Bacterial serine protease EspP induces differentiation and expansion of enteroendocrine cells in human colonoids

Abstract

EspP is a potent serine protease that functions as a cytotoxin and is secreted by enterohemorrhagic Escherichia coli (EHEC). This bacterial toxin can cause significant colonic epithelial damage and mimic the pathophysiology of EHEC infection on human colonic organoids. However, transcriptomic changes within the intestinal epithelial cells and impacts on cell proliferation and differentiation are largely unknown. Here, using human colonic organoids, we defined the molecular and cellular changes that occur in response to acute EspP exposure, independent of microbiota, stroma, and immune cells. We performed droplet-based single cell RNA-sequencing to determine changes in epithelial cell lineages in response to EspP and demonstrated that there is a disruption in proliferation and a preferential differentiation of enteroendocrine cells (EECs). Higher resolution clustering revealed that the de novo EECs were primarily enterochromaffin or EEC progenitors. Interestingly, EECs induced from EspP treatment displayed an immune profile, expressing cytokines and chemokines associated with macrophage or neutrophil recruitment. These findings demonstrate that EspP, a potent EHEC cytotoxin, stimulates transcriptomic changes previously attributed to the whole bug infection and highlights the significance of crypts-based proliferative cells and secretory cell differentiation as major colonic responses to bacterial pathogens.

NEW & NOTEWORTHY

EspP, a bacterial serine protease secreted by enterohemorrhagic E. coli, can mimic the intestinal pathophysiology of EHEC in the absence of the bacteria. Using single cell RNA-sequencing, this study identified enteroendocrine cells as early responders to EspP treatment, with preferential hyperplasia of enterochromaffin cells. EspP-induced enteroendocrine cells also expressed immune cytokines and chemokines. This highlights the importance of enteroendocrine cell responses to bacterial pathogens and intestinal pathophysiology during infection.

Graphical Abstract

INTRODUCTION

An estimated 1 in 6 Americans, or roughly 48 million people, develop foodborne illness each year. In 2024, more individuals became ill from contaminated food than in the previous year, and the number of hospitalizations and deaths doubled. Enterohemorrhagic Escherichia coli (EHEC), a strain of Shiga toxin (Stx) producing E. coli, is the primary foodborne enteric pathogen with approximately 350,000 illnesses in 2019 in the United States (1). Approximately 10% of cases will progress to hemolytic uremic syndrome (HUS), a serious condition which can lead to kidney failure, stroke, seizure, and/or death. 3–5% of these cases will prove fatal, as common treatments for bacterial infections, such as antibiotics, promote HUS development (2). Consequently, treatment for EHEC is palliative, with no approved treatment plan able to prevent bacterial colonization, progression into hemorrhagic colitis or HUS (3).

Before progressing to peripheral organs, EHEC pathophysiology first causes injuries to the intestinal epithelia, which progresses from isolated epithelial apoptosis to denuded intestinal patches (4, 5). However, severe disruptions to the colonic epithelia do not occur until approximately 3–5 days after initial infection (6). The molecular and cellular events that occur within the colonic epithelium during this early asymptomatic period remain poorly defined. Importantly, the absence of clinical symptoms does not reflect an absence of disease activity. The asymptomatic period of disease represents a stage in which EHEC establishes colonization, produces virulence factors, and initiates host responses that ultimately dictate disease progression. These early events are likely altering the epithelial state that contributes to subsequent tissue damage and inflammation. Defining the molecular mechanisms that govern these early host-pathogen interactions is essential for understanding EHEC pathogenesis and identifying novel therapeutic targets for a disease in which no targeted therapies currently exist.

While much of the morbidity associated with EHEC infection is attributed to Stx, it alone is insufficient to cause significant damage or alterations to the intestinal epithelia (7). A distinct characteristic of EHEC infections is the secretion of effector proteins and toxins that interrupt cellular processes within the host. Increasing evidence suggests that early epithelial interactions with additional bacterial virulence factors shape disease trajectory before overt toxin-mediated injury occurs (8).

One of the earliest EHEC secreted toxins is EspP (extracellular serine protease, plasmid-encoded) (9), an essential serine protease autotransporter that disrupts the intestinal barrier, ion transport, and aids in Stx uptake into intestinal epithelial cells (7, 10). Additionally, antibody response against EspP has been found during EHEC infection (11), and the presence of EspP correlates with patients suffering from severe HUS or bloody diarrhea (12). However, relatively little is known about how EspP alters epithelial cell state during the earliest stages of exposure to facilitate EHEC pathogenesis and disease.

The intestinal epithelium is a highly dynamic and heterogenous tissue composed of proliferative stem cells, transit-amplifying cells, and differentiated absorptive and secretory lineages (13). Continuous epithelial turnover is essential for maintaining barrier integrity and host defenses. In the context of enteric infection, specialized cell types such as enteroendocrine cells (EECs) serve as key mediators of gut-immune communication. EECs are rare hormone producing cells compromising approximately 1% of the intestinal epithelium (14, 15). Despite their low abundance, they comprise the largest endocrine organ and help regulate epithelial regeneration, barrier permeability, immune cell recruitment, and microbial interactions (15). Alterations to number or EEC subtype composition have been observed in inflammatory and infectious contexts, suggesting that changes to the secretory lineage contribute to mucosal adaptation during stress. In the colon, EECs primarily respond to luminal toxicants and microbial products (16). However, due to their scarcity, EEC specific responses to bacterial virulence factors are difficult to resolve in whole animal models.

