Linking flow velocity-regulated EPS production with early-stage biofilm formation in drinking water distribution systems

Populating harsh and dynamic environments, microbes, including bacteria, fungi and archaea, are prone to colonize on pipeline surfaces to form biofilms in drinking water distribution systems (DWDS) (Douterelo et al. 2014; Potgieter et al. 2018). The occurrence of surface-associated biofilms can deteriorate the distributed water quality with associated undesirable turbidity or odors (Fish et al. 2017), operation malfunction (Zacheus et al. 2001; Bertelli et al. 2018), the spread of pathogens and antibiotic resistant bacteria (Zhang et al. 2018). The DWDS often suffer from hydrodynamic fluctuations from stagnant, laminar to turbulent flow regimes, which intensifies the numerous localized biophysical interactions between individual cells and their micro-habitats (Stoodley et al. 1999; Purevdorj et al. 2002; Guasto et al. 2012; Douterelo et al. 2016; Tsagkari et al. 2017; Tsagkari & Sloan 2018b). However, the knowledge about hydrodynamics-triggered biofilm formation in DWDS remains underappreciated.

Hydrodynamics serves as the most critical factor shaping distribution of nutrients, suspended particulates, dissolved oxygen and external physical stresses on biofilms, accordingly controlling the micro-habitats of microbes and impacting biofilm dynamics (Teodósio et al. 2011; Lemos et al. 2015; Fish et al. 2017). For example, increasing flow velocity was observed to result in lower cell numbers per unit surface area due to higher biofilm detachment by elevated shear stress in a simulated DWDS (Manuel et al. 2007). In contrast, Raya et al. (2010) found that augmenting flow velocity yielded an increasing-then-decreasing trend of the extent of Pseudomonas aeruginosa PAO1 attachment onto clean and smooth surfaces, which may be attributed to the combined effects of hydrodynamics-evoked shear stress and mass transfer. Shen et al. (2015) observed that local flow conditions created by biofilm roughness asperities altered Legionella pneumophila adhesion on the developed DWDS biofilm on account of enhanced interception. Meanwhile, it was also reported that a reduction in bacterial diversity occurred subject to high shear stress, reflective of its functions on slowing down biofilm maturation (Rochex et al. 2008).

Moreover, extensive studies have yielded significant insights into the linkages between hydrodynamics and cellular-induced reactions, including cell motility, nutrient uptake, proliferation, extracellular polymeric substances (EPS) production, and quorum sensing (O'Toole & Kolter 1998; Pratt & Kolter 1998; Harshey 2003; Kirisits et al. 2007; Picioreanu et al. 2007; Guasto et al. 2012; Kim et al. 2016; Tsagkari & Sloan 2018a). Among them, the presence of EPS is essential in the processes of adhesion to substratum surfaces and biofilm formation by promoting survival strategies such as cell attachment to surfaces, protective barrier and nutrient trapping (Flemming & Wingender 2010; Lorite et al. 2013). An increase in hydrodynamic shear has been shown to stimulate bacterial exopolysaccharide secretion (Liu & Tay 2002). However, the Pseudomonas fluorescens biofilms under different flow regimes expressed distinct physiological characteristics, where turbulent flow promoted formation of higher cell density biofilms with lower amounts of matrix polysaccharides than laminar flow that generated low shear stress (Simões et al. 2007). Fish et al. (2017) observed that the biofilms developed under low-varied flow conditions had the lowest amounts of biomass and the greatest EPS volumes per cell when compared with those incubated at greater flow variations in an experimental DWDS system. Despite these important advances, the underlying mechanisms of early-stage DWDS biofilm linked to hydrodynamics-stressed EPS production are not clear, which would help unravel the complexity of hydraulics regulation on biofilm dynamics.

This study examines how flow velocity fluctuations could affect early-stage DWDS biofilm formation through determining the changes in bacterial cells, EPS production and surface properties along with variations in biofilm bacterial community with the aim of enhancing our knowledge of biofilm formation processes in DWDS.

