|Home | About | Journals | Submit | Contact Us | Français|
Recent studies highlight the diversity and significance of marine phototrophic microorganisms such as picocyanobacteria, phototrophic picoeukaryotes, and bacteriochlorophyll- and rhodopsin-holding phototrophic bacteria. To assess if freshwater ecosystems also harbor similar phototroph diversity, genes involved in the biosynthesis of bacteriochlorophyll and chlorophyll were targeted to explore oxygenic and aerobic anoxygenic phototroph composition in a wide range of lakes. Partial dark-operative protochlorophyllide oxidoreductase (DPOR) and chlorophyllide oxidoreductase (COR) genes in bacteria of seven lakes with contrasting trophic statuses were PCR amplified, cloned, and sequenced. Out of 61 sequences encoding the L subunit of DPOR (L-DPOR), 22 clustered with aerobic anoxygenic photosynthetic bacteria, whereas 39 L-DPOR sequences related to oxygenic phototrophs, like cyanobacteria, were observed. Phylogenetic analysis revealed clear separation of these freshwater L-DPOR genes as well as 11 COR gene sequences from their marine counterparts. Terminal restriction fragment length analysis of L-DPOR genes was used to characterize oxygenic aerobic and anoxygenic photosynthesizing populations in 20 lakes differing in physical and chemical characteristics. Significant differences in L-DPOR community composition were observed between dystrophic lakes and all other systems, where a higher proportion of genes affiliated with aerobic anoxygenic photosynthetic bacteria was observed than in other systems. Our results reveal a significant diversity of phototrophic microorganisms in lakes and suggest niche partitioning of oxygenic and aerobic anoxygenic phototrophs in these systems in response to trophic status and coupled differences in light regime.
Recent studies have discovered novel phototrophic organisms and pointed us to their diversity in the oceans (1, 2, 3, 27, 39, 47, 50). Microorganisms such as picocyanobacteria, picoeukaryotes, and bacteriochlorophyll- and rhodopsin-containing bacteria use diverse photopigments to photosynthesize. These organisms represent a significant fraction of marine microbial communities and are likely to be ecologically and biogeochemically significant (1, 2, 13, 27, 28, 35, 39, 47). Several molecular studies based on genes of the puf operon, coding for the bacteriochlorophyll subunits, have shown that Roseobacter and Roseobacter-like bacteria constitute a significant portion of the aerobic anoxygenic photosynthetic bacteria (AAnPB) in marine waters (37). Microscopic counts have revealed that AAnPB contribute 1 to 16% to the total marine bacterioplankton in the euphotic zone and that there are regional and temporal differences in their abundances (see, for example, references 28, 41, and 50). Still, the global significance of this functional group and the role of, for example, AAnPB in the oceanic flow of energy and carbon are controversial (13, 20, 27, 28, 41, 48). Most AAnPB isolated so far have been described as photoorganoheterotrophs that rely primarily on organic substrates for growth but can derive a significant portion of their energy requirements from solar radiation (references 13 and 26 and references therein).
AAnPB have been isolated from various freshwater habitats, ranging from cyanobacterial mats (49) to the pelagic zone of oligotrophic lakes (19, 38), but there are so far no studies of freshwater AAnPB diversity and community composition based on culture-independent techniques. A recent survey revealed some first patterns in the distribution of bacteriochlorophyll a-containing cells as well as the concentrations of the pigments in lakes ranging from oligotrophic to eutrophic. Infrared epifluorescence microscopy, fluorescence emission spectroscopy, and high-performance liquid chromatography were used to demonstrate that AAnPB may constitute up to 80% of total bacterial biomass in some low-productive lakes, implying that they are an important component of many lake ecosystems (33). In addition, genes encoding proteins for light harvesting (bacteriochlorophyll pufL and pufM gene clusters) have been identified in fosmid libraries from bacteria of the Delaware River (48) and in a functional gene survey of an Antarctic lake (24).
