Introduction

Drinking water is distributed through complicated pipe systems until it arrives at consumers′ tap (Lee et al. 2010). As taps are one of the endpoints in the drinking water distribution, citizens are exposed to and consume large amounts of tap water in their daily life (Von Baum et al. 2010). However, drinking water quality is not monitored routinely at the household level but rather directly in the distribution system, as waterworks and authorities have limited access to private homes, which increases drinking water risk (Lautenschlager et al. 2010). Even though the drinking water distribution system is well treated and strictly managed, both water and biofilms may still harbor microorganisms (Lehtola et al. 2004). It has been estimated that about 95 % of all microbial cells present in drinking water distribution systems exist as biofilms on pipe surfaces and only 5 % occur in the water phase (Moritz et al. 2010).

Microbial growth in the drinking water distribution system leads to a deterioration of aesthetic water quality like undesirable tastes, odors, and visual turbidity, and even pipe corrosion (Hammes et al. 2008). In addition, the occurrence of biofilms may also cause growth of potentially pathogenic bacteria (Lehtola et al. 2007; Giao et al. 2008; Wingender and Flemming 2011). Potentially pathogenic bacteria such as Legionella and Mycobacteria may easily cause a potential health risk, especially for immunocompromised individuals, when contaminated water pass through taps and aerosolize in indoor air (Hedayati et al. 2011). Moreover, it is well-known that bacteria in biofilms display a much higher tolerance to the environment than suspended counterparts (Flemming and Ridgway 2009). Hence, to be more efficient in the control of biofilms, a better understanding of the mechanisms involved in the biofilm development, especially those factors affecting the rate of bacterial colonization and regrowth, is necessary (Momba and Makala 2004). There are several factors which can influence the formation of biofilm, e.g., nutrients, disinfectant residual concentration, water hydraulic conditions, and materials (Momba et al. 2000; Lehtola et al. 2002). The most complicated effects on biofilm formation come from attached materials.

Many studies have focused on the effect of pipe materials on biofilm formation, which makes up drinking water distribution systems. For instance, (Kerr et al. 1999) compared biofilm accumulation and heterotrophic bacterial diversity on three pipe materials, cast iron, medium density polyethylene (MDPE), and unplasticized polyvinyl chloride (uPVC), and found that MDPE and uPVC supported the lowest number of bacteria, but the diversity of heterotrophic bacteria was greatest on cast iron. In contrast to the study, research work carried out by (Paquin et al. 1992) reported that the number of heterotrophic plate count of the biofilm grown on three types of coupons (polyvinyl chloride (PVC), polyethylene (PE), and cement-lined cast iron) did not show any difference. Therefore, there is still controversy about the effect of surface materials on biofilm development. Based on the previous studies on bacteria–surface material interaction, we wonder about the relationship between bacterial communities and attached tap materials. To our knowledge, domestic tap materials have been considered less frequently (Rudi et al. 2009). Therefore, more research is required in this field. In order to study drinking water biofilms, researchers have traditionally used culturing methods. However, only a small proportion of microorganisms can be studied using culture techniques (Phung et al. 2004; Roeder et al. 2010). Hence, molecular methods have been developed using 16S rRNA gene-based approaches to identify bacterial species and assess their abundances within drinking water communities (Henne et al. 2012).

The aim of the present study was to observe the microbe changes occurring in inflow and outflow water samples through the network and investigate bacterial community on taps with three different materials (PVC, stainless steel, and cast iron) of a more than 10-year-old drinking water network using 16S rRNA gene clone libraries.

