Introduction

Nitrate is an anion that has proven to be toxic at high concentrations. In marine environments, the nitrate concentration should not exceed 20 mg NO −3 –N/l to protect fish and invertebrates from its adverse effects [7]. In closed marine facilities such as public aquariums and aquaculture farms, the nitrate concentration often exceeds this limit [18, 36]. Maintaining a nitrate concentration below 20 mg NO −3 –N/l currently involves frequent and expensive replacement of the seawater. Biological denitrification is a valuable alternative for removing nitrate from seawater [2, 18, 38, 44, 46]. Heterotrophic denitrification is a stepwise respiratory process in which microorganisms reduce nitrate (NO −3 ) or nitrite (NO −2 ) to nitric oxide (NO), nitrous oxide (N2O), and dinitrogen (N2) gases. The four reactions of the denitrification pathway are performed by nitrate reductase, nitrite reductase, nitric oxide reductase, and nitrous oxide reductase [51].

Few studies on the microbial diversity of denitrification systems treated wastewater highly charged in salt have been reported [9, 26, 48–50]. However, these studies were mainly based on the analysis of the 16S ribosomal RNA (rRNA) gene. Heterotrophic denitrification is performed by a multitude of microorganisms belonging to diverse phylogenetic groups [42]. As a result, it is not ideal to use 16S rRNA gene sequences to identify denitrifying bacteria in the environment. During the last decade, detection of the functional genes encoding the different denitrification reductases, such as narG, napA, nirK, nirS, cnorB, qnorB, and nosZ, were shown to be a more effective way to study the diversity of denitrifying populations [4, 5, 14, 17, 20, 39].

In 1998, the Montreal Biodome established a methanol-fed denitrification reactor to control the concentration of nitrate in its 3-million-liter seawater system following an increase of the nitrate concentration to a critical level of 200 mg NO −3 –N/l [36]. This completely mixed reactor contains a fluidized bed of plastic carriers that were colonized by naturally occurring microorganisms from the seawater affluent to form a denitrifying biofilm. In previous studies, we used culture-dependent and culture-independent methods to determine the composition of the microbial biota of this biofilm. Using the 16S rRNA gene library, the number of the most abundant bacterial species present in the biofilm was estimated to be between 15 and 20 [26], among which sequences related to Hyphomicrobium spp. and Methylophaga spp. were identified. Fluorescence in situ hybridization experiments confirmed that Methylophaga spp. and Hyphomicrobium spp. made up more than 50% of the biofilm microbiota [27]. Finally, the denitrifiers Hyphomicrobium sp. NL23 and the nitrate-reducing bacteria Methylophaga sp. JAM1 were isolated from the biofilm [1, 26].

Despite the abundance of Methylophaga spp. and Hyphomicrobium spp., we believed that other denitrifying bacteria and nitrate-reducing bacteria were present in the denitrifying biofilm and that these bacteria may play an important role in the denitrification process. In this paper, we further characterized the bacterial composition of the denitrifying biofilm by detecting functional genes encoding the different denitrification reductases and by determining whether they were expressed. Finally, the concentrations of the different narG, napA, nirS, and nirK sequences were determined by quantitative PCR (qPCR) to assess the level of denitrifiers and nitrate-reducing bacteria in the biofilm.

Materials and Methods

Denitrification System

The seawater system of the Montreal Biodome contains 3,250 m3 of cold (10°C) artificial seawater (28–30‰). It comprises two life support systems and four pools, as described by Parent and Morin [36]. The denitrification system is operated in parallel to the main life support system. It consisted of a 1-m3 fixed-bed deaeration tank, a 1-m3 moving bed biofilm reactor (MBBR) for denitrification, and an overflow tank. The deaeration tank and the denitrification reactor were both methanol-fed. The denitrification reactor were filled with 0.3 m3 of Bioflow® 9 mm carriers (1.020 density; Rauschert, Steinwiessen, Germany) for biofilm colonization. In the MBBR, carriers were continuously kept in motion by hydraulic flow, assuring that the carriers in the reactor were completely mixed. The water temperature in the reactor was 16–18°C, the pH was 7.8–7.9, and the salinity was 28–29‰. Twice a day, 370 mg of FeSO4·7H2O, 130 mg of MnSO4·4H2O, and 20 mg of CuSO4·5H2O were added to the system to enhance the denitrification rate [28]. At the sampling time, phosphate and nitrate concentrations in the seawater system averaged 18 PO4–P mg/l (0.6 mM) and 50 NO3–N mg/l (3.6 mM), respectively. The system’s denitrification rates varied between 200 and 300 g day−1 NOx–N. The dissolved oxygen concentration in the denitrification reactor was recorded below 0.5 mg/l during normal operation. All the carriers used in this study were taken at the same time and were frozen at −80°C without loss of activity [30].

