ABSTRACT
The trematode Schistosoma mansoni is one of the etiological agents of schistosomiasis, a key neglected tropical disease responsible for an estimated annual loss of 70 million disability-adjusted life years. Hematophagy represents the primary nutrient acquisition pathway of this parasite, but digestion of hemoglobin also liberates toxic heme. Schistosomes detoxify heme via crystallization into hemozoin, which is subsequently regurgitated into the host's circulation. Here we demonstrate that during experimental schistosomiasis, hemozoin accumulating in the mouse liver is taken up by phagocytes at a time coincident with the development of the egg-induced T-helper 2 (Th2) granulomatous immune response. Furthermore, the uptake of hemozoin also coincides with the hepatic expression of markers of alternative macrophage activation. Alternatively activated macrophages are a key effector cell population associated with protection against schistosomiasis, making hemozoin well placed to play an important immunomodulatory role in this disease. To systematically explore this hypothesis, S. mansoni hemozoin was purified and added to in vitro bone marrow-derived macrophage cultures concurrently exposed to cytokines chosen to reflect the shifting state of macrophage activation in vivo. Macrophages undergoing interleukin-4 (IL-4)-induced alternative activation in the presence of hemozoin developed a phenotype specifically lacking in Retnla, a characteristic alternatively activated macrophage product associated with regulation of Th2 inflammatory responses. As such, in addition to its important detoxification role during hematophagy, we propose that schistosome hemozoin also provides a potent immunomodulatory function in the coevolved network of host-parasite relationships during schistosomiasis.
INTRODUCTION
Schistosomiasis, caused by trematodes of the genus Schistosoma, is one of the core neglected tropical diseases that continually reinforce the poverty trap in which the “bottom billion” of the globe live (1). Recent estimates suggest that of the 200 million individuals infected with schistosomes in Africa alone, 20 million suffer from severe debilitating manifestations of the disease, causing an estimated annual loss of 70 million disability-adjusted life years, a disease burden in excess of that caused by malaria (2).
In the experimental murine model of schistosomiasis (caused by infection with Schistosoma mansoni), the adult worms reside in the mesenteric venules of the host intestine, where they lay several hundred eggs each day (3). Approximately half of these are swept away by the circulation and become lodged in the host liver. Here they elicit a Th2 granulomatous response which ultimately causes fibrosis of the liver and much of the symptomology associated with schistosomiasis (4). However, a fully functional Th2 response to parasite eggs, although ultimately pathogenic, is also essential for host survival (5, 6). Specifically, the interleukin-4 (IL-4)-induced alternative activation of macrophages seems to play a crucial host-protective role in this regard (7). Therefore, careful modulation of the anti-egg immune response is critical for host survival and the consequent persistence of the parasite life cycle. One schistosome product that may be involved in this important immunomodulatory process is hemozoin.
Hematophagy represents the primary route of nutrient acquisition by S. mansoni adults and is a process that also liberates toxic heme (8). To compensate for this potentially lethal insult, free heme is neutralized in the schistosome gut via its crystallization into hemozoin (9). Hemozoin crystals spontaneously nucleate at the surfaces of neutral lipid droplets and grow to form characteristic spherical aggregates (10, 11), which are subsequently regurgitated into the host circulation and accumulate in the liver (12). Due to the important immune-mediated circumoval events in the liver, schistosome hemozoin seems well positioned to exert modulatory activity during chronic infection. Interestingly, hemozoin produced by another hematophagous parasite, Plasmodium spp., has already emerged as a leading candidate for the generalized immune suppression often observed in the spleen during malaria (13). In this disease, Plasmodium hemozoin is released into the host circulation as infected erythrocytes rupture. Once consumed by macrophages or dendritic cells, this apparently inert crystalline substance has been shown to exert a number of immunosuppressive as well as some proinflammatory effects in vitro and in vivo (14–16). Whether Schistosoma hemozoin has similar robust immunomodulatory effects is not yet clear (17, 18). However, analyses of its effects on alternative macrophage activation and function are likely to be important in understanding both the regulation of immunopathogenesis and the apparently immunoprotective role of this cell population during chronic schistosomiasis.
In this investigation, we demonstrate that during the deposition of S. mansoni hemozoin in the host liver, it is taken up by phagocytes coincident with the development of Th2 responses, the expression of macrophage alternative activation genes, and granulomatous pathology. Thus, we propose that hemozoin is both temporally and spatially well placed to modulate important immunopathogenic events in the liver during chronic schistosomiasis. We further demonstrate that purified S. mansoni hemozoin modulates the alternative activation of macrophages in vitro by specifically downregulating the expression and secretion of Retnla (also known as Fizz1 [19] or Relmα [20]), a factor known to be involved in the regulation of Th2 responses during experimental schistosomiasis (21–23). As such, hemozoin not only represents an important detoxification mechanism in this parasite but also may be essential in modulating the host-protective granulomatous response.
MATERIALS AND METHODS
Unless otherwise stated, all chemicals and reagents were purchased from Sigma-Aldrich, United Kingdom.
