Immunomagnetic Separation (IMS)-Fluorescent Antibody Detection and IMS-PCR Detection of Seeded Cryptosporidium parvum Oocysts in Natural Waters and Their Limitations

Abstract

Detection and enumeration of Cryptosporidium parvum in both treated and untreated waters are important to facilitate prevention of future cryptosporidiosis incidents. Immunomagnetic separation (IMS)-fluorescent antibody (FA) detection and IMS-PCR detection efficiencies were evaluated in two natural waters seeded with nominal seed doses of 5, 10, and 15 oocysts. IMS-FA detected oocysts at concentrations at or below the three nominal oocyst seed doses, illustrating that IMS-FA is sensitive enough to detect low oocyst numbers. However, the species of the oocysts could not be determined with this technique. IMS-PCR, targeting the 18S rRNA gene in this study, yielded positive amplification for 17 of the 18 seeded water samples, and the amplicons were subjected to restriction fragment length polymorphism digestion and DNA sequencing for species identification. Interestingly, the two unseeded, natural water samples were also PCR positive; one amplicon was the same base pair size as the C. parvum amplicon, and the other amplicon was larger. These two amplified products were determined to be derived from DNA of Cryptosporidium muris and a dinoflagellate. These IMS-PCR results illustrate that (i) IMS-PCR is able to detect low oocyst numbers in natural waters, (ii) PCR amplification alone is not confirmatory for detection of target DNA when environmental samples are used, (ii) PCR primers, especially those designed against the rRNA gene region, need to be evaluated for specificity with organisms closely related to the target organism, and (iv) environmental amplicons should be subjected to appropriate species-specific confirmatory techniques.

Cryptosporidium parvum, an intestinal protozoan parasite, continues to be an important cause of waterborne gastrointestinal disease worldwide. Due to oocyst robustness consisting of environmental stability and resistance to normal water disinfection processes (14, 22), the low infectious dose (10), and the lack of chemotherapy (23), detection and enumeration of this organism in both treated and untreated waters have become a focus of the water industry in order to prevent future incidents of cryptosporidiosis.

The currently accepted technique for oocyst recovery from water samples is immunomagnetic separation (IMS)-fluorescent antibody (FA) detection, as described by United States Environmental Protection Agency (EPA) Method 1623 (25). The reported IMS-FA recovery rates for oocysts seeded into previously concentrated water pellets of various turbidities have been 62 to 100% (oocyst seed density as determined by dilution, 36 to 976) (20), 55.9 to 83.1% (oocyst seed density as determined by dilution, 89.1 to 98.7) (17), 68 to 83% (oocyst seed density as determined by dilution, 525 to 870) (3), and 84.3% (oocyst seed density as determined by flow cytometry, 100) (19). In an additional study, in which oocysts were seeded into 10-liter grab samples of various turbidities, the reported recoveries ranged from <1.7 to 56.6% (oocyst seed densities, 1,615 and 2,880) (8).

However, the criteria for oocyst species identification and viability are not fulfilled by the current IMS-FA protocol (25). Therefore, several protocols have been developed by using PCR coupled with restriction fragment length polymorphism (RFLP) for species identification (3, 7, 12, 13, 16, 18) and cell culturing for viability (8, 20). As determined by the combination IMS-PCR approach, the detection limits in treated water samples have been reported to be ≤5 oocysts (13), 8 oocysts (18), and 10 oocysts (16). In addition, the detection limits in raw waters were reported to be ≤5 oocysts (12) and between 1 and 100 oocysts (16).

Due to the reported low oocyst concentrations in source water supplies (range, <0.007 to 484 oocysts/liter; geometric mean, 2.7 oocysts/liter; n = 66) (15) and in drinking waters (range, 0.001 to 0.48 oocyst/liter; n = 158) (21), the need to evaluate the IMS-FA and IMS-PCR recovery efficiencies with low oocyst numbers was deemed imperative. In this study we undertook the task of evaluating IMS-FA detection and IMS-PCR detection with low numbers of C. parvum oocysts seeded into postconcentrated samples of natural waters.