Materials and methods

Water and biofilm sampling

The water and biofilm samples were analyzed from a drinking water distribution system of Hubei Province, China. The source water entered the drinking water treatment plant and went through various treatment procedures including coagulation–flocculation, settling, and filtration. A final disinfection step to produce finished drinking water was carried out by the addition of ClO2 in the clear water tank. Also, free residual chlorine was subsequently maintained in the drinking water distribution system. The drinking water was delivered to the consumer endpoints such as taps by the stainless steel main pipes (Fig. 1). Three taps with different materials: PVC, stainless steel, and cast iron have been installed in the kitchen of one household since 2001, and all frequently provided cold water for the consumers by high usage from 6 a.m. to 10 p.m. every day. Consequently, mature biofilms were established on the inner surface of the taps. In addition, the household was located about 0.8 km from the drinking water treatment plant.

Fig. 1
figure 1

Schematic diagram of the drinking water distribution system used in this study

Full size image

Five samples were obtained from the drinking water distribution system, which were the inflow water in the clear water tank (CWTW), the outflow water in the main pipe (MPW) shared with the three taps, as well as the biofilms in the taps with three materials: PVC (TPB), stainless steel (TSB), and cast iron (TIB). Water samples were collected in the clear water tank and the main pipe, respectively, using sterile 20-L carboys. Prior to sampling the outflow water (MPW), the tap was run for 10 min to remove standing water in the pipes to obtain representative samples. When sampling, some water quality parameters including temperature, pH, and total dissolved solid were measured using YSI Professional Plus (YSI Incorporated, USA) and free residual chlorine was tested with colorimetry of orthotolidine in situ. Some other parameters including turbidity, CODMn, and nitrate and total plate count were sent to the Pony Testing International Group (Beijing, China) and strictly measured with the standard methods (GB: 5749–2006) in China. A volume of 5 L for each water samples was filtered with 0.22-μm polycarbonate membranes (47 mm diameter, Millipore, MA, USA) to get the microorganisms. Taps with different materials (PVC, stainless steel, and cast iron) were removed to collect the biofilms on them. Photographs were taken using a digital camera (Canon IXUS 220 HS, Japan) to show their shapes (Fig. 2). Both of the filter membranes and the biofilm samples were kept in an ice box during transportation. After reaching the laboratory, the outside surfaces of the taps were rinsed and cleaned carefully using sterile water. Before processing, the tap handles were turned off. Then, the biofilms on the taps were scraped out from the inner surfaces (15 mm inside; 10 cm long) at the connection point with the pipe line using a sterile scraper and then suspended in 5 mL sterile water. The biofilm suspensions were then centrifuged at 10,000 g for 10 min. The pellets were resuspended in 1 mL sterile water in 2-mL tubes (Emtiazi et al. 2004). Finally, all samples were stored at −70 °C before usage.

Fig. 2
figure 2

Tap samples with three different materials: PVC (a), stainless steel (b), and cast iron (c) taken from one household

Full size image

Nucleic acid extraction, cloning, and sequencing of 16S rRNA genes

The filter membranes were aseptically torn into several pieces with forceps and then placed into 2-mL tubes. DNA was extracted using the commercial FastDNA SPIN Kit (MP Biomedicals, Santa Ana, CA, USA) according to the manufacturer′s instructions (Roeder et al. 2010). The extracted nucleic acids were visualized by electrophoresis in a 0.8 % agarose gel in TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM ethylene diamine tetraacetic acid (EDTA), pH 8.3) and kept at −70 °C until use. These DNA preparations were used as template in polymerase chain reaction (PCR).

Bacterial 16S rRNA genes were amplified by the PCR using the forward primer 27 F (5′ AGAGTTTGATCCTGGCTCAG 3′) and the reverse primer 1492R (5′ GGTTACCTTGTTACGACTT 3′). Reactions were performed in a total volume of 50 μL comprising 200 μM of each dNTP, 0.2 μM of each primer, 10× PCR Buffer, and 2.5 U of Taq polymerase (Takara Mirus Bio Corp., Madison, Japan). DNA templates were diluted four times and then added into the reaction. The following conditions were used for DNA amplification: 5 min at 94 °C, followed by 35 cycles of 45 s at 94 °C, 45 s at 56 °C, and 1 min 30 s at 72 °C, and a final extension at 72 °C for 10 min. In addition, reactions containing sterile water (no template) served as negative DNA control. Products were screened by gel electrophoresis in 1 % agarose gel.

