attract special attention in the near future for potential applications. They could serve as models in attempts to make plants independent from a supply with combined nitrogen by incorporating an N2-fixing cyanobacterium into their cells.
The discovery of a new group of N2-fixing cyanobacteria may appear to be totally unexpected. As mentioned above, nifH is very much conserved during evolution, and probing with nifH sequences should allow one to detect all N2-fixing microorganisms in environmental samples. Recent studies showed that most of the bacterial DNA sequences from soil (for nifH as well as for nosZ in denitrification and for the 16S rRNA gene for total bacteria) could be detected with the short DNA probes available, but the gene sequences in total were entirely new (60, 180).
HYDROGENASES IN GENERAL
The subject of hydrogenases has been extensively reviewed (226, 228). Therefore, just a few general facts will be mentioned here. There are three classes of hydrogenases: (i) the [FeFe] hydrogenase, (ii) the [FeNi] hydrogenase, and (iii) the methylenetetrahydroxymethanopterin-containing enzyme. The last enzyme is a homodimer, each subunit of which contains a low-spin, redox-inactive Fe atom which is involved in H2 splitting or formation (201, 211, 229). This enzyme has been found only in some methanogenic archaea. In all other hydrogenases, iron occurs in Fe-S clusters.
The [FeFe] hydrogenases have a unique active center (the H cluster) which produces about 100-fοld higher activity than the other hydrogenases (229). The simplest [FeFe] hydrogenase occurs in green algae with only the H cluster as the prosthetic group (91). The H cluster contains two Fe atoms and the two ligands CO and CN−, which are attached to both of the Fe atoms. In green algae, the H cluster is directly reduced by ferredoxin. All other [FeFe] hydrogenases contain a relay of additional FeS centers (both 4Fe-4S and 2Fe-2S clusters) that are involved in electron transfer from the external electron source (reduced ferredoxin) to the H cluster deep inside these monomeric proteins. They possess hydrophobic channels from the surface to the active site (the H cluster) that provide access for protons and the egress of H2. [FeFe] hydrogenases function mostly in the disposal of excess reductant generated during fermentation under anaerobic conditions. However, the periplasmic [FeFe] hydrogenase of Desulfovibrio vulgaris is involved in the utilization of H2 in sulfate reduction (171). The enzyme occurs in anaerobes, such as the genera Clostridium and Desulfovibrio, and in eukaryotes (in chloroplasts of green algae or in hydrogenosomes). It has not been detected in cyanobacteria. This is true also for those cyanobacteria that synthesize starch (semiamylopectin) and could therefore be considered ancestors of chloroplasts (150). The evolutionary origin of the [FeFe] hydrogenase of green algae is a mystery yet to be resolved (132, 146).
The majority of hydrogenases in prokaryotes are Ni-containing enzymes. The core enzyme is an αβ heterodimer where the larger subunit, of ca. 60 kDa, possesses the deeply buried binuclear NiFe active site (Fig. 6). The Fe in this center binds two CN− and one CO. The whole cluster is ligated to the protein by the thiolate groups of four cysteines. The smaller subunit, of ca. 30 kDa, harbors FeS clusters (up to three) which serve to transfer electrons from or to the NiFe active site. As in the [FeFe] hydrogenases, there are hydrophobic channels from the active site to the surface of this globular αβ dimer. The Ni hydrogenases have a high affinity (low apparent Km) for H2, indicating that they act mostly in utilizing H2 in the different organisms. Indeed, they are often linked to nitrogenase, where they serve to utilize the H2 produced in N2 fixation. They are often membrane bound and feed electrons into the respiratory chain via either ubiquinone or a cytochrome at respiratory complex III. Often they are synthesized with a long signal peptide of 30 to 50 amino acid residues which is cleaved off when the hydrogenase is folded and incorporated into the membrane. They may be subdivided into four groups by their functions (227, 228).
In the oxidized form, [NiFe] hydrogenases are inactive due to a bridging hydroxo ligand between the Ni and Fe atoms (Fig. 6), and the different enzymes vary in their sensitivity to O2. When reduced, this ligand is removed by conversion to water, with the simultaneous reduction of Ni3+ to Ni2+. The enzyme can then bind H2, probably at the Fe atom, and is then able to catalyze the heterolytic cleavage to 2H+ + 2e−. Details of this enzymatic mechanism have been depicted previously (228). Remarkably, none of the [NiFe] hydrogenases transfers electrons to ferredoxin or to another low-potential electron carrier. The structure/function relationship of anaerobic gas-processing metalloenzymes has recently been summarized (73).
The biosynthesis of hydrogenase, including the synthesis of the metallocenter and the incorporation of the CO and CN− ligands, has been studied extensively for hydrogenase 3 from E. coli by Böck and colleagues in Munich and has been reviewed (226, 228). The concentration of H2 in cells is sensed by hupUV gene products, which in other organisms are termed HoxBC. These proteins also catalyze the cleavage of H2 and can therefore be considered an independent, regulatory hydrogenase, e.g., in Ralstonia eutropha (79).
HYDROGENASES IN CYANOBACTERIA
Hydrogenase Types in CyanobacteriaThe subject of hydrogenase types in cyanobacteria has been repeatedly reviewed (7, 81, 82, 91, 99, 134, 194, 199, 212, 214, 222). The reader is particularly referred to the very detailed and elaborate review by Tamagnini et al. (214). Cyanobacteria contain two different Ni hydrogenases defined by their physiological role as either an uptake or a bidirectional (reversible) enzyme. There is no evidence for an H2-sensing regulatory hydrogenase encoded by hupUV. Cyanobacterial hydrogenases do not contain Se as do some hydrogenases in anaerobic bacteria.
