Authors: Jonathan G. Pol, Bouchra Lekbaby, François Redelsperger, Sofieke Klamer, Yassmina Mandouri, James Ahodantin, Ivan Bieche, Marine Lefevre, Philippe Souque, Pierre Charneau, Noémie Gadessaud, Dina Kremsdorf, and Patrick Soussan

Published Online:28 Jan 2015https://doi.org/10.1096/fj.14-258715

Abstract

MATERIALS AND METHODS

Tg mice

To generate HBSP Tg mice, the HBSP gene (333 bp) encoding for the HBSP was cloned downstream of the weak HBx promoter/enhancer I (nt 832 and 1371 on HBV genome), restricting expression to hepatocytes (16), and a rabbit β-globin intron. Two independent strains, HBSP1 and HBSP2, were generated (Institut Clinique de la Souris, Strasbourg, France), expanded against the C57BL/6J background (12 backcrossings) and characterized (Supplementary Fig. 3B, C). All experiments were performed on heterozygous Tg males. HBSP Tg and control mice were treated with intraperitoneal injection of purified hamster anti-mouse Fas/CD95 (Jo2, 0.5 μg/g; Pharmingen/BD Biosciences, San Jose, CA, USA), or with TNF-α administered intravenously (10 μg/kg; R&D Systems, Minneapolis, MN, USA) with or without intraperitoneal injection of D-galactosamine (D-Gal; 700 mg/kg; Sigma-Aldrich, St. Louis, MO, USA) to induce acute liver inflammation. Chronic liver inflammation was induced by intraperitoneal injection of LPS (2.5 mg/kg; Sigma-Aldrich) 3 times a week for 9 months. Animals were maintained in pathogen-free conditions on 12 hour dark/light cycles and treated in accordance with EU regulations on animal care (Directive 86/609/EEC). All procedures were approved by the local animal care and use committee (Agreement A75-14-08).

Cell cultures

Human hepatoma HepG2 and HeLa cells (American Type Culture Collection, Manassas, VA, USA) were grown in DMEM with GlutaMAX (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal calf serum (FCS) and 100 U/ml penicillin-streptomycin (Invitrogen). Depending on the experiment, the cells were treated with 2 µM proteasome inhibitor MG132 (Biomol, Farmingdale, NY, USA) or with recombinant human cytokines: 5 ng/ml TGF-β1 (Sigma-Aldrich), 10 to 25 ng/ml TNF-α (R&D Systems, Minneapolis, MN, USA), 0.3 ng/ml IL-1β (Sigma-Aldrich), and treated with or without 1 µg/ml actinomycin D (ActD; Sigma-Aldrich).

Primary mouse hepatocytes from wild-type and Tg mice were isolated after liver perfusion with 0.04% collagenase and plated in 10% FCS with complete Williams medium (standard medium supplemented with 10% bovine serum albumin, 25 nM dexamethasone, 12.5 nM insulin and antibiotics). After washing, cells were cultured in 6-well plates in unsupplemented medium for Western blot analysis, or complete Williams’ medium for luciferase assay.

Plasmids and recombinant viruses

The HBSP expression vector (pHBSP) was generated from an isolated HBSP viral gene (333 bp), amplified, and cloned from a patient with chronic HBV (genotype A) infection and inserted into pcDNA3.1/myc-His (Invitrogen). Empty pcDNA3.1 was used as a control. pUS9-green fluorescent protein (GFP), expressing a membrane-bound green fluorescent protein (GFP), allowed the selection of transfected cells using fluorescence-activated cell sorting (FACS) analysis. pNF-κB-luc and pPAI-luc express the Firefly luciferase under the control of an IL-8 IκB-like site promoter and a minimal promoter of PAI gene, respectively. pGL3-luc expressing Renilla luciferase was used to normalize the luciferase assays. The lentivirus expressing HBSP with a Myc tag in its C-terminal extremity (lentivirus sense, Len.HBSPs) as well as the control lentivirus containing the HBSP gene inserted in reverse sens (lentivirus antisense, Len.HBSPas) were generated according to the procedure described previously (17). The adenovirus NF-κB-luc (Iowa gene transfer) was kindly provided by Dr. Le Goffic (Unité de Recherche UR892, Jouy-en-Josas, France).

