Evaluation of a human-associated genetic marker for Escherichia coli (H8) for fecal source tracking in Thailand

Fecal contamination by water discharged from point sources and nonpoint sources causes waterborne diseases. This is a major concern in many countries, especially in developing countries (Pandey et al. 2014). All of the waterborne pathogens are currently unrealistic for monitoring in the aquatic environment because of the diversity of pathogens, including bacteria, viruses, and protozoa (Harwood et al. 2014). Thus, fecal indicator bacteria (FIB) have been selected for monitoring due to being easy to measure and their high levels in sewage and feces (Griffith et al. 2009). For example, Escherichia coli has been commonly used as an indicator organism of fecal pollution and freshwater for many years (Anderson et al. 2005). Furthermore, FIB have been studied for water quality for decades, but the limited information cannot provide the data about the original source in wastewater due to their nature (found in warm- and cold-blooded animal feces) (McLellan 2004; Field & Samadpour 2007).

Fecal source tracking is a procedure to identify microbes or genetic markers specifically associated with their host sources, e.g., human, pig, cow (Johnston et al. 2013). This method has been successfully solving fecal pollution problems in many countries, especially in developed countries, although this method has been limited in developing countries but continues to increase (Somnark et al. 2018). In addition, control of fecal contamination becomes difficult to implement as a management plan without the identification of its source. Therefore, fecal source tracking has become an important method for understanding the source of fecal contamination and for decreasing the hazard of waterborne diseases that occur in the environment. This method will help to produce an efficient way for water quality management. Human-specific Bacteroides genetic markers have been used to detect fecal pollution in the environment (Bernhard & Field 2000; Seurinck et al. 2005; Haugland et al. 2010). However, the limitation of the usage of the Bacteroides genetic marker is that it persists for shorter periods of time (99% decay in <8 days) compared to the E. coli genetic marker (99% decay in >15 days) (Tambalo et al. 2012). While E. coli has been widely used as an indicator bacteria of fecal pollution, integrated ways to use the Bacteroides marker with E. coli concentration have not been well established yet. Recently, a human-associated genetic marker for E. coli (H8) has been developed and used in Japan, Australia, and Bangladesh (Gomi et al. 2014; Warish et al. 2015; Harada et al. 2018). The H8 marker has been validated with other human-associated E. coli genetic markers. It was found that the H8 marker had the highest performance in terms of the specificity and sensitivity to detect human fecal pollution (Warish et al. 2015). Moreover, these studies have successfully used the H8 marker to check the proportion of human-associated E. coli in wastewater and drinking water. However, this H8 marker method has not been widely applied to investigate the fecal pollution in a water environment for fecal source tracking in Thailand. Therefore, the H8 marker required further investigation into its applicability and performance in identifying fecal contamination sources as well as reducing the related waterborne diseases and better managing water quality in Thailand. This study was conducted with these objectives: (1) to evaluate the performance (specificity and sensitivity) of the H8 marker by testing with fecal and wastewater samples from eight non-target and target host groups in Thailand, and (2) to compare the performance of the H8 marker with that of other countries.

Animal feces samples including cattle, buffalo, chicken, duck, and pig (non-target host groups) were collected in sterilized plastic bags for calculating the specificity of the H8 marker. For avoiding an E. coli clone colony, 10 feces samples from each animal species were collected from different livestock farms in Nakhon Pathom, Samut Sakhon, and Ratchaburi provinces, Thailand. Cattle, buffalo, chicken, duck, and pig were chosen as animal species as they were the top five of the most prevalent livestock animals in Thailand, according to The Number of Livestock in Thailand in 2017 (Department of Livestock Development 2017).

