CD14⁺CD16⁻ monocytes exhibit NF-κB hyperactivation in biliary atresia: Clinical association and murine therapeutic validation

Abstract

Background:

Classical CD14⁺CD16⁻ monocytes are elevated in biliary atresia (BA); however, their specific role in bile duct injury and the underlying regulatory mechanisms remain unclear. This study aimed to define their contribution to BA pathogenesis, focusing on the NF-κB signaling pathway.

Methods:

Liver tissues and blood samples from patients with BA and controls were analyzed by single-cell RNA sequencing, flow cytometry, and immunofluorescence. A rhesus rotavirus–induced BA mouse model was used for anti-Ly6C monocyte depletion and NF-κB inhibition (dehydroxymethylepoxyquinomicin). Transcriptomic profiling and cytokine analysis revealed key molecular mechanisms.

Results:

Classical monocytes were significantly enriched near the damaged bile ducts in patients with BA and positively correlated with liver injury severity. These monocytes exhibited NF-κB hyperactivation, marked by the upregulation of TNF, IL-1β, Cxcl2, and NLRP3 inflammasome components. RNA-seq revealed BA-specific monocyte clusters with enriched NF-κB signatures. The depletion of classical monocytes (anti-Ly6C) in rhesus rotavirus–induced BA mice reduced biliary inflammation, restored bile duct patency, and improved survival. Pharmacological NF-κB inhibition (dehydroxymethylepoxyquinomicin) similarly attenuated inflammation and liver dysfunction and improved survival in rhesus rotavirus–induced BA mice.

Conclusions:

Classical CD14⁺CD16⁻ monocytes are spatially enriched and exhibit NF-κB hyperactivation in BA. Targeting these cells or their NF-κB axis represents a promising therapeutic strategy to mitigate disease progression.

INTRODUCTION

Biliary atresia (BA) is a severe neonatal cholangiopathy characterized by progressive extrahepatic bile duct inflammation and fibrosis, leading to impaired bile flow, liver damage, pathological jaundice, and eventual liver failure within the first year if untreated.^1,2 Despite Kasai portoenterostomy, over 50% of patients require liver transplantation by age 2 years,^3,4 underscoring the urgent need for pathogenesis-targeted therapies. Immune-mediated bile duct injury is central to BA^5,6; the specific contributions of immune cell subsets and their molecular drivers remain incompletely defined.

Monocytes, which are key orchestrators of innate immunity, are increasingly implicated in biliary inflammation. They are categorized into classical (CD14⁺CD16⁻), nonclassical (CD14⁻CD16⁺), and intermediate (CD14⁺CD16⁺) subpopulations.^7–9 Classical monocytes drive proinflammatory responses, whereas nonclassical and intermediate subsets are primarily associated with tissue repair and anti-inflammatory functions. ¹⁰ In biliary diseases, monocytes are critically implicated in pathogenesis through recruitment to the bile ducts and differentiation into macrophages that drive inflammatory and fibrotic processes.^11,12 Emerging evidence from single-cell and spatial transcriptomic analyses highlights the heterogeneity of monocyte-derived macrophages in BA livers, with distinct subsets exhibiting proinflammatory or profibrotic functional specialization.^11,13 For instance, Ly6C⁺ macrophages are enriched in BA models and correlate with severe liver injury, whereas MHC-II⁺ macrophages demonstrate a reduced prevalence. ¹³ Chemokine signaling also plays a pivotal role in monocyte recruitment. The CCL2-CCR2 axis facilitates monocyte infiltration into the portal tracts, with TNF superfamily ligands potentially contributing to this process. ¹⁴ Notably, our previous single-cell RNA sequencing (scRNA-seq) study revealed monocyte accumulation in BA livers, ⁶ subset specialization, spatial interactions, and signaling pathways governing their inflammatory activity remain unexplored.

The NF-κB pathway, a master regulator of inflammation, is activated in BA ¹⁵ and drives cytokine production. ¹⁶ However, its specific role in monocyte-mediated bile duct injury, particularly within classical subsets, and whether this activation is a causal driver or a bystander effect in BA pathology, remains unknown.

In this study, we integrated human and murine BA models to define the spatial, functional, and molecular roles of classical CD14⁺CD16⁻ monocytes. We identified NF-κB hyperactivation as the key inflammatory effector mechanism and demonstrated the therapeutic efficacy of monocyte depletion and NF-κB inhibition in mice with rhesus rotavirus (RRV)-induced BA. Our findings bridge a critical gap between BA immunopathology and actionable therapeutic strategies.

METHODS

Human samples

The children were enrolled in the Department of Pediatric Surgery at Guangzhou Women and Children’s Medical Center between March 2022 and June 2025. Patients diagnosed with BA (n=30) had a median age of 2 months (range: 1–7 mo), and had no visualization of the intrahepatic and/or extrahepatic bile ducts on intraoperative cholangiography. Age-matched controls (Control, n=30; median age: 3 mo; range: 1–8 mo) comprised patients with choledochal cysts without marked hyperbilirubinemia (primarily choledochal cysts). The exclusion criteria included active systemic infections, antibiotic use within 4 weeks, immunosuppressant therapy, or comorbid noncholestatic liver diseases. Liver tissues from patients with BA were obtained during Kasai portoenterostomy. Control tissues (choledochal cysts) were collected during the corrective surgery. All samples were collected before therapeutic intervention. The baseline demographic and clinical/laboratory parameters and confounders (including documented infections and antibiotic exposure) are shown in Supplemental Table S1, http://links.lww.com/HC9/C298.

