Respiratory protection of nitrogenase in Azotobacter species: is a widely held hypothesis unequivocally supported by experimental evidence?

Transcription

1 FEMS Microbiology Reviews 24 (2000) 321^333 Abstract Respiratory protection of nitrogenase in Azotobacter species: is a widely held hypothesis unequivocally supported by experimental evidence? Ju«rgen Oelze * Universita«t Freiburg, Institut fu«r Biologie II (Mikrobiologie), Scha«nzlestr. 1, D Freiburg, Germany Received 14 November 1999; received in revised form 22 February 2000; accepted 23 February 2000 The hypothesis of respiratory protection, originally formulated on the basis of results obtained with Azotobacter species, postulates that consumption of O 2 at the surface of diazotrophic prokaryotes protects nitrogenase from inactivation by O 2. Accordingly, it is assumed that, at increased ambient O 2 concentrations, nitrogenase activity depends on increased activities of a largely uncoupled respiratory electron transport system. The present review compiles evidence indicating that cellular O 2 consumption as well as both the activity and the formation of the respiratory system of Azotobacter vinelandii are controlled by the C/N ratio, that is to say the ratio at which the organism consumes the substrate (i.e. the source of carbon, reducing equivalents and ATP) per source of compound nitrogen. The maximal respiratory capacity which can be attained at increased C/N ratios, however, is controlled, within limits, by the ambient O 2 concentration. When growth becomes N-limited at increased C/N ratios, cells synthesize nitrogenase and fix N 2. Under these diazotrophic conditions, cellular O 2 consumption remains constant at a level controlled by the O 2 concentration. Control by O 2 has been studied on the basis of both whole cell respiration and defined segments of the respiratory electron transport chain. The results demonstrate that the effect of O 2 on the respiratory system is restricted to the lower range of O 2 concentrations up to about 70 WM. Nevertheless, azotobacters are able to grow diazotrophically at dissolved O 2 concentrations of up to about 230 WM indicating that respiratory protection is not warranted at increased ambient O 2 concentrations. This conclusion is supported and extended by a number of results largely excluding an obvious relationship between nitrogenase activity and the actual rate of cellular O 2 consumption. On the basis of theoretical calculations, it is assumed that the rate of O 2 diffusion into the cells is not significantly affected by respiration. All of these results lead to the conclusion that, in the protection of nitrogenase from O 2 damage, O 2 consumption at the cell surface is less effective than generally assumed. It is proposed that alternative factors like the supply of ATP and reducing equivalents are more important. ß 2000 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. Keywords: Azotobacter; Nitrogen xation; Nitrogenase; Respiration; Respiratory protection; Energy regeneration; N-status Contents 1. Introduction The respiratory chain of Azotobacter vinelandii Regulation of cellular O 2 consumption C/N ratio and ambient O Kind of substrate E ect of dilution rate in chemostat cultures Phosphate limitation Regulation of the composition of the respiratory electron transport system by O Control of nitrogen xation Control of nitrogenase derepression by the C/N ratio Nitrogenase activity Nitrogenase polypeptides * Tel.: +49 (761) ; Fax: +49 (761) ; oelze@uni-freiburg.de / 00 / $20.00 ß 2000 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. PII: S (00)

2 322 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^ Respiratory protection: the pros and cons Does the membrane-bound respiratory electron transport system prevent O 2 from entering the cells? Alternative strategies of protection Conclusions Acknowledgements References Introduction Biological xation of dinitrogen (N 2 ) is catalyzed by the nitrogenase enzyme complex, which is formed only by prokaryotes [1]. Nitrogenase is composed of two subunits, component 2 which reduces component 1. Component 1 catalyzes the reduction of N 2 to two NH 3 accompanied by the evolution of one H 2. Reduction of component 1 requires not only reducing equivalents but also the supply of at least 16 ATP per N 2 xed. On the basis of their metal centers, di erent types of nitrogenase complexes have been described [1,2]. Under normal growth conditions, the organisms form the so-called classical or conventional nitrogenase containing iron^molybdenum cofactors (FeMoco); while under conditions of Mo de ciency, either one of two alternative nitrogenases may be synthesized. In vitro, nitrogenases are sensitive to O 2 which irreversibly inactivates the enzyme within a few minutes [3,4]. This means that, when growing aerobically at the expense of biologically xed N 2, i.e. diazotrophically, organisms must employ mechanisms which, on the one hand, permit the supply of O 2 required for energy regeneration in the course of aerobic respiration and, on the other hand, protect nitrogenase from the deleterious e ect of O 2. These problems have been discussed extensively [5,6]. Interestingly, a new type of Mo-containing nitrogenase has recently been detected in Streptomyces thermoautotrophicus, which di ers from the three above-mentioned nitrogenases in its composition and its complete insensitivity toward O 2 [7]. Members of the azotobacters represent aerobic bacteria which are able to grow diazotrophically at the highest dissolved O 2 concentrations (about 230 WM O 2 ) attainable by aeration of the cultures [1,8,9]. Consequently, azotobacters have been the preferred organisms in studies on the protection of nitrogenase from inactivation by O 2. The high rate of O 2 consumption observed in Azotobacter chroococcum and its increase upon exposure to an increased ambient O 2 concentration gave rise to the formulation of the hypothesis of `respiratory protection' of nitrogenase [10]. According to this hypothesis, the respiratory electron transport system bound to the peripheral cytoplasmic membrane performs an O 2 -scavenging function preventing the di usion of O 2 into the cells. It is postulated that this mechanism keeps the interior of the cells anoxic even with high ambient O 2 concentrations. This means that the function of nitrogenase is actively protected. Whenever respiratory protection becomes overburdened by a sudden increase in the prevailing O 2 concentration, nitrogenase of Azotobacter