Quantitative measurement of Plasmodium-infected erythrocytes in murine models of malaria by flow cytometry using bidimensional assessment of SYTO-16 fluorescence



Flow cytometry is a powerful tool for measuring parasitemias in murine malaria models used to test new antimalarials. Measurement of the emission of the nonpermeable nucleic acid dye YOYO-1 (at 530 and 585 nm after excitation at 488 nm) allowed the unambiguous detection of low parasitemias (≥0.01%) but required prolonged fixation and permeabilization of the sample. Thus, we tested whether this issue could be overcome by use of the cell-permeant dye SYTO-16 with this same bidimensional method. Blood samples from CD1 mice infected with Plasmodium yoelii, Plasmodium vinckei, or Plasmodium chabaudi or from NODscidβ2m-/- engrafted with human erythrocytes and infected with P. falciparum were stained with SYTO-16 in the presence or absence of TER-119 mAb (for engrafted mice) in 96-well plate format and acquired in Trucount™ tubes. Bidimensional analysis with SYTO-16 was quantitatively equivalent to YOYO-1. Moreover, by combining SYTO-16 with the use of TER-119-PE antimouse erythrocyte mAb and Trucount tubes, the measurement of the concentration of P. falciparum-infected erythrocytes over a range of five orders of magnitude was achieved. Bidimensional analysis using SYTO-16 can be used to accurately measure the concentration of Plasmodium spp.-infected erythrocytes in mice without complex sample preparation. © 2008 International Society for Advancement of Cytometry

Malaria is caused by the erythrocytic stages of protozoa of the genus Plasmodium, which colonize and destroy host's erythrocytes (1). To counter this disease, murine models of malaria are essential tools for research (2), particularly for drug discovery (3). In addition to the standard rodent experimental systems, different murine models of P. falciparum malaria are currently available (4–6). These are of special interest for drug discovery because, with the exception of human subjects, these are the only experimental systems available that allow the evaluation in vivo of the real human pathogen growing inside human erythrocytes (hE) previously engrafted into immunodeficient mice. Not surprisingly, the peripheral blood of these chimeric mice [humanized mice (HM)] is a complex mixture of murine erythrocytes (mE) and hE, in which the hematological effects of massive transfusions of hE and their elimination from peripheral blood may have important effects. Hence, the specific and quantitative measurement of different erythrocytic subpopulations is crucial in HM models, particularly when these models are used to establish the relationship between the amount of an antimalarial drug in blood and the effect on parasitemia through experimental pharmacokinetic and pharmacodynamic studies (PK/PD). In this kind of studies, an accurate and precise measurement of the concentration of parasites in blood is essential (7, 8).

Flow cytometry provides higher throughput, statistical power, and precision than microscopic assessment of blood smears from Plasmodium spp.-infected mice. In spite of this, microscopy has remained as the gold standard to measure parasitemia in murine malarias because of the relatively low specificity of detection of flow cytometry in these models. To date, most cytometric techniques available employ nucleic acid dyes to detect parasite's DNA/RNA in erythrocytes otherwise lacking nuclei (9–14). However, the content of nucleic acids (residual RNA or DNA) in noninfected mE is highly variable and may be similar to the amount of nucleic acids present in infected erythrocytes. For example, since micronuclei-containing mE have quantities of DNA comparable to uninucleated parasites, P. berghei has been used as a standard to define the region of murine micronuclei in studies of genotoxicity in mice assessed by flow cytometry (15). This problem of specificity is particularly marked in the presence of reticulocytosis during recovery from anemia after treatment of an experimental infection (16) or in infections by P. vinckei or P. chabaudi, which do not have preference for reticulocytes (2). In order to overcome this hurdle, we developed a bidimensional method of measuring the appropriately compensated emission of the nucleic acid dye YOYO-1 at 530 ± 30 (FL1530) and 585 ± 42 (FL2585) nm after excitation at 488 nm (YOYO-1530/YOYO-1585) (16). The YOYO-1530/YOYO-1585 bidimensional patterns of staining allowed the distinction between infected and uninfected reticulocytes or normocytes and we could unequivocally determine whether a mouse was infected or not at parasitemias as low as 0.01%. Unfortunately, biological membranes are not permeable to YOYO-1, and a relatively complex sample preparation is required, which makes this dye unsuitable for quick and quantitative measurement of the concentration of infected erythrocytes in blood.

A number of cell-permeant nucleic acid dyes that have been used to detect Plasmodium spp. in mice are currently available. These include DNA-selective dyes such as Hoechst 33258 and Hoechst 33342 (9, 17–19). However, excitation of these dyes is achieved through the use of ultraviolet lasers which are not widely available. Among the non-DNA selective dyes, hydroethidine (13, 20) requires prior oxidation to ethidium to render it capable of binding to nucleic acids (21). Acridine orange, which is another non-DNA selective dye (22), has recently been described as a permeable dye for staining nucleic acids that does not require complex sample processing (23) despite having long been considered a nonpermeable dye that is highly susceptible to staining conditions (21). In any case, the emission spectra of Hoechst, hydroethidine or acridine orange, are different from that of YOYO-1, making them unsuitable for use in a bidimensional FL1530/FL2585 method. In contrast, SYTO-16 (i) is a cell-permeant non-DNA selective cyanine dimer, (ii) has excitation and emission spectra similar to those of YOYO-1, (iii) is essentially nonfluorescent in the absence of nucleic acids, and (iv) shows a quantum yield higher than YOYO-1 (0.65 vs. 0.52 relative to fluorescein, respectively) (24). Interestingly, SYTO-16 has already been used to detect erythrocytes parasitized with P. falciparum in vitro (25, 26) and P. berghei in vivo (26, 27), although, to the best of our knowledge, it has only been formally validated to measure parasitemias in murine models of the hemoparasite Theileria (28). Moreover, none of the authors cited used a bidimensional FL1530/FL2585 method to detect parasitized erythrocytes in mice. In consequence, we tested whether SYTO-16 could be used in murine models of malaria in a fluorescence bidimensional FL1530/FL2585 method (SYTO-16530/SYTO-16585) aimed specifically at providing a fast, specific, and quantitative measurement of the concentration of infected erythrocytes in HM.

