ABSTRACT
We evaluated the efficiency of five membrane filters for recovery of Cryptosporidium parvum oocysts and Giardia lamblia cysts. These filters included the Pall Life Sciences Envirochek (EC) standard filtration and Envirochek high-volume (EC-HV) membrane filters, the Millipore flatbed membrane filter, the Sartorius flatbed membrane filter (SMF), and the Filta-Max (FM) depth filter. Distilled and surface water samples were spiked with 10 oocysts and 10 cysts/liter. We also evaluated the recovery efficiency of the EC and EC-HV filters after a 5-s backwash postfiltration. The backwashing was not applied to the other filtration methods because of the design of the filters. Oocysts and cysts were visualized by using a fluorescent monoclonal antibody staining technique. For distilled water, the highest percent recovery for both the oocysts and cysts was obtained with the FM depth filter. However, when a 5-s backwash was applied, the EC-HV membrane filter (EC-HV-R) was superior to other filters for recovery of both oocysts (n = 53 ± 15.4 per 10 liters) and cysts (n = 59 ± 11.5 per 10 liters). This was followed by results of the FM depth filter (oocysts, 28.2 ± 8, P = 0.015; cysts, 49.8 ± 12.2, P = 0.4260), and SMF (oocysts, 16.2 ± 2.8, P = 0.0079; cysts, 35.2 ± 3, P = 0.0079). Similar results were obtained with surface water samples. Giardia cysts were recovered at higher rates than were Cryptosporidium oocysts with all five filters, regardless of backwashing. Although the time differences for completion of filtration process were not significantly different among the procedures, the EC-HV filtration with 5-s backwash was less labor demanding.
Cryptosporidium parvum and Giardia lamblia are among the major causative agents of protozoan diarrhea, with Cryptosporidium being potentially life threatening to immunocompromised individuals (2). Cryptosporidium is responsible for many waterborne outbreaks in the United States (4, 19, 34), the United Kingdom (4), and other countries (3). Although person-to-person contact and domestic animals are some of the possible sources of infection, exposure to and/or consumption of contaminated drinking water and the use of surface waters for recreational activities are among the most important routes of transmission of these parasites (8). There have been many reported outbreaks of cryptosporidiosis due to contaminated recreational water (35, 14, 22) and drinking water supplies (15, 17, 34).
Water treatment plant equipment failures have also been a potential cause of water supply contamination (18). The most significant Australian incident, which occurred during the Sydney water crisis in 1998, was due to failure of the water treatment process (32, 23).
Several epidemiological and ecological studies have been conducted to detect the presence and the abundance of these parasites in surface waters (16, 24, 25, 30). As drinking water supplies are most frequently obtained by the collection and treatment of surface water, it is important to constantly monitor the presence and seasonal fluctuation of these parasites in a given catchment area. Studies have shown that animal activity in catchment areas poses a major threat to surface water quality (1, 10, 13).
Many membrane filtration methods have been developed and used to detect the presence of both Cryptosporidium and Giardia in surface and treated water samples. Assessment of these methods in practice has shown various degrees of recovery efficiency for Cryptosporidium and/or Giardia (37, 29, 20, 27, 28, 12). The matrix variation and complexity of the procedures could explain these various results (5).
In Australia, several studies have addressed the occurrence of Cryptosporidium and Giardia in surface waters (33, 9). The Australian National Health and Medical Research Council has set no guideline value for Cryptosporidium, and no routine monitoring of distribution systems is recommended. However, the recent incidence of Cryptosporidium in Sydney (23) has led to ongoing monitoring surveillance in New South Wales. In the United Kingdom, the Drinking Water Inspectorate implemented regulations in 1999 which made it a criminal offense to supply water containing more than one Cryptosporidium oocyst per 10 liters of treated water (36).
Several membrane filters to detect Cryptosporidium have been developed and used in many laboratories in Australia and elsewhere. These include pleated-membrane capsule filters, polycarbonate membrane filters, cellulose-acetate membrane filters, compressed-foam depth filters, filters suitable for ultrafiltration, hollow-fiber filters, and yarn-wound cartridge filters (12, 20, 27, 28, 29). Procedures with these filters may require purification using either Percoll-sucrose flotation or immunomagnetic separation of oocysts and cysts. However, all procedures require immunofluorescent staining using a monoclonal antibody. The capacity for testing different volumes of water samples and the recovery efficiencies of these filters have been assessed individually or comparatively in several studies.
