Introduction

Hepatorenal fibrocystic disease (HRFCD) is a common inherited genetic disease that affects approximately 1:500 people and is characterized by cysts and fibrosis in the kidney and liver [1,2,3]. This disease is caused by mutations in proteins required for normal cilia function (Pkhd1, Pkd1, or Pkd2). Even when caused by a common genetic mutation, patients with HRFCD have varying disease severity and progression that is thought to be caused by environmental and genetic modifiers [4,5,6,7]. The phenotypic variability observed in patients can be mimicked using HRFCD mouse models inbred on different genetic backgrounds [8]. Importantly, there is a strong association between genetic background and immune cell subtypes (and their gene expression profiles) in mice and humans raising the possibility that genetic strain-driven alterations in immune cells may be contributing to the phenotype variability [9,10,11].

Current data indicate a strong association between macrophages and HRFCD [12,13,14,15,16,17,18]. The functional importance of macrophages in HRFCD is well established as general macrophage depletion reduces liver and kidney fibrocystic disease [16,17,18]. In these studies, the data suggested that M2-like macrophages were predominantly responsible for promoting cyst progression as macrophage depletion (using liposomal clodronate) during periods of maximal M2-like macrophage accumulation significantly attenuated cystic disease [19]. Further, inhibition of Arg1 activity, typically associated with M2-like macrophages, reduced cystic disease suggesting that distinct macrophage subtypes control the phenotypic outcome in the kidney [19]. However, in vivo, macrophages are more clearly defined based on their ontological origin (infiltrating vs resident) than on M1/M2 polarization status [20]. Infiltrating macrophages (CD11bhi, F4/80lo cells) originate from the hematopoietic lineage in the adult bone marrow and are recruited to tissues in response to injury or infection through cytokines, such as CCL2, binding to its cognate receptor (CCR2) expressed on the immune cell [21,22,23,24,25]. Previous work from our lab demonstrated the importance of infiltrating macrophages in regulating fibrocystic liver disease in Ift88Orpk FVB mice [26]. Further supporting the idea that infiltrating macrophages are pathogenic cells in HRFCD, two recent findings show that renal tubule-derived Ccl2 promotes macrophage accumulation and cyst formation in mouse models of cystic kidney disease [27, 28]. Surprisingly, pharmacological inhibition of CCL2 using bindarit did not reduce cyst formation in the kidney or liver of PCK rats, although the effect of bindarit on macrophage number in the kidney and liver was moderate [29]. Collectively, these data suggest that the variability in phenotypic outcome observed in human patients and in mouse models on different genetic backgrounds may be influenced by the type of macrophage present in the tissue.

Herein, we set out to define the association between infiltrating macrophage subtypes and the hepatic phenotypes observed in an HRFCD mouse model caused by cilia dysfunction (Ift88Orpk). Our data indicate that genetic strain strongly influences the subtype of infiltrating macrophage present in the liver during normal postnatal development (Ly6chi infiltrating macrophage in BALB/c vs Ly6clo infiltrating macrophage in C57BL/6) and in Ift88Orpk mice. Further, we compared infiltrating macrophage subtypes relative to the phenotypic outcomes in Ift88Orpk mice on these genetic backgrounds. While our data indicate there is a strong correlation between the hepatic phenotypes and the infiltrating macrophage subtype, we find that these associations are not causative of the phenotypic variability.

Materials and methods

Mice

Two, four, and eight week old C57BL/6J and BALB/c Ift88Tg737Rpw (hereafter referred to as Ift88Orpk) male and female mice were bred in-house. Animals were maintained in Association for Assessment and Accreditation of Laboratory Animal Care International-accredited facilities in accordance with Institutional Animal Care and Use Committee regulations at the University of Alabama at Birmingham (UAB). For all experiments, littermates were used as control mice. BALB/c Ift88Orpk CCR2−/− mice were obtained by crossing BALB/c Ift88Orpk mice with BALB/c CCR2−/− mice (generous gift of Dr. Nancy McNamara, University of California-San Francisco). Mice were fed a Purina High Energy Mouse Diet (Catalog #: 5LJ5) ad libitum.

Harvesting, fixation, and tissue processing

Mice were anesthetized with avertin (2,2,2-tribromoethyl alcohol, Sigma) and transcardially perfused with 20 mL of D-PBS. Following perfusion, mouse livers were weighed and separated into individual lobes. One lobe was used for flow cytometry, two lobes were snap-frozen in liquid nitrogen, one lobe was immersion fixed in 4% (wt/vol) paraformaldehyde 1 h at room temperature followed by cryoprotection in 30% (wt/vol) sucrose overnight at 4 °C, and one lobe was fixed in PFA overnight, dehydrated in 70% ethanol overnight, and embedded in paraffin. Paraffin-embedded tissue was sectioned at 5 µm and stained using hematoxylin and eosin (H&E). Picrosirius red staining was performed by the UAB Comparative Pathology Laboratory.

