Occurrence, Source, and Human Infection Potential of Cryptosporidium and Giardia spp. in Source and Tap Water in Shanghai, China

Abstract

Genotyping studies on the source and human infection potential of Cryptosporidium oocysts in water have been almost exclusively conducted in industrialized nations. In this study, 50 source water samples and 30 tap water samples were collected in Shanghai, China, and analyzed by the U.S. Environmental Protection Agency (EPA) Method 1623. To find a cost-effective method to replace the filtration procedure, the water samples were also concentrated by calcium carbonate flocculation (CCF). Of the 50 source water samples, 32% were positive for Cryptosporidium and 18% for Giardia by Method 1623, whereas 22% were positive for Cryptosporidium and 10% for Giardia by microscopy of CCF concentrates. When CCF was combined with PCR for detection, the occurrence of Cryptosporidium (28%) was similar to that obtained by Method 1623. Genotyping of Cryptosporidium in 17 water samples identified the presence of C. andersoni in 14 water samples, C. suis in 7 water samples, C. baileyi in 2 water samples, C. meleagridis in 1 water sample, and C. hominis in 1 water sample. Therefore, farm animals, especially cattle and pigs, were the major sources of water contamination in Shanghai source water, and most oocysts found in source water in the area were not infectious to humans. Cryptosporidium oocysts were found in 2 of 30 tap water samples. The combined use of CCF for concentration and PCR for detection and genotyping provides a less expensive alternative to filtration and fluorescence microscopy for accurate assessment of Cryptosporidium contamination in water, although the results from this method are semiquantitative.

INTRODUCTION

Waterborne cryptosporidiosis and giardiasis are significant public health concerns (6, 43). Cryptosporidium spp. and Giardia spp. are threats to water supplies because they are resistant to chlorine disinfection, have a low infectious dose, and are harbored by many animal species (57). Consequently, numerous waterborne outbreaks of cryptosporidiosis and giardiasis have been documented (62). Therefore, Cryptosporidium and Giardia are the only major pathogens specifically identified in drinking water regulations in industrialized countries and China (16, 21, 30, 48).

Farm animals, humans, and wild animals have been considered major contributors of Cryptosporidium and Giardia (oo)cyst contamination in surface water. The relative contribution of each animal species depends on the environmental setting (2, 27, 40, 59, 61). Although oocysts of all Cryptosporidium species can potentially appear in water, few of them are important human pathogens (56). Because oocysts of most Cryptosporidium spp. are morphologically similar and have the potential to be present in water, genotyping of Cryptosporidium oocysts in water is essential for source water management and risk assessment. The same is true for assemblages of Giardia duodenalis.

Currently, the identification of Cryptosporidium oocysts and Giardia cysts in environmental samples is largely performed by using the U.S. Environmental Protection Agency (U.S. EPA) Method 1623 and similar testing methods from the United Kingdom and other countries (54, 61). U.S. EPA Method 1623 requires concentration of (oo)cysts by filtration, isolation of (oo)cysts by immunomagnetic separation (IMS), staining of recovered (oo)cysts with fluorescent antibodies and 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI), and microscopic detection and enumeration of the stained oocysts or cysts (47). However, because of the high supply cost of this method (>$300), it is rarely used in water works in China. On the other hand, flocculation has been shown to be an inexpensive and convenient method for recovery of Cryptosporidium, Giardia, Toxoplasma, bacteria, and viruses in various water sources (64). In flocculation studies conducted, Cryptosporidium oocyst recoveries were 59% for alum when 1,000 oocysts/liter was used in seeding (50) and 64 to 75% for calcium carbonate when 75 to 1,000 oocysts/liter was used in seeding (11, 46, 50). Giardia recoveries were 72 to 77% for calcium carbonate when 85 cysts/liter were used to seed river or tap water (46).

PCR-based techniques have been used increasingly for detection and characterization of Cryptosporidium oocysts and G. duodenalis cysts in water (3, 13, 23, 27, 58, 61). Unlike Method 1623, recent PCR methods can differentiate Cryptosporidium and Giardia species that infect humans from those that do not infect humans. Because most Cryptosporidium species and genotypes are host specific, genotyping techniques are also used for contamination source tracking, although it is difficult to differentiate the original infected hosts from the vector hosts which carried the oocyst to the watershed (25). The most commonly used genotyping tool is a small-subunit (SSU) rRNA gene-based PCR-restriction fragment length polymorphism (RFLP) technique, which has been used effectively for genotyping Cryptosporidium oocysts in surface water, storm water, finished water, and wastewater samples (3, 13, 18, 27, 29, 45, 58, 61, 65). Similarly, genotyping methods, such as PCR combined with sequencing of the triosephosphate isomerase and glutamate dehydrogenase genes can be used to differentiate G. duodenalis assemblages A and B, which infect humans, from assemblages C to H, which do not infect humans, although the broad host ranges of assemblages A and B make the precise tracking of contamination source of G. duodenalis in water difficult (19).

