Biological nitrogen fixation is a critically important part of the nitrogen cycle in which gaseous dinitrogen from the air is "fixed" to form ammonia which can be used by micro-organisms and plants for growth. Only bacteria can carry out the nitrogen fixation process although the most important contribution of nitrogen fixation to agriculture is through the symbiosis of the soil microbes, rhizobia, with legume plants. In these symbioses the plant provides energy, in the form of photosynthate, for the microbe to fix dinitrogen to ammonia which is then utilized in plant growth processes.

John Postgate was one of the foremost microbiologists of his generation. He is most famous for his lifelong research on sulfate-reducing bacteria and nitrogen fixation and for his seminal contributions to understanding the survival and death of bacteria. John Postgate is also known for his specialist and non-specialist books on science, most notably Microbes and man , first published in 1986 and now in several editions and translated into several languages. He played an important role in the development and dissemination of microbiology and in the leadership of science in the UK. John will also be remembered warmly as the supervisor and mentor of aspiring young microbiologists, several of whom have gone on to distinguished careers in microbiology. His other great love was jazz: he was an amateur cornet player of note, the leader of several jazz groups and a highly knowledgeable writer, reviewer and author of two books on the subject.

The inactive MoFe protein of nitrogenase, NifB-Kp 1, from two distinct niB mutants of Klebsiella pneumoniae, Kp5058 (a nifB point mutant) and UNF1718 (a nijB, nifJ double mutant) has been purified and characterized. NifB-Kpl can be activated by reaction with the iron-molybdenum cofactor, FeMoco, extracted from active MoFe protein. NifB-Kpl purified from either source had similar properties and was contaminated with an approximately equimolar amount of protein of mol.wt. 21000. Like active wild-type Kp 1, it was an a2/J2 tetramer, but it was far less stable than Kpl, deteriorating rapidly at temperatures above 8°C or on mild oxidation. NifB-Kpl preparations contained 0.4-0.9 Mo and 9.0 + 0.9 Fe atoms mol-h and, when activated by FeMoco, had a specific activity of approx. 500 units mg-'. The Mo in our preparations was not associated with the e.p.r. signal normally observed from FeMoco. All preparations exhibited a weak gav = 1.95 e.p.r. signal which was probably not associated with activatable protein.

The first electrochemical and infra-red data on the binding of cyanide to the isolated iron-molybdenum cofactor of nitrogenase, FeMoco, is described. It is shown that cyanide stabilises a hitherto unrecognised, low-spin, EPR-active (S= 1/2), superoxidised form of FeMoco, and we provide the first evidence that carbon monoxide and cyanide bind synergically to the oxidised and semireduced states of the isolated cofactor, states which are unreactive to carbon monoxide alone.

The first kinetic study of a substrate (CN(-)) binding to the isolated active site (extracted FeMo-cofactor) of nitrogenase is described. The kinetics of the reactions between CN(-) and various derivatives of extracted FeMo-cofactor [FeMoco-L; where L is bound to Mo, and is NMF, Bu(t)NC, or imidazole (ImH)] have been followed using a stopped-flow, sequential-mix method in which the course of the reaction is followed indirectly, by monitoring the change in the rate of the reaction of the cofactor with PhS(-). The kinetic results, together with DFT calculations, indicate that the initial site of CN(-) binding to FeMoco-L is controlled by a combination of the electron-richness of the cluster core and lability of the Mo-L bond. Ultimately, the reactions between FeMoco-L and CN(-) involve displacement of L and binding of CN(-) to Mo. These reactions occur with a variety of rates and rate laws dependent on the nature of L. For FeMoco-NMF, the reaction with CN(-) is complete within the dead-time of the apparatus (ca. 4 ms), while with FeMoco-CNBu(t) the reaction is much slower and exhibits first order dependences on the concentrations of both FeMoco-CNBu(t) and CN(-) (k = 2.5 +/- 0.5 x 10(4) dm(3) mol(-1) s(-1)). The reaction of FeMoco-ImH with CN(-) occurs at a rate which exhibits a first order dependence on FeMoco-ImH but is independent of the concentration of CN(-) (k = 50 +/- 10 s(-1)). The results are interpreted in terms of CN(-) binding directly to the Mo site for FeMoco-NMF and FeMoco-ImH, but with FeMoco-CNBu(t) initial binding at an Fe site is followed by movement of CN(-) to Mo. Complementary DFT calculations are consistent with this interpretation, indicating that, in FeMoco-L, the Mo-L bond is stronger for L = ImH than for L = CNBu(t) and the binding of CN(-) to Mo is stronger than to any Fe atom in the cofactor.

