CSP—A Model for In Vivo Presentation of Plasmodium berghei Sporozoite Antigens by Hepatocytes

One target of protective immunity against the Plasmodium liver stage in BALB/c mice is represented by the circumsporozoite protein (CSP), and mainly involves its recognition by IFN-γ producing specific CD8+T-cells. In a previous in vitro study we showed that primary hepatocytes from BALB/c mice process Plasmodium berghei (Pb) CSP (PbCSP) and present CSP-derived peptides to specific H-2kd restricted CD8+T-cells with subsequent killing of the presenting cells. We now extend these observations to an in vivo infection model in which infected hepatocytes and antigen specific T-cell clones are transferred into recipient mice inducing protection from sporozoite (SPZ) challenge. In addition, using a similar protocol, we suggest the capacity of hepatocytes in priming of naïve T-cells to provide protection, as further confirmed by induction of protection after depletion of cross-presenting dendritic cells (DCs) by cytochrome c (cyt c) treatment or using traversal deficient parasites. Our results clearly show that hepatocytes present Plasmodium CSP to specific-primed CD8+T-cells, and could also prime naïve T-cells, leading to protection from infection. These results could contribute to a better understanding of liver stage immune response and design of malaria vaccines.


Introduction
Immunization of rodents and humans with radiation-or genetically-attenuated sporozoites (SPZ) (RAS, GAS) confers preerythrocytic stage specific protective immunity to an infectious challenge [1,2]. This protective immunity is mediated in part by CD8+ T-cells specific for the CSP and other not yet identified proteins [2][3][4][5]. Recently, it was demonstrated that both infected and sporozoite-traversed mouse primary hepatocytes can process the PbCSP and present CSP-derived peptides to a specific H-2K drestricted CD8+ T-cell clone in vitro but recognition of infected hepatocytes was the only relevant step in the elimination of infection [6]. Using bone marrow cell transfer into totally irradiated mice, it was also concluded that activation of protective CD8+ T-cell clones was due to antigen presentation by nonhematopoietic parenchymal cells [7]. Thus, the role of hepatocytes as antigen presenting cells (APCs) in the activation of primed T-cells to provide sterile protection seems to be accepted. Nevertheless, the role of liver cells and, in particular, hepatocytes in the activation of Plasmodium-specific CD8+ T-cells is not clearly elucidated. Chakravarty et al. [7] presented data supporting the role of lymph node dentritic cells (DCs) in presentation to parasite-specific T-cells while Leiriao et al. [8] supported the notion that apoptotic Plasmodium-infected hepatocytes provide antigen to liver DCs. While both pathways of antigen presentation co-exist, their role in providing a protective CD8+ T-cell response has not been established. In both studies, critical experiments such as the elimination of DCs and its consequence on sterile protection are missing. On the other hand, several publications point to the role of liver cells and hepatocytes in the mechanism of protection [2,5,[9][10][11]. In addition, intravenous (iv) injection in mice of RAS gives rise to a more robust immune response and to sterile immunity compared to intradermal (id) immunization [12]. On the other hand, this difference in protection between iv and id injection is overcome by a higher dose of sporozoites [13]. Furthermore, immunization of mice with GAS is associated with a better immune response, probably due to the development of GAS liver stages and presentation of antigens other than CSP to specific T-cells [14].
It has been shown that CD8+ T-cells are primed by the DCs in the skin, and are then thought to exit the priming site and migrate to the liver where they can eliminate infection after recognizing antigens presented by hepatocytes [2,7,9,15]. Thus, it was concluded that initiation of the CD8+ T-cell mediated immunity requires antigen presentation by only DCs [16]. On the other hand, the infected hepatocyte is the cell where SPZs develop and proliferate into the next liver forms. Furthermore, infection of hepatocytes is crucial for maintenance of anti-SPZ protective CD8+ T-cell response since protection was abrogated if hepatic stages were eliminated [2,9,15]. However, the ability of hepatocytes to prime naïve CD8+ T-cells and induce protective immunity remains unclear. It has also been illustrated that hepatocytes present CSP that is secreted directly into the cytosol, unlike DCs that cross-present Plasmodium antigens via endosomes [16]. These findings suggested that both in DCs and hepatocytes, the presentation of antigens requires the transporter associated with antigen processing (TAP) proteins [16][17][18] and that the crosspresenting DCs could be abrogated in mice after treatment with cytochrome c (cyt c) [16]. Determining if hepatocytes in vivo process and present Plasmodium antigens to naïve and primed Tcells may help in the rational identification of pre-erythrocytic vaccine candidates. To this purpose, we have established an in vivo protocol to address two questions: are hepatocytes capable of 1) in vivo stimulation of primed CD8+ T-cells and 2) priming naïve Tcells to protect mice against parasite challenge? We used different methods to address these two questions. First, we used intrasplenic (IS) transfer of PbSPZ-infected hepatocytes from BALB/c mice into naïve, TAP2/2 deficient mice (H-2K b ) and BALB/c mice (H-2K d ) in the presence or absence of H-2K d -specific CD8+ Tcells (C7 clone) in order to determine the ability of primary hepatocytes to present antigen to primed CD8+ T-cells. Secondly, to show that hepatocytes could present antigen in the absence of dendritic cells, mice treated with cyt c were injected with irradiated SPZ (iSPZ) and challenged with live SPZ. Thirdly, mice were immunized with iSPZs deficient for the sporozoite microneme protein essential for cell traversal (spect (2) iSPZs), known for being incapable of cell traversal, but capable of infection and normal development [6,19,20] in order to show the role of infected hepatocytes in inducing a protective immune response. In all cases, mice were partially or totally protected from an infectious SPZ challenge as assessed by their level of blood parasitemia up to 14 days post infection.

