Redefining IL11 as a regeneration-limiting hepatotoxin and therapeutic target in acetaminophen-induced liver injury

Inhibition of IL11 signaling limits drug-induced liver damage and promotes hepatic regeneration in a mouse model. A matter of species specificity Acetaminophen (APAP) overdose can cause liver injury; effective therapies for treating APAP poisoning beyond 8 hours after ingestion are lacking. Recombinant human interleukin 11 (rhIL11) protected rodents from liver injury; however, recent studies produced results that question the underlying mechanism. Here, Widjaja et al. used a mouse model of APAP-induced liver injury and showed that species-matched IL11 was detrimental in mice, causing hepatocyte cell death. Genetic IL11 deletion protected mice from liver damage and administration of an antibody targeting IL11 receptor reduced APAP-induced toxicity even when administered 10 hours after APAP. The results suggest that IL11 might be detrimental for hepatocytes. Additional studies will clarify the translational potential of targeting IL11 for treating liver injury. Acetaminophen (N-acetyl-p-aminophenol; APAP) toxicity is a common cause of liver damage. In the mouse model of APAP-induced liver injury (AILI), interleukin 11 (IL11) is highly up-regulated and administration of recombinant human IL11 (rhIL11) has been shown to be protective. Here, we demonstrate that the beneficial effect of rhIL11 in the mouse model of AILI is due to its inhibition of endogenous mouse IL11 activity. Our results show that species-matched IL11 behaves like a hepatotoxin. IL11 secreted from APAP-damaged human and mouse hepatocytes triggered an autocrine loop of NADPH oxidase 4 (NOX4)–dependent cell death, which occurred downstream of APAP-initiated mitochondrial dysfunction. Hepatocyte-specific deletion of Il11 receptor subunit alpha chain 1 (Il11ra1) in adult mice protected against AILI despite normal APAP metabolism and glutathione (GSH) depletion. Mice with germline deletion of Il11 were also protected from AILI, and deletion of Il1ra1 or Il11 was associated with reduced c-Jun N-terminal kinase (JNK) and extracellular signal–regulated kinase (ERK) activation and quickly restored GSH concentrations. Administration of a neutralizing IL11RA antibody reduced AILI in mice across genetic backgrounds and promoted survival when administered up to 10 hours after APAP. Inhibition of IL11 signaling was associated with the up-regulation of markers of liver regenerations: cyclins and proliferating cell nuclear antigen (PCNA) as well as with phosphorylation of retinoblastoma protein (RB) 24 hours after AILI. Our data suggest that species-matched IL11 is a hepatotoxin and that IL11 signaling might be an effective therapeutic target for APAP-induced liver damage.


INTRODUCTION
Acetaminophen (N-acetyl-p-aminophenol; APAP) is a commonly used over-the-counter drug, but APAP poisoning is a major cause of drug-induced liver injury and failure (1). The antioxidant Nacetylcysteine (NAC) is beneficial for patients presenting early with APAP poisoning (2), but there is no drug-based treatment beyond 8 hours after ingestion and death can ensue if liver transplantation is not possible (3,4).
In hepatocytes, APAP is metabolized to N-acetyl-p-benzochinonimin (NAPQI), which depletes cellular glutathione (GSH) and damages mitochondrial proteins, leading to reactive oxygen species (ROS) production and c-Jun N-terminal kinase (JNK) activation (5). ROSrelated JNK activation results in a combination of necrotic and other forms of hepatocyte cell death (1,6,7). JNK and mitogen-activated protein kinase kinase kinase 5 (MAP3K5; also known as ASK1) inhibitors have partial protective effects against APAP-induced liver injury (AILI) in mouse models, but toxicities limit its translation to the clinic (8,9). Similarly, although caspase cleavage is seen in AILI, pan-caspase inhibitors have proven ineffective and hepatocyte apoptosis is not thought to play a major role (10). Failure of caspase inhibitors could reflect caspase cleavage occurring downstream of multiple forms of cell death, a biomarker of cellular demise rather than of a specific type of cell death.
Liver regeneration after hepatic injury can be profound in both rodents and humans, as seen after partial hepatic resection (11,12). In the setting of AILI, liver regeneration is suppressed, resulting in permanent injury. Targeting the pathways that hinder the liver's regenerative capacity may trigger natural regeneration, which could be specifically useful in AILI (13,14).
