The immunomodulatory effect of microglia on ECM neuroinflammation via the PD‐1/PD‐L1 pathway

Abstract Introduction The experimental cerebral malaria (ECM) model in C57BL/6 mice infected with Plasmodium berghei ANKA (PbA) has revealed microglia are involved in the ECM immune microenvironment. However, the regulation of microglia in the ECM immune response is not clear, and there is no safe and efficient treatment clinically for the protection of the nerve cells. Aims To elucidate the negative regulation mechanism in the ECM brain mediated by microglia. Furthermore, to investigate protective effect of the appropriate enhancement of the PD‐1/PD‐L1 pathway in the brain against ECM through the intrathecal injection of the adenovirus expressing PDL1‐IgG1Fc fusion protein. Results The PD‐1/PD‐L1 pathway was induced in the ECM brain and showed an upregulation in the microglia. Deep single‐cell analysis of immune niches in the ECM brainstem indicated that the microglia showed obvious heterogeneity and activation characteristics. Intrathecal injection of recombinant adenovirus expressing PD‐L1 repressed the neuroinflammation and alleviated ECM symptoms. In addition, the synergistic effect of artemisinin and intracranial immunosuppression mediated by PD‐L1 was more efficacious than either treatment alone. Conclusion The appropriate enhancement of the PD‐1/PD‐L1 pathway in the early stage of ECM has an obvious protective effect on the maintenance of immune microenvironment homeostasis in the brain. Regulating microglia and the PD‐1/PD‐L1 pathway could be considered as a promising approach for protection against human cerebral malaria in the future.


| INTRODUC TI ON
Cerebral malaria (CM), caused by Plasmodium falciparum infection, is one of the most serious forms of malaria. Clinical studies have found that most patients with CM can quickly develop central nervous system (CNS) symptoms, and an excessive inflammatory response is the chief cause of damage to the blood-brain barrier (BBB) and the occurrence of CM. 1 The current treatment of CM is limited to artemisininbased combination therapy (ACT) and emergency support due to the lack of targeted neuroprotective and vascular therapies. Excessive inflammatory responses in CNS may prevent homeostasis of the immune microenvironment, leading to the failure of antimalarial drug therapy.
Microglia, brain-resident macrophages, 2 proliferated in the early stage in the experimental cerebral malaria (ECM) model infected with Plasmodium berghei ANKA (PbA), upregulated the inflammatory level, 3 and aggravated CM nerve inflammation 4 ; activated microglia could closely interact with the parasite-specific T cells around the microvessels 5 and impacted the physiological migration of nerve cells. 6 By regulating the immune state of microglia, a balance could be struck between the CM immune defense response and immune-pathological injury.
A murine cytomegalovirus-induced encephalitis study found that activated microglia, via the PD-L1 pathway, inhibit the IFNγ and IL-2 expression in CD8 + T cells. 7 Microglia could overexpress PD-L1, leading to T cell dysfunction and apoptosis in glioblastoma. 8 Surgical brain injury research has shown that activation of PD-1/PD-L1 signaling with exogenous PD-L1 protein could significantly reduce the inflammatory response and brain edema. 9 These results suggest that the PD-1/PD-L1 pathway on microglia played an important role in the immune regulatory mechanism of the nervous system.
We hypothesized that microglia may be involved in the regulation of ECM-immune niches through the PD-1/PD-L1 pathway and further enhancement of the PD-1/PD-L1 pathway in the brain could reduce inflammatory damage and affect the outcome of ECM. To test this hypothesis, mice were intrathecal injected (i.t) with the PD-L1 adenovirus expression vector specific and showed a neuroprotective effect against ECM. We proposed that microglia participate in the ECM intracranial inflammatory response and could inhibit the T cells by expressing PD-L1.

