In vitro effects of Type I interferons (IFNτ and IFNα) on bovine hepatocytes cultured with or without Kupffer cells

Keywords: cattle, cell culture, embryo, embryo-maternal communication, endocrinology, gestation, hepatocytes, IGF binding protein, insulin-like growth factor, interferone tau

In sheep and cattle, there is evidence that interferon (IFN)-τ reaches the blood stream and induces expression of IFN-stimulated genes (ISG) in the endometrium and other organs. In 2008, an ‘antiviral activity’ was detected in the uterine vein in sheep at Day 15 of pregnancy (Oliveira et al. 2008). Two years later this ‘antiviral activity’ was confirmed to be IFNτ, and ISG expression was detected in ovarian tissue (Bott et al. 2010). Recently, significantly higher gene expression of ISG-15 and MX dynamin like GTPase 1 (MX1) and MX dynamin like GTPase 2 (MX2) was noted in cervical and vaginal tissues on Days 17–18 after AI in cows (Kunii et al. 2018). Increased expression of ISG genes has also been detected in leucocytes between Days 15 and 32 after insemination in both sheep (Yankey et al. 2001) and cattle (Han et al. 2006). This can be assumed to be in response to IFNτ. Similarly, several studies indicated a difference in ISG-15 expression in the liver between pregnant and non-pregnant sheep on Days 14 (Romero et al. 2015) and 15 (Bott et al. 2010). In cows, studies in Holstein Friesian and Angus heifers detected increased mRNA levels of ISG (MX1, MX2, ISG-15 and oligoadenulate synthetase 1(OAS-1)) in the liver on Day 18 of pregnancy compared with non-pregnant heifers (Meyerholz et al. 2015; Ruhmann et al. 2017). Ruhmann et al. (2017) also demonstrated OAS-1 protein expression in hepatocytes via immunohistochemistry. However, there is still no direct proof (e.g. immunoassays) of IFNτ in the blood circulation. Nakamura et al. (2016) detected IFNτ in extracellular vesicles and discussed its involvement in the regulation of ISGs.

MX1, MX2, ISG-15 and OAS-1 are classical ISGs. They are induced in response to IFNτ after binding to Type 1 IFN receptors (interferon-α/β receptor (IFNAR1/2)) via induction of the Janus tyrosine kinase (JAK) signal transducers and activators of transcription (STAT) pathway. In addition, Type I IFNs stimulate increased expression of other genes via non-classical signalling pathways. Arosh et al. (2004) showed that IFNτ reduced the expression of prostaglandin F synthases on Day 16 of pregnancy in cows. In contrast, increased expression of the non-classical ISG chemokine (C-X-C motif) ligand 6 (CXCL-6) in response to IFNτ was shown in endometrial cells (Teixeira et al.1997).

The present study focused on two chemokines that are traditionally produced during an inflammatory reaction. The chemokine CXCL-10 is produced in Kupffer cells (KCs) and induces decreased angiogenesis, whereas CXCL-5 supports angiogenesis and activates neutrophil granulocytes (Fernandez and Lolis 2002; Dembic 2015). Metabolic genes also seem to be affected by IFNτ. For example, there were changes in the expression of insulin like growth factor 1 (IGF-1) and IGF binding protein 2 (IGFBP-2) mRNA expression in the endometrium (Robinson et al. 2008) and peripheral blood (Meyerholz et al. 2015) during early pregnancy in cows. An effect of IFNτ on the somatotropic axis can therefore be speculated, because IGFBP-2 is mainly produced in the liver. However, it is still unclear whether non-classical ISG expression is induced by IFNτ in bovine hepatocytes and whether it is affected by local liver KCs or progesterone (P4). KCs are located in the endothelium of liver sinusoids and displace the largest population of tissue macrophages. KCs play an important role in host defence and stay in close paracrine contact to hepatocytes (Naito et al. 2004). It remains unknown whether all IFNτ effects are specific or whether other Type I IFNs can induce comparable expression of ISGs. IFNτ belongs to the group of Type I IFNs (IFNα, IFNβ, IFNϵ, IFNκ and IFNω; Roberts et al. 1998; Capobianchi et al. 2015) and shows 45–55% homology to IFNs of the IFNα group and 70.3% homology to bovine IFNα (Imakawa et al. 1987). All Type I IFNs bind to the same receptor (IFNAR1/2), but stimulate the expression of different ISGs downstream (Mogensen et al. 1999; Uzé et al. 2007).

