Microbial source markers assessment in the Bogotá River basin (Colombia)

Pollution of water bodies with human and animal fecal wastes poses a clear risk to human health via waterborne pathogens (Ritter et al. 2002; WHO 2003; US EPA 2004). Among them, the most important ones are those following the fecal–oral route of infection, and within them, we found those causing acute gastroenteritis and diarrhea. According to the WHO (2008), infectious diarrhea is one of the main contributors to the global burden of disease with 4 × 10⁹ estimated cases, and also being one of the leading causes of death, with a little more than two million deaths yearly. Human fecal contamination is considered more significant than animal fecal pollution in terms of human health risk. However, in some instances, fecal contaminants of animal origin can also have a significant impact on microbial water quality and the risk for human health (WHO 2003). Moreover, recent studies indicate that the risk for human health associated with the fecal residues of different animals varies according to the species. Thus, risk associated with exposure to water impacted by cattle feces are greater than risks associated with exposure to water contaminated with gull, pig, or chicken feces (Soller et al. 2010).

Monitoring pathogens in water samples is unrealistic not only because it is costly and technically complex but also because pathogens are irregularly present in water bodies (Savichtcheva & Okabe 2006). For more than 100 years, the approach to evade this restraint has been to monitor for surrogate indicators, especially bacteria such as total coliforms, fecal coliforms, Escherichia coli, and enterococci (Leclerc et al. 2001; Tallon et al. 2005). These microorganisms are frequently present in the lower intestine of homoeothermic animals including humans. Hence, the use of these indicators does not provide information about the host species that are the source of the fecal contamination in a given site. Fecal pollutants can reach water bodies through the direct discharge of fecal waste or untreated wastewater as well as secondary effluents from wastewater treatment works, combined sewer and sanitary overflows, animal farming activities, the wastewaters from abattoirs and the meat industry, and wildlife (US EPA 2004). The impact of each of these fecal sources to the pathogen load varies greatly among watersheds. Therefore, determining the source of fecal contamination in aquatic environments is essential not only to better estimate the health risks associated with fecal pollution, but also to facilitate measures to remediate contaminated waterways, and to determine legal responsibility for remediation.

Different approaches are being used to track fecal sources. The suitability of chemical markers has focused on the distribution pattern of chemicals such as caffeine, fragrance substances, fluorescent whitening agents, and fecal sterol isomers (Leeming et al. 1996; Standley et al. 2000; Hagedorn & Weisberg 2011). The use of eukaryotic mitochondrial DNA has also been explored to discern sources in fecal contaminated surface water (Caldwell et al. 2011). In addition, a variety of microorganisms has been considered for fecal source tracking purposes. The use of microorganisms for fecal source tracking is known as microbial source tracking (MST). MST is at present the most widespread approach for tracking the sources of fecal pollution.

Pathogenic and commensal microorganisms determined by culture-dependent and molecular methods, associated either with library-dependent or library-independent schemes, have been studied and reviewed as potential tools to identify human fecal sources. Library-dependent schemes are those associated with a database obtained in a given area (Simpson et al. 2002; Field & Samadpour 2007; Blanch et al. 2008; Hagedorn et al. 2011; Harwood et al. 2013). However, field studies using the available methods to track sources of fecal contamination have shown that a single marker or method usually is not enough to identify the source and that ‘toolboxes' of a variable number of markers or methods are needed (Simpson et al. 2002; Stewart et al. 2003; Santo Domingo et al. 2007; Blanch et al. 2008). Among others, one of the reasons for the imperfection of many of the methods available is that their sensitivity and specificity vary among different geographical areas (Hartel et al. 2002; Moore et al. 2005; Ebdon & Taylor 2006; Hansen et al. 2009; Reischer et al. 2013; Jofre et al. 2014).