This study was conducted in accordance with the principles of the Declaration of Helsinki and the Declaration of Istanbul, and was approved by the Research Ethics Committee of Guangzhou Women and Children's Medical Center (Approval No. 130B01). Written informed consent was obtained from all participants or their legal guardians.

scRNA-seq and data analysis

For scRNA-seq, patients (n=6 BA, n=6 choledochal cysts) were selected based on tissue quality (viability >85%, RNA integrity number >8, and cell yield >10,000 cells/mg). As expected, the groups were comparable in median age and sex distribution and differed markedly in biochemical cholestasis (Supplemental Table S1, http://links.lww.com/HC9/C298). RRV-induced BA mouse models and controls were used, with n=3 in each group. Tissue collection and induction in the mouse BA model have been described in previous studies.^17,18 Library preparation adhered to the 10× Genomics protocol, with a sequencing depth of 50,000 reads per cell.

Data processing involved quality control to filter low-quality cells, with 12,000–15,000 cells per sample retained. The normalization/batch correction used was Seurat. ¹⁸ Integration uses harmonization. ⁶ Clustering and cell type identification were performed using the Seurat pipeline. ¹⁷ Pathway enrichment was performed using DAVID. ⁶ Statistical significance was determined using Wilcoxon rank-sum tests, with false discovery rate <0.05 considered significant. ¹⁸ All the analyses were performed using R (version 4.0.3) and Python (version 3.8).

Flow cytometric analysis and sorting

Single-cell suspensions were blocked with anti-CD16/CD32 (Biolegend) and stained with LIVE/DEAD 7AAD (Thermo Fisher Scientific) and monocyte surface antibodies: human (CD45, HLA-DR, CD16, and CD14) and mouse (CD45, CD11b, CD43, and Ly6C).

For NF-κB, cells were fixed and permeabilized using the Foxp3/transcription factor staining buffer set (eBioscience) before staining with p-p65 (Ser529) (BioLegend, Cat#614154) and p65 (BioLegend, Cat#653006) antibodies. Data were acquired on a BD LSRFortessa (BD Biosciences) and analyzed with FlowJo V10.8 (FlowJo). Sorting was performed using a BD FACSAria Ⅲ (≥95% purity).

Immunofluorescence analysis

Following deparaffinization and rehydration, the tissue sections were subjected to heat-induced antigen retrieval using the universal HIER antigen retrieval reagent (Abcam, Cat#ab208572). Subsequently, the tissues were incubated with 5% bovine serum albumin to block nonspecific binding. Human liver tissues were then immunostained with the following primary antibodies: EpCAM (1:400 dilution, Cat#AB223582; Abcam), CD14 (1:200 dilution, Cat#ab181470; Abcam), and CD16 (1:200 dilution, Cat#MA5-50690; Invitrogen) diluted in 5% bovine serum albumin, and incubated for 1 hour at room temperature. Secondary antibody detection was performed using a monochrome tyramide signal amplification fluorescent dye kit (PANOVUE, Cat#10002100100), which included PPD520 (FITC), PPD570 (Cy3), and PPD650 (Cy5), each diluted 1:100 in the tyramide signal amplification signal amplification solution. Nuclei were counterstained with DAPI (1:5000 dilution in PBS, Life Technologies). Immunofluorescence images were acquired using a TCS SP8 confocal microscope (Leica Microsystems). Cell counts were performed on 16 randomly selected frames, and the average number of cells per high-power field was calculated for quantitative analysis.

Mass cytometry (CyTOF)

Mouse liver tissue was minced and enzymatically digested using a Liver Dissociation Kit (Miltenyi Biotec, Cat. No. 130-105-807) at 37 °C for 1 hour and filtered through a 70-μm strainer to obtain a single-cell suspension. Approximately 2.7×10⁶ viable cells (90% of the initial 3×10⁶) were washed with 1×PBS, stained with 250 nM cisplatin (Fluidigm) for 5 minutes on ice to exclude dead cells, and incubated with Fc receptor blocking solution. Surface antibodies were added for 30 minutes on ice, washed with FACS buffer (1×PBS + 0.5% bovine serum albumin), and fixed overnight in Maxpar Fix and Perm Buffer (Fluidigm). After washing, the cells were permeabilized with Perm buffer (eBioscience) and stained with intracellular antibodies for 30 minutes on ice. Cells were then washed, resuspended in deionized water, mixed with 20% EQ beads (Fluidigm), and analyzed using a Helios mass cytometer (Fluidigm).^19,20 Raw data were debarcoded, normalized, and analyzed using FlowJo software to manually gate CD45⁺ cells, excluding debris, dead cells, and doublets. The cell frequencies in each cluster were calculated as the ratio of assigned cell events to the total number of events in the sample. T-SNE dimension reduction was performed using the R package, and cell types were identified based on canonical markers.

Murine BA model

Neonatal BALB/c mice were i.p. injected with 20 μL (titer: 1.5×10⁶ plaque-forming units) rhesus monkey MMU18006 rotavirus (RRV) within 24 hours of birth to establish a BA mouse model. Daily clinical assessment included documentation of acholic stools, jaundice incidence, and weight change. RRV (VR-1739, ATCC) was amplified using monkey kidney MA-104 cells (kindly provided by the Center for Health Protection of Hong Kong).

For Anti-Ly6C antibody (BioXCell, clone Mouts-1) treatment, mice were randomly allocated into 4 groups: (1) RRV+PBS, (2) RRV+Isotype control, (3) RRV+Anti-Ly6C, and (4) Vehicle control. Mice in group (3) received an i.p. injection of 40 μg of anti-Ly6C antibody (diluted in PBS to 20 μL). Similarly, mice in group (2) received an i.p. injection of the isotype control antibody at a dosage of 40 μg (BioXCell, clone 2A3). The remaining 2 groups (1) and (4) were administered an i.p. injection of the PBS vehicle control at a volume of 20 μL. Four hours postantibody/PBS administration, groups (1), (2), and (3) were inoculated with RRV (20 μL, 1.5×10⁶ plaque-forming unit, i.p.) to induce BA, while group (4) received a saline injection. Antibodies were administered every 2 days until day 12. Daily monitoring included body weight, survival rate, and incidence of jaundice. The mice were evaluated to collect data on liver tissue inflammation, liver function, and alterations in extrahepatic bile duct obstruction on D12.

