Plasmodium sporozoites traverse several host cells before infecting hepatocytes. In the process, the plasma membranes of the cells are ruptured, resulting in the release of cytosolic factors into the microenvironment. This released endogenous material is highly stimulatory/immunogenic and can serve as a danger signal initiating distinct responses in various cells. Thus, our study aimed at characterizing the effect of cell material leakage during Plasmodium infection on cultured mouse primary hepatocytes and HepG2 cells. We observed that wounded cell-derived cytosolic factors activate NF-κB, a main regulator of host inflammatory responses, in cells bordering wounded cells, which are potential host cells for final parasite infection. This activation of NF-κB occurred shortly after infection and led to a reduction of infection load in a time-dependent manner in vitro and in vivo, an effect that could be reverted by addition of the specific NF-κB inhibitor BAY11-7082. Furthermore, no NF-κB activation was observed when Spect−/− parasites, which are devoid of hepatocyte traversing properties, were used. We provide further evidence that NF-κB activation causes the induction of inducible NO synthase expression in hepatocytes, and this is, in turn, responsible for a decrease in Plasmodium-infected hepatocytes. Furthermore, primary hepatocytes from MyD88−/− mice showed no NF-κB activation and inducible NO synthase expression upon infection, suggesting a role of the Toll/IL-1 receptor family members in sensing cytosolic factors. Indeed, lack of MyD88 significantly increased infection in vitro and in vivo. Thus, host cell wounding due to parasite migration induces inflammation which limits the extent of parasite infection.
A crucial step in the Plasmodium life cycle within the mammalian host is the infection of hepatocytes by sporozoites. Host cell infection is preceded by the traversal of several hepatocytes by the parasite. As a consequence, these hepatocytes are wounded due to membrane rupturing (1). It has been suggested that migration through host cells is necessary to establish infection and it is thought that during the process, sporozoites become fully activated for infection (2). In addition, wounded cells secrete hepatocyte growth factor, which renders neighboring cells more susceptible for infection and protects them also from apoptosis (3, 4). However, in contrast, it has been shown that sporozoites deficient in sporozoite microneme protein essential for cell traversal, which are disabled in cell-traversal motility, infect hepatocytes as well as wild-type parasites, which questions the absolute requirement of hepatocyte traversal for infection (5). Although many wounded cells survive this migration process, a considerable number become necrotic. In general, loss of membrane integrity results in an efflux of cytosolic factors into the surrounding medium, which is an acute danger signal capable of initiating inflammation (6, 7). As previously shown in fibroblasts and macrophages, unlike apoptotic cells in which the cell membranes stay intact for a prolonged period of time, necrotic cells which totally lose their membrane integrity, induce genes involved in inflammation and tissue repair (8). Different endogenous molecules, such as heat-shock proteins (hsp),3 high mobility group box 1, and uric acid, have been identified to alert the immune system (9, 10, 11).
Toll-like receptors play a pivotal role in the initiation of an inflammatory response against various pathogens by recognizing conserved molecular patterns, called microorganism-associated molecular patterns (12). Toll-like receptors are expressed on different cell types and their importance has been investigated extensively on professional APCs, such as macrophages and dendritic cells. In addition to exogenous factors such as bacteria, viruses, and parasites, recent studies have also revealed a role of the Toll/IL-1 receptor family in regulating responses to endogenous factors (8, 13, 14, 15). Hepatocytes express multiple TLR family members and are therefore thought to be able to respond to various pathogens, but whether they are also responsive to endogenous factors is largely unknown (16). Toll/IL-1 receptor signaling involves the recruitment of common adaptor proteins, such as MyD88, IL-1 receptor-associated kinases, and TNF-α receptor-associated factor 6, which contribute to the activation of the IκB-kinase complex (12). The NF-κB heterodimer p50/p65 is held in an inactive state in the cytosol by its inhibitor IκBα (inhibitor of κBα). Toll/IL-1 receptor activated IκB-kinase phosphorylates IκBα leading to its ubiquitination and proteasomal degradation. Active NF-κB is released, which translocates to the nucleus to initiate transcription of target genes, including proinflammatory cytokines and anti-apoptotic molecules (17).
An important proinflammatory molecule regulated by NF-κB is inducible NO synthase (iNOS) and its induction has been reported upon tissue destruction (18, 19, 20). iNOS-derived NO is a pleiotropic regulatory molecule involved in homeostasis, cytoprotection, cytotoxicity, tissue repair, and infection control. In particular, NO regulates the contraction of wounds, collagen biosynthesis, and cell proliferation, although the precise mechanism is unknown. In addition, several studies have demonstrated that NO is of central importance in anti-malarial liver immunity (21, 22, 23). Due to the many dramatic effects of iNOS activity, its expression has to be tightly controlled. iNOS expression is regulated at different levels such as gene induction, mRNA, and protein stability. However, an extensively described regulation takes place at the transcriptional level via activation of transcription factors that include NF-κB, AP-1, IRF-1, NF-IL-6, and STAT-1α (24).
