Babesia microti Primarily Invades Mature Erythrocytes in Mice

Abstract

Babesia microti is a tick-borne red blood cell parasite that causes babesiosis in people. Its most common vertebrate reservoir is the white-footed mouse. To determine whether B. microti invades reticulocytes, as does the canine pathogen B. gibsoni, we infected the susceptible inbred mouse strains C.B-17.scid and DBA/2 with a clinical isolate of B. microti. Staining of fixed permeabilized red blood cells with 4′,6′-diamidino-2-phenylindole or YOYO-1, a sensitive nucleic acid stain, revealed parasite nuclei as large bright dots. Flow cytometric analysis indicated that parasite DNA is primarily found in mature erythrocytes that expressed Babesia antigens but not the transferrin receptor CD71. In contrast, CD71-positive reticulocytes rarely contained Babesia nuclei and failed to express Babesia antigens. Accordingly, the frequency of YOYO-1-positive, CD71-negative cells strongly correlated with parasitemia, defined as the frequency of infected red blood cells assessed on Giemsa-stained blood smears. Importantly, the absolute numbers generated by the two techniques were similar. Parasitemia was modest and transient in DBA/2 mice but intense and sustained in C.B-17.scid mice. In both strains, parasitemia preceded reticulocytosis, but reticulocytes remained refractory to B. microti. In immunocompetent C.B-17 mice, reticulocytosis developed early, despite a marginal and short-lived parasitemia. Likewise, an early reticulocytosis developed in resistant BALB/cBy and B10.D2 mice. These studies establish that B. microti has a tropism for mature erythrocytes. Although reticulocytes are rarely infected, the delayed reticulocytosis in susceptible strains may result from parasite or host activities to limit renewal of the mature erythrocyte pool, thereby preventing an overwhelming parasitemia.

Babesiosis is a tick-borne zoonosis caused by protozoa of the genus Babesia (8, 12, 19). In the United States, human babesiosis due to Babesia microti is considered an emerging infectious disease (19) transmitted by the hard-bodied tick Ixodes scapularis (also known as I. dammini) (27). Although infections are often subclinical, severe disease is seen in immunosuppressed individuals (20). In Europe, human babesiosis is rare but often severe (12, 19). On the basis of morphology and antigen reactivity, most cases have been attributed to the cattle pathogen B. divergens. Some of these cases may have been due to Babesia sp. strain EU1, an organism closely related to B. divergens (5, 11). Only recently has B. microti been identified as the etiologic agent of human illness in Switzerland (23). B. microti infection may be underdiagnosed in central Europe, since antibodies against B. microti have been detected in serum from residents of midwestern Germany (15) and eastern Switzerland (6). Ixodes ricinus, the tick that transmits B. divergens to cattle and people (12, 19), is also a competent vector of B. microti (10). B. microti has been detected in I. ricinus collected in regions of Switzerland (6), Slovenia (4), Hungary (17), and Poland (18).

Babesia species are obligate parasites of red blood cells (8, 12, 19). Following invasion, Babesia sporozoites and merozoites evolve into trophozoites that move freely in the host cell cytoplasm. Asynchronous, asexual budding of a trophozoite generates two to four daughter cells, or merozoites. Because egress of merozoites is accompanied by lysis of the host cell, anemia and reticulocytosis are two of the clinical features of severe babesiosis (8, 12, 19). Some Babesia species clearly differ in their tropism for red blood cells. For instance, the murine B. hylomysci has a tropism for mature erythrocytes (16), whereas the canine B. gibsoni preferentially multiplies in reticulocytes (33). B. microti, the predominant species in small rodents of endemic areas, may be ambivalent (24, 25). As B. microti appears to be a diverse species complex (9), it has yet to be determined whether zoonotic isolates of B. microti prefer immature or mature red blood cells.

In patients, Babesia microti is routinely detected by microscopic analysis of Giemsa-stained thin blood smears (8, 12, 19). The extent of infection is typically determined by analysis of 100 to 500 red blood cells in few microscopic fields most often located at the feathered edge of the smear. Although considered the gold standard of detection, this test is not ideally suited to quantitatively distinguish reticulocytes from erythrocytes. Inspired by major advances in the diagnosis of malaria using fluorescent nucleic acid stains, flow cytometric assays were developed to assess the viability and growth of B. bovis in red blood cells in vitro (32) and to quantify the percentage of red blood cells infected with B. canis or B. gibsoni in naturally (2) or experimentally infected dogs (7). Although flow cytometry is amenable to multiple and simultaneous detection of surface and intracellular molecules, these studies did not attempt to distinguish reticulocytes from erythrocytes. We recently characterized a novel mouse model of infection with B. microti (30). Using a clinical isolate maintained in ticks, we observed that DBA/2 mice develop an intense but transient parasitemia, whereas C57BL/6 and BALB/c mice present a marginal parasitemia. Using scid mice which lack T and B lymphocytes, we confirmed that adaptive immunity is required for a sustained resistance to babesiosis in BALB/c mice (22, 28). In the present study, we used these models of B. microti infection to examine the contribution of reticulocytes and erythrocytes to the parasite burden. To do so, we developed a flow cytometric assay that relies on the sensitive nucleic acid dye YOYO-1 and on the detection of the transferrin receptor, a surface antigen expressed by reticulocytes, but not by terminally differentiated erythrocytes (29).