MATERIALS AND METHODS

C. parvum oocysts.

C. parvum oocysts (Iowa isolate) were propagated and purified at the University of Arizona (2). Oocysts from a single batch, which were less than 2 months old, were sent to the Wisconsin State Laboratory of Hygiene for flow cytometric enumeration and sorting with an Epics Elite flow cytometer (Coulter Corporation, Miami, Fla.). Three oocyst preparations were flow counted and dispensed into 25 ml of 1% Tween 20 so that 1-ml aliquots nominally contained 5, 10, or 15 oocysts. The enumerated and sorted oocysts were shipped to CH Diagnostic & Consulting Service, Inc. (Loveland, Colo.) for oocyst seeding and IMS isolation.

Natural waters.

Twenty-liter grab samples were collected from two natural waters, Dowdy Lake and Cache la Poudre River, both located in the northern Rocky Mountains of Colorado. Previous work has determined the concentration of Cryptosporidium spp. in Cache la Poudre water to be 0.007 to 0.04 oocyst/liter (K. Gertig, personal communication). One-liter water concentrates were used to qualitatively identify algal species by bright-field microscopy at magnifications of ×100 and ×400.

The 20-liter grab samples were collected in carboys and concentrated by using Envirochek sampling capsules (Pall Gelman Laboratory, Ann Arbor, Mich.) at a flow rate of 0.5 gal per min. Particulate matter was eluted by using EPA Method 1623 (25). The final centrifuge pellet was measured and resuspended in 20 ml of double-distilled H₂O. One-milliliter aliquots (each equivalent to 1 liter of sample test water) were dispensed into 125-mm flat-sided Leighton tubes (Dynal A.S., Oslo, Norway) for use in IMS.

Confirmation of oocyst seed dose.

Each oocyst stock solution (5, 10, or 15 oocysts/ml) was vortexed for 2 min, which was followed by three 180° inversions just before the solutions were dispensed as 1-ml aliquots directly onto 1-μm-pore-size 25-mm-diameter black polycarbonate membrane filters on a fritted glass support. Three replicates were tested for each oocyst dose. The oocysts were labeled with fluorescein isothiocyanate by using a Merifluor C/G detection kit for Cryptosporidium and Giardia (Meridian Diagnostics, Inc., Cincinnati, Ohio). The membranes were scanned and enumerated at a magnification of ×200 by using fluorescent microscopy.

Oocyst seeding and IMS.

For seeding oocysts, the oocyst stock solutions (5, 10, or 15 oocysts/ml) were vortexed for 2 min, which was followed by three 180° inversions just before the solutions were seeded as 1-ml aliquots into Leighton tubes containing 1-liter equivalents of either Dowdy Lake or Cache la Poudre River water. A total of six Leighton tubes for each nominal oocyst value were tested; three replicates were used for microscopic analysis, and three replicates were used for IMS-PCR analysis. In addition, one sample for each nominal oocyst value was spiked into deionized H₂O, and one sample of each of the natural water 1-liter equivalents was used as a background control. All samples were subjected to IMS by using EPA Method 1623 (25) and two final dissociation steps (50 μl each). Microscopic replicates for each nominal oocyst value were scanned at a magnification of ×200 by using fluorescent microscopy, and presumptive oocysts were confirmed at a magnification of ×1,000 with the vital dye 4′,6-diamidino-2-phenylindole (DAPI) and Nomarski differential interference contrast microscopy. For replicates that were subjected to PCR detection, the final dissociation suspensions were transferred into 0.6-ml thin-wall PCR tubes, frozen at −4°C, and shipped to the University of Arizona.

PCR.