In this article, we describe a flow cytometry dual parameter procedure using SYTO-16530/SYTO-16585 to measure Plasmodium spp. parasitemias in murine models of malaria that does not require sample processing and is as sensitive and accurate as YOYO-1530/YOYO-1585. This method allows the simultaneous staining of samples with SYTO-16 and mAbs to identify specific erythrocytic subpopulations and/or avoid interfering background signals. Moreover, the method is compatible with the use of acquisition tubes containing known numbers of calibrated beads (Trucount™), which allow the measurement of the actual concentrations of infected erythrocytes in blood.



Pathogen-free 8- to 12-week-old female CD1 (Hsd:ICR) mice were obtained from Harlan (Gannat, France). A colony of NOD.Cg-Prkdcscid B2mtm1Unc/J (NODscid/β2m-/-) immunodeficient mice was raised and maintained by Charles River Laboratories (L'Arbresle, France). Female mice (8-week old) were used. The mice were provided with autoclaved tap water and γ-irradiated pelleted diet ad libitum and maintained in the GlaxoSmithKline Laboratory Animal Science animal facilities at Tres Cantos (Spain). The animal research complied with national and European Union legislation and GlaxoSmithKline policy on the Care and Use of Animals and related codes of practice.

Reagents and Antibodies

Saline solution and Dulbecco's PBS from Gibco (Paisley, UK); Giemsa solution from Merck & Co. (Whitehouse Station, NJ); SYTO-16 and YOYO-1 from Molecular Probes (Leiden, The Netherlands); biotinylated rat IgG1 isotype control, CD71 (biotinylated rat IgG1, clone C2), PE-streptavidin, PE-conjugated rat IgG2b isotype control, and TER-119 (PE-conjugated rat IgG2b antimouse erythrocyte) from Pharmingen (San Diego, CA).

Parasites and Infection of Mice

P. yoelii 17X, P. berghei ANKA (both kindly donated by Dr. E. Dei-Cas and Dr. L. Delhaes from Institut Pasteur, Lille, France), P. vinckei and P. chabaudi (Malaria Research and Reference Reagent Resource Center, ATCC, Manassas, VA) were used. P. falciparum Pf3D70087/N9 was generated from P. falciparum 3D7 as a standardized isolate capable of growing reproducibly in mice engrafted with hE in GlaxoSmithKline-Diseases of the Developing World (Tres Cantos, Spain) (6). Experiments in standard murine models (P. yoelii, P. chabaudi, or P. vinckei) were performed as described previously (16). Experiments with NODscid/β2m-/- mice engrafted with hE and infected with P. falciparum Pf3D70087/N9 were conducted as described (6). Briefly, in order to assay P. falciparum in mice, cohorts of age-matched female NODscid/β2m-/- were injected i.p. daily with hE throughout the experiment. When ≥40% chimerism in peripheral blood was reached (7–10 days after initiation of injections), the mice were infected with 20 × 106 parasites obtained from infected donors (i.v. in 0.2 ml of saline or hE suspension). In this animal model, P. falciparum Pf3D70087/N9 is not cleared from peripheral blood of mice.

Human Blood

Incomplete donations or erythrocyte concentrates of malaria-negative donors were used, generously provided by the Spanish Red Cross blood bank in Madrid, Spain. Prior to injection, blood stored at 4°C was washed twice with RPMI 1640, 25 mM HEPES (Sigma) containing 7 × 10−3 mM hypoxanthine (Sigma) at room temperature. The buffy coat was removed by aspiration (if required), and erythrocytes were resuspended at 50% hematocrit in RPMI 1640, 25% decomplemented human serum (Sigma) 3.1 mM hypoxanthine. Finally, the blood suspension was warmed at 37°C for 20 min prior to intraperitoneal injection with 1 ml of hE suspension (administered daily throughout the experiments).

Measurement of Parasitemia

Flow cytometry

In brief, 2 μl of blood from the lateral tail vein of mice were rapidly transferred into 0.1 ml/well of saline containing 2.5 or 5 μM SYTO-16 (for rodent plasmodia and P. falciparum in HM, respectively, since HM may have a significantly higher erythrocyte concentration in blood) and, if required, 10 μg/ml TER119-PE, in clean V-bottomed 96-well plates. The samples were incubated for 20 min at room temperature in the dark, after which 10 μl of saline 0.25% glutaraldehyde was added to each well. After 5 min of incubation to allow complete inactivation of plasmodia, 30 μl was transferred from each well to either normal tubes or Trucount™ tubes (Becton Dickinson, San Jose, CA) containing 270 μl of saline and acquired. Each Trucount tube contains a defined number of fluorescent beads. The samples were acquired in FACScalibur or LSRII flow cytometers (Becton Dickinson) equipped with at least a 15 mW 488 nm air-cooled Argon-ion laser, exactly as described for YOYO-1 (16). All the experiments described in this article were performed using a FACScalibur flow cytometer unless otherwise stated. Erythrocytes and leukocytes were gated in logarithmic forward/side scatter dot plots, and green and red fluorescence were collected in photomultipliers through 530 ± 30 (FL1530) or 585 ± 42 (FL2585) band-pass filters, respectively. The mean fluorescence channels in FL1530 and FL2585 were adjusted to equal values in the first decade of intensity in bivariate logarithmic scale dot-plots with noninfected control samples. Leukocytes were excluded during analysis in SSC/FL1530 dot plots because they show much higher amounts of nucleic acids than parasites. Compensation of SYTO-16 emission was performed exactly as described for YOYO-1 (16). This was a critical step to accurately define the region of infected events and must be determined empirically by comparison of the effect of increasing compensation of the emission of SYTO-16 in FL2585 in samples from uninfected and Plasmodium-infected mice. Blood samples taken 48 h after infection with P. yoelii or P. berghei were optimal for compensation, since they show normocytes containing rings and trophozoites, and reticulocytes harboring all stages of the parasite. General procedures and target patterns of staining are detailed in the Results section referring to Figures 2A and 2B. The samples were analyzed using either CellQuest-Pro or BD FACSDiva 5.0 (Becton Dickinson) software. In samples with parasitemias >0.1%, the number of erythrocytes acquired was that which allowed the gathering of at least 1,000 events (typically 2,000–5,000, depending on parasitemia) in the region established for infected erythrocytes. For parasitemias below 0.1%, 1.5 × 106 total erythrocytes were acquired. In order to determine the specifications of the technique, the minimum number of events in the region of infected events considered statistically significant was 100, because it was the minimum number of events which could provide reliably a recognizable bidimensional pattern of infected events when compared with uninfected controls. According to this criterion, the maximum sensitivity achievable by acquiring 1.5 × 106 total erythrocytes was 0.007%.

