fax: 81 82 257 5224
Liver Biology and Pathobiology
Liver NK cells expressing TRAIL are toxic against self hepatocytes in mice
Article first published online: 26 APR 2004
Copyright © 2004 American Association for the Study of Liver Diseases
Volume 39, Issue 5, pages 1321–1331, May 2004
How to Cite
Ochi, M., Ohdan, H., Mitsuta, H., Onoe, T., Tokita, D., Hara, H., Ishiyama, K., Zhou, W., Tanaka, Y. and Asahara, T. (2004), Liver NK cells expressing TRAIL are toxic against self hepatocytes in mice. Hepatology, 39: 1321–1331. doi: 10.1002/hep.20204
- Issue published online: 26 APR 2004
- Article first published online: 26 APR 2004
- Manuscript Accepted: 22 DEC 2003
- Manuscript Received: 5 JUN 2003
- Business-Academic-Public Sector Cooperation
Although it is known that activation of natural killer (NK) cells causes liver injury, the mechanisms underlying NK cell-induced killing of self-hepatocytes are not clear. We demonstrated that liver NK cells have cytotoxicity against normal syngeneic hepatocytes in mice. Polyinosinic-polycytidylic acid (poly I:C) treatment enhanced hepatocyte toxicity of liver NK cells but not that of spleen NK cells. Unlike NK cells in other tissues, approximately 30%–40% of liver NK cells constitutively express tumor necrosis factor-related apoptosis-inducing ligand (TRAIL). An in vitro NK cell cytotoxic assay revealed that hepatocyte toxicity of liver NK cells from both naïve and poly I:C-treated mice was inhibited partially by an anti-TRAIL monoclonal antibody (mAb) alone and completely by the combination with anti-Fas ligand (FasL) mAb and a perforin inhibitor, concanamycin A, indicating contribution of TRAIL to NK cell-mediated hepatocyte toxicity. The majority of TRAIL+ NK cells lacked expression of Ly-49 inhibitory receptors recognizing self-major histocompatibility complex class I, indicating a propensity to targeting self-hepatocytes. Poly I:C treatment significantly upregulated the expression of Ly-49 receptors on TRAIL− NK cells. This might be a compensatory mechanism to protect self-class I-expressing cells from activated NK cell-mediated killing. However, such compensatory alteration was not seen at all in the TRAIL+ NK cell fraction. Thus, liver TRAIL+ NK cells have less capacity for self-recognition, and this might be involved in NK cell-dependent self-hepatocyte toxicity. In conclusion, our findings are consistent with a model in which TRAIL-expressing NK cells play a critical role in self-hepatocyte killing through poor recognition of MHC. (HEPATOLOGY 2004;39:1321–1331.)
Natural killer (NK) cells are thought to provide a first line of defense against invading infectious microbes by exerting an effector function without the necessity for priming.1, 2 In the liver, both conventional NK cells and NKT cells are quite abundant, in contrast to the relatively small percentage of these cells in the peripheral lymphatics.2–6 Although the underlying reason for this anatomically biased distribution has not been fully elucidated, NK and NKT cells have been shown to mediate cytolytic activity and to secrete cytokines, enabling them to perform distinct functions in response to infected cells and neoplastic cells.4, 5, 7 Under various pathophysiological conditions, such as viral hepatitis, bacterial infection, and autoimmune liver disease, liver NK cells are activated by various cytokines, such as interleukin (IL)-2, IL-12, interferon (IFN)-α/β, and IFN-γ, secreted from a wide variety of cell types, including dendritic cells, macrophages, and NKT cells, and this activation enhances their cytotoxic capacity,1, 2, 8
Activated liver NK cells target not only infected or transformed hepatocytes but also intact hepatocytes in a nonspecific fashion.8–10 Cytotoxicity of liver NK cells against normal hepatocytes seems excessive self-defense from the point of view of innate immunity. In general, the presence of self-major histocompatibility complex (MHC) class I molecules protects target cells from lysis mediated by NK cells via binding to inhibitory receptors.1, 11, 12 Although it is not clear at present how liver NK cells kill self-hepatocytes, it has been speculated that hepatocytes do not express sufficient MHC class I to inhibit NK cell-induced killing13 or that liver NK cells include a cell population lacking sufficient inhibitory receptors. Another possibility is that liver NK cells include a cell population expressing certain effector molecules that cannot be downregulated even by self-class I recognition via NK inhibitory receptors. Enhanced cytotoxicity against infected or transformed targets is a feature of activated NK cells, which is believed to be due to the elevated expression of tumor necrosis factor (TNF) family members, including Fas ligand (FasL), and the increased production of perforin, granzymes, and cytokines by these cells.14–18 Among the TNF family members, TNF-related apoptosis-inducing ligand (TRAIL) has recently been shown to be critical for NK cell-mediated anti-tumor function.19–21 Because TRAIL is constitutively expressed by NK cells in the liver, unlike in other tissues,21 this molecule might be involved in NK cell-induced hepatocyte killing.
