Female C57BL/6 mice aged 6–8 weeks were obtained from the Laboratory Animal Center of Ningxia Medical University and maintained under specific pathogen-free (SPF) conditions (22 ± 2 °C, 12-h light/dark cycle) with ad libitum access to food and water. Two independent experimental cohorts were generated using a completely randomized design. The first cohort (n = 3 per group) was used for single-cell RNA sequencing and included an uninfected PBS-injected control group (baseline control), a 3-day post-infection group (3 dpi), and a 3-month post-infection group (3 mpi), representing acute and chronic stages of AE, respectively. The second cohort (n = 5 per group) followed the same grouping strategy and was used for flow cytometry and histological analyses.

Protoscoleces (PSCs) were isolated from laboratory-maintained E. multilocularis-infected gerbils. Harvested cyst material was minced and homogenized in sterile phosphate-buffered saline (PBS), and the homogenate was passed through a 180 μm cell strainer. The filtrate was washed three times by gravity sedimentation (10 min each, at room temperature), during which the supernatant was carefully discarded and the pellet was retained, in 40 mL sterile PBS to remove tissue debris. The recovered PSCs were then digested in 1% pepsin solution (pH 2.0) at a 1:10 (PSC suspension:pepsin, v/v) ratio for 20 min at 37 °C with gentle agitation on a rocking platform. After digestion, the suspension was washed again by gravity sedimentation (10 min, room temperature; supernatant discarded and pellet retained) in 20 mL sterile PBS and passed through a 70 μm cell strainer to remove calcareous corpuscles. PSC viability was evaluated by eosin exclusion and confirmed to be >90%. For experimental infection, mice were inoculated via the portal vein with 1,000 viable PSCs resuspended in 100 μL PBS; uninfected controls received an equal volume (100 μL) PBS via the same route. All procedures were approved by the Institutional Animal Care and Use Committee of Ningxia Medical University (IACUC-2025105).

At the designated time points, mice were deeply anesthetized via isoflurane inhalation and subsequently euthanized by cervical dislocation. Following euthanasia, the liver and spleen were promptly removed and weighed to calculate organ indices (organ weight/body weight). A portion of the right hepatic lobe was fixed in 4% paraformaldehyde for histological analysis. The remaining liver tissue was minced and digested using the gentleMACS™ Liver Dissociation Kit (Miltenyi Biotec, Germany) on a DSC-400 tissue dissociator, which combines standardized enzymatic and mechanical dissociation. The resulting cell suspension was passed through a 70 µm cell strainer and further processed with the Debris Removal Solution (Miltenyi Biotec, Germany) by density gradient centrifugation at 4 °C to remove cell debris and non-cellular contaminants. After debris removal, a mixed single-cell suspension containing immune and non-immune hepatic cell populations was obtained. Cell viability was assessed by trypan blue exclusion and exceeded 85% for all samples used for scRNA-seq.

Single-cell suspensions were prepared from liver tissues of mice in the first cohort (n = 3 per group). Single-cell capture, barcoding, and library preparation were performed using the MobiCube^® High-Throughput Single-Cell 3′ Transcriptome Kit (v2.0) and the MobiNova^®-100 system (Mojomics, China). Briefly, single-cell suspensions, barcoded beads, and reagents were loaded onto microfluidic chips, where water-in-oil droplets were generated to achieve single-cell partitioning and mRNA capture. After demulsification, reverse transcription was carried out to generate cDNA containing cell barcodes and unique molecular identifiers (UMIs). The cDNA was then pre-amplified and purified using SPRIselect magnetic beads (Beckman Coulter). Quantification and quality assessment were performed using Qubit 4.0 and the Agilent 2100 Bioanalyzer, ensuring a fragment size distribution of 0.5–2 kb. Qualified samples underwent enzymatic fragmentation, end repair, adaptor ligation, and indexed PCR to construct sequencing libraries. Final libraries were sequenced on an Illumina NovaSeq 6000 (paired-end 150 bp) with a target depth of approximately 50,000 reads per cell. Detailed scRNA-seq quality-control metrics are provided in Supplementary Data 1.