Human intestinal organoids are a tractable, epithelia-only model that recapitulates the cellular diversity and regional identity of the intestine while retaining the functional stem cell population (17), making them a valuable model for epithelia-pathogen interactions. Organoids generate differentiated absorptive and secretory lineages, including multiple EEC subtypes, and respond to stimuli in a physiologically relevant manner (18). Recent studies have shown that intestinal organoids are a physiologically relevant model to study EEC function (19–21).

In this study, we used human colonic organoids (colonoids) and single-cell RNA sequencing (scRNA-seq) to characterize the epithelial response to EspP. From the scRNA-seq analysis, we found that EspP disrupts normal differentiation by depleting proliferative cells and inducing differentiation towards secretory lineages. Colonic EECs expand and further differentiate to sub-types not present under control conditions. We also found that EspP alters the epithelial-derived immune signals, increasing expression of chemokines that function as neutrophil or macrophage chemoattractants.

MATERIALS AND METHODS

Tissue Collection and Colonoid Generation

Human colonoid studies were reviewed and approved by the University of New Mexico institutional review board (IRB #18–171, 18–626). Colonic biopsies from ascending, transverse, and descending colonic segments were obtained from healthy individuals undergoing screening colonoscopies who had given informed written consent. All colonoids were generated from colonic biopsies from patients undergoing routine, screening colonoscopy and with no pathophysiology. Colonic crypt isolation and colonoid generation were prepared as previously reported (22, 23). Briefly, biopsy tissue was minced, washed several times in freshly prepared cold chelating solution (CCS; 5.6mM Na2HPO4, 8mM KH2PO4, 96.2mM NaCl, 1.6mM KCl, 43.4mM sucrose, 54.9mM D-sorbitol, and 0.5mM DL-dithiothreitol) and incubated 1 hour at 4°C in 10 mM EDTA in CCS on an orbital shaker. Isolated crypts were resuspended in Matrigel (Corning, Tewksbury, MA) and 30 μL droplets were plated in a 24-well plate. After polymerization at 37°C, 500 μL of organoid expansion media (EM) was added for 2 days [Advanced Dulbecco’s modified Eagle medium/Ham’s F-12 (ThermoFisher, Waltham, MA), 10 mM HEPES (ThermoFisher), and 1× GlutaMAX (ThermoFisher), with 15%v/v R-spondin1 conditioned medium (cell line kindly provided by Calvin Kuo, Stanford University), 10% v/v Noggin conditioned medium (cell line kindly provided by Gijs van den Brink, Tytgat Institute for Liver and Intestinal Research (24)), 0.15 nM Wnt Surrogate-Fc fusion protein (AVS Bio, Utrecht, The Netherlands), 1× B27 supplement (ThermoFisher), 1 mM N-acetylcysteine (MilliporeSigma, Burlington, MA), 50 ng/mL human epidermal growth factor (Stem Cell Technologies, Vancouver, Canada), 10 nM [Leu-15] gastrin (Cayman Chemical, Ann Arbor, MI), 500 nM A83–01 (Cayman Chemical), 10 μM SB202190 (Cayman Chemical), 100 mg/mL primocin (InvivoGen, San Diego, CA), and 10 μM Y-27632 (Cayman Chemical)]. After 2 days, organoid EM (without Y-27632) was replaced every other day. Colonoids were passaged every 10–14 days by harvesting in Organoid Harvesting Solution (Bio-Techne, Minneapolis, MN) at 4°C with shaking for 60’. Colonoids were fragmented by trituration with a P200 pipet 30–50 times, collected and diluted in Advanced DMEM/F12, centrifuged at 300 ×g for 10’ at 4°C. The pellet was resuspended in Matrigel and plated as described for crypt isolation. All colonoid cultures were maintained at 37°C and 5% CO₂. Unless noted, colonoid lines have been passaged >20 times. No differences between passages were seen.

Generation of Colonoid Monolayers

Colonoid monolayers were generated as previously described (25). In brief, colonoids were fragmented by manual trituration and incubation in TrypLE, then seeded onto Transwell filters precoated with human collagen IV (10 μg/cm2). Approximately 50 colonoid fragments were added to the apical chamber of each filter, with organoid expansion media added to both the apical and basolateral sides and changed every other day. Confluency was confirmed by transepithelial electrical resistance (TER) measured using an EVOM2 epithelial volt ohmmeter (World Precision Instruments, Sarasota, FL).

Recombinant EspP Generation and Collection

AD202 cells transformed with the plasmid encoding wild-type EspP (pRLS5) were kindly provided by Harris Bernstein, National Institutes of Health (26). The cells were grown at 37°C in Luria-Bertani (LB) broth (ThermoFisher) overnight. They were then pelleted, washed, and grown at 37°C in fresh LB broth for approximately 15’. IPTG (100 μM) was added to induce EspP expression. The culture was grown until reaching an OD₆₀₀ 2.0. Bacterial cells were removed by centrifugation (9000 rpm, 30’, 4°C, Sorvall RC6, SLA-3000 rotor). EspP was collected from the cell-free supernatant by ammonium sulfate precipitation (60%, o/n, 4°C), followed by centrifugation (9000 rpm, 30’, 4°C, Sorvall RC6, SLA-3000 rotor). The pellet was resuspended in PBS, syringe filtered (0.2 μm), then diluted with 15% glycerol to allow for freezing. Each batch of recombinant EspP was separated on SDS-PAGE and stained with Coomassie Blue to check purity. Protein concentrations were determined by Bradford assay (Bio-Rad, Hercules, CA). Serine protease activity was determined by pepsin A-cleavage assay (27). To eliminate endotoxin contamination, each batch of EspP was processed through High Capacity Endotoxin Removal Resin spin columns (ThermoFisher), then protein purity and concentration was checked by Coomassie staining on an SDS-PAGE and BCA protein assay (ThermoFisher).

