IL-22 dampens the T cell response in experimental malaria

A tight regulation between the pro– and anti–inflammatory immune responses during plasmodial infection is of crucial importance, since a disruption leads to severe malaria pathology. IL-22 is a member of the IL-10 cytokine family, which is known to be highly important in immune regulation. We could detect high plasma levels of IL-22 in Plasmodium falciparum malaria as well as in Plasmodium berghei ANKA (PbA)-infected C57BL/6J mice. The deficiency of IL-22 in mice during PbA infection led to an earlier occurrence of cerebral malaria but is associated with a lower parasitemia compared to wt mice. Furthermore, at an early time point of infection T cells from PbA-infected Il22−/− mice showed an enhanced IFNγ but a diminished IL-17 production. Moreover, dendritic cells from Il22−/− mice expressed a higher amount of the costimulatory ligand CD86 upon infection. This finding can be corroborated in vitro since bone marrow-derived dendritic cells from Il22−/− mice are better inducers of an antigen-specific IFNγ response by CD8+ T cells. Even though there is no IL-22 receptor complex known on hematopoietic cells, our data suggest a link between IL-22 and the adaptive immune system which is currently not identified.

(P. falciparum) infection. This study brought attention to IL-22 for further investigations concerning its function in plasmodial infection. Further observations described that IL-22 plays a protective role during Plasmodium chabaudi (P. chabaudi) infection in mice 31 .
Besides P. chabaudi infection of mice there are three additional murine plasmodial species known to infect rodents: Plasmodium berghei, Plasmodium yoelii and Plasmodium vinckei. A widely established plasmodial model to investigate cerebral malaria symptoms is the infection of C57BL/6 mice with Plasmodium berghei ANKA (PbA) 32 . The plasmodial life cycle is divided into two stages: the asymptomatic liver stage and the blood-stage during which severe complications such as cerebral malaria can occur. Compared to the other strains, PbA is the most similar to P. falciparum concerning nucleotide identity (70.3%) and protein identity (62.9%) 33 and causes similar symptoms in mice compared to the P. falciparum infection in human during the blood-stage 32 . P. falciparum infections still cause the majority of complicated malaria cases 34 . Cerebral malaria is the most severe complication of a plasmodial infection and can lead to death 35 . Especially in sub-Saharan Africa P. falciparum infections are one of the leading causes of morbidity and mortality of the disease. With the murine model of PbA infection in C57BL/6 mice the importance of the immune system for disease progression was highlighted. Cerebral malaria in this model was shown to be highly promoted by CD8 + T cells that adhere to activated brain endothelium as a consequence of T H 1-dependent IFNγ production 32 . Especially IFNγ was shown to play an important role concerning the development of cerebral malaria during PbA infection 36,37 . To keep a well-balanced immune response to plasmodial infection the induction of regulatory mechanisms during malaria is indispensable, since the disruption of this balance leads to pathology. Important for these regulatory mechanisms are the cytokines TGFβ and IL-10 [38][39][40][41] .
In this study, blood-stage infection of C57BL/6 mice with PbA was used to analyse the impact of IL-22 during malaria. During the blood-stage of PbA infection mice start to develop cerebral malaria symptoms at d6-8 post infection (p.i.). Here we could observe that Il22 −/− mice showed a significantly earlier occurrence of cerebral malaria symptoms accompanied by a lower parasitemia compared to wt mice during PbA infection. In addition to this phenotype an early increase of IFNγ and an overall lower IL-17 production was observed in the absence of IL-22 during PbA infection. Taken together IL-22-deficient mice showed a stronger pro-inflammatory immune response during PbA infection resulting in an earlier occurrence of cerebral malaria symptoms but also resulting in lower parasitemia.

Results
Elevated levels of IL-22 in malaria. IL-22 is significantly increased in plasma samples of human individuals suffering from acute P. falciparum malaria compared to healthy controls (Fig. 1A). Also in the plasma of PbAinfected C57BL/6 mice, elevated levels of IL-22 at d6 p.i. were observed (Fig. 1B). Since the IL-22 level in plasma is increased in malaria, it was of interest to specify the cell type which is responsible for IL-22 production. To this end, C57BL/6 mice were infected with PbA iRBC and sacrificed at d6 p.i. Spleen cells were stimulated ex vivo, stained intracellularly for IL-22 and were subsequently analysed by flow cytometry. Even though the secretion of IL-22 was significantly increased in CD4 + T cells as well as in NK cells, the highest proportion of IL-22 secreting cells was observed in the γδ + T cell population (Fig. 1C).