Interleukin 11 (IL11) is a cytokine that is of central importance for myofibroblast activation across organs (15)(16)(17)(18). It is known that IL11 is secreted from APAP-injured hepatocytes in mice and that IL11 can be detected at very high concentration in the serum of the mouse model of AILI, where its expression is considered compensatory and cytoprotective (19). In keeping with the idea that IL11 is beneficial in the liver, administration of recombinant human IL11 (rhIL11) is effective in treating the mouse model of AILI and also protects against liver ischemia, endotoxemia, or inflammation (19)(20)(21)(22)(23)(24). As recently as 2016, rhIL11 has been proposed as a treatment for patients with AILI (25).
During our recent studies of nonalcoholic steatohepatitis (NASH), we found that IL11 appears to be detrimental for hepatocyte function, at least in some contexts (15,26). The apparent discrepancy with the previous literature prompted us to look in more detail at the effects of IL11 in the mouse model of AILI, where endogenous mouse IL11 is largely up-regulated and rhIL11 is protective (19).

IL11 drives APAP-induced hepatocyte cell death
As reported previously (19), we confirmed that AILI is associated with largely elevated concentrations of IL11 in the serum of mice (Fig. 1A). We addressed whether the elevated IL11 serum concentration in the mouse AILI model originated in the liver and found that APAP largely up-regulated hepatic Il11 expression (35-fold, P < 0.0001) (Fig. 1B). Bioluminescent imaging of a reporter mouse with luciferase cloned into the start codon of Il11 indicated IL11 expression throughout the liver (Fig. 1C and fig. S1, A and B). Western blotting confirmed IL11 up-regulation across a time course of AILI (Fig. 1D). Experiments using a second reporter mouse with an enhanced green fluorescent protein (EGFP) reporter construct inserted into the 3′ untranslated region (UTR) of Il11 (fig. S1C) showed that, after APAP, IL11 is highly expressed in necrotic centrilobular hepatocytes, the pathognomonic feature of AILI, coincident with cleaved caspase 3 (Cl. CASP3) ( Fig. 1E and fig. S1, D and E).
Having identified hepatocytes as a source of Il11 during AILI in vivo, we conducted in vitro experiments. Exposure of primary human hepatocytes to APAP resulted in the dose-dependent secretion of IL11 (Fig. 1F). Hepatocytes highly express the IL11 receptor subunit  (IL11RA), and we have observed that IL11 can be hepatotoxic (15), which we confirmed in adult human hepatocytes from additional donors ( fig. S2, A and B). IL11 activates extracellular signal-regulated kinase (ERK) in some cell types (15); hence, we explored the effect of IL11 on ERK and JNK activation in hepatocytes. IL11 induced late and sustained ERK and JNK activation that was concurrent with CASP3 cleavage (Fig. 1G). Flow cytometrybased analyses showed dose-dependent IL11-induced hepatocyte cell death ( Fig. 1H and fig. S2C).
To explore the potential role of IL11 signaling in APAP-induced hepatocyte death, we used a neutralizing antibody against IL11RA (X209) (15). We further validated X209 as reactive and specific for mouse IL11RA by Western blot using recombinant protein from two different sources In functional studies, we found that X209 dose dependently reduced mouse hepatocyte cell death [median inhibitory concentration (IC 50 ) = 54 ng ml −1 ] and inhibited hepatocyte ERK and JNK activation (Fig. 1, I and J, and fig. S2, J and K). Although these data confirm the up-regulation of IL11 in AILI, they challenge the perception that this effect is compensatory and protective.
Species-specific effects of rhIL11 rhIL11 is reported as protective across multiple rodent models of human diseases including mouse/rat models of liver damage (tables S1 and S2), which stimulated the administration of rhIL11 to patients in the hope of therapeutic effect (table S3). Yet, our studies suggested that rhIL11 has the opposite effect on human hepatocytes in vitro (Fig. 1). This prompted us to test for potential inconsistencies when rhIL11 protein is used in a foreign species, as human and mouse IL11 share only 82% protein sequence homology.
