The involvement of the fixABCX genes and the respiratory chain in the electron transport to nitrogenase in Azotobacter vinelandii

Introduction.
The work in this thesis is mainly focused on the electron transport route to nitrogenase in the free-living, obligate aerobic, nitrogen fixing organism Azotobacter vinelandii. For many years now, this topic has been the subject of research. Several hypotheses, which would explain the mechanism of electron transport to nitrogenase in obligate aerobic bacteria, have been postulated. None of these hypotheses have been proven yet.

The electron transport to nitrogenase in A.vinelandii has been investigated both biochemically and genetically. It is known in Klebsiella pneumoniae, which fixes nitrogen anaerobically or microaerobically, that the gene products of two genes are responsible for the electron transport to nitrogenase, the nifF gene and the nifJ gene. They encode a flavodoxin and a pyruvate: flavodoxin oxidoreductase, respectively. Electrons are transferred from pyruvate to flavodoxin through this oxidoreductase, and are then passed on to the nitrogenase proteins. This reaction, nor the two genes involved, have been found in A.vinelandii. There must therefore be different pathway for electrons to reduce nitrogenase in this organism.

Gubler and Hennecke [1986] discovered a number of genes, the fixA , B and C genes, which were required for nitrogen fixation in the obligate aerobic Bradyrhizobium japonicum, both in symbiotic and free-living state. Later, they found that another gene, linked to the fixBC cluster, the fixX gene, was also involved in nitrogen fixation. Since the nodules of plants, infected with mutants in these genes were normal, but no nitrogenase activity was observed, a function in the electron transport to nitrogenase was suggested. This was contradicted later by Kaminski and coworkers [1989], who found that both in vivo and in vitro nitrogenase activity was absent in mutants in the fixABCX genes in Azorhizobium caulinodans ORS571.

The fixPABCX genes of A.vinelandii: genetic analysis.
In order to find out whether these genes are also present in A.vinelandii, a 4.4 kb part of the genome of this organism, which hybridised to a heterologous fixA probe from Rhizobium leguminosarum was isolated. The nucleotide sequence of the 4.4 kb Sma I -Eco RI fragment of Azotobacter vinelandii was determined. Five open reading frames (ORF's) and the beginning of a sixth one were found. The proteins encoded by the open reading frames were investigated for their homologies with other known gene products in the Genbank ^®databank.

The nucleic acid derived protein sequence of the first open reading frame contains two cysteine patterns, [Cys-X7-Cys-X3-Cys] and [Cys-X2-Cys-X2-Cys-X3-Cys], indicative for the ligation of both a [3Fe-4S] and a [4Fe-4S] cluster. These sequences are characteristic for 7Fe-ferredoxins, such as ferredoxin I of A.vinelandii (FdI). Besides the conserved cysteine residues, no other regions of homology between the fixP gene product and ferredoxin I were found, and no apparent homology with any other protein in the database was found

The second open reading frame showed a high degree of homology with the fixA gene product of various other aerobic nitrogen fixing bacteria. The A.vinelandii fixA gene product did not have any homology with other proteins in the database, so its function could not be determined from the nucleic acid derived protein sequence.

Homology searches revealed that the product of the third open reading frame not only had a high degree of homology with the sequence of FixB proteins, but also with the protein sequence of the α-subunit of the Electron Transfer Flavoprotein (ETF) of both human and rat origin. No other significant homologies with the A. vinelandii fixB gene product were found in the database.

The fourth open reading frame was homologous with Rhizobial FixC proteins. The N- terminal domain of the A. vinelandii fixC gene product was found also to contain a sequence homologous with the consensus sequence for an ADP binding site, as found in NAD ⁺or FAD dependent enzymes. In many FAD-containing enzymes ( e.g. lipoamide dehydrogenase and mercuric reductase), the FAD-binding site is located close to the N- terminus. This suggests, that the FixC protein might be a FAD-containing protein. It is known, that the β-subunit of ETF contains FAD. However, as no primary structure of this subunit has been published yet, it is not known if FixC has homology to the β-subunit of ETF.

The protein encoded by the fifth open reading frame is also homologous with ferredoxin I of A.vinelandii and also with the FixX proteins of various Rhizobia, all ferredoxin-like proteins. The A.vinelandii FixX protein is the only FixX protein, that contains the two cysteine motifs that are found in ferredoxin I and the FixP protein, whereas the FixX proteins of other nitrogen fixing bacteria lack the cysteine motif, involved in ligation of the [3Fe-4S] cluster.

