ABSTRACT
Denitrification ability is sporadically distributed among diverse bacteria, archaea, and fungi. In addition, disagreement has been found between denitrification gene phylogenies and the 16S rRNA gene phylogeny. These facts have suggested potential occurrences of horizontal gene transfer (HGT) for the denitrification genes. However, evidence of HGT has not been clearly presented thus far. In this study, we identified the sequences and the localization of the nitrite reductase genes in the genomes of 41 denitrifying Azospirillum sp. strains and searched for mobile genetic elements that contain denitrification genes. All Azospirillum sp. strains examined in this study possessed multiple replicons (4 to 11 replicons), with their sizes ranging from 7 to 1,031 kbp. Among those, the nitrite reductase gene nirK was located on large replicons (549 to 941 kbp). Genome sequencing showed that Azospirillum strains that had similar nirK sequences also shared similar nir-nor gene arrangements, especially between the TSH58, Sp7T, and Sp245 strains. In addition to the high similarity between nir-nor gene clusters among the three Azospirillum strains, a composite transposon structure was identified in the genome of strain TSH58, which contains the nir-nor gene cluster and the novel IS6 family insertion sequences (ISAz581 and ISAz582). The nirK gene within the composite transposon system was actively transcribed under denitrification-inducing conditions. Although not experimentally verified in this study, the composite transposon system containing the nir-nor gene cluster could be transferred to other cells if it is moved to a prophage region and the phage becomes activated and released outside the cells. Taken together, strain TSH58 most likely acquired its denitrification ability by HGT from closely related Azospirillum sp. denitrifiers.
IMPORTANCE The evolutionary history of denitrification is complex. While the occurrence of horizontal gene transfer has been suggested for denitrification genes, most studies report circumstantial evidences, such as disagreement between denitrification gene phylogenies and the 16S rRNA gene phylogeny. Based on the comparative genome analyses of Azospirillum sp. denitrifiers, we identified denitrification genes, including nirK and norCBQD, located on a mobile genetic element in the genome of Azospirillum sp. strain TSH58. The nirK was actively transcribed under denitrification-inducing conditions. Since this gene was the sole nitrite reductase gene in strain TSH58, this strain most likely benefitted by acquiring denitrification genes via horizontal gene transfer. This finding will significantly advance our scientific knowledge regarding the ecology and evolution of denitrification.
INTRODUCTION
Denitrification is a biological respiratory process in which microorganisms including bacteria, archaea, and fungi reduce nitrate (NO3−) and/or nitrite (NO2−) to gaseous products (NO, N2O, or N2) to acquire energy (1). The first step in the denitrification process is the reduction of NO3− to NO2−, which is catalyzed by respiratory membrane-bound or periplasmic nitrate reductases (Nar or Nap, respectively). The second step, the reduction of NO2− to NO, is catalyzed by copper-containing or cytochrome cd1-containing nitrite reductases (NirK or NirS, respectively). The third step, the reduction of NO to N2O, is catalyzed by nitric oxide reductase (Nor). The last step of denitrification, the reduction of N2O to N2, is catalyzed by nitrous oxide reductase (Nos) (1).
Denitrification ability is widely distributed among bacteria, archaea, and fungi; however, both denitrifying and nondenitrifying strains can be present in closely related taxa (e.g., at the genus level) (2). In addition, almost identical denitrification genes, such as nirK and nirS, can be found among phylogenetically distantly related strains (3). Phylogenetically closely related strains could also harbor very different denitrification gene sequences. Disagreement between the phylogenies of denitrification genes and the 16S rRNA gene phylogeny (4–8) suggests potential occurrences of horizontal gene transfer (HGT). The potential occurrence of HGT of denitrification genes has been also suggested based on comparative genome analysis (9). However, there are only a few reports that showed direct evidence that support the horizontal transfer of denitrification genes (10, 11). These studies were done with an extremely thermophilic bacterium, Thermus thermophilus, and therefore, the occurrence of HGT in mesophilic denitrifiers is still unclear.
