A Proposal for Source Tracking of Fecal Pollution in Recreational Waters by Pulsed-Field Gel Electrophoresis

This study aimed to identify specific river sources of fecal contamination by applying pulsed-field gel electrophoresis (PFGE) to environmental water samples from a recreational beach in Japan. The genotypes of all Enterococcus faecium and Enterococcus faecalis strains used as indicators of fecal pollution on the recreational beach and rivers were analyzed by PFGE, and the PFGE profiles of the strains were classified at a 0.9 similarity level using dendrogram analysis. PFGE types of E. faecium isolated from Sakai River or urban drainage were classified in the same cluster. Therefore, the probable sources of fecal pollution on the recreational beach were Sakai River and urban drainage. The approaches for microbial source tracking employed in this study used PFGE with Enterococcus species as an indicator can be a potential tool to specify the source(s) of fecal pollution and contribute to improved public health in coastal environments.

Fecal contamination of water environments by waterborne pathogens such as Escherichia coli O157:H7, Cryptosporidium, Giardia, and noroviruses is a serious problem to public health (22,28). Concern about human infections and the detection of these pathogens in aquatic environments has been increasing (25). Coastal areas often receive many different types of water that can be affected by domestic and industrial human activities. Consequently, fecal microorganisms that flow in rivers to coastal areas have a wide variety of origins that are classified as point sources or nonpoint sources (24). On the other hand, coastal areas play an important role in human recreation and shellfish production, and the contamination of coastal surface waters could be a serious threat to human health. Therefore, identifying the source of fecal pollution is important to contribute to the improvement of bacterial water quality and ensure public health in water environments.
Potential indicators of fecal pollution include fecal coliforms, E. coli, enterococci, and other fecal microorganisms (Bacteroides spp., Bifidobacterium spp., and coliphages). Among these species, fecal coliforms, E. coli, and enterococci have been historically used as indicators of choice when monitoring coastal water quality (31). However, in Japan, there are no standards concerning bacterial water quality in aquatic environments. It has been reported that high densities of E. coli and enterococci in recreational surface waters are more strongly correlated with swimmingassociated gastrointestinal disease than high densities of fecal coliform (4). The genus Enterococcus has been recommended as an indicator of fecal pollution because of its long-term survival and low growth rate in aquatic environments (14,16).
Many studies have attempted to develop microbial source tracking (MST) methods using different indicators of fecal pollution (8,17,21,23). Various phenotypic and genotypic methods classified as library-dependent and libraryindependent methods have been examined for MST. Numerous studies provide useful knowledge and information related to MST methods, including review articles (3,22,26,28), guidance documents (34), and comparative research on genotypic and phenotypic methods (1,9,29). However, there is still no standard method for source tracking of fecal pollution.
Pulsed-field gel electrophoresis (PFGE) is a DNA fingerprinting method that can distinguish different strains within a species by comparing genotypic characteristics. It is an analytical technique in the field of molecular epidemiology (2,27), where this method has been traditionally used for identifying the route and source of infection in nosocomial infections and for specifying bacteria that are responsible for food illness (7,19). PFGE has also been applied to source tracking of Salmonella contamination in marine environments (18). PFGE has extremely high sensitivity, reproducibility, and discrimination ability compared with other MST methods (22).
We have previously proposed MST using PFGE and have strongly indicated its ready availability (5). In a case study conducted on Aoshima Beach, Japan, the fecal pollution source for a swimming beach was identified among five rivers by MST using PFGE with Enterococcus faecium as an indicator of fecal pollution (6). However, it is necessary to demonstrate its adaptability and geographical stability in coastal areas to establish PFGE as an MST method. Enterococcus faecalis is also important as an indicator of fecal pollution along with E. faecium. In certain cases, E. faecalis is the dominant Enterococcus species in river and coastal waters. Therefore, in this study, MST was performed with two species of major enterococci, E. faecium and E. faecalis, collected from a recreational beach and entirely different from those collected from Aoshima Beach and rivers near the beach. The source of fecal pollution was tracked by comparing dendrograms of PFGE types obtained from the two Enterrococcus species.

