Coastal Transient Niches Shape the Microdiversity Pattern of a Bacterioplankton Population with Reduced Genomes

Abstract

Globally dominant marine bacterioplankton lineages are often limited in metabolic versatility, owing to their extensive genome reductions, and thus cannot take advantage of transient nutrient patches. It is therefore perplexing how the nutrient-poor bulk seawater sustains the pelagic streamlined lineages, each containing numerous populations. Here, we sequenced the genomes of 33 isolates of the recently discovered CHUG lineage (~2.6 Mbp), which have some of the smallest genomes in the globally abundant Roseobacter group (commonly over 4 Mbp). These genome-reduced bacteria were isolated from a transient habitat: seawater surrounding the brown alga, Sargassum hemiphyllum. Population genomic analyses showed that: (i) these isolates, despite sharing identical 16S rRNA genes, were differentiated into several genetically isolated populations through successive speciation events; (ii) only the first speciation event led to the genetic separation of both core and accessory genomes; and (iii) populations resulting from this event are differentiated at many loci involved in carbon utilization and oxygen respiration, corroborated by BiOLOG phenotype microarray assays and oxygen uptake kinetics experiments, respectively. These differentiated traits match well with the dynamic nature of the macroalgal seawater, in which the quantity and quality of carbon sources and the concentration of oxygen likely vary spatially and temporally, though other habitats, like fresh organic aggregates, cannot be ruled out. Our study implies that transient habitats in the overall nutrient-poor ocean can shape the microdiversity and population structure of genome-reduced bacterioplankton lineages. IMPORTANCE Prokaryotic species, defined with operational thresholds, such as 95% of the whole-genome average nucleotide identity (ANI) or 98.7% similarity of the 16S rRNA gene sequences, commonly contain extensive fine-grained diversity in both the core genome and the accessory genome. However, the ways in which this genomic microdiversity and its associated phenotypic microdiversity are organized and structured is poorly understood, which disconnects microbial diversity and ecosystem functioning. Population genomic approaches that allow this question to be addressed are commonly applied to cultured species because linkages between different loci are necessary but are missing from metagenome-assembled genomes. In the past, these approaches were only applied to easily cultivable bacteria and archaea, which, nevertheless, are often not representative of natural communities. Here, we focus on the recently discovered cluster, CHUG, which are representative in marine bacterioplankton communities and possess some of the smallest genomes in the globally dominant marine Roseobacter group. Despite being over 95% ANI and identical in the 16S rRNA gene, the 33 CHUG genomes we analyzed have undergone multiple speciation events, with the first split event predominantly structuring the genomic diversity. The observed pattern of genomic microdiversity correlates with CHUG members' differential utilization of carbon sources and differential ability to explore low-oxygen niches. The available data are consistent with the idea that brown algae may be home to CHUG, though other habitats, such as fresh organic aggregates, are also possible.

Keywords: CHUG; Roseobacter; Sargassum; microdiversity; population genomics; streamlined genomes.

FIG 1

Population differentiation of the 33 CHUG isolates from a single 50 mL seawater sample surrounding a brown alga. (A) Genetically isolated populations from M1 to M5 as well as the two subpopulations, M5S1 and M5S2, within M5 that were defined by PopCOGenT are mapped to the maximum likelihood phylogenomic tree constructed by IQ-TREE, with the root position determined by the MAD algorithm. Solid circles in the phylogeny indicate nodes with bootstrap values of 100% in the 1,000 bootstrapped replicates. Three split events discussed in the main paper are marked with arrows. (B to D) The distribution of the F_ST values across 1,788 core genes between the lineages, resulting from each of the three split events. Gene families showing F_ST > 0.25 with statistical significance (P < 0.05) are marked in red, and others are marked in gray. For the dot plots (top), genes are ordered along the closed genome of HKCCA1288. For the histogram (bottom), the count represents the number of gene families. (E) The dendrogram is constructed based on gene presence and absence using the complete linkage method implemented in the R package, “pheatmap”.

FIG 2

The pangenome of the 33 CHUG isolates. All of the orthologous gene families identified by Roary are positioned according to the closed genome, HKCCA1288. From the inner to the outer circle: (1 to 33) the genomes of the 33 CHUG isolates, arranged in line with their phylogenomic tree, which is shown in the top-left corner. Genomic islands, prophages, insertion sequences, and pseudogenes are marked with different colors and are mapped to each genome; (34) M1M2-specific genes, M3M4M5-specific genes, and core genes showing unusually large d_S values are marked with different colors and are placed according to the coordinates of the HKCCA1288 genome; (35) M3M4-specific genes and M5-specific genes; (36) M5S1-specific genes and M5S2-specific genes. Gene families falling in the functional categories, “energy production and conversion”, “amino acid transport and metabolism”, and “carbohydrate transport and metabolism”, are each attached with a gene name and framed in a box with a background color corresponding to a functional category. Each box is connected with a line that is colored according to whether the gene family is part of the population-specific accessory genes or the d_S outlier core genes.

FIG 3

(A) The heat map of the respiration values of a few representative strains of the populations defined by PopCOGenT at 96 h, measured by BiOLOG. The clustering is constructed using all 190 carbon sources provided by the two microplates, PM01 and PM02, but only those 85 substrates that showed a respiration value over 10 OmniLog units in at least one strain are displayed. Substrates differentially utilized by M1M2 and M3M4M5 are framed with green boxes. (B) The respiration curves of the representative strains fed with the framed substrates in (A).

FIG 4

Oxygen uptake measurements. (A) Oxygen consumption of two CHUG isolates (HKCCA1288 and HKCCA1065) from M1M2, two CHUG isolates (HKCCA1006 and HKCCA1086) from M3M4M5, and Ruegeria pomeroyi DSS-3 under microaerobiosis. (B) The relationship between oxygen concentration and oxygen uptake rates at an oxygen level of <2 μmol/L. The solid line is a model fitted based on the Michaelis-Menten equation, and the nearby dotted line shows the standard deviation estimated by this equation. The error bars of the filled circles represent the standard deviations of three replicates. (C) The Hanes-Woolf plot of oxygen uptake as a function of oxygen concentration, ranging from 100 nmol/L to 1 μmol/L. The error bars of the filled circles represent the standard deviations of three replicates. (D) The estimated specific affinity of oxygen for the four CHUG isolates and R. pomeroyi DSS-3.