To elucidate the therapeutic potential of the NF-κB inhibitor, dehydroxymethylepoxyquinomicin (DHMEQ) (MedChemExpress, HY-14645), in the context of BA induction, a dosing regimen of 4 mg/kg body weight was initiated 24 hours after RRV inoculation and administered every 48 hours throughout the disease induction period.

Histopathological analysis

Live tissues were fixed in 4% PFA for 24 hours. Paraffin-embedded tissue sections were prepared and subjected to hematoxylin and eosin staining. Liver injury was evaluated based on the following criteria: necrosis, inflammation, ballooning degeneration, and disruption of the hepatic cord structure using a semiquantitative scoring system (0-3, 0=none, 1=mild, 2=moderate, 3=severe). ²¹ An intermediate increment of 0.5 was applied when the observed inflammation fell between 2 defined levels. Three randomly selected high-power fields were scored for each tissue section by using a Leica microscope. Scoring was independently performed by 2 blinded pathologists to ensure consistency and minimize bias.

Fluorescence cholangiography

A refined fluorescence cholangiography protocol was established to assess biliary patency in postnatal day 12 (P12) BALB/c mice. Animals were anesthetized using 2% isoflurane in oxygen (3% induction for 2 min, followed by 1.5%–2% maintenance). Following laparotomy, 5 μL of FITC-dextran (70 kDa, 20 mg/mL in PBS) was injected retrogradely into the gallbladder apex via a 33G needle. The extrahepatic bile duct was imaged for 10 minutes using a fluorescence stereo microscope (LABOMED Luxeo 6Z; GFP filter: 470 nm/530 nm), with continuous monitoring of dye flow dynamics. Biliary patency was evaluated based on (1) uniform dye distribution through the common bile duct and (2) the absence of dye pooling or stagnation. Stricture formation was defined as an abrupt interruption of dye flow or irregular luminal narrowing.

Bulk RNA-seq and data analysis

Total RNA was isolated from CD45⁺CD11b⁺Ly6C⁺CD43⁻ cells derived from the mouse liver using TRIzol reagent. RNA was reverse-transcribed using the Discover-sc template strand oligo V2, and cDNA was purified for DNA library preparation and sequencing. The reads were trimmed (Cutadapt), aligned (HISAT2), and quantified (FPKM, String Tie). Differentially expressed genes were identified using DESeq. 2, with a threshold of log2 fold-change >1.2 and a p value <0.05. Enrichment used Kyoto Encyclopedia of Genes and Genomes.

Quantification of infectious viral titer by fluorescent focus assay

Infectious rotavirus titers in mouse bile duct tissues were quantified by fluorescent focus assay on day 7 postinoculation.^22,23 Briefly, harvested bile duct tissues were homogenized in Earle’s Balanced Salt Solution and clarified by centrifugation. Serial 10-fold dilutions of the supernatants were prepared. MA-104 cell monolayers in 96-well plates were inoculated with the diluted samples. After a 2-hour adsorption period at 37 °C, the inoculum was replaced with maintenance medium containing 1.2% methylcellulose to prevent viral spread. Following 48-hour incubation, the cells were fixed with 4% formaldehyde and permeabilized.

For immunostaining, the cells were incubated with FITC-conjugated anti-rotavirus antibody (1:500, ab31435, Abcam) for 2 hours at room temperature. After washing, the fluorescent foci were visualized and counted under a fluorescence microscope (BZ-X800E, KEYENCE). Viral titers were calculated and expressed as focus-forming units per milliliter per milligram (wet weight) of tissue (FFU/mL/mg).

Statistical analysis

Data from 2–4 independent experiments are presented as the mean ± SEM. Statistical comparisons were performed using a 2-tailed unpaired Student t test for pairwise analyses or 1-way ANOVA followed by Bonferroni post hoc test for multiple comparisons. Correlation analyses were performed using the Spearman rank correlation test. All statistical analyses were performed using GraphPad Prism 8.0 software (GraphPad Software), and a p value <0.05 was considered statistically significant.

RESULTS

Increased CD14⁺CD16⁻ monocytes in the liver tissue of children with BA correlate positively with liver function injury severity

Building on our previous scRNA-seq data showing monocyte accumulation in BA livers, ⁶ we found that children with BA had significantly higher peripheral monocyte percentages and absolute monocyte counts in routine blood tests (Supplemental Figure S1A, http://links.lww.com/HC9/C298). However, their functions and underlying mechanisms have not been elucidated. Using single-cell sequencing, we classified monocyte subpopulations into classical (CD14⁺CD16⁻), nonclassical (CD14⁻CD16⁺), and intermediate (CD14⁺CD16⁺) monocytes, consistent with established immunophenotypic definitions ²⁴ (Figures 1A and B, Supplemental Figure S1B, http://links.lww.com/HC9/C298). The proportion of CD14⁺ monocytes was significantly elevated in the BA liver tissue (Figure 1C and Supplemental Figure S1C, http://links.lww.com/HC9/C298). To further investigate the role of CD14⁺ monocytes in BA, we collected liver tissue and peripheral blood samples from patients with BA and choledochal cyst controls. Flow cytometry revealed a higher proportion of classical monocytes (CD14⁺CD16⁻) in BA liver tissues than in the controls, with no significant differences in nonclassical or intermediate monocytes (Figure 1D and Supplemental Figure S1D, http://links.lww.com/HC9/C298). Similarly, peripheral blood from patients with BA showed an increased proportion of classical monocytes compared with controls (Supplemental Figure S1E, http://links.lww.com/HC9/C298). Immunofluorescence confirmed an increase in CD14⁺CD16⁻ monocytes around the damaged bile ducts in the BA liver tissue (Figure 1E).

