Identification and inactivation of Gordonia, a new chlorine-resistant bacterium isolated from a drinking water distribution system

With the emergence of disinfectant-resistant microorganisms, concerns over the safety of water supplies have increased. At present, chlorine treatment is a universally used technology owing to its reliability and relatively low cost. Chlorine-based disinfectant has been used in the forms of free chlorine and chlorine compounds – e.g., liquid chlorine, hypochlorite, and bleaching powder – in disinfection processes for water plants. Yet excessive chlorine use also promotes selection for chlorine tolerance. To ensure that disinfection is sustained, high chlorine concentrations must be maintained in finished water. An environment with surplus chlorine in water distribution systems provides survival selective pressure and selects for disinfectant-resistant cells (Falkinham 2015).

An increasing number of studies have shown the inefficacy of chlorine in pathogenic microorganism control in aquatic environments, especially in wastewater systems. Not only have Citrobacter and many Bacillus cereus species been isolated from wastewater treatment plants (Mir et al. 1997; Paes et al. 2012; Mojisola & Anthony 2017), but various reports have confirmed that microbial biomass decreases with increases in disinfectant-resistant bacteria after wastewater disinfection treatment (Diehl & Lapara 2010; Coronel-Olivares et al. 2011; Burch et al. 2013; Mojisola & Anthony 2017).

Chlorine-resistant bacteria have also been discovered in drinking water, which threatens public health. Some chlorine-resistant bacteria are pathogenic or conditionally pathogenic, such as Mycobacterium tuberculosis, Legionella pneumophila, Pseudomonas aeruginosa, and Staphylococcus aureus (Falkinham et al. 2001; Chen et al. 2012; Gomes et al. 2016). Additionally, chlorine-resistant bacteria have led to the regeneration of microbes in the urban pipe network and reduced the biological stability of drinking water. Yet studies on chlorine-resistant bacteria in the water supply system remain scarce compared with those focusing on sewage treatment systems. In particular, few studies have focused on the isolation of chlorine-resistant planktonic bacteria in urban water supply systems (Goncharuk et al. 2014).

Moreover, the quantification of microbes has traditionally been performed using plate count methods, but these have great limitations in the detection of chlorine-resistant bacteria. In fact, bacteria may enter a viable but non-culturable (VBNC) state and not be detected by traditional plate count methods when exposed to chlorine-based disinfectant. Consequently, the health risk of chlorine-resistant bacteria may be underestimated. The presence of VBNC bacteria following the disinfection process has been confirmed in previous studies (Li et al. 2014; Zhang et al. 2018). VBNC cells were found to remain animate as a result of cellular metabolic activity but were unable reproduce and form colonies in conventional culture (Oliver 2000). Hence, they may go undetected during routine colony counting. These bacteria can restore their reproductive capacity under appropriate conditions, and a considerable proportion of chlorine-resistant bacteria can enter the VBNC state under chlorine stress (Zhang et al. 2018). Therefore, detection methods for microorganisms in different states must be employed during the disinfection process. Recent flow-cytometric technological advancements provide rapid and accurate data for bacteriological determination. A cultivation-independent approach using flow cytometry (FCM) in combination with fluorescent probe technology allows for the enumeration and functional classification of the physiological state of bacteria at the single-cell level (Hammes et al. 2011). All of the bacteria, including VBNC bacteria during the disinfection process, can be collected and detected by FCM.

In China, the Yellow River flows through nine provinces, and it is one of the most important water sources in the basin. To our knowledge, there have been no reports on chlorine-tolerant bacteria in the water supply system of the Yellow River Diversion Project. In this study, we examined a method for screening and separating chlorine-resistant bacteria in the drinking water distribution system taking Yellow River as water source, and we obtained a chlorine-resistant bacterium (Gordonia JN724) identified as belonging to the genus Gordonia. Some members of the genus Gordonia are opportunistic pathogens that can cause respiratory tract diseases and other related illnesses. The abundant reproduction of Gordonia with pigment may lead to high colony counts, color, and turbidity in tap water if environmental conditions are suitable. Moreover, as there have been no reports on the chlorine resistance of Gordonia in aquatic environments, we thoroughly investigated it here. We used Escherichia coli, which is relatively sensitive to chlorine, as a comparison strain for chlorine-resistance analysis. Analytical methods used in this study included flow cytometry, scanning electron microscopy, and heterotrophic plate counts (HPC). To explore the effective countermeasures, we examined the inactivation of Gordonia JN724 via aqueous chlorine dioxide (ClO₂) and ultraviolet (UV) radiation.