In this study, we have investigated the role of sporozoite-induced hepatocyte wounding and its accompanied release of endogenous molecules on parasite development. We have found that leakage of cellular contents during the infection process with wild type but not Spect−/− parasites results in the rapid activation of NF-κB in hepatocytes bordering wounded cells. In addition, we demonstrate that NF-κB activation leads to the de novo synthesis of iNOS, which in turn eliminates a significant amount of intracellular parasites. We provide further evidence that iNOS expression is induced by members of the Toll/IL-1 receptor family, because MyD88−/− mouse primary hepatocytes showed no response. Our present data show that, Plasmodium-mediated hepatocyte wounding is unfavorable for the parasite due to the induction of an innate immune response, which partially limits infection.
Materials and Methods
Anopheles stephensi mosquitoes were fed on Plasmodium berghei (ANKA strain)-harboring mice (parasitemia 2%) 3 to 5 days after eclosion. They were kept at 70% humidity at 21°C and were fed on a 10% glucose solution. One week after infection, they received a blood meal from uninfected mice. Twenty-one days after the infectious blood meal, sporozoites (spz) are isolated as follows: mosquitoes were first washed in 70% ethanol for 2 min, followed by two washes in DMEM containing 200 U/ml penicillin and 200 μg/ml streptomycin. Infected salivary glands of Anopheles stephensi mosquitoes were then isolated under a binocular microscope. To release the parasites, the isolated, infected salivary glands were homogenized in DMEM using a glass homogenizer. Additional centrifugation steps further purified the spz. As controls, homogenized salivary glands (sgm) of uninfected mosquitoes were used. To obtain inactivated sporozoites, freshly isolated spz were incubated at 56°C for 45 min as described before (4). Spect−/− parasites (5) were maintained as described above.
Mouse primary hepatocytes were isolated by a two-step collagenase perfusion of mouse liver lobules as previously described (25) with minor modifications. In brief, liver lobules were perfused for 10 min with Ca2+-free HEPES buffer (pH 7.6) at 37°C followed by perfusion with type IV collagenase (Sigma-Aldrich) solution (Ca2+-free HEPES buffer (pH 7.6) containing 0.04% type IV collagenase and 0.075% CaCl2 ▪ 2H2O) for another 5 min at 37°C. Resulting cells were purified over a 60% percoll gradient and washed twice by centrifugation at 800 rpm for 30 s. Because collagenase isolation of hepatocytes leads to NF-κB activation and iNOS mRNA expression, the cells were first cultured in standard medium (DMEM containing 10% FCS, 1% HEPES, 2 mM l-glutamine, 500 μM 2-ME, 100 U/ml penicillin, 100 μg/ml streptomycin) containing of 2 μM dexamethasone, a known suppressor of NF-κB and iNOS (26, 27). Cells were seeded on 1% gelatin coated plates. Five hours later, cells were washed and medium was replaced by standard medium only and cells were left for an additional 48 h before use. The human hepatocarcinoma cell line HepG2 was cultured in standard medium.
Generation of conditioned media
Conditioned medium was obtained by filtering (0.2 μm pore size) supernatants of cultured cells (2 × 105) incubated with Plasmodium sporozoites (105), or sgm or Spect−/− spz at 37°C for 1–3 h. No sporozoites were found in the conditioned medium as verified by microscopy. Alternatively, conditioned medium was generated from mechanically wounded cultured cells (106). Cells were wounded by scratching and left for 1–3 h at 37°C. Supernatant was filtered and used immediately.
A total of 106 HepG2 or mouse primary hepatocytes were either untreated or pretreated with 10 μM of the specific NF-κB inhibitor BAY11–7082 (Sigma-Aldrich) or DMSO as control for 1 h, followed by washing steps. After infection by 1 × 105 Plasmodium berghei sporozoites for different time periods, the cells were lysed in 200 μl NP40 lysis buffer (1% Triton X-100, 1% NP40, 20 mM Tris (pH 7.4), 150 mM NaCl, protease inhibitors, phosphatase inhibitors). Lysis was performed on ice for 10 min. Cell debris was removed by centrifugation (12,000 rpm, 4°C). Total protein concentration was assessed using the Bradford assay and 30 μg of protein sample was mixed with an equal volume of 3 × SDS-PAGE sample buffer. After electrophoresis on a 12% SDS-polyacrylamide gel and transfer to a nitrocellulose membrane, phosphorylated IκBα (Cell Signaling Technology) and α-tubulin (Sigma-Aldrich) were detected by using the corresponding primary Abs and secondary Abs. iNOS from mouse primary hepatocytes was detected using an anti-mouse iNOS Ab (BD Biosciences) and from HepG2 cells using an anti-human iNOS Ab (Santa Cruz Biotechnology). PMA- (Sigma-Aldrich) and ionomycin- (Calbiochem) treated Jurkat cells were used as a positive control for phospho-IκBα.