The experiments were carried out in a closed DWDS, as illustrated in Figure S1 (Supplementary Material), consisting of a 3 L tank and a 10 m polyethylene (PE) pipe (inner diameter of 6 mm), where removable pipe sections (4 cm long) were inserted in the middle for biofilm collection. The water flow and internal velocities of pipelines were driven through peristaltic pumps (Lange BT300-2 J, Lange BT600-2 J, Baoding, China) for low or high velocity conditions, respectively. The pipelines and biofilm collectors were rinsed with 70% ethanol and then the system was disinfected by flushing using 1.0 mg/L sodium hypochlorite solution for 2 h (Soini et al. 2002) prior to experiments. Water samples used in the experiment were collected from a domestic drinking water tap in Hefei, China, and stored at room temperature (20 ± 1 °C) for 24 h to diminish chlorine residual before use.

To evaluate the impacts of flow velocity on early-stage biofilm formation, five parallel experimental systems were employed with various internal velocities of 0.1(laminar flow), 0.5 (transition flow), 1.0, 1.5, and 2.0 m/s (turbulent flow), respectively, mimicking the flow fluctuations regularly occurring in DWDS. At these velocities, the Reynolds numbers were 594, 2,970, 5,940, 8,910 and 11,880, respectively. Biofilm and bulk water samples were collected at 3 h, 6 h, 12 h, 24 h and 48 h after inoculation for estimating attached and suspended cell density, protein (PN) and polysaccharide (PS) contents, zeta potential and hydrophobicity of suspended cells, along with biofilm bacterial community analysis.

After wiping the outside surfaces with 70% ethanol, the biofilm collectors were gently washed three times with distilled water to eliminate unattached cells (LeChevallier et al. 1988; Percival et al. 1998). The collectors were then aseptically put into a 15.0 mL sterilized tube preloaded with 10.0 mL of phosphate buffer saline (PBS). Sonication (KQ5200DE, Shumei, Kunshan, China) at 35 kHz for 10 min was conducted to gather the suspension of attached cells in the tube (Manuel et al. 2007). In the meantime, 30.0 mL of the water sample from the systems were transferred to a 50.0 mL sterile centrifuge tube for further analysis.

Extracellular PS and PN were extracted from cell suspensions of biofilm and planktonic samples and analyzed via a modified heat method (Li & Yang 2007). Fifteen-millilitre cell suspension were transferred into a test tube and heated in a water bath at 70 °C for 1 h. Subsequently, the suspensions were sheared by a vortex mixer (CJ78-1, Jintan, China) for 1 min, followed by centrifugation at 4,000g (5804, Eppendorf, Germany) for 15 min. After filtrating the supernatant through 0.45 μm filters, PN and PS were quantified using a UV/VIS spectrophotometer (Unico, Shanghai, China) at 750 nm in accordance with Liu et al. (2017). The electrophoretic measurements of suspended cells were measured by a zeta-potential analyzer (Nano-ZS90, Malvern, UK). Surface hydrophobicity was estimated with the method developed by Rosenberg et al. (1980).

A confocal laser scanning microscope (CLSM; LSM710, Carl Zeiss, Germany) was used to characterize the physical structure of the biofilm. Specifically, the biofilm collectors were cut into three strips (1.5 cm × 0.5 cm), then transferred to PBS solution, and stored at 4 °C until analyzed. The collectors were stained with 100.0 μL SYTO63 solutions (20 μmol/L, Thermo Fisher Scientific, OR, USA) and incubated for 30 min to bind to biofilm cells. Afterwards, proteins were stained with 100.0 μL FITC solution (10 g/L, Sigma, MO, USA) for 1 h. For α-polysaccharide staining, 100.0 μL Con A (250 mg/L, Thermo Fisher Scientific, OR, USA) was added and incubated for 30 min. Finally, 100.0 μL calcofluor white solution (300 mg/L, Sigma, MO, USA) was added and incubated for 30 min to bind to β-polysaccharide. After each staining, samples were rinsed three times with PBS and then allowed to dry naturally (Chen et al. 2007; Jiang & Liu 2013). The stained samples were visualized using a ×20 objective and analyzed with ZEN 2009 light edition confocal software.