| Parasites and infection
Plasmodium berghei ANKA was maintained and used as previously reported in our laboratory. All mice were injected intraperitoneally (i.p.) with 5 × 10 6 parasitized-infected red blood cells (pRBCs).
Parasitemia was determined from Giemsa-stained thin blood smears. The mice were monitored daily for symptoms of ECM. ECM mice may have symptoms, such as loss of appetite and weight, dull hair, and sluggish movement (Video S1). To evaluate the pathological changes in the brain during the onset of ECM, the brain was divided into four parts: the olfactory bulb, cerebellum, cerebrum, and brainstem. For experiments with multiple groups, all mice were first infected, and then randomly assigned to treatment groups.

| Mice and groups
A total of 159 mice were sacrificed for this study, all of which were C57BL/6 male mice (5-6 weeks old and 18-20 g in weight), including 118 mice used for infection experiments 10 and 41 mice used as normal controls. All mice were bred and housed under specific pathogen-free conditions. Different time groups were used, respectively, on the 3, 4, 5, 6, and 7 days after PbA infection. The intrathecal injection was performed in three infected groups, separately injected with empty vector adenovirus, IgG1Fc adenovirus, or PDL1-IgG1Fc adenovirus.
The normal mice were used as blank control, and the infected mice was used as the positive control.
In the experiments combining artemisinin with PD-L1, infected mice were treated with artemisinin combined with PDL1-IgG1Fc adenovirus i.t, artemisinin combined with IgG1Fc adenovirus i.t, artemisinin combined with empty vector virus i.t, artemisinin, or PDL1-IgG1Fc adenovirus i.t, respectively. The normal mice were used as blank control, and untreated infected mice were used as the positive control. The dose of artemisinin was 0.5 mg per mouse at a time and injected intraperitoneally twice on the 4th and 5th days after infection.
All these antibodies and reagents were used with the schedules and doses indicated in the manufacturer's manual.

| Quantitative PCR
After anesthesia with 1% pentobarbital sodium i.p. (50 mg/kg), the mice were perfused through the heart with 10-ml preheated 0.9% NaCl solution to remove blood, and the whole brain was extracted and crushed. Total RNA was extracted using RNAiso Plus (TaKaRa Bio) according to the manufacturer's protocol. The PrimeScript RT Master Mix kit (TaKaRa Bio) was used to prepare cDNA from total RNA following standard protocols. Quantitative PCR was performed on a real-time PCR system (Vazyme) using synthetic primers (Table S1). Samples were subjected to 40 cycles of amplification at 95°C for 5 s and 60°C for 15 s after an initial hold at 95°C for 30 s.
Relative expression data were calculated using the 2 −ΔΔCT method.
The transcriptional level of β-action was used as control.

| Purification of leukocytes in the brain parenchyma
The mice were euthanized and perfused through the heart with 10ml 0.9% NaCl solution to remove non-adhered RBCs and leukocytes from the cerebrovascular. The brains were removed, cut into small pieces, and crushed into a single cell in an RPMI medium. The brain homogenates were prepared into a 10-ml RPMI medium with a 30% Percoll gradient medium (GE Healthcare 17-0891-02). The pelleted cells were further purified in 2 ml of 70% Percoll gradient medium.
The upper Percoll layers, containing myelin debris and cells other than leukocytes, were carefully removed. The cells at the layered interface were carefully collected using a Pasteur pipette and washed in 8-ml D-Hanks. After centrifugation, leukocytes were prepared for staining.

| Flow cytometry
The cells were resuspended in 200 µl of FACS buffer (D-Hanks with 0.5% FBS) over a 40μm strainer. The antibodies were added at the concentrations described above. Cells were incubated at 4°C for 30 min in the dark, washed twice and resuspended in 400 µl of FACS buffer. The detection was used flow cytometry (BD FACSCanto) based on fluorescent labeling of the antibody, and the data were analyzed using Flow Jo 7.6.1. Each sample collected was 0.1-1 × 10 5 cells.

| BBB integrity assay
In vivo, BBB permeability was checked using an Evans Blue (EB) permeability assay. The mice were injected with 100 µl of 1% EB in PBS i.p. After 8 h, the mice were lethally anesthetized and perfused intracardially with 20-ml 0.9% NaCl solution to remove intravascular EB. Each brain was removed, weighed, and incubated in 0.5-ml formamide overnight at 56°C. The amount of extracted EB was determined using photometric analysis of the EB-formamide solution at 620 nm and by comparison with an EB standard curve.