MATERIALS AND METHODS

Mice.

DBA/2 and B10.D2 mice were purchased from Jackson Laboratories (Bar Harbor, ME). B10.D2 mice are C57BL/10 mice that are congenic for the major histocompatibility complex locus (haplotype H^2d) of the DBA/2 strain. BALB/cBy mice were obtained from the National Institute on Aging, whereas C.B-17 and C.B-17.scid mice were purchased from Taconic, Inc. (Germantown, NY). C.B-17 mice are BALB/c mice congenic for the immunoglobulin heavy-chain Igh^b allele obtained from the C57BL/Ka strain. In addition to the Igh^b allele, C.B-17.scid mice carry the spontaneous mutation scid, which prevents differentiation of T and B cells. All mice were maintained under specific-pathogen-free conditions in clean well-tended quarters at the Division of Laboratory Animal Medicine, Tufts University. Mice were provided with water and chow ad libitum. The Division of Laboratory Animal Medicine approved all experimental designs. For preparation of parasite-infected red blood cells (pRBCs), C.B-17.scid mice were maintained at the Animal Facilities of the Harvard School of Public Health, Boston, MA.

Infection of mice with Babesia microti.

C.B-17.scid mice were exposed to 5 to 10 I. scapularis nymphs infected with RM/NS, an isolate obtained from a Nantucket Island resident diagnosed with babesiosis. This isolate was directly infectious to laboratory mice (Mus musculus domesticus). Partial sequence analysis of the 18S and β-tubulin genes (S. R. Telford III, unpublished data) indicates that RM/NS belongs to the clade of B. microti sensu stricto (9). Parasitemia was monitored by analysis of Giemsa-stained blood smears starting 15 days after ticks had detached from their hosts. When 1 to 35% of red blood cells (RBCs) were infected (sigmoid growth), blood was collected into Alsever's solution and diluted in phosphate-buffered saline (PBS). Mice were injected with 10⁵ pRBCs delivered in 0.2 ml PBS by intraperitoneal injection.

Polyclonal antibody to B. microti.

A polyclonal antibody directed against B. microti antigens was obtained by terminal bleeding of a DBA/2 mouse that had been infected with B. microti for 3 months. Whole blood was collected on EDTA, and platelet-poor plasma was separated by centrifugation at 4°C. Nonimmune plasma was obtained from an uninfected DBA/2 mouse.

Assessment of parasitemia by flow cytometry.

At 2- to 4-day intervals, a drop of blood was obtained by tail snipping and was collected in 250 μl PBS containing 16 IU/ml heparin. Cells were fixed in glutaraldehyde (0.00625%) for 30 min at room temperature, permeabilized in Triton X-100 (0.25%) for 5 min at room temperature, and treated with 100 μg/ml heat-inactivated pancreatic RNase A (Roche Diagnostics, Indianapolis, IN) for 15 min at 37°C. Following centrifugation, cells were resuspended in staining buffer, i.e., PBS containing 1% normal rabbit serum and 0.1% sodium azide. For each sample, cells were split into two reaction tubes. In the first series of tubes, cells were stained for 30 min at room temperature with 0.5 μg/ml of rat immunoglobulin G1 (IgG1) monoclonal antibody directed against mouse CD71, the transferrin receptor (BD Biosciences, San Jose, CA). In the second series of tubes, cells were incubated with a rat IgG1 monoclonal antibody directed against keyhole limpet hemocyanin, an irrelevant antigen (BD Biosciences). Upon completion of primary staining, cells were washed and resuspended in staining buffer. In all reaction tubes, cells were stained in 50 μl for 20 min at room temperature with 1.25 μg/ml of Alexa 647-labeled goat anti-rat IgG whole antibodies (BD Biosciences). The reaction volume was brought to 500 μl, and the nucleic acid dye YOYO-1 iodide (1 μl in dimethyl sulfoxide [DMSO]; final concentration, 20 nM; Molecular Probes, Eugene, OR) was added to the first series of reaction tubes. DMSO, the carrier substance for YOYO-1, was omitted from the second series of tubes, as 0.2% DMSO does not affect the fluorescence of cells stained for the transferrin receptor or keyhole limpet hemocyanin (data not shown). All tubes were incubated for at least 60 min at room temperature while protected from light under aluminum foil. Fluorescence was detected with a FACSCalibur (Becton Dickinson, San Jose, CA) using CellQuest. Upon excitation by the argon-ion laser at 488 nm, the nucleic acid dye YOYO-1 emits at 509 nm. Upon excitation by the He-Ne laser at 633 nm, the fluorochrome Alexa 647 emits at 669 nm. Because our double staining required a dual laser system, the distance between the two lasers was calibrated at each use of the FACSCalibur. Fluorescence emitted in FL1 and FL4 was analyzed using WinMDI software (Scripps Research Institute, La Jolla, CA).

