Enhancing the Detection of Giardia duodenalis Cysts in Foods by Inertial Microfluidic Separation

Abstract

The sensitivity and specificity of current Giardia cyst detection methods for foods are largely determined by the effectiveness of the elution, separation, and concentration methods used. The aim of these methods is to produce a final suspension with an adequate concentration of Giardia cysts for detection and a low concentration of interfering food debris. In the present study, a microfluidic device, which makes use of inertial separation, was designed and fabricated for the separation of Giardia cysts. A cyclical pumping platform and protocol was developed to concentrate 10-ml suspensions down to less than 1 ml. Tests involving Giardia duodenalis cysts and 1.90-μm microbeads in pure suspensions demonstrated the specificity of the microfluidic chip for cysts over smaller nonspecific particles. As the suspension cycled through the chip, a large number of beads were removed (70%) and the majority of the cysts were concentrated (82%). Subsequently, the microfluidic inertial separation chip was integrated into a method for the detection of G. duodenalis cysts from lettuce samples. The method greatly reduced the concentration of background debris in the final suspensions (10-fold reduction) in comparison to that obtained by a conventional method. The method also recovered an average of 68.4% of cysts from 25-g lettuce samples and had a limit of detection (LOD) of 38 cysts. While the recovery of cysts by inertial separation was slightly lower, and the LOD slightly higher, than with the conventional method, the sample analysis time was greatly reduced, as there were far fewer background food particles interfering with the detection of cysts by immunofluorescence microscopy.

INTRODUCTION

Giardia duodenalis is an enteric protozoan parasite which infects a wide range of hosts, including humans and a variety of domestic and wild mammals. It is the most commonly identified intestinal parasite worldwide (1), with an estimated 2.8 × 10⁸ human cases of giardiasis annually (2). G. duodenalis prevalence is much higher in children than adults and in developing than developed countries (3). Transmission of giardiasis involves the ingestion of the environmentally robust cyst life stage, either by direct contact or indirectly through cyst-contaminated water or food. It is currently estimated that 7% of giardiasis cases in the United States are foodborne (4). Foods that have previously been reported to be contaminated with G. duodenalis cysts include fresh fruits and vegetables, shellfish, and, more recently, meats (5). Fresh fruits and vegetables contaminated with G. duodenalis cysts are of particular public health concern, as they are generally consumed raw and often originate in developing countries with lower standards of water quality and hygiene. An overview by Robertson (6) of the foodborne outbreaks which occurred between the years 1984 and 2012 identified a total of nine giardiasis outbreaks worldwide. A summary of the reported incidence of Giardia spp. in fresh produce was tabulated by Dixon (7) and included 38 studies, with prevalences ranging from 0.9 to 83.3%.

Each food matrix provides its own unique challenges during testing; thus, there is currently no standard methodology for the isolation of G. duodenalis from foods. Available methods generally involve the elution of G. duodenalis cysts from the surface of the food, resulting in a parasite suspension with a low concentration of parasites and relatively high concentration of interfering food particles. The concentration of parasites from this suspension is usually performed by centrifugation. However, all the food particles present in the suspension are also concentrated by this method. Initial filtration steps can be performed to remove the larger food particles; however, as the cysts range from 6 to 10 μm (8), many of the smaller food particles cannot be removed without the loss of parasites as well.

The “gold standard” for the detection of G. duodenalis cysts is immunofluorescence microscopy, in which commercially available fluorescently labeled monoclonal antibodies specifically targeting the cyst wall are used. Flow cytometry can also be used, but morphological confirmations of cysts cannot be made. PCR is also often used for detection; however, DNA extraction from the robust cyst may be difficult, and PCR inhibitors in foods may reduce the sensitivity further.

Samples with higher numbers of food particles make identification and enumeration of protozoan parasites difficult and contribute to lower recoveries (9, 10). Food particles are capable of inhibiting the formation of the antigen-antibody complex necessary in immunofluorescence detection methods, by obstructing or chemically altering the epitope region where binding occurs (9, 11). These food particles may also nonspecifically bind the fluorescently labeled antibodies, which may lead to the detection of false positives during immunofluorescence microscopy or flow cytometry as described above. Larger food particles may also hide or obstruct cysts during microscopic examination, preventing their detection and identification. Finally, a number of PCR inhibitors, such as polysaccharides, pectin, polyphenols, phenols, glycogen, and xylans, have been found in foods that are known to be contaminated with G. duodenalis, including lettuce, berries, and shellfish (12, 13). Thus, developing better concentration and purification methods that reduce the number of food particles in the final parasite preparation, while obtaining a high recovery of parasites, is key to improving the overall detection of cysts on foods.

Inertial microfluidic separation is rapidly emerging as an efficient separation, filtration, and concentration technology in biologics, due to its low cost, high throughput, and lab-on-a-chip potential. It has recently found application in the separation of nucleated cells (14), cancerous cells (15), and stem cells (16) from whole blood. It has also been used in the separation of Escherichia coli from erythrocytes (17). Inertial microfluidic separation makes use of the hydrodynamic forces which act on particles within a fluid as it flows through a channel.

