JNK signaling prevents biliary cyst formation through a CASPASE-8–dependent function of RIPK1 during aging

Significance JNK signaling has been studied intensively in models of liver physiology and disease, but previous studies had focused on young mice. However, it had not been recognized that JNK plays a fundamental role in maintaining liver homeostasis and preventing the formation of biliary cysts in aging mice. These observations call for caution in all long-term pharmacological inhibition strategies targeting the JNK pathway. Finally, our results provide evidence of a molecular link between JNK and the cell-death mediator RIPK1. The specific overexpression of RIPK1 in cysts of a subset of patients with polycystic liver disease suggests that RIPK1 might be mechanistically involved in the pathogenesis of human biliary cysts.

window: 1.8 m/z; collision energy: 27. dd settings: minimum AGC, 5e2; dynamic exclusion: 10 seconds; only 2+ to 5+ peptides were allowed. The raw data was analyzed using MaxQuant (1.6.1.0) and the built-in Andromeda search engine (4). The spectra were searched against the mouse SwissProt database (version 06/2018) (only reviewed and canonical sequences), including the contaminants function of MaxQuant. MaxQuant default settings (including mass tolerance) were used. Specific settings included trypsin as the specific protease (2 missed cleavages); fixed modification: carbamidomethylation; variable modifications: oxidation (Met) and N-terminal protein acetylation; the false discovery rate was set to 0.01 for both peptide and protein levels; minimum peptide length, 7 amino acids. The proteinGroups.txt results derived from the MaxQuant search were further analyzed using Perseus (5). "Potential contaminants, "reverse" and "only identified by site" hits were excluded. Further requirements for protein inclusion was a minimum of two peptides (with a minimum of 1 unique peptide). Proteins were only included in the final data set if they were identified in both lanes of cyst fluid (based on the intensity value).
16S rDNA PCR. The 16S rDNA PCR was performed using a pair of primers (Forward primer (5′-ACTCCTACGGGAGGCAGCAGT -3′) and reverse primer, (5′-ATTACCGCGGCTGCTGGC -3′)) universal to all bacterial groups (6). For each sample, 10 ng of the isolated genomic DNA were added to 1x PCR reagent mixture (1 U HotStarTaq polymerase (Qiagen), 1x CoralLoad PCR buffer (Qiagen), 200 nM dNTP, 200 nM of primers). PCR was performed using a C1000 Touch Thermal Cycler (Biorad) using the protocol as follows: 95°C for 5 min, then 25 cycles of 94°C for 1 min, 55°C for 1.5 min and 72°C for 1 min. These cycles were followed by 72°C for 10 min, and storage at 4°C. The resulting amplicon has a size of 197 bp.
Analysis and quantification of immunohistochemical stainings. For the analysis for LPC-KO mice the quantification of liver cyst areas was done by scanning and analyzing whole tissue sections by ImageJ (Fiji). The pixels covered by the whole tissue section and the empty space (white area) were quantified. For quantification of proliferating hepatocytes, positive stains for Ki67 were counted manually and normalized to tissue area in five non-overlapping and randomly selected pictures (0,8 mm 2 each) per histological slide using ImageScope Software. Positively stained cholangiocytes (double positive for Ki67 and pan-CK) were counted manually and normalized to the total number of pan-CK positive counts in the corresponding images. ImageScope Software was used for the quantification of Ki67 and H2A.X. Immunofluorescence (IF) and IHC (for 3D imaging and 3D analysis). Animals were perfused transcardially with 40 ml PBS and 40 ml 4% PFA (7 ml/min). By default, the left liver lobe was processed for standard histological and IHC analysis whereas the median, right lower, and right upper liver lobes were processed for 3D IF analysis. For standard histological and IHC analysis, dissected livers were post-fixed in 4% PFA for 24 h, embedded in paraffin, and sectioned. Serial 3.5 μm-thick sections were stained with primary antibodies followed by diaminobenzidine staining (Vector Laboratories) and counterstaining with hematoxylin. For 3D imaging, perfused livers were post-fixed in 4% PFA for 2 h at 4ºC, washed in PBS, cryoprotected with increasing concentrations of 10%, 20%, and 30% sucrose solutions prepared in PBS, and embedded in O.C.T. Tissue Tek compound. For 3D fluorescent imaging liver lobes were cut into 200 μm-thick sections using a cryostat-microtome. Free-floating tissue-sections were washed 4× 15 min with 1× PBS, permeabilized with 0.5% Triton X-100 in PBS for 1h at RT, and blocked in 0.5% Triton X-100, 10% serum in PBS for 2 h at 37°C and at 4°C overnight. Pretreated tissue slices were incubated with primary antibody dilution in 0.5% Triton X-100, 5% Serum in PBS at 4°C for 2-3 days. After thorough washing, samples were incubated with fluorescent secondary antibodies and with 4,6-diamidino-2phenylindole (DAPI, Sigma) for nuclear co-staining diluted in 0.5% Triton X-100, 5% serum in PBS at 4°C for 2-3 days. Sections were finally washed with PBS and stored at 4°C in PBS until clearing. All incubation and washing steps were performed on a shaking device.
Tissue clearing, 3D imaging, and 3D analysis. Immunolabeled liver sections were cleared with a modified version of 3DISCO protocol (7). All steps were performed at RT on a shaking device. Samples were incubated successively in 50% (v/v), 75% (v/v), and 100% (v/v) tetrahydrofuran (THF, Sigma) for 15 min each in a glass vial. Finally, samples were immersed in dibenzyl ether (DBE, Sigma) for at least 10 min. Cleared liver sections were mounted in clearing solution (DBE refraction index: ~1.56). Immunofluorescence micrographs and z-stacks were acquired with confocal microscopy (Leica TCS SP8) equipped with Diode 405, white light laser, and HC PL APO CS2 20×/0.75 IMM oil objectives. Microscope settings were: 1024 × 1024 pixel frame size; 400 Hz scan speed; 1 μm z-step size; Pinhole 1.00 AU; sequential scan between stacks. 3D analysis was performed with IMARIS 8.3 software (Bitplane). For visualization of volume images, gamma corrections, and median filtering (3×3×3) were applied to optimize staining intensities and to remove noise or artifacts. Furthermore, 3D surface reconstructions and resulting number of connected surfaces were constructed using the "Surface" function.
Cell culture. Primary mouse cholangiocytes (Cell Biologics) were cultured on gelatin-coated tissue culture plastic in complete epithelial cell medium (Cell Biologics) containing FCS (fetal calf serum), L-glutamine, antibiotics, EGF (epidermal growth factor) and ITS (insulin-transferrin-serin). Inactivation of JNK was achieved by adding 50 µM JNK inhibitor SP600125 (Selleckchem) to the culture medium and incubating for 2h prior to further experiments. The corresponding amount of DMSO was used as control. For stimulation of TNF signaling cells were treated with 10 ng/ml murine TNF for 1h and then harvested by scratching in cold PBS (phosphate buffered saline). Cell pellets were used for protein isolation.
