|Home | About | Journals | Submit | Contact Us | Français|
For decades, untreated sewage flowing northward from Tijuana, Mexico, via the Tijuana River has adversely affected the water quality of the recreational beaches of San Diego, California. We used quantitative reverse transcription-PCR to measure the levels of hepatitis A virus (HAV) and enteroviruses in coastal waters near the United States-Mexico border and compared these levels to those of the conventional fecal indicators, Escherichia coli and enterococci. Over a 2-year period from 2003 to 2005, a total of 20 samples were assayed at two sites during both wet and dry weather: the surfzone at the mouth of the Tijuana River and the surfzone near the pier at Imperial Beach (IB), California (about 2 km north of the mouth of the Tijuana River). HAV and enterovirus were detected in 79 and 93% of the wet-weather samples, respectively. HAV concentrations in these samples ranged from 105 to 30,771 viral particles/liter, and enterovirus levels ranged from 7 to 4,417 viral particles/liter. The concentrations of HAV and enterovirus were below the limit of detection for all dry weather samples collected at IB. Regression analyses showed a significant correlation between the densities of both fecal bacterial indicators and the levels of HAV (R2 > 0.61, P < 0.0001) and enterovirus (R2 > 0.70, P < 0.0001), a finding that supports the use of conventional bacterial indicators to predict the levels of these viruses in recreational marine waters.
Exposure to marine recreational waters of poor microbial quality has been linked to multiple adverse health outcomes (28). Large numbers of viruses are excreted in human feces from infected individuals in a community (22). While diarrhea has been one of the primary manifestations of human infection in polluted marine bathing waters, it is now recognized that more serious chronic diseases are also associated with viral infections (10). For decades, flows of raw sewage from the city of Tijuana, Mexico, have entered the Tijuana River and posed a threat to public health at the marine recreational beaches in the coastal South Bay communities of San Diego, California (3). In this region of the U.S.-Mexico border where the incidence of endemic diarrheal diseases, as well as infectious hepatitis, has been shown to be significantly elevated compared to nonborder regions of the United States (6), the impact of sewage discharged from Mexico into coastal marine waters of the United States is of increasing concern.
In 2000, the U.S. Congress passed the Beaches Environmental and Coastal Health (BEACH) Act, which requires states with marine recreational waters to adopt water quality standards based on U.S. Environmental Protection Agency (EPA) criteria published in 1986. Changes made to California's monitoring standards (25) in 1998 required the adoption of enterococci as an indicator (based on the U.S. EPA criterion) of marine recreational water safety to supplement existing standards for total and fecal coliforms. These bacteria are meant to serve as indicators of human fecal contamination in water; however, they are not the major cause of human water-related illness. On the contrary, waterborne illness is believed to be most often due to viruses of human fecal origin (10). Because of the ability of certain viruses to persist in the marine environment longer than bacteria, risks based on bacterial standards may seriously underestimate the risk of virus-associated waterborne illness (12, 13, 21).
Currently, the detection and quantification of waterborne viruses by conventional cell culture assay is difficult and time-consuming. In addition, because hepatitis A virus (HAV) is not easily cultured, there is little published data on HAV levels in ocean waters contaminated by sewage. Brooks et al. (1) developed a SYBR green quantitative reverse transcription-PCR (qRT-PCR) method to detect and enumerate HAV in seawater. We adapted this method to measure the levels of HAV and used a commercially available qRT-PCR kit to measure enterovirus levels in coastal waters near the U.S.-Mexico border. We now present a quantitative assessment of the relationship between the levels of HAV and enterovirus and the fecal indicator bacteria, Escherichia coli and enterococci, in these marine waters.
Twenty ocean water samples were collected from two locations: 0.2 km north of the Tijuana River mouth and the south side of the Imperial Beach (IB) pier in San Diego, California (Fig. (Fig.1).1). Seven rain events were sampled (during the wet season, late October through April) in these two locations for a total of 14 samples. Each sample was collected after a rain event, which was defined as precipitation of 0.5 cm or more in a 72-h period. This definition is based on the San Diego County Department of Environmental Health general advisory, which warns the public of possible water contamination by urban runoff. In addition, six dry weather samples were collected (during the dry season, May through early October) from the IB pier in order to determine the microbial water quality at the time of highest recreation use. During the dry season, samples were not taken from the mouth of the Tijuana River, since the flow of the Tijuana River at this time is negligible or even zero. Samples ranged from 400 ml to 8 liters and were collected between 2003 and 2005. Water temperature, river flow, precipitation, and tidal height were noted for each sample at the time of collection.
