C57BL/6J mice (6–8 wk of age) were purchased from the Animal Institute of the Academy of Medical Science (Beijing, China). B cell-deficient mice (μMT, C57BL/6 background) were originally obtained from the Jackson Laboratory (Bar Harbor, ME, United States). All mice were bred and maintained in our pathogen-free animal facility. Only female mice were used in the current study. All mouse studies were reviewed and approved by the Animal Ethics Committee of the Army Medical University (Third Military Medical University) Institute of Medical Research, under the approval number AMUWEC2020135. P. chabaudi was maintained in our laboratory. For infection studies, mice were inoculated intraperitoneally with 1 × 10⁶ parasitized RBCs (pRBCs). Parasitemia was determined by examination of Giemsa-stained thin blood smears every other day after infection.

For splenocytes isolation, mouse spleens were mechanically disrupted with the back of a 10 ml syringe, filtered through a 70 μm strainer, and subjected to red blood cell lysis with 1 × red blood cell lysis solution (Miltenyi Biotec, Bergisch Gladbach, Germany). Cells were washed twice with cold Roswell Park Memorial Institute (RPMI) 1,640 medium supplemented with 2 μM glutamine, 100 U ml^–1 penicillin/streptomycin and 5–10% fetal bovine serum (FBS). Trypan blue staining showed a >90% viability of splenocytes. For scRNA-seq analysis, splenocytes isolated from P. chabaudi-infected mice (n = 5) at day 8 and day 16 post-infection, were loaded on a Chromium Controller (10 × Genomics, Pleasanton) at a concentration of 10³ cells/μl per sample, to generate Gel Bead-In EMulsions (GEMs). Barcoded sequencing libraries were constructed using the Chromium Single Cell 3′ Reagent Kits v3 (10 × Genomics) according to the manufacturer’s instructions. Following library preparation, sequencing was performed with paired-end sequencing of 150 nt at each end using one lane of the NovaSeq 6000 per sample. Raw sequencing data were processed using Cell Ranger (10 × Genomics; v6.0.1). The CellRanger function “count” was used to align raw reads with the STAR aligner against the mm10 reference genome. After counting non-redundant unique molecular identifiers (UMIs), single-cell digital-expression matrices were obtained. Library preparation and sequencing were performed at Jiayin Biotechnology Ltd. (Shanghai, China).

The output “cellranger” matrices were loaded into the Seurat package (v4.0.3) in R (v4.0.3). To remove low-quality cells, cells with unique gene counts (less than 200) and genes expressed in fewer than three cells were filtered out. Standard data pre-processing was performed to remove mitochondrial genes, residual red blood cells, cell doublets, and multiplets. The UMI count matrix for each cell was normalized using a global scaling normalization method “LogNormalize.” The “ScaleData” function implemented in the Seurat package was used to remove the unwanted sources of variation through the “vars.to.regress” argument. The 3,000 most variable genes were selected for further analysis. Before dimensionality reduction, the data were scaled such that the average expression was 0, and the variance was equal to 1. The first 10 dimensions of the principal component analysis (PCA) were used for cluster identification, and the clusters were identified using the FindClusters function with a resolution of 0.5. Datasets of multiple samples were combined into a single dataset using canonical correlation analysis with the “IntegrateData” function. Uniform manifold approximation and projection (UMAP) analysis was used to visualize the selected datasets. General cell types were annotated based on the cell type-specific marker gene expression and the genes driving heterogeneity of clusters (Cd79a, Cd79b, B cells; Cd3e, T cells; Flt3, DCs; Itgam, Adgre1, monocytes/macrophages; Klra8, NK cells; Ly6g, neutrophils; Pf4, platelets). Immune cell subpopulations were annotated using previously reported genes to identify the following populations: CD4⁺ T cells (Cd3e, Cd4), CD8⁺ T cells (Cd3e, Cd8a, Cd8b1), γδ T cells (Cd3e, Tcrg-C1, Tcrg-C2), NKT cells (Cd3e, Klrb1c, Klrc1), T helper type 1 (Th1) cells (Cd3e, Tbx21, Ifng), T follicular-helper (Tfh) cells (Cd3e, Bcl6, Cxcr5), T-regulatory (Treg) cells (Cd3e, Foxp3, Il2ra), naive CD8⁺ T cells (Cd3e, Cd8a, Cd8b1, Ccr7, Il7r), effector CD8⁺ T cells (Cd3e, Cd8a, Cd8b1, Gzma, Prf1), plasma cells (Cd79a, Cd79b, Jchain, Sdc1), GC B cells (Cd79a, Cd79b, Fas, Ighd).

