Penicillinase-resistant antibiotics induce non-immune-mediated cholestasis through HSP27 activation associated with PKC/P38 and PI3K/AKT signaling pathways

Abstract

The penicillinase-resistant antibiotics (PRAs), especially the highly prescribed flucloxacillin, caused frequent liver injury via mechanisms that remain largely non-elucidated. We first showed that flucloxacillin, independently of cytotoxicity, could exhibit cholestatic effects in human hepatocytes in the absence of an immune reaction, that were typified by dilatation of bile canaliculi associated with impairment of the Rho-kinase signaling pathway and reduced bile acid efflux. Then, we analyzed the sequential molecular events involved in flucloxacillin-induced cholestasis. A crucial role of HSP27 by inhibiting Rho-kinase activity was demonstrated using siRNA and the specific inhibitor KRIBB3. HSP27 activation was dependent on the PKC/P38 pathway, and led downstream to activation of the PI3K/AKT pathway. Other PRAs induced similar cholestatic effects while non PRAs were ineffective. Our results demonstrate that PRAs can induce cholestatic features in human hepatocytes through HSP27 activation associated with PKC/P38 and PI3K/AKT signaling pathways and consequently support the conclusion that in clinic they can cause a non-immune-mediated cholestasis that is not restricted to patients possessing certain genetic determinants.

Figure 1

Cytotoxicity and alteration of BC structures by FLX in human HepaRG cells and primary hepatocytes. (A) HepaRG cells were incubated with different concentrations of FLX (0–24 mM) for 24 h. Cytotoxicity was measured using the MTT colorimetric and caspase-3 activity assays. (B) Representative phase-contrast images of 16 mM FLX-treated HepaRG cells showing higher sensitivity of primitive biliary-like cells (white arrows) than hepatocytes. (C) HepaRG cells were treated with different concentrations of FLX for 6 or 24 h. ROS generation was detected by the DCFDA specific substrate. (D) HepaRG cells were incubated with different non cytotoxic concentrations of FLX (0–6 mM) at different time points. Immunolabelling of the junctional ZO-1 protein (green) in HepaRG cells treated for 4 h with FLX compared with control cells. F-actin was localized using rhodamine-phalloidin fluoroprobe (red). (E) Immunolabelling of the junctional protein occludin (green) in HepaRG cells treated with 2 mM FLX compared with control cells. (F) Quantification of BC area after 2, 8 and 24 h using ImageJ 1.48 software as described in the Materials and Methods. Data were expressed as the fold change in the BC mean area relative to the mean area of untreated cells arbitrarily set at a value of 1. They represented the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells. (G) Phase-contrast images and F-actin localization (red) in PHH treated with 6 mM FLX for 2 h. Nuclei stained in blue (Hoechst dye). Phase-contrast images were captured using time lapse microscopy. Orange arrows indicate BC. The fluorescent images were obtained with a Cellomics ArrayScan VTI HCS Reader (bar = 50 µm).

Figure 2

Effects of FLX on labelled bile acids and CDF clearance in HepaRG cells and PHH. (A) NBD-UDCA and CDF efflux in HepaRG hepatocytes and PHH treated for 2 h with different concentrations of FLX (0–6 mM). Orange arrows indicate fluorescence in BC (bar = 50 μm). (B) Quantification of CDF accumulation in BC of HepaRG hepatocytes after FLX treatment using ImageJ 1.48 software as described in the Materials and Methods. (C) [³H]-TA clearance in HepaRG cells treated with different concentrations of FLX (0–8 mM) for 2 h. Data were expressed relative to those of untreated cells arbitrarily set at 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells.

Figure 3

Alteration of ROCK activity and MYPT1 phosphorylation by FLX. (A) HepaRG cells were treated with FLX (0–6 mM) for 4 h, then ROCK activity was assessed using a ROCK activity assay Kit (Millipore, catalogue CSA001). (B) Representative western blots of p-MYPT1/total MYPT1 after 4 h FLX (0–6 mM) treatment. The displayed blots were cropped and the original full-length gels are included in the supplementary information. (C) Quantification of MYPT1 phosphorylation using ImageJ 1.48 software. (D) Representative phase-contrast images of HepaRG cells treated with 6 mM FLX alone or combined with 5 µM CaM, a MLCK activator. Orange arrows indicate BC (bar = 50 µm). (E) Quantification of BC area after 4 h using ImageJ 1.48 software. Data were expressed relative to those of the untreated cells arbitrary set at a value of 1. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of controls.