In the present study, we proved that liver NK cells have cytotoxicity against freshly isolated syngeneic hepatocytes expressing self-MHC class I molecules and that TRAIL on the liver NK cells contribute to their hepatocyte toxicity. We also have demonstrated that the majority of liver NK cells expressing TRAIL lack Ly-49 inhibitory receptors that recognize self-MHC molecules. These findings are consistent with a model in which TRAIL-expressing liver NK cells, which are characterized as heretics in view of the self-tolerance system, play a critical role in self-hepatocyte killing through poor recognition of self-MHC.
Materials and Methods
C57BL/6J (B6) (H-2b) mice were purchased from Clea Japan, Inc. (Osaka, Japan). B6 severe combined immunodeficient disease (SCID) (H-2b) mice were purchased from Jackson Laboratory (Bar Harbor, ME) and were bred at Hiroshima University. All animals were maintained in a specific pathogen-free microenvironment. Mice were used between 8 and 12 weeks of age.
In Vivo Analysis of NK Cell-Mediated Cytotoxicity Against Autologous Hepatocytes.
To activate NK cells, B6 normal wild-type (WT) and SCID mice were treated with intraperitoneal (i.p.) injection of polyinosinic-polycytidylic acid (poly I:C) (Sigma, St. Louis, MO) (150 μg/body). To deplete NK cells, mice were treated with i.p. injection of rabbit anti-asialo GM1 serum (anti-AsGM1) (Wako Chemicals, Osaka, Japan) (50 μL/body) two days before poly I:C injection. In our preliminary experiment, the proportions of NK cells (NK1.1+ TCRβ−) and NKT cells (NK1.1+ TCRβ+) in liver leukocytes of the B6 mice treated with anti-AsGM1 were 2.3 ± 1.1% and 20.1 ± 1.1%, respectively (n = 4) (3 d after treatment), whereas those in liver leukocytes of untreated control B6 mice were 11.5 ± 2.1% and 20.9 ± 2.8%, respectively (n = 4). Peripheral blood was collected from each mouse 1 day after poly I:C injection. The serum activity of alanine aminotransferase (ALT) was assayed by standard enzymatic methods.
Isolation of Liver and Spleen Leukocytes.
Liver and spleen leukocytes were isolated from B6 WT and SCID mice that were untreated or treated with i.p. injection of poly I:C (150 μg/body) 1 day before the assay. Liver leukocytes were prepared as described previously.22 In brief, after pre-perfusion via the portal vein using 1 mL of phosphate-buffered saline (PBS) supplemented with 10% heparin, the liver was perfused with 50 mL of PBS supplemented with 0.1% ethylenediaminetetraacetic acid (EDTA), and the perfusate was collected and subjected to erythrocyte lysis using ammonium chloride/potassium solution.
Isolation and Culture of Hepatocytes.