Most genotyping studies of Cryptosporidium in source water or drinking water were conducted in industrialized nations. In developing countries, there has been only one small-scale study in Kenya, in which Cryptosporidium oocysts in nine surface water samples were genotyped (36). Because the stringency of wastewater treatment and discharge and the intensity of animal husbandry are very different between developing countries and industrialized nations, observations made in the latter cannot be applied easily to the former. In China, there have been several studies on the occurrence of Cryptosporidium and Giardia species using microscopy (9, 51, 52, 55, 63, 66), but a study on molecular characterization of Cryptosporidium in source or drinking water has not been done. Therefore, the contamination source and human infection potential of oocysts in water in China and other developing countries are unknown.

The objectives of this study were as follows: (i) to estimate the prevalence of Cryptosporidium and Giardia spp. in source and tap water in Shanghai, China; (ii) to identify the genotypes of Cryptosporidium in water in Shanghai, China; (iii) to determine the source of contamination and human infection potential of Cryptosporidium in water; (iv) to find a less expensive method for detecting Cryptosporidium and Giardia in source water; and (v) to associate recovery of parasites with water quality characteristics.

MATERIALS AND METHODS

Study site.

Shanghai is one of the biggest cities in China. It covers 6,341 km² and has an estimated population of 20 million. In the city, there is one main river, the Huangpu River, which is 113.4 km long. This river is used for drinking water, irrigation, drainage, shipping, fishery production, and tourism/recreation. The Huangpu River has been providing source water for 70% of Shanghai residents and will provide source water for 50% of the residents after 2012 when a new reservoir with water from the Yangtze River is fully operational. The Huangpu River receives water from Taihu Lake via the Taipu River and from Dianshan Lake via the Dahei River. Although the Dahei River is within the water protection area (1,058 km²) in Shanghai, which restricts the load and concentration of sewage discharge and farming, along the Taipu River, there are 23 towns and villages with a total population of 1,200,000 persons. Some townships are major suppliers of livestock and poultry for Shanghai. The source water sampling site in the Huangpu River was located upstream of Shanghai, at the intake of the drinking water treatment plants for Shanghai.

Sample collection and processing.

A total of 50 source water samples were collected from the end of May 2009 to the middle of January 2010, with an average of one to three samples per week with the exception that no samples were collected in August and October 2009. Each source water sample was split into two 20-liter aliquots. One aliquot was processed according to U.S. EPA Method 1623 (47). Briefly, the aliquot was filtered through Filta-Max xpress filters (IDEXX Laboratories, Inc., Westbrook, ME) and eluted using a Filta-Max xpress wash station (IDEXX Laboratories, Inc.). If only a partial aliquot was filtered because of clogging, the filtered volume was recorded. Nineteen samples (38%) were not fully filtered, and the filtered volume was between 10 to 18 liters, with five samples of more than 15 liters and the remaining samples between 10 and 15 liters. After centrifugation at 1,500 × g for 15 min (Eppendorf 5810; Eppendorf, Hamburg, Germany), the pellet was used for immunomagnetic separation (IMS) using an anti-Cryptosporidium and anti-Giardia IMS kit (Dynabeads GC-Combo; Invitrogen Dynal, A.S., Oslo, Norway). Only 0.5 ml of pellet (1/2 to 1/4 of the pellet volume for water samples) was processed for IMS per sample. Fluorescence staining was performed using a fluorescein isothiocyanate (FITC)-conjugated anti-Cryptosporidium and anti-Giardia monoclonal antibodies (Aqua-Glo G/C kit; Waterborne, Inc., New Orleans, LA). The internal structure of (oo)cysts was confirmed by 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) staining. On the basis of the number of (oo)cysts detected, the portion of the pellet for microscopy and the water volume filtered, the number of (oo)cysts was reported as the number of equivalents in 10 liters of water. Another portion of the remaining pellet (0.5 ml) was used for DNA extraction and PCR analysis.