In this study, a well-established bovine liver cell culture (Ehrhardt and Schmicke 2016; Witte et al. 2019) was used and a KC coculture was established to test the endocrine effects of IFNτ. The following questions were investigated: (1) is the liver cell culture suitable for testing the effects of Type I IFNs; (2) does the Type I IFN effect differ between pure hepatocytes and hepatocytes cocultured with KCs; (3) does the expression of ISGs differ between hepatocytes stimulated with IFNτ and other Type I IFNs; and (4) does simultaneous stimulation with P4 and IFNτ affect the expression of single target genes in the maternal liver?

Between March 2017 and July 2018, 15 bovine livers (six obtained at an abattoir, nine that had been ‘patients’ of the Clinic for Cattle at the University of Veterinary Medicine Hannover; see below) were used to establish a hepatocyte coculture with KCs without dexamethasone. A further six livers from female pluriparous Holstein Friesian cows (also clinic ‘patients’) were used to investigate the effects of type I IFNs on bovine hepatocytes as a model for the metabolic effects of early pregnancy signals in ruminants.

All cows used in this study that had been patients of the Clinic for Cattle at the University of Veterinary Medicine Hannover were killed for medical reasons, not including any signs of metabolic disease and/or general inflammation with fever. Within 20 min (abattoir) or 5 min (clinic ‘patients’) of death, the caudate process of the caudate lobe of the liver was removed. It was then rinsed with 200 mL ice-cold EGTA buffer, followed by 200 mL EGTA-free perfusion buffer. Hepatocytes were isolated using a two-step collagenase perfusion as described previously (Witte et al. 2019), with one modification: in order to test the effects of Type I IFNs without bias, all buffers were used without addition of dexamethasone.

KC were also extracted from the livers using the cell suspension that was obtained after collagenase perfusion. In addition to hepatocytes, this suspension contained all non-parenchymal cells of the liver. This total cell suspension was centrifuged at 300g for 5 min at 4°C. The pellet was resuspended in William’s E medium (PAN-Biotech) and 10% fetal bovine serum (FBS; PAN-Biotech). According to the protocol of Smedsrød et al. (1985), the KCs were further purified by two-density gradient centrifugation (gradient: 5 mL of 50% Percoll (Sigma-Aldrich), 5 mL of 25% Percoll and 5 mL resuspended pellet). After centrifugation (1250g, 8 min, 4°C), the fraction between the two layers of Percoll contained KCs and fibroblasts. Further purification was based on selective adhesion properties between the two cell types on plastic surfaces (Smedsrød et al. 1985). After 20 min adhesion time at 37°C under 95% O₂ and 5% CO₂, cells were washed twice with phosphate-buffered saline (PBS) and harvested by the addition of 1 mL of 10% trypsin (Sigma-Aldrich) to the cells. Williams E Medium (PAN-Biotech) with 10% FBS (PAN-Biotech) was used to stop the enzymatic reaction, and the cells were centrifuged at 1300g for 1 min at 4°C. The cell pellet was resuspended in 1 mL Williams E Medium (PAN-Biotech) and cell vitality was tested by Trypan blue exclusion. The number of KCs was counted under a microscope (Fig. 1).

Fig. 1.  After collagenase perfusion of the liver, the cell suspension (1) containing hepatocytes, Kupffer cells (KCs) and other non-parenchymal cells (NPC) was centrifuged (3 min, 60g) and divided into an NPC supernatant (2a) and a sediment containing hepatocytes (2b). The NPC fraction was again centrifuged (5 min, 300g; 3) and the resulting pellet was loaded into two density gradients (25% and 50% Percoll; 4). The next centrifugation (5 min, 2300g) led to accumulation of KCs and endothelial cells between the two density layers (4a). After collecting this middle layer, the cells were again centrifuged (5 min, 900g; 4b) and the resulting pellet was placed on a glass Petri dish (4c). In the sediment, hepatocytes were counted (2b) and a 25% Percoll gradient was loaded with 5000–6000 cells (5). After centrifugation (8 min, 1000g), hepatocytes in the pellet (6) were seeded onto collagen-coated plates (7).

In order to find the most suitable model, two different coculture methods were tested and compared with a hepatocyte monoculture (HMC) as a control. Hepatocytes were cocultured with KCs at a ratio of 6:1 (0.5 million hepatocytes:0.15 million KCs per well) in direct (HKCd) or indirect contact (HKCid). The three different models are shown in Fig. 2.