Uptake hydrogenase.The uptake hydrogenase is encoded by the contiguous and cotranscribed genes hupSL and is associated with nitrogenase functioning. Generally, intact N2-fixing cyanobacteria show very little net H2 production due to the efficient recycling of the gas by uptake hydrogenase. This H2 consumption proceeds by the respiration- and photosystem I-dependent pathways (33). In cyanobacteria, respiration and photosynthesis share the cytochrome bc complex (respiratory complex III), from where the electrons are allocated either to the donor side of photosystem I to generate reduced ferredoxin or to respiratory complex IV accompanied by O2 consumption. Factors that control electron allocation to either photosystem I or respiration in light-grown cyanobacteria have not been elucidated. Likewise, the electron entry from H2 and uptake hydrogenase to either the plastoquinone pool or a cytochrome b, as in Xanthobacter autotrophicus (184) and presumably in Bradyrhizobium japonicum (65), is not known in cyanobacteria. Transcription starts before hupS and terminates immediately after hupL; thus, the electron acceptor is not cotranscribed on this operon. The enzyme does not couple with any other electron carrier with a redox potential more negative than −300 mV, which explains its unidirectional physiological function and name. Uptake hydrogenase is membrane bound and has never been characterized in the homogeneous form. Recent immunological studies confirmed its association with the thylakoid membranes of three cyanobacterial strains (195), which corroborates earlier studies with thylakoid preparations (reviewed, e.g., in reference 164). The sequences indicate that the larger subunit (HupL) has a molecular mass of about 60 kDa and that the smaller one (HupS) is about half that size.
In accordance with the postulates of Dixon (58), which were developed for Rhizobium nodules, H2 utilization in cyanobacteria likely functions (i) to remove O2 from the nitrogenase site via the respiratory oxyhydrogen (Knallgas) reaction, (ii) to regain ATP inevitably lost in H2 production during nitrogenase catalysis, and (iii) to prevent a deleterious buildup of a high concentration of H2 which affects nitrogenase activity. Such a situation might apply particularly to heterocysts. In addition, H2 uptake might provide additional reductant for N2 fixation, photosynthesis, and other reductive processes.
Rather simple physiological experiments, performed in student courses in the Cologne laboratory over the years, show that N2 fixation (C2H2 reduction), e.g., by Anabaena variabilis, is much less sensitive to exposure to O2 when the assay mixtures are supplemented with exogenous H2 (29, 34). Uptake hydrogenase-deficient mutants of several cyanobacteria produce roughly three times more H2 than wild-type cells (for references, see reference 214). However, their growth rates under N2-fixing conditions are essentially the same (125).
In other bacteria, a twin-arginine signal peptide at the N terminus and a hydrophobic motif, both presumably involved in translocation and anchorage, are typical for many membrane-bound hydrogenases. Such motifs are missing from the cyanobacterial HupS and HupL, which also do not contain signatures indicative of membrane insertion. As in other organisms, however, HupL contains the C-terminal extension that is cleaved off at the last step of maturation by a specific endopeptidase encoded by hupW.
In approximately half of the heterocystous strains (21, 213), hupL undergoes a rearrangement during the late state of heterocyst differentiation before it can be transcribed. The excision of the 9.5-kb element is catalyzed by the recombinase XisC, with its gene located on this element. XisC is sufficient to catalyze the site-specific recombination in hupL (43). The physiological advantage of such a site-specific recombination is not obvious. Of the two best-studied heterocystous cyanobacteria, Anabaena (Nostoc) sp. strain PCC 7120 shows this gene rearrangement but Anabaena variabilis ATCC 29413 does not.
Uptake hydrogenases occur in almost all N2-fixing microorganisms except for some Rhizobium strains (35, 36) and Herbaspirillum seropedicae (F. Pedrosa, personal communication). In cyanobacteria, the enzyme is present in all N2-fixing species with the exception of an N2-fixing unicellular strain, Synechococcus sp. strain BG 043511 (132), and some Chroococcidiopsis isolates (see below). No uptake hydrogenase and none of its genes have been unambiguously detected in non-N2-fixing cyanobacteria. It is not clear whether an uptake hydrogenase is expressed in parallel with the second Mo nitrogenase which is active in vegetative cells of A. variabilis upon transition to anaerobiosis. Low transcript levels of hupSL have been reported for A. variabilis ATCC 29413 cells grown in the presence of ammonia (231).
The formation of hupSL transcripts may be controlled by factors such as Ni availability, anaerobiosis, the presence of H2, and the absence of combined nitrogen, among others, and may proceed in parallel with heterocyst formation (92, 98, 231). The transcriptional regulator NtcA, which controls cyanobacterial genes involved in nitrogen metabolism, has also been reported to regulate hupSL expression (231). The NtcA binding site was identified 427 bp upstream of the transcriptional start site of hupSL in A. variabilis, whereas most other NtcA binding sites are located not more than 40 bp from the start site (231). The NtcA binding site identified in Nostoc punctiforme ATCC 29133 is TGTN9ACA, which differs from the optimal one, GTAN8TAC, and this might therefore indicate only weak binding (98). A shorter promoter fragment, covering 57 bp upstream of and 258 bp downstream of the transcription start site, was enough for high heterocyst-specific expression of hupSL independent of NtcA (98). Surprisingly, hupSL expression in A. variabilis ATCC 29143 was not regulated by H2 (231). This is in sharp contrast to the situation in Nostoc punctiforme and N. muscorum (10). In addition to transcriptional regulation, uptake hydrogenase synthesis could also be controlled at the posttranslational level. This enzyme, but not bidirectional hydrogenase, is activated by thioredoxin (164). N2 fixation by cyanobacteria is largely light stimulated due to the demand for reductant (reduced ferredoxin) and ATP. Therefore, activation of the uptake hydrogenase by photosynthetically reduced thioredoxin makes sense physiologically because more H2 is produced by nitrogenase in the light than in darkness. The number of proteins activated by thioredoxin is high in cyanobacteria and chloroplasts, but the target enzymes differ in the two entities (124).
Other transcription and translation cues will undoubtedly be resolved in the near future to further understanding of the signal cascade involved in the synthesis of the uptake hydrogenase. The currently available data suggest that different cyanobacteria differ markedly in their patterns of expression of this protein.
