Growth-stage-related shifts in diatom endometabolome composition set the stage for bacterial heterotrophy

Co-culture dynamics

This study was designed to enhance understanding of metabolite release and utilization across bloom stages in a simple community of phytoplankton and heterotrophic bacteria. The synthetic community was established with the diatom T. pseudonana and the bacterial strains R. pomeroyi DSS-3, Stenotrophomonas sp. SKA14, and P. dokdonensis MED152. These bacterial strains have high genetic similarity to isolates from phytoplankton cultures [14] and represent taxa that are common in phytoplankton blooms. Metabolites derived from the diatom were the sole source of carbon available for the bacteria, since no organic substrates were added. In addition, none of the bacteria can assimilate nitrate, and usable nitrogen was only available as diatom or bacterial extracellular products. The diatom had its highest specific growth rate of 1.65 d⁻¹ during days 0–3, after which the rate declined (Fig. 1A). The total abundance of heterotrophic bacteria increased steadily but there was a succession that favored P. dokdonensis through day 15, and then R. pomeroyi by day 20; Stenotrophomonas disappeared from the model system by day 3 (Fig. 1B). The presence of bacteria did not affect the growth of diatoms based on comparisons of abundance in co-cultures versus axenic cultures at day 15 (Fig. 1A), as has been found previously [14, 26]. Inorganic nutrients were not limiting (>5 μM at day 15; Table S1).

Diatom endometabolite shifts

Analyses focused on the day 3 (early bloom) and day 15 (late bloom) co-culture time points, for which a complete set of metabolomic and transcriptomic data were collected. Twenty-two diatom endometabolites that were annotated with high confidence by NMR analysis (Table S2) and quantified after normalizing to diatom cell number revealed that endometabolome composition differed substantially between bloom stages. Metabolites with significantly different cellular concentrations included nine compounds that were higher in intracellular concentration during the late bloom; these were arginine, valine, lysine, DHPS, glycerol-3-phosphate, phosphorylcholine, DMSP, glycine betaine, and homarine (T-test; P < 0.05, n = 3; Fig. 2A; Table S3), of which the last three are known to function as osmolytes [50, 51]. Elevated internal DMSP concentrations in phytoplankton cells have been linked to nitrogen, CO₂, silicate, and phosphate limitation, increasing during stationary growth phase and potentially replacing nitrogen-rich osmolytes or serving as an antioxidant under stressful conditions such as CO₂ limitation and low temperature [52]. All three osmolytes have also been identified in endometabolomes of natural plankton communities from surface seawater [53]. Eight metabolites were significantly lower in intracellular concentration in the late bloom co-cultures; these were proline, glutamate, glycine, β-1,3-glucan, aspartate, glucose, guanosine, and uridine (Fig. 2B). Five metabolites did not change between early and late bloom stages; these were alanine, leucine, isoleucine, glutamine, and acetate (Fig. 2C).

Differences in diatom endometabolome composition were also evident in comparisons of the 15 d cultures with and without a bacterial community. Six diatom endometabolites had accumulated to higher concentrations in the presence of bacteria compared to axenic cultures by day 15; these were glutamate, arginine, leucine, DHPS, glucose, and glycerol-3-phosphate; none accumulated to higher concentrations in axenic cultures (Fig. 2), similar to previous analyses [17]. Alteration of phytoplankton physiology in the presence of bacteria has been reported previously and attributed to processes such as bacterial remineralization of ammonium from dissolved organic matter [54] and release of vitamins [13] or hormones [3]. In this study, phytoplankton were both nutrient and vitamin replete and there was no evidence that co-culturing with bacteria enhanced growth. Nonetheless, the diatom cells accumulated endometabolites differently depending on the presence or absence of bacteria.

