Isolation of glnB and glnK genes from A. amazonense
The PII proteins are pivotal regulators of the nitrogen metabolism, controlling the activities of transporters, enzymes and transcriptional factors implicated in this process [9, 10]. These proteins are highly conserved and are widely distributed throughout prokaryotes [11]. In Proteobacteria in particular, there are two main types of PII proteins, GlnB and GlnK. In this work, two PII protein encoding genes from A. amazonense were isolated. Southern blot analysis utilizing a PCR-generated glnB fragment as the probe revealed two distinct signals in the genomic DNA of A. amazonense digested with SalI: the strongest at the ~2 kb DNA fragments and the weakest at the ~3 kb DNA fragments (data not shown). Based on these results, a genomic library enriched with 2-3 kb SalI fragments was constructed. The library was partially sequenced and a PII protein homolog was identified. The deduced amino acid sequence of this gene was found to be highly similar to that of the GlnZ proteins (GlnK-like homologs) from A. brasilense and Azospirillum sp. B510 (75% identity and 86% similarity), and Rhodospirillum. centenum (73% identity and 86% similarity). Arcondéguy et al. (2001) [12] suggested that the glnZ genes should be termed glnK, since their deduced proteins are highly similar to the GlnK proteins. Furthermore, there is a functional correspondence between these proteins, as both regulate the uptake of ammonium through the AmtB transporters [13–15]. Therefore, we adopted the glnK designation for this A. amazonense homolog, mainly because this nomenclature could facilitate comparisons between other bacterial systems.
The glnK gene from A. amazonense is flanked by the aat gene in the downstream region, which codes a putative aspartate aminotransferase and the ubiH gene in the upstream region, which codes an enzyme implicated in ubiquinone biosynthesis (Figure 1). This genetic organization resembles that found in other species from the Rhodospirillales order, namely A. brasilense, Azospirillum sp. B510 and R. centenum.
Since the glnB gene was not found in the genomic library, the Inverse PCR methodology was carried out to isolate this gene. A ~2 kb amplicon that contained the glnB gene was obtained (data not shown). It was found that the protein of this gene displays 92% identity and 98% similarity to the GlnB proteins from Azospirillum sp. B510 and A. brasilense, and 96% identity and 98% similarity to the GlnB protein of R. centenum. The glnB gene is located upstream of the glnA gene (glutamine synthetase), the same genetic context observed in these bacteria (Figure 1).
In A. brasilense, glnB has a key role in nitrogen fixation because its protein product regulates the activity of NifA, the transcriptional factor of nitrogen fixation [16, 17].
Furthermore, both of the GlnZ (GlnK-like homolog) and GlnB proteins are also implicated in the DraT/DraG system, which regulates dinitrogenase reductase activity by covalent modifications [15]. However, Fu et al. [18] verified that A. amazonense does not have the DraT/DraG system. Hence, in the near future, the interaction targets of the PII protein in A. amazonense should be determined to better understand their roles in the nitrogen metabolism of this microorganism.
Antibiotic minimum inhibitory concentration
Most DNA manipulation is dependent on the use of vectors containing resistance markers to antibiotics [19, 20]. In a previous work using antibiotic susceptibility test discs, Magalhães et al. (1983) [5] showed that A. amazonense is sensitive to kanamycin and gentamicin, tolerant to tetracycline, and resistant to penicillin. In this work, we determined the minimum inhibitory concentrations of A. amazonense to antibiotics that are normally used to provide a selective pressure for vectors.
The susceptibility of A. amazonense to kanamycin and gentamicin was confirmed, since no growth was observed in concentrations of these antibiotics of 0.25 μg/mL; therefore, vectors that contain selection markers for these compounds are appropriate for use.
High concentrations of ampicillin (128 μg/mL) were required for complete growth inhibition, showing that A. amazonense is also resistant to this beta-lactam antibiotic.
It is worth noting that the growth of A. amazonense was absent in a relatively high concentration of tetracycline (32 μg/mL), indicating that this species is, in fact, resistant to this antibiotic, instead of tolerant, as pointed out by Magalhães et al. [5]. These findings about the latter two antibiotics are relevant because they could be used in counter-selection procedures in conjugation experiments, as there is a variety of E. coli strains that are susceptible to them.
Conjugation
Conjugation mediated by E. coli is the standard DNA transfer technique of the Azospirillum genus [21]. Therefore, in this work the conjugation ability of A. amazonense was evaluated.
