Introduction

Cryptosporidium parvum is a waterborne parasite frequently identified in oysters intended for human consumption worldwide (Graczyk 2003). This enteric protozoan causes diarrheal disease (Graczyk et al. 1997a) and contributes significantly to the mortality of immunocompromised people and morbidity of immunocompetent individuals due to the lack of fully effective therapy (Dawson 2005). Cryptosporidium is dispersed in water via its minute (approximately 5 μm), environmentally robust transmissive stage, the oocyst.

Although multiple studies have identified human-infectious Cryptosporidium in commercial oysters worldwide (Graczyk 2003), there are no published reports of cryptosporidiosis from oyster consumption despite a long list of other infections acquired via eating oysters (OzFoodNet Working Group 2005). The epidemiological reason for this discrepancy is unclear, but it may be related to the concentration and viability of C. parvum oocysts in oysters consumed raw and the acquired immunity status of the consumers. A common technique for identification of C. parvum in oysters, immunfluorescent antibody (IFA) detection (Fayer et al. 1998), overestimates the infectious pathogen load because the antibodies cross react with species of Cryptosporidium that are not infectious to humans (Graczyk et al. 1996) and small unicellular algae (Clancy et al. 1994), and nonviable oocysts (Graczyk et al. 1997b). Most of the studies on Cryptosporidium in oysters utilize molecular identification techniques such as polymerase chain reaction (PCR), nested PCR, and/or restriction fragment length polymorphism (RFLP) (Graczyk 2003), which, although very sensitive, do not assess viability of the detected oocysts. As a consequence, the data obtained in published reports are not fully applicable to assess the risk of foodborne cryptosporidiosis, although they provide solid scientific information on contamination of oysters with human-infectious Cryptosporidium.

Recovery of oocysts from oyster tissue is a technologically complex process, but even more challenging are subsequent species-specific identification and viability assessment of oocysts recovered from tissue or environmental samples. The fluorescent in situ hybridization (FISH) technique meets both these challenges, that is, providing species-specific identification with simultaneous viability assessment of identified parasites (Smith et al. 2004). The FISH utilizes a fluorescently labeled oligonucleotide probe, that is, CRY-1, designed to hybridize with specific sequences of 18S rRNA of C. parvum (Smith et al. 2004). Because rRNA is only present in large copy numbers in viable organisms (Smith et al. 2004), FISH allows species-specific detection by providing visualization of the oocysts and facilitates enumeration of viable C. parvum oocysts (Graczyk et al. 2004). Most importantly, Cryptosporidium RNA and DNA in preserved samples is stable for years (Amar et al. 2005), which can facilitate retrospective analyses. The purposes of the present study was to determine if FISH assay for identification of viable C. parvum oocysts can be retrospectively applied to samples and to enumerate viable oocysts of C. parvum in the tissue of Eastern oysters (Crassostrea virginica) commercially harvested from the Maryland portion of Chesapeake Bay.

Materials and methods

Thirty commercial-size C. virginica (>7.6 cm shell length) were collected by dredging (Graczyk et al. 2000) from each of 53 sites in 11 regions (Tables 1 and 2) of the Maryland portion of Chesapeake Bay. Oyster collections were carried out in July 2000 at sites approved for commercial shell fishing by the Maryland Department of the Environment (MDE). Water quality parameters such as benthic water temperature, salinity, and dissolved oxygen were measured at each site (Table 2). The last three fecal coliform counts for each site were obtained from the MDE. Fecal coliform testing of shellfish harvesting waters is routinely carried out by the MDE with a frequency of 8–11 days.