Gordonia JN724 was isolated from tap water samples of water distribution system taking Yellow River as water source. Tap water was collected in triplicate along the distribution pipeline of the Quehua water treatment plant in Jinan City, China. The sampling site was 10 km from the water treatment plant. The sampling procedures were in accordance with standard examination methods for water collection and preservation (GB/T 5750.2-2006). Two liters of water were collected in pre-cleaned and sterilized glass bottles, and the samples were refrigerated during transportation and storage. The physical properties of the collected water samples are summarized in Table 1. Membrane filtration was used to isolate the bacterial cells within 4 h. The target water sample of 1 L was filtered with a polycarbonate membrane filter (0.22 μm, Millipore, UK) for cell harvesting. After filtration, the membrane was placed into a tube with 5 mL of phosphate buffered saline (PBS; 0.1 M, pH = 7.2). The bacterial cells attached on the membrane were separated out by shaking on a NP-30S vortex shaker (Suzhou Jiulian Technology Co. Ltd, Suzhou, China) for 5 min at 2,500 rpm. After that, the eluent was taken out and put into a new tube for cell collection. Fresh PBS was added to the tube for repeated vibration and elution. The subsequent eluent was combined with the previous one and centrifuged at 5,000 g for 10 min to collect the cells. After centrifugation, the supernatant was discarded and the resulting precipitate was resuspended in nutrient bouillon medium and incubated at 37 °C for 24 h. The following day, an appropriate amount of sodium hypochlorite (theoretical value of the final concentration calculated by free chlorine, 9 mg L⁻¹) was added to the eluent (initial screening of chlorine-resistant bacteria). After 2 h of inactivation, when the concentration of free chlorine is guaranteed to exceed 0.5 mg L⁻¹, the microorganisms were isolated using conventional plating methods on R₂A medium. Plates were incubated aerobically at 22 °C for 7 d for single colonies. Sodium hypochlorite was used as a disinfectant for the screening, isolation, and disinfection of the bacteria. Free chlorine was determined by the spectrophotometric N,N-diethyl-p-phenylenediamine method (Chen et al. 2012).

The separated colony that had been purified three times was used for chlorine resistance verification. The isolated strain was added to 5 mL of nutrient broth and cultured at 37 °C for 24 h. Bacteria were separated by centrifugation as before (at 5,000 g for 10 min) and washed twice with PBS to remove the remaining culture supernatant. Harvested cells were resuspended in 10 mL of PBS for bacterial suspension preparation. Sodium hypochlorite was added to the suspension for a final dose concentration of 15 mg L⁻¹ (calculated by free chlorine, reconfirming the chlorine resistance of the isolated culture). After 2 h of inactivation, sodium thiosulfate was used to terminate the reaction and the R₂A agar plate count method was used to determine the presence of chlorine-resistant bacteria. The resulting single colony was preserved using a HBPT001-1 bacterial bead preservation kit (Qingdao Hi-Tech Industrial Park Hope Bio-technology Co., Ltd, Qingdao, China) for subsequent experiments.