Experimental infection.All procedures involving the use of experimental animals were approved by the Aberystwyth University Ethics Committee and the UK Home Office. The life cycle of a Puerto Rican strain of S. mansoni was maintained via routine passage through Biomphalaria glabrata snails and C57BL/6 mice (Harlan, United Kingdom) as previously described (24). For experimental infection, 6- to 8-week-old female mice were infected percutaneously with 50 cercariae/mouse. Groups of animals (5 or 6 mice/group) were sacrificed at 2, 4, 5, 6, 7, 8, 10, 12, and 14 weeks postinfection. Blood samples were collected via cardiac puncture for the preparation of serum. Approximately 1 mg of liver tissue was collected from the periphery of the left lobe and stored in TRIzol reagent (Invitrogen, United Kingdom) for later preparation of RNA. The vasculature was then perfused for the collection of worms (24). Finally, remaining liver and spleen tissues were harvested for downstream parasitological, histological, and cellular analyses.
Parasitology.The number of adult worms collected in the vascular perfusate of each infected mouse was recorded. Liver tissue from each mouse was divided into two samples, representing both the edge and interior of each of the lobes. One sample from each animal was digested in 5% KOH at 37°C for 18 h. Schistosome eggs were then visualized by direct phase-contrast microscopy and counted. Hemozoin was extracted from the other sample via homogenization (Ultra Turrax T8; IKA, Germany) followed by proteinase K digestion (1 mg/ml overnight at 37°C). The sample was then centrifuged at 1,000 × g for 60 s at room temperature, and the supernatant was retained and then centrifuged at 5,400 × g for a further 2 h to pellet the hemozoin. The pellet was then washed sequentially with 0.1 M NaHCO3 (pH 9.1) containing 2.5% SDS, with distilled water, and then with 1:1 chloroform-methanol before being dried. For quantification, hemozoin was dissolved in 20 mM NaOH containing 2% SDS, and the optical density (OD) was compared to those of heme standards as previously described (25).
A 5-mm-thick section of liver was also cut from across the left lobe, fixed for 12 h in 10% buffered formalin, and embedded in paraffin. Five-micrometer sections were then prepared and stained with hematoxylin and eosin. Granuloma diameters around eggs containing a miracidium, and not in contact with other granulomas, were measured using an ocular micrometer, and the volumes were calculated by assuming a spherical shape for the granulomas (26).
ELISAs.Single-cell splenocyte suspensions were prepared by passing spleen tissue through a 70-μm filter followed by removal of erythrocytes with ACK lysis buffer (Invitrogen). Splenocytes were resuspended at 4 × 106 cells/ml in high-glucose Dulbecco's modified Eagle medium (DMEM) supplemented with 10% newborn calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 4 mM l-glutamine. Splenocytes were then seeded into 48-well plates (500-μl cultures) and stimulated for 72 h with 10 μg/ml of soluble egg antigens (SEA), kindly provided by Cornelis Hokke (Leiden University Medical Centre, The Netherlands). Supernatants were collected, and IL-5 secretion was assessed by sandwich enzyme-linked immunosorbent assay (ELISA) as previously described (27). Briefly, the assay was performed on Immulon 4HBX 96-well plates (Nunc, Denmark), using an antibody pair purchased from BD Pharmingen. The capture antibody (clone TRFK5) was used at a dilution of 1:500, and the biotinylated detection antibody (clone TRFK4) was used at 1:2,000. The signal was developed with 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) reagent (Kirkegaard & Perry Laboratories), and the OD (absorbance) at 405 nm was recorded immediately, using a PolarStar Omega plate reader (BMG Labtech, Germany).
Retnla secretion by macrophage cultures was also assessed by the same method, using PeproTech antibodies kindly provided by Andrew MacDonald (University of Edinburgh). Both the capture antibody and the biotinylated detection antibody were used at a dilution of 1:100.
The sera of experimentally infected mice were assessed for isotype-specific (IgG1 and IgG2b) anti-soluble worm adult protein (anti-SWAP) reactivity by ELISA as previously described (6). The signal was allowed to develop for a standardized time: 14 min for IgG1 and 20 min for IgG2b. Reactions were terminated with 1% (wt/vol) SDS, and the OD (absorbance) at 405 nm was recorded immediately, using a PolarStar Omega plate reader (BMG Labtech).
Immunohistochemistry.Formalin-fixed, paraffin-embedded liver sections (5 μm) were dewaxed and rehydrated. Antigen retrieval was performed using proteinase K digestion (20 μg/ml in Tris-EDTA buffer [50 mM Tris, 1 mM EDTA, pH 8.0] for 5 min at room temperature). Slides were washed in Tris-buffered saline (TBS; 50 mM Tris, 0.9% [wt/vol] NaCl, pH 8.4), and endogenous peroxidase activity was quenched with 4:1 methanol-3% H2O2 for 10 min at room temperature. Slides were blocked for 30 min at 37°C in 10% rabbit serum, blotted to remove excess liquid, and then stained using an anti-F4/80 primary antibody (ab6640; AbCam, United Kingdom) at a dilution of 1:500 in TBS at 37°C for 90 min. After washing with TBS, slides were incubated for 90 min at 37°C with a horseradish peroxidase (HRP)-conjugated secondary antibody (ab6734; AbCam) at a dilution of 1:1,000 in TBS. After washing, the signal was developed using 3,3′-diaminobenzidine (DAB) reagent for 4 min. Slides were then counterstained for 30 s with Meyer's hematoxylin, rinsed in distilled water, and blued with 0.1% bicarbonate solution for 1 min prior to dehydration, clearance in xylene, and mounting in distrene dibutyl phthalate xylene (DPX).