The nested PCR primers and conditions used in this study have been described previously (24), and they amplify a 590-bp region of the C. parvum 18S rRNA gene. The 100-μl dissociation suspensions were washed three times in 100 μl of 1× PCR buffer (10 mM Tris-HCl [pH 8.0], 50 mM KCl) (PE Applied Biosystems, Branchburg, N.J.) by centrifugation (16,000 × g) and resuspension. The final pellet was resuspended in a solution containing 25 μl of 1× PCR buffer and 5 μl of InstaGene matrix (Bio-Rad, Hercules, Calif.). This 30-μl suspension was subjected to six freeze-thaw cycles (2 min in liquid nitrogen, followed by 2 min in a 98°C water bath), with a 30-s centrifugation (16,000 × g) performed between the third and fourth cycles. Lastly, the 30-μl suspension was centrifuged at 16,000 × g for 3 min. One 10-μl aliquot of the freeze-thaw supernatant was used directly as the PCR DNA template. PCRs were performed with a Mastercycler gradient thermal cycler (Eppendorf Scientific, Inc., Westbury, N.Y.). PCR amplicons were visualized on 1.2% ethidium bromide-stained agarose gels with UV transillumination and photodocumented. Two negative control tubes, containing 10 μl of sterile double-distilled H₂O instead of template, were the first and last samples completed for each PCR round.

RFLP analysis.

A 10-μl aliquot of nested PCR amplicon was subjected to digestion with DraI (Roche Molecular Biomedicals, Nutley, N.J.), DraII (Roche Molecular Biomedicals), or VspI (Promega, Madison, Wis.) by following the manufacturer's recommendations. Digested PCR products were visualized on 2% ethidium bromide-stained agarose gels with UV transillumination.

Automated DNA sequencing.

Amplicons were purified by using a QIAquick PCR purification kit (QIAGEN Inc., Valencia, Calif.) and were subjected to DNA sequencing at the GATC Sequencing Facility (University of Arizona, Tucson). To identify possible matches, database searches were performed with the BlastN algorithm (1).

RESULTS AND DISCUSSION

At the time of collection, the turbidity of Dowdy Lake was 8.9 nephelometric turbidity units yielding a final 0.9-ml centrifuge pellet, and the turbidity of Cache la Poudre River was 4.6 nephelometric turbidity units yielding a final 0.2-ml centrifuge pellet.

As stated previously, the oocyst concentrations in both treated and untreated water matrices are low (15, 21). Oocyst detection methods, therefore, need to be evaluated with low and accurately counted oocyst numbers. Flow cytometry was chosen to enumerate the oocyst stock solution instead of hemacytometer counting or micromanipulation since hemacytometer counting is prone to error due to the inherent standard deviation associated with dilutions and to potential oocyst clumping and oocyst micromanipulation, while accurate (89.3% delivery rate for a single oocyst) (24), is time-consuming. Using oocysts enumerated by flow cytometry ensures a standardized stock solution that can be readily dispensed as replicate samples. Confirmation of the number of oocysts seeded per milliliter as enumerated on black polycarbonate membranes is presented in Table 1. The oocyst concentrations were at or below the nominal seed dose in all replicates except those that received the 15-oocyst/ml dose. The variation in the oocyst seed number can be accounted for by (i) the possibility that the levels of oocyst recovery with the membranes were not 100% and thus losses could have occurred or (ii) the possibility that by withdrawing 1-ml aliquots from the flow-enumerated stock solution a standard error was introduced and thus the oocyst seed dose was over- or underestimated. It has been reported that oocysts are unevenly distributed within stock suspensions (4, 9). To correct for this deficiency of the method in the future, it is recommended that oocyst seed doses be flow enumerated and sorted directly into the IMS Leighton tubes (19). Despite this, however, the seeded value used to evaluate IMS-FA and IMS-PCR recovery efficiencies in this study were at or below the nominal oocyst seed dose of interest.

TABLE 1.

Seeded oocyst detection and confirmation by IMS-FA when EPA Method 1623 was used

No. of seeded oocysts/mla	Replicate	No. of oocysts/ml
		Membrane verifiedb	IMS-FA
			Deionized water	Dowdy Lake	Cache la Poudre River
0	1	NAc	NDd	ND	ND
5	1	3	3	4	3
	2	4		3	6
	3	3		2	4
10	1	5	7	9	6
	2	8		5	4
	3	10		7	8
15	1	8	15	7	11
	2	17		7	13
	3	16		13	12

^aNominal number of seeded oocysts per milliliter.

