Dysfunction of CD169⁺ macrophages and blockage of erythrocyte maturation as a mechanism of anemia in Plasmodium yoelii infection

Significance

Malaria parasites infect hundreds of millions of people and kill over half a million annually. Severe malarial anemia (SMA) is one of the most common severe complications in malaria patients. Although the destruction of infected and noninfected red blood cells (RBCs) and inhibition of RBC production are reported to contribute to SMA, the mechanism of SMA is not completely understood. Here, we show that macrophages from Plasmodium yoelii yoelii 17XNL parasite-infected mice are defective in supporting erythroblast/RBC maturation in vitro and in vivo. Reduced expression of macrophage surface molecule CD169 may provide a marker for the erythropoiesis defect and anemia in malaria. Restoration of the functions of erythroblastic island macrophages may provide a strategy for the treatment of SMA.

Abstract

Plasmodium parasites cause malaria with disease outcomes ranging from mild illness to deadly complications such as severe malarial anemia (SMA), pulmonary edema, acute renal failure, and cerebral malaria. In young children, SMA often requires blood transfusion and is a major cause of hospitalization. Malaria parasite infection leads to the destruction of infected and noninfected erythrocytes as well as dyserythropoiesis; however, the mechanism of dyserythropoiesis accompanied by splenomegaly is not completely understood. Using Plasmodium yoelii yoelii 17XNL as a model, we show that both a defect in erythroblastic island (EBI) macrophages in supporting red blood cell (RBC) maturation and the destruction of reticulocytes/RBCs by the parasites contribute to SMA and splenomegaly. After malaria parasite infection, the destruction of both infected and noninfected RBCs stimulates extramedullary erythropoiesis in mice. The continuous decline of RBCs stimulates active erythropoiesis and drives the expansion of EBIs in the spleen, contributing to splenomegaly. Phagocytosis of malaria parasites by macrophages in the bone marrow and spleen may alter their functional properties and abilities to support erythropoiesis, including reduced expression of the adherence molecule CD169 and inability to support erythroblast differentiation, particularly RBC maturation in vitro and in vivo. Therefore, macrophage dysfunction is a key mechanism contributing to SMA. Mitigating and/or alleviating the inhibition of RBC maturation may provide a treatment strategy for SMA.

Malaria is a deadly disease that affects hundreds of millions of people worldwide. Anemia is a common complication of malaria, particularly in patients infected with Plasmodium falciparum parasites (1–6). Severe malaria anemia (SMA) is linked to increased mortality and morbidity (7). Multiple mechanisms may contribute to SMA, including the destruction of both infected and noninfected red blood cells (iRBCs and nRBCs), splenic sequestration of iRBCs/nRBCs, and inefficient and/or suppression of erythropoiesis (2, 8–12). For example, activated CD8⁺ T cells were shown to remove iRBCs in an outbred-rat model (13), and abnormal features of late erythroid progenitors and inefficient reticulocyte production were observed in children with SMA (3, 12). However, the mechanism(s) underlying the inefficient erythropoiesis remains largely unknown.

Cytokines/chemokines and immune cells play important roles in hematopoiesis and children with SMA (7, 14). IFN-γ has a strong impact on bone marrow (BM) output during inflammation and is known to inhibit hematopoiesis (15). Overproduction of TNF-α and IFN-γ has been associated with dyserythropoiesis and anemia (16, 17). In contrast, IL-12 is known to act as a hematopoietic growth factor (18, 19). Deficiency in IL-12 production has been associated with severe anemia and dyserythropoiesis (17, 20). In addition, macrophages provide signals required for the differentiation and proliferation of erythroblasts and erythroblastic island (EBI) formation (21). The role of the macrophage in malarial anemia has been controversial: Macrophages were proposed to inhibit erythropoiesis by generating oxidative stress or to be protective against a direct toxic effect of hemozoin (Hz) by inducing apoptosis of RBC progenitor cells (11, 12, 22, 23). Proper expressions of molecules such as erythropoietin receptor (EPOR) and CD169 on macrophages are critical for erythropoiesis (21, 24). Both mouse and human EBI macrophages express EPOR (24), and selective depletion of CD169(+) macrophages using granulocyte colony-stimulating factor (G-CSF) blocked erythropoiesis in the mouse BM (25). Here, we infected C57BL/6n mice with Plasmodium yoelii yoelii 17XNL parasites (17XNL) and showed a continuous reduction in RBC count up to day 18 postinfection (pi). The loss of RBCs stimulates the expansion of EBIs in the spleen leading to splenomegaly. However, a blockage in RBC maturation in the BM and the spleen due to dysfunction of EBI macrophages prevents efficient replacement of RBCs destroyed by phagocytosis and/or lysis of iRBCs/nRBCs. This study reveals an important mechanism of malarial anemia and provides valuable information for developing treatments for SMA.

Results

Characterization of 17XNL-Induced Anemia.