The presence of genes integrated into mobile genetic elements can support the occurrence of HGT. A transposon is one of the mobile genetic elements and includes insertion sequence (IS) elements (12). A transposon can be transferred within a genome, but it can be also transferred between genomes of multiple strains if it is associated with prophages or conjugative elements, such as conjugative plasmids and integrative and conjugative elements (12). Similarly, a composite transposon, which is a gene or gene cluster flanked by two IS elements, can be also mobilized within a genome, as well as between distantly related bacteria. A composite transposon system could carry genes useful for host cell survival and adaptation (e.g., antibiotic resistance genes) and therefore can enhance the adaptive capacity of the host bacteria (13, 14). Integrative and conjugative elements, also known as conjugative transposons, possess their own ability to transfer from one bacterial cell to another via conjugation with a surprisingly broad host range, suggesting that the extent of their contribution to HGT is as much as that of a plasmid system (15). Genomic islands (GIs) are defined as large genomic regions ranging from a few kilobases to >500 kb (16–19), and they are thought to have originated through horizontal transfer. GIs are discrete DNA segments which are part of a chromosome and thought to contribute to diversification and adaptation of microorganisms, with a significant impact on genomic plasticity and dissemination of genes associated with pathogenicity and antibiotic resistance (20). GIs often carry insertion elements or transposons, and as well as genes that can provide selective advantages to the host bacteria, such as those related to pathogenicity, antibiotic resistance, symbiosis, fitness, or metabolism (21, 22). The presence of GIs can be detected based on the features of GIs, such as nucleotide and codon usage bias, direct repeats, and the presence of mobility genes (23).
In this study, we looked for the evidence of horizontal transfer of denitrification genes in Azospirillum sp. denitrifiers. Azospirillum species are well known to possess multiple replicons (24). Genome sequencing of Azospirillum sp. strain B510 identified the nitrite reductase gene nirK on a 681-kbp plasmid, which is one of the seven replicons of the strain (18). Azospirillum brasilense Sp245 was reported to harbor two functional nirK genes on different replicons (25). One of the two nirK genes is located on an 85-MDa (∼142-kbp) plasmid (p85 plasmid), which is prone to rearrangement and involved in replicon fusions (25, 26). In addition, some Azospirillum sp. denitrifying strains are reported to possess nirS instead of nirK (3, 27). These results suggest that nitrite reductase genes have been transferred within or between genomes of Azospirillum sp. strains.
Previously, we isolated 41 Azospirillum sp. denitrifiers from rice paddy soils (3, 27, 28). While some strains were positive for nirK or nirS by PCR, nitrite reductase genes of the majority of the strains could not be identified, most likely due to the base mismatches to the primers used (3, 27, 28).
Therefore, the objectives of this study are to (i) identify the nitrite reductase gene sequences of the 41 Azospirillum strains, (ii) compare the nir phylogeny with the 16S rRNA gene phylogeny, (iii) identify the localization of the nitrite reductase genes in the genomes of Azospirillum sp. strains, and (iv) search for mobile genetic elements that contain denitrification genes.
RESULTS
16S rRNA gene and nirK phylogenies.Forty-one Azospirillum sp. strains used in this study were previously isolated from rice paddy soil (3, 27, 28). Among these strains, four strains possessed nirS as their nitrite reductase gene; however, no nirK or nirS genes were detected in the rest of the strains (n = 37) by PCR with commonly used primers (3, 27, 28). Based on the previous genome analysis, Azospirillum sp. B510 (18) and Azospirillum brasilense Sp245 (29) possess nirK genes with many base mismatches to these PCR primers. We therefore designed new primers to amplify nirK from Azospirillum strains. As a result, we identified nirK in the 37 Azospirillum sp. strains.
Based on the nirK phylogeny, Azospirillum strains could be divided into three groups (Fig. 1A). While most strains were clustered into group A1, strains TSO35-2 and TSO41-3 were clustered into group A2. Strain TSH58 was clustered into group A3, along with Azospirillum brasilense strains Sp7T and Sp245. Strain Sp245 possesses two nirK genes (25), both of which were clustered into group A3.
Phylogenetic trees constructed based on nirK (A) and 16S rRNA (B) gene sequences by using the maximum likelihood method. GenBank accession numbers are shown in square brackets. Bootstrap values (%) were generated from 1,000 replicates, and values of ≥70% are shown. Branch lengths correspond to sequence differences, as indicated by the scale bar. Strains shown in bold were used for genome sequence analyses.
Based on the 16S rRNA phylogeny, Azospirillum sp. strains could be clustered into four groups (Fig. 1B). While the majority of strains were clustered into group B1, all of the group B2 strains possessed nirS. Similar to the nirK phylogeny, strain TSH58 was more closely related to Azospirillum brasilense Sp7T and Sp245 strains (group B4).
