ABSTRACT
Although sunlight is an abundant source of energy in surface environments, less than 0.5% of the available photons are captured by (bacterio)chlorophyll-dependent photosynthesis in plants and bacteria. Metagenomic data indicate that 30 to 60% of the bacterial genomes in some environments encode rhodopsins, retinal-based photosystems found in heterotrophs, suggesting that sunlight may provide energy for more life than previously suspected. However, quantitative data on the number of cells that produce rhodopsins in environmental systems are limited. Here, we use total internal reflection fluorescence microscopy to show that the number of free-living microbes that produce rhodopsins increases along the salinity gradient in the Chesapeake Bay. We correlate this functional data with environmental data to show that rhodopsin abundance is positively correlated with salinity and with indicators of active heterotrophy during the day. Metagenomic and metatranscriptomic data suggest that the microbial rhodopsins in the low-salinity samples are primarily found in Actinobacteria and Bacteroidetes, while those in the high-salinity samples are associated with SAR-11 type Alphaproteobacteria.
IMPORTANCE Microbial rhodopsins are common light-activated ion pumps in heterotrophs, and previous work has proposed that heterotrophic microbes use them to conserve energy when organic carbon is limiting. If this hypothesis is correct, rhodopsin-producing cells should be most abundant where nutrients are most limited. Our results indicate that in the Chesapeake Bay, rhodopsin gene abundance is correlated with salinity, and functional rhodopsin production is correlated with nitrate, bacterial production, and chlorophyll a. We propose that in this environment, where carbon and nitrogen are likely not limiting, heterotrophs do not need to use rhodopsins to supplement ATP synthesis. Rather, the light-generated proton motive force in nutrient-rich environments could be used to power energy-dependent membrane-associated processes, such as active transport of organic carbon and cofactors, enabling these organisms to more efficiently utilize exudates from primary producers.
INTRODUCTION
Light is an abundant resource, with 5.34 × 1034 photons reaching Earth every second (calculated from http://rredc.nrel.gov/solar/spectra/am1.5/). Organisms in all domains of life use light for photosynthesis, vision, or as an environmental cue (1, 2). Photosystems capable of converting photons to stored chemical energy provide photoheterotrophic microorganisms with a way to supplement their energy metabolism, taking advantage of the abundance of light in an often nutrient-limited world (3–7). Photoheterotrophs using photosystems with bacteriochlorophyll a as the primary light-capturing pigment typically comprise up to 10% of the microbial community in aquatic and marine environments (7–12). In contrast, the much simpler rhodopsin-type light-harvesting systems are found in 30 to 60% of the microbial genomes in surface environments (13–17), even though theoretical calculations suggest that they may return significantly less energy to the cell than the bacteriochlorophyll a-utilizing photosystems (18).
Microbial rhodopsins consist of one polypeptide and a single organic cofactor, retinal, whose biosynthesis is encoded by a total of 7 genes, and most are light-activated proton pumps. Because they are so simple, it is often suggested that rhodopsins provide a metabolically inexpensive mechanism for supplementing the proton motive force when organic carbon is limiting (3, 19). If microbial rhodopsins are critical to the starvation response, they should only be expressed when cells are carbon or energy limited. Upregulation of proton-pumping rhodopsin expression under carbon- or energy-limited conditions has been observed in some bacterial isolates (3–6, 20–23). However, other work has identified bacterial species with rhodopsins that pump ions other than protons (24, 25) and proton-pumping rhodopsins that energize active transport rather than ATP synthesis (26), as well as species whose rhodopsin expression is associated with anaplerotic CO2 assimilation in the light (27–29). Rhodopsin expression may also be correlated with salinity or osmotic stress (30, 31). In sum, the existing data clearly suggest that rhodopsins may play many physiological roles and may be important under a variety of environmental conditions. Despite their array of apparent physiological roles, the distribution of functional rhodopsins in relation to environmental parameters other than nutrient availability has been underexplored.
To begin to identify the environmental conditions under which rhodopsins are most active and thus presumably most important, a method for the quantification of cells containing active rhodopsins in natural environments is necessary. The low fluorescence yield of rhodopsins has hampered the use of direct detection and counting methods. Previous estimates of abundance relied on metagenomic sequence data, amplicon sequencing, quantitative PCR (qPCR), or cultivation, which are imperfect indicators of functional rhodopsin abundance (32).
Total internal reflection fluorescence (TIRF) microscopy can be used to identify and quantify cells with active rhodopsins (33). Here, we combine TIRF microscopy, a quantitative method that we use to measure the abundance of rhodopsin-producing cells, with qualitative analyses (qPCR and metatranscriptomics) that identify the types of rhodopsins present, to assess the kinds and abundances of rhodopsins in bacterioplankton in the Chesapeake Bay. The Chesapeake Bay is a mesotrophic system with a salinity gradient along its length (34). Because prior research indicates that rhodopsin transcription in natural environments is upregulated in the light (35–38) and that rhodopsins are more abundant in marine environments than in freshwater (13–16), we predicted that functional rhodopsin abundance would be greater during the day than at night and would increase as light intensity increased, and that rhodopsin gene abundance would increase with increasing salinity. Here, we used qPCR along the length of the Bay and metagenomic data from 3 sites to confirm the hypothesis that rhodopsin genes and their transcription increase in abundance as salinity increases along the length of the estuary. We further demonstrate that functional rhodopsins are more abundant during the day than at night and that their abundance pattern is similar to the patterns of chlorophyll a (Chl a) abundance and bacterial production.
