Plasmodium products persist in the bone marrow and promote chronic bone loss

See allHide authors and affiliations

Science Immunology  02 Jun 2017:
Vol. 2, Issue 12, eaam8093
DOI: 10.1126/sciimmunol.aam8093

Plasmodium leftovers cause bone loss

Individuals who recover from malarial infection may develop long-term consequences, such as bone loss and growth retardation. Lee et al. now report that Plasmodium by-products retained in the bone marrow lead directly to bone loss. Infection with a mutant Plasmodium that lacked the by-product hemozoin did not induce bone loss. Mechanistically, these products induced MyD88-dependent inflammatory responses in osteoclast and osteoblast precursors, resulting in bone resorption. Treating infected animals with alfacalcidol, a vitamin D3 analog, could prevent this bone loss, suggesting that combining bone therapies with antimalarial drugs may prevent bone loss in infected individuals.


Although malaria is a life-threatening disease with severe complications, most people develop partial immunity and suffer from mild symptoms. However, incomplete recovery from infection causes chronic illness, and little is known of the potential outcomes of this chronicity. We found that malaria causes bone loss and growth retardation as a result of chronic bone inflammation induced by Plasmodium products. Acute malaria infection severely suppresses bone homeostasis, but sustained accumulation of Plasmodium products in the bone marrow niche induces MyD88-dependent inflammatory responses in osteoclast and osteoblast precursors, leading to increased RANKL expression and overstimulation of osteoclastogenesis, favoring bone resorption. Infection with a mutant parasite with impaired hemoglobin digestion that produces little hemozoin, a major Plasmodium by-product, did not cause bone loss. Supplementation of alfacalcidol, a vitamin D3 analog, could prevent the bone loss. These results highlight the risk of bone loss in malaria-infected patients and the potential benefits of coupling bone therapy with antimalarial treatment.


Malaria caused by Plasmodium parasites is a life-threatening infectious disease that kills at least half a million people annually while causing more than 200 million new infections. In some cases, complications can quickly develop, such as cerebral malaria, respiratory distress, and severe anemia, often leading to death. Despite these severe complications, most patients recover from the disease. However, there is evidence that malaria survivors experience long-term “hidden” pathologies that are as yet poorly defined. For instance, physical growth retardation in young children in Africa is associated with infectious diseases, with a high prevalence among malaria-infected children regardless of nutritional status (1, 2). An increased incidence of porous bone lesions has also been reported in malaria-endemic regions, suggesting that infection may compromise bone integrity (3). Despite the importance of bone tissue in health and development, and the known interaction between Plasmodium parasites and bone marrow cells whereby parasites circulate, reside (4), and infect (5), the pathology of malaria in bone is poorly understood.

Bone tissue constantly undergoes formation by osteoblasts (OBs) and resorption by osteoclasts (OCs) in a tightly regulated process. OC differentiation is initiated when receptor activator of nuclear factor ĸB (NFκB) ligand (RANKL; encoded by Tnfsf11), constitutively expressed on OBs, binds to the RANK receptor on the surface of OC precursors (OCPs) from monocyte/macrophage lineage (6). The RANKL-RANK interaction leads to recruitment of the adaptor molecule tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6) and downstream activation of NFκB, c-fos, and AP-1, leading to transcription of the master regulator of osteoclastogenesis nuclear factor of activated T cells 1 (NFATc1). This results in the formation of mature OCs, which express tartrate-resistant acid phosphatase (TRAP) (7). Aside from its role in osteoclastogenesis, RANKL is an important immunoregulatory molecule (8). Up-regulation of RANKL expression on OBs and immune cells leads to bone destruction during bacterial infections, such as osteomyelitis and gingivitis, and osteoporosis in HIV, implicating a tight relationship between immune activation and bone remodeling (911). Plasmodium infection induces robust immune activation and invasion of parasites into the bone marrow, further suggesting the harmful potential of malaria on bone.

Here, we addressed the direct effect of malaria infection on bone. We reported that the Plasmodium parasite and its associated products produced during and after malaria infection cause chronic inflammation leading to bone loss. We identified alfacalcidol, a vitamin D3 (VitD3) analog, as a successful drug for the treatment of bone problems caused by malaria infection.