The other 20-liter aliquot of source water was concentrated by the calcium carbonate flocculation (CCF) method (50). Briefly, a sample was separated into two 10-liter aliquots and stored in flat-bottomed polypropylene barrels. The water was mixed with 100 ml of 1 M calcium chloride (CaCl₂) and 100 ml of 1 M sodium hydrogen carbonate (NaHCO₃) by continuous stirring on a magnetic stirrer, and the pH was adjusted to 10 by the addition of 5 M sodium hydroxide (NaOH). The water sample was then allowed to rest undisturbed at room temperature overnight. The next day, the supernatant fluid was removed by vacuum aspiration, and the residue was dissolved by adding 200 ml of 10% sulfamic acid and shaking. Another 100 ml of sulfamic acid was added to the container if the flocculate was not fully dissolved. The dissolved flocculate was transferred to centrifuge tubes. The barrels were rinsed twice with 0.01% Tween 80; they were rinsed first with 200 ml of 0.01% Tween 80 and then with 100 ml of 0.01% Tween 80. The rinses were also transferred to the centrifuge tubes. After centrifugation at 3,000 × g for 20 min, the supernatant was removed and the pellet volume was recorded. The pellet was washed with Dulbecco's phosphate-buffered saline (DPBS) by centrifugation two or three times until the pH in 10 ml of DPBS was 7.0. A portion (0.5 ml) (1/3 to 1/4 of the pellet volume for the majority of the water samples and 1/9 and 1/12 for two samples) of the pellet was further processed with immunomagnetic separation (IMS), fluorescence staining, and microscopic observation according to U.S. EPA Method 1623 (47). The number of (oo)cysts was reported as the number of equivalents in 10 liters of water as stated above. Another portion of the remaining pellet (0.5 ml) was used for DNA extraction and PCR analysis.

Within the same period, 30 100-liter tap water samples were obtained from mid-June 2009 to mid-January 2010 from a local laboratory, with an average of one or two samples per week except that no samples were collected in August and October 2009. The tap water samples were concentrated by a filtration method using Filta-Max xpress filters (IDEXX Laboratories, Inc.) as stated above and examined for Cryptosporidium oocysts and Giardia cysts using U.S. EPA Method 1623 (47).

The recoveries of the standard Method 1623 and CCF-modified Method 1623 for Cryptosporidium oocysts and Giardia cysts in samples was monitored once every 3 months by spiking 10 liters of distilled water with Easyseed containing 100 oocysts and 100 cysts (BTF Pty Ltd., North Ryde, NSW, Australia).

For water quality analyses, total coliform and Escherichia coli were enumerated for all water samples using the Quanti-Tray/2000 (IDEXX Laboratories, Inc.) as specified by SM 9223B (16a). Heterotrophic plate count (HPC) was enumerated using the SimPlate for HPC (IDEXX Laboratories, Inc.). The turbidity of samples was measured with a Hach 2100P turbidimeter (Hach Co., Loveland, CO). The total dissolved solids (TDS) and conductivity were measured with a Hach MP-6p meter (Hach Co.).

DNA extraction and Cryptosporidium genotyping.

DNA was extracted directly from 0.5-ml water concentrate of each sample from filtration or CCF using the FastDNA spin kit for soil (Bio 101, Carlsbad, CA) and eluted in 100 μl of reagent-grade water as described previously (26). Cryptosporidium oocysts present in the samples were genotyped by nested PCR amplification of an approximately 830-bp fragment of the small-subunit (SSU) rRNA gene and restriction fragment length polymorphism (RFLP) analysis of the secondary PCR products, using the restriction enzymes SspI and VspI (27). Each sample was analyzed five times by the PCR-RFLP technique, using 2 μl of the DNA extract per PCR. Cryptosporidium baileyi DNA was used as a positive control in all SSU rRNA-based PCR-RFLP analyses. To neutralize residual PCR inhibitors in the extracted DNA, 400 ng of nonacetylated bovine serum albumin (Sigma-Aldrich, St. Louis, MO) per μl was used in the primary PCR. All positive secondary PCR products were sequenced to confirm the genotype identification.

Sequence analysis.

After purification using the MultiScreen PCR plate (Millipore, Bedford, MA), the secondary PCR products of the SSU rRNA gene were sequenced directly with the secondary PCR primers, using an ABI BigDye Terminator v. 3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) and the procedures suggested by the manufacturer. The sequences were read on an ABI3130 genetic analyzer (Applied Biosystems). Sequence accuracy was confirmed by two-directional sequencing and sequencing of all PCR products for each positive sample. Nucleotide sequences obtained were aligned with reference Cryptosporidium SSU rRNA sequences by using the ClustalX 1.81 package (ftp://ftp-igbmc.u-strasbg.fr/pub/ClustalX/) to determine the Cryptosporidium genotype.

Statistical analysis.