Bidirectional hydrogenase.After extensive controversy, the work of Houchins and Burris (100, 101) clearly showed that N2-fixing cyanobacteria may contain another hydrogenase in addition to the uptake enzyme. This reversible, bidirectional hydrogenase, which catalyzes both H2 uptake and reduced methyl viologen-dependent H2 evolution, was separated from the unidirectional, uptake enzyme in crude extracts of Anabaena (Nostoc) sp. strain PCC 7120. Later, molecular biological characterization showed that the bidirectional hydrogenase in cyanobacteria is, surprisingly, a NAD(P)H-dependent enzyme (187). This finding had been marked as a milestone in cyanobacterial hydrogenase research (212). The enzyme has a pentameric structure encoded by the genes hoxEFUYH in A. variabilis. HoxYH constitutes the hydrogenase, which contains the motifs for binding both Ni-Fe-S and Fe-S centers. HoxFU is the diaphorase part that transfers the electrons to NAD(P)+ and possesses binding sites for NAD(P)+, flavin mononucleotide (FMN), and Fe-S centers. The enzyme complex contains a further HoxE subunit, which copurifies with the active bidirectional enzyme (188). The hoxE gene possesses a motif for binding an Fe center and was therefore thought to couple the enzyme to the respiratory and photosynthetic electron transport chain on the thylakoids and also possibly at the cytoplasmic membrane. However, the role of the hoxE gene product has not been resolved yet, despite extensive research.
In organisms other than cyanobacteria, a pentameric NADH-dependent bidirectional hydrogenase is present in Thiocapsa roseopersicina (174) and in Allochromatium vinosum (108, 127). The best-studied bidirectional hydrogenase, the NADH-dependent enzyme from Ralstonia eutropha, is encoded only by hoxFUYH (75).
The locations of the five structural genes hoxEFUYH on the chromosome differ from one cyanobacterium to the next (25, 199, 214). In some cyanobacteria, they are clustered on one part of the chromosome, though interspersed with ORFs at different positions. In others, they occur in two different parts of the genome separated by several kilobases of intervening DNA. Similar to the case for HupL of the uptake hydrogenase, HoxH of the bidirectional hydrogenase undergoes maturation at the C terminus catalyzed by a specific endopeptidase encoded by hoxW. The expression of hydrogenase genes in Synechococcus sp. PCC 7942 is under the control of the circadian clock, as shown for two promoters of the gene cluster (186). When expressed, the native protein might function as a dimeric assembly complex Hox(EFUYH)2 (188). In extracts, it catalyzes both NAD(P)+-dependent H2 uptake and H2 evolution with NADP(P)H as the electron donor (190).
Bidirectional hydrogenase is widespread in cyanobacteria. It is present in unicellular, filamentous, and heterocystous species, where it occurs in both heterocysts and vegetative cells (213). The enzyme is apparently not present in marine cyanobacteria isolated from the open ocean (132). It is expressed independently of N2 fixation and thus is present in cells grown aerobically and with combined nitrogen. However, it is more O2 sensitive than uptake hydrogenase, probably due to oxidation to its inactive state (51). When reduced, it can be purified as a pentameric complex (188).
The regulation of the expression of the bidirectional hydrogenase in cyanobacteria differs with the physical location of the hox genes on the chromosome in the species. In Synechococcus sp. PCC 7942 (= Anacystis nidulans), the genes are organized into two clusters, hoxEF and hoxUYHWhypAB, and are regulated by three promoters, one before each of hoxE, hoxU, and hoxW (23, 186). In Synechocystis sp. PCC 6803, the hoxEFUYH genes are cotranscribed, with the transcription start point located 168 bp upstream of the start codon (87, 158). Taking the high diversity of the different cyanobacterial species into account, expression of the bidirectional hydrogenase in cyanobacteria seems to be species specific.
Over the last several years, significant progress has been achieved in the identification of the transcription factors regulating the expression of bidirectional hydrogenase, and details of the subject are found in a very recent review (159). NtcA does not seem to be the transcriptional activator, but a LexA-related protein (87, 158) and two members of the AbrB-like family (157) appear to be activators. In other organisms, LexA activates the expression of a cascade of genes coding for enzymes involved in either DNA repair or carbon starvation. A LexA-depleted mutant of Synechocystis sp. 6803 had lower hydrogenase activity than the wild type, indicating that LexA operates as a transcription activator of hox genes in this cyanobacterium (87). The binding site of LexA upstream of hoxE of Synechocystis sp. PCC 6803 is, surprisingly, not clear (214). LexA may bind to a region from bp −198 to −338 from the translational start point (158), to the region from bp −592 to −690 bp from the hoxE start codon (87), or to both regions (159). The two distant LexA binding regions in the hox promoter could indicate the occurrence of a DNA loop involved in gene transcription (86, 159), which warrants experimental proof. LexA may act as mediator of the redox-responsive regulation of hox gene expression (5). In Synechocystis sp. strain PCC6803, LexA binds as a dimer to 12-bp direct repeats containing a CTAN9CTA sequence in target genes (170).
Abr proteins act as transcription factors of antibiotic resistance in organisms other than cyanobacteria. An AbrB-like protein (sII0359) was recently shown to interact specifically with the promoter region of the hox genes and with its own promoter region (157). Whereas this AbrB-like protein works as a transcription activator in Synechocystis sp. PCC 6803, another one of these regulator proteins (sII0822) acts as repressor of the hox gene expression, because they were significantly upregulated in a completely segregated ΔsII0822 mutant (105). This transcription factor works in parallel to, but apparently independently from, the long-known nitrogen transcriptional control element NtcA (97) in the regulation of the expression of genes coding for nitrogen assimilation enzymes (105).
The cyanobacterial transcription factors, the LexA- and AbrB-like proteins, show significant divergences in their sequences and functions from the counterpart proteins in other organisms, and their activity may be regulated by posttranscription modifications (159). They are members of an apparently complex signal cascade that directs the expression of the bidirectional hydrogenase genes. Their expressions and interactions in responses to environmental cues might be a subject of extensive research in the near future (159). The identification of other transcription factors of bidirectional hydrogenase is to be expected (116).
Besides its inactivation by O2 and a non-light dependence (51), bidirectional hydrogenase seems to be activated by H2 on the transcriptional or translational level or even on both. The effects of H2 on bidirectional hydrogenase synthesis are not understood and appear to vary with the organism and the culture conditions employed. In some cases, high hydrogenase activity could be the result of bacterial contamination of slime-forming cyanobacterial cultures.