In view of the importance of detecting these parasites in drinking and surface waters, this study was undertaken to evaluate the recovery efficiency of five commercially available membrane filters. We also evaluated any potential for improvement for the given procedures by applying a short backwash to the filters whenever possible.
MATERIALS AND METHODS
Organisms and inoculation procedure. C. parvum oocysts and G. lamblia cysts were obtained from Biotechnology Frontiers (New South Wales, Australia) as two separate kits. An automated flow cytometry technique was used to accurately dispense cysts and oocysts, which were sealed and sterilized by gamma irradiation. These kits included an EasySeed kit containing tubes with 100 Cryptosporidium oocysts and 100 Giardia cysts as well as a ColorSeed kit containing tubes with 100 Cryptosporidium oocysts and 100 Giardia cysts. ColorSeed oocysts and cysts were modified by the attachment of Texas Red to the cell wall, which produced a red fluorescence when viewed with a Texas Red filter set. This modification differentiated these cysts and oocysts from their naturally occurring counterparts, which may have been present in environmental samples (39). Both kits contained a quality assurance certificate verifying the count obtained by flow cytometry. EasySeed tubes were added to the 10-liter distilled water samples, and ColorSeed tubes were added to the 10-liter raw water samples, according to the manufacturer's instructions.
Seeded distilled water samples.A total of 35 seeded samples of distilled water were prepared for all filtration methods. Water samples were mixed thoroughly before filtration. Upon completion of filtration, sample containers were rinsed with 200 ml of filter-sterilized reagent-grade water and 2 ml of 0.05% (vol/vol) Tween 80. This solution was also filtered to ensure removal of any cysts or oocysts that might have attached to the container's walls. Water samples were filtered by using a Watson Marlow peristaltic pump at a flow rate of 2 liters/min (purchased from Labtek, Queensland, Australia).
Seeded raw surface water samples.Duplicate 10-liter samples of raw water from a stream and associated dam were obtained and transported to the laboratory and tested within 48 h. Each container was seeded with a ColorSeed tube as described previously. Altogether, 21 seeded samples of raw water were prepared. Each 10-liter container was mixed thoroughly before filtration. Upon completion of filtration, containers were washed as just described.
Membrane filters. (i) Pall Life Science Envirochek filters.Envirochek (EC) (Pall Life Sciences, New South Wales, Australia) capsule filtration was performed in accordance with the manufacturer's instructions, except that the Laureth 12 elution buffer was replaced with a phosphate-buffered saline (PBS)-Tween-Antifoam buffer (2 liters of PBS buffer [pH 7.4], 300 μl of Tween 80, and 300 μl of Antifoam B). This replacement was based on demonstrated improvements in percent recoveries in previous trials and on the recommendation of Hunter Water Laboratories, New South Wales, Australia (21).
Duplicate 10-liter water samples seeded with target organisms were passed through the EC filter by using a peristaltic pump with a flow rate of 2 liters/min. One hundred milliliters of PBS elution buffer was added to the capsule and placed in a wrist action shaker for 5 min with the bleed valve set at the 12 o'clock position. Eluate was collected into a 200-ml centrifuge tube. The elution procedure was repeated once more, changing the orientation of the filter in the shaker to either the 3 o'clock or 9 o'clock position. The second wash eluate was combined in the same tube and centrifuged at 1,100 × g for 15 min. The supernatant was aspirated, and DYNAL immunomagnetic separation (IMS) was performed on the pellet. After the final acid-dissociated step in the IMS procedure was performed, the sample was air dried onto NaOH-prepared DYNAL Spot-on slides and fixed in methanol. Slides were stained with fluorescein isothiocyanate (FITC)-conjugated monoclonal antibodies specific to Cryptosporidium and Giardia and with 4′,6′-diamidino-2-phenylindole (DAPI) dihydrochloride hydrate; the slides were then examined under an epifluorescence microscope.
A parallel experiment was performed in which the flow direction of the pump was backwashed for 5 s (EC-R). Upon completion of filtration, the flow direction of the pump was reversed for 5 s so that a small amount of postfiltered sample, which remained in the apparatus tubing, was allowed to backwash the filter. No sample was allowed to leave the filter housing. The rest of the procedure was identical to the original experiment.