RESULTS
Environmental parameters.Samples were collected in April 2015 from the R/V Sharp along a transect from the headwaters of the Chesapeake Bay, near the Susquehanna River, to the mouth of the Chesapeake Bay (Fig. 1). An additional sample was also collected off the coast of Assateague Island (Fig. 1, site 36). This transect followed a gradient of increasing salinity, from nearly fresh (0.07 ppt salinity) to marine (35 ppt salinity) (Fig. 2A). At each site, samples were collected and analyzed for nitrate, ammonium, phosphate, and silicate contents (Fig. 2; see also Table S1 in the supplemental material), as well as total cell counts (by enumeration of 4′,6-diamidino-2-phenylindole [DAPI]-stained cells), bacterial production (by quantification of 3H-leucine incorporation), and Chl a (Fig. 2). Nitrate and silicate concentrations decreased along the length of the Bay as salinity increased, as did bacterial production (Fig. 2). Phosphate levels were below the detection limit in nearly all samples, suggesting that this was the limiting nutrient at the time of collection.
Map of cruise track. Samples were collected from 11 to 16 April 2015. Sampling sites are numbered chronologically. Samples for rhodopsin analyses were collected daily at 11:00 a.m. (white circles) and 11:00 p.m. (black circles). The Susquehanna River drains into the Chesapeake Bay just north of site 2. Site 36, since it is a coastal ocean site rather than estuarine, was excluded from most analyses. The map was created with the R package maps.
Environmental data. All data are plotted as functions of latitude. Each data point is the average of 3 measurements, and error bars indicate 1 standard deviation. (A) Salinity decreases linearly with distance from the ocean (ocean is at the lower latitudes). (B) Nitrate content is higher in samples further from the ocean. (C) Ammonium (gray squares) does not clearly vary with latitude. Phosphate content (black diamonds) was below the detection limit in most samples. (D) Cell counts are similar on average along the length of the Chesapeake. (E) Bacterial production (gray squares) and Chl a (black circles) are highest in the freshwater closer to the Susquehanna River.
The R packages rcorr and corrplot (39, 40) were used to identify and plot correlations between abiotic environmental parameters (salinity, nitrate, ammonium, silicate, and light intensity) and between abiotic and biological parameters (cell counts, bacterial production, and Chl a). Salinity was negatively correlated with nitrate and silicate in both day and night samples and with ammonium at night (Fig. 3). It was also negatively correlated with cell counts and bacterial production during the day, but no correlation between salinity and biological parameters was observed at night (Table S2). Light intensity (photosynthetically active radiation [PAR] between 400 and 700 nm) was negatively correlated with cell counts during the day, but no statistically significant correlations between light intensity and other parameters were identified. Bacterial production was positively correlated with Chl a, nitrate, and silicate during the day but not with any environmental or biological parameters at night (Fig. 3). Both Chl a and silicate are associated with primary producers, since algae, diatoms, and cyanobacteria use Chl a to capture light energy and diatoms synthesize Si-rich frustules. In total, these correlations suggest that heterotrophic activity (as indicated by bacterial production) is highest in the places with the most primary producers (as indicated by Chl a and silicate).
Correlations (Pearson's r) between environmental and biological parameters. Any correlation with a P value of <0.05 is plotted. Daytime samples are in the lower left half of the grid, and night samples are in the upper right half. Red hues indicate negative correlations, and blue hues indicate positive correlations. Salinity was measured in units of parts per thousand and is strongly correlated with most abiotic and biological parameters. Photosynthetically active radiation (PAR) was not measured for night samples.
Rhodopsin gene abundance in the Chesapeake Bay.To determine the genetic potential of the microbial communities in the Chesapeake Bay to produce rhodopsins, the abundance of rhodopsin-encoding genes was quantified using qPCR. Primers capable of amplifying SAR11-type proteorhodopsins (SAR-PR) and LG1-type actinorhodopsins (LG1; Table 1) were used in qPCR to estimate gene abundances along the Bay. Using the assumption that, on average, microbial genomes have 1.9 copies of the 16S rRNA gene (35), we estimate that the percentage of genomes in the Chesapeake Bay encoding SAR-PR increases from 0.7% at 0.1 ppt salinity to 116% at 35 ppt (Fig. S1). This change indicates that salinity strongly affects microbial community structure.
Primers used for qPCR analysisa
Trends in the qPCR data were also analyzed with rcorr. The abundance of SAR-PR genes is strongly correlated with salinity during the day (Pearson's r = ∼0.70; Fig. 3 and Table S2). Although SAR-PR gene abundance is clearly correlated with salinity, it is also strongly negatively correlated with total cell counts, bacterial production, nitrate content, and silicate content during the day and negatively correlated with nitrate, ammonium, and silicate contents in the night samples (Fig. 3 and Table S2).
In contrast, actinorhodopsin genes of the LG1 group (15, 41) are present at low levels along the entire length of the bay, decrease as salinity increases, and are consistently more abundant at night than during the day (Fig. S1).