In the present study, we used a specific primer set that amplifies the L subunit of the dark-operative protochlorophyllide oxidoreductase (L-DPOR) and its homologs (nitrogenase and chlorophyllide oxidoreductase [COR]). The dark-operative protochlorophyllide oxidoreductase (DPOR) is encoded by three genes (chlN-bchN, chlB-bchB, and chlL-bchL) and catalyzes the reductive formation of chlorophyllide from protochlorophyllide during biosynthesis of chlorophylls (chl) and bacteriochlorophylls (bch) in the dark (see reference 7 for more detail). Analysis of the deduced amino acid sequences indicated the presence of significant sequence dissimilarity in DPOR between oxygenic and anoxygenic photosynthetic organisms (5, 8, 16, 17, 18). Molecular studies have shown that AAnPB contain only DPOR (15) but that cyanobacteria, algae, and gymnosperms (nonflowering plants) contain both DPOR and a light-dependent protochlorophyllide oxidoreductase (POR) which carries one of the only two known enzymes other than photochemical reaction centers with light-driven catalysis (7).
Despite the lack of POR, AAnPB are able to modify their chlorophyllide so that their absorption spectrum is broadened to span from <350 to <1,050 nm (usually 365 to 770 nm) in the UV and near-infrared ranges. This spectral characteristic allows AAnPB to utilize light at a wavelength other than that utilized by chlorophyll-containing cyanobacteria and algae. The first step in a series of modifications is performed by COR, an enzyme that is found in anoxygenic phototrophic bacteria and that transforms chlorophyllide to bacteriochlorophyll (7).
In the present study, we assessed the diversity of L-DPOR, COR, and the coamplified homolog nifH genes in bacteria of seven different lakes by using PCR-based clone libraries parallel to a molecular fingerprinting technique to study L-DPOR gene composition in a larger data set (comprising 20 Swedish lakes). By using L-DPOR genes as our target, we simultaneously assessed the compositions of both AAnPB and oxygenic phototrophs in freshwater ecosystems and their distributions along trophic gradients.
A total of 20 Swedish lakes, situated in different climate zones with the southernmost lake, at latitude 58°28′N, and the northernmost lake, at latitude 68°26′N, were sampled on one occasion in the summer of 2006, except for one lake (Lilla Ullfjärden), which was sampled in the summer of 2002 (Table (Table1).1). For 16 of the lakes, depth-integrated samples were collected. For three stratified lakes, water samples were collected from both the epilimnion and the hypolimnion, and from oligotrophic lake Vättern, water samples were collected from three discrete depths (0.5, 10, and 28 m). These samples were analyzed separately, and hence, a total of 25 samples were used in the survey. Bacterial production was estimated from incorporation of radiolabeled leucine into proteins as described elsewhere (11). Total organic carbon (TOC) and dissolved organic carbon (DOC) levels were measured by high-temperature catalytic combustion, and nutrient concentrations were measured with standard colorimetric methods as previously described (14, 23). In situ photosynthetically active radiation (PAR) was measured at discrete depths, from the surface to the bottom or down to the depth where PAR was absent, using either a LI-193 Spherical Quantum Sensor (LI-COR Biosciences, Inc.) (for lakes at latitudes of >64°) or an International Lights IL-1400 radiometer equipped with a cosine-corrected submersible PAR sensor (for lakes at latitudes of <64°). The vertical attenuation coefficient (Kd) was calculated from the slope of the linear regression of the natural logarithm of PAR versus depth.
Bacterial cells for DNA extraction were captured on 0.2-μm membrane filters (Supor; Gelman) by vacuum filtration (<30 kPa) of between 0.25 and 2 liters of lake water. The filters were immediately frozen and stored at −80°C until analysis. DNA was extracted by bead beating using an Ultra Clean soil DNA kit (MoBio laboratories; Solana Beach, CA) in accordance with the manufacturer's instructions for high yield. Electrophoretic separation and comparison to a 1-kb DNA extension ladder (Invitrogen) on a 1% agarose gel suggested DNA concentrations of less than 10 ng μl−1.