Figure 4

Involvement of HSP27 in FLX-induced effects. (A) Representative western blots of the p-HSP27/total HSP27 forms after 2h-treatment with FLX (0–6 mM) in HepaRG cells and PHH. Quantification of HSP27 phosphorylation in HepaRG cells using ImageJ 1.48 software. The displayed blots were cropped and the original full-length gels are included in the supplementary information. (B) Representative phase-contrast images of HepaRG cells treated with 2 mM FLX alone or combined with 0.5 µM KRIBB3 after 2 h. Quantification of BC area using ImageJ 1.48 software. Orange arrows indicate BC (bar = 50 μm). (C) CDF efflux in HepaRG hepatocytes and PHH treated with 2 mM FLX alone or combined with 0.5 µM KRIBB3 compared to untreated cells. Quantification of CDF accumulation in BC of HepaRG hepatocytes and PHH after 2 h treatment with 2 mM FLX ± 0.5 µM KRIBB3 using ImageJ 1.48 software. (D) [³H]-TA clearance in HepaRG cells treated with 4 or 6 mM FLX alone or co-treated with 0.5 µM KRIBB3 for 2 h. (E) Representative western blots of p-HSP27/total HSP27 forms after 2 h-treatment with 6 mM FLX ± 0.5 µM KRIBB3 (KR). Data were expressed relative to those of untreated cells arbitrarily set at 1 or 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells, ^#p < 0.05 compared with that of cultures treated with FLX alone.

Figure 5

FLX effects in siHSP27-transfected cells. (A) Representative western blots of total HSP27 and HSC70 after 72 h in HepaRG cells: wild type (WT), transfected with scramble siRNA (siC) and HSP27 siRNA (siHSP27). Quantification of total HSP27 using ImageJ 1.48 software. (B) Representative western blots of the p-HSP27/total HSP27 forms and HSC70 after 2h-treatment with 2 mM FLX in WT and siHSP27 transfected HepaRG cells. The displayed blots were cropped and the original full-length gels are included in the supplementary information. (C) Representative phase-contrast images and CDF efflux in wild type, siC and siHSP27 transfected HepaRG cells treated with 2 mM FLX (bar = 50 μm). (D) Quantification of the BC area and (E) canalicular CDF accumulation in HepaRG cells after 2 h of FLX treatment using ImageJ 1.48 software. Data were expressed relative to those of the untreated cells arbitrarily set at a value of 1 or 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells.

Figure 6

Involvement of PKC/P38 pathway in FLX-induced effects. (A) Representative western blots of the p-P38/total P38 forms after 2h-treatment with FLX (0–6 mM) alone or combined with 20 µM PKC inhibitor (Gö6976; Gö) or 10 µM P38 inhibitor (SB203580; SB) in HepaRG cells and PHH. Quantification of p-P38 in HepaRG cells using ImageJ 1.48 software. The displayed blots were cropped and the original full-length gels are included in the supplementary information. (B) Representative phase-contrast images of HepaRG cells treated with 2 mM FLX alone or combined with 20 µM Gö6976 or 10 µM SB203580. Quantification of BC area using ImageJ 1.48 software. Orange arrows indicating BC deformation (bar = 50 μm). (C) CDF efflux in HepaRG hepatocytes and PHH treated 2 h with 2 mM FLX alone or combined with 20 µM Gö6976 or 10 µM SB203580. Quantification of CDF accumulation in BC of HepaRG hepatocytes and PHH, using ImageJ 1.48 software. (D) [³H]-TA clearance in HepaRG cells treated with 4 or 6 mM FLX alone or co-treated with 20 µM Gö6976 or 10 µM SB203580 for 2 h. (E) Representative western blots of p-HSP27/total HSP27 forms after 2h-treatment with 6 mM FLX alone or combined with 10 µM P38 inhibitor (SB203580; SB) or 20 µM PKC inhibitor (Gö6976; Gö). Data were expressed relative to those of untreated cells arbitrarily set at 1 or 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells, ^#p < 0.05 compared with that of cultures treated with FLX alone.