Functional rhodopsin abundance in the Chesapeake Bay.The abundances of functional rhodopsins were quantified using direct cell counts with TIRF microscopy. In the TIRF microscopy system, fluorophores are excited by the evanescent wave generated when the laser light reflects off the sample-coverglass interface; since this wave propagates <200 nm into the sample, background fluorescence in this system is minimal, and molecules such as rhodopsins with low fluorescent yields can be detected (33, 42). To differentiate microbial cells from autofluorescent organic matter, samples were stained with DAPI prior to analysis. All fields of view were sequentially excited with a 405-nm laser to illuminate DAPI-stained cells, a 561-nm laser to identify rhodopsin- and phycobiliprotein-containing cells, and a 641-nm laser to identify Chl a-containing cells (Fig. S2). Cells were identified as containing rhodopsins if they fluoresced when excited with the 405- and 561-nm lasers but not the 641-nm laser. This method measures the autofluorescence of the rhodopsin-retinal complex; neither retinal nor the apo-rhodopsin is autofluorescent alone under these conditions (33). Cells that fluoresced when excited with both the 561- and 641-nm lasers were interpreted as being cyanobacteria with phycobiliproteins, not rhodopsins, since this method cannot distinguish between rhodopsins and phycobiliproteins (33).
The number of cells with rhodopsin fluorescence is strongly correlated with salinity during both day and night (Fig. 3 and 4; Pearson's r coefficients of 0.76 and 0.94, respectively). At night, the percentage of cells producing rhodopsins ranges from ∼4.6% in the freshwater sample to ∼30% in the most marine sample, increasing linearly with increasing salinity. During the day, the lowest number of cells with rhodopsins was observed in the mid-salinity (7 to 15 ppt) range, but this number generally increased with increasing salinity as well.
Abundance of functional rhodopsins (TIRF microscopy). Cells producing functional rhodopsins were quantified by TIRF microscopy. Cells with active rhodopsins are consistently more abundant during the day (gray symbols, solid line) than at night (black symbols, dashed line) and increase as salinity increases. In the daytime samples, the abundance of functional rhodopsins is correlated with both salinity and with SAR-PR gene abundance (r = 0.76 and 0.7). At night, the correlation with salinity is stronger (R = 0.94), but the correlation with rhodopsin gene abundance is not significant.
The ratio of cells with functional rhodopsins to those with rhodopsin genes should indicate how many organisms with the potential to synthesize rhodopsin actually do so. A correlation of changes in this ratio with environmental parameters may thus identify environmental controls on light utilization via rhodopsin-type photosystems. At night, this ratio is not significantly correlated with any measured environmental parameter (Fig. 3). Although no consistent trend with salinity is observed, the ratio of cells with functional rhodopsins to rhodopsin-containing genomes is highest when the salinity levels are in the range of 5 to 10 ppt and 20 to 25 ppt (Fig. 5A). In the surface daytime samples, this ratio is positively correlated with bacterial production and Chl a, nitrate, and silicate concentrations, and it is negatively correlated with the percentage of genomes encoding SAR11-type proteorhodopsins (Fig. 3 and Table S2). Further, this ratio, the Chl a concentration, and bacterial production all decrease as light intensity increases (Table S2), while the TIRF-to-qPCR ratio and bacterial production both increase with increasing Chl a concentration (Fig. 5B). This combination of indicators of primary producers and heterotrophic activity strongly implies local bioavailable organic carbon and suggests in turn that rhodopsin production is associated with heterotrophic activity.
Correlation of rhodopsin production to environmental parameters. The ratio of cells producing functional rhodopsins (as quantified by TIRF microscopy) to genomes encoding rhodopsins (as quantified by qPCR) should indicate the percentage of cells capable of producing rhodopsin that actually do so. (A) This ratio varies as salinity increases, for both daytime (gray squares) and nighttime samples (black squares), but not with any consistent pattern. (B) Both bacterial production (gray squares) and the ratio of functional rhodopsins to genomes encoding rhodopsins (black triangles) are correlated with Chl a concentration in daytime samples. No correlation of this ratio with any parameter was observed in night samples (data not shown).
Metagenomic and metatranscriptomic sequence data.DNA and RNA from two cell size fractions (0.2 to 0.8 μm and >0.8 μm) were sequenced from 3 sites along the Chesapeake Bay representing low-, mid-, and high-salinity environments. The estimated number of copies of rhodopsin genes per genome varied along the Chesapeake Bay salinity gradient and with size fraction (Fig. 6 and Table 2). The number of genomes was estimated from the number of reads mapping to rplB, which encodes the 50S ribosomal protein L2 and is a single-copy gene in microbial genomes (43). In the mid-salinity smaller-size fraction (15 ppt, <0.8 μm), the ratio of rhodopsin to rplB was ∼1, suggesting that most genomes in this sample encode at least one rhodopsin (Table 3). The freshwater sample, where only the larger-size fraction was analyzed, had the fewest rhodopsin genes, as ∼30% of the genomes are estimated to encode a rhodopsin. In both cases where paired size fraction data were available (sites 17 and 33, at 15 ppt and 31 ppt salinity, respectively), the <0.8-μm-size-fraction sample had more rhodopsin genes per genome than the >0.8-μm-size-fraction sample (Fig. 6A and Table 3).
Rhodopsin gene and transcript abundances. (A) Ratio of rhodopsin genes to rplB genes in metagenomic data sets. This ratio is higher in the smaller-size fraction and suggests that nearly all of the small planktonic microbes in the mid-salinity zone of the Chesapeake Bay encode rhodopsins. (B) Ratio of rhodopsin transcripts to rplB transcripts. Rhodopsin is much more highly expressed than rplB, especially in the mid-salinity zone.