Outline of source tracking
This demonstration of fecal pollution source tracking using PFGE was conducted at a recreational beach in Oita, Japan, facing the Seto Inland Sea. The beach (latitude: 33.27°N; longitude: 131.50°E) and rivers are shown in Fig. 1. There is no official bathing season at the beach; however, it is an important recreation area because several events (festivals, fireworks) are held here during the summer. The area covered and lengths of the three rivers, i.e., Haruki, Sakai, and Asami rivers, were 8.6 km 2 and 6.27 km, 12.2 km 2 and 6.90 km, and 10.9 km 2 and 5.03 km, respectively. The distance of the Haruki River, Sakai River, Asami River, and urban drainage from the recreational beach is approximately 2.5, 0.6, 1.3, and 0.5 km, respectively.
Sampling was conducted on July 21, 2009. The coastal water sample was collected from surface water at a depth of approximately 50 cm near the water's edge on the recreational beach. River water samples were collected from the surface of each river. The sampling points of river water were within approximately 500 m of the estuary for each river. There was a slight effect of seawater mixing at the river sampling points. A water sample was also collected from urban drainage, a candidate fecal pollution source, passing through a residential area near the beach. In all, 5 surface water samples were stored in sterile 1-L polyethylene bottles and immediately transported to the laboratory, where they were tested for fecal bacteria and water quality.

Water quality parameters
We analyzed enterococci, coliforms, and E. coli in each water sample. pH, electrical conductivity (EC), and turbidity were also determined as major water parameters using a pH meter (HM-30G; TOA DKK, Tokyo, Japan), conductivity meter (CM30S; TOA DKK), and turbidity meter (SEP-PT-706D; Mitsubishi Kagaku, Tokyo, Japan), respectively. Salinity was calculated from EC (20), and water temperature was determined using a stick thermometer during sampling.

Bacterial counts
Enterococcus counts were determined by the membrane filtration (MF) method (32). One hundred milliliters of each water sample were filtered in triplicate using a 0.45-μm pore membrane filter (47 mm diameter, sterile, mixed cellulose ester; Advantec, Dublin, CA, USA) and incubated on membrane-Enterococcus indoxyl-β-d-glucoside (mEI) agar plates for 24 h at 41±1.0°C. After incubation, blue halo colonies on the filter were determined as enterococci.

Isolation of enterococci
After the Enterococcus count, a single blue halo colony from each mEI agar plate was randomly streaked on a Todd Hewitt agar plate (TH agar plate; Becton Dickinson, Franklin Lakes, NJ, USA; added agar 1.5%), followed by incubation for 24 h at 37 ± 1.0°C. One hundred and fifty colonies of Enterococcus strains from the beach and 100 colonies from each of the rivers (Haruki, Sakai, and Asami rivers) and urban drainage were isolated. Identification tests were performed on all Enterococcus strains isolated from each water sample.
Identification of Enterococcus strains E. faecium and E. faecalis were identified using PCR and Api 20 Strep (BioMerieux, Lyon, France). All the isolated Enterococcus strains were identified by PCR, and the strains identified as E. faecium or E. faecalis by PCR were subsequently confirmed using Api 20 Strep.
Subsequently, the Enterococcus strains identified as E. faecium or E. faecalis by PCR were analyzed using Api 20 Strep according to the manufacturer's instructions. Using both PCR and Api 20 Strep, Enterococcus strains were identified as E. faecium or E. faecalis, and the genotypic characteristics of these strains were analyzed by PFGE.