Cell transduction and transfection

For transduction, a suspension of 1 × 106 HepG2 cells/ml was incubated for 2 hours with Len.HBSPs or the control Len.HBSPas (80 ng/ml equivalent p24 antigen) together with 6 µg/ml polybrene (Sigma). HeLa cells were transfected using TransIiT (Mirus Bio, Madison, WI, USA) according to the manufacturer’s recommendations. After 48 to 72 hours, transduction/transfection efficiency was evaluated by immunofluorescence using an anti-Myc antibody and an Alexa488-coupled secondary antibody. Only cell populations transduced/transfected at more than 50% were selected for subsequent experiments. Primary mouse hepatocytes (6 × 105 cells), isolated from HBSP Tg and control mice, were transduced by the adenovirus NF-κB-luc (5 × 107 plaque-forming unit) in 6-well-plates. Luciferase (Dual-Luciferase Reporter Assay, Promega, Madison, WI, USA), and FACS assays were performed 48 hours post-transfection/transduction.

Cellular gene expression analysis

Cellular RNAs from transduced HepG2 cells were extracted using the RNeasy Mini Kit (Qiagen, Germantown, MD, USA), according to the manufacturer's recommendations. Reverse transcription was performed using random hexamers and the real-time PCR using the ABI Prism 7700 (Applied Biosystems, Foster City, CA, USA). Primers and probes specific of the selected genes were previously described (18). Three separate RT-qPCR experiments were performed.

Cell proliferation and viability assays

The cell proliferation assay was performed using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] Cell Proliferation Kit (Roche, Basel, Switzerland). Apoptotic activity was evaluated by TUNEL assay using the In Situ Cell Death Detection Kit according to the manufacturer’s instructions (Roche) and Evans blue counterstaining. For each condition evaluated in triplicate, TUNEL-positive cells (dUTP fluorescein stained) were counted in at least 15 successive fields. For FACS analysis, cells were harvested 48 hours after cotransfection with pUS9-GFP and pHBSP or a control. After fixation in 70% ethanol (24 hours), the cells were treated with 0.2 mg/ml RNAseA (Sigma-Aldrich) and stained with propidium iodide (0.1 mg/ml; Sigma-Aldrich). Their DNA content was analyzed using a BD/LSR automat (Becton Dickinson, San Diego, CA, USA) and the CellQuest 5.2 software. All experiments were performed at least in triplicate.

Western blot analysis

Whole cells were lysed in 300 µl of RIPA buffer (0.1% sodium dodecyl sulfate, 1% sodium deoxycholate, 1% Triton × 100, 150 mM NaCl, 50 mM Tris, 1 mM DTT) containing protease/phosphatase inhibitors (Pierce, Rockford, IL, USA). After protein quantitation (Bradford assay), 30 µg of proteins were separated in SDS-PAGE and transferred to a nitrocellulose membrane (Hybond, Escondido, CA, USA). Membranes were probed using Myc, total IκB-α or IκB-β, total p65, total JNK, and β-Actin or GAPDH antibodies as loading controls (Santa Cruz Biotechnology, Santa Cruz, CA, USA), or phospho-IκB-α, phospho-p65, phospho-JNK1/2/3, procaspase 3, and cleaved caspase 3 antibodies (Cell Signaling Technology, Danvers, MA, USA). Immunoblots were developed using secondary antibodies coupled to horseradish peroxidase and the ECL+ chemiluminescence kit (GE Healthcare, Waukesha, WI, USA). For 2-dimensional protein gel analysis, an initial migration was performed on a pH3-10 strip (Bio-Rad, Hercules, CA, USA) at 4000V/h and then proteins were separated using 12% polyacrylamide gel electrophoresis, transferred to a nitrocellulose membrane, and probed with anti-TAK1 antibody (Cell Signaling).