(ii) Pall Life Science EC-high-volume filters.EC-high-volume (HV) (Pall Life Sciences) capsule filtration was performed in accordance with the manufacturer's instructions with minor modifications, including the Laureth 12 elution buffer's replacement with the PBS-Tween-Antifoam buffer mentioned previously.
Ten-liter water samples were seeded and filtered as described previously. One hundred milliliters of PBS elution buffer was added to the capsule and placed in a wrist action shaker for 5 min with the bleed valve set at the 12 o'clock position. Eluate was collected into a 200-ml centrifuge tube. The elution procedure was repeated a second time and third time, changing the orientation of the filter in the shaker to the 3 o'clock and 9 o'clock positions, respectively; the samples were then centrifuged at 1,100 × g for 15 min between and after the second and third elutions. The supernatant was carefully aspirated, and IMS was performed on the pellet. Slides were prepared and examined as just described.
A parallel experiment was performed in which the filter was backwashed for 5 s (EC-HV-R). The rest of the procedure was identical to the original experiment.
(iii) Millipore flatbed membrane filters (MMF).Mixed-cellulose ester (cellulose acetate-cellulose nitrate) membrane filters (diameter, 142 mm) with a pore size of 3 μm were loaded onto stainless steel filter housing equipment (Millipore Australia Pty., Ltd., New South Wales, Australia). Ten-liter water samples seeded with target organisms were passed through the filter by means of vacuum. The membrane was removed from the filter housing and washed in a plastic dish by the addition of 20 ml of 0.1% PBS-Tween 80 (27, 28). A stainless steel paint scraper was then used to scrape the surface of the membrane. Washings from this procedure were collected into a 200-ml centrifuge tube. The procedure was repeated three times, and the eluates from all washes were combined and centrifuged at 1,100 × g for 15 min. The pellet was then subjected to IMS. Slides were prepared and examined as just described.
(iv) Sartorius flatbed membrane filters (SMF).Cellulose acetate membrane filters (diameter, 142 mm) with a pore size of 1.2 μm were placed onto stainless steel filter housing equipment (Sartorius Australia Pty., Ltd., Victoria, Australia). Ten-liter water samples seeded with target organisms were passed through the filter by means of vacuum. After completion of filtration, the membrane was removed and washed as described above. The three washes were combined and centrifuged at 1,100 × g for 15 min, and the pellet was subjected to IMS. Slides were prepared and examined as just described.
(v) Filta-Max depth filters.Filta-Max (FM) Automatic Wash Station, auxiliary equipment, and compressed-foam depth filters with an equivalent pore size of 1 μm were supplied by IDEXX Laboratories (New South Wales, Australia). Ten-liter water samples seeded with target organisms were passed through the filter by use of the same Watson Marlow peristaltic pump used for the EC filtration procedure (IDEXX recommends a flow rate of 4 liters/min). Filtration and elution procedures were carried out in accordance with the manufacturer's instructions.
The eluate was centrifuged at 1,500 × g for 15 min as recommended by the manufacturer. The supernatant was discarded, and the pellet was subjected to IMS. Slides were prepared and examined as just described.
IMS.IMS was performed on all concentrated samples according to the manufacturer's instructions (Dynabeads GC-Combo kit; Dynal Biotech Pty., Ltd., Victoria, Australia).
Monoclonal antibody.EasyStain containing FITC-conjugated anti-immunoglobulin (Ig) monoclonal antibodies specific to Cryptosporidium IgG1 and Giardia IgG1 (Biotechnology Frontiers) and DAPI stain (Sigma) were used to visualize the Cryptosporidium oocysts and Giardia cysts. A working strength of DAPI was prepared by mixing 10 μl of DAPI stock solution (2 mg/ml) with 10 ml of PBS solution (pH 7.4) and filter sterilized by using a 0.45-μm-pore-sized membrane.
Microscopic examination.A Zeiss Axioskop2 epifluorescence microscope was used to examine the stained slides. The entire well was examined under ×20 magnification, and confirmation was performed at ×40 magnification. Filter cubes appropriate for FITC (wavelength, 450 to 590 nm; FT 510 and LP 515), Texas Red (wavelength, 530 to 585 nm; FT 600 and LP 615) and DAPI (BP 365NM, FT 395, and LP397) were used to examine oocysts and cysts. Oocysts and cysts were located using the FITC filter set, and identification was assisted by using the DAPI filter set. Red fluorescence of ColorSeed organisms was examined by using the Texas Red filter set.