Experimental studies of flows in circular pipes have shown that randomly dispersed, suspended particles in laminar flow are able to cross streamlines and focus into a narrow annulus of 0.6 R, where R is the radius of the pipe (18). The particles are driven to migrate laterally across the streamlines due to two inertial forces, the shear-induced lift force and the wall-induced lift force. The shear-induced lift force pushes the particles toward the channel walls. This force arises from the velocity profile of the fluid as it flows (laminar flow) through a channel (pipe). The wall-induced lift force counteracts the shear-induced lift force and pushes the particles back toward the center of the channel. The equilibrium between these forces is reached somewhere between the center of the channel and the wall surface, the actual position being dependent on the channel geometry and the Reynolds number. Particles that are at least 7% the size of the channel have been shown to be affected by these inertial forces and subsequently ordered in equilibrium positions close to the channel walls (19). For rectangular channels of high aspect ratio, these focusing equilibrium positions are located at about 20% of the channel width away from the channel's vertical walls. Focused particles can be extracted at the end of the separation channel with lateral ports of an outlet trifurcation. The central port collects a portion of the liquid sample that is depleted of the focused particles.

By manipulating the dimensions of a microchannel, as well as the flow rate, particles of a desired size can be focused within the channel and separated from undesired smaller particles. Thus, particles such as G. duodenalis cysts could theoretically be targeted by designing and fabricating microfluidic chips specific for particles ranging from 6 to 10 μm. By designing a microfluidic chip that captures the particles at the specific equilibrium positions, it is possible to separate the desired particles from the undesired particles. It seemed feasible that this technology could be applied to passively separate cysts out of a suspension containing food particles without the need for specific antibodies, filters, or centrifugation and that it could be done in a high-throughput manner allowing for rapid sample processing.

The objective of this study, therefore, was to develop, integrate, and validate a high-throughput, nonclogging microfluidics-based platform for the rapid isolation of G. duodenalis cysts from foods. It is anticipated that this novel method for processing food samples will improve the speed, sensitivity, and specificity of detection over existing methods by simultaneously concentrating cysts while removing interfering particulate matter.

In this paper we outline the design and fabrication of the thermoplastic-based multilayer G. duodenalis inertial microfluidic chip, which was developed along with an optimized pumping protocol. We also demonstrate the specificity of the chip for G. duodenalis cysts in buffered suspensions, along with its efficiency in separating out nonspecific microparticles through cyclical pumping. Finally, we demonstrate the integration of the microfluidic chip into a method for the elution, concentration, and detection of G. duodenalis cysts artificially spiked onto lettuce leaves. We also compare the recovery of cysts, limit of detection, and efficiency in the removal of background food particles by this method to those of a conventional filtration-centrifugation method.

MATERIALS AND METHODS

Parasite isolates.

Suspensions of G. duodenalis cysts were purchased from Waterborne, Inc. (New Orleans, LA). They were human isolate H-3 G. duodenalis cysts and were passaged through gerbils, purified from fecal matter, and suspended in phosphate-buffered saline (PBS) with antibiotics. All parasite suspensions were stored at 4°C until required, up to a maximum storage time of 60 days.

Fluorescent microbeads.

Dragon green-dyed polymer microspheres with a diameter of 1.90 ± 0.22 μm were obtained from Bangs Laboratories, Inc. (Fishers, IN). Beads were diluted into working solutions of distilled water with 0.1% Tween 20 and enumerated with a hemocytometer.

Microfluidic chip fabrication.

The microfluidic structures of the chip were fabricated by the hot embossing of thermoplastic polymers using epoxy replication molds. The microchannel structure was initially transferred by standard photolithography to a Su-8 photoresist deposited on Si wafers that were used as master molds. The Su-8 structures were transferred to a hard epoxy mold (Conapoxy, CYTEX, USA) via an intermediate polydimethylsiloxane (PDMS) replica (mold inversion). These PDMS molds were obtained by mixing PDMS (Sylgard 184; Dow Corning, USA) and a curing agent at a 10:1 ratio and then degasing the mixture in a vacuum chamber for 30 min. The mixture obtained was poured onto the Su-8 mold and cured at 80°C for 2 h. Similarly, the final epoxy mold was obtained by pouring epoxy resin mixed with curing agent (10:8 ratio) onto the intermediate PDMS mold and curing it at 120°C for 6 h. The epoxy molds were then used to emboss microfluidic features in thermoplastic elastomers (TPE) initially extruded from pellets into continuous films with a 1.5-mm thickness. This embossing step was done with an EVG520 instrument (EV Group, Austria) operating at 135°C and 6 kN of applied force for 5 min. The final chips were sealed by using plastic injected covers with integrated luer lock ports. The injection machine used to fabricate these covers was an E110 compression injection molding system (Engel, USA). Due to its strong adhesion to the thermoplastic elastomeric films, a cyclo olefin polymer (Zeonor; Zeon Corporation, Japan) was used as material for the luer lock covers in order to provide proper seal of the microfluidic structure as well as simple connectivity to the external syringe pumping system (Harvard Apparatus, USA). The fluidic connection between the microfluidic device and the pumping system was done with laboratory tubing (Dow Corning, USA) and standard luer lock connectors (Qosina, USA). Advanced handling and sample recirculation through the chip was also done by using 3-way (T) luer lock check valves (Qosina).