Reads obtained in metagenomic and metatranscriptomic sequence data setsa
Comparison of cells synthesizing functional rhodopsins or genomes containing rhodopsin genes, quantified by different methodsa
The transcription of rhodopsin genes also appeared to vary along the salinity gradient (Fig. 6B). Gene expression ratios were clearly correlated to the number of copies of rhodopsin per genome (R = 0.95, P < 0.05). Total rhodopsin gene expression, when normalized to rplB gene expression, was highest in the mid-salinity <0.8-μm-size-fraction samples. In the >0.8-μm-size fraction, the patterns of rhodopsin expression were different from the patterns of rhodopsin gene distribution (Fig. 7). The transcription of actinobacterial rhodopsin genes in both size fractions decreases in relative abundance along the salinity gradient. The relative abundances of rhodopsin transcripts and genes from SAR11 and other Alphaproteobacteria are similar at all three sites. However, transcripts of rhodopsins from the Bacteroidetes/Chlorobi group (primarily transcripts and genes from Flavobacteria; Fig. 8) are highly abundant in the larger-size fraction at the mid-salinity site. Rhodopsin genes from eukaryotes are not detectable in the metagenomic data set, but eukaryotic rhodopsin transcripts are present in the mid-salinity and marine sites.
Taxonomic affiliations of rhodopsin genes and transcripts in the Chesapeake Bay. Relative abundances of rhodopsin genes and transcripts affiliated with specific phyla are plotted against salinity. DNA and RNA were sequenced from the large-size (>0.8 μm) fractions at three salinities (0.1, 15, and 31 ppt), and from the small-size (<0.8 μm) fractions at 15 and 31 ppt. SAR11-type proteorhodopsins are the most abundant rhodopsin type in the small-size fractions, while rhodopsin transcripts from Bacteroidetes are highly abundant in the larger-size fractions.
Phylogeny of rhodopsins found in the Chesapeake Bay. SAR11-associated proteorhodopsins in the Chesapeake Bay include at least 7 subtypes of rhodopsins, and Bacteroidetes-associated rhodopsins are similarly diverse. All identified rhodopsins in these data sets belong to families of proton-pumping rhodopsins, so sensory rhodopsins were used as the outgroup.
At the mid-salinity site (15 ppt), although only 15% of the small (<0.8 μm) heterotrophic microbes produce functional rhodopsins, the qPCR results indicate that close to 40% of genomes in this size fraction encode either SAR-PR or LG1-type rhodopsins, and the metagenome analysis suggests that nearly all genomes encode a rhodopsin (Table 3). In the higher-salinity samples, more cells produce functional rhodopsins, the qPCR results suggest that rhodopsin genes are highly abundant, and the metagenomic data set suggests that ∼74% of genomes encode rhodopsins. The discrepancies between the qPCR and metagenomics data are likely due to the variability of qPCR efficiency and to the fact that neither qPCR targeted rhodopsins from Bacteroidetes, which were abundant in the mid-salinity metagenomic data set (Fig. 7).
The phylogenetic diversity of rhodopsins varied along the salinity gradient (Fig. 7) but was consistent between the metagenomes and metatranscriptomes. The freshwater samples were dominated by rhodopsins associated with Bacteroidetes and Actinobacteria. The <0.8-μm-size-fraction samples from 15 and 31 ppt salinity levels were dominated by rhodopsins in the SAR11 clade within the Alphaproteobacteria class. In one of the mid-salinity >0.8-μm-size-fraction RNA samples, ∼88% of the rhodopsin transcripts appeared to originate from one taxon (operational taxonomic unit 23 [OTU23]) similar to Phaeodactylibacter xiamenensis (Fig. 8). This Bacteroidetes species was originally isolated from the phycosphere of a marine microalga (44, 45), providing additional support for the proposed association between rhodopsin-encoding heterotrophs and primary producers in the Chesapeake Bay.
The dominant OTUs associated with the larger-size fraction were mostly within the Bacteroidetes phylum, with some also associated with Actinobacteria, Gammaproteobacteria, and SAR11 groups (Fig. 7 and 8). In the small-size fractions, the Actinobacteria and SAR11 groups together comprised 55% of the rhodopsin genes in the high-salinity metagenome and 81% of the rhodopsin genes in the mid-salinity metagenome. At least in the mid-salinity samples, targeting only SAR-PR and LG1 captured most of the rhodopsin-encoding genes present.
Rhodopsins may either pump ions or initiate a signal cascade, and TIRF microscopy does not distinguish between the two types of rhodopsins. However, none of the rhodopsin genes identified in either the metagenomic and metatranscriptomic analyses were related to known sensory rhodopsins; all fell into clades with known light-activated ion pumps, and in fact, sensory rhodopsins were used as the outgroup in our phylogenetic analysis (Fig. 8).
DISCUSSION
The physiological role(s) of rhodopsins varies among microbial taxa and may change as environmental conditions change, but the environmental factors contributing to the distribution and expression of rhodopsins are not well understood. For this reason, we measured both rhodopsin gene abundance and functional rhodopsins along environmental gradients. If genetic potential is the primary factor controlling rhodopsin production, the ratio of organisms that produce functional rhodopsins to genomes encoding rhodopsin genes will not change as environmental conditions change. However, if environmental parameters affect rhodopsin production, changes in the functional protein-to-gene ratio will be correlated with changes in those parameters.