Kinase activity profiling microarray. Each cell culture stimulation experiment was performed three times independently and samples were measured as biological replicates with n = 3. Frozen cell pellets of stimulated cells were resuspended in M-PER mammalian protein extraction reagent supplemented with PhosphoSTOP (Roche) and cOmplete Tablets (Roche), incubated on ice for 15 minutes. After centrifugation at 20.000x g for 15 minutes at 4°C the protein concentration in the supernatant is evaluated via Bradford assay. The lysate is aliquoted, snap frozen and stored at -80°C. Only unthawed aliquots are used for the kinase activity assay. Ser/Thr Kinase (STK) activity profiling assays based on measuring peptide phosphorylation by protein kinases (PamGene International BV, The Netherlands) were performed as instructed by manufacturer. In summary, samples with 1 µg protein were applied on PamChip®4 arrays containing 144 (STK) or 196 (PTK) peptides immobilized on a porous aluminum oxide membrane. The peptide sequences (13 amino acids long) harbor phosphorylation sites and are correlated with one or multiple upstream kinases. Fluorescently labelled anti-phospho antibodies are used to detect phosphorylation activity of kinases present in the sample (8,9). Instrument operation and imaging are controlled by the EVOLVE 2.0 software and quantified using BioNavigator 6.3 (BN6; PamGene International BV, The Netherlands). Signal intensities at multiple exposure times were integrated by linear regression (S100), Log2-transformed, and normalized using a Combat correction model for batch correction where the scaling parameters (mean / sd) are estimated using an empirical Bayes approach, which effectively reduces the noise associated with applying the correction (10,11). The normalized values were used to perform statistics comparing 2 groups using unpaired t-tests or the upstream kinase analysis (UKA) tool (BN6; PamGene international BV). The phylogenetic kinome tree, useful to group the kinases into sequence families, is plotted using the online portal CORAL: http://phanstiel-lab.med.unc.edu/CORAL/ (12). The upstream kinase analysis functional scoring tool (PamGene International) rank-orders the top kinases differential between the two groups, the ranking factor being the final (median) kinase score (represented by node size). This score is based on a combined sensitivity score (difference between treatment and control groups, represented as node color) and specificity score for a set of peptides to kinase relationship that are derived from existing databases. An arbitrary threshold of a final score of 1.2 was applied, based on the specificity scores. Significant peptides (t-tests, p-value < 0.05)) or kinases (UKA, final scores > 1.2) were imported to the MetaCore pathway analysis tool (Clarivate Analytics) where an enrichment analysis was performed for pathways and networks. It consists of matching the kinases or substrates in the kinome arrays data with functional ontologies in MetaCore. The probability of a random intersection between a set of IDs in the target list with ontology entities is estimated in p value of hypergeometric intersection. The lower p value means higher relevance of the entity to the dataset, which shows in higher rating for the entity. Direct interaction network algorithms were used to build interconnected networks within each comparison, and the "Add interactions" feature was used to add the interaction between RIPK1 and the data present in the MetaCore™ database after it was built.
Statistics. Mouse data were analyzed using PRISM software (GraphPad Prism; GraphPad Software) and are expressed as mean. Gaussian distribution was tested with Kolmogorov-Smirnov test. Differences between two groups were assessed by an unpaired two-sample t test or Mann-Whitney test and multiple comparisons between more than two groups have been conducted by ANOVA with Bonferroni test or Kruskal-Wallis test for post hoc analysis.

Data availability.
