Mesenchymal stromal cells mitigate liver damage after extended resection in the pig by modulating thrombospondin-1/TGF-β

Animal trials

Animal experiments were conform to the animal welfare act and approved by the federal state authority of Saxony (file no. TVV39/13). Adult male German landrace pigs were obtained from the farm product company Kitzen (Pegau, Germany) and were housed at the Experimental Centre of the Faculty of Medicine, University of Leipzig, under a 12 h circadian rhythm at 25 °C receiving a standard pig diet for at least 3 days. Animals were starved 24 h before surgery and underwent a veterinary inspection verifying body weight (25–30 kg), temperature (below 40 °C), and general health condition. Animal housing and treatments were in accordance with the Guide for the Care and Use of Laboratory Animals. If not otherwise indicated, nine adult male German landrace pigs were randomized into groups receiving either Ringer solution after ePHx (control), or 1 × 10⁸ hepatocytic differentiated pBM-MSC via central venous transfusion. Sham-treated animals remained without liver resection and central infusion. A minimum of three animals per group was included to enable statistical analyses. However, group size was largely kept at this minimum in order to meet the 3R principles as far as possible.

Surgical procedure

Seventy percent partial hepatectomy was adapted to Arkadopoulos et al.⁴⁹ and performed essentially as described previously²². All animals, including the sham-operated pigs, underwent the same pre-, peri-, and post-surgery procedures as described in more detail in Supplementary Methods (Supplementary Fig. 11), except omission of liver resection in the sham and MSC infusion in the MSC treatment group. Post surgery, the animals were subjected to intensive care monitoring and treatment for 24 h. This time period was chosen according to the procedure described by Arkadopoulos et al.⁴⁹, which turned out to be sufficient to identify surgery-induced early changes indicative for liver damage and acute failure. Following the demands of the animal welfare act, extension of the observation period was restricted to avoid excessive suffer for the animals.

Values of AST, ALT, ammonia, lactate, and the INR were determined in blood samples taken from a horizontal arterial catheter. Depending on the downstream analysis, the various samples were submitted to different procedures immediately after collection. Blood samples collected during the 24 h post-surgery intensive care monitoring were sent directly to the Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University of Leipzig Medical Center, in the vacutainers provided for this specific analysis and immediately processed there accordingly. For the determination of plasma proteins by ELISA, blood was collected in serum vacutainers and immediately centrifuged at 4000 × g for 5 min. The supernatant was snap frozen and stored in liquid nitrogen. The ICG-clearance rate was determined before, immediately after (0 h), 12 h, and 24 h post-resection. The ICG-clearance corresponded to the plasma disappearance rate of ICG (PDR_ICG) and was measured using the LIMON^® (Pulsion Medical Systems, Munich, Germany) after a bolus application of 2 mL ICG (ICGPULSION; Pulsion Medical Systems; Rastatt, Germany) via the central venous catheter.

Human and pBM-MSC isolation and differentiation

Procedures involving hBM-MSC were approved by the Institutional Ethics Review Board Leipzig (file no. 282/11-ek). hBM was obtained from waste material during elective knee or hip joint surgery at the Department of Orthopedics, Trauma and Plastic Surgery, Division of Endoprothetic Joint Surgery/Orthopedics, University of Leipzig Medical Center, after obtaining the patients’ written consent. Isolation, differentiation into the hepatocytic lineage, and characterization of porcine (pBM-MSC) and hBM-MSC has been detailed previously^50,51. Briefly, MSCs were isolated from the Substantia spongiosa of the Os femoris using collagenase (NB4G, Serva GmbH, Heidelberg, Germany) digestion. The isolated mononuclear cell fraction was seeded on plastic culture dishes and expanded until 90% confluence in stem cell maintenance medium. Thereafter, h/pBM-MSC were differentiated into hepatocyte-like cells using hepatocyte growth medium (HGM) essentially as described⁵⁰. After 14 days, cells were collected in Ringer´s saline and 1 × 10⁸ cells were collected in a transfusion syringe. We used differentiated MSC, because they secreted a significantly broader panel of potentially hepatotropic factors than undifferentiated cells⁵². This is supposed to be advantageous, because during the post-surgery observation period of 24 h as chosen here, it is expected that hepatoprotective mechanisms are not due to MSC tissue integration and substitution. Based on our previous experience from extended PHx in the rat model, rather mechanisms depending on soluble factors were to be expected⁹.

