A bloom of a single bacterium shapes the microbiome during outdoor diatom cultivation collapse

Abstract

Algae-dominated ecosystems are fundamentally influenced by their microbiome. We lack information on the identity and function of bacteria that specialize in consuming algal-derived dissolved organic matter in high algal density ecosystems such as outdoor algal ponds used for biofuel production. Here, we describe the metagenomic and metaproteomic signatures of a single bacterial strain that bloomed during a population-wide crash of the diatom, Phaeodactylum tricornutum, grown in outdoor ponds. 16S rRNA gene data indicated that a single Kordia sp. strain (family Flavobacteriaceae) contributed up to 93% of the bacterial community during P. tricornutum demise. Kordia sp. expressed proteins linked to microbial antagonism and biopolymer breakdown, which likely contributed to its dominance over other microbial taxa during diatom demise. Analysis of accompanying downstream microbiota (primarily of the Rhodobacteraceae family) provided evidence that cross-feeding may be a pathway supporting microbial diversity during diatom demise. In situ and laboratory data with a different strain suggested that Kordia was a primary degrader of biopolymers during algal demise, and co-occurring Rhodobacteraceae exploited degradation molecules for carbon. An analysis of 30 Rhodobacteraceae metagenome assembled genomes suggested that algal pond Rhodobacteraceae commonly harbored pathways to use diverse carbon and energy sources, including carbon monoxide, which may have contributed to the prevalence of this taxonomic group within the ponds. These observations further constrain the roles of functionally distinct heterotrophic bacteria in algal microbiomes, demonstrating how a single dominant bacterium, specialized in processing senescing or dead algal biomass, shapes the microbial community of outdoor algal biofuel ponds.IMPORTANCEAquatic biogeochemical cycles are dictated by the activity of diverse microbes inhabiting the algal microbiome. Outdoor biofuel ponds provide a setting analogous to aquatic algal blooms, where monocultures of fast-growing algae reach high cellular densities. Information on the microbial ecology of this setting is lacking, and so we employed metagenomics and metaproteomics to understand the metabolic roles of bacteria present within four replicated outdoor ponds inoculated with the diatom Phaeodactylum tricornutum. Unexpectedly, after 29 days of cultivation, all four ponds crashed concurrently with a "bloom" of a single taxon assigned to the Kordia bacterial genus. We assessed how this dominant taxon influenced the chemical and microbial fate of the ponds following the crash, with the hypothesis that it was primarily responsible for processing senescent/dead algal biomass and providing the surrounding microbiota with carbon. Overall, these findings provide insight into the roles of microbes specialized in processing algal organic matter and enhance our understanding of biofuel pond microbial ecology.

Keywords: Kordia; biofuel ponds; cross-feeding; metagenomics; metaproteomics.

Fig 1

Algal host strain and decline in P. tricornutum over time influence microbiome composition within the algal raceway ponds. (A) Relative abundance of representative host 18S rRNA ASVs between ponds inoculated with M. salina versus P. tricornutum. Error bars represent the standard deviation in relative abundance between four replicate ponds. 18S rRNA amplicons assigned to either Phaeodactylum or Microchloropsis are shown in either brown or green, where all other 18S rRNA amplicons are shown as “Other” in gray. (B) NMDS ordination based on the Bray-Curtis dissimilarity matrix of 16S rRNA bacterial communities between samples inoculated with M. salina (circles) and P. tricornutum (triangles) over timepoint (indicated by color); 95% confidence ellipses (stat_ellipse) are grouped by algal host inoculum. Stress = 0.115.

Fig 2

Shift in microbiome composition during P. tricornutum demise corresponds to a bloom of an individual ASV assigned to the genus, Kordia, and co-occurring taxa. Each pond replicate is shown on the x-axis, and individual ASVs (shown with ASV number and lowest taxonomic assignment) are on the y-axis. The timepoints are grouped and color-coded by algal growth phase, which is based on NMDS clustering of the relative abundance of ASVs in Fig. S4 (growth [purple line] versus demise [orange line]). Each bubble represents the relative abundance value for each replicate, and the averaged relative abundance across all the plotted samples for each ASV is shown on the bar graph to the right. Bubbles are color-coded by the taxonomic assignment to the order level according to the SILVA v138 database.

