The objective of this work was to field test our CC-PCR C. parvum infectivity assay by using raw source water and filter backwash water grab samples obtained from 25 sites in the United States. Oocyst-seeded raw and filter backwash water samples were used to evaluate the method recovery efficiencies and the performance of the CC-PCR protocol with different water quality matrices. In addition, samples were processed by Percoll-sucrose flotation and oocysts were detected by an immunofluorescence assay (IFA) as a baseline method. CC-PCR-positive samples were confirmed by DNA sequence analysis of the hsp70 PCR products and compared to investigate the potential genetic heterogeneity of waterborne infectious C. parvum.

Oocysts and microscopy. Purified viable C. parvum oocysts (bovine isolate LA-1) were obtained from Waterborne, Inc. (New Orleans, La.). Oocyst stocks and Percoll-sucrose flotation-purified oocysts were enumerated by IFA microscopy as described in the ICR Microbial Laboratory Manual (22). Oocysts purified by IMS (method described below) were enumerated by fluorescence microscopy as follows. IMS-purified samples (40 μl of sample plus 10 μl of deionized water wash from a microcentrifuge tube) or purified oocyst stocks (50 μl) were placed into individual wells of polylysine-treated three-well microscope slides (Meridian Diagnostics, Inc., Cincinnati, Ohio) and dried at 42°C. Samples were fixed with 1 drop of room-temperature methanol and air dried; then, 75 μl of fluorescein isothiocyanate-labeled anti-Cryptosporidium monoclonal antibody (Waterborne, Inc.) was added to each well. Slides were incubated at 37°C for 30 min in a humid chamber. Excess fluorescein isothiocyanate-labeled monoclonal antibody was removed by aspiration with a micropipette followed by a single wash with 1 drop of deionized water and aspiration. Slides were placed in the dark until dry; then, 10 μl of mounting medium (10% glycerol, 80% phosphate-buffered saline [PBS], 5% 5 M NaCl, 5% formalin, 2.5% 1,4-diazabicyclo[2.2.2]octane [pH 8.6]) was added to each well. A coverslip was applied, and slides were examined by epifluorescence microscopy at a magnification of ×200. Presumptive oocysts were confirmed by 400×-magnification epifluorescence microscopy and 1,000×-magnification Nomarski differential interference contrast microscopy to determine internal morphology.

Recovery of oocysts from environmental water samples. Approximately 10-liter grab samples of raw water and filter backwash water were concentrated by centrifugation at 1,800 × g and 4°C for 10 min. Seeded 10-liter samples were spiked with 1,615 to 2,880 purified viable C. parvum oocysts prior to concentration. Cryptosporidium oocysts were recovered from up to three 0.5-ml packed-pellet portions of each water sample concentrate by IMS (Dynabeads anti-Cryptosporidium; Dynal A.S., Oslo, Norway). IMS was performed according to the manufacturer’s suggestions, with the exception of the dissociation step. The manufacturer’s acidified Hanks’ balanced salt solution (AHBSS, pH 2.75) dissociation protocol for viability testing was originally used but was found to have a variable end-point pH (ca. 3.0 to 6.0) due to residual IMS SL buffer A. To remove residual IMS buffer, a second wash with 1× PBS (pH 7.2) was performed after the samples had been separated with a microcentrifuge tube magnetic particle concentrator (MPC-M; Dynal).

To improve recovery, a combined acid-enzymatic dissociation method which may digest the antibody-oocyst complexes directly off the IMS beads and serve as an excystation trigger for subsequent cell culturing was used. For each IMS reaction, 200 μl of AHBSS–1% trypsin (type II-S porcine pancreas; Sigma Chemical Co., St. Louis, Mo.) was added, and samples were incubated at 37°C for 1 h with 10 s of vortexing every 15 min. Following separation of the samples in the MPC-M, the supernatants containing the dissociated oocysts were transferred to microcentrifuge tubes. To ensure recovery of oocysts, a second wash of the IMS beads with 100 μl of AHBSS–1% trypsin was performed. Dissociated samples were neutralized with 0.5 N NaOH (4.0 μl), and replicate IMS reactions were pooled, centrifuged at maximum speed in a microcentrifuge for 2 min with no brake and aspirated down to 20 μl for CC-PCR or 40 μl for microscopy.

