Shell Biofilm Nitrification and Gut Denitrification Contribute to Emission of Nitrous Oxide by the Invasive Freshwater Mussel Dreissena polymorpha (Zebra Mussel)

Abstract

Nitrification in shell biofilms and denitrification in the gut of the animal accounted for N₂O emission by Dreissena polymorpha (Bivalvia), as shown by gas chromatography and gene expression analysis. The mussel's ammonium excretion was sufficient to sustain N₂O production and thus potentially uncouples invertebrate N₂O production from environmental N concentrations.

TEXT

Nitrous oxide (N₂O) is a powerful greenhouse gas that contributes to stratospheric ozone destruction (3, 4). In natural systems, the production of N₂O is primarily associated with the turnover of inorganic nitrogen compounds by nitrifying and denitrifying microorganisms, often in oxic/anoxic transition zones in soil and sediment (34). Nitrifiers (both ammonia-oxidizing bacteria and archaea) produce N₂O as a by-product of ammonia oxidation (6, 28), especially under oxygen limitation, while for denitrifiers, N₂O is an intermediate in anaerobic respiration (40). Besides soils and aquatic systems, invertebrates are sites of a globally significant N₂O production, first discovered for earthworms (15, 19) and subsequently found for diverse freshwater and marine invertebrates (11, 36). This animal-associated N₂O production has been attributed to incomplete denitrification by ingested microorganisms in the anoxic invertebrate gut (13, 14, 35). In addition, biofilms covering shells and exoskeletons of marine invertebrates have been identified as sites of N₂O emission (11). Their relative contribution to animal-associated N₂O production, the pathways involved, and their distribution among marine and freshwater invertebrates are still unknown. The objective of the present study was therefore to quantify the biofilm-derived N₂O production and its mechanism(s) using the N₂O-emitting (36) freshwater bivalve Dreissena polymorpha (zebra mussel) as a model organism. This species is considered invasive in North America and Europe and can occur at extremely high abundance. Local populations in the Gudenå River system (Denmark) occasionally form large reefs at the sediment surface with more than 100,000 individuals per m² (1).

Site of N₂O production in D. polymorpha.

Mussels were sampled in April 2010 in the river Remstrup, which is part of the Gudenå system. Living animals or shells dissected from living animals were pooled in sets of 7 to 15 individuals for replicate incubations (n = 5 to 6) at 21°C in gas-tight bags (10) filled with air-saturated artificial freshwater (33) containing NH₄⁺ and NO₃⁻ (50 μM each) and a headspace of atmospheric air. Shells incubated with 50% ZnCl₂ to kill biological activity served as negative controls. N₂O emission rates were determined from linear increase of N₂O concentrations in 3-h incubations as previously described (36); in short, water samples were hourly withdrawn from the bags, transferred to N₂-flushed, ZnCl₂-containing Exetainers, and N₂O was measured by gas chromatography (36). Bags were still oxic (>50%) after the 3-h incubation, as confirmed with an O₂ microelectrode (26).

N₂O emission was approximately linear over time both in incubations of whole mussels and in biofilm-covered dissected shells; for whole mussels, the rates were similar to those for D. polymorpha collected in August 2006 in the river Rhine (36). The shell biofilm contributed approximately 25% to the total N₂O emission from D. polymorpha specimens (Fig. 1). N₂O production was an exclusively biological process, indicated by the linearity of the emissions and confirmed by the absence of N₂O emissions in the killed control.

Fig 1.

N₂O emission from living animals or shells dissected from living animals incubated in artificial freshwater with (+ ATU) or without inhibition of NH₃ oxidation by ATU (Ctrl). Error bars represent standard deviations (SD) of the mean (n = 5 to 6, and each replicate consists of 7 to 15 animals or shells). Different lowercase letters indicate significant differences between treatments (P < 0.05, t test).

Pathways of N₂O production.

Additional whole animals and dissected shells were incubated with allylthiourea (ATU; 100 μM) to inhibit NH₃ oxidation (8). N₂O emission from ATU-incubated shells was almost completely eliminated, pointing to nitrification as the dominant N₂O-producing pathway in the shell biofilm of D. polymorpha (Fig. 1). In contrast, N₂O emission from the animal itself was not reduced by ATU, which indicates that denitrification was responsible for N₂O production inside the animal, in agreement with gut-associated N₂O production via denitrification in other freshwater invertebrates (36).

