All animal experiments were approved by the Institutional Review Board of the Fourth Military Medical University (No: 20160107) and were in complete compliance with the principles of ARRIVE guidelines for reporting animal research (15). All efforts were made to minimize suffering of animals employed in this study.

C57BL/6 male mice and Pdcd1^−/− C57BL/6 male mice (6–8 weeks old) were used for infection experiments. Pdcd1^−/− mice were kindly provided by Dr. T. Honjo, and all mice were bred and housed under specific pathogen-free conditions in the animal center of FMMU. Plasmodium berghei ANKA strain (PbA) was maintained and used as previously reported in our laboratory. All mice were infected intraperitoneally (i. p.) with 1 × 10⁷ pRBCs. Parasitemia was determined by Giemsa-stained thin blood smears. Mice were monitored daily for body weight and symptoms of ECM. For experiments with multiple groups, all mice were first infected, and then randomly assigned into treatment groups.

Antibodies against mouse CD3 (SP7, ab16669), CD28 (PV-1, ab25234), and CD31 (ab28364) were purchased from Abcam. APC-labeled antibodies against mouse PD-1 (29F.1A12, 135209) and CD31 (390, 102409) were purchased from Biolegend. IL-1β (DKW12-2012-096), TNF-α (DKW12-2720-096), IFN-γ (DKW12-2000-096), IL-10 (DKW12-2100-096), and TGF-β (DKW12-2710-096) were measured using murine ELISA kits from DAKEWEI. CFSE Cell Division Tracker Kit was purchased from Biolegend. Recombinant mouse IFN-γ (50709-MNAH) and M-CSF (51112-MNAH) were purchased from SinoBiological. Cell Counting Kit (C008-3) was purchased from 7 Sea Pharmatech. LDH Cytotoxicity Assay Kit was purchased from Beyotime Biotechnology. LPS (L2630) was purchased from Sigma-Aldrich. FITC-BSA (bs-0292P-FITC) was purchased from Biosynthesis Biotechnology. A MILLIPLEX MAP KIT (MCYTOMAG-70K) was purchased from Merck Millipore. All these antibodies and reagents were used in the schedules and doses indicated.

The technique for isolating mouse BMECs was adapted from published protocols (16). Mice were euthanized and perfused with saline. And brains were finely minced with 1 ml of medium and homogenized by passing through a 23-gauge needle. The homogenate was mixed with an equal volume of 30% dextran (MW 70,000, BBI) in PBS and centrifuged at 10,000 g for 15 min at 4°C. The pellet was resuspended in PBS and passed through a 40 μm cell strainer that retained the microvessels. After washing, the cell strainer was back-flushed with 2 ml PBS over a 6-well plate to collect the microvessels, which were rocked at room temperature with 2% FBS, 1 mg/ml collagenase II (02100502, MP Biomedicals) and 20 μg/ml DNase I (10104159001, Sigma-Aldrich) for 90 min. Vessel fragments were collected and resuspended in EC medium (0.1 mg/ml EC growth supplement from ScienCell, catalog #1001) with 4 μg/ml puromycin and seeded into a collagen-coated 6-well plate. The medium was replaced (without puromycin) 3 days later and every 3–4 days thereafter. The purity of BMECs was identified with CD31 by flow cytometry. For cytokine activation of BMECs, 20 ng/ml IFN-γ was added to the cell medium 24 h prior to subsequent analysis.

Mice infected with pRBCs 7 dpi were euthanized and perfused with saline to remove non-adhered RBCs and leukocytes from the brain. Brains were removed, cut into small pieces and crushed in RPMI medium; the brain homogenates were centrifuged at 250 g for 10 min at 4°C, the pellets were dissolved in RPMI medium containing 1 mg/ml collagenase II and 10 μg/ml DNase I for 30 min at 37°C. Cell debris was removed by pushing the mixture through a 40 μm cell strainer. The tissue extract was then centrifuged at 400 g for 5 min. The pelleted cells were further purified on a 30% Percoll gradient (17-0891-02, GE Healthcare). The upper Percoll layers were carefully removed, and the cell pellet resuspended in PBS. The pellet was resuspended in RBC lysis buffer and incubated on ice for 5 min to lyse adherent pRBCs. BSLs were resuspended in PBS and counted. CD8⁺ T cells were negatively isolated from BSLs according to the manufacturer's instructions (558471, BD).