The electron-transfer chemistry of the isolated iron-molybdenum cofactor of nitrogenase (FeMoco) has been studied by electrochemical and spectroelectrochemical methods. Two interconverting forms of the cofactor arise from a redox-linked ligand isomerism at the terminal iron atom; this is attributed to rotamerism of an anionic N-methyl formamide ligand bound at this site. FeMoco in its EPR-silent oxidised state is shown to undergo three successive one-electron transfer steps. We argue that the first and second redox processes are associated with electron-transfer delocalised over the iron-sulfur core of the cofactor, whilst the third irreversible process is localised on molybdenum. This is strongly reinforced by spectroelectrochemical studies under (12)CO and (13)CO which reveal two independent carbon monoxide binding sites that are specifically associated with the second (iron core) and third (molybdenum) electron-transfer processes and which give rise to terminal nu((12)CO) bands at 1885 and 1920 cm(-1) respectively. Moreover, in parallel with earlier studies on the enzyme system, it is shown that at low CO concentration, carbon monoxide binds to the cofactor in bridging modes, with nu(CO) bands at 1835 and 1808 cm(-1) that are interconverted by single-electron transfer. Importantly we show that the contentious overall 2e difference in the assignment of the metal oxidation levels in the resting state of the enzyme-bound cofactor, arising from analysis of (57)Fe ENDOR and Mössbauer data, can be resolved in the light of the electron-transfer chemistry of the isolated cofactor described herein.

The x-ray crystal structure of NifV(-) Klebsiella pneumoniae nitrogenase MoFe protein (NifV(-) Kp1) has been determined and refined to a resolution of 1.9 Angstrom. This is the first structure for a nitrogenase MoFe protein with an altered cofactor. Moreover, it is the first direct evidence that the organic acid citrate is not just present, but replaces homocitrate as a ligand to the molybdenum atom of the iron molybdenum cofactor (FeMoco). Subsequent refinement of the structure revealed that the citrate was present at reduced occupancy.