PFGE
To elucidate the genetic relationship among the Enterococcus strains, PFGE was performed using the CHEF Bacterial Genomic DNA Plug Kit (Bio-Rad) according to the manufacturer's protocol, with a slight modification. In brief, 8-10 colonies from the TH agar plates incubated for 24 h at 37°C were suspended in 1.0 mL sterilized physiological saline in a 1.5-mL microcentrifuge tube. The tube was centrifuged for 3 min at 13,523×g (Centrifuge 5424; Eppendorf, Hamburg, Germany), and the supernatant was discarded. The pellet was resuspended in 150 μL cell suspension buffer. The suspension was mixed with 3.0 μL lysozyme (25 mg mL −1 ), 3.0 μL lysostaphin (2.0 mg mL −1 ; Wako Pure Chemical, Osaka, Japan), and 3.0 μL mutanolysin (≥4,000 U mg −1 ; Sigma-Aldrich, St. Louis, MO, USA), followed by incubation for 10 min at 37°C. Next, the suspension was mixed with 150 μL of liquid 2% CleanCut agarose before it was poured into plug molds. The sample plug was then incubated for 4 h at 37°C in 500 μL lysozyme buffer containing 10 μL lysozyme, 10 μL lysostaphin, and 10 μL mutanolysin. The plug was treated with proteinase K (>600 U mL −1 ) for 24 h at 50°C, after which the plugs were washed 5 times using 1 mL of 1×wash buffer for 30 min with rotation in a microtube rotator (MTR-103; As One, Osaka, Japan). The DNA embedded in each plug was digested with the restriction enzyme SmaI (25 U plug −1 ) in 300 μL SmaI buffer for 20 h at 25°C, after treatment with 1 mL of 10-fold-diluted (0.1×) wash buffer followed by 500 μL SmaI buffer.
DNA fragments were separated for 20 h at 14°C on 1% pulsedfield certified agarose gel (Bio-Rad) in 0.5×TBE buffer, with a switch ramp time from 5.3 to 34.9 s at a 120° angle, using a CHEF DR II system (Bio-Rad). The sizing ladder used for PFGE was a lambda DNA ladder with a range of 48.5 kb-1.0 Mb (Takara).

Dendrogram analysis of PFGE patterns
Dendrogram analysis of band-based PFGE patterns was performed using a Gene Profiler (Scanalytics, Buckinghamshire, UK). Levels of similarity between fingerprints were expressed as Dice coefficients, which were calculated by determining the ratio of twice the number of bands shared by two patterns to the total number of bands in both patterns. PFGE patterns were clustered by the unweighted pair group method with arithmetic mean (UPGMA). In the dendrogram analysis of PFGE fingerprints, some strains that belong to the same cluster with a 0.9 similarity level should be considered as related species (30).

Major water parameters and bacterial counts
The major water parameters (pH, water temperature, salinity, and turbidity) and bacterial counts (enterococci, coliforms, and E. coli) are shown in Table 1. pH (7.3-7.7) and water temperature (28.1-28.7°C) were not considerably different at each sampling station. However, turbidity varied considerably among sampling stations. The turbidity of the beach sample was low compared with that of the river water samples. High turbidity was detected in Sakai (12.2 degrees) and Asami (18.4 degrees) rivers. The salinity at the beach was 28.4 psu, and salinities of the river water samples ranged from 0.1 to 0.2 psu, indicating that the sampling points on each river were not influenced by the inflow of seawater.
A high Enterococcus count was detected in the water sample from the beach (2.6×10 2 ±10.8 CFU 100 mL −1 , mean±S.D., n=3). The counts of coliforms (63 MPN 100 mL −1 ) and E. coli (below detection limit) detected in the water sample from the beach were relatively low compared with that of the enterococci. The Enterococcus counts were high in the samples from Haruki (5.2×10 2 ±48.1 CFU 100 mL −1 ) and Asami (3.8×10 2 ±75.1 CFU 100 mL −1 ) rivers. The Enterococcus counts of the samples from Sakai River and urban drainage were low compared with those of the other three samples. The coliform counts of water samples from all rivers were extremely high, i.e., >10 3 MPN 100 mL −1 . The E. coli counts of all river water samples were <100 MPN 100 mL −1 . Enterococcus counts were lower than the coliform counts, with the exception of those of samples from the beach.
On the beach, enterococci were detected at a concentration that exceeded the present single sample limit for enterococci (104 CFU 100 mL −1 ) recommended by the US Environmental Protection Agency for marine recreational water (31). However, a low concentration of coliforms and no E. coli were detected in the water sample from the beach. The Enterococcus counts of all river water samples exceeded the single sample limit for fresh water (61 CFU 100 mL −1 ) (33). High counts of enterococci were detected in Haruki and Asami rivers. Extremely high coliform counts were also detected in all river water samples. Overall, the sampling stations where the counts of fecal bacteria in the water samples were comparatively high may be polluted by pathogenic microorganisms derived from human and/or animal feces. However, simply monitoring bacterial water quality cannot provide detailed information about the source(s) of fecal pollution that can degrade water quality in coastal recreation areas.