Gene expression analyses

Gene expression analysis from six lung and six liver samples (each two organ samples from three different animals per group) was performed using Affymetrix Porcine Gene 1.0 ST microarrays at the “KFB – Center of Excellence for Fluorescent Bioanalytics” (Regensburg, Germany; www.kfb-regensburg.de).

Total RNA extraction from liver tissue

Approximately 200 mg of porcine liver tissue was homogenized in 700 µL Trizol reagent using Precellys CK14 ceramic beads (VWR, Darmstadt, Germany) (2 cycles of 15 s at 6500 r.p.m.; 10 s break). After 5 min incubation at room temperature, 140 µL of chloroform was added and the samples were again incubated at room temperature for 2 min. Phase separation was achieved by 15 min centrifugation at 12,000 × g at 4 °C. An equal amount of 70% ethanol was added to the aqueous supernatant and the mixture was applied to RNeasy Mini spin columns (RNeasy Mini Kit, QIAGEN, Hilden, Germany), followed by an on-column DNase digestion and several wash steps. Finally, total RNA was eluted in 30 μL of nuclease-free water. Purity and integrity of the RNA was assessed on the Agilent 2100 Bioanalyzer with the RNA 6000 Nano LabChip reagent set (Agilent, Palo Alto, CA, USA).

Total RNA extraction from lung tissue

Approximately 200 mg of porcine lung tissue was homogenized in 800 µL RLT buffer (Qiagen, Hilden, Germany) using Precellys CK14 ceramic beads (2 cycles of 20 s at 6500 r.p.m.; 15 s break). Next, 1.4 mL of Trizol reagent was added to 150 µL of homogenate. After 5 min at room temperature, 280 µL of chloroform was added and the samples were again incubated at room temperature for 2 min. Phase separation was achieved by 15 min centrifugation at 12,000 × g at 4 °C. An equal amount of 70% ethanol was added to the aqueous supernatant and the mixture was applied to RNeasy Microspin columns (RNeasy Micro Kit, QIAGEN, Hilden, Germany), followed by an on-column DNase digestion and several wash steps. Finally, total RNA was eluted in 14 μL of nuclease-free water. Purity and integrity of the RNA was assessed on the Agilent 2100 Bioanalyzer with the RNA 6000 Pico LabChip reagent set (Agilent, Palo Alto, CA, USA).

GeneChip microarray assay

Sample preparation for microarray hybridization was carried out as described in the Affymetrix GeneChip WT PLUS Reagent Kit User Manual (Affymetrix, Inc., Santa Clara, CA, USA). In brief, 200 ng of total RNA was used to generate double-stranded cDNA. Then, 15 µg of subsequently synthesized cRNA was purified and reverse transcribed into sense-strand (ss) cDNA, whereas unnatural dUTP residues were incorporated. Purified ss cDNA was fragmented using a combination of uracil DNA glycosylase and apurinic/apyrimidinic endonuclease 1 followed by a terminal labeling with biotin. Next, 3.8 µg fragmented and labeled ss cDNA were hybridized to Affymetrix Porcine Gene 1.0 ST arrays for 16 h at 45 °C in a GeneChip hybridization oven 640. Hybridized arrays were washed and stained in an Affymetrix Fluidics Station FS450 and the fluorescent signals were measured with an Affymetrix GeneChip Scanner 3000 7G. Fluidics and scan functions were controlled by the Affymetrix GeneChip Command Console v4.1.3 software. Sample processing was performed by an Affymetrix Service Provider and Core Facility, “KFB – Center of Excellence for Fluorescent Bioanalytics” (Regensburg, Germany; www.kfb-regensburg.de).

Microarray data analysis

Summarized probe set signals in log2 scale were calculated by using the robust multi-array (RMA)⁵³ algorithm with the Affymetrix GeneChip Expression Console v1.4 Software. After exporting into Microsoft Excel, average signal values, comparison fold changes, and significance P-values were calculated. Probe sets with a fold-change above 2.0-fold and a Student’s t-test P-value lower than 0.05 were considered as significantly regulated.

