development of high-throughput methods to quantify cysts of toxoplasma gondii
TRANSCRIPT
Development of High-Throughput Methods to Quantify
Cysts of Toxoplasma gondii
D. Aldebert,1 * M. Hypolite,1 P. Cavailles,1 B. Touquet,1 P. Flori,1,2 C. Loeuillet,1
M. F. Cesbron-Delauw1
� AbstractToxplasma is a protozoan parasite, which forms persistent cysts in tissues of chronicallyinfected animals and humans. Cysts can reactivate leading to severe pathologies.They also contribute to the transmission of Toxoplasma infection in humans byingestion of undercooked meat. Classically, the quantification of cyst burden in tis-sues uses microscopy methods, which are laborious and time consuming. Here, wehave developed automated protocols to quantify cysts, based on flow cytometry orhigh-throughput microscopy. Brains of rodents infected with cysts of Prugniaudstrain were incubated with the FITC-Dolichos biflorus lectin and analyzed by flowcytometry and high-throughput epifluorescence microscopy. The comparison ofcyst counts by manual epifluorescence microscopy to flow cytometry or to high-throughput epifluorescence microscopy revealed a good correlation (r 5 0.934, r 50.993, P \ 0.001 respectively). High-throughput epifluorescence microscopy wasfound to be more specific and sensitive than flow cytometry and easier to use forlarge series of samples. This reliable and easy protocol allow the specific detection ofToxoplasma cysts in brain, even at low concentrations; it could be a new way todetect them in water and in contaminate food. ' 2011 International Society for Advance-
ment of Cytometry
� Key termsToxoplasma; cyst; quantification; high-throughput; microscopy; flow cytometry
THE protozoan parasite Toxoplasma gondii is a worldwide parasite. Although its
definitive host is the cat, it can virtually infect all warm-blooded animals including
birds. It is the causative agent of toxoplasmosis, an usually minor and self-limiting
disease, which can lead to dramatic effects on fetuses or on immunocompromised
individuals. Toxoplasmosis is considered as a major cause of abortion in sheep and
goats (1). This anthropozoonosis is transmitted by ingestion of vegetables or water
contaminated with oocysts from infected-cat feces or by the consumption of raw or
uncooked meat containing cysts from infected animals (2). In intermediate hosts
such as sheep, goat, pig, rodent, and human, Toxoplasma cysts persist for life-time
in tissues establishing a chronic infection. In immuno-suppressed host, cyst reacti-
vation may be responsible for severe or fatal encephalitis and/or disseminated toxo-
plasmosis (3).
The detection of cysts in meat is important to develop prevention strategies (4).
Cysts quantification is also performed in experimental animal model of toxoplasmo-
sis to determine the existence of chronic infection, to compare the ability of the
Toxoplasma strains to produce cysts, or to evaluate drugs or vaccine efficacy on the
outcome of toxoplasmosis (5). Several serological tests are available to identify
Toxoplasma infection and to do dating (6). PCR is a sensitive method to detect
infection (7). These assays detect or quantify parasite burden but not specifically
cyst burden. The aim of this study was to develop reproducible high-throughput
1Laboratoire Adaptation et Pathogeniedes Micro-organismes, UMR 5163 CNRS-UJF Grenoble I, France2Pole de Biologie-Pathologie,Parasitology and Mycology Laboratory,University Hospital, Saint Etienne,France
Received 22 December 2010; RevisionReceived 11 July 2011; Accepted 9August 2011
Grant sponsors: ANR French agency(grant IGECONTOX MIME 2007) andCNRS.
*Correspondence to: Delphine Aldebert,Laboratoire Adaptation et Pathogeniedes Microorganismes, UMR 5163 CNRS-UJF, Institut J. Roget, BP 170, 38042Grenoble cedex 9, France
Email: [email protected]
Published online 8 September 2011 inWiley Online Library(wileyonlinelibrary.com)
DOI: 10.1002/cyto.a.21138
© 2011 International Society forAdvancement of Cytometry
Brief Report
Cytometry Part A � 79A: 952�958, 2011
assays to detect and quantify cysts within tissues. These were
evaluated in comparison with standard laborious micro-
scopic observations.