is transformed into an inactive but O 2 -tolerant complex by association with a protecting [2Fe^2S] protein [1]. The latter passive protection of nitrogenase has been termed `conformational protection' [10]. During the past 30 years, more detailed information has become available on the mechanism and the physiological signi cance of conformational protection of nitrogenase [11^13]. But as yet, there is no direct experimental evidence proving that the cytoplasm is kept anoxic by cellular respiration. On the contrary, several observations have been made which, at present, cannot be interpreted in favor of respiratory protection. Nevertheless, this hypothesis has readily been adopted to explain the function of nitrogenases in other aerobic representatives of the diazotrophic bacteria [14]. Moreover, by its general inclusion in textbooks on the biology of microorganisms, the hypothesis of respiratory protection has become an accepted dogma. In the following, experimental data are compiled which, it is hoped, will stimulate a more critical view of the most widely held hypothesis on the protection of the conventional MoFe-nitrogenase from reversible inactivation by O The respiratory chain of Azotobacter vinelandii Biological xation of N 2 requires a fairly high proportion of the cellular ATP pool. In aerobic diazotrophs, ATP is regenerated in the course of respiratory electron transport reactions with O 2 as terminal electron acceptor. Yet, the exceptionally high rate of O 2 consumption characteristic of Azotobacter species growing at increased O 2 concentrations should far exceed their energy requirements. This problem does not occur because, as a response to changes in the supply of O 2, azotobacters are able to change the composition of the respiratory chain and the coupling of electron transport to ATP regeneration. In light of the hypothesis of respiratory protection of nitrogenase, uncoupling of the respiratory chain is assumed to be of primary importance in keeping the intracellular O 2

3 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^ concentration at a non-toxic level. Accordingly, studies on the mechanism of respiratory protection concentrated largely on analyses of the composition, the coupling and regulation of the respiratory chain [1,14^16]. The information on the respiratory chain relevant to the present review is summarized in Fig. 1. The initial part of the respiratory chain contains avin-dependent NADH, NADPH, succinate and malate dehydrogenases, which donate electrons to a pool of ubiquinone-8. On the assumption that NADH is the main respiratory electron donor in bacteria, Bertsova and colleagues [17] characterized membrane-bound NADH:ubiquinone oxidoreductase of A. vinelandii. Two NADH dehydrogenases (NDH I and NDH II) were identi ed catalyzing the NADH:ubiquinone oxidoreductase reaction [17]. From ubiquinol, electrons may ow into di erent branches with di erent terminal oxidases. Under high aeration, the major ow of electrons to O 2 is assumed to be via a pathway terminated by a cytochrome bd-type oxidase. This latter oxidase exhibits a high V max but a relatively low a nity for O 2 (K m = 4.5 WM). In addition, A. vinelandii exhibits the presence of two high-a nity cytochrome o-like oxidases with apparent K m values for O 2 of approximately 0.02 and 0.3 WM, respectively [18,19]. Until recently it was believed that formation of the proton motive force is associated with NADH-dehydrogenase (site I), the ubiquinone section (site II), and the high-af- nity cytochrome o-type oxidase (site III). Moreover, it was assumed that cytochrome bd-oxidase is uncoupled and functions as an `O 2 scavenger' only [14]. This view required revision when it was reported that one of the two NADH dehydrogenases (NDH II) is uncoupled whereas cytochrome bd oxidase is at least partially coupled [17,20,21]. According to calculations by Bertsova et al. [17], A. vinelandii possesses two electron transport chains, one of which has an H /e ratio of 5 comprising NDH I (H /e ratio of 2) and the ubiquinol-cytochrome o-type oxidase segment (H /e ratio of 3). The other chain comprises the uncoupled NDH II and a partially coupled cytochrome bd-type oxidase with an H /e ratio of Regulation of cellular O 2 consumption 3.1. C/N ratio and ambient O 2 Fig. 1. Diagram of the electron transport chain of A. vinelandii with special reference to the dehydrogenases, terminal oxidases and, as far as is known, to H /e 3 ratios. The diagram compiles published data [14,17^20]. DH, dehydrogenase; Q 8, ubiquinone 8; Cyt, cytochrome; cydab, cyca and cycb genes encoding cyt bd, cyt c 4 and cyt c 5, respectively; K m, apparent a nity of the terminal oxidase for O 2. In heterotrophic organisms like azotobacters, the C/N ratio de nes the ratio at which the source of organic carbon and reducing equivalents is consumed per source of compound N. While the latter is completely assimilated into biomass, part of the carbon substrate has to be dissimilated in order to provide reducing equivalents as well as energy for substrate assimilation. Under oxic conditions, dissimilation and energy regeneration are coupled to the respiratory electron transport chain with O 2 as the terminal electron acceptor. This suggests that O 2 consumption by the cells, i.e. cellular respiration, may depend on the supply of the carbon substrate. On the basis of this assumption, the dependence of cellular respiration on the C/N ratio was studied with cultures of A. vinelandii growing in ammonium-limited chemostat cultures at di erent de ned ambient O 2 concentrations [22]. Continuous chemostat systems are operated by supplying the cultures with fresh growth medium at a constant rate, i.e. the dilution rate. Since the growth medium contains the growthlimiting nutrient, the dilution rate determines the growth rate of the population. As soon as a continuous culture has reached its steady state, the organisms can reproducibly be kept under de ned growth conditions and, thus, in de ned physiological states. This is in contrast to the more Fig. 2. Dependence of the cellular O 2 consumption and nitrogenase activity on the C/N ratio (molar ratio of sucrose consumed per NH 4 )in A. vinelandii growing at 68 WM ambient O 2 and a dilution rate of 0.15 h 31 in continuous culture under steady-state conditions [22,23]. O 2 consumption by cells as assayed with samples removed from the culture system (a), in situ activities of cellular respiration (b) and acetylene reduction by nitrogenase (F).