Raw FASTQ files were processed using the MobiVision software suite for adapter trimming, sample demultiplexing, barcode and UMI identification, and alignment to the mouse reference genome (mm10), resulting in gene-by-cell expression matrices. Across qualified libraries, the primary processing pipeline identified an estimated total of 78,855 cell barcodes. One 3-mpi library yielded only 565 estimated cells and did not meet sequencing quality requirements; therefore, downstream analyses included samples from the control (n = 3), 3-dpi (n = 3), and 3-mpi (n = 2) groups. Subsequent analyses were performed in R (v4.2) using the Seurat package (v4.0). Cells with <200 or >6,000 detected genes or >10% mitochondrial transcripts were removed. Data were normalized using the LogNormalize method, followed by principal component analysis (PCA) and nonlinear dimensionality reduction with UMAP and t-SNE for clustering. Cell clusters were annotated manually based on canonical marker genes. Differentially expressed genes (DEGs) were identified using the FindMarkers function in Seurat with thresholds of |log₂FC| > 0.25 and adjusted p < 0.05. Genes displayed in violin plots were selected from DEGs additionally filtered using the following thresholds: |log₂FC| > 0.25, p_adj < 0.05, pct.1 > 0.25, and pct.2 > 0.25. Enriched Gene Ontology (GO) terms and KEGG pathways were analyzed using the clusterProfiler package to explore immunological functions and signaling pathways associated with AE infection. Full DEG lists, GO/KEGG enrichment results, and cell–cell interaction analysis tables are provided in Supplementary Data 2.

Fixed liver tissues were embedded in paraffin and sectioned at 4 µm. After deparaffinization and rehydration, sections were stained with hematoxylin and eosin (H&E) and scanned using an Olympus VS200 digital slide scanner. Pathological features, including inflammatory infiltration, necrosis, and fibrosis, were assessed according to the criteria in Supplementary Table 1.

Blood samples were allowed to clot at room temperature and then centrifuged at 3,000 rpm for 10 min; serum ALT, AST, ALP, total bilirubin (TBIL), total protein (TP), and albumin (ALB) were measured using a Chemray 240 automated analyzer (Rayto, China).

Single-cell suspensions were prepared from liver tissues as described in Section 2.2, and lymphocytes were enriched by 30%/70% Percoll density-gradient centrifugation(450 × g, 20 min, room temperature; acceleration 1, deceleration 0). Cells were stained with Fixable Viability Stain 510 (BD Horizon, 564406) to exclude dead cells. After Fc receptor blockade with anti-mouse CD16/32 (eBioscience; cat. no. C001T03F01), 1×10⁶ cells per sample for liver and 2×10⁶ cells per sample for spleen were incubated with fluorochrome-conjugated antibody panels targeting immune cell lineage markers and functional markers. Antibody information and dilution ratios are provided in Supplementary Table 2. Surface staining was performed for 30 min at 4 °C in the dark. For phenotypic analysis, cells were stained without ex vivo stimulation. For intracellular cytokine analysis, cells were stimulated in vitro with protoscolex crude antigen (15 μg/mL) plus anti-CD28 (BioLegend, San Diego, CA, USA; 1 μg/mL) for a total of 18 h at 37 °C and 5% CO₂. BD GolgiPlug™ Protein Transport Inhibitor (BD Biosciences, San Jose, CA, USA; 1 μL/100 μL) was added during the final 6 h to inhibit protein transport and promote intracellular cytokine accumulation. PMA/ionomycin cell stimulation cocktail (YaMei Biotechnology, Shanghai, China; 2 μL/100 μL) was included as a positive control. After stimulation, cells were harvested and stained for surface lineage markers, followed by fixation for 10 min and permeabilization according to the manufacturer’s instructions. Intracellular staining was then performed for 30 min at 4 °C using the Foxp3/Transcription Factor Staining Buffer Set (BD Biosciences, San Jose, CA, USA). Samples were acquired on a Cytek Aurora full-spectrum flow cytometer. Unstained, single-stained, and fluorescence-minus-one (FMO) controls were included for compensation and gating. Data were analyzed using FlowJo software (v10.8). Gating was performed on singlets and live CD45⁺ cells, followed by lineage-specific markers and functional readouts. Phenotype definitions, including positive and negative markers for each population, are summarized in Supplementary Table 3.