Luciferase reporter assay
Samples of 0.5 × 106 HepG2 cells (12-well plate) were transiently transfected using DOTAP Liposomal Transfection Reagent (Roche) with 5 μg of an NF-κB luciferase reporter construct (28) and 0.5 μg actin-β-Gal in DMEM supplemented with 10% FCS. After overnight incubation, the cells were washed, treated, and infected as described above for Western blot. Six hours later, the cells were lysed in 150 μl lysis buffer (0.2% Triton X-100, 92 mM KH2PO4, 0.91 mM K2HPO4, 1 mM DTT). Luciferase activity in the cell-free supernatant was assessed in a TD-20/20 Luminometer (BioSystems). Luciferase activity was normalized to β-galactosidase activity to correct for differences in transfection efficiency. In some experiments HepG2 cells were cotransfected with plasmids either overexpressing dominant negative, phosphorylation-deficient IκB (provided by Dr. Pascal Schneider, University of Lausanne, Switzerland), RelA (p65 subunit of NF-κB) (28) or dominant negative MyD88 (29), or with the empty vector pcDNA3.0 as control.
Determination of cellular wounding activity of sporozoites
Five × 104 sporozoites were added to 2.5 × 105 HepG2 cells in presence of 10 μg/ml propidium iodide (PI). Fluorescent nuclei of PI positive cells, which represent the wounded cells, were recorded every 20 s with the microscope Axiovert 100 (Zeiss). To achieve the final movie, the data were processed using the program MetaMorphOffice version 7.0. Alternatively, HepG2 cells were incubated with either spz, inactivated sporozoites, sgm, or Spect−/− spz for 2 h. The cells were fixed and permeabilized with 100% methanol and all nuclei were stained with 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI) for 30 min. The percentage of wounded cells was calculated as follows: 100 × (PI positive nuclei/DAPI positive nuclei).
Determination of NF-κB translocation
Six × 104 mouse primary hepatocytes were plated in 8-well Labtek chamber slides for 48 h. They were infected with 2 × 104 sporozoites for 2 h in the presence or absence of 10 μg/ml PI. Cells were washed in PBS and then fixed in 4% paraformaldehyde for 10 min. Cells were washed again and permeablized with 100% acetone for 1 min, followed by a blocking step (1% BSA in PBS for 30 min). Cells were then incubated with a polyclonal rabbit anti-p65 Ab (Santa Cruz Biotechnology) for 1 h at room temperature. Cells were washed twice and incubated with an anti-rabbit IgG Alexa Fluor 488-conjugated Ab in PBS-1% BSA for 1 h at room temperature. Cell nuclei were stained with DAPI for 30 min. Cells were examined with a fluorescence microscope (Axioplan; Zeiss).
Alternatively, cells were fixed in 1% glutaraldehyde first. Then they were incubated with a mouse Ab against the repeats of the P. berghei circumsporozoite protein (PbCSP) followed by incubation with an anti-mouse IgG Cy3 Ab (Molecular Probes) in PBS-1% BSA for 1 h at room temperature to determine sporozoites outside cells. Next, cells were permeabilized with 100% methanol and p65 was stained as described above followed by the incubation with an anti-rabbit IgG Cy3 Ab (Molecular Probes) in PBS-1% BSA for 1 h at room temperature. Sporozoites were stained again with the anti-PbCSP Ab followed by an anti-mouse Alexa Fluor 488 Ab. Sporozoites which were single stained (Alexa Fluor 488) infected cells successfully.
Determination of number of infected cells
Six × 104 mouse primary hepatocytes were plated in 8-well Labtek chamber slides for 48 h. They were infected with 2 × 104 sporozoites for 2 h. Afterward, the cells were washed and either fixed with 1% glutaraldehyde or fixed and permeabilized with 100% methanol for 10 min. Sporozoites were then stained with an Ab against the repeats of the PbCSP. The number of infected cells was calculated as follows: Number of sporozoites in methanol fixed wells − number of sporozoites in glutaraldehyde fixed wells.
Assessment of number of exoerythrocytic forms (EEFs)
HepG2 cells (105) or 6 × 104 mouse primary hepatocytes (wild type or MyD88−/−) were plated in 8-well Labtek chamber slides for 24 or 48 h respectively, followed by infection with 2.5 × 104 sporozoites for 40 h, or as otherwise indicated. Staining of EEFs was performed as previously described (3). In brief, cells were first washed in PBS and then fixed in ice cold 100% methanol for 15 min on ice, followed by a blocking step (1% BSA in PBS for 30 min). After washing twice in PBS, cells were incubated with the monoclonal mouse anti-Plasmodium hsp70 Ab (2E6) (30) for 45 min at room temperature. Cells were again washed twice and incubated with an anti-mouse IgG Alexa Fluor 488-conjugated Ab in PBS-1% BSA for 45 min at room temperature. Cells were examined with a fluorescence microscope (DC200, Leica). EEFs in each well were counted and expressed as EEFs plus SD of duplicate wells.