| Histology
To prepare paraffin sections of the brain, the mice were perfused with 10-ml 0.9% NaCl solution. The brains were carefully extracted and fixed in 4% PFA overnight. Dehydration was performed using sequential alcohol washes with 50%, 70%, 90%, and 100% ethanol. Xylene was used to perforate the brain tissue. The brains were molded using paraffin wax, and 5μm tissue sections were prepared, collected on poly-L-lysine-coated slides, and IF stained.

| Immunofluorescence staining
After dewaxing and rehydration, paraffin sections were treated with Citrate Antigen Retrieval Solution (BBI E673001), and then blocked with blocking buffer (2% FBS, 3% BSA, and 0.2% Triton X-100 in PBS). The sections were incubated with primary antibodies diluted in blocking buffer overnight at 4°C (isotype-matched antibodies were used as controls). After washing thrice with PBST (PBS with 0.2% Triton X-100), the sections were incubated with secondary antibodies diluted with blocking buffer for 2 h at 25-30°C in the dark. The nuclei were stained with a DAPI staining solution (Servicebio G1012) for 10 min. Slides were mounted with Antifade (Servicebio G1401).

| Recombinant adenovirus that expresses the PDL1-IgG1Fc fusion protein
Primers for murine PD-L1 and IgG1Fc were designed and synthe- IgG1Fc were generated using RT-PCR from mouse T cell mRNA.
Cysteines in the hinge region were replaced with serines, and the PD-L1 fusion protein has a low ADCC or antibody-mediated opsonophagocytosis effect. After digestion with EcoR I and Bgl II and gel purification, the PCR products were fused to the Fc domain of mouse IgG1Fc. The fusion genes were transfected into a recombinant adenovirus. Then, 293 T cells were infected with this recombinant adenovirus, and the supernatant was collected and detected using western blotting with anti-PD-L1 mAb to measure PDL1-IgG1Fc.

| Single-cell RNA-seq
After the heart perfusion with saline to remove non-adhered cells,

| Intrathecal injection
The back of the mice was shaved after anesthesia with 1% pentobarbital sodium i.p (50 mg/kg). Intrathecal injection of 5µl recombinant adenovirus (1 × 10 9 pfu/ml) was performed in the anterior line of the double sacrum on the midpoint of the skin (spinous process) using a 25µl microsyringe. The spot was pressed for 1 min after injection.

| Statistical analyses
All the data were processed and analyzed with GraphPad Prism software 9.0. The data were expressed as mean ± standard deviation (SD). One-way ANOVA was used for multiple-group comparisons, and an unpaired t-test was used for comparison between two groups. Statistical difference for survival curve was analyzed using the log-rank test. p-values of <0.05 were considered significant.

| PbA infection induced the PD-1/PD-L1 pathway in the brain
The expression of PD-1/PD-L1 in human neuroimmune cells was extremely low or undetectable under physiological conditions. 11 Using C57BL/6 mice infected with PbA, the transcription and protein level of PD-1 and PD-L1 were detected separately using qPCR and western blotting, and first verified to be involved in the ECM The microglia were separated from whole brain tissue using Percoll density gradient centrifugation, 12 among which the microglia were the majority in the extracted cells, while other white blood cells were also extracted. The rate of PD-1 + cells among the extracted cells increased visibly on the 5th day (p < 0.001) and 7th day (p < 0.0001) after infection; the quantity of PD-1 + cells increased on the 5th day (p = 0.0001) and 7th day (p = 0.0002) after infection, comparing with the normal mice ( Figure 1C). Further analysis showed that the proportion of CD45 + cells in PD-1 + cells increased significantly on the 5th day (p = 0.0004) and 7th day (p = 0.0006) after infection ( Figure 1C,D). The IF staining of brain tissue sections showed that PD-1 + cells and PD-L1 + cells were widely observed in the cerebrum, cerebellum, and olfactory bulb of ECM mice on the 7th day, and mainly located in the cerebral microvascular obstruction ( Figure 1E; Figure S1A).