The collected biofilm samples were placed into a 50.0 mL centrifuge tube containing sterile PBS (pH 7.0) and centrifuged at 4,000g for 5 min, followed by resuspension of the pellets. After repeating these steps three times, the total DNA extraction was performed using the E.Z.N.A Mag-Bind Soil DNA Kit (OMEGA, MO BIO Lab. Inc., Carlsbad, CA, USA) according to the manufacturer's instructions. The high-variation region of the 16S rRNA gene was subjected to polymerase chain reaction (PCR) amplification using universal primers (F-27FAGAGTTTGATCMTGGCTCAG) and (R2-CAAGTATTACCGCGGCTGCTGGCAC), and the V1–V3 regions were selected as the amplification regions. PCR products were pooled and purified using the Qubit2.0 DNA detection kit (Qubit^® ss DNA Assay Kit, Life Technologies). Biofilm bacterial communities and their relative abundance and diversity were analyzed by a high-throughput sequencing method (pyrosequencing) (Douterelo et al. 2012), which was conducted by Sangon Biotech Company (Sangon, China) with the Illumina Miseq platform (Illumina, Inc., San Diego, CA, USA). The sequence-similarity-based method was used to classify the sequences into different operational classification units at 97% similarity level using QIIME software.

All experiments were conducted at least in triplicate and the results were presented with average and standard deviations. Statistical differences were analyzed with the one-way analysis of variance (ANOVA) test, which was based on a confidence level equal to or higher than 95%. The Pearson correlation test was conducted to evaluate correlation coefficients (r) among EPS amounts, zeta potential, hydrophobicity and attaching ratios.

Figure 1 depicts the HPC and TCC variations on pipe surfaces and in bulk water across various flow velocities. After 48 h incubation, significant differences (p < 0.001) were found in biofilm cell densities. A biofilm-embedded TCC of 4.88 ± 0.03 lg(cells)/cm² was estimated at the low velocity of 0.1 m/s (Figure 1(a)), which rose up to 6.05 ± 0.04 lg(cells)/cm² responding to 1.0 m/s velocity. Elevating the velocity to 2.0 m/s yielded a gradual drop of biofilm TCC to 5.44 ± 0.11 lg(cells)/cm². Correspondingly, an opposite significant effect of flow velocity on suspended-cell growth was observed compared with biofilm formation (p < 0.001). Specifically, the suspended cell population was estimated as decreasing in TCC value from 4.60 ± 0.01 to 4.34 ± 0.04 lg(cells)/mL with the velocity increasing from 0.1 to 1.0 m/s and rising to 4.97 ± 0.02 lg(cells)/mL at 2.0 m/s (Figure 1(b)). It resulted in a similar pattern of the biofilm/total cells ratio as compared with that of biofilm cells, i.e., the TCC ratio increased significantly from 56.86% ± 1.02% at 0.1 m/s velocity to upwards of 96.91% ± 2.26% at 1.0 m/s, followed by an apparent reduction to 66.11% ± 19.29% at 2.0 m/s velocity (Figure 1(c)). Similarly, the intermediate velocity of 1.0 m/s also yielded the highest HPC attaching ratio of 45.67% ± 6.56%.

The results conformed to the previous observations of enhanced biofilm formation at elevated velocity within a certain range. For example, in both copper and plastic pipes, the biofilm cells increased with elevated flow rates up to 0.8 L/min in a pilot distribution system (Lehtola et al. 2006). As for stainless steel grades 304 and 316, higher velocities (0.96 m/s and 1.75 m/s) also yielded augmented viable and total cell counts of biofilms, associated with higher levels of extracellular polysaccharide, as compared with a lower rate of 0.32 m/s (Percival et al. 1999). Pseudomonas fluorescens was found to increase adhesive strength as the fluid velocity increased in the range of 0.6–1.6 m/s (Chen et al. 2005). This may be attributed to facilitated mass transport and excited bacterial activity at elevated flow velocity (Wang et al. 2014). Flow regimes could affect the formation and destination of aggregates that are crucial in seeding biofilm formation. Tsagkari et al. (2017) reported that aggregation was promoted in turbulent flow compared with laminar flow; inoculating with Methylobacterium further enhanced the aggregation. Tsagkari & Sloan (2018b) observed that different flow regimes resulted in substantial changes in biofilm morphology and growth and specifically the turbulent regime elevated both the initial formation and development of biofilms with the greatest roughness on the accessible surfaces compared with laminar and transition regimes. Instead of enhancing biofilm formation potentials, excessive shear forces at high velocities retarded bacterial activity and encouraged the detachment of biofilm cells, thereby contributing to reduced biofilm formation (Raya et al. 2010). In addition, Purevdorj et al. (2002) suggested that the dilution of signal molecules by mass transfer effects in faster flowing systems modified their influence on biofilm structure.

