Single-cell transcriptomic atlas of the chicken cecum reveals cellular responses and state shifts during Eimeria tenella infection

Abstract

Eimeria tenella (E. tenella) infection is a major cause of coccidiosis in chickens, leading to significant economic losses in the poultry industry due to its impact on the cecum. This study presents a comprehensive single-cell atlas of the chicken cecal epithelium by generating 7,394 cells using 10X Genomics single-cell RNA sequencing (scRNA-seq). We identified 13 distinct cell types, including key immune and epithelial populations, and characterized their gene expression profiles and cell–cell communication networks. Integration of this single-cell data with bulk RNA-seq data from E. tenella-infected chickens revealed significant alterations in cell type composition and state, particularly a marked decrease in APOB⁺ enterocytes and an increase in cycling T cells during infection. Trajectory analysis of APOB⁺ enterocytes uncovered shifts toward cellular states associated with cell death and a reduction in those linked to mitochondrial and cytoplasmic protection when infected with E. tenella. These findings highlight the substantial impact of E. tenella on epithelial integrity and immune responses, emphasizing the parasite’s role in disrupting nutrient absorption and energy metabolism. Our single-cell atlas serves as a critical resource for understanding the cellular architecture of the chicken cecum and provides a valuable framework for future investigations into cecal diseases and metabolic functions, with potential applications in enhancing poultry health and productivity.

Introduction

Chicken coccidiosis is a prevalent parasitic protozoan disease affecting the intestines of chickens, leading to significant economic losses in the poultry industry [1]. The infection causes severe physiological damage, including intestinal inflammation, diarrhea, and bleeding [2]. Eimeria tenella (E. tenella) is a major coccidial pathogen that typically colonizes the cecum of chickens, disrupting the stability of cellular junctions in the cecal epithelium [3]. E. tenella initiates host cell invasion through attachment-related proteins, including calcium-binding proteins [4], and has been shown to induce apoptosis in host cells, compromising epithelial integrity [5]. Recent in vitro studies also indicate that E. tenella infection triggers autophagy in cecal epithelial cells [6]. Furthermore, coccidial infection impacts immune cells, particularly T cells, within the ceca [7]. Given the complexity of the intestinal epithelial barrier, which comprises various cell types in synergistic communication, constructing a single-cell atlas of the chicken cecum is essential for understanding the pathogenic mechanisms of coccidial infection.

Single-cell RNA sequencing (scRNA-seq) is a powerful technique for characterizing the cell types within a tissue based on transcriptomic profiles [8]. While RNA-seq from bulk tissue samples cannot disentangle the impact of cellular heterogeneity on gene expression, and large-scale scRNA-seq remains costly, integrated analysis of scRNA-seq and bulk RNA-seq data from the same tissue has become feasible, revealing changes in cell proportions and states [9]. By reducing dimensionality of gene expression profiles and identifying cell marker gene expression patterns, scRNA-seq enables the identification of cells with similar transcriptomic profiles [10]. In addition to differential gene expression detection, scRNA-seq facilitates cell communication network construction and trajectory analysis, and its integration with bulk RNA-seq data helps deconvolute cell proportions and states within tissues [11]. Facilitated by large-scale bulk RNA-seq samples, this deconvolution method, using scRNA-seq data derived from the same tumor tissue, has been successfully employed to investigate cellular heterogeneity and prognostic progression in numerous cancer studies [12,13,14]. Although scRNA-seq data have been effectively parsed in chickens, the application of this technology in broader animal research remains limited, despite the availability of bulk RNA-seq data and the significant economic importance of mechanistic studies in farm animals [15, 16].

In this study, we performed 10X Genomics scRNA-seq on 7,394 cells from two biological replicates to construct a comprehensive single-cell atlas of the chicken cecal epithelium. We then integrated this atlas with bulk RNA-seq data from a published E. tenella infection study to analyze the impact of E. tenella on cell type composition and cell state differences in the chicken cecal epithelium [17].

