Biofilm formation and cell plasticity drive diazotrophy in an anoxygenic phototrophic bacterium

ABSTRACT

Non-cyanobacterial diazotrophs (NCDs) are widespread and active in marine waters. The carbon and low-oxygen (O₂) conditions required for their N₂ fixation may be encountered on marine particles, while a putative role of light remains uninvestigated. This study explored factors that regulate N₂ fixation in Rhodopseudomonas sp. BAL398—a anoxygenic phototrophic bacterium isolated from low-salinity surface waters. Light (250 µmol photons m⁻² s⁻¹) and anoxia (0 µM O₂) stimulated growth and N₂ fixation; however, diazotrophy in light was dependent on high organic carbon levels (35 mM, glucose:succinate). Immunolabeling revealed that cellular nitrogenase levels increased with light, decreasing inorganic nitrogen (N) and ambient O₂ (250 µM). Light and O₂ stimulated motility and biofilm formation on surfaces, and N₂ fixation rates increased compared to the control treatment. N₂ fixation rates were positively correlated with the formation of rosette-like cellular structures, and an increased concentration of nitrogenase was observed toward the center of these structures, which increased their occurrence 600 times when cultures reached maximum N₂ fixation rates vs when they had low rates. Interestingly, N₂ fixation was not completely inhibited under oxic conditions and was accompanied by increased formation of capsules and cysts. Rosettes, as well as capsules and cysts, may thus serve as protection against O₂. Our study reveals the physiological adaptations that underlie N₂ fixation in an anoxygenic phototroph, emphasizing the significance of biofilm formation for utilizing light and fixing N₂ under oxic conditions, and underscores the need for deciphering the importance of light for marine NCDs.

IMPORTANCE

The contribution of non-cyanobacterial diazotrophs (NCDs) to total N₂ fixation in the marine water column is unknown, but their importance is likely constrained by the limited availability of dissolved organic matter and low O₂ conditions. Light could support N₂ fixation and growth by NCDs, yet no examples from bacterioplankton exist. In this study, we show that the phototrophic NCD, Rhodopseudomonas sp. BAL398, which is a member of the diazotrophic community in the surface waters of the Baltic Sea, can utilize light. Our study highlights the significance of biofilm formation for utilizing light and fixing N₂ under oxic conditions and the role of cell plasticity in regulating these processes. Our findings have implications for the general understanding of the ecology and importance of NCDs in marine waters.

INTRODUCTION

Biological N₂ fixation, a process carried out by specialized prokaryotes (diazotrophs), provides fixed nitrogen to the marine water column, thereby impacting nitrogen and carbon cycling (1, 2). While cyanobacteria were traditionally considered the only important N₂ fixers in the water column (2, 3), accumulating evidence shows that non-cyanobacterial diazotrophs (NCDs) are also widespread and active in marine waters (4 – 9). The contribution of NCDs to total N₂ fixation remains unknown (9), but their importance is likely constrained by accessible labile dissolved organic matter and the limited availability of low O₂ conditions in the marine water column (6). As originally proposed (10) and recently revie (11), particles could be loci for N₂ fixation by NCDs in the water column. Particle colonization would provide NCDs with locally elevated nutritional resources and access to low O₂ microsites. Still, organic carbon availability will be an important factor limiting N₂ fixation (12). To date, organic matter has been assumed to be the only likely energy source for NCDs, whereas the potential role of light in supporting N₂ fixation remains to be investigated. However, some anaerobic NCDs containing bacteriochlorophylls, i.e., anoxygenic phototrophic bacteria, have been found in association with copepods (13), indicating that some surface-associated phototrophic NCDs in the plankton can utilize light.

Anoxygenic phototrophs such as purple non-sulfur bacteria (PNSB) are a metabolically versatile group that performs phototrophic and/or chemotrophic metabolism and uses a wide range of organic and inorganic carbon sources (14 – 16). However, their phototrophic properties are inevitably associated with illuminated anoxic environments, as their light-harvesting complexes are strongly suppressed at ambient O₂ levels (17, 18). These versatile bacteria are widely distributed in aquatic environments, including upper sediment layers of both marine and freshwater bodies, ponds, lakes, tide pools, sewage, and microbial mats (19). Some PNSBs are diazotrophs, and their growth and N₂ fixation are controlled by light, O₂, and nutrient availability (14, 20 – 23). Despite their potential contributions to marine ecosystems, the phenotype, ecology, and diazotrophy of PNSB in marine systems remain largely unknown. Although PNSBs are typically found in low O₂ zones, a few studies have shown their presence in oxic surface waters (24, 25). However, the mechanisms by which they can protect themselves from O₂ and use light to support N₂ fixation remain unexplored.

In the present study, we investigated the environmental controls on N₂ fixation in Rhodopseudomonas sp. BAL398, a PNSB isolated from the surface water of the Baltic Sea [Landsort Deep station (25)]. BAL398 was present in surface waters throughout the year [2003–2004, ca. 10⁴ nifH gene copies L⁻¹ (25)], and it was detected in nifH clone libraries generated from the same station (26). Furthermore, we recently found BAL398 in 16S rRNA gene amplicon libraries from surface waters of 12 out of 26 Baltic coastal stations sampled in September 2021. Interestingly, BAL398 was not found in the corresponding sediment samples (unpublished data, D. Riedinger). Herein, using a combination of traditional and molecular microbiology techniques together with microscale sensing of O₂ and advanced holotomographic imaging, we gained insights into the ecophysiology of a PNSB, BAL398, that is a consistent and prominent member of the diazotrophic community in the surface waters of the Baltic Sea. Our study sheds light on the adaptations underlying N₂ fixation in this phototrophic NCD, emphasizing the significance of biofilm formation for utilizing light and fixing N₂ under oxic conditions.