The concentration of cells in the acquisition tube was calculated according to the following formula: Erythrocytes/mlacquisition tube = (NE/NB) × (Beads/mlacquisition tube); where NE is the number of erythrocytes acquired and NB is the number of beads acquired. The actual concentration of erythrocytes in blood was calculated by multiplying the concentration of cells in the acquisition tube by the dilution factors employed.


Microscopic analysis of samples of peripheral blood from infected mice were performed in blood smears stained with 10% (vol/vol) Giemsa in saline buffer (0.015 M NaCl, 0.001 M phosphate buffer, pH 7.0) as described in Ref. 29.

Statistical Analysis

Experimental differences between the means or medians of different groups were analyzed by the Student's t test, Wilcoxon-signed rank test or one-factor ANOVA, followed by Dunnett's posttest. Homogeneity of variances was assessed using the Levene's test. Normality of variables was assessed by Kolmogorov–Smirnov test. All statistical analyses were performed using GraphPad Prism 4.0 (GraphPad Software, San Diego, CA) or SPSS 13.0 for Windows (SPSS, Chicago, IL). Probability values P > 0.05 were considered nonsignificant. All the experiments described were performed at least twice in order to ensure the reproducibility of results.


Optimization of Staining with SYTO-16

To assess the value of SYTO-16 in murine models of malaria, the concentration of SYTO-16 that lacked hemolytic activity was first determined, and the resolution in FL1530 between noninfected and infected erythrocytes in our assay conditions was maximized. These conditions were standardized in order to avoid sample processing before acquisition. The concentration of SYTO-16 was optimized using P. yoelii since it has a marked preference to infect reticulocytes (2), and the background due to noninfected reticulocytes is negligible 48–72 h after infection in the conditions used (16, 23).Thus, 2 μl of whole blood obtained from CD1 mice infected with P. yoelii were added to 100 μl saline (∼1.6 × 106 erythrocytes/μl, 23.7% of parasitemia at 72 h after i.v. infection with 6.4 × 106 parasitized erythrocytes) containing increasing concentrations of SYTO-16 in a 96-well plate format. After 20 min of incubation at room temperature in the dark, the samples were neutralized with glutaraldehyde for 5 min and acquired 30 μl of each sample in Trucount tubes containing a defined number of fluorescent beads. Although not necessary for the flow cytometry technique, neutralization with glutaraldehyde was included to guarantee safe manipulation of samples potentially infectious to humans. The fluorescent beads in Trucount tubes acted as internal controls to test whether the addition of SYTO-16 significantly reduced the number of erythrocytes per well in comparison to control wells without SYTO-16, as a means of assessing the potential hemolytic effect of SYTO-16 on fresh erythrocytes. The highest mean of fluorescence intensity (MFI) ratio between infected and noninfected erythrocytes was achieved at concentrations between 2.5 and 5 μM (Figs. 1A–1C). At these concentrations, neither significant reduction in the number of erythrocytes acquired (confirmed by microscopy in Neubauer hemocytometer) nor significant changes in FSC/SSC pattern were observed, compared with wells that were not exposed to SYTO-16. However, these negative changes appeared in some samples at 25 μM or higher.

Figure 1.

Establishment of the optimum concentration of SYTO-16. (A) Gating of mE in logarithmic FSC/SSC contour plots. (B) Regions used in one-dimension FL1530 to estimate the optimum concentration of SYTO-16 required maximizing the resolution between infected and uninfected erythrocytes. The MFI of infected erythrocytes was measured (Region R2) and divided by the MFI from the negative population (Region R1). The sample shown was from a CD1 mouse infected with P. yoelii and stained with SYTO-16 at 2.5 μM. (C) Resolution of infected versus uninfected erythrocytes and measurement of hemolysis. Blood from a P. yoelii-infected CD1 mouse (∼106 erythrocytes/sample, 23% parasitemia) was stained with increasing concentrations of SYTO-16 in a 96-well plate format. The figure shows the ratio of MFI from the P. yoelii-infected SYTO-16530+ population (MFIinfected) with respect to the uninfected SYTO-16530 population (MFIuninfected) (left Y-axis) and the percentage of erythrocytes recovered (relative concentration of erythrocytes in each sample with respect to control without SYTO-16, right Y-axis). The data are the mean of triplicates ± SEM pooled from two experiments. The bars for deviations are only shown if bigger than the size of symbols.

