Rapid Detection and Enumeration of Giardia lamblia Cysts in Water Samples by Immunomagnetic Separation and Flow Cytometric Analysis

Abstract

Giardia lamblia is an important waterborne pathogen and is among the most common intestinal parasites of humans worldwide. Its fecal-oral transmission leads to the presence of cysts of this pathogen in the environment, and so far, quantitative rapid screening methods are not available for various matrices, such as surface waters, wastewater, or food. Thus, it is necessary to establish methods that enable reliable rapid detection of a single cyst in 10 to 100 liters of drinking water. Conventional detection relies on cyst concentration, isolation, and confirmation by immunofluorescence microscopy (IFM), resulting in low recoveries and high detection limits. Many different immunomagnetic separation (IMS) procedures have been developed for separation and cyst purification, so far with variable but high losses of cysts. A method was developed that requires less than 100 min and consists of filtration, resuspension, IMS, and flow cytometric (FCM) detection. MACS MicroBeads were used for IMS, and a reliable flow cytometric detection approach was established employing 3 different parameters for discrimination from background signals, i.e., green and red fluorescence (resulting from the distinct pattern emitted by the fluorescein dye) and sideward scatter for size discrimination. With spiked samples, recoveries exceeding 90% were obtained, and false-positive results were never encountered for negative samples. Additionally, the method was applicable to naturally occurring cysts in wastewater and has the potential to be automated.

INTRODUCTION

Giardia spp. are ubiquitous protozoan intestinal parasites infecting the small intestine of humans and various animals. Giardia lamblia (syn. G. duodenalis or G. intestinalis) is the only Giardia species found in humans (45) and is the most common intestinal parasite of humans worldwide (27). The vegetative form of these parasites, the trophozoites, which detach from the intestinal villi, are pear-shaped, 9 to 21 μm long, 5 to 15 μm wide, and 2 to 4 μm thick and have two nuclei and eight flagella organized in four pairs (15). The noninvasive life cycle includes excystation and encystation processes, which take place during passage through the intestines. After detachment, the parasites form cysts that are ovoid, 8 to 14 μm long, and 7 to 10 μm wide and are excreted with the feces (23, 29). This encystation process involves discontinuance of cell division, but DNA replication is still performed (endoreplication), resulting in cysts with 4 nuclei (42). Thus, the exyzoite stage is prepared for direct and rapid cell division into four trophozoites (3). This could be one reason for the reported low infection doses, which can be as low as 1 to 10 cysts (36). The incubation time averages around 1 week, ranging from 3 to 20 days (31), and patients can excrete up to 10⁷ cysts per gram of stool (7, 35).

Infection occurs via the fecal-oral route, and besides direct transmission (41), the main source of infection is considered to be contaminated water (24, 26). Additionally, transmission from crops harvested from agriculture fields treated with sewage sludge has been reported (16). The zoonotic character of the disease is still under discussion (44), though many waterborne outbreaks have been associated with zoonotic transmission (22).

The prevalence of human Giardia infection in industrialized countries ranges from 2% to 5%, with a prevalence of 20% to 30% in developing countries, probably due to a less developed sanitation infrastructure in the latter (32).

Giardia cysts are ubiquitous in surface waters worldwide, their concentrations are reported to be in the range of 0.01 to 100 cysts per liter (50), and they survive for up to 2 months in water at 8°C (29). They are reported to be strongly resistant to disinfection, including chlorination, and difficult to remove by standard filtration (50). In contrast to the case for other waterborne pathogens, no tolerance values for drinking water are established (19), but based on known breakouts of giardiasis, action measures for levels of 3 to 5 cysts/100 liters are proposed (49). Thus, the requirement of detecting 1 cyst per 10 to 100 liters of water has been expressed (12), and sensitive and reliable methods are required.

Excystation and culture protocols have been developed; nevertheless, the organism cannot be cultured reliably from water samples (10, 37). Thus, classical microbiological approaches, such as plating, are not applicable for this organism.

Approved standard methods for Giardia and Cryptosporidium detection in water samples are described in the USEPA 1623 and AFNOR NF T 90-455 protocols. Both are very similar, and most existing methods are based on them (18). Generally, they rely on concentration, purification, and detection steps, and recoveries of <1% to 61% have been reported, depending on the method, seeding level, and sample turbidity (38). Recently, improved USEPA 1623-based methods showed recoveries from tap water of up to 75.4% (21).