Fig. 2.  Illustration of two different coculture methods for bovine hepatocytes (H) and Kupffer cells (KCs) in (a) direct (HKCd) or (b) indirect (HKCid) contact compared with (c) hepatocyte monoculture (HMC). H are represented by squares, whereas KC are represented by stars. The light grey line represents the collagen layers and the dark grey line represents the medium.

Six livers were used to test whether the basal expression of genes of interest is comparable between hepatocytes extracted from different donors and on different days. To this end, hepatocytes were immediately frozen at –80°C after Percoll density centrifugation. Cell samples were also taken from HMC and HKCid on Days 1, 2, 3 and 4 of culture. The mRNA expression of MX1, ISG-15, CXCL-10, CXCL-5, IGFBP-2, IGF-1 and albumin was analysed by reverse transcription–quantitative polymerase chain reaction (RT-qPCR) and compared between the six donors. Concentrations of lactate dehydrogenase (LDH) and urea in the culture supernatant were measured as indicators of vitality.

HMC and HKCid were established from five consecutive livers and cultured for 4 days. Each stimulation condition was evaluated in 12 wells. Two wells were pooled prior to analysis. The culture medium was changed daily. On Day 4, cells were stimulated for 6 h with 0.06 or 0.6 ng well^–1 IFNτ or IFNα. Other cells were stimulated simultaneously with 0.6 ng well^–1 IFNτ and 0.6 ng well^–1 IFNα or 0.06 ng well^–1 IFNτ and 0.06 ng well^–1 IFNα. Cells were also stimulated with 0.6 ng well^–1 IFNτ with or without 5 or 10 ng well^–1 P4 progesterone and were compared with cells stimulated with 5 and 10 ng well^–1 P4 alone. Cells and supernatants were stored at –80°C until further processing.

To evaluate the vitality of the hepatocytes, cells were stained using a LIVE/DEAD Viability/Cytotoxicity Kit (ThermoFisher Scientific) according to the manufacturer’s instructions and evaluated under an inverted fluorescence microscope with phase contrast (Axiovert 200M) using AxioVision 4.8.2.0 software. The proportion of live and dead cells was determined by counting 100 hepatocytes in different visual fields (various fields were used until 100 cells were counted).

For staining, hepatocytes were cultured on collagen-coated coverslips. Cells were collected on Days 1 and 5. A cryosection of bovine liver was used as a control. Cells were fixed using methanol and paraformaldehyde (PFA). Methanol (at –20°C) was added to the cultures for 10 min. Next, the cells were rehydrated in Tris-buffered saline (TBS) for 5 min, followed by treatment with 10 mL of 10% PFA. The cells were then washed three times with TBS and non-specific interactions were blocked using 5% skimmed milk with 20% Tween. After 45 min, the blocking solution was removed and primary antibodies were added to the cells (1 : 1000 dilutions; mouse anti-CK18 monoclonal antibody (C8541; Sigma-Aldrich) and rabbit anti-panCK monoclonal antibody (17916; Sigma-Aldrich). Cells were incubated overnight at 4°C, then washed with TBS for 5 min and then incubated with secondary fluorescence antibodies for 3 h at room temperature (goat anti-rabbit IgG (green; Alexa Fluor 488) and goat anti-mouse IgG (red; Alexa Fluor 546); dilution of 1:2000). Cells were incubated with 500 µL staining solution for 45 min, then washed with Tris-buffered saline Tween-20 (TBST) for 5 min, followed by 4 min incubation with Hoechst dye nucleus stain (1:1000 in TBS; 33342; ThermoFisher Scientific). After a final washing step, cover slides were gently removed from the cell culture plate and mounted on a microscope slide with Prolong Gold anti-fade reagent (P36930; Invitrogen). Cells were examined under an inverted fluorescence microscope (Zeiss Axiovert 25 with phase contrast).

To assess the vitality of the gathered KCs in the culture model, specific latex beads (2.0 µm; L2778; Sigma-Aldrich) were used as vitality markers. Vital KCs phagocytose fluorescent microspheric polystyrol particles, which can be detected microscopically. To avoid non-specific binding, latex beads were opsonised by incubation for 1 h at 37°C with 1:1 PBS/FBS. HMC and HKCid were incubated with latex beads at a ratio of 1:40 (KCs:latex beads) for 24 h. Then, plates were washed twice with PBS to remove any non-phagocytosed beads. The phagocytosed particles were counted using a fluorescence microscope with phase contrast equipment (Zeiss Axiovert 25) in four fields of view per well.

