In this study, we related the abundance of genes encoding the archaeal biotin carboxylase and archaeal ammonia monooxygenase to dark DIC fixation rates, exploring the genetic potential for chemoautotrophy. The occurrence of archaeal accA genes and their ubiquitous distribution throughout the pelagic realm of the open ocean are highly indicative of a chemoautotrophic lifestyle via the HP/HB cycle (Wuchter et al., 2006; Berg et al., 2007, 2010; Ward et al., 2007; Auguet et al., 2008; Pratscher et al., 2011) as archaeal lipids are devoid of fatty acids.
DIC fixation in the dark ocean
Dark carbon metabolism and fixation rates in the oxygenated realm of the global ocean have been of interest for some time (Prakash et al., 1991), yet the organisms responsible for it remained ambiguous. Gradually, evidence is accumulating that DIC fixation in the dark ocean might be more prevalent than hitherto assumed (Auguet et al., 2008; Alonso-Saez et al., 2010; Hu et al., 2011; Varela et al., 2011). Measurements by Reinthaler et al. (2010) indicate dark DIC fixation rates in the meso- and bathypelagic waters of the North Atlantic amounts to about 15–53% of the export primary production. Comparative studies of prokaryotic heterotrophic biomass synthesis and DIC fixation in the dark realm of the North Atlantic revealed that DIC fixation is within the same order of magnitude as heterotrophic biomass production (Baltar et al., 2009; Reinthaler et al., 2010). Whether this DIC fixation is mainly due to autotrophic or heterotrophic microorganisms remains uncertain. Some recent studies emphasize the significance of anaplerotic CO2 conversion (Alonso-Saez et al., 2010; Lliros et al., 2011) in the context of microbial carbon cycling. Laboratory studies, however, provide contradictory evidence, showing that anaplerosis amounts to < 10% of organic carbon uptake by heterotrophic bacteria (Goldman et al., 1987; Dijkhuizen & Harder, 1995; Roslev et al., 2004). Anaplerosis, as reported by Doronina & Trotsenko (1985), appears to be stimulated by the presence of easily metabolizable organic carbon and consequently might play a minor role in the DIC fixation in deep waters (Reinthaler et al., 2010) where the organic carbon pool is largely refractory (Hansell et al., 2009; Jiao et al., 2010). A recent study indicates that members of the SAR324 cluster are ubiquitous present in the mesopelagic waters of the Atlantic and Pacific using RuBisCO as DIC acquisition enzyme (Swan et al., 2011). Hence, neither the extent of chemoautotrophy nor the potential metabolic pathways are probably completely known.
In this study, we determined the abundance of the archaeal accA gene and used it as a marker for DIC fixation via the HP/HB cycle. This cycle functions in (micro) aerobic members of the crenarchaeal order Sulfolobales containing the acetyl-CoA/propionyl-CoA carboxylase for CO2 assimilation (Ishii et al., 1996; Menendez et al., 1999; Berg et al., 2007, 2010; Fuchs, 2011). The presence of genes encoding the key enzymes in the HP/HB cycle of Thaumarchaeota suggests that these and related abundant marine Archaea (Karner et al., 2001) may use a similar cycle. Key enzymes of the HP/HB cycle, including the acetyl-CoA/propionyl-CoA carboxylase, were unambiguously identified, whereas enzymes encoding genes of other autotrophic pathways were absent in the ammonia-oxidizing sponge symbiont C. symbiosum and the free-living N. maritimus (Könneke et al., 2005; Hallam et al., 2006a; Berg et al., 2007).