4 324 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^333 commonly employed closed-batch culture system where, as a result of the activity of the organisms, the growth conditions are steadily changing. Studies of the dependence of cellular respiration of A. vinelandii on the C/N ratio revealed that the rate of cellular O 2 consumption increases signi cantly when the C/N ratio is increased (Fig. 2). At higher C/N ratios, O 2 consumption approaches constancy and, at essentially the same time, nitrogenase becomes detectable (Fig. 2; for further details, see Section 5.1) [22,23]. The size of the C/N ratio and the activity at which cellular respiration becomes constant, however, depend on the ambient O 2 concentration. These results have been interpreted as follows: since growth depends on the supply of ammonium, aerobically growing cells assimilate as much of the carbon and N sources as required for the production of biomass exhibiting a constant C/N ratio (Fig. 3A). When the supply and, thus, the consumption of the carbon source is increased, the surplus amount of the carbon source not required for biomass production may be dissimilated and, consequently, the activity of cellular respiration rises. The increase in cellular respiration has been observed with the actual rate of O 2 consumption as determined in situ as well as with the maximal rate as determined with cell samples removed from the cultures. This suggests that formation of the respiratory electron transport chain is controlled by the C/N ratio [22]. However, superimposed on the e ect of the C/N ratio is an e ect of ambient O 2, which arrests the activity of cellular respiration at a de- ned maximal level (Fig. 3A). When the ambient O 2 concentration is increased up to about 70 WM, the steady-state rate of respiration increases in direct proportion to increases in ambient O 2. Upon further increases up to about 230 WM O 2, i.e. air saturation, cellular respiration increases only slightly, if at all [9]. On the other hand, when the O 2 supply is very low, the surplus of the carbon source may be consumed and stored as poly-l-hydroxybutyrate rather than be subjected to dissimilation [24]. All in all, these mechanisms are proposed to enable the cells to keep the internal C/N ratio at a level ful lling the requirements of biomass formation. The C/N ratio is generally known to be involved in the control of various physiological parameters like cellular respiration, the formation of biomass and storage material or repression and derepression of nitrogenase. However, in most of these cases, quantitative data on the relationship between the C/N ratio and such physiological activities are lacking. Recently, the internal C/N ratio required for biomass formation has been de ned by the ratio at which the carbon source, lactate, is consumed per ammonium as sole source of N [25]. The experiments were performed with the photoheterotrophic Rhodobacter capsulatus, which, under anaerobic conditions, uses the organic substrate as a source of carbon and reducing equivalents but, in contrast to aerobic azotobacters, not as a source of energy. This means that, in R. capsulatus, the e ciency of substrate Fig. 3. Model for the steady-state control of cellular respiration, N 2 xation and biomass production in A. vinelandii by the ambient O 2 concentration and the nitrogen status. The N-status is inversely proportional to the ratio (C/N ratio) at which the sources of both carbon and compound nitrogen are consumed by the organisms. A: Under conditions of NH 4 assimilation (dotted arrow), a constant proportion of the substrate (i.e. the source of carbon, reducing equivalents and ATP) is assimilated per NH 4 into biomass (dotted). With increasing substrate supply, substrate uptake and, thus, the internal C/N ratio increase. The surplus amount of substrate not required for biomass production is subject to dissimilation coupled to cellular respiration, i.e. O 2 consumption. Therefore, it is suggested that the C/N ratio is involved in the control ( ) of substrate dissimilation including the respiratory system. Superimposed on this, however, there is a higher level of control by O 2 ( ). Di erent levels of cellular respiration are shown (a), which are maximally attained at representative O 2 concentrations ( through ). At the lower O 2 concentrations ( through ), substrate dissimilation and cellular respiration increase essentially in direct proportion to increases in the ambient O 2 concentration. At moderately increased O 2 concentrations ( to ) respiration approaches constancy ( to ). B: Limitation of respiration by the prevailing O 2 concentration means that any further increase in substrate consumption augments the internal C/N ratio lowering the N-status of the organisms below the value required for biomass production. This provides the signal ( ) for nitrogenase (N 2 ase) derepression. With the aid of N 2 ase, cells become able to x N 2 (in dark gray) and to readjust the N-status to the level required for biomass production. In conclusion, when substrate consumption increases above the C/N ratio which, at a given O 2 concentration, controls respiration, respiration becomes constant while N 2 ase is derepressed and diazotrophic biomass production (in dark gray) increases in proportion to increased substrate consumption.