Hepatocytes were prepared as described previously.23, 24 Isolated hepatocytes were cultured in the hepatocyte growth medium: Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and antibiotics of 100 IU/mL penicillin G-100 μg/mL streptomycin (10% DMEM) containing 20 mmol/L N-2-hydroxyethylpiperazine-N′-2-ethane sulfonic acid, 30 μg/mL L-proline, 0.5 μg/mL insulin, 10−7 mol/L dexamethasone, 44 mmol/L NaHCO3, 10 mmol/L nicotinamide, 10 ng/mL epidermal growth factor (EGF), and 0.2 mmol/L L-ascorbic acid 2-phosphate. In some experiments, the medium was supplemented with 300 U/mL IFN-γ (Pepro Tech, Inc., Rocky Hill, NJ). Viability of hepatocytes was determined as more than 80% by the Trypan blue exclusion test.
Flow Cytometric Analysis.
All analyses were performed on a FACSCalibur® cytometer (BD Biosciences, Mountain View, CA). For phenotyping NK cell surface markers, liver and spleen leukocytes were stained with the following mAbs: fluorescein isothiocyanate (FITC)-conjugated anti-NK1.1 (PK136), phycoerythrin (PE)-conjugated anti-T cell receptor (TCR)-β (H57-597), and biotin-conjugated anti-Ly-49A (A1), anti-Ly-49C/I (5E6), anti-Ly-49G2 (4D11), anti-B220 (RA3-6B2), anti-CD69 (H1.2F3), anti-FasL (MLF4), or unconjugated anti-TRAIL mAb (N2B2) (all from BD Pharmingen, San Diego, CA). Staining with unconjugated anti-TRAIL mAb was followed by staining with biotin-conjugated anti-rat immunoglobulin (Ig)G2a mAb (RG7/1.30). In some experiments, liver leukocytes were stained with FITC-conjugated anti-NK1.1, PE-conjugated anti-TRAIL (N2B2) (eBioscience, San Diego, CA), and biotin-conjugated anti-CD69, anti-Ly-49A, anti-Ly-49C/I, anti-Ly-49G2 mAb, or a mixture of anti-Ly-49A, anti-Ly-49C/I, and anti-Ly-49G2 mAbs. The biotinylated mAb was visualized with allophycocyanin-streptavidin. For phenotyping hepatocyte MHC molecules, hepatocytes were stained with biotin-conjugated anti-H-2Db (KH95) (BD Pharmingen) and PE-streptavidin. Nonspecific FcγR binding of labeled Abs was blocked by anti-CD16/32(2.4G2) (BD Pharmingen). Dead cells were excluded from the analysis by light-scatter and/or propidium iodide.
NK Cell Cytotoxicity Assay Against Self-Hepatocytes.
Freshly isolated or 7-day cultured hepatocytes from B6 WT mice were used as target cells. As effector cells, liver and spleen leukocytes were obtained from B6 WT or SCID mice that were untreated or injected with poly I:C (150 μg, i.p.) 1 day before assay. In our preliminary studies, we confirmed that B6 SCID hepatocytes and congenic B6 WT hepatocytes express MHC class I at the same levels and are equally susceptible to NK cell-mediated cytotoxicity (data not shown). After the hepatocytes (104 cells/well) had been cultured with those leukocytes at various ratios in 96-well round-bottom microtiter plates containing 10% DMEM (total volume of 200 μL) for 4 hours at 37°C in a 5% CO2 incubator, ALT activity in each supernatant was assayed by standard enzymatic methods. Percent cytotoxicity was calculated as (experimental ALT – spontaneous ALT) × 100 / (total ALT – spontaneous ALT). Total ALT activity was determined by lysing the cells with 2.5% Nonidet® P-40 (Nacalai Tesque, Inc., Kyoto, Japan). In some experiments, the assay was performed in the presence of 10 μg/mL of anti-TRAIL (N2B2) mAb, 10 μg/mL of anti-FasL (MFL3) mAb (all from BD Pharmingen), and/or 50 nmol/L of concanamycin A (CMA) (Wako Chemicals, Osaka, Japan), which inhibited perforin-mediated cytotoxicity.
