All animal protocols were approved by the Institutional Animal Care and Use Committee of Guangzhou Medical University (S2024-282) and conducted in accordance with institutional guidelines. Every effort was made to minimize animal suffering.

Female C57BL/6 wild-type mice (6–8 weeks old) were purchased from Guangdong Zhiyuan Biomedical Technology Co., Ltd. (Guangzhou) and housed under specific pathogen-free conditions at the Animal Center of Guangzhou Medical University.

P. yoelii NSM was obtained from the Malaria Research and Reference Reagent Resource Center (MR4). To ensure the vitality of P. yoelii and the stability of the experimental results, frozen P. yoelii was thawed at 37 °C and resuscitated, and 100 μL was injected intraperitoneally into naïve mice. Tail vein blood smears were prepared daily from days 2 to 5 post-infection and stained with Giemsa (Solarbio, Beijing, China), and parasitemia was quantified. When parasitemia reached 10%–20%, mice were euthanized, and infected RBCs (iRBCs) were counted from aseptically collected ocular blood. Blood was diluted to 10⁷ iRBCs/mL in sterile Phosphate Buffered Saline (PBS), and 100 μL was injected intraperitoneally into each experimental mouse.

Roswell Park Memorial Institute (RPMI) 1640, Dulbecco's Modified Eagle Medium (DMEM), fetal bovine serum (FBS), collagenase IV, penicillin, and streptomycin were purchased from Gibco (Invitrogen, Waltham, MA). DNase I was from Solarbio, and RBC lysis buffer (Cat: C3702, 500 mL) was from Beyotime Biotech, Nantong, Jiangsu Province, China. Ampicillin (Cat: MCR003) was obtained from Dingguo Biotech, Beijing, China. Fluorescein-labeled anti-mouse antibodies were sourced from eBiosciences, San Diego, California, BioLegend, San Diego, CA, and Cell Signaling Technology, Danvers, Massachusetts. A detailed list of antibodies used in this study is provided in Supplementary Table 1.

Mice were euthanized 12–16 days post-infection. Livers and spleens from naïve and infected mice were fixed in 10% formalin for 48 hours, paraffin-embedded, sectioned, stained with hematoxylin and eosin (H&E), and examined microscopically.

Following euthanasia by 2% isoflurane, peripheral blood was collected in Ethylenediaminetetraacetic acid (EDTA)-containing tubes and diluted 1:1 with PBS. Peripheral blood mononuclear cells (PBMCs) were isolated using Mouse Lymphocyte Isolation Solution (Dakewe, Shenzhen, China) via density gradient centrifugation. Livers, spleens, lungs, tibias, and femurs were harvested. Liver cell suspensions were prepared using the Miltenyi Biotec Liver Isolation Kit, Teterow, Germany, and lymphocytes were isolated via density gradient centrifugation. Lung tissue was minced and digested with 2.4 mg/mL collagenase IV and 0.2 mg/mL DNase I at 37°C for 30 minutes, followed by grinding and filtration. Splenocytes were obtained by grinding spleens using a sterile syringe plunger and filtering through a 100-μm mesh. Bone marrow (BM) cells were flushed from tibias and femurs with RPMI 1640. RBCs in liver, lung, spleen, and BM cell suspensions were lysed using RBC lysis buffer. Isolated cells were washed twice with Hank's Balanced Salt Solution and resuspended in RPMI 1640 containing 10% FBS for subsequent experiments.

When parasitemia reached 20%, peripheral blood was centrifuged to remove serum, resuspended in an equal volume of naïve saline, and subjected to 60%/50% Percoll gradient centrifugation. The iRBC layer was collected, washed to remove Percoll, and resuspended in normal saline, and parasitemia was counted microscopically.

For surface marker detection, single lymphocytes were stained with fluorescence-labeled antibodies against FVD, CD45, CD3, CD19, CD4, CD8, NK1.1, γδT, CD103, CD69, PD-1, CD62L, ICOS, TIGIT, and CD107a at 4°C for 30 minutes. Stained cells were analyzed using flow cytometry (FCM), with data processed using the CytExpert 1.1 software (Beckman Coulter, Brea, CA).