DNA Extraction

DNA extraction from the biofilm was performed with three subsamples of the reactor using the biofilm carriers. For each subsample, the biofilm was scraped from five randomly selected carriers (to obtain a sufficient amount of DNA), pooled, and rinsed with TEN (50 mM Tris–HCl pH 8.0, 100 mM EDTA pH 8.0, and 150 mM NaCl). DNA was extracted by bead beating method, and the culture of Hyphomicrobium sp. NL23 and its DNA extraction were carried out as previously described [26].

Gene Libraries

Deduced amino acid sequences of narG, napA, nirK, and cnorB denitrification genes taken from type-strain bacteria were collected from a protein database [National Center for Biotechnology Information (NCBI) January 2006, http://www.ncbi.nlm.nih.gov/] and were aligned with ClustalW [45]. Highly degenerate primers were designed based on amino acid consensus sequences (Table 1) and were used for PCR amplifications to capture a high diversity of denitrification gene sequences. Optimization of PCR conditions (the annealing temperature, the concentration of primers, and the number of cycles) were performed with DNA extracted from Pseudomonas aeruginosa (kindly provided by Eric Déziel, INRS-Institut Armand-Frappier) for narG, napA, and cnorB, and Hyphomicrobium sp. NL23 for nirK. Further optimization was performed with the biofilm DNA. Previously published primers were used for the detection of nirS, qnorB, and nosZ (Table 1) [5, 20, 39].

Table 1 Degenerate oligonucleotides used as PCR primers
Full size table

A total of 100 ng of DNA from each biofilm subsample or 25 ng of strain NL23 DNA was used as a template for the PCR detection of narG, napA, nirS, nirK, cnorB, qnorB, and nosZ using degenerate primers (Table 1). For narG, napA, nirK, and cnorB, PCR amplifications were performed in 50 μl reaction volume with 10 mM Tris–HCl pH 9.0, 1.5 mM MgCl2, 50 mM KCl, 20 μg of bovine serum albumin (BSA), 200 μM dNTP, and 2.5 U of rTaq DNA polymerase (GE Healthcare, Buckinghamshire, UK). Amplifications were conducted as follows: 94°C for 5 min, annealing temperature for 5 min (Table 1), 30 cycles at 72°C for 1 min, 94°C for 1 min, annealing temperature for 1 min, and finally an extension period of 10 min at 72°C. Amplifications of nirS, qnorB, and nosZ were performed as described previously [5, 20, 39]. For each gene, the three PCR products from the biofilm subsamples were pooled. PCR products of the expected size were purified by agarose gel extraction with the QIAquick Gel Extraction Kit (Qiagen Inc., Mississauga, ON, Canada) before ligation.

PCR products were ligated in a T-vector (pGEM®-T Easy Vector System, Promega, Madison, WI, USA). DH5α™ Chemically Competent Escherichia coli (high efficiency 109 cfu/μg; Invitrogen Canada Inc., Burlington, ON, Canada) were transformed with DNA in accordance with manufacturer’s instructions. Plasmids were isolated by the method described by Shergini et al. [41]. In silico digestions of published sequences were performed to determine which restriction enzymes were optimal to screen the clones using the online program RestrictionMapper (http://www.restrictionmapper.org/). Plasmids were screened for different inserts by restriction fragment length polymorphism (RFLP) using the restriction endonucleases AfaI for narG, napA, nirK, and cnorB, HhaI for nirS, and Sau3AI for nosZ. Inserts of representative clones based on restriction profiles were PCR amplified and sequenced in both directions using the primers SP6 and T7. For RFLP profiles represented by numerous clones, at least two clones were sequenced.