Though improvements have been achieved in all respects compared to initially proposed methods, there is still a need for progress, as the methods are considered tedious, expensive, and only semiquantitative. A blind survey conducted in 16 different commercial laboratories performing analyses of spiked wound filter samples resulted in recoveries ranging from 0.8 to 22.3%, averaging 9.3%, which illustrates the above-mentioned drawbacks (6).

Many molecular identification methods based on PCR have been proposed, with detection limits ranging between 1 and 10⁵ cysts (data are given predominantly in cysts per liter but also in cysts per PCR), but difficulties due to small amounts of target DNA, the presence of inhibitor substances, and false-positive results were reported (4, 25, 28, 39).

Recently, different detection methods using flow cytometry (FCM) were proposed, employing different staining methods and resulting in variable recovery. They were usually accompanied by false-positive results (14, 20, 34, 47).

In this study, we developed a fast screening detection assay involving the filtration of cysts, immunomagnetic separation (IMS), and flow cytometric detection. The results presented here describe a rapid and robust method that can be applied to different water samples and has the prospect of being applicable to various matrices and to the automation potential of flow cytometry.

MATERIALS AND METHODS

Giardia lamblia cysts.

G. lamblia cysts were obtained from Waterborne Inc. (New Orleans, LA) and stored (10⁷ cysts) in 8 ml of sterile phosphate-buffered saline (PBS) (150 mM NaCl, 15 mM KH₂PO₄, 20 mM Na₂HPO₄, 27 mM KCl, pH 7.4; Sigma-Aldrich, St. Louis, MO) with 2.5% formalin (Sigma-Aldrich) at 4°C. The cysts were from an H-3 human isolate purified after passage through Mongolian gerbils.

Before dilutions were prepared, cyst samples were sonicated for 6 min in an ultrasonic bath (TUC-600; M. Scherrer AG, Wil, Switzerland) at 35 kHz and 600 W in order to dissociate the majority of the agglomerated cysts. The optimum sonication time was established by calibrating measurements and by evaluation of the results by FCM and immunofluorescence microscopy (IFM), resulting in around 90% single cysts and 10% agglomerated cysts. Agglomerates consisted mainly of 2 cysts and, rarely, 3 or more cysts sticking together.

Water samples.

One-liter tap water samples were collected from different taps in the laboratory building of the Swiss Federal Institute for Aquatic Science and Technology (Eawag). The system is fed by the drinking water supply of the town of Dübendorf, Switzerland.

Pond water was collected from a stagnant small pond of around 5 m in diameter in proximity to the laboratory building. River water samples were taken at the Chriesbach River, Dübendorf, Switzerland.

Wastewater samples were obtained from the inlet pipe of the experimental wastewater treatment plant of the Eawag, which has access to real wastewater from the community of Dübendorf.

All samples were processed directly or stored at 4°C for a maximum of 1 h.

Tap and surface water samples.

Water samples were collected in 1-liter screw-cap glass bottles and inoculated with 1 ml filtered (0.22-μm pore size) PBS containing a defined number of cysts, as indicated in the different experiments. The inoculated water samples were subsequently shaken vigorously. The water samples were prefiltered through a 30-μm nylon mesh filter (Millipore AG, Zug, Switzerland) before being vacuum filtered through a 5-μm-pore-size polycarbonate track etch filter (PCTE; Sterlitech Corporation, Kent, WA). The negative pressure was kept below 10³ Pa for tap water filtration, below 3 × 10³ Pa for river water filtration, and below 4 × 10³ Pa for pond water filtration. The different vacuum pressures were required due to the different turbidities of the water samples in order to avoid clogging of the filter (see below for more details).

The filter was carefully removed and inserted into a 50-ml centrifugation tube, followed by addition of 5 ml sterile PBS and 50 μl 1% bovine serum albumin (BSA; Fluka), strong vortexing, and addition of 1.5 μl fluorescein-labeled monoclonal antibodies (MAbs) (Giardi-a-Glo 20× solution; Waterborne Inc.). After incubation for 30 min at ambient temperature in the dark, 50 μl of superparamagnetic anti-fluorescein isothiocyanate (anti-FITC) MicroBeads (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) was added at 4°C. The assay mix was then incubated for 30 min at this temperature.

