Characteristics of N2 fixation in Mo-limited batch and continuous cultures of Azotobacter vinelandii

Diversity and Activity of Alternative Nitrogenases in Sequenced Genomes and Coastal Environments

The nitrogenase enzyme, which catalyzes the reduction of N₂ gas to NH₄⁺, occurs as three separate isozyme that use Mo, Fe-only, or V. The majority of global nitrogen fixation is attributed to the more efficient 'canonical' Mo-nitrogenase, whereas Fe-only and V-('alternative') nitrogenases are often considered 'backup' enzymes, used when Mo is limiting. Yet, the environmental distribution and diversity of alternative nitrogenases remains largely unknown. We searched for alternative nitrogenase genes in sequenced genomes and used PacBio sequencing to explore the diversity of canonical (nifD) and alternative (anfD and vnfD) nitrogenase amplicons in two coastal environments: the Florida Everglades and Sippewissett Marsh (MA). Genome-based searches identified an additional 25 species and 10 genera not previously known to encode alternative nitrogenases. Alternative nitrogenase amplicons were found in both Sippewissett Marsh and the Florida Everglades and their activity was further confirmed using newly developed isotopic techniques. Conserved amino acid sequences corresponding to cofactor ligands were also analyzed in anfD and vnfD amplicons, offering insight into environmental variants of these motifs. This study increases the number of available anfD and vnfD sequences ∼20-fold and allows for the first comparisons of environmental Mo-, Fe-only, and V-nitrogenase diversity. Our results suggest that alternative nitrogenases are maintained across a range of organisms and environments and that they can make important contributions to nitrogenase diversity and nitrogen fixation.

Nitrogen Fixation by Azotobacter vinelandii Strains Having Deletions in Structural Genes for Nitrogenase

Phenotypic reversal of Nif(-) mutant strains to Nif(+) under molybdenum-deficient conditions has been cited as evidence that Azotobacter vinelandii possesses two nitrogen fixation systems: the conventional molybdenum-enzyme system and an alternative nitrogen-fixation system. Since explanations other than the existence of an alternative system were possible, deletion strains of A. vinelandii lacking the structural genes for conventional nitrogenase (nifHDK) were constructed. These strains were found to grow in molybdenum-deficient nitrogen-free media, reduce acetylene (at low rates), and incorporate molecular nitrogen labeled with nitrogen-15. Thus it can be concluded that the phenotypic reversal phenomenon cannot be due to altered phenotypic expression of nif mutations under molybdenum-deficient conditions, but is due to the existence of an alternative nitrogen-fixation system in A. vinelandii as originally proposed.

Diazotrophy and Nitrogenase Activity in the Archaebacterium Methanosarcina barkeri 227

Nitrogen fixation (diazotrophy) has recently been demonstrated in several methanogenic archaebacteria. To compare the process in an archaebacterium with that in eubacteria, we examined the properties of diazotrophic growth and nitrogenase activity in Methanosarcina barkeri 227. Growth yields with methanol or acetate as a growth substrate were significantly lower in N(2)-grown cultures than in NH(4)-grown cultures, and the culture doubling times were increased, indicating that diazotrophy was energetically costly, as it is in eubacteria. Growth of nitrogen-fixing cells was inhibited when molybdenum was omitted from the medium; addition of 10 nM molybdate stimulated growth, while 1 muM molybdate restored maximum diazotrophic growth. Omission of molybdenum did not inhibit growth of ammonia-grown cells. Tungstate (100 muM) strongly inhibited growth of molybdenum-deficient diazotrophic cells, while ammonia-grown cells were unaffected. The addition of 100 nM vanadate or chromate did not stimulate diazotrophic growth of molybdenum-starved cells. These results are consistent with the presence of a molybdenum-containing nitrogenase in M. barkeri. Acetylene, the usual substrate for assaying nitrogenase activity, inhibited methanogenesis by M. barkeri and consequently needed to be used at a low partial pressure (0.3% of the headspace) when acetylene reduction by whole cells was assayed. Whole cells reduced 0.3% acetylene to ethylene at a very low rate (1 to 2 nmol h mg of protein), and they "switched off" acetylene reduction in response to added ammonia or glutamine. Crude extracts from diazotrophic cells reduced 10% acetylene at a rate of 4 to 5 nmol of C(2)H(4) formed h mg of protein when supplied with ATP and reducing power, while extracts of Klebsiella pneumoniae prepared by the same procedures had rates 100-fold higher. Acetylene reduction by extracts required ATP and was completely inhibited by 1 mM ADP in the presence of 5 mM ATP. The low rates of C(2)H(2) reduction could be due to improper assay conditions, to switched-off enzyme, or to the nitrogenase's having lower activity towards acetylene than towards dinitrogen.