Figure 7

Involvement of PI3K/AKT pathway in FLX-induced effects. (A) Representative western blots of p-AKT/total AKT forms after 2h-treatment with FLX (0–6 mM) alone or combined with the PI3K inhibitors LY294002 (10 µM) or WM (0.25 µM) in HepaRG cells and PHH. Quantification of p-AKT in HepaRG cells using ImageJ 1.48 software. (B) Representative phase-contrast images of HepaRG cells treated for 2 h with 2 mM FLX alone or combined with 10 µM LY294002 or 0.25 µM WM. Quantification of BC area using ImageJ 1.48 software. Orange arrows indicating BC deformation (bar = 50 μm). (C) CDF efflux in HepaRG hepatocytes and PHH treated for 2 h with 2 mM FLX alone or combined with 10 µM LY294002 or 0.25 µM WM. Quantification of CDF accumulation in BC of HepaRG hepatocytes and PHH, using ImageJ 1.48 software. (D) [³H]-TA clearance in HepaRG cells treated with 4 or 6 mM FLX alone or co-treated with 10 µM Y294002 or 0.25 µM WM for 2 h. (E) Representative western blots of p-AKT/total AKT forms after 2 h treatment with 6 mM FLX alone or combined with 0.5 µM HSP27 inhibitor (KRIBB3; KR), 10 µM P38 inhibitor (SB203580; SB), and 20 µM PKC inhibitor (Gö6976; Gö). Representative western blots of p-P38/total P38 and p-HSP27/total HSP27 after 2 h treatment with 6 mM FLX alone or combined with the PI3K inhibitors 10 µM LY294002 (LY) or 0.25 µM WM. (F) Representative western blots of p-MYPT1/total MYPT1 after 4 h treatment with 6 mM FLX alone or combined with KR, LY, WM, SB or Gö. The displayed blots were cropped and the original full-length gels are included in the supplementary information. Data were expressed relative to those of untreated cells arbitrarily set at 1 or 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells, ^#p < 0.05 compared with that of cultures treated with FLX alone.

Figure 8

HSP27-dependent cholestatic effects of penicillinase-resistant antibiotics in human HepaRG cells. (A) Phase-contrast images, NBD-UDCA and CDF efflux in HepaRG cells either untreated or treated with 6 mM cloxacillin (CLX), nafcillin (NAF), ampicillin (AMPI) and amoxicillin (AMOX) for 2 h. Phase-contrast images were captured using time-lapse microscopy. Orange arrows indicate BC deformations. Fluorescent images were obtained with a Cellomics ArrayScan VTI HCS Reader (bar = 50 µm). (B) Representative western blots of the p-HSP27/total HSP27 forms after 2h-treatment of HepaRG cells with 6 mM CLX, NAF, AMPI and AMOX. Quantification of p-HSP27 using ImageJ 1.48 software. The displayed blots were cropped and the original full-length gels are included in the supplementary information. (C) [³H]-TA clearance in HepaRG cells treated with 6 mM CLX or NAF alone or co-treated with 0.5 µM KRIBB3 for 2 h. Data were expressed relative to those of untreated cells arbitrarily set at a value of 1 or 100%. They represent the means ± SEM of 3 independent experiments. *p < 0.05 compared with that of untreated cells.

Figure 9

Schematic representation of sequential molecular mechanisms involved in FLX-induced cholestasis. FLX induces phosphorylation of HSP27 through activation of PKC/P38 (1). Activated p-HSP27 plays a central role in cholestasis; it inhibits ROCK and consequently MYPT1 phosphorylation leading to BC dilatation and cholestasis (2) and activates PI3K/AKT leading to BC dilation and cholestasis through MYPT1 dephosphorylation (3). In parallel to cholestatic insult, activated p-HSP27 could lead to cell resistance and survival through PI3K/AKT pathway activation preventing caspases induction and ROS generation (4). ROCK, Rho-kinase; MLC2, myosin light chain 2; MYPT1, myosin phosphatase target subunit 1; BC, bile canaliculi; ZO-1, zona occludens-1 protein; PKC, protein kinase C; P38, p38 mitogen-activated protein kinase; PI3K, phosphoinositide 3-kinase; AKT, protein kinase B; HSP27, heat shock protein 27; PKD, protein kinase D; ROS, reactive oxygen species; MAPKAPK, mitogen-activated protein kinase-activated protein kinase.