51Chromium release assay
A total of 105 HepG2 cells were plated per well in a 1% gelatin-coated 96-well plate and incubated overnight at 37°C. Cells were incubated with 75 μCi 51Cr for 1 h. After washing repeatedly, the cells have been incubated with either spz, sgm, inactivated spz, or Spect−/− spz for 3 h. Alternatively, the cells were wounded mechanically. Supernatant was collected and released. 51Cr was assessed with a gamma counter. The percentage of sporozoite-specific release of chromium was calculated as follows: [(experimental counts − spontaneous counts)/(maximum counts − spontaneous counts)] × 100.
Nitrite assay (Griess reaction assay)
NO is rapidly converted into the stable end products nitrite and nitrate. Total nitrite and nitrate concentrations of infected hepatocyte culture supernatants were measured by the Griess reaction. In brief, 100 μl of culture supernatant were mixed with 10 μl of 30% (w/v) ZnSO4 solution to precipitate total proteins. After 30 min of incubation at room temperature, the solution was centrifuged at 3000 rpm for 5 min. To convert nitrate to nitrite, total supernatant was taken and incubated with 0.5 g of CuSO4 coated cadmium beads for 60 min under shaking conditions. Supernatant was then centrifuged again and an equal volume of Griess reagent (0.5% sulfanilamide, 2.5% H3PO4, and 0.05% naphthylethylene diamine in H2O) was added and incubated for 10 min at room temperature in the dark. Absorbance was assessed at 550 nm and compared with a standard curve obtained using sodium nitrite.
RT-PCR and real time PCR
A total of 106 primary hepatocytes were infected with 105 sporozoites for 24 h. Total RNA was isolated using TRIzol reagents according to the manufacturer’s instructions (Sigma-Aldrich). Two μg of total RNA were reverse transcribed using a commercial kit as suggested by the manufacturer (Promega). The primers and PCR conditions have been described previously (31, 32). The primers used were β2-microglobulin (β2M) forward (5′-GGCTCGCTCGGTGACCCTAGTCTTT-3′) and reverse (5′-TCTGCAGGCGTATGTATCAGTCTCA-3′), iNOS forward (5′-GCATGGACCAGTATAAGGCAAGCA-3′) and reverse (5′-GCTTCTGGTCGATGTCATGAGCAA-3′). The amplification conditions used were: 1 min at 94°C, 1 min at 60°C, and 1 min at 72°C for 30 cycles for β2M; 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C for 35 cycles for iNOS; and a final extension of 10 min at 72°C.
In vivo assessment of parasite loads was performed with real time PCR as described previously with minor modifications (4). In brief, C57BL/6 mice were pretreated with BAY11-7082 (20 mg/kg body weight) or DMSO as control for 1 h i.p. followed by injection of 30′000 sporozoites i.v. Alternatively, C57BL/6 wild type or MyD88−/− mice (33) were injected with the same amount of sporozoites without pretreatment. Forty hours later, total livers were isolated and total RNA extracted. cDNA was synthesized as described above. The primers used were specific for the P. berghei 18S rRNA (forward: 5′-AAGCATTAAATAAAGCGAATACATCCTTAC-3′ and reverse: 5′-GGAGATTGGTTTTGACGTTTATGTG-3′). The DNA was amplified in the LightCycler 2.0 Instrument (Roche Diagnostics) using the program Roche LightCycler Run 5.32.
Assessment of parasitemia
Wild type, MyD88−/−, or iNOS−/− (C57BL/6-Nos2tm1Lau; The Jackson Laboratory) mice were injected i.v. with 103 sporozoites. At different time points, parasitemia was assessed by Giemsa-stained blood smear or alternatively by real-time PCR. Fifty microliters of total blood was taken and total RNA was isolated and reverse transcribed as mentioned above. The parasites were detected through 18S rRNA and normalization was done by β2-microglobulin (β2M) detection (as mentioned above).
Sporozoite-mediated hepatocyte wounding releases cytosolic factors into the microenvironment
Sporozoites travel through several hepatocytes before infection and in the process cells are wounded by transient rupture of their membranes (1). Cells which fail to regain their membrane integrity die by necrosis, which results in an uncontrolled release of cytosolic factors into the extracellular environment. To monitor the kinetics of sporozoite-mediated cellular wounding, HepG2 cells were incubated with spz in the presence of PI, which is taken up by cells that lose their membrane integrity either transiently, or in the case of necrosis, permanently. PI positive nuclei (red) were recorded every 20 s. Fig. 1,A shows distinct time points of the wounding kinetics. The first wounded cells appeared 30 min after addition of sporozoites and their numbers increased with time. Interestingly, the total wounding process, beginning from the appearance of the first wounded cell, lasted exactly 60 min. The percentage of wounded cells was assessed in HepG2 cells 2 h after addition of either spz, mosquito sgm as negative control, or heat-inactivated sporozoites or of sporozoites deficient in cell traversal motility (Spect−/−). Fig. 1,B shows, as expected, significant wounding occurred when incubated with spz, but not with sgm and Spect−/−. Instead, incubation with heat-inactivated sporozoites led to a moderate but significant percentage of wounded cells. To assess the release of cytosolic factors following infection, HepG2 cells were labeled with radioactive 51Cr, which was detected in the medium after cellular wounding. The amount of 51Cr, and thus cytosolic factors, released was directly proportional to the number of sporozoites added to the cells (Fig. 1,C). Compared with the incubation with spz, the treatment of HepG2 cells with sgm and Spect−/− did not lead to a significant release of cytosolic factors (Fig. 1,D). Again, treatment with heat-inactivated sporozoites led to the release of a small, significant amount of endogenous material. Due to the residual traversal activity of heat-inactivated sporozoites, further control experiments were performed with Spect−/− sporozoites. As demonstrated in Fig. 1 E, mechanical wounding of cells led to a 6-fold increase in 51Cr-release compared with the release caused by sporozoites.