The 17XNL parasite invades reticulocytes and grows slowly in early infection but stimulates the production of reticulocytes from day 8 pi leading to increased parasitemia (SI Appendix, Fig. S1). Mice infected with 17XNL had RBC counts that decreased significantly from day 6 pi, reached their lowest level on day 18 pi (very few mature RBCs), and started to increase from day 20 pi (SI Appendix, Fig. S2 A and B). With the declining RBC count, the numbers of reticulocytes increased, reaching a 1:1 ratio of RBCs:reticulocytes at approximately day 13 pi (SI Appendix, Fig. S2B). The number of reticulocytes started to decline on day 24 pi, while the number of mature RBCs increased (SI Appendix, Fig. S2B). As expected, significantly lower hematocrit from days 8 to 18 pi, hemoglobin content from days 6 to 18 pi, and body weight from days 14 to 18 pi were observed in the infected mice when compared with noninfected (NI) mice (SI Appendix, Fig. S2 C–E). RBC counts, hematocrit, and hemoglobin levels all rebounded to normal levels after clearing parasites, and all the infected mice survived (SI Appendix, Figs. S1 G and H and S2F). Phagocytosis of iRBCs, and possibly nRBCs by macrophages, dendritic cells (DCs), or other cells, could contribute to the decline in RBC counts because iRBCs and some nRBCs could trigger phagocytosis by BM-derived macrophages (BMDMs) and BM-derived DCs (BMDCs) in vitro (SI Appendix, Fig. S3 A–G). Because the 17XNL parasites preferentially invade reticulocytes, the destruction of infected reticulocytes alone cannot explain the dramatic loss of mature RBCs. Phagocytosis of nRBCs in the presence of anti-PS (anti-phosphatidylserine) antibodies was reported previously (26). We also performed phagocytosis of iRBCs and nRBCs in the presence of anti-PS and control antibodies (anti-fluorescein). A significantly higher level of phagocytosis of nRBCs from either infected or NI mice was observed in anti-PS antibody-treated groups than in the groups treated with control antibodies (SI Appendix, Fig. S3H). No significant difference was observed among the iRBC groups. These observations are consistent with the results reported previously (26). Mechanisms such as the removal of nRBCs and/or inhibition of erythropoiesis including RBC maturation can also contribute to the reduction in RBC count and hemoglobin content.

Lack of Mature RBC Production in the BM.

We next performed hematoxylin and eosin (H&E) staining of the main erythropoietic tissue, BM, from day 1 pi to day 18 pi and observed an increasing number of white spaces indicative of sinusoids with a decreasing number of mature RBCs in the infected mice (Fig. 1 A–F). The results showed increasing anemic status, which could be due to a defect in erythropoiesis and/or destruction of mature RBCs (27). Interestingly, the presence of the parasite’s Hz pigments suggested phagocytosis of iRBCs by BM macrophages (Fig. 1F, yellow arrows). The Hz could be toxic and affect the normal functions of the macrophages in erythropoiesis, leading to a defect in RBC production (28).

Fig. 1.

H&E staining of the BM and spleen from noninfected (NI) and P. yoelii yoelii 17XNL–infected mice. The procedures for tissue fixation and processing are as described in the Materials and Methods. Magnification and scale bars are as indicated. (A) BM tissue from an NI mouse. (B–E) BM tissues from 17XNL-infected mice on day 1 (B, D1), day 4 (C, D4), day 10 (D, D10), and day 18 (E, D18) pi, all at 20× magnification. (F) Partial image from E under 100× magnification. Note: increasing amounts of white spaces (sinusoids without mature RBCs) from days 1 to 18 pi. (G) Images of whole spleens from individual NI mice and 17XNL-infected mice on day 18 (D18) pi. (H) Plot of spleen weights in grams from NI and 17XNL-infected mice on day 4 and 18 pi. Mann–Whitney U test (n = 4), **P < 0.01; ***P < 0.001. (I–K) Images of H&E-stained whole spleens from NI (I), 17XNL-infected mice on day 1 pi (J, D1) and day 18 pi (K, D18) under 2× magnification. (L–P) Images of mouse spleens from NI mice (L) and 17XNL-infected mice on day 1 pi (M, D1), day 4 pi (N, D4), day 10 pi (O, D10), and day18 pi (P, D18) under 40× magnification. Black arrows indicate white pulp regions. (Q) Images of H&E-stained spleen from a 17XNL-infected mouse on D18 under 100× magnification. Light blue circles indicate cell clusters suggesting EBIs with none or a few mature erythrocytes. Yellow arrows point to malaria parasite pigments seen as brown color.

Extramedullary Erythropoiesis, Splenomegaly, and Blockage of RBC Maturation.

Malaria infection is often associated with splenomegaly, hypersplenism, or hyperreactive malaria splenomegaly syndrome (29). The spleen has been known as a key organ that responds to malaria infections early, playing an important role in “filtering” and clearing iRBC and nRBCs (or deformed RBCs), and splenomegaly is also a hallmark of malaria (30–33). Indeed, 17XNL infection led to splenomegaly starting from day 4 pi and peaked at day 18 pi with spleens weighing approximately 10 times larger than those of NI mice (Fig. 1 G and H). H&E staining of the splenic tissues did not detect obvious changes in the size or structure of the spleen on day 1 pi compared with those of NI mice, with clear separation of white pulp (black arrows) and red pulp regions (Fig. 1 I, J, L, and M). From days 4 to 18 pi, the number of RBCs in the red pulp regions decreased, and the structural separation of white and red pulps gradually disappeared (Fig. 1 K and N–P). At the same time, there were groups of cells with dark nuclei in the red pulp regions from days 10 to 18 pi, indicative of extramedullary erythropoiesis and/or expansion of immune cells (Fig. 1 O–Q; cyan circles). Similar to the BM, mature RBCs were almost absent in the spleen; again, a large number of malaria pigments were observed in the spleen (yellow arrows in Fig. 1Q) due to macrophage phagocytosis of iRBCs. Additionally, the total cell counts of T and B cells, DCs, and macrophages were also increased from day 10 pi, suggesting activation of immune cells and immune responses (SI Appendix, Fig. S4 A and B). The active extramedullary erythropoiesis and expansion of immune cells contributed to the observed splenomegaly in the infected mice.

Reduced Terminal Erythropoietic Subpopulations after Parasite Infection.

