Intestinal Microbiome‐Macrophage Crosstalk Contributes to Cholestatic Liver Disease by Promoting Intestinal Permeability in Mice

Background and Aims Mounting evidence supports an association between cholestatic liver disease and changes in the composition of the microbiome. Still, the role of the microbiome in the pathogenesis of this condition remains largely undefined. Approach and Results To address this, we have used two experimental models, administering alpha‐naphtylisocyanate or feeding a 0.1% 3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine diet, to induce cholestatic liver disease in germ‐free mice and germ‐free mice conventionalized with the microbiome from wild‐type, specific pathogen‐free animals. Next, we have inhibited macrophage activation by depleting these cells using clodronate liposomes and inhibiting the inflammasome with a specific inhibitor of NOD‐, LRR‐, and pyrin domain‐containing protein 3. Our results demonstrate that cholestasis, the accumulation of bile acids in the liver, fails to promote liver injury in the absence of the microbiome in vivo. Additional in vitro studies supported that endotoxin sensitizes hepatocytes to bile‐acid–induced cell death. We also demonstrate that during cholestasis, macrophages contribute to promoting intestinal permeability and to altered microbiome composition through activation of the inflammasome, overall leading to increased endotoxin flux into the cholestatic liver. Conclusions We demonstrate that the intestinal microbiome contributes to cholestasis‐mediated cell death and inflammation through mechanisms involving activation of the inflammasome in macrophages.


Materials
Chenodeoxycholic and deoxycholic acid (CDCA and DCA) were obtained from Sigma-Aldrich. Clodronate liposomes were obtained from Liposoma Research. MCC950, the specific inhibitor of Nrlp3, was obtained from Invivogen.

Bile acid determination
Bile acids were extracted from liver samples and analyzed using high performance liquid chromatography-tandem mass spectrometry we previously described (1). In brief, cleaned-up extracts were analysed using HPLC-mass spectrometry operated in multiple reaction monitoring (MRM) mode. Each sample (5 µl) was analysed using an Agilent 1260 binary HPLC coupled to an AB Sciex 4000 QTrap triple quadrupole mass spectrometer. HPLC was achieved using a binary gradient of solvent A (Water + 5mM Ammonium Ac + 0.012% Formic acid) and solvent B (Methanol + 5mM Ammonium Ac + 0.012% Formic acid) at a constant flow rate of 600 µl/min. Separation was made using a Supelco Ascentis Express C18 150 x 4.6, 2.7µm column maintained at 40°C. Injection was made at 50% B and held for 2 min, ramped to 95%B at 20 min and held until 24 minutes. The column equilibrated to initial conditions for 5 minutes.
The mass spectrometer was operated in electrospray negative mode with capillary voltage of -4500V at 550°C. Instrument specific gas flow rates were 25ml/min curtain gas, GS1: 40 ml/min and GS2: 50 ml/min. Mass fragmentation was monitored and quantification was applied using Analyst 1.6.2 software to integrate detected peak areas relative to the deuterated internal standards.

Mouse hepatocyte isolation and culture
Primary hepatocytes were isolated from germ free (GF) and conventionalised GF mice GF+WT by perfusion of the liver and further digestion with collagenase I (Worthington). Cells were washed, pelleted and plated on rat collagen type I (BD Biosciences) pre-coated plates with Minimum Essential Medium as we described in (1).
Microscopic images of hepatocyte cultured were taken using EVOS XL Core imaging system (Thermofisher).

Histology, immunohistochemistry and immunofluorescence
Liver tissues were immediately fixed in 10% neutral formalin and embedded in paraffin.
Tissue blocks were further sectioned, dewaxed and hydrated. Liver slides were stained with H&E for pathological analysis and fibrosis was determined with Sirius Red staining. In all cases, a representative picture of 5-10 fields per sample is shown.
Immunohistochemistry (IHC) on paraffin embedded sections was performed using an anti-CK19 (Developmental Studies Hybridoma Bank, University of Iowa) diluted in antibody diluent (Dako), after which an anti-Rat HRP labelled secondary antibody was applied. IHC was developed using the DAB + chromogen system (Dako) and nuclei were counterstained with hematoxylin.
Immunofluorescence using an anti-Occludin antibody (AbCam) was used in paraffin cut sections of duodenum, jejunum, ileum and colon. A Cy3-labeled secondary antibody was used and slides were mounted in a Dapi-containing solution (VectorLabs).

RNA isolation and Quantitative Real-Time PCR
RNA was isolated from liver samples with QiAzol Reagent (Qiagen) followed by first strand synthesis with random primers and reverse transcription using M-MLV Reverse

Characterization of apoptotic hepatocyte death
Apoptotic cell death was determined by quantifying Caspase-3 activity primary hepatocytes using a fluorescence-labelled substrate following the manufacturer's instructions (Enzo) as we previously described (1).
Whole cell lysates were further resolved in sodium dodecyl sulphate-polyacrylamide gels and transferred to nitrocellulose membranes (Whatman) and transferred to nitrocellulose blotting membranes. Membranes were probed with Caspase 1, Interleukin 1β, Occludin primary antibodies (Santa Cruz biotechnologies) as well as E-Cadherin (AbCam). As a loading control, we used b-Actin (Sigma Aldrich) antibody. As secondary antibody, we used anti-mouse IgG-

HRP-linked (Santa Cruz biotechnologies).
Determination of proinflammatory cytokines TNF and Interleukin 1β, expression was determined in intestinal protein extracts using R&D systems DuoSet.

Flow Cytometry
Immune cells were isolated from liver tissues after digestion with collagenase for 30 minutes at 37C. Samples were homogenised and passed through a mesh followed by successive washes and centrifuged for gradient separation using Percoll. Isolated immune cells were stained with CD45-APC-Cy7 (BD), CD11b-PE (BD) and F4/80-FITC (Myltenyi) antibodies. Flow cytometry analysis was performed using BD LSRFortessa and analysed using FlowJo software.

Bacterial genomic DNA isolation and 16s rRNA sequencing
Bacterial genomic DNA was isolated from faecal pellets using the MBP DNA Soil extraction kit. Genomic DNA was normalised to 5ng/µl with EB (10mM Tris-HCl) and libraries were

16S sequence analysis
The LotuS 1.36 was used pipeline (2) in short amplicon mode with default quality filtering.
Raw 16S rRNA gene reads were quality filtered to ensure a minimum length of 170 bp, not more than eight homonucleotides, no ambiguous bases, average quality >= 27 and an accumulated error below 0.5. For OTU clustering and denoising we used UPARSE (3) assigning a taxonomy using the LotuS LCA algorithms against Silva 128 reference database (5). We could assign on average 5500 ± 3108 reads to each sample that were of cyanobacterial origin. Further data analysis was conducted with R statistical language Version 3.00 (The R Foundation, https://www.r-project.org/) as described in Hildebrand et al. (6), employing the rtk software (7) or all data normalizations.

Statistical analysis
Data are expressed as mean ± standard error of the mean. Statistical significance was determined by two-way analysis of variance followed by a Student's t test. All data shown are representative of at less three independent experiments.