Fig 3

Distribution of MAGs detected in the P. tricornutum pond metagenomes. The phylogenetic tree shows the relationship between the MAGs based on orthogroups (see Materials and Methods). Cultivated isolates are denoted by their strain identification around the tree as follows: PT13A: Oceanicaulis PT13A, ARW1R1: Algoriphagus ARW1R1, ARW1Y1: Muricauda ARW1Y1, ARW7G5Y1: Arenibacter ARW7G5Y1, PTEAB7WZ: Devosia PTEAB7WZ, ARW1T: Stappia ARW1T, PT6CLA: Rhodophyticola PT6CLA, PT4BL: Yoonia PT4BL, EA3: Pusillimonas EA3, EA2: Alcanivorax EA2, PT6ES: Henriciella PT6ES, EA10: Tepidicaulis EA10, PT13C1: Roseibium PT13C1, 11-3: Thalassospira 11-3, PT3-2: Marinobacter 3-2, ARW7G5W: Muricauda ARW7G5W, PT4D: Oceanicaulis PT4D, N2S: Roseobacter N2S, N5S: Sulfitobacter N5S. The relative abundance range of MAGs averaged across all samples is shown as bubbles surrounding the tree. MAG completeness shown as a heatmap ranging from 50% to 100% completeness. Clades are color-coded by Class-level GTDB taxonomy, and to the Family level for Alphaproteobacteria.

Fig 4

Abundant protein functional categories and proteins mapped to the Kordia MAG indicate antagonism and biopolymer degradation. (A) Breakdown of the most detected protein functional categories mapped to Kordia. The "Other" category specifies categories with less than 10 detected proteins across all metaproteomics samples. Proteins with "unknown" or "ambiguous" function were excluded for visual purposes. Protein assignments are available in Table S6. (B) The topmost abundant Kordia proteins detected across the demise samples (days 29 and 38) that had an average NSAF ≥ 10. Average NSAF across all demise samples (n = 5) is shown, and their standard deviations are represented as error bars.

Fig 5

Kordia fills a distinct niche compared to Rhodobacteraceae during algal demise by encoding abundant CAZymes targeting a diverse range of predicted biopolymers. (A) All Rhodobacteraceae non-redundant MAGs at ≥75% completion (n = 30) assembled from either the ARW1 or ARW7 ponds are shown for comparison with the Kordia MAG (boldened). MAG IDs are shown in brackets, followed by their genus-level assignment. Only CAZyme hits to glycoside hydrolases (“GH”) or polysaccharide lyases (“PL”) are included in the summarized CAZyme counts per MAG, and are color-coded by their predicted substrate. Emboldened MAGs indicate taxa that were active during the demise phase, as indicated by metaproteomics (see Fig. S9). (B) Specific CAZyme HMM hits and dbCAN substrate annotations for Kordia (mARW1_16). Hits with no substrate predicted (n = 81) were removed for visual purposes.

Fig 6

Sulfitobacter sp. N5S (a representative Rhodobacteraceae strain) requires Kordia to pre-condition P. tricornutum lysate for growth in culture. (A) CAZyme summary of representative lab strains of Kordia (K. algicida OT1) and a Rhodobacteraceae (Sulfitobacter sp. N5S) used for the experiment. (B) Growth of OT1 versus N5S on P. tricornutum lysate either pre-conditioned by OT-1 (“Conditioned”) or without any added bacteria (“Abiotic” control). Growth curves show the mean and standard deviation (error bars, n = 5) of OD₆₀₀ values taken every 20 minutes. The boxplots to the right show the cell abundances at the final timepoint (n = 4). P-values are generated using unpaired two-tailed t-tests.

Fig 7

Carbon monoxide (CO) oxidation as a potential alternative carbon and energy-generating mechanism used by Rhodobacteraceae in algal ponds. (A) Metabolic potential across all P. tricornutum pond MAGs to use alternative carbon and energy sources. All screened pathways are shown (x-axis), where dark blue tiles indicate 100% of pathway detection (i.e., all known genes involved in that pathway were detected) per-MAG (y-axis, grouped by family). (B) Protein presence/absence of each CODH subunit across MAGs in the metaproteome. Only MAGs that had a detected protein mapped to CODH out of 13 samples are shown.