Characterization of diazotrophs containing Mo-independent nitrogenases, isolated from diverse natural environments

Molybdenum-independent nitrogenases were first described in the nitrogen-fixing bacterium Azotobacter vinelandii and have since been described in other diazotrophic bacteria. Previously, we reported the isolation of seven diazotrophs with Mo-independent nitrogenases from aquatic environments. In the present study, we extend these results to include diazotrophs isolated from wood chip mulch, soil, "paraffin dirt," and sediments from mangrove swamps. Mo-deficient, N-free media under both aerobic and anaerobic conditions were used for the isolations. A total of 26 isolates were genetically and physiologically characterized. Their phylogenetic placement was determined using 16S rRNA gene sequence analysis. Most of the isolates are members of the gamma subdivision of the class Proteobacteria and appear to be specifically related to fluorescent pseudomonads and azotobacteria. Two other isolates, AN1 and LPF4, are closely related to Enterobacter spp. and Paenibacillus spp., respectively. PCR and/or Southern hybridization were used to detect the presence of nitrogenase genes in the isolates. PCR amplification of vnfG and anfG was used to detect the genetic potential for the expression of the vanadium-containing nitrogenase and the iron-only nitrogenase in the isolates. This study demonstrates that diazotrophs with Mo-independent nitrogenases can be readily isolated from diverse natural environments.

Complexation of oxoanions and cationic metals by the biscatecholate siderophore azotochelin

Azotochelin is a biscatecholate siderophore produced by the nitrogen-fixing soil bacterium Azotobacter vinelandii. The complexation properties of azotochelin with a series of oxoanions [Mo(VI), W(VI) and V(V)] and divalent cations [Cu(II), Zn(II), Co(II) and Mn(II)] were investigated by potentiometry, UV-vis and X-ray spectroscopy. Azotochelin forms a strong 1:1 complex with molybdate (log K=7.6+/-0.4) and with tungstate and vanadate; the stability of the complexes increases in the order Mo<V<W (log KappMo=7.3+/-0.4; log KappV=8.8+/-0.4 and log KappW=9.0+/-0.4 at pH 6.6). The Mo atom in the 1:1 Mo-azotochelin complex is bound to two oxo groups in a cis position and to the two catecholate groups of azotochelin, resulting in a slightly distorted octahedral configuration. Below pH 5, azotochelin appears to form polynuclear complexes with Mo in addition to the 1:1 complex. Azotochelin also forms strong complexes with divalent metals. Of the metals studied, Cu(II) binds most strongly to azotochelin (log betaCuLH2-=-12.9+/-0.1), followed by Zn(II) log betaZnL3-=-24.1+/-0.14, log betaZnLH2-=-17.83+/-0.09), Mn(II) (log betaMnL3-=-29, log betaMnLH2-=-18.6+/-0.8, log betaMnLH2-=-11.5+/-0.7) and Co(II) (log betaCoLH2-=-23.0+/-0.3, log betaCoLH2-=-13.5+/-0.2). Since very few organic ligands are known to bind strongly to oxoanions (and particularly molybdate) at circumneutral pH, the unusual properties of azotochelin may be used for the separation and concentration of oxoanions in the laboratory and in the field. In addition, azotochelin may prove useful for the investigation of the biogeochemistry of Mo, W and V in aquatic and terrestrial systems.

The role of regulatory genes nifA, vnfA, anfA, nfrX, ntrC, and rpoN in expression of genes encoding the three nitrogenases of Azotobacter vinelandii

Several regulatory gene mutants of Azotobacter vinelandii were tested for ability to synthesize functional nitrogenase-1 (Nif phenotype), nitrogenase-2 (Vnf), or nitrogenase-3 (Anf). While nifA mutants were Nif-, Vnf+, and Anf+/-, and ntrC mutants were Nif+, Vnf+, and Anf+, nifA ntrC double mutants were Nif-, Vnf-, and Anf-. A vnfA mutant was Nif+, Vnf+/-, and Anf+/-, and an anfA strain was Nif+, Vnf+, and Anf-. lacZ fusions in the nifH, vnfH, vnfD, anfH, and nifM genes of Azotobacter vinelandii were constructed and introduced into wild-type and regulatory mutants of A. vinelandii. Expression of these operons correlated with the growth phenotype of the regulatory mutants. Apparently either NifA or NtrC can activate expression of nifM. Also, expression of the anf operon required the NifA transcriptional activator, although there are no NifA binding sites at appropriate locations upstream of anfH (or anfA). The results confirm previous reports that VnfA and AnfA are required for expression of vnf and anf genes, respectively, and that VnfA is involved in repression of the nifHDK operon in the absence of molybdenum and of the anfHDGK operon in the presence of vanadium.

