1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Drug-induced liver injury (DILI) is an important clinical problem. It involves crosstalk between drug toxicity and the immune system, but the exact mechanism at the cellular hepatocyte level is not well understood. Here we studied the mechanism of crosstalk in hepatocyte apoptosis caused by diclofenac and the proinflammatory cytokine tumor necrosis factor α (TNF-α). HepG2 cells were treated with diclofenac followed by TNF-α challenge and subsequent evaluation of necrosis and apoptosis. Diclofenac caused a mild apoptosis of HepG2 cells, which was strongly potentiated by TNF-α. A focused apoptosis machinery short interference RNA (siRNA) library screen identified that this TNF-α-mediated enhancement involved activation of caspase-3 through a caspase-8/Bid/APAF1 pathway. Diclofenac itself induced sustained activation of c-Jun N-terminal kinase (JNK) and inhibition of JNK decreased both diclofenac and diclofenac/TNF-α-induced apoptosis. Live cell imaging of GFPp65/RelA showed that diclofenac dampened the TNF-α-mediated nuclear factor kappaB (NF-κB) translocation oscillation in association with reduced NF-κB transcriptional activity. This was associated with inhibition by diclofenac of the TNF-α-induced phosphorylation of the inhibitor of NF-κB alpha (IκBα). Finally, inhibition of IκB kinase β (IKKβ) with BMS-345541 as well as stable lentiviral short hairpin RNA (shRNA)-based knockdown of p65/RelA sensitized hepatocytes towards diclofenac/TNF-α-induced cytotoxicity. Conclusion: Together, our data suggest a model whereby diclofenac-mediated stress signaling suppresses TNF-α-induced survival signaling routes and sensitizes cells to apoptosis. (HEPATOLOGY 2011;)

Adverse drug reactions are an important cause of morbidity and mortality in humans and drug-induced liver injuries (DILIs) are the leading cause of acute liver failure.1 In addition, DILI accounts for most of the drug attritions2 and more than 10% of the occurring liver failures happen due to idiosyncratic DILIs.1 We propose that the crosstalk between drug reactive metabolite-mediated stress responses and cytokine-mediated pro- and antiapoptotic signaling is an important component in the pathophysiology of DILI.

Diclofenac is one of the most commonly used drugs causing idiosyncratic DILI.3 Diclofenac is a nonsteroidal antiinflammatory drug (NSAID) widely prescribed to treat, for example, pain and rheumatoid arthritis. Although the most frequently occurring adverse drug reaction associated with the use of diclofenac is gastrointestinal ulceration,4 severe idiosyncratic liver injuries are reported and, due to the drug's frequent clinical use, the total number of affected patients is significant.3

The underlying cellular mechanisms that determine the susceptibility to developing DILI are incompletely understood. Due to their relatively rare occurrence, it is expected that multiple factors are involved. Increasing evidence points toward a role for the formation of reactive metabolites and the (innate) immune system.2, 5, 6 In the liver, diclofenac is metabolized into three main metabolites, 4′-OH-diclofenac, 5-OH-diclofenac, and diclofenac acylglucuronide, which are reactive towards protein thiol-groups, associated with formation of reactive oxygen species (ROS), and causally related to DILI.7 Covalent protein modifications and ROS cause cellular injury and activation of different stress signaling pathways, including c-Jun N-terminal kinase (JNK).8

The largest pool of stationary immune cells in the liver is the liver-specific macrophages, the Kupffer cells. Through intercellular communication between hepatocytes and Kupffer cells or due to direct endotoxin exposure from the intestine, Kupffer cells secrete proinflammatory cytokines, of which tumor necrosis factor-α (TNF-α) is the major component.9 TNF-α severely enhances liver damage caused by different xenobiotics.10-12 Although an involvement of the immune system in diclofenac hepatotoxicity is clear from an in vivo rat model,13 the exact role of cytokine signaling and the molecular mechanism of such an interaction are poorly defined.

TNF-α induces both pro- and antiapoptotic signaling. By formation of the complex I signalosome after TNF-α binding to its receptor (TNFR-1), the transcription factor nuclear factor kappa-B (NF-κB) is activated.14 Nuclear translocation of NF-κB occurs in an oscillatory manner following degradation of the inhibitor of NF-κB, IκBα, to induce transcription of its target genes which primarily encode survival proteins, e.g., cellular FLICE-like inhibitory protein (c-FLIP) and inhibitor of apoptosis proteins (IAPs), and proteins that negatively regulate the activation of complex I, e.g., A20 and IκBα.15 Depending on the cellular signaling context, from complex I an apoptosis activating complex II can be formed, which results in activation of caspase-8 and induction of the apoptotic pathway.15, 16 In addition to activation of transcription factor NF-κB, signaling from complex I can lead to activation of mitogen activated protein kinases (MAPKs). Activation of JNK can lead to either survival or apoptosis, depending on whether the activation is transient or prolonged.17 Activation of caspases and prolonged JNK signaling are, under normal conditions, antagonized by different NF-κB target genes.16, 18

Here we used a human HepG2 cell-based model to study the diclofenac/cytokine interaction. We show that the concentration-dependent toxicity of diclofenac is enhanced in the presence of the cytokine TNF-α, which is dependent on the activation of JNK1. Consistent with signaling from the TNFR-I downstream proapoptotic response, using an RNA interference approach targeting all apoptotic machinery components, we identified a key role for the capase-8/Bid/APAF1 route in the diclofenac/TNF-α-induced apoptosis. TNF-α-induced IκBα phosphorylation was inhibited by diclofenac in association with attenuation of the nuclear translocation of NF-κB. Inhibition of IKKβ or RNA interference-based silencing of the NF-κB subunit p65/RelA further sensitized cells to diclofenac/TNF-α-induced apoptosis. Our findings support a model whereby diclofenac perturbs prosurvival NF-κB responses during periods of inflammation, which favors proapoptotic signaling via caspase-8 and JNK upon cytokine exposure, ultimately increasing the likelihood of liver cell death.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Reagents and Antibodies.

