Thymic stromal lymphopoietin and interleukin-4 mediate the pathogenesis of halothane-induced liver injury in mice

Authors

  • William R. Proctor,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Mala Chakraborty,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Aaron M. Fullerton,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Midhun C. Korrapati,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Pauline M. Ryan,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Kenrick Semple,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Jeffrey C. Morrison,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Julia D. Berkson,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Lynette S. Chea,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Qian Yang,

    1. In Vitro ADMET Laboratories, Columbia, MD
    Search for more papers by this author
  • Albert P. Li,

    1. In Vitro ADMET Laboratories, Columbia, MD
    Search for more papers by this author
  • Rosanne Spolski,

    1. Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Erin E. West,

    1. Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Yrina Rochman,

    1. Division of Immunobiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH
    Search for more papers by this author
  • Warren J. Leonard,

    1. Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Mohammed Bourdi,

    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Lance R. Pohl

    Corresponding author
    1. Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD
    • Address reprint requests to: Lance R. Pohl, Pharm.D., Ph.D., Molecular and Cellular Toxicology Section, Laboratory of Molecular Immunology, National Heart, Lung and Blood Institute, National Institutes of Health, Building 10, Room 8N110, 10 Center Drive, Bethesda, MD 20892-1760. E-mail: pohll@nih.gov; fax: 301-480-4852.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 1473

  • This work was supported by the Intramural Research Program of the National Institutes of Health and the National Heart, Lung and Blood Institute.

Abstract

Liver eosinophilia has been associated with incidences of drug-induced liver injury (DILI) for more than 50 years, although its role in this disease has remained largely unknown. In this regard, it was recently shown that eosinophils played a pathogenic role in a mouse model of halothane-induced liver injury (HILI). However, the signaling events that drove hepatic expression of eosinophil-associated chemokines, eotaxins, eosinophil infiltration, and subsequent HILI were unclear. We now provide evidence implicating hepatic epithelial-derived cytokine thymic stromal lymphopoietin (TSLP) and type 2 immunity, in particular, interleukin-4 (IL-4) production, in mediating hepatic eosinophilia and injury during HILI. TSLP was constitutively expressed by mouse hepatocytes and increased during HILI. Moreover, the severity of HILI was reduced in mice deficient in either the TSLP receptor (TSLPR) or IL-4 and was accompanied by decreases in serum levels of eotaxins and hepatic eosinophilia. Similarly, concanavalin A–induced liver injury, where type 2 cytokines and eosinophils play a significant role in its pathogenesis, was also reduced in TSLPR-deficient mice. Studies in vitro revealed that mouse and human hepatocytes produce TSLP and eotaxins in response to treatment with combinations of IL-4 and proinflammatory cytokines IL-1β and tumor necrosis factor alpha. Conclusion: This report provides the first evidence implicating roles for hepatic TSLP signaling, type 2 immunity, and eosinophilia in mediating liver injury caused by a drug. (Hepatology 2014;60:1741-1752)

Abbreviations
ALT

alanine aminotransferase

BSA

bovine serum albumin

CCL

chemokine (C-C motif) ligand

ConA

concanavalin A

DILI

drug-induced liver injury

ELISA

enzyme-linked immunosorbent assay

FCM

flow cytometry

H&E

hematoxylin and eosin

HILI

halothane-induced liver injury

IL

interleukin

IL7Rα

interleukin-7 receptor alpha

IP

intraperitoneally

IV

intravenously

mRNA

messenger RNA

NKT

natural killer T cell

PBS

phosphate-buffered saline

PMHs

primary mouse hepatocytes

rIL-4

recombinant Interleukin-4

SEM

standard error of the mean

TFA

trifluoroacetylated

Th

T helper

TNF-α

tumor necrosis factor alpha

TSLP

thymic stromal lymphopoietin

TSLPR

thymic stromal lymphopoietin receptor

WT

wild type

Drug-induced liver injury (DILI) is a major health concern because it can lead to significant patient morbidity and mortality[1] and it is difficult to predict which new drugs will cause injury and who will be susceptible to this disease. Consequently, there is a need to better understand the mechanisms that lead to, or protect against, DILI and identify clinically relevant biomarkers of DILI. Eosinophilia, whether peripheral or hepatic, has been widely associated with instances of DILI,[2, 3] although its role in DILI remains largely unknown. In this regard, we recently reported that eosinophils can play a pathogenic role in a mouse model of halothane-induced liver injury (HILI).[4] We found that eosinophils infiltrated the liver during early stages of injury, accumulated exclusively around areas of necrosis, and increased in number proportionally to the degree of injury.[4] Severity of HILI was significantly reduced when eosinophils, but not neutrophils, were selectively removed with antibodies or when eosinophils were deleted genetically.[4] Moreover, eotaxins appeared to attract eosinophils to the liver after halothane treatment.[4] However, the signaling events that drove eotaxin expression, eosinophil infiltration, and subsequent HILI were not elucidated in that study.