Liver tissue staining

CD45 staining was performed in 10 µm frozen liver sections after methanol/ethanol (50/50) fixation. After blocking, rabbit anti-CD45 antibody (Abcam, Cambridge, MA, USA) was diluted in PBS + 0.1% Tween + 10% FCS for 1.5 hours and revealed with an Alexa488-antibody (Invitrogen). Liver cell nuclei were labeled using Hoechst (10 µg/ml, Sigma-Aldrich). The TUNEL assay was performed as previously described in mice treated with Jo2, TNF-α alone or plus D-Gal. Liver tissues were stained with hematoxylin and eosin.

Statistical analyses

Statistical analyses were performed using unpaired Student's or Mann and Whitney tests with GraphPad Prism 5 software (La Jolla, CA, USA).

RESULTS

HBSP modulates the gene expression profile of HepG2 hepatoma cells

The expression of 125 selected genes related to cell proliferation, fibrogenesis, apoptosis, viability, stress, and inflammation was quantified by RT-qPCR in HepG2 cells transduced with a recombinant lentivirus encoding for the HBSP or with a lentivirus control (Supplementary Table 1). As HBSP is a short-lived protein, transduction efficiency was evaluated in parallel on a subset of transduced cells treated with the proteasome inhibitor MG132 (Supplementary Fig. 1).

Among the 115 transcripts whose expression was detected (Ct < 35 PCR cycles), only 11 (9.6%) were differentially expressed in HBSP-transduced cells (Table 1). Six modulated genes were involved in cell proliferation, of which the genes encoding for PAK-1 and RhoA (constitutive of the Ras signaling pathway) were up-regulated, and the genes encoding for the growth factors amphiregulin (AREG), epiregulin (EREG), and the AP-1 factor (FOS, JUN) were down-regulated. In addition, 3 TNF-α-inducible genes, all known targets of the transcription factor NF-κB, appeared down-regulated: IER3 (both short IER3S and long IER3L mRNAs), MCL1L, and IL8. Finally, the TGF-β-inducible gene PAI1, encoding for the plasminogen activator inhibitor 1, and involved in liver fibrosis, was down-regulated.

Regarding the modulated expression of genes related to cell proliferation (FOS, JUN, AREG, EREG, RHOA, PAK1), we investigated the overall effect of HBSP on HepG2 cell growth. No significant effect was observed by MTT kinetic assay on cell proliferation and/or apoptosis in HBSP-transduced HepG2 cells (Fig. 1A, left panel). Additionally, DNA content analysis of HepG2 cells expressing HBSP did not show a modulation of the cell cycle and cell death activities (Fig. 1A, right panel).

We also investigated the influence of HBSP on the activation of the TGF-β and TNF-α signaling pathways by luciferase assays (Fig. 1B). In cells treated or not with TGF-β, the PAI-luc reporter did not show any transcriptional modulation related to HBSP expression (Fig. 1B, left panel). By contrast, in cells treated with TNF-α or IL-1β, stimulation of the NF-κB-luc reporter was significantly impaired in the presence of HBSP (Fig. 1B, right panel). Our results therefore suggested that HBSP down-regulates the activation of the NF-κB pathway downstream of the TNF-α and IL-1β receptors. Of note, it has been shown that the HBV polymerase can also suppress the NF-κB pathway through interaction with Hsp90 (19). However, this interaction does not involve its 47 N-terminal amino-acids shared with HBSP (20). This suggests that HBSP down-regulates NF-κB by an alternative mechanism.

HBSP down-regulates NF-κB activation while up-regulating JNK activation upon TNF-α stimulation

TNF-α is a prototypical proinflammatory cytokine that induces a variety of biological responses ranging from cell proliferation to cell death. To further characterize the impact of HBSP on the canonical NF-κB signaling cascade, we stimulated HBSP-expressing cells with TNF-α and investigated the kinetics of phosphorylation of IκB-α, the inhibitor of NF-κB, and of p65/RelA, constitutive of NF-κB (Fig. 2A). Five minutes after TNF-α induction, the level of phosphorylated IκB-α (Ser32/Ser36) was decreased in HBSP-expressing cells. Additionally, a prolonged inhibition of IκB-α phosphorylation was observed at 1, 4, and 7 hours post-treatment. Furthermore, the neosynthesis of IκB-α was altered 7 hours post-treatment. Meanwhile, p65 phosphorylation at Ser536 was also impaired at 5 minutes and then at 4 and 7 hours after TNF-α treatment in HBSP-expressing cells. These data showed that HBSP modulates both IκB-α and p65 phosphorylation status in hepatoma cells upon TNF-α treatment.

