Introduction

Proteorhodopsins (PRs) (Béjà et al., 2000, 2001) are microbial retinal-binding membrane pigments belonging to the microbial rhodopsin superfamily (Spudich et al., 2000) and were suggested to have an important role in supplying light energy for microbial metabolism in different marine ecosystems (Béjà et al., 2000, 2001; Sabehi et al., 2005b; Gómez-Consarnau et al., 2007; Martinez et al., 2007; Walter et al., 2007). PRs have been observed in a wide variety of ocean regions (Béjà et al., 2000, 2001; de la Torre et al., 2003; Sabehi et al., 2003, 2004, 2005a, 2007; Venter et al., 2004; Rusch et al., 2007; Campbell et al., 2008) and were found in diverse bacterial groups, including the ubiquitous marine gammaproteobacterial SAR86 (Béjà et al., 2000; Sabehi et al., 2004, 2005a) and alphaproteobacterial SAR11 groups (Giovannoni et al., 2005a, 2005b; Sabehi et al., 2005a), as well as in Roseobacter (alphaproteobacterium HTCC2255), marine Bacteroidetes (Venter et al., 2004; Gómez-Consarnau et al., 2007), planktonic Archaea (Frigaard et al., 2006) and other microbial taxa (de la Torre et al., 2003; Sabehi et al., 2003, 2005a; Venter et al., 2004; McCarren and DeLong, 2007; Rusch et al., 2007). While previous work did not detect light enhanced growth neither in PR-containing SAR11 (Giovannoni et al., 2005a) nor in the gammaproteobacterial SAR92 (Stingl et al., 2007) isolates grown in seawater, significant enhancement of both growth rate and yield was recently reported in PR-expressing marine Flavobacteria (Gómez-Consarnau et al., 2007).

PRs were so far detected in different marine environments, mainly in the open ocean. In the recent Global Ocean Sampling (GOS) expedition (Rusch et al., 2007), 44 samples were obtained, covering a wide range of distinct surface marine environments as well as a few estuarine and inland waters. These included three estuaries (Bay of Fundy, Canada; Delaware Bay and Chesapeake Bay, USA), one freshwater lake (Lake Gatun, Panama) and a moderate hypersaline lagoon (Punta Cormorant Lagoon, Galápagos, Ecuador). Different PRs were observed in these estuaries and non-marine stations (Rusch et al., 2007).

Freshwater ecosystems offer a wide range of variation in many ecologically relevant parameters, including the underwater light climate. Known PR genes encode pigments with two distinct light absorption spectra, commonly referred to as ‘blue-absorbing’ PR and ‘green-absorbing’ PR (Béjà et al., 2001; Man et al., 2003). Blue-absorbing PRs are particularly widespread in the blue ocean waters that have been extensively sampled in earlier PR studies (Béjà et al., 2001; Sabehi et al., 2007). The underwater light color of most freshwater ecosystems, however, is shifted towards the green and red part of the light spectrum (Kirk, 1994; Stomp et al., 2007a, 2007b). Thus, possibly, new PR genes encoding green-absorbing and perhaps even red-absorbing pigments might be awaiting discovery in freshwater ecosystems. Motivated by these ideas, we have designed general degenerate PR primers based on data from the non-marine GOS stations and searched for PR presence in different freshwater systems worldwide.

Materials and methods

Sample collection, DNA preparation and storage

Environmental samples were collected from Lake Kinneret in Israel, Lakes Großer Plöner See, Schöhsee, Plußsee in Germany, Lake L658 in Canada, Lake Zegerplas and Lake Westeinderplassen in the Netherlands and from the Baltic Sea (see Table 1 for details of sampling locations). Genomic DNA extractions were performed using the UltraClean Water DNA kit (MoBio Laboratories, Carlsbad, CA, USA). DNA was stored at −80 °C for further analysis.

Table 1 Characteristics of sampling sites
Full size table

PR PCR amplification

Proteorhodopsins were amplified from DNA extracts obtained from environmental samples using modified PR forward and reverse primers (see Table 2). The degenerate primers were designed based on known PR sequences found in NCBI and on estuary and freshwater stations in the GOS dataset (GOS stations GS06, 11, 12 and 20). A total of 48 combinations of primer sets were designed and tested against the Lake Kinneret samples. Of all the 48 combinations that were tested, only 6 combinations produced a PCR product and these 6 combinations were selected for further analysis on environmental samples. Some deep-branching actinorhodopsin-like (Sharma et al., 2008) sequence variants contain the sequence Twxxyp (for example, actinorhodopsin reads 1095521408126, 1095390008418 and 1095898194153 from groups LG1 and LG2) and would, therefore, not be amplified by our reverse primer sets, which are based on a Gwxxyp protein sequence.