Diclofenac sodium, naproxen sodium, and the selective IKK2-inhibitor BMS-345541 were obtained from Sigma (Zwijndrecht, Netherlands). Human and mouse recombinant TNF-α were acquired from R&D Systems (Abingdon, UK). The selective JNK-inhibitor SP600125 was from Enzo Life Sciences (Zandhoven, Belgium). The irreversible pan-caspase inhibitor z-VAD-fmk was from Bachem (Weil am Rhein, Germany). The irreversible inhibitors of caspases z-DEVD-fmk (caspase-3), z-IETD-fmk (caspase-8), and z-LEHD-fmk (caspase-9) were from Calbiochem (Merck, Darmstadt, Germany). AnnexinV-Alexa633 was made as described.19 The antibodies against active caspase-3 and phospho-specific JNK antibody were from New England Biolabs (Leusden, the Netherlands). The antibodies against caspase-8, caspase-9, cleaved poly (ADP-ribose) polymerase (PARP), JNK1/2, IκBα, and the phospho-specific c-Jun and IκBα antibodies were from Cell Signaling (Bioké, Leiden, Netherlands). The antibody against tubulin was from Sigma and the antibody against NF-κB (p65) was from Santa Cruz (Tebu-Bio, Heerhugowaard, Netherlands).

Cell Lines.

Human hepatoma HepG2 cells and mouse hepatoma Hepa1c1c7 cells were obtained from American Type Culture Collection (ATCC, Wesel, Germany), cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS), 25 U/mL penicillin, and 25 μg/mL streptomycin and used for experiments between passages 5 and 20.

RNA Interference.

Stable HepG2 cell lines with p65/RelA knocked down were produced using lentiviral shRNA vectors (Sigma-Aldrich, in collaboration with Dr. Hoeben, Leiden University Medical Centre, Netherlands) and selection with puromycin (2.5 μg/mL). The sequence for the nontargeting control shRNA was CCGGTCCGCAGGTATGCACGCGTG AATTC and the hRelA sequence was CCGGCACCAT CAACTATGATGAGTTCTCGAGAACTCATCATAG TTGATGGTGTTTTT. Transient knockdowns of individual target genes were achieved using siGENOME SMARTpool siRNA reagents in the primary screen or single siRNA sequences in the secondary deconvolution screen (50 nM; Dharmacon Thermo Fisher Scientific, Landsmeer, Netherlands). The negative control was siGENOME nontargeting pool #1. HepG2 cells were transfected using INTERFERin siRNA transfection reagent according to the manufacturer's procedures (Polyplus transfection, Leusden, Netherlands) and left for 72 hours to achieve maximal knockdown before treatment.

Cell Death Analysis Assays.

Apoptosis was determined by cell cycle analysis using 4′,6-diamidino-2-phenylindole (DAPI) and flow cytometry (FACSCanto II; Becton Dickinson, Erembodegem, Belgium). The amount of cells in sub-G0/G1 was calculated using the BD FACSDiva software (Becton Dickinson).

Overall cell death (loss of membrane integrity) was determined by lactate dehydrogenase (LDH) release in the medium in essentially the same manner as described.20

Induction of apoptosis in real time was quantified using a live cell apoptosis assay previously described.19 Briefly, binding of annexin V-Alexa633 conjugate to phosphatidyl serine on the membranes of apoptotic cells was followed in time by imaging every 30 minutes after drug exposure with a BD Pathway 855 imager (Becton Dickinson). The total fluorescent intensity per image or the relative fluorescence intensity per cell area was quantified using Image Pro (Media Cybernetics, Bethesda, MD). Caspase-3 activity was determined as described.21

Western Blot Analysis and Immunofluorescence.

Western blot analysis and immunofluorescent staining were essentially performed as described.20 For immunofluorescence, cells were stained for NF-κB p65 followed by goat antimouse Alexa488- (Molecular Probes, Breda, Netherlands) or Cy3-labeled (Jackson, Amsterdam, Netherlands) secondary antibodies. Hoechst 33258 (2 μg/mL) was used to visualize the nuclei. Cells were imaged using a BD Pathway 855 imager (Becton Dickinson) and the NF-κB translocation was quantified as an intensity ratio of NF-κB (nucleus): NF-κB (cytoplasm) using the AttoVision software (Becton Dickinson). Images were processed in Adobe Photoshop CS2 (Amsterdam, Netherlands).

Live Cell Imaging of GFPp65 in HepG2 Cells.

HepG2 cells stably expressing GFPp65 (NF-κB subunit) were created by 400 μg/mL G418 selection upon pEGFP-C1-p65 transfection using Lipofectamine 2000 (Invitrogen, Breda, Netherlands). Prior to imaging, nuclei were stained with 100 ng/mL Hoechst 33342 in complete DMEM. The GFPp65 nuclear translocation response upon 10 ng/mL human TNF-α challenge was followed for a period of 6 hours by automated confocal imaging (Nikon TiE2000, Nikon, Amstelveen, Netherlands). Quantification of the nuclear/cytoplasmic ratio of GFPp65 intensity in individual cells was performed using an algorithm for ImageJ (Z. Di, B. Herpers, B. van de Water, J.H.N. Meerman, and F. Verbeek, in prep.).

Luciferase Reporter Assay.

To determine the effect of diclofenac exposure and RelA inhibition/knockdown on TNF-α-induced NF-κB transcriptional activity, HepG2 cells were transiently transfected with an NF-κB promoter-luciferase reporter plasmid (Clontech, Saint-Germain-en-Laye, France) using Lipofectamine 2000 reagent according to the manufacturer's procedures (Invitrogen) and incubated for 16-18 hours. The Dual-Luciferase luciferase assay kit (Promega, Leiden, Netherlands) and a microplate luminometer (Centro XS3 LB960, Berthold Technologies) were used to monitor luciferase activity.

Statistical Analysis.

All numerical results are expressed as the mean ± standard error of the mean (SEM) and represent data from three independent experiments unless otherwise stated. Calculations were made using GraphPad Prism 4.00 (La Jolla, CA). Significance levels were calculated using unpaired Student's t test or 2-way analysis of variance (ANOVA) in the case of multiple comparisons, *P < 0.05, **P < 0.01, ***P < 0.0001.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

TNF-α Enhances Diclofenac-Induced Cell Death in Hepatocytes.

