Metagenomic profiling of microbial composition and antibiotic resistance determinants in Puget Sound

Abstract

Human-health relevant impacts on marine ecosystems are increasing on both spatial and temporal scales. Traditional indicators for environmental health monitoring and microbial risk assessment have relied primarily on single species analyses and have provided only limited spatial and temporal information. More high-throughput, broad-scale approaches to evaluate these impacts are therefore needed to provide a platform for informing public health. This study uses shotgun metagenomics to survey the taxonomic composition and antibiotic resistance determinant content of surface water bacterial communities in the Puget Sound estuary. Metagenomic DNA was collected at six sites in Puget Sound in addition to one wastewater treatment plant (WWTP) that discharges into the Sound and pyrosequenced. A total of ~550 Mbp (1.4 million reads) were obtained, 22 Mbp of which could be assembled into contigs. While the taxonomic and resistance determinant profiles across the open Sound samples were similar, unique signatures were identified when comparing these profiles across the open Sound, a nearshore marina and WWTP effluent. The open Sound was dominated by α-Proteobacteria (in particular Rhodobacterales sp.), γ-Proteobacteria and Bacteroidetes while the marina and effluent had increased abundances of Actinobacteria, β-Proteobacteria and Firmicutes. There was a significant increase in the antibiotic resistance gene signal from the open Sound to marina to WWTP effluent, suggestive of a potential link to human impacts. Mobile genetic elements associated with environmental and pathogenic bacteria were also differentially abundant across the samples. This study is the first comparative metagenomic survey of Puget Sound and provides baseline data for further assessments of community composition and antibiotic resistance determinants in the environment using next generation sequencing technologies. In addition, these genomic signals of potential human impact can be used to guide initial public health monitoring as well as more targeted and functionally-based investigations.

Figure 1. Locations of sampling sites in Puget Sound.

Refer to Table 1 for the geographic coordinates of the sampling stations. WWTP, Wastewater treatment plant.

Figure 2. Relative abundance of major taxonomic groups in the Puget Sound samples and other selected metagenomes.

Taxonomic groupings were based on BLASTX comparison to the NCBI taxonomy using MG-Rast and ≥50% identity and an alignment length ≥50 amino acids. The ‘other’ category includes bacteria taxa present in <1% of sequences in all samples, eukaryotes and viruses. Shading is proportional to the relative abundance of each taxon within a metagenome. The cladogram was displayed using hierarchical clustering and the Euclidian distance metric. See Materials and Methods for references describing the additional metagenomes used for comparison. The open Sound locations cluster with other more saline environments including open and coastal ocean samples while the Marina and WWTP effluent samples cluster within a larger clade containing other freshwater and freshwater-impacted metagenomes.

Figure 3. Over- and under-representation of predominant taxa in selected biomes relative to the open Puget Sound metagenome.

(A) Overall taxa, (B) α-Proteobacteria, (C) β-Proteobacteria, (D) Bacteroidetes and (E) Actinobacteria. The difference in proportions refers to the percent difference in the relative abundances of a given taxa between two locations. Only taxa with a difference of proportions >1% are shown. Negative values indicate higher relative abundance in the Puget Sound (p<10⁻¹⁵ for all comparisons). See Materials and Methods for references describing the additional metagenomes used for comparison. A unique taxonomic signature was identified for the Puget Sound consisting of an over-representation of α-Proteobacteria, γ-Proteobacteria and Bacteriodetes and an under-representation of β-Proteobacteria and Actinobacteria.

Figure 4. Recruitment plot of the unassembled Puget Sound reads to the alpha-Proteobacterium HTCC2255 genome assembly.

Open Sound and Marina metagenomic sequence reads were recruited to the 2.23 MB assembly (GI number: 211594581) using BLASTN and an E-value cutoff of 10⁻⁵. Over 91% of the assembly was matched at greater than 95% identity, equaling approximately 7X coverage.

Figure 5. Prevalence of antibiotic resistance gene sequences in the Puget Sound metagenomes.

Sequences were classified as antibiotic resistance genes if sharing ≥80% identity and an alignment length ≥50aa to a sequence within an expanded Antibiotic Resistance Genes Database (ARDB) . Open Sound samples were pooled together for the analysis. Bars represent 95% confidence intervals for binomial proportions. For open Sound vs. Marina and WWTP, *p<0.05 and **p<0.005. For Marina vs. WWTP, ⁺p<0.05 and ⁺⁺p<0.005. ‘n = ’ refers to the number of antibiotic resistance gene sequences. The y-axis represents a modified log scale. While the resistance gene signals were low for the samples, the signals were significantly different across the sample types, suggesting antibiotic resistance gene abundance may reflect differences in potential human health impacts.

Figure 6. Prevalence of mobile genetic elements in the Puget Sound metagenomes.

(A) Plasmid sequences, (B) transposable elements using a BLAST search against GenBank, (C) transposable elements using a hidden markov model (HMM) search against the Pfam database and (D) top ten most abundant transposable elements from the WWTP effluent by percent abundance. Bars represent 95% confidence intervals for binomial proportions. P-values are relative to the open Sound and are corrected for multiple testing. For open Sound vs. Marina and WWTP, *p<0.05 and **p<0.005. For Marina vs. WWTP, ⁺p<0.05 and ⁺⁺p<0.005. Bars represent 95% confidence intervals for binomial proportions. The dashed line represents the mean percent for the open Sound. Similar profiles for mobile genetic elements were seen for the open Sound samples while significant differences emerged when comparing to the Marina and effluent.