Terminal restriction fragment length polymorphism (T-RFLP) of PCR-amplified DNA fragments was used to assess the composition of DPOR genes in the 25 samples from 20 lakes. Primers GKGGIGKSfwd (5′GGHAARGGHGGHATHGGNAART-3′) and VCGGFAMPrev (5′GGCATNGCRAANCCVCCRCANAC-3′) (34) were used in a PCR. This primer pair amplifies L-DPOR of oxygenic (chlL) and anoxygenic (bchL) phototrophs, COR, and nifH genes from mixed genomic DNA extracts (37). The forward primer was labeled with hexachlorofluoroscein at the 5′ end (Sigma Genosys) to enable fluorescence detection of terminal restriction fragments (T-RFs). Two microliters of DNA extract was used as a template in 20-μl PCR mixtures amplified in a Peltier thermocycler (Bio-Rad Chromo4) with the following settings: initial denaturation at 92°C for 4 min; followed by 20 cycles of a touchdown protocol, with 30 s at 92°C, 30 s at an annealing temperature of 52°C (which decreased by 0.22°C for each cycle), and a 45-s primer extension at 68°C; followed by a final 10 cycles, with 30 s at 92°C, 30 s of annealing at 48°C, and a 45-s primer extension at 68°C. Each tube contained <1 to 15 ng of DNA, PCR buffer (10 mM Tris-HCl, pH 9, 50 mM KCl, 0.1% Triton X-100, and 2.5 mM MgCl2), 500 nM of each primer, 200 mM of each deoxynucleoside triphosphate, and 0.5 U Taq DNA polymerase (Invitrogen). From this PCR product, 2 μl was used in eight replicate PCRs for the final 10-cycle PCR protocol as described above. The replicate reaction mixtures were pooled, and pseudoterminal fragments were eliminated by digesting single-stranded DNA generated in the PCR with mung bean nuclease (10). The PCR products were then purified and concentrated using a QIAquick PCR purification kit (Qiagen, Hilden, Germany). For each sample, aliquots containing approximately 20 ng of PCR product were separately digested with the endonuclease HhaI for at least 10 h, according to instructions supplied by the manufacturer (Invitrogen). Fluorescently labeled T-RFs were sized with an ABI 96 capillary sequencer running in GeneScan mode (Applied Biosystems). T-RFLP electropherograms were analyzed with GeneMarker, using 1% of the total peak area as a low cutoff for inclusion of T-RFs in the comparative analysis. Only peaks identified as DPOR on the basis of parallel sequencing and T-RFLP analyses of clones (see below) were included for the normalization.
For cloning, the same two-step PCR approach as described above for the T-RFLP analyses was used, except the forward primer lacked the fluorescent label. Replicate PCRs were pooled and precipitated with ethanol and sodium acetate (40), followed by gel purification using a QIAquick gel extraction kit (Qiagen). For each sample, 5 to 10 ng of the PCR product was cloned into the pCR 4-TOPO vector and transformed into competent Escherichia. coli one-shot TOP10 by using a TOPO TA cloning kit for sequencing as recommended by the manufacturer (Invitrogen). At least 48 positive clones from each library were transferred to 96-well plates and grown under constant shaking overnight at 37°C in LB containing 50 mg liter−1 kanamycin. DNA was extracted as described elsewhere (12). Lysates were used as the template in a PCR amplification using a Hex-labeled forward primer and the 30-cycle touchdown protocol as described above. Aliquots (containing 1 ng DNA) of fluorescently labeled PCR products were separately digested with the endonuclease HhaI as described above. After inactivation of the enzymes (20 min at 85°C), the terminal fragments generated were sized with an ABI 3700 96-capillary sequencer in GeneScan mode (Applied Biosystems). Unique T-RFLP patterns were used to classify clones into operational taxonomic units (OTUs).
For each identified OTU, at least one clone was sequenced, but several randomly chosen clones were sequenced for the more abundant OTUs. For sequencing, cloned inserts were amplified with the vector primers M13f-20 (5′-GTAAAACGACGGCCAG-3′) and M13r (5′-CAGG AAACAGCTATGAC-3′) (Invitrogen), using the following PCR protocol: 94°C for 3 min; 25 cycles of 1 min at 94°C, 1 min at 55°C, and 2 min at 72°C; and a final extension of 7 min at 72°C. Clones were sequenced with an ABI 96 capillary sequencer (Applied Biosystems), using plasmid primer M13 forward and a BigDye Terminator kit (version 3.1; Applied Biosystems). This generated high-quality reads of approximately 380 to 450 bases that were deposited in GenBank under accession numbers FJ460258 to FJ460348 and FJ601371.