Gene array data processing

Pre-processing and analysis of gene expression data (available at Gene Expression Omnibus: GSE134970) was performed for each organ separately. Microarray annotation of raw data was achieved using annotation information from Brainarray⁵⁴. Background correction, quantile normalization, and probe summarization was performed using RMA method⁵⁵ as implemented in the affy-package for R⁵⁶. For the lung samples, a batch effect related to the processing of samples was corrected using ComBat⁵⁷ as implemented in the sva package for R^58,59. Differential gene expression between treatment and control samples was assessed using Limma for R⁶⁰. Based on a P-value < 0.05 and a minimal absolute log2 fold-change of at least 0.5, we identified 679 DEGs in the lung and 469 DEGs in the liver.

Identification of regulatory modules

In summary, regulatory modules were identified using ModuleDiscoverer²³. The algorithm maps a set of DEGs onto an organism-specific PPIN for the identification of a sub-network (regulatory module), which by statistics is significantly enriched with DEGs.

In order to identify a sub-network, the underlying PPIN is first fragmented into small groups of highly interacting proteins. This is driven by the assumption that within such PPINs, closely connected proteins are involved in similar biological functions⁶¹. These protein groups are then tested for their significant enrichment with DEGs based on Fisher’s exact test. The union of all significantly enriched protein groups then assembles the regulatory module, which is assumed to include all relevant biological functions and pathways affected by the treatment⁶¹. The challenge of this strategy lies in the identification of the small groups of proteins. Barrenaes et al.⁶² define them as maximal cliques in the network. A clique is a group of proteins, in which each protein is connected to every other protein within the group. A maximal clique is a clique that cannot be extended by any other protein of the network. However, the identification of all maximal cliques in a network is a non-deterministic polynomial time hard problem, which renders it computationally infeasible for large-scale, whole-genome PPINs. To overcome this problem, the randomization-based heuristic of ModuleDiscoverer enumerates maximal cliques of size three or more in an iterative approach starting from a random seed protein for each iteration. This way, the PPIN’s underlying community structure, i.e., the set of protein groups, is approximated across the full network. ModuleDiscoverer then uses permutation-based P-value calculation for the identification of all significantly enriched maximal cliques. In that, for each clique the P-value computed based on the set of DEGs is compared to a set of P-values based on sets of randomly selected genes of size |DEGs|. These genes are sampled from the statistical background, i.e., the set of genes measured on the microarray. Unification of all significantly enriched maximal cliques then assembles the regulatory module. As ModuleDiscoverer uses a randomization-based approach for the identification of regulatory modules, the stability of the result, i.e., the reproducibility of the identified regulatory module has to be assessed. To this end, ModuleDiscoverer uses bootstrap samples (model-free resampling with replacement) of the set of all enumerated maximal cliques for the identification of additional regulatory modules. The graph-edit distance between every two identified regulatory modules then provides a measure for the average similarity in terms of common edges and nodes in relation to all edges and nodes in the two regulatory modules compared.

Identification of organ-specific regulatory modules using ModuleDiscoverer

Identification of the regulatory modules for lung and liver using ModuleDiscoverer was performed as follows. The organism-specific PPIN was downloaded from the STRING database version 10⁶³. We then removed all edges with an edge-specific confidence score < 0.7 retaining a PPIN composed of 13,507 proteins connected by 184,978 high-confidence relations. Identification of maximal cliques was performed using 2,000,000 iterations starting with one random seed protein per iteration. This identified a total of 1,441,247 maximal cliques containing 133,549 unique maximal cliques. To make use of the permutation-based P-value calculation provided by ModuleDiscoverer, we further created 10,000 sets of 679/469 genes (lung/liver) randomly sampled from all 17,299 measured genes on the microarray. Translation of EntrezGene IDs for the set of DEGs and the sets of random genes into EnsemblProtein IDs was then performed using the Ensembl database⁶⁴ via the biomaRt package for R⁶⁵. Based on the set of identified maximal cliques in the PPIN, the set of DEGs and 10,000 sets of random genes, we then identified 334/310 (lung/liver) significantly enriched (P-value < 0.01) maximal cliques, which were then unified to assemble the organ-specific regulatory module.