To determine E. coli and enterococcus fecal bacterial levels, 100-ml water samples were collected and processed within 2 h of collection. Up to three 10-fold serial dilutions of each water sample were subjected to the Colilert 18 and Enterolert test methods (IDEXX Laboratories, Westbrook, ME) according to the manufacturer's instructions. The detection limit for these methods was 10(most probable number [MPN])/100 ml for both types of bacteria.
Each sample was processed within 1 to 2 h of collection according to a published protocol by Katayama et al. (15). Seawater samples were filtered at a constant rate via a vacuum pump through a series of Whatman filters (11- and 2.5-μm pore sizes) to reduce the particulate matter. Although it is well understood that viruses can adsorb to particles in the environment, the removal of particulates is necessary for PCR assays. Samples were then applied to a type HA 0.45-μm-pore-size negatively charged membrane (Millipore, Burlington, MA). The negatively charged filter was washed with 200 ml of 0.5 mM H2SO4 (pH 3.0) to remove cations, and the virus was eluted from the filter with 10 ml of 1 mM NaOH (pH 10.5 to 10.8) into a tube containing 0.1 ml of 50 mM H2SO4 and 0.1 ml of 100× Tris-EDTA buffer (pH 8.0; Sigma-Aldrich, St. Louis, MO). The filtrate was then concentrated to a 450-μl volume by centrifuging the samples in a Centriprep Concentrator (YM-30; Millipore) at 1,500 × g for 15, 10, and 5 min, consecutively. Total RNA was extracted from the 450-μl filtrate by using TRI Reagent (Molecular Research Center, Inc., Cincinnati, OH), and the RNA pellet was dissolved in 40 μl of Tris-EDTA buffer (pH 8.0).
Procedures for cDNA synthesis and SYBR green qRT-PCR were performed as described by Brooks et al. (1), except that a Bio-Rad iCycler real-time thermocycler was used instead of the Applied Biosystems GeneAmp 5700 sequence detection system. The cDNA was synthesized using random hexamer primers and the GeneAmp gold RNA PCR core kit (Applied Biosystems, Foster City, CA) in a total reaction volume of 40 μl. The cDNA samples were diluted 1:10 and 1:100 with DNase-, RNase-free water containing sonicated herring sperm DNA (5 ng/ml) as the carrier DNA (17). The SYBR green qRT-PCR amplification was carried out in a 25-μl reaction volume that contained 7.1 μl of 2× iQ SYBR green Supermix (Bio-Rad Laboratories, Hercules, CA), 300 nM each of the forward and reverse primers (Table (Table1)1) , and 1 μl of undiluted stock or diluted cDNA. qRT-PCR analyses of all samples were performed in triplicate. The thermal profile for SYBR green qRT-PCR was 95°C for 10 min, followed by 40 cycles of 95°C for 10 s and 60°C for 1 min.
The enterovirus qRT-PCR was accomplished by using a One-Step RT-PCR kit (QIAGEN, Valencia, CA) and an MGB alert enterovirus real-time PCR kit (Nanogen, San Diego, CA). The enterovirus kit contained a 20× primer mix, as well as a 20× MGB eclipse probe (Table (Table1)1) directed toward the 5′-untranslated region (5′-UTR) of enteroviruses (coxsackie A and B viruses, echoviruses, polioviruses, and enteroviruses 68 to 71). The RNA samples were diluted 1:10 and 1:100 with DNase-, RNase-free water containing sonicated herring sperm DNA (5 ng/ml) as the carrier DNA (17). Each 50-μl reaction mixture contained 17 μl of RNase-free water, 10 μl of 5× buffer, 10 μl of 5× Q-Solution, 2.0 μl of dNTP mix, and 2.0 μl of enzyme mix (all components of the QIAGEN kit); 2.5 μl of 20× forward/reverse primer mix (Table (Table1);1); 2.5 μl of 20× MGB eclipse probe (Table (Table1);1); and 4.0 μl of undiluted or diluted template RNA.
