< Back to previous page

Publication

Characterization of the bacterial community composition in drinking water production and distribution systems, emphasizing Acinetobacter species.

Book - Dissertation

The production and delivery of safe drinking water is challenging. The quality of drinking water is influenced by its chemical and microbial composition which in turn may be affected by the source water and the different processes applied in the production of drinking water. Important bacterial waterborne pathogens include members of the genera Legionella, Mycobacterium and Acinetobacter. Acinetobacter species are usually commensal organisms, but they occasionally cause infections, predominantly in susceptible patients in hospitals. Whereas Acinetobacter spp. are frequently found in soil and aquatic environments, so far only little is known about their ecology in drinking water and drinking water production and distribution facilities. The major aim of this PhD study was to assess the bacterial community composition of drinking water production and distribution systems in Flanders (Belgium), emphasizing Acinetobacter spp. In the first chapter (Chapter I), we discuss the drinking water production process, with a particular focus on drinking water bacterial communities as well as a number of bacteria of potential health concern. Furthermore, a comprehensive literature overview is given on the genus Acinetobacter, including taxonomic history, its main phenotypic traits, how Acinetobacter is adapted to thrive in diverse environments, and their clinical relevance. In Chapter II, using high-throughput sequencing of partial 16S ribosomal RNA (rRNA) gene amplicons, we investigated the bacterial diversity in different water samples from the production and distribution chain of thirteen drinking water production and distribution systems from Flanders (Belgium) that use surface water or groundwater as source water. Water samples were collected over two seasons from the source water, the processed drinking water within the production facility, and out of the tap in houses along its distribution network. Strong differences in bacterial community composition were found between processed drinking water originating from companies that use surface water as source water. Differences were less pronounced for companies that use groundwater as source water. Proteobacteria was the most abundant phylum in all samples. Yet, several phyla including Actinobacteria were significantly more abundant in surface water, while Cyanobacteria were more abundant in surface water and processed water originating from surface water. Gallionella, Acinetobacter and Pseudomonas were the three most abundant genera detected. Members of the Acinetobacter genus were even found at a relative read abundance of up to 47.5 % in processed water samples. To further investigate how bacterial communities are shaped during drinking water production and distribution as well as to further examine which Acinetobacter species occur in this ecosystem, a follow-up study was performed at the site of one particular drinking water production facility in Antwerp (Belgium) (Chapter III). More particularly, the goal of this study was to investigate the bacterial community shifts during production and distribution of drinking water, by studying a full-scale drinking water production and distribution facility that uses two series of multi-step treatment processes starting from the same source water (surface water). Furthermore, we investigated the presence and abundance of Acinetobacter at each step of the treatment chain using both culture-dependent and culture-independent methods (i.e. isolation after enrichment and quantitative real-time PCR (qPCR)). Finally, we studied the importance of the physicochemical characteristics of the drinking water on the bacterial community composition in the water delivered to the end user (household tap water). Quantification of total bacteria revealed that for both production lines the amount of bacteria decreased when a treatment was applied, and the lowest amount of bacteria were detected after chlorination. There was a clear difference between the bacterial community composition in both production lines. In the first line (line A; applied treatments: rapid sand filtration, slow sand filtration, activated carbon filtration, UV treatment and chlorination), a substantial community shift was observed after slow sand filtration, resulting in a large increase in richness in operational taxonomic units (OTUs; an operational definition used to classify groups of closely related individuals) defined by a standard cut-off of 97 % 16S rRNA gene sequence identity. In the second production line (line B; applied treatments: flotation, double layer filtration, activated carbon filtration, UV treatment and chlorination), OTU richness gradually increased after every treatment step. For both lines, OTU richness decreased after chlorination. Taxonomy assignment of the OTUs revealed that Proteobacteria was again the most abundant phylum, although changes in the relative abundance of phyla was observed between treatment steps. Strikingly, the bacterial community composition of the waters sampled at the end of production line A (tap water) and in the distribution network of line B (storage tank) were highly similar. Seasonal effects showed to be of