Picrosirius red quantification was performed using ImageJ software. Five 10X objective images from 2 to 8-week old C57BL/6 and BALB/c Ift88Orpk mice were used for the analysis. For each time point, a threshold value in which red staining was minimally detected in periportal regions in control mice by Image J software was determined and applied to all samples. Data from individual images were combined to determine the average for each mouse.

Quantification of expanded biliary structures was done using H&E stained sections and ImageJ software. Five 10X objective images from 2 and 8-week old C57BL/6 and BALB/c Ift88Orpk mice were used for the analysis. For each time point, a threshold value in which open biliary structures were considered to be holes was determined and applied to all samples. Non-epithelial structures were identified by a pathologist who was blinded to the groupings. Data collected from individual images were combined to determine the average for each mouse.

Flow cytometry

Following perfusion with PBS, one lobe of the liver was removed and put into RPMI 1640 on ice. Harvested liver tissue was mechanically disrupted through a 70-µm mesh (Falcon; BD Biosciences) yielding single-cell suspensions. Cells were centrifuged at 1300 rpm (500 × g) for 5 min and red blood cells lysed using ACK red blood cell lysis buffer (Quality Biological; 10128-802) at 37 °C for 5 min. Cells were centrifuged at 1300 rpm (220 × g), resuspended in 1 ml of 1% BSA containing Fc blocking solution (dilution 1:200), and incubated for 30 min on ice. Following cell counting with trypan blue, approximately 2 million live cells were stained for 30 min at room temperature with conjugated primary antibodies including PE rat anti-mouse CD45 (Catalog #: 12-0451-81, 30-F11, Thermo-Fisher), eFluor®450 rat anti-mouse F4/80 (Catalog #:48-4801, BM8, eBioscience), APC rat anti-mouse CD11b (Catalog #:17-0112, M1/70, eBioscience), APC/Cy7 rat anti-mouse Gr-1 (Catalog #:557661, RB6-8C5, BD Pharmingen), and PerCP-cy 5.5 rat anti-mouse Ly6c (Catalog #:560525, clone: AL-21, BD Pharmingen). Cells were washed with 1% BSA, centrifuged at 1300 rpm, and fixed with 2% PFA at room temperature for 30 min. Cells were washed with 1% BSA, centrifuged at 1300 rpm, and resuspended in 1X PBS. After immunostaining, cells were analyzed on a BD LSRII flow cytometer. Data analysis was performed using FlowJo v10 software.

Fluorescence-activated cell sorting of cholangiocytes and macrophages for RNA sequencing

For flow sorting of individual cell populations from the whole liver, tissue was harvested and digested in RPMI containing 0.5 mg/ml collagenase with 100 U/ml DNase type I for 30 min at 37 °C with agitation. Single-cell suspensions were obtained and cells were stained as described above. For isolation of cholangiocytes, we used dolichos biflorus agglutinin (DBA) (RL-1032, Vector Laboratories), a well-established, cell surface marker of bile duct cholangiocytes [30]. Cells were sorted into individual tubes containing buffer RLT (Catalog #:74104 Qiagen, Hilden, Germany) to lyse the cells. Ultrapure RNA was isolated using the RNeasy kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. The quality of RNA was determined using the Agilent 2100 bioanalyzer and only used if the RNA integrity number was above 7. We combined an equal amount of RNA from 6 to 8 individual mice per genotype (equal number of males and females) for RNA sequencing studies. Thus, the RNA sequencing data presented is an average of 6–8 biological replicates per group. The NEBNext mRNA library prep kit for Illumina with ribosome reduction was used to generate the cDNA library, which was subsequently sequenced using Illumina NextSeq500 with paired-end 75 bp sequencing and aligned to the University of California, Santa Cruz (UCSC) GRCm38/mm10 reference genome using the STAR software package [31]. For each sample, at least 25 million reads were obtained. Following alignment, HTSeq-count (version 0.9.1) was used to count the number of reads mapping to each gene [32]. Fragments per kilobase of transcript per million mapped reads (FPKM) and pairwise differential expression were calculated using the Cufflinks suite (https://github.com/cole-trapnell-lab/cufflinks) and HTSeq-Count/DESeq2 was used to identify the most highly variable genes between groups.

RNA isolation from whole liver tissue and qRT-PCR

For qRT-PCR analysis of whole liver tissue, snap-frozen liver samples were thawed and ~25–50 mg of tissue was minced and disassociated in TRIzol (Thermo-Fisher, Waltham, MA). The tissue was homogenized using a 22 gauge needle followed by removal of tissue debris by centrifugation at 14,000 rpm for 5 min. The supernatant was removed, RNA was isolated, transcribed into cDNA, and qRT-PCR performed using TaqMan real-time PCR. The following probes were used: Col3a1 (Mm01254476_m1), Col1a2 (Mm00483937_m1), Krt19 (Mm00492980_m1), and Hprt (Mm00446968_m1).