Type 2 immune responses, traditionally referred to as T helper cell type 2 responses, are characterized by their dependence upon effector cytokines, including interleukin (IL)−4, IL-5, and IL-13 and are known to drive eosinophil infiltration during allergic diseases of the skin and lungs.[5] A main regulator of type 2 immunity is thymic stromal lymphopoietin (TSLP), a cytokine primarily derived from epithelial cells that exerts its effect through a heterodimeric receptor complex comprised of the TSLP receptor (TSLPR) and IL-7 receptor alpha (IL-7Rα).[6] Because it was known that TSLP signaling plays a pathogenic role in mouse models of allergic inflammation in lung[7, 8] and skin[9] that was mediated, in part, by type 2 cytokines, including IL-4, and eosinophils, we decided to determine whether similar signaling could play a role in the mouse model of HILI. In this report, we provide evidence for this hypothesis by showing that mice deficient in TSLPR or IL-4 were resistant to HILI and that IL-4 can play a role in activating hepatocytes to secrete both TSLP and eotaxins. Similarly, it appears that TSLP signaling may play a pathogenic role in other diseases of the liver because TSLPR-deficient mice were also resistant to concanavalin A (ConA)-induced liver injury, which is an established model of natural killer T (NKT) cell-mediated hepatitis, where type 2 responses, in particular, IL-4, and eosinophils play a prominent role in mediating toxicity.[10, 11]

Materials and Methods

Animals and Treatments

Wild-type (WT) Balb/cJ (000651) and IL4−/− (002496) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Tslpr−/− mice were developed as described previously[12] and back-crossed >10 generations to the Balb/c background. All animals employed in this study were 7- to 10-week-old females, with weights ranging between 18 and 22 g. Animals were acclimated for at least 6-7 days to a 12-hour light/dark cycle in a humidity- and temperature-controlled, specific-pathogen–free environment in microisolator autoclaved cages. Mice were allowed autoclaved food and water ad libitum. All maintenance of animals conformed to the guidelines for humane treatment set by the Association for Assessment and Accreditation for Laboratory Animal Care International's Guide for the Care and Use of Laboratory Animals and by the National Institutes of Health (Bethesda, MD). Animals were injected intraperitoneally (IP) with distilled halothane (30 mmol/kg; Sigma-Aldrich, St. Louis, MO) dissolved in olive oil (Mild Olive Flavor Originale; Star Fine Foods, Fresno, CA) at a final concentration of 0.30 mmol/mL or vehicle only. In separate studies, mice were injected intravenously (IV) through the lateral tail vein with 10 mg/kg of ConA (Type V, 1.25 mg/mL dissolved in sterile phosphate-buffered saline [PBS]; Sigma-Aldrich). In other studies, 2 ng of recombinant mouse IL-4 (rIL-4; BD Biosciences, San Jose, CA) was administered IV (0.1 mL) through the lateral tail vein of WT and IL4−/− mice 12 hours after halothane treatment. rIL-4 was dissolved in sterile PBS (pH 7.4), containing 1% bovine serum albumin (BSA) Cohn Fraction V (Sigma-Aldrich) for protein stabilization.

Sera and Tissue Collection, Assessment of Liver Injury, and Detection of Trifluoroacetylated-Protein Adducts in Liver Homogenates of Mice

Mouse sera and liver tissues were collected and used for assessing liver injury by measurement of alanine aminotransferase (ALT) activities and histopathological analysis of hematoxylin and eosin (H&E)-stained liver sections, as reported previously.[4] In addition, liver homogenates from mice were prepared and sodium dodecyl sulfate polyacrylamide gel electrophoresis immunoblotting analysis of trifluoroacetylated (TFA)-protein adducts and β-tubulin was performed, as reported previously.[4]

Isolation of Hepatic Leukocytes and Flow Cytometric Analysis of Hepatic Eosinophils

Mouse hepatic leukocytes were isolated and stained to quantify eosinophils by flow cytometry (FCM), as reported previously.[4] Briefly, cells gated by FCM as CD11c CD11b+ Gr-1low Siglec-Fhigh were quantified as eosinophils. The absolute number of eosinophils was calculated by multiplying their percentage by the total number of viable hepatic leukocytes per liver.

Isolation and Treatment of Primary Mouse Hepatocytes and Hepa 1-6 Cells With Cytokines

Cultures of primary female Balb/cJ mouse hepatocytes and Hepa 1-6 cells (CRL-2026; ATCC, Manassas, VA) were used in this study (see the Supporting Information for detailed methods).