Device characterization.

The microfluidic chips consisted of an inlet (I) port leading to the main separation channel (SC), which consists of a rectangular microfluidic channel. The separation channel enlarges into the large channel (LC), which splits into three smaller channels connected directly to three output ports (C, E1, and E2) (Fig. 1a and b). In the LC, flow is slowed down to allow enhanced visualization of fluorescent particles. The SC is 70 ± 2 μm wide and 180 ± 5 μm deep in the G. duodenalis inertial separation chip, providing an aspect ratio of about 2.6. The tolerances in these dimensions originate in a certain degree of imprecision that occurs in embossing replicas due to the thermal expansion of the TPE itself, as well as to the mechanical stresses that are induced at the assembling step. The LC connecting the separation channel to the outlet trifurcation outlets is about 300 μm wide and is the same depth as the SC. The SC and the LC have lengths of 42 mm and 15 mm, respectively. All inlet and outlet collectors are 500 μm deep, which is much deeper than the SC or observation window (Fig. 1c) to avoid collapse of the relatively large TPE structures at the inlet and outlet ports.

FIG 1.

Schematic representation and characterization of the microfluidic chip. (a) The inlet (I) port allows entry into the separation channel (SC), which then widens into a large channel (LC) that splits into three outlet collectors (C, E1, and E2). (b) Scanning electron microscopy images of the transition region from the separation channel to the observation window as well as the outlet trifurcation. (c) Schematic representation of a vertical section along the separation channel showing the Zeonor-TPE chip assembly with the luer lock (LL) ports, separation channel, and input and output collectors. The regions between the Zeonor cover and the TPE film qualitatively illustrate the depth of the embossed features.

Inertial separation chip setup and procedure.

An example of a microfluidic setup for the concentration of G. duodenalis cysts consisting of two syringe pumps, two reservoirs (waste and sample), a microfluidic concentration unit, and four T check valves is shown in Fig. 2. The samples were initially loaded into a syringe (plastic pack, 10 ml; Becton Dickinson, USA) and mounted on a syringe pump (pump 1). The syringe is connected with a T valve to the chip inlet port and to the sample/recovery reservoir. Similarly, three additional syringes are connected with T valves to the chip outlet ports and to the two reservoirs as indicated in Fig. 2. The T valves are two-way valves which open on top (d1) when pushed by the syringe and open on the side (d2) when pulled by the syringe. The pumps are programmed to recycle the sample liquid through the microfluidic unit until the desired final concentration (target volume) is reached by using a specific protocol, such as the one for G. duodenalis in Table 1. As the main separation channel splits into three smaller channels of equal size, one-third of the total volume being cycled through the chip is pumped into the waste reservoir, while two-thirds of the total volume is cycled in to the recovered reservoir, where it can be recycled through the system once more. As the components of the pumping system, including the chip, occupy roughly 1 ml of volume, prior to loading the sample, 1 ml of double-distilled water (ddH₂O) was loaded into the syringe on pump 1 and pumped into the system in order to prime the chip with fluid and prevent the circulation of air in the system. The total run time for the six-cycle concentration protocol represented in Table 1 is 28 min 17 s (see Fig. S1 in the supplemental material). Any liquid remaining in the chip at the end of the run was withdrawn at 0.4 ml/min using pump 2 and pumped into the respective reservoirs. The suspensions in both the waste and sample/recovery reservoirs at the end of the run were collected for analysis.

FIG 2.

Schematic representation (a) and actual image (b) of the fluidic setup used for the concentration of parasites with the microfluidic inertial separation chip (MC). The directional arrow at the top of the diagram represents the direction of flow through the chip. The T check valve on the inlet (I) syringe is open at d2 when the sample is pulled into the syringe. During the pumping of the solution out of the syringe and into the chip, d2 is closed and d1 is open.

TABLE 1.