Concurrence was observed between the 16S rRNA gene and nirK phylogenies for most of strains, suggesting that circumstantial evidence of HGT for nirK within Azospirillum strains was hardly detected from the phylogeny comparison.
Localization of nirK in Azospirillum genomes.To examine the presence of plasmid-borne nirK, we performed pulsed-field gel electrophoresis (PFGE) analysis for 36 strains, including Sp7T and B510 strains. Four nirS-type strains were not included in this analysis. In addition, three nirK-type strains that were not recovered from frozen stocks were also excluded. Based on the PFGE analysis, all Azospirillum sp. strains examined in this study possessed multiple replicons (i.e., 4 to 11 replicons), with their sizes ranging from 7 to 1,031 kbp (see Table S1 in the supplemental material), indicating that the number and size of replicons vary depending on the strain. The PFGE patterns were reproducible among gels. The PFGE band sizes of the strain TSH7 were not significantly different by gels, with the standard deviations smaller than 1.2%. Most of the nirK genes were identified in large replicons ranging from 549 to 941 kbp by Southern blot hybridization (Table S1).
Comparative genome analysis.To identify the gene clusters around nirK, genomes of 10 representative Azospirillum strains were sequenced using an Illumina HiSeq platform. Strains selected based on the 16S rRNA gene and nirK phylogenies were TSA6c, TSH7, TSH20, TSH58, TSH64, TSH100, TSO5, TSO22-1, TSO35-2, and Sp7T. The resulting high-quality sequences were assembled to total lengths of 7.0 to 8.3 Mbp per genome, with >219× sequence coverage (Table S2). Because complete genome data became available for strain Sp7T, we used these data (GenBank accession numbers CP012914 to CP012919) instead of our draft genome data (GenBank accession number LGRK00000000).
All key functional genes for denitrification (narG, nirK, norB, and nosZ) were identified on the genomes of the Azospirillum strains except nosZ for the genome of the strain TSO5. This is probably because of the incomplete genome assembly, since nosZ was successfully detected from strain TSO5 by PCR sequencing in the previous study (3). Or, it is possible that nosZ was lost due to phase variation similar to what was seen in a previous report (30). Similar to a previous PCR analysis (27), nirS and its associated genes were identified in the genome of strain TSO22-1. This strain does not possess nirK, suggesting that cytochrome cd1-containing nitrite reductase is responsible for nitrite reduction. All the other strains possessed nirK. Interestingly, both nirK and nirS were detected in the genome of strain Sp7T.
The nirK gene was always located next to nitric oxide reductase gene (nor) cluster, except for the nirK1 of strain Sp245 (GenBank accession number HE577330) (Fig. 2). The nir-nor gene clusters were relatively well conserved among strains. Strains that had similar nirK sequences tend to have similar gene organization patterns around the gene (Fig. 2). In particular, the nir-nor gene cluster in the genome of strain TSH58 was almost identical to those in the genomes of strains Sp7T and Sp245. Strain Sp245 was reported to possess two functional nirK genes (25), one of which (nirK2) is located on the 85-MDa (∼142-kbp) plasmid p85 (GenBank accession number EU194339) (25, 26). Although the presence of two nirK genes was reported, only one nirK gene (i.e., nirK1) was detected on the genome of strain Sp245 that was sequenced previously (29), suggesting that the second nirK gene might have been lost during laboratory cultivation in the sequenced strain. The nirK2 gene of strain Sp245 showed greater sequence identity (>98%) to the nirK gene of Sp7T and TSH58 strains than another nirK gene (i.e., nirK1) of Sp245 (95%). These results suggest the occurrence of recent horizontal transfer of nirK between these strains.
Comparison of nir-nor gene clusters among the Azospirillum sp. strains. Nucleotide sequence identity (%) is shown by a heatmap generated using Genome Matcher version 2.03. Red arrows indicate denitrification genes (nirK, norC, norB, norQ, and norD). Blue arrows indicate genes encoding transposases. NA, sequences not available.