To detect the cytotoxicity of activated CD8⁺ T cells to brain endothelial cells, we constructed a BBB model in vitro with the bEnd.3 endothelial cell line. The cells (2 × 10⁴) were seeded onto the upper chamber of a 24-well Transwell system (0.4 μm, CLS3450-24EA, Corning). Transwell was checked for the formation of an intact monolayer on the insert by adding FITC-BSA (50 μg/ml) to the upper chamber and measuring the amount of FITC-BSA that passed into the lower chamber. The Transwells were used only when the intensity of fluorescence in the lower chamber was negligible, and bEnd.3 cells were stimulated with IFN-γ (20 ng/ml) and parasites (3 × 10⁶ pRBCs) 24 h. bEnd.3 were washed, and 1 × 10⁶ activated CD8⁺ T cells from PbA-infected mice were added. The extent of BBB damage by CD8⁺ T cells is reflected by the diffusion rate of the FITC-BSA.

BMECs were isolated from uninfected C57BL/6 mice as described above for an ex vivo cell-killing assay. BMECs were activated with IFN-γ (20 ng/ml) and co-incubated with pRBCs for 24 h. Then, the BMECs were incubated at various effector:target (E:T) ratios with activated/naïve CD8⁺ T cells. The cell culture supernatants were collected, and LDH release cytotoxicity assays were carried out to detect the cytotoxicity of CD8⁺ T cells for an LDH content assay. Moreover, granzyme B in the supernatants was determined using ELISA kits.

Bone marrow-derived macrophages were planted into 6-well cell culture clusters and stimulated with a sub-optimal concentration of IFN-γ (0.5 ng/ml), (Figure S4) 1 × 10⁷ pRBCs were subsequently added. Next, these wells were divided into three groups, adding PDL1-IgG1Fc and IgG1Fc as well as cell culture medium as controls. After 24 h incubation, the above mentioned stimulating factors, such as IFN-γ, pRBCs, and soluble fusion proteins were washed away via replacing the culture medium. Then, splenic CD8⁺ T cells were added to the above wells. After 24 hincubation, CD8⁺ T cells were collected, and the expression of perforin and granzyme B were detected through qPCR.

Splenic CD8⁺ T cells were negatively isolated from PDL1/PDL2-IgG1Fc-treated or IgG1Fc-treated mice by magnetic sorting (558471, BD). CD8⁺ T cells were resuspended at 1 × 10⁷ cells/ml in the 5 μM CFSE working solution; the cells were incubated for 20 min at 37°C and protected from light, and then the staining was stopped by adding 5 times the original staining volume of cell culture medium containing 10% FBS. CFSE-labeled cells were harvested, sub-optimal concentrations of anti-CD3 and anti-CD28 mAbs (0.03 μg/ml) were added to the culture system for 4 days, proliferation was assessed by CFSE dilution with flow cytometry and analyzed by Flowjo software.

In vivo BBB permeability was checked by an EB assay. Mice were injected by intravenous (i. v.) with 100 μl of 1% EB in PBS. After 8 h, the mice were lethally anesthetized and perfused with saline to remove intravascular EB. Each brain was removed, weighed, and incubated in 0.5 ml formamide overnight at 37°C. The amount of extracted EB was determined by a photometric analysis of the EB-formamide solution at 620 nm and by comparison to an EB standard curve. EB showed red fluorescence under a fluorescence microscope, which could reflect BBB damage in situ.

Brain water content was measured as previously published. Weights of brains removed at the indicated time points were compared with the dry weight after overnight incubation at 80°C.

To prepare paraffin sections of the brain, mice were perfused with saline. Brains were fixed in 4% paraformaldehyde (PFA) overnight. Dehydration was carried out using sequential alcohol washes with 30, 50, 70, 90, and 100% ethanol. Xylene was used to perforate the brain tissue. The brains were molded using paraffin wax, and 5 μm tissue sections were prepared, collected on poly-L-lysine-coated slides, and stained with H&E. Slides were checked under a light microscope to capture images.

10 μm-thick frozen sections were fixed in 4% PFA and blocked with 10% goat serum. Sections were incubated with unlabeled specific primary antibodies for 90 min at room temperature (isotype-matched antibodies used as controls) and with fluorescence-labeled secondary antibodies for 30 min at room temperature. Immunofluorescence was examined by fluorescence microscope (Olympus BX51).