MATERIALS AND METHODS
Animals
Experimental procedures were carried out in accordance
with European guidelines and approved by our local ethical
committee. Adult female outbred Swiss White mice were pur-
chased from Janvier laboratory (Le Genest-Saint-Isle, France).
Adult male Lewis, Fischer, Brown Norway, Osborne-Mendel,
Dark Agouti rats were purchased from Janvier laboratory (Le
Genest-Saint-Isle, France). They were maintained in our Spe-
cific-Pathogen-Free animal house facility.
Parasites and Infection
T. gondii Prugniaud-strain and T. gondii Prugniaud b-ga-lactosidase strain were used (8). Cyst stocks were maintained
in chronically infected mice which were sacrificed between 2
and 6 months post infection (pi). Following removal, brains
were gently homogenized with 2 ml phosphate-buffered saline
(PBS). Aliquots of this suspension were used for rat and
mouse oral infections. One month later, blood was collected
from retro-orbital sinus for the detection of anti-Toxoplasma
antibody response by immunofluorescence (9). Brains were
then recovered 2 months pi and homogenized in PBS as
described below. Brains of uninfected mice or rats were used
as controls.
Cyst Detection
Enzymatic detection. For the quantification of Prugniaud
b-galactosidase cysts, each individual mice brain was removed
and homogenized in 4 ml PBS. This suspension was fixed by
adding 4 ml of 10% formaldehyde solution and incubating at
room temperature for 20 min. After one wash in PBS, the pel-
let was suspended in 4 ml of PBS and frozen in liquid nitrogen
for 15 min then warmed at 378C to breakdown the cerebral
tissue. Cysts were labelled with b-galactosidase reagents (9).
Following one night of incubation at 378C, the suspension was
distributed into four wells of six-well culture plates and
scanned by light microscopy (magnification 3200) to count
the blue-stained cysts.
Fluorescent Staining
For the quantification of Prugniaud strain cysts, each rat
brain was removed and homogenized in 16 ml of PBS. Mice
brain was homogenized in 4 ml PBS. Brain suspension was
clarified by gentle incubation in proteinase K buffer (protein-
ase K 0.4 lg/ml, Tris pH8 10 mM, EDTA 1 mM, sodium
dodecyl sulphate 0.2%, sodium chloride 40 mM) for 15 min
at 568C. The reaction was stopped with PMSF 2 mM for 5
min at room temperature. Then, the suspension was washed
with PBS and resuspended with FITC-Dolichos biflorus agglu-
tinin (Vector laboratories, CA USA) 20 lg/ml for 30 min, at
room temperature. After one wash in PBS, rat or mice brains
were, respectively, resuspended in 6 ml or 4 ml of PBS and an-
alyzed by epifluorescence microscopy and flow cytometry.
Flow cytometry description. Data acquisition were per-
formed on a 4-colour FACSCalibur (BD Biosciences)
equipped with 488 nm argon laser and CellQuest Software.
The entire sample was analyzed. A threshold on FITC fluores-
cence was used. Staining cysts (FITC-dolichos biflorus) were
Figure 1. Correlation between two microscopic methods and flow
cytometry to quantify Toxoplasma cysts in brains of rodents.
A: Each brain homogenate from fourteen Swiss White mice
infected with the T. gondii Prugniaud b-galactosidase strain weredivided in two and each half was analyzed either by light micros-
copy after enzymatic substrate revelation (b galactosidase sub-strate) or by epifluorescence microscopy (FITC-Dolichos). B: Each
brain homogenate from Swiss White mice (n 5 14) infected with
the T. gondii Prugniaud b-galactosidase strain was divided intotwo halves and analyzed by light microscopy and flow cytometry
after FITC-Dolichos biflorus labelling. C: Brains from Lewis, Fi-
scher, Brown Norway, Osborne-Mendel, Dark Agouti rats were
treated with proteinase K and labelled with FITC-Dolichos. Plate
containing the samples was analyzed by manual and then auto-
matic epifluorescence microscopy.