Lack of IL-22 results in a more severe malaria outcome. The infection with PbA sporozoites in the absence of IL-22 resulted in a lower parasitemia ( Fig. 2A). To elucidate if this decreased parasitemia was due to a reduced parasite burden in the liver in the absence of IL-22, wt and Il22 −/− mice were infected with 1000 sporozoites i.v. and the liver load of PbA parasites was determined by qPCR 24 h p.i. (Fig. 2B). However, no differences were observed regarding the parasite burden in the liver. Hence, the effect of IL-22 on the parasitemia seems to be restricted to the blood-stage. To further analyse how the absence of IL-22 is influencing the outcome of disease, wt and Il22 −/− mice were infected with PbA iRBC and the course of the infection was monitored (Fig. 2C). Interestingly, the lack of IL-22 led to an earlier occurrence of cerebral malaria (CM). A similar result was observed by transiently neutralizing IL-22 with a specific antibody in wt mice compared to IgG treated control mice (Fig. 2D). In order to control if the lack of IL-22 during PbA infection is associated with increased

Lack of IL-22 induces a decreased parasitemia during PyNL infection in mice.
In order to follow the outcome of IL-22-deficiency on the parasitemia for a longer time period, we used Plasmodium yoelii non-lethal (PyNL) infection of C57BL/6 mice. This Plasmodium species led to a transient parasitemia, which is cleared within three weeks without inducing cerebral pathology. As in the PbA model, PyNL-infected Il22 −/− mice suffer from a decreased parasitemia at d7 p.i. compared to PyNL-infected wt mice (Fig. 3A). Furthermore, the administration of a neutralizing IL-22 antibody at the time point of infection revealed a significantly decreased parasitemia compared to the control group over the course of infection (Fig. 3B).
Increased T H 1 response in Il22 −/− mice during PbA infection. Due to the increased incidence of CM in IL-22-deficient mice and since it is known that the disruption of the tight regulation between pro-inflammatory and anti-inflammatory response during malaria leads to a severe pathology, the secretion of the pro-inflammatory cytokines IFNγ and IL-17 by T cells in Il22 −/− mice was analysed. Therefore, mice were infected with PbA iRBC and splenic cells were stimulated ex vivo with PMA/Ionomycin. An increased level of IFNγ producing cells was observed in Il22 −/− mice compared to wt mice at d3 p.i. This effect was visible in all analysed cell types including CD3 + (Fig. 4A,H), CD8 + (Fig. 4B,I), CD3 + CD8 − (Fig. 4C,J) and γδ + (Fig. 4D,K) T cells. In contrast, the IL-17 secretion was decreased in CD3 + CD8 − (Fig. 4F,J) and γδ + (Fig. 4G,K) T cells at d3 p.i., which might be due to the antagonistic effect of IFNγ on IL-17 prodcution 42 . The increased IFNγ secretion at d3 post PbA infection by CD3 + T cells was confirmed in mice in which IL-22 was transiently neutralized compared to the control group ( Fig. 4E). Even though the secretion of IFNγ was not different on d6 p.i., the IL-17 secretion of Il22 −/− mice was still significantly decreased in CD3 + CD8 − (Fig. 4F) and γδ + (Fig. 4F) T cells. Nevertheless, the IFNγ concentration in the supernatant of antigen specific stimulated splenocytes from PbA-infected Il22 −/− mice at d6 p.i. was still significantly increased compared to splenocytes from PbA-infected wt mice (Fig. 5A). To gain a closer look at the mechanism behind this, bone marrow derived dendritic cells (BMDCs) from wt and Il22 −/− mice were generated and used as antigen presenting cells (APCs). CD8 + T cells of PbA iRBC-infected wt and Il22 −/− mice were incubated with PbA antigen-pulsed BMDCs for 24 h and subsequently the IFNγ secretion in the supernatant was analysed using ELISA. This approach showed that CD8 + T cells from Il22 −/− mice secrete more IFNγ after ex vivo antigen specific stimulation and furthermore also BMDCs from Il22 −/− mice showed an improved stimulatory effect on wt and Il22 −/− CD8 + T cells compared to wt BMDCs regarding the ability to induce IFNγ secretion by CD8 + T cells (Fig. 5B).

Modified antigen presentation in the absence of IL-22.
