Biofilms in premise plumbing systems as a double-edged sword: microbial community composition and functional profiling of biofilms in a tropical region

To understand distributions of opportunistic premise plumbing pathogens (OPPPs) and microbial community structures governed by sample location, pipe materials, water temperature, age of property and type of house, 29 biofilm samples obtained from faucets, pipes, and shower heads in different households in Singapore were examined using next-generation sequencing technology. Predictive functional profiling of the biofilm communities was also performed to understand the potential of uncultivated microorganisms in premise plumbing systems and their involvement in various metabolic pathways. Microbial community analysis showed Proteobacteria, Bacteroidetes, Acidobacteria, Nitrospira, and Actinobacteria to be the most abundant phyla across the samples which was found to be significantly different when grouped by age of the properties, location, and the type of house. Meanwhile, opportunistic premise plumbing pathogens such as Mycobacterium, Citrobacter, Pseudomonas, Stenotrophomonas, and Methylobacterium were observed from the samples at 0.5% of the total reads. Functional prediction using 16S gene markers revealed the involvement of the biofilm communities in different metabolic pathways like nitrogen metabolism, biodegradation of xenobiotics, and bacterial secretion implying diverse functionalities that are yet to be studied in this environment. This study serves as a preliminary survey on the microbial communities harboring premise plumbing systems in a tropical region like Singapore. This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). doi: 10.2166/wh.2020.182 om http://iwaponline.com/jwh/article-pdf/18/2/172/709123/jwh0180172.pdf 2021 Ryan De Sotto Rena Tang Sungwoo Bae (corresponding author) Department of Civil and Environmental Engineering, National University of Singapore, 5A Engineering Drive 1, Singapore 117411, Singapore E-mail: ceebsw@nus.edu.sg This article has been made Open Access thanks to the generous support of a global network of libraries as part of the Knowledge Unlatched Select initiative.


INTRODUCTION
Premise plumbing refers to the portion of the potable water distribution system connected to the main distribution system via service lines, including both hot and cold water and devices such as water heaters, showers, faucets and filters (Wang et al. ). Premise plumbing systems differ from the public water supply networks mainly due to con- public health risks, the concern regarding the presence of biofilms in oligotrophic or nutrient-depleted environments such as pipes mainly lies in their ability to harbour organisms that may become potentially pathogenic in certain cases (Wingender & Flemming ; Wang et al. ). Proliferation of these organisms in biofilms on surfaces of pipes is possible due to reactive pipe materials, long retention times of water containing residual nutrients, warmer temperature, and low or no residual disinfectant from the water treatment plants (Feazel et al. ; Falkinham ).
Due to the ecological niche these organisms are found to be thriving in and their disease-causing nature, they are collectively called opportunistic premise plumbing pathogens (OPPPs) (Buse et al. ). Some of the most common genera belonging to OPPPs include Mycobacterium, Legionella, and Pseudomonas (Kusnetsov et al. ; Falkinham et al. ; Naumova et al. ). In contrast to traditional waterborne pathogens, OPPPs naturally colonize, persist, and multiply in the biofilms of potable premise plumbing systems due to the abovementioned conditions (Ren et al. ; Richards et al. ). Their presence has become a growing concern even in developed countries due to their prevalence, persistence and the eventual healthcare burden these organisms may cause (Feazel et al. ; Beer et al. ; Falkinham ).
Despite these negative reports on biofilms in premise plumbing systems, a study has shown the ability of biofilms to degrade disinfection by-products formed as a result of the reaction between chlorine and natural organic matter in drinking water distribution systems (Pluchon et al. ).
Since eradication of microbes in premise plumbing systems is practically impossible and unrealistic, another study has hypothesized that a rich microbiome can be harnessed to control good bacteria which might be applicable to control downstream drinking water microbiome by manipulating the microbial community in the filtration process during water treatment (Pinto et al. ; Wang et al. ). However, limited studies and the lack of knowledge on the microbes being harboured in biofilms in residential premise plumbing systems as well as their potential functional profiles hinder the applicability of this concept to potable water systems.
In this study, biofilm samples from residential areas in Singapore were collected for the identification of microbial communities in premise plumbing systems using next generation sequencing technology. We also examined how various environmental factors such as location, pipe materials, age of property, and type of house (e.g., public housing, private condominiums, and landed homes) shape microbial community composition in premise plumbing systems. Lastly, functional profiling using 16S gene markers was performed to present different pathways and potential metabolic features of organisms which aim to show a different perspective on the presence of inevitable microbial communities in our plumbing systems at home.