Isolation of NK Cells.
Liver and spleen leukocytes were obtained from B6 WT mice treated with poly I:C 1 day before the assay. NK cells were separated by high-gradient magnetic sorting using autoMACS® (Miltenyi Biotec, Auburn, CA). In brief, leukocytes were magnetically labeled by using a cocktail of biotin-conjugated anti-CD4 (RM4-5), CD5 (53-7.3), CD8a (53-6.7), CD19 (1D3), Ly-6G (RB6-8C5), and Ter-119 mAbs (TER-119) (all from BD Pharmingen) and streptavidin microbeads (Miltenyi Biotec). Isolation of unlabeled NK cells was achieved by depletion of the labeled cells. TRAIL− NK cells were further magnetically sorted using unconjugated anti-TRAIL mAb, biotin-conjugated anti-rat IgG2a mAb, and streptavidin microbeads in the negative fraction. NK cells and TRAIL− NK cells were used as effector cells in a cytotoxicity assay.
Hepatocytes isolated from B6 WT mice were labeled with 5-(and -6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes, Eugene, OR) (5 μM), as described previously.25 CFSE-labeled hepatocytes (1 × 106) in 0.5 mL of medium 199 (Sigma, St. Louis, MO) were injected through the superior mesenteric vein to B6 WT mice treated with poly I:C on day –1 with or without pretreatment of anti-AsGM1 on day –2. Those mice were sacrificed 7 days after hepatocyte transplantation. Dissected livers were examined by fluorescent microscopy to identify CFSE-labeled hepatocytes.
The results were analyzed by Student's t test of means. A P-value less than 0.05 was considered to be statistically significant.
Liver NK Cells Have Cytotoxicity Against Normal Self-Hepatocytes.
To determine whether activation of NK cells leads to injury of normal self-hepatocytes, poly I:C, which activates NK cells primarily by inducing the production of type I (α, β) IFN and IL-12 from a wide variety of cell types,26 was injected into B6 WT mice. Administration of poly I:C resulted in mild but statistically significant elevation of serum ALT values (Fig. 1). Such elevation of ALT values was abrogated by prior injection of anti-AsGM1 sera in the mice, indicating that cells expressing AsGM1 (either NK or NKT cells) are required for induction of hepatocyte injury. Similar results were obtained in B6 SCID mice lacking NKT cells, as well as T and B cells, but possessing NK cells, demonstrating that NK cells were the principal effectors causing self-hepatocyte injury induced by poly I:C.
To assess cytotoxicity of NK cells against normal self-hepatocytes in vitro, an NK cell cytotoxicity assay using liver and spleen leukocytes isolated from untreated and poly I:C-treated B6 WT mice as effectors and syngeneic hepatocytes as targets was performed. As shown in Fig. 2, the liver leukocytes were able to mediate potent cytotoxicity against syngeneic hepatocytes, whereas the spleen leukocytes were not able to mediate cytotoxicity even after poly I:C stimulation. Such hepatocyte toxicity was abrogated by pre-treatment with anti-AsGM1 sera to the mice supplying effector leukocytes, indicating NK cell-mediated cytotoxicity (data not shown). The difference in hepatocyte toxicity of liver leukocytes and that of spleen leukocytes is not due to the difference in their proportions of NK cells, because similar results were obtained from cytotoxic assays using B6 SCID mice, in which the proportions of NK cells in liver and spleen leukocytes were almost the same (approximately 20%–25%) (Fig. 3).
Difference in Phenotypic Characteristics of Liver and Spleen NK Cells.