Samples were run in duplicate on a Bio-Rad iCycler real-time PCR system. The qRT-PCR conditions were as follows: RT for 30 min at 50°C, polymerase activation for 15 min at 95°C, and 50 cycles of denaturation for 10 s at 95°C, followed by annealing and detection for 30 s at 56°C and extension for 30 s at 76°C, with a final extension step for 10 min at 76°C.
Standard curves were generated with an HM-175 strain of HAV (VR-2089; American Type Culture Collection [ATCC], Manassas, VA) and a W-2 strain of poliovirus 2 (VR-301; ATCC). Viral particles were fixed in 4% paraformaldehyde, stained with SYBR gold, and directly counted by using epifluorescence microscopy (5, 8). The concentrations of the HAV and poliovirus 2 stocks were 8.6 × 106 viral particles/μl and 1.7 × 106 viral particles/μl, respectively. Titers of the HAV and poliovirus 2 stocks were initially determined by using a 50% tissue culture infective dose approach (8) as 1.6 × 105 per μl and 5.0 × 104 per μl, respectively. Using the formula, 0.7 PFU = 1 50% tissue culture infective dose, this is equivalent to estimated concentrations of 1.1 × 105 PFU/μl and 3.5 × 104 PFU/μl for HAV and poliovirus 2, respectively.
Five dilutions ranging from 3.0 to 3.0 × 104 viral particles were used for both viruses to create standard curves. qRT-PCR was performed in triplicate for each dilution of HAV and enterovirus. Standard curves were created by plotting the log of the number of HAV and enterovirus viral particles versus their corresponding CT values and creating a best-fit line through these points (Fig. (Fig.2).2). The cycle threshold (CT) is defined as the PCR cycle at which an increase in the fluorescence above the baseline signal is first detected. The CT value is inversely related to the viral particle count. HAV and enterovirus levels in the Tijuana River mouth and IB pier samples were calculated by using the standard curves.
To determine the efficiency of the filter concentration procedure, two 1-liter seawater samples (previously found to be negative for HAV and enterovirus) were seeded with known titers of stock virus, one with HAV (strain HM-175; ATCC) and the other with poliovirus 2 (strain W-2; ATCC), prior to filtration. The same amount of each virus was also spiked directly into a paired concentrated seawater sample after filtration but before RNA extraction. qRT-PCR was performed, and viral particle counts were determined by using standard curves. The recovery assay was performed twice for each virus and the HAV and poliovirus 2 recoveries were calculated by dividing the number of viral particles in the filtered samples by the number of viral particles in the unfiltered samples.
Samples found to be positive for HAV and enterovirus by qRT-PCR were taken for cloning and sequencing. A 247-bp HAV cDNA was amplified by conventional RT-PCR according to a published protocol (1). The primers for HAV amplification are given in Table Table1.1. Amplified cDNAs were separated by electrophoresis in a 2% agarose gel and eluted from the gel by using a QIAGEN QIAQuick gel extraction kit. In order to clone the enterovirus cDNA, qRT-PCR amplicons of enteroviruses were run in a 2% agarose gel and gel purified by using a QIAGEN QIAQuick gel extraction kit. The enterovirus and HAV gel-purified cDNAs were cloned into a TOPO cloning vector (Invitrogen, Carlsbad, CA). Plasmid DNA was isolated from recombinant clones, and three to five clones were sequenced for each sample by using the vector-derived T7 primer.
Nucleotide sequences of HAV and enterovirus clones were BLAST searched and identified based on their similarity to GenBank database entries. Multiple alignments and phylogenetic analyses were performed by using MEGA version 3.0 by Kumar et al. (16). Kimura's two-parameter distance was calculated by using transitions and transversions, and a neighbor-joining tree was built. The confidence of reconstructed clusters was tested by bootstrapping with 1,000 replicates.
The Pearson correlation test was used to evaluate relationships among viral, bacterial, and environmental variables. A multiple linear regression was used to analyze the relationship between the levels of viruses and the levels of bacterial indicators and other environmental variables collected in the present study. Analyze-it (version 1.73; Leeds, England, United Kingdom) was used for linear regression and correlation analyses. Analysis of variance and subsequent comparisons to determine differences in mean viral levels between sampling sites were also performed by using Analyze-it. In all cases, the significance was determined at the 95% confidence level.