EspP Treatment

All experiments described in this study used undifferentiated colonoids (grown in the organoid EM). Colonoids were plated in diluted Matrigel (1:1 dilution with PBS) onto lysine-coated 4 well chamber slides prior to treatment. After 10 μg/mL overnight EspP treatment, colonoids were imaged or processed for further studies. Between 4–8 biologically unique colonoid lines with at least 3 independent experiments per colonoid line were used for experiments.

Immunofluorescence and Confocal Imaging

Colonoids were harvested from Matrigel using Organoid Harvesting Solution, pelleted (300 ×g, 10’, 4°C), and fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA). Colonoids were then permeabilized and blocked simultaneously for 1h using 10% Fetal Bovine Serum (Atlanta Biologicals, Flowery Branch, GA), 0.1% saponin (MilliporeSigma) solution prepared in PBS. After three PBS washes, primary antibody against CHGA (Development Studies Hybridoma Bank, Iowa City, IA), 5-HT (Abcam, Cambridge, United Kingdom), PYY (Abcam), or GLP-1 (ThermoFisher) diluted in PBS was added to the cells and incubated overnight at 4°C. Afterward, cells were washed 3 times with PBS, and 100 μL of AlexaFluor secondary antibodies, AlexaFluor-647 phalloidin, and DAPI (all ThermoFisher), diluted in PBS, were added for 1h at room temperature. All antibodies used are detailed in Table 1. After three PBS washes, 50 μL mounting media was added to the cells (50% glycerol solution, 2.5 M fructose (28)) and then mounted with a #1.5 coverslip. Confocal imaging was carried out in the University of New Mexico Comprehensive Cancer Center Fluorescence Microscopy Shared Resource using an LSM 800 AiryScan confocal microscope (Zeiss, Oberkochen, Germany) or an Olympus/Yokogawa CSU-W1 Spinning Disk confocal microscope (Olympus, Tokyo, Japan). For quantification of immunofluorescence images, at least five fields per condition were acquired. Each independent experiment included at least 100 colonoids and 1,000 cells per condition, with experiments using at least three biologically distinct colonoid lines and at least three independent experiments per line. For EdU proliferation assays, colonoids were treated with EspP or vehicle overnight and EdU (50 μM) was added to each well for 3h. Colonoids were fixed 16% PFA, then permeabilized for 20’ with 0.5% Triton X-100 in PBS. After three washes with 3% bovine serum albumin in PBS, Click-iT reaction cocktail was applied following manufacturer protocol (ThermoFisher).

Table 1.

Antibodies used

Antibody	Source	Identifier	Concentration
GLP1	Invitrogen	PA5-79303	1:200
CHGA	DSHB hybridoma	—	1:10
5-HT	Abcam	AB66047	1:200
PYY	Abcam	AB22663, AB131246, AB112474	1:150
Donkey anti-mouse AF488	Invitrogen	A32766	1:100
Donkey anti-goat AF568	Invitrogen	A11057	1:100
Donkey anti-rabbit AF568	Invitrogen	A10042	1:100
Phalloidin 647	Invitrogen	A30107	1:1000
DAPI	Invitrogen	62248	1 μg/mL

Treatment and Preparation of Colonoids for Single Cell RNA-Sequencing

Colonoids used for sequencing are described in Table 2. Colonoids in Matrigel were treated overnight with 10 μg/ml EspP. Next day, colonoids were removed from Matrigel by incubation with Organoid Harvesting Solution at 4°C for 45’, washed in Advanced DMEM/F12 and pelleted by centrifugation at 300 ×g for 10’. Colonoids were resuspended in TrypLE enzyme solution (ThermoFisher), then transferred to a 24 well plate and spun at 600 rpm at 37°C for 45’ in a spinoculator (Mixer HC, USA scientific) to digest into single cells. Single cells were washed in 10 ml of Advanced DMEM/F12, pelleted again, resuspended in 1 ml of organoid EM, and filtered using 40 μm Flowmi cell strainers (SP Bel-Art, Wayne, NJ). Cells were transported to the University of New Mexico Analytical and Translational Genomics Core for counting and processing. Average processing time from colonoids to library preparation was < 2.5hours and the cells were highly viable (>90%).

Table 2.

Description of colonoid donors used for scRNA-seq

Colonoid Identifier	Age	Race	Sex
4C	70	Hispanic, non-white	M
75C	58	White	M
82C	67	African American	F

Counting of Single Cells and Library Preparation

Using the Cell Countess II FL (Thermofisher), single cells and viability were counted by Trypan Blue exclusion, as previously described (29). Briefly, cell viability numbers were recorded (>90%), then the Next GEM Chip G was loaded and put into the Chromium Controller for processing, creating the GEMs. The GEMs were transferred into PCR tubes and cDNA synthesis was completed per the 10X Chromium Next GEM 3' Protocol. cDNA was evaluated for quality and quantity on a DNA High Sensitivity Chip on the Agilent BioAnalyzer Machine. Next, cDNA was fragmented, adaptor ligated, and sample indexed (barcoded) with PCR. Once the libraries were completed, they were processed on a DNA High Sensitivity Chip with the Agilent BioAnalyzer to ensure that the final libraries were approximately 500 bp in size. They were also processed with the KAPA Library Quantification Kit to verify the concentration of final libraries.