Plasmodium infection causes bone loss and bone growth retardation

To address the effect of Plasmodium infection on bone, we used two mouse malaria models, Plasmodium yoelii nonlethal (PyNL) and Plasmodium chabaudi chabaudi (Pcc). PyNL infection, which resembles Plasmodium vivax infection in humans and preferentially invades reticulocytes (12), was used to determine the effect of a single self-cleared Plasmodium infection on bone homeostasis, whereas Pcc infection, which more closely resembles Plasmodium falciparum infection in humans (12), was used to evaluate the effect of low-level chronic infection on bone homeostasis. Six-week-old adult mice were infected with these parasite species, and changes in bone were assessed during the acute (peak parasitemia, day 8 or 14), convalescent (clearance, day 30), and chronic (day 90) phases of infection (Fig. 1A and fig. S1A). Microcomputed tomography (μCT) analysis showed that PyNL- and Pcc-infected mice had significantly reduced trabecular bone volume and number and increased trabecular bone spacing, which continued for over 60 days after parasite clearance, as confirmed by PyNL 18S ribosomal RNA (rRNA) quantitative polymerase chain reaction (qPCR) (Fig. 1, B to D, and fig. S1, B and C). Recovery from anemia by day 30 excluded anemia’s contribution to further bone loss (Fig. 1E).

Fig. 1 Plasmodium infection results in bone loss and bone growth retardation.

(A) Parasitemia in peripheral blood of PyNL-infected mice (n = 6 to 16 mice per group). (B) Gene expression of PyNL 18S rRNA in whole bone (n = 5 per group). (C) Representative μCT images of distal femurs. (D) Bone morphometric analysis of distal femurs by μCT. Bv/Tv, bone volume/total volume; Tb.N, trabecular bone number; Tb.Th, trabecular bone thickness; Tb.Sp, trabecular bone spacing. w.o., weeks old. (E) Whole-blood count. RBC, red blood cell.; n.s., not significant. (F) Femurs of 6-week-old naïve and PyNL-infected mice. (G) Length measurement of young mice femurs. (H and I) Representative μCT images and bone morphometric analysis of distal femurs of naïve and PyNL-infected young mice at 6 (H) and 9 (I) weeks old after recovery from infection. Each data point represents individual mouse and are represented as means ± SD. Experiments were repeated at least twice. *P < 0.05, **P < 0.01, Mann-Whitney test (A, D, G, H, and I) and Kruskal-Wallis with Dunn’s posttest (E).

Increasing evidence suggests that Plasmodium infection may result in growth stunting of young children in malaria-endemic regions (summarized in table S1) (1, 2, 1319). To assess the effect of infection on bone growth at young age, we infected 3-week-old mice with PyNL (fig. S1D). At this age, mice are undergoing active bone development before epiphyseal closure. Young mice sacrificed 3 weeks after infection, at 6 weeks old, had significantly shorter femurs and lower trabecular bone volume compared with naïve littermates (Fig. 1, F to H). In mice sacrificed 6 weeks after infection, 3 weeks after parasite clearance, femur length had recovered to normal levels (fig. S1E); however, trabecular bone volume remained significantly reduced (Fig. 1I). Together, these findings suggest that both chronic low-level infection and a single self-cleared infection result in a substantial chronic effect on bone tissue, leading to bone loss and bone development defects.

Bone homeostasis is disrupted after Plasmodium infection

To further investigate bone dynamics at a cellular level, we used the PyNL infection model for the following studies. Bone histomorphometric analysis revealed total suppression of both OC and OB numbers and significantly reduced bone formation rate during acute PyNL infection (Fig. 2, A and B). The OC marker TRAP and the OB marker alkaline phosphatase (ALP) were also significantly reduced during acute infection, supporting the notion that bone remodeling is suppressed, and presumably, bone growth is halted during this phase of infection (Fig. 2C). In contrast, after the clearance of parasites on day 30, osteoblastic and osteoclastic parameters returned to normal levels, whereas by day 90 after infection, bone resorption rate was elevated compared with naïve conditions (Fig. 2, A to C). To further examine the effect of Plasmodium infection on OC differentiation, we stimulated bone marrow cells from naïve and PyNL-infected mice with RANKL in vitro. Cells from day 14 PyNL-infected mice rarely differentiated into TRAP+ multinucleated OCs based on cell number and expression of Acp5 (the gene encoding TRAP) (Fig. 2D), whereas cells taken from day 30 recovered mice differentiated into OCs more readily than cells taken from naïve mice (Fig. 2E). Together, these results suggest that both OBs and OCs were suppressed during acute infection leading to bone loss and attenuation of bone growth, whereas after parasite clearance, bone loss was most likely attributable to enhanced OC activity rather than OB impairment.

Fig. 2 Bone remodeling is suppressed during acute Plasmodium infection but is highly activated after parasite clearance.