5 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^ assimilation into biomass is not obscured by increased dissimilation and respiration. Calculation of the relative amounts of lactate and of ammonium required for biomass formation were based on an elemental composition of biomass of CH 1:83 N 0:183 O 0:5. Accordingly, the following equation describes the requirements of the assimilation of 1 mol of lactate in combination with ammonium into biomass: C 3 H 6 O 3 0:55 NH 3 0:426 NADH 2! C 3 H 5:5 N 0:55 O 1:5 1:5 H 2 O 0:426 NAD mol of NADH 2, which is required because lactate is more oxidized than biomass, is provided by the dissimilation of mol of lactate according to: C 3 H 6 O 3 3H 2 O 6 NAD! 3CO 2 6 NADH 2 Thus, biomass formation from lactate and ammonium is de ned by the following equation: 1:071 C 3 H 6 O 3 0:55 NH 3! C 3 H 5:5 N 0:55 O 1:5 1:3 H 2 O 0:213 CO 2 The corresponding atomic C/N ratio amounts to 5.8. From the above calculations it follows that the C/N ratio should increase when the substrate is more oxidized than lactate. Accordingly, the calculated C/N ratio for biomass formation from malate and ammonium is 7.8. Both of the calculated C/N ratios predict that cultures are becoming N-su cient at lower ratios and N-limited at higher ratios. In other words, the internal C/N ratio is inversely proportional to the N-status of the cells. Under conditions of N-su ciency, the cultures may be expected to consume only the amount of compound N needed for biomass formation. On the other hand, N-limitation can be prevented and constancy of the C/N ratio required for biomass production is warranted when diazotrophic organisms become able to x N 2. The experimental results obtained on the occurrence of both N-limitation and nitrogenase derepression in R. capsulatus perfectly agree with the above predictions [25] 3.2. Kind of substrate In studies on the activity of cellular respiration and on the composition of the respiratory system of azotobacters, sucrose and glucose are commonly used as carbon sources and electron donors. Both substrates, however, are exceptional in that maximal growth rates (W max ) obtained with sucrose and glucose are 0.34 h 31, while W max values obtained with other substrates like citrate and acetate range from 0.45 to 0.5 h 31 [23,26]. Moreover, at identical dissolved O 2 concentrations, sucrose and glucose give rise to cellular rates of O 2 consumption which are up to two times higher than those measurable with galactose, citrate or acetate [26,27]. Finally, yields of cell protein calculated per carbon of galactose, citrate or acetate are signi cantly higher than those obtained with glucose and sucrose. This is supported by the fact that cultures growing with acetate or citrate exhibit signi cantly lower maintenance requirements than those growing with glucose [26]. All of these di erences are supported by the occurrence of diauxic growth behavior, when A. vinelandii is supplied with mixtures of glucose and either galactose or acetate, where the latter two are the preferred substrates [28,29]. The higher rates of glucose utilization and the higher respiratory rates resulting from dissimilation of glucose prompted Liu et al. [27] to develop an interesting hypothesis on the e ects of di erent substrates on cellular respiration. According to this hypothesis, the very e cient utilization of high concentrations of glucose used in laboratory cultures might lead to the accumulation of NADH and ATP and, thus, to an inhibition of the oxidative metabolism. It has been proposed that A. vinelandii may avoid this problem by regenerating NAD in the course of an energetically largely uncoupled respiratory chain. Thus, it is assumed that high rates of O 2 consumption on glucose contribute to the protection of the organisms from over-reduction E ect of dilution rate in chemostat cultures Experiments employing substrate-limited diazotrophic chemostat cultures of A. vinelandii showed that, at given concentrations of both O 2 and substrate, the steady-state rate of cellular O 2 consumption depends not only on the type of substrate but also on the dilution rate, i.e. the rate at which the cultures are supplied with their substrate. In particular, in situ rates of cellular O 2 consumption increase as the dilution rate is increased [26,30,38] (Fig. 4). This e ect of the dilution rate under conditions of substrate limitation is supported by the fact that, under conditions of substrate saturation but sulfate limitation, the activity of cellular respiration remains constant, i.e. independent of the dilution rate (J. Kuhla and J. Oelze, unpublished ndings) Phosphate limitation Phosphate-limited diazotrophic cultures of A. chroococcum were reported to be hypersensitive to O 2 [10]. This was interpreted to result from down-regulation of cellular respiration and, therefore, from insu cient respiratory protection. Direct measurements of rates of O 2 utilization by phosphate-limited chemostat cultures of A. vinelandii, however, did not support those interpretations. On the contrary, phosphate-limited cultures exhibited up to three times higher rates of O 2 consumption than phosphate-suf- cient control cultures. This means that the sensitivity of phosphate-limited cultures to O 2 cannot be explained by insu cient respiration and, further, that the response of phosphate-limited cultures cannot be interpreted in favor of the hypothesis of respiratory protection.

6 326 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^333 Fig. 4. Comparison of in situ activities of cellular O 2 consumption (A) and nitrogenase (B) of A. vinelandii growing at di erent dilution rates (D [h 31 ]) and di erent limitations under steady-state conditions in chemostat cultures [38]. Sulfate-limited cultures (P) (Kuhla and Oelze, unpublished observations) were supplied with 1.5% sucrose and growthlimiting amounts of g FeSO 4 per liter. Substrate-limited cultures were supplied with 0.3% (w/v) of citrate (E) and sucrose (U). The ambient oxygen concentration was adjusted to 108 WM O 2 and, in the case of sucrose, in addition to 12 (a) and 192 WM O 2 (b). 