In order to determine the efficiency of our filter concentration protocol, seawater samples were seeded with known amounts of HAV or poliovirus 2 on two occasions, and virus levels were quantified by using the qRT-PCR standard curves (Fig. (Fig.2).2). The mean percent recoveries were 11% for HAV and 71% for poliovirus 2 (Table (Table22).
qRT-PCR was performed to detect and quantitate the levels of HAV and enterovirus in 20 samples from the Tijuana River mouth and IB pier. Samples that were positive for either virus using qRT-PCR were quantitated using the standard curves (Fig. (Fig.22 and and3).3). Some samples had to be diluted by 1:10 or 1:100 in order to get successful amplification. For each sample, the value from the dilution that exhibited the highest number of viral particles (i.e., that showed the least inhibition) was used in Fig. Fig.33.
HAV was detected in six of seven Tijuana River mouth samples and in five of seven wet-weather IB pier samples at levels (uncorrected for recovery efficiency) ranging from 105 to 30,771 viral particles/liter (Fig. (Fig.3).3). Enterovirus was detected in all seven river mouth samples and in six of seven wet-weather pier samples at levels (uncorrected for recovery efficiency) ranging from 7 to 4,417 viral particles/liter (Fig. (Fig.3).3). Neither virus was ever detected in any of the six samples taken at the IB pier during dry weather (Fig. (Fig.33).
The lowest viral concentrations we detected in our seawater samples via qRT-PCR and confirmed by sequencing were 3.4 and 3.0 viral particles per PCR for HAV and enterovirus, respectively (Fig. (Fig.3).3). These values corresponded to minimal detection limits of 7.0 HAV particles/liter and 4.0 enterovirus particles/liter (Fig. (Fig.33).
HAV was detected in four of six samples collected during the 2003-2004 rainy season (Fig. (Fig.3).3). Multiple alignments of the HAV sequences (Fig. (Fig.4)4) showed 100% similarity to the VP1 to VP3 genes of three strains (accession numbers AY441441, AY441442, and AY441443) previously isolated from the same region. A BLAST search showed that these sequences were significantly similar to other entries in the GenBank database (~93% with the wild-type isolate, accession number M14707; ~94% with isolate MBB, accession number M20273; and ~97% with isolate M2, accession number AY974170). A total of four nucleotide changes were found at positions 69, 77, 128, and 134 among the HAV samples amplified from the ocean water (Fig. (Fig.4).4). Of the four nucleotide changes, the mutation at position 69 resulted in an amino acid change from alanine to threonine. The remaining three nucleotide changes did not alter the amino acid sequence (Fig. (Fig.44).
Enterovirus was detected in five of six samples collected during the 2003-2004 wet-weather season (Fig. (Fig.3).3). Three to five clones were sequenced for each sample, and a BLAST search using the 151 nucleotide sequence showed that all of the clones had a similarity to the 5′-UTR of enteroviruses in the database entries. Thirteen unique sequences of nine different enterovirus types were identified among the clones. A neighbor-joining tree constructed from an alignment of the 151-base nucleotide sequence of the 5′-UTR revealed two major clusters (Fig. (Fig.5).5). The larger clade contained echoviruses 6, 11, and 30 (accession numbers AY343049, AF447476, and AY343042, respectively), as well as coxsackievirus A5 (accession number AB126201) and enterovirus B (accession number AY271469). The smaller clade included polioviruses 1, 2, and 3 (accession numbers AF111984, D00625, and L76411, respectively), as well as enterovirus 90 (accession number AY773285). The most prevalent enterovirus was echovirus 30, which was isolated from three samples. Poliovirus 2 was detected in 2 samples, and the remaining types were each detected in one sample (Fig. (Fig.55).
E. coli was detected in 13 of 14 wet-weather samples and 4 of 6 dry-weather samples with levels ranging from 41 to 601,000 MPN/100 ml and from 10 to 20 MPN/100 ml, respectively (Fig. (Fig.3).3). Enterococci were detected in 12 of 14 wet-weather samples and in 1 of 6 dry-weather samples with values ranging from 72 to 754,000 MPN/100 ml and 10 MPN/100 ml, respectively (Fig. (Fig.3).3). There was a statistically significant correlation between E. coli densities and both HAV (R2 = 0.67, P < 0.0001) and enterovirus (R2 = 0.73, P < 0.0001) levels (Table (Table3).3). There was also a significant association between enterococci and both HAV (R2 = 0.61, P < 0.0001) and enterovirus (R2 = 0.70, P < 0.0001) levels (Table (Table3).3). Finally, there was a significant relationship (R2 = 0.58, P < 0.0001) between the levels of HAV and enterovirus. For the correlation analysis, nondetectable levels of virus and bacteria were assigned the value of one-half of the limit of detection.