Urea concentrations were measured using a Urea Assay Kit (ab83362; Abcam) and LDH activity was measured using a Lactate Dehydrogenase Activity Assay Kit (Sigma-Aldrich), as described previously (Witte et al. 2019). Both assays were performed according to the manufacturers’ instructions (https://www.sigmaaldrich.com/catalogue/product/sigma/mak066?lang=deandregion=DE). The intra-assay CVs were 6.1% for urea and 5.2% for LDH.

The expression of S18, MX1, ISG-15, CXCL-10, CXCL-5, IGFBP-2, IGF-1 and albumin mRNA in cell samples was determined using RT-qPCR. To this end, total RNA was extracted using TRIzol reagent (Sigma-Aldrich), as described previously (Ehrhardt and Schmicke 2016). The extracted RNA was quantified using a NanoDrop One spectrophotometer (ThermoFisher Scientific). RNA quality was checked in HKCid samples using the 2100 Bioanalyzer System (Agilent 6000 Nano Kit). All samples had RNA integrity numbers (RIN) >8.5. A CFX96 C1000 Touch thermocycler (Real-Time PCR Detection System; Bio-Rad) was used to reverse transcribe mRNA into cDNA, as described previously (Witte et al. 2019). Each PCR contained 2.5 ng µL^–1 cDNA, 4 µL HOT FIREPol EvaGreen qPCR Supermix (5×; Solis BioDyne) and 0.4 µM forward and reverse primers (Eurofins MWG Operon). The primer sequences and Accession numbers are listed in Table 1. The PCR cycler was programmed as follows: activation of the DNA polymerase at 95°C for 10 min, followed by 40 cycles of RNA denaturation at 95°C for 15 s, annealing at 60°C for 30 s and elongation at 72°C for 30 s. A melting curve analysis was performed between 55 and 95°C in 0.5°C increments to confirm the identity of the amplicons. Gene expression levels were determined by relative quantification (ΔΔCt method) using S18 as a reference gene. Determinations were performed in duplicate.

Table 1.  Genes of interest and primers used in RT-qPCR
MX1, Myxovirus-resistance protein 1; ISG-15, interferon-stimulated gene 15 kDa; CXCL-5, chemokine (C-X-C motif) ligand 5; CXCL-10, chemokine (C-X-C motif) ligand 10; IGF-1, insulin-like growth factor-1; IGFBP-2, insulin-like growth factor-binding protein 2

Statistical analyses were performed with GraphPad Prism 8. All data were tested for normal distribution using the Shapiro–Wilk normality test. Normally distributed data were evaluated using Student’s t-test. Data that were not normally distributed were evaluated using the Mann–Whitney U-test. The mRNA data are presented as ΔΔCt relative to a calibrator with respect to the different efficiencies of the primers. Data are presented as the mean ± s.d. if data was normally distributed, or mean ± 95% CI if data was not normally distributed. Differences were considered significant at two-tailed P < 0.01.

On average, 23 827 812 hepatocytes were isolated per gram liver tissue, with 84% viability, and 387 340 KCs were isolated (Table 2). The hepatocytes showed typical cell morphology and vitality. Initially, KCs were morphologically round with a diameter of 10 µm (Fig. S1, Fig. S2). After Day 1, the cells developed typical stellate cell extensions (Fig. S1). Liver-specific markers were detected by immunostaining on Days 1 and 5 with CK18 and panCK (Fig. 3). On Day 2, 41.3% of KCs contained phagocytosed fluorescent latex beads. This decreased to 25.1% on Day 3 and to 13.6% on Day 4 (Fig. 4).

Table 2.  Results of viability tests and the quantity of hepatocytes and Kupffer cells obtained from of 21 perfused livers
Livers were obtained from an abattoir (Long warm ischaemia interval) or from cows that had been ‘patients’ of the Clinic for Cattle at the University of Veterinary Medicine Hannover and were killed for the medical reasons listed (short warm ischaemia interval). Asterisks indicate that viability was only estimated on the basis of the test procedure, but was always above 95%.

Fig. 3.  Hepatocytes on Day 1 (a–d) and Day 5 (e–h) in sandwich monoculture stained with CK18 (red) and panCK (green) to confirm liver-specific cell surface markers in culture. (i–l) A cryosection used as a control. Intact nuclei were stained blue using a Hoechst stain.