Considering the important role of TNF-α in the balance between cell death and viability, and the previous findings reporting the proapoptotic effect of HBSP in some cell lines, we therefore investigated whether the impairment of NF-κB activation in HBSP-expressing cells could sensitize HepG2 cells to programmed cell death. In this purpose, we performed a TUNEL assay and compared the sensitivity of HBSP-expressing cells to apoptosis upon TNF-α alone or TNF-α plus ActD, an inhibitor of the transcription machinery. In untreated cells (Fig. 2B) or treated with TNF-α alone (data not shown), HBSP expression did not affect cell viability: <1% dead cells in both control and HBSP-expressing cells. In contrast, cell death was significantly increased following the 6-hour treatment with TNF-α + ActD in the HBSP-transduced cells (10.3 ± 0.7%) compared with controls (4.6 ± 0.4%, P < 0.0001, Fig. 2B).

It is well documented that the balance between NF-κB and JNK activities, respectively prosurvival and pro-cell death factors, determines the outcome of TNF-α stimulation on cell viability (21, 22). Surprisingly, a prolonged overactivation of JNK was observed in HBSP-transduced HepG2 cells following TNF-α treatment (Fig. 2C). This result was unexpected considering that NF-κB activation was down-regulated in HBSP-expressing cells and that NF-κB and JNK are both coactivated following TNF-α stimulation. As lentiviral transduction might induce an interferon-α adverse effect on NF-κB activation, the 2 TNF-α-induced signaling pathways were also investigated in transfected cells. Thus, HeLa cells were transfected with an HBSP-expressing plasmid or the corresponding empty vector as control. Results confirmed the differential modulation of phosphorylation of IκB-α and JNK in the HBSP transfected cell population, 5 and 15 minutes after TNF-α treatment (Fig. 2D). Furthermore, down-regulation of the prototypical NF-κB-inducible gene, IκB-α, was confirmed by RT-qPCR 30 minutes after TNF-α treatment (Supplementary Fig. 2).

TAK1 activation upon TNF-α stimulation is modulated in the presence of HBSP

TNF-α treatment induced opposite effects on the phosphorylation of NF-κB and JNK in HBSP cells. This suggested a modulation of activation of the TAK1, belonging to the MAP3K family and located at the junction of both NF-κB and JNK signaling cascades. Typically, following TNF-α stimulation, TAK1 and its adaptors [TAK1-binding proteins (TABs) and IKK-γ] are recruited to polyubiquinated RIP. This step enables TAK1 auto- and transphosphorylation as well as ubiquitination leading to the activation of both JNK and NF-κB pathways (23, 24). We characterized the activation of TAK1 by 2-dimensional Western blot analysis using TAK1 antibody, before and 15 minutes after TNF-α treatment (Fig. 2E). In the absence of TNF-α, one spot corresponding to the inactivated TAK1 protein was detected in both HBSP-expressing and control cells (Fig. 2E, spot 0). Following TNF-α stimulation, TAK1 is activated through phosphorylation and ubiquitination events (25). By 2-dimensional Western blot, we identified 5 additional spots of TAK1 in TNF-α-treated controls (spots labeled 1–5) and a smear of ubiquitinated forms at higher molecular weight (Ub-TAK1), resulting from an accumulation of phosphorylated serine/threonine residues on TAK1. By contrast, HBSP-expressing cells treated with TNF-α still displayed unphosphorylated TAK1 (spot 0) and less hyperphosphorylated forms (absence of the spot 5, bottom right panel). This shift of phosphorylation was also observed for the ubiquitinated forms of TAK1 in HBSP-expressing cells. Taken together, our data suggested that the differential IκB-α and JNK activations with HBSP resulted from a lower phosphorylation of TAK1 following TNF-α stimulation.