| Microglial activation and upregulation of the PD-1/PD-L1 signaling pathway in the ECM brain
To clarify the obvious activation and expression profile changes in microglia in the ECM brain, qPCR was used to detect the transcriptional levels of microglial markers in the brain. There was a significant The microglia were isolated from the brain and co-incubated with anti-CD11b and anti-CD45 mAbs. Flow cytometry analysis shows that CD11b + CD45 + cells in the ECM brain increased (B), and CD45 int cells among CD11b + cells significantly increased after infection (C). CX3CR1-GFP mice were infected with PbA and the brains were captured. qPCR shows that the expression levels of macrophage chemokine CCR2 and CCL2 increased (D); the CCR2 + cells increased in the ECM brain using flow cytometry analysis (E). Flow cytometry analysis shows that the PD-L1 + cells in the brain increased at 5 dpi (F) and 7 dpi (G) with PbA. IF-stained CX3CR1-GFP brain sections at 7 dpi demonstrate that GFP + and PD-L1 + have co-localization (H) and GFP + cells are close to PD-1 + cells (I). The results in (A-G) are expressed as the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate that the differences are significant (unpaired t-test, n = 3)  Figure 3C). The expression of PD-1 and PD-L1 in the brainstem was observed through IF staining on the 7th day after infection ( Figure 3D). The expression of the inflammatory cytokine IL-6 and iNOS was observed in the ECM brain through IF staining ( Figure   S3A,S3B).

| Single-cell analysis of immune niches in the brainstem of ECM mice
Transcriptome analysis of microglia in mice with ECM has been com-  Figure 4B). In addition, astrocytes, neurons, and endothelial cells also expressed PD-L1 in the ECM group ( Figure 4B).
The peripherally infiltrating CD8 + T cells were the main cells in the brainstem of ECM mice, revealing the upper level of PD-1, and notably, microglia also expressed an amount of PD-1 ( Figure 4B). Further analysis revealed that the Ki67 of microglia in the brainstem of ECM mice was enhanced compared with that in normal mice, and the expression of CCL2 in the microglia also increased significantly ( Figure 4C). It should be a concern that the proportion of CD45 high cells increased, while the proportion of CD45 int cells decreased ( Figure 4D), showing that the expression of CD45 in microglia increased significantly after PbA infection. IF staining of brainstem tissue sections confirmed that CX3CR1-GFP + cells expressed PD-L1 on the 7th day after infection ( Figure 4E).
Further analysis of the microglia subgroups revealed that there was heterogeneity between the ECM mice and normal control mice ( Figure 4F). Clustering analysis revealed four unique microglial subgroups across ECM pathological and physiological conditions. Cluster sizes ranged from 4.91% to 46.31% of total microglia, and Cluster 3 was almost entirely composed of microglia in the ECM brain and showed obvious transcriptional characteristics that were different from those in physiological conditions ( Figure 4G). Analysis of microglia in the physiological state of the brain revealed that there were only two clusters, with the larger group (Cluster 5) accounting for more than 80% ( Figure 4H). and M2 macrophage genes almost declined and were expressed by most of the Cluster 9 cells, but only one (Ccl5) was highly expressed relative to Cluster 9 compared with Cluster 7 ( Figure 4K).

| Specific enhancement of the PD-1/PD-L1 pathway in the brain protects against ECM
Intravenous injection of PDL1-IgG1Fc fusion protein in mice can effectively inhibit the occurrence of ECM. 10 However, systemic enhancement of the PD-1/PD-L1 signaling pathway may reduce the ability of the immune system to eliminate malaria parasites.
To verify the protective effect of intracerebral-specific PD-1/ PD-L1 enhancement, an adenovirus expression vector was constructed to express the PDL1-IgG1Fc fusion protein ( Figure S4A) using adenovirus expressing IgG1Fc as controls. By intrathecal injection, viral vectors were distributed freely throughout the brain with cerebrospinal fluid (CSF), 18 infected nerve cells spontaneously, and reached its peak expression in 3-4 days after infection. Adenovirus was intrathecally injected, and the level of PD-L1 protein in the injected brain increased (p = 0.0006) compared with the non-injected brain using western blotting ( Figure S4B).

