Liver-restricted deletion of the biliary atresia candidate gene Pkd1l1 causes bile duct dysmorphogenesis and ciliopathy

Abstract

BACKGROUND AND AIMS:

A recent multicenter genetic exploration of the biliary atresia splenic malformation (BASM) syndrome identified mutations in the ciliary gene PKD1L1 as candidate etiologic contributors. We hypothesized that deletion of Pkd1l1 in developing hepatoblasts would lead to a cholangiopathy in mice.

APPROACH AND RESULTS:

CRISPR-based genome editing inserted loxP sites flanking exon 8 of the murine Pkd1l1 gene. Pkd1l1^Fl/Fl crossbred with alpha-fetoprotein-Cre expressing mice to generate a liver-specific intrahepatic Pkd1l1-deficient model (LKO). From embryonic day 18 through week 30, control (Fl/Fl) and LKO mice were evaluated with standard serum chemistries and liver histology. At select ages, tissues were analyzed using RNA-Seq, immuno-, and electron microscopy with a focus on biliary structures, peribiliary inflammation, and fibrosis. Bile duct ligation (BDL) for 5 days of Fl/Fl and LKO mice was followed by standard serum and liver analytics. Histological analyses from perinatal ages revealed delayed biliary maturation and reduced primary cilia, with progressive cholangiocyte proliferation, peribiliary fibroinflammation, and arterial hypertrophy evident in 7- to 16-week-old LKO versus Fl/Fl livers. Following BDL, cholangiocyte proliferation, peribiliary fibroinflammation and necrosis were increased in LKO compared to Fl/Fl livers.

CONCLUSIONS:

BDL of the Pkd1l1-deficient mouse model mirrors several aspects of the intrahepatic pathophysiology of BA in humans including bile duct dysmorphogenesis, peribiliary fibroinflammation, hepatic arteriopathy and ciliopathy. This first genetically-linked model of BA, the Pkd1l1 LKO mouse, may allow researchers a means to develop a deeper understanding of the pathophysiology of this serious and perplexing disorder, including the opportunity to identify rational therapeutic targets.

Graphical Abstract

INTRODUCTION

Biliary atresia (BA) is the most prevalent serious liver disease of infancy, where upwards of 75% will either die or require liver transplantation by the time they are 18 years of age (1–3). Even though BA is a prime focus of pediatric hepatologists and researchers worldwide, little progress has been made in understanding the causes and contributors to its etiopathogenesis (4, 5). Recently, several researchers have provided support for BA as a developmental cholangiopathy, primarily manifested by evidence of cholestasis at birth (6, 7). A detailed understanding of the pathophysiological contributors to BA is lacking, allowing for a wide variety of potential etiologic considerations including toxic, genetic, viral, and immunologic sources (5). Very recent work supports a focus on aberrant cholangiocyte biology as the cause of BA since isolated cholangiocyte organoids derived from BA patients have both impaired intercellular cholangiocyte tight junctions and a reactive ductular phenotype (8). These studies add to growing evidence that supports the etiologic origins of BA as a developmental or genetic disorder of the cholangiocyte, moreso than the concept that BA develops from exogenously acquired insults to an otherwise normally developing biliary tree.

Several groups from around the world, utilizing modern genome and exome sequencing technologies in BA patients, have reported contributions from ciliary and other gene variants (9–12). A focus on variants in ciliopathy genes in BA derives from several lines of evidence including the finding of aberrant cholangiocyte cilia in BA livers and the recognition that multiple genetic ciliopathy syndromes incorporate clinical features of fibrotic and cystic biliary tract diseases (13–18). Recently, the NIDDK-supported ChiLDReN consortium performed exome analysis of DNA derived from the subset of BA participants with biliary atresia splenic malformation (BASM) syndrome (12). This group, comprising ~10% of BA children, exhibits in addition to BA, laterality (heterotaxy) defects with cardiac, splenic, or intestinal manifestations (19, 20). Since there is a multi-organ developmental pattern in this subgroup, genetic causes may explain the concordance of biliary and extrahepatic laterality features. This study found several BASM participants with significant mutations in the ciliary gene PKD1L1 as a potential cause. Mutations in PKD1L1 have been associated with cardiac heterotaxy defects (21), hematologic disorders, and early prenatal demise, but were not known to lead to biliary tract disease (22). PKD1L1 is expressed in cholangiocytes, and mutations in PKD1L1 appear to be strong genetic candidates for the etiology of BASM. However, mouse models that can be used to test that hypothesis are currently lacking, as the complete Pkd1l1 knockout is embryonic lethal, albeit with clear evidence of heterotaxy (23). Given the lack of genetic mouse models of BA, we sought to pursue the findings of this human BASM genetic study by creating a mouse where Pkd1l1 is deleted in developing mouse cholangiocytes. If the livers of these conditionally Pkd1l1-deleted mice mimicked features of BA in infant human livers, then it would provide a means to explore both disease pathogenesis and potential efficacy of novel therapeutic approaches in this perplexing and serious disorder of infancy.