Materials and methods

Chicken cecum single-cell sample preparation and sequencing

Cecum samples were collected from one male and one female chicken of a Chinese local yellow-feathered breed at 90-day-old, which were euthanized by cervical dislocation. Both of chickens were confirmed to be free of E. tenella infection. All animal procedures were approved by the Animal Care and Use Ethics Committee of Hunan Agricultural University (Permit Number: CACAHU, 2021–00269). Cell viability was stringently assessed, and only samples with over 90% viable cells were processed further. Qualified cells were washed and re-suspended to achieve a concentration of 700 to 1,200 cells per microliter, meeting the requirements of the 10X Genomics Chromium™ system. cDNA libraries were prepared and subjected to PCR amplification with a calibrated cycle regimen to ensure optimal library representation. The amplified libraries were then fragmented in preparation for sequencing on the Illumina NovaSeq 6000 platform (Illumina, San Diego, USA). A targeted sequencing depth of over 50,000 read pairs per cell was sought to ensure comprehensive transcriptome coverage.

scRNA-seq data processing and analysis

Raw data were processed to create counting matrices using unique molecular identifiers (UMIs) for accurate transcript quantification with Cellranger software (version 7.1.0). Using R packages Seurat (version 4.4.0), the expression matrix was normalized to obtain count values. Cells were filtered to exclude those with fewer than 200 or more than 4,000 gene expressions, as well as those with a mitochondrial gene expression percentage higher than 10%. Only genes expressed in more than three cells were retained for further analysis. To refine the data further, we identified the top 2,000 highly variable genes per cell for downstream analysis. Principal component analysis (PCA) was employed to reduce dimensionality, using the top 14 principal components (Supplementary Fig. 1). Cell clustering was performed at a resolution of 0.5 (Supplementary Fig. 2), which facilitated the identification of distinct cell populations. These clusters were visualized in two dimensions using Uniform Manifold Approximation and Projection (UMAP), providing a graphical representation of cell type heterogeneity and cell trajectories.

Cell type annotation

Differentially expressed genes (DEGs) were identified using the R package Seurat, which compared gene expression between cells within a specific cluster and those in all other clusters [18]. Genes expressed in at least 25% of the cells within a cluster were considered for DEG analysis. Marker genes were defined as those with high expression levels in a target cell type and distinct expression across different cell types. We manually annotated cell types based on marker genes with established and published evidence, using public databases and literature. For clusters with limited marker gene evidence, we annotated cells based on the biological functions of marker genes or transcriptomic similarity to known cell types.

RNA-seq data processing and analysis

Transcriptome data (GSE160169) were retrieved from the Gene Expression Omnibus (GEO) database, available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE160169. A total of 36 Dekalb White Leghorn-type hens were obtained from a commercial hatchery. The study set an uninfected control group and 5 experimental groups orally inoculated with 1,000 live E. tenella oocysts at 17 days of age, with 6 replicates in each group. At 5 various days post-infection (dpi): 1 dpi, 2 dpi, 3 dpi, 4 dpi, and 10 dpi, the cecum samples of 6 infected chickens and 1 or 2 uninfected chickens were collected [17]. The RNA-seq data underwent quality control using fastp (version 0.20.1) to remove sequencing adapters and low-quality reads [19]. Clean reads were then aligned to the chicken reference genome (GRCg7b) using HISAT2 (version 2.2.1) [20]. Gene-level read counts were obtained with featureCounts from the Subread package (version 1.6.3) and normalized to Transcripts Per Million (TPM) to facilitate subsequent analyses [21].

Differential gene expression analysis

Differential gene expression analysis was performed using the DESeq2 package (version 1.36.0) [22]. A linear regression model was also applied with DESeq2 to account for the effect of cell proportion on expression levels [23]. Genes were considered significantly differentially expressed if they had an adjusted P value < 0.05 and a |log₂ (Fold Change)|> 2 [24]. The biological functions of the differentially expressed genes (DEGs) were annotated and enriched in Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using the clusterProfiler package (version 4.7.1.3) in R [25].

Cell differentiation and developmental trajectory analysis

CytoTRACE was used to assess and profile the differentiation potential of each cecal cell type [26]. By comparing differentiation indexes across different cell types, we inferred their developmental potential and relative states of differentiation. The developmental trajectory of APOB⁺ enterocytes was analyzed using the pseudotime algorithm in the Monocle3 package (version 1.3.4) to understand the sequential molecular changes underlying cellular differentiation [27].

Cell–cell communication analysis

Orthologous genes between human and chicken were identified using BabelGene (version 22.9). These genes were then used for cell–cell communication analysis with the R package CellChat, and annotations were based on information from CellChatDB [28].