For cytokine (granzyme B, perforin, and IFN-γ) detection, lymphocytes were stimulated with 20 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma) and 1 μg/mL ionomycin (Sigma, St Louis, MO) at 37°C for 1 hour, followed by the addition of 10 μg/mL brefeldin A (Sigma) to halt cytokine secretion. Cells were stained for surface markers, fixed with 4% paraformaldehyde, permeabilized, and incubated with fluorescence-labeled antibodies for 30 minutes. Stained cells were analyzed using FCM, with data processed using the CytExpert 1.1 software (Beckman Coulter).

For nuclear protein detection, cells were treated with the Invitrogen Foxp3 Transcription Factor Staining Kit and stained with fluorescent antibodies against LEF1, Foxp3, and Ki67 before FCM analysis.

Splenocytes from C57BL/6 mice were resuspended in cold Dulbecco's Phosphate Buffered Saline (DPBS) (0.1% Bovine Serum Albumin (BSA)) at 5 × 10⁶ cells/mL and incubated with 2.5 μM Carboxyfluorescein Succinimidyl Ester (CFSE) for 5 minutes in the dark with agitation. FBS (1/9 volume) was added to quench the reaction, and cells were washed three times with DPBS. The concentration of labeled splenic lymphocytes was adjusted to 2 × 10⁶ cells/mL by RPMI 1640 containing 10% heat-inactivated FBS and stimulated by anti-mouse CD3 (1 μg/mL) plus anti-mouse CD28 (1 μg/mL) for 72 hours in a 37°C CO₂ incubator. Cells were collected and washed twice with PBS and then stained with anti-mouse CD4-PerCP5.5, anti-mouse CD8-APC-cy7, anti-mouse CD103-APC, and FVD-PB450 for 30 minutes at 4°C in the dark. Stained cells were washed twice with PBS and then analyzed using FCM, with data processed using the CytExpert 1.1 software (Beckman Coulter).

Human embryonic kidney (HEK) 293T cells were cultured in DMEM with 10% heat-inactivated FBS at 37°C in a 5% CO₂ incubator.

Lymphocytes were resuspended in sorting buffer and centrifuged, and the supernatant was discarded. Cells were incubated with magnetic bead-conjugated antibodies in sorting buffer at 4°C for 30 minutes. Separation columns were pre-rinsed with sorting buffer, and antibody-labeled cells were loaded onto columns, followed by two washes with sorting buffer. Target cells were eluted after removing the magnetic field.

Cells were lysed with TRIzol (Invitrogen, Waltham, MA) for RNA extraction. Quantitative PCR (qPCR) was performed using a CFX96 System (Bio-Rad, Hercules, CA). Primer sequences are listed in Supplementary Table 2. β-actin is the housekeeping gene.

Splenocytes from naïve and infected mice (three each) were obtained according to the protocol mentioned above. Cells from the same group were pooled and mixed, and CD45⁺ cells were sorted using Fluorescence-Activated Cell Sorting (FACS) (Beckman MoFlo). After the viability detection and cell count, the RNA expression profile in each cell was detected using the 10x Genomics Chromium Single Cell 3’s platform in Yuanxin Biotechnology Company (Guangzhou, China). The single cell was barcoded and underwent reverse transcription in oil droplets for the preparation of the cDNA library by the Chromium Single Cell 3′Library and Gel Bead Kit v3. Libraries were sequenced on eight lanes of Illumina NovaSeq 6000 using 150-bp paired-end reads. All data processing and analysis were performed in YUANXIN Biotechnology Co., Ltd, Guangzhou, China. Single-cell RNA sequencing (scRNA-seq) data are available at NCBI under accession numbers SRR22462456 and SRR22476069. Cell Ranger (version 3.1.0) was used to align reads on the GRCm38 reference genome for mouse and generated unique molecular identifier gene expression profiles for every single cell under a standard sequencing quality threshold (Li et al., 2023). The cells were initially divided into 18 clusters by auto-annotation and then manually adjusted to 12 clusters according to the shape of the cell group and the classic gene it transcribes. The heatmap of the top 10 genes in different clusters is shown in Supplementary Figure 1, and the numbers of cells detected in each population are listed in Supplementary Table 3 (<0.2% RBC was mixed in the course of cell sorting and formed an independent cluster). Loupe Brower 6 was used to compare the gene transcription in different cell populations.