For the identification of 16S rRNA, bacterial cells were grown in R₂A medium at 22 °C for 7 days. Genomic DNA was extracted using a bacterial genomic DNA extraction kit (Biotech, China) according to product instructions. The universal primers 8F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACACTT-3′) were used for 16S rRNA gene amplification (Sun et al. 2013). PCR reactions were performed using a thermal cycler (Applied Biosystems, USA) in a total volume of 25 μL. Each reaction consisted of 2.5 μL 10 × Buffer (with Mg²⁺), 1 μL of 2.5 mM dNTP mix, 0.2 μL DreamTaq DNA Polymerase (Thermo Fisher Scientific, Waltham, USA), 0.5 μL of 10 μM 8F and 1492R primer, 0.5 μL of sample DNA and 19.8 μL of PCR-grade water. The PCR amplification conditions were as follows: an initial activation step at 94 °C for 4 min; and after heating, DNA was amplified for 30 cycles at 94 °C for 45 s, 55 °C for 45 s, and 72 °C for 60 s, with a final extension at 72 °C for 10 min. The products were gel extracted and sequenced with a 3730XL DNA analyzer (Applied Biosystems, USA). Nucleotide sequences were submitted to the BLAST search engine at NCBI GenBank and identified through similarity values (Tamura et al. 2007). The 16S rRNA gene sequences from chlorine-resistant bacteria in the previous reports (GenBank accession numbers: AJ536039 and AF480586) were also retrieved from GenBank and used as outgroups. Multiple sequence alignments were carried out by Clustal X program (Thompson et al. 1997). A phylogenetic tree was constructed by MEGA software (Molecular Evolutionary Genetics Analysis software, version 5.05), using the neighbor-joining method (Tamura et al. 2011) to conduct the bootstrap analysis (1,000 replicates).

Bacterial suspension of JN724 after the exponential phase was prepared as described in section ‘verification of chlorine resistance for the isolate’. The isolated strain was added to the nutrient broth and cultured at 37 °C for 24 h. Bacteria were separated by centrifugation as before and washed twice with PBS to remove any remaining culture supernatant. Harvested cells were resuspended in PBS for bacterial suspension preparation. Escherichia coli (CICC 10899), obtained from China Center of Industrial Culture Collection (CICC), was chosen as a representative for chlorine-resistance analysis. The culture and bacterial suspension preparation techniques for E. coli were the same as those for JN724. In line with previous work (Falkinham et al. 2001), cells were resuspended in PBS with an initial cell density of 10⁶ CFU mL⁻¹ for disinfection treatment.

To better understand chlorine resistance, E. coli was used as a comparison strain in the disinfection experiments. All of the inactivation tests were conducted in closed glassware, which had been soaked in sodium hypochlorite overnight, rinsed thoroughly, and sterilized by autoclaving. The initial densities of bacteria employed for the disinfection test were 10⁶ CFU mL⁻¹ in PBS. First, 500 mL of the suspension was added to the glassware. Sodium hypochlorite was then introduced to the bacterial suspension with an equilibrium content of 2.5 mg L⁻¹ as free chlorine (for reconfirmation of chlorine resistance). Next, the samples were incubated in a shaking incubator at 25 °C and 100 rpm for 4 h. During the incubation period, the samples were collected from the reactor and counteracted with excess sodium thiosulfate instantaneously at time intervals of 5, 10, 20, 30, 60, 120, and 240 min after inactivation for the HPC, free chlorine analysis and flow cytometry, each in triplicate. The efficacy of disinfectants was determined via assays of the initial and residual concentrations of microbes. Bacteria were estimated following dilution by the plating method (R₂A plates, incubated aerobically at 22 °C for 7 d). Two controls were set up in the experiments: bacteria in PBS without disinfectant and only PBS with disinfectant. The results of three separate trials were used for calculations.

Observations of morphology were conducted using scanning electron microscopy (SEM) (VEGA3 TESCAN, Brno, Czech Republic). To detect morphological changes in the cell, treated and untreated bacteria suspensions were pelleted by centrifuging as before. After centrifugation, the bacteria were fixed with glutaraldehyde and dried with isoamyl acetate for SEM examination. Sample pretreatments for SEM analysis were performed as described by Wen et al. (2017).

For the flow-cytometric detection, the Cell Viability Kit (BD™, USA) was used to acquire detailed information on the damage to the cell membrane after chlorine treatment. The two nucleic acid stains used were thiazole orange (TO) and propidium iodide (PI), which have different cell permeability properties that can be used to differentiate cells with different membrane integrities. TO, a permeant dye that can enter all cells and bind to nucleic acid, was used to estimate the total number of cells. PI can penetrate the cell membranes of dead cells while being excluded from live cells. The combination of TO and PI provides a reliable method for quantifying cell membrane damage. A 500-μL tested sample was mixed with 5 μL of 42 μmol L⁻¹ TO and 4.3 mmol L⁻¹ PI prepared in dimethyl sulfoxide (DMSO). It was then incubated in the dark for at least 5 min at room temperature. Liquid counting beads were also added to the tested samples for absolute counting before FCM analysis. Samples were used for FCM analysis (FACSCanto II, BD Biosciences, Franklin Lakes, USA) with 488 nm laser excitation. Data were visualized and obtained from the dot plots of green fluorescence signals versus red fluorescence signals using FACSDiva software (BD Biosciences, Franklin Lakes, USA).