Treatment of Cultured Human Primary Hepatocytes With Cytokines

Plateable cryopreserved human hepatocytes from 3 different donors (donors: HH1020, HH1026, and HH1031; In Vitro ADMET Laboratories, Columbia, MD) were used in this study (see the Supporting Information for detailed methods).

RNA Isolation and Reverse-Transcriptase Polymerase Chain Reaction Analysis

Total RNA was isolated from liver sections, freshly isolated hepatocytes, infiltrating hepatic leukocytes, and cultured cells, as reported previously.[4] Relative gene expression was determined by quantitative reverse-transcriptase polymerase chain reaction analysis using validated gene-specific assays (Applied Biosystems, Carlsbad, CA) for mouse IL1β, IL4, IL7Rα, tumor necrosis factor alpha (TNF-α), cytokine receptor-like factor 2 (referred to throughout the article as TSLPR) and mouse and human chemokine (C-C motif) ligand (CCL)11, CCL24, CCL26 (human only isoform), TSLP, and β-actin.

Measurement of Mouse and Human Cytokines and Chemokines in Cell Culture Supernatants and Sera From Mice

Mouse and human protein levels in cell-culture supernatants or sera from mice of CCL11, CCL24, CCL26, and TSLP were quantified using their respective DuoSet enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Minneapolis, MN), following the manufacturer's protocols. Serum levels of mouse IL-4, IL-1β, and TNF-α were quantified using their respective ELISA Ready-Set-Go kits (eBioscience, San Diego, CA), following the manufacturer's protocols.

Statistical Analysis

All data presented are reported as means ± standard error of the mean (SEM). Statistical significance between two groups was determined by the two-tailed Student t test, whereas statistical differences between multiple groups were determined by one-way analysis of variance with Newman-Keuls' posttest analysis. Differences were considered significant when P values were <0.05.

Results

TSLP, TSLPR, and IL-4 are Induced in Mouse Liver During HILI

TSLP and TSLPR expression was assessed in a mouse model of HILI,[13] where eosinophils play a pathogenic role.[4] As reported, treatment of female Balb/cJ mice with halothane increased serum ALT activities in a time-dependent manner, beginning at 12 hours posttreatment (Fig. 1A). In accord with the serum ALT activities, histological evaluation of liver sections from halothane-treated animals revealed centrilobular necrotic lesions that peaked at 24 hours posttreatment (Fig. 1B). TSLP messenger RNA (mRNA) increased in liver homogenates from mice treated with halothane, relative to liver homogenates from vehicle-treated control mice, at 12, 18, and 24 hours posttreatment (Fig. 1C). This was similar to the observed gene expression changes of eotaxin-1 (CCL11) in liver homogenates in response to halothane treatment.[4] Hepatocytes appeared to be the source of TSLP, because TSLP mRNA was enriched in hepatocytes, relative to liver homogenates and hepatic leukocytes isolated from naïve Balb/cJ mice (see Supporting Fig. S1A). Serum levels of TSLP were also detected 18 and 24 hours posttreatment only from mice treated with halothane (Fig. 1D). In addition, TSLPR and IL7Rα mRNA increased in liver homogenates from mice treated with halothane (Fig. 1C). We also found that IL-4 mRNA increased in liver homogenates in response to HILI at 12 hours posttreatment and then declined (Fig. 1E), whereas IL-4 protein was detected in serum only from mice treated with halothane at 12, 18, and 24 hours posttreatment (Fig. 1F). This finding may result, in part, from the fact that expression levels of mRNA and protein do not always correlate and that a significant amount of serum IL-4 may have been derived from one or more sites other than the liver.

Figure 1.

Hepatic TSLP and IL-4 are induced during HILI in mice. (A) Female Balb/cJ mice were injected IP with halothane (30 mmol/kg) or vehicle (olive oil), and serum ALT activities were determined 6, 12, 18, and 24 hours posttreatment in mice treated with vehicle or halothane. Four independent experiments were performed with N = 19-20 per group. *P < 0.05 versus vehicle-treated control mice at the same time point; #P < 0.05 versus the previous time point within the same treatment. (B) Representative photomicrographs of H&E-stained liver sections from mice showing necrotic lesions in mice treated with halothane for 24 hours. Scale bars = 100 µm. (C) Levels of TSLP, TSLPR, and IL7Rα mRNA from liver homogenates at various time points after vehicle or halothane treatment with N = 10-15 mice per group from three independent experiments. (D) Serum protein levels of TSLP in mice treated with vehicle or halothane with N = 6 per group. Shown are means ± SEM. (E) Levels of IL-4 mRNA from liver homogenates at various time points after vehicle or halothane treatment with N = 10-15 mice per group from three independent experiments. (F) Serum protein levels of IL-4 in mice treated with vehicle or halothane with N = 6 per group. All data are shown as means ± SEM: *P < 0.05 versus vehicle-treated control mice at the same time point. ND, not detected.