Paraffin-embedded liver sections were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval using EDTA buffer (pH 9.0). Endogenous peroxidase activity was quenched with 3% H₂O₂, followed by multiplex fluorescence staining using the AlphaXTSA^® multispectral immunohistochemistry kit (Aikefa Biotechnology, Beijing, China). Sections were sequentially incubated with the primary antibodies anti-CD8α (Abcam, ab217344, 1:2000), anti-Granzyme B (Abcam, ab317458, 1:1000), and anti-CD11c (CST, #97585, 1:150), each applied overnight at 4 °C. Signal amplification and fluorophore development were performed according to the manufacturer’s protocol. Nuclei were counterstained with DAPI, and images were acquired under consistent exposure settings using a Zeiss Axio Observer 7 inverted fluorescence microscope.

All statistical analyses were performed using GraphPad Prism (v9.0). Data are presented as mean ± SEM. For comparisons of continuous outcomes (e.g., flow-cytometry measurements and serum biochemical parameters) among multiple groups, ordinary one-way ANOVA was used, followed by Tukey’s multiple comparisons test. For scRNA-seq–derived proportional data, each mouse was treated as one biological replicate (one value per mouse). Because group sizes were unequal (3 mpi, n = 2 due to exclusion of one library), Brown–Forsythe and Welch ANOVA tests were applied, followed by Games–Howell’s multiple comparisons test. Statistical significance was defined as p < 0.05. Significance levels are indicated as p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), and p < 0.0001 (****); ns, not significant.

To model the temporal course of human AE infection, mice were injected via the portal vein with viable protoscoleces and evaluated across early (3 dpi) and late-stage (3 mpi) infection (Figure 1A). At 3 mpi, abdominal ultrasound revealed multiple hypoechoic nodules within the liver parenchyma (Figure 1B), consistent with space-occupying lesions. Gross inspection further demonstrated distinct stage-specific changes: scattered punctate necrotic foci were visible on the liver surface at 3 dpi, whereas by 3 mpi the liver was markedly enlarged with extensive infiltrative lesions (Figure 1C). The liver index was significantly increased in 3 mpi mice compared with both the uninfected baseline control and 3 dpi groups (p < 0.01), while a mild, non-significant increase in spleen index was observed, body weight was also elevated at 3 mpi (Figure 1D). These findings indicate successful establishment of the AE infection model and demonstrate progressive hepatic involvement.

Histopathological assessment further delineated the chronological pattern of liver injury (Figure 1E). Livers from uninfected baseline control mice exhibited intact lobular architecture. At 3 dpi, acute inflammatory responses were evident around the implantation sites, characterized by dense neutrophil and mononuclear-cell infiltration accompanied by focal hepatocellular necrosis. By 3 mpi, the lesions had evolved into granuloma-like structures with a necrotic core containing parasite remnants, surrounded by epithelioid cells, plasma cells, and lymphocytes, and encapsulated by fibrotic tissue. Normal hepatic architecture was largely replaced by these infiltrative lesions. Semi-quantitative scoring confirmed significantly increased liver injury scores in both 3 dpi and 3 mpi groups relative to uninfected baseline controls (p < 0.05), with the highest scores in 3 mpi mice (p < 0.01 vs. 3 dpi; Figure 1F), reflecting progressive tissue damage throughout infection.