Single-cell RNAseq data simulation and deconvolution

We utilized the CIBERSORT algorithm for deconvolution of bulk transcriptome data to estimate the relative cellular composition [29]. To validate this approach, we first created pseudo-bulk samples from our single-cell RNA-seq data using the R package SimBu [30]. Specifically, 100 pseudo-bulk RNA-seq samples with predefined cellular proportions were simulated, and the accuracy of the deconvolution was rigorously assessed by computing the Spearman correlation coefficient between the deconvoluted and true cell percentages. Additionally, to determine the abundance of APOB⁺ enterocytes in different cellular states, we integrated the R package MeDuSA with the pseudotime trajectory data derived from Monocle3 analysis [31].

Statistical analysis

Statistical analyses, including Analysis of Variance (ANOVA) and multiple testing corrections, were performed using R software (version 4.2.0) [32]. The inverse normal transformation was applied to normalize cell proportion data. To evaluate cell state abundances, a permutation F-test was used to determine statistical significance [33]. A threshold of P < 0.05 was set for determining statistical significance.

Results

A cell atlas of chicken cecal epithelium

The procedure for single-cell RNA-seq analysis in this study is summarized in Fig. 1 and Supplementary Fig. 3. Single-cell transcriptomic data were generated from the ceca of one female and one male chicken. Following quality control, a total of 7,394 cells were obtained. Harmony was employed to correct for sex and batch effects (Supplementary Fig. 4). The UMAP method was used to visualize the transcriptomic similarity of each cell in two dimensions. In total, 13 distinct clusters were identified, and each cluster was recognized as one cell type (Fig. 2A). These 13 cell types were manually annotated based on their marker genes (Supplementary Table 1). The identified cell types in the chicken cecum were categorized primarily into two groups: immune cells and epithelial cells (Fig. 2B, Supplementary Table 1).

Fig. 1

Brief graphical workflow of this study

Fig. 2

Single-Cell Transcriptomic Atlas of the Chicken Cecum. A UMAP visualization of single-cell profiles, with each dot representing a cell, colored according to the 13 major identified cell types. B Heatmap displaying the expression levels of selected marker genes across all identified cell types in the cecum. C Heatmap showing significantly enriched biological process terms for the top 20 specific gene signatures within the 13 major cell types, with Z-score expression levels indicated for each annotated cell type. Only terms with a p-value < 0.05 are displayed. D Prediction of cell states across the 13 cell types, with cells exhibiting higher cell cycling indices identified as more proliferative. The horizontal line in the boxplots corresponds to the median, while the box bounds indicate the 25th and 75th percentiles, and whiskers represent 1.5 times the interquartile range. E, F Cell communication networks among the 13 major cell types in the cecum, where the edge width represents the communication probability, and thicker edge lines indicate stronger signaling interactions

Immune cells

The cecal immune cell types included B cells (EBF1, PAX5), dendritic cells (CD74, IFI30, NAAA), T naïve cells (CD3E, TCF7, LEF1), cycling T cells (CD3E, SMC2, TOP2A, HMGN4), T cells (CD3E, GNLY, IL20RA, PLCL1), and specific T cell subpopulations such as GZMA⁺ T cells (GZMA, GNLY), RORA⁺ T cells (CD3D, RORA), and CCL20⁺ T cells (CD3E, CCL20, PTH2R, KCNQ1).

Epithelial cells

The cecal epithelial cell types were annotated as FABP1/2⁺ enterocytes (APOA1, FABP1, FABP2), APOB⁺ enterocytes (APOA1, APOB), goblet cells (MUC2, FER1L6, MAP2), and tuft cells (TRPM5, DCLK1, EHF). A small number of erythrocytes (HBBA, HBA1, HBAD) were also identified in the chicken cecum.

GO enrichment analysis was performed on the top 20 cell type-specific genes for each cell type. The results showed that mature T cells were enriched in immune-related pathways, whereas T naïve cells were enriched in both immune-related pathways and pathways associated with cell differentiation and cell cycle. Intestinal epithelial cells were enriched in pathways related to carbohydrate synthesis and digestion metabolism, while erythrocytes were associated with oxygen metabolism processes (Fig. 2C).

Overall, cecal epithelial cells exhibited a greater capacity of cell differentiation than that of immune cells (Fig. 2D). The cell–cell communication network and strength between cecal epithelial cells and immune cells were also characterized by identifying quantifying the ligand-receptor activity between cell types. Our results revealed that extensive cell–cell communications were detected among all cecal cell types except for erythrocyte and T naïve cells (Fig. 2E, F).