The deduced amino acid sequences of narG, napA, nirS, nirK, cnorB, and nosZ were compared against the protein databases with BLASTP (NCBI). The most related sequences of either type-strain organisms or uncultured organisms found in marine environments were selected. The deduced amino acid sequences were aligned with ClustalW [45]. The alignments were manually refined in BIOEDIT [19]. Phylogenic analysis was performed using PROTDIST (Jones–Taylor–Thornton distance method), FITCH (Fitch Margoliash method), and SEQBOOT/CONSENSE (bootstrap analysis with 1,000 replicates) programs in PHYLIP 3.65 software package [12].

Quantification of narG, napA, nirS, nirK, and nosZ in the Biofilm

Primers specific to the each of the narG, napA, nirS, and nirK sequences retrieved from the biofilm and to the Hyphomicrobium sp. NL23 nosZ sequence were designed for the qPCR assays (Table S1). Primer specificity was verified by PCR with plasmids carrying the respective gene sequences. No cross-reactivity was observed.

qPCR was performed with SYBR green in 20 μl reaction volume with 10 μl of PerfeCTa™ qPCR FastMix™ (Quanta BioSciences, Gaithersburg, MD, USA), 2 μl of DNA (biofilm or standard), and the specific primers (Table S1). The amplifications were performed at 95°C for 5 min, followed by 40 cycles at 95°C for 30 s, at the annealing temperature (Table S1) for 30 s, and at 72°C for 30 s. After PCR, specificity of the PCR products and the presence of primer dimers were verified by performing a melt curve by increasing the temperature from 65°C to 95°C by increments of 1°C per step with a pause of 5 s for each step. Specific amplifications with biofilm DNA were observed for each of the primers pairs, and no primer dimers were detected. Quantification was performed on each biofilm DNA subsample. Reactions were performed in a Rotor-Gene 6000 real-time rotary analyzer (Corbett Life Science, Sydney, Australia).

The PCR efficiency (E) for each primer pair was calculated from slope of their respective standard curve according to the following formula: \( E = \left( {{1}{0^{ - {1}/{\text{slope}}}}} \right) - {1} \). PCR-amplified fragments from plasmid DNA were used as standards. Plasmids carrying narG, napA, nirS, or nirK sequences retrieved from gene libraries or the Hyphomicrobium sp. NL23 nosZ sequence were PCR amplified with the SP6 and T7 primers as described above. PCR products were run on 1% agarose gel, and bands of expected size were extracted and purified with the QIAquick Gel Extraction Kit (Qiagen Inc.). DNA was quantified with Quant-iT™ PicoGreen® dsDNA Reagent and Kits (Invitrogen, Molecular Probe, Eugene, OR, USA) to determine the DNA copy number. Standard curves were obtained from a 10-fold dilution series of standard DNA (108–101 copies of the targeted gene).

The impact of the presence of co-extracted contaminants in biofilm DNA on qPCR efficiency was assessed. To avoid interference from the indigenous denitrification gene sequences, a DNA fragment not related to the denitrification genes sequences was spiked with biofilm DNA. qPCR assays were performed with a 10-fold dilution series of this DNA with 10 ng of biofilm DNA in each dilution. qPCR efficiencies ranged from 0.94 to 0.95 when 10 ng of biofilm DNA was present, with a linear range from 108 to 101 copies, and was 0.98 with no biofilm DNA added.

Reverse Transcriptase (RT)–PCR Assay

Hyphomicrobium sp. NL23 was cultured in Hyphomicrobium 337a medium [per liter—1.3 g of KH2PO4, 1.13 g of Na2HPO4, 0.50 g of (NH4)2SO4, 0.2 g of MgSO4·7H2O, 3.09 mg of CaCl2·2H2O, 2.0 mg of FeSO4·7H2O, 1.0 mg of Na2MoO4·2H2O, and 0.88 mg of MnSO4·4H2O] supplemented with NaNO3 (400 mg of NO −3 /l) and 0.2% methanol under denitrifying conditions for 64 h as previously described [1]. The cultures were then centrifuged at 5,000 × g for 5 min and resuspended in TEN. One volume of acid phenol (pH 4.3) was added, and RNA was extracted as described below.