Stable inactive 2 : 1 complexes of the Klebsiella pneumoniae nitrogenase components (Kp2/Kp1) were prepared with ADP or the fluorescent ADP analogue, 2′(3′)‐O‐[N‐methylanthraniloyl] ADP and AlF4− or BeF3− ions. By analogy with published crystallographic data [Schindelin et al. (1997) Nature387, 370–376)], we suggest that the metal fluoride ions replaced phosphate at the two ATP‐binding sites of the iron protein, Kp2. The beryllium (BeFx) and aluminium (AlF4−) containing complexes are proposed to correspond to the ATP‐bound state and the hydrolytic transition states, respectively, by analogy with the equivalent complexes of myosin [Fisher et al. (1995) Biochemistry34, 8960–8972]. 31P NMR spectroscopy showed that during the initial stages of complex formation, MgADP bound to the complexed Kp2 in a manner similar to that reported for isolated Kp2. This process was followed by a second step that caused broadening of the 31P NMR signals and, in the case of the AlF4− complex, slow hydrolysis of some of the excess ADP to AMP and inorganic phosphate. The purified BeFx complex contained 3.8 ± 0.1 MgADP per mol Kp1. With the AlF4− complex, MgAMP and adenosine (from MgAMP hydrolysis) replaced part of the bound MgADP although four AlF4− ions were retained, demonstrating that full occupancy by MgADP is not required for the stability of the complex. The fluorescence emission maximum of 2′(3′)‐O‐[N‐methylanthraniloyl] ADP was blue‐shifted by 6–8 nm in both metal fluoride complexes and polarization was 6–9 times that of the free analogue. The fluorescence yield of bound 2′(3′)‐O‐[N‐methylanthraniloyl] ADP was enhanced by 40% in the AlF4− complex relative to the solvent but no increase in fluorescence was observed in the BeFx complex. Resonance energy transfer from conserved tyrosine residues located in proximity to the Kp2 nucleotide‐binding pocket was marked in the AlF4− complex but minimal in the BeFx fluoride complex, illustrating a clear conformational difference in the Fe protein of the two complexes. Our data indicate that complex formation during the nitrogenase catalytic cycle is a multistep process involving at least four conformational states of Kp2: similar to the free Fe protein; as initially complexed with detectable 31P NMR; as detected in mature complexes with no detectable 31P NMR; in the AlF4− complex in which an altered tyrosine interaction permits resonance energy transfer with 2′(3′)‐O‐[N‐methylanthraniloyl] ADP.

Stable inactive 2 : 1 complexes of the Klebsiella pneumoniae nitrogenase components (Kp2/Kp1) were prepared with ADP or the fluorescent ADP analogue, 2'(3')-O-[N-methylanthraniloyl] ADP and AlF(4)(-) or BeF(3)(-) ions. By analogy with published crystallographic data [Schindelin et al. (1997) Nature 387, 370-376)], we suggest that the metal fluoride ions replaced phosphate at the two ATP-binding sites of the iron protein, Kp2. The beryllium (BeF(x)) and aluminium (AlF(4)(-)) containing complexes are proposed to correspond to the ATP-bound state and the hydrolytic transition states, respectively, by analogy with the equivalent complexes of myosin [Fisher et al. (1995) Biochemistry 34, 8960-8972]. (31)P NMR spectroscopy showed that during the initial stages of complex formation, MgADP bound to the complexed Kp2 in a manner similar to that reported for isolated Kp2. This process was followed by a second step that caused broadening of the (31)P NMR signals and, in the case of the AlF4- complex, slow hydrolysis of some of the excess ADP to AMP and inorganic phosphate. The purified BeFx complex contained 3.8 +/- 0.1 MgADP per mol Kp1. With the AlF(4)(-) complex, MgAMP and adenosine (from MgAMP hydrolysis) replaced part of the bound MgADP although four AlF(4)(-) ions were retained, demonstrating that full occupancy by MgADP is not required for the stability of the complex. The fluorescence emission maximum of 2'(3')-O-[N-methylanthraniloyl] ADP was blue-shifted by 6-8 nm in both metal fluoride complexes and polarization was 6-9 times that of the free analogue. The fluorescence yield of bound 2'(3')-O-[N-methylanthraniloyl] ADP was enhanced by 40% in the AlF(4)(-) complex relative to the solvent but no increase in fluorescence was observed in the BeFx complex. Resonance energy transfer from conserved tyrosine residues located in proximity to the Kp2 nucleotide-binding pocket was marked in the AlF(4)(-) complex but minimal in the BeFx fluoride complex, illustrating a clear conformational difference in the Fe protein of the two complexes. Our data indicate that complex formation during the nitrogenase catalytic cycle is a multistep process involving at least four conformational states of Kp2: similar to the free Fe protein; as initially complexed with detectable (31)P NMR; as detected in mature complexes with no detectable (31)P NMR; in the AlF(4)(-) complex in which an altered tyrosine interaction permits resonance energy transfer with 2'(3')-O-[N-methylanthraniloyl] ADP.