Rhodopsin production is correlated with heterotrophy in the Chesapeake Bay.We tested the hypothesis that rhodopsin protein production is associated with light and salinity. Neither of these factors was significantly correlated with the ratio of cells producing functional rhodopsins to genomes encoding rhodopsins. Instead, during the day, this ratio is positively correlated with heterotrophic bacterial production, Chl a (a general indicator of photosynthesis), and dissolved silicate, which is associated with diatom abundance (46). Based on the correlations between primary production, heterotrophy, and functional rhodopsin production, we propose that in the Chesapeake Bay, microbial rhodopsin activity is primarily associated with heterotrophic activity fueled by photosynthate. In the light, the number of primary producers increases, resulting in greater availability of organic carbon. This organic carbon in turn supports more bacterial production (47, 48). The heterotrophs that consume photosynthetically produced organic carbon in illuminated waters are exposed to light, so light energy is a potential resource for them as well. Cells with rhodopsins could use light to power processes that utilize ion (H+ or Na+) gradients, including motility, transport of organic or inorganic carbon, or cofactor import. Indeed, the abundance of proteorhodopsins of the SAR92 group has been correlated with Chl a in the Arctic Ocean, potentially indicating that proteorhodopsin-encoding SAR92-related organisms respond to algal blooms (49). Further, upregulation of ion-dependent transport functions in the light has been observed in (proteo)rhodopsin-encoding cells (4, 26, 28). Thus, light may provide energy that allows heterotrophs under carbon-replete conditions to better coordinate organic carbon import and processing with organic carbon release by primary producers. Although our data show a clear association between the presence of light and the production of rhodopsins, no effect of light intensity on rhodopsin production was observed.
Some cyanobacteria and algae have rhodopsin genes (50), so we considered the possibility that the correlation between Chl a and rhodopsin abundance was due to primary producers that also synthesize rhodopsins. However, Chl a-producing organisms observed were excluded from the TIRF analysis (see Materials and Methods), and no sequences related to cyanobacterial rhodopsins were observed in the metagenomic or metatranscriptomic data sets. For these reasons, the observed correlation is not due to organisms, such as Anabaena sp. PCC7120 or Gloeobacter violaceus, that have both chlorophyll- and rhodopsin-type photosystems (51, 52).
Although the ratio of cells producing functional rhodopsins to genomes encoding rhodopsins was not significantly correlated with salinity, this ratio is highest in a salinity range of 5 to 10 ppt, where freshwater organisms may be exposed to higher salinity than their optimum, and 25 to 30 ppt, where marine organisms are exposed to lower salinity than their optimum. Previous work in the Chesapeake Bay has suggested that salinity has the largest effect on microbial community structure in low-salinity (<5 ppt) or high-salinity (>30 ppt) regions (68). On the borders of freshwater or marine systems, rhodopsins may be expressed either by freshwater or marine organisms experiencing osmotic stress. In these cases, ion-pumping rhodopsins might be involved in maintaining osmotic balance or in conserving energy under specific stress conditions, as has been shown for the marine bacterium Psychroflexus torquis (30).
Rhodopsin gene abundance is correlated with salinity in the Chesapeake Bay.Because environmental microbial community composition is strongly dependent on salinity (53–57) and rhodopsins are common in marine microbes in surface waters (13, 14, 32, 58–60), we had hypothesized that the abundance of rhodopsin-encoding genes typical of marine microbes, such as SAR11-type proteorhodopsins, would be positively correlated with salinity, while actinorhodopsins, which are typical of freshwater Actinobacteria, would be negatively correlated with salinity (14, 15, 31, 41, 49, 57, 61). As predicted, the percentage of cells encoding SAR11-type proteorhodopsins increased with salinity in the qPCR, metagenomic, and metatranscriptomic analyses, indicating that the influence of marine waters strongly affects the composition of the microbial communities in the Bay. The percentage of cells encoding actinorhodopsin decreased as salinity increased but only in the samples collected at night, when actinorhodopsin-encoding genomes have a much greater relative abundance than during the day. These opposing trends have also been observed for Alphaproteobacteria and Actinobacteria in the nearby Delaware Bay (62) and for proteorhodopsins and actinorhodopsins in the Baltic Sea (31, 63). The discrepancy between day and night samples may be due to diurnal movement of primary producers. Algae and cyanobacteria tend to sink through the water column during the late afternoon and evening (64–66), which may make the Actinobacteria larger fractions of the surface water microbial communities at night.
The high abundance of SAR11-type proteorhodopsin genes detected by qPCR in the marine sample (over 100% of microbial genomes are predicted to encode a rhodopsin) may indicate that some genomes encode multiple proteorhodopsins. This estimate is based on the assumption that the average copy number of 16S rRNA genes in the Chesapeake Bay is ∼1.9 (67). If this assumption is incorrect, the percentage of genomes encoding rhodopsins may be overestimated by nearly 2-fold. However, the observed trends with salinity would hold true, regardless of the precise number of rhodopsin or 16S rRNA gene copies per genome. Alternatively, this high estimate may be a result of the different efficiencies of the 16S rRNA gene and SAR-PR gene qPCRs.