Diatom gene expression

Diatom gene expression provided insights into physiological changes associated with growth stage. Relative gene expression fell into two distinct expression clusters consisting of early (days 0, 3) and late (days 8, 15, 20) sample times (Fig. S3). Comparing the day 3 and day 15 time points, 6 637 of the 11 675 predicted genes in the T. pseudonana genome (59%) had significantly different relative contributions to the transcriptome. The diatom’s early bloom transcriptome was highly enriched in transcripts for CO₂ acquisition via carbonic anhydrases (Fig. 3A). Transcripts for synthesis of glycolysis products pyruvate, acetate, and acetyl-CoA (Fig. 3B) and for channeling acetyl-CoA into the tricarboxylic acid (TCA) cycle were also enriched, as were transcripts for multiple central TCA cycle steps from oxaloacetate to succinate (Fig. 3C). Transcription patterns indicated that recently fixed carbon was directed toward chitin precursor N-acetyl-D-glucosamine (GlcNAc) and chrysolaminarin backbone β-1,3-glucan (Table S4). These transcriptome features were consistent with the higher concentrations of glucose and β-1,3-glucan in the diatom endometabolome on day 3. Overall, this gene expression pattern indicated an emphasis on carbon fixation and biomass building in the early bloom (Fig. 2) and is consistent with the diatom’s highest specific growth rate occurring during days 0–3 (Fig. 1).

Early-stage transcript enrichment was also observed for genes mediating nitrate assimilation and conversion to ammonium. Glutamate is a central component of biosynthesis pathways for amino acids and the assimilation pathway for ammonium, which serves as the nitrogen homeostasis mechanism. Compared to late phase T. pseudonana cells, glutamate synthesis genes were enriched in both the cytosol and mitochondrion (Fig. 3A, B and Table S4). Indeed, higher internal endometabolome concentrations of glutamate as well as other nitrogen-rich metabolites (amino acids proline, glycine, and aspartate; and nucleosides guanosine and uridine) were evident in the early growth stage diatoms (Fig. 2). T. pseudonana’s urea cycle coordinates cellular nitrogen and carbon status [55], and several genes in this cycle were enriched during early growth (Fig. 3D). These patterns are consistent with peak concentrations of nitrogen-containing amino acids and nucleosides in the early bloom, and together indicate higher nitrogen requirements by T. pseudonana during the biomass building phase.

Diatom transcripts in the late bloom phase were enriched instead with genes that synthesize malate and oxaloacetate, two metabolites of the C₄ delivery pathway for CO₂ that performs well under low CO₂ concentrations (Fig. 3A). Whether C₄ metabolism is functional in diatoms is controversial (e.g., refs [56,57,58]) but if so, it would be beneficial in cases where inorganic carbon concentrations decrease in late phase blooms. Late-phase enriched transcripts were also found in pathways for putrescine and spermidine synthesis. These polyamines have several known functions in diatom cells, among them a role in stress response [59]. In field studies, seawater concentrations of polyamines increased during diatom bloom decline [60]. Evidence for physiological stress on late bloom diatoms also included increased expression of genes for synthesis of the osmolyte DMSP (Fig. 3); the diatom pathways for synthesizing osmolytes glycine betaine and homarine are unknown.

Genes related to sulfate assimilation and subsequent synthesis of the organic sulfur compounds cysteate, sulfopyruvate, sulfolactate, and DHPS, in addition to DMSP mentioned above, were highly enriched in the later stage cultures, with up to 100-fold increases in relative expression (Fig. 3B and Table S4). These data are congruent with higher DMSP and DHPS endometabolite concentrations in late- compared to early-stage diatom cells (Fig. 2). The temporal switch from greater investments in synthesis of nitrogen-rich compounds in the early-stage bloom to organic sulfur compounds in the late stage could reflect the higher energy costs for nitrate assimilation compared to sulfate assimilation (41 versus 33 ATP per mol assimilated; [61]). Similar to nitrate transformation to nitrite and ammonium, sulfate transformation to cysteate also consumes NADPH, and thus removes excess reductants produced during photosynthesis [62]. Genes for cysteate catabolism, believed to be an intermediate in DHPS formation [50], were significantly enriched during the late growth phase. Relatively higher sulfate usage as compared to nitrogen in aging diatom cultures or during bloom senescence is generally associated with nitrate limitation [62], but here we observed this even under nitrate replete conditions, suggesting a physiological response by the cells as their growth rate slowed. Giordano and Raven [61] suggest that low early ocean sulfate concentrations may have influenced the evolution of marine phytoplankton, manifested in divergent present-day strategies for regulation of nitrogen versus sulfur metabolism.