Unlike A. brasilense, A. amazonense cannot grow in LB medium. Furthermore, E. coli cannot grow in M79 medium; therefore, the first concern was to establish a medium that provided appropriate growth conditions for the donor and recipient strains. Hence, different medium compositions, containing distinct ratios of M79 and LB media (varying from 1:1 to 9:1), were prepared. The medium mixture of M79:LB at a proportion of 8:2 was the most suitable for culturing both bacteria and it was designated as MLB medium.
Another requisite for the conjugation procedure is to select vectors that contain proper selection markers that are mobilizable and able to replicate inside the receptor cell [19, 20]. Therefore, the pHRGFPGUS (pBBR1 replication origin) and the pPZPLACEYFP (pVS1 replication origin) plasmids were tested by tri-parental conjugation. These plasmids are mobilizable broad-host vectors harboring kanamycin resistance markers and fluorescent protein coding genes, which could promptly report achievement of the DNA transfer. The transconjugants exhibited kanamycin resistance and fluorescence. The conjugation frequencies were 3.8 × 10-8 per recipient cell for the pHRGFPGUS vector and 3.8 × 10-7 for the pPZPLACEYFP vector.
Different ratios of recipient to donor and helper strains (1:1:1, 5:1:1, 10:1:1 and 20:1:1) were also tested. The best efficiencies were obtained with the ratios 10:1:1 and 5:1:1; however, no obvious differences between these latter ratios were observed (data not shown).
In conclusion, conjugation is an appropriate method for DNA transfer to A. amazonense. Although only tri-parental mating was tested in this work, it is important to mention that bi-parental conjugation could be an alternative test, due to the possibility of increasing the conjugation efficiencies.
Electrotransformation
Since suitable vectors for A. amazonense were defined and since conjugation is a time-consuming procedure, the transformation of A. amazonense via electroporation was tested.
The eletrocompetence of the cells is greatly influenced by the growth phase [22]. Therefore, A. amazonense cells were harvested at different growth phases to evaluate their effect on electroporation efficiency. Cells from the late-log phase (OD600 1) and the stationary phase (OD600 2) were not electrocompetent. Electroporation utilizing cells from the early-log growth phase (OD600 0.12) generated a significant number of transformants. Therefore, all subsequent tests were performed utilizing cells cultivated at this growth phase.
In the electrocompetent cell preparation, the cells were harvested and washed continuously until the solution had a low-ionic strength. The MgCl2 HEPES-sucrose buffer was found to be the most suitable solution for the preparation of A. amazonense electrocompetent cells. Although 10% glycerol solution is commonly used for electrocompetent cell preparation in a diverse number of species (including A. brasilense), it was not appropriate for A. amazonense, as no transformants were obtained when this solution was used.
Different electroporation parameters were tested. The increase in electrical field strength had a positive effect on electroporation efficiency (Figure 2A). The highest electrical field strength tested was 12.5 kV/cm, and this condition was found to be the most efficient, generating about 8000 transformants/μg of pHRGFPGUS (Figure 2A). The effect of pulse length on electroporation efficiency was also investigated (Figure 2B). A pulse length of 4.3 ms (electroporation apparatus set at 200Ω) was the most efficient. The pulse lengths of 7.3 ms (400 Ω) and 10.5 ms (600 Ω) had a dramatic negative effect on transformation efficiency, where only few transformants were obtained (Figure 2B). These conditions are in agreement with the general parameters of bacterial electroporation [22–24].
In conclusion, the transfer of DNA to A. amazonense by means of electroporation was demonstrated. Although the efficiency of electrotransformation was far from desirable, this result is supported by previous works showing that bacteria closely related to A. amazonense, such as A. brasilense [25], R. rubrum [26] and Magnetospirillum gryphiswaldense [27], are recalcitrant to electrotransformation. Nonetheless, this technique is an easy and a rapid method of DNA transfer to the cells of A. amazonense.
Site-directed mutagenesis
Site-directed mutagenesis is a fundamental tool for correlating cellular functions with specific regions of the DNA. Therefore, once DNA transfer techniques were established for A. amazonense, the next step was to determine a site-directed mutagenesis protocol for this species.
Most of the A. brasilense mutants have been generated by the disruptive insertion of an antibiotic resistance cassette into the target gene [14, 28–30]. This approach is not recommended when the target gene composes an operon, since the resistance cassette could introduce a polar effect on the expression of the surrounding genes and, consequently, make it difficult to assign a mutant phenotype to the disrupted gene [31].