Cell proportion and cell stage changes of APOB ⁺ enterocytes in response to E. tenella infection

We deconvolved bulk RNA-seq transcriptome data from 36 cecal samples obtained from a published study on E. tenella infection. To validate the accuracy of our deconvolution method, we first generated simulated single-cell RNA-seq data using the SimBu software [30]. The results demonstrated a strong correlation (Spearman’s r² = 0.83 ~ 0.99, Supplementary Fig. 5) between the true cell proportions and those estimated from the pseudo-bulk RNA-seq data (Fig. 3A), supporting the feasibility and accuracy of our method and single-cell RNA-seq atlas in deconvoluting bulk RNA-seq data from chicken cecum.

Fig. 3

Decreased Abundance of APOB⁺ Enterocytes Following E. tenella Infection. A Correlation between the simulated cell proportions and the deconvolution results across 100 simulation samples for APOB⁺ enterocytes, dendritic cells, and cycling T cells. C Heatmap displaying gene expression trajectories, categorized into three distinct stages: stage_1, stage_2, and stage_3. D Top: Estimation of APOB⁺ enterocyte cell state abundance using MeDuSA, based on a dataset of 36 RNA-seq samples. P values were calculated using the permutation-based MANOVA-Pro method. Bottom: Heatmap showing significantly enriched biological process terms for the top 100 specific gene signatures across the three stages of APOB⁺ enterocytes, with Z-score expression levels indicated for each stage

We estimated the relative proportions of 13 different cell types from the 36 bulk RNA-seq samples by integrating the cecal single-cell RNA-seq data. Among these, APOB⁺ enterocytes, cycling T cells, and dendritic cells were the most abundant, with higher relative proportions (median > 0.1). Significant heterogeneity in cell proportions was observed among different groups; however, most cell types did not show substantial perturbations before day 3 post-infection (dpi) compared to the uninfected group (Fig. 3B). Notably, the relative proportion of APOB⁺ enterocytes decreased significantly (~ 22%) from 3 to 4 dpi, with a further decrease from 4 to 10 dpi (~ 19%). Conversely, the proportion of cycling T cells increased significantly by 8% and 16%, respectively, from 3 to 4 dpi and from 4 to 10 dpi. The proportion of dendritic cells first increased significantly (~ 10%) from 3 to 4 dpi, then remained stable from 4 to 10 dpi (Fig. 3B).

To further characterize the changes in cell states of APOB⁺ enterocytes, pseudotime analysis was performed to reveal the cell trajectory of these cells (Fig. 3C). According to the development of the pseudotime axis, three different stages of APOB⁺ enterocytes were clearly separated and defined as state_1, state_2, and state_3, respectively (Fig. 3C). MeDuSA was used to assess the abundances of APOB⁺ enterocytes in these three states across RNA-seq samples. Given the substantial changes in the cell percentage of APOB⁺ enterocytes at 3 dpi and 10 dpi compared to the uninfected group, we focused on these two time points. Our analysis showed no significant changes in the abundance of APOB⁺ enterocytes at 3 dpi across the three stages compared to the uninfected group. However, at 10 dpi, the relative abundance of APOB⁺ enterocytes was not significantly changed (P = 0.099) for stage_1, but significantly increased (P < 0.05) for stage_2, and significantly decreased (P < 0.05) for stage_3 compared to the uninfected group (Fig. 3D).

To understand the potential functions of the three stages of APOB⁺ enterocytes, we conducted differential gene expression analysis and enriched DEGs in KEGG pathways (Supplementary Table 2). Genes associated with stage_1 were primarily enriched in cell adhesion pathways (e.g., adherens junction and focal adhesion) and cell growth pathways (e.g., GnRH signaling and Wnt signaling). Stage_2 APOB⁺ enterocytes also expressed genes related to cell junction and growth pathways but were additionally enriched in processes related to cell death (e.g., efferocytosis and cellular senescence). Stage_3 APOB⁺ enterocytes exhibited enrichment in mitochondrial functions (e.g., oxidative phosphorylation, glutathione metabolism, and cytochrome P450-related pathways) and cytoplasmic functions (e.g., peroxisome, glycolysis/gluconeogenesis, and ribosome metabolism) (Fig. 3D).

Cellular communication shifts under E. tenella challenge

To investigate how E. tenella infection affects interactions between different cell types in the chicken cecal epithelium, we constructed a cell communication profile based on KEGG pathways identified from bulk RNA-seq data. Since traditional bulk RNA-seq data may overlook the influence of cell type mixture in the tissue, we performed differential expression analysis using two different regression models: one considering cell fraction and one without (Supplementary Table 3). Our results demonstrated that incorporating cell fraction in the regression model significantly improved DEG detection. Specifically, the cell-fraction-fitted model identified 177 new DEGs, while the unfitted model detected none when comparing gene expression profiles between the uninfected group and the 1 dpi group. Overall, 6,861 additional DEGs were detected using the cell-fraction-fitted model compared to the unfitted model (Fig. 4A).