The primary initial cell concentration was about 10⁶ CFU mL⁻¹ in PBS for both the chlorine dioxide and UV treatments.

The chlorine dioxide inactivation experiment was performed in a similar manner to the chlorine treatment. First, 500 mL of the prepared suspension was added into the treated glassware. To maintain consistency in the effective chlorine concentration, the primary dose of chlorine dioxide was 1 mg L⁻¹. The glassware was then placed in a shaking incubator at 25 °C and 100 rpm for 30 min. During the incubation period, the samples were collected from the reactor and counteracted with excess sodium thiosulfate instantaneously at time intervals of 5, 10, 20, and 30 min after inactivation for HPC analysis, each in triplicate. The residual chlorine dioxide concentration was determined through spectrophotometric N,N-diethyl-p-phenylenediamine colorimetric analysis (Chen et al. 2012). Bacteria in PBS without disinfectant were used as the control in the experiment, and the results were calculated from three separate trials.

The UV irradiation test was carried out using a collimated beam device installed with a low-pressure mercury lamp (R-CANS463RL 40 W, λ = 254 nm). Irradiance was determined by a radiometer (International Light Technologies, ILT 2400). The intensity of the incident irradiance was approximately 0.168 mW cm⁻² at the center of the exposure surface. After the lamp was warmed up, a Petri dish was positioned vertically under the collimated beam. A microbial suspension of 40 mL was added to a glass Petri dish reactor, which had a diameter of 60 mm, and the bacterial suspension was stirred. The suspension was irradiated with UV dosages of 5, 10, 15, 20, 40, 80, and 120 mJ/cm² at room temperature. Triplicate samples were taken out at scheduled times. The required exposure times were calculated by dividing the desired UV dose by the average UV radiance (Bolton & Linden 2003). The number of viable bacteria was also determined from R₂A agar plates, which had been incubated as described in section ‘Chlorine disinfection’. Results were calculated from three separate trials.

The isolates were identified by physiological, biochemical, and molecular biological techniques. Table 2 shows the parameters related to colony morphology, as well as the physiological and biochemical characteristics of JN724. The phylogenetic tree was constructed with the 16S rRNA gene sequences of JN724 and representative sequences of the genus Gordonia. The JN724 strain was found to be closely related to both Gordonia terrae and Gordonia hongkongensis, with which JN724 showed 99% homology (Figure 1). At the same time, the high chlorine resistance of Gordonia JN724, which was isolated from the drinking water distribution system, was confirmed. Although Sphingomonas, Mycobacterium tuberculosis, Legionella pneumophila, Pseudomonas aeruginosa, Staphylococcus aureus, and other species with high chlorine resistance have been isolated and identified from drinking water systems (Mir et al. 1997; Chen et al. 2012; Sun et al. 2013; Gomes et al. 2016), there have been no reports on this highly chlorine-resistant strain of Gordonia. In this study, chlorine-resistant JN724 was obtained from the water supply system for the first time. For the screening of chlorine-resistant bacteria in the local urban water supply system, we found that the occurrence frequency of Gordonia was relatively high, although it was not accurately quantified. This is one of the reasons why it was chosen as the object of the study.

It is found that the diversity of isolated bacteria may vary due to different media or isolated technology (Guan et al. 2020). In order to minimize the impact of pre-incubation, future research should focus on the two aspects: First, after further enrichment and concentration, the chlorine-resistant strains may be screened by direct chlorination treatment without pre-culture. Second, it is necessary to choose different types of media and culture conditions for meeting the growth needs of more strains in the pre-culture according to previous studies (Schroer et al. 2020; Guan et al. 2020).