Sampling sites
A total of 29 biofilm samples were collected from nine different sampling locations from central, southern, and western regions of Singapore using sterile cotton swabs in screw cap tubes ( Figure 1). Biofilms from pipes and shower heads that were accessible and did not require extensive dismantling were collected from the residential areas. Samples were kept in a À20 C freezer to preserve the DNA until further processing. The type of house, location, region, age of the property, and the average monthly water usage from the nine sampling sites are listed in Supplementary Table S1. Samples labeled as 'Heated' were collected from shower heads or hoses where the tap water has passed through the heater first while 'Normal' samples are collected from pipes without any installed heating mechanism (Supplementary Table S2).

DNA extraction
DNA was extracted using the FastDNA ® Spin Kit for Soil (Q-Biogene/MP Biomedicals, Solon, OH, USA) following instructions from the manufacturer. Extraction of DNA from drinking water environmental samples was shown to be most efficient using FastDNA ® Spin Kit for Soil (Huang et al. ). Briefly, the swabs used during sampling were cut allowing only the cotton bud area to be added into tubes containing Lysing Matrix E. 100 μL of DNA elution solution (DES) was used to elute the extracted DNA and samples were stored in À20 C until before use. DNA concentrations were quantified using Nanodrop ™ (Thermo Scientific, USA.).

16S rRNA sequencing and data processing
The 16S rRNA genes were amplified from all the DNA samples using 16S prokaryotic primers used by Takahashi

Functional profiling using PICRUSt
Phylogenetic investigation of communities by reconstruction of unobserved states or PICRUSt was used to predict metabolic and functional capabilities of the biofilm samples from the 16S gene markers obtained from the microbial community analysis by a previous study (Langille et al. ). Briefly, the OTU-containing biom file of the samples (annotated using Greengenes database) was normalized for 16S copy numbers using the default parameters of normalize_by_copy_number.py script. The resulting biom file was used as input for the predict_metagenomes.py script which produces an output with metagenome functional predictions for an OTU table. Thereafter, the biome file output was used for pathway analysis using KEGG pathways which counts the number of genes involved in several pathways per sample using the categorize_by_function.py script. KEGG pathways contain manually drawn pathway maps that represent known molecular interaction, reaction, and networks for metabolism, genetic and environmental information processing, cellular processes, organismal systems, human diseases, and drug development (Kanehisa et al. ). Data for the functional profiling and the study's metadata (supplementary information Table S2) were uploaded for visualization in statistical analysis of metagenomic profiles (STAMP) using the software's default parameters for group analysis (Parks et al. ).