To investigate the mechanisms of malarial anemia, we characterized erythroblast subpopulations to evaluate the process of erythropoiesis in vivo by staining late erythroblast stages using cell markers as described previously (34). CD71 (transferrin receptor 1) and TER119 (erythrocyte-specific antigen) are expressed at different stages of erythropoietic cell differentiation (35). Three late-stage cell populations were previously distinguished using the cell markers and the forward scatter (FSC) parameter that is a function of cell size (36). In our hands, EryA (TER119^highCD71^highFSC^high) are mostly erythroblasts with a nucleus in the spleen and BM (absent in the blood); EryB (TER119^highCD71^highFSC^low) contain infected and NI reticulocytes (mostly NI reticulocytes in the BM and spleen), whereas EryC (TER119^highCD71^lowFSC^low) are mainly mature RBCs (SI Appendix, Fig. S5 A–D). We counted these three cell populations in the blood, BM, and spleen of NI and infected mice using flow cytometry as previously described (34). In the blood, both counts and frequencies of EryA, EryB, and EryC populations were similar for NI and infected mice before day 10 pi, with most of the cells being EryC population (Fig. 2 A and B). By days 18 and 22 pi, most of the cells were EryB populations (infected and NI reticulocytes) in the infected mice. For the BM on day 1 pi, EryB and EryC were the major cell populations in both NI and infected mice (Fig. 2 C and D). The EryC population started to decrease from day 4 pi and was almost absent by day 18 pi in the infected mice. The numbers and proportions of the EryA population increased from day 10 pi, but the EryB population remained about the same. For the spleen, EryC was the major population on days 1 and 4 pi; on day 10 pi, both EryA and EryB increased; by day 18 pi, the EryC population was dramatically reduced (almost absent) (Fig. 2 E and F). The spleen in the infected mice became an active erythropoietic center with significantly increased EryA and EryB populations compared to NI mice. The lack of EryC on days 18 and 22 pi in the BM and the spleen as well as the dramatic reduction of EryC in the blood of infected mice suggest either a blockage of EryB to EryC conversion and/or rapid destruction of the EryC population. Because the peak reticulocytemia in the infected mice was approximately 20% of RBC counts before infection (SI Appendix, Fig. S2B), both destruction of mature RBCs and inhibition/reduction of erythropoiesis could contribute to the reduction of RBCs.

Fig. 2.

Flow cytometry analysis of erythroblast subpopulations of noninfected (NI) and P. yoelii yoelii 17XNL-infected mice. (A and B) Counts (A) and frequencies (B) of different developmental erythropoietic cell populations (EryA, EryB, and EryC) from the blood of NI and 17XNL-infected mice on day 1 (D1), day 4 (D4), day 10 (D10), day 18 (D18), and day 22 (D22) pi. (C and D) The same cell counts and frequencies as in A and B but from BM. (E and F) The same cell counts and frequencies as in A and B but from the spleen. Kruskal–Wallis test (n = 3): *P < 0.05; **P < 0.01 (matching the colors of the bars).

Cytokine and Chemokine Environment Supporting Hematopoiesis after 17XNL Infection.

Many cytokines such as IFN-γ, TNF-α, and IL-12 can affect erythropoiesis (37). To investigate the dynamics of inflammatory and anti-inflammatory cytokines/chemokines during 17XNL infection, we measured plasma levels of IL-1α, IL-4, IL-5, IL-10, IL-12p40/p70, IL-13, IL-17, GM-CSF (granulocyte-macrophage colony-stimulating factor), IFN-γ, FGF-basic (fibroblast growth factor), IP-10 (interferon-inducible protein 10), KC (keratinocytes-derived chemokine), MCP-1 (monocyte chemoattractant protein 1), MIG (monokine induced by gamma or CXCL9), MIP-1α (macrophage inflammatory protein 1 alpha), and TNF-α every other day up to day 18 pi. Except for IL-5, IL-12p40/p70, MIG, and MIP-1α which still had elevated levels on day 10 pi, the levels for the majority of cytokines/chemokines increased early in the infected mice (days 4 to 6 pi) and declined to background levels after days 8 to 10 pi (Fig. 3), including those known to inhibit erythropoiesis such as IL-1α, IL-10, IL-13, IL-17, IFN-γ, and TNF-α (38–40). Interestingly, the level of IL-12p40/p70 that may promote erythropoiesis (41) was elevated after infection, declined after day 5 pi, but increased again after day 10 pi, paralleling the increasing numbers of reticulocytes in the infected mice (Fig. 3 and SI Appendix, Fig. S1A). The observations of increasing IL-12 and declining levels of IFN-γ and TNF-α at day 10 pi suggest an environment promoting erythropoiesis at day 10 during 17XNL infection.

Fig. 3.

Dynamics of cytokines and chemokines during 17XNL infection. Plasma levels of cytokines and chemokines from noninfected (NI) and 17XNL-infected mice were measured using a mouse cytokine kit (Invitrogen, Waltham, MA, USA) and Luminex 200 instrument according to the manufacturer’s instructions (Invitrogen). Names of cytokines and chemokines are labeled in each subfigure. Kruskal–Wallis test (n = 5 for 17XNL infected mice and n = 3 for NI mice); *P < 0.05; **P < 0.01; ***P < 0.001.

Activation of Macrophages and Altered Surface Marker Expression after Malaria Infection.

We next investigated the potential mechanism of RBC deficiency within a cytokine environment favoring erythropoiesis. An EBI consists of different stages of erythroblasts and a central macrophage that provides necessary signals and nutrition for erythroid cell development (21). A defect in macrophage function could impair erythroblast development. To investigate the functional states of macrophages, we first stained the BM and the spleen sections from NI and 17XNL-infected mice on day 18 pi with an antibody to F4/80, a glycoprotein expressed on various mouse macrophages. F4/80 expression increased greatly in the red pulp of infected mice (SI Appendix, Fig. S6 A–D). Clusters of heavily stained cells (brown) were visible in the red pulp of infected mouse spleens (SI Appendix, Fig. S6D, green circles). Similarly, increased F4/80 expression levels were observed in the BM of infected mice at day 18 pi compared to NI mice (SI Appendix, Fig. S6 E–H). Increased expressions of iNOS (M1 macrophage marker) and CD206 (M2 marker) were also observed in the spleen and BM of infected mice (SI Appendix, Fig. S7). In particular, the increase in the staining of BM day 18 pi by anti-iNOS antibody was observed, suggesting activation of M1 macrophages (SI Appendix, Fig. S7 E and G). These observations suggest the presence of proinflammatory macrophages that phagocytose the parasites but may become defective in supporting erythropoiesis.