In some experiments, as indicated, cells were stained for nucleic acids, CD71, and Babesia antigens. In the first staining step, cells were exposed to the rat anti-mouse CD71 monoclonal antibody (or its isotype control) and to immune plasma containing B. microti-specific antibodies (or to nonimmune plasma as a control). In the second staining step, cells were incubated with Alexa 647-conjugated goat anti-rat IgG (see above) and with a biotin-conjugated goat anti-mouse IgG adsorbed with rat IgG (Southern Biotechnology Associates Inc., Birmingham, AL). In the third staining step, cells were exposed to PerCP-streptavidin (0.125 μg/ml; BD Biosciences). For controls, whole single-stain blood cells were obtained from a C.B-17.scid mouse infected with B. microti for more than 4 weeks, i.e., presenting high and sustained parasitemia levels. This triple staining took advantage of the dual laser of the FACSCalibur. Upon proper compensation using the single stains, fluorescence emitted in FL1 upon excitation of YOYO-1 reflected nucleic acid content, whereas fluorescence emitted in FL3 (PerCP) and FL4 (Alexa 647) reflected the expression of Babesia antigens and CD71, respectively. Compensation was set prior to acquisition by CellQuest and during analysis with Winlist software (Verity Software House, Topsham, ME).

Assessment of parasitemia by microscopy.

A drop of blood collected at the tip of the tail was placed on a precleaned microscope glass slide (Fisher Scientific, Pittsburgh, PA). A thin blood smear was obtained, quickly air dried, and fixed in anhydrous methanol for 1 min. Smears were exposed for 60 min at room temperature to Giemsa stain diluted in PBS. Stained smears were rinsed thoroughly in water, air dried, and read under oil immersion at a magnification of ×1,000. A trained clinical laboratory technician, blinded to the source of each sample, carried out the counting. Parasitemia was expressed as the number of erythrocytes containing at least one ring form (trophozoite or merozoite) per 100 erythrocytes analyzed. When parasitemia was below 1%, a second set of 100 erythrocytes was analyzed.

Immunofluorescence of Babesia microti-infected red blood cells.

The first step for staining of CD71 and Babesia antigens is described above. In the second step, cells were incubated with Alexa 488-conjugated goat anti-rat IgG (Molecular Probes) and with the biotin-conjugated goat anti-mouse IgG (see above). In the third step, cells were incubated for 10 min with Alexa 594-streptavidin (0.125 μg/ml; Molecular Probes). Cells were washed and resuspended in staining buffer containing 4′,6′-diamidino-2-phenylindole (DAPI; 6.7 μM; Molecular Probes). Cells were incubated in the dark for 10 min, spun, and resuspended in staining buffer. Cells were placed on a precleaned microscope slide, covered with a glass coverslip, and analyzed on a Nikon Eclipse E400 fluorescence microscope under oil immersion at ×1,000 or ×2,000. Images were captured using the Spot Advanced software. CD71 was visualized in green, Babesia antigens in red, and DNA in blue.

Fractionation of Babesia microti-infected red blood cells.

Whole blood cells were stained for detection of CD71 and nucleic acids (see above). Cells were sorted at room temperature on a MoFlo (DakoCytomation, Fort Collins, CO). YOYO-1-positive, CD71-negative cells were sorted into several fractions. An additional fraction consisted of YOYO-1-positive, CD71-positive cells. Sorted cells were spun and exposed for 30 min at room temperature to 0.5 μg/ml of a biotin-conjugated rat IgG2b directed against the mouse pan-erythroid surface marker TER119 (BD Biosciences). Cells were washed and incubated for 10 min with Alexa 594-streptavidin (0.125 μg/ml). Cells were incubated in the dark for 10 min, spun, and resuspended in staining buffer, placed on a slide, and analyzed on a Nikon Eclipse E400 fluorescence microscope.

Statistical analysis.

Data are reported as means ± standard errors of the mean (SEM) for the indicated number of replicates. Linear regression analysis was used to assess the relationship between frequencies of stained cells. Pearson's linear correlation coefficients were used to express the univariate relationships. Differences were considered significant if the two-tailed P values were less than or equal to 0.05. The statistical package used for analyses was Systat 12.0 for Windows (SPSS, Inc., Evanston, IL).

RESULTS

Babesia microti preferentially resides in mature erythrocytes.

To determine whether B. microti invades mature erythrocytes and/or immature reticulocytes, blood cells were obtained from a C.B-17.scid mouse 3 months after it was infected with 10⁵ pRBCs. As this strain lacks T and B cells, an intense parasitemia (circa 40% of infected red blood cells) persists during the second and third months postinfection (30). Blood cells were stained for CD71, the transferrin receptor found on most reticulocytes but not on mature erythrocytes (29). Parasite-derived nuclei were stained with DAPI, a DNA-specific stain. Babesia antigens were revealed by a polyclonal antibody obtained from a Babesia-infected mouse. DNA (blue) and Babesia antigens (red) were detected in CD71-negative but not in CD71-positive cells (green) (Fig. 1). Babesia antigens were detected at the membrane of infected erythrocytes and never colocalized with parasite-derived DNA. Thus, B. microti DNA and antigens are detected in mature erythrocytes.