RESULTS AND DISCUSSION
Microbial community structures in premise plumbing systems A total of around 700,000 reads were used as an input for quality filtering and 176,000 reads with lengths shorter than the truncation length of 200 bp were also excluded.
Of these, 520,000 reads having a median sequence length of 440 bp were processed in QIIME for clustering, OTU picking, and annotation (Caporaso et al. ). After quality filtering, 37 various phyla were identified and the top 9 phyla with the highest relative abundance of all samples are shown in Figure 2. The phylum Proteobacteria is found to be the most predominant group with a relative abundance of 17% of the total microbial population followed by Actinobacteria (4%), Bacteroidetes, Nitrospirae, Cyanobacteria, and teria, a phylum that is primarily recognized for oxygenic photosynthesis and one of the oldest, most morphologically diverse of the phyla, has been seen in chlorinated bulk waters while a study on biofilms from water distribution systems also reported this phylum to be one of the major In this study, Pseudomonas reads were seen mostly from metal pipe material, landed houses, residential houses in the east and west of Singapore, and in heated sources. The occurrence of Pseudomonas reads in heated sources in this study is supported by the fact that higher temperatures (<40 C) tend to be more favourable for bacterial growth.
A previous study has reported an increased percentage of this group in warm water than cold water samples (Völker et al. ). Meanwhile, since there are no published studies on the microbial community that may be found in water distribution networks in the east and west areas of Singapore (Figure 4), the link between their increased prevalence in these areas cannot be supported by any previous data and therefore needs further investigation. Of the OPPPs, the Legionella pneumophila is the least abundant with mean abundance of 0.002%. This organism is the causative agent of a life-threatening pneumonia (Yoder et al. ). The presence of this organism in water distribution networks and buildings does not necessarily indicate poor maintenance but rather just their ubiquitous presence totalling 50% in building water systems and about 30% in home water systems in the United States (Kool et al. ). This organism was only found in some of the public housing ( Figure S2) and OTU reads in metal ( Figure S3), normal water source ( Figure S4), and the west region of Singapore (Figure 4).
Citrobacter is a Gram-negative, non-spore-forming facultatively anaerobic genus and is found to be an opportunistic human pathogen which inhabits intestines of animals leading to its presence in soils, water, and sewage as a The presence of OPPPs in the biofilm samples of premise plumbing systems only accounts for less than 0.5% of the total microbial communities found in the analysis and may be deemed insignificant in terms of public health concerns. However, their prevalence from the samples collected indicates potential health risks of immunocompromised individuals, the very young and the elderly. Also, further studies of microbial ecology and potential regrowth of OPPPs in a premise plumbing pipe system are necessary for better understanding of their ecological roles and potential health risks. Particularly, the relationship between the most frequently identified taxa and the heathy conditions of the residents shall be investigated to develop water treatment strategies and maintenance practices for a premise plumbing pipe system. It is important to note, however, that the viability of these organisms was not taken into consideration in this study, but only their abundance from extracted DNA from the samples. Monitoring the viability of these opportunistic pathogens and their quantities in biofilm samples collected from the residential areas would be useful for the formulation of risk assessment frameworks that are associated with the presence of these organisms which is considered a health hazard for the public. Furthermore, quantitative PCR (qPCR) using pathogen-specific primers was not performed in this study due to the low-DNA yield from the collected biofilm samples. The authors acknowledge the need to verify the presence of these pathogenic organisms up to the species level using more accurate detection methods to confirm the preliminary results we showed in this study. Culture-dependent methods or quantitative PCR assays with propidium monoazide or DyeTox13 Green C-2 azide could be used as a confirmatory test for the viability of the pathogenic organisms in the premise plumbing systems (Lee & Bae ). Amplicon sequencing of 16S rRNA genes was used in this study as an initial

CONCLUSIONS
This study has shown the presence of various microbial communities, their metabolic potential, and potential opportunistic pathogens in the biofilm samples collected from residential areas Singapore. We illustrated that factors such as type of housing, age of property, sampling location, temperature, pipe material may affect the composition of microbial communities and the functional potential of the biofilms thriving in the premise plumbing systems. With regard to opportunistic premise plumbing pathogens (OPPPs), the lack of evidence correlating prevalence of infection and their presence in residential plumbing systems calls for further research on the health risks associated with their occurrence. Furthermore, since the viability of these organisms was not investigated in this study, culturedependent novel molecular methods such as PMA-PCR or DyeTox13-PCR should be employed in future studies that aim to address any public health-related concerns. Meanwhile, functional profiling of the biofilm communities presented interesting pathways related to membrane transport and DNA regulation and functions like xenobiotics degradation and nitrogen metabolism. The relative abundances of organisms harbouring these traits are significantly higher than the OPPPs which account for only 0.5% of the total sequencing reads. Since biofilms are ubiquitous in premise plumbing systems, it is an imperative to control microbiome to have beneficial effects on both water treatment and residential plumbing systems, while minimizing regrowth of potential pathogens. The identification of diverse microbial communities from biofilm samples in residential plumbing systems emphasizes the need for further studies that would tackle public health risks from OPPPs as well as the role of biofilms in degrading residual xenobiotic compounds in this kind of environment.