METHODS AND MATERIALS

Mouse Experiments:

All mouse experiments were approved by the Emory University Institutional Care and Use Committee (IACUC). CRISPR/Cas9 technology was used in C57BL/6J mice to insert loxP sequences in the intronic regions flanking exon 8 of Pkd1l1 (Fl/Fl). Intrahepatic Pkd1l1^{ΔExon8/ΔExon8} (LKO) mice were generated by crossing (Fl/Fl) mice with alpha-fetoprotein (Afp)-Cre expressing transgenic mice (See Supplemental Materials for detailed information on mouse lines) (24). To confirm intrahepatic localization of Cre expression, Rosa26^mTmG reporter mice, were employed (25) and crossbred to both Fl/Fl and LKO lines. Female livers were isolated for subsequent immunohistochemical analyses at embryonic day 18 (E18), postnatal day 30 (P30), week 7 (W7), W11, and W16. Bile duct ligation (BDL) was performed on W12 mice, and euthanized after 5 days, as previously described (26). See Supplemental Materials for additional mouse experimental details.

Histological and Immunohistochemical Analyses:

Female mouse liver tissues were isolated, fixed overnight in 4% paraformaldehyde, followed by sectioning, paraffin embedding, and processing for standard histology, immunohistochemistry, and immunofluorescence (27). Additional details, antibodies, and reagents can be found in Supplemental Table 1.

RNA-Seq:

Total RNA from the liver and isolated bile ducts in Fl/Fl and LKO mice was extracted using RNAeasy kit (Qiagen, Hilden, Germany). Purity and concentration were determined using a Nanodrop 2000 (Thermo-Fischer Scientific, Waltham, MA). Preparation of RNA library and transcriptome sequencing was conducted by Novogene Co., LTD (Beijing, China). The BA cohort used for gene-level comparisons was acquired from GSE159720 (28). Fl/Fl (3 female, 1 male) and LKO (4 female) W7 RNA-Seq data are available using accession number GSE201330. Gene set enrichment analysis (GSEA) Pre-ranked was performed using software provided by the Broad Institute (http://software.broadinstitute.org/gsea/index.jsp). Full results of differentially expressed genes and GSEA enrichment of the LKO versus Fl/Fl mRNA can be found in Supplemental Tables 2 and 3.

PCR & RT-qPCR:

RT-PCR of isolated female W30 Fl/Fl and LKO bile duct RNA was performed as previously detailed (29). RT-qPCR was performed as previously described (27). Relative fold change of gene expression was performed with the delta-delta CT method, using cyclophilin as an internal reference transcript (30). Primer sequences are provided in Supplemental Table 1.

Intrahepatic bile duct isolation:

30W Female Fl/Fl and LKO mice bile ducts were isolated with minor modifications as previously reported (31, 32).

Statistical Analysis:

Statistical comparisons were made using ANOVA for comparing multiple groups and standard t-tests wherever appropriate. An n=4 or more female Fl/Fl and LKO were used for every quantitative assay performed. Statistical analysis was performed using GraphPad Prism (version 9.3.1). More information can be found in Supplemental Materials.

RESULTS

Generation of liver-restricted Pkd1l1^{ΔExon8/ΔExon8} mice.