First, we compared the effects of rhIL11 versus recombinant mouse IL11 (rmIL11) in mouse hepatocytes. The species-matched rmIL11 stimulated ERK and JNK phosphorylation and induced CASP3 cleavage, but rhIL11 had no effect ( Fig. 2A). Similarly, rmIL11 induced mouse hepatocyte cell death, whereas rhIL11 did not (Fig. 2B). In reciprocal experiments in human hepatocytes, we found that rhIL11 stimulated ERK and JNK signaling and hepatocyte death, whereas rmIL11 did not ( fig. S3, A and B).
This showed that the role of IL11 signaling in hepatocyte death is conserved across species, but that recombinant IL11 protein has species-specific effects and does not activate the same pathways in other species. We tested this hypothesis in vivo by injecting either rmIL11 or rhIL11 into mice, at doses previously used by others ( Fig. 2C) (22). Injection of rmIL11 resulted in liver damage with elevated serum concentrations of alanine transaminase (ALT) and aspartate aminotransaminase (AST) as well as ERK and JNK activation (Fig. 2, D and E, and fig. S3, C and D). In contrast, rhIL11 injection into mice had no effect on ERK or JNK phosphorylation and was associated with lower serum concentrations of ALT and AST at 24 hours (ALT, P = 0.018; AST, P = 0.0017) (Fig. 2, D and E, and fig. S3, C and D). Both rmIL11 and rhIL11 equally activated signal transducer and activator of transcription 3 (STAT3) at 30 min after injection, which represents a species-agnostic effect of recombinant IL11 when injected at high dose to the mouse (Fig. 2E).
To follow up on the published protective effect of rhIL11 in the mouse, we performed a protocol similar to a previous AILI study (22), where rhIL11 was injected into the mouse before APAP dosing (Fig. 2F). We found that rhIL11 reduced the severity of AILI in mice (reduction: ALT, 52%, P = 0.0001; AST, 39%, P < 0.0001). However, and of central importance, species-matched rmIL11 was not protective ( Fig. 2G and fig. S3E). The therapeutic effect of rhIL11 was accompanied by a reduction in hepatic ERK and JNK activation (Fig. 2H), which suggests that rhIL11 blocks endogenous mouse IL11-driven signaling pathways in the liver similar to IL11RA antibody effect in vitro (Fig. 1I).
Using surface plasmon resonance (SPR), we found that rhIL11 binds to mouse IL11 receptor  chain 1 (mIL11RA1) with a K D (dissociation constant) of 72 nM, which is similar to the rmIL11:mIL11RA1 interaction (94 nM) and close to that reported previously for rhIL11:hIL11RA (50 nM), which we reconfirmed ( Fig. 2I and fig.  S3F) (27). We then performed a competition enzyme-linked immunosorbent assay (ELISA) assay and found that rhIL11 competed with rmIL11 for binding to mIL11RA1 and was an effective blocker of rmIL11 (Fig. 2J).
In mouse hepatocytes, rhIL11 acted as a dose-dependent inhibitor of rmIL11-induced signaling pathways and cytotoxicity (Fig. 2, K and L, and fig. S3G). In addition to mouse hepatocytes, rhIL11 inhibited rmIL11-driven ERK and JNK signaling and matrix metalloproteinase 2 (MMP2) production in mouse kidney fibroblasts, heart fibroblasts, skin fibroblasts, and hepatic stellate cells ( fig. S3, H and I). Thus, rhIL11 seems to act as a neutralizer of mouse IL11 in various cells from across mouse tissues.

Hepatocyte-specific expression of Il11 causes spontaneous liver damage
To test the effects of endogenous mouse IL11 secreted from hepatocytes in vivo, we expressed an Il11 transgene in hepatocytes by injecting Rosa26 Il11/+ mice (16,17) with adeno-associated virus vector serotype 8 (AAV8) virus encoding an albumin (Alb) promoterdriven Cre construct [Il11-transgenic (Tg) mice; Fig. 3A]. Three weeks after transgene induction, Il11-Tg mice had atrophied livers (38% smaller, P < 0.0001), whereas other organs were unaffected ( fig. S4A). Il11-Tg mice also had mildly elevated serum concentrations of ALT and AST, as compared to control mice (Alb-Null) (Fig. 3, B to D, and fig. S4B). Histologically, infiltrates were seen  around the portal triad and the portal veins were nonspecifically dilated (P < 0.0001) ( fig. S4, C and D). Molecular analyses of Il11-Tg livers revealed activation of ERK, JNK, and CASP3 cleavage along with increased pro-inflammatory gene expression ( Fig. 3E and fig.  S4, E and F). These data support a maladaptive effect of speciesmatched IL11 secreted from uninjured hepatocytes but do not inform as to the role of IL11 in the context of APAP toxicity, which we examined subsequently in loss-of-function experiments.