Parasites
Plasmodium berghei ANKA wild-type (wt) and spect (2) SPZ were obtained after salivary gland dissection of infected female Anopheles stephensi mosquitos raised in the mosquito facility at the Department of Biochemistry, University of Lausanne, Switzerland as described previously [6,20]. After dissection, salivary glands were homogenized in a glass grinder and released SPZ were counted and then diluted in sterile Dulbecco's Modified Eagle Medium, DMEM (GibcoH, Life Technologies TM , New York, NY).

Animals
Six-to-12-week-old BALB/c (H-2K d ) or TAP2/2 (H-2K b ) mice were obtained from Harlan Laboratories B.V. (Venray, Netherlands) or bred at the animal facility at the Department of Biochemistry (University of Lausanne, Switzerland). All mice were housed under pathogen-free conditions and handled according to the guidelines of the authorization Nu 805.7 of the Service de la consommation et des affaires vétérinaires (Lausanne, Switzerland).

Hepatocyte Isolation
SPZ-infected and naïve hepatocytes from BALB/c mice were obtained after collagenase perfusion of the liver as previously described [6,20,22]. Briefly, mice were sacrificed by CO 2 inhalation, dissected, and the biggest lobule of liver was cut out. The lobule was perfused for 10 min with Ca 2+ -free HEPES (4-(2hydroxyethyl)-1-piperazineethanesulfonic acid) buffer, pH 7.6 (GibcoH Invitrogen TM , New York, NY) at 37uC at a rate of 5 ml/min. The lobule was then perfused with type IV collagenase (Sigma-AldrichH, Steinheim, Germany) (HEPES buffer containing 0.04% type IV collagenase and 0.075% CaCl 2 -2H 2 O) for 5 min at 37uC. The perfused lobule was incubated for 10 min at 37uC in the collagenase solution. Using sterile pipettes, the tissue was gently teased apart to release cells and washed once with Ca 2+ -free HEPES buffer at 800 rpm for 30 s at 4uC. The pellet was gently re-suspended in DMEM, layered on 60% Percoll (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) and centrifuged at 2000 rpm for 2 min at 4uC. The resulting pellet was re-suspended in complete culture medium (DMEM supplemented with 10% FCS, 1% penicillin streptomycin, 1% HEPES, and 0.05 mM of bmecarptoethanol (b-ME, Sigma-AldrichH), and centrifuged again at 800 rpm for 30 s at 4uC. The pellet was finally re-suspended again in 5 ml of DMEM for counting. Viability of isolated hepatocytes was assessed by light microscopy and trypan blue dye exclusion. As reported previously [6] hepatocyte contamination with Kupffer/dendritic cell markers is less than 1% as determined by FACS analysis [6,10,23,24]. Then, mice were injected iv with 7610 5 live hepatocytes re-suspended in 200 ml of sterile DMEM. To obtain infected hepatocytes, 1610 6 of live or irradiated wt PbSPZ were injected iv through the tail vein of BALB/c mice. Two hours later, the liver was removed from the infected mouse and perfused as described above.