Fig. 4.  Inverse phase contrast images of hepatocytes in indirect coculture with Kupffer cells (HKCid), with (a, b) and without (c) fluorescence. Viable Kupffer cells can be recognised by phagocytosed red fluorescent beads.

Mean viability after Percoll density centrifugation was 91.4 ± 1.9% in the six livers used for the experiments with IFNτ. The mean urea concentration in each experiment was comparable, with concentrations of 12.43 ± 0.94, 12.26 ± 1.25 and 12.62 ± 1.55 nmol well^–1 in ES1 (HKCd), ES2 (HKCid) and ES3 (HMC) respectively. LDH concentrations were also comparable in each experiment: 3.32 ± 0.51, 1.79 ± 0.84 and 4.97 ± 2.89 mU well^–1 in ES1, ES2 and ES3 respectively. There were also no significant differences in albumin expression (9.62 ± 1.85, 10.77 ± 2.21, 11.60 ± 1.91 and 12.09 ± 1.92 for basal albumin expression and expression in ES1, ES2 and ES3 respectively).

The intra-assay CVs for basal expression were 7.5% for MX1, 11.6% for ISG-15, 5.9% for CXCL-5, 7.4% for CXCL-10, 12.5% for IGF-1 and 7.9% for IGFBP-2 (Table 3). The relative mRNA expression of the target genes did not differ significantly between the six livers analysed in the stimulation experiments.

Table 3.  Basal expression of target genes in six livers (for two samples per liver)
MX1, Myxovirus-resistance protein 1; ISG-15, interferon-stimulated gene 15 kDa; CXCL-5, chemokine (C-X-C motif) ligand 5; CXCL-10, chemokine (C-X-C motif) ligand 10; IGF-1, insulin-like growth factor-1; IGFBP-2, insulin-like growth factor-binding protein 2

The mRNA expression of ISGs (MX1, ISG-15, CXCL-5 and CXCL-10) increased in mono- and cocultures during the first days of culture until Day 3, but decreased sharply when morphological restructuring of hepatocytes was finished on Day 4 (Fig. 5). On Day 4, the expression of the inflammatory-/IFN-related genes MX1, ISG-15, CXCL-5 and CXCL-10 was back to baseline levels in HMC and HKCid. In contrast, mRNA levels of the metabolic genes (IGF-1 and IGFBP-2) increased towards Day 3 and remained higher on Day 4 than on Day 1. The overall expression profiles were comparable between mono- and cocultures (Fig. 5).

Fig. 5.  Relative mRNA expression of (a) Myxovirus-resistance protein-1 (MX1), (b) interferon-stimulated gene 15 kDa (ISG-15), (c) chemokine (C-X-C motif) ligand 10 (CXCL-10), (d) chemokine (C-X-C motif) ligand 5 (CXCL-5), (e) insulin-like growth factor 1 (IGF-1) and (f) insulin-like growth factor-binding protein 2 (IGFBP-2) in hepatocyte monoculture (HMC) and indirect coculture of hepatocytes with Kupffer cells (HKCid) on Days 1, 2, 3 and 4. The mRNA data are presented as ΔΔCt relative to a calibrator. The boxes show the interquartile range, with the median value indicated by the horizontal line; whiskers show the range (n = 6). *P < 0.01, **P < 0.001, ***P < 0.0001.

The addition of 0.6 ng mL^–1 IFNτ upregulated ISG-15 and MX1 mRNA levels in HMC and HKCid (Fig. 6a, b). CXCL-10 mRNA levels increased sharply in HMC stimulated with 0.6 ng mL^–1 IFNτ or IFNα, but not in HKCid after stimulation with 0.6 ng mL^–1 IFNτ. In HKCid, the addition of 0.06 ng mL^–1 IFNα resulted in significant upregulation of CXCL-10 mRNA expression (P = 0.0041; Fig. 6c). Stimulation of HMC with 0.06 ng mL^–1 IFNτ or IFNα resulted in higher CXCL-5 mRNA expression than without stimulation (Fig. 6d). The addition of IFNα increased IGFBP-2 mRNA expression exclusively in HKCid, whereas the addition of IFNτ reduced the expression of IGFBP-2 (Fig. 6f).