Identification of mobile genetic elements around the nir-nor gene cluster.Interestingly, we identified IS elements encoding transposase around the nir-nor gene cluster of strain TSH58. Although many IS elements were found on the genomes of Azospirillum strains, all of them except those found in strain TSH58 were distantly located from nir-nor gene cluster (Table S3). This is in contrast to the previous report, in which IS elements encoding transposase (ISAzba1 and ISAzba2) were detected near nirK2 of strain Sp245 (31). Since sequences around nirK2 of strain Sp245 were not completely available, we performed PacBio sequencing to obtain the complete genome of strain Sp245, including p85. We used the Sp245 “Russian strain” (= IBPPM 219) because this strain was reported to have p85 plasmid with nirK2 (31). The previously sequenced Sp245 “U.S. strain” contains only one nirK gene (= nirK1) (GenBank accession numbers HE577327 to HE577333) (29). In addition to the Sp245 Russian strain, we also completed the genome of the TSH58 strain by using a PacBio platform (Table S4).
Similar to the Sp245 U.S. strain, we detected only one nirK gene on the genome of the Sp245 Russian strain (GenBank accession number GCA_003119195). The nucleotide sequence identities of this nirK gene were 100 and 96% to nirK1 (GenBank accession number HE577330) and nirK2 (GenBank accession number EU194339), respectively, suggesting that nirK2 had been lost during laboratory cultivation. We therefore could not obtain sequences around the nirK2-nor cluster of strain Sp245. Complete genome sequencing also verified that there was no composite transposon structure around the nirK1 gene of strain Sp245.
The genome of TSH58 was also completed by PacBio sequencing. The nir-nor gene cluster was located on a 568-kbp-long replicon, TSH58_p05 (GenBank accession number CP022369), similar to the result from the PFGE analysis (549-kbp-long replicon). The nir-nor gene cluster was located on a putative composite transposon structure flanked by two novel IS elements we named ISAz581 and ISAz582 (Fig. 3). These IS elements belonged to the IS6 family and contained genes encoding transposases that were flanked with inverted repeats (5′-GGCTCTGTTGCAAAGTTnnnnnAACTTTGCAACAGAGCC-3′ in ISAz581 and 5′-GGCACCGTCAACTTnnnnnAAGTTGATGGTGCC-3′ in ISAz582, where “nnnnn” represents any number of intervening nucleotides). In addition, direct repeats of 5′-ACGCGCTTTTTCG-3′ were found at both ends of the composite transposon structure. These lines of evidence suggest that the nir-nor gene cluster was inserted in this position as a result of composite transposon activity. Interestingly, we found a 405-bp spacer region between the right-end inverted repeat (IRR) of ISAz582 and the direct repeat, which is not common. It is currently unknown why there is a spacer region here.
Structure of the composite transposon carrying the nir-nor gene cluster found on the replicon TSH58_p05 of Azospirillum sp. TSH58 (GenBank accession number CP022369). The composite transposon was flanked by two novel IS elements, ISAz581 and ISAz582, as indicated by orange boxes. Genes of known function are indicated by gray arrows, while those of unknown function are indicated by white arrows. Inverted and direct repeats are indicated by red and yellow boxes, respectively. Light-gray boxes indicate the region of genomic islands identified by SIGI-HMM method. IRL, internal repeat long.
The transposases in ISAz581 and ISAz582 showed 86% and 61% amino acid sequence identity to those in ISAcr2 and ISBj7 in the ISfinder database (https://www-is.biotoul.fr) as the most similar references, respectively. They were totally different from the transposases encoded by the ISs near nirK2 of Sp245 strain, ISAzba1 and ISAzba2, which belong to the ISKra4 and ISL3 families, respectively (31).
Sixteen genes, including nirK and norCBQD, were found between the two IS elements. Almost identical sequences to this region excluding the transposase genes were found in the genome of Azospirillum brasilense Sp7T (GenBank accession number CP012918). However, genes outside this region were not found adjacent to the nir-nor gene cluster of strain Sp7T. In addition, sequences near the IS elements were also identified as GIs by the SIGI-HMM method (Fig. 3), which uses a hidden Markov model (HMM) and measures codon usage to predict GIs (32, 33). These results strongly suggest that genes located between ISAz581 and ISAz582 could be transposed as a part of a composite transposon.