4. Regulation of the composition of the respiratory electron transport system by O 2 As mentioned above, the activity of O 2 consumption by A. vinelandii increases signi cantly when the ambient O 2 concentration is raised up to about 70 WM while it remains largely constant at higher O 2 concentrations [9]. Essentially the same dependence on ambient O 2 has been observed with respect to the activity of membrane-bound NADH oxidase (i.e. the complete respiratory chain with NADH as electron donor and O 2 as electron acceptor) of cells adapted to di erent O 2 concentrations [31]. Concomitantly, the activities of NADPH and malate oxidases increase as well. The latter changes, however, are less pronounced than those observed for NADH oxidase. According to more detailed analyses of O 2 -dependent changes in the composition of the electron transport system of azotobacters, O 2 positively regulates the cellular level of cytochrome d [14,16]. Control of the terminal oxidases by O 2 has recently been reinvestigated in ammonium-grown A. vinelandii [32]. When the rate of O 2 supply was raised from 3 to 90 mmol O 2 l 31 h 31, the greatest increase in the cellular cytochrome d content occurred up to 20 mmol O 2 l 31 h 31. This result is supported by corresponding results on the O 2 -dependent control of the expression of cydab genes encoding the cytochrome bd complex. In particular, the study revealed that expression of the cydb gene becomes constant when the O 2 supply is increased above 30 mmol O 2 l 31 h 31 [32]. Interestingly, a cytochrome bd-de cient mutant of A. vinelandii has been shown to exhibit the same pattern of control for a hemoprotein featuring spectral characteristics of cytochrome o of the high-a nity terminal oxidase. Unfortunately, it is not possible to estimate the ambient O 2 concentration from the supply of O 2. Therefore, it is not possible to compare results obtained at di erent ambient O 2 concentrations with those obtained at di erent O 2 supplies. Nevertheless, the above-mentioned data demonstrate that control of terminal oxidases is restricted to lower O 2 supplies. After reduction of ubiquinone, electrons may ow in parallel either via cytochrome c 5 or via cytochrome c 4 to cytochrome o-type oxidases [19] (Fig. 1). Transcription of the cyca and cycb genes encoding cytochromes c 4 and c 5, respectively, is up-regulated under conditions of N-starvation [33]. But O 2 is apparently of little e ect on the promoter activities of both genes. Considerable di erences in the regulation of two NADH dehydrogenases (NDH I and NDH II) were observed after growth of A. vinelandii at di erent dissolved O 2 concentrations in the presence of ammonium [17]. As the O 2 concentration was increased from about 1 to 70 WM, the activity of the coupled NDH I decreased by a factor of three, while the activity of the uncoupled NDH II exhibited a four-fold increase. Correspondingly, the total NADH oxidase activity approximately doubled. In cultures grown in the absence of added ammonium, the effects of O 2 on the activities of NDH I and NDH II were similar to those observed in cultures supplied with ammonium. However, in cultures grown in the absence of ammonium, the activity of NDH II started at a higher level than the activity of NDH I. In conclusion, the results obtained on the regulation of the branched electron transport pathway of A. vinelandii largely agree with the regulation of cellular O 2 consumption by the C/N ratio and O 2 (for details, see Section 3.1). This means that, in a range from about 70 to 230 WM O 2, neither the cellular capacity of O 2 consumption nor de- ned segments of the respiratory chain are signi cantly controlled by O 2. The increase in cellular respiration at increased O 2 concentrations has been interpreted by Peterson [34] as re ecting the presence of an `O 2 sensing mechanism' in A. vinelandii. This interpretation was substantiated by Wu et al. [35] who identi ed a gene, cydr, encoding the CydR pro-

7 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^ tein in A. vinelandii. The deduced amino acid sequence of CydR suggests that it belongs to the FNR class of transcriptional factors. In Escherichia coli, FNR controls the expression of genes required for the anaerobic metabolism in response to the availability of O 2 [36]. It has been suggested that, in A. vinelandii, CydR is a negative regulator of the expression of cydab encoding cytochrome bd [14]. 5. Control of nitrogen xation 5.1. Control of nitrogenase derepression by the C/N ratio When sucrose-limited diazotrophic cultures of A. vinelandii are grown at di erent O 2 concentrations, the steadystate level of cell protein, i.e. biomass, as well as the growth yield coe cient decrease with increasing the ambient O 2 concentration up to about 70 WM O 2 [9]. At higher O 2 concentrations both the steady-state biomass level and the growth yield coe cient approach constancy. Since, in the absence of a source of compound N, cell protein represents the net amount of N 2 xed by the organisms [23], it is commonly believed that the less e cient biomass formation at increased O 2 concentrations results from the O 2 sensitivity of the activity and synthesis of nitrogenase. However, continuous cultures growing under steady state either at the expense of added ammonium or at the expense of N 2 xation have been shown to exhibit essentially identical dependences of biomass formation and of growth yield on ambient O 2 [9,22]. Therefore, the lower e ciencies of biomass formation at higher O 2 concentrations should not be considered typical of N 2 - xing cultures. On the contrary, experimental evidence suggests that the adverse e ect of O 2 on biomass yields (amount of biomass formed per unit of substrate consumed) is characteristic of ammonium-assimilating cultures [22]. This follows from the above-mentioned control of cellular respiration, at a given O 2 concentration, by the C/N ratio (Fig. 3A). This situation, however, changes entirely as soon as ammonium-assimilating cells take up more of the carbon substrate than can be dissimilated at a given ambient O 2 concentration. The corresponding increase in the internal C/N ratio and decrease in the N-status provides the signal for nitrogenase derepression (Fig. 3B). As soon as nitrogenase activity occurs, further decreases in the relative supply of ammonium (by increasing the substrate supply) are compensated for by N 2 xation and the internal N- status is increased to the level required for biomass formation. Consequently, biomass formation increases with increasing C/N ratio, while cellular O 2 consumption remains constant at a level controlled by the O 2 concentration. The hypothesis detailed above on the control of cellular respiration and N 2 xation by the C/N ratio has been derived from studies with continuous cultures supplied with ammonium. Thus, the question arises whether this hypothesis also applies to cultures growing strictly diazotrophically. Obviously, under diazotrophic conditions, the cultures start at an extremely low internal N-status, leading immediately to the derepression of nitrogenase. As in the case of ammonium-assimilating cultures, respiration of strictly diazotrophic cultures depends on the availability of the respective substrate and, within the limits mentioned above, on ambient O 2 [9]. Under these conditions, respiration and yield of biomass remain constant at