We prepared the ECDs of PDL1 and PDL2 and constructed fusion proteins with the ECDs linked to the hinge-CH2-CH3 domains of murine IgG1. Primers that were designed and synthesized based on the published sequence:

cDNAs encoding the mouse PDL1/PDL2 extracellular domain and IgG1Fc were generated by RT-PCR from mouse T cell mRNA. Cysteine in the hinge region was replaced by serine. After digestion with EcoR I and Bgl II and gel purification, the PCR products were fused to the Fc domain of mouse IgG1Fc. The recombinant plasmids were transfected into the FreeStyle 293 expression system (ThermoFisher, K900010) and produced as described. Then, the proteins were purified directly from cell culture supernatants harvested by protein A-Sepharose. The protein concentrations were measured by absorbance at 280 nm, and we used SDS-PAGE and western Blot to confirm purity.

PD-1/PDL signaling can inhibit the TCR-mediated proliferation of T cells. Therefore, the effect of the PDL1/PDL2 fusion proteins was evaluated by their ability to inhibit CD8⁺ T cell proliferation stimulated by anti-CD3 and anti-CD28 mAbs.

Statistical analyses were performed using PRISM GraphPad™ software (La Jolla, CA, USA) and included Student's t-test; p-values <0.05 were considered to be significant.

We used the well-characterized PbA infection of C57BL/6 mice as the ECM model, which was confirmed to be credible in our laboratory (Figure S1). To verify the importance of the PD-1 pathway in ECM development, we used PbA-infected Pdcd1^−/− C57BL/6 mice as the ECM model. We found that Pdcd1^−/− mice showed earlier neurological signs of ECM than WT mice, and the median survival time of Pdcd1^−/− mice was 7 days, which was 2 days shorter than that of WT mice (Figure 1A). PD-1 pathway deficiency led to higher levels of T cell proliferation detected by CFSE assay (Figures 1B–D) and Cell Counting Kit (CCK-8) assay (Figure 1E). Then, we evaluated the expression of PD-1, PDL1, and PDL2 by brain microvascular endothelial cells (BMECs). BMECs were either untreated or activated with IFN-γ and pRBCs to mimic the pro-inflammatory microenvironment typically observed in the brain of ECM mice. PD-1, PDL1, and PDL2 mRNA expression levels in BMECs were examined by qPCR (Figure 1F). PDL1 was induced 4-fold at 24 h, while PD-1 and PDL2 showed little up-regulation. In addition, the expression of ICAM-1 was highly up-regulated in the activated BMECs. Furthermore, we detected the ratio of splenic PD-1⁺ CD8⁺ T cells to total CD8⁺ T cells in ECM mice from 0 day post-infection (dpi) to 10 dpi (Figure 1G) and found that the ratio of PD-1⁺ CD8⁺ T cells rose after infection (6.0%) and peaked (22.8%) at 5 dpi. All these results confirmed that the PD-1pathway participates in ECM development and that PDL1 may play a more important role than PDL2 does.

The prevailing ECM model is characterized by a severe disruption of BMECs, which is caused by PbA-specific CD8⁺ T cells. To test this model, we added activated splenic CD8⁺ T cells, which were isolated from a PbA-infected mouse immediately prior to ECM (7 dpi), to BMECs that stimulated with IFN-γ and pRBCs. The BMECs were almost entirely dead after 24 h co-incubation. When pRBCs was replaced with RBCs, the activated CD8⁺ T cells did not kill BMECs (Figure 2A). When naïve CD8⁺ T cells from an uninfected mouse were added to PbA-stimulated BMECs, BMECs showed some dysfunction caused by interaction with parasite antigens. Hence, this in vitro experiment confirmed that BMECs could be recognized and killed by the activated CD8⁺ T cells. Additionally, BMECs were incubated at various E:T ratios with naïve or activated CD8⁺ T cells, and co-culture supernatants were collected; an LDH-release cytotoxicity assay also showed profound cell damage of BMECs. These data indicated that, during ECM, the activated CD8⁺ T cell is the main contributor that kills BMECs (Figure 2B).

Based on the results mentioned above, we hypothesize that enhancement of the PD-1 pathway may down-regulate the inflammatory response and the cytotoxic effect of CD8⁺ T cells on BMECs. We constructed two soluble fusion proteins, PDL1-IgG1Fc and PDL2-IgG1Fc, which were confirmed by western blot (Figure S2A). The biologic activities of the fusion proteins were demonstrated by their inhibitory effect on the CD8⁺ T cells proliferation stimulated by sub-optimal concentrations of anti-CD3 and anti-CD28 mAbs. PDL1-IgG1Fc inhibited the proliferation of CD8⁺ T cells in a dose-dependent fashion relative to the IgG1Fc control; in contrast, PDL2-IgG1Fc had no effect on the proliferation of CD8⁺ T cells under the same conditions. There was no obvious synergistic effect between the two recombinant proteins (Figures S2B,C). These results showed that the PDL1-IgG1Fc fusion protein could attenuate the T cell proliferation stimulated by anti-CD3 and anti-CD28 mAbs.

