Plasmodium yoelii Uses a TLR3-Dependent Pathway to Achieve Mammalian Host Parasitism

Malaria is a parasitic disease due to Plasmodium sp. infection transmitted to the vertebrate host by an infected female Anopheles mosquito bite during a blood meal. In 2017, 219 million cases of malaria were reported with an assessed 435,000 deaths (1). The parasite life cycle within mammals is characterized by two obligatory replicating phases, the asymptomatic intrahepatic stage and the erythrocytic stage, which is responsible for clinical symptoms of malaria. In light of the complexity of the Plasmodium developmental cycle within the vertebrate host, different immune response components evidently operate at different stages; some of them are protective, and others help the parasite to evade those measures and induce severe pathological conditions. An optimal immune response against Plasmodium requires years of exposure to infection and the induction of innate and adaptive cellular mechanisms to achieve protection along with parasite-specific and self-reactive Abs and strongly balanced pro- and anti-inflammatory cytokines responses (2–5).

The sensing of the innate immune response during primary infection with sporozoites, including NK cells and NKT cells, during host–parasite interactions is initiated by pattern recognition receptors by accommodating ligands such as pathogen-associated molecular patterns (6–9). TLRs are the first and best characterized innate immune receptors involved in the uptake and the processing of various pathogen-associated molecular patterns expressed at the endosomal membrane (10). TLRs are composed of an extracellular leucine-rich repeats motif, a transmembrane region, and an intracellular conserved Toll and IL-1 receptor (TIR) domain (11, 12). All the TLRs, except TLR3, use MyD88-dependent pathway, which in turn induces the cascade of activation of transcription factors such as TRAF6 and NF-kB (13, 14). TLR4 is the only TLR able to use either MyD88 and/or TRIF pathway (13). TLR3 engages a MyD88-independent pathway exclusively mediated by the TRIF adapter (15), also named TIR domain–containing adaptor protein inducing IFN-β. Further, the TLR3–TRIF interaction leads to a downstream signaling of NF-kB and IFN regulatory factor (IRF) 3 (16). Induced NF-kB pathway favors the expression of a subset of primary response genes including proinflammatory cytokines, whereas IRF7/IRF3 favor type I IFNs (17, 18) as well as numerous IFN-stimulated genes.

Several studies done in vitro have shown that Plasmodium ligands such as glycosylphosphatidylinositols and hemozoin, microparticles released from infected RBCs (iRBCs), play a significant role in the induction of the immune response through MyD88, TLR2, TLR4, and TLR9 (19–22). By contrast, various observations done in vivo suggested that the deficiency in TLR4, TLR9, or TLR2/4/9 has no impact on parasitemia or survival (23–25). Therefore, the MyD88 downstream signaling pathways may involve additional receptor(s) for the recognition and the control of parasitemia in P. yoelii–infected mice. It has been shown that mice deficient in IRF3 displayed marked impairment in the control of parasite burden in the liver upon secondary P. yoelii sporozoite infection (26). Furthermore, reduced splenic IFN-β production was found in both IRF3- and IRF7-deficient mice when exposed to iRBCs or AT-rich motif proteins derived from the genome of P. falciparum. Altogether, these argue in favor of a role of IRF3 and IRF7 in IFN-β production (27). These observations also suggest a possible role for TLR3 during Plasmodium infection through the initiation of complex circuits and signals of the immune response, which may contribute to the host–parasite fitness, at the earlier steps of infection and hence deserves further attention.

In this study, we assessed molecular mechanisms implied in the sensing, initiation, and regulation of innate and adaptive cellular networks during Plasmodium infection. Using TLR-deficient mouse lines on the C57BL/6 background, we found that TLR3-triggered mechanisms play a role in the fine-tuning and function of immune responses associated with P. yoelii 265 BY primary infection induced either by sporozoites or blood stages. Clearly, we showed that early during the infectious process, P. yoelii liver stage is associated with an activation of the TLR3 pathway that promotes efficient blood stage infection through the inhibition of the production of a protective Ab response. Altogether, these data strongly suggest possible strategies implied by P. yoelii to manipulate the host to its own advantage through TLR3-dependant molecular pathways.

C57BL/6jRj (B6) mice were purchased from Janvier Labs. C57BL/6 TLR-deficient mice lines were provided by Dr. M. Chignard and Dr. M. Si-Tahar (Institut Pasteur, Paris, France). TLR-deficient lines were generated as published: TLR3 knockout (KO) (28), TLR2 KO (29), TLR4 KO (30), TLR 2–4 double KO (31), and TLR9 KO (32). Animals were bred and maintained under specific pathogen-free conditions at Institut Pasteur de Lille animal facility accredited by the French Ministry of Agriculture to perform experiments on live mice in line with the French (decree 87-848 issued on October 19, 1987) and European (directive 86/609/CEE) regulations on the care and protection of laboratory animals. Protocols were approved by the Ethical Committee of the Institut Pasteur de Lille and the French Ministry of Agriculture (permit number A 75-15-28) and performed in compliance with the National Institutes of Health Animal Welfare Assurance no. A5476-01 issued on February 7, 2007. All efforts were made to minimize animal suffering.