Whole-body Pkd1l1 gene deletion leads to situs abnormalities and embryonic lethality in mice, indicating that a mouse model capable of exploring the role of this gene in biliary function would require a liver-restricted conditional mouse knockout approach (23). LoxP sites were inserted using CRISPR-based technologies in the intronic regions flanking Exon 8 of the mouse Pkd1l1 gene (see Supplemental Materials for cloning and analytical details). Exon 8 of the mouse Pkd1l1 gene was excised between loxP sites in intrahepatic cells of the hepatoblast lineage by cross-breeding Pkd1l1^Fl/Fl mice (Fl/Fl) with alpha-fetoprotein (Afp) promoter/enhancer-Cre transgenic mice (Figure 1A) (24). Excision of Exon 8 results in a frameshift mutation after amino acid (aa) 511 leading to a novel STOP codon at aa 526. PCR analysis of cholangiocyte genomic DNA isolated from livers from Pkd1l1^Fl/Fl;Afp-Cre progeny revealed complete excision of Exon 8 in liver knockout (LKO) mice (Figure 1C). By restricting Cre expression to the hepatoblast lineage, expression of Pkd1l1 is absent in intrahepatic cholangiocytes in LKO livers.

Figure 1. Generation of an intrahepatic knockout of Pkd1l1.

(A) Schematic depiction of the Cre recombinase targeting sites for the Pkd1l1 allele. Exon 8 is excised by conditionally expressed Cre under the Afp promoter sequence causing a premature stop codon at amino acid 526. (B) Blood and livers of Fl/Fl and LKO mice were harvested at the indicated ages for serum chemistry, histology, and transcriptomic analyses. (C) PCR analysis using genomic DNA extracted from cholangiocytes isolated from Pkd1l1^Fl/Fl and LKO mice showing complete excision of Exon 8. (D) Afp-Cre driven recombination in 10-week-old Rosa26^mtmg cross-bred mouse livers revealed green fluorescence in hepatocytes and cholangiocytes (PanCK) in Rosa26^mtmg; LKO, but not Rosa26^mtmg; Fl/Fl cells. Green fluorescence represents Cre-expression. Red cells were Cre-naïve. D: Scale bar: 50μm. Fl/Fl, Pkd1l1^Fl/Fl; LKO, Pkd1l1^Fl/Fl;Afp-Cre; White (PanCK staining).

There was neither enhanced mortality nor signs of cardiac or laterality anomalies in LKO progeny. To verify intrahepatic localization of Cre expression, both Fl/Fl and LKO mice were crossbred to Rosa26^mTmG reporter mice (Figure 1D) (33). These mice exhibit excision in the expected cells derived from the hepatoblast lineage whereby Cre expression was found only within intrahepatic cholangiocytes in the LKO, and not Fl/Fl mouse livers. By 10 weeks of age, nearly all pan-cytokeratin staining intrahepatic bile duct cells were GFP+ Cre-expressors. Taken together, the development of the Pkd1l1 LKO line provided a mouse model able to explore the consequences of restricted absence of Pkd1l1 in cholangiocytes without the embryonic lethality or extrahepatic manifestations of Pkd1l1^−/− mice.

Defects in canonical developmental pathways in developing LKO livers.

In order to explore the relevance of intact Pkd1l1 signaling in developing cholangiocytes, we focused on prenatal (day E18) and early postnatal (30 days, P30) ages for histological and immunohistochemical analyses. Cellular proliferation was increased in E18 LKO livers as evidenced by increased numbers of proliferating cell nuclear antigen (PCNA) positive cells (Figures 2A, C). Cholangiocyte proliferation in particular was evident in E18 livers, as noted by increased pan-cytokeratin 8 & 19 (PanCK) and PCNA double-positive cells (Figure 2A, D). However, as mice aged to P30, the difference in proliferating cells between the LKO and Fl/Fl mice was attenuated (Figure 2C, D).

Figure 2. Duct dysmorphogenesis in Pkd1l1 intrahepatic knockout mouse livers.

(A) Representative images of PCNA (red), PanCK (white), and DAPI (blue) immunofluorescence staining of E18 Fl/Fl and LKO mice indicate increased cellular proliferation and aberrant peribiliary morphogenesis. (B) Images of P30 Fl/Fl and LKO mice stained for acetylated-alpha tubulin (Ac-Tub) (magenta), PanCK (white), and DAPI (blue). Magnified inset of the LKO bile duct highlights the absence of cilia. (C) Box plots quantifying the number of PCNA+ (proliferative cells) in Fl/Fl and LKO mouse livers. (D) Box plots quantifying cells positive for both PanCK and PCNA in whole liver sections from E18 and P30 Fl/Fl and LKO mice. (E) LKO cholangiocytes at E18 and P30 are significantly less ciliated compared to Fl/Fl. A, B: Scale bars: 50μm. **P < 0.01, ***P < 0.001. PCNA = Proliferative cell nuclear antigen, PanCK= Cytokeratin 8 & 19, Ac-Tub = Anti-Acetyl-alpha tubulin.