RESULTS

Rhodopseudomonas sp. BAL398 is a species with high metabolic versatility

Rhodopseudomonas sp. BAL398 is a rod-shaped anoxygenic photosynthetic bacterium (1.5–4 µm long and 0.5–1.1 µm wide) that exhibits large metabolic flexibility. It can use photoheterotrophic and photoautotrophic metabolism, as well as chemoheterotrophic and chemoautotrophic metabolism (Table S1). In our tests, BAL398 showed highest growth at 1% salinity. Table S1 summarizes the genomic and physiological characteristics of Rhodopseudomonas strains that are closely related to BAL398. The re-sequenced BAL398 genome was assembled into four contigs (6.2 Mb and 64.2% GC; accession numbers CP133111–CP133114), and one of these contigs represented the close chromosome of the BAL398 (5.9 Mb; Fig. S1). BAL398 contains genes coding for the photosynthetic reaction centers involved in electron transfer (pufM and pufS), RubisCo synthesis (cbbS and ccbL), bacteriochlorophyll (Bchl) synthesis (bch genes), biofilm formation (hsbR-hsbA), flagellum biosynthesis (fli, flg, and mot genes), and chemotaxis (che genes). It encodes the entire nif operon but not alternative nitrogenases. Analyses of specific genes and protein (16S rRNA, pufM, and NifH) and genomes [digital DNA-DNA hybridization (dDDH) and nucleotide identities by BLAST (AniB)] showed that BAL398 has a distinct phylogeny (Fig. S2). Based on the low similarity values found, BAL398 is potentially a new species within the Rhodopseudomonas genus (Table S1).

We carried out experiments to examine the importance of photoheterotrophy for N₂ fixation in Rhodopseudomonas sp. BAL398, specifically to address the importance of light and O₂, regulation of biofilm formation, and cellular morphological plasticity. The experimental details are outlined in Sections 1–3 of Table 1.

TABLE 1.

Summary of the experiments carried out with Rhodopseudomonas sp. BAL398 ^b

Question	Oxygen ( µM O2)	Light (µmol photons m−2 s−1)	Carbon (mM C) and nitrogen (mM N)	Duration and conditions
Section 1. How do light, O2, and nutrients impact BAL398 metabolism, growth, and N2 fixation? (Fig. 1A through F; 2A and B; Fig. S2A through C)	0 (anoxic) 100 (semi-anoxic) a 250 (oxic)	0 (dark conditions) and 250 (light conditions) Tested a light gradient: 0, 28, 70, 120, 158, 193, and 250	C: 0, 0.35, and 35 N: 0, 0.15, and 7.5	7–10 d at 20°C, shaken
Section 2. Is cell attachment and biofilm formation a strategy for the BAL398 to thrive and fix N2 in fully oxygenated waters? (Fig. 3A through F and 4A through C; Fig. S3A through D)	0 and 250	0 and 250	C: 0, and 35 N: 0, 0.15, and 7.5
I. Effect of O2 depletion on BAL398 physiology (Fig. 3A; Fig. S3A and B)	250	250	C: 35 N: 7.5	7 d at 20°C, static or shaken
II. Effect of nutrients and light on biofilm Fig. 3B	0	0 and 250	C: 0 and 35 N: 0, 0.15, and 7.5	3–7 d at 20°C, static
III. Effect of artificial surfaces on biofilm formation: polyethylene, polystyrene, and glass (Fig. 3C through F)	0 and 250	0 and 250	C: 35 N: 0.15
IV. Changes in the early biofilm (Fig. 4A through C)	0 and 250	0 and 250	C: 35 N: 0.15
Section 3. How is the morphological plasticity of BAL398 related to its diazotrophic activity? (Fig. 5A through F and 4A through I; Fig. S5A and B)	0 and 250	0 and 250	C: 35 N: 0.15 and 7.5	7 d at 20°C, shaken

^a Treatment only used to study the effect of O₂ gradient in BAL398 (Fig. 1C).

^b Note that the organic carbon and nitrogen source used, i.e., C and N, were glucose:succinate 1:1, and NH₄Cl, respectively.

Section 1: regulation of growth and diazotrophy by light, O₂, and nutrient availability

After 7 d, BAL398 reached the highest cell abundance under light and anoxic conditions (Fig. 1A). It maintained maximum growth rate (µ = 0.28 ± 0.03 d⁻¹, Fig. 1A) under 35 mM organic carbon (C, glucose:succinate) and 0.15 mM inorganic nitrogen (N, NH₄Cl)—i.e., under N₂-fixing condition for this strain (27)—yet the growth rate was significantly lower than at high N levels (7.5 mM; µ = 0.35 ± 0.04 d⁻¹, P < 0.05, Fig. 1A). Under N₂-fixing conditions, cell abundance increased with photon irradiance (Spearman’s rank-order correlation test, r = 0.78, n = 21, P < 0.05, Fig. 1B), and flavin levels, indicative of light utilization, were significantly elevated under light (Fig. S3A). Increases in O₂ levels resulted in reduced cell abundance independent of light and N availability (Fig. 1C). Indeed, ambient O₂ levels under light conditions decreased N₂ fixation ca. 400-fold (Fig. 1D). However, the nitrogenase content increased under oxic relative to anoxic conditions (Fig. 2A). Remarkably, N₂ fixation was not completely inhibited under ambient O₂ concentration (250 µM, Fig. 1D), and high C concentrations (35 mM) stimulated N₂ fixation (Fig. S3B). Under these conditions, i.e., oxic and high C concentrations, light did not stimulate N₂ fixation activities (Fig. S3C). These results show that light and O₂ influence diazotrophy in BAL398, while the nutrient regime also influences growth.

Fig 1.

Effects of light and O₂ on growth and N₂ fixation in BAL398 under varying carbon (C) and nitrogen (N). (A) Growth rates (µ) and duplication times (Tg) under different light, O₂, and N conditions. In red, the growth rate under aerobic conditions. (B) Cell abundance under different photon irradiance (from 0 to 250 µmol photons m⁻² s⁻¹) at N₂ fixing (0.15 mM N) and anoxic conditions. (C) Growth under varying O₂ levels (0, 100, and 250 O₂ µM) at 35 mM C. (D) N₂ fixation rates under different O₂ concentrations and light conditions. (E) Cell abundance. (F) N₂ fixation under different light, O₂, C (0. 0.35, and/or 35 mM), and N availability (0, 0.15, and/or 7.5 mM). Note that in (F), N₂ fixation was not measured at C = 0 mM and N = 0 mM. The values represent the mean, and the error bars indicate the standard error between the replicates (n = 3). Different letters indicate pairwise analysis among variables (P < 0.05), using a post hoc test (Wilcoxon) after Kruskal-Wallis over the whole data set. Plots (A−F) are from independent experiments. In (A), data points were recorded every day during 10 d, and in (B−F), values were measured after 7 d. For more detailed information about the conditions tested, please refer to Table 1, Section 1.