Modulation of the TNF-α-stimulated signaling pathways in hepatocytes of HBSP Tg mice

To investigate the impact of HBSP expression in liver tissue, 2 HBSP Tg mouse strains (TgHBSP1 and 2) were designed and characterized (Supplementary Fig. 3). Monitoring of the TgHBSP mice for up to 12 months did not reveal any adverse phenotype. However, we confirmed a weaker activation of the NF-κB pathway after a 6-hour treatment with TNF-α in primary mouse hepatocytes isolated from both TgHBSP strains and transduced with an adenoviral vector containing a NF-κB luciferase reporter (Fig. 3A). This result was reinforced by Western blot analysis. Indeed, we observed a differential phosphorylation of IκB-α at 5 minutes and JNK at later time points (from 1 to 10 hours) in primary hepatocytes isolated from TgHBSP mice when compared with controls after stimulation with murine TNF-α (Fig. 3B). In vivo, 10 hours after intravenous delivery of TNF-α alone, no hepatocyte apoptosis was detected. However, concomitant administration of TNF-α + D-Gal, an inhibitor of the transcription machinery, led to a significant increase of liver apoptosis in HBSP Tg mice, as illustrated by TUNEL analysis (Fig. 3C) and detection of cleaved caspase-3 (Fig. 3D). As a positive control, some mice received Jo2 antibodies mimicking Fas ligand binding and inducing apoptosis by direct activation of the caspase-8, independently of the NF-κB and JNK pathways (Fig. 3C and 3D). Unsurprisingly, both TUNEL assay (Fig. 3C) and caspase-3 Western blots (Fig. 3D) revealed massive liver apoptosis 4 hours after Jo2 administration. Interestingly, the level of hepatic apoptosis was similar in both control and HBSP Tg mice illustrating that HBSP does not sensitize hepatocyte cell death by interfering with caspase-8 activation. Taken together, these results confirmed in vivo the HBSP-mediated sensitization to apoptosis of hepatocytes by interfering with the TNF-α-stimulated signaling pathways.

HBSP expression in hepatocytes prevents extensive immune infiltration into the liver during chronic inflammation

As a next step, we investigated whether the interference of the HBSP with the TNF-α-stimulated signaling pathways affects liver disease progression. In this purpose, we induced chronic liver inflammation by repeated injections of low-dose LPS (2.5 mg/kg, 3 times a week) for 9 months and compared liver and spleen disorders in HBSP Tg and control mice. LPS is recognized by TLR4, which directly activates TAK1 in a MyD88-dependent manner. TAK1 activation then induces both NF-κB and JNK signaling cascades leading to the secretion of various chemokines (including TNF-α) by hepatocytes, Kupffer cells, and stellate cells. In return, TNF-α promotes hepatic infiltration of leukocytes and inflammation-related damages (26). First of all, we observed hepato- and splenomegalies in mice that received LPS (both control and TgHBSP) confirming the establishment of chronic inflammation (Fig. 4A). To note, despite repeated injections of LPS for a long period of time, no desensitization to LPS exposure was observed in contrast to previously reported data (27). Length and weight of both liver and spleen did not differ significantly between HBSP Tg and control mice treated with LPS (Fig. 4B). In contrast, immune infiltrates visualized on hematoxylin and eosin-stained sections of liver tissue (Fig. 4C, upper panels), confirmed with fluorescent CD45 labeling (Fig. 4C, middle panels), revealed a significantly less aggressive inflammation in HBSP Tg mice compared with control mice. The total inflammatory area per liver section was significantly lower in HBSP Tg mice (0.06 ± 0.01 mm2) compared with control mice (0.16 ± 0.02 mm2, P = 0.01, Fig. 4D, upper panel). Quantitation of the mean surface of inflammatory infiltrates per liver section confirmed the attenuated extent of the LPS-induced inflammation in HBSP Tg mice (2745 ± 339 μm2) compared to control mice (5497 ± 662 μm2, P = 0.004, Fig. 4D, middle panel). This difference was not related to the level of apoptosis in so far as a TUNEL assay showed a low and similar degree of cell death in both control and HBSP Tg (Fig. 4C, D, lower panels). Taken together, our data showed that HBSP expression in hepatocytes reduced the extent of LPS-induced chronic liver inflammation through a modulation of activation of the TNF-α-stimulated signaling pathways.