IMS oocyst recovery trials comparing 0.1 N HCl and AHBSS–1% trypsin dissociation methods were performed with deionized water and replicate 0.5-ml packed-pellet portions of raw water and filter backwash water sample concentrates. Water concentrates were seeded with oocysts, and IMS was performed as described above. The 0.1 N HCl IMS dissociation step was performed as previously described (1).

Percoll-sucrose flotation for the recovery of Cryptosporidium oocysts was performed with up to 2.0 ml of a packed pellet of each water sample concentrate as described in the ICR Microbial Laboratory Manual (22), with the exception that a 1.15-specific-gravity Percoll-sucrose solution was used.

In vitro cell culturing of C. parvum. Human ileocecal adenocarcinoma (HCT-8; ATCC CCL-244) cells were cultivated as previously described (25). Cell culture maintenance medium consisted of RPMI 1640 with l-glutamine (Gibco BRL, Grand Island, N.Y.), 5% fetal bovine serum (pH 7.2), 15 mM HEPES buffer, 100,000 U of penicillin G liter⁻¹, 100 mg of streptomycin liter⁻¹, 700 μg of amphotericin B liter⁻¹, and 12.5 mg of tetracycline liter⁻¹. Growth medium used for the in vitro development of C. parvum contained 10% fetal bovine serum, 15 mM HEPES buffer, 50 mM glucose, 35 mg of ascorbic acid liter⁻¹, 1.0 mg of folic acid liter⁻¹, 4.0 mg of 4-aminobenzoic acid liter⁻¹, 2.0 mg of calcium pantothenate liter⁻¹, 100,000 U of penicillin G liter⁻¹, 100 mg of streptomycin liter⁻¹, 700 μg of amphotericin B liter⁻¹, and 12.5 mg of tetracycline liter⁻¹. At 24 h prior to inoculation, 96-well cell culture microplates were seeded with 5 × 10⁴ HCT-8 cells per well. Plates were incubated at 37°C in a 5% CO₂ humidified incubator to allow for the development of 100% confluent monolayers. Just prior to inoculation of monolayers, 50 μl of maintenance medium was removed. Samples for CC-PCR were resuspended in 180 μl of prewarmed growth medium immediately following IMS dissociation and used to inoculate two HCT-8 cell monolayers (100 μl of inoculum for each). AHBSS–1% trypsin-treated C. parvum oocysts were used as positive controls, and single-cycle freeze-thaw killed oocysts were used as negative controls. The inoculated cell monolayers were incubated at 37°C in a 5% CO₂ humidified incubator for 72 h. After incubation, the cell monolayers were washed five times with 200 μl of PBS to remove nonexcysted oocysts. Cell monolayers were harvested by the addition of 200 μl of 1× Tris-EDTA (pH 8.0) buffer, and resuspended cells were transferred to microcentrifuge tubes. Harvested cells were centrifuged at maximum speed in a microcentrifuge for 2 min, aspirated down to a 5- to 10-μl volume, and frozen at −20°C until PCR analysis.

CC-PCR detection of infectious C. parvum oocysts. Harvested HCT-8 cells and C. parvum oocyst PCR-positive controls were lysed by eight cycles of freezing in liquid nitrogen and thawing in a 98°C heated block. Aliquots of lysed samples were used directly for PCR without further purification. PCR primers specific for the C. parvum hsp70 gene resulted in a 361-bp product (18). PCR was performed with a Perkin-Elmer model 9600 thermal cycler (PE Applied Biosystems, Foster City, Calif.). Each 50-μl PCR mixture contained 5.0 μl of 10× amplification buffer with Mg (1.5 mM final concentration; Boehringer Mannheim Biochemicals, Indianapolis, Ind.); 200 μM each dATP, dTTP, dCTP, and dGTP (Boehringer); 200 nM each forward and reverse CPHSP2 primer; 2.5 μl of bovine serum albumin (30 mg ml⁻¹; Sigma); and various amounts of C. parvum template DNA. Amplification conditions were as follows: initial denaturation at 95°C for 5 min; samples kept at 80°C while 2.0 U of Taq DNA polymerase (Boehringer) was added (hot start); 40 cycles of denaturation at 94°C for 30 s; annealing at 59°C for 1 min; extension at 72°C for 30 s; a single final extension at 72°C for 10 min; and a 4°C soak. Amplification products were separated by horizontal gel electrophoresis on a 2.0% agarose gel (Amresco, Solon, Ohio) containing 0.5 μg of ethidium bromide (Sigma) ml⁻¹ and visualized under UV light. Gel images were captured with a gel documentation system (UVP, Inc., Upland, Calif.).