Biofilm was carefully scraped from carriers in an acid phenol (pH 4.3) bath. The phenol/biofilm mix was transferred to a screw-cap microcentrifuge tube containing 250 mg of autoclaved 0.1-mm glass beads (BioSpec Products, Inc., Bartlesville, OK, USA), and 1 vol of TEN was added. The biomass was disrupted with a Fast-Prep®-24 homogenizer (MP Biomedicals, Solon, OH, USA) twice for 20 s at 4.0 m/s. The lysate was centrifuged at 14,000 × g for 15 min, and the supernatant was extracted three times with acid phenol/chloroform/isoamyl alcohol (25:24:1) and once with a chloroform/isoamyl alcohol (24:1). RNA was precipitated with 0.25 vol of 10 M ammonium acetate and 2 vol of 95% ethanol and then dissolved in diethyl pyrocarbonate-treated water. Total RNA (30 μg) was treated twice with 6 U of Turbo DNase (Ambion Inc., Austin, TX, USA) in 1× Turbo DNase buffer at 37°C for 3 h in a 300 μl volume. Turbo DNase was inactivated, after each treatment, by acid phenol/chloroform/isoamyl alcohol extraction (25:24:1) followed by a chloroform/isoamyl alcohol (24:1) extraction and RNA precipitation. The presence of DNA in the RNA samples was evaluated by performing RT–PCR without reverse transcriptase. No DNA contamination was observed.

Primers specific to the each of the narG, napA, nirS, nirK, cnorB, and nosZ sequences retrieved from the biofilm were designed for RT–PCR assays (Table S2). Primer specificity was verified by PCR with plasmids carrying the respective gene sequences. No cross-reactivity was observed. All the primers showed PCR amplification of the expected size with biofilm DNA.

Reverse transcription (RT) of RNA (1 μg) was performed in a 50 μl reaction volume containing 50 mM Tris–HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2, 200 μM dNTP, 200 U reverse transcriptase (SuperScript™ II RT, Invitrogen, Carlsbad, CA, USA), and 100 pmol of the reverse primer (Table S2). RT was performed at 65°C for 5 min and then at room temperature for 10 min, at 42°C for 60 min, and finally at 94°C for 5 min. Five microliters of the RT reaction was used for the PCR amplification. PCR was performed in 50 μl reaction volume as described before with the rTaq DNA polymerase. Primers are described in Table S2. The amplifications were performed at 94°C for 5 min, at the annealing temperature for 5 min (Table S2), followed by 30 cycles at 72°C for 45 s, at 94°C for 45 s, at the annealing temperature for 45 s, and finally at 72°C for an extension period of 10 min. RT–PCR products were verified by sequencing.

Results

Diversity of Denitrification Genes in the Denitrifying Biofilm

Degenerate primers targeting consensual sequences of narG, napA, nirS, nirK, cnorB, qnorB, and nosZ (Table 1) were used to PCR-amplify these sequences from the denitrifying biofilm and generate gene libraries. Although PCR products with the expected length were obtained for each gene, most of the PCR amplifications also contained nonspecific amplifications, which occurs often with the use of degenerate primers [4, 5, 20]. Cloning efficiency was variable from one gene to another, and a high proportion of clones with unspecific sequences was obtained for some genes (e.g., nirS and cnorB; Table S3). These problems could be related to the degenerate primers and unspecific amplifications. For instance, despite several cloning attempts, only 36 clones were obtained for napA or nirK, respectively. Among these clones, only one napA-related sequence (napA1, 14 clones; Table S3) and one nirK-related sequence (nirK1, 10 clones; Table S3) were obtained, which suggests very low diversity in the biofilm for these genes. Overall, 25 different sequences with more than 75% similarity to the different denitrification reductases were retrieved (Table S3). Seven different sequences related to narG, one related to napA, six related to nirS, one related to nirK, three related to cnorB, and seven related to nosZ were found (Fig. 1).