Statistics

Data were presented as mean ± SEM. ANOVA and Student t-tests were used for statistical analysis, and differences were considered significant for P values less than 0.05. Outliers were excluded based on the Grubbs outlier test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Results

Infiltrating macrophage subtype and gene expression are influenced by genetic strain

To determine whether genetic strain influences infiltrating macrophage subtypes, we performed flow cytometry analysis of infiltrating macrophages isolated from the liver of C57BL/6 and BALB/c inbred mice under uniform environmental conditions. For these studies, we performed flow cytometry analysis at 2, 4, and 8 weeks of age to understand how infiltrating macrophage subtypes change as a function of time. Infiltrating macrophages were further sub-gated based on differential expression of Ly6c, a cell surface protein previously reported to be associated with the rate of fibrotic disease in the liver (Ly6chi macrophages promote fibrosis) [33]. Our data show that the absolute number (as determined by the percentage of the total, live, single cells in the liver) of Ly6chi infiltrating macrophages was not different between groups at any time point (Fig. 1A, B). In contrast, livers from C57BL/6 mice had a significant increase in the number of Ly6clo infiltrating macrophages compared to BALB/c livers at all time points (Fig. 1A, B). Analysis of Ly6chi:Ly6clo ratio in both strains indicates that the majority of infiltrating macrophages in wild type (WT) BALB/c livers at 2, 4, and 8 weeks of age are Ly6chi (~70% Ly6chi) whereas C57BL/6 livers have an approximately equal distribution of Ly6chi and Ly6clo macrophages during the same time period (Fig. 1A, C). The altered ratio of Ly6chi:Ly6clo infiltrating macrophages across time points was driven by differences in the total number of Ly6clo infiltrating macrophages.

Fig. 1: Genetic strain influences the subtype of infiltrating macrophage found in the liver during postnatal development.
figure 1

A Representative flow cytometry images showing Ly6c expression in infiltrating macrophages from 2, 4, and 8-week old WT C57BL/6 and BALB/c mice. B Quantification of the number of each cell type as a percentage of total cells for each time point is shown as average ± SE. N = 6–10 mice. *P < 0.05. C Quantification of the percentage of infiltrating macrophages expressing high (left) or low (right) levels of Ly6c is shown for each genetic strain at the indicated time points. Data are shown as the mean ± SE. *P < 0.05. D Heatmap of RNAseq data obtained from 4-week old WT C57BL/6 and BALB/c Ly6chi and Ly6clo macrophages isolated from the liver showing the top 200 DEGs. E Heatmap analysis of macrophage-specific genes in WT C57BL/6 and BALB/c macrophages isolated from the liver. F List of pathways enriched in BALB/c and C57BL/6 infiltrating macrophages based on RNAseq analyses.

Full size image

Along with the quantity and subset distribution of infiltrating macrophages, we evaluated whether there were differences in gene expression when comparing Ly6chi and Ly6clo infiltrating macrophages across genetic strains. Bulk RNA sequencing analysis was performed on WT Ly6chi and Ly6clo macrophages isolated from liver tissue of C57BL/6 and BALB/c mice at 4 weeks of age. Heatmap analysis and hierarchical clustering of the top 200 differentially expressed genes (DEGs) indicate that Ly6clo and Ly6chi infiltrating macrophages segregate together independent of genetic strain (Fig. 1D). A closer examination of macrophage-associated genes in each population revealed that expression of genes associated with extracellular matrix production and fibrosis including Tgfb1, Fn1, and Thbs1 are increased in BALB/c Ly6chi macrophages compared to C57BL/6 Ly6chi macrophages (Fig. 1E). We also observed increased expression of antigen presentation and processing genes (H2-Aa, Cd74, Itgax) in BALB/c Ly6clo infiltrating macrophages compared to their C57BL/6 counterparts (Fig. 1E). Pathway analysis of genes enriched in infiltrating macrophages from each strain reveals that BALB/c infiltrating macrophages have an enrichment of genes associated with cytokine production and wound healing while C57BL/6 infiltrating macrophages have an enrichment of genes associated with phagocytosis and endosome formation (Fig. 1F). These data indicate that the ratio and gene expression of Ly6chi and Ly6clo infiltrating macrophages is different between genetic strains and thus, could differentially influence disease progression in Ift88Orpk mice.