Regulatory modules for lung and liver were identified by ModuleDiscoverer. The lung-specific regulatory module is composed of 358 proteins connected by 1883 relations. Of these 358 proteins, 262 proteins corresponded to genes measured on the microarray including 68 proteins associated to DEGs. The liver-specific regulatory module contained 364 proteins connected by 2587 relations. Two hundred and sixty of all 364 proteins were associated to genes measured on the microarray including 57 DEG-associated proteins. Both of the identified regulatory modules were significantly enriched (P-value < 1 × 10⁻⁴) with proteins associated to DEGs as well as stable (average node/edge stability greater 95%) with respect to regulatory modules identified from 100 additional bootstrap samples of sets of maximal cliques.

LASSO inferred cluster regulatory network

To identify protein clusters specifically associated with the MSC treatment, we inferred a CRN. To this end, we defined a linear model to estimate the expression value x of cluster i in sample m ((hat x_i(m))) based on the sum of the weighted ((beta _{i,j})) expression values of all remaining clusters j ((x_j(m))) as well as two weighted variables corresponding to the treatment (1, if sample m is treated with MSC; 0 else) and organ (1, if sample m is taken from liver; 0 else).

We then performed a regression analysis using the LASSO approach as implemented in the glmnet package for R⁶⁶. In this implementation, the LASSO identifies an optimal parameter set (beta ^ ast) that minimizes the difference between the observed ((x_i(m))) and predicted ((hat x_i(m))) expression values based on the mean squared error (Eq. 2). As an additional side constrain, the sum of all absolute parameter values needs to be below some user-defined threshold λ, which allows for automated variable selection. Estimation of the optimal λ was performed by 12-fold cross-validation.

The process of cross-validation involves random sampling into sets of test and training samples. In order to obtain stable variable selection, we thus performed the cross-validation step 1000 times.

Identification of upstream regulators

Upstream regulatory genes were identified by IPA (Qiagen, Hilden, Germany) using all genes significantly enriched by MSC treatment compared to the control (P-value < 0.05) from the lung and liver. TGF-β1 target genes for both the liver and lung were plotted as regulatory network.

Cultures of primary pig and mHCs, mouse non-parenchymal liver cells, MDCK II cells, HUVEC, and human THCs

Pig hepatocytes were isolated by a modified two-step collagenase perfusion protocol as described elsewhere^50,67 and cryopreserved in inactivated fetal calf serum (iFCS) and 7.5% dimethyl sulfoxide in liquid nitrogen until use. For single cultures, hepatocytes were thawed and seeded in minimal essential medium (MEM) containing 0.5 μg/mL insulin (Sigma-Aldrich, Darmstadt, Germany), 50 μg/mL gentamycin (Biochrom GmbH, Berlin, Germany), and 5% iFCS at a density of 300,000 cells/9 cm². Five hours after seeding, the medium was changed to HGM containing 2% iFCS as described⁶⁸. Medium was refreshed after 24 h and cells further grown for 2 days. Before start of the experiments, the medium was sucked off and culture continued for 1 h in serum- and growth factor-free HGM. For co-culture, pHCs were mixed with pBM-MSC (1 : 1) and grown until 80–90% confluence in HGM. TGF-β signaling was studied after stimulation with 1 ng/mL TGF-β (PeproTech GmbH, Hamburg, Germany) for 1 h and pSmad was determined by ELISA or immunocytochemistry. When pig hepatocytes were treated with CM-MSC, the stem cells were cultured for 14–18 days in HGM and CM taken at the end of culture after a medium change 72 h before harvest. CM-MSC was transferred to the hepatocytes and culture continued for 1 h with TGF-β as indicated. pSmad was determined by ELISA as described.

All mouse experiments were approved by the federal state authority of Saxony (reg. no. TVV15/16) and followed all legislation of the animal welfare act. Twelve-week-old, male immune-deficient Pfp/Rag2−/− (C57BL/6N(B6.129S6-Rag2(tm1Fwa)Prf1(tm1Clrk))) mice were housed under standard conditions with a 12 h circadian rhythm at ambient temperature with free access to food (chow diet, V1534, Ssniff, Soest, Germany) and water. mHCs were isolated essentially as described previously⁶⁹. Single cultures were grown in ECM medium (PromoCell GmbH, Heidelberg, Germany) for 48 h before starting the experiments. When hepatocytes were cultured with CM derived from human THCs collected after 10 min of culture (CM-THC), 1 mL of CM-THC medium, or ECM as control medium was added and culture continued for 1 h. THBS1, TGF-β, and pSmad were determined by ELISA.

