Partial Characterization of Nif Genes from the Bacterium Azospirillum Amazonense

Azospirillum amazonense revealed genomic organization patterns of the nitrogen fixation genes similar to those of the distantly related species A. brasilense. Our work suggests that A. brasilense nifHDK, nifENX, fixABC operons and nifA and glnB genes may be structurally homologous to the counterpart genes of A. amazonense. This is the first analysis revealing homology between A. brasilense nif genes and the A. amazonense genome. Sequence analysis of PCR amplification products revealed similarities between the amino acid sequences of the highly conserved nifD and glnB genes of A. amazonense and related genes of A. brasilense and other bacteria. However, the A. amazonense non-coding regions (the upstream activator sequence region and the region between the nifH and nifD genes) differed from related regions of A. brasilense even in nitrogenase structural genes which are highly conserved among diazotrophic bacteria. The feasibility of the 16S ribosomal RNA gene-based PCR system for specific detection of A. amazonense was shown. Our results indicate that the PCR primers for 16S rDNA defined in this article are highly specific to A. amazonense and can distinguish this species from A. brasilense.


Introduction
Bacteria belonging to the genus Azospirillum are found associated with forage grasses, cereals and agriculturally important crops from various geographical origins.This genus consists of seven species, A. lipoferum, A. brasilense, A. amazonense, A. halopraeferens, A. irakense (1), A. largomobile (2), and A. doebereinerae (Eckert B, Baller-Weber O, Kirchhof G, Halbritter A, Stoffels M and Hartmann A, unpublished results, and GenBank accession number AJ238576).Worldwide inoculation experiments using a variety of crops have shown that Azospiril-lum species can contribute to the nitrogen economy (3,4).In addition, experiments with sugarcane and other crops suggest either a synergistic or an additive effect when A. amazonense is used in combination with other diazotrophics as inoculant (5).
A. amazonense differs in several important characteristics when compared to other Azospirillum species.Its nitrogenase activity has lower oxygen tolerance, its ability to use sucrose as carbon source is different, and the most remarkable difference is its sensitivity to alkaline pH (6).The genetic relationship existing between A. amazonense and other Azospirillum species was analyzed by 16S rDNA restriction fragment length polymorphism (7).The phylogenetic analysis, based on 16S rDNA sequences, confirmed that A. amazonense and A. irakense form one cluster and that the closely related species A. brasilense and A. lipoferum form a second cluster together with A. halopraeferens (8).
During the last few years, nitrogen fixation (nif) genes have been isolated from a variety of diazotrophic organisms.In A. brasilense a DNA fragment of 25,100 bp encompassing the nif region was shown to contain the nifHDKORF1Y (9), the ORF2nifUSVORF4 (10), an mcp-like gene, the nifENXORF3,5, fdxAnifQ, and the fixABCX (Gross J, Vedoy C and Schrank IS, unpublished results) transcriptional units.However, to date, few genes were isolated from A. amazonense and none of them could be related to the biological nitrogen fixation process.
We report here the application of PCR and Southern hybridization to characterize some of the nif genes from A. amazonense.We have also analyzed the amplification products from two conserved regions within the nifD and glnB genes.Moreover, we have developed primers that specifically amplify A. amazonense DNA and allow us to distinguish between this species and A. brasilense.

DNA manipulation and sequence analysis
All DNA manipulations were performed using standard techniques (12) and instructions provided by suppliers of material, enzymes or reagents.Total DNA extraction of A. amazonense was performed as previously described (13) and Southern blot analysis was performed with the ECL Direct Nucleic Acid Labeling and Detection Systems (Amersham Pharmacia Biotech, Uppsala, Sweden).
The nucleotide sequence determination was performed by the chain-termination method of Sanger et al. ( 14) using 33 P-dNTPs and the ThermoSequenase radiolabeled terminator cycle sequencing kit (Amersham).The PCR products were purified using the GFX PCR kit from Amersham.Analysis of DNA sequences and comparison with nucleotide and deduced protein sequences from other organisms were performed using the GCG (Wisconsin Package Version 9.0, Genetics Computer Group, Madison, WI, USA) computer programs (licensed to CENARGEN-EMBRAPA, Brasília, DF, Brazil).