^bVerified number of seeded oocysts per milliliter as determined on 25-mm-diameter black polycarbonate membranes.

^cNA, not applicable.

^dND, none detected.

The results of IMS-FA microscopic detection and confirmation of oocysts seeded into deionized H₂O and naturally occurring waters are summarized in Table 1. In all but one case, IMS-FA detected less than the nominal oocyst value. The six oocysts detected and microscopically confirmed in replicate 2 of Cache la Poudre water seeded with 5 oocysts/ml can be attributed to introduced standard deviation error (see above) or to the presence of naturally occurring oocysts in the river, as Cryptosporidium spp. have been previously detected in this water source (Gertig, personal communication). These results indicate that IMS-FA is able to efficiently recover and detect low numbers of oocysts seeded into natural waters of low turbidity.

The results of IMS-PCR detection, RFLP digestion, and DNA sequencing analyses are summarized in Table 2. Like microscopic detection, IMS-PCR detection proved to be sensitive for detecting the different nominal oocyst doses in both natural waters and deionized H₂O. The negative results for the 5-oocyst/ml dose in deionized H₂O and for the third replicate of the Cache la Poudre sample seeded with a 15-oocyst/ml dose can most likely be explained by losses that occurred during repeated centrifugation of oocysts. In this protocol, oocysts recovered by IMS were washed three times by centrifugation in order to reduce naturally occurring PCR inhibitors. During this washing process, oocysts could have been inadvertently lost. Alternatively, it is possible that the limits of detection with this nested primer set were not met. However, this is unlikely since the detection limits were determined to be 100, 94, 92, 88, 76, 56, and 38% for 10, 7, 5, 4, 3, 2, and 1 oocysts, respectively (24).

TABLE 2.

Seeded oocyst detection by IMS-PCR and confirmation of identity by RFLP analysis and DNA sequencing

Sample	No. of oocysts/ml	Replicate	PCR detection	RFLP digesta			Identity determined by sequencingb
				DraI	DraII	VspI
Deionized H2O	0	1	−	NDc	ND	ND	ND
	5	1	−	ND	ND	ND	ND
	10	1	+	−	−	−	ND
	15	1	+	−	−	−	ND
Dowdy Lake	0	1	+d	+	−	−	Gymnodinium spp.
	5	1	+	−	−	−	C. parvum
		2	+	−	−	−	C. parvum
		3	+d	+	−	−	Gymnodinium spp.
	10	1	+d	+	−	−	Gymnodinium spp.
		2	+	−	−	−	C. parvum
		3	+	−	−	−	C. parvum
	15	1	+	−	−	−	C. parvum
		2	+	−	−	−	C. parvum
		3	+	−	−	−	C. parvum
Cache la Poudre River	0	1	+	−	+	−	C. muris
	5	1	+	−	−	−	C. parvum
		2	+	−	−	−	C. parvum
		3	+	−	−	−	C. parvum
	10	1	+	−	−	−	C. parvum
		2	+	−	−	−	C. parvum
		3	+	−	−	−	C. parvum
	15	1	+	−	+	−	Undetermined
		2	+	−	−	−	C. parvum
		3	−	ND	ND	ND	ND

^a+, RFLP-positive digestion; −, RFLP-negative digestion.

^bIdentity as determined by nucleotide sequencing and BlastN search.

^cND, not done.

^dThe PCR amplicon (approximately 625 bp) was larger than the expected 590-bp amplicon of C. parvum.

Interestingly, the two natural water background control samples, while negative as determined by the IMS-FA technique, were both positive as determined by the IMS-PCR (Table 2), and the amplicon from Dowdy Lake was visibly larger (approximately 625 bp) than the amplified C. parvum positive control DNA (Fig. 1A). While the positive amplification results were not a surprise for the Cache la Poudre River water, since Cryptosporidium spp. have been detected previously in this water (Gertig, personal communication), the positive PCR results for Dowdy Lake water were unexpected. Subsequently, the Cache la Poudre River and Dowdy Lake background PCR amplicons were DNA sequenced and determined to be 584 and 624 bp long, respectively. BlastN searches (1) identified the 584-bp Cache la Poudre amplicon as a Cryptosporidium muris amplicon (GenBank accession no. AF093496) with a pairwise identity value of 98.8% and the 624-bp Dowdy Lake amplicon as a dinoflagellate amplicon (Gymnodinium spp.; GenBank accession no. AF274260) with a pairwise identity value of 90.5% (Table 2). Microscopy analysis of Dowdy Lake water identified the dinoflagellates Ceratium spp. and Peridinium spp. but not Gymnodinium spp.

