Management of contaminated sediments has focused predominately on chemical agents, overshadowing risks posed by pathogenic microorganisms. Current accepted bacterial indicator methods do not provide defensible data with respect to the occurrence and types of pathogens in sediments. In an effort to adapt new defensible methods for assessing the risk posed by pathogens in sediments, we evaluated the sensitivity of a commercially available real-time polymerase chain reaction TaqMan® Escherichia coli 0157:H7detection kit. The lower limit of linear quantitation of this assay was experimentally determined in sediment and sediment extract samples spiked with known amounts of E. coli 0157:H7 DNA. Parallel control experiments were conducted in pure water samples spiked with known amounts of Escherichia coli 0157:H7 DNA. The lower limit of quantification of the TaqMan® assay was 1000 colony forming units when interrogating 100 mg sediment samples. In contrast, the assay was 20-fold more sensitive with a lower limit of quantification of 50 colony forming units in pure water and sediment extract samples. These results suggest that the sensitivity of the TaqMan® Escherichia coli 0157:H7detection kit is more dependent on recovery of the desired target from the sediment matrix than efficiency of polymerase chain reaction amplification. The potential human health risk associated with the lower limit of quantification of the TaqMan® assay in the spiked sediment samples was estimated using a Beta-Poisson dose-response model. Using this approach, lower limit values corresponded to exposure levels of Escherichia coli 0157:H7 that meet United States Environmental Protection Agency accepted illness rates for recreational swimming.
As the number of beach closings and advisories continue to rise, so does the public's concern regarding microbial pathogens in recreational waters. In a survey of more than 230 U.S. coastal and Great Lake communities, there were at least 13,410 days of beach closings or advisories during 2001 (NRDC, 2002). This figure represents a 19% increase from the previous year. The majority of beach closings and advisories were based on the presence of elevated levels of fecal contamination as measured by bacterial indicators, such as Escherichia coli and Enterococci. Indicator species are easily sampled, nonpathogenic bacteria that are frequently associated with pathogens transmitted by fecal pollution. Elevated bacterial indicators in recreational waters translate into significant health and economic costs. The Centers for Disease Control and Prevention (CDC) estimates that close to a million cases of illnesses occur annually in the United States as a result of exposure to waterborne microbial pollution (ASM, 1999). Most of these illnesses will be self-limiting but as many as 1% may result in death. In addition to the human health cost, significant economic losses can be attributed to microbial pollution. Coastal waters are estimated to support 28.3 million jobs and generate 54 billion in goods and services each year (USEPA, 2000a). In response to the increased threat posed by microbial pollution to recreational waters, the Clean Water Act was amended on October 10, 2000, by the Beaches Environmental Assessment and Coastal Health Act (BEACH) of 2000 (USEPA, 2000b). The BEACH Act requires that states adopt new or revised water quality standards by April 2004 for pathogens and bacterial indicators for which EPA has published criteria under the Clean Water Act.
Frequently cited sources of microbial contamination in recreational waters are the result of the expanding human population (Young and Thackston, 1999; Jiang et al., 2001; Boehm et al., 2002). For example, in a study of bacterial loading in urban streams, it was determined that fecal bacteria densities were directly related to the density of housing, population, development, percent impervious area, and domestic animals (Young and Thackston, 1999). This situation is being exacerbated in coastal regions where it is estimated that half of the human population of the United States will reside by 2010 (NRDC, 2002). Sewage from outdated, combined and sanitary sewage overflows, along with malfunctioning sewage treatment plants and pump stations, continue to serve as main point sources of pathogens in coastal waters. On-site septic systems are also a significant source of waterborne pathogens. Nationwide there are some 25 million septic tanks receiving 175 billion gallons of human wastewater (ASM, 1999), 10 to 30% of which are not functioning properly (USEPA, 2001a). Aside from human sewage inputs, animal waste is also a significant source of waterborne pathogens (Cole et al., 1999). Collectively, corporate farming operations generate 220 billion gallons of waste annually (NRDC, 2002). Storm water events further magnify the levels of pathogens and bacterial indicators introduced into watersheds through various nonpoint sources. Pollution by urban storm water accounts for about a quarter of the nations contaminated recreational waters (NRDC, 2002). Agricultural waste also contains high concentrations of nutrients and pesticides. Nutrient pollution can lead to eutrophication of waterways and stimulate the growth of organisms like Pfiesteria, which was responsible for the death of more than a billion fish in the Chesapeake Bay region (ASM, 1999).