All PCR products were further purified with the PCR purification Kit (Fermentas, USA) following the manufacturer's instructions. About 1.5-kb long fragments were ligated with 3 U of T4 DNA ligase and 1× T4 DNA ligase buffer overnight at 4 °C. Cloning reactions were carried out as follows: transformed Escherichia coli (Takara Mirus Bio Corp., Madison, Japan) were spread-plated onto LB agar containing 100 μg/mL ampicillin and incubated overnight at 37 °C (Williams et al. 2004). Selected colonies were screened by carrying out PCR amplification with M13 forward primer (5′ GTTTTCCCAGTCACGAC 3′) and M13 reverse primer (5′ CAGGAAACAGCTATGAC 3′), followed by gel electrophoresis on 1 % agarose gel at 180 V for 20 min. The positive clones were incubated in LB broth containing 100 μg/mL ampicillin for 12 h and then sent to Majorbio Bio-pharm Technology Co., Ltd., China for sequencing.

Phylogenetic and statistical analyses

Sequencing reads were trimmed of vector and inspected for ambiguities using Sequencher 4.1 software (Gene Codes, Ann Arbor, MI, USA). All 16S rRNA gene sequences were examined for chimeras with the Mallard program version 1.02 and screened for putative anomalies (Ashelford et al. 2006). The 16S rRNA gene sequences were affiliated to bacterial taxa using SeqMatch on the RDP (Ribosomal Database Project) website. Multiple sequence alignment and clustering into Operational Taxonomic Units (OTUs) were performed with mothur (http://www.mothur.org), using a 97 % similarity cutoff. Representative clones were selected from each OTU and compared with available rRNA gene sequences in GenBank (http://www.ncbi.nlm.nih.gov/) using the NCBI Blast program. Mothur was also used to generate rarefaction curve and community tree and to calculate richness estimators and diversity index including abundance-based coverage estimator (ACE), Chao 1 and Shannon's and Simpson's indices. The coverage of clone libraries was calculated according to the reference (Moissl et al. 2007) using the formula [1 − (ni / N), where ni is the number of OTUs represented by only one clone (singletons) and N is the total number of the clones examined in each library]. Additionally, the shared OTUs were used to estimate similarity between communities based on membership and structure. For further downstream analysis, Origin7.0 (OriginLab Corporation) was used to draw the figs. The 16SrRNA gene sequences determined in this study are available in the GenBank database under accession numbers JQ923506 to JQ924029.

Results

Water quality analysis

The standard water chemistry and bacteriology data at the time of sampling are summarized in Table 1. All the physical and chemical qualities of drinking water met the national drinking water standard (GB: 5749–2006) in China. The results revealed that total plate count (nutrient agar, 37 °C, 48 h), CODMn, and NO3 −–N slightly increased during transportation. However, free residual chlorine concentration became low, about 20 % of the initial concentration.