Transcription of nirK located on the composite transposon system.Active transcription of the nirK gene located on the composite transposon system on the strain TSH58 genome was verified by reverse transcription-quantitative PCR (RT-qPCR). The transcription level was significantly higher (P < 0.05 by analysis of variance) in the cells grown under anoxic conditions with nitrite or nitrate addition (i.e., denitrification-inducing conditions) than those grown under oxic conditions without nitrite/nitrate addition (Fig. S1). Since this nirK is the sole nitrite reductase gene in TSH58, the composite transposon system most likely carries functional nirK.
Conjugative potential of the composite transposon.The composite transposon system containing a nir-nor cluster was located on a 568-kbp-long replicon, TSH58_p05 (GenBank accession number CP022369). On this plasmid or other replicons of strain TSH58, we did not detect integrative and conjugative elements (ICEs) or key genes required for plasmid conjugation (e.g., those encoding relaxosome and conjugative and type IV pili). Although there is potential that conjugative or mobilizable plasmid had been lost after introducing the composite transposon system to the cells, it is difficult to examine this by genome sequencing alone. However, intact prophage sequences were found on a 3,020-kbp-long replicon (GenBank accession number CP022364). Prophage sequences were also found in other Azospirillum sp. genomes (Table S5). Therefore, transfer of a nir-nor cluster could occur between cells if the composite transposon is moved to a prophage region and the phage becomes activated and released outside the cells.
DISCUSSION
HGT is a key pathway for acquiring genes related to condition-specific advantages facilitating adaptation to environments, such as virulence and antimicrobial resistance (34). While HGT of denitrification genes has been suggested (4–8), integration of the genes in mobile genetic elements has not been clearly presented, except in an extremely thermophilic bacterium (10, 11).
In this study, we looked for the evidence of horizontal transfer of denitrification genes in Azospirillum sp. denitrifiers. Azospirillum spp. are commonly found in various soil environments and are well studied as plant growth-promoting rhizobacteria which can fix atmospheric N2 (35) and enhance lateral root development (36). Nitric oxide, a product of dissimilatory nitrite reduction, plays an important role in stimulating lateral root growth (37). A nitrite reductase nirK gene of A. brasilense Sp245 was upregulated when the strain colonized wheat roots (38), suggesting that nitrite reductase is important for adaptation and survival of Azospirillum species in the environment. Therefore, acquisition of nir via HGT could increase the fitness of the Azospirillum host cells in the environments.
The majority of the Azospirillum sp. strains tested in this study possessed nirK (37 out of 41), and nirS was only found in four strains. The presence of nirK or nirS was confirmed by sequencing the genomes of 10 representative strains. We identified both nirK and nirS on the genome of A. brasilense Sp7 (GenBank accession number CP012918), similar to some Bradyrhizobium spp. and other denitrifying strains (39–41). The rest of the strains had either nirK or nirS as the sole nitrite reductase gene.
Three out of four nirS-harboring strains formed a distinct cluster from nirK strains based on the 16S rRNA gene phylogeny, implying different evolutionary histories between nirS- and nirK-possessing Azospirillum sp. strains. While nirK and nirS may have redundant roles in terms of nitrite reduction (39), we observed higher growth rates of nirK-harboring Azospirillum strains than nirS-harboring strains under denitrification conditions (data not shown). Similarly, Yoshida et al. (42) showed a faster response of nirK-harboring bacteria than nirS-harboring ones to denitrification-inducing conditions, by using culture-independent analyses done with rice paddy microcosms.
Clustering patterns of most of the Azospirillum strains were similar between phylogenetic analyses performed using the 16S rRNA gene sequence and those performed using the nirK sequences. Therefore, it is difficult to predict the occurrence of HGT based on phylogenetic comparison alone, although several previous studies suggest the occurrence of HGT events based on the disagreement between phylogenies of 16S rRNA and denitrification genes (4–7, 27).
All Azospirillum sp. strains used in this study had multiple replicons, similar to previous studies (18, 24). The nirK genes were located on relatively large replicons (549 to 941 kbp). Similar results were obtained between PFGE and complete genome sequencing, although the sizes of the replicons were slightly different between the two methods, most likely due to the lower resolution power of the PFGE analysis. This is especially notable in large (>1 Mbp) replicons; we could not reliably predict the size of large replicons because the largest fragment size of the molecular size marker used in the PFGE was 1 Mbp.