Cytosolic factors activate NF-κB in cells bordering wounded hepatocytes
Tissue damage and the accompanied release of endogenous factors provoke an inflammatory response mediated by NF-κB (8, 9). NF-κB is tightly regulated by the IκBα. From a variety of signals, IκBα is phosphorylated, ubiquitinated, and thus inactivated, resulting in the release and translocation of active NF-κB from the cytosol to the nucleus where transcription is initiated. To assess whether sporozoite infection activates NF-κB, HepG2 cells were transfected with a NF-κB luciferase reporter construct and either remained untreated, or were incubated with spz or sgm. Sporozoite infection caused a significant induction in NF-κB activity, unlike sgm treated cells (Fig. 2,A). In addition, compared with wild-type spz, Spect−/− spz failed to activate NF-κB (Fig. 2,B). To test whether sporozoite-induced hepatocyte damage, and thus the release of endogenous factors, activates NF-κB, conditioned media derived from filtered supernatant of infected hepatocytes (SN spz) or filtered supernatant of mechanically wounded cells (SN MC) were added to uninfected cells. Both conditions led to a significant activation of NF-κB (Fig. 2,C). To exclude the introduction of unspecific activators during the generation process of the conditioned media used above, conditioned media from sgm- and Spect−/−-treated cells were tested. None of these media activated NF-κB significantly (Fig. 2 D). As positive control TNF-α-treated cells were used.
Overexpression of dominant negative, phosphorylation-deficient IκBα blocked NF-κB completely (Fig. 2,E). This demonstrates that NF-κB activation evolves via the classical pathway including phosphorylation and ubiquitination of IκBα followed by its degradation. In contrast, overexpression of RelA, the p65 subunit of the NF-κB heterodimer, led to a 2-fold increase in sporozoite-induced NF-κB activation over that observed in control (pcDNA3.0) transfected cells (Fig. 2 F).
The kinetics of sporozoite-induced phosphorylation of IκBα in hepatocytes showed that IκBα phosphorylation could be detected as early as 30 min after infection of HepG2 cells with sporozoites (Fig. 2,G). As a positive control for IκBα phosphorylation, we used Jurkat cells treated with PMA/ionomycin for 5 min (Fig. 2,E, first two lanes). In agreement with our observation in HepG2 cells, sporozoites but not sgm or Spect−/− sporozoites led to phosphorylation of IκBα in mouse primary hepatocytes (Fig. 2, H and I).
To demonstrate that wild-type and Spect−/− sporozoites equally infect hepatocytes, and that NF-κB activation does not depend on infectivity, the number of infected cells was assessed 2 h post-infection. No differences in the number of infected hepatocytes was scored between wild type and Spect−/− sporozoites (Fig. 2 K).
To assess the ability of hepatocytes bordering injured cells to activate NF-κB in response to infection, we analyzed the subcellular localization of the p65 subunit of the NF-κB heterodimer in infected mouse primary hepatocytes. In uninfected cells, p65 (green) is predominantly located in the cytoplasm (Fig. 2,L). However, 2 h after addition of sporozoites, nuclear translocation of p65 (white arrows) was observed in cells neighboring wounded cells (PI stained nuclei; red) but not in distant cells. Interestingly, >90% of wounded cells did not show any nuclear p65. In addition, their nuclei were often smaller than the nuclei of healthy cells. We observed that cells containing translocated p65 (red) were potential target cells for final parasite (green) infection (Fig. 2 M). In fact, we found that NF-κB was activated in around 35% of infected hepatocytes.
These observations showed that activation of NF-κB is an immediate early response induced by hepatocyte-derived factors which are released during the process of sporozoite cell traversal.