The qPCR analysis described here did not amplify rhodopsin genes associated with Bacteroidetes, Alphaproteobacteria other than SAR11, or Gammaproteobacteria. However, in the metagenomic data from the mid-salinity small-size fraction sample, rhodopsins from Actinobacteria and SAR11-type Alphaproteobacteria accounted for 81% of the observed rhodopsins, suggesting that the qPCR likely detected the majority of the rhodopsin genes present. In the highest-salinity small-size-fraction sample, rhodopsins from Bacteroidetes and Alphaproteobacteria other than SAR11 were larger fractions of the rhodopsin pool in the metagenomic and metatranscriptomic data sets. The Bacteroidetes primers that we tested were developed for amplification of rhodopsins from marine Bacteroidetes (35). After we obtained the metagenomic data sets, these primers were tested against the assembled rhodopsin genes in silico and would not have successfully amplified the rhodopsin genes associated with Bacteroidetes in the Chesapeake Bay samples. Thus, in the mid-salinity samples, the ratio of cells with rhodopsin fluorescence to cells with rhodopsin genes may be lower than what we calculated here. If this is indeed the case, the correlation of this ratio with bacterial production would also be greater, strengthening the argument that production of functional rhodopsins is associated with heterotrophic activity rather than nutrient limitation in the Chesapeake Bay. Although we lack metagenomic data for the small-size fraction in the freshwater sample, we would predict based on the other metagenomic data that a greater fraction of detectable rhodopsins would be actinorhodopsins and fewer would be SAR11-type proteorhodopsins. The fraction of rhodopsins associated with Bacteroidetes might be larger than in the mid-salinity sample, but because Bacteroidetes cells are not typically as small as the SAR11-type Alphaproteobacteria and Actinobacteria, they might not have passed through the 0.8-μm-pore-size filter. In a 2007 study of Chesapeake Bay microbial communities, SAR11-type Alphaproteobacteria and Actinobacteria together comprised 70% of the microbial community near the headwaters of the Bay, while Bacteroidetes were ∼11% of the community (68). For these reasons, we conclude that the qPCR data presented here, while not a comprehensive analysis of all possible rhodopsin genes, reflect the main groups of rhodopsins in the small-size fraction of the Chesapeake Bay microbial community.
Trends in the rhodopsin-encoding genes in the metagenomic data are generally consistent with trends in the qPCR and TIRF data, though only two of the metagenomic and metatranscriptomic samples are from the same size fraction as the TIRF microscopy and qPCR samples. Between the mid- and high-salinity samples, the relative abundance of SAR-PR gene decreases, likely reflecting the relative increase in the abundance of rhodopsins associated with Alphaproteobacteria of the Rhodobacterales group. The metagenomic data show that the qPCR analysis missed rhodopsins from Alphaproteobacteria other than SAR11 and from the Bacteroidetes, which encode a variety of rhodopsins (69, 70). Additionally, the metagenomic data suggest that in the mid- and high-salinity samples (15 and 31 ppt, respectively), rhodopsins from the Bacteroidetes group are at least as abundant as actinorhodopsins in the small-size-fraction (<0.8 μm) samples. The Bacteroidetes-type rhodopsins are much more abundant in the larger-size fraction, suggesting that these cells may be common in multicellular and/or particle-associated aggregates. Although rhodopsins from marine Bacteroidetes have been well characterized recently (4, 26, 27, 30, 35, 70–72), rhodopsins in freshwater Bacteroidetes have not, though rhodopsin genes associated with Bacteroidetes (specifically Flavobacteria) have been observed (73). Given the abundances of genes and transcripts from Bacteroidetes in the freshwater and mid-salinity samples, this group of rhodopsins clearly merits more study.
Rhodopsin abundance patterns in the Chesapeake Bay compared to other environments.In the Chesapeake Bay, a mesotrophic estuary with generally high concentrations of nutrients and salinity that decrease with increasing distance from the ocean, rhodopsin gene abundance seems to be primarily controlled by salinity. Overall, the results described here are similar to those of studies of rhodopsin gene abundance and expression in the Baltic Sea, an estuary with a salinity gradient similar to that of the Chesapeake Bay. In the Baltic Sea, salinity affected the abundance of rhodopsin genes but not rhodopsin expression (31, 63). Instead, quality and bioavailability of organic carbon also contribute to bacterial growth efficiency (74, 75), and the availability of dissolved organic carbon may control rhodopsin expression there (31), suggesting that light may be linked to the regulation of heterotrophy.
In contrast to estuaries, with their steep salinity gradients, the eastern Mediterranean Sea has fairly constant salinity and steep nutrient gradients. Recent work by Gómez-Consarnau et al. using retinal as a proxy for functional rhodopsin concentrations in the Mediterranean Sea and eastern Atlantic Ocean found that retinal concentration was inversely proportional to Chl a concentration, and that the highest concentration of retinal was found in the most oligotrophic areas of the Mediterranean (19). All of their samples were marine, removing salinity as a major driver of rhodopsin gene abundance; the major gradient in their samples was nutrient concentrations, and they conclude that rhodopsins are abundant enough in the oligotrophic regions to meet cellular maintenance energy requirements (19). Similarly, previous analysis had shown that the number of rhodopsin-encoding genes in metagenomic data sets in the Mediterranean increases as nutrient concentrations decrease (60).
Perhaps in oligotrophic environments, such as the eastern Mediterranean or open ocean, energy supplementation is the most important physiological role of these rhodopsins, while in environments with higher levels of nutrients, rhodopsins power active transport (26) or other processes, enhancing the ability of heterotrophic bacteria to take advantage of organic carbon or small molecules produced by phytoplankton. Since the supplemental energy provided by rhodopsins may be used for physiological activities other than maintenance energy in environments where C, N, and P are not limiting, rhodopsin production in the Chesapeake Bay may be controlled by different factors from those in typical marine environments.