These results were supported by the detection of transcripts for bacterial ammonia monooxygenase (amoA), the key enzyme of ammonia oxidation, and for nitrite reductase (nirK and nirS), a key enzyme of denitrification. RNA was extracted from dissected whole guts and from biofilm material (sampled in June 2010) with the FastRNA Pro Soil-Direct kit (MP Biomedicals) and DNase treated (Ambion) for 30 min to remove DNA, as confirmed by (lack of) 16S rRNA gene-specific PCR amplification. Reverse transcription-PCR (RT-PCR) (35 cycles) was performed with the OneStep RT-PCR kit (Qiagen). Published protocols and primers specific for bacterial amoA, amoA1F-amoAR-TC (24, 27), and for nirK and nirS, F1aCu-R3Cu (9) and Cd3aF-R3cd (21, 39), respectively, were used. Bacterial amoA mRNA was only detected in biofilm samples, while mRNAs of nirK and nirS were only detected in gut samples (Table 1). Since archaeal amoA genes were never detected by PCR (12) in any of the samples (see below), detection of archaeal amoA transcripts was not attempted.

Table 1.

Expression of genes encoding ammonia monooxygenase (amoA) and nitrite reductase (nirK and nirS) in animals collected in June and December 2010

Material	Expression ofa:
	amoA		nirK		nirS
	June	December (cDNA copies/mg wet wt)	June	December (cDNA copies/mg wet wt)	June	December (cDNA copies/mg wet wt)
Gut	−	−	+	205–1,585b	+	<240c
Shell biofilm	+	200–2,000b	−	−	−	−

^a−, not detected by RT-PCR or RT-qPCR; +, detected by RT-PCR.

^bAbove the limit of detection but below the limit of quantification for RT-qPCR.

^cBelow the limit of detection for RT-qPCR but detected and cloned after RT-PCR.

Additional animals were collected in December 2010 and analyzed by reverse transcription-quantitative PCR (RT-qPCR). Mussels were incubated for 4 h under similar conditions to those used during N₂O rate measurements. Then total nucleic acids were extracted in triplicate by a phenol-chloroform protocol (7, 25), and one aliquot of the nucleic acid extract was DNase treated as described above. cDNA synthesis with the Omniscript reverse transcription kit (Qiagen) was primed by random hexamers, and cDNA copy numbers of bacterial amoA, nirK, and nirS were quantified in a LightCycler 480 (Roche) as described previously (12). Annealing temperatures were adjusted to 55°C for nirS and to 57°C for bacterial amoA and nirK; detection limit (10 to 13 cDNA copies) was defined as 3× the standard deviation (SD) of the nontemplate control, while the limit of quantification was defined by the lower limit of the linear range of the standard curves (85 to 100 cDNA copies).

Copy numbers of all cDNAs were low (always below the limit of quantification, for nirS always below the detection limit), but confirmed the results of the qualitative RT-PCR assay for bacterial amoA and nirK: bacterial amoA cDNA was only detected in biofilm samples, while nirK cDNA was only detected in gut samples (Table 1).

To test for the metabolic potential of the biofilm and gut microbial community, gene copy numbers of bacterial amoA, nirK, and nirS were quantified in the nucleic acid extracts from December 2010. Amplification of archaeal amoA (12) was attempted several times, but the result was always negative, indicating that archaeal ammonia oxidizers were not relevant in these samples. qPCRs were performed as described above, and functional gene copy numbers were normalized against 16S rRNA gene copy numbers amplified with primer pair 341F-907R (22, 23), with annealing at 57°C. 16S rRNA gene copy numbers (per mg wet weight) were 4.83 × 10⁶ ± 6.2 × 10⁵ in the gut and 3.48 × 10⁸ ± 1.7 × 10⁷ in the shell samples. Copy numbers of all functional genes were above the limit of quantification. Relative abundance ± SD was low in gut samples: i.e., 1.6 × 10⁻³ ± 1.5 × 10⁻³ for bacterial amoA, 2.7 × 10⁻¹ ± 6.5 × 10⁻² for nirK, and 2.5 × 10⁻¹ ± 3.8 × 10⁻² for nirS. Biofilm samples showed higher relative abundances: i.e., 2.0 × 10⁻² ± 2.4 × 10⁻³ for bacterial amoA, 1.6 × 10¹ ± 9.6 × 10⁻¹ for nirK, and 1.6 × 10⁰ ± 1.2 × 10⁻¹ for nirS. These data indicate a potential for ammonia oxidation and denitrification in both gut and biofilm, if environmental conditions allow. Expression of bacterial amoA, nirK, and nirS is affected by a variety of environmental factors, including O₂ partial pressure and availability of N substrates (29, 40). Inside the mussel gut, O₂ will most likely be depleted (35). In accordance with the data presented here, denitrification will therefore be induced and ammonia oxidation repressed when denitrifiers and ammonia oxidizers, respectively, enter the gut. Mussel biofilms analyzed in this study, on the other hand, were relatively thin and presumably fully oxic, as indicated by preliminary O₂ microsensor measurements (data not shown). N₂O is therefore mainly produced by nitrification, while denitrification is repressed. However, high nir gene abundance indicates that denitrification may contribute to N₂O production, if anoxic microsites develop within the biofilm (30).

Diversity of expressed amoA, nirK, and nirS.