Evidence showed that nutrient limitation, toxic chemicals, hydrodynamic shear force and other stressful conditions could enhance microbial EPS production for self-protection (Flemming & Wingender 2010). After 48 h incubated in the experimental systems, the biofilms were collected and analyzed using a CLSM. As illustrated in Figure 2, the biofilms were dominated by EPS rather than cells (blue), while the EPS was predominantly composed of proteins (red) and α-polysaccharides (green) along with particularly sparse β-polysaccharides (yellow), regardless of hydrodynamic conditions. Interestingly, greater cell adhesion and more extensive EPS were observed for biofilms grown at 1.0 m/s velocity compared with those at lower (0.1 m/s) or higher (2.0 m/s) velocities.

To further quantify hydrodynamics-stressed EPS release and its impacts on biofilm formation in a DWDS environment, the EPS amounts (characterized in terms of PN and PS) at various flow velocities were analyzed, with results depicted in Figure 3. Statistically, flow velocity affected EPS production significantly (p < 0.001). With velocity increasing from 0.1 to 1.0 m/s, the shear-induced EPS release of biofilm cells at 48 h was observed with an increase of PN content from 8.99 ± 0.11 to 9.78 ± 0.09 μg/lg(cells) and of PS amount from 1.46 ± 0.09 to 1.64 ± 0.09 μg/lg(cells) (Figure 3(a)). In contrast, a decay in PN and PS production to 8.77 ± 0.21 and 1.13 ± 0.04 μg/lg(cells) was observed at 2.0 m/s velocity. The suspended cells experienced similar EPS variations as compared with biofilm cells in response to flow variations. During 48 h incubation, biofilm and suspended cells exhibited an analogous variation pattern to the PN and PS production rates, i.e., the rates increased and then decreased with velocity increasing, reaching peak PN and PS rates of 0.20 and 0.07 μg/lg(cells)·h for biofilm cells and 0.07 and 0.02 μg/lg(cells)·h for suspended cells at 1.0 m/s, respectively (Figure 3(b)). Our results revealed that the intermediate velocity level (1.0 m/s) considerably promoted EPS production, while at high velocity levels, inhibition of cell activity reduced EPS release.

Statistical analysis showed a positive correlation between EPS amounts of biofilm and attaching ratios (r = 0.88, p < 0.05) reflecting favored surface aggregation by EPS production at intermediate velocity in DWDS. The mechanism of environment-stressed EPS production shaping cell–surface interactions and thereby surface aggregation is consistent with previous reports of pure culture conditions and a two-species system. For instance, Ducret et al. (2012) reported that slime secretion of gliding Myxococcus xanthus cells was associated preferentially with the outermost components of the motility machinery and promoted cell adhesion to the substrate. For the photosynthetic bacterium Chloroflexus aggregans, extracellular protease secreted from Bacillus licheniformis mediated interspecies interaction and encouraged its aggregation (Morohoshi et al. 2015).

Surface potential and hydrophobicity are generally considered the triggering forces for surface adhesion. Figure 4 depicts the zeta potentials and hydrophobicity of suspended cells 48 h after inoculation. With velocity elevating from 0.1 to 1.0 m/s, zeta potential value decreased from 17.34 ± 1.02 to 11.62 ± 1.57 -mV, accompanied with hydrophobicity boosting from 52.90% ± 4.38% to 58.16% ± 2.19%. Interestingly, a continuous elevation to 2.0 m/s yielded an ascending zeta potential back to 17.60 ± 0.62 -mV associated with a hydrophobicity of 37.54% ± 2.50%. For suspended cells, zeta potential and EPS amounts were negatively correlated with a coefficient of −0.986 (p < 0.01), albeit a positive moderate correlation between surface hydrophobicity and EPS contents (r = 0.729). Similarly, a higher negative charge density was found to be linked to the reduced production of PS and PN induced by D-tyrosine (Xu & Liu 2011). Cell-surface-attached EPS components encompassing an assortment of usually charged functional groups could considerably impact the physicochemical properties of cell surfaces (Tielen et al. 2010; Das et al. 2011).