Intrahepatic Pkd1l1 absence results in cholangiocyte proliferation and ciliary defects.

One of the central features of mature cholangiocytes is the presence of a single apical primary cilium (34). Bile ducts in BA livers have reduced numbers of cells with primary cilia (14, 15). Immunostaining with the ciliary-specific acetylated-alpha tubulin (AcTub) antibody identified a reduction in the number of cilia in LKO cholangiocytes at E18 and P30 (Figures 2B, E). Taken together, Pkd1l1-deficient cholangiocytes exhibit significant early cholangiocyte proliferation, disruption of primary cilia formation, and impairments in bile duct development.

Fibroinflammatory cholangiopathy in week 16 LKO mouse livers.

The livers of syndromic and non-syndromic BA patients are characterized by peribiliary inflammation, fibrosis, and ductular proliferation (35). We sought to determine if similar features would develop in LKO adult mouse livers. Although there were no differences in serum liver chemistries between LKO and control mice (Supplemental Figure 1), significant peribiliary pathology developed in LKO compared to Fl/Fl livers. Firstly, Sirius red staining revealed an increase in collagen deposition and fibrosis in LKO versus Fl/Fl livers (Figure 3 A, B). Hematoxylin and Eosin (H&E) staining revealed increased numbers of inflammatory cells in the peribiliary region within LKO livers (Figure 3A, bottom panel). In particular, macrophage (F4/80+) and stellate cell (desmin+) infiltrations were significantly greater in LKO compared to Fl/Fl livers (Figure 3C & D). The number of cholangiocytes was increased in LKO versus Fl/Fl livers (Figure 3E), in agreement with the increased cholangiocyte proliferation noted at earlier developmental stages (Figure 2D). Finally, a decrease in cilia in the livers of LKO mice at week 16 complemented the findings for E18 and P30 ciliary staining and quantitation (Figure 2B, 3F). Scanning electron microscopy (SEM) was able to visualize representative ductular apical surfaces showing reduced or absent primary cilia in LKO livers (Figure 3G)(Supplemental Materials). In sum, LKO livers exhibit increased biliary proliferation, ciliary dysmorphogenesis, peribiliary inflammatory cell recruitment, and enhanced fibrosis. This indicates that Pkd1l1 is indispensable for proper intrahepatic bile duct formation and that its absence leads to a reactive ductular phenotype.

Figure 3. lntrahepatic fibroinflammatory and ciliary phenotypes in W16 Pkd1l1 LKO mouse livers.

(A) Increased peribiliary fibrosis and cellular infiltration in a portal triad from a 16W LKO mouse liver compared to Fl/Fl livers. Sirius Red (top), IF including PanCK (green), desmin (white), and DAPI (blue; middle), and H&E (bottom) were used to analyze female Fl/Fl and LKO mice. (B) lntrahepatic collagen content of LKO is significantly higher compared to Fl/Fl mouse livers. (C-E) The mean densities of F4/80+ cells (macrophages), desmin+ cells, and PanCK+ cells are increased in LKO versus Fl/Fl. (F) LKO cholangiocytes are significantly less ciliated compared to Fl/Fl. (G) Scanning electron microscopy at 10,000x magnification illustrates paucity of cilia (red arrowheads) in the apical surface LKO bile ducts. A: scale bars: 50μm. G: Scale bars: 1 μm. *P < 0.05, **P < 0.01, ****P < 0.0001. PanCK = Cytokeratin 8 & 19, H&E = Hematoxylin & Eosin.

Transcriptome analysis of LKO vs Fl/Fl livers.