Figure 1
figure 1

Phylogenetic analysis of the deduced reductase amino acid sequences retrieved from the denitrifying biofilm and the isolated strains. An unrooted phylogenetic tree demonstrated the evolutionary relationship of NarG (a), NapA (b), NirS (c), NirK (d), cNorB (e), and NosZ (f) sequences obtained in this study with representative reductase sequences affiliated with Proteobacteria. The scale bar represents amino acid substitutions per position. Bootstrap values (1,000 replicates) above 50% are indicated at the branches. GenBank accession numbers are indicated in parentheses beside the species name. Trees were inferred from a matrix of pairwise distances using aligned sequences containing 343, 189, 261, 109, 116, and 362 positions for NarG, NapA, NirS, NirK, cNorB, and NosZ, respectively. Underlined sequences are from the biofilm, and boxed sequences from the isolated strains. UB: uncultured bacteria

Full size image

The narG1 and narG2 sequences were identical to the corresponding narG sequences found in Methylophaga sp. JAM1 (also named narG1 and narG2; [1]). Deduced amino acid sequences of the five other narG sequences (Fig. 1a) were related to NarG sequences found in bacterial species affiliated with different sub-divisions of Proteobacteria. The napA1 sequence was related to NapA sequences found in bacterial species affiliated with Alpha-Proteobacteria (Fig. 1b).

The deduced amino acid sequences of nirS2 and nirS34 were related to NirS sequences found in the Gamma-Proteobacterium Kangiella koreensis species. This bacterial species is a moderate halophilic bacterium known to perform denitrification [47]. NirS13 and NirS36 were related to NirS found in the denitrifying bacterium Roseobacter denitrificans and with NirS sequences retrieved from sediments in a mariculture environment (GenBank annotation). Finally, NirS5 and NirS32 were related to NirS sequences found in a marine environment (GenBank annotation). NirK1 was related to NirK sequences found in bacterial species affiliated with Alpha-Proteobacteria, including Hyphomicrobium zavarinii.

The three cNorB sequences identified in the biofilm were related to cNorB sequences found in bacterial species affiliated with Alpha-Proteobacteria. cNorB21 was nearly identical (99%) to the corresponding sequence in Pseudovibrio sp. JE062, a bacterium retrieved from a marine sponge [11]. No PCR amplification products were obtained from the biofilm DNA with primers used to amplify qnorB. NosZ3 was related to the Hyphomicrobium denitrificans NosZ. The six other sequences (NosZ1, NosZ5, NosZ8, NosZ21, NosZ25, and NosZ39) were related to the CLEM group of NosZ sequences retrieved from Pacific Ocean sediment (GenBank annotations and [40]).

As deduced amino acid sequences of nirK1 and nosZ3 were closely related to the corresponding sequences found in Hyphomicrobium species, we performed PCR assays with the degenerate primers on Hyphomicrobium sp. NL23. Sequences similar to napA, nirK, cnorB, and nosZ were retrieved. The napA, cnorB, and nosZ sequences were identical to the ones found in the biofilm (napA1, cnorB9, and nosZ3). There was only one nucleotide difference between nirK from strain NL23 and the nirK1 retrieved from the biofilm. This difference could be an artifact introduced by PCR.

Quantification of the Nitrate- and Nitrite-Reductase Genes in the Biofilm

To assess the level of the denitrifying bacteria in the biofilm, the gene copy number of all the narG, napA, nirS, and nirK retrieved from the biofilm was determined by qPCR. We established that the Methylophaga sp. JAM1 narG1 and narG2 were dominant among all the narG genes found in the biofilm, with 1.0 × 104 and 1.6 × 104 copies/ng biofilm DNA, respectively (Table 2). The other narG genes were approximately four to 500 times less abundant. The level of the Hyphomicrobium sp. NL23 napA was similarly abundant of narG1 and narG2, with 4.5 × 104 copies/ng biofilm DNA. The abundance of the different nirS ranged from 8.9 × 100 to 4.4 × 102 copies/ng biofilm DNA, which were 100 to 10,000 less abundant than the Hyphomicrobium sp. NL23 nirK, with 8.4 × 104 copies/ng biofilm DNA. The Hyphomicrobium sp. NL23 nosZ was assayed to confirm the results obtained with napA and nirK, with 5.4 × 104 copies/ng biofilm DNA.

Table 2 Quantification of the denitrification gene sequences in the biofilm by qPCR
Full size table

Transcription of Denitrification Genes in the Bacterial Strains and in the Denitrifying Biofilm

The four denitrification genes carried by Hyphomicrobium sp. NL23 were transcribed in pure cultures (Table 3). However, in the biofilm, napA transcripts were not detected, but the other three were detected. The two Methylophaga sp. JAM1 narG were transcribed in pure cultures [1], but only narG1 transcripts were detected in the biofilm (Table 3). Among the 19 denitrification-related gene sequences associated with the uncultured bacteria in the biofilm, only the narG7 and cnorB21 transcripts were detected in the biofilm (Table 3).