BRIEF REPORT
Cytometry Part A � 79A: 952�958, 2011 953
analyzed on a bivariate dot plot of FSC versus FITC (Fig. 2).
The logarithmic mode was used for each one.
High-throughput epifluorescent microscopy detection. The
imaging station supports the IX2 system from Olympus
microscope. The illumination system MT20 allows fast wave-
length switching and attenuation for a rapid image acquisition
using a sensitive digital camera (ORCA B&W firewire 1.4MP).
The station is equipped with a 43 objective and filter blocks
BP 460-495/DM505/BA510-550. Automated image acquisition
and data analysis were performed with ScanR software.
Statistical analysis. Spearman’s rank correlation test (r) was
done to determine the association between different techni-
ques, the online BIOSTATGV software was used, P\ 0.01 was
considered significant. As both the sensitivity and specificity
of a test vary with the threshold, the XLSTAT software (version
2010.3.05) was used to draw the ROC curve.
RESULTS
Toxoplasma cysts are usually detected by light microscopy.
However, it is difficult to detect small cysts among brain
aggregated components. The in situ enzymatic detection of b-galactosidase in Toxoplasma expressing the b-gal gene has beenused to improve cyst numeration (8,9). This published
method is, however, limited to b-gal transgenic parasites.
Here, we used the Dolichos biflorus lectin properties to bind
the N-acetylgalactosamine of the cyst wall (10) and to develop
two fluorescence methods to stain cysts from any Toxoplasma
strains. Before labelling brain cysts with the fluorescent Doli-
chos lectin agglutinin, a step of mild proteinase K digestion
was added for cerebral tissue clarification without cyst wall
destruction. Under these conditions, the lack of very low non-
specific labelling allowed easy detection of cysts. Cyst quantifi-
cation was performed on 14 brains of infected mice. Each
brain was homogenized, then, one half was used for enzymatic
labelling, the other half for fluorescent labelling. Comparison
of the two methods revealed a significant correlation between
cyst numeration by light and fluorescent microscopy (r 50.806, n 5 14, P\ 0.0017, thus validating our protocol of cyst
labelling with FITC Dolichos lectin (Fig. 1A).
Microscopic methods for cyst quantification are labori-
ous and the reproducibility is user-dependant. We, therefore,
developed an automated method to count cysts by flow cyto-
metry. The preparation of samples were similar to that used
for fluorescent microscopy except that samples diluted a fur-
ther fourfold in PBS before flow cytometry analysis to avoid
plugging the lines. Brains of noninfected mice were used to
determine analysis parameters. Brains from noninfected mice
determined the background (Fig. 2A). Cyst number was
obtained in the gate. (Fig. 2B)
Flow cytometry cyst count was compared with count
obtained by the epifluorescent microscopic method. The analysis
was performed on a total of 14 individual infected mice brains. A
high correlation was found (r 5 0.934, n 5 14, P\ 0.001) indi-
cating that flow cytometry can be used for cyst quantification
(Fig. 1B). We further assessed the repeatability and reproducibil-
ity with the brain of four and five infected mice, respectively. Ten
samples from each brain were analyzed for the repeatability assay.
The coefficients of variation ranged between 10.0 to 24.1% (Fig.
3A). To assess the reproducibility, each individual infected brain
was homogenized with PBS and divided into four samples which
were labelled and analyzed on 4 consecutive days. The coeffi-
cients of variation ranged between 8.0 and 35.7% (Fig. 3B).
Figure 2. Flow cytometry numeration of Toxoplasma cysts. Brain homogenates from Swiss White mice were labelled with FITC-Dolichos
biflorus agglutinin after proteinase K treatment. Samples were diluted fourfold in PBS and analyzed by flow cytometry. Threshold was
done on fluorescence. The region chosen to determine the cyst number was established on FITC fluorescence and forward scatter (FSC)
parameters. Samples from uninfected mice were used to establish the background (A). The cysts were counted in the positive gate (B).
