Microbial nitrogen cycling in Arctic snowpacks

Samples were processed immediately after collection in the field laboratory and were left to melt (6 h) at room temperature prior to being filtered onto sterile 0.22 μm 47 mm filters (Millipore) using a sterile filtration unit (Nalge Nunc International Corporation). Between 2 and 3 l of melted snow were filtered per sample. Filters were stored and transported to the laboratory in France in sterile bead-beating tubes at −20 ° C until further analysis. Procedural blanks were carried out by filtering Nanopure water (Siemens) using the same procedure. DNA was obtained from the filters using Fastprep^® bead-beating tubes (Lysing matrix E, MP Biomedicals) and a Fastprep^® bead-beater (MP Biomedicals). DNA was extracted from the aqueous phase with an equal volume of chloroform: isoamyl alcohol (24:1) and precipitated with isopropanol. Detailed descriptions of the method are provided in [24].

The Agilent Sureprint Technologies microarray format was used, and consisted in 8 identical blocks of 15 000 spots each on a standard glass slide format 1'' × 3'' (25 mm × 75 mm). A total of five slides were used for the hybridization of all samples. Probes were designed to target the rrs gene and to cover different taxonomic levels from a wide part of the Bacteria and Archaea phylogenic tree using the ARB software package. Details concerning microarray design and DNA amplification are provided in [24, 25]. Briefly, rrs genes were amplified by PCR from total DNA extracted from each sample, using universal primers pA (5' AGAGTTTGATCCTGGCTCAG 3') and pH-T7 (5' AAGGAGGTGATCCAGCCGCA 3') (universal for most Bacteria and some Archaea) and the illustra Hot Start Mix RTG (GE Healthcare) PCR kit. The PCR conditions used were 3 min at 94 ° C, followed by 35 cycles of 45 s of denaturation at 94 ° C, 45 s of annealing at 55 ° C, and 90 s of elongation at 72 ° C. After a final 5 min extension 72 ° C, PCR products were separated by 1%-agarose gel electrophoresis, purified using the NucleoSpin^® Extract II kit (Clontech) and transcribed. In vitro transcription was carried out at 37 ° C during 4 h in 20 μl reactions that contained 8 μl of the purified PCR product (50 ng μl⁻¹) and 12 μl of the following mix: T7 RNA buffer (5×), DDT (100 mM), 10 mM of each of the four NTPs, RNasin (40 U μl⁻¹), T7 RNA polymerase (1 μl) and UTP-Cy3 (5 mM). In vitro transcription was carried out at 37 ° C during 4 h. RNA was purified using the Quiagene RNeasy mini Kit according to the manufacturer's instructions and quantified with a nanophotometer before undergoing chemical fragmentation. The RNA solution was then diluted to 5 ng μl⁻¹ and a hybridization mix was prepared (v/v ratio) in a 50 μl reaction with 2× GeX Hyb Buffer (Agilent). A total of 100 ng of RNA were then placed on the slide and incubated at 60 ° C overnight in the Agilent Hybridization Oven. Microarray were washed with the GE wash Buffer kit (Agilent) according to the manufacturer's instruction, and dried by one minute incubation in acetonitrile. An Innoscan 700 scanner (Carbonne, France) was used for scanning microarray slides according to the manufacturer's instructions.

The DNA (2 μg 50 μl⁻¹) extracted from environmental samples was pyrosequenced by GATC (Constance, Germany) using a Roche 454 Titanium pyrosequencer. Nine samples (seven surface and two basal samples) distributed throughout the sampling season were selected for pyrosequencing. The selection process was based on clustering data from both the chemical and 16S rRNA gene microarray data sets such that members of different clusters were represented in the pyrotag run. Since the required amount of DNA for pyrosequencing is high (2 μg 50 μl⁻¹), the metagenome extracted from each sample for pyrosequencing was amplified using multiple displacement amplification with the illustra™ GenomiPhi™ V2 DNA Amplification Kit (GE Healthcare). Amplification was carried out according to the manufacturer's instructions. The fasta sequences obtained were filtered for errors using cd-hit, blasted against the NCBI-NR database using the BLASTX default settings [26, 27] and analyzed using MEGAN 4.69 [28]. Data were also annotated in MG-RAST [29] to allow for comparisons with publicly available metagenomes from various ecosystems (see [30] for details of the methodology and a description of the different ecosystems compared).