E. coli is a microorganism susceptible to chlorine at the normal dosage required for water disinfection, whereas the isolated strain in this study is known to be chlorine resistant. Consequently, we anticipated a significant difference in chlorine demand and inactivation effects between the two strains. For comparisons of chlorine consumption, disinfectants were mixed into the microbial suspension at an identical initial concentration of 2.5 mg L⁻¹ as free chlorine. The primary initial cell concentrations for Gordonia JN724 and E. coli were 4.0 × 10⁶ CFU mL⁻¹ and 3.6 × 10⁶ CFU mL⁻¹, respectively. Free chlorine concentration and chlorine demand were measured and compared. Chlorine demand was expressed as the difference between the concentration of free chlorine before treatment and the amount of chlorine residual at a given exposure time (Helbling & Vanbriesen 2007).

The inactivation kinetics of Gordonia JN724 and E. coli exposed to chlorine are illustrated in Figure 2. For both species, the concentration of free chlorine decreased as contact time increased. In the first 5 min, the chlorine concentration decreased very quickly, accompanied by a rapid inactivation velocity. Compared with Gordonia JN724, chlorine consumption rate of E. coli was significantly faster and chlorine demand was greater at the end of the experiment (Table 3). In order to determine the effect of organic matter on chlorine consumption, total organic carbons in the supernatants of Gordonia JN724 and E. coli were measured, and it was shown that there was no significant difference. The faster chlorine consumption rate of E. coli indicated that E. coli contained more cellular substance that could easily react to the chlorine than Gordonia JN724 (Chen et al. 2012). In the present study, a higher chlorine demand was observed in the E. coli treatment than in previous studies (Zhao et al. 2001). The ultimate chlorine demand of free chlorine exceeded 1 mg/L for 99.99% inactivation. It was previously reported that chlorine consumption would increase with the growing initial dosage of chlorine disinfection in E. coli treatment (Helbling & Vanbriesen 2007). The free chlorine concentration used in the present study was much higher than those in other studies, and that may have been a reason for the observed high chlorine consumption. Despite the continuous free chlorine consumption, there was no significant progress in the Gordonia JN724 inactivation. The subsequent depletion of residual chlorine may be interpreted as a secondary response to the substances released in the dead cells. Additionally, the concentration of free chlorine decreased when HClO was added to the sterilized PBS buffer solution (Wang et al. 2019). Therefore, more detailed investigation is needed to explore the continued consumption of free chlorine.

Reduction of E. coli ranged from 6 log units to 1 log unit after 10 min of exposure (LR was about −5), and no viable count was detected in the plate after 20 min of exposure. E. coli was considered to have been entirely inactivated by chlorine in the traditional sense. Within the first 5 min (LR was about −3), chlorine eliminated about 3 log units of Gordonia JN724. The number of viable bacteria remained relatively constant from 5 min to the end of the experiment (LR was about −3), and the final determination of Gordonia JN724 in the plate was above 3 log units (for a contact time of 240 min). There was a clear demarcation point at 5 min in the survival curves, as an initial speedy inactivation period before and a slow or an ineffective inactivation stage after. The result is in accordance with many previous studies on the two stages in the response curve (Luh & Mariñas 2007; Chen et al. 2012). These two stages may be attributed to the existence of two populations – one consisting of susceptible cells and the other of tolerant cells. One possible explanation for the presence of two populations may be that the cultures contained multiple colony variants. The existence of colony variants and the spontaneous transition between colony types have been reported in the past (McCarthy 1970; Woodley & David 1976; Stormer & Falkinham 1989). Similar to that reported for free chlorine, the two-population analysis appeared following chlorine dioxide and ultraviolet radiation treatment. The idea that a ‘weak’ group and a ‘strong’ group exist in disinfection experiments has also been proposed (Benito et al. 2002; Vicuna-Reyes et al. 2008).

SEM photos of Gordonia JN724 and E. coli before and after chlorine inactivation are shown in Figure 3. Prior to treatment, both were rod-shaped and their surfaces were glossy. Following inactivation, the surface of E. coli became incomplete, and holes appeared on the surface of the membrane. With increased contact time, the cell membrane of E. coli deteriorated, and the complete cell morphology could not be observed in the later experiment. However, no major changes in membrane integrity were seen in Gordonia JN724; only its shape changed. This difference in morphology further confirmed that Gordonia JN724 is much more resistant to chlorine than E. coli.