Genetic strain influences infiltrating macrophage recruitment in Ift88 Orpk mice

To address how infiltrating macrophage subtypes changed in the context of HRFCD, we analyzed infiltrating macrophages in the Ift88Orpk mouse model of HRFCD by flow cytometry. Ift88Orpk mice have hypomorphic levels of Ift88 leading to short, malformed, dysfunctional primary cilia and HRFCD [2]. These analyses were done on mice at 2 and 4 weeks of age in order to understand how early changes in infiltrating macrophage subtype recruitment correlated with late-stage disease. In 2 and 4-week old mice, we did not observe any significant difference in the number of Ly6clo macrophages in either Ift88Orpk strain compared to controls at any time point (Fig. 2A–D). In contrast, there was a significant increase in the number of Ly6chi infiltrating macrophages in BALB/c Ift88Orpk livers compared to C57BL/6 Ift88Orpk and BALB/c control livers (Fig. 2A–D). Similar to their WT counterparts, the ratio of Ly6chi:Ly6clo infiltrating macrophages was increased in BALB/c Ift88Orpk mice compared to C57BL/6 Ift88Orpk mice (Fig. 2E, F). Since Ly6chi infiltrating macrophages in the liver are associated with inflammation and fibrosis and Ly6clo infiltrating macrophages are associated with tissue repair and matrix degradation [33], our data suggest that BALB/c mice may be more prone to developing hepatic fibrosis due to the increased proportion of Ly6chi infiltrating macrophages while C57BL/6 mice may be resistant to hepatic fibrosis due to the increased proportion of Ly6clo infiltrating macrophages in the liver.

Fig. 2: Accumulation of infiltrating macrophage subtype in Ift88Orpk mice is influenced by genetic strain.
figure 2

A, B Representative flow cytometry plots of Ly6chi and Ly6clo macrophages from (A) 2-week old or (B) 4-week old control and Ift88Orpk C57BL/6 and BALB/c mice. C, D Quantification of the number of Ly6chi and Ly6clo infiltrating macrophages as a percentage of total cells in (C) 2-week old and (D) 4-week old livers. Data are plotted as the average ± SE. *P < 0.05 **P < 0.01. E, F Quantification of the percentage of infiltrating macrophages expressing high (left) or low (right) levels of Ly6c is shown for each genetic strain at (E) 2 and (F) 4 weeks of age. Data are shown as the mean ± SE. *P < 0.05; **P < 0.01.

Full size image

Infiltrating macrophage subtypes correlate with the severity of biliary hyperplasia and fibrosis in Ift88 Orpk mice

To test the hypothesis that differences in the ratio of infiltrating macrophage subtypes at early stages of disease correlate with late-stage phenotypic outcome, we analyzed the liver phenotype across the two inbred strains at 2, 4, and 8 weeks of age. H&E staining of 2-week-old Ift88Orpk liver sections showed that the liver pathology and severity of biliary hyperplasia (as measured by bile duct area/portal area) was similar between the two genetic strains at this time point (Fig. 3A, B). To confirm a similar biliary pathology between strains at 2 weeks of age, we performed qRT-PCR for Krt19, a marker of bile duct cholangiocytes [34], on whole liver tissue isolated from control and Ift88Orpk mice on the C57BL/6 and BALB/c backgrounds. Our data show that there is no difference in the expression of Krt19 mRNA in 2-week-old Ift88Orpk mice between genetic strains, confirming our histological analyses (Fig. 3C). In contrast to what was observed in 2-week old Ift88Orpk mutants, H&E staining of 8-week old mice shows the expansion of biliary regions in all Ift88Orpk mice with C57BL/6 Ift88Orpk mice having increased biliary hyperplasia compared to BALB/c Ift88Orpk mice (Fig. 3A, B). In agreement with these data, analysis of whole liver mRNA shows that Krt19 expression was higher in 8-week old C57BL/6 Ift88Orpk mice compared to 8-week old BALB/c Ift88Orpk mice (Fig. 3C). Four-week-old mice appear to have an intermediate phenotype in both strains.

Fig. 3: C57BL/6 Ift88Orpk mice have increased bile duct expansion in the liver compared to BALB/c Ift88Orpk mice.
figure 3

A Control and Ift88Orpk livers were harvested at the indicated time points and stained with hematoxylin and eosin. A representative 10X objective image from each mouse is shown. N = 5–8 per group. B Quantification of biliary hyperplasia (bile duct area/portal area) in 2 and 8-week-old C57BL/6 and BALB/c Ift88Orpk mice are shown as average ± SE. **P < 0.01. C qRT-PCR analysis of Krt19 mRNA in whole liver tissue from 2, 4, and 8-week-old control and Ift88Orpk C57BL/6 and BALB/c mice. Data are shown as the average ± SE. *P < 0.05.