Implementation of the SYTO-16530/SYTO-16585 Method

As is true for all of the nucleic acid dyes tested in murine models so far, the one-dimensional assessment of emission of SYTO-16 in FL1530 may have serious problems of specificity at relatively low but significant parasitemias. A common finding in infected mice is exemplified in Figure 2A (upper panels). This figure shows representative histograms of blood samples from uninfected control CD1 and P. yoelii-infected mice 24 h after infection with 107 parasites. The estimated parasitemia, adjusting the region of infected events with samples of infected mice with higher parasitemias as shown in Figure 1B, matched the parasitemia measured by microscopy (1%). However, the background signal in a sample taken from an uninfected CD1 mouse processed in the same conditions was 0.8%. Previously we showed that the staining differences between complex cells invading erythrocytes and residual macromolecular structures in uninfected erythrocytes were unveiled using YOYO-1530/YOYO-1585 by addressing the effect of compensation on the light measured by the FL2585 detector, which was found to be different for infected and uninfected erythrocytes (16). In the present study, tests were performed to ascertain whether or not a similar approach could be employed with SYTO-16. As shown in Figure 2A (lower panels), in the absence of compensation, SYTO-16530/SYTO-16585 did not allow the discrimination between uninfected and infected erythrocytes. Conversely, compensation of emission SYTO-16585 (∼48% in our experimental set-up) revealed a clearly identifiable population of infected erythrocytes. The gates of infected events for SYTO-16530/SYTO-16585 were established by comparing samples from uninfected and infected mice when the compensation of the emission of SYTO-16 in the FL2585 detector was increased from 0% (Fig. 2A, lower panels) to 48% (Fig. 2B). CD1 mice were infected with the rodent parasites P. yoelii, P. chabaudi, and P. vinckei, which exhibit high, indifferent and low preference for infecting murine reticulocytes, (2), respectively. Humanized NODscid/β2m-/- mice with ∼50% of hE in peripheral blood were infected i.v. with P. falciparum. All plasmodial species could be gated with the same region, although each species produced characteristic patterns of staining that correlated with their affinity for reticulocytes and degree of natural synchronization (Fig. 2B). As expected, P. yoelii infection reduced the percentage of uninfected high FL2585 reticulocytes because of its preference for these cells and showed all erythrocytic stages (nonsynchronized growth) (Fig. 2B). Conversely, neither P. chabaudi nor P. vinckei significantly diminished the percentage of uninfected high FL2585 reticulocytes and showed an apparent high degree of synchronization (2). Interestingly, P. falciparum, which does not productively infect mE (30), lacked high FL2585 reticulocytes, probably because of the inhibition of murine erythropoiesis, and showed all erythrocytic stages in peripheral blood (nonsynchronized growth). Applying the gate of acquisition defined previously, the background signal in samples taken from uninfected CD1 or humanized NODscid/β2m-/- mice was <0.05% in all cases. Importantly, since the pattern of staining was similar in samples not neutralized with glutaraldehyde, it was decided that the neutralization step would be maintained for safety reasons.

Figure 2.

Gates of infected erythrocytes in bidimensional SYTO-16530/SYTO-16585 dot plots for murine models of malaria. (A) Specificity of detection of P. yoelii at low parasitemias using a one-dimensional pattern of staining SYTO-16530. We measured blood samples from control uninfected CD1 mice (left panels) or 24 h after infection with P. yoelii (i.v., 10 × 106 infected erythrocytes/mouse, n = 5 mice/group, right panels). Upper panel on the left shows the background signal in uninfected mice, whereas the upper panel on the right shows the percentage of infection measured as indicated above. Lower panels show the very same samples in bidimensional FL1530/FL2585 dot plots without compensation. The arrows show the putative infected events. Data are from a representative individual of each experimental group. (B) Effect of compensation of FL2585 on bidimensional SYTO-16530/SYTO-16585 dot plots and establishment of gates of infected events. We took blood samples from control uninfected CD1 mice or 48 h after infection with P. yoelii, P. chabaudi or P. vinckei (i.v., 10 × 106 infected erythrocytes/mouse, n = 5 mice/group). We also took blood samples from control uninfected humanized NODscidβ2m-/- or 6 days after infection with P. falciparum Pf3D70087/N9 (i.v., 20 × 106 infected erythrocytes/mouse, n = 3 mice/group). The plots shown are from one representative mouse of each experimental group. The arrow indicates the population of erythrocytes stained with SYTO-16 (mostly reticulocytes), which is present in blood from uninfected CD1 mice but not in P. yoelii-infected CD-1 and humanized NODscidβ2m-/- mice.

The next step was to ascertain whether parasitemias measured using SYTO-16530/SYTO-16585 were accurate. Accuracy is defined as the capability of the cytometric technique to give true values of parasitemia by comparison with a reference standard. YOYO-1530/YOYO-1585 was employed as the reference method, since it has been validated independently for standard rodent models of malaria (16, 31) and P. falciparum cultures (32) and utilizes the same strategy of analysis (16). Thus, we measured parasitemias in parallel using YOYO-1530/YOYO-1585 and SYTO-16530/SYTO-16585 in 212 samples taken at different times from peripheral blood of CD1 or humanized NODscid/β2m-/- mice infected with rodent Plasmodium spp. or P. falciparum, respectively. The rodent malaria samples were taken between days 0 and 6 after infection and showed a range of parasitemia between 0.02 and 92%.The samples from the murine P. falciparum model were taken between days 0 and 10 after infection and showed a range of parasitemia between 0.02 and 10%. As shown in Figure 3A, there was a strong correlation between both methods (Spearman's coefficient of correlation r = 0.997, P < 0.0001). Moreover, SYTO-16530/SYTO-16585 measurements were deemed accurate (mean 91.9 ± 1.1%, median 92.6%) in the range assayed (0.07–92% of parasitemia), although a tendency for SYTO-16 to give slightly lower values than YOYO-1 was detected (Fig. 3B, Wilcoxon-signed rank test, P < 0.0001 for departure of the theoretical median 100%). The next step was to determine the range of linearity and sensitivity, as well as to obtain an estimation of the precision of measurements in the detection range. In order to do this, triplicate 1/2 serial dilutions (vol/vol) of blood from P. yoelii-infected CD1 mice with blood from uninfected CD1 controls were tested. As expected, the data were found to be linear (r2 = 0.95) for parasitemias above 0.07% (Fig. 3C). This defined the suitable range of parasitemias over which changes in murine parasitemia could be accurately measured. However, the sensitivity of the method, that is, the lowest detectable parasitemia, was 0.01% and the precision, expressed as coefficient of variation, was 4.3 ± 2.8. Therefore, the cell-permeant nucleic acid dye SYTO-16 provided a fast SYTO-16530/SYTO-16585 method of measuring Plasmodium spp. parasitemias in murine models of malaria that did not require complex sample processing and is as accurate, sensitive, and precise as YOYO-1530/YOYO-1585 (16).

Figure 3.