(A and B) Bone histomorphometric analysis of tibia sections and OB and OC numbers per bone surface (A) and bone formation and bone resorption rates (B). (C) Serum levels of TRAP and ALP. OD405nm, optical density at 405 nm. (D and E) Number of OCs and mRNA expression of Acp5 in OC culture of naïve and PyNL-infected mice on day 14 (D) and on day 30 (E). (F) Schematic diagram of OC signaling. (G) Gene expression of cultured OCs. (H) Serum cytokine levels. (I) Acp5 mRNA expression of a 3-day differentiated OC culture incubated with hemin for 6 hours. (J) Course of PyNL infection in Jdp2+/− and Jdp2−/− mice (n = 7 to 8 mice per group). (K) Representative μCT images and bone morphometric analysis of distal femurs of naïve and PyNL-infected Jdp2+/− and Jdp2−/− mice. (L) Gene expressions of Jdp2, Nfatc1, and Acp5 in indicated mouse tibia. Naïve control groups were indicated as day 0 PyNL infection. Data are means ± SD; each point represents individual mouse and is repeated at least twice with similar results. *P < 0.05, **P < 0.01, ***P < 0.001, Kruskal-Wallis with Dunn’s posttest (A to C, G, H, K, and L) and Mann-Whitney test (D and E).

TRAF6, Jdp2, and NFATc1 signaling are disrupted during acute Plasmodium infection

The RANK-RANKL signaling pathway is essential for the differentiation of OCs; we therefore investigated whether modulation of this pathway could explain changes in osteoclastogenesis observed during infection (Fig. 2F). The expression of Tnfrsf11a (encoding RANK) was comparable between OCPs derived from naïve and PyNL-infected mice at all time points tested (Fig. 2G). However, Traf6 expression was down-regulated in cells derived from acutely infected mice, with concomitant suppression of downstream signaling molecules Jdp2, Nfatc1, and Acp5 (Fig. 2, D and G).

We next addressed possible Plasmodium-related factors involved in the suppression of OCs in vivo. Several cytokines, as well as hemin, have been reported to be involved in the inhibition of OC differentiation (20). We detected increased interferon-γ (IFNγ), interleukin-4 (IL-4), IL-10, and IL-13 levels in the serum during acute infection compared with naïve controls (Fig. 2H). In addition, acute hemolysis during Plasmodium infection results in substantial heme release from ruptured erythrocytes (20, 21). We confirmed in vitro that OC differentiation was directly suppressed by hemin as indicated by reduced Acp5 expression, highlighting a potential role for heme in the suppression of osteoclastogenesis during acute infection (Fig. 2I).

To further evaluate the role of OCs in bone remodeling during malaria infection, we infected Jdp2−/− mice with PyNL. As previously reported, Jdp2−/− mice were osteopetrotic due to osteoclastogenesis defect (22). Despite similar course of infection, bone loss was not observed in these mice during or after infection (Fig. 2, J and K). Moreover, mice exhibited increased bone volume during convalescence (Fig. 2K). This highlights a key role for OCs in malaria-induced bone loss. In accordance with this, OC markers and inflammatory cytokines were up-regulated in the bone during convalescence, indicating OCs activation (Fig. 2L and fig. S2). Together, these findings suggest that Plasmodium infection causes suppression of bone remodeling during acute infection due to increased cytokines and acute heme release, whereas elevated Jdp2, Nfatc1, and Acp5 expression in the bone during convalescence leads to enhanced OC activity and chronic bone loss.

Accumulation of Plasmodium products into the bone marrow alters bone marrow niche

We next investigated possible factors contributing to increased osteoclastogenesis during convalescence. Black discoloration of femurs was prominent during the convalescent and chronic phases of infection, but not during the acute phase, suggesting that Plasmodium products gradually accumulate in the bone and remain there long term (Fig. 3, A and B). To further understand the interaction of these Plasmodium products with bone marrow cells, we cultured OCs and calvarial OBs in vitro with P. falciparum crude extract (PfCE), which consists of parasite proteins, nucleic acid, and hemozoin. PfCE did not inhibit OC maturation on the basis of TRAP expression (Fig. 3C); however, both OCPs and primary calvarial OB precursors (OBPs) responded strongly to PfCE and PyCE, with dose-dependent induction of the inflammatory cytokines IL-1α, IL-1β, IL-6, and TNFα (Fig. 3, D and E, and fig. S3). These inflammatory cytokines robustly promote OC formation via OB RANKL induction (23, 24). PfCE significantly induced Tnfsf11 expression (encoding RANKL) in mature OBs (Fig. 3F and fig. S3). Together, these data suggest that sustained infiltration of Plasmodium products in the bone after parasite clearance directly induces proinflammatory and osteoclastogenic cytokines from both OCPs and OBPs, which in turn may promote osteoclastogenesis via amplification of RANKL.

Fig. 3 Persistence of Plasmodium products in the bone elicits inflammation and amplifies RANKL.