PGL3-basic plasmids containing the Itgae promoter and luciferase, pCDNA3.1(+) plasmids expressing Lef1, and empty vectors were constructed by Tsing Ke Bio-Tech (Beijing, China). Plasmid sequences were verified by enzymatic digestion and sequencing. Prof. Ming-Sheng Cai kindly offered the pRL-TK plasmid.

HEK 293T cells were seeded in 24-well plates at 1 × 10⁵ cells/well and cultured for 24 hours. When confluency reached 70%–80%, cells were transfected with plasmids using Lipofectamine 3000 (Invitrogen, Cat: L3000015, Waltham, MA) for 48 hours. Cells were lysed, and luciferase activity was measured using the Dual-Luciferase Reporter Assay System.

Comparisons between two groups were performed using unpaired two-tailed Student’s t-tests. Multiple comparisons were analyzed using one-way analysis of variance (ANOVA) followed by Dunnett’s or Tukey’s post-hoc tests. Non-parametric data were analyzed using the Kruskal–Wallis or Mann–Whitney U tests with Dunn’s post-hoc test. A p-value < 0.05 was considered statistically significant.

To investigate the distribution of the integrin CD103 in lymphocytes, splenocytes from naïve C57BL/6 mice were analyzed using flow cytometry and single-cell RNA sequencing. Based on the gating strategy (Supplementary Figure S2), varying degrees of integrin CD103 expression were observed in splenic lymphocytes of naïve mice via FACS (Figure 1A). While a CD103⁺ subset was also detected in γδT cells, CD8⁺CD103⁺ T cells were more prominent, accounting for approximately 65.2% of CD8⁺ T cells (Figure 1B). At the same time, splenocytes were isolated from naïve mice, and FACS was performed to isolate CD45⁺ cells. scRNA-seq was performed to detect the expression of the Itgae gene (which encodes CD103) in different cell populations (Figure 1C). The results indicated that the transcription of the Itgae gene was primarily concentrated in the CD8⁺ T-cell population (Figure 1C). This implied that CD8⁺ T cells were the dominant subset within the CD103⁺ cell population. Additionally, lymphocytes isolated from the peripheral blood, spleen, lung, liver, and bone marrow of naïve C57BL/6 mice were analyzed using flow cytometry. CD103⁺ cells were gated first, and the expression of CD8a was counted (Figure 1D). As shown in Figure 1E, the results showed that approximately 80% of CD103-expressing cells in the spleen of naïve mice were CD8⁺ T cells, and a similar percentage was found in the lung, which was higher than that in the peripheral blood, liver, and bone marrow. Altogether, these results indicated that the CD103 molecule was mainly expressed on CD8⁺ T cells in different tissues from naïve mice.

P. yoelii NSM infection could destroy RBC and induce inflammatory changes in the spleen of C57BL/6 mice, especially on 12–16 days post-infection (dpi) (Supplementary Figure S3). To explore the change of CD103 expression in CD8⁺ T cells in the course of P. yoelii NSM infection, splenic lymphocyte suspensions were prepared from both naïve and infected mice. RNA was extracted and reverse-transcribed to cDNA, and Itgae transcription was detected using qPCR. As shown in Figure 2A, Itgae transcription in the spleens of infected mice was significantly decreased (p < 0.05). At the same time, lymphocytes were stained with different fluorescence-labeled antibodies to detect the expression of CD103 on CD45⁺ T cells, too. Flow cytometry analysis confirmed a reduction in the percentage of CD103⁺ cells within the CD45⁺ lymphocyte population in infected mice (p < 0.05, Figures 2B, C). Further examination revealed that the proportion of CD103 on CD8⁺ T cells continued to decrease from 4 to 16 dpi, and then the percentage increased from 16 to 24 dpi (Figure 2D). A significant reduction in the rate of CD103-expressing CD8⁺ T cells was also observed in the blood, liver, lung, and bone marrow 12–16 days after infection (p < 0.05, Supplementary Figure S4A). Longitudinal analysis of CD103 expression on blood CD8⁺ T cells supported this result, too (p < 0.05, Supplementary Figure S4B). Single-cell RNA sequencing of FACS-sorted splenic CD45⁺ cells was used to detect the distribution of Itgae in different cell populations. The results showed that Itgae transcription was notable in CD8⁺ T cells from naïve mice. However, The level of IItgae transcription in CD8+ T cells from infected mice was decreased compared to naive mice (Figure 2E). These results indicated that P. yoelii NSM infection could reduce CD103 expression on splenic CD8⁺ T cells in mice.