Hemozoin purification and β-hematin synthesis.Approximately 5 ml of packed adult S. mansoni worms was resuspended in 20 ml of phosphate-buffered saline (PBS) and thoroughly disrupted using a probe sonicator (Microson XL ultrasonic cell disrupter; Misonix) at 14 W in bursts of 30 s. This mixture was then centrifuged twice at 1,000 × g for 60 s at room temperature, and the supernatants were retained and centrifuged for a further 2 h at 5,400 × g. The hemozoin pellet was then resuspended in PBS, and lipid contamination was removed via chloroform-methanol extraction according to the method of Bligh and Dyer (28). Hemozoin was retrieved from the aqueous-solvent interface, resuspended in 10 ml of PBS, gently sonicated (5 W for 30 s), and then treated for 18 h with proteinase K (1 mg/ml at 37°C). The hemozoin was then pelleted at 21,000 × g for 5 min, resuspended in PBS, and centrifuged as described above. This washing process was repeated five times, and the hemozoin was then treated for 18 h with DNase and RNase A (100 U/ml and 1 mg/ml, respectively) at 37°C. After primary nucleic acid digestion, the hemozoin was pelleted and washed as described above, five times in PBS containing 2% SDS and five times in PBS, before a secondary 18-h treatment with DNase was performed (as described above). The hemozoin was again pelleted and washed as described above, twice in PBS containing 2% SDS, five times in 0.1 M NaHCO3 (pH 9.1) containing 2.5% SDS, and five times in distilled water. Finally, purified hemozoin samples were mixed with Triton X-114 (1% [vol/vol] in distilled water), followed by shaking (200 rpm) at 4°C for 30 min and then at 37°C for 10 min as previously described (29). Samples were then centrifuged at 20,000 × g at 25°C for 10 min, and the aqueous and detergent phases were discarded. The hemozoin was resuspended in distilled water, and this endotoxin removal treatment was repeated eight times. Samples were then washed five times in distilled water, and endotoxin levels were measured using the endpoint chromogenic Limulus assay (Lonza, Switzerland) according to the manufacturer's instructions. Finally, hemozoin was quantified as described above and resuspended in sterile PBS at 1 mg/ml. Hemozoin purity was assessed using routine DNA gel electrophoresis, SDS-polyacrylamide gel electrophoresis, and Western blotting methodologies (30), as well as thin-layer chromatography on a silica-coated plate (Merck), using heptane-diethyl ether-acetic acid (70:30:2 by volume) as the running solvent and staining with 0.2% (wt/vol) amido black dissolved in 1 M NaCl.
β-Hematin was crystallized under sterile conditions, using porcine-derived heme, via the aqueous acid-catalyzed reaction described by Bohle et al. (31).
Scanning electron microscopy (SEM).Both hemozoin and β-hematin samples were washed and resuspended in distilled water, gently sonicated to disrupt aggregates (5 W for 30 s), and then slowly dried on platinum-palladium-coated glass coverslips. Samples were then sputter coated with 4 nm of platinum-palladium and imaged with a model S-4700 field emission scanning electron microscope (Hitachi, Japan) at 1 kV, using the secondary-electron upper detector.
Raman and Fourier-transformed infrared (FTIR) spectroscopy.Raman spectra of purified hemozoin and β-hematin were collected using a LabRam high-resolution confocal Raman microscope (Horiba-Jobin Yvon, Japan) equipped with an integral Olympus microscope. Samples were resuspended in distilled water, and a droplet was dried on a glass microscope slide. Visualized at a magnification of ×10, an area of even coating was selected and excited using a 632.8-nm helium-neon laser. Power at the sample was 20 mW for each spectrum, and 25 scans were accumulated between 1,700 and 200 cm−1. The laser exposure for each scan was 10 min. The Raman shifted light was collected through a notch filter and recorded using a charge-coupled device camera cooled to −70°C. Data were compiled and analyzed using Labspec software, version 5.25.15 (Horiba-Jobin Yvon).
Samples for FTIR spectroscopy were washed several times in distilled water and then dried under vacuum. The dried powder was then finely ground with potassium bromide, dried in an oven, and compressed at 8 tons under vacuum into a disk through which a light beam was passed using an Equinox 55 FTIR spectrophotometer (Bruker, United Kingdom). The hemozoin/β-hematin absorbance spectra were collected using Opus software, version 4.2, and the absorbance spectrum of a control potassium bromide disk was subtracted.
BMDMΦ culture and activation.Bone marrow was collected from the femurs of male C57BL/6 mice and passed through a 70-μm filter. Cells were resuspended at a concentration of 2 × 106 cells/ml in high-glucose DMEM supplemented with 10% newborn calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, and 20% L929 conditioned medium and then cultured in petri dishes for 7 days, with replacement of the medium on days 3 and 5. The macrophage monolayer (>98% macrophages by May-Grünwald-Giemsa staining) was then detached with 5 mM EDTA and resuspended at a density of 0.5 × 106 cells/ml in high-glucose DMEM supplemented with 10% newborn calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. Bone marrow-derived macrophage (BMDMΦ) cultures (500 μl) were seeded into 48-well culture plates and maintained at 37°C and 5% CO2 in a humidified incubator.
Cultures were allowed to equilibrate for 16 h and then treated for 20 h with either recombinant IL-4 (20 ng/ml) or a combination of gamma interferon (IFN-γ) (10 U/ml) and lipopolysaccharide (LPS) (10 ng/ml). Macrophages undergoing activation switching were allowed to equilibrate for 2 h before being treated with IFN-γ–LPS, as described above, for 18 h. The cell monolayer was then washed and the medium replaced prior to counterstimulation with IL-4, as described above, for a further 18 h. Where required, hemozoin or β-hematin (25 μg/ml) was added to these cultures alongside the activating cytokines. In the case of cultures undergoing switching, hemozoin/β-hematin was added at the point of IL-4 counterstimulation. At the end of these stimulation regimens, cells were either lysed in TRIzol reagent (Invitrogen) for RNA extraction or processed for arginase activity quantification using a QuantiChrom assay kit (BioAssay Systems). Cell culture supernatants were also retained for quantification of secreted Retnla (as described above).