The biosynthesis and maturation of the [NiFe] hydrogenase have been characterized for the enzyme from E. coli (20). The hyp genes required for the synthesis of the hydrogenase are similar in E. coli and cyanobacteria and are scattered throughout the genomes of those cyanobacteria in which their occurrence was examined (reviewed in reference 214). Both uptake and reversible hydrogenases appear to utilize the same hup gene products for their biosynthesis. However, the last step, the maturation at the C terminus by endopeptidase, seems to be specific for the two enzymes, with HupW catalyzing the final cleavage of uptake hydrogenase and HoxW involved in processing the bidirectional enzyme (233). Both endopeptidases are transcribed from their own promoters (67) and are under similar regulatory control as the hydrogenases they cleave (54).
In contrast to uptake hydrogenase, the bidirectional enzyme is soluble after breaking cyanobacterial cells. The exact location of the enzyme inside the cells is unknown (Fig. 7). Immunological (109) and membrane solubilization (110) studies indicated a location at/on the cytoplasmic membrane in Anacystis nidulans (Synechococcus PCC 6301). Other researchers with different antibodies found a location in the cytoplasm, with some preferential association to the thylakoids (213, 214). However, all these investigations with antibodies were performed before the true nature of the hydrogenase as a pentameric NAD(P)H-dependent complex was recognized. Clearly, this issue needs to be reexamined with newly raised antibodies.
Possible coupling of the bidirectional hydrogenase to the cytoplasmic membrane in cyanobacteria. The HoxE subunit may serve as a device for coupling to the membrane, but this has not been verified experimentally. Solubilization experiments indicate that the bidirectional hydrogenase is loosely membrane bound (110).
The physiological function of this constitutively expressed bidirectional hydrogenase in photosynthetic, aerobic cyanobacteria has been hotly debated but remains controversial. Work with mutants of Anabaena (Nostoc) sp. PCC 7120 (139) showed that the bidirectional hydrogenase is unable to support N2 fixation. Its high affinity (low apparent Km value) for H2 suggests that the enzyme functions in H2 utilization under physiological conditions (99). Indeed, H2 uptake catalyzed by the bidirectional hydrogenase can support photosynthetic reactions such as CO2 fixation and also, to some extent, nitrite or sulfite reduction (215). The rates of these reduction reactions with H2 as the only electron donor are low, however, compared to these same photosynthetic activities with H2O as the electron source (31). Bacteria, such as Ralstonia eutropha or Xanthobacter autotrophicus (185), are able to grow autotrophically with H2 as the sole source of reductant and energy, and some of them, such as Bradyrhizobium japonicum, can do so even under N2-fixing conditions (204). H2-dependent growth in darkness has never been demonstrated for any cyanobacterium. Such anoxygenic growth is possible when energy is provided by cyclic photophosphorylation and when the electrons are provided from Na2S or H2S in some cyanobacteria, such as Oscillatoria limnetica (78). However, to our knowledge, H2- and photosystem I-supported growth has not yet been demonstrated in cells of Anabaena, Nostoc, or other autotrophic unicellular species when photosystem II is impaired by use of dichlorophenyldimethylurea (DCMU). In the two facultative anoxygenic cyanobacteria Oscillatoria limnetica and Aphanothece halophytica, however, H2 was described to substitute for H2S in supporting CO2 fixation in a photosystem I-driven reaction (13).
In all organisms, the respiratory complex I consists of at least 14 subunits, but only 11 in the cyanobacterial NADPH-dehydrogenase complex I have as yet been identified. The diaphorase genes hoxEFU show high sequence homologies to the missing three genes. Although it has been suggested that the hoxEFU gene products are used simultaneously by both the bidirectional hydrogenase and respiratory complex I (189), the experimental evidence is against this suggestion. Mutants with mutation either in hoxF (102) or in hoxU (22) do not show bidirectional hydrogenase activity but have unimpaired respiratory activity. Furthermore, Nostoc PCC 73102 has no bidirectional hydrogenase activity at all but respires with rates comparable to those of other cyanobacteria (22). This could mean that cyanobacterial respiration partly circumvents respiratory complex I and utilizes the succinate dehydrogenase complex instead, as may be inferred from studies with mutants (49). Then the fate of the NAD(P)H generated in carbon catabolism has to be determined. The electron input pathway into respiratory complex I in cyanobacteria remains unknown (11).
Some authors consider the bidirectional hydrogenase to work in the transition from anaerobiosis in the dark to aerobic conditions in the light (6, 51, 88, 132). In order to avoid an overload of reducing equivalents, the organisms react to dispose of the excess by generating a burst of H2 via photosynthetic electron transport, ferredoxin, FNR, NADPH, and hydrogenase. Such sudden H2 production that lasts for only seconds up to few minutes, has been observed repeatedly. However, the physiological relevance of this observation is questionable, because the sun does not rise so suddenly in the morning that it overreduces soil cyanobacteria. Furthermore, in aqueous habitats, turbulences are hardly so effective that they expose cyanobacteria to extremely high light intensities within a very short time scale. Cyanobacteria may, however, be overreduced when continuously exposed to too bright a light on a very sunny day and then be forced to use hydrogenase as a valve for disposing of the excess of photosynthetically produced reductants, as shown in laboratory cultures of Anabaena cylindrica (119).