The stain combination of TO and PI was used to evaluate the permeability of the cell membrane and thus distinguish between live cells (with an intact membrane) and dead cells (with an injured membrane) after inactivation. According to the distinction in fluorescence species and intensity, the stained individuals are represented in different areas of the dot plot. There was adequate separation between the background noise and the stained individuals (Hammes & Egli 2010). The gate applied for the quantification of the total cell concentration was labeled ‘E. coli’ or ‘Gordonia JN724’ and another was ‘beads’ in FCM dot plots (FSC and SSC). According to the distinctions in the TO and PI stains, the quantification region was divided into three regions: ‘live’, ‘injured’, and ‘dead’. Live (bacteria which were cultured in nutrient broth at 37 °C for 24 h) and completely dead cells (heat-killed, 100 °C hot water for 10 min) were used as controls to designate the shapes of the ‘live’ and ‘dead’ regions. The ‘injured’ region was obtained by cells from the fluorescence mixture of the ‘live’ and ‘dead’ regions (Zhang et al. 2018). Cells in the ‘injured’ region consisted of VBNC individuals that could not be detected by culture methods.

Cells stained before and after the inactivation are shown in Figures 4 and 5. The data provide support for the use of TO and PI double staining associated with flow cytometry as a cultivation-independent approach for evaluating both total cell concentrations and cells with damaged membranes in E. coli inactivation. Before the inactivation, the proportion of ‘live’ cells was over 98%, whereas the sum of ‘injured’ and ‘dead’ cells was less than 2%. However, there was a marked increase in the quantity of ‘injured’ and ‘dead’ bacteria with prolonged contact time. The percentage of cells in the ‘injured’ and ‘dead’ regions rose to 81.85% and 14.4%, respectively, after 5 min of chlorine treatment. After 20 min of exposure, only about 20% of the cells appeared in the ‘injured’ region (VBNC state), and no viable counts were detected by traditional methods. All of the cells were ‘dead’ within 2 h. Bacteria entered a VBNC state in the chlorine treatment that was not detectable by the cultivation method.

In the present assay, abnormal cell membrane permeability for Gordonia JN724 was observed. Only total cell concentrations and ‘dead’ cells were measured, while cells with damaged membranes were not detected. However, we discovered that normal microbial cells could be stained by TO and PI at the same time, with all of the cells located in the ‘injured’ region. Red fluorescence can be observed in living cells with a combination of PI and nucleic acid in the repeated test. Thus, it is impossible to identify the real state of cells using FCM, as ‘live’ and ‘injured’ appear in the same region. The number of cells in the ‘dead’ region increased with exposure time. From 20 min exposure to the end of the treatment, the proportion of dead cells remained relatively stable (20–30%), which also confirmed the chlorine resistance of Gordonia JN724 on the other sides. For VBNC cell detection, it may be possible to try other types of dyes in future.

FCM analysis has been used to detect bacterial activity in water disinfection, and more challenges have been uncovered, such as the existence of VBNC cells (Nie et al. 2016; Zhang et al. 2018). A past study found that while pathogens in the VBNC state could not trigger disease directly, potential virulence was retained, and they were able to initiate infections with recovery to a metabolic state (Zhang et al. 2018). Our study also confirmed the existence of a large number of VBNC cells following chlorine disinfection. The fraction of VBNC cells decreased with increases in available chlorine and contact time. Because of the potential to cause harm later, the VBNC state must be examined in further detail.

Previous studies showed that the disinfectant resistance of bacteria may be attributed to differences in cytoarchitecture, components of the cell wall, cell size, and other aspects (Sisti et al. 2012; Wen et al. 2016). FCM analysis revealed that the number of cells that PI completely penetrated was only less than 30% for Gordonia after 30 min of chlorine exposure, while that of E. coli was more than 80%. The difference in membrane integrity and permeability between the two strains were likely responsible for the difference in chlorine tolerance and are consistent with the SEM observations. Gram-positive bacteria are generally more resistant to gram-negative bacteria because their thicker cell walls can withstand the oxidation effects of hypochlorous acid and hypochlorite (Nie et al. 2016). This phenomenon may explain the strong resistance of the cytomembrane of Gordonia JN724 to chlorine. It remains unclear whether the special blocking structure of the bacteria lead to TO and PI double staining and contribute to its chlorine tolerance. To learn more about the resistance mechanisms expressed in Gordonia JN724, further research on its surface structure is needed, including the protein and fatty acid composition, as well as intracellular interactions.