Fig 2.

Nitrogenase content in BAL398 under different environmental conditions. (A) Effect of O₂ concentration (0 and 250 µM O₂). (B) Impact of N concentration (0, 0.15, and 7.5 mM) and light availability on the nitrogenase content of BAL398. In panels (A and B) BAL398 cells were cultured under high C concentration (35 mM). In panel (A) BAL398 cells were harvested under light and N₂-fixing conditions. The values represent the mean, and the error bars indicate the standard error between the replicates (n = 3). Different letters indicate pairwise analysis among the variables (P < 0.05), using a post hoc test (Wilcoxon) after Kruskal-Wallis over the entire data set. A.U is defined as arbitrary units. (A and B) are from independent experiments, and values were measured after 7 d. For more detailed information about the conditions tested, please refer to Table 1, Section 1.

Under anoxic conditions, only high C (35 mM) and N (7.5 mM) levels stimulated cell abundance of BAL398 (Fig. 1A and E). Light stimulated N₂ fixation at high carbon availability (up to 40-fold), regardless of the N level (Fig. 1F). At N₂-fixing conditions, light also stimulated nitrogenase content (Fig. 2B). The addition of C under light and dark conditions increased the N₂ fixation rates by 30- and 15-fold, respectively (Fig. 1F). However, significant N₂ fixation rates were measurable in darkness and anoxic conditions with high C availability, indicating that dark fermentation can support N₂ fixation (Fig. 1F). Although N promoted higher cell abundance, it hampered N₂ fixation (100-fold and 2-fold decrease under light and dark conditions, respectively; Fig. 1E and F), and the nitrogenase content was negatively correlated with N (n = 3, r = 0.98, P < 0.05, Fig. 2B). Hence, under anoxic conditions, photoheterotrophy supports N₂ fixation in BAL398, but the N₂ fixation activity is regulated by the level of N.

Section 2: influence of nutrient regimes, light, and O₂ on pigmentation and biofilm formation

The in vivo absorption spectrum of BAL398 cells under phototrophic conditions, i.e., light and 0 µM O₂, showed peaks associated with carotenoids (484 nM) and BChl a (806 and 874 nM) (Fig. 3A), and indeed, a high BChl a content was measured under these conditions (Fig. S4A). In contrast, when cells were grown under oxic conditions, the absorption spectrum was almost flat (Fig. 3A), and BChl a was undetectable (Fig. S4A). However, when cells settled at the bottom of a cultivation flask under stagnant conditions, the O₂ levels decreased (from 200 to 35 µM O₂, Fig. S4B), and BChl a became detectable (Fig. S4A). This prompted us to investigate the mechanisms in the water column by which cells can decrease O₂ levels, ultimately enabling BAL398 to fix N₂.

Fig 3.

Biofilm formation on different hydrophobic surfaces under varying environmental conditions. (A) In vivo absorption spectrum of BAL398 under anoxic and oxic conditions. (B) Normalized biofilm formation on polystyrene (PS) well plates at different light and nutrient regimes, under anoxic conditions. (C) Normalized biofilm formation measured on different surfaces, i.e., polyethylene (PE), polystyrene, and glass, under different light and O₂ availability. (D) O₂ profile onto the PS surface with and without cells. (E) N₂ fixation rates of BAL398 cells growing in media with PS under light and oxic conditions. Controls without PS were added. (F) Motility assay of BAL398 under standard conditions. Cells were inoculated with an inoculation loop from the bottom to the top of the tube. Positive motility was considered when cells diffused in the agar. In panels (C, D, and E) experiments were carried out under N₂-fixing conditions, i.e., 35 mM C and 0.15 mM N. The values represent the mean, and the error bars indicate the standard error between the replicates (n = 3). Different letters indicate pairwise analysis among the variables (P < 0.05), using a post hoc test (Wilcoxon) after Kruskal-Wallis over the entire data set. In (C) asterisks indicate pairwise differences between the PS and PE. All plots (A–F) are from independent experiments and were measured after 7 d. For more detailed information about the conditions tested, please refer to Table 1, Section 2.

Under different nutrient regimes, light/dark, and anoxic conditions, cell abundance was positively correlated with biofilm formation (Spearman’s rank-order correlation test, r = 0.85, n = 144, P < 0.05, Fig. S4C). N stimulated biofilm formation (when N was increased from 0 to 0.15 mM), while the effect of C was ambiguous (Fig. 3B). Light availability roughly doubled biofilm formation, irrespective of the nutrient level (Fig. 3B). Indeed, biofilm formation correlated with light intensity under N₂-fixing conditions (Spearman’s rank-order correlation test, r = 0.78, n = 7, P < 0.05, Fig. S4D).

Biofilm formation under oxic conditions was tested on surfaces with different chemical structures and polarities [determined using the water-substrate contact angle, (ϴ°), as a proxy]: glass (51 ϴ°), polystyrene (PS, 87 ϴ°), and polyethylene (PE, 101.7 ϴ°). Biofilm formation was higher on PS than on PE (Fig. 3C). Under N₂-fixing conditions and light, O₂ promoted biofilm formation on PS and PE by six- and twofold, respectively, while no biofilm formed on glass (Fig. 3C). Under oxic conditions and after 7 d, the O₂ concentration decreased from 270 to 150 µM, from the surface to the base of the PS biofilm (Fig. 3D), and N₂ fixation rates increased compared to the control treatment without PS (Fig. 3E). Since motility is conceivably a prerequisite for efficient surface colonization, this was examined for BAL398. A motility assay in soft agar suggested that O₂ values above 80 µM O₂ stimulated BAL398 motility (Fig. 3F). Thus, the formation of biofilm, which can be mediated by the increase in the motility, is a critical process that enhances the N₂ fixation rates of BAL398 by decreasing the levels of O₂.