the highest and lowest levels, respectively, reached by ammonium-assimilating cultures. In other words, the e ciency of net xation of N 2 by diazotrophic cultures of A. vinelandii is essentially a consequence of the e ciency of biomass formation by cultures growing at the expense of ammonium as the sole N source at the highest possible C/N ratio which does not yet provide the signal for nitrogenase derepression Nitrogenase activity According to common practice, the cellular activity of nitrogenase is determined by the acetylene reduction assay with cell samples removed from the culture vessel. Since this assay is routinely performed in the absence of O 2 and at saturating concentrations of both an electron donor and an ATP-regenerating system, it measures the maximal activity of nitrogenase. This, however, does not necessarily re ect the actual activity in situ. To avoid this problem, Drozd and Postgate [37] measured acetylene reduction in situ by whole chemostat systems of A. chroococcum. In cultures adapted at the same dilution rate (i.e. the same growth rate) to di erent O 2 concentrations, the rates of acetylene reduction were largely identical. Subsequent studies employing chemostat cultures of A. vinelandii con- rmed and extended these results by demonstrating that, under steady-state conditions, the in situ activity of nitrogenase increases in direct proportion to increases in the dilution rate (Fig. 4). The signi cance of the data obtained by the acetylene reduction assay was con rmed after measuring in situ N 2 xation [23,38]. Under all of the conditions tested, the molar ratio of N 2 xed (plus H 2 evolved) per acetylene reduced remained constant at 1:3.8, which agrees with the theoretical ratio of 1:4 [23,38]. Importantly, when increasing the dilution rate, the level and the increment of the in situ activity of nitrogenase has been shown to be independent not only of the ambient O 2 concentration but also of the carbon source/electron donor (i.e. sucrose, glucose, acetate, and citrate) and of the type of limitation (i.e. substrate, phosphate and sulfate limitations) [30,38,39]. In contrast to the in situ activity of nitrogenase, the corresponding activity of cellular O 2 consumption is signi cantly in uenced by all of the abovementioned growth factors (Fig. 4). The activity of nitrogenase is reversibly switched o when diazotrophically growing cultures of A. chroococcum and A. vinelandii are exposed to an O 2 stress by suddenly

8 328 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^333 increasing the prevailing O 2 concentration [1]. Nitrogenase activity resumes, however, when the stress is released or when the cultures are allowed to adapt to the increased O 2 concentration. Investigations of the mechanism underlying the switch-o reaction revealed an association of the nitrogenase complex with a low-molecular-mass FeSII protein. Formation of the three-component complex depends on the oxidation of nitrogenase and of its electron donor [11,12]. Studies on the physiological signi cance of the FeSII protein revealed that a mutant of A. vinelandii devoid of the FeSII protein was not impaired in its ability to grow diazotrophically in the presence of O 2 [13]. Under O 2 stress, however, the presence of the FeSII protein conferred signi cant protection from irreversible destruction of nitrogenase in vitro and in vivo. Switch-o of nitrogenase activity occurs when those mechanisms which keep the enzyme active in cultures growing diazotrophically in the presence of O 2 become overburdened. Consequently, the switch-o e ect may be used to estimate the e ciency of mechanisms protecting nitrogenase from inactivation. On this assumption, the e ciency of respiratory protection was tested with cultures of A. vinelandii adapted at a constant dilution rate to di erent ambient O 2 concentrations [40]. In each case, complete switch-o required a constant stress size of 23 WM O 2. This implies that the e ciency of protection is independent of the respiratory capacity of the organisms, because, as mentioned above, the respiratory capacity increases under steady-state conditions up to about 70 WM O 2 and approaches constancy at higher O 2 concentrations [9]. The e ciency of protection, however, signi cantly increases when the dilution rate is raised. Numerous control experiments proved that this latter e ect depends on the in situ nitrogenase activity, while it is independent of the actual rate of O 2 consumption or of the respiratory capacity of the cells [38]. Studies on the response of the cellular ATP pool to conditions of O 2 stress showed a signi cant decrease which was reversed as soon as the stress was released [41]. This type of response could not be observed in ammoniumassimilating cultures but only in cultures exhibiting nitrogenase activity, although both types of culture exhibited essentially the same rates of O 2 consumption. Therefore, it has been suggested that nitrogenase is either directly or indirectly involved in the enhanced ATP utilization under stress [41]. Moreover, under steady-state conditions, nitrogenase activity requires a critically high ATP level. These suggestions are supported by the following observations. During the adaptation of wild-type cells to the higher O 2 concentration, both the activity of nitrogenase and the cellular ATP pool concomitantly return to their respective levels before stress. In contrast and unexpectedly, the cytochrome bd deletion mutant MK5 of A. vinelandii, which cannot grow diazotrophically at increased O 2 concentrations [42], is unable to recover from O 2 stress by readjusting its ATP pool to the original size [41] Nitrogenase polypeptides Constancy of nitrogenase activity, as determined in A. vinelandii growing under steady-state conditions at di erent O 2 concentrations, has been con rmed after quanti cation of the cellular amounts of component 1 and component 2 of nitrogenase [39]. Within the range from 3 to 216 WM O 2, the ratio of component 1 to component 2 stays constant at 1:(1.45 þ 0.12) and the amounts of both polypeptides constitute a constant proportion of about 10% of the total cell protein [39,43]. Importantly, no signi cant turn-over of both components of nitrogenase could be registered. This was true even when the cultures were subjected to an O 2 stress for 2 h [39]. Upon increasing the dilution rate, cellular nitrogenase activity also increased, while the cellular amounts of the two nitrogenase polypeptides remained essentially constant. This suggests an increase in the catalytic activities of both components of nitrogenase. 