Involvement of NK1⁺ T Cells and Their IFN-γ Production in the Generalized Shwartzman Reaction

IFN-γ, TNF-α, and IL-1 are reportedly involved in the pathogenesis of the generalized Shwartzman reaction (1, 2, 3), which is known as a lethal shock syndrome of mice elicited by two consecutive injections of LPS into mice. IFN-γ is important for the priming phase, while TNF-α and IL-1 are effector molecules in mice that were already sensitized by IFN-γ (2, 3). Recently, it was revealed that IL-12 is a pivotal factor in the priming phase of the Shwartzman reaction because of its induction of IFN-γ. In addition, i.p. injection of either IL-12 or IFN-γ can replace the priming effect of an LPS injection into the footpads of mice (2, 3). IL-12 is a heterodimeric cytokine with various biologic effects produced mainly by phagocytic and/or APCs; it plays important roles in the protection against bacterial and parasitic infections (4). IL-12 also acts against malignant tumors by activating CD8⁺ cytotoxic T cells and NK cells (4, 5, 6). These IL-12 effects are mainly, but not exclusively, mediated through IFN-γ production (4).

NK1⁺ T cells are a recently identified T cell population that consists of CD4⁺ as well as CD4⁻8⁻ double negative cells. This population is dependent on MHC class I or related molecules for its development (7, 8, 9). We recently reported that NK1⁺ T cells in the liver (and also presumably in other sites) are one of the responders to IL-12. IL-12 endows potent antitumor and antimetastatic functions to these cells (10, 11, 12, 13). LPS also activates NK1⁺ T cells in the liver through IL-12 production from Kupffer cells. This LPS-induced effect is blocked by anti-IL-12 Ab and partly inhibited by anti-IFN-γ Ab (12). These findings led us to determine whether or not NK1⁺ T cells are involved in a generalized Shwartzman reaction. In the present study, we demonstrate that NK1⁺ T cells are potent producers of IFN-γ in the priming phase of an IL-12-induced Shwartzman reaction and are involved in the mortality of this phenomenon. We also demonstrate that excessively activated liver NK1⁺ T cells, as well as NK cells in the liver, can be cytotoxic even against hepatocytes.

Male C57BL/6 (B6²)^+/+ mice, 6 to 8 wk of age, were purchased from CLEA Japan (Tokyo, Japan). Male B6 nu/nu mice were maintained in the Tohoku University School of Medicine (Sendai, Japan). B6 β₂-microglobulin-deficient (β₂m^−/−) mice, obtained by backcrossing original β₂m^−/− mice eight times with B6 mice (14), were kindly provided by Dr. M. Taniguchi at Chiba University School of Medicine (Chiba, Japan). Mice were fed under specific pathogen-free conditions.

Escherichia coli-derived LPS was purchased from Sigma (St. Louis, MO). Recombinant murine IL-12 (15, 16) with an activity of 4.9 × 10⁶ U/mg was kindly provided by Dr. M. Kobayashi of Genetics Institute (Cambridge, MA). The preparations were diluted in sterile PBS(−) briefly before use.

Mice were anesthetized and bled from the subclavian artery and vein. The liver, spleen, and kidneys were removed immediately, fixed with 10% formalin for 12 h, and embedded in paraffin. For histologic observation, several tissue sections were deparaffinized and stained with hematoxylin and eosin.

Under ether anesthesia, mice were bled from the subclavian artery and vein. The livers were removed from the mice. Hepatic lymphocytes were prepared essentially as previously described (12). Briefly, the liver was passed through stainless steel mesh and suspended in HBSS. After one washing, cells were resuspended in osmolarity and pH adjusted 30% Percoll containing 100 U/ml heparin and centrifuged at 2000 rpm for 15 min at room temperature. The pellet was resuspended in RBC-lysis solution (0.17 mM NH₄Cl, 0.01 mM EDTA, 0.1 M Tris, pH 7.3), and then washed twice in 10% FCS-RPMI 1640 medium.

As recently reported (2), the priming IL-12 injection was given i.p. and was followed 24 h later by a challenge i.v. LPS injection. The occurrence of the generalized Shwartzman reaction was evaluated by mortality.