Giardia duodenalis cyst pumping protocol for the concentration of a 10-ml sample with the Giardia microfluidic inertial separation chip

Cycle no.	Pump 1 program	Pump 2 program
1	Infuse 10.0 ml at 1.2 ml/min.	Withdraw 3.333 ml at 0.4 ml/min.
	Wait for 1.0 min.	Infuse 3.333 ml at 3.333 ml/min.
	Withdraw 6.667 ml at 6.667 ml/min.	Wait for 1.0 min.
2	Infuse 6.667 ml at 1.2 ml/min.	Withdraw 2.222 ml at 0.4 ml/min.
	Wait for 0.667 min.	Infuse 2.222 ml at 3.333 ml/min.
	Withdraw 4.444 ml at 6.667 ml/min.	Wait for 0.667 min.
3	Infuse 4.444 ml at 1.2 ml/min.	Withdraw 1.481 ml at 0.4 ml/min.
	Wait for 0.444 min.	Infuse 1.481 ml at 3.333 ml/min.
	Withdraw 2.963 ml at 6.667 ml/min.	Wait for 0.444 min.
4	Infuse 2.963 ml at 1.2 ml/min.	Withdraw 0.988 ml at 0.4 ml/min.
	Wait for 0.296 min.	Infuse 0.988 ml at 3.333 ml/min.
	Withdraw 1.975 ml at 6.667 ml/min.	Wait for 0.296 min.
5	Infuse 1.975 ml at 1.2 ml/min.	Withdraw 0.658 ml at 0.4 ml/min.
	Wait for 0.198 min.	Infuse 0.658 ml at 3.333 ml/min.
	Withdraw 1.317 ml at 6.667 ml/min.	Wait for 0.198 min.
6	Infuse 1.317 ml at 1.2 ml/min.	Withdraw 0.439 ml at 0.4 ml/min.
		Infuse 0.439 ml at 3.333 ml/min.

Efficiency of the inertial separation chip in concentrating G. duodenalis cysts.

Approximately 1,000 G. duodenalis cysts were isolated by serial dilution of the stock 1.25 × 10⁶ cyst per ml suspension in 1% PBS. The cysts were then added to 10 ml of elution buffer containing 1% PBS and 0.01% Tween 80, pH 7.4. The 10-ml sample was vortexed thoroughly, loaded into a 10-ml syringe, and cycled through the inertial microfluidic separation chip as described in Table 1. The final suspensions pumped into the recovered reservoir (approximately 1.5 ml) and waste reservoir (approximately 9.5 ml), at the end of the run, were centrifuged at 3,724 × g for 15 min. The supernatant was removed, leaving each sample with a final suspension of approximately 500 μl. A new inertial microfluidic separation chip and setup, including syringes, check valves, and tubing, were used for each sample. The efficiency of the inertial separation chip in concentrating cysts from buffer was determined by enumerating a portion of the initial inoculum, the final concentrate suspension, and the waste suspension. The enumeration was performed by epifluorescence microscopy as described below.

Efficiency of the inertial separation chip in eliminating fluorescent beads.

A 2.653 × 10⁹/ml stock solution of microbeads was serially diluted down to approximately 6 × 10³ in 1% PBS. These 1.90-μm fluorescent beads along with 2,000 G. duodenalis cysts were spiked into 10 ml of elution buffer containing 1% PBS and 0.01% Tween 80, pH 7.4. The 10-ml sample was vortexed thoroughly, loaded into a 10-ml syringe, and cycled through the Giardia inertial separation chip for either one, three, or six cycles. The final concentrated parasite suspension of approximately 1.5 ml and the approximately 9.5 ml of waste solution were centrifuged at 3,724 × g for 15 min. The supernatant was removed, leaving each sample with a final suspension of approximately 500 μl. The efficiency of the inertial separation chip in separating out nonspecific microparticles and concentrating cysts over a range of cycles was determined by enumerating a portion of the final concentrate suspension and the waste suspension. The enumeration was performed by epifluorescence microscopy as described below.

Enumeration by epifluorescence microscopy.

A 100-μl portion of the final 500-μl sample was incubated with Crypto/Giardia Cel reagent fluorescein isothiocyanate (FITC)-labeled monoclonal antibodies (Cellabs, Brookvale, Australia) for 1 h at room temperature. The suspension was rinsed of any unbound antibodies by dilution in 1 ml of PBS. The sample was centrifuged once more and the supernatant was aspirated off, leaving the labeled parasites in 100 μl of solution. Twenty-microliter aliquots of the resuspended suspension were pipetted onto microscope slides (Fisher Scientific, Pittsburgh, PA). The cysts and/or fluorescent microbeads observed on each of three slides were enumerated on a Nikon 80i epifluorescence microscope (Nikon Canada, Inc., Mississauga, ON, Canada) at a 200× magnification, using a blue filter with excitation at 450 to 490 nm. The total numbers of cysts and/or fluorescent microbeads on the three slides were added and multiplied by the appropriate factor to obtain the total number per sample.

Preparation of artificially contaminated food samples.

Prepackaged salad kits were purchased at retail. The contents of these kits were prechopped and prewashed by the manufacturer. Salad kits were stored at 4°C and utilized prior to their expiration dates. Twenty-five (±0.5)-gram samples of romaine lettuce, taken from the salad kits, were weighed out in stomacher bags (Seward, Worthing, United Kingdom). Purchased G. duodenalis suspensions were diluted in ddH₂O to obtain an initial suspension with a concentration of approximately 300 cysts per 750 μl. Six 2-fold serial dilutions of the 300-cyst parasite suspension were performed. Each of the six dilutions was used to inoculate duplicate lettuce samples. Each lettuce sample was spiked with 750 μl of the parasite suspension, added dropwise to the surface of the lettuce leaves in each stomacher bag. Two additional lettuce samples were inoculated with 750 μl of ddH₂O and used as negative controls. Samples were air dried at room temperature for 2 h and then refrigerated at 4°C overnight prior to analysis.

