Seed Imbibition and Metabolism Contribute Differentially to Initial Assembly of the Soybean Holobiont

Abstract

Seed germination critically determines successful plant establishment and agricultural productivity. In the plant holobiont's life cycle, seeds are hubs for microbial communities' assembly, but what exactly shapes the holobiont during germination remains unknown. Here, 16S rRNA gene amplicon sequencing characterized the bacterial communities in embryonic compartments (cotyledons and axes) and on seed coats pre- and post-germination of four soybean (Glycine max) cultivars, in the presence or absence of exogenous abscisic acid (ABA), which prevented germination and associated metabolism of seeds that had imbibed. Embryonic compartments were metabolically profiled during germination to design minimal media mimicking the seed endosphere for bacterial growth assays. The distinction between embryonic and seed coat bacterial microbiomes of dry seeds weakened during germination, resulting in the plumule, radicle, cotyledon, and seed coat all hosting the same most abundant and structurally influential genera in germinated seeds of every cultivar. Treatment with ABA prevented the increase of bacterial microbiomes' richness, but not taxonomic homogenization across seed compartments. Growth assays on minimal media containing the most abundant metabolites that accumulated in germinated seeds revealed that seed reserve mobilization promoted enrichment of copiotrophic bacteria. Our data show that seed imbibition enabled distribution of seed-coat-derived epiphytes into embryos irrespective of germination, while germinative metabolism promoted proliferation of copiotrophic taxa, which predominated in germinated seeds.

Keywords: Bacillus; Buchnera; Glycine max (soybean); Rhodococcus; abscisic acid; imbibition; metabolism; microbiome; seed germination.

Fig. 1. Overview of soybean germination and material analyzed.

A, Progress of germination (red circles) and imbibition (blue squares) overlaid with images of typical seeds at time points of analysis. B, Images of typical seed structures analyzed at each time point. t-dry, dry seeds; t-imb, imbibed seeds just before any had germinated; t-germ, germinated seeds 5 days after imbibition started.

Fig. 2

Constrained analysis of principal coordinates (CAP) performed on Bray–Curtis dissimilarities on normalized data, considering “time of germination progress” and “seed compartment” (see key, top) as constraining factors for each of the four soybean cultivars indicated in gray boxes. t-dry, dry seeds; t-imb, imbibed seeds just before any had germinated; t-germ, germinated seeds 5 days after imbibition started. Each symbol represents a unique replicate (n = 3 of 30 seeds each). The proportions of variance explained by constrained eigenvalues are reported along the figure axes.

Fig. 3

Relative abundance of the top 10 bacterial indicator genera influencing the composition of microbiomes during the progress of soybean seed germination. Vertical bars represent replicates (n = 3 of 30 seeds) ordered by the factors (i) “time of germination progress”: dry, dry seeds; t-imbibition, imbibed seeds just before any had germinated; t-germination, germinated seeds 5 days after imbibition started. (ii) “Seed compartment”: cotyledons, seed coats, and embryonic/seedling axes (axes) that were subdivided into radicle and nonemerged plumule (left and right of dashed lines, respectively) at t-germination. (iii) “Cultivar”: Abelina (Abe), Amadea (Ama), Amadine organic (Am-org), and Cordoba (Cor). Distinctive colors denote genera (bottom key).

Fig. 4

Metabolite changes during seed germination of the soybean cultivar Abelina. Seed metabolites are grouped into classes of amino acids (with γ-aminobutyric acid, GABA), saccharides, organic acids, and sugar alcohols. Seeds were analyzed before imbibition (t-dry), after 50 h imbibition before any had germinated (t-imb), and germinated 5 days after imbibition started (t-germ). Fold changes in the concentrations of metabolites between t-imb and t-dry, and t-germ and t-dry, respectively, are shown for cotyledon and embryonic/seedling axis (axis) on a log₂ basis with colored scale (bottom). Asterisks denote significant differences at P value < 0.05 (single) and < 0.01 (double) for each metabolite relative to concentrations at t-dry.

Fig. 5

Correlation between changes in metabolite concentrations 5 days after imbibition started (t-germ) relative to either seeds imbibed with 800 μM abscisic acid (t-germ-ABA) or dry seeds (t-dry). Symbols denote biochemical classes of metabolites in embryonic/seedling axes (closed symbols; Ax) and cotyledons (open symbols; Cot). The linear fit considers all data (n = 5 replicates of 30 axes or cotyledons for each time point of germination progress). Triangles, amino acids (AA); circles, saccharides (Sug); inverted triangles, organic acids (OA); squares, sugar alcohols (SOH).

Fig. 6. Effect of exogenous abscisic acid (ABA) on the richness and diversity of bacterial communities of soybean seeds 5 days after imbibition started.

A, Decision tree of bacterial α-diversity based on recursive partitioning analysis. The factors are (i) “treatment”: ABA-treated seeds (t-germ-ABA) and untreated control (t-germ); (ii) “cultivar”: Abelina (Abe), Amadea (Ama), Amadine organic (Am-org), and Cordoba (Cor). Richness, as represented by number of detected amplicon sequence variants (ASVs), and diversity, calculated as Simpson’s index, are shown as boxplots with outliers denoted by open circles. B, Relative abundance of the top 10 bacterial indicator genera (bottom key) influencing the microbiomes’ composition in embryonic and seedling axes (axes), cotyledons, and seed coats of each cultivar, with t-germ seedling axes subdivided into radicle and nonemerged plumule (left and right of dashed lines, respectively). Each bar represents a unique replicate (n = 3 of 30 seeds each).

Fig. 7. Correlations of seed metabolite changes with shifts in bacterial taxa during the progress of soybean seed germination, and influence of metabolites on bacterial growth.

A, Correlogram summarizing correlations between individual metabolites and amplicon sequence variants (ASVs) selected according to their abundance (rank test) and importance (random forest), as given in Supplementary Table S7. Significant correlations are depicted by circles, whose diameter is proportional to Spearman’s r coefficients (P value < 0.05), and colors refer to the correlation direction (key on the right). Empty cells denote lack of significant correlation. Taxonomic assignment: ASV 2, Erwiniaceae; ASV 3, Pantoea agglomerans; ASV 4, Bacillus; ASV 6: Buchnera aphidicola; ASV 8, Curtobacterium; ASV 9, Staphylococcus; ASV 15, Bacillaceae; ASV 17, Paenibacillus; ASV 21, Pseudomonas; ASV 69, Enterobacterales; ASV 88, Rhodococcus fascians; ASV 100, Bacillus halmapalus; ASV 136, Bacillales; ASV 177, Paenibacillus. B, Changes over time in the number of colony-forming units (CFU) of selected strains on tryptic soy agar (TSA) and M9 minimal saltsbased media supplemented with the most abundant metabolites targeted in each biochemical class, as indicated in gray headings, at concentrations detected in the cotyledons of germinated seeds 5 days after imbibition started. Data are means ± SE (n = 4) for Rhodococcus fascians (ASV 88, blue) and Pantoea agglomerans (ASV 3, red), and different letters denote significant differences (nonparametric Kruskal–Wallis rank tests followed by the Bonferroni correction; P value < 0.05).