Cloning and DNA sequence analysis of CC-PCR products. CC-PCR products were cloned and sequenced for confirmation of homology to the C. parvum hsp70 gene. Products were cloned with a TOPO TA cloning kit (Invitrogen, Carlsbad, Calif.) according to the manufacturer’s instructions. Cloned products were sequenced commercially (ACGT, Northbrook, Ill., and DNA Sequencing Service, The University of Arizona, Tucson), and sequence homology to the C. parvum KSU-1 hsp70 gene (10) was confirmed with Gene Runner version 3.0 (Hastings Software, Inc., Hastings, N.Y.). Duplicate clones of each PCR product were sequenced and analyzed to identify potential sequencing errors. Clones obtained from independent CC-PCR products from the same sample were analyzed to exclude random nucleotide misincorporation by Taq DNA polymerase.

Statistical analyses. Statistical analyses of seeded-sample oocyst recoveries were performed with a two-sample t test and SYSTAT version 8.0 software (SPSS Inc., Chicago, Ill.). Results are given at the 95% confidence level.

Nucleotide sequence accession numbers. The C. parvum KSU-1 hsp70 reference sequence used in this study has GenBank accession no. U11761. The hsp70 sequences described in this study have the following GenBank accession numbers: C. parvum LA-1, AF150831; C. parvum AWS-1, AF150816; C. parvum AWS-2, AF150817; C. parvum AWS-3, AF150818; C. parvum AWS-4, AF150819; C. parvum AWS-5, AF150820; C. parvum AWS-6, AF150821; C. parvum AWS-7, AF150822; C. parvum AWS-8, AF150823; C. parvum AWS-9, AF150824; C. parvum AWS-10, AF150825; C. parvum AWS-11, AF150826; C. parvum AWS-12, AF150827; C. parvum AWS-13, AF150828; C. parvum AWS-14, AF150829; and C. parvum AWS-15, AF150830.

Oocyst recoveries and CC-PCR with seeded water samples. Flotation and IMS recoveries of oocysts from raw and filter backwash water samples seeded with viable C. parvum LA-1 are summarized in Table 1. Results of CC-PCR analysis of IMS-purified seeded samples are also included in Table 1. While IMS had higher mean recoveries than flotation for both raw and filter backwash water samples, differences were not significant (P values were 0.235 and 0.362, respectively). Mean IMS and flotation oocyst recoveries were lower for filter backwash water than for raw water. This result was most likely due to interference by the large amounts of debris in the filter backwash water samples. However, it is important to note that IMS oocyst recoveries for several filter backwash water samples (A, D, and E) were higher than the recoveries for two raw water samples (C and D).

CC-PCR results agreed with all IFA oocyst counts for IMS-purified seeded samples (Table 1). These results included a positive CC-PCR assay for seeded filter backwash water sample C, which had an IMS-IFA count of only three oocysts. IMS-purified filter backwash water samples B and F had no detectable oocysts by IFA and were found negative by CC-PCR. Deionized water negative controls (n = 8) were all found negative by both IFA and CC-PCR.

Effect of IMS dissociation method on oocyst recovery. IMS trials revealed significantly higher oocyst recoveries from deionized (P, 0.025), raw (P, 0.005), and filter backwash (P, 0.020) water samples with the 0.1 N HCl dissociation method than with the AHBSS–1% trypsin dissociation method (Table 2).