Serum biochemical measurements were consistent with the histological alterations (Figure 1G). At 3 mpi, levels of ALT, AST, and ALP were significantly elevated, indicating hepatocellular injury and cholestasis. Mild fluctuations in TP, ALB, and globulin were also detected. In contrast, biochemical markers at 3 dpi remained largely comparable to controls, except for a slight increase in total protein, suggesting that early lesions had not yet progressed to systemic biochemical abnormalities.

Together, these results demonstrate that AE infection induces a continuous progression from acute localized inflammation to extensive chronic hepatic pathology, accompanied by increasing tissue injury and systemic immune activation.

To characterize the dynamic changes in the hepatic immune microenvironment during AE infection, single-cell RNA sequencing was performed on liver tissues from uninfected baseline control, 3 dpi, and 3 mpi mice (Figure 2A). Following primary processing and integration with batch correction, gene-by-cell expression matrices comprising a total of 78,290 cells were obtained, with an average of 1,323 detected genes per cell. Thirteen major cell populations were identified, including T cells, B cells, DCs, macrophages, Kupffer cells, neutrophils, NKT cells, plasma cells, and non-immune populations such as hepatic stellate cells, fibroblasts, hepatocytes, and cholangiocytes. Integrated UMAP visualization revealed marked stage-dependent shifts in cell distribution (Figure 2B), and canonical marker expression confirmed accurate annotation (Figure 2C).

Focusing on immune compartment dynamics, proportional analysis demonstrated a pronounced temporal reorganization of immune composition during AE infection (Figure 2D). In uninfected baseline control mice, macrophages (37%) and B cells (33%) constituted the dominant immune subsets. At 3 dpi, the immune profile shifted toward an acute inflammatory response, characterized by increased macrophages (48%) and reduced B cells (29%) and T cells (6%), indicating reliance on innate immunity at early stages. By 3 mpi, the immune landscape underwent a profound transition, with substantial expansion of T cells (28%) and DCs (32%), accompanied by sharp declines in macrophages (24%) and B cells (1%). Minor fluctuations were also observed in other innate and adaptive subsets.

Given the marked expansion of T cells in the chronic phase and their functional importance, this subset was examined in greater detail. Quantitative scRNA-seq analysis showed a significant increase in T-cell abundance at 3 mpi compared with both uninfected baseline control and 3 dpi groups (Figure 2E), suggesting a potential central role for T cells in chronic-stage immune regulation. This pattern was independently validated by flow cytometry, which similarly revealed reduced T-cell proportions at 3 dpi and a significant rebound at 3 mpi (Figures 2F, G). Proportional changes in other major immune subsets were likewise confirmed by flow cytometry (Supplementary Figure S1), supporting the robustness of the single-cell dataset.

In summary, integrated single-cell transcriptomic and flow cytometric analyses demonstrated a stage-specific reorganization of the hepatic immune landscape during AE infection. The acute phase was characterized by macrophage-dominated inflammatory activity and a transient reduction in adaptive immune cells. In contrast, during the chronic phase, DCs and T cells became the predominant immune populations, indicating a shift from inflammation-driven innate defense toward antigen presentation and T cell–mediated adaptive immunity.

To delineate the dynamic evolution of T-cell states during AE infection, we performed high-resolution clustering of liver-derived T cells. UMAP analysis identified 13 transcriptionally distinct subsets (Figure 3A), including CD4⁺ T-cell lineages (Th1, Th2, Th17, Tfh, Treg, naïve, and memory) and CD8⁺ T-cell lineages consisting of cytotoxic CD8⁺ T cells (CTL), effector-memory CD8⁺ T cells (Tem), exhausted CD8⁺ T cells (Tex), and naïve CD8⁺ T cells, together with a DNT population. Canonical marker expression supported the annotation of each subset (Figure 3B).