Accuracy, linearity, and sensitivity for detection of infected erythrocytes (A) Correlation between parasitemias measured using SYTO-16530/SYTO-16585 and YOYO-1530/YOYO-1585. We randomly took blood samples from CD1 mice infected with P. yoelii (n = 66 samples), P. berghei (n = 18 samples), P. chabaudi (n = 24 samples), P. vinckei (n = 24 samples), or humanized NODscidβ2m-/- infected with P. falciparum Pf3D70087/N9 (n = 105 samples) at different times after i.v. infection. The Spearman's coefficient of correlation was r = 0.997 (P < 0.0001). (B) Bland–Altman plot to detect any bias in the measurements of parasitemia using SYTO-16530/SYTO-16585 compared to YOYO-1530/YOYO-1585. The plot shows the ratio between the % of parasitemia measured by SYTO-16530/SYTO-16585 and YOYO-1530/YOYO-1585 versus the average % of parasitemia measured by both methods. Data plotted are the same used in correlation studies shown in the previous section. (C) Sensitivity and linearity of detection of SYTO-16530/SYTO-16585Plasmodium yoelii-infected erythrocytes at low parasitemias. We serially diluted (1/2 vol/vol) erythrocytes from a mouse having 0.5% of parasitemia with erythrocytes from a control CD1 mouse. The theoretical parasitemias were calculated for the different number of erythrocytes in each type of blood. The data shown are the mean ± SEM of the decimal logarithm of triplicate samples for each dilution tested, pooled from three different experiments.

Simultaneous Detection of Erythrocyte Subpopulations Using Monoclonal Antibodies

With the exception of some nonhuman primates, P. falciparum only productively infects hE. In order to evaluate the therapeutic efficacy of antimalarial compounds in P. falciparum-infected humanized NODscid/β2m-/- mice, it is mandatory to measure both parasitemia and the percentage of hE in peripheral blood of mice to discard the possibility that any drop in the former is due to an enhanced clearance of hE, as found in other analogous models (33). To assess this practical problem, we tested whether the nonhemolytic rat anti-mouse mAb TER-119, which is specific for the murine erythroid lineage (34), could be used for the simultaneous detection of mE and parasites by the SYTO-16530/SYTO-16585 technique. First, experiments were performed to determine the concentration of antibody necessary to achieve minimum background, maximum discrimination of murine and hE while showing no hemolytic activity and no significant erythrocyte aggregation (35). Thus, 2 μl of blood from an uninfected humanized NODscid/β2m-/- mouse with ∼50% hE in peripheral blood (∼20 × 106 erythrocytes) was added to 100 μl saline containing SYTO-16 at 5 μM and increasing concentrations of TER-119-PE (from 0.1 to 50 μg/ml) in 96-well plate format. Each sample was then incubated, neutralized, and acquired in Trucount tubes as described. At 5–10 μg/ml, TER-119-PE reached saturation of staining without any hemolytic activity (Fig. 4A) and negligible background staining with PE-conjugated isotypic control (not shown). To ascertain whether staining with TER-119-PE would render accurate percentages of erythrocytes, blood of known concentration of erythrocytes obtained from NODscid/β2m-/- mice and human donors were mixed at various proportions (vol/vol) and the relative percentage of each cell type was measured. The results shown in Figure 4B indicated that TER-119-PE measurements were accurate (93.2% ± 2.5%), although the values of percentage of mE were slightly lower than the predicted ones (One sample t test, P = 0.021) in the range assayed (0.7–100%). Next, the specificity of TER-119-PE/SYTO-16530–585 for detection of P. falciparum-parasitized erythrocytes was determined. Most of the background signal detectable in peripheral blood of mice was due to mE, as shown in Figure 4C. Accordingly, the use of TER-119-PE enabled us to reduce background signal to <0.01% in uninfected HM. Using serial dilutions of infected samples with blood from uninfected NODscid/β2m-/- mice, in order to increase the relative percentage of interfering reticulocytes, the sensitivity of detection was observed to be <0.01% and the measurements were linear (r2 = 0.9882), at least for parasitemias ≥ 0.01% (Fig. 4D). Thus, in the humanized mouse model of P. falciparum, TER-119-PE/SYTO-16530–585 staining significantly increased the specificity, range of linearity, and sensitivity of detection of parasitized erythrocytes by at least one order of magnitude compared with SYTO-16530/SYTO-16585 alone.

Figure 4.

Simultaneous staining with SYTO-16 and TER-119 mAb. (A) Titration of TER-119-PE mAb for staining 2 μl of blood from peripheral blood of a humanized NODscidβ2m-/- mice with about 50% of hE in peripheral blood. (B) Accuracy of measurement of mE in samples stained with TER-119-PE mAb. We mixed known volumes of human blood (5.2 ± 0.1 × 109 erythrocytes/ml) with blood from a control NODscidβ2m-/- mouse (6.1 ± 0.4 × 109 erythrocytes/ml). Data are the mean ± SEM of triplicate measurement per data point. (C) Flow cytometry pattern for samples of human blood, NODscidβ2m-/- mice, uninfected HM or Pf3D70087/N9-infected HM stained simultaneously with TER-119-PE mAb and SYTO-16 (TER-119-PE/SYTO-16530–585). Data are representative of n = 44 infected mice. (D) Linearity and sensitivity of detection of P. falciparum-infected erythrocytes in samples stained simultaneously with TER-119-PE/SYTO-16530–585. We serially diluted (1/2 vol/vol) blood from a humanized NODscidβ2m-/- mouse having 0.5% of parasitemia (7.7 ± 0.3 × 109 erythrocytes/ml) with erythrocytes from a control nonchimeric NODscidβ2m-/- mouse (5.2 ± 0.2 × 109 erythrocytes/ml). The theoretical parasitemias were calculated for the different number of erythrocytes in each type of blood. The data shown are the mean ± SEM of the decimal logarithm of triplicate samples for each dilution tested.