The Seurat package in R (v4.0.3) was used to perform differential gene expression analysis and compare the expression profiles between the two groups. Differential expression analysis was conducted using the default two-sided non-parametric Wilcoxon rank-sum test with Bonferroni correction, using all genes in the dataset. P-values were adjusted for multiple testing using the false discovery rate (FDR). Genes with a P-value of < 0.05 was considered differentially expressed genes. The enhanced volcano plots were used to display the differentially expressed genes between the two cell clusters. Genes of interest were labeled.

The following antibodies were used for flow cytometry analysis: anti-mouse CD45R/B220 (APC, clone RA3-6B2, BioLegend), anti-mouse CD3 (FITC, clone 17A2, BioLegend), anti-mouse NK1.1 (PE, clone PK136, BioLegend), anti-mouse CD11c (APC, clone N418, BioLegend), anti-mouse CD11b (Percp/Cy5.5, clone M1/70, BioLegend), anti-mouse Ly6G (APC, clone 1A8, BioLegend), anti-mouse F4/80 (FITC, clone BM8, BioLegend), anti-mouse Ly6C (PE, clone HK1.4, BioLegend), anti-mouse CD11c (APC/Cyanine7, clone N418, BioLegend), anti-mouse TCRγ/δ (APC, clone GL3, BioLegend), anti-mouse CD3 (PE/Cyanine7, clone 17A2, BioLegend), anti-mouse CD4 (FITC, clone GK1.5, BioLegend), anti-mouse CD8α (APC, clone 53–6.7, BioLegend), anti-mouse CD45R/B220 (Pacific Blue, clone RA3-6B2, BioLegend), anti-mouse CD19 (PerCP/Cyanine5.5, clone 6D5, BioLegend). The LIVE/DEAD™ Fixable Violet Stain kit (Invitrogen, Grand Island, NY, United States) was used to determine the viability of cells. For flow cytometry staining, spleens were collected on the appropriate days, and single-cell suspensions of splenocytes were prepared as described above. A 1 × 10⁶ splenocytes were first incubated with anti-CD16/CD32 antibodies (clone 93, BioLegend) to block Fc receptors and then incubated with the antibodies of interest for 45 min. The cells were analyzed on a FACSCanto II instrument (BD Biosciences, San Jose, CA, United States), and the data were analyzed with FlowJo software.

For CD4 and CD8⁺ T cell depletion assay, 200 μg anti-CD4 mAb (clone GK1.5, BioXcell), anti-mouse CD8β (clone 53–5.8, BioXcell) or the corresponding isotype control IgG was injected intraperitoneally into P. chabaudi-infected mice at days -1, 1, 3, 5, 7 or days 14, 16, 18, 20, 22 after the challenge. Depletion of B cells was accomplished by intraperitoneal injection of 500 μg of anti-CD19 mAb (clone 1D3, BioXCell) and 500 μg of anti-B220 mAb (clone RA3.3A1/6.1, BioXCell) or the corresponding isotype control IgG at 12, 14, 16, 18, and 20 days post-infection. CD4⁺ T cells, CD8⁺ T cells and B cells depletions were verified by staining blood samples with anti-mouse CD4 (APC, clone GK1.5, BioLegend), anti-mouse CD8α (FITC, clone 53–6.7, BioLegend), anti-mouse CD19 (PerCP/Cyanine5.5, clone 6D5, BioLegend), and anti-mouse B220 (Pacific Blue™, clone RA3-6B2, BioLegend). For phagocytes depletion assay, 200 μl of clodronate liposomes (Liposoma BV) or control liposomes were injected intravenously into P. chabaudi-infected mice at days 14, 17, 20 after infections.

The data were analyzed using GraphPad Prism version 8.0 (GraphPad Software, La Jolla, CA, United States). Data are shown as the means ± SD unless otherwise stated. Non-parametric tests (Mann-Whitney test) and two-way ANOVA were used to determine the statistical significance between groups. P-values < 0.05 were considered statistically significant.

To investigate the predominant immune effectors that control the recurrence of rodent malaria, the splenic lymphocyte populations at the peak of the recurrent stage (day 16 post-infection) were compared to those at the peak of the acute infection stage (day 8 post-infection) using scRNA-seq (Figure 1A). UMAP analysis classified the splenic lymphocytes into eight clusters based on the cell-specific molecular markers, including T cells (Cd3e), B cells (Cd79a, Cd79b), NK cells (Klra8), DCs (Flt3), RPMs (Adgre1), monocytes (Itgam), neutrophils (Ly6g), and platelets (Pf4) (Figures 1B,C). The main lymphocytes were the B and T cells. However, a cluster of red blood cells was found in the initial UMAP analysis (data not shown), which was due to insufficient lysis of red blood cells during splenocyte preparation. The clusters of red blood cells were removed in the subsequent analysis.