The distribution of T-cell subsets across groups is visualized in Figure 3C, and their relative enrichment is quantified using the Ro/e (ratio of observed to expected) index in Figure 3D. At 3 dpi, Th17 (1.54) and Treg cells (1.93) were enriched, reflecting simultaneous induction of pro-inflammatory and regulatory programs. The overall proportion of CD8⁺ T cells remained comparable to controls at this stage. By 3 mpi, however, the T-cell landscape had undergone marked restructuring. CD4⁺ T-cell responses shifted toward a Th2-skewed profile (1.42), whereas CD8⁺ T cells became the predominant adaptive population. Significant enrichment was observed in CTL (1.41), Tem (1.51), and Tex (1.78), indicating the emergence of a diversified cytotoxic–memory–exhaustion configuration during chronic infection. Functional pathway scoring further demonstrated pronounced changes in CD8⁺ T-cell programs at 3 mpi (Figure 3E). CD8⁺ T cells exhibited increased enrichment of TCR-signaling, activation, inflammatory, cytotoxic, effector-memory, and exhaustion signatures, reflecting the simultaneous upregulation of effector and inhibitory transcriptional modules under sustained antigen exposure.

Flow cytometry provided quantitative validation of CD8⁺ T-cell subset dynamics (Figure 3F). CTL frequencies showed a mild reduction at 3 dpi and higher values at 3 mpi. Tem frequencies were lower at 3 dpi and increased at 3 mpi, whereas Tex frequencies were elevated predominantly at 3 mpi. These flow cytometry-based frequency changes in CTL, Tem, and Tex populations were consistent with the enrichment patterns observed in the scRNA-seq-based Ro/e analysis.

Together, these data demonstrate stage-dependent remodeling of CTL, Tem, and Tex subsets during AE infection, with chronic infection characterized by coordinated changes across cytotoxic, memory-associated, and exhaustion-associated CD8⁺ T-cell programs.

To dissect the molecular processes underlying CD8⁺ T-cell adaptation across AE infection, we performed differential expression and functional enrichment analyses for the CTL, Tem, and Tex subsets. Detailed GO and KEGG enrichment tables with the contributing genes (geneID) for each term are provided in Supplementary Data 1.

During early infection (Ctrl vs. 3 dpi), CTLs exhibited robust induction of chemotaxis- and cytoskeleton-associated genes (Ccl3, Ly6c2, Tmsb4x), together with upregulation of regulatory and metabolic markers such as Lgals1 and Apoe, indicating rapid recruitment and activation. A subset of signaling-related genes including Inpp4b and Atp8a2 showed early downregulation (Supplementary Figure S2A). By the chronic stage (Ctrl vs. 3 mpi), CTLs acquired a distinct transcriptional profile characterized by strong induction of feedback-regulatory transcription factors and NF-κB modulators, including Junb, Jund, Btg1, Tnfaip3, Bhlhe40, and H3f3b, accompanied by reduced expression of inflammatory mediators such as S100a9 (Supplementary Figure S2B). Enrichment analysis indicated significant activation of pathways associated with immune regulation, apoptosis-related signaling, and IL-17/TNF-modulated inflammatory circuits (Supplementary Figure S2C). Longitudinal comparison (3 dpi vs. 3 mpi) further demonstrated a coherent temporal trajectory: early-induced genes such as Apoe, S100a9, and Lyz2 declined over time, whereas chronic-stage regulators Btg1 and Tnfaip3 were progressively upregulated (Figures 4A, B). GO analysis highlighted sustained involvement of antigen-presentation, efferocytosis, and TNF/IL-17 axis pathways (Figure 4C). Together, these patterns suggest that CTLs transition from early migratory and cytotoxic priming toward a more immune-regulatory, feedback-controlled state during chronic infection.