Real-time quantitative reverse-transcription PCR (qRT-PCR).Total RNA was prepared from cells or tissues lysed in TRIzol reagent (Invitrogen), and the final concentration was adjusted to 0.1 μg/μl. For each sample, 1 μg of total RNA was converted into cDNA as previously described (32). Negative-control reaction mixtures (lacking reverse transcriptase) were also prepared in parallel. All cDNA reaction mixtures (including negative controls) were diluted to a total volume of 200 μl in RNase- and DNase-free water.
Real-time qRT-PCR primers (Table 1) were designed for the alternatively activated macrophage transcripts Ym1 (GenBank accession no. NM_009892.1), Retnla (GenBank accession no. NM_020509.3), and Arg1 (GenBank accession no. NM_007482.3), the classically activated macrophage transcript Nos2 (GenBank accession no. NM_010927.3), and an endogenous, constitutively expressed control transcript, Hprt1 (GenBank accession no. NM_013556.2) (33). All amplicons were sequenced (IBERS Sequencing Facility, Aberystwyth University) to confirm PCR specificity.
Primers used for real-time qRT-PCR gene expression analysis
Triplicate real-time qRT-PCR mixtures (12.5 μl) were prepared with Fast SYBR green master mix supplemented with ROX passive reference dye (Applied Biosystems), 2.5 μl of cDNA template, and 0.4 μM (each) PCR primers. PCR was conducted in 96-well MicroAmp PCR plates on a StepOnePlus real-time thermal cycler (Applied Biosystems) using a fast protocol as follows: 95°C for 20 s followed by 40 cycles of 94°C for 5 s and the annealing temperature (Tm) (Table 1) for 30 s. A melting curve analysis of PCR products was conducted over 0.2°C increments, and samples with double melt peaks were automatically excluded from the analysis.
Relative gene expression was calculated using the Pfaffl method (34), where expression of each target gene in a test sample is calculated as the fold change with respect to a calibrator cDNA population (unstimulated macrophage cultures or uninfected mouse liver tissue) and normalized to expression of the reference gene Hprt1. An expression ratio of 1 demonstrates no change in gene expression between the compared cDNA populations.
Statistical analyses.Statistical comparisons between treatments were performed using one-way analysis of variance (ANOVA) followed by post hoc testing with Fisher's least-significant-difference test. All significant comparisons are expressed as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.005. Where appropriate, error bars are displayed and represent the standard errors of the means.
RESULTS
Immunological context of hemozoin accumulation in the host liver.Analysis of tissues collected from a longitudinal experimental murine infection revealed that accumulation of S. mansoni hemozoin in the liver (Fig. 1A) commenced at approximately 5 weeks postinfection. This was coincident with the detection of schistosome eggs in the same organ and followed by circumoval granulomatous pathology from 6 weeks postinfection (Fig. 1B). Granuloma development, peaking at 8 to 10 weeks postinfection prior to being downmodulated (Fig. 1B), was mirrored by a systemic Th2 immune phenotype characterized by SEA-induced secretion of IL-5 by splenocyte cultures ex vivo (Fig. 1C), as well as rapidly increased anti-SWAP IgG1 titers from 6 weeks postinfection (Fig. 1D). Inspection of liver tissue sections revealed that the vast majority of hemozoin deposited in this organ was internalized by phagocytic, F4/80+ cells (macrophages) dispersed in the parenchyma peripheral to the egg-induced granulomas (Fig. 1E and F).
Immunological context of hemozoin accumulation in the host liver. Groups of experimentally infected mice were harvested at 2 to 14 weeks postinfection for parasitological and immunological analysis. Samples of liver tissue were assessed for hemozoin content (A), egg burden (B), and granuloma size (B). (C) Splenocyte cultures were stimulated ex vivo for 72 h with SEA (10 μg/ml), and supernatants were then analyzed for IL-5 secretion via sandwich ELISA. Results were normalized against unstimulated negative-control cultures. (D) Sera were collected, and SWAP-specific IgG1 and IgG2b titers were quantified by isotype-specific ELISA. (E and F) Formalin-fixed sections of liver tissue were probed with an anti-F4/80 monoclonal antibody (brown). Black hemozoin granules are visible without staining. A representative section from 12 weeks postinfection is shown. The arrow in panel E indicates a granuloma, and arrows in panel F indicate F4/80+ cells containing hemozoin. (G and H) Hepatic expression of macrophage activation genes was quantified using real-time qRT-PCR. Both the alternative activation transcripts Ym1 and Retnla (G) and the classical activation transcript Nos2 (H) were analyzed. Gene expression is presented as the fold change (f.c) over the expression in uninfected liver tissue and was normalized against Hprt1. All results are representative of two replicate experimental infections.