An in-depth discussion on the snowpack chemical characteristics and analysis is provided in [6]. We used the nitrogen chemical species data from this data set to explore N cycling in more detail. Briefly, inorganic ion (NO₃⁻,NO₂⁻,Cl⁻,SO₄²⁻,NH₄⁺,Ca²⁺,Na⁺,K⁺ and Mg²) concentrations were measured at the Laboratoire de Glaciologie et Géophysique de l'Environnement by conductivity-suppressed Ion Chromatography using a Dionex ICS 3000.

JMP 9.0 software (SAS Institute, 2003) was used to detect associations between the different chemicals and relative probe fluorescence at the genera level. Nitrogen cycling gene abundance analysis was carried out in MEGAN version 4.69 [28]. Ecosystem comparisons were carried out using STAMP [31].

Based on previous work describing the interaction between chemistry and microbial community structure [24], we were able to observe that the snowpack is stratified in terms of its chemistry and metabolic potential. A detailed description of the evolution of the snowpack is described elsewhere [6], but the snowpack was 45 cm thick at the beginning of sampling and 20 cm thick when the last basal samples were taken before the snowpack melted completely. Probes targeting organisms with the potential to carry out nitrogen cycling were more represented at the base of the snowpack [24] and basal samples were enriched in N chemical species relative to surface samples [6]. Based on the pyrosequencing data, functional genes associated with N cycling were identified in all sample types (table 1). From the chemical data, we observed atmospheric transport and both dry and wet deposition of NH₄⁺,NO₃⁻ and NO₂⁻ to the surface of snowpacks. For example, the late peak in NO₃⁻ levels observed in surface samples was related to a fresh snowfall event (figure 2). The sampling frequency, however, did not allow for the observation of rapid photochemical processes occurring at the snow–atmosphere interface (such as NO₃⁻ photolysis and redeposition as HNO₃, or NO₂⁻ photolysis). The nitrogen species varied as a function of time in the snow cover (figure 2). This is consistent with other reports that suggest that the main N input to snow systems is atmospheric deposition, specifically wet deposition [33, 34]. The changes in N chemical species at the base of the snowpack are likely due to a combination of physical, chemical and potentially microbial processes, including transfer from the surface of the snowpack (by elution, burial and wind pumping) and in situ production (figure 2). The nitrogen cycling observed at the base of the snowpack is probably unrelated to photochemistry given that UV wavelengths are quenched within the first few centimeters of the snow [35]. However, photosynthetically active radiation (PAR) has been measured up to a depth of 2 m, implying that photoactive responses, such as photosynthesis, can occur at the base of snowpacks [36].

