P R O T O Z O A A N D O T H E R PAT H O G E N S

Using flow cytometry to detect protozoa For routine detection of Cryptosporidium and Giardia, flow cytometry with cell sorting bests IFA in terms of relative sensitivity as well as average cost, assay volume, assay time, and turnaround time. Rebecca M. Hoffman, Jon H. Standridge, Audrey F. Prieve, Joseph C. Cucunato, and Mat Bernhardt

T

he American Society of Testing and Materials (ASTM) immunofluorescence assay (IFA) currently used for routine detection of Cryptosporidium and Giardia is fraught with problems. The procedure is tedious, requires high levels of technical expertise, cannot be reproduced among laboratories,1 and lacks sensitivFlow cytometric cell sorting (FCCS), followed by microscopic ity.2 LeChevallier et al3 verification, was evaluated as an alternative to examined each compoimmunofluorescence assay (IFA). Two hundred and sixty-two nent of the ASTM method water samples were stained with either indirectly conjugated or hoping to elucidate the directly conjugated anti-Cryptosporidium and anti-Giardia sources of cyst and oocyst antibodies and were examined by IFA and FCCS, respectively. loss and to combine optiRelative sensitivities of each assay were calculated by dividing the mal conditions to create number of positive samples detected by one assay by the number an improved test. Alof positive samples detected by either assay. When samples though the authors unanalyzed within five months of the initial IFA analysis were covered several sources of compared, the relative sensitivities of FCCS for detection of organism loss, modificaCryptosporidium and Giardia were 94.1 and 78.6 percent, tions of the method rerespectively, whereas the relative sensitivities using the IFA were sulted in variable recovery 35.3 and 61.4 percent. Identification of Cryptosporidium and Giardia of Cryptosporidium oocysts in water samples using FCCS takes less time, costs less, and and Giardia cysts, with analyzes a greater sample volume than the IFA. Given the nearly rates averaging 75 percent threefold increase in relative sensitivity for Cryptosporidium (range 7–129 percent) for detection and the equal relative sensitivities for Giardia detection, Cryptosporidium and 50 the authors support the use of FCCS for routine detection of percent (range 31–68 perCryptosporidium and Giardia in environmental samples. cent) for Giardia in six spiked samples. Copyright (C) 1997 American Water Works Association

104

VOLUME 89, ISSUE 9

JOURNAL AWWA

Inside the flow cytometer (below), an electrical charge (left) is applied to green fluorescing particles. The particles are pulled out of the stream by the oppositely charged deflection plates and directed onto a three-welled microscope slide.

Flow cytometry and cell sorting basics Flow cytometry coupled with cell sorting (FCCS) is used in the United Kingdom and Australia for detection of Cryptosporidium and Giardia in surface waters.4 Flow cytometry relies on the ability of the instrument to individually analyze particles in a suspension. As each particle passes through the flow cytometer’s laser beam, several light scatter properties are simultaneously collected. The flow cytometer measures the amount of light scattered in the forward direction and in the 90 degree–angle directions. These measurements correlate with size and internal complexity, respectively. Additionally, the instrument measures the fluorescent light emitted by each particle. Each parameter can be collected as a linear, log, or peak signal. The instrument collects each of these signals in the form of light energy, converts them to electrical energy, and plots them on user-defined histograms. Flow cytometer–cell sorters have the ability to sort particles of interest from unwanted particles. The cell sorter is equipped with a bimorph crystal and an ultrasonic transducer that vibrate the sample stream, causing it to break into droplets. Using light scatter and fluorescence intensity measurements as a guide, the flow