Intrasplenic (IS) Transfer
Hepatocellular transplantation was carried out by direct injection of 7610 5 hepatocytes in 200 ml of sterile DMEM with a syringe into the splenic parenchyma of recipient mice that were anaesthetized with Isoflurane (Provet AG, Berne, Switzerland). Briefly, under aseptic and anaesthesia conditions, cautiously, with a chisel and a clip, the abdomen of the mouse was opened on the left flank. Using a clamp fitted, the spleen was gently pulled out and placed on a sterile piece of paper. The injection of cell suspension in the spleen parenchyma was carefully carried out at a rate of 10 ml per 10 respiratory cycles of the mouse. Suturing and clamping of the skin minimized leakage from the site of operation. Four hours later, 20 million PbCSP-restricted CD8+ T-cells (C7clone) or irrelevant CD8+ T-cells (clone S14) were injected iv through the tail vein in a volume of 500 ml of DMEM. Control mice received only hepatocytes.

T-cell Clone Re-stimulation
The C7 and irrelevant S14 clones [3,21] were re-stimulated weekly, maintained at 37uC and used as effector cells. Briefly, P815 cells (mastocytoma cells as antigen presenting cells, APC) were re-suspended (1610 6 cells/ml) in complete culture medium (DMEM supplemented with 10% FCS +1% of pyrimethaminestreptomycin +0.1% of b-ME and 1% of HEPES) and pulsed for 1 hour with PbCSP-epitope peptides (1 mg/ml) specific for C7-and S14 clones. C7 and S14 were washed and re-suspended (2610 6 / ml/well) in CTL culture medium (complete culture medium supplemented by 30 U/ml of mouse IL-2) in a 6-well plate. For restimulation, the respective clones were added to the 1610 6 pulsed and irradiated P815 cells (10000 rads/20 minutes) in the presence of 15610 6 /well of irradiated-BALB/c spleen cells (5000 rads/10 minutes) in 6-well flat bottom plates. In the intrasplenic experiment, clones were re-stimulated and kept for two weeks in culture to make sure they were resting at the time of injection.

Cytochrome c Treatment and Induction of CD8+ T-cells
BALB/c mice were depleted of cross-presenting DCs by iv treatment for 3 days with 15 mg/ml (5 mg/ml per day) of horse cyt c (Sigma-AldrichH, St Louis, LA) in 100 ml of PBS (GibcoH Invitrogen TM ). Control mice received 100 ml of PBS alone. On the last day of treatment, mice were immunized iv with 1610 5 wt PbiSPZ). Seven-to-ten days later, the frequency of both PEconjugated SYIPSAEKI-tetramer and FITC-conjugated antimouse CD8 antibody (BD Biosciences, Allschwil, Switzerland) specific CD8+ T-cells was measured in different mouse tissues (blood = PBL, spleen, liver and lymph nodes (LN)) by flow cytometry.