EBI macrophages from BM and fetal liver expressing EPOR have been shown to have higher levels of CD106 (vascular cell adhesion protein 1 or VCAM1), CD163, CD169, CD11b (integrin alpha M), and ER-HR3 (antibody recognizing hematopoiesis associated macrophages) than macrophages that do not express EPOR (42). In another study, F4/80, CD106, and CD169, but not CD11b, were found to be heterogeneously expressed by the central macrophages within the EBIs (43). Although not all the F4/80⁺CD106⁺CD169⁺ macrophages are EBI macrophages, EBI macrophages are often characterized by the expression of EPOR in both mice and humans (24). We, therefore, counted cells expressing F4/80, EPOR, CD106, and CD169 as potential EBI macrophages from the spleen and BM of NI and 17XNL-infected mice on days 4, 10, 18, and 40 pi (Fig. 4). For the BM, the percentages of live cells expressing all four markers significantly increased on days 4, 10, and 40 pi compared with those of NI mice (Fig. 4A). For individual markers, the percentages of EPOR⁺ cells were also significantly increased in the infected mice compared with those of NI mice (Fig. 4B). In contrast, the percentages of CD169⁺ cells were significantly decreased on days 10 and 18 pi (Fig. 4C). The percentages of F4/80⁺ and CD106⁺ were about the same between NI and infected mice (Fig. 4 D and E). For the spleen, the percentages of cells expressing all four markers were lower than those of BM (~1 to 3% vs. 0.01 to 0.02%), and parasite infection appeared to reduce the percentages of cells expressing all four markers (Fig. 4F). Similar to those in the BM, the percentages of EPOR⁺ cells were significantly higher in the infected mice (Fig. 4G) but lower for CD169⁺ cells (days 4, 10, and 18 pi) (Fig. 4H). The percentages of F4/80⁺ and CD106⁺ cells were significantly higher in the spleen of the infected mice than in NI mice on days 10 and 18 pi (Fig. 4 I and J). The dynamics of cell counts from BM were generally similar to the dynamics of cell percentages (SI Appendix, Fig. S8 A–E); the counts for cells expressing all four markers in the spleen were generally higher for the infected mice than NI mice, particularly on days 10 and 18 pi due to splenomegaly (SI Appendix, Fig. S8 F–K). However, the numbers of potential EBI macrophages expressing all four markers in the BM samples were much higher than those of the spleen, and the increase in EBI macrophages in the spleen did not make up for the loss of EBI macrophages in the BM (SI Appendix, Fig. S8 A and F; note only BM cells from the femur and tibia of rear legs and cells from the whole spleen were counted). The decrease in BM cells expressing all four markers was largely due to the loss of cells expressing CD169 (Fig. 4C and SI Appendix, Fig. S8C). Selective depletion of CD169⁺ macrophages using G-CSF concomitantly depleted F4/80⁺VCAM-1⁺CD169⁺ER-HR3⁺Ly-6G⁺ EBI macrophages and blocked erythropoiesis (25). The loss of cells expressing CD169 could affect erythropoiesis and contribute to anemia during 17XNL infection.

Fig. 4.

Dynamics of erythroblast island (EBI) macrophages from the spleen and BM during 17XNL infection. Cells from the spleen and BM of noninfected (NI) and P. yoelii yoelii 17XNL-infected mice (17XNL) were isolated on days 4, 10, 18, and 40 pi and were stained using antibodies specific for surface proteins for flow cytometry analysis as indicated. (A–E) Percentages of live cells expressing F4/80⁺EPOR⁺CD106⁺CD169⁺ (indicative of EBI macrophages) or individual markers in the BM of NI and 17XNL-infected mice. (F–J) Percentages of live cells expressing F4/80⁺/EPOR⁺CD106⁺CD169⁺ (indicative of EBI macrophages) or individual markers in the spleen of NI and infected mice. Mann–Whitney U test (n = 5): *P < 0.05; **P < 0.01.

Macrophages from Infected Mice Are Defective in Supporting MEL Cell Differentiation In Vitro.

After 17XNL infection, the spleen becomes an active erythropoietic organ. To functionally investigate the ability of macrophages to support erythropoiesis, we isolated macrophages from the spleens of NI and 17XNL-infected mice and observed the formation of cell clusters each containing a macrophage and differentiating mouse erythroleukemia (MEL) cells in vitro. Day 4 after the addition of MEL cells (1 × 10⁶), differentiating MELs (black arrows) adhering to healthy macrophages with clear nuclei (blue arrows) and well-defined cytoplasm were observed in cultures with macrophages from NI mice (Fig. 5 A and B). In contrast, most macrophages from 17XNL-infected mice contained Hz particles (red arrows) and were either breaking apart (Fig. 5C) or displaying a shrinkage of cytoplasm (Fig. 5D), suggesting dying or dead macrophages. On day 6, most of the macrophages from NI mice still looked healthy with differentiating MEL cells (Fig. 5 E and F), but most of the macrophages from 17XNL-infected mice disappeared, leaving residual malarial pigments, cell debris, or no traces of macrophages (Fig. 5 G and H). We counted cell clusters containing a macrophage and three or more MEL cells under a microscope. On day 4 culture with cells from NI mice, 73 out of 86 cell clusters (84.9%) contained healthy-looking macrophages that featured a well-defined nucleus and cytoplasm, but only seven out of 77 (9.1%) cell clusters had macrophages with identifiable nucleus and cytoplasm in the culture with cells from 17XNL-infected mice, leading to significant differences in surviving EBIs (χ² = 93.4, df = 1; P < 0.0001). On day 6, 81.0% (85/105) of the cell clusters from splenic cells of NI mice looked healthy with well-defined nucleus and cytoplasm, whereas only 5.7% (5/88) of cell clusters in the culture with cells from 17XNL-infected mice had identifiable macrophages, again having significant differences in surviving EBIs (χ² = 109, df = 1; P < 0.0001). These results show that macrophages from the infected mice were damaged and were unable to support normal differentiation of erythroblasts. To investigate the mechanisms of macrophage death, we counted apoptotic and necrotic macrophages isolated from the spleens and BMs of NI and infected mice using an apoptosis/necrosis assay kit (Abcam). Significantly more macrophages from the spleens and BMs of infected mice were apoptotic than those of NI mice (Fig. 5 I and J). Similarly, significantly more macrophages from the infected spleen were necrotic than those of NI mice, although the percentage of necrotic macrophages was small (Fig. 5J). Very few necrotic macrophages were observed in the infected and NI BMs. The results suggest that apoptosis was the main mechanism of macrophage death in the spleen and BM.