Many studies have demonstrated that liver NK cells have higher levels of NK activity than do spleen or peripheral blood NK cells, suggesting that liver NK cells are naturally activated in liver microenvironments.2, 4–6 To address this possibility, we analyzed expressions of various activation makers and effector molecules on liver and spleen NK cells of B6 SCID mice that were either non-treated or treated with poly I:C by flow cytometry (FCM). As shown in Fig. 4, the naïve liver NK cells were larger than the naïve spleen NK cells and expressed CD69 at a higher level than did the naïve spleen NK cells. However, such differences were no longer seen after poly I:C stimulation; i.e., both liver and spleen NK cells equally expressed B220 and CD69. On both liver and spleen NK cells, the expression of FasL was not detected in the naïve condition but was faintly detected after poly I:C stimulation. In contrast, approximately 30%–40% of the liver NK cells expressed TRAIL even in the naïve condition, but spleen NK cells did not. Poly I:C stimulation did not significantly alter the positive-frequency and intensity of TRAIL expression on liver and spleen NK cells. Similar results were obtained from B6 WT mice, although percentages of CD69+ and TRAIL+ NK cells in total NK cells were relatively low (50% and 25%, respectively) compared to those in B6 SCID mice (70% and 35%, respectively). Thus, liver NK cells are naturally activated and include a TRAIL+ subset in the naïve condition, whereas spleen NK cells are almost devoid of TRAIL expression even after poly I:C stimulation.
TRAIL Is Involved in Self-Hepatocyte Toxicity of Liver NK Cells.
We attempted to determine whether TRAIL+ NK cells exclusively mediate cytotoxicity against self-hepatocytes. By magnetic sorting, NK cells were isolated from spleen and liver leukocytes of poly I:C-treated B6 WT mice (Fig. 5A and B). TRAIL+ and TRAIL− NK cells were further sorted from the isolated liver NK cells (Fig. 5C), and the resulting populations were then analyzed for cytotoxicity against self-hepatocytes. As shown in Fig. 5D, liver NK cells including both TRAIL+ and TRAIL− cells mediated significantly higher hepatocyte toxicity when compared to spleen NK cells, whereas the hepatocyte toxicity of TRAIL− liver NK cells and that of spleen NK cells were comparable. These findings indicate that TRAIL+ liver NK cells predominantly mediate cytotoxicity against self-hepatocytes.
To determine the contribution of TRAIL molecules to the hepatocyte toxicity of liver NK cells, the effect of a neutralizing anti-TRAIL mAb was examined in a cytotoxicity assay using isolated liver NK cells including both TRAIL+ and TRAIL− liver cells as effectors and syngeneic hepatocytes as the target. Hepatocyte toxicity mediated by isolated liver NK cells was inhibited partially by the anti-TRAIL mAb alone and more profoundly by the combination of the anti-TRAIL mAb and CMA (Fig. 6). In addition, the presence of anti-FasL mAb together with anti-TRAIL mAb and CMA resulted in complete inhibition of liver NK cell-induced hepatocyte toxicity, indicating that FasL is also involved in self-hepatocyte killing by NK cells.
Reinforcement of MHC Class I-Expression on Hepatocytes Partially Abated but Did Not Eliminate Hepatocyte Toxicity of Liver NK Cells.
It has been thought that the expression of MHC class I molecules on hepatocytes is almost undetectable but is up-regulated by IFN-γ.13 In conflict with such an accepted belief, the significant expression of MHC class I molecules was detected on freshly isolated murine hepatocytes by FCM (Fig. 7B). The MHC class I expression on hepatocytes was downregulated by cultivation but upregulated in the presence of IFN-γ . To determine the relationship between expression of class I molecules on hepatocytes and susceptibility of the hepatocytes to NK cell-mediated killing, hepatocytes that were either freshly isolated or had been cultivated for 7 days with/without IFN-γ were subjected to NK cell cytotoxic assays using liver leukocytes obtained from poly I:C-stimulated B6 SCID mice as effectors. As shown in Fig. 7C, the level of MHC class I expression on the target hepatocytes influenced their susceptibility to NK cell-mediated killing; i.e., the cultivated class Idim hepatocytes were sensitive but the IFN-γ-stimulated class Ihigh hepatocytes were relatively resistant to liver NK cell-mediated killing. However, reinforcement of class I-expression on hepatocytes by IFN-γ could not completely eliminate hepatocyte toxicity of liver NK cells. This raises the question as to how liver NK cells kill hepatocytes expressing self-MHC class I molecules, which are known to inhibit NK cell function via binding to NK inhibitory receptors.
