Transcriptional analysis of murine biliary atresia identifies macrophage heterogeneity and subset-specific macrophage functions

Abstract

Introduction: Macrophages play an important role in disease progression of pediatric cholestatic liver disease, particularly biliary atresia (BA); however, the restorative versus pathogenic role for precise macrophage subsets remains poorly defined. We aimed to distinguish the transcriptional profiles and roles of defined macrophage subset(s) in murine BA.

Methods: We used multiparameter flow cytometry and RNA-sequencing analysis to profile recruited CD11b^hiCD64+ hepatic macrophages by cell surface expression of MHCII and Ly6c in the Rhesus rotavirus (RRV)-induced murine model of BA versus saline controls. Modulation of macrophage numbers via intra-peritoneal injections of clodronate-loaded liposomes was performed to determine the association between macrophage numbers and histologic injury (Ishak score).

Results: Ly6c+ macrophages demonstrated the greatest increase in numbers and percent of total macrophages in murine BA versus saline controls whereas MHCII+ macrophages decreased. Transcriptional changes in murine BA MHCII+ macrophages included reduced expression of the Kupffer cell gene signature, lower expression of genes involved in homeostatic processes, and increased expression of genes involved in inflammatory processes. Ly6c+ macrophages in murine BA showed increased expression for Hif1a and other genes involved in the cellular response to hypoxia. Among all subsets, the number of Ly6c+ macrophages exhibited the strongest correlation with severity of histologic liver injury by Ishak score.

Conclusions: Our data identify specific pathways upregulated in Ly6c vs MHCII+ macrophage subsets in murine BA. Transcriptional similarities between murine BA and human cholestatic macrophages may enable translation of future mechanistic studies to new macrophage subset-specific therapies.

Keywords: cholestasis; hepatic macrophages; innate immunity; obstructive cholangiopathy; pediatric liver disease.

Figure 1

Compositional changes in murine BA macrophage subsets. (A) Overview of experimental design for flow cytometry and RNA-sequencing analyses in murine BA mice and saline controls. (B, C) Increased F4/80+ macrophage infiltration is present in murine BA with associated hepatic necrosis. Hematoxylin and eosin staining shows inflammation and liver injury characterized by hepatic necrosis. (D, E) Flow cytometry analysis over the first 14 days of life shows changes in numbers of macrophage subsets in murine BA versus saline controls as defined by cell surface expression of MHCII and Ly6c. The average number of macrophages across replicates was calculated and the proportion of macrophages compared between experimental groups. At day of life 14, KCs and CD11b^hi MHCII+ macrophages comprised the majority of total macrophages in saline controls as compared to CD11b^hi Ly6c+ macrophages in murine BA (D). ANOVA with Bonferroni correction was performed and p-values < 0.05 for comparisons between experimental groups and prior time points within experimental groups are shown (E). DOL, day of life; DN, double negative; DP, double positive; RRV, Rhesus rotavirus.

Figure 2

Reduced transcriptional heterogeneity is present across murine BA macrophages. (A) Pearson correlation of expressed genes demonstrates similar transcriptional profiles between corresponding macrophage subsets in saline controls and murine BA. (B) Visualization of expressed genes by principal component analysis shows reduced heterogeneity among murine BA macrophages. (C) K-means clustering of 3,200 highly variable genes identified 6 gene clusters that differentiated saline macrophage subsets but were expressed at lower levels in murine BA macrophages. (D) Representative genes that differentiate each saline cluster are shown for both experimental models. (E) Gene ontology enrichment analysis for genes of each saline cluster identified distinct processes for each cluster. (F) Comparison of all saline gene clusters to previously defined differentially expressed genes (DEGs) of circulating monocytes and KCs showed greatest enrichment for KC genes in clusters 1-3 as compared to enrichment for monocyte genes in clusters 4-5 (31). n=4 for all transcriptional comparisons except DN subsets where n=3. * indicates adjusted p-value < 0.05 by DESeq. DN, double negative; DP, double positive; KC, Kupffer cell; Mono, monocyte; RRV, Rhesus rotavirus; VEGF, vascular endothelial growth factor.

Figure 3

Inflammatory processes are upregulated in murine BA macrophages. (A) The number of differentially expressed genes (DEGs) with increased expression in murine BA versus saline controls and the overlap between subsets (solid line) was calculated using the UpsetR package. Overall, MHCII+ and DP macrophage subsets showed the greatest number of increased DEGs. (B) Evaluation of the overlap between DEGs in gene sets 1-4 and previously identified genes in macrophages from a murine model of MASH (33) demonstrated the highest proportion of unique genes present in gene set 4 that defined murine BA Ly6c+ macrophages. (C) Immune signaling and pro-inflammatory genes increased in murine BA MHCII+ and DP macrophages included Cxcr4, Ccl12, and Cxcl16. (D) Genes present in set 4 that defined murine BA Ly6c+ macrophages that were also present in healthy KCs included Ccl24, Hmox1, and Mertk. (E) Set 4 genes not present in any MASH subsets included Hif1a, Nos2, C5ar1. (F) Evaluation of enriched processes from gene set 4 identified processes related to hypoxia and angiogenesis. n=4 for all transcriptional comparisons except DN subsets where n=3. * indicates adjusted p-value < 0.05 by DESeq. DN, double negative; DP, double positive; KC-H, healthy KC; KC-N, NASH KC; KN-RM, recruited macrophage occupying the KC niche; RM, recruited macrophage; RRV, Rhesus rotavirus.

Figure 4

Murine BA macrophages are transcriptionally similar to their human counterparts. (A) The top 20 genes from pediatric cholestatic macrophage subsets (11) were converted to their murine orthologs. Evaluation of the mean expression of these genes across murine BA macrophage subsets showed a high level of similarity between MLM and murine BA Ly6c+ macrophages, and LAM and murine BA DP and MHCII+ macrophages. (B) Characteristic genes for each previously identified pediatric cholestatic macrophage subset are visualized for our mouse models. n=4 for all transcriptional comparisons in murine experiments except DN subsets where n=3. * indicates adjusted p-value < 0.05 by DESeq. AM, adaptive macrophage; DN, double negative; DP, double positive; LAM, lipid associated macrophage; MLM, monocyte-like macrophage; RRV, Rhesus rotavirus.

Figure 5

Modulation of macrophage composition in murine BA demonstrates a direct association between Ly6c+ macrophages and severity of histology liver injury. (A) Murine BA mice were treated with either 3 intra-peritoneal injections of clodronate-loaded liposomes (n = 6) or PBS (n = 8). Outcome analyses were performed at day of life 12. (B) In murine BA mice given intra-peritoneal clodronate versus saline, the total number of CD11b^hi macrophages was directly related to the severity of histologic injury by total Ishak score (Pearson’s r = 0.768, p = 0.001). Among all subsets, the number of Ly6c+ macrophages was most strongly associated with liver injury by the Ishak score (Pearson’s r = 0.835, p < 0.001). (C) Representative hematoxylin and eosin staining is shown for a clodronate-treated mouse with high macrophage numbers and increased Ishak score (Left, purple triangle in panel B) versus low macrophages and Ishak score (Right, green triangle in panel B). White arrows show portal inflammation and black arrow shows lobular inflammation. DOL, day of life; PBS, phosphate buffered saline; RRV, Rhesus rotavirus.