P. yoelii yoelii (nonlethal 265BY strain) sporozoites were obtained by dissecting the salivary glands of infected A. stephensi mosquitoes as previously described (33). The mosquitoes were bred, maintained, and infected at the insectarium of Pasteur Institute of Lille and at the INSERM U1135 (Centre d’Immunologie et des Maladies Infectieuses, Paris, France). For the liver stage study, mice were infected i.v. with 40,000 sporozoites diluted in PBS. For the erythrocytic stage study, mice were infected i.v. with 4000 sporozoites diluted in sterile PBS or i.p. with 10⁶ iRBCs. Parasitemia were measured on Giemsa-stained thin blood smears. Results are expressed as the percentage of iRBCs.

Mice were infected with sporozoites (40,000) or iRBCs (1 × 10⁶) transgenic PyGFP-LUC strain that constitutively expresses luciferase (replacement of PyP230 nonessential gene by a GFP-LUC cassette) (34) and were imaged using an IVIS Lumina XR system (PerkinElmer, Villebon-sur-Yvette, France). Mice were injected i.p. with d-luciferin (150 mg/kg in DPBS; PerkinElmer) and were anesthetized using the isoflurane anesthesia system (Tem Sega, Pessac, France) at 42 h postinfection (pi) (hepatic stage) and 7, 14, and 21 d pi during blood stage. Measurements were performed within 5 min of injection, and bioluminescence imaging was acquired with an exposure time of 60 s. Bioluminescence generated by luciferase-expressing transgenic parasites was measured using the same region of interest for all samples being compared prior to imaging. To standardize imaging and to allow comparison between mice, the images presented in the figures were taken once luminescence plateaued. For liver bioluminescence imaging, mice were sacrificed on 42 h pi, and the whole livers were excised after imaging whole mice. Imaging data were analyzed using Living Image 4.3.1 (PerkinElmer).

RNA was isolated from total hepatic cells 42 h post–sporozoite infection and splenic cells using RNeasy Mini Kit (QIAGEN, Hombrechtikon, Switzerland) at days 7, 14, and 21 pi. Reverse transcription was performed using SuperScript VILO cDNA Synthesis Kit (Life Technologies, St. Quentin Fallavier, France) with the following program: 10 min at 25°C, 60 min at 42°C, and 5 min at 85°C. Quantitative PCR analyses were done using the SYBR Green PCR Master Mix (Applied Biosystems, Villebon-sur-Yvette, France). The thermal cycling conditions were composed of 50°C for 2 min followed by an initial denaturation step at 95°C for 10 min, 45 cycles at 95°C for 15 s, and 60°C for 60 s. We used the respective primers listed in Supplemental Table I.

Liver lysate extracted from C57BL/6 and TLR3 KO mice, infected or not, were resuspended in 500 μl in lysis buffer (RIPA lysis buffer; Interchim, Montluçon, France) with a mixture of antiproteases (Roche, Boulogne-Billancourt, France) and then crushed. Proteins concentrations were evaluated by the BC Assays method using the Protein Quantification Kit (Interchim). Each sample was resuspended in Laemmli buffer (Bio-Rad Laboratories, Marnes-la-Coquette, France) at the protein concentration of 1 mg/ml. Proteins were separated on 10% acrylamide gels (Bio-Rad Laboratories) by SDS-PAGE and electrotransferred on 0.2 μm nitrocellulose membrane (Bio-Rad Laboratories). A Precision Plus Protein All Blue Standard (Bio-Rad Laboratories) was used to follow the separation of proteins. Membranes were blocked 1 h with TBS 1×–milk 5% and then incubated with primary Abs anti-IRF3 (1:8000), anti-IRF7 (1:1000), anti–NF-κBp65 (1:500), anti-TRIF (1:500), anti-NAK/TBK-1 (1:500), anti-TLR3 (1:500) (Abcam, Paris, France), and anti–β-actin (1:1000) (Cell Signaling Technology) using an immunoblot cassette. After overnight incubation at 4°C, membranes were washed twice with TBS 1× Tween 20 0.1% buffer for 5 min and secondary Abs anti–IgG-rabbit and/or anti–IgG-mouse coupled at a HRP (1:10,000) (Abcam) were added for 1 h at room temperature. The blots were revealed with chemiluminescent substrate Clarity Western ECL Substrate (Bio-Rad Laboratories) and analyzed by the Molecular Imager ChemiDoc XRS+ system (Bio-Rad Laboratories) using the Image Lab software (Version 5.0, Image Lab software; Bio-Rad Laboratories). The densitometry of proteins was also quantified by the Image Lab software (Version 5.0, Image Lab software; Bio-Rad Laboratories) whose intensities were normalized at the β-actin intensity.

Procarta Plex Mouse IFN-β Simplex was used to quantify IFN-β in sera from B6 and TLR3 KO controls and infected mice. Steps were followed from the guide kit.