Based on the genome sequence analysis, nirK was almost always located next to a nor cluster, which encodes nitric oxide reductase. This makes sense because nitric oxide, a product of nitrite reductase, is toxic to cells and therefore needs to be transformed to the nontoxic compound N2O (43). Similar to this study, the close proximity of nir and nor genes is also found in other denitrifiers (1, 9), supporting the importance of detoxifying NO that is produced by nitrite reductases. In addition to nor, Sp7T and TSH58 possessed a gene encoding nitric oxide dioxygenase, which converts NO to NO3− and serves as another NO detoxification machinery (44).
Azospirillum strains that had similar nirK sequences also shared similar nir-nor gene arrangement. This was especially notable between the TSH58, Sp7T, and Sp245 strains. In addition to the high similarity between nir-nor gene clusters among the three Azospirillum strains, a composite transposon structure was identified in the genome of strain TSH58, which contains the nir-nor gene cluster and novel IS elements (ISAz581 and ISAz582). Although the presence of either (i) almost-identical sequences between multiple strains or (ii) a composite transposon system containing the gene of interest does not necessarily indicate the occurrence of HGT, the presence of both (i) and (ii) strongly suggests the occurrence of HGT. Therefore, our results suggest that strain TSH58 horizontally acquired the nir-nor gene cluster from the ancestor of either the Sp7T or Sp245 strain. Although the composite transposon system containing the nir-nor cluster in strain TSH58 was not closely associated with conjugation systems, intact prophage sequence was found on the genome of strain TSH58, which could act as a vehicle for the composite transposon. Prophage sequences were also found in other Azospirillum sp. strains examined in this study. Similarly, Boyer et al. (45) also report the widespread occurrence of prophages in various Azospirillum sp. strains. These phages were released to the environment after mitomycin C induction. These results suggest that phage could play an important role in HGT between Azospirillum sp. cells.
The nir-nor gene cluster within the composite transposon system encodes functional nitrite reductase. The nirK in TSH58 strain was transcribed in the presence of nitrate or nitrite. Since this nirK gene is the sole nitrite reductase gene in TSH58, the composite transposon system most likely carries functional nirK.
Although an IS was previously identified near the nirK2-nor cluster of strain Sp245 (31), it is not clear whether its nirK2-nor is part of the composite transposon because only one IS element was identified. To clarify this, we sequenced the complete genome of Sp245 Russian strain; however, we could not identify nirK2 in the genome. In addition, the plasmid p85 (∼142 kb), which was reported to contain nirK2 (31), could not be identified, indicating that the plasmid had been lost or cointegrated with other replicons during laboratory cultivation (25, 26). Because strain Sp245 was reported to possess two functional nirK genes (nirK1 and nirK2) (25), the loss of nirK2 from the genome most likely had no major impact on its denitrification ability and anaerobic growth.
The IS elements identified in strain TSH58 were totally different from those identified in strain Sp245 (ISAzba1 and ISAzba2). While ISAz581 and ISAz582 in strain TSH58 belong to the IS6 family, ISAzba1 and ISAzba2 belong to the ISKra4 and ISL3 families, respectively (31). This suggests that potential horizontal transfer of the nir-nor gene cluster can occur by using a variety of IS elements and therefore could occur more frequently than previously thought.
In conclusion, this study supports the occurrence of horizontal transfer of denitrification genes between Azospirillum sp. strains. Although occurrence of HGT in other genera is still unclear, heterologous expression of the entire composite transposon system, including the denitrification genes, should further confirm the HGT potential of these genes.
MATERIALS AND METHODS
Bacterial strains.Forty-one Azospirillum sp. strains were previously isolated from rice paddy soil in Tokyo, Japan (3, 27, 28). Azospirillum brasilense Sp7T (= JCM 1224T) was obtained from the Japan Collection of Microorganisms (JCM). Azospirillum brasilense strain Sp245 (= IBPPM 219) was obtained from the Collection of Rhizosphere Microorganisms of the Russian Academy of Sciences’ Institute of Biochemistry and Physiology of Plants and Microorganisms (IBPPM RAS). The strains were anaerobically grown at 30°C in nutrient broth supplemented with 5 mM nitrate and 10 mM acetate.