Full size image

Since patients with HRFCD often develop severe hepatic portal fibrosis, we also analyzed fibrosis in C57BL/6 and BALB/c mice by performing qRT-PCR analysis of Col1a2 and Col3a1 from whole liver tissue isolated at 2, 4, and 8 weeks of age followed by confirmation of RNA data using picrosirius red staining and quantification. Similar to our previous finding in Ift88Orpk FVB mice [26], 2 weeks old C57BL/6 and BALB/c Ift88Orpk mice do not have increased levels of the extracellular matrix genes Col1a2 and Col3a1 or picrosirius red positive area compared to control mice (Fig. 4A–D) [26]. In contrast, analysis of 4-week old Ift88Orpk mice shows that BALB/c Ift88Orpk mice have a significant increase in Col1a2 and Col3a1 gene expression compared to C57BL/6 Ift88Orpk mice and BALB/c control mice (Fig. 4A, B). Increased expression of Col3a1 was also observed in 8-week old BALB/c Ift88Orpk mice compared to 8-week old Ift88Orpk C57BL/6 and BALB/c control mice although not to a significant level (Fig. 4B). The increased level of fibrosis in BALB/c Ift88Orpk mice compared to C57BL/6 Ift88Orpk mice was confirmed using picrosirius red staining and quantification (Fig. 4C, D). Collectively, these data indicate that C57BL6 mice have increased biliary hyperplasia while BALB/c Ift88Orpk mice have increased periportal fibrosis.

Fig. 4: BALB/c Ift88Orpk mice have increased hepatic fibrosis compared to C57BL/6 Ift88Orpk mice.
figure 4

A, B qRT-PCR analysis of (A) Col1a2 and (B) Col3a1 mRNA in whole liver tissue from 2, 4, and 8-week old control and Ift88Orpk C57BL/6 and BALB/c mice. Data are shown as the average ± SE. *P < 0.05. C Control and Ift88Orpk livers were harvested at the indicated time points and stained with picrosirius red. A representative 10X objective image from each mouse is shown. N = 5–8 per group. D Quantification of picrosirius red area for 2 and 8-week old C57BL/6 and BALB/c Ift88Orpk mice are shown as average ± SE. *P < 0.05.

Full size image

Genetic strain influences infiltrating macrophage gene expression in Ift88 Orpk mice

To understand how infiltrating macrophage subtypes may be influencing phenotypic outcomes in the liver of Ift88Orpk mice, we performed heat map analysis of RNA sequencing data from Ly6chi and Ly6clo infiltrating macrophages that were isolated from WT and Ift88Orpk mice on the C57BL/6 and BALB/c backgrounds. Sequencing and analyses were done on 4-week old mice with intermediate phenotypes so that we could identify changes in infiltrating macrophage gene expression prior to the presence of severe disease. A broad comparison of gene expression between C57BL/6 and BALB/c WT and Ift88Orpk Ly6chi (top) and Ly6clo (bottom) infiltrating macrophages indicates that Ift88Oprk mice have several DEGs compared to their respective WT controls (Fig. 5A). Since C57BL/6 Ift88Orpk mice have an increased ratio of Ly6clo infiltrating macrophages and BALB/c Ift88Orpk mice have an increased ratio of Ly6chi infiltrating macrophages, we performed a GO pathway analysis of genes that were enriched in Ly6clo macrophages from C57BL/6 Ift88Orpk mice and Ly6chi macrophages from BALB/c Ift88Orpk mice to gain insight into their function. Our data show that Ly6clo macrophages isolated from C57BL/6 Ift88Orpk mice have an enrichment of genes associated with the vascular wall, angiogenesis, and endothelial development (green arrows, Fig. 5B). We focused on these biological processes because several of the genes found within, including Vegf, promote cholangiocyte proliferation and bile duct expansion in the liver, thereby matching our observed liver pathology in these mice [35]. In contrast, Ly6chi macrophages from BALB/c Ift88Orpk mice have enrichment of genes associated with the inflammatory response, cytokine production, and response to wounding, similar to their WT counterparts (blue arrows, Fig. 5B).

Fig. 5: BALB/c infiltrating macrophages express pro-fibrotic genes while C57BL/6 infiltrating macrophages express genes associated with cholangiocyte proliferation.
figure 5

A Heatmap analysis of RNA sequencing data obtained from 4-week old WT and Ift88Orpk C57BL/6 and BALB/c mice. The top 200 DEGs for Ly6chi and Ly6clo macrophages are shown. B GO pathway analysis of the genes enriched in C57BL/6 Ly6lo macrophages and BALB/c Ly6chi macrophages. Green arrows (top panel) indicate pathways containing genes associated with cholangiocyte proliferation and biliary hyperplasia. Blue arrows (bottom panel) indicate pathways containing genes associated with extracellular matrix production and fibrosis. C, D Heatmap analysis of RNA sequencing data from Ly6chi and Ly6clo macrophages showing expression of genes known to (C) promote cholangiocyte proliferation or (D) promote extracellular matrix production and fibrosis.

