New insights in acetaminophen toxicity: HMGB1 contributes by itself to amplify hepatocyte necrosis in vitro through the TLR4-TRIF-RIPK3 axis

Extracellular release of HMGB1 contributes to acetaminophen-induced liver injury. HMGB1 acts as a danger-associated molecular patterns during this toxic process but the mechanisms of action and targeted cells are incompletely defined. Here we studied, in vitro, the role of HMGB1 in amplifying the acetaminophen-induced hepatocyte necrosis process. Using cultured HepaRG cells, primary human hepatocytes and selective chemical inhibitors we evaluated acetaminophen-induced toxicity. We confirmed that addition of acetaminophen induced HepaRG cell death and HMGB1 release. We showed that inhibition of HMGB1 decreased acetaminophen-induced HepaRG cell death, suggesting a feedforward effect. We provide the first evidence that exposure of HepaRG cells to recombinant human HMGB1 (rhHMGB1) also resulted in cell death. Moreover, we found that both acetaminophen and rhHMGB1 induced programmed HepaRG cell necrosis through a RIPK3-dependent mechanism. By using TLR4 blocking antibody, we demonstrated the reduction of the HepaRG cell death induced by acetaminophen and rhHMGB1. Furthermore, inhibition of TRIF, known to induce a RIPK3-dependent cell death, reduced rhHMGB1-induced HepaRG cell death. Our data support that released HMGB1 from acetaminophen-stressed hepatocytes induced necrosis of neighboring hepatocytes by TLR4-TRIF-RIPK3- pathway. This in vitro study gives new insights in the role of HMGB1 in the amplification of acetaminophen-induced toxicity.

A release of danger-associated molecular patterns (DAMPs) by dying hepatocytes during APAP-induced liver injury was demonstrated in both murine models and patients. The current paradigm suggests that cell death leads to the release of DAMPs alerting immune cells and triggering a self-amplifying circuit resulting to tissue damage and finally organ failure 7 . However, the current knowledge on those DAMPs that activate hepatic inflammatory cells is controversial.
High-mobility Group Box 1 (HMGB1) is a DAMP originating from the nucleus and released during the necrotic process 8 . HMGB1 protein is a small ubiquitous non-histone nuclear protein that acts as an architectural chromatin-binding factor involved in the maintenance of nucleosome structure and regulation of gene transcription by binding the minor groove of DNA 9,10 .
Two types of cells that can release HMGB1 have been reported. Firstly, immune cells (such as macrophages, mature dendritic cells, and natural killer cells) can actively secrete HMGB1 as part of the inflammatory response 11,12 . Secondly, necrotic cells can also release HMGB1. Although HMGB1 is a nuclear protein, cytoplasmic HMGB1 can be detected in necrotic cells and it can be released into the extracellular space by cellular damage 13 . Serum HMGB1 levels are increased in humans APAP hepatotoxicity and correlated with the severity of liver failure 14,15 . In mice, HMGB1 blockade significantly reduces the severity of APAP-induced hepatotoxicity 16,17 .
The mechanism of action of HMGB1 in APAP-induced liver injury remains unclear. In this study, we hypothesized that HMGB1, released during the hepatotoxic process induced by APAP, could act directly on hepatocytes independently of any intervention by inflammatory cells. (b) HepaRG cells were treated with APAP (10 mM) for 24 hours and GSH levels were measured. The enzyme concentration obtained is expressed as nanomoles of enzyme per milligram of protein using bovine serum as a standard. ** p < 0.01 vs vehicle. (c) HepaRG cells were exposed to increasing concentrations of APAP (0-20 mM) and cell viability was assessed by MTT (expressed as a percentage of unexposed cells (0)) at 24 hours. Results are expressed as mean ± SEM. ** vs. 0, p < 0.01. Experiments were reproduced three times.
www.nature.com/scientificreports www.nature.com/scientificreports/ To investigate this hypothesis, in vitro models are indispensable tools. In this study, HepaRG cells were used. HepaRG cell line is an immortalized hepatic cell line who exhibit many characteristics of primary human hepatocytes including morphology, expression of key metabolic enzymes, expression of nuclear receptors, and drug transporters. Thanks to all these features, HepaRG cells are a recognized tool to investigate APAP hepatotoxicity [18][19][20] .