(A) Images of femurs from naïve and PyNL-infected mice. (B) Representative images of bone histology sections and bone marrow smears of naïve and PyNL-infected mouse tibias. Scale bars, 20 μm (B). Arrows show the accumulated birefringent brownish hemozoin. (C) Acp5 expression of a 3-day differentiated OC culture with PfCE containing 0, 4, 10, and 20 × 106 infected RBCs (iRBCs) for 24 hours. (D) Cytokine expression in OCP and a 3-day differentiated OC culture after a 6-hour stimulation with PfCE containing 0, 4, 10, and 20 × 106 iRBCs. (E) Cytokine expression in OBP and differentiated OB culture after a 6-hour stimulation with PfCE containing 0, 10, and 20 × 106 iRBCs. (F) Tnfsf11 expression in OB culture after a 24-hour stimulation with PfCE containing 20 × 106 iRBCs. All experiments were repeated at least twice. Data are means ± SD. *P < 0.05, Mann-Whitney test.

MyD88 controls Plasmodium-induced chronic bone inflammation

Given the role of Plasmodium products in inducing inflammatory cytokines and RANKL expression, consequently leading to the activation of OBs, OCs, and their precursors, we investigated the possible involvement of Plasmodium-immune recognition machinery in bone loss. Plasmodium products have been reported to induce immune responses involving MyD88, TLR9 (Toll-like receptor 9), NLRP3, and IL-1R (25, 26). Infection of mice deficient in these genes showed that Myd88−/− mice, but not Tlr9−/−, Il1r−/−, or Nlrp3−/− mice, were protected from PyNL-induced bone loss (Fig. 4, A and B). Although bone volume negatively correlated with parasitemia, such correlation was not found in Myd88−/− mice (fig. S4A). The lack of bone loss in Myd88−/− mice was unlikely because of osteoclastogenesis or osteoblastogenesis impairment because Myd88−/− OCs and OBs formed normally in vitro (fig. S4B). Rather, we found that Plasmodium product–induced proinflammatory cytokine expression in OCPs and OBPs was Myd88-dependent (Fig. 4, C and D), leading to a lack of RANKL expression in OBs under stimulatory conditions (Fig. 4E). This indicates that MyD88 signaling is essential for Plasmodium-induced bone loss through its role in bone cell inflammation and concomitant RANKL expression.

Fig. 4 MyD88 controls malaria-induced chronic bone inflammation and bone loss.

(A and B) Bone morphometric analysis of femurs of naïve and PyNL-infected Myd88+/− and Myd88−/− (A) and Tlr9−/−, Il1r−/−, and Nlrp3−/− (B) mice. (C) Relative mRNA expression of cytokines in Myd88+/− and MyD88−/− OCP culture stimulated with PfCE at 0, 200, 500, and 1000 μg/ml for 6 hours. (D) Cytokine expressions in Myd88+/− and Myd88−/− OBP culture stimulated with PfCE at 0, 200, 500, and 1000 μg/ml for 24 hours. (E) Tnfsf11 expression in Myd88+/− and Myd88−/− OB culture stimulated with PfCE at 0, 200, 500, and 1000 μg/ml for 24 hours. All experiments were repeated at least twice. Each data point represents individual mouse. Data are means ± SD. *P < 0.05, **P < 0.01, Kruskal-Wallis test and Dunn’s post-test (A) and Mann-Whitney test (B).

Plasmodium products are involved in malaria-induced bone pathology

Because it is difficult to determine the full moiety of accumulated Plasmodium products in the bone, we further investigated one of the components, hemozoin, a heme detoxification by-product of Plasmodium, which we identified abundantly in the bone (Fig. 3B). In contrast, parasites and parasite nucleic acids could not be detected in the bone during convalescence (Fig. 1B). To investigate the involvement of hemozoin in chronic bone loss, we used a mutant Plasmodium berghei ANKA (PbA) parasite strain, Δpm4Δbp2, which is deficient in the pm4 and bp2 genes encoding the plasmepsin-4 and bergheipain-2 hemoglobinases, respectively (27). This mutant has impaired hemoglobin digestion and produces little or no hemozoin. Unlike the PbA strain, this mutant strain loses virulence and mice are able to clear infection (27); therefore, its course of infection resembles that of PyNL (Fig. 5A). To compare bone loss in PbA versus Δpm4Δbp2 infection, it was necessary to treat PbA-infected mice with chloroquine to control parasitemia to prevent death. Bone marrow smears taken during convalescence confirmed a marked reduction in hemozoin accumulation in PbAΔpm4Δbp2-infected mice (Fig. 5A). Compared with PbA-infected mice that had significant bone loss despite chloroquine treatment (fig. S5), Δpm4Δbp2 infection did not cause bone loss, despite a higher parasitemia during acute infection (Fig. 5, B and C). To assess the exact role of Plasmodium products but not other factors such as cytokines or anemia on the bone pathology, we used Rag2−/− mice that were unable to resolve the infection because of lack of T and B cells. Similar to the observation in wild-type (WT) mice, Δpm4Δbp2 infection in Rag2−/− mice did not cause bone loss despite having similar parasitemia, chronic anemia, and cytokines with PbA infection (Fig. 5, D to F, and figs. S6 and S7). Furthermore, in vitro studies showed that Δpm4Δbp2 crude extract induced lower inflammatory responses and RANKL expression in OB culture compared with PbA crude extract (PbACE) stimulation (Fig. 5G).