Summary.Light is a ubiquitous resource in surface environments. This work demonstrates that light is actively captured via functional rhodopsins in the Chesapeake Bay, where up to 40% of the microbes in the surface water produce active rhodopsins. Salinity controls the distribution of microbial rhodopsin genes in the Chesapeake Bay, while time of day and bacterial production appear to control the percentage of cells that synthesize rhodopsins. The association of functional rhodopsin abundance with Chl a and bacterial production, as proxies for locally available photosynthate, suggests that in the Chesapeake Bay, rhodopsins are utilized by active heterotrophic microbes that do not suffer from nutrient or energy limitation. We hypothesize that the light-dependent proton motive force supplied by rhodopsins contributes to heterotrophy not solely by enabling additional ATP synthesis but also by powering proton motive force-dependent transport of organic carbon and/or cofactors released by primary producers. It is clear that combinations of genetic and environmental factors work uniquely in different microbes to control rhodopsin gene expression, possibly separately from synthesis of the retinal cofactor (30). Future work in this field would likely benefit from high-throughput cultivation methods (71) or functional metagenomic screens (76) that link rhodopsins with physiological traits or specific genetic pathways.
MATERIALS AND METHODS
Sample collection and storage.Surface water samples were obtained with a 12 Niskin bottle rosette sampler with a conductivity, temperature, depth (CTD) tool on the R/V Sharp sampled along the length of the Chesapeake, from its source at the Susquehanna River to the Atlantic Ocean, from 11 to 16 April 2015. Water quality data, including temperature, salinity, dissolved oxygen content, turbidity, and fluorescence, were measured for each cast using a Sea-Bird data sonde. Samples were also collected for later determination of nutrient concentrations (NO32−, NH4+, PO43−, and SiO42−) and bacterial production, as described previously (62, 77, 78).
Rhodopsins were quantified by TIRF microscopy and qPCR in the samples collected at 11:00 a.m. and 11:00 p.m. Eastern Standard Time (EST) each day. Samples (100 ml) for TIRF microscopy were prefiltered through 1-μm-pore-size filters, fixed in 4% paraformaldehyde, and stored at 4°C until analysis. Samples (100 ml) for qPCR analysis were prefiltered through 1-μm-pore-size filters, collected on 0.2-μm-pore-size filters provided with the Mo Bio PowerWater kits (catalog no. 14900; Mo Bio, Carlsbad, CA), and stored at −20°C until analysis. An equal volume of RNAlater (Invitrogen) was added to a liter of water immediately after the rosette sampler was placed on the ship for metagenomic and metatranscriptomic analyses. Water was filtered through 0.8- and 0.22-μm-pore-size filters, and filters were frozen at −80°C within 1 h of collection.
DNA extraction and qPCR.DNA extractions were performed using a Mo Bio PowerWater kit, according to the manufacturer's instructions. Preliminary PCR was performed on positive controls using LG1 (41), SAR-PR (35), and 16S (79) primer sets to optimize conditions for the amplification of actinorhodopsins, SAR11-type proteorhodopsins, and 16S rRNA genes, respectively (Table 1). Additional primer sets were tested but not used because they did not amplify anything from the Chesapeake Bay samples. These included primers designed to amplify proteorhodopsins from Bacteroidetes, specifically sphingomonads and flavobacteria (35). The positive-control template for SAR-PR was a SAR11-type proteorhodopsin amplified from water collected from the Delaware River and cloned into pCR4 (33); the positive control for LG1 was Rhodoluna lacicola genomic DNA, since R. lacicola has a proton-pumping actinorhodopsin (41, 80, 81).
Quantitative PCR (qPCR) was performed in triplicate with 5 μl of DNA (2.5 to 5.4 ng · μl−1) in a final volume of 20 μl using the Quanta Biosciences PerfeCTa SYBR green FastMix for iQ (catalog no. 95071; Quanta Biosciences, Gaithersburg, MD). All primer concentrations were 0.25 μM. Standard curves were made using genomic DNA from R. lacicola (81) as the template for the actinorhodopsin and 16S primers and pCR4-SAR11 as the template for the SAR11-type proteorhodopsin. Average amplification efficiencies were as follows: 16S rRNA, 54%; LG1, 39%; and SAR11 PR, 73%. Rhodopsins were amplified using the following program: 95°C 2.5 min, followed by 40 cycles of amplification at 95°C for 15 s, the indicated annealing temperature (Table 1) for 30 s, and 72°C for 30 s.
The number of copies of each gene was calculated using the threshold cycle (CT) values and the standard curves for each reaction. To estimate the percentage of genomes encoding rhodopsins, we assumed no more than one rhodopsin and 1.9 copies of the 16S rRNA gene per genome (14).
Nucleic acid extraction for metagenomic and metatranscriptomic analysis.Samples for metagenomic and metatranscriptomic sequencing were collected from sites 1, 17, and 33. Samples were filtered through 0.8- and 0.22-μm-pore-size filters to separate cells into large- and small-cell-size fractions, respectively. DNA and RNA were extracted from the filters simultaneously using the Qiagen AllPrep kit, according to the manufacturer's instructions. DNA was removed from the resulting RNA preps with the Ambion Turbo DNA-free DNase kit (Invitrogen), and the RNA was checked for DNA contamination using standard PCR with universal bacterial primers 1369F and 1492R targeting the 16S rRNA gene (79).