In primary human hepatocytes, ROS dose dependently induced IL11 secretion and cell death ( fig. S5, A and B) and IL11 also stimulated ROS production, which was diminished, in part, by NAC (fig. S5, C and D). We observed an additive effect of H 2 O 2 -derived ROS and IL11 on hepatocyte death (flow cytometry, ALT) and maladaptive signaling (JNK, CASP3, and NOX4) ( fig. S5, E to H). IL11 dose dependently stimulated hepatocyte GSH depletion that mirrored ERK and JNK activation and NOX4 up-regulation (Figs. 1G and 3, H and I). As expected, only species-matched IL11 induced NOX4 up-regulation and lowered the amount of GSH ( Fig. 3J and fig. S6, A to D). APAP stimulated NOX4 and ROS up-regulation as well as GSH depletion, all of which were dependent, in part, on IL11 signaling (Fig. 3, K and L, and fig. S6E). These data link APAP toxicity in hepatocytes with IL11-stimulated, NOX4-dependent ROS production in a feed-forward manner downstream of APAP-induced mitochondrial ROS.
We reconsidered the effect of rhIL11 in inhibiting endogenous mouse IL11-induced mouse cell death and observed a dosedependent effect of rhIL11 on restoring GSH concentrations in rmIL11-stimulated mouse hepatocytes ( fig. S7A). Similarly, in vivo, rhIL11 was associated with improved GSH concentrations in APAPtreated mice, whereas rmIL11 was not ( fig. S7B). GKT-13781, a NOX1/NOX4 inhibitor, prevented IL11-stimulated GSH depletion, ERK, JNK, and CASP3 cleavage, as well as hepatocyte death (ALT), in a dose-dependent manner (Fig. 3, M to O, and fig. S8A). The specificity of inhibition of NOX4 was confirmed using small interfering RNA (siRNA) against NOX4 that prevented IL11-induced hepatotoxicity ( fig. S8, B to E), and we also showed no effect of IL11 on NOX1 expression ( fig. S8F). Together, these data show that IL11-stimulated NOX4 activity is important for GSH depletion in hepatocytes.

Hepatocyte-specific deletion of Il11ra1 prevents APAPinduced liver failure
Previous studies have shown that mice with global germline deletion of Il11ra1 are not protected from AILI (19), which we confirmed ( fig. S9, A to C). Shortcomings of germline gene deletion relating off-target effects (and/or developmental compensation) are recognized, and we used RNA sequencing (RNA-seq) to examine the expression of genes at the targeted locus in the Il11ra1 null mouse (30). This revealed that, in addition to Il11ra1, the expression of C-C Motif Chemokine Ligand (Ccl) 27a was also disrupted at the locus ( fig. S9, D and E). Given this potential confounding factor, we decided to use conditional and temporal deletion of Il11ra1 to better address the impact of Il11ra1 loss of function in mouse hepatocytes in AILI.
We created Il11ra1 conditional knockouts (CKOs) by injecting AAV8-Alb-Cre virus to mice homozygous for LoxP-flanked Il11ra1 alleles (Il11ra1 loxP/loxP ), along with wild-type controls. Three weeks after viral infection, control mice and CKOs were administered APAP (400 mg kg −1 ) (Fig. 4A). At baseline, both control and CKO groups had equivalent expression of hepatic cytochrome P450 2E1 enzyme (CYP2E1), a key enzyme responsible for the conversion of APAP to its active hepatotoxic metabolite, NAPQI (fig. S10A). One hour after APAP dosing, both CKO and control mice had equivalent plasma concentrations of APAP and a range of APAP metabolites, including NAPQI ( Fig. 4B and fig. S10B). Correspondingly, both strains had large depletion of hepatic GSH, the molecular fingerprint of NAPQI-mediated oxidative stress ( fig. S10C). Thus, Il11ra1 deletion in hepatocytes does not affect APAP metabolism or GSH depletion.