All raw data is included in the corresponding tables and datasets.     Phylogenetic kinase mapping illustrates the distribution of kinases analysed in a serine/threonine kinase activity profiling array and their regulation. Ser/Thr kinases activity, that differed upon TNF stimulation, are predominantly clustered in the CMGC-kinase family in cholangiocytes pre-treated with solvent (DMSO). The coloring scale indicating the changes in kinase activity ranges from −1 (strong decrease of kinase activity upon TNF stimulation, blue) over 0 (no change of kinase activity upon TNF stimulation, grey) to +1 (strong increase of kinase activity upon TNF stimulation, red). The node size was based on final (median) kinase score (0 to 3; 3 being largest). Serine/threonine kinase activity profiling was performed with protein lysates of cells (pre-treated with solvent (DMSO) for 2h) with or without TNF (1h) (biological replicates; n = 3). CMGC kinase group: cyclin-dependent kinase (CDK) family, mitogen-activated protein kinase (MAPK) family, glycogen synthase kinase (GSK) family and CDC-like kinase (CLK) family. Phylogenetic kinase mapping illustrates the distribution of kinases analysed in a serine/threonine kinase activity profiling array and their regulation. Ser/Thr kinase activity, that differed upon TNF stimulation, are predominantly clustered in the AGC-kinase family, CAMK-kinase family and CMGC-kinase family in cholangiocytes pre-treated with JNK inhibitor (SP600125). The coloring scale indicating the changes in kinase activity ranges from −1 (strong decrease of kinase activity upon TNF stimulation, blue) over 0 (no change of kinase activity upon TNF stimulation, grey) to +1 (strong increase of kinase activity upon TNF stimulation, red). The node size was based on final (median) kinase score (0 to 3; 3 being largest). Serine/threonine kinase activity profiling was performed with protein lysates of cells (pre-treated with JNK inhibitor SP600125 for 2h) with or without TNF (1h) (biological replicates; n = 3). AGC kinase group: cAMP-dependent protein kinase (PKA), the cGMP-dependent protein kinase (PKG), and the protein kinase C (PKC) families; CAMK group: Ca2+/calmodulin-dependent protein kinase class of enzymes; CMGC kinase group: cyclin-dependent kinase (CDK) family, mitogenactivated protein kinase (MAPK) family, glycogen synthase kinase (GSK) family and CDC-like kinase (CLK) family.

Fig. S7. Direct interaction network analysis of substrates and kinases being activated upon TNF stimulation in primary mouse cholangiocytes pre-treated with solvent (DMSO).
Significant peptides (t-tests, p-value < 0.05) or kinases (UKA, final scores > 1.2) were imported to the MetaCore pathway analysis tool (Clarivate Analytics) where an enrichment analysis was performed for pathways and networks. It consists of matching the kinases or substrates in the kinome arrays data with functional ontologies in MetaCore. Direct interaction network algorithms were used to build interconnected networks within each comparison, and the "Add interactions" feature was used to add the interaction between RIPK1 and the data present in the MetaCore™ database after it was built. Node represent kinase activity increase upon TNF stimulation (TNFstimulated versus unstimulated cells). The intensity of red color reflects the increase in kinase activity. Green arrows indicate positive interaction, red arrows indicate negative interactions and gray arrows indicate unspecified interactions. See MetaCore website for detailed legend at https://portal.genego.com/legends/MetaCoreQuickReferenceGuide.pdf .

Fig. S8. Direct interaction network analysis of serine/threonine kinases being activated upon TNF stimulation in primary mouse cholangiocytes pre-treated with JNK inhibitor (SP600125).
Significant peptides (t-tests, p-value < 0.05) or kinases (UKA, final scores > 1.2) were imported to the MetaCore pathway analysis tool (Clarivate Analytics) where an enrichment analysis was performed for pathways and networks. It consists of matching the kinases or substrates in the kinome arrays data with functional ontologies in MetaCore. Direct interaction network algorithms were used to build interconnected networks within each comparison, and the "Add interactions" feature was used to add the interaction between RIPK1 and the data present in the MetaCore™ database after it was built. Node represent kinase activity increase upon TNF stimulation (TNFstimulated versus unstimulated cells). The intensity of red color reflects the increase in kinase activity. Green arrows indicate positive interaction, red arrows indicate negative interactions and gray arrows indicate unspecified interactions. See MetaCore website for detailed legend at https://portal.genego.com/legends/MetaCoreQuickReferenceGuide.pdf . Table S1. Proteome analysis of the cyst fluid of JNK1/2 LPC-KO mice. Mass spectrometry analysis of the proteome of cyst fluid from two JNK1/2 LPC-KO mice. Table S2. Evaluation of histological RIPK1 staining on liver tissue of human PLD patients. ADPLD: autosomal dominant polycystic liver disease; ARPLD: autosomal recessive polycystic liver disease (congenital hepatic fibrosis); vMC: von Meyenburg complex; staining: -= negative, + = weak, ++ = moderate