RESULTS
All membrane filters showed various degrees of recovery of Cryptosporidium oocysts (mean ± standard error of the mean percent recovery between 0.2 ± 0.2 for EC standard filter and 28.2 ± 3.8 for FM membrane filter) and Giardia cysts (percent recovery between 4 ± 1.5 per 10 liters for EC standard filter and 49.8 ± 5.4 per 10 liters for FM membrane filter) (Table 1). The percent recovery of Giardia cysts was much higher than that of oocysts with all membrane filters, except the EC-HV method for distilled water samples.
Number of water samples and the range of recovery of Cryptosporidium oocysts and Giardia cystsa
When a 5-s backwash was applied, the recovery efficiencies of EC-standard filter (reversed [EC-R]) and EC-HV membrane filter (reversed [EC-HV-R]) increased significantly for both the oocysts (11.6 ± 4.3 versus 0.2 ± 0.2, P = 0.0305 for EC-R and 53 ± 6.9 versus 18.2 ± 3.6, P = 0.0022 for EC-HV-R) and cysts (32.8 ± 10.3 versus 4 ± 1.5, P = 0.0079 for EC-R and 59 ± 5.1 versus 28 ± 4.6, P = 0.0021 for EC-HV-R), giving the highest recovery values to EC-HV membrane filter compared with those of other filters (Table 1).
The three most effective filtration procedures (yielding the highest percent recovery for seeded distilled water samples) were used to assess their efficiencies to recover Cryptosporidium oocysts and Giardia cysts in raw surface water. Using the same procedure, we spiked surface waters with 10 organisms/liter. The results showed that the EC-HV membrane filter, when subjected to a 5-s backwash, had a significantly (P = 0.0002 for oocysts and P = 0.0281 for cysts) higher percent recovery (i.e., 51.1 ± 4 for oocysts and 37.4 ± 3.8 for cysts) than did the FM depth filter (19.4 ± 2.2 for oocysts and 23.8 ± 3.5 for cysts). The percent recovery of the SMF was less than 10 for both Cryptosporidium oocysts (3.2 ± 2.7) and Giardia (8.2 ± 3.4) (Table 1).
The recovery efficiencies of all membrane filters were tested in replicates of 5 to 10 for both distilled and surface raw waters. The highest variations in percent recovery for both oocysts and cysts were observed with EC and EC-HV with and without backwash and FM depth filter (only with cysts) (Table 1).
The time required for completion of filtration process was also evaluated. It was found that, while the FM depth filter required one extra step for completion of the process, both EC and EC-HV took more time in the elution and centrifugation processes. Table 2 shows the steps and time required for each filtration process. The SMF, MMF, and FM procedures were cumbersome and time consuming, with flatbed membranes often becoming clogged during filtration of the raw water samples, resulting in a maximum filtration time of 2 h. There was potential for cross-contamination of samples filtered using the SMF, MMF, and FM techniques due to the excessive manual handling of the membranes postfiltration. The EC membranes remained enclosed in the capsule housing throughout the duration of the testing procedure, thus reducing the opportunity of cross-contamination.
Comparison of steps and time required for processing and preparation of five filtration processes
DISCUSSION
Detection of Cryptosporidium oocysts and/or Giardia cysts in surface waters, especially in reservoirs for drinking water supply, is of great public health importance. Several filtration processes have been developed and/or evaluated with respect to their efficiency for recovery of these organisms in water samples. In evaluating the efficacy of these filters, we thought it would be important to establish a baseline for the detection level of each filtration system under laboratory conditions. Therefore, we utilized distilled water spiked with a known number of oocysts and cysts. None of the filtration processes yielded 100% recovery when distilled water was used. Similar results have been reported by others using raw or distilled water (5, 12, 20, 21, 29). It has been previously shown that factors such as the age of the oocysts or cysts and the technique of spiking the water sample may play an important role in recovery of the organisms and that inorganic debris interferes with counting the organisms (11).