Inertial separation method for the elution and concentration of cysts from lettuce.

Twenty-five-gram lettuce samples were submerged in 200 ml of elution buffer containing 1% PBS and 0.01% Tween 80, pH 7.4, in a stomacher bag. Samples were agitated on an orbital shaker at 120 rpm for 15 min. The contents of the bag were then vacuum filtered through a sheet of polyester monofilament with a 10-μm pore size (IFC Fabrics, Minneapolis, MN) to remove any large particles. The filtered suspension was then centrifuged at 2,000 × g for 15 min at 4°C. A total of 190 ml of supernatant was removed, leaving a final concentrated suspension of 10 ml. The 10-ml suspension was vortexed thoroughly, loaded into a 10-ml syringe, and cycled through an inertial microfluidic separation chip for six cycles as described in Table 1. A new inertial microfluidic separation chip and setup were used for each sample. The final collected suspension of approximately 1.5 ml was centrifuged at 1,500 × g for 10 min, and the supernatant was removed, leaving a final suspension of approximately 500 μl.

Conventional method for the elution and concentration of cysts from lettuce.

G. duodenalis cysts were eluted from spiked lettuce and initially concentrated according to the method of Dixon et al. (20), which is specific for the detection of Cyclospora, Cryptosporidium, and Giardia from ready-to-eat packaged leafy greens. In brief, the method involved weighing 25-g samples into 1-liter stomacher bags and adding 200 ml of elution buffer containing 1% PBS and 0.01% Tween 80, pH 7.4. The bags were next placed on an orbital shaker for 15 min at 120 rpm. The elution buffer was poured into centrifuge tubes through four layers of gauze within a funnel to trap larger particles. Particles in the elution buffer were concentrated by a series of centrifugation steps at 2,000 × g for 15 min at 4°C, followed by removal of the supernatant. The remaining suspension containing the pellet of cysts and background particles was transferred to a 1.5-ml microcentrifuge tube and brought to a final volume of 500 μl with elution buffer.

LOD and percent recovery of G. duodenalis cysts.

The limit of detection (LOD) and percent recovery of cysts were assessed on lettuce samples artificially contaminated with G. duodenalis cysts (described above) for both the inertial separation method and a conventional method used by Dixon et al. (20), considered the gold standard. Of the 14 total samples in each trial, half of the samples were eluted and concentrated using the inertial separation method, while the second half were eluted and concentrated in parallel using the conventional method of Dixon et al. (20). Each sample was screened for the presence of G. duodenalis cysts by epifluorescence microscopy as described above. Three slides were examined, for a total volume of 60 μl. The percent recovery was determined by enumerating a portion of the initial inoculum and the concentrate recovered from the samples spiked with the highest inoculum as described above. The LOD and percent recovery were determined by three replicate experiments.

Particle enumeration of final concentrate by flow cytometry.

Fifty-microliter portions of the remaining recovered parasite suspensions from the 25-g lettuce samples inoculated with G. duodenalis were analyzed by flow cytometry in order to estimate the efficiency of the inertial separation method in removing background lettuce and food particles. Sample suspensions were stained with Crypto/Giardia Cel reagent, diluted in 200 μl of PBS, and 10 μl of Absolute Count Standard Beads (Bangs Laboratories, Fisher, IN) was added. The Absolute Count Standard Beads were used to ensure that equal portions of each sample were analyzed. Sample analyses were performed using a Becton Dickinson FACSCalibur equipped with a 488-nm air-cooled laser (BD Biosciences, Mississauga, ON, Canada). The total number of background particles (events) per sample was determined by subtracting the number of G. duodenalis cyst events and bead events detected from the total number of events detected by the cytometer. The average proportion of background particles in the final concentrated parasite suspension for both the inertial microfluidic method and the conventional method were compared and used to estimate the purity of the parasite suspensions.

Statistical analysis.

Comparative analysis was performed using the Mann-Whitney nonparametric test to determine if there was a significant difference among the cyst percent recoveries. A level of significance of 5% (α = 0.05) was used to test for statistical differences. Analysis was done using GraphPad Prism 5, version 5.03 (GraphPad Software, Inc.).

RESULTS

Specificity and concentration efficiency of the inertial separation chips for cysts suspended in buffer.