Real Time PCR (RT-PCR)
In vivo assessment of parasite loads was performed by RT-PCR as described previously [25]. Briefly, BALB/c mice were injected iv with 1610 5 PbiSPZ (wt or spect (2)). Two hours later, total livers and spleens were isolated, perfused with PBS alone and total RNA was extracted. Then, cDNA was synthesized using specific primers

Parasitemia Assessment
At different time points after the IS transfer or challenge with live PbSPZs (iv), parasitemia was assessed by 10% Giemsa (FlukaH, Sigma-AldrichH)-stained blood smears. Blood smear slides were air-dried and read by light microscopy (Olympus CH-2, Microscope Company, Hicksville, NY) to determine infected red blood cells (iRBC). Animals were protected against malaria if they remained negative in Giemsa-stained blood smears 2 weeks after receiving infected hepatocytes or live SPZ. Control animals were included to verify infectivity of SPZ or infected hepatocytes. In each control animal, parasitemia was detectable 7 days after IS transfer or challenge.

Statistical Analysis
Different statistic tests were performed using GraphPad Prism software (version 6). Fisher's exact test compares the proportion of mouse protection among various groups. The Mann-Whitney test was performed to compare parasite DNA load and frequency of PbCSP-specific CD8+ T-cells in PBL in two independent experiments or in different organs in the same experiment in which mice underwent different treatments (wild-type and Spect (2) iSPZ or PBS and cyt c groups). All p-values equal to or lower than 0.05 were considered significant.

Infected Hepatocytes Present the PbCSP-specific Epitope to Cloned CD8+ T-cells with Subsequent Protection against Malaria
To show that infected hepatocytes process and present PbCSP, TAP2/2 mice received BALB/c PbSPZ-infected hepatocytes by IS transfer followed by iv injection of PbCSP specific (clone C7) or irrelevant (S14 clone) CD8+ T-cells 4 h later. TAP 2/2 mice were selected to bypass the possibility that presentation could be performed by professional APCs through processing of apoptotic, infected hepatocytes or live SPZs possibly externally associated with BALB/c hepatocytes. In addition, the host vs graft immune response is minimized. All mice (8/8; 100%) that received infected hepatocytes and the specific C7 clone were protected from an infective sporozoite challenge (Table 1, group A). In contrast, all mice receiving the irrelevant S14 clone were infected (Table 1, group B). In addition, 75% of naive TAP2/2 mice treated with infected BALB/c hepatocytes were infected ( Table 1, group C).
In order to show that the protective immune response is specific and not due to a bystander effect or a continuous secretion of cytokine by the C7 clone, TAP2/2 mice were first infected with live PbSPZ (1000/mouse in iv) or not. Ten hours later, mice received infected or naïve BALB/c hepatocytes together or without C7 clone as indicated in Table 2. Parasitemia determination of group C showed that only 1/7 (14%) of mice was protected ( Table 2, group C). Other controls showed that protection occurred only in the group that received infected BALB/c hepatocyte and C7 clone ( Table 2, group A).