ing them onto a glass microscope slide while the droplets containing uncharged particles and debris flow into the waste collection tank (Figure 1). Uses of flow cytometry. The uses of flow cytometry have greatly expanded since development of the technology in the 1950s. Initially used to detect and identify surface antigens on blood cells, flow cytometry today is used in such applications as detection of chromosomal abnormalities in he goal of this study was to evaluate tumor cells, detection of DNA damage and repair, bacterial detection FCCS performance in comparison and viability studies, DNA sequencwith that of IFA. ing, and molecular phenotyping. In May 1993, Vesey et al4 first cytometer operator designates the population to be reported the use of FCCS for detection of Cryptosorted by enclosing the particles of interest in a defined sporidium and Giardia in surface water. With this method, a fluorescein- (FITC-) conjugated antibody sort region. When a particle meeting the criteria specis incubated with a sample and binds to the antigens ified in the sort region passes through the laser beam, on any Cryptosporidium and Giardia present. The the instrument electrically charges the droplet carrying that particle. Oppositely charged deflection plates pull stained sample suspension is aspirated by the flow cytometer, and each particle in the sample is interthe particles of interest out of the uncharged sample rogated by the instrument’s laser beam. When excited stream and toward the charged plate, ultimately deflect-

T

Copyright (C) 1997 American Water Works Association SEPTEMBER 1997

R.M. HOFFMAN ET AL

105

FIGURE 1

Schematic of flow cytometer–cell sorter operation

Electrical charge

Flow cell

Sample stream

Deflection plate

– 2,000 v

+ 2,000 v

by the 488-nm laser light, the FITC molecule emits light at 525 nm. The light energy is detected by the flow cytometer, converted to electrical energy, quantitated, and plotted. As Vesey et al4 reported, the authors found plotting the log 90 degree light scatter versus log green fluorescence signals generated by the particles in the suspension gave the most defined separation of background from organisms based on differences in mean fluorescence channel measurements (Figure 2). The cysts and oocysts were sorted out of the sample onto a glass microscope slide. The slide, free of the majority of the debris obscuring membrane IFA filters, was briefly examined using a fluorescent microscope. The goal of this study was to evaluate FCCS performance in comparison with that of IFA.

ents, water treatment plant finished water, agriculturally affected rivers and streams, urban rivers and streams, and pristine wildlife area streams. These samples have been described elsewhere.6 The remainder of the samples were routine environmental samples from diverse sites in Wisconsin. The samples were processed (described later) and archived in the dark at room temperature between IFA testing and flow cytometric analysis. Sampling methods and procedures were adapted from established procedures.7–8 Samples were collected by filtration through a 25.4-cm (1-in.) wound polypropylene cartridge filter* with a nominal porosity of 1 µm. Samples were packed on ice and immediately shipped to the laboratory for processing. All samples were processed for IFA analysis within 72 h of collection. Additionally, 26 packed pellets from geographically diverse sites in the United States and Canada were purchased from a contract laboratory, floated as described later, and simultaneously tested with the membrane IFA and FCCS. FCCS was performed on both the floated sample and the pellet slurry that had been filtered through a 53-µm nylon mesh filter prior to staining and sorting. The latter manipulation was performed to determine whether sample flotation, which is known to be a source of organism loss, could be eliminated using the FCCS assay. Sample concentration. Filters were tested for Cryptosporidium and Giardia using the accepted ASTM procedure9 with the modifications described later. Filters were separated from housing water and cut in half lengthwise to produce fibers approximately 5-cm (2-in.) long. Fibers were teased apart and placed in a 3,500-mLcapacity stomacher bag† with 1,500 mL of elution medium. The filter material was agitated using a stomacher‡ for 10 min to release oocysts and cysts. The stomacher bag was manually pressed, and the eluate was collected and added to the housing water. Samples were concentrated by centrifugation at 1,050 ˘ G in a swinging-bucket centrifuge.§ Combined pellets were resuspended by vortexing and sonicated** for 10 minutes. Samples were underlaid with Percoll-sucrose†† (specific gravity 1.10) and centrifuged at 1,050 ˘ G for 10 minutes. The top 20 mL and 5 mL of the Percollsucrose interface were drawn off, producing 25 mL of concentrated sample. At this point, the concentrated samples were analyzed by either IFA or FCCS assay or archived for FCCS analysis at a later date. IFA. The concentrated sample, along with positive and negative controls,‡‡ were pipetted directly onto prewetted 25-mm- (0.9-in.-) diameter cellulose–acetate filters§§ (0.2-µm-pore-size) that were supported by a second cellulose–acetate filter with a stainless-steel