Next, to determine whether PDL1/PDL2-IgG1Fc could modulate malaria pathogenesis, C57BL/6 mice were infected with PbA and treated with PDL1/PDL2-IgG1Fc and IgG1Fc at day 0. Body weight, parasitemia and survival were monitored daily. The PDL1-IgG1Fc treatment was most effective in delaying the onset of ECM, and compared with PDL2-IgG1Fc-treated and IgG1Fc-treated mice, PDL1-IgG1Fc-treated mice displayed prolonged survival times (Figure 3A), reduced body weight loss (Figure 3C) and splenomegaly (Figures 3D–E), although there was no obvious difference in peripheral parasitemia between PDL1-IgG1Fc-treated mice and PbANKA group (Figure 3B). Moreover, brain sections were analyzed histopathologically following H&E; consistent with previous results, mice treated with PDL1-IgG1Fc exhibited markedly less cerebral edema, fewer pRBCs sequestrations and less lymphocyte infiltration than PbANKA group (Figure 3F). These findings confirm that PDL1-IgG1Fc can protect mice from ECM.

Furthermore, ECM is associated with excessive systemic inflammation, which involves pro-inflammatory cytokine production that leads to BMECs activation and vascular permeability. In the serum of PbANKA group pro-inflammatory cytokines, including IFN-γ, TNF-α, and IL-1β, rose at day 7, as did anti-inflammatory cytokine IL-10. The levels of serum pro-inflammatory cytokines, including IL-1β, and IFN-γ, in PDL1-IgG1Fc-treated mice were much lower than those in IgG1Fc-treated mice; besides, the levels of anti-inflammatory cytokines, such as IL-10 were also reduced in PDL1-IgG1Fc-treated mice compared with those in PDL2-IgG1Fc-treated and IgG1Fc-treated mice (Figure 3G). Nevertheless, the levels of serum TGF-β were barely detectable in all groups of mice. In summary, PDL1-IgG1Fc treatment protected mice from ECM via relieving the systemic inflammation.

As BBB damage is the most important pathogenetic mechanism of ECM, we used an Evans blue (EB) permeability assay to evaluate BBB integrity. When the BBB is intact, EB cannot enter the brain parenchyma, after BBB breakdown, EB can enter the brain and shown the localization of BBB injury. The integrity of the BBB was compromised in PbANKA group, as demonstrated by EB leakage into the brain parenchyma at 7 dpi. Mice of PbANKA group were injected intravenously with EB 8 h before being killed, and the EB penetration was notably alleviated in PDL1-IgG1Fc-treated mice compared with that in PDL2-IgG1Fc-treated and IgG1Fc-treated mice (Figure 4A). The spectrophotometric quantification of EB revealed nearly 4-fold less EB in PDL1-IgG1Fc-treated mice than in other mice (Figure 4B). The fluorescence of EB in the cerebral cortex showed less BBB damage in PDL1-IgG1Fc-treated mice, with fewer BBB damage (Figure 4C). PDL1-IgG1Fc-treated mice showed less cerebral edema than did other mice (Figure 4D). All these results confirmed that PDL1-IgG1Fc plays a protective role against BBB damage during ECM.