Fig. 5.

Erythroblast island (EBI) differentiation in vitro and in vivo supported by macrophages from noninfected mice (NI) and 17XNL-infected mice. (A–H) Macrophages from NI and 17XNL-infected mice day 15 pi were seeded on coverslips in culture plates for 1 to 2 d to allow adherence to the coverslips. MEL cells (1 × 10⁶) were added to the cultures, and images of cell clusters were taken on day 4 and day 6 after the addition of MEL cells. Blue arrows point to nuclei of macrophages, red arrows point to malaria Hz pigments, and black arrows indicate differentiating erythroblasts. (A–D) Images from day 4 post addition of MEL cells; (E–H) images from day 6 post addition of MEL cells. (A, B, E, and F) Cell clusters using splenic cells from NI mice; (C, D, G, and H) cell clusters from 17XNL-infected splenic macrophages. (I and J) Percentages of apoptotic and necroptotic cells, respectively, from the NI and 17XNL-infected spleens (SP) and BM. (K and L) Representative images of EBIs isolated from NI mice. (M and N) Representative images of EBIs from infected mice on day 14 pi. (O) Plots of erythroblast counts per macrophage from NI and 17XNL-infected mice on days 4, 10, 18, and 42 pi. Mann–Whitney U test (n = 5); *P < 0.05; ***P < 0.001; ns, not significant.

Lower Erythroblast Binding Capability of Macrophages from Infected Mice.

We next isolated EBIs from the BM of NI and infected mice on days 4, 14, 18, and 42 pi using methods described previously (44). For cells isolated from NI mice, many cell clusters with differentiating and maturing erythroblasts surrounding a central macrophage were observed, suggesting normal maturing or mature EBIs (Fig. 5 K and L). For cells from infected mice on day 14 pi, the isolated EBIs had fewer erythroblasts attached to the central macrophages, and there were more free erythroblasts detaching from the macrophages during the isolation process (Fig. 5 M and N). We randomly counted the erythroblasts attached to the macrophages (a macrophage with at least three or more erythroblasts) from NI and infected mice on days 4, 14, 18, and 42 pi and showed significantly fewer erythroblasts from the 17XNL-infected BM than those of the NI mice on days 4, 14, and 18 pi (Fig. 5O). These observations suggest potentially lower binding affinity of the erythroblasts to the macrophages from infected mice, which could be due to lower or lack of CD169 expression on the macrophages of infected mice. CD169 is an adhesion molecule expressed by EBI macrophages, and CD169⁺ macrophages have been shown to promote late erythroid maturation (24, 45).

Macrophages from Infected Mice Were Defective in Supporting Erythropoiesis In Vivo.

To further confirm the functional defect of macrophages in vivo, we depleted macrophages by treating NI mice with clodronate liposomes, reconstituted with equal numbers of BM or splenic macrophages from NI and infected mice, treated the mice with pyrimethamine (PYR) to clear parasites, and then induced erythropoiesis using phenylhydrazine (PHZ) (Fig. 6A). Before the experiments, the effectiveness of PYR in drinking water (70 mg/L) was tested to make sure no live parasites remained in the mice reconstituted with macrophages from infected mice (Fig. 6B). Water with PYR was then provided to all mouse groups on the day of reconstitution, including mice receiving macrophages from NI mice to control for potential effects of PYR on erythropoiesis. Additionally, we also stained the macrophages with the DiR dye (1,1′-Dioctadecyl-3,3,3′,3′-Tetramethylindotricarbocyanine Iodide; Thermo Fisher, Waltham, MA) and injected them into mice to show proper homing of injected macrophages to the BM (~8% of injected cells. Note only cells from femur and tibia of rear legs were counted) and spleen (~23%) (Fig. 6 C and D and SI Appendix, Fig. S9). After PHZ treatment, hemoglobin content decreased in all mouse groups until day 4 when hemoglobin levels in the PBS-liposome control mice began to increase (Fig. 6E). Hemoglobin levels in the group reconstituted with BM macrophages from NI mice began to increase day 6 post-PHZ treatment, but not those reconstituted from BM macrophages of infected mice; the differences in hemoglobin levels between these two groups were significant on days 8, 10, and 12 post-PHZ treatment (Fig. 6E). No difference was observed for mice reconstituted with splenic macrophages from NI, infected, and non-reconstituted mice (Fig. 6F). This observation is not surprising because the NI macrophages are not activated for supporting erythropoiesis in the spleen without active infection. Similarly, RBC counts were significantly different between the mice reconstituted with BM macrophage from NI and infected mice on day 8 and day 10 post-PHZ treatment (Fig. 6G); again, no significant difference was observed between mice reconstituted with splenic cells of NI and infected mice (Fig. 6H). Similar results were observed for reticulocyte counts (Fig. 6 I and J). These results show that reconstitutions with BM or spleen cells from 17XNL-infected mice do not improve hemoglobin levels or RBC cell counts (similar to those from mice without reconstitution), supporting the in vitro finding that macrophages from infected mice are defective in supporting erythropoiesis.