Amplification conditions
PCR amplification of the target sequences was performed using a DNA thermal cycler (MJ Research, Waltham, MA, USA).The PCR mixture contained the reaction buffer (50 mM KCl, 10 mM Tris-HCl, pH 8.3, 2.5 mM MgCl 2 ), 200 µM of each dNTP, 30 pmol of each primer, 1 U of Taq polymerase (CenBiot enzymes, Centro de Biotecnologia, UFRGS, Porto Alegre, RS, Brazil), template DNA, and distilled water to a final concentration of 25 µl.The reaction mixture was subjected to PCR under the following conditions: heat denaturation at 94 o C for 5 min and then an additional 35 cycles with heat denaturation at 94 o C for 30 s, annealing (at a temperature defined for each set of primers, see Table 2) for 30 s, and DNA extension at 72 o C for 30 s.After the last cycle, samples were maintained at annealing temperature for 5 min followed by 72 o C for 10 min.PCR products were analyzed by gel electrophoresis (12).Primers listed in Table 2 were purchased from Oligo ETC.& Oligo Therapeutics Inc. (Wilsonville, OR, USA).

A. amazonense nif genes: hybridization and PCR
In order to understand the molecular organization of some of the nif genes from A. amazonense and to establish their relationship with the well-known nif genes from A.
brasilense, we used two different approaches.
Initially, DNA fragments from different nif operons already characterized in A. brasilense were used in Southern blot hybridization to analyze the relatedness between the two Azospirillum species.The hybridization patterns of A. amazonense DNA are shown in Figure 1.The nif structural genes from A. amazonense are localized in a 6.5-kb EcoRI DNA fragment (Figure 1A) as previously found for A. brasilense genome (13).In A. brasilense and other diazotrophic bacteria the nifHDK genes are organized in a single transcriptional unit (9).To determine whether the A. amazonense nif genes are clustered in a similar manner as found in A. brasilense and other nitrogen-fixing organisms, we used sequences from nif genes (Table 1) from A. brasilense as DNA hybridization probes.Three different probes, which consist of the entire nifHDK operon (data not shown), the nifH gene alone (data not shown) from A. brasilense, and a region of the nifD gene (Table 1) from A. amazonense, showed homology with the same DNA fragment from A. amazonense (Figure 1A, lane 1E).This result suggests that the region comprising a 6.5-kb EcoRI DNA fragment probably contains the nifHD homologous in A. amazon-ense and that the genes are organized in a single operon.Nevertheless, the hybridization pattern of the nifD gene amplified from A. amazonense revealed a different restriction pattern when A. amazonense and A. brasilense total DNA was digested with PstI and SalI restriction enzymes (Figure 1A, lanes 1P and 1S).A hybridization signal specific to SalI DNA fragments was obtained when A. amazonense total DNA was hybridized with A. brasilense nifENX and fixABC probes, respectively (Figure 1B, lane 1 and Figure 1C, lane 1).
In order to understand the nature of the A. amazonense genes that regulate nitrogen fixation we have isolated and characterized the nifA and glnB genes, which are responsible for the regulation of other nif genes and operons in A. brasilense (15,16).The presence of genes homologous to glnB and nifA was detected by hybridization of A. amazonense DNA with two different probes (Table 1).The entire nifA gene from A. brasilense is present within a 2.5-kb SalI DNA fragment (15).Heterologous hybridization revealed that the nifA homologue in A. amazonense is also found in a SalI DNA fragment of approximately 3 kb (Figure 1D, lane 1).