Samplea	Mean (range) % oocyst recovery with:
	0.1 N HCl	AHBSS–1% trypsin
Deionized water	94.3(85.4–104.0)	61.8(56.0–77.3)b
Raw water	57.3(56.1–58.8)	36.3(33.3–40.4)
Backwash water	25.4(21.9–29.8)	13.8(11.7–15.3)

Detection of Cryptosporidium oocysts by flotation-IFA and infectious C. parvum by CC-PCR in environmental water samples. A total of 122 raw water and 121 filter backwash water samples were analyzed. Flotation-IFA detected oocysts in 13.1% (n = 16) of the raw water and 5.8% (n = 7) of the filter backwash water samples, while CC-PCR detected infectious C. parvum in 4.9% (n = 6) of the raw water and 7.4% (n = 9) of the filter backwash water samples. Raw water mean equivalent volumes assayed were 1.57 liters by flotation-IFA and 2.68 liters by IMS. Filter backwash water mean equivalent volumes assayed were 0.75 liter by flotation-IFA and 1.01 liters by IMS. Sites with samples found positive for Cryptosporidium oocysts by flotation-IFA and infectious C. parvum by CC-PCR are listed in Table 3. Several sites had multiple positive samples, and overall Cryptosporidium was detected at 19 of the 25 sites. From only sites 11 and 25 did the same sample test positive by both flotation-IFA and CC-PCR.

Site found positivea by:
Flotation-IFA		CC-PCR
Raw water	Filter backwash water	Raw water	Filter backwash water
2 (2)	11(2)	7	3
3	13	9	8
4	17	11b	9
5	25(3)b	12	10
8		13	15
9		14	17(2)
11b			21
14			25b
16
18
20
21
25

DNA sequence analysis of CC-PCR products. All CC-PCR products were confirmed by cloning and DNA sequence analysis to be >98% homologous to the C. parvum KSU-1 hsp70 gene. Five of the CC-PCR products were 100% homologous to the C. parvum KSU-1 hsp70 reference sequence, while the other products contained nucleotide substitutions which represented six different hsp70 genotypes (Fig. 1). The sequence of C. parvum LA-1, which was used as our laboratory quality control strain, differed from the C. parvum KSU-1 hsp70 reference sequence at three positions, and a single environmental C. parvum strain (AWS-11) had this hsp70 genotype (Fig. 1).

FIG. 1 DNA sequence alignment of hsp70 CC-PCR products of infectious C. parvum recovered from raw (R) and filter backwash (B) water samples. States (postal abbreviations) from and dates (month/day/year) on which samples were recovered are indicated. C. parvum (more ...)

To allow the water industry to make accurate human health risk assessments, it is crucial to have methods to detect viable, infectious C. parvum oocysts in water samples. Here we present field test results of an integrated CC-PCR C. parvum infectivity assay. This study is the first to report the recovery of naturally occurring C. parvum oocysts from environmental water samples with IMS and the determination of their infectivity.

Comparison of oocyst recoveries from seeded samples with Percoll-sucrose flotation and IMS revealed that although mean IMS oocyst recoveries were higher than flotation oocyst recoveries for both raw water and filter backwash water samples, the differences were not statistically significant (P values were 0.235 and 0.362, respectively) due to sample-to-sample variation in IMS oocyst recoveries. The IMS oocyst recoveries in this study are not comparable to those previously reported by others with the Dynal IMS system (1, 2, 16), since the previous studies did not include sample concentration. Thus, the IMS oocyst recoveries in this study are the first reported for the entire IMS method (concentration of seeded 10-liter samples and IMS). In previous studies, IMS recoveries from 10-ml deionized water samples seeded directly with C. parvum oocysts and dissociated with 0.1 N HCl were approximately 91% (2) and 77% (1). Our trials revealed significantly higher oocyst recoveries from deionized, raw, and filter backwash water samples with 0.1 N HCl dissociation than with AHBSS–1% trypsin dissociation (Table 2). Therefore, it is important to note that if the 0.1 N HCl dissociation method had been used for the IMS recovery trials in this study, it is likely that the IMS oocyst recoveries would have been significantly higher than the flotation recoveries. One possible explanation for the differences in oocyst recoveries from deionized water among the laboratories may be the protocols used for microscopic enumeration. Recoveries reported by Campbell et al. (2) (91%) were based on both hemacytometer and well slide IFA counts, while those of Bukhari et al. (1) were based on well slide IFA counts with multiple washes. The well slide protocol used by our laboratory minimizes the numbers of washes, since washing steps may cause loss of oocysts.