Zoom In Zoom Out Reset image size

Nitrogen fixation is a high energy process requiring 16 molecules of ATP to break the triple N bond. This process is carried out by only a few organisms including Cyanobacteria (e.g. Anabaena, Nostoc) and certain genera that form mutualistic associations with plants (e.g. Rhizobium) [37]. While we were able to detect phylogenetic probes that target genera capable of nitrogen fixation in most samples (SI table 1 available at stacks.iop.org/ERL/8/035004/mmedia), only a low number of reads associated to N₂ fixation (about 2% of annotated reads) were detected in the pyrosequencing data set (table 1). This might be linked to the elevated NH₄⁺ concentrations in the snow. Telling et al [38] also reported a lack of N₂ fixation in cryoconite holes despite the presence of N₂ fixing populations, which were attributed to high ammonium concentrations [38]. Ammonium, even at concentrations below 8 μmol l⁻¹, has been shown to inhibit N₂ fixation in some Cyanobacteria genera through the repression of nitrate reductase synthesis [39]. This falls well below the range of NH₄⁺ concentrations observed in our snowpack (figure 2). In addition, microorganisms will preferentially use alternative forms of bioavailable N over N₂ [40]. The anthropogenic N delivered by atmospheric transport appears to be preferentially metabolized, thus leading to a shift from N₂ fixation to bioavailable N assimilation and metabolism. However, the genetic and taxonomic potential for N₂ fixation is present in snow and this pathway may be initiated under the appropriate environmental conditions (i.e. low ammonium concentrations).

Based on pyrosequencing results, almost half of the reads involved in N cycling were associated with ammonium assimilation (43.5%) and over a quarter (25.7%) with nitrate and nitrite ammonification (figure 1). The assimilatory process includes the conversion of nitrate to ammonia and starts with the reduction of nitrate to nitrite and then ammonium, which is used by the cell to incorporate nitrogen into biomolecules [41]. Reads corresponding to assimilatory nitrate reduction enzymes were detected in all but one sample (SI figure 1 available at stacks.iop.org/ERL/8/035004/mmedia), however, the enzymes involved varied across samples. For example, in samples dominated by eukaryotic and photosynthetic sequences (SVN35, SVN40), reads corresponding to EC 1.7.1.1 (plant nitrate reductases) and 1.7.1.3 (fungal nitrate reductases) were identified, while bacterial nitrate reductases (EC 1.7.7.2 and 1.7.99.4) were detected in all other samples. Once NO₃⁻ is converted to NO₂⁻, nitrite can undergo either denitrification or ammonification to ammonium (figure 1). Based on the pyrosequencing results, reads associated with both plant/cyanobacterial (EC 1.7.7.1) and microbial (EC 1.7.1.4) nitrite reductases were identified in all samples (SI figure 1) and organisms able to carry out these reactions were observed using both microarray and pyrosequencing approaches. Assimilation appears to be a ubiquitous process in the snow, carried out by different phyla in all snow types.

Two ammonia/ammonium incorporation pathways have been identified in microorganisms: the NADP-dependent glutamate dehydrogenase (GDH) (EC 1.4.1.3), and the two-step energy dependent glutamine synthetase (GS) and NADP- (or NAD-)dependent glutamate synthase (GOGAT) (EC 6.3.1.2 and EC 1.4.1.13, respectively) [42]. Both pathways, which may be common to bacteria, cyanobacteria, algae, yeasts and fungi, lead to the net synthesis of one molecule of glutamate from one molecule of ammonia [43], while GS catalyzes the only pathway that leads to the formation of glutamine [44, 45]. Both these molecules constitute central nitrogen intermediates and provide N for the synthesis of all other N-containing components: ≈88% of cellular N comes from glutamate, and the rest comes from glutamine [46]. At low NH₄⁺ concentrations, the GS-GOGAT pathway is essential for glutamate synthesis and for regulation of the glutamine pool, and is used when the cell is not under energy limitation [44, 45]. The GDH pathway is used in glutamate synthesis when the cell is limited for energy (and carbon; i.e. glucose-limited growth) but ammonium and phosphate are present in excess [44, 45]. Reads related to both pathways were identified in all samples and corresponded to enzymes from multiple phyla (SI figure 1). We also observed changes in the relative abundance of probes targeting efficient ammonium scavengers such as Caulobacter [47] that mirror those of ammonium concentrations at the base of the snowpack (figure 3). This co-variation may be due to a source effect rather than active microbial processes in the snowpack, however, most of the data presented in figure 3 represent sample points from the basal snow layer before the snowpack went isothermal. A thick ice layer was present throughout, making potential transfers from the soil to the base of the snowpack unlikely, but transfer from the surface of the snow is a possibility.