Fig. 4

E. tenella Infection Induces Changes in Cellular Communication and Cellular Signaling. A Venn diagram showing the overlap of differentially expressed genes (DEGs) when cellular proportions are considered as a covariate, before and after adjustment. B Significantly enriched KEGG pathway terms identified in the analysis. C Frequency of overrepresented signaling pathways in cellular communication. D Inferred signaling pathway networks for ncWNT, CCL, RANKL, and PARs among the major cell types. In the chord diagram, edge width represents communication probability, with thicker lines indicating stronger signaling. E Bar chart illustrating the expression levels of ligand and receptor genes within the four signaling pathways. Lowercase letters indicate significant differences among the six groups, determined by one-way ANOVA (p-value < 0.05)

The DEGs identified by the cell-fraction-fitted regression model were enriched in several KEGG pathways. Notably, the WNT signaling pathway (KEGG: hsa04310), cytokine-cytokine receptor interaction (KEGG: hsa04060), and neuroactive ligand-receptor interaction (KEGG: hsa04080) were significant (P < 0.05) and were also observed in ligand-receptor interactions by CellChat analysis from scRNA-seq data (Fig. 4B and C). We examined the cell communication networks for these KEGG pathways, focusing on ligand-receptor interactions between different cecal cell types. In the ncWNT pathway, tuft cells were the sole source of ligands for both APOB⁺ enterocytes and tuft cells themselves. In the CCL pathway, both RORA⁺ T cells and CCL20⁺ T cells both sent and received ligands. For the RANKL pathway, RORA⁺ T cells and CCL20⁺ T cells sent ligands unidirectionally to goblet cells, APOB⁺ enterocytes, and dendritic cells, respectively. In the PARs pathway, GZMA⁺ T cells, cycling T cells, and T cells sent ligands to goblet cells, APOB⁺ enterocytes, tuft cells, and dendritic cells (Fig. 4D).

Analysis of ligand and receptor expression from RNA-seq data revealed significant changes in expression levels at different infection stages. In the ncWNT pathway, ligand expression increased significantly, while receptor expression decreased at 10 dpi compared to the uninfected group. In the CCL pathway, ligand expression significantly increased and receptor expression decreased at 3 dpi and 4 dpi, but the trend reversed at 10 dpi with decreased ligand expression and increased receptor expression. For the RANKL pathway, only ligand expression was significantly increased at 10 dpi compared to the uninfected group. In the PARs pathway, ligand expression significantly increased at 1, 2, 3, and 4 dpi, with a sharp increase at 10 dpi, while receptor expression remained unchanged (Fig. 4E).

Discussion

Chickens, with an estimated global population of around 20 billion, play a crucial role in global agriculture and the economy [34]. They differ from mammals in several unique aspects, including nucleated erythrocytes, a muscular proventriculus, a reverse sex chromosome system (ZZ for males and ZW for females), and the presence of the bursa of Fabricius. Additionally, chickens have a distinctive digestive system characterized by double-developed ceca, which are crucial for nutrient digestion and host diverse microbial communities [35]. The cecum is a key site for various pathogenic infections, including bacterial pathogens like Campylobacter and coccidial parasites such as Eimeria [36,37,38]. One possible reason these pathogens target chicken ceca is the unique microenvironment, which not only nourishes commensal microbes but also supports the propagation of disease-associated bacteria and parasites [39]. Despite its importance, the cell type composition of the chicken cecal epithelium has not been well-characterized.

In this study, we utilized single-cell RNA-seq (scRNA-seq) to create a cell atlas of the chicken cecal epithelium, identifying 13 distinct cell types categorized into immune and epithelial cells. Among the epithelial cells, we identified two types of enterocytes: FABP1/2⁺ enterocytes and APOB⁺ enterocytes. FABP1/2⁺ enterocytes are involved in general nutrient metabolism, while APOB⁺ enterocytes are implicated in nitrogen metabolism, essential for the recycling of urea and other nitrogenous compounds in the cecum [40, 41]. This distinction highlights the specialized functions of the cecal epithelium in nutrient processing and microbial interaction. The cell communication network constructed in our study revealed complex interactions within the cecal epithelium. Goblet cells and tuft cells were identified as special cell types with relatively strong incoming and outgoing interactions, potentially indicating their role as intermediaries in signaling between different cell types. Most immune cells, such as cycling T cells and T cells, exhibited stronger outgoing interactions but weaker incoming interactions, which may reflect their roles in maintaining intestinal homeostasis and modulating inflammatory responses [42].