All eight clusters were found in splenocytes collected from either uninfected mice or P. chabaudi infected mice at day 8 or day 16 post-infection, but the frequency of several clusters fluctuated significantly. Compared to baseline (day 0), the number of T cells decreased at day 8 post-infection but returned to the baseline level at day 16 post-infection. However, the number of B cells was significantly increased in the acute phase, but then significantly declined in the recurrence phase. In contrast, innate immune cells, including RPMs, DCs, and neutrophils, significantly increased during the recurrence stage compared to the acute stage (Figures 1D,E). The change trends of the infiltrated lymphocytes were well confirmed by flow cytometry analysis of splenocytes from P. chabaudi-infected mice at day 8 and day 16 post-infection (Figures 1F,G). Therefore, our single-cell analysis revealed significant changes in both adaptive and innate immune cells between the acute phase and the recurrence phase of P. chabaudi infection.

It is well known that parasite specific CD4⁺ T cells are responsible for controlling blood-stage malaria infection (Meding and Langhorne, 1991). The acute infection of P. chabaudi drove the differentiation of Th0 into CD4⁺ Th1 cells and CD4⁺ Tfh cells (Lonnberg et al., 2017). Interferon-gamma (IFN-γ) secreted by Th1 cells then promoted macrophages to degrade phagocytosed parasites (Stephens et al., 2005), while CD4⁺ Tfh cells helped the B cells to generate antibodies (Perez-Mazliah et al., 2017). Additionally, acute malaria infection increased Treg cells, which reduce the CD4⁺ T cell response, to maintain immune hemostasis (Hisaeda et al., 2004; Nie et al., 2007; Haque et al., 2010). However, the dynamic change of CD4⁺ T cell responses in the whole process of P. chabaudi infection is still largely undefined.

T cell population was firstly classified into clusters of CD4⁺ T cells (Cd3e and Cd4) and other T cells, according to the specific markers (Figures 2A,B). The frequencies of CD4⁺ T cells were comparable at the three time points (Figure 2C). Then, the CD4⁺ T cells were further divided into Th1 (Cd3e, Tbx21, and Ifng), Tfh (Cd3e, Bcl6, and Cxcr5), Treg (Cd3e, Foxp3, and Il2ra), and other CD4⁺ T cells, based on their specific markers (Figures 2D,E). Th1 cells were greatly increased during the acute stage as compared to the control, but then significantly decreased in the recurrent phase. An increased number of Tfh cells were also observed during the acute P. chabaudi infection stage, but no significant reduction in Tfh cells was observed during the recurrence stage. As expected, the number of Treg cells was also significantly increased during the acute stage and declined during the recurrence stage (to a level much lower than the baseline) (Figures 2F,G). Differential gene expression analysis showed that the signals of T cell exhausting markers, such as CTLA-4 (Ctla4), LAG3 (Lag3), Tim3 (Harvcr2), and PD-1 (Pdcd1) were significantly downregulated, but the signal of the main effector of Th1 cells, IFN-γ (Ifng), was also significantly reduced in Th1 cells in the recurrence phase, as compared to the gene expression in the acute phase (Figure 2H). No significant change of T cell exhausting markers in Treg cells was observed between acute stage and recurrence phase of P. chabaudi infection (Figure 2H). To compare the different roles of CD4⁺ T cell responses in the control of acute infection and recurrence of P. chabaudi, CD4⁺ T cells were depleted during the acute stage and before the onset of recurrence, respectively. After CD4⁺ T cells were depleted during the acute phase (Supplementary Figure 1), the parasitemia of P. chabaudi was greatly increased during this phase, and then fluctuated at high levels. The mice could not clear the parasite, and succumbed to the infection. In contrast, mice could clear the parasites after CD4⁺ T cells were depleted during the recurrence stage, although the depletion of CD4⁺ T cells also significantly elevated the P. chabaudi parasitemia (Figure 2I). Our data thus illustrate the more important role of CD4⁺ T cell response in the control of acute infection than in the control of recurrent infection of P. chabaudi, which was closely associated with the frequency change of CD4⁺ T cell subsets.