Table 2 The proteorhodopsin forward and reverse primers used in this study, designed based on known PR sequences found in the NCBI and Global Ocean Sampling datasets
Full size table

Polymerase chain reactions were performed using high-fidelity proof-reading polymerase mix (TaKaRa Ex Taq from Takara Shuzo). PCR amplification was carried out in a total volume of 25 μl containing 10 ng of template DNA, 200 μM dNTPs, 1.5 mM MgCl2, 0.2 μM primers and 2.5 U TaKaRa Ex Taq polymerase. The amplification conditions comprised steps at 92 °C for 4 min, 35 cycles at 92 °C for 1 min, 49.8 °C for 1 min and 72 °C TaKaRa Ex Taq polymerase for 1 min. PCR products were cloned using the QIAGEN-PCR cloning kit (Qiagen, Hilden, Germany). Unique EcoRI and RsaI restriction fragment length polymorphism groups were sequenced. All PCR products identical to samples handled or amplified previously in the lab were omitted from the analyses to avoid being influenced by possible contamination.

PR phylogeny

The PR tree was constructed according to Sabehi et al. (2005a). PR proteins were aligned using CLUSTALx (Thompson et al., 1997) and 110 amino-acids alignment was used to construct a neighbor-joining phylogenetic tree as inferred using the neighbor programs of PAUP* (Swofford, 2002). Bootstrap resampling (1000) of the distance and parsimony trees were performed in all analyses to provide confidence estimates for the inferred topologies.

Nucleotide sequence accession numbers

PR gene sequences were deposited in GenBank under accession numbers EU563268–EU563333.

Results and discussion

Design of new PR primers

Proteorhodopsin sequences retrieved from different estuaries and freshwater GOS stations (Stations GS06, 11, 12 and 20) were aligned and new primers were designed to conserved regions in the PR protein (amino-acid positions 94–98 and 196–201 based on PR BAC31A8 numbering (Béjà et al., 2001)) to cover possible new PR diversity in brackish and freshwater ecosystems. The same PR regions were previously used to recover PRs from different marine environments and from BAC libraries (Sabehi et al., 2005b, 2007; Campbell et al., 2008). Four forward primers and twelve reverse primers (Table 2) were used to perform multiplex PCR reactions on Lake Kinneret samples. Six out of the possible forty-eight combinations gave positive PCR products (combinations RYIDW+GWAIYP, RYIDW+GWSIYP, RYIDW+GWAVYP, RYVDW+GWSIYP, RYVDW+GWVIYP and RYVDW+GWAVYP; see Table 3) and were later used with other samples.

Table 3 Pairs of primers that gave positive results in the different sites
Full size table

A recent in silico search (Sharma et al., 2008) for rhodopsins performed on several aquatic non-marine GOS stations has identified unique groups of deep-branching PR-like sequences linked to Actinobacteria (coined ‘actinorhodopsins’ by the authors). It is important to note that the primers used in this study would not amplify some actinorhodopsin-like sequence variants (see Materials and methods section).

Screening of different freshwater lakes and the Baltic Sea

To explore the possible existence of PRs in Lake Kinneret, Israel, we screened different season and depth samples with the 48 different primer combinations. Out of these 48 primer combinations, only 6 gave positive PCR signals. The PRs amplified from Lake Kinneret differed from known marine PRs and clustered separately or with other non-marine GOS PRs (Figure 1). Using these six primer sets we have amplified PR genes from DNA extracts from clear lakes in Germany (Lakes Großer Plöner See, Schöhsee, Plußsee) and Canada (Lake L658), peat lakes from the Netherlands (Zegerplas and Westeinderplassen), as well as from the Baltic Sea.

Figure 1
figure 1

Neighbor-joining phylogenetic tree of aquatic non-marine proteorhodopsins (PR) proteins (102 amino acids position alignment length; without the corresponding primer regions). Estuarine global ocean-sampling (GOS) sequences are represented by green bands while freshwater GOS sequences are marked with blue. PR sequences retrieved via PCR in this study are marked by boldface letters and their station position color is indicated in the box on the top left. Bootstrap values (distance/parsimony) greater than 50% of major clades are indicated above the branches. For color figure, see online version.

Full size image

In total, we found a tremendous diversity of freshwater PRs that show as low as 42% identity (based on amino-acid level) between different distanced clades. Several PR groups were unique to Lake Kinneret and were not amplified in other freshwater environments (freshwater clades 2, 3 and 4 in Figure 1), whereas other groups contained PR representatives from different environments (non-marine clades 1 and 6 as well as freshwater clades 7, 8 and 9 in Figure 1). Several PR clades clustered in proximity to PRs from cultured marine Flavobacteria and possibly represent PRs from freshwater Flavobacteria. Another two PR clades (6 and 7) clustered with representatives from the marine SAR11 clade. Freshwater PR clades 8 and 9 did not cluster with any known PR and their affiliations remain unknown.