As stated in an extensive review (207), the majority of cyanobacteria are obligate photoautotrophs. Only few species are able to grow chemoautotrophically at the expense of a limited number of organic carbon compounds, and they do so with O2 as the terminal respiratory electron acceptor. Anaerobic chemoorganic growth is exceptional in cyanobacteria. Thus, most species accumulate glycogen in the light, which they then have to degrade in darkness. Glucose residues from glycogen are utilized via the oxidative pentose phosphate pathway, finally resulting in pyruvate (208). Its further degradation is hampered by the fact that the tricarboxylic acid cycle is incomplete in cyanobacteria because neither an oxoglutarate dehydrogenase complex nor an oxoglutarate:ferredoxin oxidoreductase is present (208), which has been confirmed by recent large-scale proteomic studies (162, 163). This prevents the complete degradation of the C2 moiety to CO2 and NAD(P)H. Cyanobacteria apparently prefer to utilize NADP+ rather than NAD+ in catabolism (51), since several enzymes, such as isocitrate dehydrogenase (165) and glyceraldehyde-3-phosphate-dehydrogenase (166), are NADP+ rather than NAD+ dependent. In darkness, most cyanobacteria have to generate their energy via the oxidative pentose phosphate pathway: pyruvate, pyruvate:ferredoxin oxidoreductase, reduced ferredoxin, FNR, and NADPH (Fig. 8). By using the lux reporter system, it was shown that the pyruvate:ferredoxin oxidoreductase is constitutively expressed, even in aerobically grown A. variabilis (191). In dense cultures, biofilms, mats, or cyanobacterial blooms, the amount of O2 may rapidly become insufficient to oxidize all NAD(P)H by respiration. Thus, the NADPH generated via pyruvate:ferredoxin oxidoreductase and FNR must then be reoxidized via the bidirectional hydrogenase in order to avoid overreduction in the cells. The generation of H2 (E0′ = −420 mV for H2/2H+) from NAD(P)H [E0′ = −320 mV for NAD(P)H/NAD(P)+] is thermodynamically unfavorable. It requires a 1,000-fold excess of reduced pyridine nucleotides, but this can rapidly be generated in dark-kept cells under anaerobic conditions. To prevent overreduction of the cells during the night, reducing equivalents must be disposed of as H2 (Fig. 8). Similar to the case for pyruvate:ferredoxin oxidoreductase, bidirectional hydrogenase is also constitutively expressed under aerobic growth conditions. When cyanobacteria such as Synechocystis, Anabaena, or Nostoc sp. are transferred to darkness and anaerobiosis, H2 production begins immediately without a distinct lag phase. High hydrogenase activity under anaerobic conditions was described long ago (99), and an increase in the hoxH (67, 68) or hoxEF (116) transcription levels during dark periods was recently detected in different cyanobacteria.
Roles of bidirectional and uptake hydrogenases in cyanobacterial hydrogen metabolism. Bidirectional hydrogenase is active mainly in the dark and under anaerobic conditions to dispose of reductants, whereas uptake hydrogenase functions in recycling the hydrogen lost during nitrogen fixation.
Thus, cyanobacteria might have retained the genes coding for these enzymes (hydrogenase and pyruvate:ferredoxin oxidoreductase) of anaerobes because of their obligatory autotrophy (of many species). The essential role of hydrogenase during fermentation of cyanobacteria has also been suggested by others (224). As recently shown (218), cyanobacteria contain one petH gene that encodes two isoforms of FNR, one of which accumulates under heterotrophic conditions. It needs to be shown whether the latter is specifically involved in the fermentative degradation of pyruvate. The same question also applies to the two isoforms of pyruvate:ferredoxin oxidoreductase in heterocystous species. As mentioned above, acetyl coenzyme A formed in pyruvate fermentation may be converted to ATP by phosphotransacetylase and acetate kinase, but this also remains to be shown. ATP formation by this pathway must be accompanied by the formation of acetate, but the fate of any acetate produced remains unknown.
In photosynthetic eukaryotic algae, hydrogenase is located in plastids (210). The ancestors of plastids are believed to be organisms similar to the filamentous, heterocyst-forming, N2-fixing species of class IV of the cyanobacteria, related to the current Nostoc or Anabaena spp. (53). If so, it is surprising that, during evolution, plastids have lost not only N2 fixation genes but also both gene sets that encode the bidirectional and uptake hydrogenases. When hydrogenase occurs at all in plastids, it is an [FeFe] hydrogenase of a completely unknown origin.
Similarly, it is totally unclear how both hydrogenases have been acquired by cyanobacteria from bacteria over evolutionary time. With respect to photosynthetic bacteria, the green nonsulfur bacterium Chloroflexus aurantiacus possesses both uptake and bidirectional hydrogenases, which has led to the assumption that a Chloroflexus-like bacterium is the ancestor of C. aurantiacus and cyanobacteria (132). On the other hand, the first phototrophs may have been anoxygenic procyanobacteria from which the Chlorobiaceae, Heliobacillaceae, Chloroflexaceae, purple sulfur bacteria, and cyanobacteria descended in parallel and independently of each other (148). The gene sets of both cyanobacterial hydrogenases may have been acquired vertically or laterally. A lateral gene transfer is particularly difficult to conceive for the bidirectional enzyme because its genes may be scattered throughout the genome of a species. Similarly, the loss of hydrogenase from one cyanobacterial isolate but not from another may be difficult to explain.
The unicellular cyanobacterium Chroococcidiopsis sp. (Fig. 9 A to C) is regarded as a fossil relict which may have properties related to those of the first O2-evolving cyanobacterium developed some 3 × 109 years ago (69). Chroococcidiopsis is being proposed as the organism best suited to go on exploratory missions to Mars (48). Today, Chroococcidiopsis thrives at sites with extremely hostile conditions (24). The strains Chroococcidiopsis thermalis ATCC 29380 (1) and CALU 758 (197) were found to possess the bidirectional, but not the uptake hydrogenase and to fix N2 (reduce C2H2) under microaerobic conditions. However, experiments performed in the Cologne laboratory (106) showed that the hydrogenase activities of Chroococcidiopsis sp. strain PCC 7203 exhibited some unusual features. Southern hybridizations and PCR experiments with probes from hupL and hoxH, hoxF, or hoxE developed from A. variabilis sequences indicated the presence of the bidirectional hydrogenase but the absence of the uptake enzyme in Chroococcidiopsis PCC 7203. In this cyanobacterial strain, H2 and the bidirectional hydrogenase can support nitrogenase activity (C2H2 reduction) but only at a rather low concentration of 0.3 to 0.5% O2 in the gas phase. Above that concentration, O2 is completely inhibitory, presumably by oxidizing the NiFe center of the enzyme to its inactive oxidized state or (less likely) by affecting an extremely O2-sensitive nitrogenase in this organism. In more than 100 different experiments performed in air-free vessels, about 50% showed no H2-supported C2H2 reduction activity, whereas the outcome was positive in the other half. However, the optimal O2 concentration was 0.3% in one experiment and 0.5% in the next, depending on the concentration of cells in the assay vessels, the photosynthetic O2 production activity of the cells, and the success in getting the vessels air free. The activity in the positive experiments must come from bidirectional hydrogenase, since any uptake hydrogenase is not so sensitive toward O2. No C2H2 reduction activity was seen in the dark. The results indicate that the bidirectional hydrogenase of Chroococcidiopsis PCC 7203 can only poorly protect nitrogenase from damage by O2. Thus, the bidirectional hydrogenase may be a fossil relict together with the organism itself. In early geological times, it may have served in fermentation and may have effectively supplied reducing equivalents to nitrogenase. However, when the concentration of O2 in the atmosphere rose above 0.3 to 0.5%, bidirectional hydrogenase may have been inactivated. Then, heterocysts that could better accommodate and protect their nitrogenase had to be developed. Indeed, Chroococcidiopsis has been discussed as an ancestor of heterocyst-forming species (69).