Sequence data mining was initially performed using 31 reference sequences from cultured species. We then used BLAST to collect 38 GenBank and 124 global ocean survey gene fragments (39) with significant similarities to the reference sequences. With ARB (32), all sequences were translated into amino acids and aligned using MAFFT version 6 (25) on the Bioportal server at the University of Oslo. From this protein alignment, a phylogenetic tree was reconstructed using RaxML (43) with bootstrapping on the CIPRES server, using nifH sequences as outgroups. The trees were visualized and further edited with MEGA (version 3.1) (29) and iTOL (30).
Nonmetric multidimensional scaling (MDS) was performed with Primer 6 (Primer, Ltd., Plymouth, Great Britain). A dissimilarity matrix including L-DPOR fingerprints was obtained using the Bray-Curtis distance measure. A total of 500 iterations were run and resulted in a two-dimensional solution with a final stress value of 0.09.
One-way analysis of similarity (ANOSIM) (Primer 6; Primer, Ltd., Plymouth, Great Britain) using the Bray-Curtis distance measure were applied to test whether lakes with certain trophic statuses differ significantly from each other in their L-DPOR gene compositions. ANOSIM is based on the null hypothesis that there is no difference between such contrasting lake systems. The significance was examined by a permutation test where R was recomputed for 999 permutations. If the observed R value was found to exceed 95% of the simulated R values, then the null hypothesis can be rejected at the 5% significance level. Results from pairwise ANOSIM were analyzed to infer where the main between-group differences have arisen.
To identify environmental variables correlating with L-DPOR composition in the most consistent manner, a BIO-ENV procedure (Primer 6; Primer, Ltd., Plymouth, Great Britain) was applied. Environmental variables were normalized, and all possible combinations of these variables were transformed into Euclidean matrices. A proceeding RELATE test allowed testing for the null hypothesis that there is no relationship between environmental and DPOR matrices, i.e., that ρ is effectively 0. The significance was examined by a permutation test where ρ was recomputed for 999 permutations. If the observed ρ value was found to exceed 95% of the simulated ρ values, then the null hypothesis can be rejected at the 5% level. The combinations of variables are then ranked depending on ρ, i.e., the highest ρ value represents the environmental variables that best matched the L-DPOR fingerprints.
Sequences were deposited in GenBank under accession numbers FJ460258 to FJ460348 and FJ601371.
All 20 lakes included in the study were sampled during summer 2006, except for Lilla Ullfjärden, which was sampled in 2002. The sample from the latter lake was also used to establish sampling procedures and PCR-based methods. The lakes are located at different latitudes and vary broadly in productivity, water color, and other characteristics (Table (Table1).1). Total phosphorus content in the lakes ranged from 2 to 81 μg liter−1, and TOC or DOC ranged from 1 to 17.4 mg liter−1. The variable organic carbon content was also reflected in variable water color, with extinction coefficients for PAR [Kd(PAR) values] varying from 0.4 to 3.8 m−1 (Table (Table1).1). On the basis of total phosphorus and organic carbon contents, the lakes were classified into four trophic states (eutrophic, mesotrophic, oligotrophic, and dystrophic) (Table (Table1).1). Total phosphorus was also correlated with other parameters associated with lake productivity, such as total nitrogen and bacterial production (Table (Table1).1). Dystrophic lakes were characterized by high DOC concentrations and also feature high Kd(PAR) values.