The association between detecting Cryptosporidium and Giardia and physical-chemical (turbidity, conductivity, and TDS) and microbiological parameters (log HPC, log E. coli count, and log total coliform count) of water was compared using Student's t test. Differences in detecting Cryptosporidium by different methods were compared using chi-square analysis. P values of <0.05 were considered statistically significant. All statistical analyses were performed with SPSS 10.0 for Window (SPSS Inc., Somers, NY).

RESULTS

Occurrence of Cryptosporidium oocysts and Giardia cysts.

Of the 50 source water samples, FITC staining indicated that 16 samples (32%) were Cryptosporidium positive and nine (18%) were Giardia positive using the standard U.S. EPA Method 1623 (Table 1). In the positive samples, the Cryptosporidium concentration ranged from 1.8 to 22 with an average of 5.2 oocysts in 10 liters of water; and the Giardia concentration ranged from 2 to 8 with an average of 4 cysts in 10 liters of water. DAPI staining indicated that 14 of the 16 Cryptosporidium-positive samples were DAPI positive, with the concentration of intact oocysts ranging from 1.8 to 9 and with an average of 3.5 oocysts in 10 liters of water. In contrast, three of nine Giardia-positive samples were DAPI positive, with the concentration of intact cysts ranging from 2.5 to 3 and an average of 2.8 cysts in 10 liters of water. No correlation was found between the positive rates of the two parasites and filtered water volume. The recovery efficiencies by the standard method, Method 1623, was 28 to 42% for Cryptosporidium oocysts and 25 to 35% for Giardia cysts (n = 9), and DAPI staining found that more than 96% of the seeded (oo)cysts were DAPI positive.

Table 1.

Occurrence of Cryptosporidium and Giardia spp. in source and tap water samples in Shanghai, China

Water sample and detection technique	Cryptosporidium				Giardia
	Filtration		CCF		Filtration		CCF
	Detectiona	Intensity (mean ± SD)b	Detection	Intensity (mean ± SD)	Detection	Intensity (mean ± SD)	Detection	Intensity (mean ± SD)
Source water
Microscopy	16/50c	5.2 ± 6.5	11/50c	4.5 ± 4.3	9/50d	3.8 ± 2.1	5/50d	2.2 ± 0.5
PCR	6/50	1.8 ± 0.8	14/50	2.1 ± 1.2	NDe		ND
Tap water and microscopy	2/30	0.2 ± 0.0	ND		0		ND

^aNumber of samples positive for Cryptosporidium or Giardia/number of samples analyzed.

^bFor microscopy, the intensity is the calculated concentration of Cryptosporidium oocysts or Giardia cysts for 10 liters of water is shown. For PCR, the intensity is the number of PCR-positive replicates/5 replicates for all positive samples.

^cEight samples were positive for Cryptosporidium by both methods.

^dNo sample was positive for Giardia by both methods.

^eND, not done.

For the source water samples analyzed by microscopy of concentrates by calcium carbonate flocculation (CCF), FITC staining indicated that 11 samples (22%) and five samples (10%) were positive for Cryptosporidium and Giardia, respectively (Table 1). In the positive samples, the Cryptosporidium concentration ranged from 2 to 16 with an average of 4.5 oocysts in 10 liters of water, whereas the Giardia concentration ranged from 1 to 4.5 with an average of 2.2 cysts in 10 liters of water (Table 1). DAPI staining indicated that 10 of the 11 Cryptosporidium-positive samples were DAPI positive, with the concentration of intact oocysts ranging from 2 to 16, with an average of 4.4 oocysts in 10 liters of water. In contrast, three of the five Giardia-positive samples were DAPI positive, with the concentration of intact cysts ranging from 1 to 2, with an average of 1.7 cysts in 10 liters of water. Eight source water samples were positive for Cryptosporidium by both filtration- and flocculation-based methods, while no sample was positive for Giardia by both methods. The recovery efficiencies by the CCF-modified Method 1623 was 54 to 58% for Cryptosporidium oocysts and 62 to 66% for Giardia cysts (n = 9), and DAPI staining found that more than 95% seeded (oo)cysts were DAPI positive.

Cryptosporidium was detected by PCR in 17 source water samples, with six PCR-positive samples following concentration by filtration and 14 PCR-positive samples following concentration by CCF. Three samples were positive by PCR of concentrates from both filtration and CCF. One sample concentrated by filtration and one sample concentrated by CCF produced PCR products that were noticeably larger than the expected amplicons (∼1,000 bp versus ∼830 bp) and therefore were not considered Cryptosporidium positive. For the samples concentrated by filtration, all those positive by PCR were negative by microscopy. For the samples concentrated by CCF, two of the samples positive by PCR were also positive by microscopy, and the remaining 12 samples positive by PCR were negative by microscopy. Because only 6 of the 20 cysts observed were DAPI positive and Giardia PCR is known to have lower sensitivity, PCR analysis for Giardia was not performed.