Fig. 6.  Relative mRNA expression of (a) Myxovirus-resistance protein-1 (MX1), (b) interferon (IFN)-stimulated gene 15 kDa (ISG-15), (c) chemokine (C-X-C motif) ligand 10 (CXCL-10), (d) chemokine (C-X-C motif) ligand 5 (CXCL-5), (e) insulin-like growth factor 1 (IGF-1) and (f) insulin-like growth factor-binding protein 2 (IGFBP-2) in hepatocyte monoculture (HMC) and indirect coculture of hepatocytes with Kupffer cells (HKCid) stimulated with 0.6 or 0.06 ng mL^–1 IFNα (α) or bovine recombinant IFNτ (τ) for 6 h on Day 4 of culture. The mRNA data are presented as ΔΔCt relative to a calibrator. The boxes show the interquartile range, with the median value indicated by the horizontal line; whiskers show the range (n = 6). *P < 0.01, **P < 0.001, ***P < 0.0001.

The addition of P4 alone had no effect on MX1 and CXCL-10 mRNA expression in HMC or HKCid (Fig. 7a, c). However, CXCL-5 mRNA levels increased significantly after stimulation with 10 ng mL^–1 P4 (P = 0.0055; Fig. 7d). The addition of IFNτ increased MX1, ISG-15, CXCL-10 and CXCL-5 mRNA expression, whereas additional stimulation with 5 ng mL^–1 P4 reduced the expression of these four genes. This reduction was more pronounced when the P4 concentration was increased to 10 ng mL^–1 (Fig. 7a–d). In HKCd, IGF-1 and IGFBP-2 mRNA expression was comparable in the presence and absence of P4, but decreased after stimulation with 0.6 ng mL^–1 IFNτ alone. However, this decrease was reversed by the addition of 5 ng mL^–1 P4 to the HKCd (Fig. 7e, f).

Fig. 7.  Relative mRNA expression of (a) Myxovirus-resistance protein-1 (MX1), (b) interferon (IFN)-stimulated gene 15 kDa (ISG-15), (c) chemokine (C-X-C motif) ligand 10 (CXCL-10), (d) chemokine (C-X-C motif) ligand 5 (CXCL-5), (e) insulin-like growth factor 1 (IGF-1) and (f) insulin-like growth factor-binding protein 2 (IGFBP-2) in hepatocyte monoculture (HMC) and indirect coculture of hepatocytes with Kupffer cells (HKCid) stimulated with 5 or with 10 ng mL^–1 progesterone (P4) and 0.6 ng mL^–1 bovine recombinant IFNτ (τ) for 6 h on Day 4 of culture. The mRNA data are presented as ΔΔCt relative to a calibrator. The boxes show the interquartile range, with the median value indicated by the horizontal line; whiskers show the range (n = 6). *P < 0.01, **P < 0.001, ***P < 0.0001.

The first aim of the present study was to establish a hepatocyte coculture with KCs to test whether hepatocytes respond to IFN directly or whether KCs transduce the signal instead. The hepatocyte sandwich culture had been established previously (Witte et al. 2019). The present work confirmed that, in a sandwich culture system, the shape of hepatocyte cells changed to cuboidal and polygonal, and hepatocytes formed cords reminiscent of the hepatic trabeculae with bile canaliculi, which can be detected as bright elucidations between hepatocytes. This has already been reported by Witte et al. (2019). The separation of sessile KCs from bovine livers was established using a previously established protocol (Smedsrød et al. 1985). KCs in culture were viable, as indicated by the phagocytosis of fluorescent latex beads. In HKCid cocultures, KC could be detected based on phagocytosed latex beads on Days 2, 3 and 4, suggesting immunologically active cells. KCs are located in the endothelium of liver sinusoids and are responsible for the phagocytosis of pathogens before they enter the liver, as well as the elimination of apoptotic endothelial cells. KCs are normally in an inactive state, but can be activated by inflammatory signals (Duque and Descoteaux 2014). The activation of sessile KCs can be differentiated into classical activation (e.g. by IFNγ or toll-like receptors (TLRs)/interleukin (IL)-1 receptors). This results in the production of proinflammatory cytokines, such as IL-1, IL-6, tumour necrosis factor-α, transforming growth factor-β, IFNγ, epidermal growth factor and leukaemia inhibitory factor (Gordon and Martinez 2010). In addition to the classical activation, IL-4 and/or IL-13 can also activate M2 macrophages (Stein et al. 1992; Wynn et al. 2013) to exhibit anti-inflammatory functions. The third group of activated KCs are regulatory KCs, which need two activation signals: a priming-signal from apoptotic cells, IL-10, prostaglandins or glucocorticoids and a second signal via their TLRs (Mosser and Edwards 2008). It would be interesting to determine whether the sessile KCs in the hepatocyte culture are inactivated or activated, but unfortunately mRNA expression analyses of factors differentiating between the two states were not performed in the present study.