Identification of fecal indicator bacteria
Identification of E. faecium and E. faecalis was performed with PCR and Api 20 Strep, using the 150 Enterococcus strains isolated from the water sample of the beach ( Table  2). Among these 150 strains from the beach, 94 and 22 strains were identified as E. faecium and E. faecalis, respectively, by PCR. Then, these 94 E. faecium and 22 E. faecalis strains were confirmed using Api 20 Strep. Eighty-nine of the 94 E. faecium strains were also identified as E. faecium using Api 20 Strep. Eleven of the 22 E. faecalis strains were also identified as E. faecalis using Api 20 Strep.
All 100 strains isolated from the river water samples (Haruki, Sakai, and Asami rivers, and urban drainage) were analyzed using the same identification tests. In the Haruki River, the number of strains identified as E. faecalis was 6 times higher than that identified as E. faecium, thus showing a considerable difference. On the other hand, similar numbers of E. faecium and E. faecalis were identified in Sakai and Asami rivers and urban drainage. The smallest number of E. faecium and E. faecalis strains was identified in the Asami River. E. faecium and E. faecalis strains were identified from  (12). Fecal pollution can be broadly classified into human and nonhuman pollution. The most serious threat to human health is thought to be from human rather than animal feces in water environments (3). Therefore, E. faecium and E. faecalis are very important species for evaluating fecal pollution in water environments. Identification of E. faecium and E. faecalis using PCR and Api 20 Strep showed that these two Enterococcus species were found in water samples from all stations. The total proportion of E. faecium and E. faecalis in each water sample was >50%, except in the water sample from Asami River. This indicated that all sampling stations (the beach, Hariki and Sakai rivers, and urban drainage), with the exception of Asami River, might be polluted by fecal bacteria. E. faecium was the dominant species in samples from the beach. In contrast, E. faecalis was identified at much higher levels than E. faecium in the samples from the three rivers and urban drainage. This suggests that the tolerance to fluctuating environmental conditions, i.e., freshwater and seawater, may differ between these two Enterococcus species. Further studies are required to determine the environmental fate of these species.
Genotypic analysis of E. faecium and E. faecalis using PFGE All 155 E. faecium and 138 E. faecalis strains isolated from the water samples were analyzed by PFGE using SmaI restriction enzyme. Many of the E. faecium strains isolated from the beach were analyzed by PFGE, but these strains only produced 12 PFGE types (Table 3). Of the 12 PFGE types of E. faecium isolated from the beach (72 strains), the largest number had the Fm-Rb-3 type, i.e., the dominant PFGE type from the beach. In contrast, E. faecium strains isolated from each river water sample produced a wider variety of PFGE types compared with those isolated from the beach. In Haruki and Asami rivers, although the number of E. faecium strains isolated from both rivers was relatively low, various PFGE types were observed. E. faecium strains from both rivers had individual genotypes.
E. faecalis strains isolated from the beach (11 strains), Haruki River (63 strains), Sakai River (28 strains), urban drainage (28 strains), and Asami River (8 strains) produced 9, 31, 25, 22, and 6 PFGE types, respectively (Table 3). Thus, 93 PFGE types were found for the E. faecalis strains. The number of E. faecalis strains isolated from all stations was lower than that of E. faecium strains. However, the number of PFGE types of E. faecalis was more than that of the PFGE types of E. faecium.