Bacterial gene expression

The early versus late bloom differences evident in the diatom metabolome and transcriptome set the stage for responses by associated heterotrophic bacteria in their substrate acquisition patterns. The substrates available to the co-cultured bacteria were inferred from significant differences in relative expression of genes diagnostic for uptake of organic matter (i.e., those mediating transport or initial catabolism of exogenous molecules). We focus on R. pomeroyi and P. dokdonensis in this analysis as Stenotrophomonas sp. was not present in the co-cultures at day 15 (Fig. 1B).

Expression patterns of R. pomeroyi genes suggested higher relative availability of taurine, glycerol-3-phosphate, lactate, DHPS, putrescine, alanine, and glycine betaine in the early bloom; and of N-acetyltaurine, glycine, choline, spermidine, glycolate, trimethylamine, and sugars during late bloom (Tables 1 and S5). Transcription patterns suggested that DMSP was available at both time points but degraded by different pathways in the early (demethylation pathway) versus late (cleavage pathway) bloom [63], a temporal switch in pathway dominance consistent with previous studies in cultures [26, 64] and in a natural phytoplankton bloom [65], and potentially driven by differences in reactive oxygen stress between pathways [65]. Enriched bacterial gene expression for spermidine transport coincided with enriched diatom gene expression for its biosynthesis. Overall, however, there were few positive relationships between shifts in R. pomeroyi gene expression for uptake of a metabolite and shifts in phytoplankton intracellular concentration or biosynthetic gene expression for that same metabolite. For example, of eleven metabolites found in the diatom endometabolome and for which the R. pomeroyi transporters are known, we found that glycine betaine, DHPS, and glycerol-3-phosphate had higher concentrations in phytoplankton cells on day 15 but their transporters were enriched in the bacterial transcriptome on day 3; glycine had higher concentrations in phytoplankton cells on day 3 but the transporter was enriched on day 15; alanine and acetate did not differ in concentration between early versus late bloom phytoplankton cells but their transporters were enriched on day 3 and 15, respectively; and proline, aspartate, glucose, glutamate, and phosphorylcholine were either more abundant in day 3 or day 15 phytoplankton cells but their transporter expression in R. pomeroyi was not different. Assuming bacterial transporter expression is induced primarily by substrate availability [66], these mismatches argue for a minor role for passive leakage in metabolite release and instead support active release mechanisms that cannot be predicted from endometabolite concentrations. An additional ten metabolites detected in the phytoplankton metabolome do not have confirmed transporters in the R. pomeroyi genome: lysine, glutamine, arginine, valine, leucine, isoleucine, homarine, guanosine, uridine, and β-1,3-glucan.

Flavobacteriia member P. dokdonensis increased in abundance through time in the co-cultures, as has been observed for marine flavobacteria in natural diatom bloom progression [19]. This pattern has been attributed to specialization by this taxon for glycan utilization, release of which increases in aging phytoplankton cells [67]. Gene expression patterns by P. dokdonensis were also consistent with increasing glycan importance in the late bloom, particularly for genes active in polysaccharide utilization loci (PULs) [68]. These genomic regions enable hydrolysis of polysaccharides to monomers, which are transported into the periplasm and subsequently transported across the cell membrane. Several genes within PUL3 (as designated in the CAZy database; Taxonomy ID 313598 [14, 69]) were significantly enriched during the late growth phase, including two annotated as fucose permeases (Table 1 and S6). Polaribacter species were shown to peak concurrently with fucosidase activity in a natural bloom [19], indicating an association with fucose degradation within the genus. Genes in PUL5 hypothesized to transport and degrade chitin subunits [14] exhibited up to 40-fold enrichment in late bloom P. dokdonensis transcriptomes. Chitin is a common organic molecule in the ocean [70,71,72], and serves as component of cell walls and a locus for bacterial attachment in diatoms [73, 74].