The fact that A.vinelandii contains at least five different 7Fe-ferredoxins (FdI, FdN (7Fe-ferredoxin in the nif gene cluster), FdV (7Fe-ferredoxin in the alternative nitrogenase gene region), FixP and FixX) is indicative for a special function of these proteins in this organism. Recently, a hypothesis has been proposed by Thomson [1991], stating that the function of the 7Fe- ferredoxins in A.vinelandii is to regulate gene expression by binding to the DNA. This binding is controlled by the iron(II) levels and the redox state of the cell. When the 7Fe-ferredoxin-DNA complex binds iron(II), it becomes an 8Fe-ferredoxin. The affinity of the 8Fe-ferredoxin for the DNA is less than that of the 7Fe ferredoxin, so the ferredoxin no longer binds to the DNA, and mRNA synthesis is possible. The fact that A.vinelandii contains five genetically distinct 7Fe-ferredoxins could be clue for this model, whereas the existence of two different 7Fe-ferredoxins in what is most likely one operon, might give reason to assume that the fixPABCX cluster is involved in regulation of some kind of process, possibly involved in nitrogen fixation.

Recently, the fixABCX genes have also been found in the 0-2.4 min region of the Escherichia coli genome. E.coli is unable to grow diazotrophically. The E.coli fixABCX genes were followed by an open reading frame encoding a NAD(P)H dehydrogenase and preceded by a number of genes encoding proteins, involved in fatty acid metabolism. This might be a clue for a function of these genes in fatty acid metabolism, which could be of vital importance for nitrogen fixing organisms.

Downstream of the fixX gene, the start of a sixth open reading frame was found, but the N-terminal sequence did not show any homology to other proteins in the database.

A sequence motif with high homology to the promoter consensus for RNA polymerase complexed with sigma factor 54 (σ ⁵⁴) was found 63 bp upstream of the ATG start codon of the fixP gene. A putative binding site for the regulatory NifA protein (TGT-N ₉ -ACA) is found 164 bp upstream of the start codon. However, the spacing between the TGTand ACA- elements of this sequence is one base shorter than the consensus (TGT-N ₁₀ -ACA). The reason for this mismatch is not known, but it is known that a mismatch like this still functions in other organisms. No terminator sequence was found downstream of the stop codon of any of the genes.

It is concluded that amongst the fixPABCX genes of Azotobacter vinelandii at least three genes encode proteins, which are possibly involved in electron transport: FixB is highly homologous to the α-subunit of ETF and both FixP and FixX are homologous to A. vinelandi ferredoxin I. Whether these genes are actually involved in an electron transport system, fatty acid metabolism or whether they fulfil a function in the proposed regulation of gene expression in A.vinelandii, is the objective of the research described in chapter 3.

the fixPABCX genes of A.vinelandii: physiological analysis
Chapter 3 describes the construction and characterisation of mutants of A.vinelandii with alterations in the fixA, fixB, fixC and/or fixX genes. The gene of interest was exchanged with a plasmid derived copy that had been interrupted by insertion of the gene encoding kanamycin-resistance. A mutant lacking the fixABCX genes was constructed by replacing these genes by a DNA fragment containing the gene encoding kanamycin resistance.

The mutants were tested under a large number of conditions. All fix^-mutants showed normal growth characteristics in nitrogen-free medium under all conditions tested. In vivo and in vitro activities of acetylene reduction of mutants were comparable to wildtype activities. Growth on several sugars, dicarboxylic acids and fatty acids or amino acids was not different from wild type bacteria, which indicated that the fixABCX cluster is not obligatory in the catabolism of these components.

Our results indicate that there is a major difference between the fixABCX genes of various Rhizobia and those of A.vinelandii. In contrast to A.vinelandii, deletion of the genes in symbiotic nitrogen fixing organisms results in loss of both in vivo and in vitro nitrogenase activity. The finding, that in R.leguminosarum, the polypeptides from which the nitrogenase enzyme complex is composed, are present in FixA ^-, B ^-, and C ^-mutants but inactive, suggests that a step in the biosynthesis of active nitrogenase enzyme is hampered in these bacteria or that the proteins are inactivated by oxygen damage during growth. In A.caulinodans, a FixC ^-mutant still had 10% of wild type nitrogenase activity, which could be elevated to 36% by adding saturating amounts purified nitrogenase Fe protein. This is an indication that the fixC gene product is necessary for the maturation of nitrogenase of A.caulinodans. In A.vinelandii no similar function for the fixABCX genes could be demonstrated.

The hypothesis of Thomson that the 7Fe-ferredoxin of Azotobacter might be a DNA binding protein, involved in regulation of protein synthesis in response to iron(II) levels in the cell, could be an indication for the function of the fixPABCX genes in A.vinelandii. The fixPABCX gene cluster contains two 7Fe-ferredoxins, but evidence for the iron(II) dependent regulation has not been found.

the fixPABCX genes of A.vinelandii: promoter analysis.
The expression of the fixPABCX genes was investigated using two methods. A chromosomally integrated fixA :: lacZ gene fusion was made and it was observed that expression of the fixABCX genes of A.vinelandii was not significantly increased when cells, grown in the presence of ammonium, were transferred to a nitrogen free medium. It was concluded that the expression of the fixABCX genes, if occurring, was very low. From experiments, in which the promoter activity was investigated by immunological techniques, using antibodies against the purified FixA protein, similar observations were made. A very low signal of the FixA protein on the Western blots was found. Approximately half of this signal was found in cell extracts, grown in the presence of ammonia, and even in a FixA ^-mutant, a weak signal was detected. This signal was probably caused by a-specific binding of the antibodies, since 300 μg of total cellular protein was loaded in one slot of the gel.