Table 1 Summary of water quality at the time of sampling
Full size table

Sequence diversity analysis

A total of 440 partial 16S rRNA gene sequences were analyzed in this study. The sequences were obtained from five different clone libraries representing five different samples: CWTW (81 clones), MPW (86 clones), TPB (85 clones), TSB (85 clones), and TIB (103 clones). Diversity analysis including rarefaction curves, community tree, richness estimators, and diversity indices was used to assess the level of richness, evenness, and community membership of the clones using 97 % as the taxonomic unit cutoff (Table 2). The number of observed OTUs for the five samples was different (CWTW = 23; MPW = 20; TPB = 17; TSB = 28; TIB = 37). The expected OTUs (using Chao 1) were higher than the observed OTUs (CWTW = 36; MPW = 98; TPB = 20; TSB = 43; TIB = 83), suggesting a large degree of bacterial diversity for these five samples. The coverage values (>78 %) confirmed that the core communities were covered in all samples. These results were consistent with comparisons of richness based on calculations using rarefaction curve (Fig. 3). Apparent differences of bacterial diversity were observed between inflow water (CWTW) and outflow water (MPW). The species richness estimators (Chao1 and ACE) showed MPW had higher diversity than CWTW. Metallic materials (TSB and TIB) supported higher diversity than plastic ones (TPB). When OTU pairwise comparisons were performed, it was found that TSB and TIB shared a higher number of OTUs (four shared OTUs) than TPB and TSB (two shared OTUs), and TPB and TIB (two shared OTUs). The number of OTUs shared between groups TPB, TSB, and TIB is one OTU (Venn diagram data not shown). Finally, community tree (Fig. 4) analysis was used to show the similarity of different samples. Clone libraries from two metallic taps resembled each other and clustered together. However, sample TPB clustered with samples CWTW and MPW.

Table 2 Sequence diversity and richness estimators
Full size table
Fig. 3
figure 3

Rarefaction curves of OTUs based on a similarity threshold of 97 %. CWTW water from clear water tank, MPW water from main pipe, TPB biofilms on tap of PVC, TSB biofilms on tap of stainless steel, TIB biofilms on tap of cast iron

Full size image
Fig. 4
figure 4

Cluster analysis of the population profiles from water and biofilm samples to show their relationships according to the Yue and Clayton theta structural diversity measure. CWTW water from clear water tank, MPW water from main pipe, TPB biofilms on tap of PVC, TSB biofilms on tap of stainless steel, TIB biofilms on tap of cast iron

Full size image

Microbial community analysis of drinking water and biofilms

Community profiles were prepared, using these assignments, for each sample (Fig. 5). Proteobacteria represented the majority of microbes within the bacteria community, accounting for 96.3 %, 98.9 %, 91.8 %, 95.3 %, and 87.4 % for CWTW, MPW, TPB, TSB, and TIB, respectively. The Proteobacteria were affiliated to the classes Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria, as well as some unclassified Proteobacteria. Differences in the relative abundance of Proteobacteria classes were observed in the two water samples. For example, Alphaproteobacteria were more abundant in MPW (66.7 %) than in CWTW (27.5 %). In contrast, Beta- (i.e., 20.6 % vs. 30 %) and Gammaproteobacteria (i.e., 11.5 % vs. 38.8 %) were less abundant in MPW as compared to CWTW. The analysis of the microbial community structures in the three tap biofilms suggested that TSB sample was similar to TIB sample, whereas TPB sample was different from TSB and TIB samples. Betaproteobacteria sequences predominated in TSB (56.5 %) and TIB (47.6 %) clone libraries, while these did not appear in TPB clone library. Further evaluation at the order level (Fig. 6) showed that Alphaproteobacteria represented by Sphingomonadales and Rhizobiales in all samples. Moreover, Burkholderiales order was found to be predominant within the Betaproteobacteria. Within the Gammaproteobacteria class, bacterial groups comprised potential pathogen species, including Legionellales, Enterobacteriales, and Pseudomonadales orders. Among these, the biofilms were tested for the presence of Legionella spp. at the samples TPB, TSB, and TIB. The Enterobacteriales were only found at the sample TSB. Additionally, the Pseudomonadales were detected at all samples. Other bacteria included Firmicutes, Acidobacteria, Bacteroidetes, Cyanobacteria, and Deinococcus-Thermus. One clone related to Deinococcus-Thermus was found merely in the CWTW sample. Similarly, one clone that showed high sequence identity (99 %) with Cyanobacteria was barely detected in MPW sample. Also, it was quite evident from this work that bacterial biofilms in TIB sample harbored a diverse group of bacteria. Within Gammaproteobacteria, 11 clones showed 87 % similarity with the Legionellales; thus, we could not exactly confirm them. Acidobacteria and Bacteroidetes were found in TIB samples. Additionally, it was interesting to note that a small proportion of sequences were identified as belonging to unclassified bacteria from TPB (3.5 %) and TIB (10.6 %), respectively. It also should be noted that 17 clones showed high sequence identity (97 %) with iron-reducing bacterium (FJ269052) in TIB samples.