BRIEF REPORT
954 Automatic Assays to Count Toxoplasma Cysts
Finally, both the sensitivity and specificity of the flow
cytometric assay were determined in comparison with results
obtained using the b-galactosidase detection. Twenty four
individual brains were analyzed, i.e., 12 from infected and 12
from noninfected mice. The 12 brains from infected mice were
found positive by flow cytometry, and 11 brains from nonin-
fected mice were also detected positive. In the negative sam-
ples, flow cytometry revealed between 1 to 4 cysts. These
results gave a sensitivity of 100% and a specificity of 8%. As
both the sensitivity and specificity of a test vary with the
threshold, the XLSTAT software (version 2010.3.05) was used
to draw the ROC curve. A specificity of 90.9% and a sensitivity
of 100% were obtained if the brains contained more than 70
cysts. Hence, it appeared that flow cytometry is a good auto-
matic method to count cysts in brains with high cyst burden
([70 cysts).
To improve the lack of specificity of flow cytometry for
brains with low cyst burden, we next examined the perform-
ance of high-throughput microscopy. Brain samples from
inbred rat strains known to produce either very few Toxo-
plasma cysts or no cyst burden were analyzed (11). Rat brains
were labelled as previously described with FITC-Dolichos, and
the Olympus ScanR screening station was used for automated
image acquisition and data analysis of samples. To obtain
accurate autofocus for successful automated image
acquisition, Human Foreskin Fibroblasts stained with Hoechst
were plated onto six-well plates and overlaid with labelled
brain samples. The automated image acquisition was per-
formed at 340 magnification. The entire area of the wells was
observed and data analysis was performed using a flow cyto-
metric approach. All FITC events were detected (Figs. 4A and
4D), and cysts were selected on the basis of their area and cir-
cularity (Figs. 4B and 4E). A gate corresponding to cysts was
drawn and quantification was automatically realized for each
well. In contrast to flow cytometry, a gallery view of all images
of the gated data population can be created to allow a visual
control of data point (Figs. 4C and 4F). We could manually
exclude nonspecific events as images are linked reciprocally to
any data point. The performance of high-throughput micros-
copy was evaluated by comparing the cyst number with man-
ual fluorescent microscopy. Eleven rats were infected with the
Prugniaud Toxoplasma strain. The brain of each rat was ana-
lyzed successively with manual and then automatic micros-
copy. We detected between 2 and 413 cysts per brain, and a
high correlation was observed between automated and manual
assays (r 5 0.993, n 5 11, P \ 0.001) (Fig. 1C). We also
observed a good correlation for the five brains containing less
than 70 cysts per brain (r 5 0.903). Thirteen negative brains
were screened with both assays. All were negative using the
manual technique, and the few false positives seen with the
automated method could easily be eliminated through obser-
vation of the control gallery (Figs. 4C and 4F).
We next performed some retention tests to find out
whether the availability of the instruments or the screening of
a large series of samples could be a limitation. One half of
infected brain was immediately stained and analyzed, whereas
the other half was stored in liquid nitrogen with 90% fetal calf
serum and 10% dimethyl sulfoxide. These frozen samples were
labelled and analyzed 1 month later. The important lost of
cysts after freezing led us to try an alternative conservation
process (Fig. 5A). Brains were immediately labelled after
recovery. One half of the brain was analyzed immediately,
whereas the other half was stored at 48C for 1 or 2 weeks
before analysis. We also tested whether labelling could be per-
formed on brain samples after 1 month of storage at 48C (Fig.
5B). The labelling and the analysis were performed on these
samples. A high preservative performance was observed with
storage at 48C prior or after labelling.
DISCUSSION
Here, we have developed and validated a sensitive high-
throughput assay which is, to our knowledge, the first report
of an automated count of Toxoplama cysts from brain tissue.