FIG. 1.

Babesia antigens and DNA colocalize to mature erythrocytes but not to reticulocytes. Blood was obtained from an infected C.B-17.scid mouse. Cells were fixed in glutaraldehyde, permeabilized in Triton X-100, and treated with 100 μg/ml DNase-free RNase A. Cells were stained for the transferrin receptor CD71 (green), DNA (blue) with DAPI, and Babesia antigens (red) with a polyclonal antibody obtained from a DBA/2 mouse 3 months after it was infected with B. microti. Images were captured using the Spot Advanced software.

Because fluorescence microscopy of Babesia-infected red blood cells is not ideally suited for quantitative assessment on a large scale, a flow cytometric assay was developed. Blood cells were obtained from an infected C.B-17.scid mouse. Following fixation and permeabilization, cells were stained with YOYO-1, a cyanine dimer that binds both DNA and RNA (1). Whereas DAPI stained infected CD71-negative cells only (Fig. 1), YOYO-1 stained both CD71-negative and CD71-positive cells (Fig. 2). Since reticulocytes are rich in RNA, cells were treated with DNase-free RNase A. As the concentration of RNase A was increased from 3 to 300 μg/ml, the intensity of YOYO-1 staining decreased in CD71-positive cells (Fig. 2). In contrast, YOYO-1 staining remained unchanged in CD71-negative cells. Thus, the flow cytometry assay corroborated the observations made by fluorescence microscopy, i.e., parasite-derived DNA accounts for the nucleic acid staining in erythrocytes but not in reticulocytes.

FIG. 2.

Nucleic acid staining is sensitive to RNase in reticulocytes but not in Babesia microti-infected erythrocytes. Blood was obtained from an infected C.B.-17.scid mouse. Upon fixation and permeabilization, whole blood cells were treated with increasing concentrations (from 3 to 300 μg/ml) of DNase-free RNase A. Cells were stained for nucleic acids with YOYO-1 and for the transferrin receptor CD71. For each RNase A concentration, control cells were exposed to an irrelevant monoclonal antibody directed against keyhole limpet hemocyanin (not shown). Since the fluorescence of control cells remained unchanged despite increasing concentrations of RNase A, thresholds (solid lines) were set at 1% of control cells untreated with RNase A. Dashed arrows mark the most intense YOYO-1 staining in CD71-negative cells untreated with RNase A. Note that the intensity of YOYO-1 staining in CD71-negative cells (lower right quadrants) was not affected by RNase A treatment. In contrast, the intensity of YOYO-1 staining in CD71-positive cells decreased as the RNase A concentration was increased. Data are representative of two separate experiments.

To verify that B. microti has a predilection for mature erythrocytes, red blood cells were sorted according to nucleic acid and CD71 staining (Fig. 3, central panel). Sorted cells were stained for the pan-erythroid surface marker TER119. When examined under fluorescent light (Fig. 4), the parasite nuclei appeared as green dots while the red blood cell membrane appeared red. Most CD71-negative cells with low YOYO-1 staining contained one nucleus (Fig. 3, fraction F1). As the intensity of YOYO-1 staining increased, the number of nuclei per cell increased (Fig. 3, fractions F2 and F3). A greater total amount of nuclear material allowed a wider range of nucleus numbers per cell because nucleus size varied (Fig. 4C, E, and F). In some cells, the close proximity of nuclei revealed two daughter cells generated by binary fission (Fig. 4C to E). In others, nuclei were arranged in a tetrad or Maltese cross (Fig. 4G), a morphology pathognomonic of babesiosis (26). Scattered nuclei were frequent in CD71-negative cells with high YOYO-1 staining, suggesting multiple infections per cell (Fig. 4D to G). CD71-positive cells were sorted as a single fraction (F4), as they were stained homogenously by YOYO-1. Most CD71-positive cells (82%) contained no parasite nucleus (Fig. 3, fraction F4). Only 12% of CD71-positive cells contained one parasite nucleus, while less than 5% contained two nuclei. CD71-positive cells that contained three nuclei or more were very rare. In cells devoid of a parasite nucleus, the YOYO-1 stain appeared as dim tiny green dots scattered uniformly throughout the cell (Fig. 4A), suggesting that YOYO-1 stains residual RNA which remained undigested by RNase A. These observations indicate that (i) B. microti has a predilection for mature erythrocytes, and (ii) nucleic acid staining in erythrocytes is a function of the number and size of parasite nuclei.

FIG. 3.

Babesia microti primarily resides in mature erythrocytes. Blood was obtained from C.B-17.scid mice 3 months after infection with 10⁵ pRBCs. Whole blood cells were fixed, permeabilized, and treated with 100 μg/ml DNase-free RNase A. Cells were stained for nucleic acids with YOYO-1 and for CD71. YOYO-1⁺ cells were fractionated by fluorescence-activated cell sorting (central panel). CD71⁻ cells were sorted into fractions (marked by vertical rectangles) according to their content in nucleic acids. CD71⁺ cells were sorted as a single fraction, represented by the horizontal rectangle. Fractionated cells were stained for the pan-erythroid surface marker TER119. Each fraction was examined under fluorescent microscopy, and the number of nuclei per cell (ranging from 0 to 8) was recorded for 100 cells (see inserts). The median number of nuclei per cell in CD71⁻ cells ranged from one in the fraction with the lowest YOYO-1 staining (F1) to two and three in the fractions with higher intensity, namely F2 and F3, respectively. CD71⁺ cells were rarely positive for nuclei (F4). Data are means ± SEM for cells from three mice.