minor influence in shaping the bacterial community composition. In contrast to Chapter II, Acinetobacter was found at low relative abundance in the water samples investigated in this study, reaching a maximum relative abundance of 2.8 %. Absolute abundance of Acinetobacter was the lowest after chlorination for both lines. Yet, for line A, a significant increase of Acinetobacter was observed in the household tap water in comparison to the finished product within the drinking water facility. Plating of the water samples on agar media revealed a total of 14 different species (based on rpoB (RNA polymerase beta subunit) gene similarity), among which three potential pathogenic Acinetobacter species (i.e. A. guillouiae, A. johnsonii and A. lwoffii) were found in finished drinking water or household tap water. Water chemical parameters were more significantly different between seasons than between sampling locations, with the exception of trihalomethanes. Trihalomethanes concentrations were lower in the tap water samples taken at one of the locations that received drinking water from line A. Interestingly, the bacterial community in the drinking water sampled at this location was enriched in Methylophilus species which are capable of utilizing chlorinated methanes. Next, in Chapter IV we determined the level of genetic and phenotypic relatedness between Acinetobacter isolates from environmental (mainly (drinking) water) and clinical environments using 58 isolates belonging to four Acinetobacter species that are associated with human infections (A. calcoaceticus, A. guillouiae, A. johnsonii en A. lwoffii). Isolates were phenotyped using Biolog's GENIII identification microplate, which analyzes the performance of a microorganism in 94 phenotypic tests, including 71 carbon source utilization assays and 23 chemical sensitivity assays. A Spearman rank correlation analysis was performed to determine whether the ability to use different carbon sources co-varied between the different isolates. Additionally, antibiotic susceptibility testing was performed using 15 antibiotic compounds. Further, isolates were genotyped by partial sequencing of the rpoB gene, and a Mantel test was performed to assess correlations between phenotypic and rpoB gene sequence similarity. When results obtained for the different carbon sources were evaluated based on the origin of the isolates (i.e. aquatic/environmental vs. clinical/veterinary) (irrespective of species classification), significant differences were obtained for six carbon sources. On the other hand, 34 carbon sources and 14 chemical sensitivity assays were significantly different based on grouping by species classification, suggesting that phenotypic traits are more species- than habitat-dependent. The Spearman rank correlation test showed that several carbon sources co-varied between the different isolates. Strong significant correlations were found for sugars and tend to be species-dependent. Antibiotic susceptibility testing revealed that especially isolates from finished drinking water (i.e. after chlorination) displayed resistance to one or more antibiotics. A significant relationship between the pairwise phylogenetic distance and trait differentiation among isolates was found for four carbon source assays (i.e. L-arginine, L-histidine, citric acid and γ-amino-butyric acid) and one chemical stressor (i.e. sodium bromate). Finally, in Chapter V we tested the hypothesis whether bacterial phylogeny reflects molecular functions and phenotypic characteristicsfor a large set of Acinetobacter strains. A total of 133 strains belonging to 33 Acinetobacter species with validly published names and eight genomic species were analyzed using the GENIII technology of Biolog. We estimated the strength and significance of the phylogenetic signal of each trait across phylogenetic reconstructions based on partial rpoB and core genome sequences. Secondly, we tested whether phylogenetic distance was a good predictor of trait differentiation by a Mantel test. And finally, evolutionary model fitting was used to determine if the data for each phenotypic character was consistent with a phylogenetic or an essentially random model of trait distribution. Results revealed that some key phenotypic traits related to substrate assimilation and chemical sensitivity are linked to the phylogenetic placement of Acinetobacter species. The strongest phylogenetic signals found were for utilization of carbon sources such as some organic acids, amino acids and sugars, thus suggesting that in the diversification of acinetobacters carbon source assimilation has had a relevant role. In conclusion, this PhD study has provided new insights on the bacterial community composition in drinking water production and distribution systems which use groundwater or surface water as source water. Furthermore, we studied how bacterial community composition shifts can be attributed to different treatment steps and which Acinetobacter species are present at the different stages of drinking water production and distribution. Additionally, we provide new insights in the phenotypic traits of different Acinetobacter species which can partially be influenced by the habitat of isolation. And finally, using Acinetobacter as a study object, we showed that the phylogeny of bacteria can reflect phenotypic characteristics.
Publication year:2019