The method is based on the Dolichos biflorus lectin property to
bind to N-acetylgalactosamine on the Toxoplasma cyst wall of
any Toxoplasma strain (10). A commercial FITC-Dolichos
biflorus lectin is available. The labelling may be done in one
Figure 3. Repeatability and reproducibility of flow cytometric
method. A: Repeatability: Each infected Swiss White mouse brain
(n 5 4) was separated into 10 samples which were labelled with
FITC-Dolichos and analyzed by flow cytometry. B: Reproducibility
: Each mouse brain (n 5 5) was separated into four samples,
labelled with FITC-Dolichos and analyzed by flow cytometry at 4
consecutive days. Each sample is represented by one dot. The
line represents the mean.
BRIEF REPORT
Cytometry Part A � 79A: 952�958, 2011 955
step with low nonspecific labelling if appropriate digestion of
cerebral tissue is performed. We have found that mild protein-
ase K treatment is a critical step to obtain highly sensitive
labelling without loss of cysts.
The performance of flow cytometry was evaluated first,
as this equipment is currently available in many laborato-
ries. We demonstrated that flow cytometry analysis gave a
good correlation with the standard microscopic method.
Repeatability and reproducibility were also comparable to
microscopy. The variability of cyst distribution in the sample
required to analyse 60% of the sample (data not shown). We
also showed that flow cytometry is a powerful tool to count
cysts from brain containing more than 70 cysts. These results
were comparable to those obtained from studies of bacterial
counts in food samples, water, and environment. Those reports
showed that although flow cytometry allows a rapid detection
of microorganisms and avoids most of the problems encoun-
tered with culture, it cannot be applied to the enumeration of
rare events (12).
To quantify low cyst burden in brain, we have developed
an alternative method based on high-throughput microscopy,
which permits acquisition of thousands of images. Automa-
tion of all aspects of microscope control due to advances in
computer hardware and software, made the development of
this method possible. The high-throughput technology has
been already used to measure host cell attachment and inva-
sion by the human pathogen Toxoplasma gondii (13). Here, we
described for the first time an automatic microscopic assay to
enumerate Toxoplasma cyst from brain. An entire rat brain
could be analyzed, and during the scanning of the six-wells
plate, 1680 images could be acquired. The images were then
automatically analyzed with the software (Olympus, ScanR).
The cysts were automatically detected by their fluorescence,
area, and circularity. The possibility to obtain a gallery of
Figure 4. High-throughput method. Brains from Lewis, Fischer, Brown Norway, Osborne-Mendel, Dark Agouti rats were treated with pro-
teinase K and labelled with FITC-Dolichos. Plate containing the samples was scanned by automatic epifluorescence microcope. The fluo-
rescence events were detected (A, D) and cysts were identified with circularity factor and area (B, E). The number of events in the box was
automatically obtained. To verify that all events found in the box were cysts, a gallery corresponding to selected events could be obtained
(C, F). (A, C) a positive brain. (D, F) a negative brain. events corresponding to no cysts.
BRIEF REPORT
956 Automatic Assays to Count Toxoplasma Cysts
selected events permitted to exclude nonspecific events. The
high-throughput microscopy gave a good correlation with
manual epifluorescence microscopy even for brains containing
rare events making it a highly specific and sensitive method.
Moreover, the manual microscopic method is time consuming,
laborious, and user-dependant. It can be replaced by high-
throughput microscopy that gives comparable results and can
easily be used on a large series of samples. In addition, the con-
servation test demonstrated that samples could be stored at 48Cfor several weeks before analysis.
In summary, we have developed two complementary
automated fluorescence methods to count Toxoplasma cysts in
brain samples. Flow cytometry, largely present in laboratories,
appeared to be limited to samples with high cyst number and
adequate size to be aspirated by the flow cytometry needle.
The high-throughput microscopy is also quantitative, and
more specific and sensitive than flow cytometry. These techni-
ques should be preferred to bioassay or PCR when determina-
tion of cyst number is critical. These assays could be a new
way to analyse Toxoplasma cysts from water or meat.
ACKNOWLEDGMENT
The text was re-read by Jean Gagnon and Barhie Bellete.