To elucidate why BMDCs from Il22 −/− mice have an improved ability to stimulate IFNγ secretion by CD8 + T cells, the expression of costimulatory ligands on antigen presenting cells from the spleen were analysed during PbA infection. For this purpose wt and Il22 −/− mice were infected with PbA iRBC and CD11c + CD11b − dendritic cells were analysed for their CD86 expression at d3 p.i. (Fig. 6A). Dendritic cells from IL-22-deficient mice express significantly more CD86 compared to dendritic cells from wt mice at this early time point of infection (Fig. 6B). This result was confirmed by the restimulation of BMDCs with LPS over 24 h. In this case the expression of CD86 and CD80 was significantly higher after stimulation on BMDCs from Il22 −/− mice compared to BMDCs from wt mice (Fig. 6F). However, recombinant IL-22 was unable to inhibit LPS-induced expression of CD86 and CD80 on BMDCs (Fig. 6D). Since T cells are stronger activated in Il22 −/− mice during PbA infection it was of interest to analyse if the proliferation of T cells in Il22 −/− mice is altered in vivo. Therefore, splenic OT1 T cells were labelled with CFSE and transferred into Il22 −/− and wt recipient mice. Subsequently the recipients were infected with a transgenic OVA expressing PbA strain (PbA-OVAtg) or with a transgenic GFP expressing PbA strain (PbA-GFP) as control. Even though the proliferation of CFSE + CD8 + T cells was highly synchronous in Il22 −/− and wt mice (Fig. 7A), the lack of IL-22 resulted in an increased fraction of CD8 + T cells that start to proliferate. Hence this led to a significant increase in the recovery of proliferating OT1 T cells in Il22 −/− recipient mice after infection with PbA-OVA on d5 p.i. (Fig. 7B).   The infection of C57BL/6 mice with PbA is one of the closest models to mimic the pathology of P. falciparum induced cerebral malaria in humans 32,43 . Hence this murine model was applied to elucidate the effect of IL-22 on the outcome of malaria. Here we could show that the absence of IL-22 in murine malaria exacerbates cerebral malaria symptoms. That IL-22 plays a protective role was already found during P. chabaudi infection 31 . Also in other diseases like multiple sclerosis 44 or hepatitis 24,45 IL-22 were shown to exert protective functions. During a P. chabaudi infection CD4 + and CD8 + T cells were shown to produce IL-22 31 , whereas in a PbA infection γδ + T cells beside others showed the highest ability to produce IL-22. γδ + T cells are known to play a crucial role in the development of cerebral malaria 46 . They expand early during plasmodial infection 47 and human γδ + T cells are the major producers of IFNγ upon in vitro re-stimulation with schizonts antigens 48 . Similarly we observed an early production of IFNγ by γδ + T cells. Interestingly the early IFNγ production by γδ + T cells was significantly increased in IL-22-deficient mice, which might lead to the lower parasitemia in IL-22-deficient mice. The lower parasitemia in the absence of IL-22 signalling was also found using infection with the PyNL strain, which indicate an effect of IL-22 independent of the plasmodial strain used. In the absence of the closely related cytokine IL-10 a similar result was already described 49 . The infection of IL-10-deficient mice with P. chabaudi lead to a lower parasitemia and an enhanced mortality 38,39 .
Also the transient neutralization of IL-22 by antibodies induced a similar phenotype as seen in Il22 −/− mice. This excludes that compensatory mechanisms in Il22 −/− mice are responsible for the observed phenotype. It is already known that IL-22 deficiency may alter the microbiome of mice and therefor may influence the course of malaria indirectly. However it is unlikely that the transient neutralization of IL-22 alters the microbiome till the occurrence of cerebral symptoms 25,26 .