The specificity of the Giardia inertial separation chip for G. duodenalis cysts suspended in 10 ml of 1% PBS and 0.01% Tween 80 elution buffer and the efficiency of concentration were initially assessed to determine the ability of the chip to focus cysts from a relatively particle-free cyst suspension. A 20-fold concentration was achieved, with 72.44% ± 1.77% (n = 3) of G. duodenalis cysts recovered (Fig. 3). An average of 16.49% ± 5.40% (n = 3) of G. duodenalis cysts were not focused toward the outside of the channel and were thus cycled out of the chip into the waste reservoir. There was also an average of 11.07% ± 7.14% (n = 3) of G. duodenalis cysts that were lost to the components of the system (i.e., channels, valves, tubing, or syringes).

FIG 3.

Giardia duodenalis cysts recovered from 10 ml of spiked elution buffer following concentration with the inertial separation chip. A portion of the spike, recovered concentrate, and waste suspension cycled out of the chip was stained with Crypto/Giardia Cel reagent FITC-labeled monoclonal antibodies and excited at 450 to 490 nm for enumeration by immunofluorescence microscopy. The cysts that were not found in the recovered or waste suspensions were deemed as experimental loss. Error bars represent SDs from three separate experiments.

Efficiency of the inertial separation chip in eliminating nonspecific microparticles.

To (i) further investigate the G. duodenalis cysts that were being lost in the waste reservoir, (ii) determine the efficiency of the chips in cycling out nonspecific microparticles, and (iii) ultimately determine an optimal cycling number for use with food samples, 10-ml buffer suspensions containing G. duodenalis cysts and nonspecific microparticles (beads) with a mean diameter of 1.90 μm were pumped through their respective microfluidic inertial separation chips for one-, three-, and six-cycle protocols. The small bead particles were similar in size to some lettuce and food particles, which may be inhibitory to detection methods and were not expected to focus toward the channel walls and were thus expected to be separated out into the waste reservoir during the cyclical concentration process.

Suspensions from the recovered and waste reservoirs were collected and enumerated following one, three, and six cycles of focusing. The proportions of cysts in each of the reservoirs were compared. Following one cycle, 92.14% ± 3.93% (n = 2) of cysts were found in the recovered reservoir, 87.47% ± 5.53% (n = 2) following three cycles, and 81.93% ± 7.87% (n = 2) following six cycles (Fig. 4). Thus, initially during the first cycle, there was a greater rate of loss of cysts, with 7.86% ± 3.93% being found in the waste reservoir. This may be attributed to unfocused cysts that were outliers in terms of size or to cysts clumping together and becoming larger particles. Over one to six cycles, it appeared that there was a constant rate of loss of 2.04% cysts per cycle, which is much lower than the loss observed over the first cycle.

FIG 4.

The proportion of Giardia duodenalis cysts and green-dyed polymer microbeads in the recovered concentrate following one, three, or six cycles of inertial separation. Portions of the recovered and waste suspensions were stained with Crypto/Giardia Cel reagent FITC-labeled monoclonal antibodies and excited at 450 to 490 nm for enumeration by immunofluorescence microscopy. Error bars represent SDs from two separate experiments.

With respect to the fluorescent beads, following a one-cycle Giardia inertial microfluidic separation run, 54.96% ± 14.88% (n = 2) of the beads were found in the recovered reservoir, 45.11% ± 10.69% (n = 2) following three cycles, and 29.52% ± 3.17% (n = 2) following six cycles (Fig. 4). Thus, as expected, cycling the recovered suspension through the chip additional times helped to eliminate more nonspecific microparticles from the recovered parasite suspension. As seen with the cysts, initially there was a large elimination of beads (45.04% ± 14.88%) in the first cycle, followed by a constant rate of elimination of 5.09% of beads per cycle over the remaining cycles.

Concentration and separation of G. duodenalis cysts recovered from spiked lettuce samples using an inertial separation integrated method.

To directly assess the potential of microfluidic inertial separation chips as a concentration and separation device in food microbiological analyses, the Giardia inertial separation chip was incorporated into a method for the recovery and detection of G. duodenalis cysts from lettuce. Lettuce was chosen as an appropriate food matrix for initial testing, due to its relatively high prevalence of contamination (20, 21), published methodologies, and relatively low number of background food particles, thus theoretically allowing for greater ease of cyst detection. A direct comparison to an effective, yet more conventional, lettuce sampling method (20) was performed in parallel. The effectiveness of this new method incorporating inertial separation into the elution, concentration, and separation of G. duodenalis cysts from artificially contaminated lettuce samples was assessed by comparing the LOD and percent recovery of the two methods and comparing the numbers of background particles in the final suspensions.

The LOD of the inertial separation method for G. duodenalis cysts in spiked lettuce samples was 38 cysts per 25-g sample. As shown in Table 2, this method enabled the identification of a minimum of one G. duodenalis cyst in each of the three samples spiked with 38 cysts or more, while the samples spiked with 19 or less were negative in each trial. The conventional method (20) enabled the identification of at least one G. duodenalis cyst in the samples spiked with 19 cysts or more but did not yield any positive results for the samples spiked with only 9 cysts. Thus, the conventional method (20) had a slightly higher sensitivity. Parasites were not detected in any of the negative-control samples that were spiked with ddH₂O. It may be possible to lower these detection limits even further by analyzing a larger proportion of the total sample volume (i.e., >60 μl of the total 500-μl sample). However, as immunofluorescence detection is very time-consuming, analyzing larger volumes of samples would not likely be efficient or cost-effective for routine laboratory testing.