Table 3 Detection of denitrification gene transcripts in the bacterial strains and in the biofilm by RT–PCR
Full size table

Discussion

Twenty-five putative denitrification gene sequences originating from the methanol-fed denitrifying biofilm were retrieved. This number of gene sequences may seem relatively low for an environmental microbiota, but it is probable that other denitrifying genes were present but not detected. At the time that the degenerate PCR primers were designed, the sequences available in the gene databases were not necessarily representative of all respective denitrifying genes, such as those found in Archaea. In addition, the degeneration of the primers could have affected the sensitivity of detecting such genes. On the other hand, the low number of denitrification gene sequences could reflect the level of diversity of the biofilm, which was estimated to have between 15 and 20 bacterial species. The biofilm has developed under controlled conditions (affluent from a closed-circuit aquarium made with artificial seawater, 18°C, methanol as carbon source, and denitrifying conditions), which may have limited the number of denitrifying bacterial species that could have been established. All these gene sequences are more closely related with corresponding genes found in bacteria affiliated with Proteobacteria. Moreover, most, if not all, of these sequences were related to corresponding reductases retrieved from marine environments.

Among the 25 gene sequences, six are identical to the corresponding genes in Hyphomicrobium sp. NL23 and Methylophaga sp. JAM1. Our study shows that Hyphomicrobium sp. NL23 does possess a complete set of denitrifying genes (napA, nirK, cnorB, and nosZ). Very few studies have reported a complete survey of denitrifying genes in Hyphomicrobium spp. Kloos et al. [25] and Fesefeldt et al. [13] reported the presence of narG, nirK, and nosZ, but not that of nirS, in several Hyphomicrobium species and strains. However, narG was not detected in several of these strains, and, as napA had not been discovered at the time these studies were conducted, it is possible that such strains contained napA. Recently, two complete genomes from H. denitrificans and Hyphomicrobium sp. MC1 were available in gene databases. Gene sequences related to the nitrate reductase/molybdopterin oxidoreductase narG and nasA (involved in nitrate assimilation) are present in each genome, but not napA. Therefore, our study shows for the first time that napA is associated with a Hyphomicrobium strain.

Seven sequences related to narG were retrieved from the biofilm. Some of the narG could come from the same microorganisms since multiple copies of distantly related narG have been reported for some bacterial species [35]. Indeed, two narG sequences retrieved from the biofilm are identical to the two narG found in Methylophaga sp. JAM1. Furthermore, some evidence supports the coexistence of both narG and napA in the same microorganism [37]. DNA from Hyphomicrobium sp. NL23, which contains the only napA found in the biofilm, did not amplify narG sequences with the narG primers used in this study.

Sequences related to nitrite reductases, key enzymes of the denitrification pathway, were detected in the biofilm DNA, of which six sequences were related to nirS and one to nirK. By compiling data of diversity studies in which genes encoding the nitrite reductases were used as functional markers of denitrification, Jones and Hallin [23] found that most nirK sequences were derived from soil but that most nirS sequences were prominently derived from marine and estuarine environments. Moreover, Hallin et al. [21] observed that the addition of methanol to a pilot-scale plant treating wastewater resulted in an increase of the nirS-type denitrifiers diversity. Yoshie et al. [50] observed that the diversity of nirS was higher than the diversity of nirK in two acetate-fed anaerobic reactors used to remove nitrate from metallurgic saline wastewater. Our results are consistent with these observations, as more nirS than nirK was found in the marine methanol-fed denitrifying biofilm. nirS13 and nirS36 were related to nirS belonging to the Roseobacter clade, which is prominent in coastal waters and in surface waters of the open ocean [32] and was associated with an industrial saline wastewater denitrification system [9]. With our degenerate primers, only the Hyphomicrobium sp. NL23 nirK was found in the biofilm. Hyphomicrobium spp. has been associated with marine methanol-fed denitrification systems [16, 33, 43].