Target cells used were P815 mastocytoma cells, L5178Y T lymphoma cells of DBA/2 (H-2d) origin, and TLR-2 hepatocyte cells. P815 mastocytoma cells and L5178Y T lymphoma cells were propagated in RPMI 1640 medium supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, and 25 mM NaHCO₃ in humidified 5% CO₂ at 37°C. TLR-2 hepatocytes were propagated as previously described (17). This hepatocyte cell line was established from SV40 T-Ag gene transgenic mice of B6 background. It is considered to be very similar to normal hepatocytes because these cells have intracellular p450IA and albumin (17). Target cells were labeled with 100 μCi of Na₂⁵¹CrO₄ for 60 min at 37°C in RPMI 1640 medium containing 10% FCS, washed three times with medium, and subjected to cytotoxicity assays as previously described (10). Labeled targets (10⁴/well) were incubated in a total volume of 200 μl with effector cells in 10% FCS-RPMI 1640 in 96-well round-bottom microtiter plates. The plates were incubated for 4 h after they were centrifuged, after which the supernatant was harvested and counted in a gamma counter. The cytotoxicity was calculated as the percentage of releasable counts after subtraction of spontaneous release. The spontaneous release was less than 15% of the maximum release.

Liver MNC were depleted of specific cell populations by using mAbs and rabbit serum as a source of complement (C) (LOW-TOX-M; Cedarlane, Hornby, Ontario, Canada). As previously described (10), cells (5 × 10⁷) were incubated with the respective Ab at optimal concentrations (45 min, 4°C), washed, and incubated (45 min, 37°C) in 10 ml of a 1:10 dilution of rabbit serum as a source of complement. Optimal concentrations for each Ab were determined by treating isolated MNC with various dilutions of Ab and C, followed by flow cytometric analysis to assay the success of the treatment. The mAbs used, anti-NK 1.1 (PK136) and anti-CD3 (500-A-2), were prepared from hybridomas grown in our laboratory. These Abs were purified from ascites, which were prepared from hybridomas grown in our laboratory, by affinity chromatography on protein G-Sepharose (Pharmacia, Piscataway, NJ). Under these conditions, Ab treatment depleted >95% of the appropriate cell population in MNC.

Monoclonal mouse anti-NK1.1 Ab (PK136) (200 μg/mouse) or polyclonal rabbit anti-asialo GM1 Ab (50 μg/mouse) was injected into mice twice a week before sacrifice to eliminate NK-type cells. Polyclonal rabbit anti-asialo GM1 (AGM1) Ab was purchased from Wako (Tokyo, Japan). The elimination of NK-type cells was confirmed by flow cytometry and cytotoxicity assays against an NK-sensitive YAC-1 target.

Liver and/or spleen MNC were obtained from B6 or B6 nu/nu mice that had been injected with IL-12 (0.5 μg/mouse) 6 h before sacrifice. MNC were stained with anti-CD3 Ab and anti-NK1 Ab. NK cells and NK1⁺ T cells were separated by FACS Vantage (Becton Dickinson, Mountain View, CA). Contaminations by other types of cells in each population were less than 5%. In all, 2 × 10⁶/ml (10% FCS-RPMI 1640) of NK1⁺ T cells and NK cells were cultured with low amounts of IL-2 (5 U/ml) in 96-well flat-bottom plates for 48 h. Culture supernatants were subjected to ELISA.

Serum levels of IFN-γ and TNF-α were evaluated using the cytokine-specific ELISA, commercially available from Endogen (Boston, MA).

Differences between groups were analyzed by the Mann-Whitney U test. Differences were considered significant if p < 0.05.

Priming of mice by a 0.5 μg IL-12 injection (i.p) followed by a challenge with an LPS injection (i.v.) could induce the Shwartzman reaction. Mortality depended on the dose of the second LPS injection (Table I). However, neither a 0.25 μg IL-12 nor a 5 μg LPS injection could prime the reaction. Similarly, two consecutive injections of 0.5 μg of IL-12 with a 24 h time interval or a 5 μg LPS injection followed 24 h later by a 0.5 μg injection of IL-12 were ineffective. Similar mortalities were observed in nude mice by 0.5 μg of IL-12 priming followed by an LPS challenge.

We recently reported that anti-AGM1 Ab injection into mice could delete NK cells but did not delete NK1⁺ T cells (intermediate TCR cells), while anti-NK1 Ab (PK136) depleted both NK cells and NK1⁺ T cells in the liver and other organs (18, 19). Depletion of NK cells and NK1⁺ T cells by anti-NK1 Ab diminished serum IFN-γ levels in the priming phase (24 h after IL-12 injection) of the Shwartzman reaction (Table II), but anti-AGM1 Ab treatment had only a moderate effect on serum IFN-γ levels (Table II). These findings are further supported by the fact that β₂m^−/− mice, which have NK cells but fewer NK1⁺ T cells, are also low producers of IFN-γ (Table II). However, since anti-NK1 Ab-treated mice produce low but significant amounts of IFN-γ, conventional T cells may be partly responsible for IFN-γ production.