Fig. 5 Involvement of Plasmodium products in malaria-induced bone disorder.

(A) Bone marrow smears of PbA- and Δpm4Δbp2-infected mice during convalescence. Scale bars, 20 μm. Arrows show the accumulated birefringent hemozoin. (B) Course of infection with WT and mutant parasites. Red arrows indicate the time of chloroquine (CQ) treatment. (C) Representative μCT images and bone morphometric analysis of distal femurs of WT mice. (D to F) Bone volume of distal femurs (D), blood parasitemia (E), and blood count (F) of Rag2−/− mice upon sacrifice. (G) Cytokine and Tnfsf11 expressions in OB culture after a 16-hour stimulation with crude extracts containing 1, 2, and 5 × 107 infected erythrocytes. (H) Transmission electron microscopy (TEM) images of phagocytosis of sHZ by OCP and mature OC in in vitro cultures. Red arrows show the phagocytosed sHZ. (I and J) Cytokines expression in Myd88+/− and Myd88−/− OCP and OC cultures (I) and OBP and OB cultures (J) after a 6-hour stimulation with sHZ at 20, 50, and 100 μg/ml. (K) Tnfsf11 expression in OB culture after a 24-hour stimulation with sHZ. Data are means ± SD, repeated at least twice with similar results. *P < 0.05, **P < 0.01, ***P < 0.001, Kruskal-Wallis test and Dunn’s post-test (C, D, and F) and Mann-Whitney test (E).

To specifically investigate the direct role of hemozoin in bone, we used synthetic hemozoin (sHZ) crystals (28) as a proxy for Plasmodium-produced hemozoin. We found that both OCPs and mature OCs were capable of phagocytizing sHZ in vitro (Fig. 5H). sHZ stimulated inflammatory cytokine production in OCPs and OBPs, which further induced RANKL expression in OBs in a Myd88-dependent manner (Fig. 5, I to K). Overall, these data suggest the involvement of hemozoin in the MyD88-dependent bone inflammation and resultant chronic bone loss after malaria infection.

Malaria-induced chronic bone loss is rescued by alfacalcidol supplementation

VitD is an important molecule for bone metabolism and calcium homeostasis and is also involved in immunomodulation (29). VitD insufficiency occurs in human malaria (30), and possible beneficial role of VitD supplementation for cerebral malaria has been reported (31). To address whether VitD supplementation would alleviate bone loss caused by Plasmodium infection, we treated mice with alfacalcidol, a VitD3 derivative used to treat osteoporosis because of its high efficacy and safety (32, 33). Treatment of alfacalcidol orally at 100 to 400 ng/kg was sufficient to prevent malaria-induced bone loss, whereas higher doses (400 to 1000 ng/kg) resulted in enhanced bone growth despite infection (Fig. 6, A to C). VitD treatment also suppressed PfCE-induced inflammation in OCPs and OBPs in vitro, suggesting that it may act by preventing overactivation of osteoclastogenesis during infection (Fig. 6, D and E). Alfacalcidol treatment not only improved bone status, it also suppressed parasite growth at higher doses (Fig. 6F). In vitro P. falciparum inhibition assay showed that alfacalcidol reduced parasite growth dose-dependently (Fig. 6G). Together, these results suggest that alfacalcidol administration might be beneficial to reduce parasite burden while preventing bone loss.

Fig. 6 Malaria-induced bone loss can be prevented by alfacalcidol treatment.

(A) Experimental design of alfacalcidol treatment after PyNL infection. (B and C) Representative μCT images (B) and bone morphometric analysis (C) of distal femur of naïve and PyNL-infected mice with and without alfacalcidol treatment. (D and E) Cytokine expressions of OCP (D) and OBP (E) culture stimulated with PfCE at 1 mg/ml) for 6 hours in the presence of 10, 50, and 100 μM alfacalcidol. (F) Parasitemia of PyNL-infected mice with and without alfacalcidol treatment. (G) In vitro antimalarial effect of alfacalcidol on P. falciparum (3D7) culture at indicated doses. Each data point represents individual mouse. Data are means ± SD, repeated at least twice. *P < 0.05, **P < 0.01, Kruskal-Wallis test with Dunn’s posttest.