Most quantification studies performed to date have used universal primers to quantify the entire denitrifying population and then compare the sampling sites [6, 22, 24, 31, 34]. Here, we performed a quantification of specific genes in an attempt to establish the contribution of each gene to the overall functional population of the biofilm. Assays have targeted narG and napA, which can represent either nitrate-reducing or denitrifying bacteria, nirS and nirK, which were used as markers of denitrifying populations. The most abundant genes in the biofilm were associated with the genes found in Methylophaga sp. JAM1 (narG1 and narG2) and in Hyphomicrobium sp. NL23 (napA, nirK, and nosZ). The other genes from the uncultured bacteria were 10 to 10,000 less abundant than both strains. These results confirmed our previous studies [1, 26, 27] that Methylophaga spp. and Hyphomicrobium spp. are predominant in the biofilm. The low abundance of the other genes suggests minor involvement of those in the denitrification process in the biofilm at the time of the sampling.

napA, nirK, cnorB, and nosZ of Hyphomicrobium sp. NL23 were transcribed in pure cultures under denitrifying conditions, which suggests its capacity to reduce nitrate into N2. However, only the Hyphomicrobium sp. NL23 nirK, cnorB, and nosZ transcripts were detected in the biofilm. Contrary to pure cultures, the napA transcripts were not observed in the biofilm, suggesting that strain NL23 could use the nitrite produced by the nitrate reducers present in the biofilm to support its metabolism.

Methylophaga sp. JAM1 has been shown to reduce nitrate into nitrite in denitrifying conditions, and this activity was related to the presence of both narG genes [1]. We previously showed that the two genes were transcribed in pure cultures regardless of the incubation conditions (denitrifying or aerobic conditions), suggesting a constitutive expression of these genes [1]. Furthermore, the narG1 gene product is 88% identical to NarG of the obligate chemolithoautotrophic, sulfur-compound-oxidizing Beta-Proteobacterium Thiobacillus denitrificans, one of the rare bacteria species known to couple denitrification and sulfur-compound oxidation. It has been proposed that Methylophaga sp. JAM1 acquired narG1 by horizontal gene transfer [1]. The “recent” acquisition of narG1 could have allowed Methylophaga sp. JAM1 to adapt to the Biodome denitrification system. Curiously, only narG1 transcripts were detected in the denitrifying biofilm, suggesting that it is differently regulated from narG2. Similarly, Escherichia coli has two homologous membrane-bound nitrate reductases, NRA and NRZ, which are encoded by narGHJI and narZYWV, respectively [3]. Both enzymes are expressed at basal levels, with higher expression observed at stationary growth phase, but they are regulated differently. NRA expression is stimulated by anaerobiosis and nitrate under the control of oxygen-responsive regulator FNR and the nitrate/nitrite-responsive regulator NarXL, respectively. NRZ expression is rather dependent of the stationary phase regulatory factor RpoS and is not influenced by anaerobiosis and nitrate [8]. RpoS is a stress response regulator associated with the stationary phase and is responsible for the expression of numerous genes conferring more resistance to bacteria under environmental stress conditions [29].

Among the 19 putative denitrification genes associated with the uncultured bacteria, only the narG7 and cnorB21 transcripts were detected by RT–PCR. The 17 other genes were not transcribed, or the levels of their transcripts were below the detection limit. RT–PCR primers used for these genes were able to detect the corresponding sequences in the DNA biofilm, which thus excludes a problem related to the primers’ sensitivity.

Conclusions

Our results confirmed that Methylophaga sp. JAM1 and Hyphomicrobium sp. NL23 are predominant in the biofilm, with strain JAM1 expressing narG1 and strain NL23 expressing nirK, cnorB, and nosZ. Bacteria only able to reduce nitrate to nitrite are widespread in many environments and are abundant in wastewater treatment plants [10, 15]. The low level of napA transcripts of Hyphomicrobium sp. NL23 may suggest that this strain directly used the nitrite present in the biofilm, generated by nitrate respirers such as Methylophaga sp. JAM1. Other sequences related to denitrification genes from uncultured bacteria were found in the biofilm. Although present in low amounts, bacteria carrying these genes could play an important role in the denitrification process of the biosystem, thus allowing the biofilm to adapt quickly upon changing conditions.