Similarity analysis of genotypes using dendrogram
In this study, we analyzed the similarities of all SmaIdigested band patterns obtained from Enterococcus strains using dendrogram analysis. Dendrograms were constructed containing all the PFGE types of each E. faecium ( Fig. 2A) and E. faecalis (Fig. 2B) strain. We determined the PFGE profiles at a 0.9 similarity level in the dendrogram containing E. faecium strains by classifying the PFGE types of Fm-Sa-1 obtained from Sakai River and those of Fm-Rb-3 (dominant type) obtained from the beach in the same cluster. The PFGE types of E. faecium that were classified into the same clusters at a 0.9 similarity level were closely associated with each other. The two PFGE types of the strains isolated from urban  drainage were also classified in the same clusters as those from the beach at a 0.9 similarity level (Fm-Ud-9 and Fm-Rb-2, and Fm-Ud-3 and Fm-Rb-12). Thus, it was highly possible that fecal pollution of the beach was closely related with Sakai River and urban drainage. The PFGE types of the strains isolated from Haruki and Asami rivers were not classified into the same cluster as those from the beach. There were a variety of PFGE types for E. faecalis strains, but these were not very similar. Only the PFGE type of Fs-Sa-1 obtained from Sakai River had more than 0.9 similarity with that of Fs-Rb-4 obtained from the beach. Meanwhile, some genotypes of E. faecalis strains isolated from different river water samples were classified into the same clusters (Fs-H-25, Fs-Sa-6 and Fs-Ud-17, Fs-A-5 and Fs-H-31, Fs-H-18 and Fs-Sa-12, Fs-H-15 and Fs-Sa-5, and Fs-Sa-21 and Fs-Ud-9). It was suggested that E. faecalis strains were derived from similar hosts in different rivers. Therefore, it seemed difficult to estimate the source of fecal pollution using the dendrogram of the genotypes obtained from E. faecalis strains in this coastal urban area.
Based on the results of the dendrogram analysis of E. faecium genotypes, the most likely sources of fecal pollution of the beach were Sakai River and urban drainage, although only one sampling was conducted for this demonstration of the MST approach. It was clear that this MST approach using PFGE can discriminate within the same species regardless of the differences among the Enterococcus spp. and the numbers of analyzed strains. A larger number of enterococci were detected from Haruki and Asami rivers than were detected from Sakai River and urban drainage, which were presumed to be the source of fecal pollution based on genotypic analysis using PFGE. Thus, rivers with the highest concentration of enterococci are not always the source of fecal pollution in coastal recreation areas.
An important consideration common to library-dependent methods is the response to temporal changes in the genotypes of fecal pollution indicators. Temporal change in genotypes is a factor that may complicate the determination and understanding of fecal pollution sources (10). A previous study reported that E. coli varied in PFGE banding patterns, which might be caused by mutation or recombination over the course of 8 weeks (15). It is important to clarify that temporal changes in the genotypes of indicators because of chemical and bacterial water quality and bacterial flora in rivers and coastal areas may be markedly changed by various factors such as rainfall, wind, and water current. In this study, to eliminate the problem of temporal changes in PFGE banding patterns of Enterococcus spp. as an indicator of fecal pollution, sampling was performed on a single day (July 21, 2009). As a result, it was shown that MST using PFGE could identify rivers that were fecal pollution sources of the beach. Bacterial water quality and Enterococcus flora in river water and seawater might change with the season, even in the same water environments, and therefore, multiple demonstrations of MST using PFGE will be essential in further studies. In addition, we will also try to identify more detailed sources of fecal pollution by the same MST approach within several stations such as municipal sewage, septic tanks, and livestock, which were selected as candidate sources in rivers that were considered as fecal pollution sources in the present study.