Sporozoite infection-mediated NF-κB activation negatively interferes with parasite development
Recently, the chemical compound Bay11-7082 has been described as being a selective inhibitor of cytokine-induced IκBα phosphorylation (34). We therefore analyzed its effect on Plasmodium-induced IκBα phosphorylation. Bay11-7082 potently inhibited parasite-mediated IκBα phosphorylation in mouse primary hepatocytes, while treatment with solvent alone (DMSO) showed no reduction of phosphorylated IκBα (Fig. 3,A). As expected, Bay11-7082 but not DMSO inhibited parasite-induced NF-κB luciferase activity in HepG2 cells (Fig. 3,B). To investigate the influence of Plasmodium-mediated NF-κB activation on liver-stage development of the parasite, we infected HepG2 cells, which were either untreated or treated with BAY11-7082 or DMSO and assessed the number of Plasmodium-infected cells at different time points after infection. Results demonstrated that in untreated or control-treated (DMSO) hepatocytes, the number of EEFs decreased with time (Fig. 3,C). To the contrary, in BAY11-7082-treated hepatocytes the loss of infected hepatocytes was less pronounced. In contrast, the number of Spect−/− EEFs did not change between untreated, BAY11-7082 or DMSO treated cells 45 h post-infection (Fig. 3,D). Because Spect−/− sporozoites fail to activate NF-κB, we compared the number of EEFs between wild-type and Spect−/− sporozoites 40 h post-infection. Interestingly, the number of Spect−/− EEFs was significantly higher compared with the number of wild-type EEFs (Fig. 3 E).
To assess the significance of NF-κB in vivo, we infected mice with 30′000 sporozoites, which were either pretreated with DMSO as a control or with the IκBα phosphorylation inhibitor Bay11-7082 for 1 h. Liver infection was assessed 40 h post-infection by real-time PCR. Bay11-7082 significantly enhanced liver infection compared with control (DMSO)-treated mice (Fig. 3 F).
Taken together, these results demonstrated that there is a hepatocyte-mediated response against Plasmodium infection which is regulated by NF-κB, and that NF-κB activation can contribute to the reduction of parasite infection.
NF-κB-regulated iNOS expression in hepatocytes reduces the number of infected cells
Increased iNOS expression has been observed upon tissue wounding (18, 19, 20). Indeed, iNOS-derived NO regulates wound contraction, collagen formation, and cell proliferation. In addition, NO has strong antimicrobial activity and its importance in antimalarial immunity has been well documented (22, 23). To assess the ability of sporozoites to induce iNOS expression in hepatocytes, HepG2 cells were incubated with either sporozoites or salivary gland material of uninfected mosquitoes (sgm) and iNOS expression was assessed at different time points. Significant sporozoite-mediated iNOS expression was detectable in HepG2 cells as early as 12 h post-infection, and became much more pronounced at 24 h after infection, while iNOS induction in cells treated with sgm only was close to background levels (Fig. 4,A). In addition, incubation with Spect−/− sporozoites failed to activate iNOS expression (Fig. 4,B). Although it had been previously shown that P. falciparum induces iNOS expression in human primary hepatocytes (35), the molecular regulation of Plasmodium-mediated iNOS expression was not explored. To determine the extent to which this iNOS expression and activity is dependent on NF-κB, we infected primary mouse hepatocytes, either in presence or absence of BAY11-7082 or DMSO, and analyzed iNOS induction by RT-PCR (Fig. 4,C), Western blot (Fig. 4,D), and iNOS enzyme activity by the nitrite assay (Fig. 4,E). All three methods revealed that iNOS expression in hepatocytes after sporozoite infection was NF-κB dependent. Furthermore, Fig. 4,E showed that sporozoite-induced iNOS enzyme activity can be blocked selectively by the chemical inhibitor S-methylisothiourea sulfate (SMT). To assess whether sporozoite-induced NO production kills intrahepatic parasites, we infected mouse primary hepatocytes with sporozoites and treated the cells with SMT, 12 h post-infection. The total number of infected hepatocytes was assessed by immunostaining 40 h later. As illustrated in Fig. 4,F, the number of wild-type EEFs was significantly higher in cells treated with SMT compared with untreated cells, most likely because SMT prevented the antimicrobial activity of iNOS. In contrast, no difference in the number of Spect−/− EEFs was observed between untreated cells and cells treated with SMT (Fig. 4 G).
MyD88-dependent activation of NF-κB and iNOS expression upon hepatocyte wounding
TLRs play a crucial role in the initiation of an effective immune response against various pathogens (12). In addition, TLRs sense endogenous self Ags such as necrotic cell-derived factors, which provoke inflammation (8). Signaling of distinct TLRs is dependent on the adapter protein MyD88 (12). Therefore, to test whether sporozoite-mediated NF-κB activation is TLR-dependent, we overexpressed dominant negative (DN) MyD88 in HepG2 cells and infected them (spz), or treated them with conditioned media of either infected hepatocytes (SN spz) or mechanically wounded cells (SN MC). Unlike in control transfected cells (pcDNA3.0), overexpression of DN MyD88 inhibited spz-, SN spz-, and SN MC-induced NF-κB activation completely (Fig. 5,A). NF-κB activation after sgm and Spect−/− treatment was compared between control cells and DN MyD88 overexpressing cells. As expected sgm and Spect−/− treatment did not lead to activation of NF-κB (Fig. 5,B). As positive control, TNF-α-treated cells were used. To further confirm the results observed in HepG2 cells, we infected wild-type or MyD88−/− mouse primary hepatocytes and detected endogenous levels of P-IκBα (Fig. 5,C) and iNOS protein expression (Fig. 5,D). Compared with wild-type hepatocytes, in MyD88−/− cells the levels of P-IκBα and iNOS were similar to background levels after spz infection or SN spz or SN MC treatment. The number of infected cells was assessed in wild-type and MyD88−/− hepatocytes 40 h post-infection. Consistent with our expectations, the number of wild-type EEFs in MyD88−/− cells was significantly higher compared with wild-type cells (Fig. 5,E), whereas the number of Spect−/− EEFs did not differ between wild type cells and MyD88−/− cells (Fig. 5F). To assess the significance of MyD88 in vivo, wild-type and MyD88−/− mice were infected with 30′000 sporozoites and the liver infection rate was assessed 40 h post-infection by real-time PCR. As predicted, the level of infection was significantly higher in MyD88−/− mice than in wild-type mice (Fig. 5 G). Our data clearly demonstrate that sporozoite-mediated cellular wounding and its associated release of cytosolic factors result in activation of NF-κB and iNOS expression through the activation of TLRs, thereby limiting the number of infected cells.