Table 1 Descriptive statistics of mean cumulative number of viable Cryptosporidium parvum oocysts identified by combined FISH and IFA, and Cryptosporidium sp. oocysts identified by IFA in six commercial-size (>7.6 cm shell length) Eastern oysters (Crassostrea virginica)
Full size table
Table 2 Mean±standard deviation of water quality parameters at all 53 commercial Eastern oyster (Crassostrea virginica) harvesting sites in the 11 regions (listed from North to South) of the Maryland portion of Chesapeake Bay
Full size table

Thirty oysters from each site were placed separately in mesh bags and transported to the laboratory in insulated coolers. Approximately 5 ml of hemolymph was collected from each oyster (Graczyk et al. 1998). Hemolymph samples of six oysters from the same site were combined into plastic 50-ml tubes to yield five groups from each sample of 30 oysters. Each of the tubes were vortexed for 2 min, centrifuged (3,000×g; 10 min), the supernatant discarded, and the pellet resuspended in 50 ml of sterile phosphate-buffered saline (PBS; pH 7.4). The oysters were shucked, and the gills were excised from each oyster and placed separately into a 15-ml plastic tube containing 5 ml of sterile PBS. The tubes were vortexed for 2 min, gills removed, and the resulting suspensions from each of six oysters from the same site were placed together into 50-ml plastic tubes. The tubes were centrifuged (3,000×g; 10 min), the supernatant was discharged, and the pellet resuspended in 50 ml of sterile PBS. The remaining meat from each set of six oysters was placed together in a plastic bag and weighed. Gill- and hemolymph-derived samples were processed separately by the cellulose acetate membrane (CAM)-filter dissolution method (Graczyk et al. 1997b). The resulting pellets were placed on three lysine-coated immunofluorescent slide wells (1.0-cm diameter). The slides were air-dried, fixed with methanol, and processed with fluorescein isothiocyanate (FITC)-conjugated fluorescent antibodies (IFA) against the cell wall antigens of Cryptosporidium (i.e., MERIF LUOR; Meridian Diagnostic, Inc., Cincinnati, OH) as described previously (Graczyk et al. 1996). Testing included positive and negative controls provided by the manufacturer. Cryptosporidium oocysts were enumerated with the aid of an Olympus BH2-RFL epifluorescent microscope, dry ×60 objective, and BP450-490 exciter filter. The slides were stored at −70°C.

Small aliquots, approximately 150 μl, of oyster hemolymph and gill washing samples from each area of Chesapeake Bay were used for PCR analysis. DNA was extracted from these aliquots using proteinase K digestion and glass bead disruption of the oocysts as described elsewhere (Pieniazek et al. 1999). The PCR amplification was performed using primers designated to amplify specific for the Cryptosporidium oocyst wall protein (COWP) and SSUrRNA genes following previously described protocols (Johnson et al. 1995; Pieniazek et al. 1999). Amplicons were subjected to DNA sequence analysis to confirm the species of Cryptosporidium. Amplified products from all PCR reactions were purified by using the Strataprep PCR purification kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Sequencing analysis was performed using cycle sequencing using the ABI BigDye version 3.0 sequencing kit (ABI, Foster City, CA). Sequencing reactions were analyzed on the PerkinsElmer ABI 3100 sequence analyzer, and the sequences were assembled using SeqMan II (DNASTAR Inc., Madison, WI).

In 2004, glass slides were removed from the freezer, thawed, coded, and used for combined FISH and direct IFA analysis (Graczyk et al. 2004). An oligonucleotide probe, CRY-1 (5′ CGG TTA TCC ATG TAA GTA AAG 3′) 5′ labeled with a single molecule of a fluorochrome [i.e., hexachlorinated 6-carboxyfluorescein (HEX)] was used (Graczyk et al. 2004; Smith et al. 2004). The probe hybridizes to positions between 138 and 160 on the C. parvum 18S rRNA (Deere et al. 1989). The probe was synthesized by the DNA Analysis Facility of the Johns Hopkins University, Baltimore, MD, in 1.0 μM scale and purified by high-performance liquid chromatography (HPLC). The slides were placed in a humidified chamber, and all combined FISH and IFA reactions were carried out in a total volume of 200 μl of hybridization buffer at 48°C for 1 h (Graczyk et al. 2004). FISH and IFA testing included positive and negative controls. The concentration of oligonucleotide probe was 1 mmol, and the fluorescently labeled antibodies were diluted 1:1 (Graczyk et al. 2004). After hybridization, the area of each well was examined as described above for the IFA testing, and only oocysts that showed both FISH and IFA reaction were enumerated.