Quantitative Determination of Blood Populations in HM Using Trucount Tubes

The use of Trucount tubes containing a defined number of fluorescent beads has been validated in clinical practice for measurement of cells and platelets in human blood (36–38). Since the use of SYTO-16 involves no loss of erythrocytes due to sample processing before acquisition in flow cytometry, experiments were carried out to determine whether Trucount tubes could be used to measure the concentration of erythrocytes in blood in our assay conditions. Erythrocytes were counted in electronic hemocytometer Coulter AcT.5.Diff (Beckman Coulter, Fullerton, CA) and in Neubauer hemocytometer as reference methods. In these experiments, blood from uninfected CD1 was used, since the Coulter AcT.5.Diff could not unambiguously resolve the erythrocyte populations present in samples of HM. Hence, 2 μl of blood from an uninfected CD1 mouse was added to 100 μl saline containing SYTO-16 at 2.5 μM in 96-well plate format (n = 10 replicates). The samples were incubated for 20 min at room temperature in the dark, neutralized with glutaraldehyde for 5 min, and 30 μl of each sample was acquired in Trucount tubes containing 270 μl of sterile saline. Trucount beads were distinguished from erythrocytes by their fluorescence detected at 660 ± 20 (FL4660) in SSC/FL4660 dot plots after excitation at 633 nm. Leukocytes were excluded on the basis of their high fluorescence on FL1530. Although all three methods rendered similar results in terms of accuracy and precision, it was found that flow cytometry counting gave consistently lower values than the hemocytometer (P < 0.01, one-factor ANOVA followed by Bonferroni posttest), as shown in Table 1. Next, these results were confirmed by measuring in parallel erythrocyte concentrations using microscopy (performed by different technicians with different degrees of training), and TER-119-PE/SYTO-16530–585 plus Trucount beads (TSC) in 26 samples taken from peripheral blood of humanized NODscid/β2m-/- mice having between 2.2 × 109 and 11.9 × 109 erythrocytes/ml. As expected from previous results, both sets of data correlated significantly (Spearman's coefficient of correlation r = 0.76, P < 0.0001) and TSC measurements showed an accuracy of 117% ± 28% (median 114.7), indicating a tendency to give slightly higher values than microscopy under routine laboratory conditions (Wilcoxon-signed rank test, P < 0.0028 for departure of the theoretical median 100%). Finally, the linearity and sensitivity of counting by TSC was determined using a LSRII digital flow cytometer in order to maximize the sensitivity of counting infected erythrocytes using 2 μl blood samples. First, the concentration of erythrocytes was measured in samples obtained by 1/2 (vol/vol) triplicate serial dilutions of blood from a P. falciparum-infected humanized NODscid/β2m-/- mouse with human serum. The data were linear over the entire range tested (2.6 × 109 down to 4.1 × 107 erythrocytes/ml, which modeled an extreme anemia), as shown in Figure 5A. The accuracy with respect to the theoretical values in the range tested was 104 ± 3.3% (one sample t test, P = 0.26 for departure from a theoretical accuracy of 100%) and the limit of sensitivity was 4.0 × 105 erythrocytes/ml due to the high background found in human serum alone (3.7 ± 0.1 × 105 erythrocytes/ml). Second, we measured the concentration of infected hE (ihE) in samples obtained by 1/2 (vol/vol) triplicate serial dilutions of blood from a P. falciparum-infected humanized NODscid/β2m-/- mouse with blood from an uninfected NODscid/β2m-/- mouse to maximize the unspecific background due to murine reticulocytes. Since the measurement of the concentration of cells depends on the concentration of beads in the tube, we eliminated the uninfected erythrocytes by establishing a threshold for acquisition in FL1530. This strategy dramatically diminished the number of events to be acquired, limiting this number, in practical terms, to the number of beads per Trucount tubes (Fig. 5B). As shown in Figure 5C, the measurements were linear over the whole range tested (83.1 × 106 down to 1.6 × 103 ihE/ml, which encompassed a range of parasitemias from 1.2% down to 0.0018%). The accuracy with respect to the theoretical values in the range assayed was 111% ± 7.1% (one sample t test, P = 0.14 for departure of the theoretical 100% of accuracy) and the limit of sensitivity was 7.0 × 102 erythrocytes/ml (background was 0.0006 ± 0.00005 × 106 erythrocytes/ml). In summary, these data indicated that Trucount tubes can be used in combination with SYTO-16 and TER-119-PE mAb to accurately measure the concentration of P. falciparum-infected or uninfected erythrocytes in a range of five orders of magnitude while generating small size listmode files.

Figure 5.

Measurement of the concentration of erythrocytes in HM by TSC. (A) Linearity and accuracy of measurement of concentration of hE using TSC. We serially diluted (1/2 vol/vol) blood from a humanized NODscidβ2m-/- mouse infected with P. falciparum (7.0 ± 0.2 × 109 erythrocytes/ml) with decomplemented human serum. The data shown are the mean ± SEM of the decimal logarithm of the concentration of hE from triplicate samples for each dilution tested. (B) Flow cytometry patterns for samples of P. falciparum-infected HM (upper and middle panels) or uninfected HM (lower panel) mice, following acquisition of all SYTO-16530+ and FL4660+ events. SYTO-16530+ events encompass ihE and background cells, whereas FL4660+ events are beads. (C) Linearity and accuracy of measurement of the concentration of ihE using TSC. We serially diluted (1/2 vol/vol) blood from a humanized NODscidβ2m-/- mouse infected with P. falciparum (1.1% parasitemia, 7.0 ± 0.2 × 109 erythrocytes/ml) with blood from a control nonchimeric NODscidβ2m-/- mouse (8.8 ± 0.4 × 109 erythrocytes/ml). The data shown are the mean ± SEM of the decimal logarithm of the concentration of parasitized erythrocytes in blood in triplicate samples for each dilution tested. All samples were acquired in a Becton Dickinson LSRII flow cytometer and analyzed using BD FACSDiva 5.0 software.

Table 1. Comparison of three methods for measurement of the concentration of erythrocytes in murine blooda
  • a

    Concentration of erythrocytes in samples of blood from the same CD1 mouse in each experiment. Data are the mean erythrocytes/ml × 109 ± SEM of 10 replicates for measurements performed in Coulter AcT.5.Diff or using Trucount™ beads. For Neubauer counting, four fields containing at least 100 erythrocytes were counted.

  • b

    The difference of the mean compared to Neubauer counting was significant (P < 0.05, one-factor ANOVA, Dunnett's posttest).