Although data on the non-discriminating indicators tested here are available from Colombia and South America (Lucena et al. 2003), data on the adequacy of common microbial source tracers from these areas are non-existent or scarce. Here, we selected a number of the markers used elsewhere and evaluated their adequacy in the area of Bogotá. Bacteriophages infecting some strains of Bacteroides (Jofre et al. 2014), one of them isolated and selected in this work, and the sorbitol-fermenting bifidobacteria (SFBIF) (Blanch et al. 2006) were examined as culture methods. Targeted polymerase chain reaction (PCR) with Bifidobacterium adolescentis (ADO) and Bifidobacterium dentium (DEN) markers (Bonjoch et al. 2004; Blanch et al. 2006), Bacteroidetes marker HF183F (Bernhard & Field 2000), which are specific to human sources, and Bacteroidetes marker CF128F, which is specific to ruminant sources (Bernhard & Field 2000) were assayed as molecular methods. All these host-specific markers and E. coli, somatic coliphages (IAWPRC 1991) and bacteriophages infecting strain RYC2056 of Bacteroides fragilis (Jofre et al. 2014), which are markers of general fecal pollution, were determined in different human and animal wastewater sources and in the Bogotá River. Then, the numerical ratios of the concentrations of somatic coliphages/phages infecting selected strains of Bacteroides (Muniesa et al. 2012) and the ratio SFBIF/total bifidobacteria (TBIF) (Blanch et al. 2006) were also calculated as fecal source markers.

Municipal raw sewage samples, assumed to contain mainly human fecal contamination, were collected at the influents of different municipal wastewater treatment plants (WWTPs) in Bogotá. Untreated wastewater was also collected from various abattoirs, slaughtering mostly cows, in the Bogotá area. River water samples were gathered from either the Bogotá River or its tributaries. Eleven of the sampling sites were located in a section of the river unaffected by contaminating inputs from the Bogotá metropolitan area receiving the fecal contaminants from small towns, small slaughterhouses, and scaled-down cattle-raising activity. The other four sampling sites were downstream from the city of Bogotá where the waterway has received high inputs of municipal wastewaters and wastewater from three very large slaughterhouses. Water samples were kept at 4 °C and tested within 12 h post-collection.

Escherichiacoli was enumerated by membrane filtration using Chromocult coliform agar (Merck, Darmstadt, Germany) supplemented with antibiotics (E.coli/coliform selective supplement; Merck, Germany), and incubated for 24 h at 37 °C. The dark blue/purple colonies on the Chromocult agar were confirmed as E. coli by the addition of Kovac's reagent. Prior to filtration, the wastewater samples were diluted 1:10 in saline peptone, while the cattle manure samples were submitted to an elution step made according to Ebdon et al. (2007).

Human bifidobacteria sorbitol-fermenting agar (HBSA) (Mara & Oragui 1983) was used for the enumeration of the TBIF and SFBIF. Water samples were filtered through 0.45-μm pore size membranes (Millipore, Bedford, MA, USA) that were transferred onto HBSA according to the described procedures and incubated for 48 h at 37 °C under anaerobic conditions (GasPak; BBL, Hampshire, UK) with atmosphere generators of CO₂ (AnaeroGen; Oxoid, Basingstoke, UK). Yellow colonies were counted as SFBIF as described previously (Bonjoch et al. 2005).

Bacteriophages were enumerated after filtration of the sample through low protein-binding hydrophilic Polyethersulfone 0.22-μm pore size membrane filters (Millipore, Bedford, MA, USA) and the phages were enumerated as plaque forming units (PFUs) by the double agar layer technique. Somatic coliphages (SOMCPH) were enumerated on host strain E. coli WG5 in accordance with the ISO standardized procedure (Anonymous 2000). Bacteroides infecting bacteriophages were enumerated using the double agar layer plaque assay according to the ISO standardized method (Anonymous 2001). The host strains of Bacteroides tested were Bacteroides fragilis RYC2056 (Puig et al. 1999), Bacteroides tethaiotaomicron GA17 (Payan et al. 2005), both isolated in Spain; Bacteroides tethaiotaomicron HB13, isolated from fecal samples from Colombia (Payan et al. 2005), and Bacteroides fragilis CA8 (this work).

Bacteriophages were concentrated from 1 l of river water samples according to the method described by Méndez et al. (2004). MgCl₂ was added to the water samples to a final concentration of 0.05 mol l^–1. Briefly, the sample was then passed through an acetate-nitrate cellulose ester membrane filter (Millipore, Bedford, MA, USA), 0.22-μm pore size, and 47-mm diameter, at a filtration rate of about 2 l h^–1. Then, the membrane was cut into eight fragments and placed in a glass flask containing 5 ml eluting solution (1% beef extract, 0.05 mol l⁻¹ NaCl and 3% Tween 80). The flask was placed in an ultrasound-cleaning bath for 4 min and the eluted bacteriophages were counted as described previously.