Mice were bled at the indicated days, and sera were stored at –20°C before use. Mouse IFN-γ ELISA Set (catalog no. 555138) from BD Biosciences was used to quantify the cytokine following the manufacturer’s instructions. Duplicate serial dilutions were performed for each serum, and OD means were used to determine the concentrations of cytokines in the samples according to the standard.

Spleen from controls and infected mice were collected at days 7, 14, and 21 pi and were then treated for 30 min at 37°C with collagenase IV (1 mg/ml). After homogenization, cells were washed by centrifugation at 1600 rpm for 5 min at 5°C. Erythrocytes were then lysed using ammonium–chloride–potassium lysis buffer (0.15 M NH₄Cl, 10 mM KHCO₃, and 0.1 mM Na₂EDTA), and cells were resuspended in PBS–3% FCS before counting living cells in trypan blue. Splenic cells were stained at 4°C in PBS–FCS 3% using different a mix of mAbs purchased from BD Biosciences (unless otherwise specified): LIVE/DEAD (Aqua LIVE/DEAD), anti-CD19 (allophycocyanin), anti-CD3 (PeCF594), anti-NK1.1 (PeCy7), anti-CD8 (V450), and anti-CD4 (Alexa Fluor 700). Fixation and intracellular staining were performed following BD Biosciences instruction. Briefly, cells were incubated for 30 min in a Cytofix/Cytoperm buffer (BD Cytofix/Cytoperm Kit 554514) and, intracellular staining with PE-labeled anti-TLR3 mAb (BioLegend) or isotype-matched control mAb PE (rat IgG2a) were incubated with cells in Perm Wash Buffer. The cell pellet was then resuspended in PBS FCS 10% before the acquisition on a LSRFortessa Flow Cytometer with a stopping gate at 5 × 10⁵ events.

Blood of infected mice was enriched in iRBC using a Percoll solution centrifuged at 2500 × g for 30 min. After verification of the iRBC enrichment by Giemsa staining on blood smears, cells were incubated with a lysing buffer containing Triton X-100 and protease inhibitor tablet (Roche cOmplete) for 10 min at 4°C, and then the solution was sonicated at 4°C for 15 s twice. Protein quantity was determinate by Bradford assay.

Total IgG and specific IgG against P. yoelii 265 BY proteins were quantified by sandwich ELISA. Briefly, 96-well plates were coated, respectively, by an anti-mice total IgG (ab97042) at 1 μg/ml at 2 μg/ml and RBC or iRBC proteins extract at 5 mg/ml. Plates were saturated by 1% gelatin and incubated with serums from infected mice diluted in PBS–0.1% Tween following 3-fold dilution from 1/10,000 for IgG. The revelation step was performed with an mAb against mice IgG (A4312 dilution 1/10,000; Sigma) with alkaline phosphatase. Plates were washed between each step by PBS–0.1% Tween. IgG optical densities were reported to a standard curve obtained by dilutions of mouse IgG1, kappa monoclonal [MOPC-21] - isotype control ab18443, respectively.

To analyze IgG antigenic-specific repertoire against P. yoelii 265 BY, we used proteins extracted from iRBC. Then, 1 mg of iRBC proteins was first denaturized for 5 min at 95°C with Laemmli buffer and separated by NaDodSO₄ polyacrylamide 8% gel electrophoresis (SDS-PAGE). Second, quantitative immunoblotting (35) (Panama blot method) was performed with different samples justified at 1 mg/ml of IgG and diluted at 1/20 in PBS Tween 20 (0.1%). Samples were incubated overnight at 4°C. A pool of all serums was used to normalize as a positive control, and PBS was used as a negative control. After incubation with mAb against mouse IgG coupled to alkaline phosphatase (A4312 dilution 1/10,000; Sigma), the membrane was incubated with 75 μl of NBT-BCIP (2:1) substrate (S3771; Promega) diluted in 15 ml of alkaline phosphatase buffer (100 mM NaCl, 5mM MgCl₂, and 100 mM Tris [pH 9.5]) to reveal a quantitative signal. Proteins were revealed using PROTOGROLD solution. Membranes were analyzed using IGOR 3.16 software (WaveMetrics, Lake Oswego, OR).

Protein band was localized after 10% acrylamide gel SDS-PAGE separation of the iRBC protein extract. Asymptomatic individuals exhibiting the reactivity of interest were tested. Half of the gel was used to transfer the protein onto a nitrocellulose membrane for Western blot analysis. The other half was stained with Coomassie Brilliant Blue R-250 (Bio-Rad Laboratories), the respective bands were cut, and proteins were digested as previously described (36).

UltiMate 3000 RSLCnano System (Thermo Fisher Scientific) was used for separation of the protein digests. Peptides were automatically fractionated onto a commercial C18 reversed phase column (75 μm × 150 mm, 2 μm particle, PepMap100 RSLC column, temperature 35°C; Thermo Fisher Scientific). Trapping was performed for 4 min at 5 μl/min with solvent A (98% H₂O, 2% ACN, and 0.1% formic acid [FA]). Elution was performed using two solvents A (0.1% FA in water) and B (0.1% FA in ACN) at a flow rate of 300 nl/min. Gradient separation was 3 min at 5% B, 37 min at 5% B to 30% B, and 5 min at 80% B and maintained for 5 min. The column was equilibrated for 10 min with 5% buffer B prior to the next sample analysis.