Section 3: environmental regulation of cellular morphology and capsule formation in BAL398

We assessed cellular morphology during the process of biofilm formation on the PS surface. Remarkably, our observations revealed that the frequency of cellular rosette structures adhering to the plastic surface was up to threefold higher under oxic conditions compared to anoxic conditions (Fig. 4A through C). To further investigate this phenomenon, we conducted a detailed examination of the morphological changes that occurred in the free-living cells. Holotomographic microscopy revealed that the cellular morphology was highly dependent on the environment (Fig. 5A through F; Fig. S5A through I). Under N₂-fixing conditions with variable light and O₂, BAL398 was found both as single cells (Fig. 5A; Fig. S5A through C) and rosette-like structures (Fig. 5B; Fig. S5D through F). These rosette-like structures were about 600 times more prevalent under N₂ fixation conditions when cells were exposed to light and anoxic conditions (i.e., at the maximum N₂ fixation rates) than when exposed to anoxic and dark conditions or light and oxic conditions (i.e., at the lowest N₂ fixation rates) (Fig. 5C). However, high N concentrations inhibited rosette formation (Fig. 5C). Immunolabeling revealed that the nitrogenase was allocated toward the center of the rosette (Fig. 5B). Thus, the morphology of BAL398 changed according to the environmental conditions tested, which was accompanied by changes in diazotrophic activity.

Fig 4.

Cell attachment and rosette-like structures of BAL398 on polystyrene under (A) oxic and (B) anoxic and N₂-fixing and light conditions, using an epifluorescence microscope under 100× magnification. (C) Number of rosettes per square millimeter on PS depending on the light and O₂ availability. White arrows indicate the rosette-like structures. Different letters indicate pairwise analysis among the variables (P < 0.05), using a post hoc test (Wilcoxon) after Kruskal-Wallis over the entire data set. In (A through C) variables were measured after 3 d. For more detailed information about the conditions tested, please refer to Table 1, Section 2.

Fig 5.

Effect of light, O₂, and nutrients on cell structure and bacterial capsule in BAL398 under N₂-fixing conditions. (A) A rod-shaped BAL398 cell under anoxic conditions. (B) Rosette-like structures. Note that the green colors represent the nitrogenase detected by the immunolabeling approach. (C) Number of rosettes depending on the light, O₂, and N levels. (D) Cyst formation under aerobic and light and nitrogen-fixing conditions. (E) Rod-shaped BAL398 cells under oxic conditions. Note that blue, red, and gray colors represent the inner membrane, outer membrane, and capsule, respectively. The blue color in (D) represents possible large intracellular granules. (F) Length, width, and capsule thickness depending on the light and O₂ conditions. In panels (C and F) the values represent the mean, and the error bars indicate the standard error between the replicates (n = 3). Different letters indicate pairwise analysis among the variables (P < 0.05), using a post hoc test (Wilcoxon) after Kruskal-Wallis over the entire data set. Panels (A, B, D, and E) were taken by holotomographic microscopy. Panels (A through F) are from the same experiments, and variables were measured after 7 d. For more detailed information about the conditions tested, please refer to Table 1, Section 3.

BAL398 also produced cellular capsules, and their thickness was controlled by O₂ availability (Fig. 5A and E). Under oxic and N₂-fixing conditions, the capsule thickness and width increased significantly from 0.13 ± 0.02 µM to 0.15 ± 0.03 µM and from 0.77 ± 0.10 µM to 0.91 ± 0.11 µM, respectively (P < 0.05, Fig. 5A and E). In aerobic cultures, some cells formed a cyst-like structure, consisting typically of two rounded cells, sometimes forming long chains, containing large cytoplasmatic granules (Fig. 5D and F and Fig. S5G through I). A schematic representation of the differences between the described morphological shapes is shown in Fig. 5F. The layer surrounding the cyst was composed of polysaccharides, as evident from staining with alcian blue (Fig. S6A and B). Hence, BAL398 produces cellular capsules, and some cells form cyst-like structures in response to ambient O₂ levels.

DISCUSSION

Non-cyanobacterial heterotrophic diazotrophs in marine and fresh waters are often considered challenged by their need for carbon and energy from dissolved organic matter (6, 9). Herein, we describe the lifestyle of Rhodopseudomonas sp. BAL398, a photoheterotrophic bacterium isolated from low-salinity surface waters. Genomic and physiological characterization showed that this strain has a highly versatile metabolic repertoire capable of performing phototrophic, chemotrophic, and dark fermentation metabolism, utilizing N₂ fixation to sustain growth under light and low O₂ conditions. Our study reveals how N₂ fixation in BAL398 is influenced by light, O₂, and nutrient conditions (C and N), guided by its unique ecophysiology and plasticity in terms of cell aggregation and capsule production.

Light, O₂, and nutrients control diazotrophy in Rhodopseudomonas sp. BAL398

Rhodopseudomonas sp. BAL398 is an example of an NCD supporting growth and N₂ fixation by utilizing light under anoxic conditions. Dark fermentation can support N₂ fixation in BAL398, as described for other PNSB (16, 20), but N₂ fixation was stimulated 40-fold in the presence of light. In contrast, oxic conditions reduced N₂ fixation ≅ 400-fold, although it was not completely inhibited. In the presence of light and under low O₂ levels, extensive production of BChl a was observed, as also observed for other PNSB under such conditions (17, 18). Chromophores, such as flavins, were also detected, which may facilitate energy acquisition from light-driven electron transfer or catalysis of redox reactions, like N₂ fixation, within the cell (28). Furthermore, such chromophores can promote nitrogenase synthesis (29), which could be the reason the nitrogenase content was significantly higher under light conditions. Thus, elevated growth and N₂ fixation rates in BAL398 are dependent on simultaneous availability of light and low O₂ conditions.

Both C and N availability influenced N₂ fixation in BAL398. Here, photoheterotrophy may confer BAL398 with a selective advantage over heterotrophic diazotrophs that rely on energy from carbon sources to produce ATP, while photoheterotrophs can use light energy to produce ATP and convert organic compounds into biomass (30). Increased availability of N decreased the N₂ fixation rates of BAL398, accompanied by a decrease in the nitrogenase levels. We speculate that high levels of N inactivated the transcription factor nifA, resulting in decreased nif gene expression and N₂ fixation (31). These results show that the N₂ fixation of BAL398 is linked to its photoheterotrophy, but diazotrophy can be modulated by nitrogen availability.