Non-parenchymal mouse liver cells (NPCs) were isolated from the supernatant of the first centrifugation step in the hepatocyte isolation protocol. The supernatant was re-centrifuged (5 min, 47 × g, 4 °C) followed by centrifugation for 10 min at 317 × g. The pellet was resuspended in a final volume of 30 mL phosphate-buffered saline (PBS), and centrifuged (10 min, 317 × g, 4 °C). This step was repeated until the pellet was void of erythrocytes by visual inspection. The final cell pellet was resuspended in 12 mL ECM containing 1.8 × 10⁶ hepatocytes and plated at 2 mL per well in a six-well plate for co-culture.

To address a direct impact of MSC on TGF-β-induced EMT, we used in vitro co-cultures of hBM-MSC and the epithelial cell line MDCK II (Madin-Darby canine kidney; Sigma-Aldrich as supplied by European Collection of Authenticated Cell Cultures (ECACC 00062107)). MDCK II cells and hBM-MSC (1 : 1) were seeded in alpha MEM (Biochrom GmbH, Berlin, Germany) supplemented with 5% iFCS for 3 days until confluence. After a medium change, cells were stimulated with TGF-β (5 ng/mL) for 20 h where indicated.

Then, 33,000 cells/cm² HUVEC (C-003-5C; Thermo Fischer Scientific GmbH, Dreieich, Germany) were grown in ECM medium (PromoCell GmbH, Heidelberg, Germany) until 80%–90% confluence. For co-culture, HUVECs and hBM-MSCs were seeded at a ratio of 1 : 1 in ECM and grown until 80%–90% confluence for at least 3 days before starting the experiments. Where indicated, HUVECs were treated with CM-MSC collected after 72 h of culture in ECM supplement as described above. Supernatants and cells were collected after 0.25, 2, and 6 h for determination of THBS1 by ELISA or immunocytochemistry, RT-PCR, and western blotting, respectively.

THCs were isolated from plasma donations at the Institute of Transfusion Medicine, University of Leipzig Medical Center, as approved by the Institutional Ethics Review Board Leipzig, after receiving the donors’ written consent. The blood was collected in acid citrate dextrose solution as used for routine blood donations. In order to minimize activation, the platelet concentrates were stored at ambient temperature in storage solution containing 30% plasma with a final citrate concentration of about 3 µM. This procedure still allowed for platelet activation by external stimuli in vitro, since for downstream applications the platelets were suspended or diluted in non-citrate containing media. Immediately after the isolation procedure, THC suspensions (11–14 × 10⁸ THC) were centrifuged (800 × g, 5 min, 24 °C) and resuspended 1 : 1 (wt/v) in ECM medium (PromoCell GmbH, Heidelberg, Germany), or in CM (CM-MSC) from hBM-derived hepatocyte-differentiated MSC (hBM-MSC) collected after 72 h of culture in ECM. After 10 min of incubation at 37 °C with or without 10 U/mL thrombin⁷⁰, the medium was collected, centrifuged (5 min, 1000 × g, 24 °C), and used for cultures with mHCs as described above, or determination of THBS1 by ELISA.

Immunohistochemical detection of E-cadherin, N-cadherin, ZO-1, HS, CD31, CD42b, α-SMA, THBS1, TGF-β, and Smad 2/3