The long-term pathological consequences of chronic malaria infection are poorly understood. Recent study has suggested that chronic exposure to Plasmodium parasites promotes prolonged immune activation and enhances the risk of lymphoma formation (34). Another unforeseen outcome of malaria is the evidence that chronically infected birds experience enhanced telomere degradation (35). These are likely only few among many potential deleterious outcomes triggered by chronic malaria infection. In particular, the possible association between growth retardation and malaria infection in humans (table S1) suggests that infection may have a negative impact on bone remodeling and growth. However, there is a limitation in studying the direct effect of malaria infection on bone in humans due to various coexisting factors such as malnutrition and other infections. Using well-established mouse models mimicking various aspects of human Plasmodium infection, we showed here that infection causes significant and long-term bone loss in adult mice and growth retardation in young mice. Bone remodeling is completely suppressed during acute infection but is highly activated immediately after parasite clearance, with increased osteoclastic activity skewing the balance toward bone resorption. OCs are activated by the key osteclastogenic cytokine RANKL, which was up-regulated in OBs through MyD88-dependent inflammation, triggered by the persistence of parasite products in the bone marrow (Fig. 7).

Fig. 7 Proposed schematic mechanism of malaria-induced bone loss.

During acute Plasmodium infection, several malaria infection–related factors may have a deleterious effect on bone tissue, resulting in the suppression of both OC and OB differentiation. One example molecule is heme, which is released abundantly during hemolysis of infected and noninfected erythrocytes and strongly inhibits both OC and OB. Also, cytokines such as IL-4, IL-10, IL-13, and IFNγ released from activated T cells in response to Plasmodium infection disrupt the signaling in osteoclastogenesis. During convalescent and chronic phase of Plasmodium infection, although parasites are cleared systemically, parasite products remain and gradually accumulate in the bone marrow, leading to MyD88-dependent chronic inflammation in OCPs, OBPs, and mature OBs. Plasmodium product–induced inflammatory responses synergistically up-regulate RANKL expression on OBs, which further promotes osteoclastogenesis. The robust activation of OCs skews bone remodeling toward bone resorption and causes chronic bone loss.

The alteration of bone remodeling has been observed in other infectious and inflammatory conditions (36), suggesting that the mechanism of bone loss in malaria might be multifactorial. Induction of IL-4, IL-10, IL-13, and IFNγ by activated T cells in response to infection likely disrupts OC signaling. The suppression of RANK-RANKL signaling from the upstream TRAF6 during acute Plasmodium infection could be due to increased IFNγ (37). An additional factor unique to Plasmodium infection might be the release of heme due to malaria-induced hemolysis (21), which may directly contribute to the impairment of OC and OB differentiation (3840). Hence, multiple infection-related factors likely cooperate to cause suppression of bone remodeling during acute malaria.

A key finding of our study is that, although Plasmodium infection is resolved systemically, the bone marrow evolves into a state of chronic inflammation that could be associated with the persistent accumulation of Plasmodium products in the bone marrow. Although low-level chronic infection is a common feature across many pathological conditions, persistence of pathogen by-products in tissues is a unique feature of malaria (41). Plasmodium products (i.e., proteins, hemozoin, and nucleic acids) are biologically active compounds capable of modulating immune responses (25, 26, 42). Here, we showed that bone marrow OCs and OCPs engulf Plasmodium products including hemozoin, thereby eliciting strong inflammatory cytokine production. Although secretion of IL-1α, IL-1β, IL-6, and TNFα is vital in the regulation of immune cell infiltration and activation for the elimination of pathogens, persistent production of these cytokines can lead to inflammatory osteoporosis by increasing OC activity (43). These cytokines act synergistically with minimal amounts of RANKL to promote robust osteoclastogenesis (23, 24, 44) and may additionally induce osteoclastogenesis independently of RANK-RANKL signaling (44). In support of the hypothesis that enhanced osteoclastogenesis is a key factor in malaria-induced bone loss, Jdp2−/− mice, which lack mature OCs (22, 45), did not experience bone loss during infection. Overall, phagocytosis of Plasmodium products, even in the absence of active infection, can lead to chronic proinflammatory cytokine release from OCPs and OBs in bone. These cytokines may synergistically amplify RANKL expression, leading to overproduction of OCs and an imbalance in bone resorption and bone formation, ultimately resulting in bone loss.

