Isolation of Oligotrophic Denitrifiers Carrying Previously Uncharacterized Functional Gene Sequences

Abstract

Oligotrophic denitrifying bacteria, including those belonging to the genera Herbaspirillum, Azospirillum, and Bradyrhizobium, were obtained using a single-cell isolation technique. The taxonomic composition of the denitrifier population was similar to those assessed by previous culture-independent studies. The sequencing of nitrite reductase and N₂O reductase genes of these strains revealed previously unknown links between 16S rRNA and the denitrification-functional gene phylogenies. In particular, we identified Bradyrhizobium strains that harbor nirS sequences previously detected only in culture-independent studies.

Denitrification is a microbial respiratory process in which nitrogen oxides (nitrate and nitrite) are reduced stepwise to gaseous end products (NO, N₂O, and N₂) (19, 32). Denitrification ability is sporadically distributed among the taxonomically diverse groups of bacteria, archaea, and fungi (8, 19, 27); therefore, it is difficult to identify denitrifiers based only on their 16S rRNA gene sequences (19). Nitrite reductase genes (nirK and nirS) and N₂O reductase genes (nosZ) have frequently been used as markers for denitrifiers instead of 16S rRNA gene sequences (2, 21, 26). However, denitrifiers cannot be identified based solely on this functional gene sequence information, because phylogenetically distantly related bacteria may carry highly similar functional gene sequences (9).

Single-cell analysis and isolation techniques have the potential to link strain phylogeny with function because both the 16S rRNA gene and the functional genes originate from the same cell (10, 11, 20). In the present study, we applied a functional single-cell (FSC) isolation technique (1) to obtain oligotrophic denitrifying bacteria from rice paddy soil, an environment with strong denitrification activity.

In a soil microcosm setup using 1 g rice paddy soil (12, 23), 0.1 mg nitrate N and 0.5 mg succinate C were added. The vial was then anaerobically incubated at 30°C for 24 h to enhance the denitrifying activity. Microbial populations responsive to this condition have previously been investigated by culture-independent studies (12, 23, 31). In the present study, cell division inhibitors (20 μg nalidixic acid, 10 μg pipemidic acid, 10 μg piromidic acid, and 10 μg cephalexin) (13) were added to the soil, together with nitrate and succinate, to elongate cells responsive to denitrification-inducing conditions. Unlike the FSC isolation method originally proposed by Ashida et al. (1), denitrifier enrichment was not performed prior to the addition of cell division inhibitors; therefore, fast-growing denitrifiers did not overwhelm the microbial population. These elongated cells were stained with 5-carboxyfluorescein diacetate acetoxymethyl ester (CFDA-AM), individually captured with a micromanipulator (1), and transferred and grown in 100-fold-diluted nutrient broth supplemented with 3 mM nitrate and 4.4 mM succinate (DNB-NS medium) at 30°C for 2 weeks. The diluted nutrient broth was previously shown to be useful for obtaining oligotrophic bacteria (7, 18). To obtain well-isolated single colonies, the cultures in the DNB-NS medium were streaked onto DNB-NS agar and anaerobically incubated at 30°C for 2 weeks. The denitrifying activity of the strains was determined using the acetylene block method (27). Strains that reduced >20% of nitrate to N₂O in the medium were considered denitrifiers (27). This cutoff value is lower than that proposed by Mahne and Tiedje (17); however, their proposal was based on the experiments performed with nutrient-rich medium, unlike the nutrient-poor DNB-NS medium used in the present study. Because some oligotrophic denitrifiers did not show sufficient growth in the DNB-NS medium, a portion of nitrate may remain unused. Therefore, we believe that a cutoff value of 80% (17) is too strict for the identification of oligotrophic denitrifiers. In the present study, strains that harbored nirK or nirS, as determined using PCR (see below), were also considered to be denitrifiers even if their nitrite-reducing levels were less than 20%.

Genomic DNA was extracted from bacterial cells, and PCR was performed to amplify the 16S rRNA gene as described previously (1). PCR targeting nirK, nirS, and nosZ was also performed using the primers F1aCu and R3Cu (26), cd3aF and R3cd (26), and nosZ-F-1181 and nosZ-R-1880 (21), respectively. For Bradyrhizobium strains, PCR targeting the internal transcribed spacer (ITS) region between the 16S and 23S rRNA genes was also performed using the primers BraITS-F and BraITS-R (22). PCR was performed using a Veriti 96-well thermal cycler (Applied Biosystems, Foster City, CA) under conditions described elsewhere (1, 21, 22, 26). PCR products were purified using a Wizard DNA cleanup system (Promega, Madison, WI) and directly sequenced as described previously (1). The nucleotide sequences were trimmed and assembled as described previously (1). Taxonomic assignment of the strains was performed based on their 16S rRNA gene sequences, using the Ribosomal Database Project classifier program (28). Nucleotide or deduced amino acid sequences from multiple strains were aligned with reference sequences obtained from the DDBJ/EMBL/GenBank databases. A phylogenetic tree was constructed based on the neighbor-joining method, using MEGA software version 4 (25).