Pooling and Sequencing Quality Control

Sample concentration was determined using AgilentTape Station 4200 and Invitrogen Qubit 4.0 reagents. Samples were diluted, normalized, and pooled at 4nM prior to qPCR to determine cluster efficiency. Samples were loaded onto Illumina NovaSEQ 6000 instrument and sequenced on S4 Flow cells at specified read length. Data was delivered in FASTQ file format. Raw FASTQ reads were processed using Cell Ranger (V2.0.0) software from 10X Genomics. Briefly, sequencing reads were demultiplexed, unique molecular identifier (UMI) were collapsed and reads aligned to 10X human transcriptome GRCh38–3.0.0 as the reference.

scRNA-seq Quality Control and Processing of 10x Sequencing Data

The raw data created by Cell Ranger was read into the Seurat R package (v4.1.2) for quality control and downstream processing. For each sample, gene expression matrices were imported with the Read10x function and converted into Seurat objects using (CreateSeuratObject). Low quality cells expressing <200 genes and genes expressed in fewer than 3 cells were filtered out. After looking at the violin plots of the number of counts, features (genes), and percent mitochondria per sample, we further filtered cells expressing >2000 genes in control samples and cells expressing >4000 genes in EspP treated samples to avoid probable doublets. Cells expressing >20% of mitochondrial genes were filtered out in control and EspP samples. To account for differences in sequencing depth across samples, the data was log-normalized and scaled using Seurat’s (NormalizeData) function with the (LogNormalize) method and a scale factor of 10000. After quality checks to test sample variability, samples were merged with Seurat’s (merge) function with cell identities added and split by sample.

Clustering, Visualization and Cell Annotation

Initial clustering analyses were performed separately for control and EspP-treated colonoids to visualize condition-specific cell populations. For each condition, samples from 3 biologically distinct colonoid lines were merged using Seurat’s merge function after normalization and quality control. Highly variable genes were identified using the variance-stabilizing transformation method (FindVariableFeatures), retaining the top 2000 features. Dimensionality reduction was performed by principal component analysis (PCA) on these variable genes. The elbow plot was then used to determine the number of principal components (PCs) to retain, and the top 15 PCs were selected for clustering. Cells were grouped into unsupervised clusters using a shared nearest neighbor modularity optimization-based algorithm (FindClusters) at resolutions of 0.4 or 1.0 depending on dataset size and interest. Visualization of the resulting clusters was performed using Uniform Manifold Approximation and Projection (UMAP) (30).

Differential Expression and Cluster Annotation

Differentially expressed genes (DEGs) between clusters were calculated using the Wilcoxon rank-sum test implemented in Seurat. Cluster annotation was based on canonical marker expression and supported by literature references. Marker gene expression was visualized using violin and dot plots, where dot size reflects the percentage of cells expressing the gene and color reflects relative expression. DEGs identified through Seurat were ranked by log2 fold change for pathway enrichment analysis. Ranked DEG lists were submitted to g:Profiler for enrichment of Gene Ontology (GO) biological processes and Reactome pathways. Enrichment results were exported in GMT and GEM formats and visualized in Cytoscape using the EnrichmentMap application with a p-value cutoff of 0.05, a q-value cutoff of 1, and an overlap coefficient of 0.5. For EEC analysis, the pan-EEC markers CHGA and CHGB were used for marker-based gating and targeted subclustering.

Data Integration

For direct comparison between control and EspP treated populations, datasets from all biological replicates and treatment conditions were integrated using the Seurat integration workflow. For integrated analyses across biological replicates, datasets were normalized independently, and variable genes were identified prior to integration. Integration anchors were identified using the top 30 dimensions (FindIntegrationAnchors, dims = 1:30) and datasets were integrated with (IntegrateData). The integrated dataset was subsequently scaled, reduced by PCA, clustered, and visualized using UMAP as described above.

Cell Cycle Determination

Cell cycle phase determinations were made using (CellCycleScoring) in Seurat (v5) and established cell cycle gene sets (Seurat::cc.genes.updated.2019). Cells were scored using log normalized expression values from the RNA assay and phase distributions were calculated as the percent of cells per condition and per cluster.

Statistics

The relative abundance of each cell type was calculated as percentage of cells per condition (control and EspP). Data are represented as mean ± SEM. Statistical significances were calculated using Student’s t-test or Mann-Whitney U test. Significance was represented as at least p < 0.05. All experiments were performed on a minimum of 3 different colonoid lines derived from separate normal human subjects, with a total of 8 colonoid lines used throughout these studies. N refers to numbers of independent replicates performed. All analyses were performed on GraphPad Prism 10.4.1.

RESULTS

EspP Exposure Induces a Distinct Epithelial Cell Cluster in Human Colonoids

EspP treatment led to a nearly 2-fold increase in gene expression per cell, thus we sought to investigate the transcriptomic changes that occur in the epithelia following EspP treatment. Since the gut epithelium is made up of many cell lineages with distinct functions (31, 32), we performed droplet based single cell RNA-sequencing (scRNA-seq) on three biologically distinct human colonoid lines under vehicle (control, organoid EM) or EspP-treated (10 μg/mL) conditions.

Initial graph-based clustering using the Louvain algorithm and marker-gene analysis was performed separately for each donor to assess potential genetic variability (Fig S1). Similar grouping patterns and gene expression profiles were observed across all lines under both control and EspP-treated conditions. Therefore, replicates within control and EspP treated datasets were merged for downstream visualization and clustering analysis. UMAP visualization of each donor further confirmed similar donor contributions across groups, with no evidence of donor specific grouping (Fig S2).