Statistical analysis was carried out with Statistix 7.0 (Analytical Software, St. Paul, MN). Cryptosporidium parvum load is presented by cumulative numbers of oocysts in hemolymph and gill washing samples of six oysters. This rationale was used because oysters are usually offered for human consumption in a raw form by the half-dozen. Variables were tested by Wilk–Shapiro/ranking plots to determine whether their distribution conformed to a normal distribution, and if so, parametric tests, were used. Results are presented as the mean±SD for continuous variables and as the number or percentage for categorical data. The statistical significance of association between variables was assessed by Pearson correlation coefficient. Statistical significance was considered to be P<0.05.

Results

Testing of positive controls processed by IFA, stored at −70°C, and then reprocessed by combination of FISH and IFA yielded similar and not statistically different enumeration values for C. parvum oocysts.

FISH and IFA labeling clearly differentiated viable C. parvum oocysts from nonviable or non-C. parvum oocysts (i.e., oocysts that were stained only with IFA). Oocysts labeled by FISH and IFA were predominantly intact, revealing a small gap between the oocyst wall and internal structures, and in most of them, the sporozoites were visible. In comparison, dead C. parvum oocysts, that is, oocyst shells, or non-C. parvum oocysts frequently had discernable damage to their walls. No fluorescence of other particles in the samples was observed except for weak autofluorescence of nonstructural debris.

Oocyst isolates represented C. parvum as determined by PCR amplification confirmed by DNA sequencing analysis.

The cumulative numbers of oocysts identified in oysters are presented in Table 1. As determined by the combined FISH and IFA technique, the mean cumulative numbers of viable C. parvum oocysts in six oysters varied from 2 to 106; IFA testing revealed from 4 to 125 Cryptosporidium sp. oocysts (Table 1). Of a total of 265 groups of six oysters, 221 (83.4%) contained viable C. parvum oocysts as determined by combined FISH and IFA technique, and 235 (88.7%) were Cryptosporidium sp.-positive by IFA. Oysters from Wye River, Choptank River, Upper Bay, and Eastern Bay had the highest oocyst amounts, and those from Manokin River and Nanticoke River had the lowest overall levels, irrespective of the testing technique (Table 1). In general, the mean cumulative numbers of viable C. parvum oocysts in six oysters were significantly lower (two-sample t test; t=1.96, P<0.01) for samples processed by combination of FISH and IFA (overall, 28.0±2.9) than the number of Cryptosporidium sp. oocysts in samples processed by IFA alone (overall, 42.1+4.1) (Table 1). Testing of oysters by combined FISH and IFA techniques yielded lower numbers of viable C. parvum oocysts by approximately 34% compared with Cryptosporidium sp. oocysts enumerated by IFA; however, the numbers of oocysts enumerated by these two methods were statistically correlated (Pearson correlation; R=0.84, P<0.01). Overall, using FISH and IFA, from 10 to 32% (mean, 23%) of the total oocyst number was identified in hemolymph in six oyster pools compared with 25 to 52% (mean, 39%) by IFA.

The mean wet weight of oyster meat varied considerably among the 11 Bay subregions, from 92.8 g at Nanticoke River to 216.0 g at Chester River (Table 1). The weight of oysters was not related to the level of oocysts in their tissue within the region or a site.

The fecal coliform counts varied from 1.0 at Manokin River to 7.7 most probable number (MPN)/100 ml at the Wye River site. The highest mean fecal coliform values (Table 2) coincided with the highest C. parvum load only at the Wye River site (Table 1). The relation between fecal coliform counts and C. parvum level in oysters was not statistically significant in any of the 11 Chesapeake Bay regions.