Two of the 30 tap water samples analyzed by Method 1623 were Cryptosporidium positive, each containing two oocysts in 100 liters of tap water. None of the tap water samples were positive for Giardia (Table 1).

Comparison of Cryptosporidium and Giardia detection by the standard and CCF-incorporated Method 1623.

Statistically, the differences between filtration and CCF in detecting Cryptosporidium and Giardia by microscopy were not significant for both Cryptosporidium and Giardia, although Cryptosporidium was detected more often higher detection than Giardia was (P ≥ 0.25). The difference was also not related to the water volume filtered in the standard Method 1623. In comparing Cryptosporidium detection by microscopy and PCR, significantly more Cryptosporidium were obtained by microscopy than by PCR when the samples were concentrated by filtration (P = 0.02), although similar positive rates were obtained by the two methods when the samples were concentrated by CCF (P = 0.49) (Table 1). The number of Cryptosporidium detected by PCR of CCF concentrates was comparable to that detected by the standard Method 1623 (P = 0.66).

Water quality parameters and detection of Cryptosporidium and Giardia by microscopy and PCR.

Of the source water samples, Cryptosporidium-positive samples (by microscopy or PCR) had a significantly lower mean total coliform count (P = 0.003) and mean E. coli count than Cryptosporidium-negative samples (P = 0.02) (Table 2). In contrast, Cryptosporidium-positive samples had significantly higher conductivity (P = 0.0001) and total dissolved solids (TDS) (P = 0.001) than Cryptosporidium-negative samples (Table 3). We did not find any significant correlation between the detection of Giardia and water quality parameters.

Table 2.

Correlation of total coliform, E. coli, and heterotrophic plate count with Cryptosporidium and Giardia detection in source water samples^a

Source water sample	No. of samples	Log HPC count (mean ± SD)b	P valued	Log total coliform count (mean ± SD)c	P value	Log E. coli count (mean ± SD)c	P value
Cryptosporidium
Positive	30	3.98 ± 0.44	0.24	3.18 ± 0.67	0.003	1.96 ± 1.01	0.02
Negative	20	4.12 ± 0.31		3.65 ± 0.20		2.55 ± 0.34
Giardia
Positive	14	3.94 ± 0.50	0.30	3.43 ± 0.68	0.63	2.15 ± 1.00	0.81
Negative	36	4.07 ± 0.35		3.34 ± 0.55		2.21 ± 0.82

^aCorrelation of total coliform, E. coli, and heterotrophic plate count (HPC) with detection of Cryptosporidium and Giardia in source water samples that were positive by microscopy or PCR analysis of concentrates from filtration or CCF.

^bNumber per 1 ml of sample.

^cNumber per 100 ml of sample.

^dThe P values result from a comparison of the values (counts) for source water samples that are positive and negative for Cryptosporidium or Giardia.

Table 3.

Correlation of turbidity, conductivity, and total dissolved solids (TDS) with Cryptosporidium and Giardia detection in source water samples^a

Source water sample	No. of samples	Turbidity (NTUb) (mean ± SD)	P valuec	Conductivity (μS/cm) (mean ± SD)	P value	TDS (mg/liter) (Mean ± SD)	P value
Cryptosporidium
Positive	30	38.20 ± 25.63	0.71	965.23 ± 218.42	0.0001	509.33 ± 154.21	0.001
Negative	20	35.37 ± 26.31		759.60 ± 122.69		380.15 ± 62.02
Giardia
Positive	14	41.79 ± 34.52	0.42	879.71 ± 218.85	0.95	498.36 ± 204.02	0.20
Negative	36	35.24 ± 21.63		884.25 ± 210.72		441.83 ± 105.52

^aCorrelation of turbidity, conductivity, and total dissolved solids (TDS) with Cryptosporidium and Giardia detection in source water samples that were positive by microscopy or PCR analysis of concentrates from filtration or CCF.

^bNTU, nephelometric turbidity unit.

^cThe P values result from a comparison of the values (turbidity, conductivity, and TDS) for source water samples that are positive and negative for Cryptosporidium or Giardia.

Occurrence of Cryptosporidium species by PCR.