Materials and methods Samples. Many of the 262 samples for this project were obtained from the Wisconsin Cryptosporidium Surveillance Study. This study was initiated by the Wisconsin legislature in response to the massive Cryptosporidium outbreak in Milwaukee in the spring of 1993.5 Samples included raw sewage, sewage treatment plant effluents, water treatment plant influ-

*M39R10A, Furey Filter and Pump, Milwaukee, Wis. †Tekmar Co., Cincinnati, Ohio ‡Model 3500, Tekmar Co., Cincinnati, Ohio §Model IEC 266, International Equipment Co., Needham Heights, Mass. **Model 8891, Cole-Parmer, Niles, Ill. ††Sigma Chemical Co., St. Louis, Mo. ‡‡Sigma Cote, Sigma Chemical Co., St. Louis, Mo. §§Epics Elite, Coulter Corp., Miami, Fla.

Copyright (C) 1997 American Water Works Association 106

VOLUME 89, ISSUE 9

JOURNAL AWWA

FIGURE 2

1,000

A Giardia 100

Log FITC

Cryptosporidium

10

Debris

1

0.1 0.1

perature for 25 min. Following incubation, 10 µL of directly conjugated anti-Cryptosporidium and anti-Giardia antibody (detection reagent) from kit 2** were added. Samples were vortexed and allowed to incubate in the dark at room temperature for 25 min. Prior to sorting, each sample received 50 µL PBS + 1 percent goat serum. On each run, the positive control was processed. Samples were sorted on a flow cytometer–cell sorter†† equipped with a sorter option that enabled the cytometer to sort defined numbers of fluorescing particles into each of the wells on threewell microscope slides.‡‡ The flow cytometer, with the laser set to 7 mW, was allowed to warm up for 30 min prior to operation. A fluorescent latex microbead standard§§ was used to check instrument alignment and sorter settings. Samples were analyzed using a protocol designed specifically for recovery of cysts and oocysts. Log 90 degree light scatter voltage was set to 500 V, and log green 1,000 fluorescence voltage was set to 840 V. The discriminator setting, initially 50 on the forward scatter signal, was later changed to 5 on the peak green fluorescence signal to enhance recovery. Instrument setup for the flow cytometer–cell sorter was performed according to manufacturer’s instructions. A three-droplet sort was performed with the coincidence abort set to off for optimal recovery. Cell sorter setup took approximately 35 min and could be performed while samples were incubating with blocking agent, goat serum, and antibody. The positive control was sorted at the beginning and end of each run to ensure instrument stability. Samples were loaded onto the sample stage and run at approximately 2,000–4,000 events. The events were

Histogram depicting log 90 degree light scatter versus log green fluorescence of the control stained with anti-Cryptosporidium and anti-Giardia antibodies

1

10 Log 90 degree Light Scatter

100

manifold.* Each well was rinsed with filter-sterilized elution solution, and vacuum was maintained below 12.7 cm (5 in.) of mercury by using a bleeder valve. All IFA analyses were performed using commercial kit antibodies and controls† (kit 1). The primary antibody for Cryptosporidium oocysts and Giardia cysts was diluted 1:10 with phosphate-buffered saline (PBS), applied to each membrane, allowed to react for 25 min, and washed with 2 mL PBS. The kit 1 labeling reagent was added to each membrane and incubated for 25 min. The samples were washed with PBS before dehydration using an ethanol glycerol series. The stained filters were layered on a slide containing antifade agent‡ glycerol soluhe IFA currently used for routine detection tion, allowed to clear, coverslipped with an additional drop of Cryptosporidium and Giardia is fraught of glycerol, and sealed with nail with problems. polish. Slides were stored in a refrigerated desiccant box until microscopic examination. Residual concentrated samallowed to accumulate until the background populaples were archived in the dark at room temperature. tion was discernible. Sample flow was stopped, and a FCCS. The floated material from archived samples sort region was drawn, encompassing the entire region or samples recently processed for IFA analysis was vig- fluorescing greater than the background. The sample orously vortexed, and 0.5–10 mL of sample (dependwas restarted, and the sorter option was activated. The ing on the amount of debris in the pellet) were trans*Cel-line Assoc. Inc., Newfield, N.J. ferred to a siliconized§ 15-mL polypropylene tube. The †DNACheck, Coulter Corp., Miami, Fla. samples were centrifuged for 3 min at 1,050 X G, and ‡Meridian Diagnostics, Cincinnati, Ohio the supernatant was discarded. After resuspension in 90 §Sartorius Inc., Edgewood, N.Y. **Hoefer Scientific, San Francisco, Calif. µL PBS + 1 percent goat serum, the pellets were trans††Hydrofluor, Ensys Inc., Research Triangle Park, N.C. ferred by siliconized pipette tips to a siliconized 12 X 75‡‡DABCO, Aldrich Chemical, Milwaukee, Wis. §§Meridian Diagnostics Inc., Cincinnati, Ohio mm glass tube and allowed to incubate at room tem-