DNA extraction and PCR amplicon sequencing.DNA was extracted from the cells by using an alkali-boiling method (46). To amplify nirK from Azospirillum sp. denitrifiers, new primers were designed using genome sequences of Azospirillum sp. strain B510 (18) and Azospirillum brasilense Sp245 (29) as references. Primers nirK_247F (5′-ACACCTAYTGGACSTTCAAC-3′) and nirK_818R (5′-GACSTTGTCGAAGATCTCGC-3′) can amplify 572-bp fragments from nirK of Azospirillum sp. B510 (Fig. S2). The PCR mixture (50 μl) contained 1× Ex Taq buffer (TaKaRa Bio), 0.2 μM each primer, 0.2 mM each dinucleoside triphosphate (dNTP), 1 U of Ex Taq DNA polymerase (TaKaRa Bio), and 1 μl of a DNA template. PCR was performed using a Veriti 96-well thermal cycler (Applied Biosystems) with the following conditions: initial annealing at 96°C for 5 min, followed by 35 cycles at 95°C for 30 s, 56°C for 30 s, and 72°C for 1 min. After the final extension at 72°C for 7 min, the PCR mixtures were stored at 4°C. The PCR products were purified and sequenced by the Sanger method, as described previously (46).
RNA extraction and reverse transcription-quantitative PCR.To verify active transcription of nirK, reverse transcription-quantitative PCR (RT-qPCR) was performed. RNA was extracted by using TRI Reagent (Invitrogen) and the Direct-zol RNA MiniPrep kit (Zymo Research) from cells grown under anoxic conditions at 30°C in R2A broth supplemented with 10 mM acetate and 5 mM nitrate (or nitrite) for 24 h. RNA was also extracted from the cells grown under oxic conditions at 30°C in R2A broth without any supplements. The absence of genomic DNA in the RNA samples was verified by PCR targeting the 16S rRNA gene, as described previously (46). Complementary DNA (cDNA) was synthesized from the RNA samples by reverse transcription-PCR using the PrimeScript RT reagent kit (TaKaRa Bio) with random hexamers. Primers nirK_247F (5′-ACACCTAYTGGACSTTCAAC-3′) and nirK_918R (5′-GATARTCSACCTTGAACTCC-3′) were used to quantify nirK transcripts, while Eub338 (5′-ACTCCTACGGGAGGCAGCAG-3′) and Eub518 (5′-ATTACCGCGGCTGCTGG-3′) were used to quantify 16S rRNA. The reaction mixture for RT-qPCR (10 μl) contained 1× SYBR Premix Ex Taq ROX plus (TaKaRa Bio), 0.2 μM each primer, and 100 ng cDNA. RT-qPCR was performed using the StepOnePlus real-time PCR system (version 2.3; Applied Biosystems) with the following conditions: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Levels of transcription were normalized using the quantity of 16S rRNA.
Pulsed-field gel electrophoresis and Southern blotting.Replicons in the Azospirillum strains were separated by PFGE, as previously described (47), with some modifications. In brief, Azospirillum strains were anaerobically grown at 30°C in nutrient broth supplemented with 5 mM nitrate and 10 mM acetate. Cells were embedded in 1 % (wt/vol) agarose gel plugs created by using plug molds (Bio-Rad). The plugs were incubated in 4 % (wt/vol) sodium dodecyl sulfate (SDS) in 0.25 M EDTA for 20 min. The plugs were washed with Tris-EDTA (TE) buffer (pH 8.0) and then incubated in lysozyme solution (4 mg/ml) in TE buffer containing 0.2% (wt/vol) SDS at 37°C for 2 h. The plugs were washed again and incubated with 0.2 mg/ml proteinase K in TE buffer containing 1% (wt/vol) SDS at 50°C overnight, as previously described (47). After incubation, the plugs were washed again and used for PFGE analysis.
The DNA in the plugs was separated using a Chef-DR II PFGE apparatus (Bio-Rad) in 0.5× Tris-borate-EDTA (TBE) buffer at 14°C for 22 h, with the following parameters: gradient, 6.0 V/cm; included angle, 120°; initial switch time, 2.98 s; final switch time, 57.62 s; and ramping factor, linear. λ Ladder (Bio-Rad) and λ DNA-HindIII digestion (Nippon Gene) were used as the external size standards. In addition, strain TSH7 was included in all gels as the positive control for the Southern hybridization (see below). The gel was stained with SYBR green I and scanned by using the Molecular Imager PharosFX system (Bio-Rad). The size of each DNA fragment (i.e., replicon) was estimated by using external size standards. Due to the resolution of the PFGE analysis, only fragments smaller than 1 Mbp were analyzed.