With 0.5 mg L⁻¹ and 1 mg L⁻¹ chlorine dioxide exposure, no colonies were detected by the HPC method after a short time exposure. Subsequently, 0.38 mg L⁻¹ of chlorine dioxide treatment was performed. Gordonia JN724 exposed to 0.38 mg/L of chlorine dioxide for 5 min produced more than 2 log inactivation (Figure 6). The CT value for 99.9% inactivation was calculated as 2.04 mg min L⁻¹, which was notably lower than that of free chlorine. This suggests that Gordonia JN724 is more sensitive to chlorine dioxide than chlorine.

Many studies have found chlorine dioxide to be more effective than free chlorine in the inactivation of bacteria (Chen et al. 2012; Wen et al. 2017). The oxidation potential if chlorine dioxide oxidation potential (redox potential: 435 1.51 V) is higher than that of chlorine (redox potential: 1.36 V) (Hosni et al. 2013), which will lead to a more rapid reaction with intracellular materials. Additionally, chlorine dioxide, a neutral molecule in water, can easily diffuse through the cell wall membrane (Gagnon et al. 2004). The higher oxidation potential and permeabilization of chlorine dioxide will thus result in more efficient inactivation compared with chlorine.

The UV dose-response curve in Figure 7 indicated that UV irradiation was effective in Gordonia JN724 inactivation. When the UV dosage increased to 40 mJ cm⁻², approximately 4 log 10 (i.e., 99.99%) removal was detected. Complete inactivation (about 5 log 10) was practically reached after exposure to UV fluences of 80 mJ cm⁻². The UV dose-response curve for Gordonia JN724, which was similar to that of other organisms inactivated by UV (Sun et al. 2013), converged with the polynomial regression model (USEPA 2006).

According to the inactivation experiment, chlorine dioxide and UV irradiation are fast and effective disinfection methods for Gordonia JN724 in water. Many studies have shown that a portion of chlorine-resistant microbes are sensitive to UV irradiation, chlorine dioxide inactivation, and other disinfection methods (Linden et al. 2002; Drescher et al. 2011; Sun et al. 2013). Similarly, Gordonia JN724, which was isolated from a water-pipe system in this study, showed a high sensitivity to UV irradiation. The inactivation rate reached 99.99% at a dose of 40 mJ cm⁻², which is suggested by the USEPA as the optimal density in drinking water treatment (USEPA 2006). The CT value for 99.9% inactivation by chlorine dioxide was approximately 20 times lower than that of chlorine disinfection. Therefore, chlorine dioxide or UV disinfection may be alternative approaches for the control of chlorine-resistant microorganisms.

Gordonia JN724 is a highly chlorine-resistant, gram-positive bacterium obtained from the drinking water distribution system taking Yellow River as the water source. FCM analysis confirmed the presence of a great majority of VBNC cells and damage to cell membranes in E. coli inactivation. The structural properties of chlorine-resistant bacteria must be studied systematically for chlorine-resistant mechanisms. Gordonia JN724 can be effectively inactivated by ClO₂ and UV treatments, which suggests that they may serve as alternative approaches for the control of chlorine-resistant microorganisms. Meanwhile, water treatment plants using chlorine disinfection should be aware of the existence of chlorine-resistant microbes and consider different disinfection methods based on their own unique needs.

This study was supported by National Major Projects on Water Pollution Control and Management Technology (No. 2017ZX07501003, No. 2017ZX07502003-06), National Natural Science Foundation of China (51808512) and the Special Project of Taishan Scholar Construction Engineering (ts200640025). We thank Natalie Kim, PhD, from Liwen Bianji, Edanz Editing China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.

No conflict of interest declared.

All relevant data are included in the paper or its Supplementary Information.