Once introduced into a watershed, the fate and transport of microbial pathogens are of concern. Numerous studies have been conducted looking at survival rates of different pathogens and their indicators in various environmental matrices (Burton, 1985). Microbial survival is commonly expressed using exponential decay equations such as dN/dt = – KN, where t = time, K = a rate constant, and N = the number of microorganisms (Hurst, 1997). In practice, however, long-term survival of pathogens and bacterial indicators is longer than predicted by first order rate decay equations. Various environmental factors including sunlight, temperature, pH, salinity, and predation can affect survival (Roszak and Colwell, 1987). In addition, sediments have been shown to greatly extend the survival of pathogens and bacterial indicators (LaLiberte and Grimes, 1982; Burton et al., 1987; Sherer et al., 1992; Davies et al., 1995). Once microbial pathogens enter into the water column they may become associated with suspended solids that eventually settle out and accumulate in the underlying sediments. Both bacteria and viruses possess electrostatic charges, which facilitate their adsorption onto fine-grained, charged clays and mud. Some reports indicate that sediments can contain 100 to 1000 times as many bacterial indicators as the overlying water (Van Donsel and Geldreich, 1971; Ashbolt et al., 1993). Sediments likely extend survival of pathogens and bacterial indicators because they provide nutrients as well as protection from predation and sunlight. Resuspension of sediments can result in desorption of pathogen indicators and the subsequent contamination of the overlying surface waters (Grimes, 1975, 1980).
Thus far, management of contaminated sediments has predominately focused on chemical contamination, overshadowing the risks posed by pathogenic microorganisms. While programs and guidelines are in place to regulate discharges of biological wastes (USEPA, 1999), programs or guidelines to evaluate the impacts of pathogens in sediments are rudimentary. Part of the reason for this limitation is that current standard methods recommended for measuring the levels of bacterial indicators in recreational waters are generally not appropriate for sediments (USEPA, 2000c). Furthermore, current accepted bacterial indicator methods do not provide defensible data with respect to the occurrence and types of pathogens in sediments. These methods have been largely criticized based on some studies that show that the environmental fate of pathogen indicators differs from the pathogens they proxy for (Griffin et al., 1999; Jiang et al., 2001; Kong et al., 2002). In addition, advances in the bacteriology of the coliform group indicate that thermotolerant fecal coliforms can have an environmental origin as well as exhibit a high re-growth potential once introduced into the environment (Jimenez, 1989; Leclerc et al., 2001). This can result in the over estimation of some pathogens. Recovery of bacterial indicators from the environment is also problematic because they can enter a viable but non-culturable state (Rollins and Colwell, 1986). Under these conditions indicator methods underestimate the actual number of indicator organisms present.
In an effort to adapt new defensible methods for assessing the risk posed by pathogens in sediments, we evaluated the utility of a commercially available real-time polymerase chair reaction (PCR) TaqMan® E. coli 0157:H7 detection kit using E. coli 0157:H7 spiked sediment and sediment extract samples. We chose E. coli 0157:H7 as our model organism because it has very low infective dose, that is, 10 to 100 organisms, and therefore should require a sensitive method for detection (Keene et al., 1994). Furthermore, this pathogenic coliform has been found to be frequently associated with waterborne-disease outbreaks of gastroenteritis associated with recreational waters (Barwick et al., 2000) and is likely to impact sediments. As part of the evaluation, we employed a risk-based approach to determine whether or not the sensitivity of the real-time PCR TaqMan E. coli 0157:H7 method was sufficient to detect pathogen levels at or below EPA accepted levels of risk of illness.