At least 48 clones from each of the eight libraries were screened by T-RFLP and grouped into 77 different T-RFs. Of the 480 screened clones, a total of 192 clones were sequenced and compared to GenBank entries by using BLAST. This revealed 92 sequences closely related to L-DPOR, COR, or nifH that could be translated into amino acid sequences without any stop codons. The rest of the sequences either contained stop codons or could not be annotated. Phylogenetic analysis using RaxML of the partial protein sequences, including the ATP-binding domain of the iron protein and the cysteine residues that function as a ligand for the 4Fe-4S cluster (6), revealed three major clades, corresponding to COR, DPOR, and nifH. Of the environmental clones recovered in this study, 11 COR sequences, 61 L-DPOR sequences, 17 nifH sequences, and 3 sequences that could not be clearly resolved (Fig. (Fig.1A)1A) were identified. Of the 61 L-DPOR sequences, 39 were most closely related to oxygenic phototroph sequences (chlL), mainly cyanobacteria belonging to Synechocystis, Anabaena, and Microcystis but also chloroplasts of eukaryote algae such as those belonging to Scenedesmus and Chlamydomonas (Fig. (Fig.1B).1B). The remaining 22 freshwater DPOR protein sequences were more closely related to Roseobacter and other marine proteobacteria (bchL gene) but still clustered separate from the marine representatives and formed an exclusively freshwater bchL clade (Fig. 1A and B). These sequences were closely related to bchL genes from bacteriochlorophyll-containing isolates belonging to the genera Erythrobacter, Rhodobacter, and Rhodopseudomonas (Fig. (Fig.1B).1B). Also, the 11 COR sequences were related to marine proteobacteria, but similar to the bchL sequences, lake sequences formed two clusters that were clearly separated from marine COR sequences (Fig. (Fig.1A).1A). As shown in Fig. Fig.1A,1A, the primer pair used in this study does not target the full diversity of L-DPOR and COR genes identified in the PCR-independent global ocean survey data set (39). Almost all mismatches were against the reverse primer. Still, with allowances for one mismatch to the reverse primer, all except 15 of the L-DPOR entries would be amplified by the applied primer pair (Fig. (Fig.1A).1A). Since most COR genes have more than two mismatches to the reverse primer, PCR amplification has probably been highly biased against certain COR gene variants.
A total of 82 different T-RFs were detected, and 18 of these were putatively identified as L-DPOR genes, 1 was putatively identified as a COR gene (T-RF size, 88), and a single T-RF was putatively identified as nifH related (T-RF size, 267) (Fig. (Fig.2).2). The remaining 62 T-RFs could not be linked to any known protein with certainty, either because no clones with matching T-RF size were found in the clone libraries generated from the lakes or because matching clones could not be classified, for lack of matching database entries. The average relative contribution of these 62 T-RFs to the total T-RFLP peak area was 29% (range, 0 to 68%). These nonannotated T-RFs were excluded from further analyses. The remaining T-RFs were grouped into either nifH, aerobic anoxygenic COR (bchX), and L-DPOR (bchL) genes or L-DPOR genes related to oxygenic phototrophs (chlL) on the basis of phylogenetic analysis of clones matching T-RFs (Fig. (Fig.1).1). The numbers of L-DPOR T-RFs detected for individual lakes ranged from one to nine. T-RFs affiliated with chlL were found in all lakes, whereas bchL genes were missing in several systems, e.g., half of the eutrophic lakes and all three lakes from the northernmost region sampled (Fig. (Fig.2).2). The ratio of bchL genes to chlL genes was on average 0.15, with the highest values observed for humic lakes (range, 0.1 to 1.4) and for Lake Vättern at a 26-m depth (ratio, 2.4).
In some samples, L-DPOR T-RFs that could not with certainty be categorized as either chlL or bchL were detected, and in a few instances (e.g., for two of the northernmost clearwater systems), these contributed to ~70% of the total detected L-DPOR pool (Fig. (Fig.2).2). In bacteria of three out of four humic lakes (Lilla, Nedre, and Övre Björntjärn), bchL contributed around 10% to the total L-DPOR amplicons at the surface. In bacteria of eutrophic lakes, chlL genes related to cyanobacterial sequences were the dominant L-DPOR components, contributing up to 100% of the total peak area for individual samples. For the four lakes sampled at multiple depths, bchL genes increased with increasing depth in three cases, whereas the contribution of T-RFs putatively identified as chlL decreased in parallel (Fig. (Fig.22).