It cannot be ruled out, that downstream of the fixX gene, one or more genes are located that are co-transcribed with the fixPABCX genes. A mutation in the fixPABCX genes might cause polar effects on the downstream genes. The fact however, that no effect of any of the mutations was found, is either evidence that no polar effect is present, or that the gene(s) downstream of the fixX gene is/are a negative regulatory gene(s).

It is concluded, that, the fixABCX genes, which are of vital importance for nitrogen fixation in symbiotic organisms, are not essential for nitrogen fixation in A.vinelandii. Despite all investigations the function of the fixABCX genes is not known.

Electron transport to nitrogenase: biochemical investigations.
Biochemical investigations on the electron transport to nitrogenase are the subject of chapter 4. In order to elucidate the electron pathway to nitrogenase, a model system was used, in which the flavodoxins were replaced by artificial low potential electron carriers. The respiration and the reduction of viologens by different substrates, catalysed by Azotobacter vinelandii cytoplasmic membranes was investigated. Only with NADH, viologen oxidoreductase activity could be detected; NADPH, malate, succinate and lactate were unable to reduce viologens. From the oxygen consumption experiments with different substrate combinations it is concluded, that NADH and NADPH are oxidised by different dehydrogenases, although they have the same output site in the respiratory chain towards ubiquinone. Malate dehydrogenase on the other hand, has a different output site to ubiquinone than the dehydrogenases that oxidise NADH and NADPH.

The kinetic parameters of the NADH:ubiquinone oxidoreductase of the respiratory chain were investigated and compared with the NADH:ferricyanide oxidoreductase activity, which is an activity of complex I of the respiratory chain. In contrast to the NADH:ferricyanide oxidoreductase activity, the NADH:viologen oxidoreductase activity did not show double substrate inhibition, which indicates that the viologen reducing site of the NADH dehydrogenase complex is different from the NADH binding site. EPR studies on the presence of paramagnetic centers in cytoplasmic membranes demonstrated a 1owpotential" electron accepting site, which could only be reduced using dithionite. This indicates that either the site is not accessible for NADH, or the redox potential is too low to enable NADH to reduce this site.

Viologen reduction did not only take place under conditions of an inhibited respiratory chain, but also under aerobic conditions with an active respiratory chain. This shows that electrons from NADH are transferred to the viologens, when at the same time electrons from NADH are transferred through the respiratory chain to oxygen.

The NADH:viologen oxidoreductase activity was modulated by using different viologens and by changing the viologen concentration. It was observed that the hydrogen peroxide formation increased linearly with the NADH:viologen oxidoreductase activity. It was also observed that maximally 50% of the electrons from NADH were transferred to the viologens and that the NADH-dehydrogenase activity (modulated by the NADH/NAD ⁺) had no influence on the distribution over viologens and ubiquinone.

The results of the experiments can be used as an example for a model, as proposed by Haaker and Klugkist [1987]. Some modifications should be made to update the model to the current knowledge. In the model, a NADPH dehydrogenase is the central part, which, according to the results of this work, should be altered to a NADH dehydrogenase, since no viologen reduction was observed when NADPH was used as electron donor. No statements can be made to whether or not the 29kDa protein is involved in the reduction of low potential mediators. The concomitant flow of electrons through the respiratory chain and electron transport to low potential redox mediators, the central dogma of the model of Haaker and Klugkist, is shown in this work. This supports the observation of Klugkist et al. [1986], that electron transport to nitrogenase and respiration are coupled. According to the revised model, shown in figure 1, two electrons from NADH are accepted by the FMN group of the NADH dehydrogenase, operating at -320mV. During respiration, these electrons are subsequently distributed over respiration and a route, leading to the reduction of viologen. In the presence of cyanide and oxygen, the electrons can only be directed to the viologen reducing cluster. The viologens are oxidised efficiently by dioxygen to form H ₂ O ₂ , thereby maintaining a high concentration of oxidised viologen. The reduced FMN group of the NADH dehydrogenase can also be oxidised by ferricyanide.

The electron transport pathways as suggested in Figure 1 explain why viologen reduction and respiration are coupled and why not more than 50% of the electrons from NADH are used for viologen reduction. Unfortunately, the formation of H ₂ O ₂ was not observed when the viologens were replaced by purified flavodoxin. The fact that flavodoxin reduction is not observed under the conditions applied, could be an indication

for the existence of a factor, which is absent under the experimental conditions. This factor could be an oxygen sensitive and/or a soluble protein. A possible candidate for an O ₂ sensitive, membrane bound protein could be the 29kDa protein, found by Klugkist and coworkers. It is also possible that this factor is present in a complex of proteins, which is lost apart during the isolation of the membranes. Future research is required to investigate the possibility to link flavodoxin reduction to respiration.