Fig. 5
figure 5

Microbial composition of water and biofilms in drinking water distribution determined by 16S rRNA gene sequence analysis. CWTW water from clear water tank, MPW water from main pipe, TPB biofilms on tap of PVC, TSB biofilms on tap of stainless steel, TIB biofilms on tap of cast iron

Full size image
Fig. 6
figure 6

The taxonomic distribution of Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria and Deltaproteobacteria orders for each sample. CWTW water from clear water tank, MPW water from main pipe, TPB biofilms on tap of PVC, TSB biofilms on tap of stainless steel, TIB biofilms on tap of cast iron

Full size image

Discussion

There have been numerous studies that showed that biofilms can form on the surface of pipelines and revealed the differences among the materials including biofilm formation capacity, fixed bacterial biomass, and morphology feature (Niquette et al. 2000). This study represented a molecular survey using 16S rRNA clone libraries on water and biofilms to describe the microbial communities. In particular, the taps with three different materials have been used for more than 10 years. The biofilms grown on them should be very stable and mature, supplied with the same water source, treated by the same processes, and located in one household with analogous conditions.

Analysis of bacterial diversity suggested that MPW had higher diversity than CWTW. The reason might be involved with the biofilms' growth on the surfaces of pipelines and free chlorine loss in distribution system, which was consistent with variational water quality such as increased total plate count and lower free chlorine (Bachmann and Edyvean 2005; Al-Jasser 2007; Zhang et al. 2008). Comparing TSB and TIB to TPB, higher shared OTUs were retrieved from both metallic materials (stainless steel and cast iron). Moreover, cluster analysis of the population profiles, samples TSB and TIB cluster together and samples TPB, CWTW, and CWTB cluster together. These results suggested that bacterial community seems dependent on the tap materials. However, it has been shown that sample TPB clustered with samples CWTW and MPW, which might be explained by bacteria transmitted from the clear water tank to consumer taps during drinking water distribution.

Our results showed that the microbial communities in all samples consisted principally of Proteobacteria. This is in good agreement with earlier studies from drinking water systems, suggesting that these organisms were well suited to survive in potable water supplies (Kormas et al. 2010; Revetta et al. 2011). When comparing samples of MPW to CWTW, the Alphaproteobacteria increased from 27.5 % to 66.7 %. Beta- and Gammaproteobacteria decreased from 30 % to 20.6 % and 38.8 % to 11.5 %, respectively. Differences in the Proteobacteria percentages within our drinking water might be explained by free chlorine concentration changes during distribution. The proportion of Alphaproteobacteria declined when chlorine concentrations increased (Mathieu et al. 2009). One unexpected finding, however, is the striking lack of Betaproteobacteria among our identified TPB clone library, given that members of this division are usually abundant in freshwaters as well (Manz et al. 1999; MacDonald and Brozel 2000). The reason for this apparent absence is unknown, but it might be related to the habitat characteristic, particularly attached materials (Buesing et al. 2009). (Schwartz et al. 1998) suggested that plastic materials such as PVC, widely used in domestic drinking water distribution systems, appeared to be colonized very frequently by Beta- and Gammaproteobacteria, whereas the Beta subclass was detected in higher percentages on metallic materials (steel and copper).