Our data further suggest that Myd88 is involved in this bone pathology. Plasmodium products fail to elicit an inflammatory response in Myd88−/− OBs and OCPs, and Myd88−/− mice had intact bones. TLR9, NLRP3, or IL-1 signaling alone had a minimal role in bone loss, suggesting that an as yet unknown Plasmodium recognition machinery upstream of MyD88 may be involved or possibly several simultaneous signaling pathways may act through MyD88 to induce bone loss. One possible Plasmodium component that contributes to this pathological bone loss is hemozoin. The absence of bone loss in mice infected with a mutant Plasmodium parasite with strongly reduced hemozoin production and reduced virulence (27) gives rise to two possible mechanisms: (i) a direct effect of hemozoin on bone loss or (ii) the involvement of Plasmodium antigens as virulence factors in bone loss. Plasmepsin-4 and bergheipain-2 proteases are involved in the degradation of hemoglobin. Deficiency of these two proteases causes a significant reduction of hemozoin production due to inefficient hemoglobin digestion, yet the parasites can still survive by possibly acquiring proteins from other sources (27). Besides the activation of bone cells by hemozoin, the concurrent recognition of parasite proteases by another receptor may also involve Myd88 signaling (46), which may further synergize to contribute to the pathologic bone disorder. Because hemozoin persists in the bone marrow, it can possibly sustain the immune response initiated by the parasite antigens, given that our group had previously shown the adjuvanticity of hemozoin (28, 42). Further work is needed to determine which mechanism is involved.

Although chloroquine is a common antimalarial and has a protective effect on bone (47), it failed to prevent malaria-induced bone loss. Alternatively, we showed that alfacalcidol, a VitD3 derivative, protects Plasmodium-infected mice from losing bone. Alfacalcidol has been safely and effectively used to treat osteoporosis by promoting osteoblastic activity and suppressing OC formation (32), possibly by decreasing the RANKL expression on OB (48), and restricting the pool of OCPs in the bone marrow (49). VitD has also been shown to have immunomodulatory effects, including suppression of inflammation (50). Consistent with this, our data showed the suppression of Plasmodium product–induced inflammatory cytokines in vitro. VitD derivatives, including alfacalcidol, can restrict parasite growth directly (51). Our in vivo and in vitro data support this notion, but further work is needed to understand the mechanism(s) involved in the antiplasmodial effect of alfacalcidol.

On the basis of our current findings, we postulate that malaria infection and resultant Plasmodium products accumulation in the bone marrow modulate both the immune response and bone homeostasis, which eventually leads to bone loss. Sustained bone loss does not seem to ameliorate even after recovery from malaria, suggesting that malaria-infected patients are likely at risk of bone deterioration. These observations suggest that antimalarials coupled with bone therapy may be beneficial in improving bone health in malaria-infected individuals.


Study design

This study aimed at the characterization of the effects of malaria infection on bone by using various Plasmodium spp. in mice. Animal experiments were carried out according to the guidelines of the Immunology Frontier Research Center and Research Institute for Microbial Diseases of Osaka University and the National Institutes of Biomedical Innovation, Health and Nutrition. Age- and gender-matched littermates of the same strain were used in each experiment. Experimental replicates were indicated in the figure legends. Blinded bone histomorphological analysis was conducted by the Niigata Bone Science Institute. No other randomization was performed in animal experiments.


Female BALB/c and male C57BL/6J WT mice were purchased from CLEA Japan. Female BALB/c Myd88−/− and Tlr9−/− mice as well as male C57BL/6J Rag2−/−, Il1r−/−, and Nlrp3−/− and female Jdp2−/− mice were used as described (22, 52).


sHZ was produced as described (28). Hemin (Fluka) was dissolved in 0.01 M NaOH. PfCE was prepared from schizont stage–infected erythrocytes (42). PbACE and PbAΔpm4Δbp2 crude extract (Δpm4Δbp2CE) were prepared from infected mice blood at high parasitemia. Leukocytes and platelets were removed using Plasmodipur filters (EuroProxima). Crude extracts were subjected to five cycles of freeze-thaw to release the Plasmodium components. Alfacalcidol (Wako) was dissolved in absolute ethanol to 20 mg/ml and further diluted in medium-chain triglyceride (Nisshin OilliO).

Plasmodium infection and treatments

Mice were inoculated intraperitoneally with 1 × 105 PyNL-, Pcc-, PbA- (12, 53), or PbAΔpm4Δbp2-infected erythrocytes (27). PbA-infected mice were treated with chloroquine (50 mg/kg) (Sigma-Aldrich) intraperitoneally, as indicated. PyNL-infected mice were sacrificed on days 14, 30, and 90 after infection; Pcc-infected mice on days 8, 30, and 90 after infection; and PbA- and PbAΔpm4Δbp2-infected WT mice on day 30 after infection. PbA- and PbAΔpm4Δbp2-infected Rag2−/− mice were sacrificed on days 27 and 33 after infection, respectively, at similar parasitemia. EDTA-treated blood was analyzed by complete blood counter (Horiba LC-662). Three-week-old PyNL-infected mice were treated with chloroquine (25 mg/kg), as indicated. Femur length was measured using a digital micrometer caliper (As One Japan). Mice were orally administered with alfacalcidol (400 and 1000 ng/kg) daily from day 2 to day 14 after infection, and the doses were then reduced to 100 and 400 ng/kg from day 11 up to day 30.