TABLE 2.

Limit of detection of Giardia duodenalis cysts artificially inoculated on lettuce samples, based on the number of positive samples in three trials at each concentration

No. of G. duodenalis cysts		No. of positive samples/trial
Per 25-g lettuce sample	Expected in proportion analyzedb	Microfluidic method	Conventional method
300 (225–425)a	36	3/3	3/3
150	18	3/3	3/3
75	9	3/3	3/3
38	5	3/3	3/3
19	2	0/3	3/3
9	1	0/3	0/3

^aRange of cysts for samples spiked with the highest dilution in three trials. Actual average of three trials was 296 cysts.

^bTheoretical total number of possible G. duodenalis cysts present on the three 20-μl slides analyzed from the total 500 μl of final concentrate.

The average percent recovery by each method was determined by enumerating the cysts recovered in the final sample and the cysts in the spike used to inoculate the lettuce samples. The lettuce samples spiked with the highest concentration of G. duodenalis cysts in each trial of the LOD experiment were used to calculate the percent recovery. The inertial separation method recovered an average of 68.39% ± 20.67% (n = 3) of cysts, while the conventional method (20) recovered 80.00% ± 8.16% (n = 3). The slightly lower percent recovery of cysts from the inertial separation method supports the slightly higher LOD. The results of the statistical test demonstrated that the mean percent recovery obtained from the conventional method was not significantly higher (P = 0.35). Initial experiments with buffered cyst suspensions demonstrated that 11.07% of the cysts are lost in the chip and pumping system itself (Fig. 3), while another 16.49% is lost during the six-cycle concentration process. The additional loss in recovery from the expected 72.44% obtained with buffered suspensions, down to 68.39%, can be attributed to the loss of cysts from the elution, vacuum filtration, or initial centrifugation steps of the method.

Particle enumeration of final concentrate.

To determine the concentration of food particles and related debris in the final recovered concentrate, flow cytometry was performed on samples in each of the three trials which were processed by either the conventional or inertial separation methods. The concentration of food particles and debris in each sample was correlated with the total number of background particles (events) analyzed by the cytometer. An approximately 10-fold decrease in background particles (Fig. 5a) was observed in the fluorescence by fluorescence dot plots of the lettuce samples that underwent the inertial separation method (Fig. 5c) versus the conventional method (Fig. 5b). The 10-fold reduction in background particles determined by flow cytometry suggested that the inertial separation method of analysis greatly reduces the concentration of food debris in the final sample.

FIG 5.

Tenfold reductions in background particles in the samples recovered from the inertial separation method (a). A background particle was defined as any event analyzed by the cytometer that was not a Giardia duodenalis cyst or a flow count bead. Representative fluorescence by fluorescence dot plots of samples concentrated by either the conventional method (b) or the inertial separation method (c) is also depicted. Fifty-microliter subsamples of the recovered concentrate from the inertial separation and conventional method were analyzed by flow cytometry. Flow count standard beads ensured that equal proportions of the samples were analyzed by the cytometer.

The representative microscopic images in Fig. 6 qualitatively illustrate the additional difficulty background particles posed in the detection of fluorescently tagged G. duodenalis cysts. The samples processed with the conventional method (Fig. 6A and C) contained a number of larger particles (>10 μm in diameter) which fluoresced red, green, and yellow under blue light. These fluorescent background particles were present at high concentrations and could mask fluorescing G. duodenalis cysts, preventing their detection and leading to an increase in false-negative results. The sample processed by the conventional method also contained a number of smaller particles (≤10 μm in diameter) which nonspecifically fluoresced and could be mistakenly identified as G. duodenalis cysts (6 to 10 μm) if analysis was not thorough, leading to the possibility of false-positive results. In comparison, the samples processed by the inertial separation method (Fig. 6B and D) had many fewer background fluorescent particles. This allowed for a greater ease and speed of detection of cysts during the LOD and enumeration assays and reduced the probability of false-negative and false-positive results. These observations made by immunofluorescence microscopy support the quantitative results obtained by flow cytometry.

FIG 6.

Nonspecific immunofluorescent emissions from lettuce particles impede Giardia duodenalis cyst detection. Shown are representative microscopical images of the recovered concentrate from the conventional method at 200× (A) and 400× (C) magnifications and from the inertial separation method at 200× (B) and 400× (D) magnifications. All samples were stained with Crypto/Giardia Cel reagent FITC-labeled monoclonal antibodies and excited at 450 to 490 nm. Green bars adjacent to the G. duodenalis cyst in panels A and B are 11.97 μm and 10.76 μm in length, respectively.