Results
HMGB1 is released by APAP-damaged hepatocytes. As previously shown, HepaRG cells were able to metabolize APAP. Indeed, a time-dependent decrease of APAP trace quantities in the supernatant (reflected by absorbance decrease) and reduced glutathione (GSH) levels in HepaRG cells at 24 hours of APAP exposure confirmed it (Fig. 1a,b). Moreover, a significant dose-dependent HepaRG death was observed from 10 mM of APAP with a cell viability percentage of 35.4 ± 3.34% (p < 0.001) (Fig. 1c).
HMGB1 is expressed in the nucleus of HepaRG cells (Fig. 2a) and its release in the supernatant was observed when the cells were exposed to APAP for 24 hours in a dose-dependent manner as shown in Fig. 2b (significantly (p = 0.009) from 10 mM of APAP with HMGB1 supernatant concentration at 56.6 ± 1.74 ng/ml). HMGB1 amplifies by itself APAP-induced hepatocyte death. Supernatant transfer experiments from APAP-exposed HepaRG cells to naïve HepaRG cells (cells non previously exposed to APAP) were performed (Suppl Fig. 1). HepaRG cells were exposed to APAP (10 mM) for 6 hours, then washed to remove cell debris and new culture medium was added for an additional 6 hours. The latter supernatant, containing cell components and HMGB1 (Fig. 3a), was then added to naïve HepaRG cells for 12 hours. As shown in Fig. 4b, this supernatant induced a 18.4% mortality rate in naïve HepaRG cells (p = 0.02). However, this toxic effect is preventable by the addition of glycyrrhizin in the new culture medium (100 µM) who reduced the HepaRG cells mortality to 3.1% (p = 0.021) (Fig. 3b).
The addition of drugs known to act on HMGB1 protein, as glycyrrhizin (GL; 100μM) or ethyl pyruvate (EP; 4 mM), significantly improved HepaRG cells survival (Fig. 4a). Indeed, compared to APAP alone, addition of GL or EP increased cell viability to around 75% (p < 0.001) and 60% (p < 0.001), respectively, at 24 hours. In parallel, decreased HMGB1 concentration in the cell supernatant was observed (Fig. 4b). HMGB1 concentration dropped from 61.36 ±1.73 to 24.52 ±0.94 ng/ml (p = 0.001) after GL treatment and 42.26 ±1.26 ng/ml (p = 0.014) after EP treatment. Due to the design of the experiments, the concentration of HMGB1 in the supernatant were significantly lower than in experiments with APAP treated HepaRG (Fig. 2b). This can explain the relatively small mortality rate (18%) induced by the supernatant on naïve HepaRG cells. www.nature.com/scientificreports www.nature.com/scientificreports/ To test a direct effect of HMGB1, HepaRG cells were exposed to increasing concentration of recombinant human HMGB1 (rhHMGB1). rhHMGB1 mediated HepaRG cell death in a dose-dependent manner as shown in Fig. 5. A significant decrease of HepaRG cells viability was observed with 50 ng/mL (p = 0.001) of rhHMGB1 (in the same range of concentrations of native HMGB1 measured in the supernatant of APAP-treated HepaRG cells in Fig. 2b).
This experiment was reproduced on primary human hepatocytes. We have previously analyzed the release of HMGB1 by HepaRG cells as a function of APAP exposure time (Fig. 6a). Thereafter, rhHMGB1 at 300 ng/ml was added during 24 hours and cell viability was assessed. As shown in Fig. 6b, rhHMGB1 also induced significant cell death in primary human hepatocytes with a drop of cell viability to 56.41 ± 3.16% (p = 0.001).

HMGB1 and APAP induce hepatocytes necroptosis by RIPK3-dependent pathway.