The goal of the present study was to quantify the levels of HAV viral particles and other indicators of fecal pollution in coastal ocean waters in the vicinity of the U.S.-Mexico border. Relatively little is known regarding the levels of human enteric viruses in southern California coastal waters near the U.S.-Mexico border. In one study, among samples collected from 12 beach locations from Malibu to the border of Mexico, 33% (4 of 12) of marine samples were positive for adenoviruses (12). MPN concentration estimates indicated that there were 880 to 7,500 adenoviruses per liter of water. These marine sites were located outside of river discharge points, and the authors of that study noted that bacterial indicators did not correlate with the presence of viruses. Using qRT-PCR, Brooks et al. (1) detected HAV in all eight samples taken during rain events from either the mouth of the Tijuana River (near the U.S.-Mexico border) or the nearby surf zone at IB at levels ranging from 90 to 3,523 copies/liter and 347 to 2,656 copies/liter, respectively. These relatively high levels of HAV measured during wet weather were attributed to the inadequate sewage collection infrastructure in the region of Tijuana, Mexico (1).
In the present study, 86 and 100% of wet weather samples collected from the surf zone adjacent to the Tijuana River mouth were positive for HAV and enterovirus, respectively. The concentrations of HAV in these samples ranged from 951 to 30,771 viral particles/liter (geometric mean = 7,569), and the values of enterovirus ranged from 30 to 4,417 viral particles/liter (geometric mean = 327) (Fig. (Fig.3).3). Due to the close proximity of this sampling site to the Tijuana River, these levels were anticipated to be higher than levels at the IB pier. The levels of HAV and enterovirus were significantly higher (P < 0.05) for the Tijuana River mouth than for samples collected at the same time from the IB pier (Fig. (Fig.33).
HAV and enterovirus were detected in 71 and 86%, respectively, of wet-weather IB pier samples. The HAV concentrations in these samples ranged from 105 to 3,445 viral particles/liter (geometric mean = 715), and enterovirus levels ranged from 7 to 375 viral particles/liter (geometric mean = 45) (Fig. (Fig.3).3). Concentrations of HAV and enterovirus were below the limit of detection for all six samples collected at Imperial Beach during the dry-weather season (Fig. (Fig.3).3). All of the dry-weather samples contained fewer than 7.0 viral particles per liter of HAV and 4.0 viral particles per liter of enterovirus (Fig. (Fig.33).
qRT-PCR offers improved sensitivity and specificity over traditional viral culture (18, 27); however, an inherent limitation of qRT-PCR is its inability to discriminate between infectious and noninfectious viral particles (2). However, Gantzer et al. (9) noted the instability of infectious enterovirus particles in wastewater and concluded that the presence of enterovirus viral particles (detected by qRT-PCR) can still be valuable as an indicator of recent viral contamination. Furthermore, for viruses such as HAV, which are difficult and time-consuming to culture, qRT-PCR may be an invaluable tool for rapid environmental monitoring. In the present study, we were able to estimate the relationship between the number of viral particles (as indicated by SYBR gold staining) and PFU for our stock control viruses. Using the virus titers provided by the ATCC, we found this relationship to be 78 viral particles/PFU for HAV (strain HM-175) and 48 viral particles/PFU for poliovirus 2 (strain W-2). Similar results for poliovirus were obtained by Donaldson et al. (5), who found a relationship of 55 viral particles/PFU and Fuhrman et al. (8), who found a relationship of 66 viral particles/PFU. No comparable results exist in the literature for HAV.