It is known that IL-10-defiency leads to hepatic pathology during murine malaria 40 . Thus liver damage was analysed in Il22 −/− mice. Even though hepatocytes express high levels of the IL-22Rα1 chain 29 , we observed no different levels of liver transaminases in PbA-infected wt and Il22 −/− mice. Moreover IL-10 is known to have a regulatory effect on development of T H 1 cell 50,51 . Thus we analysed if the closely related cytokine IL-22 5 may have similar effects on the T cell response during murine malaria. One key player during the antimalarial immune response is IFNγ which is the hallmark cytokine of a T H 1 dominated immune response. The high IFNγ production of different splenic lymphocyte subsets at an early time point of PbA infection and the enhanced susceptibility for cerebral malaria symptoms in the absence of IL-22 might be due to an enhanced T H 1 response during PbA infection 52 . Even though the IFNγ production is comparable between wt and Il22 −/− mice at d6 p.i., at d3 p.i. T cells from Il22 −/− mice produce significantly more IFNγ. Interestingly a significantly reduced production of IL-17 is detectable regardless of the time point after infection. The decreased IL-17 production in Il22 −/− mice might be a consequence of the early increased IFNγ production by T cells in the absence of IL-22, since IFNγ is known to suppress T H 17 differentiation 53,54 . Hence we assume that the T H 1 differentiation is more favoured in IL-22-deficient mice. This increased IFNγ production was also observed in the supernatant of antigen-specifically stimulated splenocytes from PbA-infected Il22 −/− mice using PbA-specific peptides 55,56 . To further identify the reasons for an increased activation of T cells we studied the function of antigen-presenting cells. Interestingly, CD8 + T cells from PbA-infected Il22 −/− mice showed already a higher IFNγ production upon re-stimulation with wt BMDCs compared to CD8 + T cells of PbA-infected wt mice. This difference in IFNγ production was even more pronounced if the CD8 + T cells were re-stimulated with BMDCs of Il22 −/− mice. Moreover, the re-stimulation of CD8 + T cells by BMDCs generated from Il22 −/− mice induced a stronger IFNγ release regardless of the origin of the CD8 + T cells. This finding goes along with the observations that the transient neutralization of IL-22 in the beginning of plasmodial infection leads to the same malaria pathology, concluding that IL-22 has an influence on the activation of the adaptive immune system. The increased ability to stimulate T cells of Il22 −/− BMDC might be explained by their increased expression of the costimulatory molecule CD86 on dendritic cells in the spleens of PbA-infected Il22 −/− mice compared to wt mice. Thus IL-22 may dampen directly or indirectly the expression of costimulatory ligands of DCs during PbA infection. However, the addition of recombinant IL-22 during LPS stimulation of BMDCs was unable to inhibit the expression of CD86 and CD80, which might be due to an overly strong stimulatory effect of LPS. In addition in was shown recently that several subtypes of DCs produce IL-22BP constitutively 10 , this might interfere with the direct interaction between IL-22 and BMDCs during this experiment. An alternative explanation would be that IL-22 influences maturation of DCs indirectly by inducing factors on yet unknown cells that interfere with DCs. Interestingly, using the infection with Listeria monocytogenes no increase of the antigen-specific T cell response in Il22 −/− mice was observed 24 . However in this infection model it remains unclear if IL-22 is induced and the absence of any effect of IL-22-deficieny on adaptive immunity might be due the lack of IL-22-induced signalling during Listeria monocytogenes infection.
Considering the enhanced IFNγ release and the increased expression of costimulatory ligands by DCs, it was of interest if the proliferation of PbA-OVAtg specific CD8 + T cells in vivo is improved in the absence of IL-22. Even though the proliferation of CFSE + CD8 + T cells was highly synchronous in wt and Il22 −/− mice at d5 p.i. the fraction of cells that start to proliferate was higher in IL-22-deficient mice leading to the assumption that the lack of IL-22 leads to an increased antigen specific activation of T cells and to a better survival or an altered migration of activated T cells.
Taken together the low parasitemia and the earlier onset of cerebral malaria in the absence of IL-22 might be due to the enhanced expression of costimulatory receptors by DCs followed by an increased activation of antigen specific CD8 + T cells and the earlier IFNγ production of different lymphocyte subsets.
Nevertheless it remains unclear how IL-22 modulates the immune response during PbA infection. In other diseases the expression of the IL-22Rα1 chain was already shown on hematopoietic cells 3,4 although it was for long postulated that these cells do not express the IL-22Rα1 chain 2,29 . Therefore the idea that cells of the immune system are completely absent of an IL-22Rα1 chain expression has to be reconsidered. However, further investigations are needed to elucidate that IL-22 directly acts on hematopoietic cells expressing the IL-22Rα1 chain during malaria. It is also possible that IL-22 interacts with non-hematopoietic cells and indirectly modulates the immune system via yet unidentified factors.

Materials and Methods
Ethics. C57BL/6J (wt), Tg(TeraTerb)1100MjB (OT1), NMRI and Il22 −/− mice were bred in the animal facility of the Bernhard Nocht Institute for Tropical Medicine. Il22 −/− mice were described previously 24 and were used when backcrossed 12 times on C57BL/6 background. For all Il22 −/− and wt comparisons co-housed mice were used. For experiments using only wt mice, animals were purchased from Charles River (Sulzfeld, Germany). All animal experiments were performed in agreement with the German animal protection law under the supervision of a veterinarian. The responsible federal health Authorities of the State of Hamburg (Behörde für Gesundheit und Verbraucherschutz) has reviewed and approved the experimental protocol under the permission number 32/15. Mice were sacrificed by cardiocentesis under deep CO 2 narcosis. All animals were used at the age of 7 to 10 weeks and were kept in individually ventilated cages under specific pathogen free conditions. Human samples. Plasma from 30 patients with acute P. falciparum malaria was used. These were obtained from travellers treated at the University Hospital Eppendorf in Hamburg. Samples were obtained after the first and fourth day of the treatment and display a parasitemia ranging from 0.01 to 5%. Diagnosis was performed in the diagnostic department of the Bernhard Nocht Institute for Tropical Medicine. 9 control samples were obtained from healthy donors. Ethical approval was obtained from the Ethikkommission Hamburg and all experiments were carried out in accordance with the relevant guidelines. Written informed consent was obtained from all enrolled participants prior to inclusion in the study.