Unsupervised clustering using the Louvain algorithm and marker-gene analysis revealed few clusters in control colonoids that primarily represented cells found in the deep crypt (Fig 1A, left). Clusters were manually identified using marker genes and published datasets (31, 33, 34) as: cycling transit amplifying (TA) cells, stem cells, colonocytes, goblet cells and secretory progenitor cells (Tables S1, S2). Cycling TA cells were defined by strong expression of cell cycle and proliferation-associated genes (such as MKI67, TOP2A, CENPF, PCNA), whereas secretory TA cells retained expression of cell cycle genes while also expressing secretory lineage commitment transcripts. Secretory progenitors predominantly express secretory differentiation-associated genes (such as ATOH1) with reduced expression of proliferation or cell cycle-associated genes. Differentiated cell types (such as goblet or enteroendocrine cells) express unique, identifiable transcripts. These colonoids were representative of nondifferentiated epithelia, since they were maintained in organoid EM that contained Wnt surrogate/Rspo-1/Noggin.

Fig 1. Acute EspP treatment on human colonoids leads to epithelial differentiation and expansion of enteroendocrine cells.

(A) Representative Uniform Manifold Approximation and Projection (UMAP) of epithelial cell clusters identified from merged control and EspP-treated colonoids. Putative cluster identification is listed on each cluster.

(B) The top 10 differentially expressed genes found in EspP-treated colonoids, showing enteroendocrine markers CHGA and TPH1 have the highest fold change.

(C) Quantification of the percentage of CHGA+ cells compared between the pooled control and EspP-treated colonoids in the scRNA-seq datasets.

(D) Heatmap showing expression of enteroendocrine lineage markers across both control and EspP-treated epithelial clusters.

(E) Representative immunostaining for CHGA in control and EspP-treated colonoids. CHGA, green; nuclei, blue; scale bar = 20 μm.

(F) Quantification of percent CHGA+ cells relative to total epithelial cells, in control and EspP-treated colonoids. p = 0.002, n≥ 5. Data are presented as mean ± SEM. p-values are based on Student’s t-test.

(G) Representative immunostaining for CHGA in control and EspP-treated colonoid monolayers. EspP was added to the apical surface of the monolayers. CHGA, green; GLP-1, red; nuclei, blue; scale bar = 20 μm.

(H) Quantification of percent CHGA+ cells relative to total epithelial cells, in control and EspP-treated colonoid monolayers. p = 0.0411, n = 4. Data are presented as mean ± SEM. p-values are based on Mann-Whitney test.

In contrast, EspP-treated colonoids maintained in identical media, had a higher degree of cellular differentiation. Unsupervised clustering of the treated dataset revealed epithelial populations common to both conditions, as well as a distinct EspP-associated cluster not present in control, shown in the representative EspP-treated colonoid UMAP (Fig 1A, right). EspP-treated clusters were manually annotated and identified as: stem cells, cycling TA, secretory progenitors, secretory TA, goblet cells, enteroendocrine cells (EECs), and colonocytes (Tables S1, S3). This indicates that acute EspP treatment diversified the differentiated epithelial cell lineages in human colonoids. These clustering patterns were observed in all three EspP-treated colonoid lines (Fig S1). Additionally, these patterns were confirmed at higher resolution 1.0, with control colonoids still lacking enteroendocrine clusters (Fig S3, Table S4)

Differential gene expression analysis using the Seurat Wilcoxon rank-sum test identified chromogranin A (CHGA) as the top upregulated gene globally (avg_log2FC = 5.12), representing a 34.7-fold increase in gene expression compared to control (Fig 1B, Table S5). Alongside the increased transcript density per cell, we observed an expansion in the proportion of cells expressing this marker. Analysis of the merged scRNA-seq datasets showed a 3.3-fold increase in the percentage of CHGA+ cells in the EspP-treated group compared to control (Fig 1C). Heat map analysis of CHGA and other EEC markers within merged treated clusters showed strong localization of positive cells within newly formed EspP-treated cluster 6, with minimal expression seen in either control or other EspP-treated clusters (Fig 1D).

To validate expansion of the EEC lineage, immunofluorescence staining for CHGA across ≥5 distinct colonoid lines was performed. Automated quantification using the Leica AI Aivia platform revealed a significant increase in the percentage of CHGA+ cells relative to total cells (Fig 1E, F). This effect was similarly observed in colonoid monolayers apically treated with EspP, with automated quantification showing a significant increase in the percentage of CHGA+ cells relative to total cells (Fig 1G, H).

EspP Exposure Alters Epithelial Cell Cycle Dynamics

To enable direct comparison of epithelial states between control and EspP-treated clusters, datasets from both conditions were integrated together prior to downstream analysis. To determine whether EspP-induced transcriptional changes were associated with altered proliferative states, cell cycle phase scoring was performed on control and EspP-treated integrated datasets. EspP induced global remodeling of the cell cycle, characterized by a nearly three-fold decrease in the S-phase population (34.9% vs 12.5% in control) and a corresponding increase in G1 (37.8% vs 60.1% in control) (Fig 2A). This shift was supported by the dysregulation of canonical proliferation markers. The percentage of cells expressing G1/S cyclins (CCND1, CCND2) increased while DNA replication machinery (PCNA, MCM2) decreased across EspP-treated samples (Fig S4). Cluster specific analysis determined that proliferative activity was decreased in all treated clusters compared to control. While S-phase percentage exceeded 30% across control clusters, it decreased to ~12% in all EspP-treated clusters. Although there was a marginal shift in overall G2M percentages with EspP treatment, there was cluster-specific cell-cycle distribution, with cluster 3 (identified as stem cells) showing a subtype specific increase and an increase in the mitotic marker UBE2C (Fig 2B, C). To confirm these computational findings, EdU staining was performed on control and EspP treated colonoids, showing a significant decrease in EdU incorporation (representing S phase) with EspP treatment (Fig 2 D, E).