FIGURE 1.

Increased CD14⁺CD16⁻ monocytes in the liver tissue of children with biliary atresia correlate positively with liver function injury severity. (A) UMAP illustrating the 8 annotated cell populations of liver tissue–derived myeloid cells from BA and control samples. (B) Violin plots showing marker gene expressions across myeloid cell clusters. (C) Bar plots showing monocyte subset distributions across samples within different groups, with blocks denoting individual samples. The graph shows the percentage of classical monocytes between the 2 groups. (D) Gating scheme of monocyte subsets from livers of control (n=9) and BA (n=12) subjects. Classical monocytes (red): CD45⁺HLA-DR⁺CD14⁺CD16⁻; Intermediate monocytes (green): CD45⁺HLA-DR⁺CD14⁺CD16⁺; Nonclassical monocytes (blue): CD45⁺HLA-DR⁺CD14⁻CD16⁺. Bar graphs present absolute frequencies of monocytes as a percentage of total viable cells. (E) Immunofluorescence images of liver tissues from control and BA subjects, stained for Epcam (green), CD14 (cyan), CD16 (red), and DAPI (blue). Counts of CD14⁺CD16⁻ cells from 16 frames were used to calculate counts per HPF (bars, 25 µm). (F) Bubble plot showing the expression of proinflammatory cytokines in CD14⁺CD16⁻ monocytes. The color represented the average expression of each gene. (G) Correlation between the frequencies of CD14⁺CD16⁻ monocytes and clinical liver function parameters (TBA, ALP, ALT, and AST) in patients with BA (n=30). Correlations were analyzed using the Spearman rank correlation test. Error bars show mean ± SEM. **p<0.01, ***p<0.001 by unpaired Student t test. Abbreviations: BA, biliary atresia; HPF, high-power field; ns, not significant; TBA, total bile acid; UMAP, Uniform Manifold Approximation and Projection.

Expression analysis revealed elevated levels of proinflammatory cytokines (TNF, IL-1β, CXCL8, CCL3, CCL4, and CXCL2) and the NLRP3 inflammasome in classical monocytes from patients with BA (Figure 1F), suggesting their active involvement in the pathogenesis of cholestatic liver injury in BA. Correlation analysis revealed positive associations between the proportion of CD14⁺CD16⁻ monocytes and liver function markers, including total bile acid, ALP, ALT, and AST (Figure 1G). However, no significant association was observed with GGT, direct bilirubin, or total bilirubin (Supplemental Figure S1F, http://links.lww.com/HC9/C298). These findings associate classical monocytes with the severity of cholestatic liver injury in patients with BA.

Increased inflammatory monocytes and enhanced proinflammatory functions in the liver of the BA mouse model

To further explore the role of CD14⁺CD16⁻ monocytes in BA, we employed a well-established RRV-induced mouse model of BA, ²⁵ with technical refinements as previously established by our group. ¹⁷ Mass cytometry (CyTOF) analysis of hepatic CD45⁺ cells revealed a significant increase in monocytes in RRV-induced BA mice compared with controls, with no marked change in macrophage populations (Figures 2A and B). Flow cytometric analysis of Ly6c^high classical monocytes (counterpart to human CD14⁺CD16⁻ monocytes)^24,26 was quantified as CD45⁺CD11b⁺Ly6C⁺CD43⁻ (Supplemental Figure S2A, http://links.lww.com/HC9/C298), confirming marked elevation in hepatic Ly6C⁺CD43⁻ classical monocytes in RRV-induced BA mice. In contrast, Ly6C⁻CD43⁺ nonclassical monocytes and Ly6C⁺CD43⁺ intermediate monocytes remained unchanged (Figure 2C). We further analyzed macrophage changes in 6-day and 12-day RRV-induced BA mice. In 6-day BA mice, neither macrophage numbers nor apoptosis in the liver differed significantly from that in the control group. In contrast, 12-day BA mice exhibited significantly increased hepatic macrophage numbers and elevated apoptosis compared with the controls (Supplemental Figures S2B and S2C, http://links.lww.com/HC9/C298). This likely explains the absence of significant changes in the macrophages in our CyTOF analysis. Single-cell RNA sequencing indicated significant upregulation of Ly6C expression (Supplemental Figures S2D–S2F, http://links.lww.com/HC9/C298) and enrichment of proinflammatory signaling pathways in RRV-induced BA mouse classical monocytes (Figure 2D). In addition, the expression levels of proinflammatory cytokines (TNF, IL-1β, Cxcl2, Ccl4, Nlrp3, and Ccl5) were notably increased in these monocytes (Figure 2E). These findings suggest a proinflammatory role for classical monocytes in the progression of BA.

FIGURE 2.

Increased inflammatory monocytes and enhanced proinflammatory functions in the liver of the biliary atresia mouse model. (A) T-SNE illustrating the annotated cell population of concatenated CD45⁺ cells from saline- and RRV-treated mouse livers by mass cytometry (CyTOF) (left). Feature plots show the expression of the monocytes and macrophages on the same integrated cells (right). (B) The expression levels of Ly6C in monocytes and macrophages were measured and compared between the saline and RRV groups. (C) Gating scheme of monocyte subsets from the livers of the saline and RRV groups. Classical monocytes (red): CD45⁺CD11b⁺Ly6C⁺CD43⁻; Intermediate monocytes (green): CD45⁺CD11b⁺Ly6C⁺CD43⁺; Nonclassical monocytes (blue): CD45⁺CD11b⁺Ly6C⁻CD43⁺. Bar graphs present absolute frequencies of monocytes as a percentage of total viable cells. (D) The enriched and upregulated pathways in Ly6C⁺CD43⁻ monocytes of mice with RRV-induced BA were identified. (E) The bubble plot depicts the expression levels of proinflammatory cytokines in Ly6C⁺CD43⁻ monocytes from the saline and RRV groups, with colors indicating the average expression of each gene. Error bars show mean ± SEM. **p<0.01, ****p<0.0001 by unpaired Student t test. Abbreviation: ns, not significant; RRV, rhesus rotavirus.