Paraffin-embedded tissue slices (1.5 µm) were incubated with TRIS-buffer (10 mM Tris-HCl, 1 mM EDTA, pH 9.0) and blocking solution (5% goat serum) was added thereafter for 20 min followed by blocking for 60 min in blocking solution made of 5% bovine serum albumin (BSA) and 0.5% Tween 20. Then, slices were incubated with the primary anti-E-cadherin, anti-N-cadherin, anti-ZO-1, anti-HS, anti-CD31, anti-CD42b, or the anti-α-SMA antibody (Supplementary Table 1) overnight at 4 °C. Subsequent to three washing steps with PBS, the secondary antibodies labeled with Cy3 or AlexaFluor488 (Supplementary Table 1) were applied for 70 min at room temperature. After another washing with PBS, slices were counterstained with DAPI solution (Roth GmbH, Karlsruhe, Germany) and embedded in glycerol solution (50%; Roth GmbH, Karlsruhe, Germany) for microscopic analysis. For detection of THBS1 or TGF-β, tissue slices were incubated in citrate buffer (100 mM citric acid, 100 mM sodium citrate pH 6.0). After treatment with H₂O₂ (3%), BSA (5%), and Avidin/Biotin (Vector- Kit SP-2001, BIOZOL GmbH, Eching, Germany) blocking solution, slices were incubated with the anti-THBS1 or the anti-TGF-β antibody overnight at 4 °C (Supplementary Table 1). In addition to three washing steps with PBS, the biotin-labeled secondary antibody (Supplementary Table 1) was incubated with the tissue slices for 60 min at room temperature. Subsequent to three washing steps with PBS, ABC-reagent (Vector- Kit PK-6100, Vectorlabs, Burlingame, CA, USA) was applied for 30 min. After another PBS washing step, DAB-solution (Pierce™ DAB Substrate Kit, Thermo Fischer Scientific GmbH, Dreieich, Germany) was used for color development of THBS1 and NovaRed (Vector- Kit SK-4800, Vectorlabs, Burlingame, CA, USA) for TGF-β. Slices were counterstained with nuclear fast red solution (Roth GmbH, Karlsruhe, Germany) for THBS1 and Mayer’s hemalum solution (Merck, Darmstadt, Germany) for TGF-β, and embedded in Entellan (Merck GmbH, Darmstadt, Germany).

For color development of Smad 2/3, a horseradish peroxidase (HRP)-linked secondary antibody was used instead of avidin/biotin in combination with Histogreen (Linaris- Kit- E-109, Dossenheim, Germany).

Slices were counterstained with nuclear fast red solution (Roth GmbH, Karlsruhe, Germany) and embedded in Entellan (Merck GmbH, Darmstadt, Germany).

Immunocytochemistry

For staining of Smad 2/3 and Phalloidin in co-cultures of pHCs and pBM-MSC, cells were seeded on collagen-coated coverslips. Coverslips were washed twice in PBS and fixed with formalin (3.7%) for 15 min. In addition to two further PBS washing steps, blocking solutions were added (5% goat serum for 20 min and 5% BSA for another 60 min). The anti-Smad 2/3 antibody (Supplementary Table 1) was administered overnight at 4 °C. After incubation with the secondary antibody AlexaFluor488 (Supplementary Table 1) for 50 min, the cytoskeleton was visualized with Phalloidin 568 (1 : 500) (Thermo Fisher Scientific GmbH, Dreieich, Germany) for 50 min. Following another washing with PBS, cells were counterstained with DAPI solution (Roth GmbH, Karlsruhe, Germany) and embedded in glycerol (50%; Roth GmbH, Karlsruhe, Germany) for microscopic analysis.

For staining of ZO-1 in co-cultures of MDCK II cells and hBM-MSC, cells were seeded on fibronectin-coated coverslips. After two washings with PBS, cells were fixed with 3.7% formalin, again washed twice with PBS, and further incubated with 5% goat serum in PBS for 25 min at room temperature, followed by 60 min incubation in BSA blocking solution (5% BSA and 0.5% Tween 20 in PBS). The primary anti-ZO-1 antibody (Supplementary Table 1) in 1% BSA in PBS was added overnight at 4 °C, followed by three washings with PBS at room temperature for 10 min each. The secondary antibody AlexaFluor488 (Supplementary Table 1) in 0.5% BSA in PBS was added for 60 min followed by three washings with PBS at room temperature. Coverslips were embedded in Prolong with DAPI (Thermo Fisher Scientific GmbH, Dreieich, Germany) for microscopic analysis. For quantification of cell contacts, ZO-1-stained membranes on fluorescence microscopy pictures were manually delineated and total lengths quantified using ImageJ 1.46 (National Institutes of Health, Bethesda, MD, USA).