Nucleic acids (RNA and DNA) for large- and small-size fractions were obtained from sites 17 and 33, and for the large-size fraction alone from site 1. All samples were sent to the Joint Genome Institute (JGI) for library preparation and sequencing on the Illumina HiSeq 2000 following their standard protocols and as outlined separately (see Appendix in the supplemental material). Sequences averaged 150 bp in length, and 46 million to 275 million reads were obtained from each sample (Table 2).
Metagenomic and metatranscriptomic sequence analyses.Two genes (the rhodopsin gene and rplB) were assembled using the Xander program, which utilizes hidden Markov models (HMM) of the gene of interest in directing gene assemblies (43). The HMM used for the rplB gene was downloaded from the functional gene repository (http://fungene.cme.msu.edu/). The HMM for the rhodopsin gene was generated from a seed database of 17 phylogenetically diverse rhodopsins, and alignments and taxonomic identities were generated compared to a reference database of 1,000 rhodopsins. Default assembly parameters were used, and no chimeras were identified from the assembled gene contigs. Xander coverage files were used to estimate the total number of sequences per gene and mean coverage, as described in the original program (43). The number of rhodopsins per genome equivalent was determined as the ratio of the rhodopsin to rplB abundance, since rplB is a single-copy gene encoding the 50S ribosomal protein L2, which is found in most, if not all, bacteria, and has been utilized previously in this manner (43). Translated contigs were clustered at 0.2 distance to generate OTUs and abundance data for diversity and phylogenetic analyses. The phylogenetic positions of representative rhodopsin OTUs were determined via BLASTP analysis and assignment in MEGAN, as previously described (82). Phylogenetic relationships of the rhodopsin OTUs were further resolved using MEGA version 6 (83) to generate a ClustalW alignment, followed by construction of a maximum likelihood tree with the default parameters (JTT model, 500 bootstraps, partial deletion).
TIRF microscopy.Fixed water samples were concentrated by filtration onto a 0.2-μm-pore-size filter, stained with DAPI (NucBlue fixed-cell ReadyProbes reagent, catalog no. R37606; Life Technologies), and adhered to gelatin-coated coverslips, as described previously (33, 84). Samples were sequentially excited with 405 nm-, 561 nm-, and 641 nm-lasers and viewed using a lab-built TIRF imaging system to visualize all cells, rhodopsin- and phycobiliprotein-containing cells, and Chl a-containing cells, respectively (33). Positive controls for TIRF microscopy were analyzed during each microscopy session and included an algal isolate for Chl a fluorescence and Escherichia coli expressing actinorhodopsin and grown with retinal for rhodopsin fluorescence (33, 84); both were stained with DAPI prior to microscopy. Objects that fluoresced when excited with the 405-nm and 561-nm lasers only were counted as rhodopsin-containing cells. Three independent slides were prepared from each sample, and 15 to 20 fields of view were imaged for each slide.
Statistical analysis.Statistical analyses comparing samples collected at all sites were performed using PRIMER 7 (85, 86). Environmental and biological (TIRF, LG1-qPCR, and SAR11-qPCR) data were log-transformed and normalized, and then a Bray-Curtis similarity matrix was calculated. The nonmetric multidimensional scaling routine (NMDS) in PRIMER 7 was used to identify overall trends in the data. Since the day and night samples were clearly differentiated in the NMDS plot (Fig. S3), they were analyzed separately in subsequent analyses.
Correlation coefficients (Pearson and Spearman) between individual variables were calculated on both untransformed and log-transformed data using the R package rcorr. Most iterations of this analysis identified the same variables as being correlated. Therefore, for simplicity, Pearson correlation coefficients (r) calculated based on untransformed data are presented here. The sample at site 1 had an anomalously high TIRF-to-qPCR ratio (∼24, implying that far more cells produced rhodopsins than had detectable rhodopsin genes) and was not included in the statistical analyses.
Accession number(s).All sequences are found on the JGI Genome Portal with the following project identification numbers: 1110833, 1110835, 1110838, 1110849, 1110851, 1110854, 1110841, 1110843, 1110846, 1110865, 1110867, 1110870, 1110857, 1110859, 1110862, 1110881, 1110883, 1110886, 1110873, 1110875, 1110878, 1110897, 1110899, 1110902, 1110889, 1110891, 1110894, 1110905, 1110907, 1110910, 1110921, 1110923, 1110926, 1110913, 1110915, and 1110918.
ACKNOWLEDGMENTS
The research reported in this publication was supported in part by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under grant number 5 P30 GM103519. This research cruise was supported by National Science Foundation grants to B.J.C. (grant OCE-082546), and metagenome and metatranscriptome sequencing were supported by a DOE/JGI grant to B.J.C. (grant CSP-1621). Microscopy access was supported by the INBRE program with a grant from the NIH-NIGMS (grant P20 GM103446) and the State of Delaware.
We thank the crew of the R/V Hugh R. Sharp for sample collection and David Kirchman, Matt Cottrell, and Liying Yu for technical and sampling support.
FOOTNOTES
- Received 17 January 2018.
- Accepted 23 April 2018.
- Accepted manuscript posted online 27 April 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AEM.00137-18.
- Copyright © 2018 American Society for Microbiology.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