Using the FSC isolation method, 68 elongated single cells were captured and transferred to the DNB-NS medium. After subsequent single-colony isolation and measurement of the denitrifying activity, 51 strains were shown to convert >20% of nitrate in the medium to N₂O gas (see Table S1 in the supplemental material) and were, therefore, considered to be denitrifiers. Two strains, TSA30 and TSA63w, reduced <20% of nitrates to N₂O but harbored nitrite reductase genes and were, therefore, also considered to be denitrifiers. Consequently, a total of 53 denitrifier strains were obtained.

Strains closely related to the genus Herbaspirillum were most frequently obtained in the present study (18 strains). The 16S rRNA gene sequences of these strains were 99.8% to 100% similar to the environmental clone sequences (e.g., clone TS33 [AB378594]) that were frequently obtained in the culture-independent studies performed using the same rice paddy soil (12, 23) (Table 1). These results suggest that the Herbaspirillum denitrifiers are most likely responsible for denitrification under the specified conditions. Strains classified as belonging to Azospirillum spp. were the second most frequently obtained (14 strains), followed by those classified as belonging to Bradyrhizobium spp. (six strains), Cupriavidus spp. (six strains), and Mesorhizobium spp. (three strains). The remaining six strains belonged to the genera Arthrobacter, Bacillus, Bosea, Pelomonas, Stenotrophomonas, and Denitratisoma. The 16S rRNA gene sequence of the Denitratisoma sp. strain TSA61 was 96.3% and 99.9% similar to the Denitratisoma oestradiolicum strain AcBE2-1^T (4) and the TS48 clone (GenBank accession number AB378598) obtained in a previous stable-isotope-probing (SIP) study (23), respectively.

TABLE 1.

Taxonomic identification of clones/strains obtained from the same rice paddy soil using culture-independent and -dependent methods

Taxona	No. (proportion [%]) of clones or strains
	SIPb	Comparative rRNA analysisc	This study
Alphaproteobacteria
Rhizobiales
Bradyrhizobium	0 (0)	0 (0)	6 (11)
Other Rhizobiales	0 (0)	0 (0)	4 (8)
Rhodospirillales
Azospirillum	1 (2)	0 (0)	14 (26)
Other Rhodospirillales	1 (2)	0 (0)	0 (0)
Betaproteobacteria
Burkholderiales
Herbaspirillum	8 (16)	82 (53)	18 (34)
Other Burkholderiales	8 (16)	14 (9)	7 (13)
Rhodocyclales
Denitratisoma	22 (43)	0 (0)	1 (2)
Other Rhodocyclales	11 (22)	11 (7)	0 (0)
Gammaproteobacteria
Stenotrophomonas	0 (0)	0 (0)	1 (2)
Deltaproteobacteria
Geobacter	0 (0)	14 (9)	0 (0)
Bacilli
Bacillus	0 (0)	8 (5)	1 (2)
Clostridia
Symbiobacterium	0 (0)	27 (17)	0 (0)
Actinobacteria
Arthrobacter	0 (0)	0 (0)	1 (2)
Total	51	156	53

^aTaxonomic identification was done based on the 16S rRNA gene by using the Ribosomal Database Project classifier program (28).

^bClones obtained from [¹³C]succinate assimilating populations under denitrification-inducing conditions, as determined by stable isotope probing (23).

^cClones specifically enriched under denitrification-inducing conditions, as revealed by comparative 16S rRNA gene analysis (12).

Of the 53 denitrifier strains obtained in the present study, 50 strains (94%) carried one or more denitrification-functional genes (see Table S1 in the supplemental material). While nirK was detected only in Mesorhizobium strains (3 strains), nirS and nosZ were detected in broad phylogenetic groups of bacteria (35 and 38 strains, respectively). nosZ was not detected in some denitrifiers (e.g., several Cupriavidus strains), while only nosZ was detected in some strains (e.g., many Azospirillum strains). PCR amplicons were not obtained in these strains, even when other primers targeting nirK, nirS, or nosZ (2, 14) were used. Sequencing the genome of Azospirillum sp. strain B510 (15) revealed that this strain possesses nirK with no annealing sites for the currently available PCR primers, suggesting the need to develop other primers to amplify nirK in this strain. It is also possible that some of these strains lack part of the denitrification pathway. For example, some denitrifiers do not have the ability to perform the final step (N₂O reduction) in the denitrification pathway (32). Strains possessing N₂O-reducing ability but no nitrite-reducing ability are also known (33). Further studies focused on designing new PCR primers, hybridization, or genome sequencing may be necessary to understand the denitrification pathway of strains shown to be negative for nirK, nirS, or nosZ in the present study.