To stain HUVEC for THBS1, cells were treated as described before for ZO-1. After blocking, the primary anti-THBS1 antibody (Supplementary Table 1) in 1% BSA in PBS was added overnight at 4 °C, followed by three washings with PBS at room temperature for 5 min each. The secondary Cy3-labeled goat anti-mouse antibody (Supplementary Table 1) in 0.5% BSA in PBS was added for 45 min at 37 °C followed by three washings with PBS at room temperature. Phalloidin iFluor 488 (Supplementary Table 1) in 0.5% BSA in PBS was added for 45 min at 37 °C followed by three washings with PBS and embedding in in Prolong with DAPI (Thermo Fisher Scientific GmbH, Dreieich, Germany).

Western blot analyses

Western blottings after electrophoretic separation under denaturing conditions were performed according to standard protocols. In brief, liver tissue was homogenized in RIPA buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 1% Na-deoxycholate) supplemented with protease- and phosphatase inhibitor tablets (Roche, Basel, Switzerland). Twenty micrograms of protein lysates were separated on 10% SDS-polyacrylamide gels and transferred onto nitrocellulose membranes according to standard protocols. Primary antibodies were applied overnight at 4 °C and proteins detected by HRP-linked secondary antibodies (Supplementary Table 1).

ELISA assays

TGF-β1

Plasma and culture medium levels were determined in duplicates by using the porcine TGF-β1-ELISA (DlDevelop, Kelowna, BC, Canada) according to the provider’s manual. Then, 100 µL of porcine blood plasma or hepatocyte culture medium were incubated for 120 min at 37 °C in the ELISA microtiter plate. Following three washing steps (300 µL 1× wash buffer), 100 µL detection reagent A were added for 60 min at 37 °C. After another four washing steps (300 µL washing buffer), the chromogen solution (90 µL) was applied for at least 15–25 min at 37 °C. After incubation with 50 µL stop solution, chromogen formation was measured (450 nm) using the GloMax®-Multi Detection System (Promega, Mannheim, Germany). Standard dilutions of TGF-β1 in the range of 0–1000 pg/mL were treated analogously to the samples and used for the calculation of TGF-β1 concentrations.

Antithrombin III

Plasma levels were determined in duplicates by using the porcine Antithrombin-ELISA Kit (MyBioSource, San Diego, CA, USA) according to the instructions. Fifty microliters of plasma and 100 µL of HRP-conjugate were incubated in the ELISA microtiter plate for 60 min at 37 °C. After four washings (250 µL washing buffer each), 50 µL each of chromogen solution A and B were added and incubated for 15 min at 37 °C in the dark. Reaction was stopped by adding 50 µL stop solution and chromogen formation was measured at 450 nm using the GloMax®-Multi Detection System (Promega, Mannheim, Germany). Values shown in Fig. 3b were corrected for the amount of AT III at the zero time point.

Thrombospondin-1

Levels in porcine plasma during 24 h post surgery and in HUVEC and THC culture supernatants were determined in duplicates by the human THBS1-ELISA (human Quantikine®ELISA R&D Systems, Abingdon, UK) according to the instructions. In brief, 100 µL Assay-Diluent and 50 µL plasma (1 : 20 in Calibrator Diluent RD5-33) or supernatant (1 : 2 in Calibrator Diluent RD5-33) were filled into the wells of the ELISA plate and incubated for 2 h at room temperature. After four washing steps, 200 µL THBS1-conjugate solution was added to each well for 2 h. After another four washing steps, 200 µL substrate solution was added for 30 min. Reaction was stopped by adding 50 µL stop solution and chromogen formation was measured at 450 nm using the GloMax®-Multi Detection System (Promega, Mannheim, Germany).

Phospho-Smad 2/3

Phospho-Smad 2/3 was determined using the PathScan® Phospho-Smad 2 (Ser465/467)/Smad 3 (Ser423/425) Sandwich ELISA (Cell Signaling Technology, Frankfurt, Germany) according to the manufacturer’s instructions. In brief, 50 mg of porcine liver tissue or cell pellets (mouse or pig) of one well of a six-well plate were lysed in 500 µL of 1× lysis buffer. Lysates were diluted 1 : 2 by using sample-diluent. Next, 100 µL of diluted samples were incubated overnight at 4 °C in the microwells. After four washing steps (200 µL of 1× wash buffer), 100 µL of detection antibody were added to the wells for 1 h at 37 °C. After another four washing steps, 100 µL of HRP-linked secondary antibody were added for 30 min at 37 °C. Following four washings, 100 µL 3,3′, 5,5”-tetramethylbenzidine substrate were added for 10 min. After incubation with 100 µL stop solution, absorbance was measured at 450 nm using the GloMax®-Multi Detection System (Promega, Mannheim, Germany). Relative absorbance was normalized to 100 µg of total protein where indicated. Each sample was run in duplicate in the assay.