Full size image

To further investigate how infiltrating macrophage subtypes may be influencing phenotypic outcome (i.e., proliferation-dependent bile duct expansion or fibrosis), we analyzed the expression of specific cytokines that are known to promote either cholangiocyte proliferation and bile duct expansion or extracellular matrix production and fibrosis in infiltrating macrophage subtypes. Our analysis shows that Ly6clo macrophages from C57BL/6 WT and Ift88Orpk mice have increased Vegfc transcripts compared to their Ly6chi counterparts whereas expression of other proliferative cytokines was minimal in these cells and virtually absent in Ly6chi macrophages (Fig. 5C). Analysis of pro-fibrotic growth factor gene expression reveals that BALB/c Ly6chi and Ly6clo WT and Ift88Orpk infiltrating macrophages expressed these genes (Fig. 5D). However, expression of Tgfb1, a gene associated with extracellular matrix production and fibrosis in the liver [36, 37], was enriched in Ly6chi infiltrating macrophages compared to Ly6clo macrophages (Fig. 5D).

Genetic strain influences epithelial cell response to environmental cytokines in Ift88 Orpk mice

In these studies, we also analyzed how ciliated, bile duct epithelial cells (cholangiocytes) respond to alterations in environmental cytokines by performing RNA sequencing and analysis of cholangiocytes isolated from 4-week-old WT and Ift88Orpk mice on the C57BL/6 and BALB/c background. Our data identified several changes in gene expression patterns between control and Ift88Orpk cholangiocytes on both the C57BL/6 and BALB/c background (Fig. 6A). Pathway analysis of the genes that were enriched in cholangiocytes isolated from each Ift88Orpk strain shows that C57Bl/6 Ift88Orpk cholangiocytes have an enrichment of genes associated with VEGFR signaling pathways (green arrows, Fig. 6B) whereas cholangiocytes isolated from BALB/c Ift88Orpk mice have an enrichment of genes associated with extracellular matrix organization and response to wounding (blue arrows, Fig. 6B). VEGFR signaling is known to promote bile duct expansion in the liver [35], which is consistent with our data showing that C57BL/6 mice have enhanced bile duct pathology and increased Vegfc transcripts in Ly6clo infiltrating macrophages.

Fig. 6: BALB/c Ift88Orpk cholangiocytes express genes associated with a mesenchymal and fibrotic phenotype while C57BL/6 Ift88Orpk cholangiocytes express receptors associated with cholangiocyte proliferation.
figure 6

A Heatmap of the top 200 DEGs from WT and Ift88Orpk cholangiocytes isolated from C57BL/6 and BALB/c mice. B GO pathway analysis of the genes enriched in C57BL/6 and BALB/c Ift88Orpk cholangiocytes. Green arrows (top panel) indicate pathways containing genes associated with cholangiocyte proliferation and biliary hyperplasia. Blue arrows (bottom panel) indicate pathways containing genes associated with extracellular matrix production and fibrosis. C–F Heatmaps showing gene expression of (C) cholangiocyte receptors that bind to pro-fibrotic growth factors, (D) mesenchymal markers, (E) receptors known to promote or inhibit cholangiocyte proliferation, and (F) chemoattactants that promote immune cell recruitment. In panel (E), receptors that inhibit cholangiocyte proliferation are indicated by the red bar on the left side of the graph whereas receptors that promote cholangiocyte proliferation are indicated by a green bar on the left side of the graph.

Full size image

To further investigate ligand:receptor crosstalk between infiltrating macrophage subtypes and cholangiocytes that may cause the divergent phenotypic outcome in Ift88Orpk mice from each genetic background, we analyzed gene expression of receptors known to drive extracellular matrix production, mesenchymal transition, and cholangiocyte proliferation in the RNA sequencing data. Analysis of pro-fibrotic growth factor receptor expression shows that C57BL/6 Ift88Orpk cholangiocytes have decreased Fgfr2 and Met expression compared to C57BL/6 WT cholangiocytes (Fig. 6C). In contrast, Ift88Orpk BALB/c cholangiocytes have increased expression of multiple pro-fibrotic growth factor receptors including Fgfr2, Tgfbr1, Pdgfrb, and Pdgfra compared to WT controls (Fig. 6C). Notably, the ligand for Tgfbr1 (Tgfb1) was specifically enriched in Ly6chi infiltrating macrophages from BALB/c Ift88Orpk mice suggesting crosstalk between these cell types. We also observed increased gene expression of multiple mesenchymal markers including Cdh2 (N-cadherin), Des, Acta2 (SMA), and Vim as well as ECM genes including Fn1, Col1a1, Col1a2, and Col3a1 in BALB/c Ift88Orpk cholangiocytes compared to WT BALB/c cholangiocytes and C57BL/6 Ift88Orpk cholangiocytes (Fig. 6D). These data suggest that BALB/c, but not C57BL/6, Ift88Orpk cholangiocytes increase expression of profibrotic growth factor receptors and produce extracellular matrix in response to cytokines (TGFB1) produced by BALB/c Ly6chi infiltrating macrophages.