Expression of both the macrophage alternative activation genes Ym1 and Retnla (Fig. 1G) and the classical activation gene Nos2 (Fig. 1H) was monitored in liver tissue throughout the experimental infection. Expression of Arg1 in the liver was not assessed, as it is transcribed constitutively in this organ (35). From 7 weeks postinfection, expression of both Ym1 and Retnla was increased significantly (P < 0.005) compared to that at weeks 2 to 6. Expression of Ym1, but not Retnla, was then somewhat downmodulated by 10 weeks postinfection, consistent with other markers of the Th2 immune phenotype (Fig. 1B to D). In contrast, expression of the classically activated macrophage transcript Nos2 peaked between 5 and 7 weeks postinfection before being downregulated. Low-level expression was then maintained throughout the rest of the experimental period (Fig. 1H). This dynamic expression pattern is consistent with nitric oxide secretion by restimulated splenocytes ex vivo (36).
To explore the potential modulatory effects mediated by hemozoin in this immunologically shifting hepatic system, we designed in vitro models of alternative macrophage activation and tested the effects of hemozoin uptake on these cells' activation phenotypes.
Alternative activation of BMDMΦs in vitro.As previously described, stimulation of naïve BMDMΦs with IL-4 (20 ng/ml) was sufficient to induce expression of the three key markers of an alternatively activated macrophage (aaMΦ) phenotype: Ym1 (Fig. 2A), Retnla (Fig. 2B), and Arg1 (Fig. 2C) (37). Ym1 was upregulated >32,000-fold compared to untreated cells, Retnla was upregulated approximately 7,000-fold, and Arg1 was upregulated approximately 8,000-fold. The relative abundances of these IL-4-induced transcripts (Ym1 ≫ Retnla/Arg1) were comparable to those observed in aaMΦs derived in vivo from mice experimentally infected with the Th2-inducing parasites Fasciola hepatica, Litmosoides sigmodontis, and Nippostrongylus brasiliensis (38, 39). Stimulation of BMDMΦs with the classically activating regimen of IFN-γ and LPS did not upregulate the expression of Ym1, Retnla, or Arg1 (Fig. 2A to C) but did upregulate the expression of Nos2, approximately 2,000-fold compared to untreated cells (Fig. 2D).
Activation of BMDMΦs in vitro. BMDMΦ cultures were either classically activated with IFN-γ (10 U/ml) and LPS (10 ng/ml) or alternatively activated with IL-4 (20 ng/ml). Cultures were also switched from classical to alternative activation via sequential stimulation with IFN-γ–LPS followed by IL-4. Expression of Ym1 (A), Retnla (B), Arg1 (C), and Nos2 (D) was quantified using real-time qRT-PCR. Gene expression was normalized against Hprt1 and is presented as the fold change (f.c) over the expression in untreated cultures. Data are derived from 2 or 3 biological replicates per treatment and are representative of duplicate experiments.
Stimulation of BMDMΦs with IFN-γ and LPS followed by stimulation with IL-4 resulted in a switched phenotype with modified alternative activation characteristics (Fig. 2). In this case, expression of Ym1 was elevated compared with that in naïve cultures stimulated with IL-4 (Fig. 2A). Similarly, expression of Arg1 by switched cultures was significantly elevated compared to that in IL-4-stimulated naïve populations (Fig. 2C) (P < 0.005). In stark contrast, the expression of Retnla by switched cultures was significantly less than that induced by IL-4 stimulation of naïve cultures (Fig. 2B) (P < 0.005). However, this highly reduced Retnla expression (only 11-fold over that in untreated cells) was still significantly greater than the background level induced by the classically activating IFN-γ–LPS regimen (P < 0.005). Expression of the classically activated transcript Nos2 was significantly reduced in switched cultures compared to IFN-γ–LPS-treated naïve cells (Fig. 2D) (P < 0.05). However, this level of Nos2 expression was significantly greater than the background level induced during IL-4 stimulation of a naïve population (P < 0.05).
Preparation and validation of experimental hemozoin.In order to test the effects of hemozoin uptake on the alternative activation of BMDMΦs as described above, experimental hemozoin was purified from S. mansoni worms and validated for use in cell culture systems. We also prepared the structurally identical, synthetic β-hematin (11) as a convenient experimental control for biologically derived hemozoin.
Synthesis of β-hematin via the aqueous acid-catalyzed method of Slater et al. (40) yielded elongated tapered crystal aggregates that were approximately 200 nm long (Fig. 3A). The crude hemozoin preparation, extracted from sonicated adult S. mansoni worms, was coated with a heterogeneous, noncrystalline substance (Fig. 3B), which could be removed by biochemical purification steps (Fig. 3C). Analysis of these preparations revealed significant contamination of the crude S. mansoni hemozoin with lipids (Fig. 3D), proteins (Fig. 3E) (many of which retained a high degree of immunogenicity) (Fig. 3F), and nucleic acids (Fig. 3G). Furthermore, this crude preparation was also contaminated with >2.5 endotoxin units (EU)/mg of endotoxin. The source of endotoxin contamination was not clear but may be attributed to the original supply of worm material dating from the 1970s. Importantly, however, purified hemozoin and β-hematin were shown to be free of detectable lipids, proteins, or nucleic acids (Fig. 3D to G). Residual endotoxin detectable in the purified hemozoin samples was acceptably low for use in a cell culture system (0.175 ± 0.065 EU/mg; n = 3 preparations).