The day after APAP administration, gross anatomy revealed small and discolored livers in control mice, whereas livers from APAP-treated CKO mice looked similar to livers from mice receiving saline injection (Fig. 4C). Histology showed typical and extensive centrilobular necrosis in control mice, which was lesser in CKOs ( Fig. 4D and fig. S10D). It was striking that CKO mice had markedly lower serum concentrations of ALT and AST, as compared to controls and GSH concentrations that had largely returned to baseline (Fig. 4, E to G). ERK, JNK, and CASP3 activation was observed in control mice but not in the CKOs (Fig. 4H). Deletion of Il11ra1 in hepatocytes reduced cytokine/chemokine markers and increased F4/80 expression but had no effect on cluster of differentiation (Cd) 68 or Cd11b expression ( Fig. 4I and fig. S10E).

Mice deleted for IL11 are protected from AILI
We recently found that, although Il11ra1 KO mice have similarities with a globally deleted Il11 mouse (Il11 KO), they also have differences in some phenotypes (31) and also in expression of Ccl27a ( fig.  S9, D and E). To investigate further the effects of AILI in a second model of genetic loss of function in IL11 signaling, we subjected Il11 KO mice to AILI (Fig. 4J). We found that Il11 KO mice are protected from liver damage and that the injured livers phenocopied the signaling patterns seen in the CKO mice (Fig. 4, K to M). Thus, germline loss of function of Il11 or hepatocyte-specific deletion of Il11ra1 in the adult is protective against AILI.

Anti-IL11RA given early during AILI is beneficial
We next tested if therapeutic inhibition of IL11 signaling was effective in reducing AILI by administering either anti-IL11 (X203) or anti-IL11RA (X209) (15,17). Initially, we used a preventive strategy by injecting X203, X209, or control antibody (20 mg kg −1 ) 16 hours before APAP (Fig. 5A) and found both X203 or X209 to be protective ( fig. S11, A and B). X209 proved most effective in protecting the liver, as seen previously in NASH studies (15), and was prioritized for subsequent experiments. We quantified APAP, NAPQI, and other APAP metabolites in plasma of the immunoglobulin G (IgG)-or X209-treated mice 1 hour after APAP by mass spectrometry and found equivalent concentrations (Fig. 5B and fig. S12A). Despite normal APAP metabolism and large acute GSH depletion (fig. S12, A and B), mice receiving X209 had lower serum markers of liver damage, largely restored hepatic GSH concentrations, and lesser centrilobular necrosis by 24 hours after APAP (Fig. 5, C and D, and  fig. S13, A and B).
Next, we gave anti-IL11RA therapy in a therapeutically relevant mode at 3 hours after APAP, a time point by which APAP metabolism and toxicity is established and at which most interventions have no effect in the mouse model of AILI (Fig. 5E) (9). X209 (2.5 to 10 mg kg −1 ) inhibited all aspects of AILI with dose-dependent improvements in the degree of hepatocyte death (ALT and AST), ERK/JNK activation, GSH concentrations, and extent of centrilobular necrosis (Fig. 5, F to I, and fig. S13C).
We also determined whether inhibiting IL11 signaling had added value when given in combination with the current standard of care, NAC, 3 hours after APAP dosing (Fig. 5E). Administration of NAC alone reduced serum concentrations of ALT and AST (Fig. 5, F and G). However, NAC, in combination with X209, was even more effective than either NAC or X209 alone (ALT reduction: NAC, 38%, P = 0.0007; X209, 47%, P < 0.0001; NAC + X209, 75%, P < 0.0001). The degree of ERK and JNK inhibition with NAC or NAC together with X209 mirrored the magnitude of ALT and AST reduction in the serum and the restoration of hepatic GSH concentrations (Fig. 5, F to H and J). As such, anti-IL11RA therapy has added benefits when given in combination with the current standard of care in this mouse model.

Effects of X209 on AILI across mouse strains
There are instances in the literature where a pharmaceutical or genetic intervention has been associated with protection against AILI but has been difficult to replicate in follow-on studies (32). This may reflect the fact that liver phenotypes are susceptible to microbiota-, strain-, and sex-associated differences, with some of these factors having profound effect (33,34).