Early after infection (Ctrl vs. 3 dpi), Tem cells exhibited modest activation, including upregulation of B2m and Arap2, accompanied by reduced expression of Tmem163 and Pag1 (Supplementary Figure S2D). These features suggest initiation of memory-associated signaling but without large-scale effector activation. At 3 mpi, Tem cells displayed a transcriptional program enriched for negative feedback and immune-regulatory molecules, including Tnfaip3, Btg1, Nfkbia, and Junb, together with reduced expression of Frrs1, Fut8, and Sh3bgrl3 (Supplementary Figure S2E). KEGG analysis revealed enrichment of TNF-associated and regulatory pathways (Supplementary Figure S2F). Temporal comparison (3 dpi vs. 3 mpi) revealed downregulation of a set of early-elevated genes (Cyb5a, Lyz2, Smarcad1, Cept1, Ifitbl1) and increased enrichment of chromatin-regulatory, stress-response, and cytolytic processes (Figures 4D–F). These data indicate that Tem cells acquire a low-inflammatory, persistence-oriented transcriptional state during chronic infection, marked by reduced early-phase metabolic activation and strengthened transcriptional control.

In the Tex subset, early infection (Ctrl vs. 3 dpi) induced upregulation of genes involved in cellular stress response and activation (Slc39a11, Plac8, Ly6a) (Supplementary Figure S2G). By 3 mpi, Tex cells expressed higher levels of chronic-stimulation markers including Ccl4, Junb, Btg1, and Nr4a1, and exhibited reduced expression of inflammatory mediators such as S100a8/9 and Tmem260 (Figures 4G, I). Enrichment analysis highlighted chemotaxis-related and cytoskeletal-remodeling pathways (Supplementary Figure S2H). Longitudinal comparison (3 dpi vs. 3 mpi) revealed progressive upregulation of inhibitory and feedback-regulatory modules (Btg1, Junb) along with downregulation of early-induced inflammatory genes, and enrichment of pathways related to cell–matrix interactions, cytoskeletal regulation, and adaptive immune signaling (Figures 4H–J). These features suggest that Tex cells in chronic AE maintain a hybrid state combining inhibitory signals with persistent activation modules rather than undergoing uniform functional collapse.

In summary, chronic AE infection induces coordinated transcriptional remodeling across CTL, Tem, and Tex subsets. Each subset displays stage-dependent changes involving effector-associated, memory-associated, and inhibitory gene programs. Rather than a unidirectional transition, chronic infection establishes a heterogeneous CD8⁺ T-cell landscape in which cytotoxic, memory, and exhaustion signatures are concurrently enhanced under prolonged antigen exposure.

Multiplex immunofluorescence imaging revealed a consistent spatial organization of immune cells within AE lesions (Figure 5A). CD11c⁺ dendritic cells formed dense, circumferential layers surrounding the parasitic structures, while CD8⁺ T cells—including cytotoxic and effector-memory subsets—were distributed immediately adjacent to the DC-rich regions. This arrangement placed CD8⁺ T cells in direct proximity to DCs across both early and chronic stages of infection.

To complement the spatial observations, ligand–receptor prediction based on scRNA-seq data was performed to infer putative DC–T communication within lesions, with a particular focus on cytotoxic CD8⁺ T cells (Figure 5B; Supplementary Data 1). Among the predicted pairs linking DCs and cytotoxic CD8⁺ T cells, Thy1–Adgre5 showed a prominent interaction signal and met the more stringent significance tier indicated in the plot (p < 0.01). In addition, several adhesion- and trafficking-related axes—such as Selplg–Sell, Icam1–Itgal, and integrin-associated interactions (Itgav+Itgb1/Itgb3–Adgre5)—were also detected, supporting potential mechanisms for DC–CD8 proximity and contact within the lesion niche. Although the prediction framework included CD4⁺ Th1/Th2/Treg subsets as contextual references, the Thy1–Adgre5 axis represented a prominent inferred interaction involving cytotoxic CD8⁺ T cells and DCs (Figure 5B).

Integrating spatial and molecular analyses, CD11c⁺ DCs and cytotoxic CD8⁺ T cells exhibited close anatomical localization together with inferred ligand–receptor connectivity, suggesting a spatially adjacent and transcriptionally coupled DC–CD8 interaction landscape within AE lesions.