Plasmodium parasites and infections. PbA and PyNL strains were stored in liquid nitrogen in a solution containing 0.9% NaCl, 4.6% sorbitol and 35% glycerol. Pb-GFP and Pb-OVAtg were a kind gift of W.R. Heath 57 .
To start an experiment, mice were injected with the respective Plasmodium strain intraperitoneal (i.p.) and the blood was collected after 5 to 6 days post infection (p.i.). Infected red blood cells (iRBCs) were subsequently used to infect the experimental mice with either 1 × 10 5 iRBC PbA or 2 × 10 4 iRBC PyNL i.p. To maintain the life cycle of plasmodia NMRI mice with a high parasitemia were anesthetized and used for a blood meal of Anopheles stephensi (A. stephensi) mosquito at the mosquito colony of the Bernhard Nocht Institute for Tropical Medicine. After 18 to 25 days the sporozoites were isolated by manual dissection of the salivary glands of mosquitos. For further experiments 1000 sporozoites were injected i.v. into each mouse. PbA-infected C57BL/6 mice develop cerebral malaria symptoms, which include the occurrence of neurological symptoms such as ataxia, convulsion and coma. These symptoms usually appear between day 6 and 8 p.i. and ultimately lead to dead. Mice developing cerebral symptoms were scored daily as described previously (score 0: no symptoms; score 1: deceleration and weight loss <10%; score 2: ruffled fur; score 3: weight loss <20%, strong deceleration, paralysis of one leg; score 4: coma; score 5: dead) 58 . To avoid unnecessary suffering, mice were euthanized when reaching score 3 and were registered as dead in Kaplan-Meier survival curve statistics, since mice developing score 3 never cleared Scientific RepoRts | 6:28058 | DOI: 10.1038/srep28058 symptoms in our model and would progress to score 5 within several hours. All experiments were in accordance with the local Animal Ethics Committee Regulations.

Analysis of transaminases.
PbA-infected mice were sacrificed at d6 p.i., the blood was collected in EDTA coated tubes and centrifuged at 5000 rpm, 15 min at RT. The plasma was stored at −20 °C. ALT and AST levels out from the plasma were analysed in 1 to 3 dilution in PBS with the Reflotron test system (Roche).

Preparation of bone marrow derived dendritic cells (BMDCs).
Cells were collected out of the femur and tibia. Subsequently, these cells were treated with erythrocyte lysis buffer, the cell number determined and 3 × 10 6 cells were seeded per 10 cm dish. Bone marrow derived dendritic cells were generated in complete DMEM medium in the presence of granulocyte macrophage colony-stimulating factor (GM-CSF) for 7 days. GM-CSF containing complete DMEM medium was added to the culture each third day additionally. At d7 BMDCs were used for further experiments.
In vitro CD8 + T cell stimulation. 2000 BMDCs were seeded per well in a 96 round bottom-plate and pulsed with 1 μg/mL PbA Peptide for 4 h in complete DMEM medium at 37 °C and 9% CO 2 . After one washing step with PBS, 5 × 10 4 splenic CD8 + T cells, which were enriched through magnetic cell separation (MACS, Miltenyi) to a purity of >95%, were added to each well. The culture was incubated for 24 h at 37 °C at 9% CO 2 in complete DMEM medium. The supernatant was stored at −20 °C for further analysis.
Isolation of primary Hepatocytes. Primary Hepatocytes (pHepa) were isolated as described in Burghardt et al. 60 . Subsequently isolated cells were stored in TRIzol reagent (Life Technologies) at −70 °C.
Scientific RepoRts | 6:28058 | DOI: 10.1038/srep28058 Statistical analysis. The GraphPad Prism 5 software was used to perform statistical analysis. Statistical differences in survival were analysed using Kaplan-Meier survival curve statistics. Additional statistical differences were identified using an unpaired t-test. For parasitemia testing the Mann-Whitney test was performed, assuming no Gaussian distribution for tested experiments. A regular two way ANOVA with a Bonferroni post-test was performed for pooled data. p-Values were indicated as *p < 0.05; **p < 0.01; ***p < 0.001. All experiments were performed at least twice and indicated error bars are based on SEM values.