To study putative strain-specific factors in AILI, we performed an additional set of blinded experiments in female C57BL/6NTac (InVivos) mice and in male and female mice from four additional mouse strains (C57BL/6J, B6.129S1, 129X1/SvJ, and C3H/HeNTac). These studies revealed notable strain-related variation in the degree of liver injury after APAP ( fig. S14 and table S4). This said, in all experiments, we found that inhibition of IL11 signaling using X209 reproducibly reduced AILI in both male and female mice (Fig. 5,  C and K; fig. S14; and table S4).
Last, we studied C57BL6/NTac mice from a second provider to assess for within-strain variation, as genetic drift and/or differences in the microbiome or pathogen load might also influence AILI severity or the response to inhibition of IL11 signaling in AILI. As compared to the C57BL6/NTac mice used throughout this manuscript (InVivos), the degree of AILI in the additional C57BL6/NTac strain (Taconic Biosciences) was much greater even at a lower APAP dose (300 mg/kg) (~3-fold, males; ~20-fold, females) (Fig. 5C,  fig. S14, and table S4). Despite this, administration of X209, as compared to IgG, significantly reduced liver damage in both female and male mice (ALT reduction: male, 41%, P < 0.0001; female, 21%, P = 0.0127), although the magnitude of effect was diminished as compared to other strains, notably in female mice (Fig. 5, C and K;  fig. S14; and table S4).

Liver regeneration with anti-IL11RA dosing
For patients presenting to the emergency room 8 hours or later after APAP poisoning, there is no effective treatment. This prompted us to test anti-IL11RA 10 hours after APAP (400 mg kg −1 ) administration to mice (Fig. 6A). Analysis of gross anatomy, histology, and serum concentrations of IL11, ALT, and AST revealed that X209 reversed liver damage by the second day after APAP, whereas IgG-treated mice had sustained liver injury (Fig. 6, B to E, and fig.  S15, A and B). X209 effectively blocked ERK and JNK activation throughout the course of the experiment, and this preceded a reduction in Cl. CASP3 at 24 hours (Fig. 6F and fig. S15C).
Interventions promoting liver regeneration have been suggested as a new approach for treating AILI (13), and we assessed the status of genes important for liver regeneration. Inhibition of IL11 signaling was associated with a signature of regeneration with up-regulation of proliferating cell nuclear antigen (PCNA), cyclin D1/D3/E1, and phosphorylation of retinoblastoma protein (RB) (Fig. 6F), as seen during regeneration after partial hepatectomy (11).
EdU (5-ethynyl-2′-deoxyuridine) injection and histological analyses showed large numbers of nuclei with evidence of recent DNA synthesis in X209-treated mice as compared to controls (Fig. 6G  and fig. S15D). Effects of X209 administration on cytokine gene expression were variable, whereas inflammatory cell markers (Cd68, Cd11b, and F4/80) were generally increased ( fig. S15E). We reassessed the adjunctive effects of X209 and NAC given 3 hours after APAP to see whether regeneration was also associated with inhibition of IL11 signaling at earlier time points. This proved to be the case, and the combination of X209 and NAC was more effective than NAC alone, notably for cyclin D1 and D3 (Fig. 6H).
We then administered X209 (20 mg kg −1 ) 10 hours after a higher and lethal acetaminophen dose (550 mg kg −1 ) at a time point when mice are moribund and livers undergo fulminant necroinflammation (Fig. 6I). X209-treated mice recovered and had a 90% survival by the study end. In contrast, IgG-treated mice did not recover and succumbed with a 100% mortality within 48 hours (Fig. 6J). On day 8 after the lethal dose of APAP, X209-treated mice appeared healthy with normal liver morphology and serum ALT concentrations were comparable to controls that had not received APAP (Fig. 6K and  fig. S16, A and B).
Taking our data together, we propose a mechanism for APAP toxicity whereby NAPQI damage of mitochondria results in ROSrelated IL11 up-regulation, subsequent IL11-dependent NOX4 expression, and further ROS production ( fig. S17). This drives dual pathologies: killing hepatocytes via activation of downstream signaling and preventing hepatocyte regeneration, through mechanisms yet to be defined.

DISCUSSION
APAP poisoning is common, with up to 50,000 individuals attending emergency departments every year in the United Kingdom, some of whom develop liver failure requiring transplantation (1). Here, we show that IL11, previously reported as protective against APAP-induced liver failure (19,22), liver ischemia (20,23), endotoxemia (24), or inflammation (21), is a hepatotoxin and of importance for APAP-induced liver failure.