Experiment 17.84 ± 0.747.71 ± 0.447.01 ± 0.42b
Experiment 27.32 ± 0.448.15 ± 0.8b7.89 ± 0.26b
Experiment 38.08 ± 0.528.67 ± 0.41b7.19 ± 0.33b


In this article, we describe a simple, sensitive, and accurate quantitative method to measure parasitemias in peripheral blood of mice infected with Plasmodium spp. using SYTO-16530/SYTO-16585.

Any flow cytometry technique for murine models of malaria should have adequate specificity, that is, the ability to discriminate between infected and background erythrocytes in the range of its intended use. In this respect, the enhanced specificity of the YOYO-1530/YOYO-1585 method for detection of parasitemias in mice allowed the unequivocal identification of parasites in many circumstances in which the previously described methods had failed (16). Despite the quality of this technique, it had two major shortcomings. First, it did not allow a quick determination of parasitemia since a minimum of 2 h of fixation in glutaraldehyde was required before sample processing. Second, measuring the concentration of infected erythrocytes in blood samples was not possible, because of the extensive manipulation of the blood samples. Therefore, SYTO-16, a membrane-permeant nucleic acid dye with spectral properties similar to YOYO-1, was tested because of its potential to maintain the level of specificity achieved with YOYO-1530/YOYO-1585 while allowing a fast and quantitative measurement of infected erythrocytes in mice infected with Plasmodium spp.

Before implementing the SYTO-16530/SYTO-16585 bidimensional method, in which compensation is of critical importance, we determined the concentration of SYTO-16 that maximized the MFIinfected/MFIuninfected ratio at 530 nm in samples showing a negligible percentage of uninfected reticulocytes (e.g., P. yoelii 48–72 h after i.v. infection with 107 parasites). SYTO-16 has been previously used to detect P. falciparum in vitro (25, 26), and P. berghei (26, 27) and Theileria sergenti (28) in mice by measuring its emission around 530 nm after excitation at 488 nm. The concentrations of dye used by these authors ranged from 20 nM (26) to 3 μM (27). Unfortunately, no information has been published on the criteria and the titration curve results in the different conditions of staining with SYTO-16. Most of the studies used PBS as staining buffer (26–28) although there seems to be evidence indicating that phosphate salts may affect the binding of SYTO-16 to DNA (24). Moreover, there is experimental evidence indicating that changing the staining buffer and/or the interactions with other reagents may markedly reduce the MFI after staining with SYTO-16 (26). Therefore, it is very difficult to compare the concentrations deployed in the aforementioned studies with the relatively high concentrations that were shown to be required to maximize the MFIinfected/MFIuninfected ratio. In summary, it is likely that the wide differences in the concentrations of SYTO-16 employed reflect not only different conditions of staining but also different criteria of optimization. Since SYTO-16 has a quantum yield that is higher than most nucleic acids dyes, it is not surprising that sufficient resolution for some uses was also obtained at lower concentrations of SYTO-16. In fact, in our experimental conditions, the MFIinfected/MFIuninfected ratio for SYTO-16 at 0.5 μM (MFIinfected/MFIuninfected ratio ∼ 100) was actually higher than the highest ratio obtained for YOYO-1 at 0.25 μM (optimal MFIinfected/MFIuninfected ratio ∼ 60) (16) and appeared to be very similar to the resolution described by Huber et al. (27) using 3 μM in PBS. Regardless of the differences between findings reported in the present study and those reported by other authors, SYTO-16 showed a MFIinfected/MFIuninfected that was five times higher than YOYO-1 (at optimum concentration), hydroethidine or thiazole orange and even higher when compared to Hoechst, propidium iodide, or acridine orange, according to data presented by other authors (20, 23, 28, 39–41). This trait of SYTO-16 was crucial to reproducibly measure very low concentrations of parasites (below <0.01%) since the background signal was minimized.

The detection specificity of the SYTO-16530/SYTO-16585 method is dependent on the differential distribution of values of the bidimensional vector (SYTO-16530, SYTO-16585) for the populations of uninfected erythrocytes containing residual nucleic acids and infected erythrocytes containing Plasmodium spp. Although SYTO-16 is not a DNA-selective dye, its peaks of absorption (488 or 494 nm bound to DNA or RNA, respectively) and emission (518 and 525 nm, bound to DNA or RNA, respectively) were slightly shifted to longer wavelengths, and its quantum yield was significantly lower (0.65 vs. 0.24, bound to DNA or RNA, respectively) when bound to RNA (24). Thus, the first component of the vector (SYTO530), which is close to the emission maximum of SYTO-16, probably reflects the amount and type of predominant nucleic acid inside erythrocytes, that is, DNA in infected erythrocytes and RNA in uninfected reticulocytes. Consequently, as occurred in one-dimensional methods, SYTO-16530 alone (the first component of the bidimensional vector), this component may not discriminate between infected and uninfected erythrocytes containing similar amounts of nucleic acids. This discrimination was achieved by measuring the compensated SYTO-16585 component. The relative contribution of SYTO-16 bound to DNA, RNA, other undefined sources of fluorescence, and autofluorescence to the emission in FL2585 is not currently known, but it may differ significantly from YOYO-1. The SYTO-16530/SYTO-16585 and YOYO-1530/YOYO-1585 methods were equivalent in terms of measurement of the percentage of parasitemia in blood from infected mice, but some experimental evidences indicate that the patterns of staining might be qualitatively different. First, in contrast with the YOYO-1530/YOYO-1585 method, the SYTO-16530/SYTO-16585 procedure does not involve digestion of RNA. It is likely that SYTO-16 bound to RNA may be more relevant in the SYTO-16585 component. This would explain why the SYTO-16530/SYTO-16585 method did not permit a clear differentiation between infected reticulocytes and infected normocytes, in contrast with the pattern of infected erythrocytes observed with YOYO-1530/YOYO-1585 (16). Consequently, the presence of significant amounts of RNA in infected erythrocytes may mask detectable differences in fluorescence between infected reticulocytes and infected normocytes that are unveiled in YOYO-1530/YOYO-1585 upon fixation with glutaraldehyde and RNA digestion. Second, the SYTO-16530/SYTO-16585 procedure does not require fixation with glutaraldehyde. Even though a short 5-min period of fixation with 0.25% glutaraldehyde may induce significant autofluorescence in erythrocytes, our results suggest that autofluorescence does not interfere in the pattern of staining. Therefore, it is likely that the autofluorescence of uninfected background events may be more important for discrimination between infected and uninfected events in YOYO-1530/YOYO-1585 than in the SYTO-16530/SYTO-16585 procedure. This might explain why the RNA present in noninfected erythrocytes could be compensated in FL2585 as the specific signal in infected erythrocytes in SYTO-16530/SYTO-16585 (Fig. 2), whereas the effect of compensation (up to 70%) on background erythrocytes compared to infected erythrocytes was negligible in YOYO-1530/YOYO-1585 (16). Third, since the staining with SYTO-16 was performed using fresh erythrocytes, SYTO-16 may accumulate in mitochondria, the parasitophore vacuole, or cytoplasmic structures of parasites, as found in many eukaryotic cells (24, 42). As a whole, the aforementioned characteristics make the interpretation of SYTO-16530/SYTO-16585 patterns of staining difficult in terms of unequivocally identifying the different stages of the parasite in vivo, as has been addressed recently for monitoring P. falciparum in vitro using Hoechst and thiazole orange (41). A potential way to overcome staining pattern interpretation problems might be to test permeable DNA-selective dyes that can be excited at 488 or at 532 nm (e.g., Vybrant Dye-Cycle™ Green or Orange stains), as has been suggested recently (43). These reagents emit fluorescence when bound to double-stranded DNA, and the intensity of emission is proportional to the amount of DNA in the cell (24). Although a systematic comparison using a bidimensional method FL1530/FL2585 is still in progress, our preliminary results indicate that SYTO-16530/SYTO-16585 and Vybrant Dye-Cycle™ Green530/Vybrant Dye-Cycle™ Green585 produce similar patterns of staining, SYTO-16 being brighter than Vybrant Dye-Cycle Green while the latter dye renders lower levels of background.