Characterization of genes involved in molybdenum transport in Azotobacter vinelandii

Expression of alternative nitrogenases in Azotobacter vinelandii is repressed by molybdenum. Two strains with Tn5 insertion mutations showed alternative nitrogenase-dependent diazotrophic growth in the presence of Mo. The mutations were in a region which contained four open reading frames (ORFs 1-4). The genetic structure and predicted products of ORFs 2, 3 and 4 are typical of the membrane-associated elements of the ATP-binding cassette (ABC) superfamily of transport systems. The products of ORF3 and ORF4 are homologous with the products of the Escherichia coli genes chlD and the partially sequenced chlJ, respectively, both of which are implicated in molybdenum transport. ORF1, which is in the relative position of bacterial permease genes commonly specifying periplasmic binding proteins, encodes a 29 kDa protein with a novel primary structure. It lacks a potential signal sequence, and its C-terminal half consists of a tandem repeat of a segment which is homologous with the M(r) 7 kDa molybdenum-pterin binding protein Mop from Clostridium pasteurianum. This suggests that a substituted pterin may be involved in the initial capture or early metabolism of molybdenum.

The nifU, nifS and nifV gene products are required for activity of all three nitrogenases of Azotobacter vinelandii

Strains with mutations in 23 of the 30 genes and open reading frames in the major nif gene cluster of A. vinelandii were tested for ability to grow on N-free medium with molybdenum (Nif phenotype), with vanadium (Vnf phenotype), or with neither metal present (Anf phenotype). As reported previously, nifE, nifN, nifU, nifS and nifV mutants were Nif- (failed to grow on molybdenum) while nifM mutants were Nif-, Vnf- and Anf-. nifV, nifS, and nifU mutants were found to be unable to grow on medium with or without vanadium, i.e. were Vnf- Anf-. Therefore neither vnf nor anf analogoues of nifU, nifS, nifV or nifM are expected to be present in A. vinelandii.

Transcriptional regulation by metals of structural genes for Azotobacter vinelandii nitrogenases

Azotobacter vinelandii has three nitrogenases: a molybdenum (Mo) nitrogenase, a vanadium (V) nitrogenase, and a third nitrogenase (nitrogenase-3), which apparently lacks Mo and V. Mo represses synthesis of both V nitrogenase and nitrogenase-3, and in the absence of Mo, V represses synthesis of nitrogenase-3. We have investigated transcriptional regulation of the three nitrogenases by metals using Northern analysis and probes specific for transcripts of each of the three nitrogenases. Our results confirm that Mo is required for expression of the Mo nitrogenase structural genes (nifHDK), and substantiate the notion that Mo represses transcription of the structural genes for both V nitrogenase and nitrogenase-3. We show that repression by V of nitrogenase-3 is also effected at the level of transcription. Unexpectedly, V only represses transcription of the nitrogenase-3 structural genes (anfHDGK) if the V nitrogenase structural gene cluster vnfDGK is present. Further, deletion of nifHDK allows low expression of anfHDGK in the presence of Mo. Repression by Mo or V is independent of cofactor synthesis and therefore of enzyme activity.

Selective removal of molybdenum traces from growth media of N2-fixing bacteria

A new method for the selective removal of traces of molybdenum from growth media of N2-fixing bacteria (Rhodobacter capsulatus and Klebsiella pneumoniae) was developed. This method is based on the filtration of nutrient solutions through a layer of activated carbon (pulverized charcoal). The adsorption of Mo (molybdate) to activated carbon was optimal if a charcoal suspension (50 g/liter) was degassed by boiling before use and if the pH of the solutions, which had to be purified, was adjusted to values between 1.5 and 4. In this pH region no or only negligible amounts of other metal ions were adsorbed. The activated carbon method was compared with other Mo-eliminating procedures, including 8-hydroxyquinoline/dichloromethane extraction, Chelex 100 chromatography, and treatment with Mo-starved Azotobacter vinelandii cells. The activated carbon filtration appeared to be the most effective, specific, and rapid method. Whereas the untreated Rhodobacter growth medium was contaminated with 1.2 ppb Mo, as analyzed by inductively coupled plasma mass spectrometry (ICP-MS), the activated carbon-treated medium was below the ICP-MS detection limit (less than 0.05 ppb). A similarly effective removal of Mo impurities was obtained by the Azotobacter treatment. Even at low optical densities (2-5 at 436 nm) Mo traces were removed very rapidly within 10-15 min. However, because the Mo uptake/Mo adsorption capacity of A. vinelandii depended on freshly cultivated cells and on the growth phase at which the cells were harvested, this microbiological method was generally more time-consuming and less reproducible than the activated carbon method.(ABSTRACT TRUNCATED AT 250 WORDS)