L-DPOR composition comparison was performed using Bray-Curtis dissimilarity measures and nonparametric MDS. MDS yielded a two-dimensional solution with a stress value of 0.09 after a total of 500 iterations (Fig. (Fig.3).3). A stress value of 0.09 suggests that this is a useful model even if the detailed features of the plot should not be overinterpreted. To further test if differences in L-DPOR fingerprints were related to the trophic state, ANOSIM was applied. ANOSIM using lake trophic state to group lakes (dystrophic, oligotrophic, mesotrophic, and eutrophic) revealed that the four groups occupy different regions in the DPOR community space. The global P value shows that we can reject the null hypothesis of no difference between L-DPOR fingerprints from groups of lakes with contrasting trophic statuses (Table (Table2).2). Pairwise comparison revealed that the differences are mainly due to L-DPOR composition in dystrophic lakes. But oligotrophic and eutrophic lakes also seem to be significantly different in their L-DPOR compositions (Table (Table2).2). Using BIO-ENV procedures, we identified total nitrogen, DOC, and NH4 concentrations as the variables giving the patterns most consistent with the L-DPOR patterns. A RELATE test revealed a significant matching between these three variables and L-DPOR patterns (ρ = 0.277; P = 0.007; n = 22). A significant matching was also observed between Kd(PAR) and L-DPOR patterns by using the RELATE test (ρ = 0.205; P = 0.048).
Earlier studies have shown that bacteriochlorophyll-containing cells can make up a significant portion of the total bacterial community in lakes (33). Results from the present lake survey confirmed that L-DPOR and COR genes closely related to AAnPB were present in a wide range of freshwater environments even though their relative contributions to the total amplified gene pool varied significantly. L-DPOR genotyping by T-RFLP suggested that AAnPB were more-consistent and -significant members of the phototrophic microbial community in humic and oligotrophic lakes than in eutrophic lakes. As expected, cyanobacteria-related chlL sequences dominated the L-DPOR gene pool in the eutrophic systems (Table (Table11 and Fig. Fig.2).2). Mašin and coworkers (2008) observed a similar pattern in a microscopy- and pigment-based lake survey where the contribution of AAnPB to the total microbial community was insignificant for the most-eutrophic systems (33).
At first inspection, results from L-DPOR T-RFLP did not suggest any evident and systematic differences in the compositions of chlL- and bchL-related sequences between bacteria of lakes with different trophic statuses. However, MDS analysis revealed a certain separation of L-DPOR composition in bacteria of lakes with different trophic statuses (Fig. (Fig.3).3). ANOSIM corroborated the significant differences in L-DPOR community composition between bacteria of lakes with different trophic statuses (Table (Table2).2). In particular, the L-DPOR gene composition in bacteria of the dystrophic lakes with high inputs of colored terrestrial organic matter was distinct from those in bacteria of all the other systems. We suggest that humic substances may affect L-DPOR composition via the qualitative and quantitative differences in light climate in dystrophic/humic systems in comparison to clearwater lakes. The former systems feature a shallow photic zone and a higher proportion of red light available for the photosynthetic community than clearwater systems with low contents of chromophoric dissolved organic matter (46). Further support for this is found in the significant correlation between in situ PAR extinction and L-DPOR composition as indicated by the RELATE analysis. Because AAnPB are able to efficiently capture radiation in the far red portion of the solar spectrum (46), it can be assumed that these organisms may have a competitive advantage in humic systems, at least with regard to the availability of energy in the form of solar radiation. Such light spectrum-dependent niche partitioning between photosynthetic populations has previously been reported to occur in marine plankton communities (2, 22, 45).
The four- to sevenfold-higher ratio of bchL to chlL in bacteria of hypolimnetic water than in those of epilimnetic water in humic lakes Nedre and Övre Björntjärn and lake Vättern is noteworthy but may simply reflect the inability of oxygenic phototrophs to maintain a viable population at the low light levels characteristic of these deeper strata. In contrast, AAnPB might be capable of strictly organotroph energy acquisition when solar energy is scarce (13). This will certainly be a competitive advantage in humic systems where solar radiation decreases rapidly with depth and is shifted toward the red end of the wavelength spectrum.
Further clues to the main environmental drivers causing changes in L-DPOR composition are provided in the BIO-ENV analysis, where total nitrogen, DOC, and NH4+ appeared to be significant state variables coupled to L-DPOR composition. The reason for the significance of DOC in this regard may be that in temperate lakes, the bulk of the dissolved organics consists of highly colored humic substances having a strong impact on light climate (4). Additionally, high total nitrogen and NH4+ concentrations are characteristic for the most-eutrophic lakes where cyanobacterial chlL genes dominated the DPOR gene pool, and this may also explain the separation of the L-DPOR compositions in the eutrophic systems in comparison to what was found for the least productive (oligotrophic) lakes (Fig. (Fig.33 and Table Table22).