FIG. 4.

Budding and multiple infections in Babesia microti-infected erythrocytes. Blood cells from Babesia-infected C.B-17.scid mice were fractionated on the basis of CD71 surface expression and nucleic acid content (see Fig. 3). Nucleic acids were stained in green (YOYO-1), whereas the surface marker TER119 was red (Alexa 594). In CD71⁺ cells (A), the uniform distribution of numerous tiny dim green dots reflected the residual RNA content despite RNase treatment. These tiny dots were not seen in CD71⁻ cells (B to G). In these cells, parasite nuclei appeared as bright large green dots. As YOYO-1 staining became brighter, the number of nuclei per cell increased (B to G). In CD71⁻ cells with intense YOYO staining, nuclei varied in number (C to G) and size (C, E, and F). Some nuclei were in close proximity, suggestive of binary fission (C to E). Other nuclei were distant from each other, suggestive of multiple infections per cell (D to F). The majority of cells in the brightest YOYO-1 staining (far right fraction) contained four or more nuclei (data not shown), thereby increasing the chances of visualizing four daughter cells arranged in a tetrad or Maltese cross (G). Images were captured using the Spot Advanced software.

Reticulocytes remain refractory to B. microti during the course of infection in the susceptible DBA/2 and C.B-17.scid strains.

Because the host cell of choice may change during the course of B. microti infection, blood samples were obtained at 2- to 4-day intervals following the inoculation of DBA/2 mice with 10⁵ pRBCs (Fig. 5). A first drop of blood was dedicated to Giemsa staining on thin blood smears. A second drop was used for flow cytometric analysis of blood cells stained for nucleic acids, CD71, and Babesia antigens. As previously reported (30), parasitemia (defined as the percentage of infected red blood cells counted on Giemsa-stained blood smears) started to rise on day 10, peaked on day 17 (10%), and gradually decreased thereafter to become undetectable on day 28 (Fig. 5A). The frequency of red blood cells expressing Babesia antigens demonstrated similar kinetics. The rise in parasitemia was followed by a reticulocytosis that was detected on day 17, peaked on day 20, and gradually returned to basal levels thereafter. At each time point, virtually all CD71-positive cells were stained by YOYO-1 but were negative for Babesia antigens (Fig. 5B). Very few CD71-positive cells expressed Babesia antigens or failed to stain with YOYO-1. Thus, CD71-negative cells stained by YOYO-1 had the kinetics of parasitemia determined on Giemsa-stained smears and of cells expressing Babesia antigens.

FIG. 5.

Reticulocytes remain refractory to Babesia microti despite severe host susceptibility. Mice from the DBA/2 (n = 4) and C.B-17.scid (n = 3) strains were infected by intraperitoneal injection of 10⁵ pRBCs (day 0). One additional mouse of each strain served as an uninfected control. Blood samples were obtained at 2- to 4-day intervals from day 10 to day 31. (A and C) On each of these days, a drop of blood was placed on a glass slide, a thin blood smear was obtained, and nuclear material was revealed by Giemsa stain. A second drop of blood was placed in heparinized PBS. Blood cells were stained for Babesia antigens (Bab pAb), nucleic acids (YOYO-1), and CD71. (B and D) CD71⁺ cells were analyzed for nucleic acid content and Babesia antigen expression during the course of infection. For each day, staining for the uninfected mouse was subtracted from the staining for each infected mouse. Data are means ± SEM of stained cells as percentages of total counted cells. Note that virtually all CD71⁺ reticulocytes in DBA/2 and C.B-17.scid mice failed to express Babesia antigens. Despite RNase A treatment, YOYO-1 stained residual RNA in reticulocytes, as illustrated in Fig. 4A.

Since parasitemia and reticulocytosis are moderate and reversible in DBA/2 mice, we next investigated whether reticulocytes become infected with B. microti in hosts that develop a severe and sustained infection. C.B-17.scid mice were infected with 10⁵ pRBCs. Parasitemia determined on Giemsa-stained blood smears rose on day 19 to reach a peak (57%) on day 26 and declined to moderate levels (22%) on day 31 (Fig. 5C). A similar time course was observed for Babesia antigen-positive cells. In contrast, reticulocytosis trailed parasitemia by 5 days (Fig. 5C). Although CD71-positive cells represented some 40% of the blood cells on day 31, only 2% expressed Babesia antigens (Fig. 5D). As seen in DBA/2 mice, nearly all CD71-positive cells were stained by YOYO-1. Thus, even under conditions of sustained infection with high levels of parasitized red blood cells, reticulocytes are rarely infected.