Infected Hepatocytes Prime PbCSP-specific CD8+ T-cells and Protect Mice against SPZ Challenge
Since the infected hepatocytes can reactivate resting CSPspecific CD8+ T-cells and induce protective immunity (Table 1 and 2), the next step was to determine if they could also prime naïve T-cells to protect mice against live SPZ challenge. To this effect, we injected iSPZ-infected BALB/c hepatocytes into naive BALB/c mice before challenging with live SPZ. Considering that the immunization procedure may give rise to a sub-optimal immunity (about 700 infected hepatocytes injected IS if the overall infection efficacy is estimated to be 10%), mice were challenged with a sub-optimal or optimal dose of live SPZ (2610 3 and 5610 3 ) that led to 60% and 100% infection in naïve mice, respectively, with an overall protection of 13% (Table 3). In contrast, mice receiving iSPZ-infected hepatocytes were protected at 64% (7/11; p = 0.014) ( Table 3), suggesting that the infected hepatocytes could contribute to the priming of naïve T-cells and protection of mice against infection.
To further corroborate that hepatocytes could prime naïve Tcells and induce protection, BALB/c mice were treated with cyt c to delete cross-presenting DCs before and during iSPZ immunization. This protocol was established according to previous studies which showed that cross-presenting DCs can be largely depleted after in vivo cyt c treatment [16,26]. The first experiment (as pilot) showed a significant reduction of 60% of PbCSP 245-253 specific CD8+ T-cells in PBL after cyt c treatment (data not shown). In the second experiment, the analysis was extended to other organs (LN, spleen and liver) (Fig. 1). Thus, FACS analysis showed that induction of PbCSP-specific CD8+ T-cells was highly reduced in the cyt c-compared to the PBS -treated mice in PBL and liver (86% and 67%, respectively), while this reduction was about 50% Figure 2. Relative comparative wt and spect (2) parasite DNA load in spleen and liver of immunized mice. To compare relative load of wt and spect (2) parasite DNA in liver and spleen, BALB/c (3 mice per group) were immunized iv on tail with 1610 5 iSPZ (wt or spect (2)) in 500 ml of RPMI. Two hours later, a real-time PCR was performed following extraction of respective parasite RNA. Every sample was done in duplicate. To avoid any contamination by eventual parasite from blood, the liver and spleen were perfused with PBS before performing the RT-PCR. Figures a and b represent wt and spect (2) parasite DNA load, respectively, in the liver and spleen 2 h after iSPZ injection (iv). Naïve mice (receiving only 500 ml iv of DMEM) were used as a control. doi:10.1371/journal.pone.0051875.g002 In Vivo Presentation of Antigens by Hepatocytes PLOS ONE | www.plosone.org in LN and spleen (Fig. 1, inserted panel). Taking the PBL data from the two experiments or from all organs in the second experiment, normalizing and combining them, we obtained p values of 0.003 (PBL) and 0.001 (organs) for the cyt c-treated compared to PBS groups. In spite of these differences, both groups were protected after SPZ challenge (Table 4).
Protective Immune Response Induced in BALB/c Mice Immunized with spect (2) iSPZ Wild-type SPZs are known to be able to cross several cells (leaving a trail of the CSP behind) before infecting a single hepatocyte, unlike spect (2) SPZ that infect hepatocytes without cell traversal. It has been shown that both infected and traversed hepatocytes are able to process and present PbSPZ CSP to primed CD8+ T-cells to induce IFN-c secretion in vitro but only the infected hepatocytes were responsible for their own elimination [6,20]. Similar experiments were then performed in vivo. Thus, it was expected that spect (2) SPZ would activate a lower number of the PbCSP-specific CD8+ T-cells in the periphery and present a lower parasite load in the liver. RT-PCR clearly showed that the relative parasite DNA load in liver and spleen was significantly higher in mice receiving wt iSPZ compared to the spect (2) iSPZ parasite group (p = 0.026 and 0.008, respectively) ( Fig. 2a and b). In addition, FACS analysis showed that the level of the PbCSPspecific CD8+ T-cell response was significantly higher in the wt compared to the spect (2) group in PBL and spleen (p = 0.029) suggesting a role of cross-presenting DCs in these compartments (Fig. 3). In contrast, in the liver, the frequency of CD8+ T-cells was similar (p = 0.886) for the two kinds of iSPZ (Fig. 3) in spite of a lower parasite load for spect (2). Both groups of mice were protected against live SPZ challenge (Table 5). Together, these data indicate that T-cell traversal by SPZ induces a higher level of immune response in PBL and spleen, but not in the liver, further  (2) iSPZ. BALB/c mice (4 mice per group) were immunized (iv) with wt or spect (2) ANKA PbiSPZ (one dose of 1610 5 iSPZ). Seven (7) days later, PBL, liver and spleen cells of mice were isolated to determine frequency of PbCSP-specific CD8+ T-cells by flow cytometry (FACScan). The CD8+ T-cells were double stained with PE-conjugated PbCSP epitope tetramer and FITC-conjugated anti-mouse CD8b antibody. The naive group received neither wt nor spect (2) iSPZ. p-value compares statistically significant mean of CD8+ T-cell frequency between wt and spect (2) iSPZ-immunized groups. doi:10.1371/journal.pone.0051875.g003 Table 5. Both wt and spect (2) PbiSPZ immunization protect mice against SPZ challenge. supporting the key role of infected hepatocytes in antigen presentation and induction of protective immunity.