Halothane-Induced Liver Injury is Attenuated in Mice Deficient in TSLP Signaling

To determine whether TSLP/TSLPR signaling plays a role in HILI, we assessed HILI in Tslpr−/− mice. This mouse model has been used extensively to study the role of TSLP and TSLPR signaling in leukocyte development and differentiation[12] as well as in the pathogenesis and progression of allergic lung[7, 8] and skin inflammation.[9] Severity of HILI was significantly reduced at 12, 18, and 24 hours posttreatment in Tslpr−/− mice, relative to WT control animals (Fig. 2A). The decrease in serum ALT activity 24 hours posttreatment with halothane corresponded with a significant reduction in size of necrotic lesions in livers of Tslpr−/− mice, relative to WT controls (Fig. 2B). Similar results were obtained when littermate Tslpr+/+ controls were used in the place of WT mice (data not shown). TFA-protein adducts, which appear to initiate HILI,[14] are formed when liver proteins are covalently labeled by the trifluoroacetyl chloride metabolite of halothane. Consequently, we verified that the decrease in severity of HILI was not the result of a decrease in TFA-protein adduct formation, because there was no substantial difference in TFA-protein adducts detected in liver homogenates from WT and Tslpr−/− mice 8 hours after halothane treatment (Fig. 2C). Eight hours was a time point at which there was no noticeable liver damage (Fig. 2A), where TFA-protein adducts could be leaked from necrotic hepatocytes.

Figure 2.

Severity of HILI was significantly attenuated in mice deficient in TSLPR. (A) Serum ALT activities at 8, 12, 18, and 24 hours posthalothane treatment in WT and Tslpr−/− mice with N = 5 for 8 hours and N = 14-15 for 12, 18, and 24 hours. (B) Representative photomicrographs of H&E-stained liver sections 24 hours after halothane treatment in WT and Tslpr−/− mice (scale bar = 50 μm). (C) Immunochemical detection of TFA-protein adducts and β-tubulin in homogenates from individual livers excised from WT and Tslpr−/− mice 8 hours post–halothane treatment (N = 4 per group). Molecular markers (kDa) are indicated on the left and the relative abundance of TFA-protein adducts per group, as measured by densitometry, below. (D) Representative FCM density dot plots for hepatic eosinophils (CD11c CD11b+ Gr-1low Siglec-Fhigh) from WT mice at 24 hours posttreatment with halothane with a summary of the total number of hepatic eosinophils per liver from WT and Tslpr−/− mice 12, 18, and 24 hours post–halothane treatment with N = 9-10 per group from two independent experiments. (E) Levels of IL4, CCL11, and CCL24 mRNA from liver homogenates from WT and Tslpr−/− mice sacrificed 12 hours post–halothane treatment with N = 9-10 per group from two independent experiments. *P < 0.05 versus WT mice. (F) Serum protein levels of IL-4, CCL11, and CCL24 in WT and Tslpr−/− mice 12 and 24 hours post–halothane treatment with N = 9-10 per group from two independent experiments. *P < 0.05 versus WT mice and #P < 0.05 versus the previous time point within the group. All data are reported as means ± SEM.

Next, we examined the levels of hepatic eosinophils in Tslpr−/− and WT mice during HILI. Hepatic eosinophils were quantified by FCM as viable CD11c, CD11b+, Gr-1low, and Siglec-Fhigh cells[4] (representative gating scheme is depicted in Fig. 2D). Similar to serum ALT activities, the number of hepatic eosinophils was significantly lower at 12, 18, and 24 hours after halothane treatment in Tslpr−/− mice, relative to WT mice (Fig. 2D), consistent with eosinophils contributing to HILI[4] and being regulated by TSLP.[8, 9] There was no difference in the numbers of hepatic eosinophils in vehicle-treated Tslpr−/− and WT mice at any time point (data not shown).

To determine whether the decrease in severity of HILI and numbers of hepatic eosinophils observed in Tslpr−/−, relative to WT, mice may result, in part, from decreases in the type 2 effector cytokine, IL-4, as well as eotaxins CCL11 and CCL24, we examined hepatic expression of these factors in Tslpr−/− mice in relation to WT controls 12 hours post treatment with halothane. This time point was selected because it is the first time at which liver injury is observed (Fig. 1A) and also the point at which the greatest hepatic induction of eotaxins[4] and IL-4 (Fig. 1E) occurs. Liver mRNA levels of IL4, CCL24, and, particularly, CCL11 were reduced in Tslpr−/− mice 12 hours post–halothane treatment, relative to WT controls (Fig. 2E), as were serum protein levels of CCL11 and CCL24 at 12 and 24 hours and IL-4 at 24 hours posttreatment with halothane of Tslpr−/− mice, relative to WT animals (Fig. 2F).