In addition to utilization of structural polysaccharides, gene expression in P. dokdonensis PUL2 suggested utilization of the storage polysaccharide chrysolaminarin ([14]; Table 1). This parallels the high transcriptional signal and endometabolite concentration for the β-1,3-glucan building block of chrysolaminarin in early-stage T. pseudonana (Fig. 3B and Table S4) [75, 76]. Phytoplankton were recently shown to exhibit daily cycles of internal chrysolaminarin concentrations synchronized with diel light patterns [17, 76], with this storage polysaccharide comprising up to 80% of diatom carbon under certain conditions [75]. The utilization of both structural and storage polysaccharides represented a distinct niche for P. dokdonensis in the T. pseudonana co-cultures. As found previously [14], R. pomeroyi and P. dokdonensis transcriptomes indicated little overlap in resource use, with each species responding to different suites of organic compounds in early and late bloom stages.

There was temporal matching of peaks in diatom endometabolite concentrations and P. dokdonensis gene expression for early bloom peaks of proline, glucose, and β-1,3-glucan. Of these, proline is a candidate for passive diffusion because of its small size, but the higher molecular weight metabolites glucose and β-1,3-glucan are less likely to diffuse through the diatom membrane, suggesting instead a link to export for physiological balance [18, 23, 25, 77]. Previous studies have found glucose release from diatoms when carbon fixation rates are high [78], likely due to photosynthetic overflow pathways. Yet the poor match between peaks in diatom endometabolites and peaks in both P. dokdonensis and R. pomeroyi transcription indicate that endometabolite abundance cannot predict the release of metabolites into surrounding seawater (Table S2), and argues for a key role for active (i.e., non-diffusive) mechanisms of diatom metabolite release. Also of interest are the six metabolites (arginine, DHPS, glycerol-3-phosphate, glutamate, glucose, and leucine) with higher concentrations in phytoplankton cells in co-culture compared to axenic culture (Fig. 2). Although the mechanism is unclear, bacteria appeared to directly or indirectly affect the internal concentrations of these metabolites.

Community vs. individual bacterial gene expression

These same three bacterial species were co-cultured individually with T. pseudonana in a prior study [14], providing the opportunity to explore gene expression as a member of a bacterial community versus when alone. Several conditions differed between the studies; in Ferrer-González et al. [14], exometabolites built up in axenic T. pseudonana cultures for 7 days prior to inoculation of the bacteria, followed by sampling after 8 h. In the present study, the diatom and bacteria were inoculated at the same time, with sampling after 3 d. In both studies, however, the diatom was in exponential growth at the time of bacterial sampling. Significantly enriched bacterial genes were identified based on a reference dataset collected from bacteria growing in a glucose medium. Of the 24 enriched transporters in R. pomeroyi, 12 were enriched only in the individual co-cultures, including those with experimentally-verified substrates N-acetyltaurine and urea; and four were enriched only in the community co-cultures, including those with experimentally-verified substrates choline, acetate, and DMSP (Table S7). There was no difference in distribution of the three dominant compound classes represented by the transporters (organic acids, amino acids, and organic sulfur compounds), nor in the distribution of nitrogenous compound transporters (Table S7). For P. dokdonensis, the enrichment of genes in PULs 2, 3, 5, 6, and 7 seen in this study matched the Ferrer-González study [14]. However, PUL4 (no annotation available) had only one enriched gene matching and PUL8 (annotated for siderophore uptake and potentially colicin toxins [79]) had none (Table S8). Because the study designs differed in factors other than just the number of bacterial species, these comparisons are viewed as preliminary. Nonetheless, they suggest differences in uptake when bacteria are members of a community, with potential mechanisms including altered metabolite release by the diatom, or antagonism or competition for resources by the bacteria.

Source: Ecology - nature.com