The eluted peptides from the C18 column were analyzed by Q Exactive instruments (Thermo Fisher Scientific). The electrospray voltage was 1.9 kV, and the capillary temperature was 275°C. Full mass spectrometry (MS) scans were acquired in the Orbitrap mass analyzer over mass-to-charge ratio (m/z) 300–1200 range with resolution 35,000 (m/z 200). The target value was 5.00 × 10⁵. Ten most intense peaks with a charge state between two and four were fragmented in the higher-energy collisional dissociation collision cell with a normalized collision energy of 27%, and tandem mass spectrum was acquired in the Orbitrap mass analyzer with resolution 17,500 at m/z 200. The target value was 1.00 × 10⁵. The ion selection threshold was 5.0 × 10⁴ counts, and the maximum-allowed ion accumulation times were 250 ms for full MS scans and 100 ms for tandem mass spectrum. Dynamic exclusion was set to 30 s.

Raw data collected during nano–liquid chromatography–tandem MS analyses were processed and converted into *.mgf peak list format with Proteome Discoverer 1.4 (Thermo Fisher Scientific). Tandem MS data were interpreted using search engine Mascot (version 2.4.0; Matrix Science, London, U.K.) installed on a local server. Searches were performed with a tolerance on mass measurement of 0.2 Da for precursor and 0.2 Da for fragment ions, against a composite target decoy database (36,244 total entries) built with Mus musculus Swiss-Prot database (TaxID = 10,090; July 01, 2015; 16,716 entries) and three strains of P. yoelii yoelii http://PlasmoDB.org database (strains 17×; 17XNL and YM; July 01, 2015; 19,411 entries) fused with the sequences of recombinant trypsin and a list of classical contaminants (46 entries). Cysteine carbamidomethylation, methionine oxidation, protein N-terminal acetylation, and cysteine propionamidation were searched as variable modifications. Up to one trypsin missed cleavage was allowed. For each sample, peptides were filtered out according to the cutoff set for proteins hits with one or more peptides taller than 10 residues, ion score >25, identity score >−2, and a false positive identification rate less than 1%.

Statistical analyses were performed using Prism GraphPad and StatView 5.0 software (SAS Institute, Cary, NC). For quantitative comparisons between groups, we used either the Mann–Whitney U test (two groups) or Kruskal–Wallis with Dunn post hoc test (more than two groups). For quantitative analysis of patterns of reactivity between groups, principal component analysis (PCA) was used as previously described (37, 38). PCA factor 1 is, by definition, the linear combination of single reactivity measurements that represent the maximum of information about a multivariate dataset in terms of total variance (37, 38). In this study, factor 1 represented 62% of total variance and had exclusively positive factor loads so that it can be seen as a univariate approximation of IgG repertoire reactivity. A p value <0.05 was considered significant.

The STRING database (https://string-db.org/) allowed us to generate a network between candidate proteins. The nodes represent proteins, and the lines represent the level of confidence of interactions among them. The thickness of the line indicates the degree of confidence prediction of the interaction. Unconnected proteins were not shown in the interactome.

According to the STRING Web site, “the minimum required interaction score puts a threshold on the confidence score, such that only interactions above this score are included in the predicted network.” We choose to show high confidence (e.g. a score of 0.7 minimum). The p < 0.05 was considered significant.

To evaluate the impact of TLR expression in P. yoelii 265 BY development within the mammalian host, we followed P. yoelii 265 BY sporozoite infection in TLR2, MyD88, TLR4, TLR9, TLR2–4 KO, and TLR3 KO mice. Surprisingly, only TLR3 KO mice were able to better control parasite burden compared with B6 and the other TLR KO mice (Fig. 1A). In addition, a significant lower parasitemia level was observed from day 15 pi in TLR3 KO mice, rising to nearly 30% in TLR3 KO compared with 60–70% in B6 and the other TLR-deficient mice at day 18 pi (Fig. 1A). We found reduced parasitemia levels in TLR3 KO at day 21 pi, whereas (white) parasitemia levels were still high until day 28 pi in other KO mice groups (Fig. 1A). Additionally, we also monitored the course of infection induced by sporozoites in MyD88-deficient mice compared with B6 mice. As is known, MyD88 is a critical adaptor molecule used by all TLRs except TLR3. We also found a significant decrease of parasitemia levels in MyD88 KO mice starting at day 18 after P. yoelii 265 BY sporozoite infection compared with B6 mice, although parasites still persisted and cleared several days later compared with TLR3 KO mice (Fig. 1A). However, even if the parasitemia tends to be higher initially in TLR3 KO mice with iRBCs instead of sporozoites, a similar trend was observed in TLR3 KO mice, showing a significant decrease in parasite load at 21 d pi compared with B6 mice (Fig. 1B).