T

Copyright (C) 1997 American Water Works Association SEPTEMBER 1997

R.M. HOFFMAN ET AL

107

TABLE 1

Method FCCS IFA FCCS IFA

Comparison of FCCS and IFA performance using samples assayed by FCCS within five months of the initial IFA analysis*

Organism

Number Positive

Number Detected

Relative Sensitivity†— 95 percent Confidence Interval

Cryptosporidium Cryptosporidium Giardia Giardia

17 17 70 70

16 6 55 43

94.1 (69.2–99.7) 35.3 (15.3–61.4) 78.6 (66.8–87.1) 61.4 (49.0–72.6)

Negative Predictive Value Percent‡— 95 percent Confidence Interval 99.1 (94.6–100) 91.3 (84.6–95.3) 80.5 (69.6–88.3) 69.7 (58.9–78.7)

*Number of samples—130 †Relative sensitivity percent—the number of samples in which Cryptosporidium or Giardia were detected divided by the number of samples positive for Cryptosporidium or Giardia by either method; p values for the differences in Cryptosporidium and Giardia relative sensitivities were 0.006 and 0.091, respectively. ‡The number of true negatives divided by the number of true negatives plus false negatives

than true sensitivity is explained in the discussion section.) Negative predictive values, which measure the likelihood that a sample reported as negative is truly negative, were calculated for each assay. McNemar’s test was used to compare differences in relative sensitivities. A Student’s t-test was used to evaluate sample volume differences.

Results

Sample evaluation. The authors tested 262 fresh or archived samples for the presence of Cryptosporidium cysts and Giardia oocysts using IFA and FCCS assay. The average archive time was 5.5 months (range 0–16 months). Overall, Cryptosporidium positives were detected in 25 of 262 samples by IFA and 24 of 262 samples by FCCS. Giardia positives were detected in 116 of 262 samples by IFA and 114 samples by FCCS. Studies by Vesey et al,10 Vesey and Slade,11 and Rose et al12 have demonstrated the loss of surface epitopes as cysts and oocysts age and during proce-