To determine whether the frequency of mature erythrocytes containing nucleic acids (YOYO⁺ CD71⁻ cells) reliably reflects parasitemia over the course of infection, we compared it to the frequency of infected erythrocytes detected on Giemsa-stained blood smears and to the frequency of cells expressing Babesia antigens. Microscopically determined parasitemia strongly correlated with the frequency of YOYO⁺ CD71⁻ cells in DBA/2 (r = 0.90, P < 0.001; Fig. 6A) and C.B-17.scid mice (r = 0.99, P < 0.001; Fig. 6B). These linear relationships had a slope of 1.13 and 1.03 in DBA/2 and C.B-17.scid mice, respectively, indicating that an increase in parasitemia reliably translates into an increased frequency of YOYO⁺ CD71⁻ cells. Since the origins were 1.73 and 1.81 in DBA/2 and C.B-17.scid mice, respectively, the flow cytometric assay displayed a slight gain in sensitivity. We next tested for an association between microscopically determined parasitemia and the frequency of cells expressing Babesia antigens. Although correlations were strong in DBA/2 (r = 0.95, P < 0.001; Fig. 6A) and C.B-17.scid mice (r = 0.97, P < 0.001; Fig. 6B), slopes were below 1. Likewise, the frequency of YOYO⁺ CD71⁻ cells strongly correlated with the frequency of Babesia antigen-positive cells in DBA/2 (r = 0.92, P < 0.001) and C.B-17.scid mice (r = 0.98, P < 0.001), but slopes remained below 1 (data not shown). These results indicate that the polyclonal antibody directed against Babesia antigens detected fewer infected cells than the microscopic analysis of Giemsa-stained blood smears or the flow cytometry assay based on CD71 and nucleic acid staining.

FIG. 6.

Frequency of YOYO⁺ CD71⁻ cells is an accurate measure of parasitemia in Babesia microti-infected mice. DBA/2 (A) and C.B-17.scid (B) mice were infected with B. microti (see the legend to Fig. 5). Infection was monitored from day 10 to day 31. Parasitemia defined as the frequency of infected red blood cells assessed by microscopic analysis of Giemsa-stained blood smears was tested for an association with the frequency of YOYO⁺ CD71⁻ cells (open squares) or Babesia antigen-positive cells (filled triangles) determined by flow cytometry. Coefficients of correlation are reported as r². Slope and origin are reported for each linear regression. Bab pAb, polyclonal antibody against Babesia antigens.

Reticulocytosis is induced in mouse strains resistant to B. microti.

Whereas infection of immunocompetent BALB/c mice with B. microti is followed by a low-level and short-lived parasitemia, an intense and sustained parasitemia develops in infected C.B-17.scid mice (30). In addition to the spontaneous scid mutation, C.B-17.scid mice carry the Igh^b allele from the C57BL/Ka strain on an otherwise BALB/c background. To determine whether the delayed reticulocytosis in C.B-17.scid mice is a function of the genetic background or is an attribute of susceptible mice, we monitored reticulocytosis and parasitemia in infected C.B-17 and C.B-17.scid mice (Fig. 7). In this experiment, parasitemia was defined as the frequency of YOYO⁺ CD71⁻ cells. In C.B-17.scid mice (Fig. 7A), parasitemia rose on day 17, peaked for a first time on day 21, and oscillated until day 45, stabilizing thereafter. Reticulocytosis was delayed as it rose on day 21 to stabilize on day 33. In C.B-17 mice (Fig. 7B), parasitemia was highest (4%) on day 21 but receded within 5 days. In these mice, reticulocytosis was early and transient (11% at peak). A similar but more striking pattern was seen in the BALB/cBy mice (Fig. 7C). Parasitemia reached a modest plateau (3 to 4%) between days 17 and 21, whereas reticulocytosis peaked at 16% on day 19. We next monitored parasitemia in B10.D2 mice, which are congenic for the major histocompatibility complex haplotype H^2d of the DBA/2 strain (Fig. 7D). B10.D2 mice were highly resistant to infection but developed a modest and short-lived reticulocytosis that also peaked (5%) on day 21. We conclude that the intensity of reticulocytosis varies with the degree of parasite burden, but the timing of reticulocytosis appears to differ between strains, i.e., reticulocytosis is delayed in the susceptible DBA/2 and C.B-17.scid strains, whereas parasitemia and reticulocytosis develop simultaneously in resistant strains such as C.B-17, BALB/cBy, and B10.D2.

FIG. 7.

Early reticulocytosis in resistant mice but delayed and sustained reticulocytosis in the absence of adaptive immunity. C.B-17.scid (A) (n = 4), C.B-17 (B) (n = 4), BALB/cBy (C) (n = 4), and B10.D2 (D) (n = 7) mice were infected by intraperitoneal injection of 10⁵ pRBCs (day 0). One additional mouse of each strain served as an uninfected control. A drop of blood was collected in heparinized PBS at 2- to 4-day intervals from day 7 until day 91 (A to C) or day 79 (D). Blood cells were stained for nucleic acids (YOYO-1) and for CD71. For each day, staining for the uninfected mouse was subtracted from the staining of cells from each infected mouse. Data are means ± SEM of stained cells as percentages of total counted cells. Note that the scale of the y axis is larger in panel A than in panels B to D, as C.B-17.scid mice have high levels of parasitemia.