To investigate the role of TNF-α signaling in enhancement of diclofenac-induced hepatotoxicity, we first pretreated human hepatoma HepG2 cells with increasing concentrations of diclofenac for 8 hours to allow formation of reactive metabolites (Supporting Data S1), followed by treatment with TNF-α (10 ng/mL). Importantly, during this 8-hour period diclofenac was metabolized into both acylglucuronide and hydroxymetabolites (Supporting Data S1) which further accumulated over time. After 24 hours, cells were collected to determine cell death. Diclofenac alone induced a mild concentration- dependent increase in cell death and although no apoptosis was observed by TNF-α alone, in combination with diclofenac, TNF-α addition resulted in a 2-fold increase of apoptosis (Fig. 1A). This enhancement effect could be abrogated by cotreatment with the pan-caspase inhibitor z-VAD-fmk (50 μM), indicating that the TNF-α-enhanced diclofenac-induced DNA fragmentation is a caspase-executed apoptotic process (Fig. 1A). A structurally different nonsteroid antiinflammatory compound, naproxen, did not lead to concentration-dependent induction of apoptosis after 24 hours of exposure, even in the presence of TNF-α (Fig. 1B). This indicates that the diclofenac/TNF-α-induced apoptosis is independent from cyclooxygenase inhibition.

thumbnail image

Figure 1. TNF-α enhances diclofenac-induced apoptosis in hepatocytes. Treatment with diclofenac (A), but not the structurally different NSAID naproxen (B), induces apoptosis in HepG2 cells after 24 hours, which is significantly enhanced by addition of TNF-α (10 ng/mL) after 8 hours of drug exposure. The apoptosis was determined by cell cycle analysis (A,B) or followed in time using AnxV-Alexa633 staining and automated imaging (500 μM diclofenac; C,D). LDH activity in the medium after 24 hours of diclofenac exposure ± TNF-α was used as a measurement of overall cytotoxicity in human (E) and mouse hepatocytes (F; 1 ng/mL TNF-α added after 8 hours). The data are presented as means of at least three independent experiments ± SEM. The percentage of total LDH activity under control conditions was <5% and z-VAD-fmk (50 μM) was included where indicated. ***P < 0.001, **P < 0.01, *P < 0.05, #P < 0.05 compared to vehicle-treated cells.

Download figure to PowerPoint

To gain insight in the temporal onset of the TNF-α-enhanced diclofenac-induced apoptosis, we applied a live apoptosis microscopy assay, based on Alexa633-labeled annexin V (AnxV) binding to cells that present phosphatidyl-serine in the outer layer of their plasma membrane.19 The diclofenac/TNF-α-induced apoptotic response was initiated 4 hours after addition of TNF-α (Fig. 1D; Supporting Movies M1-4). This enhanced accumulation of AnxV-positive cells correlated with enhanced release of LDH in the medium, which was prevented by z-VAD-fmk treatment (Fig. 1E), supporting a model whereby TNF-α promotes diclofenac-exposed HepG2 cell killing primarily by apoptosis, followed later by secondary necrosis. Importantly, the apoptosis-inducing synergism between diclofenac and TNF-α was not restricted to HepG2 cells, because the mouse hepatoma cell line Hepa1c1c7 was also susceptible to the combined treatment (Fig. 1F).

Induced Apoptosis by TNF-α in Diclofenac-Treated HepG2 Cells Is Dependent on the Extrinsic Apoptotic Pathway.

Diclofenac alone can induce apoptosis in hepatocytes via the intrinsic, mitochondrial, apoptotic pathway involving caspase-9.22 However, TNF-α is well known to induce the apoptotic cascade via the extrinsic, death receptor-mediated pathway involving caspase-8.16 To investigate which initial apoptotic process dominates in diclofenac/TNF-α coexposed HepG2 cells, we assessed the activation of initiator caspases-8, -9, and effector caspase-3. Diclofenac-induced apoptosis involved activation of caspase-8, which started at 16 hours after drug exposure. Activation of caspase-8 by TNF-α coexposure was already observed between 10 and 12 hours after drug treatment (Fig. 2A). However, whereas enhanced initiation of caspase-8 cleavage correlated with caspase-3 activation and cleavage of the endogenous caspase-3 substrate PARP, caspase-9 activation was not significantly affected. Caspase-inhibition by z-VAD-fmk blocked activation of caspase-3 and PARP cleavage (Fig. 2A) as well as caspase-3 activity (Ac-DEVD-AMC cleavage) (Fig. 2B). Caspase-8 and caspase-9 cleavage was also partially inhibited by z-VAD-fmk and diclofenac/TNF-α coexposure (Fig. 2A, right panel). Together these data indicate that under diclofenac conditions TNF-α induces direct cleavage of caspase-8, which is the essential step in promoting the diclofenac-induced apoptotic response.

thumbnail image

Figure 2. Diclofenac/TNF-α-induced apoptosis is dependent on caspase-8 and -3 activities. (A) Expression levels of the active forms of caspase-8, -9, and -3 and the cleavage of PARP after 500 μM diclofenac exposure ± TNF-α (10 ng/mL) were measured by western blot analysis. “C,” controls exposed to vehicle for 24 hours ± TNF-α after 8 hours. (B) Activity of effector caspase-3 after exposure to diclofenac ± TNF-α (10 ng/mL) was measured by Ac-DEVD-AMC cleavage assay. The activity is presented as the fold of control where the control activity is ≈1 pmol AMC/min/mg protein. (C-E) The effect of specific caspase inhibitors on apoptosis induced by 500 μM diclofenac ± TNF-α (10 ng/mL) was investigated by following AnxV-Alexa633 binding to apoptotic cells over time using automated imaging. (C) Images, representative of three independent experiments (see Supporting Information for complete movies M1-M7); IETD, caspase-8 inhibitor z-IETD-fmk; LEHD, caspase-9 inhibitor z-LEHD-fmk; DEVD, caspase-3 inhibitor z-DEVD-fmk. (D) Quantification of the fluorescent AnxV-Alexa633-labeled apoptotic cells. (E) The difference in percent increase in AnxV-Alexa633 staining after 24 hours of drug exposure (+TNF-α) is shown, with the three independent experiments illustrated separately.

Download figure to PowerPoint

To further determine the individual roles of caspase-8, -9, and -3, we exposed HepG2 cells with selective irreversible inhibitors (z-IETD-fmk, z-LEHD-fmk, and z-DEVD-fmk, respectively) and measured the onset of apoptosis by AnxV-Alexa633 live cell imaging (Fig. 2C; Supporting Movies M5-7). All three caspase-inhibitors delayed the initiation of the diclofenac/TNF-α-induced apoptotic process (Fig. 2D). At the 24-hour timepoint, the inhibition of caspase-3 and caspase-8 significantly reduced the apoptotic response to diclofenac/TNF-α treatment, whereas the capase-9 inhibitor z-LEHD-fmk did not (Fig. 2E).