E. tenella, a coccidial parasite, primarily infects the cecum, causing epithelial damage and digestive dysfunction in chickens [43]. E. tenella infection disrupts microbiota balance, competes with the host for nutrient absorption, and even destroys the integrity of the epithelial villus structure in chicken ceca [44]. However, the specific cell type in the cecal epithelium most affected by E. tenella has not been explicitly identified. Understanding the transcriptomic profile or cell trajectory of each cell type via scRNA-seq allows for the deconvolution of bulk RNA-seq data into cell proportions or cell states [45]. Our study deconvoluted bulk RNA-seq data from 36 cecal samples obtained from E. tenella infection studies to estimate cell proportions and states. Results showed that APOB⁺ enterocytes were the predominant cell type in the chicken cecum, with the highest percentage, but sharply decreased upon E. tenella infection, aligning with previous findings that E. tenella preferentially targets enterocytes [46]. Additionally, we noted a marked increase in cycling T cells, consistent with observations of T cell aggregation and immune activation in response to E. tenella infection [7].

Intestinal epithelial cells typically exist in a dynamic state, transitioning from proliferation and growth to death. The cell trajectory analysis of APOB⁺ enterocytes revealed three distinct stages along the pseudotime axis, each associated with specific cellular functions. The stage associated with cell death increased significantly at 10 dpi, while the stage related to cell stress decreased, reflecting the impact of E. tenella on epithelial cell function. This is consistent with previous reports of epithelial degradation and apoptosis following E. tenella infection [47]. The reduction in APOB⁺ enterocytes associated with mitochondrial metabolism and nutrient synthesis further underscores the detrimental effect of E. tenella on nutrient utilization and energy production [48].

Traditional DEG analysis from bulk RNA-seq data often fails to account for cell type heterogeneity. By incorporating cell fraction information into our regression models, we significantly improved DEG detection [23]. This approach allowed us to identify 6,861 additional DEGs compared to models that did not account for cell fraction. Notably, the KEGG pathways enriched by DEGs identified using the cell-fraction-fitted model, such as Wnt signaling, RANKL, and PARs, were consistent with pathways identified in our scRNA-seq data. Wnt signaling, which is critical for maintaining intestinal stem cells and regulating cell fate decisions, was particularly affected by E. tenella infection. We observed increased expression of the WNT5B ligand at 10 dpi, suggesting that E. tenella infection activates Wnt signaling pathways through enhanced ligand expression. This finding aligns with studies demonstrating the role of Wnt signaling in intestinal development and homeostasis [49, 50]. RANKL and PARs pathways, involved in inflammation and apoptosis, also showed significant changes, correlating with the observed decline in APOB⁺ enterocytes [51, 52]. Our study highlights the importance of integrating single-cell and bulk transcriptomic data to understand cellular responses to infection. Future studies should include single-cell transcriptomic data from E. tenella-infected cecal samples to validate and further explore these findings.

Conclusions

This study provides a comprehensive single-cell atlas of the chicken cecal epithelium, revealing significant insights into the cellular landscape and its response to E. tenella infection. We identified 13 distinct cell types, including critical immune and epithelial cell populations, and characterized their gene expression profiles and communication networks. The infection by E. tenella led to notable changes in cecal cell composition, particularly a significant reduction in APOB⁺ enterocytes and an increase in cycling T cells, reflecting the parasite's impact on epithelial integrity and immune activation.

Our cell trajectory analysis of APOB⁺ enterocytes highlighted a shift in cellular states, with an accumulation of cells associated with cell death functions and a reduction in those linked to mitochondrial and cytoplasmic protection when the birds were infected by E. tenella. These findings underscore the detrimental effects of E. tenella on nutrient absorption and energy metabolism within the cecum. Furthermore, by integrating single-cell and bulk transcriptomic data, we demonstrated the utility of combining these approaches to uncover the complexities of host–pathogen interactions at the cellular level.

The single-cell atlas developed in this study serves as a critical resource for understanding the cellular architecture of the chicken cecum. It also provides a valuable framework for future research into the pathophysiology of cecal diseases and the metabolic functions of the cecum, with potential applications in improving poultry health and productivity.