All PCR products identical to samples handled or amplified previously in the lab (based on the DNA level) were omitted from the analyses to avoid influence from possible contamination. This step is extremely important when different lake samples are compared at once. In Figures 2a and b, alignments of different PRs found in different lake environments and in the Baltic Sea are shown. As can be seen in Figure 2a, non-marine clade 1 (top Figure 1 above freshwater clade 2), which contained PRs from different samples is composed of very different PR proteins, which could not have been amplified from possible lab contaminations. Freshwater clade 8 includes PRs from a wide range of different and geographically distant lakes (Lake Kinneret in Israel, several clear-water lakes in Germany and Canada as well as peat lakes in the Netherlands). In contrast to non-marine clade 1, the PRs in freshwater clade 8 were closely related (Figure 2b). However, there are amino-acid differences between the PRs in this clade. This echoes similar findings of Zwart et al. (1998), describing 16S rRNA clades consisting of highly related proteobacteria isolated from different continents. The similarity of PRs isolated from lakes in different parts of the world suggests that these genes are dispersed globally and that they may encode unique functional capabilities enabling successful competition in a wide range of freshwater environments.

Figure 2
figure 2

Multiple alignment of proteorhodopsins (PR) amino-acid sequences. (a) Non-marine clade 1; (b) freshwater clade 8. The phylogenetic tree shown on the left-hand side corresponds to the tree in Figure 1. Multiple alignments of PR amino-acid sequences are shown. Position 105 is marked with a red box. For color figure, see online version.

Full size image

Organismal origins of PRs from freshwater lakes and the Baltic Sea

Similar to taxon prediction performed in previous studies (de la Torre et al., 2003; McCarren and DeLong, 2007; Sharma et al., 2008), we tried to identify the organisms containing unique non-marine PRs by carrying out a phylogenomic analysis of adjacent sequences found on similar freshwater and estuary GOS scaffolds. Using this approach, we assigned the affiliation of clades 1, 2 and 3 with Flavobacteria (based on GOS sequences GS06 1096686103557 and GS11 1096687482507, which correspond to GOS scaffold-1096626019272 and scaffold-1096627448042, respectively). We speculate that the same is true for clades 4 and 5 but with lower confidence.

Clades 6 and 7 form distinct clades closely related to Pelagibacter ubique and the SAR11-type rhodopsins. Different GOS scaffolds (GS12 singleton-1096686052401, scaffold-1096627380104) that clustered within clade 6 and 7 were highly similar to P. ubique based on gene content but the genomic organization was different with an operon for carotene biosynthesis adjacent to the PR gene. Similar operon organization is observed in many uncultured PR-carrying organisms (Sabehi et al., 2005b, 2007) but is different from the organization in P. ubique, where an acetyl-CoA synthetase and a ferredoxin are adjacent to the PR gene and the retinal biosynthesis operon is found elsewhere in the genome (Giovannoni et al., 2005a, 2005b). We, therefore, assigned clades 6 and 7 as SAR11-related clades (Figure 1). Non-marine SAR11s (alphaproteobacterial cluster LD12) were first reported in an arctic lake (Bahr et al., 1996) and are readily detected in other freshwater systems (Zwart et al., 1998, 2003; Crump et al., 1999). PRs from clades 6 and 7 could be from bacteria related to alphaproteobacterial cluster LD12. However, because of previously reported PR lateral gene transfer (Frigaard et al., 2006), our predicted affiliations should be dealt with caution and are provisional.

Spectral tuning in estuaries and freshwater PRs

All predicted PR proteins retrieved in this study contained leucine (L) or methionine (M) in position 105 (cf. Figure 2). This makes them suspected green-absorbing PRs (Man et al., 2003). The predominance of green-absorbing PRs can be explained by the underwater light spectrum in coastal and freshwater ecosystems. In these waters, green-light conditions prevail (Stomp et al., 2007b; Haverkamp et al., 2008) and hence green-absorbing PRs will have a selective advantage compared to blue-absorbing PRs. Yet, the PRs described here show more variability in their protein sequence than in previous studies (Man et al., 2003) and are largely distinct from the marine clades (Figure 1). Therefore, prediction of the color of light absorbed by these freshwater PRs based on a single position (such as position 105) in the overall sequence should be made with great caution.

One remaining open question is the role of the PR-like molecules. Several of the PRs reported here are phylogenetically quite distant from ‘known’ PRs (as low as 42% amino-acid identity). Some of these PRs might indeed function as proton pumps involved in phototrophy but others might be sensory rhodopsins in the classical sense (reacting with transducers—but may appear to be a ‘proton pump’ by sequence) or might have other unknown functions. Our work, along with the recent GOS observations (Rusch et al., 2007; Sharma et al., 2008), suggests that PR and PR-like pigments are common in different aquatic environments worldwide, including marine, brackish and freshwater ecosystems. Thus, PR-based phototrophy might be a common theme in different aquatic environments worldwide.