RNA-seq analysis of liver samples from W7 Fl/Fl and LKO mice identified a total of 113 differentially-expressed genes (DEGs) (p-adj value <0.05) (GSE#201330, Supplemental Table 2, 3). There were 71 upregulated and 42 downregulated genes in LKO compared to Fl/Fl livers (Figure 4A). Volcano plot analysis indicated a reduction in cholangiocyte developmental transcription factors (Hhex, Onecut1) and marked upregulation of both metallothionein genes (Mt1, Mt2). Pre-ranked Gene Set Enrichment Analysis (GSEA) using the entire gene set indicated that downregulated genes were enriched for Wnt and TGF-ß signaling pathways, while upregulated genes were enriched for interferon alpha and gamma response pathways (Figure 4B). RT-qPCR of RNA derived from isolated bile ducts revealed increased expression in LKO cholangiocytes of sentinel genes that comprise the reactive ductular phenotype: Cxcl1, Vegfa, and Tnfα (Figure 4D) (18). These gene expression findings from both whole liver and isolated bile ducts indicate that Pkd1l1 deficiency in cholangiocytes induces pro-inflammatory and pro-fibrotic alterations in gene expression.

Figure 4. Activation of transcriptional signaling pathways in livers of Pkd1l1 knockout mice.

(A) Heat map indicating dysregulated genes in 7W Fl/Fl (n=4) and LKO (n=4) mouse livers. Red and blue indicate high and low gene expression, respectively. (B) Hallmark gene pathways enriched in Fl/Fl versus LKO were identified by GSEA. (C) Volcano plot of log₂-fold change and P values between control and knockout transcripts. Red dots indicate transcripts that had both a P-adj < 0.05 and log2fold change > 1. Grey dots are transcripts that remain relatively unaffected between Fl/Fl and LKO mice. (D) RT-qPCR of RNA purified from the isolated bile ducts of Fl/Fl and LKO mice indicate elevated levels of 3 genes central to the inflammatory arm of the reactive ductular phenotype—Cxcl1, Tnf-α and Vegfa. (E) Heat map of common differentially expressed genes (P-adj <0.05) in livers of human BA patients and LKO mice. *P <0.05.

In order to explore the possibility that gene expression changes in LKO mice reflect those from human BA livers, RNA-Seq results from a cohort of BA and control patients (GSE159720) were used as a model transcriptome for comparison to LKO mouse livers (28). This gene-level analysis revealed several areas of significant similarity with 9 DEGs shared amongst both the BA cohort and LKO data (Figure 4E). Of note, expression of Forkhead Box A2 (Foxa2) and Foxa1 was significantly decreased within both BA and LKO livers, in agreement with the proposed abnormal bile duct development in both cholangiopathies (Figure 4E). These results delineate that a knockout of intrahepatic Pkd1l1 in mice elicits transcriptional reprogramming that partially overlaps with that seen in BA.

Pkd1l1 knockout mice exhibit hepatic arterial hypertrophy.

While histological evaluation of BA livers provides evidence of progressive thickening of the hepatic arterial medial layer, this has not been a notable feature of cholestatic mouse models to date (36). We sought to determine if hepatic arterial hypertrophy was present in LKO livers. Both visually and via quantification of arterial wall thickness, LKO mouse livers revealed a pronounced increase in arterial thickness (Figure 5A, B). Supporting this finding is a significant increase in Fibroblast growth factor 1 (Fgf1) and Vascular cell adhesion molecule 1 (Vcam1) (Figure 5C), prominent markers of angiogenesis (37).

Figure 5. Arteriopathy of the LKO liver.

(A) PanCK (green), αSMA (red), and DAPI (blue) staining of a representative portal triad from a female Pkd1l1 16W Fl/Fl and LKO mouse liver identifies arterial hypertrophy in LKO livers. (B) Arterial wall thickness was significantly increased in female W16 LKO compared to Fl/Fl mouse livers. (C) Relative RNA expression of Fgf1 and Vcam1 was increased in LKO (n=4) versus Fl/Fl (n=4) acquired from W7 RNA-seq data; t-test. A: White arrows indicates artery. A: scale bar: 100 μm. *P < 0.05, **P < 0.01, ****P < 0.0001. αSMA = alpha-smooth muscle actin, PanCK = Cytokeratin 8 & 19.

Three-dimensional reconstructions of portal triads reveal in vivo biliary and arterial distinctions between LKO and Fl/Fl livers.