The observation that species-matched IL11 is pathogenic is surprising, as more than 30 publications have reported cytoprotective, anti-inflammatory, and/or anti-fibrotic effects of rhIL11 across a range of rodent models of human disease. Here, we show that, unexpectedly and paradoxically, rhIL11 is a competitive inhibitor of mouse IL11 binding to its cognate receptor. Furthermore, after binding to murine IL11RA1, rhIL11 does not stimulate the maladaptive signaling seen with species-matched IL11 (NOX4, JNK, and Caspase3 cleavage) but instead transiently activates STAT3. Although it could be argued that activation of STAT3 is protective in itself, we think this unlikely as rmIL11 also activates STAT3 when injected to mice but is hepatotoxic; thus, the STAT3 effect may be a bystander/nonspecific event.
The fact that rhIL11 turns out to be an inhibitor of mouse IL11 activity challenges our understanding of the role of IL11 in AILI and in disease more generally, as we found that the inhibitory effect of rhIL11 on IL11 signaling in the mouse is conserved across cells and tissues. This implies that IL11 signaling may be relevant for a range of diseases where rhIL11 has had protective effects in mouse models, which include rheumatoid arthritis (35) and colitis (36) (tables S1 and S2). We highlight that rhIL11 has been administered to patients in clinical trials for diseases where rhIL11 was found protective in mouse models of disease (table S3).
Although mice globally and germline deleted for Il11ra1 (30) are not protected from AILI (19), which we confirmed, we believe that this may, in part, be explained by off-target effects at the Il11ra1 locus, which we documented. Furthermore, there are unique features seen in the Il11ra1 KO mouse that are not apparent in Il11 null mice (31). Our studies advise caution against using a single genetic model on one genetic background for the study of AILI, and we suggest that loss-and gain-of-function approaches across genetic models are preferred, especially if complemented by specific pharmacologic interventions.
It is apparent from the published literature that the effect of APAP on liver damage in mice can vary across strains, which reflects influences of genetic background, microbiome, and pathogen load (32)(33)(34). We documented notable variability in the severity of AILI across strains and observed surprisingly large within-strain differences in AILI in genetically identical mouse strains from two different sources. The magnitude of effect of anti-IL11RA dosing also varied across and within strains, with some strains showing far greater reductions in ALT concentrations as compared to others (ALT reduction in female mice: C57BL6/NTac, 21%; C3H/HeNTac, 93%). Thus, the power to detect an effect associated with inhibition of IL11 in AILI is dissimilar between mouse strains, and this is an important experimental consideration.
Our study stimulates questions and has a number of limitations. We show that ERK is co-regulated with JNK in APAP injured livers, yet ERK's role in AILI is not well characterized. The role of apoptosis in AILI is contentious, and although IL11 stimulates caspase cleavage in hepatocytes, the functional relevance of this is not clear and studies of IL11 in hepatocyte lipotoxicity suggest that IL11 is important for more than one form of cell death (26). The mechanism by which anti-IL11 administration stimulates liver regeneration and the nature of the replicating cells remain unknown. The effect of IL11 on cell types other than hepatocytes in AILI was not dissected, and we did not address the role of IL11 on the immune response, which is an important issue that requires further study. Whether variation in the microbiome and/or pathogen load affects the IL11 axis in AILI, which is inferred from our within-strain studies if genetic drift is excluded, appears profound but has not been studied here. Although we believe that it is unlikely that rhIL11-stimulated STAT3 activity plays a role in the protective effects of rhIL11 in AILI, which we suggest instead reflects competitive inhibition of mouse IL11 binding to IL11RA1, this was not formally excluded. Why the anti-IL11RA approach was more effective in reducing liver damage as compared to anti-IL11 dosing was not determined but is consistent with other studies of liver disease in the mouse (15,37).
We point out that although the toxic effects of IL11 appear conserved in mouse and human hepatocytes, we did not study human biospecimens from patients with AILI. Measuring IL11 expression in humans is difficult, as the concentration of IL11 in the serum is very low and liver biopsy is not part of routine clinical care in patients with AILI. Whether or not inhibition of IL11 signaling is beneficial in patients with AILI can only be tested formally in randomized and blinded clinical trials, which might be envisaged now that therapeutic molecules are being developed.