Chroococcidiopsis sp., as isolated from the gypsum rock “Sachsenstein” near Bad Sachsa, Harz Mountains, Germany (24). This cyanobacterium is regarded as a fossil record ancestor of heterocystous cyanobacteria (69) (see the text). It now occupies ecological niches such as the fissures in gypsum, where it might be exposed to light intensities that are low but still sufficient for photosynthesis. It forms packages of 16 cells or multiples thereof (A). The gypsum shards can easily be peeled off by hand (B), and the greenish-blue layer consisting almost exclusively of Chroococcidiopsis below the shards then becomes visible (C).
POTENTIAL FOR EXPLOITING CYANOBACTERIA IN SOLAR ENERGY CONVERSION PROGRAMS FOR PRODUCTION OF COMBUSTIBLE ENERGY (HYDROGEN)
Of all organisms, cyanobacteria have the simplest nutrient requirement in nature. They thrive photoautotrophically on simple inorganic media, and many of them do not need combined nitrogen in their medium. They can be grown with a reasonably fast generation time of 2 to 3 h for unicellular forms (though not as fast as fermentative bacteria, such as E. coli, where the half-life [t1/2] can be close to 10 min). A laudable goal is to generate clean energy, without generating greenhouse gases such as CO2 or NOx, by exploiting the photosynthetically produced reductant (ferredoxin) for H2 production. To do so demands the separation of the photosynthetically produced O2 from H2 production. Research in this area started around 1973 during the first global energy crisis and has found renewed interest currently due to the concerns over global warming. Success in this area demands the continuous production of H2 over weeks or months, followed by effective utilization of the cyanobacterial cells produced. Cyanobacterial proteins are not optimal to feed to cattle but can be used as dietary supplements with various positive effects for humans and animals (77, 114). One obstacle is that neither cyanobacterial hydrogenase couples with the reduced ferredoxin generated photosynthetically. Presumably based on their own research interests, different researchers favor the use of either hydrogenase or nitrogenase in solar energy conversion programs.
A comparison of the published rates of H2 formation suffers from the fact that different laboratories refer their data to different units. As a basis for comparing the various results, the following gross estimates are made (Table 1). In all photosynthetic organisms, chlorophyll a constitutes 1 to 2% of the dry weight. Taking the average of 1.5%, the cyanobacterial dry weight can be estimated by multiplying the chlorophyll a content by a factor of 67 (http://www.chebucto.ns.ca/ccn/info/Science/SWCS/DATA/PARAMETERS/CHA/cha ). Moreover, chlorophyll a has a molecular weight of slightly less than 1,000, and 1 mg of chlorophyll corresponds to 20 to 25 mg of cell protein (20 mg is used here). In photosynthesis, the unit commonly used since the time of Willstätter and Stoll (232) is mg chlorophyll per h. The C/N ratio is around 6 in cells, and the maximal photosynthetic CO2 fixation rates are roughly 100 μmol/h·mg chlorophyll. Thus, the N2 fixation rate is unlikely to exceed 20 μmol NH4+ produced/h·mg chlorophyll. If all electrons transferred to nitrogenase were reallocated to reduce H+, H2 production by cyanobacteria would be around 40 μmol H2 produced/h·mg chlorophyll, based on the fact that four electrons are needed for NH4+ production (with concomitant H2 evolution) but only two electrons are needed for H2 formation (equations 1 and 2). The data in Table 1 also use a gas molar volume of 24 liters at 25°C.
Examples of published rates of cyanobacterial H2 formation
On the basis of the considerations described above, the few significantly higher activities reported in the literature (Table 1) seem to require reassessment. If the experiments were not done with great care, less than total chlorophyll could be released from the cyanobacterial cells, which would lead to the high activities reported. With artificial photosystem I/Pt or Au nanoparticle biconjugates, maximal H2 production activities were 49 μmol/mg chlorophyll·h (85), which are in the same range as the theoretically achievable formation with cyanobacteria.
To overcome the problem that cyanobacterial hydrogenases do not couple with ferredoxin, the clostridial, ferredoxin-dependent hydrogenase I was heterologously expressed in the unicellular cyanobacterium Synechococcus PCC 7942 (8). Cell extracts of the genetically engineered isolate showed about 3-fold-higher activity than the wild type. An alternative genetic approach was to modify the photosystem I PsaE subunit from Thermosynechoccocus elongatus so that it linked to the O2-insensitive membrane-bound hydrogenase of Ralstonia eutropha and PSI from Synechocystis sp. PCC 6803 (104). This artificial hydrogenase-PSI complex displayed light-driven H2 production, but only at low rates, and this activity was suppressed by ferredoxin and FNR (104). The latter problem was circumvented by modifying the ferredoxin-binding site of PsaE (103). There have been other attempts with limited success to express a foreign hydrogenase in cyanobacteria (8) or a cyanobacterial hydrogenase in a foreign organism (135). Approaches with cyanobacteria are based on the assumption that the membrane-bound [NiFe] hydrogenases from Ralstonia eutropha, R. metallidurans, Allochromatium vinosum, or others are more O2 tolerant than the cyanobacterial enzymes (64). Since both the bidirectional and uptake hydrogenases of cyanobacteria have never been biochemically characterized in the pure form, this assumption may not necessarily be true, particularly for the bidirectional hydrogenase. This enzyme, with its complex of five HoxEFUYH subunits, may easily fall apart upon purification, and not necessarily due to any inferred O2 lability. The current state of attempts to develop heterologous and recombinant expression of hydrogenases for improving H2 formation by organisms has been summarized and reviewed (64, 131).