MyD88 and iNOS influences early blood-stage infection
To assess whether the MyD88- and iNOS-dependent decrease in liver infection influences the course of the disease, we infected either wild-type, iNOS-, or MyD88-deficient mice with 1000 sporozoites and assessed parasitemia. As shown in Fig. 6,A, parasitemia is similar in all three mouse strains tested except at early time points where parasitemia is significantly higher in MyD88 and iNOS mice compared with wild-type mice as confirmed by real-time PCR (Fig. 6 B).
The role of hepatocyte traversal by sporozoites in malaria infection has been addressed in the past. A current model claims that cell traversal is beneficial and necessary for final infection, because it triggers the switch from cell passage to cell invasion motility (2). In support of this model, studies have shown that hepatocyte growth factor, secreted by wounded cells, binds its receptor Met on neighboring cells and subsequently renders them susceptible for infection (3). However, the dependence of infection on host factors released during cell traversal has been disproved by Spect−/− sporozoites that are deficient in cell traversal, but infect hepatocytes just as well as wild-type sporozoites (5). In support of the latter findings, we show in this study that cell traversal by sporozoites has a detrimental effect on parasite survival, because nontraversed hepatocytes have the ability to respond to released host factors resulting in a NF-κB-dependent innate immune response and reduced infection. Indeed, NF-κB activation was found to be an early event that occurred at the same time as appearance of the first wounded cells (Figs. 1,A and 2,G), suggesting that rather than the parasite itself, the cell traversal activity of the sporozoites is the essential stimulus for the immune response. This is supported by the observation that compared with wild-type sporozoites, Spect−/− sporozoites failed to activate NF-κB significantly (Fig. 2, B and I). Surprisingly, heat-inactivated sporozoites still retain residual traversing activity and could not serve as a proper control in these studies (Fig. 1, B and D). Our observations therefore exclude the possibility that sporozoite-secreted proteins alone, such as the CSP and thrombospondin-related adhesion protein, or a simple contact of parasite and host cells may be a sufficient trigger to activate NF-κB. The liver has long been considered a tolerogenic organ that favors the induction of peripheral tolerance because it is in constant contact with microbial products derived from the gut. This may explain the unresponsiveness of hepatocytes to Spect−/− sporozoites. In contrast, tissue damage can break this tolerance, generating an environment that promotes tissue recovery and a response against pathogens. Indeed, supernatant derived from mechanically wounded cells activated NF-κB to a similar level as supernatant from sporozoite-damaged hepatocytes, which showed a 6-fold lower 51Cr release (Figs. 1,E and 2 C). We can therefore hypothesize that cell wounding initiates NF-κB activation and breaks tolerance, thus sensitizing hepatocytes to the presence of parasites, leading to more NF-κB activation in a synergistic manner. The difference in 51Cr release between mechanically wounded cells and sporozoite-damaged hepatocytes may not be indicative for the actual amount of NF-κB activating material. Alternatively, mechanical wounding of cells differs from parasite-mediated wounding which may result in the release of different hepatocyte factors with distinct efficiencies in activating NF-κB. A proteomic approach to address this question is envisaged.
It is interesting to note that NF-κB was not activated in wounded cells (Fig. 2 L), thus excluding an autocrine activation, and this might be an indication of lethal damage. Although it has been shown that many cells are able to reseal their membranes upon injury, their subcellular architecture is disrupted (36). Therefore, it is very doubtful whether sporozoite-traversed cells fully recover. In addition, wounded cells contain free sporozoite-derived proteins in the cytoplasm, which might potently interfere with important host cell functions. Indeed, recently it has been shown that overexpression of CSP antagonizes NF-κB translocation to the nucleus in infected cells upon TNF-α stimulation, which consequently inhibits the expression of proinflammatory molecules (37). This seems to contradict our observations. However, whereas the CSP in wounded cells is freely in the cytosol and thus able to act immediately, the situation in an infected neighboring cell is different in the sense that the CSP has to be actively transported through the parasitophorous vacuole to block the translocation of activated NF-κB in the cytosol. Our data suggest that the signal induced by endogenous factors evolves chronologically earlier than that associated with CSP. CSP however may control the magnitude of the NF-κB response and block later NF-κB-inducing stimuli.