To investigate whether the parasite levels are impacted by TLR3 signaling during the hepatic stage, we infected luciferase-expressing P. yoelii sporozoites in B6 and TLR3 KO mice, and our data confirmed lower parasite biomass at 42 h pi in the TLR3 KO mouse as well as in the liver compared with infected B6 mice (Fig. 1C–F). Interestingly, these results were further supported by the reduced bioluminescence day 21 pi TLR3 KO mice compared with B6 mice when infected with blood stage P. yoelii GFP-Luc parasites (Fig. 1G–I). As expected, the parasite bioluminescence at 7 and 14 d pi were found similar to peripheral parasitemia in TLR3 KO mice compared with B6 infected (Supplemental Fig. 1A, 1B). These data indicate that TLR3 may have a beneficial effect on processes contributing to the host–parasite fitness.

It is known that TLR3 directly recruits TRIF and initiates signaling through its TIR that leads to the activation of the serine/threonine kinase TBK1, which in turn phosphorylates IRF3/IRF7 that enables dimerization, nuclear translocation, and transcription of type I IFNs (39, 40). Highly homologous sequences of transcription factors IRF3 and IRF7 are considered important in the induction of antiviral immunity (41), but little has been known about their role during the hepatic stage. To explore the potential function of TLR3 during the liver stage, we first examined the molecular level expression of TLR3 downstream signaling molecules such as Trif, Tbk1, Irf7, Irf3 and NF-kB in the liver at 42 h pi with sporozoites and other proteins involved in innate immunity in infected cells. In addition to the induction of IRFs, proinflammatory cytokines are also produced by TLR3 mediated by NF-kB activation via TRIF (13, 42). Along with a significant increase in parasite load (Fig. 2A) and Tlr3 gene level expression (Fig. 2B) in infected B6 mice, Trif, Tbk1, Irf7, and NF-kB levels were significantly lower in the liver of TLR3 KO at 42 h pi compared with B6 mice (Fig. 2C). In addition, Irf7 levels were nearly 5-fold higher and Trif, Tbk1, and NF-kB levels nearly 1.5-fold higher in B6 mice compared with control, although Irf3 levels were unaltered in B6 and TLR3 KO infected and uninfected liver. Further, except for Irf7, the expressions of all TLR3-mediated signaling molecules were unaltered in P. yoelii 265 BY TLR3 KO infected mice when compared with control (Fig. 2C). To further investigate the protein expression of TLR3, TBK1, TRIF, IRF7, and IRF3 and NF-κB p65 signaling in B6 and TLR3 KO mice in the hepatic stage after P. yoelii 265 BY sporozoite infection, Western blotting was performed. We also noticed a concomitant upregulation of TLR3, TBK1, IRF7, IRF3, and NF-κBp65 protein expression in the liver of B6 mice at 42 h pi compared with control uninfected (Fig. 2D–F). During the hepatic stage, TLR3 cleavage could modify the sensitivity of the receptor and/or modify its specificity for different ligands and somehow increase the affinity for dsRNA (42) and thereby facilitate the recruitment of TRIF (43). No significant changes in Trif, Tbk1, Irf7, and NF-kB protein levels were observed in TLR3 KO, supporting the above observation Supplemental Fig. 1C). To confirm the functional significance, we quantified the expression of inflammatory cytokines induced by the NF-kB signaling pathway in the liver 42 h pi. We found an elevated level of Tnfa and Ifng, which are also controlled by TBK1, (Fig. 2G) in B6 liver 42 h pi compared with TLR3 KO (42). Type I and II IFN expression are known to be regulated by IRF7, which plays an important role in the TLR3-mediated signaling pathway (44–46). Interestingly, significant differences were observed in Ifna genes at 42 h pi in TLR3 KO compared with B6 mice, whereas we did not find any change in the Ifnb level (Fig. 2G). Further, the expression of only Tlr4, Myd88, and Sting (Supplemental Fig. 1D) genes levels in liver 42 h pi of B6 mice were found unaltered compared with TLR3 KO mice. These observations strongly suggest that involvement of the MyD88-independent signaling pathway is shared by TLR3 cascades during the hepatic phase of P. yoelii 265 BY infection.

To get an insight into the functional role of the TLR3-dependent pathway triggered during infection in parasite clearance, we followed the expression of TLR3 downstream inflammatory response genes NF-kB, Tnfa, Ifna, Ifnb, and Ifng at 7, 14, and 21 d pi in the spleen. Interestingly, a significant increase was observed in NF-kB and Tnfa gene levels at 7 d pi compared with B6 infected, whereas no further alteration in NF-kB (Fig. 3A) and Tnfa (Fig. 3B) splenic gene levels were found at day 14 and 21 pi in B6- and TLR3 KO–infected mice. In addition, higher splenic type I Ifna and Ifnb responses observed in B6 mice at day 14 and 21 pi compared with TLR3 KO (Fig. 3C, 3D) were possibly related to delayed parasite clearance. Although at 7 d pi, no modification was observed for type I IFN-β at gene expression (Fig. 3D) and protein levels (Supplemental Fig. 2A), but serum levels were found to be higher on day 14 pi in B6 mice compared with TLR3 KO. We also observed a significant downregulation of Ifng (Fig. 3E) in infected B6 mice compared with TLR3 KO at day 21 pi (Fig. 3E). Interestingly, we also found early serum IFN-γ levels on 5 d pi in B6 mice compared with TLR3 KO (Supplemental Fig. 2B).