For 100-liter water samples, two inline filter cartridges were attached sequentially with rubber tubes to the tap, with the closer one to the tap comprising the 30-μm nylon mesh filter and the downstream one comprising the PCTE membrane. The flow rate was set to 2 liters per minute. Spiking was maintained by injecting the cysts through a syringe and a hollow needle through the rubber tube into the stream upstream of the filters. Injection was done after 10 min of a total filtration time of 50 min. Subsequently, the PCTE membrane was carefully removed and treated as described above.

Wastewater preparation.

Wastewater samples were prepared as described above for the surface water samples, except for the following modifications. Prior to the 30-μm nylon mesh filtration, the samples were filtered through a 1-cm-thick layer of glass wool (Sigma-Aldrich). The negative pressure was maintained below 6 × 10³ Pa, and the sample volume was 100 ml. Due to the very rapid clogging of the filter, it was impossible to process 1 liter of sample.

IMS.

Immunomagnetic separation using MACS MS columns (Miltenyi Biotec GmbH) was conducted at a defined speed by a peristaltic pump (Alitea U1-Midi-D pump; Bioengineering AG, Wald, Switzerland) at a flow rate of 2.05 ml/min. The sample volume of 5 ml was run through the column after placing the column in the separator magnet. Subsequently, the column was washed twice by rinsing with 3 ml of PBS, followed by two washes with 3 ml of PBST (PBS plus 0.01% Tween 20; Fluka) and finally with 3 ml PBS, conducted at a flow rate of 2.39 ml/min. The washing buffers were added to the column just before the columns ran dry according to the manufacturer's instructions. After removal of the column from the magnet, the purified cells were recovered by flushing 1 ml PBS through the column into a 15-ml Falcon tube by firmly applying the provided plunger.

Turbidity evaluation.

The turbidity of samples was measured with a Hach 2100 turbidimeter (Hach Company, Loveland, CO) following the instructions of the manufacturer. Results are given in nephelometric turbidity units (NTU).

Microscopic cyst enumeration.

Samples for microscopic cyst enumeration were stained as described above with 1.5 μl specific FITC-coupled surface antibody (Waterborne). Stained cysts were filtered onto a 13-mm, 0.2-μm-pore-size black PCTE filter (Isopore; Millipore), using a 3-ml syringe and a 13-mm filter holder cartridge. The filter was dried on a glass slide and a small drop of antifade mounting medium (Waterborne Inc.) was applied before a cover glass was placed over the filter. Microscopic enumeration was performed using an Olympus IX51 inverted microscope equipped with filters for FITC at a magnification of ×400. The provided software (Olympus DP-Soft, version 3.2) was applied to analyze 10 pictures taken with a defined image area of 1.34 mm², and the average for counted cells was extrapolated to the filter perimeter after subtracting the border area covered by the filter cartridge sealing, resulting in an effective filter area of 86.59 mm².

The average number of cysts was calculated according to the equation C_t = (ΣC_i × r²π)/(n × A_i), where C_t is the total number of cysts in the sample, C_i is the number of cysts counted within the image area, n is the number of images, r is the effectively exposed filter radius, and A_i is the image area.

FCM detection.

Flow cytometric detection was performed with a Partec CyFlow space flow cytometer (Partec GmbH, Muenster, Germany) equipped with a 200-mW blue solid-state laser emitting light at 488 nm. The volumetric counting properties of the flow cytometer measured the events in 200 μl of sample volume and extrapolated the result to 1 ml of sample volume. Optical filters were adjusted to measure green fluorescence at 520 nm, red fluorescence at 630 nm, and the side scatter (SSC) at 488 nm. The trigger was set on green fluorescence. Events were defined based on SSC, 520-nm and 630-nm fluorescence. Results were presented by plotting green fluorescence (520 nm) versus red fluorescence (630 nm) as well as versus SSC and by applying defined gating regions.