RFLP analysis of the PCR products using restriction enzymes SspI and VspI indicated the presence of at least five different Cryptosporidium species/genotypes. DNA sequencing of all PCR products confirmed the results of the RFLP analysis. The five Cryptosporidium species were C. andersoni, C. suis, C. baileyi, C. hominis, and C. meleagridis (Tables 4 and 5). The most common species in source water samples was C. andersoni, which was found in 26 PCR products from 14 samples (Table 5). In contrast, C. suis was found in nine PCR products from seven samples, followed by C. baileyi in three PCR products from two samples, C. hominis in one PCR product from one sample, and C. meleagridis in two PCR products from one sample (Table 5). Of the 17 PCR-positive samples, seven (41.2%) contained more than one species, including six samples containing two species and one sample containing three species (Table 4).

Table 4.

Cryptosporidium species in water samples in Shanghai, China

Sample(s)	Species (no. of positive PCR resultsa)
	Filtration-based method	Flocculation-based method
29765		C. andersoni (1)
29412		C. andersoni (1)
29416		C. baileyi (2) and C. andersoni (1)
29418		C. suis (1)
29426		C. andersoni (2)
29427	C. suis (1)
29430		C. andersoni (1)
29432		C. andersoni (3)
29433, 29434	C. andersoni (2)	C. suis (1)
29436		C. andersoni (3) and C. suis (1)
29437, 29438	C. andersoni (2) and C. suis (1)	C. andersoni (4)
29439	C. suis (1) and C. andersoni (1)
29441	C. hominis (1)
29443, 29444	C. suis (2)	C. suis (1) and C. andersoni (1)
29446		C. andersoni (1)
29448		C. andersoni (2)
29452		C. meleagridis (2), C. andersoni (1), and C. baileyi (1)
Total positive	6	14
Total species	C. andersoni (5), C. suis (5), and C. hominis (1)	C. andersoni (21), C. suis (4), C. baileyi (3), and C. meleagridis (2)

^aNumber of positive PCR results in five replicate analyses of each sample.

Table 5.

Cryptosporidium species in water concentrates by both filtration and flocculation-based methods

Cryptosporidium species	Major known host	Pathogenic in humans?	No. of samples positive	No. of positive PCR resultsa
				Filtration	CCF
C. andersoni	Cattle	Rarely	14	5	21
C. suis	Pigs	Rarely	7	5	4
C. baileyi	Birds	No	2	0	3
C. hominis	Humans	Yes	1	1	0
C. meleagridis	Birds	Yes	1	0	2

^aTotal number of positive PCR results in five replicate analyses of each sample.

Source water samples concentrated by CCF produced more PCR-positive results than those concentrated by filtration (14 samples versus 6 samples or 30 PCR replicates versus 11 replicates) (Tables 4 and 5). As a result, more Cryptosporidium species were found in water samples concentrated by CCF (C. andersoni, C. suis, C. baileyi, and C. meleagridis) than in samples concentrated by filtration (C. andersoni, C. suis, and C. hominis). In water samples concentrated by CCF, there were 12 samples with C. andersoni, 4 samples with C. suis, 2 samples with C. baileyi, and 1 sample with C. meleagridis. In contrast, in samples concentrated by filtration, there were three samples with C. andersoni, three samples with C. suis, and one sample with C. hominis (Table 4).

DISCUSSION

Using the standard U.S. EPA Method 1623, 32% of the 50 source water samples were positive for Cryptosporidium and 18% were positive for Giardia. Finding Cryptosporidium and Giardia (oo)cysts in source water samples was expected because the study area was adjacent to highly populated and developed land areas. The prevalence of Cryptosporidium in this study was within the range obtained in other studies conducted in China, although the prevalence of Giardia was lower (34, 51, 52). The rate of Cryptosporidium-positive samples in source water was similar to data (20 to 40%) presented in recent studies in France, Finland, Hungary, Portugal, and Spain (4, 15, 17, 22, 40, 41, 59). Likewise, the prevalence of Giardia in this study was also similar to rates reported in Finland, Norway, and Portugal (11.8% to 16.7%) (3, 22, 42).

A higher occurrence of Cryptosporidium and Giardia in source water was obtained by microscopy of concentrates from filtration than from calcium carbonate flocculation (CCF), although the differences were not significant (P ≥ 0.25). Because the recovery rates for Cryptosporidium oocysts and Giardia cysts in this study were similar to data previously reported for CCF (64) or the standard Method 1623 (47), this difference could be associated with the high turbidity of source water samples. The turbidity of the source water samples in this study was between 12 and 157 nephelometric turbidity units (NTU). Previously, recovery rates of Cryptosporidium oocysts and Giardia cysts by filtration were positively correlated with the turbidity of samples (20, 24), whereas those by CCF were negatively correlated with the turbidity (64). In addition, a higher packed pellet volume was obtained by the CCF method (2 to 4 ml) than the standard filtration method (1 to 2.5 ml), which resulted in a higher proportion of the pellet being processed for immunomagnetic separation (IMS) and microscopy for filtration than for CCF. In addition, the flocculants could have some inhibitory effect on the IMS isolation of oocysts and cysts, although this issue has not been studied in this study or in earlier studies.