Majority of TRAIL-Expressing Liver NK Cells Lacked Expression of Ly-49 Inhibitory Receptors.
The cytotoxicity of liver NK cells against hepatocytes expressing self-MHC class I indicated the possibility that liver NK cells include a subset devoid of ability for recognizing self-MHC class I. Taken together with the finding that TRAIL contributes to the self-hepatocyte toxicity of liver NK cells, this suggests that cytotoxicity of liver NK cells expressing TRAIL might not be able to be inhibited even in the presence of self-class I recognition or that they might lack sufficient inhibitory receptors. To address these possibilities, we analyzed expressions of various inhibitory receptors recognizing MHC class I (Ly-49A, Ly-49C/I, and Ly-49G2) on liver NK cells from untreated and poly I:C-treated B6 SCID mice together with expression of TRAIL by multi-color FCM. As shown in Fig. 8, liver TRAIL+ NK cells expressed CD69 at high levels, indicating their naturally activating state. The majority of TRAIL+ NK cells lacked expression of Ly-49 inhibitory receptors, whereas approximately half of the TRAIL− NK cells expressed either Ly49-A, Ly-49C/I, or Ly-49G2 in untreated naïve B6 SCID mice. In the naïve condition, the majority of TRAIL− NK cells did not express CD69, indicating their inactivated state, in which NK cells might have limited cytotoxic activity, regardless of their expression of Ly-49 receptors. Poly I:C treatment significantly upregulated the expression of CD69 and Ly-49 inhibitory receptors on TRAIL− NK cells, i.e., 77% of TRAIL− NK cells expressed Ly49 and the intensity of expression was significantly upregulated. It is likely that upregulation of Ly-49 expression is a compensatory mechanism to protect self-class I-expressing intact cells from activated NK cell-mediated killing while maintaining cytotoxic activity against cells lacking self MHC molecules. However, such compensatory alteration (upregulation of Ly-49 expression) was not seen at all in the TRAIL+ NK cells (78% of TRAIL+ NK cells did not express Ly-49 even after Poly I:C treatment). Thus, TRAIL+ liver NK cells have less capacity for self-recognition, and this might be involved in NK cell-dependent self-hepatocyte toxicity.
Liver NK Cell-Mediated Cytotoxicity Caused Insufficient Intrahepatic Engraftment of Transplanted Autologous Hepatocytes.
Because liver NK cells are located predominantly within the sinusoidal lumen in close contact with sinusoidal endothelial cells, it is likely that hepatocytes neighboring on the liver sinusoid are targeted by liver NK cell-mediated cytotoxicity but those in other zones are not. This can explain why administration of poly I:C resulted in only mild elevation of serum ALT values in our studies (Fig. 1). In the case of hepatocyte transplantation, however, NK cell-mediated cytotoxicity would be a major obstacle to engraftment of autologous hepatocytes because direct interaction of the hepatocytes transplanted via the portal vein with liver NK cells is inevitable. To address this possibility, CFSE-labeled syngeneic hepatocytes (1 × 106) were infused into the liver via the portal vein in B6 mice. Seven days after hepatocyte transplantation, the fluorescence of the CFSE-labeled hepatocytes was undetectable in the liver by fluorescent microscopic analysis, indicating failure in engraftment of inoculated hepatocytes (Fig. 9A). In contrast, when the recipient mice were treated with anti-AsGM1 sera before transplantation, hepatocyte grafts were clearly detected in the perisinusoidal area of the liver (Fig. 9B). These findings indicate that liver NK cell-mediated hepatocyte toxicity causes insufficient intrahepatic engraftment of transplanted hepatocytes.