Preparation and validation of experimental hemozoin. (A) SEM of synthetic β-hematin crystallized from porcine heme via the aqueous acid-catalyzed method. (B) Crude S. mansoni (S. m) hemozoin was extracted from adult worms disrupted by sonication and was heavily contaminated with a heterogeneous noncrystalline material. (C) SEM of purified S. mansoni hemozoin. (D to G) The contamination of hemozoin/β-hematin samples was assessed for lipids via thin-layer chromatography (D), for proteins by silver-stained SDS-PAGE (E), for immunogenic proteins via Western blot analysis with infected mouse sera (from 7 weeks postinfection; 1:5,000 dilution) (F), and for nucleic acids by use of ethidium bromide-stained agarose gels (G). Hz, hemozoin; ChE, cholesterol ester; TAG, triacylglycerol; FA, fatty acid; Ch, cholesterol; DAG, diacylglycerol; MAG, monoacylglycerol; Ph, phospholipids; MWM, molecular size markers.
Biophysical comparison of S. mansoni hemozoin and β-hematin.The Raman spectra of β-hematin and S. mansoni hemozoin (Fig. 4A) displayed good peak-to-peak matching and were consistent with Raman spectra previously published for both β-hematin and Plasmodium falciparum hemozoin (41, 42). Both spectra included the expected key vibrational modes (Fig. 4A, arrows a to d) descriptive of metallo-porphyrin ring structures such as heme {ferriprotoporphyrin-IX [Fe(III)PPIX]} (43, 44). A symmetrical Fe-N “doming” mode (Fig. 4A, arrow e) descriptive of the reciprocally bound Fe(III)PPIX dimers that comprise hemozoin (45) was also observed.
Biophysical comparison of S. mansoni hemozoin and β-hematin. Postpurification comparison of S. mansoni hemozoin (black lines) and β-hematin (gray lines), using Raman spectroscopy (A) and FTIR spectroscopy (B). (A) The characteristic Raman-active modes of metalloporphyrin structures (arrows a to e), including the symmetrical Fe-N doming mode (arrow e) of reciprocally bound Fe(III)PPIX dimers, are indicated. (B) Infrared stretching frequencies (arrows f and g), characteristic of the bond structures of the propionic side chains of Fe(III)PPIX dimers incorporated into a crystal lattice, are also indicated.
Similarly, the FTIR spectra of β-hematin and S. mansoni hemozoin (Fig. 4B) also displayed good peak-to-peak matching and were consistent with published FTIR analyses of both β-hematin and hemozoin derived from S. mansoni and P. falciparum (9, 40, 41). Typical stretching absorptions (Fig. 4B, arrows f and g), characteristic of the way the propionic side chains of Fe(III)PPIX behave when reciprocally bound and incorporated into a crystal lattice, were both present (40, 46).
Hemozoin uptake by BMDMΦs.Having successfully purified native S. mansoni hemozoin, we next investigated the effects of its uptake on the BMDMΦ models described above (Fig. 2).
After 30 min of coculture with hemozoin, 50.7% of naïve BMDMΦs had taken up some hemozoin granules (Fig. 5A and D). After 2 h, this had increased to 82.6% of cells (Fig. 5B and D), and it increased to 95.7% of cells by 18 h (Fig. 5C and D). Similar dynamics of β-hematin uptake were also observed (data not shown). Uptake of both hemozoin and β-hematin was variable, ranging from the appearance of sparse individual granules in the cytoplasm to granules completely filling the cytoplasmic space (as indicated by arrows in Fig. 5A to C). At a concentration of 25 μg/1 × 106 macrophages (25 μg/ml), neither β-hematin nor S. mansoni hemozoin had any innate alternatively activating effects on BMDMΦs as assessed by Ym1, Retnla, and Arg1 expression (Fig. 5E to G).
Hemozoin uptake by BMDMΦs has no innate alternatively activating effect in vitro. (A to C) Hemozoin (25 μg/ml) was added to cultures of BMDMΦs, and after 30 min, 2 h, and 18 h, cells were stained with May-Grünwald and Giemsa stains. (D) The percentage of hemozoin-positive cells (such as those indicated with arrows) at each time point was assessed by manual counting. Similar uptake dynamics were observed with β-hematin (data not shown). (E to G) Hemozoin or β-hematin (25 μg/ml) was added to parallel BMDMΦ cultures and incubated for 20 h. Expression of the alternative activation transcripts Ym1 (E), Retnla (F), and Arg1 (G) was quantified using real-time qRT-PCR. IL-4 (20 ng/ml) was used as a positive control. Gene expression was normalized against Hprt1 and is presented as the fold change (f.c) over the expression in untreated cultures. Data are derived from 3 biological replicates per treatment and are representative of duplicate experiments.
Effects of hemozoin and β-hematin on IL-4-induced alternative activation of BMDMΦs in vitro.To address the effects of hemozoin uptake on the alternative activation of BMDMΦs, we added β-hematin or S. mansoni hemozoin, at the same time as IL-4 stimulation, to both naïve (Fig. 6A to E) and switched (Fig. 6F to J) macrophage models. In general, the addition of β-hematin or hemozoin to naïve BMDMΦ cultures (Fig. 6A to E) potentiated the IL-4-induced expression of alternatively activated transcripts. Both hemozoin and β-hematin significantly enhanced IL-4-induced expression of Ym1 (Fig. 6A) (P < 0.005). Hemozoin also significantly increased IL-4-induced Arg1 expression (Fig. 6B) (P < 0.005) and arginase activity (Fig. 6C) (P < 0.05), while β-hematin had no effect on these markers of alternative activation. β-Hematin also significantly upregulated IL-4-induced expression of Retnla (Fig. 6D) (P < 0.01). In stark contrast, however, the addition of hemozoin resulted in a highly significant reduction of IL-4-induced expression of Retnla (Fig. 6D) (P < 0.005), a result that was also confirmed at the level of Retnla protein secretion (Fig. 6E) (P < 0.05).