To characterize the phenotypes of CD103⁺CD8⁺ T cells before and after infection, single splenocytes were stained with fluorescent antibodies against surface markers. In naïve mice, CD103⁻CD8⁺ T cells and CD103⁺CD8⁺ T cells showed no significant differences in the expression of CD69, ICOS, CD62L, or TIGIT (p > 0.05). However, after infection, CD103⁻CD8⁺ T cells were activated, with upregulated expression of CD69, ICOS, PD-1, and TIGIT (p < 0.05). While CD103⁺ cells also showed increased CD69 and ICOS expression post-infection, the changes were less pronounced than in CD103⁻ cells. These results indicate that infection enhances the activation of CD103⁻CD8⁺ T cells, whereas CD103⁺CD8⁺ T cells retain more resting characteristics (Figures 3A, B).

Moreover, flow cytometry analysis of effector and resting subsets in CD103^+/−CD8⁺ T cells post-infection (Figure 3C) revealed that CD103⁻CD8⁺ T cells contained a higher proportion of CD44^hiCD62L^low effector subsets, while CD103⁺CD8⁺ T cells were predominantly in the CD44^lowCD62L^hi resting subset (p < 0.05).

Comparison of cell proliferation capacity by flow cytometry showed a higher percentage of Ki67⁺ cells in CD103⁻CD8⁺ T cells from infected mice (p < 0.05), indicating stronger proliferative ability compared to CD103⁺CD8⁺ T cells (p < 0.05, Figure 3D). In vitro stimulation of CFSE-stained naïve splenocytes with CD3 and CD28 mAbs was performed for 3 days; cells were washed twice and stained with different fluorescent-labeled antibodies for mouse CD4, CD8, CD103, and FVD. FACS results showed that CD103⁻CD8⁺ T-cell proliferation peaks were concentrated at 4× and 3×, while those of CD103⁺ cells were at 2× and 1×, suggesting delayed proliferation in CD103⁺CD8⁺ T cells (Figure 3E).

Additionally, flow cytometry detection of function-associated molecules showed increased percentages of CD107a, IFN-γ, granzyme B, and perforin-expressing cells in CD8⁺ T cells from infected mice (p < 0.05). However, most of these cytotoxic molecules were expressed in CD103⁻CD8⁺ T cells rather than CD103⁺CD8⁺ T cells, particularly in infected mice (Figures 3F, G). These results suggested that CD103⁻CD8⁺ T cells, rather than CD103⁺CD8⁺ T cells, contribute to the antimalarial immune response.

Next, the RNA transcription differences in CD8⁺ T cells from naïve and infected mice were compared, both with and without the Itgae gene transcribed, using scRNA-seq. The transcription of function-, activation-, suppression-, and proliferation-related genes between Itgae⁻CD8⁺ and Itgae⁺CD8⁺ T cells, in both naïve and infected, was compared (Figures 4A, B). The results showed that most pro-inflammatory markers, such as granzyme, Lamp1, Ifng, Prf1, Ccl5, Emoes, and Id2 transcription, increased after infection. However, compared to that in CD103⁺CD8⁺ T cells, the gene transcription is more active in CD103⁻CD8⁺ T cells. At the same time, the transcription of genes related to naïve or memory T cells, such as Sell, Il7r, Tcf7, and Lef1, decreased after infection. Compared to Itgae⁺CD8⁺ T cells, the reduction of this gene transcription in Itgae⁻CD8⁺ T cells in infected mice is more obvious. These results are consistent with the report that Itgae⁻CD8⁺ T cells may be a population of effector cells. In contrast, reduced function, as observed using FACS, suggests that Itgae⁻CD8⁺ T cells could represent a population of effector cells with diminished functionality.