| Intracerebral enhancement of PD-L1 affected the immune microenvironment of each brain region
To further clarify the effect of the PD-L1 fusion protein in different brain regions, qPCR was used to detect the transcriptional levels of ( Figure S5D).

| Intracerebral enhancement of PD-L1 affects the activation and inflammation of microglia
On the 7th day after infection, the transcriptional levels of microgliaspecific markers were detected using qPCR. Compared with the  Figure 7H; Figure S6D).
Similarly, the quantity of CD11b + cells decreased observably in the combined treatment group relative to the artemisinin group (p = 0.0097) or the PD-L1 group (p = 0.0012) ( Figure 7H; Figure   S6D). Further analysis showed that the proportion of CD45 + in CD11b + cells was upregulated in the combined treatment group compared with the artemisinin group (p = 0.0048) or the PD-L1 group (p = 0.0074) ( Figure 7H; Figure S6E) and the quantity of CD11b + CD45 + cells showed no difference ( Figure S6F).

| DISCUSS ION
CM is one of the most serious complications caused by Plasmodium falciparum infection, with a mortality rate of 25%. 19  However, the occurrence of many nervous system diseases was also related to the infiltration of peripheral monocytes. The monocyte-derived macrophages could cross the BBB to infiltrate the brain; however, the expression profiles of myeloid macrophages and microglia were different. 17 More importantly, microglia, as the main cells expressing PD-L1 in the brain immune microenvironment, were activated by the upregulation of PD-L1 expression. However, the existing PD-L1 level in the brain was difficult to inhibit the occurrence of ECM, so we tried to specifically enhance the PD-L1 function in the brain to alleviate the inflammation. Owing to the lack of efficient and specific methods to regulate PD-L1 expression in microglia, we enhanced the PD-1/PD-L1 signaling pathway through the intrathecal injection of recombinant adenovirus expressing PDL1-IgG1Fc. Our results showed that increased expression of PD-L1 in the brain significantly increased the survival rate of mice; reduced the incidence of ECM and the degree of BBB damage; downregulated the transcription levels of the inflammatory cytokines TNFα, IL-6, and IL-10; and decreased nerve cell death. PbA infection also induced microglia to upregulate the expression of PD-1, suggesting that it could also be negatively regulated by immunity.
Aging and chronic inflammation could activate microglia; increase the pro-inflammatory response, oxidative damage, and phagocytosis; and deliver antigens to infiltrating CD4 + T or CD8 + T cells through MHC-I/II molecules. 30 In response to effective communication with T cells, the phenotype of microglia could change from the active M1 type to the neuroprotective M2 type, which corresponded to tissue remodeling and homeostasis. 31 The singlecell RNA-seq results suggested that microglia showed significant heterogeneity, and the expression of M1 polarized related genes of macrophages was upregulated. We found that intrathecal injection

CO N FLI C T S O F I NTE R E S T
The authors declare they have no conflicts of interest.

AUTH O R S' CO NTR I B UTI O N S
YS, YZ, XAW, and YHL designed and supervised the study. YS, YHL, QHZ, JW, JL, and YXH performed the experiments and collected the data. YS, YHL, and QHZ summarized, analyzed, and plotted the data.
YS wrote and finalized the manuscript. All authors have approved the final version of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset generated and analyzed during the current study is available from the corresponding author upon reasonable request.
Our original single-cell RNA-seq data have been submitted to the database of the NCBI Sequence Read Archive (http://www.ncbi.nlm. nih.gov/bioproject/) under the BioProject ID: PRJNA735877.