Halothane-Induced Liver Injury is Attenuated in Mice Deficient in IL-4

Because IL-4 is often associated with TSLP signaling[9, 15] and can stimulate eotaxin expression and resultant eosinophil infiltration in mouse liver,[11] we investigated whether IL-4 may be playing a pathogenic role during HILI. Similar to Tslpr−/− mice, severity of HILI was reduced in IL4−/− mice, relative to WT animals (Fig. 3A,B) and was not the result of diminished TFA-protein adduct formation (Fig. 3C). Moreover, hepatic eosinophils in IL4−/− mice were reduced at 18 and 24 hours post–halothane treatment, relative to WT control mice (Fig. 3D), whereas levels of hepatic eosinophils remained unchanged between the groups when animals were treated with vehicle only (data not shown). As expected, hepatic IL-4 mRNA was undetected in IL4−/− mice (Fig. 3E), whereas liver mRNA expression (Fig. 3E) and serum levels (Fig. 3F) of CCL11 and CCL24 12 hours post–halothane treatment did not differ between IL4−/− and WT mice. However, there were decreases in serum levels of CCL11 and CCL24 between IL4−/− mice and WT controls 24 hours posttreatment with halothane (Fig. 3F).

Figure 3.

Severity of liver injury, numbers of hepatic eosinophils, and eotaxins were significantly attenuated in IL4−/− mice during HILI. (A) Serum ALT activities at 8, 12, 18, and 24 hours post–halothane treatment in WT and IL4−/− mice with N = 5 for 8 hours and N = 12-13 for 12, 18, and 24 hours. (B) Representative photomicrographs of H&E-stained liver sections 24 hours after halothane treatment in WT and IL4−/− mice (scale bar = 50 μm). (C) Immunochemical detection of TFA-protein adducts and β-tubulin in homogenates from individual livers excised from WT and IL4−/− mice 8 hours post–halothane treatment (N = 4 per group). Molecular markers (kDa) are indicated on the left and relative abundance of TFA-protein adducts per group, as measured by densitometry, below. (D) Summary of total number of hepatic eosinophils (CD11c CD11b+ Gr-1low Siglec-Fhigh) from WT and IL4−/− mice at 12, 18, and 24 hours post–halothane treatment with N = 9-10 per group from two independent experiments. (E) Levels of IL4, CCL11, and CCL24 mRNA from liver homogenates from mice sacrificed 12 hours post–halothane treatment with N = 9-10 per group from two independent experiments. ND, not detected. (F) Serum protein levels of CCL11 and CCL24 in WT and IL4−/− mice 12 and 24 hours post–halothane treatment with an N = 9-10 per group from two independent experiments. *P < 0.05 versus WT mice at the same time point; #P < 0.05 versus previous time point within the same group. All data are reported as means ± SEM.

To confirm that IL-4 has a pathologic role in HILI, we attempted to enhance the susceptibility of IL4−/− mice to HILI by injecting them with rIL-4 12 hours post–halothane treatment, a time point at which hepatic expression and serum levels of IL-4 rise in response to HILI (Fig. 1E and F, respectively). rIL-4 increased the severity of HILI (Fig. 4A,B) and numbers of hepatic eosinophils (Fig. 4C) to levels greater than those observed in IL4−/− mice treated with PBS. Injection of rIL-4 failed to cause liver injury or increase the number of hepatic eosinophils in the absence of halothane treatment (data not shown).

Figure 4.

Recombinant IL-4 increased the susceptibility of IL4−/− mice to HILI. IL4−/− and WT mice were injected with 2.0 ng of rIL-4 12 hours post–halothane treatment or vehicle only (PBS) and sacrificed 12 hours later. (A) Serum ALT activities obtained 24 hours after halothane treatment from IL4−/− and WT mice treated with rIL-4 or PBS. (B) Representative photomicrographs of H&E-stained liver sections 24 hours after halothane treatment in WT and IL4−/− mice treated with rIL-4 or PBS (scale bar = 50 μm). (C) Summary of total number of hepatic eosinophils (CD11c CD11b+ Gr-1low Siglec-Fhigh) from PBS- or rIL-4-treated WT and IL4−/− mice at 24 hours post–halothane treatment. All data are reported as means ± SEM with N = 13-15 from three independent experiments. *P < 0.05 versus IL4−/− PBS-treated mice; #P < 0.05 versus WT PBS-treated mice.