Biofilm formation enables light-dependent N₂ fixation in BAL398

Free-living BAL398 cells colonized, proliferated, and formed biofilms on surfaces, depending on their polarity and chemistry. This ability stimulated N₂ fixation in BAL398 under oxic and C-rich conditions while also reducing the O₂ levels in the biofilm stimulating BChl synthesis and consequently phototrophic metabolism. Biofilm formation provides protection against stressors, e.g., by having lower O₂ levels due to a reduced diffusion of O₂ from outside to inside in the biofilm, accompanied by extensive microbial respiration (32). Some free-living rhizobacteria can convert carbon sources into exopolysaccharides, enabling biofilm formation and N₂ fixation (33). In BAL398, light and O₂ availability stimulated biofilm formation. Light may have increased the level of c-di-GMP (a signal molecule involved in the synthesis of the biofilm matrix), which in turn stimulated the formation of biofilms, similar to what has been reported for Rhodopseudomonas palustris (34).

Interestingly, motility and maximum biofilm formation for BAL398 were observed under oxic conditions. Similarly, the purple bacterium Marichromatium gracile increased its motility in response to >100 µM O₂ (35). We speculate that motility enables efficient surface detection and subsequent colonization by BAL398—a feature considered critical for the onset of biofilm formation for other bacteria (36, 37). Interestingly, it was recently shown that chemotaxis is a prevalent trait among marine heterotrophic diazotrophs in the water column conceivably allowing for efficient surface colonization and biofilm formation (38). For BAL398, this could be combined with negative aerotaxis, i.e., bacterial movement away from increasing O₂ concentration, as seen in Desulfovibrio (39). Although the exact mechanisms are unknown, our findings indicate that N₂ fixation is sustained by photoheterotrophy, and surface attachment of BAL398 may guide this metabolic process. However, the activity also appears to be facilitated by a remarkable capacity for morphological adaptation.

Cell morphology of BAL398 is controlled by light and O₂

The morphology of BAL398 is highly adaptable and is influenced by light, O₂ levels, and nutrient availability. For instance, the frequency of rosette-like structures increased 600-fold under light and anoxic conditions, which coincided with the maximum N₂ fixation rates. The rosettes, ranging from a few to hundreds of cells, showed compartmentalization of nitrogenase with a high concentration toward their center. Interestingly, under ambient O₂ conditions, where maximum biofilm formation was reached and N₂ fixation rates were higher compared to free-living cells, the proportion of rosette-like structures increased, indicating their potential role in biofilm formation, as previously found within the Rhodopseudomonas genus (40).

Compartmentalization of nitrogenase may be a mechanism to protect it from irreversible inactivation by O₂. In this context, our finding that O₂ stimulated the nitrogenase content, while N₂ fixation remained relatively low, may be seen as a need for continuous synthesis of nitrogenase as it is rapidly inactivated by O₂. Similarly, the soil bacterium Azotobacter vinelandii can maintain high nitrogenase content even at ambient O₂ levels (41). Rosette formation would consequently reduce energy costs for replenishment of nitrogenase and could in addition minimize the diffusive loss of NH₄ ⁺ generated by N₂ fixation (40). Hence, rosettes are likely critical structures for N₂ fixation by BAL398 under oxic conditions, which is also supported by their absence when N was replete. Rosette-like structures have also been observed in a cultivated marine diazotroph within the genus Sagittula (42). We speculate that such cell aggregation may be a more widely used mechanism to evade O₂ for NCDs, but we are not aware of reports of rosettes from in situ.

BAL398 also showed other morphological features, likely representing adaptations to cope with and survive under O₂ stress. As mentioned, N₂ fixation was not completely inhibited by O₂, and thicker cellular capsules were observed under oxic conditions. This may be an adaptation to avoid O₂ and maintain some N₂ fixation, as is known from Azotobacter vinelandii (43). Under oxic conditions and low nutrient availability, non-motile oval-shaped cells surrounded by polysaccharides appeared, i.e., so-called cysts. These cells are quiescent forms allowing survival under adverse environmental conditions (44, 45). To the best of our knowledge, Rhodospirillum centenum is the only PNSB that can form cysts (45), and it requires light to germinate (46). We speculate that cyst formation in BAL398 is a strategy that ultimately facilitates the survival of the population until favorable environmental conditions are encountered.

Ecology and diazotrophy of the photoheterotrophic Rhodopseudomonas sp. BAL398

We consider BAL398 an important member of the surface bacterioplankton as it was isolated from surface waters of the Baltic Sea at a station with a total depth of 459m, located 30km offshore, where it has been consistently found at the sampled station (25, 26). Furthermore, it is widespread in 16S rRNA gene amplicon libraries from Baltic Sea surface waters (Riedinger et al., unpublished data). The photoheterotrophic lifestyle could be a key adaptation that allows BAL398 to thrive in environments where other N₂-fixing bacteria struggle, yet O₂ and carbon will limit its activity. We speculate that aggregates in the water column would be suitable loci for N₂ fixation by BAL398. As free-living bacteria in the aerobic water column, BAL398 would be motile and non-pigmented, and the formation of cysts would ensure long-term survival. It is conceivable that it would be able to locate and colonize carbon-rich aggregates and, subsequently, form an extensive biofilm in this environment. Microbial respiration would exceed diffusion of O₂ from outside the aggregate, leading to the formation of microenvironments with low O₂ levels. Under these conditions, BAL398 could exploit light, facilitated by the synthesis of BChl a, and carry out efficient N₂ fixation when forming rosette-like cellular structures. The period where O₂ conditions are suitable for N₂ fixation would be governed by the amount of bioavailable carbon and nitrogen in the aggregate (12), but the ability to acquire energy from light would expand the period.