In intact cyanobacterial cells, H2 produced by nitrogenase is more or less completely recycled by hydrogenase so that often almost no net H2 production is detectable. Uptake hydrogenase, but not the bidirectional enzyme, is effective in recycling the gas (139). Mutants defective in uptake hydrogenase show a much higher H2 production than wild-type cells. This was shown some years ago with mutants of Anabaena variabilis obtained by classical N-methyl-N′-nitro-N-nitrosoguanidine (NTG) mutagenesis (147) and more recently with strains that were defective in uptake hydrogenase due to site-directed mutagenesis (92, 140).
As recently published (34), Anabaena variabilis and A. azotica produce large amounts of H2 when incubated under high concentrations of H2 and C2H2 (Fig. 10 A). This H2 production, on top of the H2 added, is higher in V- than in Mo-grown cultures of A. azotica (34). The amount of H2 formed increases and C2H4 production decreases in parallel with the concentration of H2 added to the vessels (Fig. 10B). In line with these findings, a 2- to 4-fold increase of light-induced H2 production was observed in Nostoc muscorum preincubated under argon and H2 (182). Although added C2H2 is known to inhibit the uptake hydrogenase (205), this observation does not explain the effect of increasing amounts of H2. The effects of H2 and C2H2 on nitrogenase itself and/or photosynthetic electron flow to nitrogenase cannot mechanistically be explained as yet. However, the meaning of these findings is that all electrons coming to nitrogenase can be directed to produce H2, particularly in V-grown cells. The rate of ∼40 μmol H2 produced reflects the maximal photosynthetic H2-forming potential of cyanobacterial suspension cultures.
(A) H2 production by Anabaena azotica (V or Mo grown) and A. variabilis. The lower parts of the columns indicate the amount of H2 added to the vessels by syringes and determined by gas chromatography at the start of the experiments. The gas phase was 85% argon and 15% C2H2 (vol/vol). Complete means gas-phase H2 (about 1 bar). (B) Inhibition of C2H2 reduction by increasing concentrations of H2 added to the assays, using Mo-grown A. azotica. The inhibition pattern was the same for V-grown A. azotica and for Mo-grown A. variabilis (not shown). The data are from reference 34.
Such an interpretation of the data indicates that further genetic engineering of cyanobacteria, either by transferring an alien hydrogenase or nitrogenase or by genetically manipulating the acceptor side of photosystem I, is unlikely to enhance the rate of cyanobacterial H2 production. The compilation of the data in Table 1 shows that maximal H2 production in suspension cultures is already achieved by coupling either nitrogenase or hydrogenase to the cyanobacterial photosystem I. A temporal separation of the photosynthetic organic carbon formation (glycogen) in light followed by a fermentative degradation of these carbohydrates in the dark (3) is unlikely to enhance H2 production rates, although it would separate H2 and O2 production from each other. Apart from this, rates of H2 production in strict fermentative bacteria (clostridia) are at least 3 orders of magnitude higher than those in cyanobacterial fermentations. Therefore, clostridia or other fermentative bacteria with a much more efficient [Fe-Fe] hydrogenase could possibly be coupled and exploited to degrade the cyanobacterial photosynthetically produced organic carbon for maximal H2 production.
The transfer of a hydrogenase which is insensitive to exposure to O2, either produced by genetic modification or taken from an alien organism, may facilitate but may not be obligatory for commercially acceptable rates of cyanobacterial H2 production. Genetic alterations of amino acids in the gas-substrate channel of hydrogenases changes their intramolecular gas transport kinetics (121). Substitutions of two amino acids at the end of the channel (valine and leucine, both with methionine) make [NiFe] hydrogenase O2 tolerant, as shown for the enzyme from Desulfovibrio fructosovorans (50). Similar genetic engineering of an [Fe-Fe] hydrogenase could be rewarding, since such an enzyme heterologously transferred to cyanobacteria could couple directly with ferredoxin and the photosynthetically generated reducing power while being insensitive to the photosynthetically produced O2. However, as pointed out previously (64), heterologous expression of any such genetically modified hydrogenase in a cyanobacterium also requires transcription of host-specific response regulators, and, as outlined above, transcription factors likely show a degree of specificity for cyanobacteria, as is evidenced for LexA- and AbrB-like proteins of bidirectional hydrogenase (159) (see above).
A realistic chance of improving H2 production by using either nitrogenase or hydrogenase lies in optimizing the photosynthetic electron flow for the generation of reductants, as outlined by the late David Hall and coworkers (90) some years ago. The light energy conversion efficiencies for H2 production in suspension cultures are only ca. 1 to 2% and thus very low (136). However, these values refer to the radiant energy incident on the cells rather than the energy absorbed, which is difficult to determine. These efficiencies can hardly be improved in dense cyanobacterial suspension cultures with self-shadowing effects. However, immobilization of cyanobacteria by adsorption on solid matrices or by entrapment in gels or polymers may enhance the functional lifetime of cells and may also increase the number of heterocysts in filamentous cyanobacteria. Indeed, immobilized cells were reported to show sustained high rates of H2 production (90, 134, 175) (Table 1). The light energy conversion efficiencies for H2 production may also be higher in immobilized cells than in suspension cultures. In addition, such an approach may enhance the lifetime of the cyanobacterial cells and thus may result in longer-lasting H2 production (90).
Sulfur deprivation leads to inactivation of photosystem II activity, resulting in anaerobiosis in the cultures and subsequently enhanced H2 production, as shown first for the green alga Chlamydomonas reinhardtii (145) and subsequently for cyanobacteria (4, 241). Cyanobacterial H2 production may also be augmented by altering the PSII/PSI ratio and by reducing the content of phycobilisome antennae in the cells (16). Both cyanobacterial hydrogenases are Ni enzymes. Cyanobacterial H2 production could also be altered by the supply of Ni to the cells (2, 10, 164, 169). Limitations could prevent synthesis of uptake hydrogenase, resulting in higher net H2 production from nitrogenases in the cells. Excess Ni could favor bidirectional hydrogenase synthesis and H2 production by this enzyme. In addition, culture conditions can be optimized for maximal cyanobacterial H2 production (40, 41).