A direct negative correlation between zeta potential and attaching ratio (r = −0.928, p < 0.05) was found at various velocities. The lower zeta potential associated with augmented EPS production corresponded to enhanced biofilm formation at intermediate flow velocity, reminiscent of the classical DLVO theory that suppression of electrorepulsive forces facilitated cell–cell and cell–surface interactions (Zita & Hermansson 1994). Consideration of augmented hydrophobicity provides an alternative explanation for promoted adhesion behavior at intermediate flow velocity. Previous studies reported that presence of naturally produced extracellular DNA on both Gram-negative and Gram-positive bacterial cell surfaces affected surface hydrophobicity and promoted attractive acid–base interactions between cells and surfaces (Das et al. 2010, 2011).

The Shannon diversity index for biofilm bacterial communities revealed a community shift from high (2.45) to low (1.70) diversity at velocities from 0.1 to 1.0 m/s followed by an increase in diversity (4.47) at 2.0 m/s velocity. It suggests that DWDS bacterial diversity could be lowered by intermediate flow velocity (1.0 m/s). The high diversity at lower or higher velocity (lower flux or higher shear stress) indicates a high metabolic versatility of bacterial communities, which would allow them harsh-environment adaptation (Lautenschlager et al. 2014). Meanwhile, surface colonization was lowered accompanying such uneven community. Accordingly, the existence of a higher diverse community may lead to higher functional stability confronting environmental stresses (Wittebolle et al. 2009).

Shifts in biofilm bacterial community were recorded under various velocities at phylum level (Figure 5). Proteobacteria dominated within all sampled biofilm community, and has been repeatedly detected as predominant in drinking water systems (Williams et al. 2004; Lautenschlager et al. 2014; Lin et al. 2014). The abundance of this phylum increased from 55.11% to 78.16% with the velocity rising from 0.1 to 1.0 m/s followed by a decrease to 62.78% at 2.0 m/s. It suggests that intermediate velocity stimulates the dominance of Proteobacteria in a DWDS. The second abundant phylum was Actinobacteria, accounting for 42.27%, 20.62% and 32.31% at the velocities of 0.5, 1.0 and 1.5 m/s, respectively, while Firmicutes occupied the second large phylum with a content of 37.50% at 0.1 m/s and of 32.78% at 2.0 m/s. Sampling from an urban water distribution system in Greece, Kormas et al. (2010) observed that Proteobacteria dominated in the bacterial community of the water samples collected from pumping wells and water treatment tanks, while Actinobacteria was detected as predominant in household waters. Figure 6 depicts the variability of biofilm composition at genus level along with hydrodynamics fluctuations. Specifically, Pseudomonas (24.13%), Rhodococcus (19.49%) and Aquabacterium (15.05%) were the most dominating groups at 0.1 m/s velocity. Confronting 1.0 m/s flow environment, the major groups changed to Pseudomonas (37.69%) and Rhodococcus (32.17%). In contrast, Rhodococcus (41.95%) was dominant followed by Pseudomonas (29.31%) in the biofilm compositions at 2.0 m/s velocity. Overall, Pseudomonas was the genus dominant in the composition of most sampled biofilms, particularly at 1.0 m/s velocity. These species have higher capacity to form biofilm due to their enhanced ability of EPS secretion, thereby favoring biofilm formation (Douterelo et al. 2014). Cyclic patterns in planktonic composition have been observed associated with hydrodynamics fluctuations in a small, looped water distribution system (Sekar et al. 2012). Douterelo et al. (2014) suggested similar planktonic species were involved in the process of surface attachment and initial biofilm formation.

Results showed that intermediate velocity (1.0 m/s) yielded increasing biofilm cells, associated with enhanced EPS production. Correlation between EPS release, zeta potential and hydrophobicity of suspended cells suggested that EPS release might alter surface properties, thus affecting microbial aggregation onto pipe surfaces. Shifts in biofilm bacterial community were observed along with hydrodynamics fluctuations. Proteobacteria and Actinobacteria dominated the biofilm samples incubated at the velocities of 0.5, 1.0 and 1.5 m/s, while Proteobacteria and Firmicutes were the most abundant at 0.1 and 2.0 m/s after 48 h incubation. Pseudomonas was the genus predominant in the composition of most sampled biofilms, particularly at 1.0 m/s velocity. Overall, this study offers insights into microbial responses and the mechanistic understanding of biofilm dynamics populating the hydrodynamically fluctuating DWDS environment.

This study was financially supported by the Natural Science Foundation of China (51479046) and the National Key Research and Development Program of China (2016YFC0401303).

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/ws.2020.039.