Therefore, in this work, a site-directed mutagenesis methodology that generates in-frame mutants without the disruptive insertion of a resistance cassette was evaluated. The glnK gene was selected for this methodology because subsequent studies of our laboratory will aim to determine the role of the PII proteins in A. amazonense metabolism.
The mutagenesis methodology is depicted in Figure 3A. Firstly, an amplicon containing an in-frame deletion of the glnK gene was generated through Crossover PCR, and it was subsequently cloned in the suicide replacement vector pK19MOBSACB, generating the pKΔK plasmid. This vector contains a kanamycin resistance gene (positive selection marker) that allows the selection of bacteria that would have integrated the plasmid into the chromosome. This vector was delivered to A. amazonense by means of conjugation (the carbon source utilized was maltose instead of sucrose) and one colony resistant to kanamycin was obtained, suggesting that the integration of the plasmid was successfully accomplished. The sacB gene (negative marker selection) of the vector is lethal in the presence of sucrose; therefore, the merodiploid strain (containing both wild-type and mutant alleles) was unable to grow in M79 (containing 10 g/L of sucrose). Subsequently, expecting that a recombination event could replace the wild-type allele, the merodiploid strain was cultured for many generations in M79 containing maltose instead of sucrose. Finally, this culture was plated in M79 containing sucrose to eliminate the bacteria that did not accomplish the second recombination event. Seven sucrose-resistant/kanamycin-sensitive colonies were chosen for PCR evaluation of the substitution of the mutant allele for the wild-type gene. Four colonies presented a band of 121 bp, indicating that the wild-type glnK was successfully substituted, whereas three colonies presented the 361 bp band, corresponding to the wild-type allele (Figure 3B). Furthermore, an additional PCR with primers flanking the recombination sites was performed, and it also demonstrated a reduction of the amplicon sizes originated from the glnK mutants in relation to the wild type strain (Figure 3C). This latter result demonstrates that recombination occurred in the target site.
Altogether, these results show that an in-frame glnK gene mutant strain of A. amazonense was successfully generated by this mutagenesis system.
Reporter gene system
The study of promoters is fundamental to elucidation of the genetic regulatory mechanisms of bacterial species. Up until now, there has been neither a report of heterologous gene expression in A. amazonense, nor a reporter system designed for this species. In this work, a reporter system based on expression of the Enhanced Yellow Fluorescent Protein (EYFP) was developed to analyze the regulatory regions of A. amazonense genes in vivo.
In silico analysis using a Sinorhizobium meliloti sigma 70 promoter weight matrix revealed that the genes aat, glnK, and glnB of A. amazonense have putative promoter sequences in their upstream regions (Figure 4). In E. coli, sigma 70 is considered to be the vegetative sigma factor, as it is responsible for the expression of the majority of genes [32, 33]. Therefore, one could expect that these putative A. amazonense sigma 70 promoters could act under standard laboratory growth conditions (aerobic environment, 35°C and M79 medium). Consequently, different vectors were constructed to determine the activity of the upstream regulatory sequences of A. amazonense genes in the expression of EYFP.
The lac promoter was utilized as a positive control since there is a report showing that this promoter has high activity in A. brasilense [34]. Two different vectors were constructed with the lac promoter, one derived from pPZPLACEYFP (pVS1 replicon) and the other derived from pHRGFPGUS (pBBR1 replicon). The upstream regions of the genes glnB, glnK, and aat were cloned into the pHRGFPGUS derivative.
The lac promoter had the best score in the in silico analysis from among the promoters detected, and, as expected, the highest fluorescence levels were observed in the lac constructions (Figure 5). The difference in the fluorescence levels between the pHRLACEYFP and pPZPLACEYFP transformants could be a product of the difference in the copy number between these vectors.
Although the in silico analysis revealed that the glnK promoter had a higher score than the aat and glnB promoters, its in vivo activity under the conditions tested did not differ significantly from the negative controls (without promoter and without plasmid) (Figure 5). One of the possible reasons for this is that this gene was repressed under these conditions. The reporter gene analysis also demonstrated that the aat and glnB promoters were active under the conditions tested, although the aat promoter showed a higher activity than the glnB promoter.
These observations show that a reporter system based on EYFP can be used for in vivo promoter analyses in A. amazonense.