Fig 2. EspP treatment disrupts proliferation and alters cell cycle dynamics.

(A) Distribution of epithelial cells across cell cycle phases G1, S, and G2/M in control and EspP-treated colonoids.

(B) Violin plot of cell cycle gene UBE2C, with cluster-specific expression in cluster 3 (stem cells).

(C) Comparison of cell percentages for cell cycle phase distribution comparing control and EspP- treated colonoids. n = 3 colonoid lines.

(D) Representative immunostaining for EdU in control and EspP-treated colonoids. EdU, white; nuclei, blue; scale bar = 20 μm.

(E) Quantification of percent EdU+ cells relative to total epithelial cells, in control and EspP-treated colonoids. p = 0.0014, n ≥ 4. Data are presented as mean ± SEM. p-values are based on Student’s t-test.

EspP Treatment Alters Hormone Expression Patterns and Promotes Subtype-Specific Enteroendocrine Cell Response

Enterochromaffin and L cells are the two most abundant EEC subtypes in the colon, however, there are at least 12 known subtypes in the small bowel and colon (35). We further analyzed EEC subtype transcripts in control and EspP-treated colonoids. Dot plots revealed that EEC subtype-associated transcripts were diffusely expressed at low levels across control clusters (Fig 3A, left), while EspP-treated clusters showed more spatially restricted and cluster specific hormone expression patterns (Fig 3A, right). Comparison of EEC transcripts between control and EspP-treated revealed selective remodeling of EEC subtypes rather than a global expansion of all EEC subtypes. EspP treatment led to a 3.7-fold increase in peptide tyrosine tyrosine, PYY (L-cells) and 2-fold increase in tryptophan hydroxlase 1, TPH1 (enterochromaffin cells) (Fig 3B).

Fig 3. EspP-treatment induces subtype-specific enteroendocrine cell differentiation.

(A) Dot plots of enteroendocrine-related transcripts in control (left) and EspP-treated (right) colonoids (n = 3).

(B) Quantification of enteroendocrine-related transcripts relative to total cells in control and EspP-treated colonoids.

(C) Higher resolution UMAP subclustering (1.0) of EECs from pooled control and EspP-treated colonoids. Putative cluster identification is listed on each cluster.

(D) Heatmap of EEC markers across putatively identified EEC subclusters for control (left) and EspP-treated (right) colonoids.

To further resolve EEC heterogeneity that resulted from acute EspP exposure, we reanalyzed EEC clusters by subclustering at a higher resolution of 1 instead of 0.4, then identified each subcluster by comparing the differentially expressed genes (Table S6) with published EEC datasets (18, 36, 37). The control colonoid EECs separated into six clusters identified as: EEC progenitors and crypt-based EEC (Fig 3C, left UMAP). The EspP-treated EECs subclustered into eight clusters, identified as: EEC progenitors, early enterochromaffin (EC) cells, EC cells, and a mature mixed EC and L-cell population (Fig 3C, right UMAP). The transcriptional identity of these subclusters was further characterized by the expression of canonical EEC markers (Fig 3D). In EspP-treated EECs, subclusters 0 through 6 showed varying levels of CHGA and chromogranin B (CHGB), consistent with progenitor and early stage EC cells, however, subcluster 7 exhibited the highest maturation profile. This population was defined by marked expression of CHGA and TPH1, confirming its identity as a mature EC population. Notably, subcluster 7 had unique co-expression of PYY not seen in any other subcluster.

To validate the subtype specific EEC changes identified by our scRNA-seq dataset, immunostaining was performed for representative EEC hormones and peptides, including serotonin (5-HT; EC cells), PYY (L-cells), and glucagon-like peptide 1 (GLP-1; L-cells). In control colonoids, 5-HT, PYY, or GLP-1-positive cells were rare and sparsely distributed (Fig 4A–C, left). However, EspP-treated colonoids exhibited a specific marked increase (Fig 4A–C, middle). Quantification across ≥5 independent colonoid lines showed statistically significant increases in 5-HT and PYY-positive cells but not GLP-1-positive cells (Fig 4A–C, right).

Fig 4. Expansion of specific EEC subtypes upon EspP treatment.

(A) (Left) Representative immunostaining for serotonin (5-HT) and CHGA in control and EspP-treated colonoids. CHGA, green; 5-HT, red; nuclei, blue; scale bar = 20 μm. (Right) Quantification of percent 5-HT+ cells relative to total epithelial cells, in control and EspP-treated colonoids. p = 0.0008, n ≥ 5.

(B) (Left) Representative immunostaining for peptide tyrosine tyrosine (PYY) and CHGA in control and EspP-treated colonoids. CHGA, green; PYY, red; nuclei, blue; scale bar = 20 μm. (Right) Quantification of percent PYY+ cells relative to total epithelial cells, in control and EspP-treated colonoids. p = 0.0033, n ≥ 5.