HAV cDNA obtained from the four positive 2003-2004 wet-weather samples was cloned and sequenced, and a BLAST search identified three highly similar HAV strains (Fig. (Fig.4).4). All three strains were at least 98% identical. Ticehurst et al. (26) reported that different human HAV strains of diverse geographic origin were remarkably closely related. The isolates in the present study were significantly similar to isolates from southern Italy (~99% similar to isolate IT-DAL-00, accession number AJ505803), Argentina (~97% similar to isolate Arg873, accession number AF452067), and Japan (~97% similar to isolate FH3, accession number AB020569) (Fig. (Fig.4).4). Unlike the present study, in which the virus types were fairly evenly distributed among the samples, a previous study on the Venice Lagoon (23) found that a single strain of HAV (accession number AY441443) was present in a majority of the samples.
Nine different enterovirus types were isolated from the five positive 2003-2004 wet-weather samples. A neighbor-joining tree grouped the enterovirus isolates into two major clades (Fig. (Fig.5).5). One clade contained echoviruses 6, 11, and 30, coxsackievirus A5, and enterovirus B, whereas the second clade contained polioviruses 1, 2, and 3 and enterovirus 90. This is in general agreement with previously published enterovirus phylogeny (20). While the enterovirus types in the present study were relatively evenly distributed (Fig. (Fig.5)5) among the samples, a previous study on the Venice Lagoon (23) found that a single virus, poliovirus 2, was present in a majority of the samples. Likewise, Donaldson et al. (5) found that coxsackievirus A9 was the dominant enterovirus type. Since we found poliovirus 2 (the same type of enterovirus as our positive control), one might argue that this resulted from a contamination event in the laboratory. However, this is unlikely because negative controls run in parallel with positive samples were consistently negative by both PCR and sequencing.
Twelve of fourteen wet-weather samples (86%) exceeded the California state water quality standard for one or both of the bacterial indicators, E. coli and enterococci (Fig. (Fig.3).3). In contrast, there were no bacterial exceedances for any of the dry-weather samples (Fig. (Fig.3).3). Regression analyses of the viral densities (as measured by qRT-PCR) and indicator levels showed a significant correlation between the densities of both bacterial indicators and the levels of HAV (R2 > 0.61, P < 0.0001) and enterovirus (R2 > 0.70, P < 0.0001) (Table (Table3).3). These results suggest that bacterial indicator levels may be predictive of the levels of viruses at the Tijuana River mouth and IB pier.
A multiple regression analysis showed that HAV and enterovirus levels were directly (positively) related (P < 0.05) to the levels of the fecal indicator bacteria, E. coli and enterococci, one another, river flow rate, and precipitation, and inversely related to water temperature (Table (Table3).3). There was no correlation between tidal height and the levels of either virus. Collectively, these variables were able to predict HAV levels in a sample 35 to 67% of the time (P ≤ 0.002) and enterovirus levels 19 to 73% of the time (P ≤ 0.025) (Table (Table33).
Although the association (percent concordance) with viral levels was slightly higher for E. coli than for enterococci, these differences were not significant (P > 0.05), suggesting that both bacterial indicators were similarly able to predict levels of virus. A series of large-scale epidemiological studies carried out by the U.S. EPA (29) found that, among the indicator organisms, only two—E. coli (r = 0.51) and enterococci (r = 0.81)—exhibited a strong correlation to swimming-associated gastroenteritis. Based on these findings, the U.S. EPA's draft of the Implementation Guidance for Ambient Water Quality Criteria for Bacteria (30) recommended criteria for marine waters solely based on enterococci. Our results suggest that E. coli might, as well, be a suitable indicator of viral contamination in sewage-contaminated marine waters.
In contrast to our results, a number of recent studies using PCR have shown that viral contamination cannot be well assessed by using bacterial indicators (2, 7, 24). There may be a number of important reasons for the differing results. All of these studies were conducted on urban rivers where a variety of nonpoint sources of contamination, including animal fecal sources, could have contributed to the lack of a relationship between fecal indicator bacteria and human virus levels. In our study, the existence of a predominant source of human fecal contamination resulting from the inadequate sewage infrastructure in Mexico may have facilitated our finding of statistically significant correlations between bacterial indicator and human virus levels in coastal waters. As such, the present study presents the first quantitative assessment of the statistical relationship between levels of HAV, enterovirus, E. coli, and enterococci in marine waters.
Financial assistance for this research was provided by the Southwest Consortium for Environmental Research and Policy and the U.S. Environmental Protection Agency.
Published ahead of print on 15 September 2006.