Sequencing the 16S rRNA gene and the denitrification-functional genes (nirK, nirS, and/or nosZ) of the strains obtained in the present study revealed previously unknown links between the phylogenies of these genes (see Fig. S1 in the supplemental material). In particular, we identified Bradyrhizobium strains that harbor nirS sequences previously detected only in culture-independent studies (Fig. 1). The nirS sequences from the Bradyrhizobium strains were 100% similar to those from the TAS046 clone (GenBank accession number AB377772) obtained from a comparative clone library analysis targeting nirS (31). In addition to the microcosm experiments, similar nirS sequences were also recovered from rice paddy soils (30). For example, the nirS clone 727S150 (GenBank accession number AB453437) obtained previously was 100% similar to the nirS clones of Bradyrhizobium strains isolated in the present study. These nirS clones occupied relatively large proportions of the above-mentioned libraries (26% [30] and 45% [31]). Similar nirS sequences were also recovered from other culture-independent studies using forest (16) and arable (26) soils, indicating that these nirS sequences are ubiquitous in various environments.

FIG. 1.

Relationship between the nirS sequences of the denitrifier strains obtained in the present study. Phylogenetic trees were constructed based on the deduced nirS amino acid sequences, using the neighbor-joining method. Strains isolated in the present study are shown in bold. The accession numbers of the reference strains and the environmental clone sequences in the DDBJ/EMBL/GenBank databases are indicated in brackets. The bootstrap values (>70%) for 1,000 replicates are indicated next to each branch.

Until the present study, Bradyrhizobium strains with denitrification ability were known to carry nirK for the nitrite reductase gene (24), and no Bradyrhizobium strains were known to harbor nirS. Because nirK was not detected in our Bradyrhizobium strains, nirS in these strains most likely encodes the functional nitrite reductase NirS. This is also supported by the presence of a conserved cytochrome d₁ heme domain in the deduced amino acid sequences of these nirS genes. The deduced NirS sequences also contain amino acids equivalent to His-345, His-388, and Arg-391 of NirS of Paracoccus pantotrophus (5). While both His-345 and His-388 of NirS of P. pantotrophus form hydrogen bonds with nitrite and nitric oxide (29), Arg-391 forms hydrogen bonds with the d₁ heme (5). Further studies (e.g., mutagenesis or expression studies) are necessary to clarify the exact function of the nirS sequences in Bradyrhizobium strains.

In some genera (e.g., Azospirillum and Pseudomonas), both nirK-harboring species and nirS-harboring species are present. Similar to these genera, the genus Bradyrhizobium may also contain both nirK-harboring and nirS-harboring denitrifiers. Because the phylogenetic relationship between nirK-harboring and nirS-harboring Bradyrhizobium strains is of interest, detailed phylogenetic analysis was conducted. Based on the ITS sequence analysis, nirS-harboring Bradyrhizobium strains could be divided into two groups (Fig. 2). While nirS-Brady-I group strains were closely related to Bradyrhizobium japonicum and other nirK-harboring bradyrhizobia, the nirS-Brady-II group strain was distantly related to known nirK-harboring Bradyrhizobium strains. Because nearly identical nirS sequences were present in phylogenetically distantly related Bradyrhizobium strains, these nirS sequences may be acquired via horizontal gene transfer. The acquisition of nirS may be advantageous under certain environmental conditions because habitat-selective factors can influence nirK-harboring and nirS-harboring denitrifiers in different ways (6).

FIG. 2.

Phylogeny of nirS-harboring Bradyrhizobium strains based on the ITS sequences. The accession numbers of the reference strains in the DDBJ/EMBL/GenBank databases are indicated in brackets. The presence (+)/absence (−) of nirK and nosZ (24) is indicated next to the accession numbers.

In conclusion, the present study provides examples of oligotrophic denitrifiers carrying previously uncharacterized functional gene sequences. The environmental clone sequences of nirK, nirS, and nosZ have largely populated the databases (3); however, the denitrifiers carrying many of these sequences are unknown. Single-cell analysis and isolation would further unveil novel links between denitrifier phylogeny and functional gene phylogeny.

Nucleotide sequence accession numbers.

The nucleotide sequences of the 16S rRNA gene, nirK, nirS, nosZ, and the ITS region were deposited into the DDBJ/EMBL/GenBank databases under accession numbers AB542368 to AB542421 and AB572349, AB542297 to AB542300, AB542301 to AB542335, AB542259 to AB542296, and AB542422 to AB542428, respectively (see Table S1 in the supplemental material).