To obtain further insight into which players in the apoptotic pathways have important roles in the induction of diclofenac/TNF-α-induced apoptosis, we created an siRNA SMARTpool library containing 50 siRNAs against known pro- and antiapoptotic genes (Supporting Data S2). Following knockdown in three independent experiments, as a primary screen, HepG2 cells were treated with diclofenac/TNF-α, and onset of apoptosis was analyzed by live cell imaging. Candidate gene knockdown resulting in a >2.0-fold reduction in apoptosis after 24 hours of diclofenac/TNF-α exposure were defined as target genes involved in the onset of diclofenac/TNF-α-induced apoptosis (Fig. 3A, Supporting Data S2); knockdown resulting in a >2.0-fold increase in apoptosis were defined as apoptosis suppressors (Fig. 3B, Supporting Data S2). Knockdown of caspase-8, Bid, p53 up-regulated modulator of apoptosis (PUMA), apoptosis inducing factor (AIF), and apoptotic protease activating factor 1 (APAF1) inhibited diclofenac/TNF-α-induced apoptosis; knockdown of cellular FLICE-like inhibitory protein (c-FLIP), Bcl10, Bcl-B, and RIP-associated protein with a death domain (RAIDD) enhanced the apoptosis. Knockdown of caspase-9 and -3 partially inhibited apoptosis (Supporting Data S2). To further validate the effect of the knockdown of the mentioned genes, a deconvolution secondary screen was performed with the four individual siRNA sequences that were present in the siRNA SMARTpool mix used in the primary screen. Deconvolution analysis of caspase-3 and -9 were also included, to verify the mitochondrial-dependent APAF1-mediated apoptosome formation resulting in caspase-3 activation. Knockdown that resulted in a >1.5-fold increase or decrease in apoptosis compared to siControl transfected cells after 24 hours exposure of diclofenac/TNF-α in at least two out of four individual siRNA sequences were considered true mediators of diclofenac/TNF-α-induced apoptosis (Fig. 3A,B; Supporting Data S3). These included caspases-8, −3, and −9, Bid, APAF1, and PUMA. The caspase-8 inhibitory protein c-FLIP was identified as a reducer of diclofenac/TNF-α apoptosis.

thumbnail image

Figure 3. siRNA-mediated knockdown of central apoptotic machinery components. siRNAs targeting 50 apoptotic machinery genes were used to investigate their individual roles in diclofenac/TNF-α-induced apoptosis (500 μM/10 ng/mL). AnxV-Alexa633 staining was followed in time using automated imaging. (A) Genes defined as proapoptotic regulators of diclofenac/TNF-α based on the average >2.0-fold decrease compared to siControl transfected cells in the primary screen and two out of four single siRNA sequences showing the phenotype in the secondary deconvolution screen. (B) Genes defined as antiapoptotic components of diclofenac/TNF-α-induced apoptosis based on average >2.0-fold increase compared to siControl transfected cells in the primary screen and two out of four single siRNA sequences showing the phenotype in the secondary deconvolution screen. For both (A) and (B) the tables present the fold-change induction of apoptosis in the primary screen compared to siControl transfected HepG2 cells from three independent experiments as well as from averages of all three and the number of constructs in the secondary screen that resulted in the expected phenotype. The graphs present the average time-dependent diclofenac/TNF-α-induced apoptosis as well as the averages of the 24-hour timepoints for the hits from the primary screen that could also be validated in the secondary deconvolution screen. Data are the means of three independent experiments ± SEM. **P < 0.01, *P < 0.05.

Download figure to PowerPoint

Together, these data indicate that the route of apoptosis induction after diclofenac/TNF-α exposure is dependent on the extrinsic apoptotic pathway involving caspase-8. In addition, the siRNA screen revealed that the apoptosis was dependent on the mitochondrial apoptotic pathway initiated by the activation of Bid and included the release of APAF1-dependent activation of caspase-9 and -3.

Diclofenac and Diclofenac/TNF-α Coexposure Leads to JNK Activation in HepG2 Cells.

Diclofenac-induced small intestinal injury is dependent on JNK-activation.23 In addition, TNF-α-induced apoptosis can result from sustained JNK activation.24 Therefore, we determined the activation status of JNK after diclofenac and diclofenac/TNF-α treatment. Both diclofenac alone and diclofenac/TNF-α coexposure caused sustained JNK phosphorylation, although only little sustained activation was seen with TNF-α alone (Fig. 4A). The diclofenac/TNF-α-induced JNK activity was associated with increased phosphorylation of c-Jun, an important downstream target of JNK (Fig. 4A). Inhibition of JNK-activation by a specific inhibitor of JNK, SP600125, almost completely inhibited the onset of diclofenac/TNF-α-induced apoptosis (Fig. 3B-D).

thumbnail image

Figure 4. Diclofenac/TNF-α-induced apoptosis is dependent on JNK activity. (A) JNK activation in time after diclofenac exposure ± TNF-α was assessed by western blot analysis. Tubulin was used as loading control. “C,” controls exposed to vehicle for 24 hours ± TNF-α after 8 hours. The ratios represent the fold increase of phosphorylated protein expression over total protein compared to C (-TNF-α). (B-D) Cells were treated with the c-Jun N-terminal kinase (JNK)-specific inhibitor SP 600125 (20 μM) and its effect on apoptosis induced by 500 μM diclofenac ± TNF-α was followed by AnxV-Alexa633 staining and automated imaging (B) and by cell cycle analysis after 24 hours of drug exposure (D). The endpoint quantification of the AnxV-Alexa633 signal increase after 24 hours of drug exposure in the live apoptosis assay (B) is shown in (C). (E) Effect of siRNA-mediated knockdown of the individual isoforms JNK1 and JNK2 on apoptosis induction by diclofenac/TNF-α was followed by AnxV-Alexa633 staining and automated imaging. Data are the means of three independent experiments ± SEM. ***P < 0.001, **P < 0.01, *P < 0.05.