To understand how HMGB1 acts on HepaRG cells, we explored the potential pathway involved in APAP and rhHMGB1-induced cell death. Significant lactate dehydrogenase (LDH) release was observed after addition of rhHMGB1 (p < 0.001), as well as APAP (p < 0.001), in absence of significant caspase-3 activation as shown in Figs. 7a and b, respectively. These results suggest that necrosis represent the main death process in our model. However, different forms of regulated caspase-independent cell death have been described and necroptosis, which requires receptor interacting protein kinases (RIPK), is one of these regulated mechanisms. As demonstrated in the literature, RIPK3 and RIPK1 are implicated in the necroptosis pathway and the expression of these two proteins by HeparRG cells was confirmed at both protein (Figs. 8a and 9a) and mRNA levels (Figs. 8b and 9b).
In order to investigate if necroptosis is implicated in our model, inhibitors of RIPK3 and RIPK1 were used. HepaRG cells were pretreated 1 hour with inhibitors before addition of APAP or rhHMGB1 and cell viability was assessed at 24 hours. The addition of dabrafenib, a potent inhibitor of RIPK3 kinase activity, rescued significantly both APAP-and rhHMGB1-mediated HepaRG cell death in a dose-dependent manner (Fig. 8c,d). On Naïve HepaRG cells were exposed to supernatant of HepaRG cells previously stressed by APAP (10 mM) in presence or absence of glycyrrhizin (100 µM) for 12 hours. Cell viability was assessed by MTT method. Results are expressed as mean ± SEM. * p < 0.05 vs control. # p < 0.05 vs APAP. Experiments were reproduced three times. (2020) 10:5557 | https://doi.org/10.1038/s41598-020-61270-1 www.nature.com/scientificreports www.nature.com/scientificreports/ the contrary, the addition of necrostatin-1, a selective inhibitor of RIPK1 kinase activity, to HepaRG cells had no significant effect on APAP-and rhHMGB1-mediated HepaRG cell death (Fig. 9c,d).
Necroptosis requires RIPK3-mediated activation of mixed lineage kinase domain-like protein (MLKL) which induces necrotic membrane rupture 21,22 . Alternatively, RIPK3 can induce mitochondrial dynamin-related protein 1 (Drp1) translocation and fission 23 . Inhibitors were added 1 hour prior APAP or rhHMGB1 and cell viability was assessed at 24 hours. Significant reduction of APAP-mediated cell death was observed in Fig. 10a after addition of NSA, an MLKL inhibitor, at 2.5 μM (p = 0.025). Similarly, rhHMGB1-mediated cell death was also reduced as observed in Fig. 10b (p = 0.045). If an inhibitor of Drp1, Mdivi-1, was added at 50 μM to HepaRG cells, significant reduction of both APAP-and rhHMGB1-mediated HepaRG cell death was observed as shown in Fig. 10c (p = 0.004) and Fig. 10d (p = 0.045), respectively. HMGB1 induces hepatocyte necroptosis through TLR4 adaptor TRIF pathway. After the demonstration of RIPK3 implication in our model, we explored which signaling pathway is involved upstream.
Signaling by TLRs involves cytosolic Toll/Interleukin-1 receptor (TIR) domain-containing adaptor, such as TRIF. A noncanonical necroptosis has been described in which TRIF activates RIPK3 without the requirement of RIPK1 24 . Expression of TRIF at mRNA level by HepaRG cells was confirmed in Fig. 12a. Using a peptide inhibitor (PepInh), the potential implication of TRIF has been investigated in our model. HepaRG cells were pretreated PepInh-TRIF at 40 µM 6 hours prior rhHMGB1 (300 ng/ml). Cell viability was assessed 24 hours after rhHMGB1 addition and improved HepaRG cell viability by 20.27% (p = 0.009) was observed as shown in Fig. 12b.

Discussion
Our study suggests a new explicative mechanism for the propagation of hepatocyte death during APAP-induced liver injury. First, our experiments reveal that released HMGB1 from dying hepatocytes contributes by a feedforward circuit to the amplification of hepatocyte necrosis. Actually, many DAMPS are probably released in the supernatant of APAP-treated-cells and not only HMGB1. HMGB1 protein is known to bind to other DAMPS to promote various signaling pathways 25 . We have not explored this part. However, we have been able to prove the existence of HMGB1 in the supernatant exposed to naïve cells (Fig. 4a). Second, we observed that HMGB1 www.nature.com/scientificreports www.nature.com/scientificreports/ induces hepatocyte necrosis by RIPK3-dependent pathway suggesting necroptosis implication. Third, we observed that released HMGB1 directly acts on neighboring hepatocytes through TLR4-TRIF pathway.