Discussion
Data presented in this manuscript provide further evidence for the role of Plasmodium infected hepatocytes in the stimulation of protective secondary CD8+ T-cells leading to the elimination of P. berghei pre-erythrocytic stages. In addition, they indicate that hepatocytes can indeed prime sporozoite-specific, protective naïve T-cells. While the antigen-presenting role of infected hepatocytes in the secondary CD8+ T-cell response seems to be accepted, the mode of priming naïve CD8+ T-cells is still, in our opinion, not yet established. With regard to the first point, the in vivo data presented here are fully consistent with our previous in vitro results [6,20] and with in vivo data published by Zavala and collaborators [7] via bone marrow cell transfer experiments. In our case, we transferred either BALB/c infected hepatocytes or a CS specific Tcell clone or both to TAP-deficient H-2k b mice. Tap2/2mice were chosen to bypass the possibility that antigen presentation is mediated by professional antigen presenting cells which might have ingested apoptotic infected hepatocytes [8] or live sporozoites possibly externally associated with BALB/c hepatocytes, and minimize host vs graft immune responses. In addition, we have determined protection as lack of infection in mice 14 days postchallenge, which represents a stringent, but the only significant standard for protection for pre-erythrocytic vaccines. Our results show that protection from infection is antigen-specific since mice are not protected if an irrelevant CD8+ T-cell clone is used. In addition, as observed in vitro [6], our results indicate that infected hepatocytes are directly killed by the antigen-specific T cells and not by a bystander effect through continuous secretion of IFN-c or other lymphokines by the injected T-cell clones or host vs graft immune response since concomitant infection of TAP-deficient mice does not lead to protection after treatment with the CS specific T-cell clone and/or naïve BALB/c hepatocytes. These and the previous data [6,7] clearly establish the central role of infected hepatocytes in the total clearance of Plasmodium infection in vivo once an immune response has been induced (secondary response). However, in our opinion, the role of hepatocytes in priming naïve, T-cells is not yet elucidated. Zavala and collaborators [7,16] claim that peripheral dendritic cells prime CD8+ T-cells, while Leiriao et al. [8] suggest that apoptotic infected hepatocytes provide antigens to liver dendritic cells. In addition, these claims seem to be supported by the notion that hepatocytes act as tolerizing cells [27,28]. On the other hand, given the large number of hepatocyte genes affected by sporozoites and salivary gland components, including some related to antigen processing and presentation and chemokine production it is not far fetched to hypothesize that infected hepatocytes become fullfledged antigen presenting cells upon infection [29,30]. This would allow the activation of T-cells specific to sporozoite and late liver stage antigens (14). This would optimize the balance between infection and immune responses that parasites and hosts have developed through co-evolution. Evidence that the liver is central to obtaining an optimal immune response was provided early on by Renia et al [10] in which similar results as obtained here were presented, where immunization with non-parenchymal cells did not result in protection. Recent data by Epstein et al. [12] in which a better immune response was obtained by immunization of mice with irradiated sporozoites via iv than id or sc injection also point to the antigen-presenting role of the liver. In this study, we specifically target the role of infected hepatocytes in the immune response by immunizing (iv) mice with spect (2) iSPZ that infect hepatocytes without prior traversal. We show that, in spite of a significant decrease in parasite DNA load, the liver CSP-specific CD8+ T-cell response was similar to that in wt iSPZ-immunized groups. This and cyt c treatment of mice prior to immunization with iSPZs further support the key role of infected hepatocytes in T-cell priming to provide protective immune responses.
In conclusion, the data provided show that hepatocytes can indeed present Plasmodium berghei CSP epitopes to primed CD8+ Tcells and strongly suggest that they could also prime parasitespecific naïve T-cells to fully protect mice against a live parasite challenge. But in our opinion, formal proof of the role of hepatocytes in antigen presentation can only be obtained by isolating malaria parasite infected hepatocytes for in vivo and in vitro experiments.