Zoom In Zoom Out Reset image size

Nitrification is a two-step process carried out by a restricted number of autotrophic bacteria and Archaea. Although both steps are tightly coupled, each step is carried out by distinct microbial populations. The first step, carried out by organisms such as Nitrosomonas and Nitrospira (among others), is the oxidation of ammonium to nitrite, while the second involves the oxidation of nitrite to nitrate, which has been observed in Nitrobacter [37]. Both reactions are strictly aerobic and sensitive to acidic conditions. We were unable to detect probes that target any of the nitrifying genera in the microarray data (SI table 1) or any of the genes encoding the enzymes involved (ammonia monooxygenase, hydroxylamine oxidoreductase and nitrite oxidoreductase) in the pyrosequencing data set. This lack of nitrification evidence implies that it was not an important process in the snowpack at the time of sampling. This is consistent with reports by others who were unable to observe nitrification in the snowpack [48, 49], although Amoroso et al did propose two pathways for explaining the microbial isotopic signature observed in winter snow at Ny-Alesund: nitrification and denitrification [14]. If nitrification does occur in snowpacks, it might during another period of the year or it might not occur at any significant level due to anthropogenic inputs of nitrate and nitrite.

The complete denitrification of nitrate to N₂ is catalyzed by separate nitrogen oxide reductases and generates nitrite, nitric oxide (NO), and nitrous oxide (N₂O) intermediates [50, 51]. This process is facultative and expression is initiated in the cell by low oxygen tension and N oxide availability [50]. Reads associated to denitrification represented 7.2% of sequences assigned to N metabolism in our pyrosequencing data (figure 1). We detected genes encoding all the required enzyme complexes, but also probes that targeted genera whose members are able to carry out the denitrification reactions such as Roseomonas (figure 3), whose relative probe abundance was anti-correlated to nitrate concentrations and correlated to nitrite concentrations, suggesting that nitrate is being assimilated in order to produce nitrite. Similar patterns were observed for other known denitrifiers such as Bacillus [37]. Again, it is possible that this co-variance is linked simply to co-deposition or to a source effect (see discussion above for Caulobacter). As denitrification is carried out in environments with reduced oxygen concentrations, the snowpack likely contains some anaerobic microniches. We reported that the highest proportion of anaerobes was observed at the base of the snowpack [24]. Rohde and Price reported the existence of separate, isolated microenvironments within the same ice that explained the apparent concomitant aerobic and anaerobic metabolism [52]. Some dissimilatory ammonification from nitrate and nitrite has been observed in other environments under both aerobic and anaerobic conditions (e.g., Bradyrhizobium [53] and Bacillus [54]). The snow pack microbial community appears to be as rich in genes associated with this function as any other ecosystem analyzed (figure 4) based on comparative metagenomic analyzes described previously [30]. The combination of these observations and the significant deposition of nitrate and nitrite support the notion that the snowpack is actively transforming these compounds within the nitrogen cycle.

Zoom In Zoom Out Reset image size

We were also able to detect genes involved in nitrosative stress response in our pyrosequencing data set (9.6% of N metabolism associated reads). The presence of these genes supports the hypothesis that microorganisms may be actively involved in denitrification, since nitric oxide (NO), a free radical, is a by-product of the denitrification process. NO is mutagenic, inhibits respiration, and can interact with a wide variety of cellular targets in eukaryotes and prokaryotes such as electron transport chains by binding metalloenzymes [50]. The NO produced during denitrification has been shown to be toxic to neighboring microorganisms in the environment and was suggested to influence the metabolic activity of microbial communities [55]. Upon NO exposure, a variety of genes that encode protective molecules can be induced [56]. In addition, biologically produced NO may be more effective in eliciting a stress response than other NO-generating compounds [55]. The presence of these genes suggests a functional response within the community to NO stress.