Samples of municipal wastewater were examined for new Bacteroides host strains specific to bacteriophages of human origin. The isolation of presumptive Bacteroides strains, the preparation of phage suspensions, and the screening of specific strains was performed as in Payan et al. (2005). First, black colonies with a black or dark halo grown on Bacteroides bile esculine agar were isolated as presumptive Bacteroides. Then, the isolates that were Gram-negative rods growing only under strict anaerobic conditions, propagating satisfactorily in BPRM (Bacteroides phage recovery medium) broth (Anonymous 2001), detecting bacteriophages in phage suspensions from municipal wastewater and failing to detect phages from abattoir wastewaters were selected as Bacteroides host strains potentially able to detect preferably bacteriophages present in waters contaminated with human fecal residues.

The CA8 strain, which was a Gram-negative rod, showing strict anaerobic growth, adequate growth in BPRM, and positive detection of phages in municipal wastewater and not in slaughterhouse wastewater was characterized by sequencing the 16S RNA gene. The DNA of CA8 strain was extracted and purified with Wizard SV Gel Kit and PCR Clean-up System (Promega, Madison, WI, USA) was used as template. PCR amplification of the 16S rRNA gene was performed according to Weisburg et al. (1991). The primers used were Upper Bact 16S (GCTACCTTCTTACGACTT) and Lower Bact 16S (GAGTTTGATCCTGGCTC) (Weisburg et al. 1991) with an annealing temperature of 50 °C. A 5 μl volume of each PCR product was analyzed by agarose (1.5%) gel electrophoresis, and bands were visualized by staining with ethidium bromide. The 1,488-kDa PCR products obtained were used directly for sequencing. Sequencing was performed with the Sanger method (Sanger & Coulson 1975). Searches for homologous DNA sequences in the EMBL and GenBank database libraries were performed with the BLAST tool (http://www.ncbi.nlm.nih.gov).

DNA was directly extracted from 250 μl of slaughterhouse and wastewater samples using the QIAamp^®DNA Blood Mini Kit (Qiagen GmbH, Hilden, Germany) in accordance with the manufacturer's instructions. The DNA was eluted in 100 μl of the elution buffer provided by the manufacturer.

Specific PCR primers, HF183F and CF128F, designed by Bernhard & Field (2000) were used to discriminate Bacteroidetes from human and ruminant sources of fecal contamination, respectively. These primers were designed as a function of the uncultured markers obtained from an analysis of the Bacteroidetes communities in feces using the terminal restriction fragment length polymorphism (T-RFLP) technique. The PCR amplification was carried out according to Ballesté et al. (2010). Each 25 μl PCR mixture contained the following: 1X buffer (Kit TucanTaq, CorpoGen, Colombia), each primer at a concentration of 0.2 μM, 0.625 U of Taq polymerase (Kit TucanTaq, CorpoGen, Colombia), 1.5 mM MgCl₂, 0.2 mM each nucleotide, and 1 μl of the extracted DNA. The reaction was performed in an MJ Research PTC-100 thermal cycler. The conditions were as follows: an initial denaturation at 94 °C for 2 min, 35 cycles consisting of 94 °C for 1 min, a suitable annealing temperature for every marker for 1 min (62 °C for the ruminant marker and 63 °C for the human marker) and 72 °C for 1.5 min, followed by a final 7 min extension at 72 °C. The reaction mixture was cooled down to 4 °C. The amplicons were separated after electrophoresis on 3% agarose gels and visualized with ethidium bromide staining.

The procedure consisted of two amplification steps. First, a PCR was performed using Bifidobacterium genus-specific primers, lm26 and lm3 (Kaufmann et al. 1997). Then, PCR products were used for PCR with sets of primers specific for Bifidobacterium spp. ADO-DEN multiplex PCR was used to detect B. adolescentis and B. dentium in all samples as described previously (Bonjoch et al. 2004); briefly, 1X buffer (Kit TucanTaq, CorpoGen, Colombia), the species-specific primers for B. adolescentis at 0.25 μM concentration (Matsuki et al. 1999), the species-specific primer for B. dentium at 0.50 μM concentration, 0.04 U of Taq polymerase (TucanTaq, CorpoGen), 0.5 mM MgCl₂, 0.2 mM each nucleotide, and 1 μl of the extracted DNA. PCR parameters in the MJ Research PTC-100 thermal cycler were the following: an initial denaturation at 94 °C for 5 min, 35 cycles consisting of 94 °C for 20 sec, a suitable annealing temperature for 20 sec at 55 °C and 72 °C for 30 sec, followed by a final 5 min extension at 72 °C. The reaction mixture was cooled down to 4 °C. The amplicons were separated after electrophoresis on 3% agarose gels and visualized with ethidium bromide staining. Positive controls were performed using B. adolescentis DSM 20083 and B. dentium DSM 20084.