6. Respiratory protection: the pros and cons The high rate of respiration observed in Azotobacter species growing diazotrophically at ambient O 2 concentrations may certainly be interpreted in favor of the assumed O 2 scavenging and, thus, protective function of respiration. On this basis the following proposal appears logical: ``If respiration is protective, the nitrogenase activity of living organisms ought to depend on their respiration'' [8]. Results of detailed analyses of the membrane-bound respiratory chain of A. vinelandii have been interpreted to suggest the involvement of NDH II and cytochrome bdtype oxidase in respiratory protection, because (i) their levels are increased when the O 2 supply is increased, (ii) the uncoupling of both segments explains the low e ciency of biomass production at increased O 2 supplies, and (iii) mutant MK5 of A. vinelandii, unable to synthesize cytochrome bd-type oxidase, has been reported to be unable to grow diazotrophically in the presence of air [14,42]. Yet, if O 2 consumption is a prerequisite for protection, it remains unexplained how this is possible at concentrations above 70 WM O 2 where, as detailed above, cellular O 2 consumption and the respiratory system exhibit no signi cant variations in activity or composition. Since the cultures are able to grow diazotrophically up to at least 230 WM O 2, it might be argued that the respiratory activity attained at the lower O 2 concentrations is su ciently high to extend protection to nitrogenase in cells growing at the higher O 2 concentrations. This possibility, however, can largely be excluded because it has been shown with continuous cultures growing under steadystate conditions that switch-o of nitrogenase requires an O 2 stress of constant size, irrespective of the O 2 concentration to which the organisms are adapted [40]. But

9 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^ when the dilution rate is raised, the sensitivity of the switch-o behavior toward an O 2 stress decreases without any apparent relationship to the actual rate of O 2 consumption [38]. From these results the conclusion can be drawn that, in the protection of nitrogenase, the supply of the limiting substrate (represented by the dilution rate) and, thus, the regeneration of reducing equivalents and energy are more important processes than the mere reduction of O 2. The respiratory capacities of cells as well as of de ned segments of the respiratory chain are being assayed with culture samples under optimal test conditions. Thus, it might be postulated that, under culture conditions, a more direct relationship may exist between the actual rates of O 2 consumption and of acetylene reduction. This argument can be ruled out by the observation that, upon increasing the C/N ratio, the actual cellular activity of acetylene reduction by nitrogenase increases steeply while the corresponding rate of O 2 consumption remains constant (Fig. 2) [22,23]. Conversely, when at a given O 2 concentration the type of growth limitation (i.e. substrate, sulfate, phosphate limitation) is changed, up to three-fold variations in the actual rate of O 2 consumption do not signi cantly a ect the actual cellular activity of nitrogenase [30,38] (Fig. 4). In this context, it is also of interest to recall that a mutant of A. vinelandii exhibiting increased levels of cytochrome bd but no detectable amounts of cytochrome o is able to grow diazotrophically in air, although its respiration is only half as active as that of the wild-type [44]. Finally, it should be noted that respiratory protection does not explain why net xation of N 2 is enhanced with substrates which are metabolized at decreased rates of O 2 consumption. Taken together, these results show that there is no direct dependence of the cellular activity of nitrogenase on the activity of cellular O 2 consumption. In light of these results, it even remains questionable whether the increase in respiratory activity in response to increasing the ambient O 2 concentration to up 70 WM ful lls a signi cant function in the protection of nitrogenase. The inability of mutant MK5 of A. vinelandii to increase its respiratory activity and to grow diazotrophically at moderately increased O 2 concentrations is presently considered as the most convincing evidence in favor of the hypothesis of respiratory protection [14,42]. Nevertheless, it remains unexplained why after a shift from 2.3 to 38 WM O 2 nitrogenase activity of MK5 eventually resumes at a low, yet signi cant, level [41]. More importantly, however, the ATP pool of mutant MK5 has been shown to be unable to recover from exposure to increased O 2 concentrations [41]. This e ect is not explained by the lack of the largely uncoupled cytochrome bd-type oxidase in mutant MK5. It appears plausible that mutant MK5 may be unable to adapt to higher O 2 concentrations because it is unable to readjust its energy metabolism to the level required for reactivating nitrogenase. Obviously, control of nitrogenase in mutant MK5 is more complex than originally assumed. 7. Does the membrane-bound respiratory electron transport system prevent O 2 from entering the cells? According to the hypothesis of respiratory protection, cellular respiration should be high enough to keep the intracellular concentration of O 2 at a level which is vaguely referred to as `non-toxically low'. Clearly, cellular respiration requires O 2, which has to enter the cell when serving as terminal electron acceptor at the inner face of the cytoplasmic membrane. However, with respect to the e ciency of protection, it should make essentially no difference whether O 2 is removed at the periplasmic or the cytoplasmic face of the cytoplasmic membrane. The e ciency of respiratory protection in preventing the di usion of O 2 into the cytoplasm must be questioned, if substrate amounts of O 2 are required for metabolic activities taking place in the cytoplasm. This is apparently true when aerobic organisms are growing at the expense of aromatic substrates, where oxidative cleavage of the aromatic ring is catalyzed by soluble oxygenases requiring O 2 as co-substrate. As a matter of fact, A. vinelandii has been reported to grow diazotrophically using para-hydroxybenzoate as the sole carbon source and electron donor [45]. Under these conditions, diazotrophic growth occurs at even the highest O 2 concentrations as assayed up to 180 WM O 2 (Oelze and Steindorf, unpublished ndings). An alternative line of evidence showing that O 2 is able to di use into the cytoplasm refers to the regulation of gene expression by O 2.InE. coli, intracellular O 2 is proposed to provide a direct signal to the transcriptional regulator FNR [36]. In A. vinelandii, the expression by O 2 of cydab genes encoding cytochrome bd requires the transcriptional regulator CydR which is closely related to FNR [35]. To support the hypothesis that relatively high amounts of O 2 are available in the cytoplasm of aerobic cells, the rate of O 2 di usion into cells has been estimated in Pseudomonas putida growing aerobically on benzoate as substrate [46]. The calculated rate of O 2 di usion of 360 mmol of O 2 min 31 g (dry weight) 31 exceeds not only the rates of cellular O 2 consumption in E. coli and P. putida by at least two orders of magnitude but also those in A. vinelandii and A. chroococcum ranging from about 0.2 to 4.0 mmol O 2 min 31 g (dry weight) 31 [9,30,37,38]. 8. Alternative strategies of protection The isolation of O 2 -sensitive mutants of A. vinelandii, which are not impaired in their respiratory activities, suggests that N 2 xation under aerobic conditions involves additional strategies of the protection of nitrogenase activity [47]. A variety of such strategies have been discussed