The glnB homologous gene was local- ized within a 4.5-kb PstI DNA fragment from A. amazonense (Figure 1E, lane 1P).Faint hybridization signals were detected in A. amazonense total DNA and also with PstI-digested A. brasilense total DNA (Figure 1E, lane 2).Genes homologous to glnB have been found in A. brasilense, Herbaspirillum seropedicae and other bacteria, suggesting the presence of two copies of glnBlike genes in these organisms (17,18).Since no PstI restriction site was found within glnB or glnZ genes from A. brasilense or in the glnB gene from A. amazonense, the presence of two hybridization fragments in both strains may represent similarity between the probe and the two glnB-like genes.
The results presented here are the first to reveal homology between A. brasilense nif genes and A. amazonense total DNA.Taken together, these results suggest that the nifHDK, nifENX and fixABC operons and the nifA and glnB genes may be structurally homologous to A. amazonense counterpart genes.
The second approach to the understanding of the nature of nitrogen fixation in A. amazonense was based on PCR amplification of regions within the nif/fix genes.Several primers were selected from the sequences of the nif/fix genes from A. brasilense.The primers listed in Table 2 represent, except for nifA, those that revealed positive and conclusive results after PCR amplification and DNA sequencing.Three sets of primers were designed from regions within the A. brasilense nifHDK operon.The promoter region of this operon shows two overlapping upstream activator sequences (UAS) as the only potential NifA-binding sites (19).To date this organization seems to be unique to A. brasilense and may be an atypical NifAbinding site interacting with two dimers of NifA, as proposed to occur in other nif promoters (20).UAS primers define a region of approximately 200 bp in the nifH promoter sequence encompassing the two overlapping UAS.The second pair of IGR primers is derived from the intergenic region between nifH and nifD and amplifies a 310-bp region.In A. brasilense three inverted repeat structures have been found downstream from the nifH stop codon (9).Transcription analysis revealed the presence of one 1.1-kb transcript corresponding to the nifH gene (21).The nifD primer pair is specific for a 710-bp region of the nifD gene containing a highly conserved amino acid sequence and a serine residue which is present only in the nifD gene of A. brasilense (9).Amplification of A. amazonense DNA with the above selected primers (Table 2) resulted in fragments of the predicted size (Figure 2A,B,C).All three amplified DNA fragments were used to probe total A. amazonense DNA, revealing homology with the same EcoRI fragment encompassing the nifHD genes (Figure 1A).To further characterize the PCR products, all DNA fragments were purified and sequenced.The sequences for the UAS and IGR products showed very little similarity to the A. brasilense counterparts (data not shown).Only the region amplified within the nifD gene is similar to other nifD gene sequences (discussed below).These results indicate that the A. amazonense non-coding regions described above differ from related A. brasilense regions even among nitrogenase structural genes which are highly conserved among diazotrophic bacteria.
A set of primers was designed (Table 2) from previously published sequences of nifA and glnB genes from A. brasilense (15,22).A PCR product of approximately 250 bp was detected in A. amazonense DNA (Figure 2D, lane 1).This product hybridized with total A. amazonense DNA (Figure 1E, lanes 1P and 1S) and, after DNA sequencing, revealed homology with P II proteins from different organisms (discussed below).Surprisingly, no amplification product was visualized when nifA target primers were applied to A. amazonense DNA although these primers were designed on the basis of a highly conserved region found among NifA proteins from A. brasilense, A. lipoferum and related bacteria.This unexpected result suggests that the A. amazonense nifA gene may have greater differences in DNA sequence than found among other nifA genes previously characterized.