Depletion of inflammatory monocytes significantly alleviates symptoms and biliary inflammation in BA mice

To investigate the contribution of the significantly expanded classical monocyte population to hepatic inflammation and bile duct injury in BA, we employed a targeted depletion strategy in our murine model. Neonatal mice received i.p. injections of the Anti-Ly6C antibody within 24 hours after birth, followed by RRV inoculation 4 hours later. Subsequent Anti-Ly6C antibody injections were administered intraperitoneally every 48 hours, and the analysis was performed on day 12 after infection (Figure 3A). Flow cytometric assessment confirmed the efficacy of anti-Ly6C treatment for depleting Ly6C⁺ monocytes. Importantly, while both classical (Ly6C⁺CD43⁻) and intermediate (Ly6C⁺CD43⁺) monocyte subsets were significantly reduced (Supplemental Figures S3A and S3B, http://links.lww.com/HC9/C298), depletion was more pronounced in classical monocytes than in their intermediate counterparts. This differential depletion allowed us to probe the specific role of classical monocytes while acknowledging the partial reduction in intermediate monocytes. Depletion resulted in improved overall appearance, increased body weight, enhanced survival rates, and reduced jaundice incidence in the RRV + Anti-Ly6C group compared with the RRV + Isotype control group (Figures 3B–E). Macroscopic evaluation revealed that livers from the RRV + PBS and RRV + isotype groups were characteristically yellowish, hardened, and displayed clear extrahepatic bile duct obstruction. In contrast, livers from the RRV + Anti-Ly6C group appeared similar to the vehicle controls, with a substantial reduction in visible bile duct obstruction (Figure 3F). Fluorescence imaging corroborated this finding, demonstrating near-intact fluorescence signals in the extrahepatic bile ducts of anti-Ly6C–treated mice, indicative of restored patency (Figure 3F). Histopathological analysis via hematoxylin and eosin staining further confirmed the significant attenuation of bile duct inflammation in the depletion group (Figures 3G and H). Moreover, liver function tests indicated severe impairment in RRV-infected mice receiving the isotype control antibody, whereas anti-Ly6C treatment led to notable functional recovery (Figure 3I). Collectively, these findings strongly implicate Ly6C⁺ monocytes, particularly the classical subset, as major drivers of pathogenesis and symptomatology in this RRV-induced BA model.

FIGURE 3.

Depletion of inflammatory monocytes significantly alleviates symptoms and biliary inflammation in biliary atresia mice. (A–I) Neonatal BALB/c WT mice were administered an i.p. injection of anti-Ly6C antibody (40 μg) within 24 hours postpartum. Four hours later, groups 1, 2, and 3 were inoculated with RRV (20 μL, titer 1.5×10⁶ PFU) to establish the BA mouse model. Anti-Ly6C antibody treatment was repeated every 48 hours until day 12, with assessments performed on day 12. (A) Schematic representation of the mouse model establishment. (B) Gross appearance of the BA model mouse. (C) Effect of anti-Ly6C antibody on mouse body weight. (D) Effect of anti-Ly6C antibody on mouse survival rate. (E) Effect of anti-Ly6C antibody on mouse jaundice rate. (F) Effect of anti-Ly6C antibody on the appearance of mouse liver and bile ducts. (G) Representative H&E staining of liver sections (bars, 50 μm). (H) Semiquantitative scoring of liver tissue inflammation. (I) Effect of anti-Ly6C antibody on mouse liver function. Error bars represent mean ± SEM. ***p<0.001, ****p<0.0001 by unpaired Student t test. Abbreviations: H&E, hematoxylin and eosin; ns, not significant; PFU, plaque-forming unit; RRV, rhesus rotavirus.

To directly investigate the cytotoxic capacity of classical monocytes toward biliary epithelial cells (BECs) and address potential contributions from other immune effectors, we performed direct ex vivo coculture experiments. Classical monocytes (Ly6C⁺CD43⁻) were isolated from RRV-induced BA model mice and age-matched controls. These cells were cocultured with BECs (EpCAM⁺) isolated from control mice (postnatal day 12, D12). Coculture with RRV-derived classical monocytes significantly induced cholangiocyte apoptosis and elevated the secretion of the proinflammatory cytokines TNF-α and IL-1β compared to coculture with control-derived monocytes (Figures 4A and B). To specifically assess the relative contribution of CD8⁺ T cells, which have been implicated in BA pathogenesis, parallel coculture experiments were performed using CD8⁺ T cells isolated from the same RRV-induced BA model mice. Strikingly, RRV-derived CD8⁺ T cells induced significantly less cholangiocyte apoptosis and lower levels of TNF-α and IL-1β than RRV-derived classical monocytes did (Figures 4A and B). This direct comparison highlights the potent and specific cytotoxic and proinflammatory effector functions of classical monocytes toward cholangiocytes in this context. Supporting a potential mechanism for monocyte-cholangiocyte crosstalk in RRV-induced BA, analysis of our single-cell RNA sequencing data using CellPhoneDB revealed BA-specific enrichment of interaction pairs between monocytes and cholangiocytes. Notably, the Tnf-α—Tnfrsf1a ligand-receptor pair, which is directly implicated in proapoptotic signaling and inflammation, was significantly upregulated in the RRV-induced BA cohort compared with that in the controls (Figure 4C). These data provide mechanistic insights into how monocytes may directly inflict damage on cholangiocytes.

FIGURE 4.