As shown in the preceding results, PbA-specific CD8⁺ T cells play a central role in BBB breakdown. Splenic CD8⁺ T cells were isolated and stained with CFSE, then co-incubation with sub-optimal concentrations of anti-CD3 and anti-CD28 mAbs (0.03 μg/ml) for 4 days. The proliferation of T cells was detected by flow cytometry (Figure 5A) and analyzed by Flowjo software, divided index (Figure 5B) and percentage of divided cell (Figure 5C) demonstrated that the proliferation of splenic CD8⁺ T cells was much down-regulated in PDL1-IgG1Fc-treated mice compared with that in control mice (Figures 5A–C). Additionally, to detect the cytotoxicity of activated CD8⁺ T cells to brain endothelial cells, we constructed a BBB model in vitro with a Transwell system. Briefly, bEnd.3 cells were seeded into the upper chamber of a 24-well Transwell system, activated with IFN-γ, and then incubated with pRBCs. Activated splenic CD8⁺ T cells from PDLs-IgG1Fc-treated groups were added to the upper chamber. Using FITC-BSA as a permeability tracer, we evaluated the BBB leakage by the diffusion rate of the FITC-BSA from the upper chamber to the lower chamber. By this BBB model, we showed that the cytotoxicity of splenic CD8⁺ T cells to activated ECs was attenuated in the PDL1-IgG1Fc-treated mice compared with that in the PDL2-IgG1Fc-treated and IgG1Fc-treated mice (Figure 5D). To further confirmation the killing effect of activated CD8⁺ T cells on brain endothelial cells, primary BMECs were isolated, then co-incubation with pRBCs, IFN-γ, and splenic/brain-sequestered CD8⁺ T from different groups of mice. LDH assay showed the cytotoxicity of splenic CD8⁺ T cells notable reduction in PDL1-IgG1Fc-treated mice compared with that in control mice (Figure 5E). As CD8⁺ T cell-secreted granzyme B is known to drive CTL-mediated cerebral pathology, our ELISA data showed that the secretion of granzyme B was down-regulated in PDL1-IgG1Fc-treated mice compared with that in control mice, both by splenic (Figure 5F) and brain-sequestered CD8⁺ T cells (Figure 5G). All these results confirmed that PDL1-IgG1Fc protects the BBB against the cytotoxicity of CD8⁺ T cells via limiting T cell proliferation and relieving their targeted killing of BMECs.

We found that PD-1 was also expressed on the surface of macrophages at different polarization statuses (Figure S3). Thus, it was necessary to study the possible influence of PDL1-IgG1Fc on macrophages via the PD-1 pathway. At 4 dpi, the mice from different groups (“Control,” “PbANKA,” “PbANKA + PDL1-IgG1Fc,” and “PbANKA + IgG1Fc”) were euthanized. We detected the inflammatory cytokines secreted by splenic macrophages that were isolated from the mice of different groups, and the serum cytokines of these mice were also detected simultaneously. Among these serum cytokines, the level of IL-1β, one of the most important pro-inflammatory cytokines during PbA infection, was significantly higher in PbANKA group than in control group but was notably reduced by PDL1-IgG1Fc (Figure 6A). Meanwhile, the levels of other serum cytokines exhibited no difference between PbANKA group and PbANKA + PDL1-IgG1Fc group. Interestingly, we found that the IgG1Fc-treated mice showed a significant decline in pro-inflammatory cytokines (including IL-1β, IL-6, IL-12p70, TNF-α, and MIP-1α/β) and in anti-inflammatory cytokine IL-10 compared with PbANKA group, according to the data from ex vivo trials (Figure 6B). However, we found that either the pro-inflammatory cytokines (including IL-1β, IL-6, IL-12p70, TNF-α, and MIP-1α/β) or anti-inflammatory cytokine IL-10 in PbANKA + PDL1-IgG1Fc group exhibited no difference compared with PbANKA group. In sum, it appeared that in the early stage of PbA infection, the effect of PDL1-IgG1Fc treatment on cytokine secretion of macrophage was mild and partial.

To further investigate the effects of PDL1-IgG1Fc on macrophages, we constructed a macrophage-CD8⁺ T cell co-incubation model. Immediately before the CD8⁺ T cells were added, we first verified the macrophage polarization influenced by PDL1-IgG1Fc via detecting IL-6, IL-10, IL-12p70, and TNF-α secretion (Figures 7A–D). At this timing, all these cytokines secreted by IgG1Fc-treated macrophages decreased compared with PDL1-IgG1Fc-treated macrophages and “pRBC + IFN-γ” groups; furthermore, PDL1-IgG1Fc-treated macrophages showed increased IL-10 secretion and decreased secretion in IL-6, IL-12p70, and TNF-α, while anti-PD1 mAb was observed to oppose the effect of PDL1-IgG1Fc on IL-10 and IL-12p70. These results implied that PDL1-IgG1Fc repressed macrophages M1 polarization, and the effects of IgG1Fc on macrophages were analyzed in Discussion. Besides, the perforin and granzyme B mRNA expression levels in CD8⁺ T cells co-incubated with macrophages after being treated with PDL1-IgG1Fc were decreased markedly compared with the levels in the “pRBC + IFN-γ” group (Figure 7E), which indicated that the cytotoxicity of the CD8⁺ T cells was suppressed due to the decreased antigen presentation capability of the macrophages.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.