NF-κB activation evolved in a MyD88-dependent manner indicating the so-far unknown competency of hepatocytes to respond to endogenous factors via TLR family members (Fig. 5, A and C). However, further work is needed to elucidate which members of the TLR family are implicated in NF-κB activation during sporozoites infection. In addition, we cannot exclude a role of IL-1R and IL-18R in NF-κB activation, because downstream signaling pathways of these receptors also require MyD88.
The activation of NF-κB and subsequent expression of iNOS upon hepatocyte wounding eliminated about a third of all exoerythrocytic forms. This decrease is in accordance with the observation that NF-κB was activated in ∼35% of infected hepatocytes (Fig. 2,M). These cells were predominantly directly bordering wounded cells (Fig. 2,L). Consistently, because Spect−/− sporozoites do not activate NF-κB, their number of EEFs was significantly higher than that of the wild-type-derived EEFs. This result conflicts the observations by Ishino et al. (5) where the number between wild-type and Spect−/− EEFs did not differ. This discrepancy can be explained by the fact that different reagents were used to assess the number of exoerythrocytic forms. Whereas Ishino et al. (5) used the CSP as an indicator for EEFs, parasite-expressed Hsp was used in this study. Hsp is expressed in sporozoites in low quantities and we observed that during EEF development the expression is greatly enhanced (data not shown). Therefore, using the Hsp allows a more accurate assessment of EEF development than using the CSP, which is expressed exclusively on sporozoites. The NF-κB- and iNOS-mediated decrease in liver infection is reflected in the significantly higher parasitemia in iNOS- and MyD88-deficient mice compared with wild-type mice at early time points (Fig. 6 B).
The partial inhibition of parasite development discussed above is in line with the notion of concomitant evolution of pathogens and hosts. In fact, various mechanisms have evolved to strike a balance between survival and death of pathogens and hosts. One of these mechanisms is the interference with TLR signaling to control the magnitude of the immune response and so avoid detrimental outcomes, such as severe host tissue destruction or total eradication of the pathogen. Therefore, TLR-mediated signaling events contain negative feedback mechanisms. For example, soluble decoy TLRs (sTLR-2/-4) antagonize ligand binding and potently attenuate TLR-induced effector functions (38, 39). Different intracellular negative regulators have been uncovered such as MyD88s, IRAKM, TOLLIP, A20, SOCS1, PI3K, and NOD2 (40, 41, 42). Thus, pathogens may target these molecules to ensure their survival. Interestingly, PI3K is activated in hepatocytes early upon infection (4). To date, it is unclear whether it might be a candidate to limit a TLR-mediated immune response against the parasite. As already discussed above, an additional candidate is the CSP (37).
The Spect−/− protein is essential for the sporozoite to cross the sinusoidal cell layer to reach the hepatocytes. However, the reason why sporozoites continue traversing hepatocytes remains unclear. Even though Spect−/− sporozoites infect hepatocytes just as well as wild-type sporozoites (Fig. 2 K), we cannot exclude the possibility that sporozoites become activated for final infection through host cell traversal activity as suggested by a previous study (2). It is possible that regulated exocytosis of proteins induced by hepatocyte traversal may be necessary to antagonize the effect of Spect. An alternative hypothesis is that traversal activity is used by sporozoites to divert the immune system from the final infected hepatocytes. Indeed, traversed hepatocytes have been shown to present the CSP to specific T cells and undergo apoptosis (43). The price to pay for this possible decoy mechanism is a lower parasite infection rate.
In summary, pathogen-mediated tissue destruction induces an innate immune response and Plasmodium sporozoite-mediated hepatocyte wounding is no exception, with the consequence of activating nearby hepatocytes that respond by limiting the extent of parasite infection.
We thank Dr. Thomas Brunner and Dr. Pascal Schneider for providing expression and reporter plasmids, Dr. Olivier Gaide and Dr. Moriya Tsuji for Abs, Dr. Jürg Tschopp for providing the MyD88−/− mice and Shizuo Akira for the generation of these mice, Dr. Robert Ménard for the Spect−/− parasites and Masao Yuda for generation of these parasites, and Dr. Stephan Duss for technical assistance.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Lausanne University Institutional funds.
Abbreviations used in this paper: hsp, heat-shock protein; iNOS, inducible NO synthase; spz, sporozoites; sgm, homogenized salivary gland; PI, propidium iodide; DAPI, 4′,6-diamidino-2-phenylindole, dihydrochloride; PbCSP, P. berghei circumsporozoite protein; EEFs, exoerythrocytic form; SN spz, supernatant of infected hepatocyte; SN MC, filtered supernatant of mechanically wounded cell; SMT, S-methylisothiourea sulfate; DN, dominant negative; CSP, circumsporozoite protein.