In order to better explore the arrangements of vascular and biliary portal structures in the LKO mouse, we created three-dimensional (3D) reconstructions of isolated portal triads from Fl/Fl and LKO livers (Figure 6). Reconstructive stacking of 20 consecutive slides (for a total z-axis distance of ~200 μm) was performed using a modified method previously described in Suzuki et al. 2020 (see Supplemental Materials)(38). Using anti-Smooth muscle actin (SMA) antibodies to identify vascular structures and anti-PanCK antibodies for bile ducts, it was apparent that both the bile duct structures and arteries in LKO livers are larger and more spatially separated from the portal venous structures than in Fl/Fl livers. This “increased space” seen in immunostaining 2-dimensional slices is more evident in the 3D reconstruction and is likely a consequence of infiltrating peribiliary inflammatory cells surrounding the LKO bile duct (Figure 6B, inset). To better understand the biliary and vascular arrangements of the reconstructed portal tracts, see videos (Supplemental videos 1 & 2).

Figure 6. 3D reconstructions of intrahepatic portal triads from Fl/FI and LKO mouse livers.

Serial consecutive 2D images of portal triads at 5–10 μm thick are reconstructed into 3-D models. Cholangiocytes (green) (PanCK), hepatic arteries (red) (αSMA), and portal veins (purple) (αSMA) are shown as intact signals. (A) Representative 3D images of a portal triad from an 11 W female Fl/Fl mouse liver. The inset image was taken from the indicated region in white dashes. (B) 3D images of a portal triad from an 11W female LKO mouse liver. The inset image (IF) was taken from the indicated white dash restricted region. A, B: top scale bar: 150μm, inlet scale bar: 25 μm. αSMA = alpha Smooth muscle actin, PanCK = Cytokeratin 8 & 19.

Biliary obstruction due to bile duct ligation (BDL) reveals enhanced peribiliary fibroinflammation, cholangiocyte proliferation, and senescence in LKO compared to Fl/Fl livers.

LKO livers have normal and patent extrahepatic bile ducts given the lack of expression of the Afp-Cre transgene in these cells due to their developmental origin distinct from those that contribute to intrahepatic bile ducts (39). To reveal potential differences in the response of Pkd1l1-expressing versus non-expressing intrahepatic cholangiocytes to static pressure and retention of bile after biliary tract obstruction (i.e., a central feature of BA), Fl/Fl and LKO mice were subjected to BDL for 5 days (40). After BDL, serum chemistries were increased to a similar extent in Fl/Fl and LKO mice (Supplemental Table 4). However, there were significant and distinct effects on biliary morphology and peribiliary fibroinflammation between the two genotypes in response to BDL. After BDL, there was a significant increase in fibrosis, macrophage and stellate cell infiltration, cholangiocyte proliferation, necrosis, and mRNA expression of inflammatory markers (Figure 7 A-G). Low-magnification sirius red images highlight the exacerbated differences seen in LKO BDL livers (Supplemental Figure 2). There were indications of biliary rupture and peribiliary necrosis secondary to BDL in LKO livers to a much greater degree than seen in Fl/Fl livers (Figure 7 A, H&E panels). As a result of Pkd1l1 loss, LKO cholangiocytes display an increased number of dual positive senescent (nuclear p16+) cholangiocytes compared to Fl/F (Figure 7 H, I). Further, LKO cholangiocytes also displayed a decrease in primary cilia, and enhanced arterial hypertrophy (Supplemental Figures 3, 4). In total, compared to Fl/Fl livers, LKO livers with absent intrahepatic cholangiocyte Pkd1l1 elaborated enhanced necrosis, cholangiociliopathy, cholangiocyte senescence, peribiliary fibroinflammation, and biliary proliferation after BDL, with significant histologic similarities to features seen in BA livers prior to alleviation of the obstruction after KPE (41).

Figure 7. Bile duct ligation (BDL) exacerbates LKO hepatic fibroinflammation and necrosis.