The R1 gate of the green (520 nm; y axis) versus red (630 nm; x axis) bivariate dot plot and the RN1 gate of the sideward scatter histogram (488 nm) were combined into an RN1-R1 multigate, thus counting only cysts that lay inside both restriction gates (Fig. 1). The specific instrumental gain settings used for green fluorescence, red fluorescence, and sideward scatter channels were 259, 345, and 191.1, respectively. The speed setting was 3, implying a counting rate of <500 events s⁻¹.

Fig. 1.

FCM detection of FITC-labeled Giardia lamblia cysts in 1 ml of PBS. (Top left) Histogram of green fluorescence (520 nm) count. (Top middle) Histogram of red fluorescence (630 nm) count. (Top right) Histogram of SSC count. (Bottom left) Bivariate dot plot of green fluorescence (520 nm) versus SSC. (Bottom right) Dot plot of 520 nm versus 630 nm. Events are defined based on SSC, 520 nm, and 630 nm. The events of gate R1 were restricted to the events inside gate RN1, establishing the multigate RN1-R1. SSC versus 520 nm and gate R2 are displayed because this distribution pattern and gating are often used in the literature as the only discrimination parameters.

Data treatment.

Giardia cyst enumeration is expressed as the mean for the experiments (n = 3) ± the standard deviation. The recovery or agreement quotas, expressed as percentages, represent the quotients of the means of the respective experiments (n = 3) ± the normalized root mean square deviations, taking into account the error propagation included in the procedure. Calculations and charts were established with Microsoft Excel 2003.

RESULTS

Spiking experiments. (i) Agreement between flow cytometric and immunofluorescence microscopic counting.

A suspension of G. lamblia cysts in sterile filtered PBS was stained with FITC-coupled surface antibodies and enumerated by microscopy and flow cytometry (Table 1). The labeled cysts could be identified clearly by the fluorescence microscope (see Fig. S1 in the supplemental material). An examination of the samples with conventional and immunofluorescence microscopy showed that all observed cysts were stained with FITC-coupled antibodies. Since microscopic enumeration is generally considered the accepted standard method, we defined this as the 100% recovery value. The agreement ratio was defined as the quotient between the FCM count and the IFM count, displayed as a percentage, i.e., X_a = (C̄_FCM/C̄_IFM) × 100, where X_a is the agreement ratio (%), C̄_FCM is the average FCM count, and C̄_IFM is the average IFM count. This agreement quota was established for three different assays, namely, for 1 ml spiked PBS, for 1 ml spiked PBS and subsequent immunomagnetic separation, and for the complete procedure for detection of cysts in tap water and of naturally occurring cysts in wastewater (Table 1).

Table 1.

Quantification of Giardia lamblia cysts in PBS and tap water by either flow cytometry or fluorescence microscopy, using different assay procedures

Matrix	Procedure	Mean cyst count ± SD		Xa (%)a
		FCM	IFM
1 ml spiked filtered PBS	Detection only	16,248 ± 163	16,782 ± 273	96.8 ± 1.9
5 ml spiked filtered PBS	IMS and detection	16,057 ± 202	16,413 ± 179	97.8 ± 1.6
1 liter spiked tap water	Filtration, resuspension, IMS, and detection	12,348 ± 824	12,159 ± 966	101.6 ± 10.5
100 ml wastewater	Filtration, resuspension, IMS, and detection	827 ± 25	773 ± 16	106.9 ± 3.96

^aDefined in Results. Values are means ± normalized root mean square deviations (n = 3 experiments).

For the first two assays, X_a was around 96 to 98%, with a low standard deviation of <2%. For spiking of 1 liter of tap water, the recovery with the complete method was still high (102%), but the standard deviation was much higher (Table 1). False-positive results were never encountered for the negative controls of these experiments. The slightly higher FCM/IFM recovery ratio for tap water might be due to particles in tap water that are not all eliminated during IMS and thus form a layer on the filter area, eventually covering some cysts and impeding microscopic enumeration. This effect was even more pronounced with wastewater samples (Table 1).

In conclusion, flow cytometric enumeration seems to be equivalent to enumeration by fluorescence microscopy, and thus all control measurements for spiking experiments were conducted by employing flow cytometry.

(ii) Quantification limit for detection of cysts in tap water.

We tested the method with spiking levels between 0 and 10,000 cysts in order to determine the detection and quantification limits of the method (Fig. 2).