Materials and methods
Bottom lake sediment from a fresh water lake located on-site, known as Browns Lake (Vicksburg, Mississippi), was collected using a grab sampler and placed in a five gallon bucket. Sediments were immediately transported back to the laboratory, where they were kept refrigerated. Particle distribution was characterized as 22.8% sand, 69.4% silt, and 7.8% clay. The sediment had a pH of 7.4 and organic matter was estimated at 1.7% from the weight–loss of dried sediment upon overnight combustion in a muffle furnace at 550°C. A 500-gram sub-sample of the sediment collected was homogenized and subsequently autoclaved for 90 min on three successive days. Sterile water was added as needed to keep the sediment moist. One g and 100 mg sediment samples were individually spiked directly with 10-fold serial dilutions of E. coli 0157:H7 DNA (ATCC # 700927D, ATCC, Manassas, VA) ranging from 1 × 10−8 to 1 × 10−14 g. Corresponding colony forming units (CFUs), ranging from 1.8 × 106 through 1.8, were derived from the mass of a single E. coli 0157:H7 genome based on published genomic DNA sequences (Perna et al., 2001). Recently the European Community has reclassified E. coli 0157:H7 from a Hazard Group 2 to a Hazard Group 3 pathogen (equivalent to BSL-3) in response to a number of laboratory-acquired infections (Coia, 1998; European Parliament, 2000). For this reason we chose to work with E. coli 0157:H7 DNA. Following spiking, samples were gently mixed and immediately processed. Total DNA was extracted from spiked sediments using the alternative protocol available for the Mo Bio UltraClean Soil DNA extraction kit (Solana Beach, California) according to the manufactures instructions. The same extraction kit was used to extract a panel of individual 1 g and 100 mg non-spiked autoclaved sediments. The resulting sediment extracts were then spiked with 10-fold serial dilutions of E. coli 0157:H7 DNA as described above. To determine baseline sensitivity of the assay in the absence of sediment, the same serial dilution scheme was used to spike pure water samples. All serial dilutions were done in triplicate and subjected to Taqman PCR analysis.
A commercially available real-time PCR TaqMan® E. coli 0157:H7 detection kit (Applied Biosystems, Foster City, CA) was employed in this work. The real-time PCR method is based on the use of the 5′ nuclease assay (Holland et al., 1991), which has been optimized by the use of fluorescent TaqMan® methodology (Applied Biosystems, Foster City, California) (Heid et al., 1996). Briefly, the assay requires the standard PCR forward and reverse primers, in addition to a TaqMan® probe that hybridizes between them. The TaqMan® probe is a fluorescent-labeled oligonucleotide with a 5′-reporter dye and a 3′-quencher dye. When the probe is intact, the proximity of the reporter dye to the quencher dye suppresses the reporter fluorescence. During the elongation step of each PCR cycle, the 5′ nuclease activity of DNA polymerase degrades the annealed Taqman probe. As a result, the reporter and the quencher become separated, leading to an increase in fluorescence emission. Fluorescence increases logarithmically as each cycle of the PCR proceeds, until reagents become limiting. The point at which the fluorescence level rises appreciably above background levels is considered the threshold cycle, or CT. There is a linear relationship between the log of the starting amount of template and the corresponding threshold cycle during real-time PCR. A standard curve can be constructed from known starting amounts of target DNA by plotting the log of starting amount versus the threshold cycle.
All real-time PCR amplification reactions were performed using an optical thermal cycler (BioRad, Hercules, CA). Thermal cycling conditions were initiated at 95°C for 9 min followed by 50 cycles of 95°C for 20 sec, 60°C for 1 min and 72°C for 30 sec. Five μl of spiked samples were added to 45 μl of a pathogen PCR cocktail (Applied Biosystems, Foster City, California). Amplification negative controls and template negative controls were added as recommended by the manufacturer.
Lower limit of linear quantitation (LLQ) calculations were based on the divergence from linearity observed when CT values were plotted relative to known E. coli 0157:H7 CFU's in spiked and control samples. A least-squares linear fit was made to the four most concentrated serial dilutions and the trend line subsequently extended for visualization. The linear fit to semilog data is based on the amplification process being exponential in this range, such that the detector signal depends on the initial amount of DNA undergoing amplification (A) for a defined number of cycles(N), expressed as constant = log[initial] + N log A.