Amplification of A. amazonense 16S rDNA
To overcome problems with misinterpretation of the PCR results, a PCR amplification system was developed to specifically detect A. amazonense.The 16S rRNA sequence from Azospirillum species was obtained from the GenBank database.Multiple sequence alignments to other 16S rRNA gene sequences revealed one region of considerable sequence divergence from the closest relatives, as shown in Figure 3. Two primers (16S-up and 16S-do) were constructed, one based on the variable region and the other based on a lower sequence divergence region (Table 2).The PCR product generated with A. amazonense was 400 bp long (Figure 2E, lane 2) and was confirmed by direct sequencing (Figure 3).No PCR product was detected when A. brasilense DNA was the template (Figure 2E, lane 3).Using this methodology we could specifically detect and differentiate DNA templates from these two Azospirillum species.

Interspecies conservation of nif/gln genes
A sequence comparison of the A. amazonense and A. brasilense nifD genes revealed a high degree of similarity at both the DNA and amino acid levels (Figure 4).One relevant difference between A. amazonense and A. brasilense NifD proteins is located at position 275 (relative to A. brasilense NifD sequence; 9).The A. brasilense nifD gene product contains a total of five cysteine residues, four of which are highly conserved   among the other nifD genes with respect to both positions and adjacent sequences (9).Of the five conserved cysteine residues proposed to act as ligands to the iron-clusters of dinitrogenase, the cysteine at position 275 is substituted by a serine on the A. brasilense predicted sequence (Figure 4) (9).The highly conserved nifD region amplified from A. amazonense shows a cysteine residue at position 275 (Figure 4), suggesting that the substitution observed in the A. brasilense nifD gene is not conserved among other Azospirillum species and appears to be unique among diazotrophic bacteria.The regulation of nitrogen fixation in A. lipoferum and A. amazonense is less well understood than in A. brasilense.The regulation is controlled both transcriptionally and post-translationally in A. brasilense and A. lipoferum (23).In A. brasilense, the expression of the nifHDK operon is positively regulated by NifA.In this bacterium nifA is expressed under conditions both compatible and incompatible with nitrogen fixation.NifA activity is modulated by the P II protein (encoded by glnB), the intracellular signal transmitter, in response to the nitrogen status of the cell.The glnB gene of A. lipoferum was isolated, but the P II protein has not yet been characterized (24).The isolation of part of the glnB gene from A. amazonense is the first evidence that the nitrogen regulatory path-* way of this microorganism may be similar to that of other Azospirillum species.The P II protein from A. amazonense revealed a high level of similarity when compared to P II from other diazotrophic bacteria (Figure 5).
The occurrence of duplicate copies of genes encoding P II -like proteins now appears to be common among members of the = (Azospirillum) and C (Klebsiella) subdivisions of the Proteobacteria class.In A. brasilense, P II and P Z proteins are involved differently in nitrogen-dependent regulation of various physiological functions (25).The P II amino acid sequence is about 60% identical to that of P Z and could explain the detection of two hybridization signals in A. brasilense and A. amazonense Southern blots (Figure 1E, lane 1S and lane 2).
In this report, we present distinct lines of evidence showing that the nif gene organization and regulation in A. amazonense differ, in some aspects, from those of the best characterized A. brasilense.Although our results indicate the presence of counterparts of genes nifHDK, nifENX, fixABC, nifA, and glnB in A. amazonense, we found differences in restriction sites and non-coding sequences suggesting that the organization of these genes may differ from that of the related A. brasilense.These results support the suggestion by Fani et al. (8) that A. amazonense isolates may be members of a taxonomic cluster that is clearly distinct from the closely related A. brasilense and A. lipoferum species.(26,27).

Figure 1 .
Figure 1.Hybridization pattern of Azospirillum amazonense total DNA with EcoRI (1E), PstI (1P), and SalI (1S) in A, SalI in B, C and D, and PstI (1P) and SalI (1S) in E with the following probes: nifD in panel A; nifEN in panel B; fixABC in panel C; nifA in panel D, and glnB in panel E. Lane 2 represents A. brasilense total DNA digested with PstI in A, C and E, or Sal I in B and D. Lane M contains l HindIII as molecular marker.

Figure 2 .
Figure2.Specific PCR for amplification of nif genes from Azospirillum amazonense.Amplified products on 1.5% agarose gel were visualized with a UV transilluminator after ethidium bromide staining using the following primers: panel A, UAS; panel B, IGR; panel C, nifD; panel D, glnB, and panel E, rRNA from A. amazonense (lane 2) and A. brasilense (lane 3) total DNA.The negative control (lane 1) contains no template DNA in the reaction mixture.Lane M contains the 100-bp DNA ladder (Gibco/BRL).

Figure 3 Figure 4 .BFigure 3 .
Figure 3 continued on the next page

Figure 5 .
Figure 5. Azospirillum amazonense glnB partial deduced amino acid sequence alignment with A. brasilense (GenBank accession No. X51499), Rhodobacter sphaeroides (GenBank accession No. AF032116), R. capsulatus (GenBank accession No. U25953), Rhodospirillum rubrum (GenBank accession No. X84158), Azorhizobium caulinodans (GenBank accession No. Y10213), Rhizobium meliloti (GenBank accession No. U50385), Bradyrhizobium japonicum (GenBank accession No. M26753), and Herbaspirillum seropedicae (GenBank accession No. U86073) P II proteins.A black background indicates conserved residues in all aligned sequences, a dark grey background indicates conserved residues in at least 80% of the aligned sequences, and a light grey background indicates conserved residues in at least 6% of the aligned sequences.Multiple alignments were done with the PILEUP program, University of Wisconsin Genetics Computer Group, and the alignment editing was done with the GENDOC program considering the Dayhoff PAM 250 score table(26,27).