Demonstration of a molybdenum- and vanadium-independent nitrogenase in a nifHDK-deletion mutant of Rhodobacter capsulatus

In Rhodobacter capsulatus there exists, in addition to a conventional Mo-containing nitrogenase, a second, Mo-indendent nitrogenase which was demonstrated in wild-type cells as well as in cells of a nifHDK- mutant. To construct this R. capsulatus mutant, a 4-kb BglII-HindIII fragment encompassing nifK, nifD and most of the nifH coding region was substituted by an interposon coding for kanamycin resistance. The alternative nitrogenase is repressed by molybdenum. Mo concentration greater than 1 ppb in the growth medium prevented diazotrophic growth of nifHDK- cells and the expression of nitrogenase activity. The Mo-independent nitrogenase was maximally derepressed in activated carbon-treated media which contained less than 0.05 ppb Mo, high concentrations of iron (1 mM ferric citrate) and serine as N source. Under N2-fixing and optimal Mo-deficient conditions, nifHDK- cells grew with a doubling time of 9 h. The highest activity achieved with whole cells was 1.2 nmol ethylene.min-1.mg protein-1. Vanadium neither stimulated nor inhibited growth and activity. The alternative nitrogenase reduced acetylene to both ethylene and ethane. With whole cells (nifHDK-) the proportion of ethane varied over 2-5% depending on the amount of residual traces of Mo in the medium. The addition of Mo to a growing, nitrogenase-active culture resulted in a slow decrease of total activity but also in a simultaneous increase of ethane production up to 40%. In contrast, cell-free extracts and the purified enzyme did not show any or only very little ethane formation (0-0.4%). Both enzyme components appeared to be very labile proteins. Component 2 lost almost all its activity during cell breakage. With component 1 in crude extracts, if complemented with the stable component 2 of the Mo-nitrogenase from Xanthobacter autotrophicus, a recovery of 50% of the original whole cell activity could be achieved. During purification, component 1 (from the nifHDK- mutant) remained remarkably stable. The partially purified component 1 had a pH optimum (acetylene reduction) of 7.8-8.0, relatively high affinity to acetylene (Km = 0.055 mM) and was analyzed to contain 20 mol Fe atoms/mol protein, 0.2 mol Mo atoms and negligible amounts of V, W and Re. The dithionite-reduced dinitrogenase appeared to be ESR-silent. The results indicate that the alternative nitrogenase of R. capsulatus is not a vanadium enzyme but rather a heterometal-free Fe-nitrogenase or a nitrogenase with an as-yet-unidentified heterometal atom.

Temperature-Dependent Regulation by Molybdenum and Vanadium of Expression of the Structural Genes Encoding Three Nitrogenases in Azotobacter vinelandii

Temperature affects the expression of the three different nitrogenases in Azotobacter vinelandii. Molybdenum repressed the vnfH and anfH operons relatively more at 30°C than at 20°C; at 14°C molybdenum did not repress these genes at all. Similarly, V repressed the anf operon at 30°C but not at 20 or 14°C. Mo was poorly transported into cells grown at the lower temperatures. A. vinelandii thus has the potential to synthesize any of the three nitrogenases at 14 to 20°C regardless of the presence of Mo or V.