Fig. 6.

Macrophage depletion and reconstitution with BM and spleen cells show defective macrophages in erythropoiesis. (A) Diagram of the experimental procedure for macrophage depletion and reconstitution. NI, noninfected; pi, postinfection; Macro reconst, macrophage reconstitution; PYR, pyrimethamine; PHZ, phenylhydrazine. 1, Infect mice (n = 5) with 17XNL parasites and set up NI control (n = 5); 2, Day 16 pi, collect cells from the BM and spleen; 3, One day before cell harvest, treat mice (n = 10) with clodronate to deplete macrophages; 4, Inject macrophages from step 2 into mice with depleted macrophages; 5, Treat mice with PYR in drinking water; 6, Treat mice with PHZ to stimulate erythropoiesis and measure hemoglobin level and blood cells on the day of PHZ treatment and every other day after. (B) PYR in drinking water (7 mg/L) completely killed parasites in the reconstituted mice. (C and D) Cell counts and percentages of macrophages recovered from the spleen and BM after DiR (1,1′-Dioctadecyl-3,3,3′,3′-Tetramethylindotricarbocyanine Iodide) staining in vitro and injection (iv, 1 × 10⁶) into mice. Cells from the spleen and BM were counted using flow cytometry. (E) Hemoglobin levels (grams/deciliter) day 0 to day 12 post-PHZ treatment. Clo_17XNL BM, clodronate treated and reconstituted with BM cells (4 × 10⁷) from 17XNL-infected mice; Clo_NI BM, clodronate treated and reconstituted with BM cells (4 × 10⁷) from NI mice; Lipo control, liposome in PBS (no macrophage depletion); Clo_no rec, clodronate treated without reconstitution. (F) The same treatments as in (E) but reconstituted with splenic cells (Sp). (G and H) the same experiments as in (E) and (F), but blood cells were counted. (I and J) The same treatments as in (E) and (F), but reticulocytes were counted on days 8 and 10 post-PHZ treatment. Mann–Whitney test U (n = 4 to 5); *P < 0.05.

CD169 Blockage Impairs Erythroblast Binding to Macrophages and Erythrocyte Maturation.

CD169⁺ macrophages were found to promote late erythroid maturation and control erythroblast retention in the BM (45). To investigate the roles of CD169 in erythroblast binding to macrophages and in erythrocyte maturation, we also measured the expression levels of CD169 on F4/80⁺EPOR⁺ macrophages from the BMs and spleens of NI and 17XNL-infected mice days 4, 10, 18, and 40 pi. Significantly reduced mean fluorescent intensities were observed in the BM of infected mice on days 10 and 18 pi and in the spleen on days 4, 10, 18, and 40 pi (Fig. 7 A and B). We next performed in vitro assays of MEL binding to BM macrophages isolated from NI and 17XNL-infected mice (SI Appendix, Fig. S10A) on day 15 pi at the presence of anti-CD169 (4 mg/mL antibody 3D6.112 plus 4 mg/mL antibody SER4) or control antibodies (anti-fluorescein, #Ab00102-7.1). Significantly lower numbers of bound MELs were observed in the groups treated with anti-CD169 antibodies than the groups of nontreated or treated with the control antibodies on days 4 and 6 postincubation for macrophages isolated from NI mice but not from 17XNL-infected mice possibly due to low-level expression of CD169 on macrophages from infected mice (Fig. 7C). We also isolated EBIs from BM of infected mice, treated the EBIs with EDTA to strip erythroblasts from the macrophages, and cultured the cells with anti-CD169 antibodies (4 mg/mL or 8 mg/mL) or control antibodies for 24 h after washing off EDTA. EDTA treatment stripped erythroblasts from macrophages successfully (SI Appendix, Fig. S10B), and many erythroblasts were reattached to the macrophages after the removal of EDTA (SI Appendix, Fig. S10C). We then randomly counted macrophages with three or more erythroblasts and found that macrophages from both NI and 17XNL-infected mice treated with anti-CD169 antibodies had significantly fewer erythroblasts than those without treatment or treated with control antibodies (Fig. 7D and SI Appendix Fig. S10D). To investigate the role of CD169 in erythropoiesis in vivo, we injected anti-CD169 antibodies (40 μg SER-4 plus 40 μg 3D6.112 per mouse on days 3 and 5 pi) or control antibodies (the same amounts as anti-CD169) into mice and counted EryA, EryB, and EryC using flow cytometry on day 10 pi; no difference was observed between groups treated with anti-CD169 and control antibodies (SI Appendix, Fig. S10 E and F). The failure of in vivo blockage was likely due to insufficient amounts of antibodies used because 200 mg SER-4 plus 100 mg 3D6.112 antibodies were used in a previous study of CD8⁺ T cell and macrophage interaction (46). In a repeat experiment, we used higher dosages of anti-CD169 (50 mg SER-4 plus 100 mg 3D6.112 and 60 mg SER-4 plus 120 mg 3D6 per mouse on days 3 and 5 pi, respectively) or control antibodies (the same amounts as anti-CD169). Significantly higher percentages of EryB (reticulocyte) but lower percentages of EryC (RBC, despite not being significant) were found in anti-CD169 antibody-treated mice than in control antibody-treated mice (Fig. 7 E and F), suggesting an impaired transition from reticulocytes to RBCs after anti-CD169 treatment. We also isolated EBIs from BM of the antibody-treated mice, treated the cells with EDTA, cultured the cells for 24 h after removing EDTA, and counted erythroblasts on macrophages randomly. Again, significantly fewer erythroblasts per macrophage were observed in the in vivo anti-CD169 antibody-treated group than in the group treated with control antibodies (Fig. 7G). These results show that CD169 indeed plays an important role in erythropoiesis (or anemia), particularly in reticulocyte maturation into RBCs during malaria parasite infections.

Fig. 7.