To assess the diversity of the active ammonia oxidizers and denitrifiers, clone libraries were constructed from cDNA of bacterial amoA (biofilm samples) and nirK or nirS (gut samples) of animals collected in June and December 2010. RT-PCR products were cloned using the pGEM-T cloning kit (Promega), with approximately 30 randomly picked clones per sample, the genes were sequenced (GATC Biotech; Macrogene), and the cDNA clone sequences were deposited in GenBank. Sequences were aligned by the integrated aligner tool in the ARB software (18) together with sequences of their closest relatives found by nucleotide BLAST, translated into amino acid sequences and used for phylogenetic tree construction in ARB using neighbor-joining and maximum likelihood analysis with 1,000 bootstrap replications. Both methods resulted in identical tree topologies.

Sequences of expressed bacterial amoA were in June 2010 affiliated with the Nitrosomonas europaea and Nitrosomonas oligotropha lineage and a lineage without cultured relatives, while in December 2010, they were affiliated with the Nitrosospira and N. oligotropha lineage (Fig. 2a). Since both clone libraries were well covered (Good's coverage of >98% based on a 97% nucleotide similarity threshold), the most probable explanation is differential activity of ammonia oxidizers at the different sampling times, possibly related to their different substrate affinities (16).

Fig 2.

Neighbor-joining tree of amino acid sequences deduced from cDNA obtained in June 2010 (red) or December 2010 (green). Sequences of bacterial amoA (a) are from shell biofilms, while sequences of nirK (b) and nirS (c) are from the gut of D. polymorpha. The number of identical sequences (at least 97% nucleotide identity) is shown in parentheses. Scale bar, 10% amino acid sequence divergence. Node symbols indicate bootstrap support by maximum likelihood analysis: closed circles, >75%; open circles, >50%.

In contrast, diversity of expressed nirK and nirS was very low, and sequences retrieved from animals collected in June and December were highly similar or identical. nirK was affiliated with Dechloromonas aromatica (87% DNA sequence similarity), while nirS genes were only distantly related (70% DNA sequence similarity) to various members of the Alphaproteobacteria: e.g., Rhodopseudomonas palustris or Rhodobacter sphaeroides (Fig. 2b and c). This limited diversity of active denitrifiers in the gut may be explained by the fact that mussels are capable of feeding on a diet of bacteria due to the high lysozyme content in their digestive organs (20, 32). Consequently, only a minor part of the ingested denitrifiers may survive and induce their denitrification genes during the gut passage in D. polymorpha.

Ammonium excretion by D. polymorpha.

The availability of NH₃ (as a substrate for ammonia oxidation) is usually low in natural freshwater systems but can be high in environments infested with D. polymorpha (5, 17). Ammonium excretion rates of D. polymorpha were measured by incubating groups of 1 to 8 living mussels (n = 6) in artificial freshwater without amendment of any N sources. NH₄⁺ concentrations were quantified spectrophotometrically (2) every half hour for a total of 3 h. The average excretion rate ± SD was 0.128 ± 0.063 μmol NH₄⁺ individual (ind) ⁻¹ h⁻¹, which is >1,000 times the N needed to explain the N₂O production by nitrification in shell biofilms. Therefore, a significant part of the mussels' N₂O emission is sustained by the animals' N excretion.

Environmental implications.

The results presented here are important on three accounts. First, they provide quantitative data for the contribution of shell biofilms to the overall N₂O emission by a benthic freshwater invertebrate, hence extending earlier qualitative observations on marine invertebrates (11). Second, with a substantial part of N₂O produced via nitrification, which can be entirely fueled by the mussels' own ammonia excretion, the data suggest that invertebrate-associated N₂O emissions can be decoupled from environmental nitrate concentrations, one of the main drivers of gut denitrification (19, 37, 38); in addition, biofilm nitrification may not only directly produce N₂O but also provide nitrate for denitrification-derived N₂O production inside the mussel.

The data also show a considerable potential of the invasive D. polymorpha to contribute to overall N₂O emissions from zebra mussel-infested ecosystems. Maximum densities of up to 100,000 individuals per m² in the river Gudenå and a potential emission rate of 144 pmol N₂O ind⁻¹ h⁻¹ amount to an emission potential of 28 μmol N₂O-N m⁻² h⁻¹ for D. polymorpha, or up to 400 times the areal N₂O fluxes reported for (noninfested) freshwater environments (31). Finally, shell biofilms, ammonium excretion, and coupled nitrification-denitrification are likely to combine also for other freshwater and marine invertebrates into significant N₂O emission potentials (I. M. Heisterkamp, A. Schramm, L. H. Larsen, N. B. Svenningsen, G. Lavik, D. de Beer, and P. Stief, unpublished data). It should however be noted that for assessment of their true environmental impact, in situ studies will be necessary, combining activity measurements and molecular analyses throughout the seasonal cycle.

Nucleotide sequence accession numbers.

cDNA clone sequences obtained in this study have been deposited in GenBank under accession no. JF820296 to JF820311.