Analysis of receptors that drive cholangiocyte proliferation shows that Ift88Orpk C57BL/6 cholangiocytes have increased expression of pro-proliferative receptors including the Vegfc receptor (Flt4) while Ift88Orpk BALB/c cholangiocytes have decreased expression of receptors that promote cholangiocyte proliferation compared to their respective controls (Fig. 6E) [38]. Likewise, there is a downregulation of gene expression for receptors known to inhibit cholangiocyte proliferation in Ift88Orpk C57BL/6 cholangiocytes and upregulation of these same pathways in Ift88Orpk BALB/c cholangiocytes compared to controls (Fig. 6E). These data suggest that cholangiocytes from C57BL/6 Ift88Orpk mice respond to proliferative cytokines including Vegfc (produced by Ly6clo infiltrating macrophages) due to the presence of the Vegfc receptor (Flt4) whereas BALB/c Ift88Orpk cholangiocytes do not possess the receptors necessary to bind Vegfc and drive cholangiocyte proliferation.

Finally, to assess the interplay between cholangiocytes and infiltrating macrophages, we analyzed cholangiocyte RNA sequencing data for the expression of chemoattractants that might recruit Ly6chi infiltrating macrophages and other immune cells to the liver. Our analysis shows that BALB/c Ift88Orpk cholangiocytes have increased expression of Ccl2, Cxcl1, Cxcl2, and Csf1 compared to BALB/c WT mice and C57BL/6 Ift88Orpk mice (Fig. 6F). In contrast, there was no overt upregulation of chemoattractant cytokines in cholangiocytes from C57BL/6 Ift88Orpk mice, which agrees with data showing that C57BL/6 Ift88Orpk mice do not have a significant increase in the number of Ly6chi macrophages compared to control C57BL/6 mice (Fig. 2). The lack of increased chemoattractant gene expression in C57BL/6 Ift88Orpk mice may also explain why most of the infiltrating macrophages in this strain are Ly6clo (since they are unable to recruit in new, Ly6chi infiltrating macrophages from the circulation).

Blockade of Ly6chi infiltrating macrophage recruitment to the liver does not affect biliary hyperplasia or fibrosis in BALB/c Ift88 Orpk mice

To test whether BALB/c Ly6chi macrophages drive the fibrotic phenotype, we generated BALB/c Ift88Orpk CCR2−/− (CCR2mut) mice. These mice have a marked reduction in the number of Ly6chi macrophages in the liver due to their inability to emigrate out of the bone marrow in the absence of CCR2 [24]. Our data indicate that mice lacking CCR2 have significantly reduced numbers of Ly6chi macrophages in the liver in steady-state and in Ift88Orpk mice compared to CCR2 control mice (Fig. 7A). In contrast, CCR2 deficiency did not significantly affect the number of Ly6clo macrophages in control or Ift88Orpk mice (Fig. 7A).

Fig. 7: Genetic inhibition of Ly6chi infiltrating macrophage recruitment to the liver does not affect biliary hyperplasia or fibrosis in BALB/c Ift88Orpk mice.
figure 7

A Representative flow cytometry plots of Ly6chi and Ly6clo macrophages from 8-week-old BALB/c livers. Quantification of the number of Ly6chi and Ly6clo macrophages as a percentage of total cells is shown as the mean ± SE. *P < 0.05. B qRT-PCR data showing Col1a2 and Col3a1 gene expression in 8-week old BALB/c livers. Values represent the mean + SE. **P < 0.01. C Representative picrosirius red images from 8 weeks old BALB/c Ift88Orpk CCR2Ctrl vs Ift88Orpk CCR2mut mice. Quantification of the percentage picrosirius red positive area is shown as the mean ± SE. Also shown is the quantification of hydroxyproline content (µg/g liver tissue) isolated from Ift88Orpk CCR2Ctrl vs Ift88Orpk CCR2mut mice. D Representative H&E stained sections from 8-week-old BALB/c Ift88Orpk CCR2Ctrl vs Ift88Orpk CCR2mut mice. Quantification of biliary hyperplasia (bile duct area/portal area) and bile duct number is shown as the mean ± SE. E Quantification of the number of cytokeratin 19 + (K19 + ) cholangiocytes as determined by flow cytometry. Values represent the mean ± SE. F qRT-PCR data showing the expression level of Krt19 mRNA from 8-week-old BALB/c CCR2 mice. Values represent the mean ± SE.