Bone morphometric analysis

Ethanol-fixed femurs were scanned by a three-dimensional μCT using Scan-Xmate RB080SS110 scanner (Comscan) and analyzed using TRI/3D-Bon software (Ratoc System Engineering). For bone histomorphometric analysis, mice were injected intraperitoneally with calcein (16 mg/kg) (Dojindo) in 2% sodium hydrogen carbonate at 96 and 24 hours before sacrifice. Methylmethacrylate-embedded tibia samples were sectioned and analyzed by the Niigata Bone Science Institute, Japan.

In vitro culture of bone marrow OC

Bone marrow–derived macrophages (OCPs) were maintained with macrophage colony-stimulating factor (M-CSF) (10 ng/ml) (PeproTech) in α-minimum essential medium (α-MEM) supplemented with 10% fetal calf serum (54). OCs were formed after a 4-day stimulation with M-CSF (10 ng/ml) and RANKL (50 ng/ml) (R&D Systems). Cells were fixed and stained with an acid phosphatase staining kit (Sigma-Aldrich). TRAP+ multinucleated OCs (more than three nuclei) were counted under light microscope, and images were captured at ×10 magnification. Cells were stimulated for 6 hours, as indicated. In some experiments, alfacalcidol was added into the culture 1 hour before stimulation.

In vitro calvarial OB culture

Calvarial OB was isolated from newborn mice by sequential digestion in α-MEM containing 0.1% collagenase and 0.2% dispase at 37°C with continuous shaking (54). Calvarial cells were cultured in RPMI 1640 supplemented with an OB inducer reagent (Takara). Culture medium was replenished every 3 days. After 20 days, cells were stimulated for 6 or 24 hours as indicated.

Gene expression analysis by quantitative reverse transcription PCR

RNAs were extracted from cultured cells and whole bone. Gene expressions were analyzed by qPCR and normalized to the expression level of 18S rRNA, as described previously (12). The expression level of P. yoelii 18S rRNA (Py18S) was measured using a customized probe and primers (53, 55).

Statistical analyses

Data are expressed as means ± SD. Data points represent individual mouse. Statistical significance between two groups was analyzed using Mann-Whitney test. Kruskal-Wallis test and Dunn’s posttest were performed when more than two groups were compared. Significances are represented as *P < 0.05, **P < 0.01, and ***P < 0.001. All statistical analyses were performed using GraphPad Prism 5.0.


Materials and Methods

Fig. S1. Bone loss in P. chabaudi infection and suppression of bone growth in PyNL-infected young mice.

Fig. S2. Gene expression of inflammatory cytokines in the bone.

Fig. S3. PyNL crude extract elicits inflammation in OCPs, OBPs, and OBs.

Fig. S4. Comparison between Myd88+/− and Myd88−/− mice.

Fig. S5. Chloroquine does not rescue Plasmodium-infected mice from bone loss.

Fig. S6. Serum cytokines of Rag2−/− mice infected with WT PbA and mutant PbAΔpm4Δbp2.

Fig. S7. Bone marrow infection of PbA-infected and mutant Δpm4Δbp2-infected mice.

Table S1. Possible association between malaria infection and child growth.

Table S2. Excel file containing source data in tabular format for all figures.


Acknowledgments: We thank K. Honjo for insights into public data evaluation, and K. Matsuda and H. Omori for help with TEM. Funding: This work was supported by the Grants-in-Aid for Scientific Research (B) (grant no. 16H05181 to C.C.), the Japan Agency for Medical Research and Development (to C.C. and K.J.I.), the Uehara Memorial and Yamada Science Foundations (to C.C.), the Grant-in-Aid for Young Scientists (A) (grant no. 15H05686 to K.M.). M.S.J.L. is the recipient of the Japanese Government Scholarship (Ministry of Education, Culture, Sports, Science and Technology). Author contributions: M.S.J.L. conducted all experiments and statistical analyses. K.M. helped with bone studies; Y.F. with qPCR; and A.K., P.M.L., and S.I. with infections. K.M., T.H., E.K., S.A., K.J.I., J.-w.L., and S.M.K. contributed critical reagents. M.S.J.L. and C.C. wrote the manuscript, and C.C. provided overall supervision. All authors contributed to critical revision of the manuscript. Competing interests: The authors declare that they have no competing financial interests.

Stay Connected to Science Immunology

Navigate This Article