Effects of hemozoin/β-hematin uptake on the alternative activation of naïve and switched BMDMΦs. BMDMΦ cultures were either left untreated for 18 h (A to E) or stimulated for 18 h with IFN-γ and LPS (10 U/ml and 10 ng/ml, respectively) (F to J). All cultures were then stimulated with IL-4 (20 ng/ml) for 20 h in the presence or absence of hemozoin or β-hematin (25 μg/ml). Expression of the alternative activation transcripts Ym1 (A and F), Arg1 (B and G), and Retnla (D and I) was quantified using real-time qRT-PCR. Expression is presented as the fold change (f.c) over the expression in untreated cultures and was normalized against Hprt1. (C and H) Arginase activity was also assessed using a chromogenic endpoint assay. (E and J) Retnla secretion was assessed via ELISA. Data are derived from 2 or 3 biological replicates per treatment and are representative of duplicate experiments.
Neither S. mansoni hemozoin nor β-hematin had any effect on the IL-4-induced expression of Ym1 (Fig. 6F), expression of Arg1 (Fig. 6G), or arginase activity (Fig. 6H) in switched macrophage cultures. Despite an already low level of IL-4-induced Retnla expression by switched macrophage cultures (Fig. 2B), however, hemozoin, but not β-hematin, caused a further reduction in the expression of this gene (Fig. 6I). Again, this result was confirmed at the level of Retnla protein secretion (Fig. 6J).
DISCUSSION
Schistosoma and Plasmodium parasites cause some of the most debilitating and chronic diseases of mankind. Both parasites rely upon host hemoglobin as a food source, and both tackle the toxic insult posed by free heme via its crystallization into hemozoin. The deposition of “malaria pigment” (hemozoin) in the tissues of malaria patients has been noted for centuries (47), and more recently, a number of immunomodulatory effects on antigen-presenting cells have been ascribed to it (14–16, 48, 49). It was not until the early 2000s, however, that the pigment regurgitated by schistosomes was recognized as hemozoin and confirmed to share the same molecular crystal structure as Plasmodium hemozoin (9, 11).
Despite their identical molecular structures, Schistosoma hemozoin differs greatly from Plasmodium hemozoin in its ultrastructural appearance. Furthermore, its interaction with the host occurs in a very different immunological environment. Our study comprehensively places the accumulation of S. mansoni hemozoin in its Th2 immunopathological context and demonstrates, through the use of appropriate in vitro cell model systems, that it exerts specific modulatory effects upon phagocytic macrophages (BMDMΦs) undergoing IL-4-induced alternative activation.
Macrophages in the host liver encounter S. mansoni hemozoin in the context of the developing egg-induced Th2 immune response (Fig. 1). Our study reveals that prior to maximal hepatic expression of the alternative activation markers Ym1 and Retnla (Fig. 1G), a peak of Nos2 expression, characteristic of classically activated macrophage activity, is also observed (Fig. 1H). This is consistent with the murine model of schistosomiasis whereby the early immune response, occurring between 3 and 6 weeks postinfection, appears to be Th1 in nature, with nitric oxide production and a cytokine milieu dominated by IFN-γ (36, 50–52). Recent evidence suggests that aaMΦs accumulate in tissues not as a result of monocyte precursor recruitment from the blood but rather via IL-4-stimulated in situ proliferation of the resident macrophage population, even when this population was initially elicited and activated under Th1-type inflammatory conditions (53). This suggests that the shift in hepatic macrophage activation markers observed here (Fig. 1G and H) may represent a functional plasticity of the resident population. As such, BMDMΦs switched from a classically to an alternatively activated phenotype in vitro represent the most pertinent experimental model with which to test the effects of hemozoin uptake. Indeed, the relative gene expression by these switched cultures (Fig. 2) (high Ym1, relatively low Retnla, and residual Nos2 expression) was consistent with the pattern of gene expression in the host liver from 8 weeks postinfection onwards (Fig. 1G and H).
The alternatively activated BMDMΦ phenotype derived in the presence of hemozoin in vitro maintained normal or increased Ym1 expression (Fig. 6A and F), Arg1 expression (Fig. 6B and G), and arginase activity (Fig. 6C and H), depending on the state of the cell prior to IL-4 stimulation. In stark contrast to this general trend of neutral to positive effects on the IL-4-induced alternatively activated phenotype, hemozoin mediated the suppression of Retnla expression (Fig. 6D and I) and resultant Retnla protein secretion (Fig. 6E and J) to near background levels. This effect was not mediated by β-hematin and was consistently observed, regardless of the background against which IL-4 stimulation was applied. Furthermore, the same effect was also observed using the transformed macrophage cell line RAW264.7 (data not shown). While the functions of Retnla are largely unknown, evidence suggests that during a schistosome infection, one of its roles is to limit the Th2 inflammatory response via a negative-feedback loop. In Retnla−/− mice, increased granulomatous inflammation and elevated fibrosis have been observed in response to parasite eggs and related to increased Th2 cytokine production (22, 23). Secretion of IL-4 by Retnla−/− splenocytes stimulated with SEA ex vivo was exaggerated and could be returned to wild-type levels by the addition of exogenous Retnla (23). Similarly, the direct binding of Retnla to T cells and macrophages was shown to specifically inhibit these cells' production of IL-4, IL-13, and IL-5, while Retnla−/− aaMΦs promoted exaggerated antigen-specific Th2 cell differentiation (22). Furthermore, Retnla derived from alternatively activated dendritic cells has been shown to promote IL-10 production by T cells (21). Thus, Retnla represents one of the many host mechanisms employed to limit and modulate the pathological Th2 inflammatory response to parasite eggs.