Depending on composition and size (47), marine aggregates can have high sinking velocities out of the photic zone (48). However, some particles are neutrally buoyant (49), and others even ascend to the sea surface (50). Hence, at least some aggregates would remain in the photic zone where BAL398 and other NCDs could use light for growth and N₂ fixation. In the upper meters of the water column, infrared light is available, whereas blue-green light penetrates deeper (51). These fractions of the light could be exploited by Bchl a and carotenoids, respectively, possessed by BAL398. The outlined scenario is a first attempt to unify the obtained results in a suggestive life strategy for BAL398. Future studies should investigate how BAL398 and other putative phototrophic NCDs carry out N₂ fixation in situ in the water column.

MATERIALS AND METHODS

Bacterial, genomic sequencing, and analysis

Rhodopseudomonas sp. BAL398 (nifH nucleotide accession number: KC140365) was originally isolated from 3 m depth at the Landsort Deep in the Baltic Sea [58°36 = N, 18°14 = E, March 2009 (25)]. Prior to the DNA extraction and experimental procedures, cultures were grown in filtered seawater (Durapore Membrane Filter, 0.22 µM, Sigma-Aldrich, Burlington, MA, USA) supplemented with 0.05% peptone and 0.1% yeast extract at 20°C, under anoxic conditions in constant light (OSRAM Parathom DIM PAR38 120 30° 15.2 W/2700K E27 lamp) under a photon irradiance (400–700 nM) of 250 µmol photons m⁻² s⁻¹, as measured with ULM-500 meter (WALZ, Effeltrich, DE) using a LI-COR Quantum Sensor (Cambridge, UK).

The genome of Rhodopseudomonas sp. BAL398 [old accession number: GCA_000935205.1, (27)] was re-sequenced using both short- (Illumina) and long-read (Nanopore) technologies: (i) a NextSeq 550 platform (Illumina) using the Nextera XT DNA Library Preparation Kit and v 2.5 Mid Output Sequencing Reagent Kit (300 cycles) and (ii) a GridION platform (Oxford Nanopore Technologies) using the Rapid Barcoding Sequencing Kit and a R10.4.1 flow-cell at Statens Serum Institute, Denmark. For short-read data, quality control of the raw fastq sequences was performed via a modified pipeline for WGS analysis, bifrost (https://github.com/ssi-dk/bifrost, last accessed 10 June 2022), based on the SKESA assembler (52). Long-read data were base-called with Guppy (v. 6.4.6) (53), and the quality of the fastq data was evaluated using NanoPlot and NanoStat (54). Three separate genome assemblies were generated: (i) Flye (v. 2.9.1) (55) followed by medaka (v. 1.9.1) error correction and short-read polishing using Polypolish (v. 0.5.0) (56) or (ii) hybrid and (iii) long-read-only assembly using SPAdes (v. 3.15.3) (57), and all assemblies were assessed for completeness and contamination using CheckM (v. 1.0.18) (58). The quality of each assembly was analyzed using Bandage (v. 0.8.1) (59). Genomic analysis was performed using the polished Flye assembly. The genome was annotated using Bakta (v. 1.7.0) (60) and visualized using G (v. 1.7) (61).

The closest available 16S rRNA, pufM, and genome sequences of BAL398 were retrieved from NCBI (September 2022). Comparative 16S rRNA and pufM gene sequence analyses were done using the maximum composite likelihood to calculate evolutionary distances. Cluster analyses and phylogenetic trees were built by neighbor-joining using the MEGA7 software. Bootstrap values were determined for 500 replications. Maximum likelihood and maximum parsimony trees were inferred from the alignment. For the genomic analyses, average nucleotide identities by BLAST (ANIb) were determined using the JSpeciesWS tool (62). The dDDH similarity values were determined with the genome-to-genome distance calculator (63). In addition, to evaluate similarities in the BAL398 NifH protein with other bacteria, the closest NifH amino acidic sequences were downloaded from Uniprot (64), aligned with MUSCLE (65), and processed as above.

Experimental culture conditions

All the experiments were carried out in triplicate (n = 3), using a modified photosynthetic medium (MPM; Na₂HPO₄ (12.5 mM), KH₂PO₄ (12.5 mM), MgSO₄ × 7H₂O (0.24 mM), CaCl₂ × 2H₂O (0.045), Na₂S₂O₃ × 5H₂O (0.1 mM), KBr (0.13 mM), MgCL₂ × 6H₂O (9.39 mM), following reference (66)), and amended with 10 g NaCl, 1 mL vitamins (1 g L⁻¹ p-aminobenzoic acid, 1 g L⁻¹ thiamine, and 0.1 g L⁻¹ biotin), and 2 mL trace-element solution per liter medium (25). Axenic cultures in the exponential phase ≅ 1 optical density (O.D.)] were washed with MPM by centrifugation (5 min, 6,000 × g) to remove any traces of carbon and nitrogen from the growth medium and were inoculated into either 6-well (8 mL MPM), 24-well plates (2 mL MPM) (Sigma-Aldrich, Burlington, MA, USA), or 20/50 mL glass vials (8/25 mL MPM) at an O.D. of 0.1–0.2. The cultures were then exposed to the experimental manipulation for 7–10 d at 20°C (Table 1). Organic carbon (C, 1:1, glucose:succinate) and/or inorganic nitrogen (N, NH₄Cl) were added according to the treatment (Table 1). Table 1 specifies whether the cultures were incubated under static or shaken (Labline 4635 shaker, rocking motion at a constant 10 oscillations per minute) conditions. Cultures were handled aseptically. Parallel incubation of medium blanks (without cells added) confirmed that there was never any microbial contamination in the conducted experiments.

BAL398 cells were cultured at 0, 100, and 250 µM O₂ (ambient) under light or dark conditions (Table 1). We also tested a light gradient from 0 to 250 µmol photons m⁻² s⁻¹, by shading with nylon mesh layers. Aluminum foil was used to create darkness. Desired O₂ levels were achieved by incubation in gas-tight sealed bags (Anaerocult A and C mini, Sigma-Aldrich, Burlington, MA, USA) with an anoxic (0 µM O₂) or low O₂ (100 µM O₂) environment, or by flushing with pure N₂ gas in the case of glass vials (capped with butyl rubber stoppers and crimp sealed). Complete anoxia was confirmed by adding 1 mg L⁻¹ of resazurin as an O₂ indicator, which is colorless under anoxic conditions (67). Cells were incubated under three organic C levels combined with three N levels (see Table 1). Unless stated otherwise, optimal and N₂-fixing conditions were defined as 35 mM C and 7.5 mM N, and 35 mM C and 0.15 mM N—based on an earlier study with BAL398 (27).

