TEXT

Top Abstract Text References

Rhizobial FixK proteins belong to the large and phylogenetically widespread family of FNR- and CRP (CAP)-like proteins of bacteria (8, 27). First identified in Sinorhizobium (Rhizobium) meliloti (4), FixK has been discovered in all rhizobial species examined, and additional copies or homologs of the corresponding gene exist in certain species (e.g., see references 11 and 21). The FixK proteins are transcriptional regulators that usually function as activators of their target genes. They bind to a conserved, symmetric nucleotide sequence, 5'-TTGA(N₆)TCAA-3' (the FixK box [8, 28]), whose axis of symmetry is located between positions 41 and 40 upstream of the transcriptional start site of the regulated genes. FixK in S. meliloti is part of a regulatory cascade in which the membrane-bound hemoprotein FixL senses a low oxygen concentration, phosphorylates itself, and then transfers the phosphate moiety to the response regulator FixJ, which finally activates the expression of fixK (3). The FixK protein then activates a number of genes or operons, including the fixNOQP and fixGHIS operons, which are responsible for the synthesis of a high-affinity terminal oxidase for respiration under microaerobic conditions, such as in legume root nodules (3, 12). FixK negatively regulates its own expression in S. meliloti by activating the fixT gene, whose product inhibits the FixLJ system (10).

The mechanisms by which fixK genes are regulated and the nature and number of genes controlled by FixK differ by various extents in different rhizobial species (8). For reasons of space restrictions, these differences are not discussed in this article. A particularly puzzling case has been described previously for the nitrogen-fixing soybean symbiont Bradyrhizobium japonicum (1, 2). This species possesses a FixLJ-dependent fixK gene (here called fixK₁) which, when mutated, does not cause the negative nitrogen fixation (Fix) and anaerobic nitrate respiration (NR) phenotypes that are characteristic for B. japonicum fixLJ mutants. Constitutive overexpression of fixK₁ in a fixJ mutant background, thus bypassing the FixJ dependency, partially restored NR activity (2) and completely restored Fix activity (18). It thus appeared as if the FixK₁ protein could substitute for a FixLJ-dependent function. One interpretation of this result was that, in the wild-type B. japonicum, FixLJ might regulate a second fixK-like gene whose product would be the real activator for some nitrogen fixation and nitrate respiration genes and which would also activate fixK₁, a gene that itself is not involved in these activities. This would explain why constitutive fixK₁ expression could compensate for the lack of expression of such a second fixK-like gene in a fixJ mutant. We report here that these assumptions turned out to be correct.

Identification, sequencing, and mutational analysis of a second B. japonicum fixK-like gene, fixK₂. Using either B. japonicum fixK₁ or S. meliloti fixK DNA fragments as probes, a second fixK-like gene (here called fixK₂) was discovered in Southern blot hybridizations to previously isolated DNA clones of the B. japonicum fixLJ-open reading frame 138 (ORF138) region (1). Thus, fixK₂ was identified as part of the so-called fix cluster III (8) (Fig. 1A). The gene was sequenced on both DNA strands. The putative fixK₂ start codon was located 104 bp downstream of the ORF138 stop codon (Fig. 2A), and the fixK₂ ORF had a length of 696 bp and encoded a protein of 232 amino acids. The positional amino acid sequence identity between the deduced FixK₂ and FixK₁ proteins (Fig. 2B) is 34%. This identity is less than that between FixK₂ and the FixK proteins of Azorhizobium caulinodans (44% [14]) and S. meliloti (40% [4]), therefore suggesting a function for FixK₂ that is distinguishable from the function of FixK₁. The difference between FixK₂ and FixK₁ is further emphasized by the fact that not only the cysteine-rich domain present at the FixK₁ N terminus but also cysteine 114, both rendering this type of protein oxygen sensitive (2, 15, 27), is missing in FixK₂ (Fig. 2B). Each of these proteins contains near its C terminus a helix-turn-helix motif potentially involved in DNA binding (Fig. 2B).

View larger version (21K): [in this window] [in a new window]   FIG. 1.   B. japonicum fixK2 gene and fixK2 mutations. (A) Map of fix gene cluster III showing the fixK2 locus at the left end. The insertion in mutant 9043 containing the streptomycin-spectinomycin resistance gene ( str, spc) is marked. The FixJ box in front of fixK2 is symbolized with a closed square, and the FixK boxes in front of the fixNOQP and fixGHIS operons are marked with open squares. The combined 13,018-bp DNA sequence of the entire cluster III, including the newly sequenced fixK2 region (open bar), was deposited in the EMBL/GenBank database under accession no. AJ005001. (B) Genomic structure of strain 9039K2, which carries the same fixK2 mutation as strain 9043 plus an insertion in fixJ. This inserted fragment carries fixK1 transcriptionally fused to the bleomycin resistance gene of the Tn5-derived aphII-ble-str operon. Constitutive expression of the fixK1 gene is thus controlled by the promoter of the kanamycin resistance gene aphII. B, BamHI; E, EcoRI; S, SalI; X, XhoI.