LITERATURE CITED
1. Frenkel JK. Patho-Physiology of toxoplasmosis. Parasitol Today 1988;4(10):273–278.
2. Tenter AM, Heckeroth AR, Weiss LM. Toxoplasma gondii: From animals to humans.Int J Parasitol 2000;30(12–13):1217–1258.
3. Dzierszinski F, Nishi M, Ouko L, Roos DS. Dynamics of Toxoplasma gondii differen-tiation. Eukaryot Cell 2004;3(4):992–1003.
4. Aubert D, Villena I. Detection of Toxoplasma gondii oocysts in water: Proposition of astrategy and evaluation in Champagne-Ardenne Region, France. Mem Inst OswaldoCruz 2009;104(2):290–295.
5. Gormley PD, Pavesio CE, Minnasian D, Lightman S. Effects of drug therapy on Toxo-plasma cysts in an animal model of acute and chronic disease. Invest Ophthalmol VisSci 1998;39(7):1171–1175.
6. Petersen E, Borobio MV, Guy E, Liesenfeld O, Meroni V, Naessens A, SpranziE, Thulliez P. European multicenter study of the LIAISON automated diagnos-tic system for determination of Toxoplasma gondii-specific immunoglobulin G(IgG) and IgM and the IgG avidity index. J Clin Microbiol 2005;43(4):1570–1574.
7. Burg JL, Grover CM, Pouletty P, Boothroyd JC. Direct and sensitive detection of apathogenic protozoan, Toxoplasma gondii, by polymerase chain reaction. J ClinMicrobiol 1989;27(8):1787–1792.
8. Dao A, Fortier B, Soete M, Plenat F, Dubremetz JF. Successful reinfection of chroni-cally infected mice by a different Toxoplasma gondii genotype. Int J Parasitol2001;31(1):63–65.
Figure 5. Storage of mouse brain. Brains from Toxoplasma infected Swiss White mice, collected 2 months after infection were homoge-
nized in PBS. A: The brains (n 5 5) were separated in halves. One part was labelled and analyzed immediately ( ), whereas the other was
store in liquid nitrogen and analyzed 1 month later ( ). B: Each brain (n 5 4) was separated into four parts. One part was labelled and ana-
lyzed immediately ( ), two others were labelled immediately and stored at 48C for 1 week ( ) or 2 weeks ( ) before analysis. The fourth partwas stored for 4 weeks at 48C before labelling and analysis ( ).
BRIEF REPORT
Cytometry Part A � 79A: 952�958, 2011 957
9. Sergent V, Cautain B, Khalife J, Deslee D, Bastien P, Dao A, Dubremetz JF, FournieGJ, Saoudi A, Cesbron- Delauw MF. Innate refractoriness of the Lewis rat to toxoplas-mosis is a dominant trait that is intrinsic to bone marrow-derived cells. InfectImmun 2005;73(10):6990–6997.
10. Zhang YW, Halonen SK, Ma YF, Wittner M, Weiss LM. Initial characterization ofCST1, a Toxoplasma gondii cyst wall glycoprotein. Infect Immun 2001;69(1):501–507.
11. Cavailles P, Sergent V, Bisanz C, Papapietro O, Colacios C, Mas M, Subra JF, LagrangeD, Calise M, Appolinaire S, et al. The rat Toxo1 locus directs toxoplasmosis outcome
and controls parasite proliferation and spreading by macrophage-dependent mechan-isms. Proc Natl Acad Sci USA 2006;103(3):744–749.
12. Aurell H, Catala P, Farge P, Wallet F, Le Brun M, Helbig JH, Jarraud S,Lebaron P. Rapid detection and enumeration of Legionella pneumophila in hotwater systems by solid-phase cytometry. Appl Environ Microbiol 2004;70(3):1651–1657.
13. Mital J, Schwarz J, Taatjes DJ, Ward GE. Laser scanning cytometer-based assays formeasuring host cell attachment and invasion by the human pathogen Toxoplasmagondii. Cytometry Part A 2006;69A(1):13–19.
BRIEF REPORT
958 Automatic Assays to Count Toxoplasma Cysts