(C) (Left) Representative immunostaining for glucagon-like peptide-1 (GLP-1) and CHGA in control and EspP-treated colonoids. CHGA, green; GLP-1, red; nuclei, blue; scale bar = 20 μm. (Right) Quantification of percent GLP-1+ cells relative to total epithelial cells, in control and EspP-treated colonoids. p = 0.4352, n ≥ 5. Data are presented as mean ± SEM. p-values are based on Student’s t-test.

EspP-Induced EECs Display Immune Associated Profiles

To further understand the altered transcriptional state of EspP-induced EECs, we examined the genes with the highest avg_log2FC within the EspP-treated EEC subclusters. This revealed enrichment of GDF15, NEAT1, and MALAT1, along with transcripts associated with stress responses. Top downregulated genes within this EEC subcluster included KRT18, FABP1, and CTNNB1 (Fig 5A).

Fig 5. EspP-indued enteroendocrine cells exhibit immune-associated signatures.

(A) The top upregulated (left) and downregulated (right) genes in EspP-treated EECs compared to control EECs.

(B) KEGG pathway enrichment analysis of differentially expressed genes in EspP-induced EECs.

(C) Representative cytokine signaling inference graph generated from the CytoSig platform.

The broader biological processes represented by this EEC gene expression profile were determined using pathway enrichment analysis performed via KEGG. EspP-treated EECs showed enrichment of pathways related to infectious disease response, epithelial junctions, and intracellular signaling (Fig 5B). To determine broad alterations in signaling pathways in other cell populations, KEGG enrichment analysis was performed across all remaining EspP treated clusters. Cluster 3 (identified as stem cells) showed enrichment of antigen processing and presentation pathways, consistent with recent work that identified human colon stem cells as the primary epithelial responders to bacterial antigens (38). Interestingly, cluster 5 (identified as goblet cells) showed enrichment of nucleotide sugar biosynthesis pathways, the precursor machinery required for mucin O-glycosylation. Cluster 7 (identified as colonocytes) displayed the most immune specific response of the non-EEC clusters, with enrichment of NFkB/TNF/NOD-like receptor signaling pathways alongside epithelial cell signaling in Helicobacter pylori infection as the top hit (Fig S5).

To determine broader cytokine/chemokine signaling profiles associated with EspP treatment, top DEGs from control and treated were examined using CytoSig profile mapping. This revealed selective enrichment of chemokine-related signaling patterns in EspP-treated colonoids compared to control. Specifically, CCL2, CCL7, and CCL25 emerged as key transcripts associated with EspP treatment (Fig 5C). Overall, these results demonstrate that EspP, a potent EHEC cytotoxin, mimics EHEC infection by stimulating transcriptional changes previously attributed to the whole bug infection and induces EEC differentiation as an acute response to this injury.

DISCUSSION

Gut epithelial responses to microbial toxins have been linked to disrupted epithelial homeostasis, altered differentiation, and immune-associated signaling (39–42). However, the transcriptional and lineage-specific epithelial responses that underlie these outcomes remain poorly defined. In this study, human colonic organoids (colonoids) and single-cell RNA sequencing were used to understand how exposure to the EHEC secreted serine protease EspP alters intestinal epithelial cell fate and immune relevant signaling pathways. Single cell analysis revealed a reduction in proliferative cells alongside a pronounced shift toward secretory lineage differentiation following EspP treatment. Similar reductions in proliferative capacity have been observed following exposure to non-microbial environmental insults (29), suggesting that diverse stressors converge on conserved epithelial stress responses.

Cell cycle analysis of EspP-treated colonoids showed a reduction in cells in S-phase, increase in G1, and altered expression of genes associated with cell cycle progression and DNA replication. However, these changes do not reflect uniform suppression of proliferation in all epithelial populations. UBE2C, a key regulator of mitotic progression responsible for degradation of mitotic cyclins and facilitating the metaphase-to-anaphase transition, was among the top upregulated genes following EspP treatment. UBE2C was concentrated within a specific stem cell subcluster rather than evenly distributed across epithelial populations. This cluster exhibited increased representation of cells in G2/M, despite the global reduction in proliferative capacity observed in EspP-treated colonoids.

Alongside this disruption in proliferation, EspP exposure prompted differentiation toward the secretory lineage, with a marked expansion of enteroendocrine-related transcripts, and a correlative increase in enteroendocrine cell (EEC) numbers. These enteroendocrine-associated transcripts were concentrated within a newly emerged epithelial cluster, suggesting that EspP drives the formation of an epithelial state enriched for enteroendocrine identity rather than uniformly altering all epithelial lineages. EECs are rare under homeostatic conditions but act as dynamic regulators of gut-brain communication (43, 44).

A major finding of this study was the diversification of EECs following EspP treatment. EspP-treated colonoids exhibited significantly higher expression of 5-HT, a marker of enterochromaffin cells, and of PYY, a marker of L-cells. These findings suggest that EspP does not broadly increase EEC numbers but instead alters EEC state and subtype composition.

In the intestinal epithelium, 5-HT plays a central role in gut motility and secretion (45), enhances mucosal immune responses, and has been implicated in epithelial wound healing and regeneration (46). Increases in luminal 5-HT has been shown to decrease virulence of both EHEC and Citrobacter rodentium infection by inactivating the bacterial transcription factor CpxR, which controls many bacterial virulence genes (47). Thus, an EspP-induced increase in enterochromaffin cells may have a beneficial effect against bacterial toxin cytotoxicity.

PYY producing L-cells are involved in hunger and satiety, barrier function, and epithelial growth responses (48–50). PYY has also been implicated in modulation of inflammatory signaling within the gut (51), suggesting that EspP-induced expansion may influence epithelial-immune communication during early toxin exposure.