To minimize the effect of uneven distribution of oocysts and cysts, we analyzed 10-liter water samples. This volume of water has also been used by other researchers to investigate the recovery efficiency of filters (6, 38) and has proved sufficient for this purpose. Under such an ambient situation, the recovery efficiency of the FM depth filter was higher than those of the other available methods. The FM procedure is currently the method of choice in the United Kingdom and is endorsed by the Drinking Water Inspectorate as the standard operating protocol for monitoring treated water supplies. Higher recovery levels have been obtained in other previous studies (31, 36). On the other hand, the United States Environmental Protection Agency has evaluated and recommended the use of EC filters for the detection of oocysts and cysts in surface water samples (37). Surprisingly, we found very few studies that indicated higher efficacy of the FM filter than that of the EC (EC-HV) filter in recovery of the organisms from both distilled and raw water samples (31). In this experiment, the FM method yielded a higher recovery for both oocysts and cysts. In fact, the recovery of Giardia cysts was often higher than that of Cryptosporidium oocysts with all filters tested using distilled water, a result which is supported by the previous findings of other investigators (11, 26). This result could probably be due to the larger size of the Giardia cysts, which makes them easier to capture in the filtration material (7).
The backwash procedure was adapted in our laboratory after obtaining several low percent recovery incidents using EC filters and our constant observation of the compression status of the filter membrane during the process. A 5-s backwash was carefully applied to avoid breaking the membrane and to prevent the sample from leaving the capsule. This modification yielded a significant improvement in the recovery from the EC filters, making the EC-HV filter superior to the FM depth filter. Attempts to apply similar backwashing procedures to the other filters failed mainly because of the design of these filters and their vulnerability to backwashing.
In order to evaluate the efficacy of filters against factors that influence recovery of organisms in surface water, such as turbidity, we chose to use the three filtration processes that yielded highest percent recovery when distilled water was used (EC-HV-R, SMF, and FM). The 5-s backwash was shown to yield the highest percent recovery for the EC-HV. Interestingly, we found that the recovery of Cryptosporidium oocysts in raw water was much higher than that of Giardia cysts when a 5-s backwash was applied. This outcome was mainly due to a drop in recovery of cysts rather than an increase in the recovery of oocysts. The SMF become clogged when filtering raw water samples, often requiring several replacements of filters. This factor could be partially responsible for the lower percent recovery exhibited by this filter, as some of the organisms may be damaged or lost during the scraping procedure. During microscopic examination, broken and distorted oocysts were observed on slides prepared from the SMF and MMF procedures. This was perhaps due to the membrane scraping procedure used in the elution process. These disrupted oocysts may make definitive identification of the organisms more difficult. It should be noted that the effective filtration areas for the MMF and SMF filters (i.e., 158 cm2) and FM filter (i.e., 52 cm2) are much less than that of the EC filters (i.e., 1,300 cm2). We suspected that the smaller surface area of these filters might contribute to clogging of the membrane when filtering raw water samples and reducing the percent recovery of the organisms. The pleated membrane of the EC filters ensures a greater surface area. However, cysts and oocysts could become trapped in the folds of these pleats, thereby reducing the percent recovery of the organisms. The unique structure of the FM filter ensures that target organisms trapped in the pores of the foam are released by the disks expanding during the elution procedure.
The minimum acceptable requirement for accreditation from the National Association of Testing Authorities in Australia is recovery of at least 10%. This requirement was not met by the EC procedure, either for the seeded distilled water or for the raw water samples assessed during this study. Similar low percent recoveries were obtained when the SMF procedure was used for raw water samples. This finding is in contrast to previous studies showing a high recovery percentage for these filtration techniques (11, 27, 28, 38).
Regardless of the filtration procedure used, recovery efficiencies need to be determined for each type of water matrix likely to be analyzed by a laboratory. With each batch of water samples analyzed, the inclusion of an internal control (ColorSeed) is recommended to enable a recovery efficiency level to be quoted for each sample type analyzed. In our study, all filtration procedures underestimated the actual count of organisms present in a water sample.
In conclusion, we found that EC-HV filters, when subjected to a 5-s backwash, yield a higher percent recovery of both oocysts and cysts in distilled and raw surface waters. Although the time differences for completion of filtration process were not significantly different among procedures, the EC-HV filtration with 5-s backwash was also less labor demanding.
ACKNOWLEDGMENTS
We thank Leonie Cover, Trudy Graham, Robin Woodward, and Graham Vesey for their technical assistance and helpful comments. We also thank Rosemary Santangelo (IDEXX Laboratories), Paula Phan (Sartorius Australia), and Sally Campbell and Janelle Matfin (Millipore Australia) for supplying filters and for their technical assistance and advice. We are also grateful to the South East Queensland Water Board for providing replicates of raw water samples.
FOOTNOTES
- Received 28 August 2003.
- Accepted 6 January 2004.
- Copyright © 2004 American Society for Microbiology