To determine the sensitivity of the H8 marker, 10 samples from each influent and effluent (target host groups) were collected in 100 mL sterilized bottles from a domestic wastewater treatment plant serving approximately 566,000 people in Bangkok, Thailand. In addition, septage samples (target host groups) from 10 individual households in Samut Sakhon province, Thailand were also collected in 100 mL sterilized bottles and analyzed for evaluation of the sensitivity of the H8 marker. Ten individual samples per animal and human sources were selected for analysis as advised by the US Environmental Protection Agency (EPA) microbial source tracking guide document (US EPA 2005). Figure 1 shows examples of septic tanks which were used for collecting the samples in this study. All the samples were kept in the dark and transported on ice to the laboratory, then stored at 4 °C and processed within 12 hours.

For livestock feces, samples were directly streaked on HiCrome Chromogenic Coliform Agar (recommended for detection of E. coli, HIMEDIA, India) by using a sterilized loop. A loop was dipped into and picked up a small amount of feces, and then dragged across the surface of the agar. This agar was incubated at 37 °C for 22 hours.

Moreover, each of 10 samples from influent, effluent, and septage samples was mixed carefully and 30 μL of each sample was used to make a dilution in 30 mL of phosphate buffered saline. Then, the samples were filtrated by using a 100 mL sterilized funnel and disposable filtration devices (0.45 μm, white MCE membrane, Merck, Germany). After filtration, the filter was placed on HiCrome Chromogenic Coliform Agar and it was incubated at 37 °C for 22 hours. For E. coli analysis, E. coli isolates (blue colonies) were picked up by using sterilized toothpicks and transferred into each well of a 96-well microtiter plate filled with 50 μL MilliQ water. These 96 well plates were stored and kept at −20 °C for a maximum of 24 hours prior to polymerase chain reaction (PCR) analysis.

To check the performance of specificity and sensitivity, SYBR Green based real-time PCR assays were performed on each E. coli isolate with a human-associated E. coli genetic marker (H8). Table 1 shows E. coli isolates which were used in this study. Two hundred and fifty E. coli isolates from non-target host groups and 150 E. coli isolates from target host groups were used to perform the analysis. In addition, the primer and target DNA sequence of the H8 genetic primer sets were used for SYBR Green based PCR assays, and are shown in Table 2.

In this study, the colony PCR method was used as a powerful tool for rapid and easy screening of the potentially large numbers of colonies, without DNA extraction, with high sensitivity (Bergkessel & Guthrie 2013). Each isolate of E. coli was screened for the presence of H8. Selective agar was used to grow E. coli, and isolates were collected by picking up a colony identified as E. coli based on its color profile. The mechanism used to identify E. coli from various environmental sources was based on β-glucuronidase as the target enzyme due to very high sensitivity and specificity of this method (Rice et al. 1990). Later, the separated isolates were directly transferred to a PCR master mix.

For the real-time PCR assays, the PCR mixture (15 μL) was composed of 4.9 μL of MilliQ water, 7.5 μL of QuantiFast SYBR Green PCR (QIAGEN, Germany), 0.3 μL each of forward and reverse primers, and 2 μL of E. coli samples. All PCR reactions were performed in a 96-well plate using the CFX96 touch real-time PCR system (BIO RAD, Singapore). Positive (DNA from control strains) and negative (sterile water) controls were included for each PCR assay. The Real-time PCR conditions were set at 95°C × 5 min + (95°C × 10 sec + 60 °C × 30 sec) × 40 cycles + melting curve analysis.

To analyze the results from SYBR Green based PCR assay, the Ct value (cycle threshold) of the amplification curve was used to determine the presence or absence of PCR amplification from each sample. The H8 positivity control was observed by the Ct value of the amplification curve as below 25 cycles. Moreover, the Tm value (melting temperature) was also determined based on the results of dissociation curve. The Tm value of the dissociation curve is shown within 91 ± 1 °C (Tm value of positive control) (Harada et al. 2018).

Finally, all the results were evaluated by using the specificity and the sensitivity test. Specificity was calculated by using true negatives and false positives of the non-target host samples. Sensitivity was calculated by using true positives and false negatives of the target host samples (Odagiri et al. 2015).

To differentiate the positive proportion of the H8 marker between human and animal, a proportion test was used to confirm the positive samples between the target and the non-target host groups by using RStudio (R 3.4.0), a programming language for statistical analysis and data science.