Female Fl/Fl and LKO mice were subjected to BDL for 5 days. (A) Sirius Red (top), IF including PanCK (green), desmin (white), & DAPI (blue; middle), and H&E (bottom) staining. Representative images displaying enhanced fibroinflammation and necrosis of LKO compared to Fl/Fl mouse livers after BDL. (B-F) Quantification fibrosis (Sirius red) macrophage infiltration (F4/80 staining), desmin+ cells, cholangiocyte proliferation (PanCK+ cells, and increased necrosis in LKO compared to Fl/Fl livers. (G) RT-qPCR of RNA purified from whole liver tissue of Fl/Fl and LKO BDL mice indicate elevated levels of Cxcl1, Tnf-α and Ifn-γ. (H, I) P16 staining identified an increase in senescent cholangiocytes. A: scale bar: 100 μm. *P < 0.05, **P <0.01, ****P < 0.0001. PanCK = Cytokeratin 8 & 19, CK19 = Cytokeratin 19, P16 = cyclin-dependent kinase inhibitor 2A, H&E = Hematoxylin & Eosin.

DISCUSSION

BA, the principal serious liver disease of infancy affecting ~1:5–15,000 live births, is without confirmed etiologies or agreed-upon pre-clinical models. These seminal issues contribute to several significant knowledge gaps: mechanisms of disease; pathogenic pathways; and the drivers of the broad spectrum of BA phenotypes (1, 4–6). Although there are murine models of BA that utilize post-natal viral infections or toxins to induce damage to normal murine cholangiocytes, there are currently no mouse models based upon genetic findings from BA patients (5). In this study, we present the liver-restricted Pkd1l1-deficient mouse as the first model of BA based upon genetic discoveries—one that exhibits several features of BA pathology, especially in the setting of distal biliary obstruction.

Generated through CRISPR/Cas9 introduction of loxP sites flanking Pkd1l1 Exon 8, and excision using Cre expression of the liver-specific Afp promoter, Pkd1l1 was conditionally knocked out in developing hepatoblasts, precursors to intrahepatic cholangiocytes. During development (E18 & P30), there was an increase in proliferative cholangiocytes and a reduction of ciliated cholangiocytes within intrahepatic ducts; similar to findings in BA livers, and other cholangiociliopathies (17, 18). As mice aged to 16 weeks, the fibroinflammatory phenotype seen in LKO mice was readily apparent with pathophysiological features specific to the peribiliary region. Data from LKO transcriptomes outlined significant differences between LKO and Fl/Fl mice, with notable pathways and gene expression patterns in LKO livers and isolated bile ducts that partially overlap with those seen in BA patients. In addition to biliary and peribiliary inflammatory features in this Pkd1l1-deficient model, there was an associated arterial hypertrophic phenotype, also seen in BA patients. Finally, by using BDL of LKO mice as a model to determine Pkd1l1-deficient cholangiocytes’ response to distal obstruction, we found markedly enhanced necrosis and fibroinflammation in LKO livers. Further, these LKO cholangiocytes displayed evidence of senescence and loss of cilia. Taken together, this new model based on genetic findings in BA patients demonstrated that disruption of Pkd1l1 expression led to a collection of pathophysiological features seen in BA: i) abnormal cilia, ii) senescence, iii) proliferation of cholangiocytes, iv) increased fibrosis, v) augmented recruitment of inflammatory cells, vi) hepatic arterial thickening, and vii) enhanced peribiliary necrosis (see schematic in Figure 8).

Figure 8. Schematic depiction of the fibroinflammatory biliary consequences in Pkd1l1-deficient cholangiocytes.

On the left side of the schematic is a normal Pkd1l1 expressing bile duct, mirrored by Pkd1l1-deficient bile duct on the right. Pkd1l1-deficient cholangiocytes have fewer primary cilia and are in a reactive ductular state, expressing various proinflammatory, angiogenic, and profibrotic molecules. (This figure was made with BioRender.com under a purchased license agreement.)

Recent data suggest that the pathophysiology of BA is primarily linked to a developmental or genetic defect of cholangiocytes (5, 6, 8), although the genetic underpinnings remain poorly understood. The lack of Mendelian inheritance and discordance among twins indicates a higher level of genetic complexity and incomplete penetrance (12). However, there are genetic diseases, including biliary tract diseases (e.g., Bardet-Biedl Syndrome and other cholangiociliopathies) that have complex inheritance patterns and BA may fall within that spectrum (18, 42). To explore potential monogenic causes of BA, we focused upon the findings from BASM genetic analyses, as the coexistence of multi-organ heterotaxic phenotypes and biliary dysmorphology had a significantly greater chance of discovering the contribution of a single gene defect to the cholangiopathy of BA. We believe that the pathophysiology evident in this mouse model, the Pkd1l1-deficient cholangiocyte, supports this line of thinking with regards to BA pathogenesis—that there are select causes of BA due to single genes.