The clear separation of L-DPOR and COR gene sequences derived from bacteria of marine and freshwater systems (Fig. (Fig.1)1) can be interpreted either as adaptations of the proteins to the different salinities or as adaptations of the organisms carrying the gene to these conditions, combined with limited genetic exchange via horizontal gene transfer. This is in agreement with results from a recent study of rhodopsin diversity, revealing a clear separation between marine and freshwater sequences (42). In fact, multiple studies of bacteria and algae have revealed similar patterns characteristic for separations between freshwater and marine taxa (21, 31), suggesting that salinity is likely to be a strong environmental filter selecting for specific photosynthetic populations.
The T-RFLP survey corroborated the observation of marked differences in L-DPOR composition among the studied lake clone libraries. Our results indicate that cyanobacteria dominated oxygenic phototrophs in the studied lakes, at least at the time of sampling, since L-DPOR clone libraries in most lakes were dominated by cyanobacterial sequences. Only a few sequences related to chloroplasts were identified, and these were mainly recovered from the humic lakes.
For many cyanobacteria, microscopy-based differentiation based on morphological traits is a feasible strategy for assessing the distribution of these cyanobacteria in aquatic systems and in relation to environmental gradients (9). Such approaches have been applied extensively to bloom-forming cyanobacteria, highlighting the roles of hydrodynamic stability combined with sufficient availability of light and nutrients (particularly P) as triggers of bloom events (36). Such blooms are common in the eutrophic systems included in the current study (12). In less productive systems, non-bloom-forming unicellular cyanobacteria tend to play a more significant role (44). Microscopic analyses have also been applied to identify and quantify many such unicellular or colonial cyanobacteria with contrasting habitat preferences (reviewed in reference 44). Usually, nonblooming cyanobacteria make up the largest portion of the phytoplankton (eukaryote and cyanobacterial oxygenic phototrophs) in less productive lakes (44). In these microscopy-based comparisons, the contribution of AAnPB to the phototrophic community in lakes has so far been neglected. Our study points to AAnPB as abundant members of the phototrophic community in humic and less productive lakes, whereas a lower frequency of bchL genes than of chlL genes was observed in the most-productive (eutrophic) systems (Table (Table11 and Fig. Fig.2).2). This agrees with a recent survey of chlorophyll and bacteriochlorophyll pigments (33) and emphasizes the need to quantify AAnPB if we want to determine the true biomass of phototrophic organisms in lakes.
It should be acknowledged that the primers employed in the current gene survey might be biased against specific bchL genes revealed in PCR-independent large-scale metagenomic sequencing of the global ocean survey (Fig. (Fig.1A).1A). Also, their unspecific amplification of COR genes, nifH, and nonannotated gene fragments may call for a redesigning of the primer pair in future studies directed at AAnPB. It should also be stressed that the presence of genes required for bacteriochlorophyll synthesis by no means guarantees the presence or activity of the gene product. So far, there are few surveys of bacteriochlorophyll concentrations in lakes (see, for example, reference 33), and there is certainly a need for future studies combining analysis of pigment concentrations with molecular identification of phototrophic organisms like AAnPB. Because of their widespread occurrence, we need to consider AAnPB potentially important players in biogeochemical cycles. In particular, the possible dual role of AAnPB as primary and secondary producers calls for further investigation. Isolation of abundant members of freshwater bacterioplankton communities in combination with culture-independent studies should allow us to define their trophic status also in relation to environmental conditions and a future changing environment.
This study was funded by Formas, Uppsala Micorbiomics Centre, the Swedish Research Council (grants to S.B.), and the Olsson-Borg Foundation (grant to A.E.).
We thank Ramiro Logares for assistance with phylogenetic analyses, Jenny Ask and Jan Johansson for field and laboratory assistance, and Eva Lindström and Jürg Brendan Logue for kindly contributing DNA samples and environmental state variables for some of the lakes included in the survey.
Published ahead of print on 2 October 2009.