The AHBSS–1% trypsin dissociation method was used in this study instead of the 0.1 N HCl method because of its compatibility with the CC-PCR infectivity assay. AHBSS-trypsin dissociation also served as an excystation trigger and pretreatment for cell culturing and was anticipated to have a less adverse impact than 0.1 N HCl on the infectivity of environmentally stressed oocysts. Recently, it was reported that 0.1 N HCl-treated C. parvum oocysts retained their in vitro cell culture infectivity (17), although large numbers (1,000) of fresh oocysts were used and infections were not quantitated. In another study at the same laboratory, sporozoites released (without the use of 0.1 N HCl) from IMS-recovered fresh oocysts (>100) retained their infectivity, but the infections were not quantitated (16). Recent results obtained with a quantitative infectivity assay in our laboratory (5a) revealed a significant reduction (range, 51 to 69%) in the infectivity of 0.1 N HCl-treated oocysts compared to AHBSS-trypsin-treated oocysts. Since 0.1 N HCl has an adverse effect on the infectivity of freshly purified oocysts, the effect is likely more pronounced with environmentally stressed oocysts. Therefore, the AHBSS-trypsin dissociation method is preferred for CC-PCR, while the 0.1 N HCl dissociation method is recommended for IMS-IFA studies.

The unique chemical and physical properties (matrix) of a water sample may have an effect on oocyst recovery with IMS. We attempted to investigate what appeared to be matrix effects which resulted in low oocyst recoveries from some water samples by performing additional IMS recovery trials. We used a replicate, unseeded sample concentrate of backwash water sample F (which had <2.9% recovery by IMS; Table 1). When oocysts were added directly to the sample concentrate and immediately processed by IMS, a mean recovery of 67.4% (n = 3; range, 56.0 to 77.3%) was obtained. However, when the sample concentrate was seeded, centrifuged, and resuspended prior to IMS, the mean recovery was only 1.0% (n = 3; range, 0.7 to 1.3%). These results suggested that during centrifugation, the oocysts became associated with particulates, a situation which hampered oocyst recovery.

Despite the similar oocyst recoveries with flotation and IMS for seeded samples in this study (Table 1), several advantages of IMS over flotation are critical for the CC-PCR infectivity assay. First, IMS is superior to flotation in removing debris and allows larger equivalent volumes of water concentrate to be assayed. This difference was particularly evident for the raw water samples assayed in this study. This difference was not as evident for the filter backwash water samples assayed in this study, since the experimental design limited IMS to the purification of 1.5 ml of a packed pellet and flotation to 2.0 ml of a packed pellet. Flotation was more effective at removing the large debris particles from the filter backwash water samples than the fine particles from the raw water samples. Another advantage of IMS over flotation is that the final IMS-purified sample volume is typically 50 μl, compared to 5 ml for flotation-purified samples. This factor is critical for the CC-PCR assay, since large-volume purified samples are not amenable to analysis. CC-PCR results for IMS-purified seeded samples agreed with all IFA oocyst counts. These results included a positive CC-PCR assay for a seeded filter backwash water sample which had an IMS-IFA count of only three oocysts. In addition, no cytotoxic effects for cell monolayers were observed for any of the samples due to trace water debris present in IMS-purified samples.

It is difficult to compare IFA and CC-PCR, since each method detects different types of oocysts. IFA detects all oocysts (including those of other Cryptosporidium species), dead, viable, or infectious. In contrast, the CC-PCR assay detects only infectious C. parvum oocysts. Therefore, it was anticipated that there would be a larger number of IFA-positive samples than of CC-PCR-positive samples. Indeed, IFA detected oocysts in 13.1% of the raw water samples; CC-PCR detected infectious C. parvum in 4.9% of the raw water samples. Unexpectedly, more filter backwash water samples were found positive for infectious C. parvum by CC-PCR (7.4%) than to contain total oocysts by IFA (5.8%). Only 2 of the 15 CC-PCR-positive water samples were IFA positive for oocysts (Table 3). These results are likely a reflection of the splitting of samples containing very small numbers of oocysts between the IFA and CC-PCR assays; the high fluorescent background interference of flotation-purified samples; the similar volumes of filter backwash water samples examined by IFA and IMS (0.75 and 1.01 liters, respectively); and the high sensitivity of the CC-PCR assay.