A number of techniques aimed at simultaneously staining membrane antigens and intraerythrocytic parasite's nucleic acids have been described. All of them were intended to characterize specific subpopulations of infected erythrocytes (44–46) or detect antibodies against membrane antigens found on the surface of infected erythrocytes (47, 48). In addition to this approach, the present study demonstrates that mAb targeted against erythrocytic populations responsible for background signals can be used to significantly enhance the specificity and sensitivity of cytometric measurements. This strategy seemed more reliable than trying to select the target population (e.g. P. falciparum-infected erythrocytes) using antibodies against human glycophorin A, since this is a high-density antigen that might cause artefacts due to agglutination of erythrocytes (35), thereby reducing the quality of measurements of ihE in HM at very low parasitemias. The sensitivity achieved using TSC in HM was remarkable (≤0.001%). Previously, the most sensitive procedures for detection of parasites in peripheral blood had required the lysis of erythrocytes and detection of released parasites (22, 49). With these procedures, the sensitivity in human samples was 0.005% and significantly lower (49), or even unacceptably unspecific (22), in rodent malaria models. Compared with these methods, TSC does not require sample lysis, employs smaller quantities of blood [2 μl vs. 50 μl (49) or 25 μl (22), respectively] and gives accurate values of concentration of ihE over the entire range tested. The main disadvantage is the relatively long time of acquisition per sample required to achieve maximum sensitivity. Indeed, our results suggest that there is still room for improvement in TSC procedure, since platelets are a major source of background in mice at very low parasitemias. Thus, mAbs against murine platelets could be employed to enhance the sensitivity and specificity of measurements in expensive PK/PD studies in P. falciparum HM models.

The technique described in the present article does not require complex sample processing. As there are no changes in sample volume, the measurement of the actual concentrations of erythrocytes in blood was possible. The use of Trucount tubes was strongly advocated because of its demonstrated reliability in clinical practice to measure the concentration of leukocytes (36), red cell vesicles in thalassemic patients (37), and particularly, platelets or residual cells in plasma products (38). However, given the relatively high cost of these tubes, other approaches could be validated for each particular analytical purpose. Therefore, methods based on the stability of the flow rate of bench cytometers (50) or, preferably, with standards prepared with biotin-labeled erythrocytes (51), might provide cheaper alternatives.

Murine models of malaria may become key tools to understand the in vivo PK/PD relationships that govern the therapeutic efficacy of antimalarials (52). These studies demand accurate quantitative measurements of parasite concentration over wide ranges of parasitemia (7, 52) and, as in the case of P. falciparum HM models, the capacity to resolve complex cellular dynamics. The TER-119-PE/SYTO-16530–585 technique described here was developed to measure ihE with high specificity and sensitivity in HM used in PK/PD studies of antimalarial therapeutic efficacy. This procedure provided information on the hematological status of HM (e.g. erythrocyte concentration and degree of engraftment with hE), significantly improved the specificity of detection of P. falciparum-ihE over SYTO-16530/SYTO-16585 alone and enabled us to accurately measure the concentration of ihE in blood over the entire range of detection (five orders of magnitude). Moreover, this technique can be modified to measure the reticulocytosis and the degree of anemia induced by different strains of rodent Plasmodium spp. in immunocompetent mice using anti-mouse CD71 mAb instead of TER-119 (data not shown). Hence, it may become a useful tool in any experimental design in murine models of malaria requiring quantitative assessment of erythrocytic subpopulations.


This work was supported in part by Medicines for Malaria Venture (Geneva, Switzerland) through GSK-Diseases of the Developing World/MMV agreement. We acknowledge the support of Antonio Martínez and all the staff at Laboratory Animal Science Department in GlaxoSmithKline-Diseases of the Developing World in Tres Cantos (Spain) for providing and maintaining the mice used in this study. The authors are also indebted to Elena Jiménez and the staff from Experimental Pharmacophysiology at GSK DDW-Biology for their contribution counting blood samples with the electronic hemocytometer Coulter AcT.5.Diff. The authors are indebted to Joseph J. Campo (CRESIB, Barcelona, Spain) for critical review of the manuscript. The authors are also grateful to Nick Guthrie for the thorough review of the English grammar and syntax of the manuscript.