Genetic evidence for an Azotobacter vinelandii nitrogenase lacking molybdenum and vanadium

We have constructed a strain of Azotobacter vinelandii which has deletions in the genes for both the molybdenum (Mo) and vanadium (V) nitrogenases. This strain fixed nitrogen in medium that did not contain Mo or V. Growth and nitrogenase activity were inhibited by Mo and V. In highly purified medium, growth was limited by iron. Addition of other metals (Co, Cr, Cu, Mn, Ni, Re, Ti, W, and Zn) did not stimulate growth. Like the V-nitrogenase, the nitrogenase synthesized by the double deletion strain reduced acetylene to both ethylene and ethane (C2H6/C2H4 ratio, 0.046). There was an approximately 10-fold increase in ethane production when Mo was added to the deletion strain grown in medium lacking Mo and V. This change in reactivity may be due to the incorporation of an Mo-containing cofactor into the nitrogenase synthesized by the double-deletion strain. A strain synthesizing the V-nitrogenase did not show a similar increase in ethane production. The growth characteristics of the double-deletion strain, together with the metal composition reported for a nitrogenase isolated from a tungstate-tolerant strain lacking genes for the molydenum enzyme grown in the absence of Mo and V (J. R. Chisnell, R. Premakumar, and P. E. Bishop, J. Bacteriol. 170:27-33, 1988) show that A. vinelandii can synthesize a nitrogenase which lacks both Mo and V. Reduction of dinitrogen by nitrogenase can therefore occur at a center lacking both these metals.

Identification and characterization of two nitrogen fixation regulatory regions, nifA and nfrX, in Azotobacter vinelandii and Azotobacter chroococcum

Five Tn5-induced Nif- mutants of Azotobacter vinelandii were characterized as regulatory mutants because they were restored to Nif+ by the introduction of constitutively expressed nifA from Klebsiella pneumoniae. The mutants fell into two different classes on the basis of hybridization to a Rhizobium leguminosarum nifA gene probe and by complementation with cosmids isolated from pLAFRI gene banks of A. vinelandii and Azotobacter chroococcum. One mutant, MV3, was located in or near a nifA gene. The others, MV12, MV16, MV18 and MV26, defined a new regulatory gene, which has been called nfrX. The lack of expression of different nif-lacZ fusions confirmed the regulatory phenotype of all five mutant strains. The ability of both nifA and nfrX mutants to grow on nitrogen-free medium with vanadium, but not on medium with molybdenum, suggests that neither gene is required for expression of the alternative V-containing nitrogenase of A. vinelandii. A fragment carrying Tn5 and flanking DNA from MV3 was used as a probe to isolate the nifA region of A. chroococcum. Ligation of two adjacent EcoRI fragments of A. chroococcum yielded an intact nifA gene that activated expression of nifH-lac fusions and also restored MV3 to Nif+. The four nfrX mutants were complemented by pLAFR1 cosmids pLV163 and pLC121. The nfrX gene was subcloned from pLV163 and located within a 3.2 kb fragment. To determine whether nfrX might be found in other nitrogen-fixing organisms, DNA from 13 different species was hybridized to an nfrX probe. The failure to observe hybridization suggests that nfrX may be specific to nif regulation in Azotobacter.

Bacterial alternative nitrogen fixation systems

The introduction briefly reviews some of the salient features of the well-characterized conventional molybdo-enzyme system for N2 fixation. This is followed by a brief account of the discovery of an alternative N2 fixation system that does not require molybdenum in the N2-fixing bacterum Azotobacter vinelandii. The next section cites observations from the early literature on N2 fixation suggesting may not always require molybdenum. Next, recent evidence for an alternative N2 fixation system in A. vinelandii is discussed. A brief description of our discovery of an alternative nitrogenase which is not a molybdenum or vanadium enzyme is presented, followed by a summary of recent papers describing an alternative vanadium-containing nitrogenase. Available information on the genetics and regulation of alternative N2 fixation systems is discussed. Finally, the possible/probable presence of alternative N2 fixation systems in bacteria other than Azotobacter species is covered.

Purification of a second alternative nitrogenase from a nifHDK deletion strain of Azotobacter vinelandii