instrument was set to sort 1,500 fluorescing particles into each of the nonstick-coated slide’s three 14-mm (0.5-in.) wells. If more than 4,500 fluorescent particles were present, additional slides were loaded onto the slide holder. Sample sorting was continued until the sample tube was empty or 20 min had elapsed. The authors believed that decreased assay time was an asset of this procedure, and a sort time greater than 20 min began to discount this advantage. Sample volume analyzed was obtained by subtracting the volume remaining in the sample tubing and sample tube from the original 150 µL. The slides were allowed to air-dry in the low cytometer–cell sorters have dark and were mounted with the ability to sort particles of interest antifade glycerol solution prior to microscopic verification perfrom unwanted particles. formed as described earlier. Between samples, the flow cytometer was rinsed for 2 min with a 10 percent bleach dures such as centrifugation and sonication. In these solution and 2 min with a detergent solution.* studies, 130 of the samples were archived more than Microscopy. Slides for both the FCCS method five months. To examine recovery of samples assayed and the IFA were initially examined at 200X by epi- within a shorter time frame, the authors stratified fluorescence microscopy using a research-grade micro- data by time elapsed between the initial IFA analysis and FCCS analysis (Table 1). scope.† Cryptosporidium oocysts and Giardia cysts were Of the 132 samples analyzed by FCCS within five identified at a magnification of 1,000X using characteristic fluorescence, shape, and size. Fluorescence months of the initial IFA analysis, 17 were positive for intensity grading was done on a 0– +4 scale (0 = not Cryptosporidium. IFA analysis detected 6 of these 17, visible, 1 = dim, 2 = dull, 3 = green, 4 = bright green) whereas FCCS analysis detected 16 of 17 positives. Of by an experienced microscopist. Oocysts and cysts the 132 samples, 70 were positive Giardia samples. IFA were also examined for internal structures by phase analysis detected 43 of 70 positives, and FCCS analycontrast and differential–interference–contrast sis detected 55 of 70 positive samples. The longer the microscopy at 1,000X magnification. samples were archived prior to FCCS analysis, the Statistical analysis. Data were captured in closer the relative sensitivities became (Table 2). Contract laboratory samples. Twenty-six packed spreadsheets‡ and imported into a statistical analysis pellets purchased from a private laboratory were software program.§ Stratified data analyses were perfloated as described in the methods section and formed, relative sensitivity defined as assayed by FCCS and IFA. The purpose was not to Number of samples positive by assay A compare the contract laboratory’s detection effiency

F

Number of samples positive by assay A or B

X 100

was calculated, and 95 percent confidence intervals were computed. (The use of relative sensitivity rather

*Coulter Clenz, Coulter Corp., Miami, Fla. †BX50, Olympus Optical Co. Ltd., Tokyo, Japan ‡QuattroPro, Borland Intl., Scotts Valley, Calif. §EpiInfo,Centers for Disease Control, Atlanta, Ga.

Copyright (C) 1997 American Water Works Association 108

VOLUME 89, ISSUE 9

JOURNAL AWWA

TABLE 2

Comparison of the number of positive Cryptosporidium and Giardia samples stratified by time elapsed between the initial IFA and the FCCS assay Total Positive

IFA Positive

FCCS Positive

p Value

Storage Time months

Cryptosporidium

Giardia

Cryptosporidium

Giardia

Cryptosporidium

Giardia

Cryptosporidium

Giardia

1 5 10 16

11 17 38 43

36 70 120 158

4 6 22 25

25 43 82 116

11 16 22 24

25 55 88 116

0.016 0.006 0.860 1.0

0.823 0.040 0.396 0.914

with the IFA and FCCS performed at the Wisconsin State Laboratory of Hygiene (WSLH), Madison, Wis., but to evaluate the ability of the FCCS assay and IFA to detect positives when the methods were performed simultaneously. FCCS was performed on both the floated sample and the pellet that had been filtered through a 53-µm nylon mesh filter. Relative sensitivity (calculated as described in the methods section) was used because the pellets were stored 2–10 months (average = 4.6 months) since their analysis at the contract laboratory, and several of the pellets originally reported as negative were positive by one or both of the WSLH methods and vice versa. The data are summarized in Table 3. FCCS performed on floated material detected 10 of 10 Cryptosporidium-positive samples. IFA analysis detected 3 of 10 Cryptosporidium positives. FCCS performed on the filtered pellet detected 5 of 10 Cryptosporidium-positive samples and did not offer an advantage over FCCS on the floated sample. FCCS performed on floated material detected 13 of 17 Giardia-positive samples, whereas IFA analysis detected 11 of 17 positives. Data from this comparison of analyses performed simultaneously are comparable to the data the authors obtained in larger analyses, once archive time was taken into account. When samples were analyzed simultaneously, FCCS detected more Cryptosporidium-positive samples and an equal number of Giardia-positive samples. Sample features. Average sample volume analyzed by IFA was 1.5 mL of floated pellet, whereas the

average volume analyzed by FCCS was 2.9 mL. These volumes were compared using a Student’s t-test, and the differences were found to be statistically significant (p =