In contrast to observations by microscopy, when PCR was used, a lower detection rate (12% versus 28%) for Cryptosporidium was obtained in the source water samples concentrated by filtration than in the samples concentrated by CCF (Table 1). Even though the recommended FastDNA spin kit for soil for DNA extraction and 400 ng/μl of bovine serum albumin in the PCR were used to reduce PCR inhibition, this strategy is known to have a lower sensitivity than DNA extraction of Cryptosporidium oocysts isolated from water concentrates by IMS (26). It is unclear whether some of the chemical treatments in the process of flocculation can lead to better neutralization of PCR inhibitors, thus improving PCR detection of Cryptosporidium oocysts in water samples.

In this study, most samples that were positive for Cryptosporidium by microscopy were negative by PCR. Multiple reasons could have contributed to the inconsistent results obtained by PCR and microscopy for the same samples. The number of Cryptosporidium oocysts in water samples was low. Only a portion of the pellet was examined by microscopy or PCR, and only one oocyst was detected in the majority of microscopy-positive portions. It is possible that the stochastic distribution of oocysts in samples led to the absence of oocysts in many aliquots analyzed by PCR or microscopy. Furthermore, the number of empty Cryptosporidium oocysts in environmental samples can be high (31). In this study, 15 of the 37 oocysts and 3 of the 25 oocysts detected in pellets concentrated by filtration and CCF, respectively, were DAPI negative. The presence of more empty oocysts (ghosts) in the filtration method was possibly also partially responsible for the lower occurrence of Cryptosporidium oocysts detected by PCR in this study, because the ghosts could still be detected by microscopy but not by PCR. In a recent study in Luxembourg, only 25% of the samples positive by microscopy were confirmed by PCR (23).

The combination of CCF and PCR was shown to be a cost-effective method for detection of Cryptosporidium in water and for contamination source tracking. Although both Cryptosporidium and Giardia are listed in the water quality standards of China (16), the high cost of using standard Method 1623 ($300 per sample for the material used, including a filter, chemicals, and IMS and staining reagents) has restricted its usage in routine detection of these pathogens. As a result, the expensive set of Method 1623 equipment ($75,000 for a centrifuge, a filtration and elution system, IMS equipment, and a fluorescence microscope) purchased by many laboratories in China is seldom used because of the high running cost. In this study, although the number of Cryptosporidium detected by microscopy of concentrates by CCF was lower than that by filtration in Method 1623, when CCF was combined with PCR, Cryptosporidium detection was similar to that of the standard Method 1623 (P = 0.66). Currently, regulatory agencies and the water industry are trying to develop appropriate Cryptosporidium genotyping procedures downstream of Method 1623. Flocculation combined with PCR may be a viable option. A shortcoming of the CCF method combined with PCR may be its inability to quantify Cryptosporidium oocyst as Method 1623 does. This can be partially overcome by estimating the contamination intensity by the number of positive PCR results in multiple replicates (Table 1) (37, 60). Since many studies have shown that Method 1623 overestimates the human infection potential of Cryptosporidium in water because of its inability to differentiate Cryptosporidium species that are pathogenic in humans from Cryptosporidium species and genotypes that are not pathogenic in humans (27, 45, 61), CCF combined with PCR has some advantage because of its genotyping capability. Compared to conventional detection by filtration, IMS, and immunofluorescence microscopy, the low cost of CCF combined with PCR ($100 per sample for the material used, including chemicals, PCR reagents, and sequencing and $25,000 for equipment, including a centrifuge and a PCR machine) makes it cost-effective for detecting Cryptosporidium in source water in China. A drawback of the CCF technique is its inapplicability to finished water samples because of their low turbidity and requirement for processing much larger volume (100 liters or more) of samples. Like filtration-based methods, they also cannot be used in processing of source water with extremely high turbidity.