DISCUSSION

Long considered a disease of cattle, babesiosis is now recognized as an emerging infectious disease in humans (19). Babesiosis is routinely diagnosed by microscopic analysis of Giemsa-stained thin blood smears (8, 12). Additional tests include the indirect fluorescent antibody test (IFAT), enzyme-linked immunosorbent assay (ELISA), and PCR (8, 12). Because B. divergens infections are already fulminant when antibody titers rise, IFAT and ELISA are of limited diagnostic value. In B. microti infections, serology is positive prior to the onset of parasitemia (21). However, since antibodies persist even after parasites are cleared, IFAT and ELISA fail to distinguish an active from a resolved infection. PCR detects the overall presence of babesial DNA in a blood sample (8, 12) but provides no information on the number of infected red blood cells or on the number of parasites per cell. Flow cytometric assays avoid these limitations. Hydroethidine has been used in studies of B. bovis- and B. canis-infected red blood cells (2, 32). This assay relies on the uptake and metabolic conversion of hydroethidine into ethidium by live parasites. Because conversion does not occur in reticulocytes (2), this assay fails to detect those Babesia species that prefer reticulocytes, such as B. gibsoni (7). In the present study, we adapted the flow cytometric assay that Barkan et al. (1) developed to assess parasitemia in mouse models of malaria. Our assay takes advantage of the strong fluorescence signal generated by the sensitive nucleic acid dye YOYO-1 and distinguishes reticulocytes from erythrocytes on the basis of transferrin receptor surface expression (29). Using mouse models of babesiosis, we have now established that reticulocytes are rarely infected with the human pathogen B. microti. We propose that the frequency of YOYO-1-positive erythrocytes may be used as a surrogate measure of parasitemia in infections with Babesia species that do not invade reticulocytes.

Relying on Giemsa staining of thin blood smears, early studies analyzed the tropism of Babesia species for erythrocytes and reticulocytes (34). Whether B. microti has a tropism remains unclear. In rats infected with a B. microti isolate obtained from the bank vole Clethrionomys glareolus, reticulocytes appeared to be the preferred host cell during early parasitemia (24). Once reticulocytes started to accumulate in the blood, B. microti preferentially invaded mature erythrocytes. When tested in laboratory mice, this isolate was not infectious (24). Examining a blood film from an infected hamster, a host of exquisite susceptibility, Dammin (3) noted that B. microti shows some preference for immature erythrocytes. The duration of infection was not mentioned. A subsequent study in hamsters indicated that B. microti invades reticulocytes and erythrocytes to the same extent and with the same kinetic (25). Differences in experimental hosts and origin of B. microti isolates may explain, in part, these contradicting observations. It has recently been recognized that zoonotic isolates of B. microti form a phylogenetic clade related to but separate from the clade of isolates found in microtine rodents inhabiting endemic areas where human babesiosis is absent (9). In this regard, it is noteworthy that the isolate used by Nowell (24) was obtained from a vole in Surrey, United Kingdom. In the present study, we used a clinical isolate of B. microti that was readily infectious to laboratory mice. We (30) and others (22) reported that BALB/c.scid mice show exquisite susceptibility to B. microti. In this model of severe chronic infection, parasitemia sharply increases to reach high and sustained levels within a month. We now report that, despite an intense parasite burden and a sustained reticulocytosis, reticulocytes are rarely infected. Reticulocytes were sorted on the basis of high nucleic acid content and high CD71 surface expression. When red blood cells were collected in the second month of sustained and persistent parasitemia, few CD71-positive cells contained one parasite nucleus and even fewer had two nuclei. However, even in the absence of parasite nuclei, CD71-positive cells had an intense staining of nucleic acids by YOYO-1. This staining appeared in the form of tiny dim dots scattered uniformly throughout the CD71-positive cell, indicating that YOYO-1 is sensitive enough to detect residual RNA which remained undigested by DNase-free RNase A. As these dots are numerous, the fluorescence emitted on a cell basis is equivalent to that emitted by one or two parasite nuclei typically found in mature erythrocytes (CD71-negative cells). In these cells, parasite nuclei were stained by YOYO-1 (or DAPI) as large dots. Therefore, the low frequency of parasite-derived nuclei in CD71-positive cells supports our conclusion that, in a model of severe chronic infection, reticulocytes are not the host cell of choice for invasion by and budding of B. microti.

The preferred host cell may vary over the course of infection (24). Of the two mouse strains we examined, C.B-17.scid mice fail to resolve B. microti infection, whereas DBA/2 mice develop an acute but transient parasitemia (30). Parasite burden was assessed at regular intervals during the first month postinfection. For both strains, the frequencies of YOYO⁺ CD71⁻ cells strongly correlated with parasitemia values determined by microscopic analysis of Giemsa-stained blood smears. The slopes of linear regression were nearly 1, indicating that B. microti preferentially, if not exclusively, resides in mature erythrocytes during the early phase of infection. Accordingly, CD71-positive cells that expressed Babesia antigens were very rare. We conclude that reticulocytes do not significantly contribute to parasite burden in the early phase of B. microti infection, despite differences in susceptibility pattern between mouse strains. Since B. microti resides in a small subset (18%) of CD71-positive cells 3 months after infection of C.B-17.scid mice, B. microti may partially adapt to a cellular environment where reticulocytes now account for as much as a quarter of the circulating red blood cells. It remains to be established whether this pattern of invasion is a feature of zoonotic isolates of B. microti.