These observations were supported by the increase in splenic Tlr3 expression (Supplemental Fig. 1E) along with the decrease in Tlr4 and Sting, whereas Myd88 (Supplemental Fig. 1F) remains unaltered in B6 mice at day 21 pi compared with control. In addition, Tlr4 and Sting were significantly upregulated in TLR3 KO mice compared with B6-infected mice (Supplemental Fig. 1F). Interestingly, no significant changes were observed in the Trif, Tbk1, Irf7, and Irf3 expression in B6 and TLR3 KO compared with their respective controls (Supplemental Fig. 1G). Further, decreased expression of the Irf7 level was noted in infected TLR3 KO mice compared with B6 (Fig. 3B). Interestingly, the increase of parasitemia concomitantly to the decrease of Ifng (Fig. 3E) gene level also supports its role in delayed parasite clearance in B6 compared with TLR3 KO mice at day 21 pi (2).

The modifications of innate and adaptive cell subsets were analyzed in the spleen of wild type compared with TLR3 KO mice after P. yoelii 265 BY infection. During P. yoelii 265 BY infection, we observed higher splenic weight at day 7 pi in TLR3 KO compared with B6 (Supplemental Fig. 2C), but there was no change in terms of the absolute number of nucleated cells between the respective groups (Supplemental Fig. 2D). We also marked that both TLR3 KO and B6 mice developed splenomegaly later during the infection as shown by the rise of the weight and absolute number of cells at day 14 pi (Supplemental Fig. 2C, 2D). Interesting, after 3 wk of infection, splenomegaly in B6 mice was significantly higher compared with TLR3 KO (Supplemental Fig. 2C, 2D), supporting further the fact that those TLR3 KO mice eliminate the parasite more efficiently than B6 mice. We then assessed the cellular modification elicited in the absence or presence of TLR3 signaling during P. yoelii 265 BY infection by following the phenotypic changes of splenic cell subpopulations on day 0, 7, 14, and 21 pi. To investigate the possible association of innate (NK and NKT) and adaptive (B and T) immune cell subpopulations with the TLR3 sensing, the frequency of these cells in the spleen was evaluated (2, 8, 9, 47). We observed a significant change in the total cell numbers of NK, NKT, T, and B cells only in TLR3 KO at day 14 pi, but no bias in the distribution of the different subpopulations was observed (Fig. 4). However, the majority of cell subpopulations showed a tendency to decrease around day 14 pi except NK and NKT cells, which rose during the peak parasitemia. Interestingly, the absolute number of NK cells and NK⁺TLR3⁺ cells were significantly higher in B6 than in TLR3 KO mice at day 21 pi (Fig. 4A, 4B), whereas no change was observed in the percentage of NK1.1⁺TLR3⁺ cells (Fig. 4C). In contrast, we did not observe any significant alterations in the absolute number of CD4⁺, CD8⁺, and NKT cells (Fig. 4D–G). However, the total number of CD4⁺ and CD8⁺ T cells dramatically decreased in both groups at day 14 compared with day 7 and 21 pi (Fig. 4E, 4F). It is noteworthy that the number and the frequency of B cells differentially increased in the spleen at day 21 pi but to a lesser extent in B6 than in TLR3 KO mice (Fig. 4H, 4I). As shown in Supplemental Fig. 2C, the 2-fold decrease in spleen size in TLR3 KO mice compared with B6 mice at day 21 pi correlates with the decrease in the total number of lymphocytes in TLR3 KO mice. In contrast, the absolute number and frequency of CD19⁺TLR3⁺ were significantly increased on day 14 pi in B6 mice (Fig. 4J, 4K). Consequently, this suggests that TLR3 might trigger a NK and B cell response that may possibly help in the control of P. yoelii 265 BY multiplication during the erythrocytic stage.