In this study, Cryptosporidium-positive samples detected by microscopy or PCR were associated with higher conductivity and TDS but lower total coliform and E. coli counts than Cryptosporidium-negative samples. The biological or environmental basis for this observation, however, is not clear, and the small sample size of the study could have contributed to some of the observed associations. Nevertheless, the correlation between the occurrence of Cryptosporidium and Giardia and conductivity was reported in samples of recreational rivers in Malaysia (5) and in a drinking water reservoir in Luxembourg (23). In addition, organic matter was found to be significantly correlated with Cryptosporidium and Giardia occurrence in surface water in Argentina (1). Total coliform, fecal coliform, or E. coli was reported to correlate with Cryptosporidium and Giardia occurrence in source water samples in several studies in Italy, Luxembourg, Spain, United States, Canada, and Argentina (1, 12, 23, 31, 39, 49). In some other studies, no correlation between total coliform or E. coli and Cryptosporidium was reported (7, 8). However, the negative correlation between microbiological parameters and Cryptosporidium occurrence as shown in this study was unexpected. The high turbidity of the source water in this study and small sample size could have contributed somehow to the discrepancy between this study and previous studies.

Genotyping results suggested that Cryptosporidium oocyst contamination of source water in Shanghai, China, was mainly from animal farms, especially cattle farms. In this study, C. andersoni was the dominant species, as it was found in more than 80% (14/17) of the Cryptosporidium-positive samples detected by PCR. Cryptosporidium andersoni is predominantly a parasite of juvenile and adult cattle and has been found only on a few occasions in small numbers of other farm animals, such as sheep, pigs, and horses (56). Interestingly, several other common bovine Cryptosporidium genotypes, such as C. parvum in preweaned calves and C. bovis and C. ryanae in older calves, were not found in water samples, indicating that adult cattle contributed to the contamination more than juvenile cattle and calves. Similar to this study, a predominance of C. andersoni was also found in the Potomac River and other watersheds in the United States (59, 61) and in other studies in industrialized nations (3, 14, 38, 44).

Genotyping results also suggested that pigs and birds were also important contamination sources in the source water in Shanghai, China. Cryptosporidium suis found in this study is a dominant species in pigs, whereas C. baileyi and C. meleagridis are mainly parasites of birds (57). Previously, C. suis has not been found in source water, whereas C. baileyi was found in one sample each in Canada (45) and Germany (53), and C. meleagridis was found in two samples in Japan (28). Finding Cryptosporidium from farm animals in a short river spanning mostly urban and suburban areas in one of the world's largest metropolitan areas was unexpected. Nevertheless, this result is in agreement with the fact that several upstream townships and villages are major producers of farm animals.

Surprisingly, urban wastewater discharge was not a significant contributor to Cryptosporidium oocyst contamination in this study. Although C. hominis is the most common Cryptosporidium genotype in untreated urban wastewater in the study area (18), only one source water sample was positive for C. hominis in this study. This is possibly because the sampling point is within the source water protection area of Shanghai, China. As expected, Cryptosporidium genotypes from wildlife were not found in this study, indicating that unlike source water supplying some major metropolitan areas in industrialized nations (25, 27, 33, 45, 53, 61), wild animals do not appear to be a source for Cryptosporidium contamination in source water in the Shanghai area. The river studied, the Huangpu River, is located in one of the highly developed regions in China with a high population density and intensive animal husbandry.

Among the five Cryptosporidium species found in water, only C. hominis and C. meleagridis are major human pathogens. The majority of human Cryptosporidium infections are caused by C. hominis and C. parvum (56). Even though C. meleagridis is also responsible for some human infections, it was found in this study at a very low prevalence. Human infections with C. andersoni and C. suis have each been reported in only two cases and three cases, respectively (10, 32, 35). Therefore, most the Cryptosporidium oocysts detected in source water in this study were not from major species that infect humans. On the other hand, finding Cryptosporidium oocysts in two drinking water samples highlights the need for more stringent treatment of drinking water in Shanghai.

In conclusion, the occurrence of Cryptosporidium oocysts and Giardia cysts in source water and the presence of Cryptosporidium in tap water in Shanghai, China, represent some potential water quality and public health problems. They also suggest the need for regular monitoring of the two pathogens in source water and treated water, which is currently impractical because of the expense of using Method 1623. The data generated by Method 1623 are also not very useful in accurate risk assessment of contamination. In contrast, the combined use of CCF for pathogen concentration and PCR for pathogen detection and genotyping can provide data that are useful in assessing the extent, source, and human infection potential of Cryptosporidium oocysts in water samples. The use of CCF and PCR in this study led to the identification of the farm animal origin and non-human-pathogenic nature of Cryptosporidium oocysts in source water in Shanghai. With further development, this technique would be a less expensive alternative to the standard Method 1623 for monitoring Cryptosporidium and Giardia contamination in raw source water in developing countries.