Liver NK cells constitute a unique NK population of cells that reside predominantly in hepatic sinusoids and partly in the parenchyma between hepatocytes.27, 28 Compared to freshly isolated peripheral blood and spleen NK cells, liver NK cells were found be more cytotoxic against various tumor cells.29, 30 Unlike freshly isolated spleen NK cells, liver NK cells have high levels of expression of mRNAs encoding for perforin, granzymes, and cytokines.30 These facts suggest that liver NK cells have characteristics comparable to those of activated NK cells. The present study has demonstrated that liver NK cells have significantly higher cytotoxicity against self-hepatocytes when compared to spleen NK cells even after activation with poly I:C, suggesting that the functional difference between liver NK cells and spleen NK cells is not merely derived from the difference in their activating states. Although the phenotype of liver NK cells resembled that of spleen NK cells activated with poly I:C, liver NK cells exclusively included cells expressing TRAIL (Fig. 4). TRAIL can induce either apoptosis by a Fas-associated death domain-dependent mechanism or necrosis via a receptor-interactive peptide-dependent cascade through ligation of its death domain-containing receptors under physiological conditions.19–21 Soluble TRAIL proved unique in that it frequently induced apoptosis of transformed cells in vitro without causing apoptosis of normal cells.31 TRAIL, for example, induced apoptosis in human hepatocellular carcinoma and cholangiocarcinoma cells.32, 33 However, results regarding the hepatotoxicity of the soluble form of TRAIL are conflicting.34 A recombinant soluble form of TRAIL lacked hepatotoxicity in cynomolgus monkeys and mice31 but was reported to induce apoptosis in cultured human hepatocytes.35 These observations raise the possibility that human hepatocytes are uniquely sensitive to TRAIL compared to nonhuman primate and rodent hepatocytes. There are also conflicting results regarding the hepatotoxicity of the membrane-bound form of TRAIL delivered by gene-transfection using an adenoviral vector.36 Despite a number of studies dealing with exogenous TRAIL of either the soluble form or membrane-bound form expressed by genomic manipulation, hepatotoxicity of TRAIL endogenously expressed on NK cells has not yet been investigated. In the present study, hepatocyte toxicity of liver NK cells was inhibited partially by anti-TRAIL mAb alone and further profoundly by the combination of anti-TRAIL mAb and CMA, indicating that TRAIL contributes to NK cell-mediated hepatocyte toxicity.
In addition to NK cell-mediated hepatocyte toxicity, NKT cells also cause hepatocyte injury in certain immunopathological conditions. A recent study has indicated that NKT cells activated by α-galactosylceramide (α-GalCer) induce hepatocyte damage through the Fas-FasL pathway.10 Different from hepatocyte damage induced by α-GalCer, NK cells were predominant effectors of hepatocyte damage induced by poly I:C kit given that poly I:C-treatment triggered hepatocyte injury in SCID mice lacking NKT cells. In the present study, poly I:C treatment did not increase the expression level of Fasl on NKT cells (data not shown) but slightly increased the expression of Fasl level of FasL on both spleen and liver NK cells (Fig. 4). It is likely that the Fas-FasL pathway plays a role in the hepatocyte toxicity mediated by NK cells, including TRAIL− NK cells. In support of this possibility, the presence of anti-FasL mAb in addition to anti-TRAIL mAb and CMA resulted in complete inhibition of liver NK cell-induced hepatocyte toxicity (Fig. 6). Thus, although TRAIL+ NK cells are predominant cells that have the ability to kill self-hepatocytes, it is likely that TRAIL molecules per se partially contribute to their killing. Either perforin and FasL are also involved in TRAIL+ NK cell-mediated hepatocyte toxicity or TRAIL− NK cells partially contribute to hepatocyte toxicity via perforin and FasL. To distinguish these possibilities, additional studies are needed.