The basal expression of the genes of interest varied less than 13% between the six livers from different donors. This leads to the assumption, that regardless of the donor animal, mono- and cocultures of hepatocytes lead to comparable results. During the 4 days of incubation, the expression profiles of MX1, ISG-15, CXCL-5 and CXCL-10 were also comparable. The expression of inflammatory genes was initially increased, which paralleled the morphological changes noticeable in the cultures and is suggestive of a proinflammatory state. Separation of hepatocytes leads to an increase in nuclear factor (NF)-κB expression (Li and Verma 2002; Vinken et al. 2014). NF-κB is an important regulator of inflammation (Li and Verma 2002) and may indirectly induce ISG via IFNβ (Sen and Sarkar 2007; Oeckinghaus and Ghosh 2009). Although not further examined in the present study, this may be a possible pathway for the upregulation of the chemokines investigated in this study.

Within the first 3 days of incubation, hepatocytes developed the specific polygonal cell shape, cell–cell contacts and canaliculi (Ehrhardt and Schmicke 2016; Witte et al. 2019). In parallel with an end to the morphological changes, the expression of inflammatory genes decreases. In agreement with Ehrhardt and Schmicke (2016) and Witte et al. (2019), this led us to conclude that Day 4 of culture is most suitable for experiments in primary bovine hepatocytes. Moreover, the continuously increasing IGFBP-2 mRNA expression levels throughout the 4 days of incubation indicate that the highest rate of metabolic synthesis is achieved on Day 4.

Once established, the HMCid model was used to investigate the following questions: (1) is liver cell culture suitable for testing the effects of Type I IFNs; (2) does the effect of Type I IFNs differ between pure hepatocytes and hepatocytes cocultured with KCs; (3) does the expression of ISGs differ between hepatocytes stimulated with IFNτ and other Type I IFNs; and 4) does simultaneous stimulation with P4 and IFNτ affect the expression levels of individual target genes in the maternal liver?

As suspected, and previously shown in hepatocytes, stimulation with IFNτ induced classical ISGs (MX1 and ISG-15; Ruhmann et al. 2017). The stimulation was dose dependent and did not depend on the presence of sessile KCs because the expression levels were comparable between the HMC and HKCid models. Others have previously reported an increase in CXCL-10 expression in response to IFNα in human thyrocytes (Antonelli et al. 2010) and in response to IFNτ in sheep endometrial cells (Imakawa et al. 2006). In line with these findings, CXCL-10 mRNA expression was increased in response to the higher concentration of 0.6 ng mL^–1 IFNτ in hepatocytes in HMC in the present study. CXCL-10 expression has previously been identified in KCs and in the uterus of sheep during the peri-implantation period (Fernandez and Lolis 2002; Imakawa et al. 2006; Dembic 2015). It has previously been shown that the endometrial IGF system changes in early pregnancy, and IFNτ has been speculated as a possible factor for changes in the endometrial gene expression (Robinson et al. 2008). Following on from this, Meyerholz et al. (2015) reported a decrease in IGFBP-2 during early pregnancy (towards Day 18 of gestation) in dairy heifers. IGFBP-2 isolated from blood samples is mainly produced in the liver. Therefore, the hypothesis that IFNτ directly affects hepatic IGFBP-2 production was confirmed in the present study. The addition of a low concentration (0.06 ng mL^–1) of IFNτ to the HMC for 6 h led to a reduction in IGFBP-2 mRNA. Furthermore, the present study examined differences in responses to IFNτ compared with other Type I IFNs. For this comparison, IFNα was chosen as an alternative Type I IFN because it is normally produced by nearly all nucleated cells, especially leucocytes (Ohmann et al. 1987).

IFNα was shown to induce CXCL-5 and CXCL-10 mRNA expression but not the expression of MX1 and ISG-15. This appears to be a specific IFNτ pattern despite the fact that all Type I IFNs bind to the same receptor, namely IFNAR1/2 (Li and Roberts 1994; Mogensen et al. 1999; Uzé et al. 2007). Future studies will need to show why ISG-15 and MX1 mRNA expression is exclusively induced by IFNτ in primary hepatocytes and whether comparable effects can be observed in vivo. However, should be noted that in the present study only mRNA expression was evaluated, and not protein expression. This aspect should also be further investigated in future studies.