Download figure to PowerPoint

TNF-α-induced apoptosis is associated with prolonged activation of JNK1.25 However, other studies demonstrated a role for JNK2 as well.26, 27 We performed knockdown of the two major JNK isoforms, JNK1 and JNK2. Knockdown was verified by western blotting (Supporting Data S4). Live cell imaging of the apoptosis onset identified JNK1 as the main JNK isoform involved in diclofenac/TNF-α-induced apoptosis (Fig. 4E). Together, these data suggest an important role for JNK1 activation by diclofenac in sensitizing HepG2 cells towards enhanced cell injury induced by TNF-α.

Diclofenac Inhibits TNF-α-Induced NF-κB Signaling.

Under normal physiological conditions hepatocytes are resistant to TNF-α-induced apoptosis due to NF-κB-mediated transcriptional regulation of antiapoptotic target genes such as c-FLIP, thereby suppressing TNF-R-mediated caspase-8 activation.16 Activation of the IKK-complex by TNF-α causes phosphorylation of the inhibitor of NF-κB, IκBα, followed by its degradation by the proteasome allowing translocation of NF-κB to the nucleus. Because caspase-8 was central in diclofenac/TNF-α-induced apoptosis (see Figs. 2, 3), we next investigated whether the NF-κB signaling induced by TNF-α was affected by diclofenac. Diclofenac caused a concentration-dependent disturbance of the TNF-α-induced NF-κB signaling (Fig. 5). Diclofenac both decreased and delayed IκBα phosphorylation in association with an inhibition of its degradation (Fig. 5A, top and middle panels, respectively). The delay in IκBα degradation was associated with a disruption of NF-κB nuclear translocation shown by immunofluorescence staining of wildtype HepG2 cells and live cell imaging of HepG2 cells stably expressing the NF-κB subunit p65 coupled to GFP (GFPp65) (Fig. 5B,D; Supporting Movies M8-10). Automated image quantification of the p65 signal intensity ratio of nucleus/cytoplasm showed that diclofenac delays the onset of the second nuclear entry of NF-κB (Fig. 5C,E). This diclofenac-induced reduction of the TNF-α-induced NF-κB nuclear translocation response was associated with a decrease in the transcriptional NF-κB activity after TNF-α treatment (Fig. 5F). These data indicate that diclofenac interferes with the TNF-α-mediated NF-κB response dynamics.

thumbnail image

Figure 5. Diclofenac inhibits NF-κB signaling induced by TNF-α. (A) HepG2 cells were preexposed to diclofenac for 8 hours before adding TNF-α (10 ng/mL). The phosphorylation status of inhibitor of NF-κB (IκBα) and the subsequent degradation of the IκBα protein were determined by western blot analysis. Tubulin staining was used as loading control. Results are representative of three independent experiments and the fold changes (FC) represent the tubulin-normalized FC in protein expression compared to 0 minutes of TNF-α exposure. (B) The effect of 8 hours diclofenac exposure on NF-κB (p65) translocation into the nucleus was investigated over time by immunofluorescence staining of p65 (green). The nuclei were stained with Hoechst 33258 (blue). The cells were imaged using a BD Pathway 855 imager. The results are representative of three independent experiments. (C) The nuclear translocation pattern of NF-κB was determined using the BD AttoVision software. The graphs represent the means of three independent experiments ± SEM. (D,E) HepG2 GFPp65 cells were used to follow the translocation of NF-κB after TNF-α exposure live ± diclofenac pretreatment using automated confocal imaging with pictures taken every 6 minutes. (E) The graph represents the quantification of the GFPp65 nuclear/cytoplasmic intensity ratio normalized to the highest intensity ratio/cell. (F) NF-κB transcriptional activity was investigated using a NF-κB luciferase reporter construct. The luciferase activity was measured after 12 hours drug exposure ± 10 ng/mL TNF-α for the last 4 hours. Results are expressed as ratios of the luciferase activity measured after TNF-α exposure in nonpreexposed cells over diclofenac preexposed cells and represent the means from three independent experiments ± SEM. *P < 0.05.

Download figure to PowerPoint

NF-κB Signaling Is Essential to Prevent TNF-α-Mediated Enhancement of Diclofenac-Induced Apoptosis.

We anticipated that interference of the NF-κB signaling pathway is causally associated with diclofenac/ TNF-α-induced apoptosis. To inhibit NF-κB activation we used an inhibitor of IKKβ, BMS-345541 (BMS, 2 μM). Pretreatment with BMS for 6 hours prevented the TNF-α-induced nuclear entry of NF-κB (Fig. 6A) in association with decreased NF-κB transcriptional activity (Fig. 6B). Under these conditions, BMS strongly sensitized the HepG2 cells for diclofenac/TNF-α-induced apoptosis (Fig. 6C).

thumbnail image

Figure 6. Inhibition of IKKβ prevents nuclear NF-κB translocation and enhances diclofenac/TNF-α-induced apoptosis. (A) Immunofluorescence staining of HepG2 cells for NF-κB (red) and Hoechst (blue) after exposure to 10 ng/mL TNF-α. Preexposing the HepG2 cells to 2 μM of the IκB kinase (IKK)-inhibitor BMS-345541 for 6 hours, but not 1 hour successfully inhibits the NF-κB nuclear translocation. (B) HepG2 cells transiently expressing an NF-κB luciferase reporter were pretreated for 6 hours with 2 μM BMS-345541 followed by 10 ng/mL TNF-α for 4 hours before measuring luciferase activity. The graph shows the average of five independent experiments ± SEM. (C) HepG2 cells were preexposed to diclofenac and IKK-inhibitor BMS-345541 before adding TNF-α (10 ng/mL). After 24 hours of drug exposure cells were collected for cell cycle analysis by fluorescence activated cell sort (FACS). The graph shows the means from three independent experiments ± SEM. ***P < 0.001, **P < 0.01, *P < 0.05.