The design of our flow cytometry assay is based on the absence of mammalian nuclei in reticulocytes and erythrocytes. As YOYO-1 stains nucleic acids without sequence specificity, the visualization of condensed nuclei does not warrant the presence of B. microti in red blood cells. YOYO-1-stained CD71-negative erythrocytes were indeed infected by B. microti, since their frequency strongly correlated with the frequency of infected red blood cells assessed by microscopy and with the frequency of cells expressing Babesia antigens detected by flow cytometry. However, in both DBA/2 and C.B-17.scid mice, the slopes of these linear regressions were below 1. Our immunofluorescence studies indicated that a fraction of CD71-negative cells fails to express Babesia antigens, despite the presence of a parasite nucleus (data not shown). Such discordance may be attributed to the time required for the parasite to express antigens once it has invaded the host cell. Interestingly, Babesia antigens were detected at the surface of infected erythrocytes but not at the parasite itself. Flow cytometric analysis of infected red blood cells indicated that Babesia antigens are localized to the inner leaflet of the red blood cell membrane, since they are not detected in unfixed (and nonpermeabilized) cells (data not shown). Our results confirm that YOYO-1-stained erythrocytes are infected by B. microti and indicate that parasitized erythrocytes, in their majority, present Babesia antigens at their cytoplasmic membrane.

Defining the host cell in babesiosis may be critical to the understanding of resistance and/or susceptibility. In a mouse model of malaria infection using Plasmodium yoelii, which preferentially infects reticulocytes, gamma interferon confers protection by suppressing erythropoiesis, i.e., by decreasing the numbers of circulating reticulocytes (31). Conversely, the failure of gamma interferon to prevent or reduce infection of mice with P. vinckei petteri has been attributed to the fact that this parasite invades solely mature red blood cells (31). In the present study, we have identified the mature erythrocyte as the main host cell of B. microti in two susceptible mouse strains. Because reticulocytes are rarely infected throughout the course of severe infection, an early erythropoiesis may contribute to resistance by increasing the frequency of nonhost reticulocytes while decreasing the frequency of erythrocytes, the host cell. In this regard, it is intriguing to consider that a significant and short-lived reticulocytosis was concomitant to a modest, if not marginal, parasitemia in mice of two resistant strains, namely BALB/cBy and B10.D2. Likewise, the kinetics of reticulocytosis and parasitemia overlapped in resistant C.B-17 mice (on a BALB/c background). In striking contrast, reticulocytosis was delayed in C.B-17.scid mice, which lack peripheral T and B cells, and displayed an exquisite susceptibility to infection with B. microti. Likewise, the susceptible DBA/2 strain developed a delayed reticulocytosis. Thus, the delayed reticulocytosis in susceptible strains appears to result from an inefficient or deficient immune response rather than from allelic variations that would directly affect the generation of reticulocytes. Whether Babesia itself reduces erythropoiesis in susceptible strains remains to be investigated. Studies of B. gibsoni infection have shed some light on how an intraerythrocytic pathogen can hijack erythropoiesis (13, 14). B. gibsoni inhibits the activity of 5′ nucleotidase, an enzyme that degrades rRNA in reticulocytes (13). By doing so, B. gibsoni prevents the maturation of reticulocytes (13). The reduced 5′-nucleotidase activity leads to an accumulation of pyrimidine and purine nucleotides, such as CMP and IMP. The former inhibits parasite replication and retards reticulocyte maturation, whereas the latter inhibits parasite replication (14). As reticulocyte maturation is halted and parasite growth curtailed, this scenario favors survival of both parasite and host. Because B. microti preferentially resides in mature erythrocytes, the regulation of erythropoiesis, if any, should differ. By delaying the generation of reticulocytes, B. microti may protect susceptible hosts from an overwhelming parasitemia that would lead to massive hemolysis and ultimately compromise the survival of the host and the parasite itself.

Our studies demonstrate that Babesia microti primarily infects mature erythrocytes in mouse models of human babesiosis. We developed a flow cytometric assay that relies on the detection of nucleic acids by the sensitive dye YOYO-1 and on the identification of reticulocytes as CD71-positive cells. As reticulocytes are rarely infected, even in severe chronic infection, we established that the frequency of YOYO-positive, CD71-negative cells is a reliable surrogate measure of parasitemia in B. microti-infected mice. We anticipate that our assay will be a powerful tool in future genetic studies aiming at identifying chromosomal regions and genes that confer resistance and/or susceptibility to B. microti infection. The present studies are the prelude to a flow cytometric assay that will help to diagnose babesiosis and monitor efficacy of therapy, whether in human or veterinary medicine.