View larger version (44K): [in this window] [in a new window]   FIG. 2.   Sequence analyses. (A) Nucleotide sequence of ORF138-fixK2 intergenic region showing the putative ribosome binding site sequence (rbs), the transcription start site (indicated by the arrow at position +1), and the conserved FixJ binding site motifs at positions 35 and 65 (indicated by boldfaced underlining). The EcoRV site used to construct deletion clone pRJ9053 (see the text) is designated. (B) Amino acid sequence alignment of the FixK1 and FixK2 proteins. Identical amino acids are connected by vertical lines. The stars denote the essential cysteines characteristic for the oxygen-responsive FNR protein class. The probable DNA binding domain (helix-turn-helix motif) is underlined.

The fixK₂ gene is regulated positively by FixLJ and negatively by its own product. A translational fusion of lacZ was constructed within the 43rd fixK₂ codon (at an EcoRI site [Fig. 1A]) and integrated at the chromosomal fixK₂ locus of the B. japonicum wild type and the fixJ, fixK₁, and fixK₂ mutants. For this purpose, the fusion was first constructed in vector pNM480B and then transferred into pSUP202 for cointegration into the chromosome (18). -Galactosidase (-Gal) activity expressed from the reporter fusion was determined in cells grown aerobically or microaerobically as described previously (1, 2). Table 1 shows that fixK₂ expression is induced about 10-fold in microaerobically grown cells compared with aerobically grown cells and that this induction requires FixJ but not FixK₁. A very strong induction (60-fold) in the fixK₂ mutant suggests that FixK₂ negatively regulates expression of its own structural gene in the wild type. Whether this autoregulation is direct or indirect, e.g., via activation of a repressor gene, is not known. In this context, it is of interest that ORF138 (Fig. 1), a gene for a FixJ homolog lacking the C-terminal DNA-binding domain (1), is induced (sixfold) under microaerobic growth conditions and that this induction partly depends on FixK₂ but not FixJ (18). More work is needed to explore the possibility that the ORF138 protein is a player in the negative autoregulatory circuitry involving FixK₂.

FixK₂ activates the fixK₁ gene. A previously constructed translational fixK₁'-'lacZ fusion (2) was integrated into the chromosomes of the B. japonicum wild type and the fixJ, fixK₁ and fixK₂ mutants at the fixK₁ locus for measurements of -Gal activity in aerobically and microaerobically grown cells (Table 1). Appreciable -Gal activity was produced only in microaerobically grown cultures of the wild type and the fixK₁ mutant, whereas there was background activity in the fixJ and fixK₂ mutants. The FixLJ dependency of fixK₁ expression has already been shown previously (2); however, this effect must have been indirect, since in the present study we demonstrated an involvement of the FixJ-dependent fixK₂ gene in this control. In fact, a perfect FixK box (5'-TTGATCTGGGTCAA-3'), present at a proper distance (between positions 41 and 40, at the axis of symmetry) from the transcription start site of fixK₁ (2, 18), might serve as the binding site for FixK₂. To support this idea, we tested plasmid-encoded fixK₁'-'lacZ fusions in microaerobically grown wild-type cells. A plasmid (pRJ9025) with 96 bp of DNA present upstream of the transcription start site (up to a BamHI site [2]) produced 58 U of -Gal activity, whereas a deletion clone (pRJ9056) with its 5' deletion end point at position 33 produced only 1 U of -Gal activity. This corroborates that the conserved FixK box is an important part of the fixK₁ promoter region. All lines of available evidence now suggest the existence of a regulatory cascade in which FixJ first activates the fixK₂ gene, whose product is then the activator of the fixK₁ gene (Fig. 3).

View larger version (23K): [in this window] [in a new window]   FIG. 3.   The crucial position of B. japonicum FixK2 as a distributor of the low-oxygen-level signal transduced through FixLJ. See the text for further explanations. Abbreviations and symbols: cm, cytoplasmic membrane; i, inside; o, outside; P, phosphate moiety; +, positive control; , negative control; , FixJ box.

Other FixK₂-regulated B. japonicum genes. In the course of studying symbiotically relevant B. japonicum genes that are induced at a low oxygen tension, we found several that are most likely controlled by the FixLJ-FixK₂ system (Fig. 3). The evidence for this inference that has been obtained is discussed below briefly for each individual gene or operon.

(i) rpoN₁, one of two ⁵⁴ genes. Having found a FixLJ dependency for rpoN₁ expression previously (16), we showed in the present study that this expression was mediated by FixK₂ since a chromosomally integrated rpoN₁'-'lacZ fusion was not expressed in a fixK₂ mutant background (Table 1). A nearly perfect FixK box (5'-TTGCGCGACATCAA-3' at position 75 relative to the rpoN₁ start codon [16]) is present in the region upstream of rpoN₁, but unfortunately its precise distance to the transcription start site is not known because, despite repeated attempts, we failed to map it. However, the importance of the FixK box was demonstrated by testing upstream deletion derivatives of a plasmid-borne rpoN₁'-'lacZ fusion. A plasmid (pRJ8042) with a 5' deletion end point in the middle of the FixK box produced only background -Gal activity (4 U), whereas a clone (pRJ8041) with an additional 21 bp of upstream DNA allowed for full expression (287 U) in microaerobic culture.