The replicons separated by PFGE were transferred to a nylon membrane, as described previously (48), and used for Southern hybridization to identify the localization of nirK. The digoxigenin (DIG)-labeled nirK PCR product from strain TSH7 was used as a probe.
Genome sequencing.Ten representative Azospirillum sp. strains were selected based on the 16S rRNA gene and nirK phylogenies. Genomic DNA were extracted using the PowerSoil isolation kit (Mo Bio). Sequence libraries were prepared using TruSeq DNA sample prep kit (Illumina), according to the manufacturer’s instructions. Genome sequencing was done using the Illumina HiSeq 2000 platform with a 101-bp paired-end library. The resulting sequence reads were assembled de novo by using Velvet version 12.0.8 (49).
In addition to HiSeq sequencing, genomes of Azospirillum brasilense IBPPM219 (= Sp245) and Azospirillum sp. TSH58 strains were also identified by using the single-molecule real-time (SMRT) sequencing technology of Pacific Biosciences (PacBio). SMRTbell libraries were constructed according to the PacBio SMRTbell 10-kb library preparation protocol. Genome sequencing was performed using the Pacific Biosciences RSII sequencing platform with a single SMRT cell per genome. De novo assembly after quality filtering of reads was done using the Hierarchical Genome Assembly Process (HGAP3) in the SMRT Link portal (version 2.3.0).
Gene prediction and annotation were done using the NCBI Prokaryotic Genome Annotation Pipeline (50). The sequence assembly and gene annotation results are summarized in Tables S2 and S4.
Bioinformatic and statistical analyses.Phylogenetic trees were constructed based on the 16S rRNA gene and the nirK sequences by using maximum likelihood method with bootstrap analysis (n = 1,000) by the MEGA6 software (51). Pairwise blastn results were visualized using the GenomeMatcher software (52). Genomic islands were identified using the web-based IslandViewer software integrating multiple accurate methods for genomic island prediction (23). Insertion sequences were identified by blastp against the ISfinder database (53) and by using ISsaga (54). Prophages were detected on the genomes of the Azospirillum strains by using the web-based tool PHASTER (55). Other genes of interest (e.g., those encoding transposase, relaxosome, and conjugative and type IV pili) were identified on the genomes by searching homologous sequences in the NCBI’s nonredundant protein database by using blastp algorithm and in-house perl scripts. In addition, we also used our sequenced genomes as the databases for the blastp search to identify homologous sequences to the genes of interest. An E value of 10−5 was used as a cutoff. Analysis of variance was performed by using R version 3.3.2 to test statistical significance in transcript quantities across samples.
Data availability.The genome sequences identified in this study were deposited to the DDBJ/EMBL/GenBank databases under accession numbers GCA_003116055, GCA_003119195, GCA_003115895, GCA_003115945, GCA_003115935, GCA_003115995, GCA_003119115, GCA_003116035, GCA_003115975, GCA_003116065, GCA_003116015, and GCA_003116095 for strains Sp7, Sp245, TSA6c, TSH7, TSH20, TSH58 (by HiSeq), TSH58 (by PacBio), TSH64, TSH100, TSO5, TSO22-1, and TSO35-2. The nirK sequences identified in this study were also deposited to the DDBJ/EMBL/GenBank databases under accession numbers MK108919 to MK108955.
ACKNOWLEDGMENTS
We thank Nisha Vishwanathan for technical assistance. Azospirillum brasilense strain IBPPM 219 (= Sp245 Russian strain) was kindly provided by Elena I. Katsy and Olga V. Turkovskaya at the Institute of Biochemistry and Physiology of Plants and Microorganisms Russian Academy of Sciences (IBPPM RAS). Genomic DNA of Azospirillum sp. strain B510 was kindly provided by Kiwamu Minamisawa at Tohoku University.
This work was supported, in part, by the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry (BRAIN) and the Science and Technology Research Promotion Program for Agriculture, Forestry, Fisheries and Food Industry, and by the Minnesota’s Discovery, Research and InnoVation Economy (MnDRIVE) initiative of the University of Minnesota. This work was done in part using computing resources at the Minnesota Supercomputing Institute.
We declare no conflicts of interest.
FOOTNOTES
- Received 9 October 2018.
- Accepted 5 November 2018.
- Accepted manuscript posted online 9 November 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AEM.02474-18.
- Copyright © 2019 American Society for Microbiology.