Anti-CD169 antibody treatment reduces erythroblast binding to macrophages and erythrocyte maturation. (A and B) CD169 expression on F4/80⁺EPOR⁺ macrophages from the BM and spleen of noninfected (NI) and 17XNL-infected mice, as detected using flow cytometry. (C) The numbers of MEL cells bound to BM macrophages isolated from NI and infected mice. The MEL cells were added to macrophage cultures with or without the treatment of anti-CD169 or control antibodies (anti-fluorescein). Macrophages with three or more MEL cells were counted randomly. (D) The numbers of erythroblasts bound to macrophages isolated from the BM of NI and infected mice. EBIs were isolated from NI and infected mice, treated with EDTA, and cultured in the presence of anti-CD169, control antibodies, or no antibody (None) for 24 h after washing off EDTA. Erythroblasts were counted as in C. (E and F) Anti-CD169 antibody treatment in vivo reduces reticulocyte to erythrocyte maturation in the BM and spleen. Mice were injected with anti-CD169 or control antibodies on days 3 and 5 pi. The percentages of EryA, EryB, and EryC populations were counted and calculated using flow cytometry as described in the Materials and Methods. (G) The numbers of erythroblasts bound to macrophages isolated from the BM of mice treated with anti-CD169 or control antibodies. The assays were done as in D, except for no in vitro antibody treatment. Mann–Whitney test U (n ≥ 5); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Discussion

This study investigated the mechanism of malaria anemia in C57BL/6n mice after infection with 17XNL parasites and revealed that defective EBI macrophages play an important role in anemia and splenomegaly (Fig. 8). Destruction or removal of RBCs from circulation will trigger erythropoiesis; indeed, the spleen responded to the RBC loss by extensive cell replication in the red pulp from days 4 to 18 pi. However, a defect in erythropoiesis due to dysfunctional EBI macrophages prevented the production of sufficient erythrocytes to replenish the lost RBCs. With active parasite infection, the body continued to receive signals for more RBC production and responded by expanding erythroblasts and other immune cells in the spleen, leading to splenomegaly. Suppression of late erythroid precursors was also observed in mice infected with blood-stage Plasmodium chabaudi AS parasites (47). The impaired erythroblast maturation was associated with a shift in CD71 expression from the TER119(+) population to the B220(+) population (47), although the molecular mechanism of this shift is unknown. Dyserythropoiesis has also been observed in human malaria patients infected with P. falciparum and Plasmodium vivax (10, 48), but the mechanism of dyserythropoiesis in human infections remains elusive. Here, we showed that splenic macrophages from infected mice could bind to MEL cells but were not able to support MEL cell differentiation in vitro. In vivo, macrophages from NI mice could significantly enhance hemoglobin levels and reticulocyte counts as well as RBC production in clodronate-treated mice, but not those from infected mice. Additionally, a reduced level of CD169 expression was observed in 17XNL-infected mice, which could alter the functional state or capability of the macrophages to support erythropoiesis. Depletion of CD169(+) macrophages was shown to block erythropoiesis in mouse BM (25), and CD169-CD43 interaction plays a role in EBI formation and erythroid differentiation (49). Indeed, we also showed that antibody blockage of CD169 on macrophages reduced the binding of erythroblasts and MEL cells in vitro and impaired RBC maturation from reticulocytes. However, there are some differences between human and murine erythrocytes and erythropoiesis such as variations in RBC life span, oxygen affinity, membrane proteins, signaling pathways, and regulation of ion content and stress erythropoiesis (50). Although further investigations are required to clarify whether the macrophage dysfunction, RBC maturation blockage, and splenomegaly observed in this 17XNL-C57BL/6n mouse model are relevant to human malaria, the frequent reports of anemia and splenomegaly in patients of malaria-endemic regions suggest some common mechanisms between rodent and human malaria (12, 51).

Fig. 8.

A proposed mechanism of malarial anemia by inhibition of erythroblast maturation and macrophage dysfunction. (A) In noninfected mice, there is normal erythropoiesis in the BM without splenomegaly. (B) In malaria, parasite-infected mice, phagocytosis of iRBCs and nRBCs triggers immune responses, cytokine production, dysfunction of macrophages, and a blockage of erythroblast maturation in the BM, leading to anemia. (C) The reduction in RBCs induces extramedullary erythropoiesis in the spleen with replication of erythroblasts and expansion of erythroblast islands (EBIs). Similar to BM, phagocytosis of iRBCs triggers cytokine production and modifies macrophage functions, manifested as increased expression of F4/80 and decreased expression of molecules such as CD169, which also impair EBI macrophage’s ability to bind erythroblasts and support erythrocyte maturation (EryB to EryC transition). As a feedback mechanism, the continuous decline in RBCs signals for more erythroblasts and EBI expansion, leading to splenomegaly. Clearance of parasites restores macrophage’s ability to support RBC maturation and resolution of anemia.

In addition to erythropoiesis, macrophages play an important role in the phagocytosis of iRBCs and the production of inflammatory cytokines and chemokines (52). The elevated levels of F4/80, iNOS, and CD206 on macrophages, particularly in the spleen, suggest an active role of these cells in immune responses. Large amounts of parasite Hz pigments were observed in the macrophages of infected mice, and malaria Hz has been shown to inhibit erythropoiesis leading to anemia (28). Phagocytosis of malaria digestive vacuoles can also drive monocytes into a state of immunological exhaustion and reduced microbicidal activity (53). High-level expression of F4/80 in the BM (although the surface expression of F4/80 detected using flow cytometry was lower on days 10 and 18 pi) and spleen macrophages may also enhance immune suppression through activation of regulatory T cells (Tregs). F4/80 plays an essential role in the development of antigen-specific Tregs in immune tolerance and immunosuppression (54). The increase in F4/80 expressions can contribute to malaria-induced immune suppression by promoting Treg activity (55, 56). Indeed, serum levels of many cytokines and chemokines including IFN-γ, IL-10, IL-17, and TNF-α were down-regulated on day 10 pi with 17XNL parasites.