Download figure to PowerPoint

Next we determined whether knockdown of the NF-κB subunit p65 affected diclofenac/TNF-α cell killing. Lentiviral shRNA-based stable knockdown of p65/RelA reduced the levels of p65 as determined by immunofluorescence and western blotting NF-κB (Fig. 7A,B). Moreover, importantly, TNF-α-induced activation of a NF-κB-luciferase reporter construct was inhibited in shRelA-HepG2 cells to approximately the same extent as preexposure to the IKKβ inhibitor BMS (compare Figs. 6B and 7C). Knockdown of p65/RelA did not affect the basic level of apoptosis under both control and TNF-α conditions (Fig. 7D,E, respectively). Next, we exposed the different cell lines to a concentration range of diclofenac for 8 hours followed by TNF-α treatment. Loss of p65/RelA did not significantly enhance the induction of apoptosis by diclofenac alone (Fig. 7D). However, knockdown of p65/RelA sensitized HepG2 cells towards diclofenac/TNF-α-induced apoptosis (Fig. 7E). Together, these data confirm the important role of functional NF-κB signaling and nuclear translocation in protecting hepatocytes from TNF-α-enhanced diclofenac-induced apoptosis.

thumbnail image

Figure 7. Knockdown of p65 sensitizes HepG2 cells for TNF-α-enhanced diclofenac-induced apoptosis. (A) Immunofluorescence staining of NF-κB (green) on untreated wildtype (Wt) HepG2 cells and cells stably transduced with control or human RelA shRNA constructs. Nuclei were detected with Hoechst 33258 (blue). (B) Quantification of RelA/p65 levels in Wt, shControl, and shRelA HepG2 cells by western blot analysis. (C) Control shRNA and p65/hRelA shRNA transduced HepG2 cells transiently expressing an NF-κB luciferase reporter were treated with 10 ng/mL TNF-α for 4 hours before measuring luciferase activity. The graph shows the average of five independent experiments ± SEM. (C,D) Wt, control shRNA transduced and HepG2 cells stably knocked down for p65 (hRelA shRNA) were exposed to diclofenac for 24 hours without (C) or in the presence of TNF-α (10 ng/mL; D). The graphs represent the percentage of apoptosis detected by cell cycle analysis and shows the means from three independent experiments ± SEM. ***P < 0.001, *P < 0.05.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Drug-induced liver injuries are a significant problem in clinical practice as well as during drug development. There is increasing evidence for a role of the innate immune system in the pathophysiology of drug-induced liver injury which seems to involve an interaction between reactive metabolite formation and cytokine signaling.2, 5, 6 Diclofenac can cause drug-induced liver injury in an idiosyncratic manner. Here we studied in detail the role and mechanism of diclofenac/cytokine-induced stress signaling on the onset of hepatocyte apoptotic cell death. Our data indicate that: (1) TNF-α enhances diclofenac-induced cell death of HepG2 cells; (2) this cell death involves the onset of apoptosis which is dependent on a caspase-8/Bid/APAF1 cascade and the activity of caspase-3; (3) the diclofenac/TNF-α synergy involves an inhibition of TNF-α-induced NF-κB activity which is related to a disruption of IκBα phosphorylation and degradation and subsequent differential translocation of NF-κB to the nuclear compartment. Our results support a model in which diclofenac affects the TNF-α signaling program, thereby preventing the antiapoptotic actions of NF-κB and allowing a JNK- and death receptor-mediated mitochondrial onset of apoptosis.

Our data demonstrate that TNF-α enhances the cytotoxicity of diclofenac in both human HepG2 and mouse Hepa1c1c7 cells. Our data fit with a model whereby activated Kupffer cells in the liver release TNF-α and thereby aggravate liver injury.10-12 Although a role for TNF-α in diclofenac-induced liver injury has not previously been demonstrated, lipopolysaccharide (LPS) treatment severely enhances liver failure induced by diclofenac.13 Because LPS induces direct activation of Kupffer cells with the subsequent release of high levels of TNF-α,9 a direct interaction between diclofenac and TNF-α in the liver can be anticipated. Importantly, oral administration of diclofenac can cause severe injury to the intestine,28 thus creating a condition for increased systemic levels of endotoxins and activation of liver macrophages with a subsequent deleterious interaction with diclofenac in the liver. A role of polymorphonuclear neutrophils (PMNs) infiltration in the above-mentioned animal model of diclofenac/ LPS-induced liver injury was demonstrated by PMN-depleting antiserum.13 We suggest that this observation does not exclude the importance of (LPS-induced) TNF-α secretion in this model, as we simulate in our in vitro system. Thus, the PMN-depletion only provided partial protection against diclofenac/LPS liver injury, whereas the deleterious effect of the PMNs themselves could involve the release of TNF-α by these cells within the liver.28

The concentrations of diclofenac used to achieve diclofenac/TNF-α synergy (>100 μM) in this in vitro study exceed the maximal plasma concentration reached in humans after a single dose of 50 mg, 5-10 μM29 (with the potentially higher concentration in the liver after oral intake). This is also expected because HepG2 cells are poor in a drug-metabolizing capacity, but do form identical relevant diclofenac metabolites (see Supporting Data S1). Indeed, using HepG2 cells that express GFP-Nrf2, we could demonstrate stabilization of Nrf2 upon diclofenac treatment in a time- and concentration-dependent manner supportive for drug-reactive metabolite formation that has cell biological consequences (Herpers, Fredriksson, and Van de Water, unpubl. obs.). Although the observed patient concentrations are well below the concentration used in our studies, higher levels of drug metabolism in hepatocytes in the liver as well as chronic treatment of patients with diclofenac are likely to lead to equal levels of drug metabolite formation, including accumulation of diclofenac metabolite covalent modification of cellular proteins in hepatocytes, in particular after overdosing, conditions of liver function insufficiency, or reduced capacity of systemic diclofenac excretion.