In accordance with others, we have confirmed that the administration of GL reduces the severity of APAP-induced liver injury in murine model 13,26,27 . In parallel, in mice treated with GL, we observed a decrease in the release of HMGB1 from hepatocytes: immunostaining quantification and serum HMGB1 levels were measured. GL is known to have many effects including inhibition of HMGB1. Based on these results, the present study was initiated to assess whether HMGB1 released from dying hepatocytes could act on neighboring hepatocytes independently of inflammatory cells. We confirmed this hypothesis in vitro using cultured HepaRG cells. As expected, and without interference with APAP bioactivation (Suppl Fig. 4), we observed that the addition of GL or EP reduced APAP-induced HepaRG cell death, suggesting that HMGB1 contributed to the APAP-induced death process.
We demonstrated 28 that hepatocytes expressed the cognate HMGB1 receptors TLR2, TLR3, TLR4, and TLR9 allowing for a direct response to released HMGB1 and that HMGB1 by itself can induce cell death. Several previous experiments using knock-out mice have suggested that TLR2, TLR3, TLR4, TLR9, and RAGE could contribute to APAP-induced liver injury although some controversy remains pertaining to possible dual or opposite effects according to which cell types are studied [29][30][31] . We focused on TLR4 receptor in our experiments. Inhibition of TLR4 receptor using blocking antibody, prevents APAP induced HepaRG cells death. These results suggested the implication of TLR4 receptor in the hepatotoxicity process induced by APAP.
We observed that both APAP and HMGB1-induced HepaRG death was characterized by LDH release without implication of caspase-3. Different forms of regulated necrosis have been described and to distinguish www.nature.com/scientificreports www.nature.com/scientificreports/ RIPK-dependent necrosis from other forms, the term necroptosis is used. We know that the activation of RIPK1 and RIPK3 is required for the activation of the necrosome. In this study, we investigated the inhibition of each one. We showed that RIPK3 inhibition by dabrafenib significantly attenuated APAP and rhHMGB1-induced HepaRG cell death without interference with APAP bioactivation (Suppl Fig. 4). Previous studies have shown that RIPK3 genetic ablation or inhibition by dabrafenib partially protected mice from APAP-induced liver injury although this observation was not reproduced in RIPK3−/− mice [32][33][34] . The reasons of these discrepancies are uncertain. In the other hand, the inhibition of RIPK1 by necrostatin-1 is still controversial: protective 35 and deleterious 36,37 roles have been reported. In our in vitro experiments, inhibition of RIPK1, had no impact on rhHMGB1-and APAP-induced cell death in our experiments. Taken together, our in vitro data support a model in which HMGB1 causes hepatocyte cell death via RIPK3-dependent, RIPK1-independent, noncanonical activation of necroptosis.
As previously mentioned, hepatocytes express a part of TLR repertoire which is able to bind HMGB1. An intracellular adaptor implicated in TLR signaling, called TRIF, have a Receptor interaction protein kinase (RIPK) homology interaction motif (RHIM) which activates in some instances, in cooperation with RIPK3, a noncanonical programmed cell necrosis 24 . Interestingly, we showed that inhibition of TRIF by Pepinh-TRIF significantly attenuated rhHMGB1-induced HepaRG cell death. In addition, we previously observed that Trif mutant mice are partially protected against APAP-induced liver damage 28 .
In conclusion, we observed in vitro that, in APAP-induced hepatotoxicity, released HMGB1 from damaged hepatocytes contributed to the propagation of necrosis by a direct effect on neighboring hepatocytes in a feed-forward circuit. Our results give new insights in the acetaminophen toxicity by suggesting that HMGB1 amplifies by itself the hepatocyte necrosis trough TLR4 -TRIF-RIPK3-dependent pathway.