During early pregnancy, ISG expression is dependent on steroid hormones like P4 (Spencer et al. 2016). In heifers supplemented with P4 on Day 16 of pregnancy, ISG-15 and CXCL-10 mRNA expression was higher in uterine epithelial cells than in heifers with lower endogenous P4 concentrations (Forde et al. 2009). In the present study, based on physiological P4 concentrations on Day 16 of pregnancy (5–10 ng mL^–1; Matsuyama et al. 2012), concentrations of 5 and 10 ng mL^–1 P4 were used in the HKCid cultures to detect potential effects not only in reproductive tissues, but also in the liver. In liver cells, only ISG-15 mRNA expression was slightly increased in response to P4. There was no change in MX1 and CXCL-10 mRNA expression after stimulation with P4. After the addition of 5 ng mL^–1 P4 and 0.6 ng mL^–1 IFNτ to cultures, ISG-15 expression was higher than after the addition of 10 ng mL^–1 P4 and 0.6 ng mL^–1 IFNτ. This is in contrast to the results in endometrial tissue reported by Forde et al. (2009), who detected higher ISG-15 and CXCL-10 mRNA expression in samples stimulated with higher concentrations of P4. However, the P4 concentrations used by Forde et al. (2009) were higher in vivo than the concentrations used in the present in vitro study. Moreover, different cell types were used and the reaction to P4 may differ between the endometrium and liver. Further studies should also focus on respective hormone receptor expression and intracellular signal transduction pathways, which may differ in different tissues. Interestingly, CXCL-5 mRNA expression was only stimulated by P4 and the additional effect of IFNτ was lower compared with that on the other genes. Increased CXCL-5 expression in response to P4 was previously reported in vaginal tissue of mice after ovariectomy (Gillgrass et al. 2005). This led to speculation about the presence of a protection mechanism for pregnant animals against infectious particles during pregnancy. This is imaginable for the liver as well, because CXCL-5 attracts and activates neutrophils and induces an immune response (Dembic 2015). However, P4 receptors were not analysed in the present study, and therefore the comparison with reproductive tissue is debatable. After stimulation of the HKCid with IFNτ and P4, IGF-1 mRNA expression decreased, which is in line with previous studies. McCarthy et al. (2012) showed that IGF-1 expression on Day 16 after ovulation decreased significantly in pregnant but not cycling cows compared with expression on Day 5. Several other studies also detected lower IGF-1 expression during the luteal phase in endometrial (Meikle et al. 2001) and serum (Mense et al. 2015) samples.

The expression of IGFBP-2 was lower in HMC and HKCid after stimulation with P4 and IFNτ. As mentioned above, Meyerholz et al. (2015) also detected a decrease in blood IGFBP-2 in early pregnant cows (Day 14). However, McCarthy et al. (2012) detected an increase in IGFBP-2 mRNA expression in the endometrium from Day 7 towards Day 13. Because the liver is the main source of IGFBP-2, it can be hypothesised that lower peripheral IGFBP-2 may lead to higher free IGF-1 concentrations, whereas higher local IGFBP-2 may bind the additional IGF-1, which is essential for fetal growth (Lassarre et al. 1991).

In conclusion, primary bovine hepatocyte cultures were proven to be a suitable model for stimulation experiments with Type I IFNs and investigation of extrauterine embryo–maternal communication. The results of the present study show that there is a significant increase in MX1, ISG-15, CXCL-5, CXCL-10, IGF-1, IGFBP-2 mRNA expression in hepatocytes in response to stimulation with IFNτ. Thus, IFNτ has been confirmed to have a direct effect on the mRNA expression of target genes in hepatocytes. The KCs in cocultures seemed to make no major contribution to the effect of IFNτ on hepatocytes. In contrast, hepatocytes seem to respond differently when P4 was added to the cultures. Although MX1 und ISG-15 mRNA expression was stimulated by IFNτ in both hepatocytes and endometrial cells, the downregulation of IGFBP-2 mRNA expression was only detected in hepatocytes and not in the endometrium. The proposed endocrine action of IFNτ in the maternal liver may affect maternal metabolism, as well as maternal immune function in the liver. Further investigations are needed to elucidate the underlying mechanisms, which may be relevant for pregnancy establishment, and further research is required to examine potential systemic and endocrine effects on other organs during the early pregnancy in ruminants.

The authors declare no conflicts of interest.