A cytotoxic function of classical monocytes defines cholangiocyte injury in experimental biliary atresia. (A) Representative flow cytometry plots (left) and quantification (right) of apoptosis in BECs (EpCAM⁺) isolated from naïve control mice (D12) after 48-hour coculture with liver-derived classical monocytes (Ly6C⁺CD43⁻) sorted from RRV-induced BA model mice or age-matched control mice. (B) Levels of TNF-α and IL-1β measured by ELISA in supernatants from 48-hour cocultures of BECs (EpCAM⁺, D12) with liver-derived classical monocytes (Ly6C⁺CD43⁻) sorted from RRV-induced BA model mice or age-matched control mice. (C) Analysis of ligand-receptor interactions using CellPhoneDB on single-cell RNA sequencing data from liver cells of RRV-induced BA and control mice. Expression levels of the significantly upregulated Tnf (ligand, monocytes)—Tnfrsf1a (receptor, cholangiocytes) pair specifically enriched in the RRV group compared to controls. Error bars show mean ± SEM. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001 by unpaired Student t test. Abbreviations: BA, biliary atresia; BEC, biliary epithelial cell; RRV, rhesus rotavirus; ns, not significant.

Therefore, our results demonstrated that classical monocytes are key effector cells that directly mediate bile duct epithelial injury in RRV-induced BA, exhibiting potent cytotoxic and proinflammatory capabilities exceeding those of CD8⁺ T cells in this model system. Furthermore, the therapeutic efficacy of anti-Ly6C–mediated depletion underscores the potential of targeting this monocyte subset to mitigate BA pathology.

NF-κB signaling pathway regulates classical monocyte-mediated bile duct inflammatory damage in a mouse model of BA

To elucidate the molecular mechanisms by which classical monocytes mediate bile duct inflammatory damage in BA, we performed bulk RNA-seq transcriptional analysis on Ly6C⁺CD43⁻ monocytes isolated from control and RRV-induced BA pups. Principal component analysis demonstrated distinct clustering of transcriptomes between Ly6C⁺CD43⁻ monocytes from the RRV-induced BA and control groups (Figure 5A). Differential gene expression analysis identified elevated transcript levels of key effector molecules in RRV-induced BA-derived Ly6C⁺CD43⁻ monocytes, including proinflammatory mediators (Nos2, IL1b, and Tnf), chemotaxis regulators (Ccr2 and Vcan), efferocytosis markers (Mertk and Cd68), and Ly6c1 (the gene encoding Ly6C protein that was transcriptionally upregulated in RRV monocytes), compared with those in control monocytes (Figure 5B). Kyoto Encyclopedia of Genes and Genomes–based enrichment analysis and Gene Set Enrichment Analysis demonstrated significantly enhanced NF-κB signaling in Ly6C⁺CD43⁻ monocytes from RRV-induced BA mice (Figures 5C and D). Further investigation revealed the upregulation of specific positive regulators of NF-κB signaling, such as Cxcl2, Tnf, IL1b, Ptgs2, Ccl4, Bcl2a1a, Cd40, Myd88, and Tnfrsf11a, in RRV-induced BA monocytes. Conversely, the expression of negative regulators, including Cxcl12 and Card10, was downregulated (Figure 5E). Importantly, these observations were consistently validated by RT-PCR in an RRV-induced BA model (Figure 5F).

FIGURE 5.

NF-κB signaling pathway regulates classical monocytes-mediated bile duct inflammatory damage in a mouse model of biliary atresia. (A) PCA of the RNA-seq data. (B) Heatmap of the expression levels of selected effector genes related to the classical monocyte function. Note that all sorted populations were Ly6C⁺ by surface protein. Elevated Ly6c1 transcript in RRV-group monocytes reflects increased inflammatory gene expression within this subset. (C) KEGG pathway analysis revealed significantly enriched signaling pathways in the classical monocytes of the RRV group. (D) GSEA highlighting NF-κB signaling pathway activation. (E) Heatmap of selected genes involved in the NF-κB signaling pathway. (F) mRNA expression levels of selected NF-κB pathway genes analyzed by qRT-PCR in Ly6C⁺CD43⁻ monocytes. (G) Flow cytometric analysis of phosphorylated NF-κB (p-p65) expression in hepatic Ly6C⁺CD43⁻ classical monocytes from saline- or RRV-treated mice. The y-axis shows the MFI for p-p65. (H) Flow cytometric analysis of phosphorylated NF-κB (p-p65) in hepatic CD14⁺CD16^- classical monocytes from control subjects or children with BA. The y-axis shows the MFI for p-p65. Error bars represent mean ± SEM. *p<0.05; **p<0.01, ***p<0.001，****p<0.0001 by unpaired Student’s t test. Abbreviations: GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; MFI, mean fluorescence intensity; ns, not significant; PCA, principal component analysis; RRV, rhesus rotavirus.

To validate these observations in vivo, we systematically evaluated NF-κB activation dynamics across distinct monocyte populations. In RRV-induced BA mice, Ly6C⁺CD43⁻ classical monocytes exhibited significantly elevated phosphorylated NF-κB (p-NF-κB) levels compared with controls (Figure 5G), whereas steady-state NF-κB expression remained unchanged in classical monocytes from both saline-treated and RRV-infected liver tissues (Supplemental Figure S4A, http://links.lww.com/HC9/C298). No significant alterations were observed in other monocyte subsets (Supplemental Figure S4B, http://links.lww.com/HC9/C298). These findings were corroborated in human studies, where classical monocytes isolated from BA patient liver tissues showed parallel p-NF-κB upregulation compared with our murine model (Figure 5H and Supplemental Figure S4C, http://links.lww.com/HC9/C298). Similarly, no significant differences in p-NF-κB expression were observed in CD14⁺CD16⁺ intermediate or CD14⁻CD16⁺ nonclassical monocytes from clinical samples (Supplemental Figure S4D, http://links.lww.com/HC9/C298). Collectively, these analyses identified constitutive NF-κB activation, specifically within classical monocytes, as a key signature associated with bile duct inflammation in BA, identifying this pathway as a critical therapeutic target.