Fig. 2.

Calibration of spiked Giardia lamblia cysts by applying the complete method. (a) Number of spiked cysts versus number of recovered cysts from 1 liter of tap water. The full line indicates the linearization function y = 0.9293x + 19.23 (r² = 0.9998). (b) Zoomed-in version of the 0- to 260-cyst count region of panel a. The full line indicates the linearization function y = 1.0275x − 4.71 (r² = 0.9984), limited to the data points for up to 260 Giardia cysts per liter. Areas of the estimated detection, determination, and quantification limit ranges are marked. Point A is the highest seeding level encountering a false-negative measurement. Point B is the lowest one not encountering a false-negative measurement.

The average recovery of cysts from 1 liter of tap water, with turbidities ranging from 0.2 to 0.4 NTU, was 92.9% ± 17.4% for all tested seeding levels. Spiking levels above 100 cysts per liter resulted in an average recovery of 97% ± 8.4% (see Table S1 in the supplemental material).

Employing our adapted column-washing protocol, the FCM dot plots showed low background signals and clear separation of the cyst cluster from the background for both spiked PBS and recovery from 1 liter of tap water (Fig. 3). Additionally, there were no false-positive results encountered for all of the 1-liter tap water negative controls (Fig. 3c).

Fig. 3.

FCM detection of spiked Giardia lamblia cysts in buffer and tap water. Bivariate FCM dot plots (520 nm versus SSC and 520 nm versus 630 nm) are shown. (a) One milliliter of PBS spiked with 200 Giardia cysts. (b) One liter of tap water spiked with 200 Giardia cysts, with the whole method applied (filtration/resuspension, IMS, and FCM). (c) One liter of tap water (negative control), with the whole method applied (filtration/resuspension, IMS, and FCM).

Different definitions for detection, determination, and quantification limits are used in validation documentation. In this case, the detection limit was defined as the spiking level at which 50% of the measurements were falsely negative, and the determination limit was defined as the spiking level at which reliable detection involved a high statistical security (>95%). The quantification limit corresponds to the seeding level at which a statistically safe quantification can be performed. In order to establish an estimated determination limit, we used two different approaches for validation purposes (43). In spiking 10 to 20 cysts, we encountered one false-negative result (Fig. 2b; see Table S1 in the supplemental material), so we considered the mean value of 10 cysts (Fig. 2b, data point A) the detection limit due to the lack of more data points, and this value was adopted for calculation purposes.

The first approach (46) for estimating the determination limit consisted of adding 3 times the standard deviation to the detection limit, resulting in around 40 cysts per liter. Analogously, the quantification limit was established by adding 10 times the standard deviation to the average detection limit, resulting in 110 cysts per liter.

The second approach (43) considered the determination limit to be the lowest value (Fig. 2b, data point B) lacking false-negative measurements where it was possible to statistically significantly distinguish between positive signals and negative control signals. For statistical significance, 2 times the standard deviation of this determination limit may not interfere with 2 times the standard deviation of the negative control or background noise. Based on this criterion, the determination limit was found to be 35 Giardia lamblia cysts per liter. In this approach, addition of 3 standard deviations to the determination limit leads to a quantification limit of 71 cysts per liter (43).

Combining both approaches, the estimated determination limit ranges from 35 to 40 cysts per liter, and the quantification limit ranges from 71 to 110 cysts per liter.

In our opinion, this conservative consideration may allow safe quantification for water samples. Of course, this should be confirmed experimentally for samples containing naturally occurring cysts, which may have lower binding affinities for the specific surface antibodies. Reduced binding may be due to either damaged or stressed cyst walls derived from either mechanical or chemical stress encountered in water supply networks (30) or from attachment to particles sterically hindering antibody binding.

(iii) Recovery from 100 liters of tap water.

Three trials were conducted with the inline filter setup. The average spiking level was 140 ± 30 cysts, resulting in an average recovery of 67% ± 5.13%. Figure 4 illustrates the increase in background signal for this large water volume. Fortunately, the cysts were still well distinguishable from background signals.

Fig. 4.