Our data indicate that diclofenac/TNF-α-induced apoptosis involves a mitochondrial pathway. This is in accordance with previous observations that diclofenac alone can induce apoptosis via the mitochondrial death route, which involves an increase in free cytosolic calcium and induction of the mitochondrial pore transition (mPT),31 most likely through Bax-mediated mitochondrial outer membrane permeabilization (MOMP).22 We used a small siRNA library that targets all individual components of the apoptotic machinery. Thereby, we identified two Bcl-2 family members that contribute to the control of diclofenac/TNF-α-induced apoptosis, Bid and PUMA, two proapoptotic family members.32 Bid is activated through direct caspase-8-mediated cleavage, thereby inducing onset of the MOMP in mitochondria.33 Indeed, diclofenac/TNF-α activates caspase-8 and knockdown of caspase-8 as well as an inhibitor of caspase-8 effectively inhibited diclofenac/TNF-α-induced apoptosis (Figs. 2, 3A). Furthermore, knockdown of c-FLIP, an endogenous inhibitor of caspase-8, strongly enhances apoptosis (Fig. 3B). Diclofenac-induced permeabilization of the mitochondria is followed by release of proapoptotic factors which results in activation of caspase-9 and caspases-3-mediated apoptosis.22, 31, 34 In our hands only a weak activation of caspase-9 could be seen with diclofenac and diclofenac/TNF-α exposure (Fig. 2A), and in accordance with this, the specific caspase-9 inhibitor was not fully effective in inhibiting diclofenac/TNF-α-induced apoptosis (Fig. 2C-E). Nevertheless, knockdown of caspase-9 as well as of APAF1 was confirmed as essential for the diclofenac/TNF-α-induced apoptosis (Fig. 3A, Supporting Data S3). Together, this suggests a clear role for the apoptosome formation in this process.

We show that diclofenac directly affected the efficient TNF-α-induced activation of NF-κB. Diclofenac inhibited the TNF-α-mediated phosphorylation and degradation of the inhibitor of NF-κB, IκBα. This was directly associated with a shift in the oscillatory NF-κB translocation pattern and a reduced TNF-α-induced NF-κB transcriptional activity (Figs. 5, 6). In our model, NF-κB signaling is essential in the control of diclofenac/TNF-α-induced apoptosis. Thus, inhibition of the NF-κB activation using an IKKβ-inhibitor, BMS-345541, or stable shRNA-based RelA knockdown increases the cell death induced by diclofenac/TNF-α (Figs. 6C, 7E). The inhibition of NF-κB signaling by diclofenac most likely affects the antiapoptotic program typically induced by TNF-α in normal cells.15 This would result in the reduced expression of antiapoptotic molecules. Indeed, in cutaneous squamous cell carcinoma cells, diclofenac enhances death ligand-induced apoptosis which was associated with downregulation of c-FLIP,35 an NF-κB target gene that inhibits caspase-8.36 Knockdown of c-FLIP in our hands enhances diclofenac/TNF-α induced apoptosis (Fig. 3B). TNF-α alone did not cause apoptosis when NF-κB signaling was inhibited by BMS-345541 or RelA knockdown (Figs. 6C, 7E). This indicates that TNF-α signaling itself is not the main contributor to the onset of apoptosis, further supporting a synergistic action between TNF-α and the toxic properties of diclofenac, most likely involving formation of diclofenac reactive metabolites2, 5, 6 (Supporting Data S1).

In summary, we show that TNF-α enhances hepatocyte injury caused by diclofenac. We propose a mechanism by which diclofenac inhibits the TNF-α-induced nuclear translocation of NF-κB, thereby affecting the transcriptional activation of antiapoptotic molecules and allowing a caspase-8/Bid/APAF1 dependent onset of apoptosis (Fig. 8). These results shed new light on the interaction of hepatotoxic drugs and proinflammatory cytokines in drug-induced liver cell injury.