The sensitivity of a commercially available real-time PCR TaqMan® E. coli 0157:H7 detection kit was evaluated using sediment. To generate a sensitivity curve, increases in relative fluorescent units (RFU) were monitored over the duration of the polymerase chain reaction (Figure 1). Taqman CT values were defined across the logarithmic increase phase of each PCR reaction such that fluorescence from no-template controls was avoided. These values were plotted relative to the corresponding log of E. coli 0157:H7 CFUs to generate a sensitivity curve (Figure 2). Linearity between the CT values and CFUs was observed over 6 orders-of-magnitude of the dilution series for the water-spiked control samples (Figure 2). An amplification factor of 1.94 was derived from the slope of the linear fit. This value approaches the maximum theoretical PCR amplification factor 2. A lower limit of linear quantitation (LLQ) of 50 CFUs was estimated based on the data's graphical divergence from linearity. In contrast, linearity between the Taqman CT values and CFUs for spiked 100 mg sediment samples was reduced approximately an order-of-magnitude (Figure 3). The estimated LLQ of the assay in sediments corresponded to 1000 CFUs. An amplification factor of 1.83 was derived from the slope of the linear fit, suggesting that efficiency of PCR was reduced in the presence of sediments. At higher spiked sediment sample sizes (1 g), TaqMan® assay performance was poor due to low sensitivity and high variability (Figure 4). Because of the variability in the data, LLQ values were not calculated for this data. Interestingly, E. coli 0157:H7 post-spiked sediment extracts, derived from 1 g sediment samples, yielded similar results as the water spiked controls seen in Figure 1 (data not shown).
The health risk associated with the LLQ for the E. coli 0157:H7 TaqMan® detection kit in spiked sediments was evaluated. Currently, the accepted levels of risk promulgated by the EPA for recreational waters are no more than 14 illnesses per 1000 swimmers (0.014) for fresh waters, and no more than 19 illnesses per 1000 swimmers (0.019) for marine waters (USEPA, 1986). These values were derived from epidemiological studies in which bacterial indicator levels were correlated with gastrointestinal illness and therefore, likely represent daily risk of illness. For this exercise, we relied on a Beta-Poisson dose-response model:Haas et al., 2000). Based on these values, the probability of illness associated with the LLQ of the E. coli 0157:H7 TaqMan® method in sediments is 0.0025. This value is significantly lower than EPA risk guidelines for freshwaters (0.014) and marine waters (0.019), suggesting that, under the conditions we tested, this method is sufficiently sensitive to detect E. coli 0157:H7 at levels corresponding to EPA accepted illness rates. Efforts to extrapolate risk values for recreational water exposure on an annual basis have been proposed (Loge et al., 2002). However, the appropriateness of this method for sediments is questionable given the lack of information of exposure data for pathogens in sediments.
In an effort towards adapting new defensible methods for assessing the risk posed by pathogens in sediments, we evaluated the sensitivity of a commercially available real-time PCR TaqMan® detection kit in sediments. Assays based on PCR technologies offer a powerful alternative for pathogen detection because they are rapid and extremely sensitive, capable of detecting a single organism (Sachse and Frey, 2003). In comparison, traditional indicator methods are time-consuming, laborious and provide only indirect evidence for the presence of pathogenic organisms. Water resource managers need rapid monitoring tools so that they can quickly respond to changes in recreational water quality. We chose E. coli 0157:H7 as our model organism because it is frequently associated with waterborne-disease outbreaks of gastroenteritis in recreational waters and has a low infective dose (Keene et al., 1994; Barwick et al., 2000). In comparison to other microbial pathogens associated with gastroenteritis (Table 1), as few as 10 E. coli 0157:H7 organisms may be sufficient to cause illness. Given such a low infective dose, a sensitive method, such as the one applied in this study, is required to effectively determine E. coli 0157:H7 concentrations. Previously, a similar TaqMan® detection kit from the same commercial vendor was evaluated for its ability to detect Salmonella in meat samples (Kawasaki et al., 2001). This study found that the Taqman PCR method was superior to the conventional culture method for routine detection of Salmonella from meat samples. We therefore wondered how such a detection kit would perform using a complex environmental sample like sediment. Soils and sediments have been found to be particularly inhibitory to PCR due to the presence of humic acids and other high molecular weight compounds (Wilson, 1997; Watson and Blackwell, 2001). Commercially available DNA soil extraction kits, such as the one used in this study, offer additional extraction steps to eliminate or reduce these compounds. In addition to removing PCR inhibitors, complex environmental samples have the added complication of genetic heterogeneity. TaqMan probes and primers specific for one gene target may cross-react with another similar gene target. Recently, an analysis of environmental E. coli isolates using the TaqMan PCR system found that the assay generated a number of false positive identifications when compared to serotyping analysis (Davis et al., 2003). While this is of concern for field application of the TaqMan® E. coli 0157:H7 detection kit, we circumvented this problem in our study by choosing to work with a genetically homogeneous sample of E. coli 0157 DNA.