Discussion

The present study, together with other studies on C. parvum in oysters for human consumption (as reviewed in Graczyk 2003), leave no doubt that such oysters can be contaminated with this pathogen for which fully effective therapy is currently unavailable (Dawson 2005). Because depuration of oocysts from oysters is ineffective (Gomez-Couso et al. 2003), and they remain viable in estuarine waters for over a year (Tamburrini and Pozio 1999), reports from the world’s highest seafood-producing regions suggest implementation of a rational plan to prevent or reduce Cryptosporidium contamination in edible molluscan shellfish (Gomez-Couso et al. 2003, 2004). However, lack of evidence of human cryptosporidiosis linked to oysters generates a great deal of skepticism on the epidemiological importance of these reports and the necessity of reducing or eliminating the presence of waterborne Cryptosporidium in oyster harvesting waters.

A set of host-related, environment-derived, and Cryptosporidium-specific factors interplay in the epidemiology of foodborne cryptosporidiosis. Cryptosporidium parvum isolates differ in their virulence to humans (Okhuysen et al. 1999; Messner et al. 2001). Volunteer challenge trials that utilized healthy C. parvum-seronegative individuals demonstrated that the median infective dose (ID50) for the three main C. parvum isolates (i.e., TAMU, IOWA, and UCP) varied from 9 to 18, from 87 to 190, and from 1,042 to 2,980 oocysts (Okhuysen et al. 1999; Messner et al. 2001), respectively. Considering the worst case scenario, when oysters are contaminated with the most virulent C. parvum isolate (i.e., TAMU; ID50=9 oocysts) (Okhuysen et al. 1999), 205 (77.4%) of 265 groups of six oysters in the present study potentially can cause infection in 50% of the exposed individuals. This means that when consumed by 265 people (i.e., meal per person), 265 oyster meals can potentially cause infection in approximately 100 people, assuming that all these individuals are healthy (i.e., immunocompetent) and do not have preexisting anti-C. parvum serum immunoglobulin G (IgG). For partially immune immunocompetent people with systemic IgG, which is indicative of previous exposure, the ID50 becomes more than 20-fold higher (Chappell et al. 1999; Dann et al. 2000; Okhuysen et al. 2004). Thus, our results indicate that infections are not likely to develop in oyster consumers who were partially immune, even upon consumption of raw oysters contaminated with the most virulent isolate of C. parvum. In addition, as demonstrated in the present study, even the most heavily contaminated oysters did not contain more viable oocysts than the ID50 value of a moderately virulent C. parvum isolate, that is, IOWA, ID50=190 oocysts (Messner et al. 2001). Cryptosporidiosis has life-threatening consequences in people with various immune deficiencies, and it is thought that in this population, a single oocyst can cause infection (Rose et al. 2002). However, it is also believed that people with immunological impairments are usually more aware of the potential health hazards associated with consumption of raw food items, particularly molluscan shellfish, and are therefore more likely avoid this type of food (Rippey 1994).

As demonstrated in the present study, approximately 23% of the infectious C. parvum oocysts were found in oyster hemolymph. Therefore, it is reasonable to assume that the amount of oocysts ingested by a person eating raw oysters is actually lower than reported. Shucking of oysters severs both sides of the adductor muscle, a hemolymph-rich organ, causing hemolymph leakage.

The assessment of FISH assay as suitable for identification of C. parvum oocysts was initiated in 1989 (Deere et al. 1989); however, the present study demonstrated that this particular assay can be applied retrospectively. The storage of dry, methanol, fixed samples at −70°C did not affect the quality of rRNA, as demonstrated by the similar number of C. parvum oocysts identified in positive controls of IFA- and FISH-processed slides. Although RNA in dry smears is practically indestructible, fixation with alcohol and freezing has been shown previously to be particularly beneficial in long-term preservation of pathogen RNA or DNA, particularly for Cryptosporidium (Amar et al. 2005; Vincek et al. 2005; Perlmutter et al. 2004). It is recommended to examine IFA-processed smears without oil immersion and without coverslips because otherwise, such slides are not suitable for long-term storage, and the fluorescence will deteriorate rapidly.