Eotaxins and TSLP Are Secreted by IL-4 and IL-1β or TNF-α Treatment Respectively From Mouse Liver Cells and Human Hepatocytes

One way IL-4 may promote HILI is by inducing hepatic secretion of eotaxins, as suggested by the results in a previous study where IL-4 was found to induce the expression of eotaxin genes in primary hepatocyte cultures from Balb/c mice.[11] We have now extended this finding by showing that IL-4 can induce protein secretion of eotaxins in primary mouse hepatocyte (PMH) cultures. The mRNA expression (Fig. 5A) and protein secretion (Fig. 5B) of CCL11 and CCL24 increased in mouse hepatocytes after treatment with 10 ng/mL of IL-4 for 24 hours, which was the minimum level of IL-4 at which maximal induction of eotaxins was observed (data not shown).

Figure 5.

Eotaxins and TSLP are secreted by IL-4 and IL-1β or TNF-α treatment, respectively, from mouse liver cells in vitro. Mouse hepatocytes were isolated from naïve WT mice and cultured for 48 hours. Gene expression (A) and protein levels in culture supernatant (B) of CCL11 and CCL24 were determined at 24 hours posttreatment with recombinant mouse IL-4 (10 ng/mL) or vehicle (N = 4). *P < 0.05 versus untreated control. (C) Levels of IL-1β and TNF-α mRNA from liver homogenates of mice treated with vehicle or halothane (N = 10-15 mice per group from three independent experiments). *P < 0.05 versus vehicle animals at the same time point. (D) Serum protein levels of IL-1β and TNF-α in mice treated with vehicle or halothane with N = 6 per group. ND not detected. (E) Expression of TSLP mRNA (N = 3 per group) and protein levels (N = 6 per group) in cell-culture supernatant were assessed from mouse Hepa 1-6 cells treated individually or in combination for 24 hours with IL-1β (1 ng/mL), TNF-α (10 ng/mL), and IL-4 (10 ng/mL). *P < 0.05 versus same IL-4 treatment in the absence of IL-1β and TNF-α; #P < 0.05 versus same IL-1β and TNF-α treatment in the absence of IL-4. (F) Expression of TSLP protein levels in cell-culture supernatant were assessed from PMHs treated individually or in combination for 24 hours with IL-1β (1 ng/mL), TNF-α (10 ng/mL), and IL-4 (10 ng/mL). *P < 0.05 versus same IL-4 treatment in the absence of IL-1β and TNF-α; #P < 0.05 versus same IL-1β and TNF-α treatment in the absence of IL-4. All data are reported as means ± SEM with an N = 6 comprised of N = 3 wells from hepatocytes isolated from 2 individual mice.

It is known that early inflammatory cytokines IL-1β and TNF-α can induce TSLP secretion from cultured epithelial cells[16-18] and that IL-4 can augment the activities of IL-1β or TNF-α in inducing TSLP secretion in vitro.[16, 18] Based on these findings and our discovery that gene expression of IL-1β and TNF-α were elevated in liver homogenates as early as 12 hours after halothane treatment (Fig. 5C), as were serum levels relative to vehicle-treated mice (Fig. 5D), it seemed plausible that IL-1β, TNF-α, and IL-4 could have similar effects on inducing hepatic TSLP during HILI. We tested this idea initially in a mouse Hepa 1-6 cell line. The concentrations of recombinant IL-1β (1 ng/mL) and TNF-α (10 ng/mL) were selected based on previous reports.[17, 18] IL-1β and TNF-α had little effect on TSLP mRNA and protein secretion when treated individually, but when combined together induced TSLP in Hepa 1-6 cells (Fig. 5E). IL-4 also had no effect alone on TSLP mRNA and protein secretion from Hepa 1-6 cells, but synergistically induced TSLP levels over 25-fold in relation to untreated wells in the presence of IL-1β or TNF-α and 70-fold in the presence of both IL-1β and TNF-α (Fig. 5E). In contrast to Hepa 1-6 cells, treatment of PMHs with either IL-1β or TNF-α alone or in combination did not induce TSLP protein secretion (Fig. 5F). However, similar to the results observed in Hepa 1-6 cells, IL-4 treatment alone did not significantly increase TSLP protein secretion, but, when combined with either IL-1β or TNF-α and in the presence of both IL-1β and TNF-α, increased TSLP levels, relative to the same cytokine treatments, in the absence of IL-4.

We also examined the effects of recombinant human IL-1β, TNF-α, and IL-4 treatment on the production of TSLP and eotaxins from cultured human hepatocytes from 3 separate donors. Unlike mice, the human TSLP gene encodes for two splice variant transcripts. The long transcript represents the inducible variant that encodes the fully functional TSLP protein,[17, 19] whereas the function of the truncated protein remains unknown. Treatment of human hepatocytes with IL-1β or TNF-α induced TSLP mRNA 3-fold, whereas a 10-fold increase was observed when cells were treated with both IL-1β and TNF-α, relative to untreated controls (Fig. 6A). Protein levels of TSLP in cell-culture supernatants were also induced by IL-1β and TNF-α (Fig. 6B) analogous to the influence these cytokines had on TSLP mRNA. In contrast, IL-4 did not appear to augment the ability of IL-1β or TNF-α to induce TSLP mRNA production and protein secretion from human hepatocytes (Fig. 6A,B). Closer examination revealed that, similar to mouse liver cells (Fig. 5E,F), IL-4 treatment augmented the activities of IL-1β and TNF-α in inducing TSLP mRNA and secretion in hepatocytes from donor HH1026 (see Supporting Fig. S2).