Our results indicate that in the presence of as little as 100 mg sediment, the LLQ of the Taqman® assay was inhibited 20-fold to 1000 CFUs. At higher spiked sediment sample sizes (1 g) TaqMan® assay performance was less reliable due to low sensitivity and high variability. Reduction in TaqMan® assay performance, in the presence of sediments, is the likely result of two factors, 1) inhibition of the PCR reaction and 2) poor recovery of E. coli 0157:H7 DNA from spiked sediments. The decrease in amplification efficiency of the PCR reaction observed in sediments (1.83) suggests that the PCR reaction was slightly inhibited. However, the observation that sediment extracts spiked with E. coli 0157:H7 were minimally inhibited in the assay suggests poor recovery of E. coli 0157:H7 from spiked sediments. Great care was taken to extract DNA from sediment samples immediately following addition of E. coli 0157:H7 DNA. It is possible that a fraction of the spiked DNA irreversibly bound to the sediment making it unavailable. Therefore, the sensitivity of the TaqMan E. coli 0157:H7 detection kit is more dependent on sample processing rather than PCR amplification efficiencies when processing complex environmental samples such as sediments. Similarly, a related study aimed at interpreting PCR results from water matrices in a human health risk context concluded that significant improvements were needed in sample processing (Loge et al., 2002). Because E. coli 0157:H7 concentrations can vary substantially in natural sediment samples (Grimes, 1975, 1980; Jimenez, 1989), further evaluation of the real-time PCR TaqMan® E. coli 0157:H7 detection kit is warranted. Detection of low infectious dose pathogens, such as E. coli 0157:H7, will likely require larger sample sizes (>1 g) to be processed. Increased sample size can reduce sensitivity, reliability, and reproducibility of the real-time PCR TaqMan® assay, as seen in the spiked 1-g sediment samples of this study.
As part of our evaluation of the TaqMan® E. coli 0157:H7 detection kit, we employed a risk-based approach to determine whether the sensitivity of the method was sufficient to detect pathogen levels at or below EPA's acceptable levels of risk of illness. For this exercise we relied on the Beta-Poisson dose-response model, a model that has been shown to describe accurately dose-response data for E. coli 0157:H7 (Haas, 2000). Two assumptions are commonly made when using this model, 1) the host must ingest at least 1 organism capable of causing disease, and 2) a portion of the ingested organism must survive in the host to cause illness (Haas, 1999). From this exercise we concluded that our LLQ values for spiked sediment corresponded to exposure levels that meet EPA risk guidelines. While this result is encouraging, there were a number of knowledge gaps we encountered during the process. For example, exposure pathways and dose-response data for most waterborne pathogens in sediment matrices is either lacking or incomplete. Information, such as the temporal and spatial distribution of pathogens in sediments is not known. As a result, current risk assessments for pathogens rely heavily on computer modeling. Furthermore, EPA risk estimates are derived from poorly defined epidemiological data resulting from exposure to bacterial indicators in water, not sediment.
While the current methods for monitoring microbial pathogens are imperfect, there are no universally accepted alternatives to replace existing ones. The future of monitoring for microbial pathogens will likely rely on a matrix approach placing less emphasis on any one single parameter. The toolbox of potential new methods for detecting waterborne pathogens is extensive yet largely untested for environmental applications (Rose and Grimes, 2002). Technologies currently used in clinical medical research offer powerful alternatives to traditional indicator methods. New technologies, like the one tested in this study, potentially offer a high-throughput format for analyzing multiple pathogen species in a single assay. Nonetheless, interpretation of PCR data in a risk assessment context is complicated by the lack of basic information required to conduct risk assessments. In the mean time, the current imperfect methods are being relied upon more and more to make increasing complex regulatory decisions. Issues such as pathogen total maximum daily loads (TMDLs) are likely to come to the forefront as State and Federal agencies struggle to address controls of microbial contamination in watersheds (USEPA, 2001b).
This work was funded by the Long-term Effects of Dredging Operations (LEDO) program at the U.S. Army Engineer Research and Development Center (ERDC). Permission was granted by the Chief of Engineers to publish this information.