Activity may also be increased by artificially enhancing the number of heterocysts within filaments and thus nitrogenase concentrations, e.g., by use of chemicals such as 7-azatryptophan (30) or by site-directed mutagenesis (144, 123). A high number of 600 to 1,000 genes are estimated to be specifically expressed in recently differentiated heterocysts (42, 133). The master gene controlling the expression of heterocysts is hetR, and their suppression is regulated by the patS and hetN gene products (38, 42, 236). Overexpression of the hetR gene leads to an enhancement of heterocyst frequency up to 29% in Anabaena (Nostoc) PCC 7120, but the remaining vegetative cells cannot perform CO2 fixation fast enough to meet the demand of the filaments for organic carbon and reductants (38). Research over the next several years following up in such directions will reveal whether cyanobacteria can ever be exploited for the realistic generation of new energies.
ACKNOWLEDGMENTS
We are indebted to Gudrun Boison (Mariefred, Sweden) for helpful discussions and to Stefanie Junkermann (University of Cologne) for expert technical assistance with some of the experiments.
- Copyright © 2010 American Society for Microbiology
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.↵
- 115.↵
- 116.↵
- 117.↵
- 118.↵
- 119.↵
- 120.↵
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.↵
- 132.↵
- 133.↵
- 134.↵
- 135.↵
- 136.↵
- 137.
- 138.↵
- 139.↵
- 140.↵
- 141.↵
- 142.↵
- 143.↵
- 144.↵
- 145.↵
- 146.↵
- 147.↵
- 148.↵
- 149.↵
- 150.↵
- 151.↵
- 152.↵
- 153.↵
- 154.↵
- 155.↵
- 156.↵
- 157.↵
- 158.↵
- 159.↵
- 160.↵
- 161.↵
- 162.↵
- 163.↵
- 164.↵
- 165.↵
- 166.↵
- 167.↵
- 168.↵
- 169.↵
- 170.↵
- 171.↵
- 172.↵
- 173.↵
- 174.↵
- 175.↵
- 176.↵
- 177.↵
- 178.↵
- 179.↵
- 180.↵
- 181.↵
- 182.↵
- 183.↵
- 184.↵
- 185.↵
- 186.↵
- 187.↵
- 188.↵
- 189.↵
- 190.↵
- 191.↵
- 192.↵
- 193.↵
- 194.↵
- 195.↵
- 196.↵
- 197.↵
- 198.↵
- 199.↵
- 200.↵
- 201.↵
- 202.↵
- 203.↵
- 204.↵
- 205.↵
- 206.↵
- 207.↵
- 208.↵
- 209.↵
- 210.↵
- 211.↵
- 212.↵
- 213.↵
- 214.↵
- 215.↵
- 216.↵
- 217.↵
- 218.↵
- 219.↵
- 220.↵
- 221.↵
- 222.↵
- 223.↵
- 224.↵
- 225.↵
- 226.↵
- 227.↵
- 228.↵
- 229.↵
- 230.↵
- 231.↵
- 232.↵
- 233.↵
- 234.↵
- 235.↵
- 236.↵
- 237.↵
- 238.↵
- 239.↵
- 240.↵
- 241.↵
- 242.↵
Author Bios
Hermann Bothe received his Ph.D. from Göttingen University and his habilitation from Bochum University. He was Professor of botany and microbiology at the University of Cologne, Germany, from 1978 and is now retired. As a student of A. Trebst, Göttingen/Bochum, Germany, he started to work on photosynthetic electron transport before he switched to nitrogen fixation, both in cyanobacteria. He also studied aspects of nitrogen fixation by associative bacteria, denitrification, arbuscular mycorrhiza, and heavy metal and salt resistance in plants. He has almost 200 publications in refereed journals.
Oliver Schmitz studied biology in Cologne, Germany, with his main focus on botany, genetics, and biochemistry, and completed his diploma thesis on arbuscular mycorrhiza in 1991. In the course of his dissertation in the laboratory of Professor Bothe, he specialized in hydrogen metabolism in cyanobacteria and obtained his Ph.D. in 1995 by characterizing the bidirectional hydrogenase in unicellular and in N2-fixing cyanobacteria by means of protein purification and applying molecular biology, resulting in the first identification of cyanobacterial hydrogenase genes at that time. He worked as postdoctoral fellow in Susan Golden's group at Texas A&M University, performing research on photosynthesis and the circadian clock in cyanobacteria. In 2001, he joined Metanomics GmbH, a BASF Plant Science company specialized in applying metabolomics in the fields of plant biotechnology, pharmacology, diagnostics, and toxicology. Currently, he is member of the management team and head of the Data Interpretation Health group at Metanomics.
M. Geoffrey Yates received his B.Sc. from the University College of North Wales, Bangor, United Kingdom, and his Ph.D. from the University of Nottingham, United Kingdom, and then was Research Associate at Unilever Research Colworth House, Bedford, United Kingdom, at the Biochemistry Department of John Hopkins University, Baltimore, MD, and then at the Department of Biochemistry of Oxford University. For almost 30 years, he was Principal Scientific Officer at the BBSRC Unit of Nitrogen Fixation, University of Sussex, United Kingdom. For the last 15 years, he was Visiting Research Fellow at the Department of Biochemistry and Molecular Biology, Federal University of Paranã, Curita, Brazil. In recent years he worked on nitrogen fixation and hydrogen uptake in Azotobacter chroococcum, Azospirillum brasilense, and Herbaspirillum seropedicae.
William E. Newton received his B.Sc. from the Nottingham University and his Ph.D from London University (both in the United Kingdom), and he then spent a postdoctoral year at Harvard before spending 15 years at the Charles F. Kettering Research Laboratory in Yellow Springs, OH, as a member of its nitrogen fixation group. He then became Research Leader for Plant Productivity at the Western Regional Research Center (USDA-ARS) in Berkeley, CA, where he was awarded the USDA Certificate of Merit. He also served as Adjunct Professor at UC-Davis. In 1990, he moved to Virginia Polytechnic Institute and State University (Virginia Tech) as Director of the Biotechnology Center and Professor of Biochemistry. He later served as head of both the Biochemistry Department and the Department of Anaerobic Microbiology. He was elected Fellow of the Royal Society of Chemistry in 1992 and Fellow of the American Association for the Advancement of Science in 1996.