Figure 6.

IL-4, IL-1β, and TNF-α stimulated eotaxin and TSLP expression and secretion from primary cultured human hepatocytes. Plateable cryopreserved hepatocytes from 3 donors (HH1020, HH1026, and HH1031) were cultured for 2 days. On day 3, cells were treated for 24 hours individually or in combination with recombinant human IL-1β (1 ng/mL), TNF-α (10 ng/mL), and IL-4 (10 ng/mL). Expression of TSLP, CCL11, and CCL26 mRNA in cells (A) and protein levels of TSLP, CCL11, and CCL26 from cell-culture supernatant (B) were assessed 24 hours post-treatment. All data are reported as means ± SEM with an N = 9 comprised of N = 3 wells from hepatocytes isolated from 3 individual donors. *P < 0.05 versus same IL-4 treatment in the absence of IL-1β and TNF-α; #P < 0.05 versus same IL-1β and TNF-α treatment in the absence of IL-4; †P < 0.05 versus the two bracketed treatment groups.

IL-4 treatment of human hepatocytes also induced mRNA of CCL11 by 50-fold over the untreated control, with a corresponding 10-fold increase in protein levels in cell-culture supernatant (Fig. 6A,B). The effect of IL-4 on CCL11 protein secretion was enhanced by the presence of TNF-α, but was ablated when IL-1β was added to TNF-α and IL-4. IL-4 treatment significantly induced CCL26 mRNA expression (upward of 10,000-fold) and protein secretion (upward of 100-fold) in human hepatocytes, which was enhanced by IL-1β alone and TNF-α when in combination with IL-1β (Fig. 6A,B). Unlike CCL11 and CCL26, CCL24 mRNA expression increased only 2-fold in response to IL-4 treatment (data not shown).

The Severity of ConA-Mediated Hepatitis is Significantly Attenuated in Mice Deficient in TSLPR

To determine whether TSLP signaling plays a pathogenic role in other diseases of the liver, we evaluated the susceptibility of Tslpr−/− mice to ConA-mediated hepatitis, an established mouse model in which IL-4 and eosinophils play a critical role in its hepatotoxicity.[10, 11] The severity of ConA-mediated hepatitis was decreased even more in Tslpr−/− mice, relative to WT mice (Fig. 7A,B) than that found in studies of HILI (Fig. 2A,B).

Figure 7.

Mice deficient in TSLPR are protected against ConA-mediated liver injury. Female WT or Tslpr−/− mice were administered ConA (10 mg/kg) IV and sacrificed 8 hours posttreatment Serum ALT activities (A) and representative photomicrographs of H&E-stained liver sections (B) from WT and Tslpr−/− mice treated for 8 hours with ConA (scale bar = 50 µm). *P < 0.05 versus WT mice. All data are reported as means ± SEM with N = 19-20 from four independent experiments.

Discussion

TSLP functions as an initiator and regulator of allergic inflammatory diseases of the skin and lung,[6] which is attributed, in mice, to its prominent role in driving and sustaining type 2 responses, including production of IL-4, and infiltration of eosinophils.[8, 9] Although the signaling mechanisms that lead to secretion of TSLP in vivo are not known, early inflammatory signals, including IL-1β and TNF-α, can induce TSLP expression and secretion from epithelial cells[16-18] and may play a role in this process. TSLP, in turn, can induce expression and secretion of type 2 cytokines, including IL-4, from leukocytes.[9, 15]