The combined FISH and IFA approach has some advantages over IFA or various PCR methods, with the main advantage being the assessment of viability of even a single oocysts (Vesey et al. 1998; Smith et al. 2004). Such resolution is not available, or impractical, in any other technique. For example, using mouse bioassay and highly sensitive RT-PCR for RNA detection, the lowest number of C. parvum oocysts that can be assessed for their viability was 103 (Jenkins et al. 2000). The practicality of the FISH method used in the present study relies on the elimination of the permeabilization step (Smith et al. 2004) because this was done through the CAM-filter dissolution method protocol (Graczyk et al. 1997b).

The mouse bioassay (Fayer et al. 1998, 2002a,b) has limited value for risk assessment of foodborne cryptosporidiosis related to consumption of oysters because the assay is based on manipulated pathogens. In this bioassay, mice are inoculated with oocysts extracted from oysters, purified, and delivered directly to their gastric region. However, under natural situations, the oocysts enter the gastrointestinal tract of a person while deeply buried in oyster tissue, which can prevent, inhibit, or delay the arrival of a set of stimuli, triggering oocyst excystation and subsequent infection. In addition, the mouse bioassay will produce false-negative results for infectivity assessment of Cryptosporidium hominis, which is a highly human-virulent species (Morgan-Ryan et al. 2002). For risk-assessment purposes, a bioassay that utilizes piglets is much more applicable because piglets can be fed with whole oyster meat which would mimic natural circumstances. The piglet model is commonly used for assessment of pathogenicity of Cryptosporidium (Enemark et al. 2003).

The present study provides useful information for formulation of risk assessment of foodborne cryptosporidiosis due to consumption of contaminated oysters in a raw form. We conclude that the lower numbers of C. parvum oocysts enumerated by combined FISH and IFA were due to elimination of dead oocysts and oocysts of other Cryptosporidium species, and elimination of small unicellular algae which are detected by IFA (Clancy et al. 1994). We further conclude that IFA should not be used for purposes of oocyst enumeration from oysters because this assay significantly overestimates the level of C. parvum oocysts. However, irrespective of the identification technique, the recovery efficiency of oocysts from oyster tissue is not 100%. A previous study that utilized the tissue of molluscan shellfish from Chesapeake Bay spiked with known numbers of C. parvum oocysts demonstrated a 50% recovery efficiency (Graczyk et al. 1999). Thus, we conclude that the numbers of viable C. parvum oocysts reported in the present study were potentially underestimated.

Crassostrea virginica is a suspension feeder with a clearance rate (i.e., cleared particles in a unit time) proportionally related to their size (Newell and Langdon 1996), which means that larger oysters filter more than smaller ones, and the size of Cryptosporidium oocysts fall within the range of filtered particles. The present study showed that oyster size was not related to the amount of Cryptosporidium in their tissue, which indicates involvement of determinants other than biological factors (i.e., environmental) in recovery of waterborne oocysts by oysters.

The present study confirms previous findings that the standard water quality parameters such as fecal indicator coliform count does not correlate with the Cryptosporidium contamination in edible oysters (Freire-Santos et al. 2000; Gomez-Couso et al. 2003, 2004). Foodborne illnesses following consumption of raw oysters can occur even when fecal coliform testing of shellfish harvesting waters demonstrates compliance with the National Shellfish Sanitation Program criteria and the sanitation at the oyster harvesting facilities met standards set by the state authorities (Graczyk and Schwab 2000). This clearly demonstrates that monitoring of shellfish waters for fecal coliforms is not sufficient to indicate the presence of waterborne Cryptosporidium. We conclude that oysters harvested from natural waters and identified as suitable for human consumption in a raw form can be a potential vehicle for infectious C. parvum oocysts and serve as a source of human cryptosporidiosis. This epidemiological possibility should be addressed by formulation of the risk assessment model accounting for Cryptosporidium-specific, consumer-related, and environment-derived cofactors.