In total, 400 fecal and wastewater samples were collected from five non-targets (cattle, buffalo, chicken, duck, and pig) and three (influent, effluent, and septage) host groups in Thailand. H8 positive percentages of isolates from non-target and target host groups are summarized in Table 3. For the specificity and sensitivity analysis, we screened 50 E. coli isolates from each sample to avoid false positive detection of the H8 marker. From all of 250 non-target host groups tested in this study, 239 (96%) E. coli isolates were PCR negative for the H8 marker. However, one chicken, one pig six cattle, and three buffalo fecal samples were PCR positive. In addition, it was found that in a total of 150 E. coli isolates from target host groups the human excreta samples from septage (18/50) had the highest positive percentage of 36% with the H8 marker. Moreover, influent (9/50) and effluent (6/50) samples from a wastewater treatment plant showed the positive percentage only at 18 and 12%, respectively. The percentage of positive proportion of the H8 marker from the wastewater treatment plant was lower than the human excreta samples from septic tanks because wastewater treatment plants in Thailand have been normally constructed to use a combined sewerage system (Boontanon & Buathong 2013). This may cause some nonpoint sources that can potentially mix with the target wastewater samples such as animal feces, rain water, and toxic chemicals from urban runoff. Based on the highest positive percentage from target host groups, the positive proportion from septage samples were selected to calculate the difference between non-target and target host groups. The results showed that the H8 marker was found to be significantly different in positive proportions from human to all animal samples consisting of cattle (6/50), buffalo (3/50), chicken (1/50), pig (1/50), and duck (0/50) (p ≤ 0.01).

Among the non-target samples, cattle feces samples were found to have the highest number of H8 positive isolates at six positive samples (12%). Such data were projecting a similar trend to that from a previous study in Australia which found five wastewater samples (50%) from cattle were positive with the H8 marker from their non-target host groups (Warish et al. 2015). These results show that cattle are the most frequent animal species in which the H8 marker has been detected (from two studies). In addition, another study conducted in Japan showed that pig (6.7%), cow (3.1%), and chicken (11.1%) fecal samples indicated positive results with the H8 marker (Gomi et al. 2014). These results illustrate the variation of the H8 marker performance in the non-target host samples. This variation may be attributed to differences in livestock feeding or livestock living conditions.

Specificity and sensitivity are commonly used to determine and validate the host-specificity of genetic markers (Bernhard & Field 2000). The performances by specificity and sensitivity of genetic markers are crucial characteristics for the fecal source tracking method because low specificity and sensitivity of genetic markers may result in false positives and false negatives in the identification of the origin of nonpoint source fecal pollution (Stoeckel & Harwood 2007). According to universal performance criteria, it has been recommended that a specificity percentage with values below 80% need to be considered carefully and may not be suitable as a fecal source tracking method. Therefore, it has been suggested that the specificity and sensitivity of a genetic marker be tested for its applicability and performance in identifying fecal contamination sources for environmental surface water, especially for new geographical studies (US EPA 2005).

In this study, the specificity and sensitivity of the H8 marker were evaluated by using 250 fecal samples from five non-target host groups and 150 fecal/wastewater samples from three target host groups. According to Equations (1) and (2), the specificity and sensitivity of the H8 marker to human-associated E. coli strains were 96 and 36%, respectively. The specificity of the H8 marker was found to be at a high-performance level for using this marker for a fecal source tracking method as suggested by the US EPA microbial source tracking guide document (US EPA 2005). However, the number of H8 positive samples between influent and effluent were low and not significantly different. This indicated that the survivability of E. coli in influent and effluent was not significantly different for H8 positive E. coli and negative E. coli. The samples were collected from a domestic wastewater treatment plant with activated sludge process and the effluent discharged from the plant without disinfection. In addition, the sensitivity of the H8 marker in this study was a quite low percentage. The reason for the low sensitivity of the H8 genetic marker was that the H8 sequence was quite sensitive to cattle. For example, in the living conditions of a community with close livestock animal contact it was found that the evident components of the gut microbiome were being shared between children and cattle (Mosites et al. 2017). In addition, the consumption of cow's milk products can transfer dairy cattle-associated bacteria to the human gut (Milani et al. 2019). On the other hand, Yersinia enterocolitica is often a contaminant in meat foods such as beef, lamb, pork, and animal products such as raw milk, which mainly causes gastrointestinal manifestations in humans (Rahman et al. 2011). The genera like Yersinia which are found in animal intestines, whose sequences are also assimilative by BLAST searches with the H8 maker and positive PCR amplifications, may originate from them (Gomi et al. 2014).