It is well known that the presence of MHC class I molecules protects normal self-cells from NK cell-mediated cytotoxicity via binding to NK inhibitory receptors.11, 12 A “missing-self” hypothesis has been proposed to explain the protective effect of target cell MHC class I on NK cell-mediated lysis; when tissues lack expression of MHC class I or express abnormal forms of it, such as during tumorigenesis or infection, this inhibitory influence is released, permitting target lysis mediated by NK cells. Inconsistent with the missing-self hypothesis, we demonstrated in the present study that normal hepatocytes suffered from liver NK cell-induced killing even when the hepatocytes expressed normal MHC class I. We assumed that liver NK cells include a cell population lacking sufficient inhibitory receptors that can recognize self-MHC class I molecules, and we analyzed the expressions of various inhibitory receptors recognizing class I on liver NK cells. Ly-49A binds the class I molecules H-2Dd, Dk, and Dp, and Ly-49G2 binds H-2Dd and H-2Ld.37–41 These bindings mediate a negative signal to the NK cells, resulting in inhibition of lysis. Recent studies have suggested that these members of the Ly-49 family may also interact with H-2b molecules.42–45 TRAIL+ liver NK cells almost completely lacked expression of Ly-49A and Ly-49G2 inhibitory receptors, whereas approximately 10% and 30% of TRAIL− liver NK cells expressed Ly-49A and Ly-49G2, respectively. In addition, TRAIL+ liver NK cells less frequently express Ly-49 C/I, which binds H-2b molecules, compared to TRAIL− NK cells. These findings indicate TRAIL+ liver NK cells have less capacity for self-recognition, and this might be involved in NK cell-dependent self-hepatocyte toxicity. Notably, poly I:C treatment significantly upregulated expression of Ly-49 inhibitory receptors on TRAIL− NK cells but did not alter Ly-49 expression on TRAIL+ NK cells. It is likely that normal hepatocytes may be protected from killing mediated by activated TRAIL− NK cells via binding to upregulated Ly-49 inhibitory receptors but are susceptible to killing mediated by activated TRAIL+ NK cells because of the insufficiency of MHC class I recognition.
Such mechanisms for NK cell-mediated toxicity against autologous hepatocytes need to be taken into account in a case of autologous hepatocyte transplantation, because liver NK cells caused a major problem obstructing its success, i.e., insufficient intrahepatic engraftment of transplanted hepatocytes. Hepatocyte transplantation has been proposed as an alternative to whole-organ transplantation to support many forms of hepatic insufficiency. Previous studies have demonstrated therapeutic efficacy of hepatocyte transplantation through a variety of liver disease models.46, 47 Particularly for the treatment of inherited metabolic disease, a technique for transplantation of genetically modified autologous hepatocytes, which have the advantage of no T cell-mediated rejection, is being developed. Although the host liver represents an ideal site for transplanted hepatocytes in terms of the unique hepatic organization and interactions with nonparenchymal liver cells,48 insufficient intrahepatic engraftment of transplanted hepatocytes, even those that are autologous, remains a major obstacle. It has been demonstrated that transplanted hepatocytes become stacked at the portal vein radices, resulting in deposition of some of the cells at the hepatic sinusoid.49 Although the majority of hepatocytes are cleared from these areas, a portion of the cells start to translocate into the space of Disse by disrupting the sinusoidal endothelium and finally join adjacent host hepatocytes. Because liver NK cells are located predominantly within the sinusoidal lumen, direct interaction of the transplanted hepatocytes with liver NK cells is inevitable. Therefore, to establish intrahepatic placement of hepatocyte grafts, inhibition of liver NK cell-mediated hepatocyte toxicity might be necessary.
The authors thank Drs. T. Itamoto and H. Tashiro for their advice and encouragement.
- 1Natural killer cells. In: PaulEW, ed. Fundamental Immunology. 4th ed. Philadelphia, Pa: Lippincott-Raven; 1999: 575–603..