It should be noted that the knockout of Pkd1l1 was restricted to intrahepatic cholangiocytes, and expression of Pkd1l1 was unaffected in the extrahepatic tree. Although the full spectrum of BA was not present, there was evidence of a reactive ductular cholangiopathy (Figures 2, 3, 7) with defective biliary morphogenesis and an enhanced fibro-inflammatory and necrotic response. Importantly, this ductular responsive phenotype was significantly more evident in response to obstruction and pressure placed upon Pkd1l1-deficient cholangiocytes after BDL (Figure 7). Molecularly, the reactive ductular state in Pkd1l1-deficient cholangiocytes likely led to a greater degree of peribiliary recruitment of inflammatory cells and fibrosis via an increase in the expression of genes like TNFα and Cxcl1. Evidence of a cholangiociliopathy in LKO mice is present (Figures 2, 3, and Supplemental Figure 3) and helps explain the developmental dysmorphology of the intrahepatic ducts. Together, histological and transcriptomic comparisons support this model as a cholangiopathy with pathogenesis similar to that of BA.

Often induced by cholangiocyte proliferation, cholangiopathies are followed by activation of healing and regenerative pathways in nearby fibroblasts, stellate cells, and resident macrophages, causing an increase in collagen deposition (fibrosis) (18). In the BDL LKO model, most of the necrotic patches were located adjacent to bile ducts, a response to intrahepatic bile acid accretion and leakage of bile acids into the parenchyma (43). Since cholangiocytes in LKO mice have a pro-inflammatory signature, this finding supports a general development of the reactive ductular phenotype, which further recruits immune cells. We believe that until we or others can study the consequences of a pan-biliary knockout of Pkd1l1, the employment of LKO with BDL is a candidate new model to explore the pathophysiology of BA, where one can then explore the responses of Pkd1l1-deficient intrahepatic cholangiocytes to biliary pressure from distal obstruction.

A large body of evidence supporting aberrant ciliary gene expression as a cause of fibrotic and fibrocystic liver disease provides the intriguing suggestion that the consequences of absence of Pkd1l1 signaling in this mouse model, combined with the human BASM PKD1L1 findings provide support for select ciliopathy conditions as significant genetic contributors to BA. While ciliary genes have been suggested to be involved in the etiology of BA, none have been tested empirically to this extent. Supporting the proposal that defective cilia are linked to BA pathogenesis, we demonstrated that LKO mice have decreased number of cilia, although the direct links between absent Pkd1l1 and reduced ciliary development remain to be understood. Among the possibilities are that since Pkd1l1 is shown in select models to be present in primary cilia and plays a role in ciliary calcium signaling (44), its absence could plausibly lead to aberrant intracellular signaling that induces the development of the reactive ductular phenotype (45). Further, repression of other members of the polycystin family (Pkd1 & Pkd2) causes an elongated/twisted cilia (46) rather than an overall decrease in cilia, as seen here. This hints at a potential role for Pkd1l1 in ciliary functioning in cholangiocytes, perhaps whereby its absence leads to aberrant signaling through the Pkd1l1 partner proteins Pkd2l1 or Pkd2. Further studies exploring the molecular consequences of absent Pkd1l1 in cholangiocytes may lead to discoveries of novel pathogenic pathways underlying cholangiociliopathies.

In sum, the cholangiocyte-deficient Pkd1l1 conditional knockout mouse model provides support for the genetic findings of PKD1L1 mutations as a genetic cause of the BASM syndrome. This first genetic mouse model of BA, a rapidly progressive developmental cholangiopathy without effective medical therapies, can be studied as a means to begin to unravel the complex reasons for BA’s profound fibroinflammatory pathogenesis. It is anticipated that future studies based upon this mouse will help improve our understanding and perhaps identify rational therapeutic targets to address the ongoing pathogenesis of this complex disease.

ABBREVIATIONS:

BA

Biliary Atresia

KPE

Kasai portoenterostomy

KO

Knockout

OR

Odds ratio

TB

Total bilirubin