FIG. 1.

Nested PCR (A) and RFLP digestion with DraI ( B) or DraII (C) of Cryptosporidium spp. and two natural water matrixes with or without seeded C. parvum oocysts following IMS. Lanes1 to 4, amplicons from known isolates; lanes 5 to 9, amplicons from water matrixes; lane 1, C. parvum Iowa isolate; lane 2, C. andersoni; lane 3, C. muris; lane 4, G. fuscum; lane 5, Dowdy Lake background sample; lane 6, Cache la Poudre background sample; lane 7, Dowdy-5-3; lane 8, Dowdy-10-1; lane 9, Poudre-15-1; lanes M, 50-bp molecular weight marker (Gibco BRL, Grand Island, N.Y.).

Recognizing the possibility that all the oocyst-seeded water samples could be PCR positive due to background DNA and not seeded C. parvum oocysts, we performed RFLP analysis and DNA sequencing of all samples. DraII digestion of the Cache la Poudre background amplicon resulted in two fragments of 298 and 286 bp, while the positive control C. parvum amplicon was undigested (Fig. 1C), a result that would be expected if the amplicon was derived from C. muris. Next, the 18S rRNA sequences of Gymnodinium spp. and C. parvum were examined for a restriction enzyme able to distinguish the two organisms. DraI was identified, and subsequent digestion of the Dowdy Lake background sample produced predicted 352- and 272-bp fragments, while the positive control C. parvum amplicon was undigested (Fig. 1B).

All seeded PCR-positive water samples were then subjected to digestion with DraI, DraII, and VspI. Restriction digestion with VspI was performed to confirm that PCR amplicons were from C. parvum genotype 2 and not genotype 1, and subsequent VspI digestions of all background and seeded water samples were negative (Table 2). However, three samples were digested with either DraI or DraII; replicate 3 of Dowdy Lake water seeded with 5 oocysts (Dowdy-5-3) was digested with DraI, replicate 1 of Dowdy Lake water seeded with 10 oocysts (Dowdy-10-1) was digested with DraI, and replicate 1 of Cache la Poudre River water seeded with 15 oocysts (Poudre-15-1) was digested with DraII (Fig. 1B and C). In addition, the PCR amplicons of samples Dowdy-5-3 and Dowdy-10-1 were both visually larger than the positive control C. parvum amplicon (Fig. 1A), as was the Dowdy Lake background amplicon. Therefore, preliminary data suggested that DNA from a dinoflagellate was amplified in samples Dowdy-5-3 and Dowdy-10-1 and DNA from C. muris was amplified in sample Poudre-15-1. DNA sequencing and BlastN (1) searches identified both Dowdy-5-3 and Dowdy-10-1 amplicons as a dinoflagellate amplicon, possibly Gymnodinium spp., like the amplicon in the Dowdy Lake background sample. The pairwise levels of identity between the Dowdy-5-3 and Dowdy-10-1 samples and Gymnodinium spp. were 91.3 and 90.1%, respectively. The pairwise level of identity between samples Dowdy-5-3 and Dowdy-10-1 was 93%. Unfortunately, there was not sufficient Poudre-15-1 amplicon to successfully sequence it (Table 2). Two additional attempts were made to PCR amplify a nested amplicon from the Poudre-15-1 external master mixture, followed by DraII digestion and DNA sequencing. Both attempts resulted in undigested amplicons, and sequence data identified the new amplicons as C. parvum amplicons. These results suggest that Poudre-15-1 contained DNA of both C. muris and C. parvum. This finding has important implications. Since multiple Cryptosporidium spp. can be present in both environmental and treated water samples, the PCR methods used for detecting C. parvum in water (i) need to be confirmed with C. parvum-specific primers, (ii) should incorporate species-specific RFLP analysis, and/or (iii) should include DNA sequencing for proper identification.