Q-PCR profiles of accA genes assessed in the western and eastern part of the longitudinal transect of the RFZ revealed highly similar tendencies in their distribution along the depth strata (Fig. 3). The accuracy of determining the gene abundance of the accA subunit by Q-PCR depends on the efficiency of the Q-PCR and a reliable standardization of the method. The observed Q-PCR efficiencies varied depending on the primer pair used and were lower amplifying the accA gene (Yakimov et al., 2009) than other genes quantified in this study (Table 2). Determination of gene numbers and differences in efficiencies can be caused by a number of factors, such as initial extraction of nucleic acids, preparation and amplification of the standard curve template (Love et al., 2006), the presence of reverse transcriptase, reaction reagents, fragment size of the target gene and PCR inhibitors. AccA primers, applied in DGGE analyses (Yakimov et al., 2009) as well as Q-PCR assays (Hu et al., 2011), yield rather long amplicons (Table 2), possibly impairing PCR efficiencies. However, amplification efficiencies were constant throughout our measurements and hence reproducible. We are confident that reliable estimates of the different gene abundances were achieved in this study. DNA extraction efficiencies were consistent, yielding DNA concentrations proportional to total cell counts (Fig. 2) within discrete depth layers (Table S2). Dilution series were performed a priori to check on PCR inhibitors empirically (see Supporting Information, Data S1). Variance in Ct values of triplicate reactions was higher when gene abundance was high and ranged between 102 and 104 gene copy numbers (Supporting Information, Table S1). Fierer et al. (2005), however, showed that varying DNA concentrations do not result in significant changes in gene abundances, supporting our conclusion of accurate gene numbers using the Q-PCR assay.
Spatial variations in gene abundances
Prokaryotic abundance (Fig. 2) and dark DIC fixation (Table 3, Fig. S1) declined exponentially from surface to abyssopelagic layers as shown in previous studies from the North Atlantic (Reinthaler et al., 2006; Teira et al., 2006a; Varela et al., 2008). Also, a pronounced stratification in the composition of prokaryotic communities in the meso- and bathypelagic realm in the Atlantic and the Pacific has been reported (DeLong et al., 2006; Aristegui et al., 2009; Agogue et al., 2011). In this study, thaumarchaeal 16S rRNA gene abundance in the deep waters of the tropical Atlantic was within the range of abundances previously found (Varela et al., 2008; De Corte et al., 2009). However, thaumarchaeal 16S rRNA gene abundance profiles revealed a patchy spatial distribution in the upper mesopelagic layer and a more homogenous distribution below 1750 m depth (Fig. 2b). The percentage of thaumarchaeal 16S rRNA genes to total picoplankton counts (determined by flow cytometry) ranged between ~7% and < 23% in the AABW, the oxygen minimum layer and in the lNADW. Similar abundances were also reported using CARD-FISH (catalyzed reporter deposition-fluorescence in situ hybridization) (Varela et al., 2008). Hence, independent of the quantification method used, a coherent pattern in the distribution of Thaumarchaeota in the different depth layers of the Atlantic is evident.
In the central-eastern part of the RFZ, the high abundance of thaumarchaeal 16S rRNA genes down to 1750 m depth (Fig. 2b, Sts. 15–23) coincided with high DIC fixation rates (10.5–14 μmol C m−3 days−1) and a high AOU (~172 μmol O2 kg−1). While the high AOU may indicate an increased heterotrophic activity, the high DIC fixation rates are likely due to the activity of chemoautotrophs rather than anaplerotic reactions of heterotrophs because the autotrophic activity is similar to the measured heterotrophic prokaryotic biomass production (De Corte et al., 2010).
Correlating the abundance of biotin carboxylase α-subunit genes to the 16S rRNA gene abundance of Thaumarchaeota, ratios close to one were obtained in the central parts of the RFZ. The high abundance of accA genes below 2000 m depth at Sts. 23 and 25 (Fig. 2c), however, is not reflected in enhanced DIC fixation rates (measured down to 4550 m depth, Fig. S1) or elevated concentrations of ammonia and/or nitrite (data not shown). This might indicate that the ecological function of Crenarchaeota is changing across the depth strata or alternative pathways of DIC fixation are followed. Similar dynamics have been shown by Grzymski et al. (2012), reporting seasonal shifts of chemolithoautotrophic organisms in surface waters of the Antarctic Peninsula.
A positive relation of thaumarchaeal 16S rRNA gene abundance and the concentration of ammonia (Wuchter et al., 2006; Kirchman et al., 2007; Varela et al., 2008) as well as nitrite (Teira et al., 2006a; Lam et al., 2007) has been reported for surface and mesopelagic waters, tentatively indicating chemolithoautotrophy of ammonia-oxidizing Thaumarchaeota. Dark DIC fixation rates normalized to accA gene abundance varied over three to five orders of magnitude over all depths sampled (data not shown) reflecting either large variations in the expression of ACCase and/or a large variability in the enzymes responsible for CO2 fixation in the deep ocean.