Moreover, the number of differentially expressed genes (DEGs) between the Itgae⁺CD8⁺ and Itgae⁻CD8⁺ T cells in naïve and infected mice was compared. A total of 62 DEGs (47 downregulated and 15 upregulated) were found in naïve mice (Figure 4C), while 1,050 DEGs (719 downregulated and 331 upregulated) were found in infected mice (Figure 4D). These results indicated that noticeable differences exist in gene transcription between Itgae⁺CD8⁺ and Itgae⁻CD8⁺ T cells, especially in the infected mice. It supports the suggestion that the CD103⁺CD8⁺ T cell population is different from the CD103⁻CD8⁺ T cells.

Furthermore, DEGs between Itgae⁺CD8⁺ T cells and Itgae⁻CD8⁺ T cells from both naïve mice and 12-dpi mice were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses; the enriched gene pathways are shown in Supplementary Figures S5A–D. “Interferon-γ production” and “T-cell differentiation” pathways were enriched between Itgae⁺CD8⁺ T cells and Itgae⁻CD8⁺ T cells from infected mice. DEGs in these two pathways were listed, and the related contents were compared. As shown in Figure 4E, most of the interferon-γ production-related genes, such as Cd160, Pglyrp1, Il18r1, Il18rap, Eomes, Xcl1, and Klrk1, were downregulated in Itgae⁺CD8⁺ T cells compared to Itgae⁻CD8⁺ T cells. It is consistent with the FACS results, which showed that most of the IFN-γ expression was secreted by CD103⁻CD8⁺ T cells (Figure 3F). Similar results were found in DEGs in “T-cell differentiation”, “cytokine–cytokine receptor interaction”, “cytokine binding”, “leukocyte cell–cell adhesion”, and “regulation of immune effector process” pathway (Supplementary Figure S5E).

At the same time, Gene Set Enrichment Analysis (GSEA) of DEGs in the infected group showed high scores for “cytoplasmic translation”, “ribonucleoprotein complex biogenesis”, “mRNA binding”, and “ncRNA metabolic process” (Figure 4F), indicating active biological processes in Itgae⁺CD8⁺ T cells. Conversely, the chemokine signaling pathway and cytokine–cytokine receptor interaction gene sets showed negative enrichment (Supplementary Figure S5F). Altogether, these results indicated that Itgae⁺CD8⁺ T cells may be in a pre-mobilization state despite exhibiting some practical function.

To identify genes regulating CD103, DEGs between CD103⁺CD8⁺ and CD103⁻CD8⁺ T cells from naïve and infected mice were intersected, and a Venn diagram was generated (Figure 5A). Lymphocyte enhancer-binding factor 1 (Lef1, encoded by Lef1) showed significant differences in both groups. Lef1, a high-mobility group (HMG) family transcription factor, is a homolog of T-cell factor 1 (TCF1, encoded by Tcf7). Transcription factor analysis of sequencing data revealed the differential transcription of Tcf1 and Lef1 between the two groups (Figure 5B). Although Lef1 transcription decreased after infection, it remained higher in Itgae⁺CD8⁺ T cells than in Itgae⁻CD8⁺ T cells (Figure 5C). Flow cytometry confirmed high LEF1 expression on CD103⁺CD8⁺ T cells in both naïve and infected states (Figure 5D).

To verify Lef1-mediated regulation of Itgae transcription, potential Lef1 and Tcf1 binding sites on the CD103 promoter were predicted using JASPAR (Figure 5E). Lef1 was predicted to have three binding sites in the region from −2,000 bp upstream to 100 bp downstream of the CD103 transcription start site, while Tcf1 had only one. Dual-luciferase reporter assays were performed by co-transfecting 293T cells with a pGL3 plasmid containing the CD103 promoter and luciferase gene, a pCDNA3.1(+) plasmid expressing Lef1 (or empty vector), and the pRL-TK plasmid (Figure 5F). Lef1 transfection strongly induced CD103 promoter activity compared to the empty vector. Dose-dependent co-transfection of Lef1-expressing plasmids with the CD103 promoter luciferase reporter vector in 293T cells confirmed the specific activation of CD103 promoter activity by Lef1 (Figure 5G).