With the help of Tfh cells, germinal center (GC) B cells were activated and then differentiated into memory B cells and long-lived plasma cells (LLPC) (Deenick and Ma, 2011). High-affinity antibodies were produced by the LLPC after the Ig type switch and destroyed malaria parasites through neutralization, activation of the complement system, mediation of antibody-dependent cellular cytotoxicity (ADCC) or opsonic phagocytosis mediated by complement fragments and antibodies (Chua et al., 2013; Julien and Wardemann, 2019). Using B cell-deficient mice (μMT) (Supplementary Figure 2), B cells were found to be necessary for the final elimination of the parasites but were not required for the control of early blood-stage malaria infection (von der Weid et al., 1994; Brugat et al., 2017). Using μMT mice in these studies, B cells were absent at the beginning of infection and thus their definitive role in the control of P. chabaudi recurrence was unknown.

Therefore, we further classified splenic B cells into GC B (IgD^– FAS⁺) and plasma cells according to their specific markers (Cd79a, Cd79b, Jchain, Sdc1, plasma cells; Cd79a, Cd79b, Fas, Ighd, GC B cells) (Figures 3A,B). GC B cells were significantly increased during the recurrence stage, but not during the acute stage. In contrast, the number of plasma cells greatly increased in the acute phase, but sharply declined to a level slightly higher than that of the baseline, in the recurrence phase (Figures 3C,D). Consistently, the expression levels of either Ighg2b (IgG2b), Ighg2c (IgG2c) or Ighg3 (IgG3) were significantly increased at the acute stage, but decreased in the recurrence phase (Supplementary Figure 3). Similar to previous studies, parasitemia in μMT mice was elevated at day 10 post-infection, and mice could not clear the parasite and succumbed to infection (Figure 3E). To determine the role B cells play in the control of recurrence of P. chabaudi, B cells were depleted with anti-CD220 and anti-CD19 (Supplementary Figure 1). However, no significant difference in parasitemia was found between mice depleted of B cells and mice with B cells (Figure 3F). Overall, B cells thus do not play a role in the control of the parasitemic recurrence of P. chabaudi, which was also supported by the declined plasma cells during the recurrence stage.

Recently, parasite specific CD8⁺ T cells were demonstrated to recognize and destroy reticulocytes infected with P. vivax and cooperate with macrophages to protect against blood-stage rodent malaria (Imai et al., 2015). However, the role of CD8⁺ T cells during P. chabaudi infection is still unknown. UMAP analysis showed that effector CD8⁺ T cells (Cd3e, Cd8a, Cd8b1, Gzma, Prf1) were increased in the acute phase and further increased in the recurrence phase (Figures 4A–G). Differential gene expression analysis also found that the signal of T cell exhausting markers, such as CTLA-4 (Ctla4) and Tim3 (Harvcr2), as well as granzyme B (Gzmb) and perforin (Prf1) were significantly downregulated in effector CD8⁺ T cells at the recurrence stage, as compared to the acute infection stage (Figure 4H). No significant effect on parasitemia was observed after CD8⁺ T cells were depleted during the acute or recurrence stage (Figure 4I). Thus, in contrast to its protective effect against Plasmodium yoelii (P. yoelii) infection, CD8⁺ T cells had no significant role in the control of either acute or recurrent P. chabaudi infection, which might be attributed to the exhaustion and dysfunction of CD8⁺ T cells during malaria infection (Horne-Debets et al., 2013).

Based on the results obtained for the role of CD4⁺ T and B cells in the control of parasite recurrence in this study, we wondered whether the control of P. chabaudi recurrence was dependent on an innate immune response. UMAP analysis showed that the frequency of innate immune cells, including γδT cells, RPMs, DCs, and neutrophils, were significantly higher in the recurrence phase than in the acute phase (Figures 1E, 5A–C).

As a high dose of clodronate liposome (CL) has been shown to deplete most phagocytic cells in spleen tissue (Fontana et al., 2016), the P. chabaudi infected mice were administered a high dose of CL at the onset of the recurrent infection stage. As a result, almost 100% of the RPMs (CD11b⁺ F4/80^high), 80% of the monocytes (CD11b⁺ F4/80^low), and approximately 30% of the DCs (CD11c⁺), were depleted after the injection of CL. However, no neutrophil depletion was observed (Figures 5D,E). The parasitemia in the recurrent phase sharply increased to a level higher than the peak of the acute phase, and all mice died after treatment with CL (Figure 5F). Recently, γδT cells were reported to kill malaria parasites through direct phagocytosis (Junqueira et al., 2021). However, phagocytosis by γδT cells might result in its depletion by high-dose CL treatment. In contrast, we found that high-dose CL treatment did not reduce the frequency of γδT cells in the mice spleen (Figures 5G,H), therefore excluding the possibility that γδT cell depletion plays a role in parasitemia in mice treated with CL. Thus, our data strongly suggest the critical role of innate immune responses in the control the P. chabaudi recurrence.