thumbnail image

Figure 8. Working model for diclofenac-TNF-α hepatotoxicity synergy. Reactive metabolites formed by diclofenac (DCF) metabolism inhibit IkBα phosphorylation in association with the NF-κB nuclear translocation oscillatory response and transcriptional activity. In the absence of the antiapoptotic NF-κB signaling, the synergistic activation of both the death receptor pathway and JNK signaling pathway mediate the onset of apoptotic cell death. This apoptosis is dependent on the caspase-8/Bid/APAF1/caspase-9/3 pathway with an additional involvement of the proapoptotic Bcl-2 family member PUMA.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
  • 1
    Lee WM. Etiologies of acute liver failure. Semin Liver Dis 2008; 28: 142-152.
  • 2
    Antoine DJ, Williams DP, Park BK. Understanding the role of reactive metabolites in drug-induced hepatotoxicity: state of the science. Expert Opin Drug Metab Toxicol 2008; 4: 1415-1427.
  • 3
    Aithal GP, Day CP. Nonsteroidal anti-inflammatory drug-induced hepatotoxicity. Clin Liver Dis 2007; 11: 563-575, vi-vii.
  • 4
    Laine L, Smith R, Min K, Chen C, Dubois RW. Systematic review: the lower gastrointestinal adverse effects of non-steroidal anti-inflammatory drugs. Aliment Pharmacol Ther 2006; 24: 751-767.
  • 5
    Kaplowitz N. Drug-induced liver injury. Clin Infect Dis 2004; 38( Suppl 2): S44-S48.
  • 6
    Ulrich RG. Idiosyncratic toxicity: a convergence of risk factors. Annu Rev Med 2007; 58: 17-34.
  • 7
    Boelsterli UA. Diclofenac-induced liver injury: a paradigm of idiosyncratic drug toxicity. Toxicol Appl Pharmacol 2003; 192: 307-322.
  • 8
    Han D, Shinohara M, Ybanez MD, Saberi B, Kaplowitz N. Signal transduction pathways involved in drug-induced liver injury. In: UetrechtJ, ed. Handbook of Experimental Pharmacology. Volume 196. 2009/12/19 ed. Heidelberg: Springer; 2010: 267-310.
  • 9
    Roberts RA, Ganey PE, Ju C, Kamendulis LM, Rusyn I, Klaunig JE. Role of the Kupffer cell in mediating hepatic toxicity and carcinogenesis. Toxicol Sci 2007; 96: 2-15.
  • 10
    Shaw PJ, Hopfensperger MJ, Ganey PE, Roth RA. Lipopolysaccharide and trovafloxacin coexposure in mice causes idiosyncrasy-like liver injury dependent on tumor necrosis factor-alpha. Toxicol Sci 2007; 100: 259-266.
  • 11
    Barton CC, Barton EX, Ganey PE, Kunkel SL, Roth RA. Bacterial lipopolysaccharide enhances aflatoxin B1 hepatotoxicity in rats by a mechanism that depends on tumor necrosis factor alpha. HEPATOLOGY 2001; 33: 66-73.
  • 12
    Shaw PJ, Ganey PE, Roth RA. Tumor necrosis factor alpha is a proximal mediator of synergistic hepatotoxicity from trovafloxacin/ lipopolysaccharide coexposure. J Pharmacol Exp Ther 2009; 328: 62-68.
  • 13
    Deng X, Stachlewitz RF, Liguori MJ, Blomme EA, Waring JF, Luyendyk JP, et al. Modest inflammation enhances diclofenac hepatotoxicity in rats: role of neutrophils and bacterial translocation. J Pharmacol Exp Ther 2006; 319: 1191-1199.
  • 14
    Hoffmann A, Baltimore D. Circuitry of nuclear factor kappaB signaling. Immunol Rev 2006; 210: 171-186.
  • 15
    Beyaert R, Van Loo G, Heyninck K, Vandenabeele P. Signaling to gene activation and cell death by tumor necrosis factor receptors and Fas. Int Rev Cytol 2002; 214: 225-272.
  • 16
    Micheau O, Tschopp J. Induction of TNF receptor I-mediated apoptosis via two sequential signaling complexes. Cell 2003; 114: 181-190.
  • 17
    Wullaert A, Heyninck K, Beyaert R. Mechanisms of crosstalk between TNF-induced NF-kappaB and JNK activation in hepatocytes. Biochem Pharmacol 2006; 72: 1090-1101.
  • 18
    Papa S, Bubici C, Zazzeroni F, Franzoso G. Mechanisms of liver disease: cross-talk between the NF-kappaB and JNK pathways. Biol Chem 2009; 390: 965-976.
  • 19
    Puigvert JC, de Bont H, van de Water B, Danen EH. High-throughput live cell imaging of apoptosis. Curr Protoc Cell Biol 2010; 47: 18.10.11-18.10.13.
  • 20
    van de Water B, Houtepen F, Huigsloot M, Tijdens IB. Suppression of chemically induced apoptosis but not necrosis of renal proximal tubular epithelial (LLC-PK1) cells by focal adhesion kinase (FAK). Role of FAK in maintaining focal adhesion organization after acute renal cell injury. J Biol Chem 2001; 276: 36183-36193.
  • 21
    de Graauw M, Tijdens I, Cramer R, Corless S, Timms JF, van de Water B. Heat shock protein 27 is the major differentially phosphorylated protein involved in renal epithelial cellular stress response and controls focal adhesion organization and apoptosis. J Biol Chem 2005; 280: 29885-29898.
  • 22
    Siu WP, Pun PB, Latchoumycandane C, Boelsterli UA. Bax-mediated mitochondrial outer membrane permeabilization (MOMP), distinct from the mitochondrial permeability transition, is a key mechanism in diclofenac-induced hepatocyte injury: multiple protective roles of cyclosporin A. Toxicol Appl Pharmacol 2008; 227: 451-461.
  • 23
    Ramirez-Alcantara V, Loguidice A, Boelsterli UA. Protection from diclofenac-induced small intestinal injury by the JNK inhibitor SP600125 in a mouse model of NSAID-associated enteropathy. Am J Physiol Gastrointest Liver Physiol 2009; 297: G990-G998.
  • 24
    Deng Y, Ren X, Yang L, Lin Y, Wu X. A JNK-dependent pathway is required for TNFalpha-induced apoptosis. Cell 2003; 115: 61-70.
  • 25
    Chang L, Kamata H, Solinas G, Luo JL, Maeda S, Venuprasad K, et al. The E3 ubiquitin ligase itch couples JNK activation to TNFalpha-induced cell death by inducing c-FLIP(L) turnover. Cell 2006; 124: 601-613.
  • 26
    Kodama Y, Taura K, Miura K, Schnabl B, Osawa Y, Brenner DA. Antiapoptotic effect of c-Jun N-terminal kinase-1 through Mcl-1 stabilization in TNF-induced hepatocyte apoptosis. Gastroenterology 2009; 136: 1423-1434.
  • 27
    Wang Y, Singh R, Lefkowitch JH, Rigoli RM, Czaja MJ. Tumor necrosis factor-induced toxic liver injury results from JNK2-dependent activation of caspase-8 and the mitochondrial death pathway. J Biol Chem 2006; 281: 15258-15267.
  • 28
    Cassatella M. Neutrophil-derived proteins: selling cytokines by the pound. Adv Immunol 1999; 73: 369-509
  • 29
    Willis JV, Kendall MJ, Flinn RM, Thornhill DP, Welling PG. The pharmacokinetics of diclofenac sodium following intravenous and oral administration. Eur J Clin Pharmacol 1979; 16: 405-410
  • 30
    Bjarnason I, Hayllar J, MacPherson AJ, Russell AS. Side effects of nonsteroidal anti-inflammatory drugs on the small and large intestine in humans. Gastroenterology 1993; 104: 1832-1847.
  • 31
    Lim MS, Lim PL, Gupta R, Boelsterli UA. Critical role of free cytosolic calcium, but not uncoupling, in mitochondrial permeability transition and cell death induced by diclofenac oxidative metabolites in immortalized human hepatocytes. Toxicol Appl Pharmacol 2006; 217: 322-331.
  • 32
    Willis SN, Adams JM. Life in the balance: how BH3-only proteins induce apoptosis. Curr Opin Cell Biol 2005; 17: 617-625.
  • 33
    Garrido C, Galluzzi L, Brunet M, Puig PE, Didelot C, Kroemer G. Mechanisms of cytochrome c release from mitochondria. Cell Death Differ 2006; 13: 1423-1433.
  • 34
    Gomez-Lechon MJ, Ponsoda X, O'Connor E, Donato T, Castell JV, Jover R. Diclofenac induces apoptosis in hepatocytes by alteration of mitochondrial function and generation of ROS. Biochem Pharmacol 2003; 66: 2155-2167.
  • 35
    Fecker LF, Stockfleth E, Braun FK, Rodust PM, Schwarz C, Kohler A, et al. Enhanced death ligand-induced apoptosis in cutaneous SCC cells by treatment with diclofenac/hyaluronic acid correlates with downregulation of c-FLIP. J Invest Dermatol 2010; 130: 2098-2109.
  • 36
    Karin M, Lin A. NF-kappaB at the crossroads of life and death. Nat Immunol 2002; 3: 221-227.

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Additional Supporting Information may be found in the online version of this article.

HEP_24314_sm_suppinfo1.doc21KSupp Info 1
HEP_24314_sm_suppinfo2.doc28KSupp Info 2
HEP_24314_sm_suppinfo3.doc94KSupp Info 3

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.