Abs are known to be major efficient players in the clearance of blood stage parasites (48, 49). We analyzed the parasite-specific Ab response produced in TLR3 KO and B6 mice infected with P. yoelii 265 BY. Levels of total IgG were increased and peaked at 5 mg/ml at day 14 pi in B6 mice, whereas they reached up to 10 mg/ml in TLR3 KO mice at day 21 pi (Fig. 5A). This rise in the level of total IgG in TLR3 KO mice were possibly associated with the decrease in blood parasite levels. In addition, we found higher P. yoelii 265 BY–specific IgG in B6 and TLR3 KO mice on day 14 pi (Fig. 5B). It is worth mentioning that in TLR3 KO mice, the highest production of parasite-specific IgG is associated with the increase of B cells and the clearance of parasitemia. Quantitative immunoblot analysis against P. yoelii 265 BY blood stage extract at day 0, 7, 14, and 21 pi allowed us to gain an insight into the diversity of antigenic specificities produced in the two lines of mice. Typical examples of the immune profiles obtained with the iRBCs extract from B6 and TLR3 KO mice are depicted in Supplemental Fig. 2E. We found an increase in the diversity of IgG reactivity patterns with the progress in the course of infection in both strains B6 and TLR3 KO (Supplemental Fig. 2F). Remarkably, the analysis of patterns of reactivity between groups using PCA distinguished TLR3 KO mice day 21 pi from all the other groups of mice (Fig. 5C). At least five parasite proteins (130, 120, 80, 45, and 25 kDa) corresponding to four sections of reactivity differently recognized by TLR3 KO mice compared with B6 mice were then identified by a Kruskal–Wallis test (Fig. 5D).

Reactivity of interest determined by Panama blot was then analyzed by MS. Four of the sections correspond to section 2 (130 and 120 kDa), 4 (80 kDa), 14 (45 kDa), and 16 (25 kDa), which characterized the response of B6 mice compared with TLR3 KO during days 21 pi. Sections 2, 4, and 16 were more recognized in the sera of TLR3 KO mice, whereas section 14 was highly expressed in the infected B6 group (Fig. 5E–H, Supplemental Table II).

In further analysis of section 2, we obtained six candidates, most represented an aminopeptidase, a papain, a falcilysin-related protein, and a dehydrogenase. Among the 10 candidates identified in the section 4 of interest with molecular mass around 80 kDa, we noticed that heat shock protein (hsp) 70 and hsp70 homolog Pfhsp70-3 were the most represented proteins in the band. Concerning a protein of interest around 45 kDa, better recognized by B6 sera in section 14, we identified four candidates: a dehydrogenase, a deaminase, a proteasome subunit P40.5, and a putative calcium-binding protein. Finally, few proteins were identified in section 16; the most represented were proteasome β-subunit and 40S ribosomal protein (Supplemental Table II). We then used STRING to generate Protein Protein Interaction networks to identify key signatures associated with protective mechanisms triggered during the TLR3 absence. The reactome displayed by STRING shows G3PDH as a key signature and an interaction between G3PDH and seven other proteins including hsp70 and transketolase with a high confidence score along with ketolase (Fig. 5I).

In this study, we described differential roles of TLR3 during P. yoelii infection in B6 mice compared with deficient mice. We showed that the intrahepatic and the ensuing blood stage development of P. yoelii could be affected by the expression of TLR3 molecules through direct or indirect mechanisms. We also showed a reduction in parasite load in the liver and blood pi with P. yoelii sporozoite in TLR3-deficient mice. Our observations were consolidated by the involvement of MyD88, TLR2, and TLR4 in the survival but not in the control of parasite load during blood or liver stages in mice infected with P. berghei sporozoites (50).

However, how P. yoelii manipulate the host to its own advantage through TLR3-dependant molecular pathways remains elusive. Our results obtained in TLR3 KO mice and by analogy with viral infection suggest that the TLR3 sensing in hepatic stage by sporozoite products induces a TRIF-dependent pathway, leading to the activation of different transcription factors such as IRF3, IRF7, NF-kB, and TBK1 (39, 40, 42). In addition, these transcription factors promote the production of proinflammatory cytokines such as TNF-α, IFN-γ, and IFN-β (51–53) that were able to inhibit the development of the liver stage (53) at its very late phase (54, 55). However, the underlying mechanisms and the possible downstream TLR3 pathways involved during malaria infection remains challenging. Altered Ifna gene expression was noted in B6 mice compared with TLR 3 KO; nevertheless, I Ifnb gene expression was found unaltered at 42 h pi of liver stage infection in B6 and TLR3 KO mice. The importance of TLR3 signaling pathway in Plasmodium spp. parasite fitness is further reinforced by the association of type I IFN-β in the extracellular matrix by Capuccini et al. (56). Such disparity between our results and published results could be attributed to either the strain of the parasite or the kinetics of type I IFN production at different time windows during the liver stage. Type I IFN induction via TLR3-induced TRIF-dependent pathway was also reported in Neospora and Toxoplasma infections (57). It is noteworthy that despite the lower level of parasitemia, there were no significant differences in the survival and the onset of experimental cerebral malaria in TLR3 KO mice infected with PbA. In the TLR3 KO group, 20% mice survive until day 20 pi, overcoming the experimental cerebral malaria phase during PbA infection compared with B6 mice (Supplemental Fig. 2G, 2H).