Detection of spiked Giardia lamblia cysts in a 100-liter tap water sample by FCM. Examples of bivariate FCM dot plots (520 nm versus SSC and 520 nm versus 630 nm) are shown for a 100-liter tap water sample spiked with 150 Giardia cysts, with the whole method applied (filtration/resuspension, IMS, and FCM). An average of 140 ± 30 cysts were spiked, resulting in an average recovery of 67% ± 5.13%.

(iv) Recovery from spiked pond and river water.

Spiking experiments with samples of different water quality were conducted in order to test the robustness of the proposed method for use with different matrices. Seeding levels for pond and river water samples were in the range of 800 ± 100 cysts per liter, representing 20 times the determination limit for validation purposes according to the instructions of the Swiss Accreditation Service (43).

The recovery from 1 liter of river water with a turbidity of 5 to 6.5 NTU was 94.9% ± 9.7%. Similar results were obtained for pond water (see Table S2 in the supplemental material). As expected, the background of such samples was much increased compared to that for tap water. No false-positive results were observed with the negative controls (Fig. 5).

Fig. 5.

FCM detection of Giardia lamblia cysts spiked into 1-liter samples of different surface waters. Examples of positive and negative samples for application of the complete method (filtration/resuspension, IMS, and FCM) are shown. Bivariate FCM dot plots (520 nm versus SSC and 520 nm versus 630 nm) are shown. (a) Pond water spiked with 800 cysts. (b) River water spiked with 800 cysts. (c) River water negative control. (d) Pond water negative control.

Although the turbidity of the pond water was lower than that of river water, the filtration process for the pond water was much more difficult and time-consuming. A higher filtration pressure was applied to the pond water samples (∼−5 × 10³ Pa) because the filter clogged more rapidly. This higher pressure may have resulted in a lower recovery and a higher standard deviation for the pond water samples (see Table S2 in the supplemental material).

Application of the method to naturally contaminated wastewater samples.

In order to test the method for detection of naturally occurring Giardia cysts and to determine its applicability to samples with higher turbidities, raw wastewater was analyzed by flow cytometry and fluorescence microscopy. The samples had turbidities ranging from 250 to 300 NTU. The cyst concentrations were 826.7 ± 25.2 and 773.3 ± 16.3 cysts per 100 ml of sample by FCM and IFM, respectively, resulting in an X_a value of 106.9% ± 4% (Table 1). As already observed for tap water, the number of detected cysts was higher when FCM was used. This supports the hypothesis that particles cover and hide cysts on the filter, leading to reduced visible fluorescence and thus to false-negative results. When we spiked these samples, recovery of 84% of the spiked cysts was achieved (Fig. 6).

Fig. 6.

Comparison of method recoveries for spiking with Giardia lamblia cysts, depending on the matrix.

DISCUSSION

Today's standard immunofluorescence detection methods based on the USEPA 1623 method require skilled personnel and are thus influenced by associated factors, for example, viewer fatigue (17). They are considered to be tedious, expensive, and only semiquantitative and often provide low and variable recoveries (6), usually ranging from 0.5% to 53% for environmental water samples (11).

Molecular approaches based on PCR can be highly sensitive and specific down to a single cyst per sample if appropriate concentration and purification are achieved. Nevertheless, detection is often hampered by PCR-inhibiting substances, and results are often semiquantitative (40). Additionally, PCR is likely to be affected by nonspecific DNA contaminations or carryovers from previously amplified DNA leading to false-positive results (25, 39).

Flow cytometry with fluorescence-activated cell sorting (FACS) is used increasingly as an alternative method for separation and enumeration, but difficulties are encountered when autofluorescent algae are present and cross-reactions of the MAbs lead to unspecific binding (2, 8, 48). Furthermore, expensive and difficult-to-operate cell sorters are required, which inhibits wide application. Some of these problems were addressed by Ferrari and Veal (14), who developed a two-color detection-only assay for Giardia cysts leading to no false-positive results but also having a moderate recovery of 39% for a seeding level of 90 ± 28 cysts in 10 liters of backwash water.