(ii) fixNOQP operon, coding for a high-affinity cbb₃-type oxidase complex. A translational lacZ fusion to the sixth codon of fixP, the last gene of the fixNOQP operon (23, 25), was constructed and chromosomally integrated into the B. japonicum wild type and the fixJ mutant. A clear FixJ dependency was found for fixP'-'lacZ expression (Table 1). Although expression in the fixK₂ mutant was not tested, regulation by FixK₂ is likely in view of the optimal FixK box (5'-TTGATTTCAATCAA-3') that is present at a perfect distance (between positions 41 and 40, at the axis of symmetry) upstream of the fixN transcription start site (22, 23).

(iii) fixGHIS operon. These four genes code for a redox-coupled cation transport system (P-type ATPase) that is essential for the biogenesis of the cbb₃-type oxidase (13, 24). Therefore, coregulation with the fixNOQP operon would appear sensible. In fact, the fixG transcription start site that was mapped by primer extension (24) is preceded at an appropriate distance by a FixK box (5'-TTGAGCTGGATCAA-3'). No primer extension products were detected in fixJ and fixK₂ mutant backgrounds (22).

(iv) Heme biosynthesis genes. Page and Guerinot (20) reported a FixLJ-dependent microaerobic induction of the -aminolevulinic acid synthase gene (hemA) and demonstrated that this induction required a functional FixK box (5'-TTGATCGGGATCAA-3') at a proper distance from the transcription start site. Likewise, the hemB gene for the next enzyme of the heme biosynthesis pathway, -aminolevulinic acid dehydratase, is induced under conditions of oxygen limitation in a FixJ-dependent manner (6), but evidence for the involvement of FixK₂ is not yet available. Recently we cloned a hemN-like gene, coding for a putative coproporphyrinogen III dehydrogenase, and obtained preliminary results suggesting a partial regulation by FixK₂ (9). In general, oxygen limitation in B. japonicum triggers an increase in cytochrome synthesis which is accompanied by an increased demand for heme (19).

(v) Nitrate respiration genes. As shown previously (1) and in this work (see above), B. japonicum fixLJ and fixK₂ mutants are defective in anaerobic growth with nitrate as the terminal electron acceptor, suggesting that some critical genes for nitrate reduction or for the entire denitrification pathway are subject to control by the FixLJ-FixK₂ cascade. An attractive target gene for further studies is the cycA gene for cytochrome c₅₅₀, because cycA was also shown to be essential for nitrate respiration (5). Furthermore, B. japonicum DNA sequences for a nitrite reductase gene (nirK) and some genes for N₂O respiration (nosZDF) have been deposited in the EMBL database (accession no. AJ002516 and AJ002531); these genes should now be amenable for regulatory studies. The involvement of FixK- or FNR-like regulators in the control of denitrification genes of nonrhizobial denitrifiers is well documented (29).

Concluding remarks. The existence in bacteria of a gene expression cascade with three transcriptional regulators interconnected in series (FixJ-FixK₂-FixK₁ or FixJ-FixK₂-RpoN₁ [Fig. 3]) is quite remarkable. The physiological reason for such complex hierarchies is not fully understood, even though their logic is persuasive. One function of the sophisticated cascade might be to sense oxygen gradients, since not only FixL but also FixK₁ is an oxygen sensor. In that case, the threshold level for efficient response of FixK₁ ought to be at a lower oxygen concentration than that for FixL, and the autoregulatory FixK₂ system would be in an optimal strategic position to fine-tune between these values. Likewise, a gradual lowering of the ambient oxygen concentration would trigger a gradual increase in the synthesis of ⁵⁴ (RpoN), which is needed in the extremely oxygen-sensitive NifA-dependent activation of nitrogenase gene expression (8). The FixK₂ protein may thus be regarded as a distributor and as an amplifier or silencer of the input signal arriving at FixL (Fig. 3). It makes sense that activation of the genes for the high-affinity cbb₃-type oxidase proceeds directly via FixLJ-FixK₂, apparently without an additional oxygen-sensitive switch such as FixK₁ or NifA, because this gives cells a chance to exploit intermediate to low oxygen concentrations for respiration. The same rationale might apply for the hem genes. Too little is currently known about the complete set of transcription factors needed in the control of B. japonicum denitrification genes. It is intriguing to learn from studies with other denitrifiers that up to three FNR-like proteins may be employed in this process (29).

Nucleotide sequence accession number. The nucleotide sequence reported here has been submitted to the EMBL/GenBank database and assigned accession no. AJ005001 (see the legend to Fig. 1 for further details).