It will be critical to further investigate the metabolic and immunologic states of the BM and splenic macrophages after malaria parasite infection to dissect the molecular mechanisms of macrophage dysfunction in erythropoiesis and immune responses.

Another important contributor to malaria anemia is the loss of RBCs. In our model, RBC counts declined continuously until day 18 pi, even though the 17XNL parasites generally do not invade mature RBCs. Various mechanisms have been shown to contribute to malaria anemia, including the removal of iRBCs and nRBCs (2, 3, 12, 57). Except for P. falciparum and Plasmodium knowlesi infections that may have parasitemia higher than 10%, P. vivax and other species infecting humans rarely exceed 2% parasitemia (58), suggesting that the loss of iRBCs may not be the major mechanism for the rapid decrease in RBC counts or SMA (3). More nRBCs than iRBCs are lost during P. falciparum and P. vivax infections (59, 60). nRBCs can be removed by phagocytosis due to reduced RBC deformability, changes to RBC surface molecules, binding of parasite components, immunoglobulin (anti-PS), or complement to nRBCs, and/or immune mechanisms such as CD8⁺ T cell–dependent clearance (13, 26, 61–64). In the 17XNL-C57BL/6n mouse model, reduction in mature RBC counts and hematocrit from days 4 to 18 pi suggests removal of nRBC as the parasites preferentially invade reticulocytes. Our data show that iRBCs can trigger significantly higher rates of phagocytosis by BMDMs or BMDCs than RBCs from NI mice in vitro. Additionally, anti-PS antibodies could enhance the phagocytosis of nRBCs. Therefore, clearance of iRBCs (or infected reticulocytes) and/or nRBCs are potential mechanisms contributing to the total loss of RBCs. However, the mechanism of nRBCs removal is not clear because our in vitro phagocytosis showed only limited phagocytosis of nRBCs by BMDCs, which may activate some immune signaling pathways. Phagocytosis through increased surface exposure of PS or lysis by neutrophils, CD8⁺ T cells, and NK cells may also contribute to nRBC loss (13, 65, 66). Nonetheless, our study reveals a mechanism of malarial anemia through the inhibition of late-stage erythropoiesis by dysfunctional macrophage with reduced CD169 surface expression (Fig. 8). Restoration of EBI macrophage functions to support RBC maturation may represent an effective treatment of malarial anemia.

Materials and Methods

Malaria Parasites and Mice.

The 17XNL parasite and the procedures for infecting C57BL/6n mice were as described previously (67). All animal procedures were performed following the protocol approved (#LMVR11E) by the Institutional Animal Care and Use Committee at the National Institute of Allergy and Infectious Diseases (NIAID).

RBC Count, Hemoglobin Content, and Hematocrit.

Mice were injected with 1 × 10⁶ parasites; blood samples from mouse tails were collected every other day, stained 1:1 with Trypan blue, and counted on a Nexcelom cell counter (Nexcelom Bioscience, Lawrence, MA). Infected blood cells, reticulocytes, and RBCs were also measured by counting Giemsa-stained thin blood smears. For hemoglobin measurements, ~10 µL whole blood from the mouse tail was collected into a microcuvette and measured on the HemoCue Hb 201 System (HemoCue AB, Ängelholm, Sweden). Hematocrit levels were measured by filling one-quarter of BD Clay Adams™ SurePrep capillary tubes (VWR International, Radnor, PA, USA) with whole blood, centrifuged on the Hematocrit Centrifugette 4203 for 3 min, and read on a Critocaps microhematocrit tube card.

Flow Cytometry.

Cell preparations from the blood, spleen, and BM were processed for flow cytometry using antibodies listed in SI Appendix, Table S1. Stained cells were analyzed on an LSRII flow cytometer (BD Biosciences, Billerica, MA, USA) using BD FACSDiva and BD FlowJo (V10) or on MACSQuant Analyzer 16 (Miltenyi Biotec, Bergisch Gladbach, Germany) with MACSQuantify Software. The gating strategies for different cell populations in various experiments are summarized in SI Appendix, Fig. S11. Immune cells were identified using various molecular markers listed in SI Appendix, Table S1. Populations of erythroblasts EryA, EryB, and EryC were gated using cell markers and criteria reported previously (34).

Histopathology.

For immunohistochemistry and histopathology, spleen and BM samples were processed according to methods described previously (68) using antibodies listed in SI Appendix, Table S1.

In Vitro Phagocytosis.

In vitro, phagocytosis of iRBCs and nRBCs was performed as described previously (26, 69).

Isolation of EBIs.

Potential EBIs were isolated from NI and 17XNL-infected BM using the density gradient method as described previously (44).

In Vitro Erythroblast Differentiation Assay.

Macrophages isolated from the BM and spleens of NI and 17XNL-infected mice were allowed to adhere to coverslips in a 6-well plate with 2 mL DMEM with 30% FBS and 1% antibiotic–antimycotic overnight. Murine erythroleukemia (MEL, 1 × 10⁶) cells were added after washing off the nonadhered cells. EBI formation and MEL differentiation were observed under a microscope on days 4 and 6 after MEL cell addition.

Macrophage Reconstitution and Erythropoiesis Assay.

Mice were infected with 17XNL parasites, and cells from the spleen (5 × 10⁷) or BM (4 × 10⁷) of NI and infected mice were injected into clodronate-liposome-treated mice (200 µL 1 d before reconstitution) after lysis of RBCs. The mice were treated with PYR (70 mg/L in drinking water) starting on the day of cell injection to kill parasites from infected mice, and blood smears were examined to confirm the absence of parasites before PHZ treatment (one-time injection of 300 µL at 6 mg/mL in PBS). Hemoglobin levels and RBC/reticulocyte counts were measured every other day from the day of the PHZ injection.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.