Statistical computations and tests were performed using Statgraphics Plus software (version 5.1; Rockville, MD, USA). Differences were considered significant at P < 0.05 as determined by the appropriate comparative test.

Thirty-six black colonies on Bacteroides bile esculine agar were selected from two samples of municipal sewage. All the corresponding isolates were Gram-negative rods, but only 15 were strict anaerobes. Six of them grew in BPRM and four detected bacteriophages in municipal wastewater. One of them, termed CA8, detected phages in five of five municipal sewage samples with values expressed in log₁₀ units ranging from 2.0 to 4.8 PFUs per 100 ml, averaging 3.6 log₁₀ units. In contrast, phages were not detected by strain CA8 in samples of slaughterhouse wastewaters tested.

The BLASTn analysis of the sequence (deposited in GenBank with accession numbers BankIt1777297 Seq1 KP188833) of the 1,488-kDa fragment amplified from DNA of strain CA8 showed a 97% similarity with Bacteroides fragilis 16S rRNA X83940.1 and B. fragilis strain NCTC 9343.

This strain detects lower numbers of phages than other Bacteroides host strains such as Bacteroides tethaiotaomicron GA17 and HB13 do in the area of Bogotá and consequently it is of little practical interest. However, its isolation confirms that with little effort it is possible to isolate host strains for phages of Bacteroides discriminating between human and animal fecal pollution in a given geographical area, as shown before (Payán et al. 2005; Jofre et al. 2014).

Escherichia coli, SOMCPH, phages infecting strain RYC2056 of Bacteroides fragilis and TBIF, which are markers of general fecal pollution, were present in all the samples tested, independently of their origin (Table 1). As well, SFBIF were detected in all the samples tested herein (Table 1). This results regarding SFBIF are in agreement with results reported in Blanch et al. (2008), but not with those reported in Mara & Oragui (1983), where they were reported to be mostly present in samples of human origin. In contrast, the culturable markers able to discriminate fecal sources, this is bacteriophages infecting strains GA17, HB13, and CA8, were found in all samples of human origin but in none of the samples of animal origin tested. The highest values were detected by strain GA17. PFU values counted on strain GA17 ranged from 4.0 to 5.6 log₁₀ units per 100 ml of municipal wastewater. Despite being isolated in Spain, strain GA17 performed better than strains HB13 and CA8 isolated from Colombian raw sewage samples. The ratios between the concentration of somatic coliphages and bacteriophages detected by strain GA17 of Bacteroides expressed in log₁₀ units were lower than 2 in the human sources and greater than 4.6 in animal sources. These differences are statistically significant (ANOVA (analysis of variance) P < 0.05) and similar to those described elsewhere overseas (Muniesa et al. 2012). The SFBIF/TBIF ratios ranged from 0.15 to 0.30 in human sources and from 0.05 to 0.17 in animal sources. These values in human sources are lower than values reported elsewhere, whereas values in animal sources are higher (Blanch et al. 2006).

Regarding the molecular markers (Table 2), they all provided absolute specificity. Indeed, none of them was found in the non-corresponding source samples. Regarding sensitivity, ADO and HF183F were identified in all the corresponding source samples, whereas CF128F failed to be detected in 25% samples of slaughterhouse and DEN in 60% of municipal sewage samples. These results indicate that regarding specificity, all these molecular markers comply with the expected results in highly contaminated water samples in Colombia.

The values of markers with quantitative results are displayed in Table 3. They clearly show the high impact of the fecal inputs contributed by the conurbation of Bogotá. Upstream from Bogotá, only E. coli, SOMCPH, and TBIF were detected in all the samples, with maximum values expressed in log₁₀ units per 100 ml of 3.5, 3.4, and 4.4, respectively. All the values were between three and four orders of magnitude lower than those detected in raw municipal sewage and abattoir wastewater samples. SFBIF were detected in 45% of samples, and the ratio SFBIF/TBIF ranged from <0.25 to 0.6. In contrast, bacteriophages infecting Bacteroides strains RYC2056, which is non-discriminating, GA17 and HB13 were only found in one out of eleven samples when only 10 ml was analyzed. In the positive sample, the ratio SOMCPH/GA17PH was 1.2 log₁₀ units, indicating that human fecal contamination is predominant. In the rest of the samples of this group, the values of the ratio ranged from >1.0 to >2.4. These values do not provide sufficient information about the source according to available information (Jofre et al. 2014). No phages infecting CA8 were detected at concentrations >10 PFUs per 100 ml in this set of samples.