In addition, environmental strains such as Rhizobiales were commonly found in soil environments, and it was suspected that they were initially present in the water sources (Poitelon et al. 2009). The Sphingomonas belonging to Alphaproteobacteria were isolates from water distribution system tap water, and it is known that the species can survive chlorination by forming bacterial aggregates (Yu et al. 2010). It is intriguing to find Cyanobacteria population detected in MPW when the source of water was surface water. Though most of these organisms depend on light for survival, it was possible that some of the Cyanobacteria populations detected in this study can temporarily survive the darkness under anaerobic conditions (Revetta et al. 2010). Recent studies have shown that the Bacteroidetes are frequently isolated in aquatic environments (Poitelon et al. 2010). Other minor sequences closely related to Acidobacteria and Firmicutes have recently been found in a distribution network. In general, these lineages are widely detected in terrestrial environments or other locations. This suggests that microorganisms might take root in the aquifer supplying the distribution network (Martiny et al. 2005). Nonetheless, the significance of the bacterial groups is not currently discernible and requires much further research.

Molecular analysis of microbial communities clearly showed that the microbial richness and diversity are significantly related to the pipe materials by the order: PVC (TPB) < stainless steel (TSB) < cast iron (TIB). There are some reasons for it. Initially, microbial adhesion to surfaces is a complex process influenced by several physicochemical properties of both microorganism and substratum. Strongly adherent bacteria may play a determinant role in the primary colonization of surfaces, seeding biofilms which will subsequently develop by cellular proliferation and immigration (Simoes et al. 2007). For example, Sphingomonas and other biofilm-associated slime producers, like Rhizobiales bacteria are responsible for bacteria colonization (Pang and Liu 2007). Furthermore, the influence that attached materials exert on bacterial adhesion and biofilm formation is rooted in characteristics such as surface structure and chemical composition (Waines et al. 2011). For instance, roughness of pipe surface and corrosion products affects bacterial attachment on the surface. Hence, biofilm regrowth on pipes made of rough surface materials such as cast iron and galvanized steel was greater than that on smooth surface PVC pipe (Yu et al. 2010; Chowdhury 2011). The stainless steels, as an alloy metal, depend for their corrosion resistance on a thin film, but it may corrode when environmental conditions become aggressive enough to take advantage of the weakness in the film (Percival et al. 1997; Percival et al. 1998). In addition, pipe corrosion contributed to the bacterial regrowth in a distribution system. Conversely, biofilm can also promote corrosion of metals (Teng et al. 2008). Bacteria identified on corroding surfaces encompass a range of species, including sulfate-reducing bacteria (SRB), sulfur-oxidizing bacteria (SOB), and bacteria-secreting organic acids and slime (Dong et al. 2011; Wang et al. 2012). Thus, it was not surprising to find that two OTUs (17 clones) in sample TIB was closely related to iron-reducing bacterium (FJ269052). Based on this study, PVC taps supported a lower level of bacterial diversity than the other two materials, which suggested that PVC may be a potentially suitable material for use as a major component in drinking water distribution. The result agreed with previous studies by traditional culture technology showing that plastic-based material supports less fixed biomass than metal materials (Kerr et al. 1999; Niquette et al. 2000). Finally, to explore the impact of attached materials on biofilms, a model experiment will be designed in the next paper.

The occurrence of pathogenic bacteria in the drinking water distribution environment has been reported previously (Felfoldi et al. 2010; Wingender and Flemming 2011). In this study, pathogenic bacteria, like Legionellales, Enterobacteriales, and Pseudomonadales, were detected. However, it should be noted that it is impossible to determine if the bacteria that we identified represented dead or live cells especially for the pathogens. So the significance of the bacterial groups detected in this study clearly requires much further research. Moreover, in recent years, researchers have reported that amoebas could act as vehicles for multiplication and shelter from antibiotic and disinfection treatments (Hsu et al. 2009; Lau and Ashbolt 2009). Our continuing objective is to elucidate the active members of drinking water microbial communities via reverse transcriptase (RT)-PCR, particularly for pathogen and symbiotic protozoa.