The positive PCRs for the environmental background samples prompted continued evaluation of the nested primer set. DNA from Cryptosporidium andersoni KSU-3 and C. muris 108735 (both donated by Steve Upton, Kansas State University), as well as Gymnodinium fuscum CCMP1677 (Provasoli-Guillard National Center for Culture of Marine Phytoplankton McKown Point, West Boothbay Harbor, Maine), were extracted by using a QIAamp DNA stool mini kit (QIAGEN Inc.) as recommended by the manufacturer. The PCR conditions were as previously described (24), and restriction enzyme digestion with DraI or DraII followed amplification. Both C. andersoni and C. muris were PCR positive, and digestion with DraII was able to discriminate these organisms from the C. parvum positive control (Fig. 1A and C). Isolated DNA from G. fuscum also was amplified, and digestion with DraI produced the predicted bands, as seen in the Dowdy Lake background sample (Fig. 1A and B). Finally, DNA sequencing of G. fuscum revealed pairwise levels of identity with the Dowdy Lake background amplicon, Dowdy-5-3 amplicon, and Dowdy-10-1 amplicon of 90.3, 89.7, and 88.9%, respectively.

The finding that DNA from an unknown Dowdy Lake dinoflagellate and isolated DNA from cultured G. fuscum both were amplified by the nested primer set protocol emphasizes the need to further evaluate primer sets designed specifically to detect C. parvum in environmental samples. Gymnodinium spp. and Cryptosporidium spp. are members of the protozoan infrakingdom Alveolata sensu Cavalier-Smith (6). Alveolata is a robust, monophyletic taxon, as confirmed by rRNA phylogenies (5, 11), and contains three phyla: Apicomplexa (synonym, Sporozoa; apicomplexans), Ciliophora (ciliates), and Dinozoa (dinoflagellates). Furthermore, it is recognized that the levels of ribosomal DNA homology are high among species, genera, and even families, and similar levels of homology can be observed among rapidly diverging lineages, as is the case for the three phyla constituting the Alveolata. In fact, relevant to this study, alignment of complete 18S rRNA gene sequences of C. parvum (GenBank accession no. AF108864) and Gymnodinium spp. revealed 86.5% pairwise identity. Because C. parvum oocysts, dinoflagellates, and ciliates are common in aquatic habitats and because of the close phylogenetic relationship among these phyla, it is not entirely surprising to observe that a given 18S rRNA primer set amplifies DNA derived from multiple alveolate members present together in an environmental sample. Therefore, we believe that other defined 18S rRNA primer sets employed for environmental screening and detection of C. parvum should also be tested with DNA templates derived from common aquatic microorganisms (e.g., Gymnodinium spp.) that are closely related to C. parvum.

In summary, this study demonstrated that IMS-FA and IMS-PCR both were able to detect low numbers of oocysts (i.e., 5 oocysts) seeded into natural waters with low turbidities. We recommend that ongoing studies evaluating C. parvum recovery and detection methods incorporate low oocyst seed doses. Furthermore, to obviate introduction of a standard deviation, we strongly believe that oocysts should be enumerated by flow cytometry and sorted directly into designated experimental vessels for use in a selected method. Lastly, the recommendation stated above concerning proper identification of a Cryptosporidium sp. in a water sample also applies to the entire water matrix. Water samples contain diverse assemblages of organisms, and DNA liberation and recovery techniques do not discriminate between the different organisms; thus, DNA isolated from a water sample is representative of all the organisms present in the sample. Therefore, we recommend that PCR detection of a target organism from an environmental sample be based upon (i) highly specific and tested PCR primers, (ii) PCR amplicon size, (iii) species-specific RFLP digestion, and/or (iv) DNA sequencing for proper species identification.