Further characterization of EspP-induced EECs revealed enrichment of stress-associated and long noncoding RNA transcripts, suggesting that these cells adopt an active stress-response state after EspP treatment. Among the highest expressed genes was GDF15, a stress-responsive cytokine within the TGFβ family. Although the functions of GDF15 in normal colonic epithelium are not fully understood, its known immunomodulatory roles (52) suggest that EspP-induced EECs may actively signal epithelial distress rather than dying. EspP-induced EECs also showed broad upregulation of the long noncoding RNA NEAT1, which has been implicated in transcriptional reprogramming during cellular stress and has been linked to immune modulation and differentiation in other epithelial contexts (53). The widespread induction of NEAT1 across the treated EEC population further supports the conclusion that EspP exposure drives coordinated transcriptional remodeling.

In addition, a subset of EECs exhibited increased expression of CYP1A1, a gene classically associated with epithelial sensing of xenobiotic and microbe-derived small molecules (54, 55). While EECs are not traditionally considered detoxifying cells, they are recognized as luminal sensors, and induction of CYP1A1 suggests engagement of these pathways. Notably, this response occurred in an epithelial-only system lacking microbiota or intact bacteria, indicating that EspP alone is sufficient to induce these sensing programs. Conversely, genes downregulated in EspP-induced EECs included CTNNB1 (β-catenin), a central mediator of WNT signaling, and TUBA1B, DYNLL1, and KRT18, which are involved in cytoskeletal organization and intracellular transport. This suggests that EspP-induced EECs may undergo changes in hormone packaging or trafficking.

Lastly, pathway enrichment analysis of EspP-induced EECs identified pathways associated with epithelial junctions, infectious disease and immune signaling. The enrichment of epithelial junction pathways is consistent with the well established role of EspP in disrupting barrier integrity and suggests that disruptions in these epithelial barrier pathways are an early response to EspP exposure. Enrichment of pathways associated with infectious disease and immune-related pathways validates prior studies that described an immunoendocrine axis within the gut, with EECs secreting cytokines, hormones, and peptides in response to microbial metabolites (15, 56). Together, these findings suggest that EspP-induced EECs adopt a stress-responsive, immune-associated transcriptional state, consistent with a role for EECs as active epithelial sensors during early acute toxin exposure.

In summary, this study demonstrates that acute exposure to the EHEC-secreted serine protease EspP induces rapid epithelial remodeling characterized by altered proliferation, selective expansion and diversification of EECs, and activation of immune-associated transcripts. EspP has previously been shown to be the critical determinant of Shiga toxin internalization (7), the primary virulence factor responsible for severe EHEC disease; however, these data indicate that EspP also reshapes epithelial lineages during early stages of asymptomatic exposure. These changes may lower the threshold for subsequent toxin uptake, amplify immune dysregulation, or alter disease trajectory once Shiga toxin exposure from EHEC occurs. Understanding how EspP-driven epithelial remodeling influences immune signaling provides important insight into the early, asymptomatic phase of disease. Defining these early epithelial responses is critical for understanding EHEC disease trajectory and identifying intervention points before irreversible pathology can develop.

Supplementary Material

Supplemental Figs S1–5; Supplemental Tables S1–6: 10.6084/m9.figshare.31719943

Supplemental Fig 1. Comparison of epithelial cell lineages on individual control and EspP-treated colonoids.

(A) UMAP of Control (left) and EspP-treated (right) colonoids from line 4C. Putative cluster identification is listed on each cluster.

(B) UMAP of Control (left) and EspP-treated (right) colonoids from line 75C. Putative cluster identification is listed on each cluster.

(C) UMAP of Control (left) and EspP-treated (right) colonoids from line 82C. Putative cluster identification is listed on each cluster.

Supplemental Fig 2. Comparison of epithelial cell lineages on merged control and EspP-treated colonoids separated by donor.

UMAP of Control (left) and EspP-treated (right) colonoids from merged control and treated samples.

Supplemental Fig 3. Comparison of epithelial cell lineages on merged control and EspP-treated colonoids at resolution 1.0.

UMAP of Control (left) and EspP-treated (right) colonoids from merged control and treated samples. Putative identifications listed on each cluster.

Supplemental Fig 4. EspP exposure alters the percent expression of cell cycle markers.

Bar graph showing percent expression of several cell cycle markers in both control and treated transcripts.

Supplemental Fig 5. KEGG pathways enriched on each treated cluster.

Top enriched KEGG pathways for clusters 0: stem cell; 1: secretory progenitor; 2: cycling TA; 3: Stem cell; 4: secretory progenitor; 5: goblet cell; 7: colonocytes.

Supplemental Table 1. Differential expressed genes for each cluster in control and EspP-treated colonoids, separated by donor.

Supplemental Table 2. Merged differential expressed genes for the control colonoids.

Supplemental Table 3. Merged differential expressed genes for the EspP-treated colonoids.

Supplemental Table 4. Representative UMAP (1.0 resolution) of epithelial cell clusters identified from merged control and EspP-treated colonoids. Putative cluster identification is listed on each cluster.

Supplemental Table 5. Top 50 globally upregulated genes in the EspP-treated colonoids.

Supplemental Table 6. Higher resolution subclustering of enteroendocrine clusters in merged control and EspP-treated colonoids.

DATA AVAILABILITY

All unique reagents generated in this study are available from the Lead Contact without restriction. All scRNA-seq datasets are available in Gene Expression Omnibus (GEO GSE239584) and Sequence Read Archive (SRA) and will be released upon publication.