To our knowledge, we showed in this study for the first time that P. yoelii infection possibly triggers a TLR3-dependent signaling pathway involving the TRIF/TBK1/IRF7–IRF3 during the liver stage in B6 mice. In addition, this phenomenon may favor the host–parasite fitness within hepatocytes and RBCs by interfering directly or indirectly with molecular processes implied in the parasite growth. This event can be best explained by the TLR3 expression in the endosomal compartment of cell types within the liver, including Kupffer cells and hepatocytes (58–60); as TLR3 is known to accommodate and cross-link dsRNA endocytosed during viral infection, our findings raise the question of dsRNA existence in P. yoelii parasites (61, 62). In this context, we can speculate that the traversal journey of sporozoites into cells before hepatocyte invasion may lead to the production of parasite dsRNA favoring TLR3 (63) sensing and IFN-γ production during malaria (64). Interestingly, in the case of Leishmania, a dsRNA virus acting as a potent innate parasite stimulator recognizes TLR3, which influences the host immune response by inducing an hyperinflammatory reaction and exacerbated disease phenotype (65–67) Remarkably, all these strategies initiated during infection could be considered as numerous ones developed by protozoan parasites to escape the immunity machinery of the mammalian host and to achieve parasitism.

Revealing an intriguing parallel with the mechanism associated with TLR3 signaling pathway, our notion was further reinforced by the finding that TLR3 KO mice were able to better control parasite growth during P. yoelii sporozoite infection compared with TLR2-, TLR4-, TLR9-, TLR2–4–, and MyD88-deficient mice. Furthermore, another endosomal nucleic acid sensor TLR7 studied in P. chabaudi AS infection showed no effect in parasitemia, although an early burst of type 1 IFN was observed (68). Our study highlights the role of TLR3 in parasite fitness. Although we observed an early clearance of the parasite in TLR3 KO mice during P. yoelii infection, a recent report suggests that the TLR2-deficient mice result in increased parasitemia during P. yoelii infection compared with wild type (69). By contrast, disrupting type I IFN signaling in the spleen also affects serum IFN-γ production (70) (Supplemental Fig. 2B), which may account for differential TLR responses early during the blood stage infection.

Although TLRs are expressed by a wide range of immune cells, they are predominantly present on innate cells. We analyzed the impact of TLR3 signaling on the fate of innate and adaptive immune cell responses during infection; interestingly, we found a significant increase of NK1.1, NKT, and NK⁺TLR3⁺ cells during a late stage of infection but not during the early phase. In addition, the differential response of type I and II IFN responses may be correlated with the subpopulation of NK cells expressing TLR3 in B6 mice during a later stage of infection (71, 72). During the blood stage, T cell responses were similar in the presence or absence of TLR3, but B cell responses were found altered, which may be an essential factor in controlling the erythrocytic stage (48). Surprisingly, we observed a lower production of total IgG in the presence of CD19⁺TLR3⁺ cells, which possibly leads to a delay in the clearance of blood stage parasite in B6 mice. Moreover, in TLR3 KO mice, the rise in the amount of total IgG was concomitant with higher total CD19⁺ B cells frequency and the production of parasite-specific Abs during P. yoelii 265 BY.

PCA and the interactome analyses allow us to identify various key Ags possibly allied with the clearance of parasitemia during P. yoelii infection. G3PDH was found to be more recognized in the sera of B6 mice, whereas transketolase, hsp70, 40S ribosomal protein S5, and eIF3 as in sera of TLR3 KO mice. Different hypotheses have been proposed in malaria during G3PDH deficiency, such as host immune system–mediated recognition/destruction of iRBCs and impaired infection due to oxidative stress in erythrocytes (73–75). P. falciparum development and its association with the pathology leads to structural and functional remodeling of the host cell due to exports of parasite-encoded proteins (76, 77). Furthermore, there is a least homology of P. falciparum transketolase (PfTk) with its human host, but they are known to play different roles in the malaria parasite, including pentose sugar supply for nucleotide synthesis (78). In contrast, hsp are reported to be highly adapted to the life cycle of the malaria parasite, playing an important role as molecular chaperones (75, 79). Furthermore, eIF3 individually interacts with viral mRNAs, 40S ribosomal subunit, and multiple initiation factors (80). Therefore, we can hypothesize that IgG anti-Ab production in B6 and TLR3 KO may lead to a differential recognition of parasite-specific Abs affecting parasite clearance (81, 82).

Upon infection with P. yoelii in wild type mice, the induction of a virus-like mechanism through the activation of TLR3 and its downstream signaling pathways promotes an early innate inflammatory immune response that favors an efficient host–parasite fitness, possibly resulting in critical interplay between type I/II IFN response, which may warrant further investigations. Further, we hypothesized that, later during the infection, possibly high T-independent B cells respond, avoiding production of specific IgG in B6 mice. Altogether, these findings argue in favor of our hypothesis that by activating the TLR3 pathway, P. yoelii is able to induce a proinflammatory state beneficial for the parasite and linked to the higher parasitemia in B6 mice. Exploration of this pathway will be the opening of a new horizon for the therapeutics of malaria.

We acknowledge the support of Dr. Michel Chignard (Institut Pasteur Paris, France). We are grateful to Catherine Ronet for critical analysis of the manuscript and Jean-François Franetich and Maurel Tefit for the production of GFP-Luc sporozoites.

The authors have no financial conflicts of interest.