The effect of C, N, and light availability on biofilm formation was examined in 24-well polystyrene plates under light/dark and anoxic conditions after 3 and 7 d (Table 1). BAL398 colonization was tested on surfaces of different polarity [water-substrate contact angle (ϴ°)] and chemical structure: glass (51 ϴ°), polystyrene (PS, 87 ϴ°), and polyethylene (PE, 101.7 ϴ°). For these tests, triplicate pieces (2 cm²) of neat plastic (Goodfellow, Lille, FR) or glass were UV sterilized in the flow bench for 20 min and placed into the 6-well plates with MPM and BAL398 inoculum under different light and O₂ availabilities (Table 1).

Growth, O₂ concentration profiles, bacteriochlorophyll a (BChl a), N₂ fixation rates, nitrogenase content, biofilm formation, cell plasticity, and extracellular polysaccharides were measured as response variables after 7 d unless stated otherwise.

Growth and physiological measurements

Bacterial growth was followed by measuring the O.D. at 590 nM using a spectrophotometer (FLUOstar OPTIMA, Ortenberg, DE) every day (for the growth curves) or after 7 d (for the endpoint analysis). For the normalization of the N₂ fixation rates and nitrogenase content, O.D. was converted to cell counts using the linear relation (r = 0.99): cells (mL⁻¹) = (O.D./8 × 10⁻¹⁰) − 0.0285, which was established via comparing O.D. measurements with cell enumeration using Sybr Green and flow cytometry, as done earlier (27). Growth rate (μ, d) and generation time (Tg, d) were calculated during the exponential phase, i.e., between days 2 and 6:

$${\displaystyle \mu =\frac{\mathit{L}\mathit{n}\left(\mathrm{O}.\mathrm{D}.6\right)-\mathit{L}\mathit{n}\left(\mathrm{O}.\mathrm{D}.2\right)}{\mathit{T}\mathit{6}\mathit{-}\mathit{T}\mathit{2}}}$$ (1)

$${\displaystyle Tg=\frac{Ln2}{\mu }}$$ (2)

where O.D.6 is the O.D. at day 6 (T6), and O.D.2 is the O.D. at day 2 (T2).

The O₂ concentration in liquid media was measured with a retractable fiber-optic O₂ microsensor (OXR50; Pyroscience, Aachen, DE) connected to a FirestingPro meter (accuracy ±0.31 µM O₂, Pyroscience, Aachen, DE). Measurements of O₂ concentration profiles over the PS surface were done with an electrochemical microsensor (OX25; Unisense A/S, Aarhus, DK) connected to a multichannel microsensor meter (accuracy ±0.3 µM O₂, Unisense A/S, Aarhus, DK) and mounted on a motorized micromanipulator system. The O₂ microsensors were linearly calibrated by recording measuring sensor signals, i.e., 100% air-saturated seawater and anoxic water (using an alkaline ascorbate solution). The surface position (z = 0) of the PS material was determined by manually positioning the microsensor while watching it under a stereo microscope. The O₂ consumption rate was calculated as the local flux of O₂ according to Fick’s first law of diffusion:

$${\displaystyle J=-D\cdot \frac{dC}{dz}}$$ (3)

where dC/dz is the slope of the linear O₂ concentration gradient in the diffusive boundary layer above the sample and D (= 1.93 × 10⁻⁵ cm² s⁻¹) is the diffusion coefficient of O₂ in seawater at the experimental temperature (20°C) and salinity (10). All O₂ measurements were done in triplicate.

Motility was assessed by inoculating BAL398 cells into 10 mL soft agar [0.4% (wt/vol)] in filtered seawater supplemented with 0.05% peptone and 0.1% yeast extract agar, from the bottom to the top in 15 mL glass tubes. After 14 d, motile cells could be observed as they spread within the agar. Vertical O₂ concentration profiles in the tube were measured as in liquid media.

BChl a was quantified after 7 d, following reference (68). Briefly, BChl a was extracted from cell pellets (previously centrifuged, 6,000 × g, 5 min) with 1 mL of acetone: methanol (7:2 vol/vol) and measured at 775 nm using an extinction coefficient value of 75 mol⁻¹ m³ cm⁻¹. The in vivo absorption spectrum from 300 to 900 nm was recorded using a visible scanning spectrophotometer (Shimadzu UV 1800, Kyoto, JP). Fluorescence was measured using a spectrophotometer (FLUOstar OPTIMA, Ortenberg, DE) in cell-free supernatant at 525/440 nm emission/excitation, a characteristic signal of flavins (69).

Nitrogen fixation

The acetylene reduction assay (ARA) was used as a proxy for N₂ fixation following a modified protocol of reference (70) and performed in 20 mL gas-tight glass vials containing 8 mL of culture under 20% (vol/vol) acetylene (purity >99.6%; AlphaGaz, Taastrup, DK). Prior to the ARA, the PM medium was saturated with acetylene by flushing with pure acetylene gas. After 7 d, 2 mL of PM-acetylene was added to the samples. All vials, including the oxic treatments, were capped with butyl rubber stoppers and crimp sealed. The vials were incubated for 6 h with gentle shaking in the Labline 4635 shaker (10 oscillations per minute) to ensure proper mixing of saturated acetylene and sample. The incubation was terminated with the injection of trichloroacetic acid [final conc. 2% (wt/vol)], and vials were kept at 4°C until gas analysis. Before measurements, the vials were incubated overnight under 37°C. Ethylene and acetylene were determined using a gas chromatograph (GC-2010, Shimadzu, Tokyo, JP), equipped with a flame ionization detector and a portepak T SUS column (Shimadzu, Tokyo, JP). Ethylene production was calculated according to reference (71) using ethylene (0.5% balance with nitrogen, CALGAZ, Staffordshire, UK) and acetylene standard curves. The detection limit was 1.78 nM. Pure acetylene was used as an internal standard in the calculations, bypassing potential inaccuracy due to gas loss during incubations. Samples without cells served as blanks.