Furthermore, small livestock farms in Thailand have run their businesses ranging from backyard animal farming to industrial livestock farming enterprises (FAO 2002). The large concentration of animals and large farm sizes are causing environmental stress. Livestock animals around the villages therefore have more opportunity to be exposed to human feces in such living conditions, which probably increases the false negative rate with the H8 marker.

The performance of the H8 marker was compared to the three previous studies carried out in Japan (sensitivity: 30%, specificity: 99%), Australia (sensitivity: 45% and specificity: 94%), and Bangladesh (sensitivity: 16%, specificity not confirmed) (Gomi et al. 2014; Warish et al. 2015; Harada et al. 2018). Table 4 shows comparison of the H8 performance from the four studies.

According to this study, this marker seems to have good host-specificity performance (more than 80%) and is comparable to the performance in Japan, Australia, and Thailand. However, there is no data of the specificity in Bangladesh and other regions of the developing world. The sensitivity of the H8 marker in this study was 36% (septage samples from septic tanks), which is comparable to other studies such as in Japan and Australia which used sanitary wastewater samples. However, the sensitivity of the H8 marker in Bangladesh, where E. coli isolates from wastewater samples from a community toilet were analyzed, was quite low. The reasons behind this could be that animal feces (e.g. mice and ruminants) might accidentally mix into the community toilet, and human distal gut microbiota in Bangladesh could be different from other countries (Harris et al. 2016; Harada et al. 2018). Therefore, the performance of the fecal source tracking method has been significantly affected by gut microbiomes (Wu et al. 2011) and different geography (Yatsunenko et al. 2012). From these results, the performance of the H8 marker in Thailand was projecting a similar trend as compared with other countries. Thus, there is a good possibility of the usefulness of a human-associated E. coli genetic marker for fecal source tracking in water environments in Thailand.

The performance (specificity and sensitivity) of the H8 marker in Thailand was evaluated and compared with three previous studies in Japan, Australia, and Bangladesh. The further purpose was to find a suitable human-specific E. coli genetic marker for identifying a human source of fecal pollution in environmental water for fecal source tracking in Thailand. In the present study, the specificity and sensitivity of the H8 marker for detecting human-associated E. coli strains were about 96 and 36%, respectively. Among non-target and target host group samples, the H8 marker was found to have good host-specificity and sensitivity compared with the studies in the three other countries. However, the sensitivity of the H8 marker was quite low because target samples were mixed with nonpoint sources from a combined sewerage system, which is a common practice used in Thailand at wastewater treatment plants. Furthermore, cattle's gut microbiome and cow milk consumption could easily deliver cattle-associated bacteria to humans. In this study, the E. coli isolates were screened and selected from eight host groups (animal and human samples). Therefore, this approach for the assessment of the H8 genetic marker is recommended for various types of animal and human sources from new geographical areas, because of the similarity of H8 sequences with Yersinia genera and closeness to cattle. In addition, some other factors may be different in each area such as gut microbiomes, animal husbandry practices, and living conditions, causing possible detection of false positives of the H8 marker. Therefore, it is crucial to evaluate the performance and suitability of this E. coli genetic marker for identifying human sources of fecal pollution in surface water.

This study was supported by the Faculty of Graduate Studies, Mahidol University and the Graduate School of Global Environmental Studies, Kyoto University.

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