Antibiotic Resistomes and Microbiomes in the Surface Water along the Code River in Indonesia Reflect Drainage Basin Anthropogenic Activities

Abstract

Water and sanitation are important factors in the emergence of antimicrobial resistance in low- and middle-income countries. Drug residues, metals, and various wastes foster the spread of antibiotic resistance genes (ARGs) with the help of mobile genetic elements (MGEs), and therefore, rivers receiving contaminants and effluents from multiple sources are of special interest. We followed both the microbiome and resistome of the Code River in Indonesia from its pristine origin at the Merapi volcano through rural and then city areas to the coast of the Indian Ocean. We used a SmartChip quantitative PCR with 382 primer pairs for profiling the resistome and MGEs and 16S rRNA gene amplicon sequencing to analyze the bacterial communities. The community structure explained the resistome composition in rural areas, while the city sampling sites had lower bacterial diversity and more ARGs, which correlated with MGEs, suggesting increased mobility potential in response to pressures from human activities. Importantly, the vast majority of ARGs and MGEs were no longer detectable in marine waters at the ocean entrance. Our work provides information on the impact of different influents on river health as well as sheds light on how land use contributes to the river resistome and microbiome.

Keywords: 16S rRNA amplicon sequencing; Antimicrobial resistance; bacterial communities; quantitative PCR; river health.

Figure 1

Map of the Code River displaying the sampling sites.

Figure 2

Most abundant bacterial orders and ARGs and MGEs in different sampling sites. Samples on the x-axis are grouped according to the sampling sites and color-coded. Sample names are shown in the legend in the middle. (A) Stacked bar plot showing 16 most abundant bacterial orders. (B) Venn diagram showing the OTUs that are shared between samples belonging to different sampling areas. (C) Venn diagram showing the ARGs and MGEs that are shared between samples belonging to different sampling areas. (D) Most abundant ARGs and MGEs (n = 85). Each row represents the results of each primer set (assay) (Supplementary Table S1) displayed on the y-axis. Assays are grouped according to the antibiotic group to which the target genes confer resistance. MLSB is the abbreviation for Macrolide, Lincosamide, Streptogramin B, and MGE for mobile genetic elements.

Figure 3

Bar plots showing the ARGs and MGEs with highest and statistically significant fold changes between sampling areas. Gamma distribution generalized linear models (GLMs) with false discovery rate control for p-value adjustment were used in statistical testing. (A) Comparison between rural and city areas. (B) Comparison between city and estuary areas. (C) Comparison between rural and estuary areas. See Supplementary Table S3 for fold changes of all differently abundant ARGs and MGEs.

Figure 4

Clustering of different samples in nonmetric multidimensional scaling (NMDS) plots. (A) NMDS ordination of OTUs. (B) NMDS ordination of ARGs and MGEs. Circles show 95% confidence area for standard error of the centroids of the sampling sites. Samples in the different sampling sites are not significantly different if these confidence areas overlap.

Figure 5

Comparisons of sum abundances and Shannon’s diversity indexes of different sampling areas. The asterisks “*”, “**”, “***”, and “****” denote statistical significance levels at p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively. (A) Comparison of sum abundance of ARGs and MGEs in different sampling areas using the Wilcoxon rank-sum test. (B) Comparison of Shannon’s diversity indexes of ARGs and MGEs in different sampling sites against the mean Shannon’s diversity index (dashed line) using Student’s t test. ARGs’ and MGEs’ r were not detected in the Spring Water sampling site. (C) Comparison of Shannon’s diversity indexes of OTUs in different sampling sites against the mean Shannon’s diversity index (dashed line) using Student’s t test. (D) Relationships between Shannon’s diversity indexes of OTUs and ARGs and MGEs in different sampling sites.

Figure 6

Comparison of detection frequencies (proportion of qPCR positive assays to the total number of targeted assays) of ARGs and MGEs in different sampling areas (color-coded) against published data from other studies (shades of gray) (Table S4). Samples are in increasing order, and the asterisks “*”, “**”, “***”, and “****” denote statistical significance levels at p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively, in the detection frequency of ARGs and MGEs against the mean detection frequency (dashed line) using Student’s t test. The sample names are as follows: FIN Agriculture Ditch Water = water from ditches receiving leachate and runoff from agricultural fields in Finland, Antarctic Soil Gondwana Station, Antarctic Soil Farm From Station, and Antarctic Soil Gondwana Station = soil samples from Antarctic research stations, FIN Agriculture Soil = Finnish soils before and after manure fertilization, FIN Aquaculture Sea Outside, FIN Aquaculture Fish Farm1, and FIN Aquaculture Fish Farm2 = sediments from Baltic sea outside fish farm and from under fish nets in two fish farms, respectively, FIN Agriculture Manure = Finnish production animal manure, CHN Pig Farms Soil, CHN Pig Farms Compost, and CHN Pig Farms Manure = manure fertilized soil, composted manure, and fresh manure from Chinese pig farms.

Figure 7

Comparison of potential pathogens (human, animal, or plant) and possible AMR traffickers in different sampling areas. The relative abundance (y-axis) of each genus in sampling areas (comparison groups “Rural”, “City”, and “Estuary”) was compared against the relative abundance of each genus in the Spring Water site (reference group) using the Wilcoxon rank-sum test. The asterisks “*” and “**” denote statistical significance levels at p < 0.05 and p < 0.01, respectively. Genera are color-coded according to the order displayed on the right.