Despite the very low levels of Retnla expression by hemozoin-loaded switched BMDMΦs in vitro (only approximately 2-fold over that in untreated cells) (Fig. 6I), expression of Retnla in the infected mouse liver was significantly increased from 7 weeks postinfection (approximately 50-fold over that in uninfected tissue) (Fig. 1G). These data may be explained by the finding of Pesce et al., who demonstrated that in the livers of schistosome-infected mice, eosinophils, not aaMΦs, were the primary source of Retnla (23). The presence of hemozoin in their model system (experimentally infected mice) would explain the lack of macrophage-derived Retnla and reconcile their findings with those of another research group who demonstrated that in the lung model of granulomatous pathology (intravenous injection of parasite eggs), where no hemozoin is present, both eosinophils and aaMΦs were equal cellular sources of Retnla (22). Similarly, dermal cell populations collected from the pinnae of mice after multiple topical infections comprised a Retnla-positive eosinophil population and an aaMΦ population specifically lacking in Retnla (54). This is a finding that, again, may now be explained by the presence of hemozoin, which seems to appear in the guts of schistosomula as early as 2 days posttransformation (55). On the basis of these distinct murine studies and the results presented here, we speculate that in a normal Th2 inflammatory milieu, both eosinophils and aaMΦs are capable of expressing Retnla in an IL-4-dependent manner, contributing to the downmodulation of the Th2 response and the resolution of inflammation via the Retnla negative-feedback loop. Schistosomes moderate this pathway via the hemozoin-mediated suppression of aaMΦ-derived Retnla, resulting in an environment relatively permissive toward the Th2 response, which is nonetheless kept in check by the unmodulated Retnla derived from the nonphagocytic eosinophil population.
The mechanism by which hemozoin might specifically downregulate Retnla (but not Ym1 and Arg1) expression is unknown. Expression of all three of the alternatively activated transcripts measured in this study has been shown to be dependent upon IL-4 signaling and the consequent binding of various transcription factors, including STAT6, at IL-4-responsive elements (56–58). Oddly, despite being affected by hemozoin in different ways, the IL-4-responsive element of Retnla closely resembles that of Arg1, and transcription of both of these genes is dependent upon the binding of the same set of transcription factors (56, 57, 59). A global microarray analysis of IL-4-inducible macrophage genes and a bioinformatic comparison of those affected or unaffected by hemozoin might help to highlight potential mechanistic explanations for further study.
Interestingly, β-hematin did not have the same effects on the alternative activation of BMDMΦs as did the isolated S. mansoni hemozoin employed here. Broadly, it either enhanced or had no effect on the IL-4-induced expression of aaMΦ markers. It remains a possibility that the specific effects mediated by experimental S. mansoni hemozoin could be attributed to undetectable biological contaminants. However, the facts that hemozoin had no innate alternatively activating effects on these cells (Fig. 5E to G), that it produced similar Raman and FTIR spectral profiles to those of β-hematin (Fig. 4), and that its modulation of IL-4-induced gene expression was target specific suggest that this is not the case. Previous evidence suggests that morphological differences in various preparations of β-hematin can have profound effects on the way the cell responds to its uptake (60). The pronounced ultrastructural differences observed between β-hematin and S. mansoni hemozoin may affect their modulatory capacities, and we argue, therefore, that β-hematin should not be used as a direct experimental proxy for schistosome hemozoin. Understanding the reasons behind the differing effects of β-hematin and schistosome hemozoin presents a fascinating biophysical proposition.
This study provides a foundation for further investigations into the immunomodulatory effects mediated by schistosome hemozoin, in particular, and of hemozoin in the context of Th2-type immune responses, in general. The effect of purified S. mansoni hemozoin on the aaMΦ phenotype was strikingly specific, suggesting a genuine hemozoin-mediated modulatory event. Furthermore, the target of hemozoin modulation, Retnla, is itself a critical regulatory component in the balancing of the host response away from fatal extremes (23). We thus propose that S. mansoni hemozoin represents not only a detoxification pathway for this parasite but also a parasite-directed contribution to the complex coevolved network of mechanisms that drive, modulate, and moderate host responses. This complex moderation is essential for a response that both enables continuation of the parasite life cycle and parasite survival in the exposed circulatory niche and also protects the host from the extremes of immune-mediated pathology.
ACKNOWLEDGMENTS
This work was supported by an Aberystwyth University Ph.D. studentship to Martha Truscott.
We thank Iain Chalmers, Emily Peak, and Julie Hirst for assisting with schistosome life cycle maintenance and experimental infections. We acknowledge Fred Lewis for providing adult S. mansoni worms for hemozoin extraction, Andrew MacDonald for L929 cells, Cornelis Hokke for SEA, Alan Cheever for detailed discussions of murine pathology, Steve Wade for SEM work, Tony Pugh for photomicroscopy, and Gordon Alison for assistance with FTIR spectroscopy.
FOOTNOTES
- Received 9 July 2012.
- Returned for modification 28 August 2012.
- Accepted 15 October 2012.
- Accepted manuscript posted online 22 October 2012.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