10 330 J. Oelze / FEMS Microbiology Reviews 24 (2000) 321^333 [5,6]. Some of them, like di usion barriers, particularly occur in heterocystous cyanobacteria or in symbiotic bacteria rather than in free-living vegetative cells. In the following, those mechanisms are summarized which are likely to provide protection of nitrogenase activity in azotobacters as representatives of the free-living aerobic diazotrophs. An interesting mechanism of the protection of nitrogenase activity in A. vinelandii and Klebsiella pneumoniae from O 2 inactivation has been proposed by Thorneley and Ashby [48] and outlined by Bergman and colleagues [6]. This mechanism, termed `autoprotection', implies that nitrogenase reduces O 2 without becoming inactivated. Thorneley and Ashby [48] suggested that, depending on the prevailing relative concentrations of O 2 and component 2, O 2 is reduced by component 2 to the superoxide radical or to H 2 O 2. These products can be removed by superoxide dismutase (SOD) or catalase/peroxidase. The hypothesis has been put forward that SOD may be important in the protection of the process of N 2 xation [49,50]. Investigations of cellular levels of SOD and catalase in A. vinelandii revealed that both enzymes increase when the prevailing O 2 concentration is raised [51]. Autoprotection as a possible mechanism of protection of nitrogenase has been discussed to occur not only in A. vinelandii and K. pneumoniae but also in Azorhizobium caulinodans and various members of the cyanobacteria [6,30,48,52]. Mechanisms like autoprotection or the cleavage of aromatic carbon sources by O 2 -consuming cytoplasmic oxygenases may help remove O 2 from the cytoplasm. In doing so they certainly contribute to whole-cell O 2 consumption. However, they should not contribute to the `respiratory protection' postulated to prevent the di usion of O 2 into the cells. Mutants of A. chroococcum lacking uptake hydrogenases have been reported to be out-competed by the wild-type strain when growing diazotrophically in sucrose-limited mixed cultures [53,54]. As a result of these observations, it has become widely accepted that the recycling of hydrogen evolved by nitrogenase may bene t the carbon-limited growth of diazotrophs [15,54]. In addition, it has been postulated that nitrogenase is protected from O 2 inactivation by respiration at the expense of hydrogen evolved by the same nitrogenase [55]. Since, however, rates of cellular O 2 consumption are generally about two orders of magnitude higher than cellular nitrogenase activities (Figs. 2 and 4), it can a priori be excluded that hydrogen evolved by nitrogenase signi cantly contributes to cellular respiration. This postulate was con rmed after determination of cellular O 2 consumption and biomass formation by an uptake hydrogenase-de cient mutant of A. vinelandii which showed essentially no di erences when compared to the wild-type [56]. Similar results were reported for a hydrogenase-de cient mutant of A. caulinodans growing diazotrophically ex planta [57]. In some cyanobacteria, continuous synthesis and, thus, high turn-over of nitrogenase may contribute to a constant level of N 2 xation under oxic conditions [5]. As mentioned above, however, no signi cant destruction of the in vivo nitrogenase of A. vinelandii by air could be observed [39]. On the other hand, the cellular level of nitrogenase may be important to maintain enzyme activity. This conclusion is derived from the observation that, upon raising the dilution rate in continuous cultures of A. vinelandii, inactive nitrogenase is activated and its stability toward O 2 inactivation increases [38]. This indicates that a surplus of nitrogenase may augment the capacity of cells to protect nitrogenase from inactivation [38,58]. The actual extent of protection, however, depends on the dilution rate representing the supply of the carbon source, i.e. the supply of the source of reducing equivalents as well as of ATP. Both are required for keeping nitrogenase at a su ciently low redox state required for its function [59]. As reported above, cellular respiration depends not only on the ambient O 2 concentration but also on the type of carbon source. Importantly, substrates giving rise to lower rates of respiration yield higher levels of biomass and, thus, of N 2 xed per carbon consumed [26,27]. Liu et al. [27] proposed that lower rates of respiration by the more coupled electron transport terminated by cytochrome o-type oxidases provide the cells with higher amounts of ATP than higher rates of respiration proceeding via a less coupled electron transport pathway terminated by cytochrome bd oxidase. This implies that net xation of N 2 depends on the regeneration of ATP rather than on the rate of cellular O 2 consumption. More recently, the dependence of net N 2 xation on respiration and on the cellular ATP pool has been studied in continuous cultures of A. vinelandii [30]. A single linear relationship was observed between the amount of N 2 xed and the cellular ATP concentration. The latter was true in spite of considerable di erences in cellular O 2 consumption, the kind of substrate, the type of growth limitation and ambient O 2. A role for the energy metabolism in the protection of nitrogenase activity is also suggested by the results mentioned above on the inability of mutant MK5 of A. vinelandii to adapt its nitrogenase activity and its ATP pool to increased ambient O 2 concentrations. In addition, the inability of citrate synthase-de cient mutants of A. chroococcum to grow diazotrophically in the presence of air [60] can be interpreted to result from the inability of these mutants to regenerate reducing equivalents and ATP. Essentially the same interpretation explains the observation that a mutant of A. vinelandii impaired in glucose uptake exhibits a diazotrophic growth defect unless the glucose supply is increased or the ambient O 2 concentration is decreased [61].