The method presented here consists of a prefiltration step in order to eliminate large particles and contaminants. One-liter water samples are filtered through a 5-μm membrane. The cysts are resuspended in 5 ml PBS and labeled with a specific FITC-coupled surface antibody. Subsequently, the Giardia cysts are separated immunomagnetically with commercial anti-FITC MACS MicroBeads with a diameter of 50 nm, resulting in a 1-ml sample analyzed by detection-only flow cytometry. This entire process requires only 100 min. Due to their small size and lack of autofluorescence, the beads do not need to be removed from the microorganisms after IMS and can remain in the sample for enumeration. The MicroBeads do not create problems that have been reported previously (9, 33), such as cyst detachment from large beads due to shear forces or interference with immunofluorescence detection. Thus, it is unnecessary to dissociate beads and cysts before detection, leading to a simple protocol. The short incubation time, increased recovery, and simple handling of the MACS technology were reported previously for Cryptosporidium detection (9).

Surprisingly, in contrast to the case for other detection methods relying on fluorescent dyes and flow cytometry (13), we did not encounter any false-positive results among the nonspiked samples, and furthermore, potential cross-reactions of the antibodies did not obviously impair the demonstrated robustness of the method. Furthermore, we achieved recoveries exceeding 90% for all drinking water samples when we seeded 75 cysts or more; a similar performance of our method was observed with environmental water samples, with time requirements of less than 2 h. As shown in the comparison with IMF analysis, flow cytometric enumeration seems to be less prone to signal hiding, because during sample analysis, cysts and particles are separated and not deposited together on a membrane surface.

Because all commercially available antibodies target the same epitopes on Giardia sp. cysts (30), noncompetitive surface staining with two different dyes is not yet possible. According to our results, this might not be necessary in order to obtain reliable results when monitoring for Giardia sp. contamination in drinking, surface, and wastewater.

The determination limit was estimated to be around 40 cysts per liter for drinking water, and thus the goal of detecting 1 cyst per 10 to 100 liters of water (50) is not reached by the proposed method. Nevertheless, adaptation of the filtration and resuspension process or an alternative preconcentration method might enhance the performance. The first assays filtering 100 liters of tap water indicate that the method could be capable of robustly detecting Giardia cysts at 100-times-lower concentrations.

Due to the 200-μl volumetric sampling port of the FCM used in this study, the probability of deviation was much higher, as the flow cytometer actually analyzes 200 μl of the 1-ml sample. This increased statistical deviation adds to the loss of cysts during the process, resulting in the observed deviations. This becomes especially significant for seeding levels below 100 cysts. It can be speculated that the use of an FCM equipped with a larger volumetric sampling port or a nonrestricted sampling port, thus lowering the statistical deviations compared to those with the cytometer used here, could also result in a lower determination limit. Combining this with the above-mentioned 100-liter sample preconcentration, the detection of 0.1 cyst per liter might be achieved, thus meeting the previously mentioned detection requirements. Nevertheless, this would be, in agreement with all available methods, still far too low to meet one theoretical maximum acceptable concentration of 1.7 × 10⁻² cyst/1,000 liters (which would require the processing of at least 60,000 liters of water to detect a single cyst [19]).

The drawbacks of the proposed method relative to PCR detection might be overcome by employing peptide nucleic acid (PNA) probes, as proposed by Ferrari and Veal (14), for species-specific detection. Combining this method with other staining methods in order to distinguish physiological states, such as potential infectivity (e.g., propidium iodide live-dead staining), would be an important improvement, and earlier reports indicated a good correlation with in vitro excystation protocols for Cryptosporidium parvum (5).

Considering the application of our method to complex matrices, the wastewater experiment can be considered successful. Naturally occurring cysts were detected, despite the fact that detection can be difficult due to environmental influences on the cyst wall (30). To date, we have not been able to elucidate the impact of these effects, as both detection methods rely on the same principle, i.e., antibody binding to the cell surface.

In summary, this method has shown reliable results for various matrices and has the potential to find wide applications, which are not limited to use with a flow cytometer only, as a combination with PCR or IFM detection is also possible. The high recoveries (Fig. 6) for different matrices and the low dependence on highly skilled laboratory personnel are additional benefits. The convenient handling and about 100-min time requirement appear very promising, and the costs per measurement range at the moment from CHF 25 to 32 (US$28 to 36 or EUR 20 to 25, based on exchange rates as of May 2011) for consumables only. Finally, the applicability to various matrices and the potential to adapt the method to the detection of virtually every microorganism show the opportunities related to this method.