The NF-κB inhibitor DHMEQ significantly ameliorates the symptoms and biliary inflammation in mice with RRV-induced BA

So far, our data demonstrate that classical monocyte depletion ameliorates BA pathology, potentially by modulating NF-κB signaling within these cells. We hypothesized that direct pharmacological inhibition of the NF-κB pathway could mitigate bile duct inflammation in BA. To test this hypothesis, we administered DHMEQ, a well-established inhibitor of NF-κB nuclear translocation,^27,28 to RRV-induced BA mice according to the regimen outlined in Figure 6A. Critically, to address the possibility that either anti-Ly6C antibody or DHMEQ treatment might exert their beneficial effects by suppressing RRV replication itself, thereby preventing the induction of the BA model, we quantified viral titers in extrahepatic bile duct tissues on day 7 after infection. Neither the anti-Ly6C antibody nor DHMEQ treatment significantly altered RRV replication levels compared with the relevant control groups (Figure 6B and Supplementary Figure S5A, http://links.lww.com/HC9/C298), confirming that therapeutic effects are attributable to modulation of the host immune response rather than inhibition of viral pathogenesis. To delineate the cellular targets of NF-κB activation in BA and DHMEQ action, we first investigated the single-cell RNA sequencing data. This analysis revealed significantly elevated NF-κB pathway activity, specifically within classical monocytes and BECs in RRV-infected livers compared to saline controls, while activity in HSCs remained relatively unchanged (Supplemental Figure S5B, http://links.lww.com/HC9/C298). Consistent with this expression profile, flow cytometric assessment of NF-κB activation confirmed that DHMEQ treatment significantly suppressed NF-κB pathway activity in classical monocytes and BECs, with more pronounced inhibition in classical monocytes and no significant effect in HSCs (Supplemental Figures S5C and S5D, http://links.lww.com/HC9/C298). This differential inhibition profile suggests that while DHMEQ can affect BECs, monocytes are the primary cellular targets in this context. Furthermore, consistent with the role of NF-κB in orchestrating chemokine production, we assessed whether DHMEQ treatment affects the recruitment of natural killer (NK) cells, a key immune population implicated in BA pathogenesis. ²⁹ Flow cytometric analysis on day 12 revealed a significant increase in hepatic NK cells (lineage⁻CD45⁺CD49a⁻CD49b⁺) in RRV-infected mice, which was markedly attenuated by DHMEQ treatment (Supplemental Figure S5E, http://links.lww.com/HC9/C298). This finding indicates that the therapeutic mechanism of NF-κB inhibition extends to normalizing the broader inflammatory landscape by reducing pathogenic recruitment of NK cells. Corresponding with this specific NF-κB inhibition profile, systemic DHMEQ-treated mice exhibited substantial improvements in clinical presentation, including increased body weight, enhanced survival, reduced jaundice incidence (Figures 6C–E), attenuated peribiliary inflammation (Figure 6F), and significantly improved liver function parameters (Figure 6G) compared with vehicle-treated RRV controls. It should be noted that the increase in mean body weight in the RRV + DHMEQ group beyond day 15 coincided with a reduction in survival and may therefore be influenced by survival bias, reflecting the growth of the remaining healthier pups rather than the entire original cohort.

FIGURE 6.

The NF-κB inhibitor DHMEQ significantly ameliorates the symptoms and biliary inflammation in mice with RRV-induced BA. (A–G) Neonatal BALB/c WT mice were treated with the NF-κB inhibitor DHMEQ (4 mg/kg body weight) via i.p. injection, starting 24 hours after RRV inoculation and repeated every 48 hours throughout the disease induction period. On day 12, comprehensive analyses were performed to evaluate the hepatic tissue inflammation and liver function parameters. (A) Schematic representation of mouse model establishment. (B) Quantification of RRV viral load in extrahepatic bile duct tissues at day 7 after infection. Neonatal mice were treated with anti-Ly6C antibody, DHMEQ, or respective controls following RRV induction. Viral titers were assessed by FFA. (C) Gross appearance of the RRV-induced BA mouse model. (D) The effect of DHMEQ on mouse body weight (left) and survival rate (right). Higher curves denote improved survival. (E) Effect of DHMEQ on the mouse jaundice rate. (F) Representative H&E staining of liver sections (bars, 50 μm) and semiquantitative scoring of liver tissue inflammation. (G) The effect of DHMEQ on mouse liver function. Error bars represent mean ± SEM. Statistical significance was determined using an unpaired Student t test: *p<0.05, ***p<0.001, ****p<0.0001. ns, not significant. Abbreviations: BA, biliary atresia; DHMEQ, dehydroxymethylepoxyquinomicin; FFA, fluorescent focus assay; H&E, hematoxylin and eosin; RRV, rhesus rotavirus.

To determine whether NF-κB inhibition, specifically in classical monocytes, mitigates cholangiocyte cytotoxicity—independent of the potential direct effects of DHMEQ on BECs, we conducted ex vivo coculture experiments. Classical monocytes (Ly6C⁺CD43⁻) from RRV-induced BA mice were pretreated with DHMEQ or DMSO for 24 hours, washed to remove the residual drug, and cocultured with naive EpCAM⁺ BECs from healthy control mice for 48 hours. DHMEQ-pretreated RRV-monocytes significantly reduced cholangiocyte apoptosis and secretion of proinflammatory cytokines TNF-α and IL-1β compared with DMSO-pretreated RRV-monocytes (Supplemental Figure S5F, http://links.lww.com/HC9/C298), indicating that NF-κB blockade specifically in monocytes attenuated their capacity to damage BECs. Collectively, these results underscore the pivotal role of NF-κB signaling in BA-related bile duct injury and highlight its therapeutic potential.