Scientific RepoRtS |
(2020) 10:5557 | https://doi.org/10.1038/s41598-020-61270-1 www.nature.com/scientificreports www.nature.com/scientificreports/ 24-well plate and exposed to APAP for only 6 hours; the time required for HepaRG cells to become stressed (minor inducer of death). Cells were then washed to remove cell debris and new culture medium was added for an additional 6 hours without APAP. The latter supernatant was added to the naïve cells for 12 hours. In this Neutralizing monoclonal antibody to TLR4 at 50 μg/ml was added to HepaRG cells 1 hour prior APAP. Cellular viability was assessed by MTT method 24 hours after APAP (10 mM). Results are expressed as mean ± SEM. * vs. ctrl p < 0.05; # vs. APAP, p < 0.05. Experiments were reproduced three times. (d) Neutralizing monoclonal antibody to TLR4 at 50 μg/ml was added to HepaRG cells 1 hour prior rhHMGB1 (300 ng/ml). Cellular viability was assessed by MTT method 24 hours after rhHMGB1 addition. Results are expressed as mean ± SEM. ** vs. ctrl p < 0.01; # vs. HMGB1, p < 0.05. Experiments were reproduced three times. In vitro, HepaRG cells were exposed to a TRIF inhibitor peptide (Pepinh-TRIF, 40 µM) or to a control peptide (Pepinh-Control, 40 µM), 6 h prior rhHMGB1 (300 ng/ml) and cell viability was evaluated by MTT after 24 h. Results are expressed as mean ± SEM. ** vs. HMGB1, p < 0.01.

Species Direction Sequences
Classic PCR www.nature.com/scientificreports www.nature.com/scientificreports/ experiment, the aim was to investigate whether compounds (DAMPs) released by stressed cells was capable of stress/killing naïve cells. In all other experiments performed, HepaRG cells were seeded on 96-well plate and exposed to APAP for 24 hours (major inducer of death).
For HepaRG cells differentiation, cells were cultured in the medium for 2 weeks and then in presence of 2% dimethyl sulfoxide (DMSO) for 2 more weeks. Thawing, seeding, culture, differentiation and maintenance of HepaRG cells for toxicity study were performed as described in the HepaRG TM Cell User Guide provided by the firm and as previously described in the literature 18 .
Primary human hepatocytes (HU1539, Life Technologies, Waltham, MA, USA) were thawed and plated in collagen I-coated 24-wells plates following the manufacturer's protocol. Cells were cultured 24 hours in incubation medium (Williams' E medium (1×, no phenol red) and Hepatocyte Maintenance Supplement Pack (serum free)) at 37 °C with 5% CO2. After that, primary human hepatocytes were exposed to rhHMGB1 for 24 hours at 300 ng/ml (concentration measured in the supernatant of HepaRG cells APAP-treated as shown in Fig. 6a).
Cell viability assay. Cell viability was evaluated using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). Briefly, 10 µL of MTT diluted in 90 µL of culture medium were added to each well and incubated for 3 hours at 37 °C. After removing MTT solution, 100 µL of hydrochloric acid/isopropanol was added to each well for 10 min. The absorbance was read at 540 nm. The relative cell viability (%) related to control wells containing cells and culture medium without treatment was calculated by ((A) test/(A) control) × 100.
Cell death process. Lactate dehydrogenase (LDH) activity was determined using the Pierce LDH Cytotoxicity Assay Kit (ThermoFisher Scientific, Belgium) per the manufacturer's instructions. Caspase-3 activity was quantified using the Caspase-3 colorimetric assay (R&D Systems, Minnesota, USA) per the manufacturer's instructions.
Enzyme-linked immunosorbent assay (ELISA). HMGB1 concentrations in the supernatant of cultured HepaRG cells was measured by a sandwich-enzyme immunoassay (IBL International GmbH, Hamburg, Germany) following the manufacturer's protocol.
Trace quantities of acetaminophen in the supernatant of cultured HepaRG cells was measured using Neogen Corporation's test kit (Lexington, KY, USA). It operates based on competition between the drug in the sample and the drug-enzyme conjugate for a limited number of antibody binding sites. The presence of bound drug-enzyme conjugate is recognized by the addition of K-Blue ® Substrate. The extent of color development at 450 nm is inversely proportional to the amount of drug in the sample.