We end by noting that because IL11 neutralizing therapies are not dependent on altering APAP metabolism and stimulate liver regeneration, they could be useful for patients presenting late with AILI, in addition to having added value when given together with NAC. Overall, our studies question the premise that IL11 is protective in the liver and suggest instead that species-matched IL11 is a hepatotoxin.

Study design
In this study, we used primary human and mouse hepatocytes and in vivo mouse experiments to investigate the effects of IL11 in hepatocytes and of inhibiting IL11 signaling (genetically and therapeutically) in a mouse model of AILI. Animal procedures were performed according to the protocols approved by the SingHealth Institutional Animal Care and Use Committee (IACUC). IL11 RNA and protein expression was examined in primary human hepatocyte supernatants and in liver and serum from the mouse model of AILI by quantitative polymerase chain reaction (qPCR), Western blotting, and ELISA. Further confirmation of IL11 protein expression in liver was assessed by bioluminescent imaging and immunofluorescence analysis of liver tissue from Il11-luciferase and Il11-EGFP reporter mouse, respectively. The effects of species-specific IL11 on hepatocyte injury and death were examined by gain-of-function approaches in vitro in primary human and mouse hepatocyte cultures and in vivo by systemic administration of rhIL11/rmIL11 and in hepatocyte-specific Il11-expressing Tg mice (AAV8-Alb-Cre-Rosa26 Il11/+ ; Il11-Tg). Binding of human/mouse IL11 to mouse IL11RA was assessed by competitive ELISA, and their binding affinities were determined by SPR. Loss-of-function experiments were performed in Il11 and Il11ra1 global KO mice and in hepatocytespecific Il11ra1-deleted mice to investigate the effects of genetic inhibition of Il11 or Il11ra1 in AILI. Binding of neutralizing antibody against IL11RA (X209) to mouse IL11RA was validated by immunofluorescence analysis and Western blotting of primary mouse hepatocytes and mouse IL11RA protein from two different commercial sources. X209 specificity to mouse IL11RA was evaluated by immunohistochemistry of liver tissue isolated from wild-type and l11ra1 KO mice. We used X209 in primary hepatocytes experiments and in pharmacologic prevention and reversal studies in the AILI mouse model. Colorimetric assays (ALT, AST, GSH measurement) and Western blot analysis were performed on hepatocytes or hepatocyte supernatants as well as liver/serum samples from mice subjected to AILI. Plasma concentrations of APAP and APAP metabolites were assessed by liquid chromatography-tandem mass spectrometry. Sample size for cell-based assays was determined on the basis of sample availability and technical needs. The in vivo experiments were designed to detect differences between treatment groups or genotype-dependent effects at 80% power (ɑ = 0.05), but sample sizes may vary depending on animal availability. Outlier tests were performed using the ROUT method (GraphPad Prism). Sample sizes are detailed in figure legends. Mice were randomly assigned to experimental groups on the day of the treatment, except for KO mice in which randomization was assigned within the same genotypes. For in vitro experiments, investigators were not blinded to group allocation during data collection and analysis. For in vivo experiments, investigators were not blinded other than for the gender and strain studies, which were performed double blind, for the large part. Histological analysis of liver tissue samples was performed blinded to treatments and genotypes. Further details are described in Supplementary Materials and Methods.

Statistical analysis
Statistical analyses were performed using GraphPad Prism software (version 8). Datasets were tested for normality with Shapiro-Wilk tests. For normally distributed data, statistical significance between control and experimental groups was analyzed by two-sided Student's t tests or by one-way analysis of variance (ANOVA) as indicated in the figure legends. P values were corrected for multiple testing according to Dunnett's (when several experimental groups were compared to a single control group) or Tukey (when several conditions were compared to each other within one experiment). Nonparametric tests (Kruskal-Wallis with Dunn's correction in place of ANOVA and Mann-Whitney U test in place of two-tailed t test) were conducted for nonnormally distributed data. Comparison analysis for two parameters from two different groups was performed by twoway ANOVA and corrected with Sidak's multiple comparisons when the means were compared to each other. Survival curves were analyzed by Gehan-Breslow-Wilcoxon test. The criterion for statistical significance was P < 0.05.