DISCUSSION

Classical monocytes, which exhibit NF-κB hyperactivation in both human BA livers and murine models, have been identified as key effectors of bile duct injury in the latter. In human patients, these cells are spatially enriched in injured portal tracts and secrete proinflammatory mediators (TNF, IL-1β, and CXCL8), consistent with their pathogenic role identified in mice. Critically, in the RRV-induced BA model, depleting monocytes or inhibiting NF-κB alleviated biliary injury and improved survival, indicating their pathogenic contribution.

In patients with BA, classical monocytes were enriched in the periportal regions near the damaged bile ducts, suggesting niche-specific roles. Their high phagocytic capacity and robust secretion of proinflammatory cytokines^30–32 contribute to immune cell recruitment and amplify inflammatory responses.^33–35 Unlike the heterogeneous functions of liver macrophages,^36,37 classical monocytes in BA (strictly defined as CD14⁺CD16⁻ in humans and Ly6C⁺CD43⁻ in mice) exhibit a conserved proinflammatory phenotype across species,^24,26 characterized by the NLRP3 inflammasome and NF-κB activation. In mice, this was further reflected at the transcript level by the elevated expression of activation-associated genes, such as Ly6c1. This extends prior findings on CCL2-CCR2–mediated recruitment by defining effector mechanisms. ¹⁴ The correlation between circulating classical monocytes and liver injury markers (total bile acid, ALP, ALT, and AST) further underscores their systemic relevance and shows promise as contextually informative biomarkers. In parallel, we observed a concomitant decrease in the nonclassical monocyte subset (Ly6C⁻CD43⁺ in mice) in the RRV model. Although a reduction in these patrolling cells may theoretically disrupt immune homeostasis, our functional depletion studies unequivocally identified the expanded classical subset, and not the diminished nonclassical subset, as the primary effector of biliary injury. This finding underscores the dominant pathological role of classical monocyte-driven inflammation in this experimental model.

NF-κB activation in classical monocytes from RRV-induced BA mice is a key pathogenic driver in this experimental model, rather than a bystander effect. Our data revealed BA-specific upstream regulators of canonical NF-κB signaling in classical monocytes. These molecules amplify NF-κB activity by bridging Toll-like and B-cell receptor pathways,^38,39 suggesting that microbial or endogenous ligands may initiate monocyte activation in this experimental context. Although Myd88 was upregulated in our RRV model, its known dispensability for ease of development ²² suggests that alternative pathways initiate the process, with NF-κB activation driven by other upstream activators. DHMEQ-mediated NF-κB inhibition reduces monocyte-driven inflammation and improves liver function and survival. However, incomplete prevention of mortality suggests that other parallel mechanisms may contribute to disease lethality. Notably, our dual approach, monocyte depletion versus NF-κB inhibition, provides complementary translational insights specific to the RRV model. While anti-Ly6C ablation broadly removes proinflammatory monocytes, DHMEQ targets their dominant signaling axis, offering a more precise strategy with fewer off-target effects. While our findings in the RRV model identify classical monocytes and NF-κB as critical mediators of inflammatory bile duct injury, the extent to which this mechanism operates in human BA, which may involve distinct or noninflammatory etiologies, requires further study.

Our findings align with and extend previous work on immune-mediated mechanisms in BA pathogenesis.^6,40–42 Although our study identified classical monocytes as direct effectors of biliary injury, their ablation may also indirectly modulate NK cell activity, a cell type implicated in BA pathogenesis.^43,44 Monocyte depletion likely reduces cytokine (eg, IL-12 and TNF-α) and ligand (eg, MIC/CD40) signals required for NK cell activation and cytotoxicity toward cholangiocytes. This synergizes with our observed reduction in direct monocyte-mediated epithelial damage, collectively mitigating bile duct obstruction. Although macrophages are important,^45–47 our findings emphasize the specific role of NF-κB in classical monocytes, suggesting that targeted anti-monocyte therapies may yield greater specificity. However, key questions remain. First, what recruits classical monocytes to the bile ducts? Although CCL2-CCR2 has been implicated, ¹⁴ our scRNA-seq data suggest that additional chemotactic signals (eg, CXCL8/CXCR1) warrant exploration. Second, how do monocyte-derived cytokines interact with the bile duct cells? TNF-α and IL-1β may directly induce ductular senescence or activate stromal fibroblasts,^48,49 perpetuating fibrosis—a critical area for future study. Third, although DHMEQ improves outcomes in mice, its long-term safety and efficacy in pediatric BA require rigorous evaluation, particularly given the pleiotropic roles of NF-κB in development.

Clinically, our study bridges the critical gap between immunopathology and actionable therapies for BA. The conserved NF-κB signature across species supports translational relevance, and the feasibility of pharmacological inhibition (via DHMEQ) or cellular targeting (via anti-Ly6C antibody) offers multiple therapeutic entry points. Although NF-κB inhibition holds therapeutic potential for BA, its safety profile in neonatal populations warrants rigorous evaluation. Future preclinical investigations should prioritize assessing DHMEQ as a post-Kasai portoenterostomy adjuvant therapy to minimize the systemic immunomodulatory effects. Combinatorial therapeutic strategies incorporating targeted anti-inflammatory agents may synergistically enhance efficacy and reduce toxicity profiles. Critically, this approach would enable precise intervention during the narrow therapeutic window of early fibro-inflammatory progression, thereby attenuating biliary damage without compromising the systemic immune competence essential for neonatal defense mechanisms.

In conclusion, this study establishes that classical monocytes exhibiting NF-κB hyperactivation associate with bile duct injury in patients with BA and functionally drive this pathology in a murine model, identifying this cellular and molecular axis as a promising therapeutic target.

Supplementary Material

hc9-10-e0941-s001.pdf ^{(2.2MB, pdf)}