We now provide evidence, for the first time, that TSLP signaling plays a pathological role in liver inflammation initiated by the hepatotoxic drug, halothane. First, halothane treatment of mice resulted in liver injury that was accompanied by increased liver expression of TSLP, its receptor components, and IL-4 as well as elevated serum levels of the cytokines (Fig. 1). We also showed that hepatic expression and serum levels of IL-1β and TNF-α increased during the early phase of HILI in mice (Fig. 5C,D) and that mouse and human hepatocytes secreted TSLP in response to these factors (Figs. 5E,F and 6B, respectively). It is possible that TSLP in the liver exerts its effect on various leukocyte populations to confer a type 2 immune response, which may play a role in mediating HILI. In this regard, TSLPR and IL-7Rα were detected on the surface of CD4+ T cells, CD8+ T cells, NKT cells, and eosinophils isolated from naïve mice and those treated with vehicle and halothane (see Supporting Fig. S1C and S1D, respectively). Second, severity of HILI in mice deficient in TSLPR was lower than in WT controls, as were hepatic expression and serum levels of IL-4 and eotaxins as well as hepatic eosinophils (Fig. 2A, B, E, F, and D, respectively). These findings are consistent with previous reports that Tslpr−/− mice exhibited attenuated disease severity, levels of type 2 cytokines, including IL-4, and numbers of eosinophils at the site of inflammation in mouse models of asthma[8] and atopic dermatitis.[9] Third, we provided evidence indicating that IL-4 also had a pathogenic role in HILI by showing that IL-4-deficient mice were less susceptible than WT mice to HILI and had depressed levels of hepatic eosinophils (Fig. 3A, B, and D, respectively) that were reversed significantly when mice were treated with recombinant IL-4 (Fig. 4A, B, and C, respectively). IL-4 was similarly reported to be involved in antigen-induced lung inflammation when IL-4-deficient mice experienced decreased disease severity and numbers of eosinophils in lungs, compared to WT controls.[20] Fourth, our finding that ConA-induced liver injury was nearly abolished in Tslpr−/− mice (Fig. 7) provided additional support for a role of TSLP signaling in injury to the liver because elevated levels of IL-4, eotaxins, and eosinophils play prominent roles in its pathogenesis.[10, 11, 21] Although these findings provide significant evidence for TSLP having a pathologic role in HILI, additional mechanistic studies will be needed to establish this role for TSLP.

The mechanism by which IL-4 caused liver injury in the mouse model of HILI was investigated in hepatocyte cultures. We found that IL-4 directly stimulated expression and secretion of eosinophil eotaxins in cultured mouse (Fig. 5A and B, respectively) and human hepatocytes (Fig. 6). IL-4 also augmented the activities of IL-1β and TNF-α to induce TSLP expression and secretion in mouse liver cells (Fig. 5E,F) and, to a lesser extent, in human hepatocytes from 1 of the 3 donors (see Supporting Fig. S2), supporting previous reports of synergistic effects of IL-4, IL-1β, and TNF-α on production of TSLP from epithelial cells.[16, 18] These findings suggest that TSLP and IL-4 may act in a feed-forward inflammatory cascade in the liver to drive eosinophil infiltration during HILI. At this point, it remains unknown what leukocytes are signaled by TSLP and what the source of IL-4 is during HILI. Although NKT cells appear to be the main source of IL-4 in ConA-mediating liver injury,[21] the role these cells play in HILI in mice remains controversial.[22, 23]

TSLP signaling may be involved in instances of DILI caused by other drugs, because IL-4 is associated in the pathogenesis of diclofenac-, flutamide-, and methimazole-induced liver injury in mice.[24-26] In contrast to these reports, IL-4 plays a protective role in a mouse model of acetaminophen-induced liver injury.[27] In addition, genetic variants that resulted in high IL-4 transcription were associated with cases of diclofenac-induced hepatotoxicity,[28] suggesting that increased type 2 responses may lead to enhanced susceptibility to DILI. Similarly, in diseases with type 2–induced pathology, genetic variants and polymorphisms in the TSLP gene are associated with disease susceptibility, such as polymorphisms in the promoter region of the TSLP gene that enhance mRNA expression in asthma,[29] as well as variants in the genes encoding TSLP, TSLPR, and CCL26 in eosinophil esophagitis.[30] It is possible that aberrant levels of TSLP, other type 2 cytokines, and/or eotaxins mediated by genetic and other modulators of gene expression may serve as potential risk factors for DILI. In addition, TSLP and type 2 cytokines may play similar roles in other diseases of the liver where hepatic eosinophilia are associated, in particular, hepatic allograft rejection,[31] primary sclerosing cholangitis,[32] primary biliary cirrhosis,[33] and chronic viral hepatitis C.[34] For example, a recent report showed that TSLP expression was increased in patients with chronic hepatitis C viral infection and that hepatocytes infected with hepatitis C virus secrete TSLP in vitro.[35] In conclusion, this report provides the first evidence implicating TSLP in liver inflammation and injury caused by a drug, and suggests that other liver diseases where eosinophilia and type 2 immunity have been associated may also be mediated by TSLP signaling.

Acknowledgment

The authors acknowledge John George for maintaining the IL4−/− mouse colony and the National Heart, Lung and Blood Institute Flow Cytometry core facility for their help with this work. The authors thank Dr. Kimberly Dyer (National Institute of Allergy and Infectious Diseases) for her helpful discussions pertaining to eosinophils. The authors also thank Tami McCoy Graf for her careful review of the manuscript.

Ancillary