Relating the gene abundance of archaeal amoA and thaumarchaeal 16S rRNA genes in the upper 250 m revealed a ratio of accA/thaumarchaeal 16S rRNA gene of ~ 1.2 averaged over the whole transect of the RFZ (Table 3). Similar values were reported by Alonso-Saez et al. (2012) in polar oceans and Mincer et al. (2007) for the HOTS station in the Pacific, whereas Beman et al. (2008) and Wuchter et al. (2006) obtained accA/thaumarchaeal 16S rRNA gene ratios of about 2.5 and 2.8 in the Gulf of California and the North Atlantic, respectively. Significantly lower accA/thaumarchaeal 16S rRNA gene ratios (Table 3) were obtained from the lower meso- and bathypelagic waters of the RFZ, tentatively indicating that other energy sources than ammonia might be utilized in the lower meso- and bathypelagic waters.
Archaeal amoA and accA gene abundances were only weakly correlated over the whole depth range (Fig. 4). A metagenomic survey by Hallam et al. (2006a) reported a gene ratio of accA/amoA close to one in surface waters. In our study, the exclusion of the 100-m samples, however, resulted in a tighter relationship of archaeal amoA to accA gene abundances. These differences might be caused by the specificity of the primers designed based on a few sequences (Yakimov et al., 2009), possibly failing to target accA homologues of epipelagic Archaea. Data indicative of a vertical stratification of Crenarchaeota in a ‘shallow-’ and ‘deep-water clade’ have recently been suggested based on accA gene diversity (Hu et al., 2011; Yakimov et al., 2011). A genomic and transcriptomic survey in the deep waters of the Gulf of California also revealed multiple populations of Crenarchaeota (Baker et al., 2012). Based on our data, the rather weak correlation between genes encoding the archaeal ammonia monooxygenase and acetyl-CoA carboxylase suggests that the HP/HB cycle might be fuelled at least to a certain extent by energy sources other than ammonia.
The overall ratio of the acetyl-CoA/propionyl-CoA carboxylase gene to thaumarchaeal 16S rRNA gene abundance supports the assumption that a major fraction of Thaumarchaeota is capable of fixing DIC via the HP/HB cycle. Assuming the accA gene is present as a single copy in the crenarchaeal genome (Könneke et al., 2005; Hallam et al., 2006a), its abundances matches roughly those of thaumarchaeal cells in the SACW and AAIW (Fig. 3). In the water layers of the NADW and AABW, the ratio of accA/thaumarchaeal 16S genes increased to 3.8 and 5.5, respectively. Highest accA gene abundance (1.14 × 105 genes mL−1) was obtained in the oxygen minimum zone (200–750 m depth) corresponding to the highest abundance of thaumarchaeal 16S rRNA genes. In a previous study, high accA gene abundance (2.08 × 103 ± 312 mL−1) was found at 3000 m depth in the Tyrrhenian Sea (western Mediterranean Sea) and a ratio of accA/thaumarchaeal cells of about unity (Yakimov et al., 2009).
Overall, the gene abundances obtained by Q-PCR of archaeal accA, archaeal amoA and thaumarchaeal 16S rRNA genes reveal similar trends of gene abundances with depth. The high abundance of thaumarchaeal 16S rRNA genes in the oxygen minimum layer might reflect the importance of Thaumarchaeota in the biogeochemical processes, particularly nitrification, occurring in this layer. However, also some groups of mesopelagic Bacteria such as members of the SAR324 or the SUP05 cluster contribute to autotrophy in the dark ocean (Swan et al., 2011).
Taken together, this study, along with some other recent work, demonstrates that there is a substantial genetic predisposition of DIC fixation present in the dark ocean. In this context, it adds to the emerging notion that dark ocean DIC fixation might be more important than hitherto assumed. The extent to which this DIC assimilation is due to anaplerotic reactions of heterotrophic prokaryotes or true autotrophy remains to be shown. Additional studies, including the quantification of other related genes and proteomic approaches, are needed to shed light on the modes of archaeal carbon metabolism. The sheer magnitude of DIC fixation, roughly equalling heterotrophic production in the North Atlantic deep waters, however, might indicate that the anaplerotic metabolism of heterotrophic prokaryotes is of minor importance and that autotrophic processes prevail in the DIC fixation pathway in the Atlantic's interior.