Previous studies have shown that Cryptosporidium oocysts present in raw surface water may still be present in treatment plant filter backwash water (3, 8, 15). Oocysts present in raw surface water may be concentrated by water treatment with sand filtration; the degree of concentration has been reported to range from less than 1 (8) up to 3 orders of magnitude (3, 15). In this study, IFA-positive raw water samples had a mean concentration of 190 oocysts/100 liters (range, 37 to 1,463), while filter backwash water samples had a mean concentration of 220 oocysts/100 liters (range, 37 to 556). If the IFA data are adjusted on the basis of oocyst recovery efficiencies determined from seeded samples, then positive filter backwash water samples contained about 3.3 times as many oocysts as positive raw water samples.

DNA sequence analysis of the CC-PCR products revealed a total of six different C. parvum hsp70 genotypes (Fig. 1). It is possible that the single nucleotide substitution observed for AWS6 was due to random misincorporation by Taq DNA polymerase. The sequence of our laboratory quality control strain, C. parvum LA-1, differed from the C. parvum KSU-1 hsp70 reference sequence at three nucleotide positions. These differences were reproducible with different lots of oocysts obtained from the supplier and harvested from different experimentally infected immunosuppressed mice. These results suggested that hsp70 sequences may be useful for differentiating strains of C. parvum. Five of the CC-PCR products (AWS-1 to AWS-5) were identical to the C. parvum KSU-1 hsp70 reference sequence and came from samples from five different states, suggesting widespread occurrence of this C. parvum hsp70 genotype. Similarly, AWS-9 and AWS-10, which represented a novel C. parvum hsp70 genotype, came from samples collected in two different states. In contrast, other C. parvum hsp70 genotypes appeared to have limited geographic distributions. For example, the C. parvum hsp70 genotype represented by AWS-13, AWS-14, and AWS-15 came only from samples collected in Indiana. There was also evidence of the occurrence of mixed C. parvum hsp70 genotypes within watersheds. Two samples from site 9 (AWS-4 and AWS-11) were collected at the same time yet contained two different C. parvum hsp70 genotypes. In contrast, two samples from site 17 (AWS-14 and AWS-15) were collected at different times and contained identical C. parvum hsp70 genotypes. These results exemplify the complex ecology of C. parvum, and analysis of additional environmental strains is necessary to clarify the significance of our findings.

Using the PCR primers of Johnson et al. (7), we attempted to amplify Cryptosporidium 18S rRNA genes from the genomic DNA present in the completed hsp70 CC-PCR samples to gain further information about the infectious C. parvum detected in this study. Sequence polymorphisms within this amplified region differentiate C. parvum from C. baileyi, C. muris, and C. wrairi (7) as well as from human and animal C. parvum strains (14, 27). Thus far, we have been able to successfully amplify and sequence only one sample (AWS-12), and the 18S rRNA gene fragment was identical to that of a novel human C. parvum strain characterized by colleagues at the Centers for Disease Control and Prevention (26). The AWS-12 C. parvum strain also represented a novel hsp70 genotype (Fig. 1).

The detection of infectious C. parvum in filter backwash water samples with the CC-PCR assay in this study is significant in that it provides the first evidence that infectious C. parvum may penetrate water treatment barriers. Previous studies revealed that oocysts may be present in treatment plant effluents (11, 12), including two sites from this study which had samples positive for infectious C. parvum. Determination of the occurrence of infectious C. parvum in finished drinking water is critical for the water industry to make accurate human health risk assessments. The results of this study support the utility of IMS for oocyst recovery and purification of samples of diverse water qualities and the sensitivity and specificity of the C. parvum CC-PCR infectivity assay. Future research is aimed at the optimization of oocyst recovery and the application of the CC-PCR infectivity assay to finished water samples.

This research was supported by grants from the American Water Works Research Foundation and the American Water Works Service Co., Inc.

We gratefully acknowledge the efforts of Ramon Aboytes-Torres and Dale Young in sample processing.