PF4

PF4 in pig plasma and liver tissue was determined with the PF4 ELISA kit obtained from ELK Biotechnology (Hölzel Diagnostika Handels GmbH, Köln, Germany) using the manufacturer´s instructions. Next, 100 mg of liver tissue were homogenized in 1 mL of ice-cold PBS and centrifuged at 10,000 × g for 5 min. One hundred microliters of the supernatant or 100 µL porcine plasma were incubated on the ELISA plates at 37 °C for 2 h, while shaking. Supernatants were sucked off and 100 µL of the diluted biotin-antibody conjugate were added and incubation continued for 1 h. Following three washing steps for 2 min each with wash buffer, 100 µL of HRP-Avidin conjugate were added and incubation continued for 1 h. Following another three washings, 90 µL of the substrate solution were added and incubation continued in the dark for 15–30 min. After adding 50 µL stop solution, absorbance was measured at 450 nm using using the GloMax®-Multi Detection System (Promega, Mannheim, Germany).

PDGF-AB

PDGF-AB in pig plasma and liver tissue was determined with the PDGF-AB ELISA kit obtained from CUSABIO (Hölzel Diagnostika Handels GmbH, Köln, Germany) using the manufacturer’s instructions. Liver samples were prepared and the procedure continued as described above for PF4.

THBS1

THBS1 at 24 h after ePHx was determined using the ELISA Kit for THBS1 obtained from Cloud-Clone Corp. (Katy, TX, USA) according to the manufacturer’s manual. Liver tissue samples were prepared as described above for PF4 and 100 µL of tissue extract or 100 µL of plasma were incubated on the ELISA plates at 37 °C for 2 h, while shaking. Supernatants were sucked off and 100 µL of the diluted detection reagent A were added and incubation continued for 1 h. Following three washing steps for 2 min each with wash buffer, 100 µL of diluted detection reagent B were added and incubation continued for 30 min. Following another three washings, 90 µL of the substrate solution were added and incubation continued in the dark for 10–20 min. After adding 50 µL stop solution, absorbance was measured at 450 nm using the GloMax®-Multi Detection System (Promega, Mannheim).

Semi-quantitative RT-PCR

Total RNA was extracted using a standard Trizol protocol. The cDNA synthesis was performed with the Maxima-H-Minus-First-Strand kit (Thermo Fisher, Dreieich, Germany). Then, 100 ng of cDNA were amplified and PCR products analyzed after electrophoretic separation by quantifying the relative intensity of specific bands using ImageJ (v1.42, National Institutes of Health, Bethesda, MA, USA) where indicated. A list of primers is presented in Supplementary Table 2.

Cytokine array analysis

Cytokines in supernatants of THC were detected using 100 µl of medium according to the manufacturer’s procedure using the Proteome Profiler Human XL Cytokine Array Kit (R&D Systems, Abingdon, UK). Next, 100 µL of CM-THC either with or without CM-MSC were applied according to the manufacturer’s instructions. The array detected proteins as listed in Supplementary Table 3. Labeled proteins were visualized with the Micro Chemi 4.2 using the gel capture software (Biostep, Burkhardtsdorf, Germany) by taking serial pictures with an exposure time of 9 min. Signal intensities were quantified using the ImageJ 1.46 software (NIH, Bethesda, MD, USA) and background correction by the negative controls. Mean pixel density of reference spots was set to 100, to which all other values are relative. Proteins below an abundance of 5 were ignored. For better clarity, Fig. 9b does only comprise cytokines known to be contained in THC α-granules.

Statistics

If not otherwise indicated in the figure legends, the SPSS software (v24, IBM, Ehningen, Germany) was used for statistical analyses. To verify statistical standard distribution, the t-test or the one-way analysis of variance test were applied. Values were considered different at P ≤ 0.05.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.