Full size image

To assess the impact of reduced Ly6chi infiltrating macrophages on fibrosis in BALB/c Ift88Orpk mice, we performed the qRT-PCR analysis of mRNA isolated from whole liver tissue at 8 weeks of age. Surprisingly and in contrast to our initial hypothesis, the data indicate that Col1a2 and Col3a1 mRNA levels were not significantly different between BALB/c Ift88Orpk CCR2Ctrl and BALB/c Ift88Orpk CCR2mut mice (Fig. 7B). Likewise, quantification of picrosirius red-stained liver sections or analysis of hydroxyproline levels (a measure of total collagen protein [39]) shows no difference between BALB/c Ift88Orpk CCR2Ctrl and BALB/c Ift88Orpk CCR2mut mice (Fig. 7C).

Quantification of biliary hyperplasia and bile duct number from H&E stained sections shows that loss of Ly6chi macrophages does not affect the biliary phenotype in BALB/c Ift88Orpk mice (Fig. 7D). Further, the number of cytokeratin 19+ cholangiocytes as assessed by flow cytometry (Fig. 7E) and the expression of Krt19 mRNA did not differ between Ift88Orpk CCR2Ctrl and Ift88Orpk CCR2mut mice (Fig. 7F). These data show that while infiltrating macrophage subtypes correlate with phenotypic outcome, they do not directly cause the observed phenotypic differences between BALB/c and C57BL/6 Ift88Orpk mice.

Discussion

Patients with HRFCD present with variable levels of inflammation, cyst progression, and fibrosis despite a common underlying genetic mutation suggesting that both extrinsic environmental and intrinsic genetic alterations may influence the rate and severity of disease progression. Herein, we use Ift88Orpk mice on different, inbred genetic backgrounds to test the hypothesis that the subtype of infiltrating macrophage in the liver is a major determinant of phenotypic outcome. Although our data indicate a significant correlation between the subtype of infiltrating macrophages and the phenotypic outcome, Ly6chi infiltrating macrophages do not directly cause the increased fibrosis observed in BALB/c Ift88Orpk mice. Despite the lack of causality, our data indicate that early differences in infiltrating macrophage phenotype may be predictive of long-term phenotypic outcome in the liver.

We observed a strong correlation between the subtype of infiltrating macrophage in the liver at the early stages of the disease and the phenotypic outcome at late stages. Quite surprisingly, we did not see an effect on phenotypic outcome when we blocked the infiltrating macrophage subtype (Ly6chi) that we proposed was driving the fibrotic outcome in BALB/c Ift88Orpk mice. One possible reason for the lack of causality is that the effect of blocking Ly6chi infiltrating macrophage accumulation in BALB/c Ift88Orpk mice may be hidden due to compensation by other inflammatory cells, such as Kupffer cells, producing pro-fibrotic growth factors. This possibility seems plausible due to the relative paucity of Ly6chi infiltrating macrophages in the liver in comparison to Kupffer cells. Despite the lack of a causative role, our data are still important as they indicate that infiltrating macrophage subtypes at early periods of the disease may be predictive of late-stage phenotypic outcomes in the liver.

Our data show that genetic strain has a strong influence on epithelial cell gene expression. During steady-state conditions, there is a moderate difference in the gene expression profile between WT C57BL/6 and BALB/c cholangiocytes. However, there is a clear difference in gene expression patterns between cholangiocytes isolated from C57BL/6 and BALB/c Ift88Orpk mice despite the presence of a common genetic mutation. BALB/c Ift88Orpk cholangiocytes display upregulate pro-fibrotic growth factor receptors and mesenchymal markers while downregulating receptors associated with epithelial cell proliferation. In contrast, C57BL/6 mice have no change in mesenchymal marker expression but upregulate receptors associated with epithelial proliferation. Why cholangiocytes express different receptors during disease settings despite a common genetic mutation remains to be determined.

Previous studies in the kidney [16, 18] and liver [17] indicate that global depletion of phagocytic cells using liposomal clodronate ameliorated HRFCD pathology. Since Kupffer cells are the major phagocytic cell in the liver, and general depletion of phagocytic cells reduced liver hepatic fibrocystic disease [17], it is likely that these effects are the result of Kupffer cell depletion and not loss of infiltrating macrophages. It is also possible that other phagocytic cells such as liver dendritic cells (type I or type II) may be involved in regulating liver pathology. This is in good agreement with the data in this manuscript showing that, although correlative, Ly6chi infiltrating macrophages do not influence the liver phenotype in the Ift88Orpk model of HRFCD. Future studies specifically targeting Kupffer cells or dendritic cells in mouse models of HRFCD are needed to test this hypothesis.

Overall, our data show that genetic strain impacts infiltrating macrophage subtype, gene expression, and phenotypic outcome, although the effect on phenotypic outcome is not dependent on the subset of infiltrating macrophage present in the liver. Therefore, while targeting specific subsets of macrophage may be beneficial in regards to renal fibrocystic disease [27, 28], this does not appear to be the case in the liver.