To gain insight on the limitations of the method in samples with intermediate to low concentrations of fecal indicators, bacteriophages were concentrated from 1 l volumes taken from sampling points where phages infecting Bacteroides were not detected in 10 ml volumes. Values of the parameters tested in these samples are shown in Table 4. The number of positive detections increased, thus showing that testing phages in 1 l of river water with a very simple method is significantly improving the sensitivity of the procedure. With these data, the ratio SOMCPH/GA17PH was calculated in three samples, showing two values of 2.7, which indicates the presence of human contamination as being predominant (Muniesa et al. 2012). The third sample gave a value greater than 3.5, which is unlikely to indicate the predominance of human fecal contamination (Jofre et al. 2014).

Regarding molecular markers, the only positive detections (Table 2) were ADO and DEN. They were found in a single sample, the same one in which phages infecting Bacteroides strains GA17 and HB13 were detected. In addition, this sample was one of the few samples with E. coli above 10³ colony forming units (CFUs) per 100 ml. In this particular sample, all the markers, except HF183F and phages infecting CA8, indicate a human origin of fecal contamination. None of the molecular markers was detected in the rest of the samples in this set of samples.

In contrast, and due to the great impact caused by the input of the city of Bogotá with depurated or non-depurated municipal sewage and the wastewater of several abattoirs, the number of parameters with positive detection was significantly greater (P < 0.05, ANOVA) than those observed in samples collected upstream of the Bogotá conurbation (Table 3). Values of E. coli rose to over 10⁶ CFUs per 100 ml in the city area and decreased to 4 × 10⁴ at some distance downstream. Bacteriophages infecting strain GA17 were detected in all samples, with values ranging from 1.6 to 3.4 log₁₀ PFUs per 100 ml and the ratios SOMCPH/GA17PH ranged from 1.4 to 2.2, which are characteristic values of human source contamination. Phages infecting HB13 and CA8 were not detected in one of the samples, confirming a better performance of strain GA17. In contrast, values of the ratio SFBIF/TBIF ranged from 0.11 to 0.22, which are more typical of animal fecal contamination, thus confirming the poor performance of this ratio observed in wastewaters. Regarding the molecular markers, ADO and DEN were detected in only one sample that had more than 10⁶E. coli CFUs per 100 ml. HF183F was detected in two of the samples that were also those with E. coli exceeding 10⁶ CFUs per 100 ml. CF128F was also detected in the sample where all the other molecular markers were detected.

With the exception of the ratio SFBIF/TBIF, the markers chosen are applicable for MST in the area studied. To our knowledge, this is the first study in South America providing information on such a number of MST markers. The bacteriophages infecting discriminating hosts of Bacteroides, particularly those infecting strain GA17 of Bacteroides tethaiotaomicron, as well as the molecular markers, show a good specificity and sensitivity in the municipal sewage and slaughterhouse wastewater. Only the ratio SFBIF/TBIF failed regarding specificity.

When these MST markers were evaluated in samples from a contaminated river with concentrations of E. coli below 10³ all methods showed low sensitivity and their detection will require concentration. Concentrating phages in a 1 l sample for host-specific phages infecting Bacteroides is sufficient to enable the calculation of the ratio SOMCPH/GA17PH that seem to be one of the best predictive markers, according to the data reported abroad (Muniesa et al. 2012).

Findings reported herein contribute to upgrade the current MST methodology in the region, and very likely by extension in South America, and hence to the implementation of management practices of water resources in a way aimed to decrease the health risks associated with fecal contamination. In addition, data provided herein will help to reduce health risks associated with water by allowing policy-makers to better focus correcting measures aimed at improving the sanitation of the Bogotá River, which is the main source of water for human consumption and irrigation for the most populated region of the country.

The present work was supported by Colciencias Program for Young Researchers and Innovators ‘Virginia Gutiérrez de Pineda’ and the Pontificia Universidad Javeriana, Colombia through the project ‘Isolation and identification of Bacteroides and Bifidobacterium strains for potential use to differentiate water fecal contamination’.