DISCUSSION

The six-cycle microfluidics concentration protocol is designed to remove approximately 9 ml (90%) of the initial 10-ml volume into the waste reservoir. Therefore, assuming homogeneity, if focusing by inertial separation was not occurring, approximately 90% of the cysts would be expected in the waste reservoir and 10% in the recovered reservoir. The Giardia inertial separation chip appears to be very specific for G. duodenalis cysts, as 72% of the cysts were found in the recovered reservoir following the six-cycle run.

The lost cysts that were not found in either of the reservoirs (waste or recovered) may be attributed to dead volume, which remained within the system and was not pumped out into either of the reservoirs following the run. These lost cysts represent losses in the concentration efficiency, as their loss directly affects the number of parasites concentrated in the process. This was addressed in early optimization experiments through the implementation of an additional priming and withdrawal step at the end of the run. However, it appeared that a small percentage of parasites still remained trapped within components of the system, i.e., channels, valves, tubing, or syringes.

The small polymer fluorescent beads do not fall in the theoretical range of particles expected to be focused by the microfluidic chip, as they are too small (1.90 μm). They acted, therefore, as nonspecific particles which could represent food debris found in a cyst suspension following the elution process. The efficiency of the method in eliminating these undesired particles was high. After cycling the suspensions through the Giardia inertial separation chip six times, greater than 70% of the beads were removed.

The enumeration of cysts and beads in the recovered and waste suspensions following one, three, and six cycles demonstrated that the overall loss of parasites to the waste reservoir occurred due to the gradual loss of unfocused cysts. With each additional cycle, however, a greater proportion of the undesired bead particles was eliminated. Thus, depending on the concentration of lettuce or food particles in the suspension, the protocol can be tailored to eliminate either more or fewer particles, by either increasing or decreasing the number of cycles in the protocol.

In comparison to the conventional method, the microfluidic method proved to be very effective, as both flow cytometry and fluorescence microscopy demonstrated a significant reduction in the number of lettuce particles and other food debris in the final recovered suspensions. Protozoan parasite detection and enumeration by fluorescence microscopy can be a tedious and time-consuming method, especially when analyzing samples with high concentrations of background particles. The sensitivity of detection may also be lower in such samples due to analyst fatigue and the possible masking of target organisms by debris particles. One of the advantages of the inertial separation method was that it resulted in decreased sample analysis times by fluorescence microscopy. The method was also very efficient and specific for the G. duodenalis cysts, as it enabled the recovery of a large proportion (68%) of cysts initially spiked on the lettuce samples, and had a limit of detection of 38 cysts per 25-g sample. In comparison to those of the conventional method, the percent recovery was not significantly different and the limit of detection was only slightly higher.

While the conventional concentration and separation method for foods used in the present study made use of centrifugation and filters, other methods, such as that of Cook et al. (22), incorporate immunomagnetic separation (IMS), which also specifically targets G. duodenalis cysts. By incorporating IMS following centrifugation, Cook et al. (22) obtained an average recovery of 46.0% ± 19% for artificially cyst-contaminated leafy green products. A method developed by Robertson and Gjerde (9) involving two washing procedures by rotating drum and sonication, followed by centrifugation and IMS, resulted in an average recovery of 67%. Thus, the average recovery of 68% of spiked Giardia cysts from lettuce samples obtained in the present study, through the use of inertial separation, demonstrates the potential of this technology as an alternative to the more costly and time-consuming IMS-based approaches. Concentration and separation of 10-ml samples using the Giardia microfluidic inertial separation chip required 28 min to complete, while IMS procedures require 1 h of antibody incubation alone, not including the additional processing time.

No previous LOD assays using immunofluorescence have been performed for G. duodenalis on food; thus, the LOD of the method developed by Dixon et al. (20) was determined in parallel with the inertial separation method. The slight additional loss of cysts in the new method resulted in a slightly higher LOD for use with immunofluorescence microscopy. However, the additional speed in which the slides were analyzed could allow for the analysis of a greater portion of the sample (>3 slides), which would lower the 38-cyst LOD obtained from the inertial separation method and approach that of the conventional method.

The promising results of this integration of inertial microfluidic separation into the detection of G. duodenalis cysts have recently led to the fabrication of a second inertial separation chip, specifically for Cryptosporidium oocysts. Cryptosporidium is another prevalent food- and waterborne protozoan parasite for which current detection methodologies are very similar to those of Giardia spp. and are thus subject to many of the same obstacles. In addition, the majority of detection methods target both protozoan parasites. Initial testing of the Cryptosporidium microfluidic chip on fluorescent beads and C. parvum oocysts showed the specific inertial separation of oocysts (data not shown).

In summary, a novel technique for the concentration and separation of Giardia duodenalis cysts was created. This study demonstrated the efficiency of the inertial separation protocol in concentrating cysts and eliminating nonspecific particles. The microfluidic chip was successfully integrated into a detection method for G. duodenalis cysts on lettuce samples and was shown to effectively separate the parasites from food particles. The reduction in background particles resulted in increased speed and accuracy in sample analysis by immunofluorescence microscopy.