A second alternative nitrogenase complex (nitrogenase 3) was purified from a nifHDK deletion strain of Azotobacter vinelandii. The active complex is made up of two components, dinitrogenase 3 and dinitrogenase reductase 3. Dinitrogenase 3 contains two protein subunits (alpha, Mr 58,000, and beta, Mr 50,000) which assemble into at least two active configurations: alpha 2 beta 2 (dinitrogenase 3s) and alpha 1 beta 2 (dinitrogenase 3F). Dinitrogenase 3s contains 24 Fe and 18 acid-labile S2-ions per Mr 216,000, and dinitrogenase 3F contains 11 Fe and 9 acid-labile S2-ions per Mr 158,000. Dinitrogenase reductase 3 is composed of two protein subunits of identical Mr (32,500) and contains four Fe and four acid-labile S2- ions per Mr 65,000. On two-dimensional gels, the protein subunits of the nitrogenase 3 complex comigrated with the four Mo-, V-, and NH4+-repressible proteins originally designated as N2ase B: the nitrogenase hypothesized to exist in the alternative N2 fixation system first described in 1980 (P.E. Bishop, D. M. L. Jarlenski, and D. R. Hetherington, Proc. Natl. Acad. Sci. USA 77:7342-7346, 1980). Neutron activation analysis indicated that the nitrogenase 3 complex lacked significant amounts of Mo, V, Cr, Re, and W. Some Zn, however, was found in the dinitrogenase 3S and dinitrogenase 3F preparations. The pattern of substrate reduction efficiency was H+ greater than N2 greater than C2H2. The maximum specific activity found for N2 reduction was 38 nmol of NH3 per min per mg of protein (dinitrogenase 3S). Nitrogenase 3 was found to be extremely sensitive to O2, and activities could not be reproducibly maintained during freezing and thawing.

The vanadium nitrogenase of Azotobacter chroococcum. Purification and properties of the VFe protein

1. Nitrogenase activity of a strain of Azotobacter chroococcum lacking the structural genes for conventional nitrogenase (nifHDK) was separated into two components: an Fe-containing protein and a vanadoprotein. 2. The larger protein was purified to homogeneity by the criterion of electrophoresis of 10% (w/v) acrylamide gels in the presence of SDS. Two types of subunit, of Mr 50,000 and 55,000, were present in equal amounts. 3. The protein had an Mr of 210,000 and contained 2 V atoms, 23 Fe atoms and 20 acid-labile sulphide groups per molecule. The Mo content was less than 0.06 g-atom/mol. All the common amino acids were present, with a predominance of acidic residues. Ultracentrifugal analysis gave a maximum sedimentation coefficient of 9.7 S and a symmetrical boundary at 5 mg of protein X ml-1; dissociation occurred at lower concentrations. The specific activities (nmol of product/min per mg of protein), when assayed under optimum conditions with the complementary Fe protein from this strain, were 1348 for H2 evolution, 350 for NH3 formation and 608 for acetylene reduction. Activity was O2-labile, with a t1/2 of 40 s in air. At low temperatures the dithionite-reduced protein showed e.p.r. signals at g = 5.6, 4.35, 3.77 and 1.93, consistent with an S = 3/2 ground state with an additional S = 1/2 centre giving rise to the feature at g = 1.93. The u.v. spectra of dithionite-reduced and thionine-oxidized protein were very similar. Oxidation resulted in a general increase in absorbance in the visible region. The shoulder at 380 nm in the spectrum of reduced protein was replaced with shoulders near 330 nm and 420 nm on oxidation.

Nitrogen fixation in molybdenum-deficient continuous culture by a strain of Azotobacter vinelandii carrying a deletion of the structural genes for nitrogenase (nifHDK)

Steady-state chemostat cultures of Azotobacter vinelandii strain CA11, carrying a deletion of genes encoding the structural polypeptides of nitrogenase nifHDK, were established in a simple defined medium chemically purified to minimize contamination by Mo. The medium contained no utilizable N source. Growth was dependent on N2 (1.1 X 10(8) viable cells X ml-1 at D = 0.176 h-1), and was inhibited by Mo (20 nM). DNA hybridization showed the deletion to be stable during prolonged (55 days) growth in the chemostat (132 doublings). Since batch cultures, using unsupplemented 'spent' chemostat medium, showed good growth (1.9 X 10(8) cells X ml-1), no requirement for subnanomolar concentrations of Mo was found. The biomass yield, as the dilution rate (D) was varied, showed that the N content of the culture, protein and dry wt. increased as D was decreased, indicating that neither N2 nor O2 were limiting growth. The limiting nutrient was not identified. Substantial amounts of H2 were evolved by the chemostat cultures, probably as the result of inhibition of O2-dependent hydrogenase activity by nitrilotriacetic acid present in the medium. Over a range of D values approx. 50% of the electron flux through the alternative system was allocated to H+ reduction. C2H2 was a poor substrate, being reduced at 0.14-0.1 times the rate of N2 fixation, calculated from the N content of the cells. SO4(2-)-limited steady-state continuous cultures of strain UW136 (wild-type for nifHDK) had a 2-fold greater biomass in the presence of MoO4(2-) (1 microM). The significance of this finding for 'Mo-limited' continuous cultures [Eady & Robson (1984) Biochem. J. 224, 853-862] is discussed.