Nitrogenase immunolabeling

Nitrogenase levels were quantified by immunolabeling bacteria according to reference (72). Briefly, after 7 d, a 2-mL bacterial culture of known O.D. was centrifuged (6,000 × g, 5 min), resuspended in 1 mL phosphate buffer saline (PBS), and fixed and permeabilized with 700 µL of 1% (vol/vol) glutaraldehyde (5 min) and 0.1% (vol/vol) triton X-100 in PBS (15 min), respectively. Cells were then incubated for 60 min with 2% (wt/vol) bovine serum albumin (blocking agent) and discarded by centrifugation (2,000 × g, 5 min). Also, 500 µL of the primary antibody was added (6 µg mL⁻¹, Anti-NifH:AS01 021A, Agrisera, Vännäs, SE) and incubated overnight at 4°C. After washing the primary antibody-stained cells with PBS (2,000 × g, 5 min), 500 µL of secondary antibody (goat anti-chicken IgY, H + L, Cross-Adsorbed Secondary Antibody, dyLight 488, Thermo Fisher Scientific, Massachusetts, USA) was added in a 1:800 proportion (BSA:PBS), and the samples were incubated for 45 min. After washing with PBS via centrifugation (2,000 × g, 5 min), the pellet was resuspended in 300 µL of PBS and transferred to a black fluorometry 96-well plate. The fluorescence of the secondary antibody bound to the primary one was quantified (480 nM excitation, 520 nM emission) using a spectrophotometer (FLUOstar OPTIMA, Ortenberg, DE). Three blanks were used to ensure that the fluorescence obtained was from the conjugation of the primary and secondary antibody together with the nitrogenase protein: (i) PBS only, (ii) PBS and primary antibody, and (iii) PBS and secondary antibody. Results were expressed in arbitrary units per cell.

Biofilm quantification

After 7 d, cells cultured in the 6- or 24-well plates were shaken in a KS 4000 i control (IKA, Staufen, DE) at 150 r.p.m. for 3 min, and crystal violet (CV) was added to a final concentration of 0.02% (vol/vol), following reference (40). In treatments involving plastic/glass pieces, these “surfaces” were transferred to a fresh 6-well plate, covered with sterile PM, and subjected to the same procedure as above. The samples were then incubated statically for 15 min, followed by three Milli-Q washes to eliminate excess CV. Next, 10% (vol/vol) acetic acid was added, and the dissolved CV was measured at 570 nM using a spectrophotometer (FLUOstar OPTIMA, Ortenberg, DE). Wells with/without plastic containing PM but without cells (treated as above) served as blanks. Results are presented as relative biofilm formation normalized to O.D. (A₅₇₀/O.D.₅₉₀). The early biofilm formation, i.e., after 3 d, and the formation of rosette-like structures were examined on PS surfaces (Table 1). The plastic pieces, previously washed with PM without C and N, were placed in a 2-mL tube containing PM and Sybr Green solution for 30 min. Cells were then inspected on an Olympus BX50 epifluorescence microscope (Olympus, Tokyo, JP) under 100× magnification. A minimum of 12 fields were considered, and the number of rosette-like structures was counted in each field, ranging from 0 to 30. Subsequently, the values were converted to the number of rosettes per square millimeter of the PS surface.

Cell plasticity

Cell sizes and morphology were examined after 7 d using a digital holotomographic microscope (DHM; HT-2, Tomocube Inc., Daejeon, South Korea). DHM employs optical diffraction tomography, where multiple 2D holographic images of a sample are measured to quantify the phase shift of light at various illumination angles followed by a calculation of 3D tomograms of refractive index (RI) via inversely solving the measured scattering in the sample (73, 74); see Fig. S7A through F. The images were obtained by mounting 30 µL of bacterial culture in a Tomodish (Tomocube Inc., Daejeon, South Korea) covered with a coverslip. The tomographic imaging data were visualized and analyzed using TomoStudio and TomoAnalysis software (Tomocube Inc., Daejeon, South Korea), which enabled segmentation of the data sets to highlight specific cell structures (e.g., cell wall, capsule, and cyst size) via their specific RI signatures. Using the built-in fluorescence capability of the Tomocube 2HT system, we could combine 3D RI imaging with imaging of the nitrogenase in antibody-stained cells via excitation at 460 nm and emission of the antibody fluorescence at 510 nm. More than 30 cells for each treatment were imaged and analyzed, and the resulting images were exported to Image J software (75) for further analysis and preparation of figures.

Rosette-like cells were counted using a CKX53 inverted microscope (Olympus, Tokyo, JP) and counting chambers (Graticules Optics, Tonbridge, UK). To visualize rosette-like cells and cysts, the cultures were fixed using a 15% Lugol’s iodine solution after 7 d. A total of 12 fields were evaluated, ranging from those with no rosettes to those with over 300 rosettes per field at 100× magnification.

Staining of extracellular polysaccharides

To evaluate the nature of the cyst capsules, they were stained with alcian blue 8 GX (Sigma-Aldrich, Burlington, MA, USA) ((76). One milliliter of culture was filtered onto a black 0.2-µM polycarbonate filter (Merck Millipore, Massachusetts, USA) and stained with 0.02% alcian blue 8 GX diluted in 0.06% (vol/vol) acid acetic. The filter was washed with MilliQ water to remove excess dye before inspection on an Olympus BX50 epifluorescence microscope (Olympus, Tokyo, JP) under 100× and 400× magnifications.

Statistical analyses

Kruskal-Wallis non-parametric tests were used for non-normally distributed data, and an unpaired two-sample Wilcoxon test was used to determine the significance of differences among the variables tested. Spearman’s rank-order correlation tests were performed between cell abundance vs light intensity, cell abundance vs biofilm formation, and biofilm formation vs light intensity. The statistical significance level for all the analyses was set at P < 0.05. All analyses were done in R-Studio, R v. 4.1.0.

DATA AVAILABILITY

Genome sequencing data have been made available in NCBI (https://www.ncbi.nlm.nih.gov; accession numbers CP133111–CP133114).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01027-23.

Supplemental material. aem.01027-23-s0001.pdf.

Fig. S1 to S7.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.