Sorting and culture experiments of liver MNC of IL-12-injected mice confirmed that NK1⁺ T cells produced greater amounts of IFN-γ than did NK cells (Table III). Similar results were obtained from liver MNC from B6 nude mice (Table III). While NK1⁺ γδ T cells of liver T cells of B6 mice are negligible (10), NK1⁺ T cells of the liver and spleen of nude mice contain 40 to 50% γδ T cells (not shown). Therefore, NK1⁺ γδ T cells of nude mice may also have participated in IFN-γ production. Although we do not deny that anti-CD3 Ab staining itself may have somewhat stimulated NK1⁺ T cells through their CD3/TCR complex, it probably did not have a great effect because sorted liver NK1⁺ T cells of B6 mice without IL-12 injection produced only a small amount of IFN-γ in same culture condition (data not shown).

In contrast to IFN-γ production, elevated serum TNF-α levels in the second phase (1 h after LPS challenge) were not affected by either Ab pretreatment (Table IV). We followed the time course of TNF-α levels in the second phase and found that they showed maximal levels at 1 h after the second LPS challenge.

The above results suggest that NK1⁺ T cells are major IFN-γ-producing cells in the priming phase of the Shwartzman reaction. To investigate the role of NK1⁺ T cells in the pathogenesis of the Shwartzman reaction, either anti-NK1 Ab or anti-AGM1 Ab was injected into mice twice before IL-12 priming. It was found that anti-NK1 Ab treatment reduced the mortality significantly, while anti-AGM1 treatment was ineffective (Table V). Further, β₂m^−/− mice all survived in a Shwartzman reaction (Table V).

Subsequently, we evaluated the histologic changes induced in the liver by IL-12 injection. The result revealed that IL-12 injection (i.p.) induced lymphocyte infiltration and focal necrosis in the liver in a dose-dependent manner (Fig. 1). Although IL-12 and subsequent LPS injection induced more focal necroses in the liver, serum transaminases were not dramatically elevated (data not shown).

Next, we examined whether activated liver MNC cause hepatocytes damage. Treatment of normal mice with a single injection of IL-12 augmented the cytotoxic activity of hepatic MNC against the P815 target (Table VI), as we reported previously (10, 11, 12, 13), while IL-12-activated hepatic MNC possessed lower cytotoxicities against Fas Ag-negative L5178Y cells (20) and the hepatocyte cell line, TLR-2 (17) (Table VI). However, among combinations of two consecutive injections of IL-12 (0.5 μg) or LPS (5 μg) with a 24 h time interval, IL-12 injection followed 24 h later by a challenge of LPS induced greatly augmented in vitro cytotoxicity against L5178Y and TLR-2 48 h after LPS challenge (Table VI).

Subsequently, we attempted to determine whether NK1⁺ T cells were responsible for the cytotoxity against hepatocytes. Hepatic MNC isolated from B6 or B6 nu/nu mice injected with IL-12 followed 24 h later by an LPS challenge were treated with anti-NK1.1 Ab or anti-CD3 Ab and C in vitro before cytotoxic assays. The results showed that treatment with anti-NK1.1 Ab almost completely reduced cytotoxity and anti-CD3 Ab significantly reduced cytotoxity (Table VII).VII These results show that although activated NK cells also seem to be responsible for cytotoxicity against hepatocytes, activated NK1⁺CD3⁺ cells are a main population causing hepatocyte damage. It was also shown that NK1⁺ T cells exert cytotoxicity, at least partly, in a Fas-Fas ligand-independent fashion. NK1⁺ γδ T cells of nude mice MNC also may participate in cytotoxicity.

Although it was recently reported that liver NK1⁺ αβ T cells are potent responders to IL-12 (10, 11, 13) or LPS (via production of IL-12 from Kupffer cells) (12), the role of these cells in the Shwartzman reaction has not been explored. In the present study, we demonstrate that NK1⁺ T cells are major producers of IFN-γ, which is an essential cytokine for priming of the Shwartzman reaction (2). IL-12 administration and extended doses of subsequent LPS injection induced a lethal Shwartzman reaction. Depletion of NK cells and NK1⁺ T cells before IL-12 priming greatly decreased IFN-γ production and thereby reduced the mortality of mice, whereas depletion of NK cells alone only partially decreased IFN-γ production by IL-12 and did not improve mouse mortality. The result of β₂m^−/− mice also supports these findings. In addition, the fact that the mortality of athymic nude mice was comparable to that of normal mice and liver high TCR cells did not produce a substantial amount of IFN-γ suggests that thymus-derived T cells are not significantly involved in the pathogenesis of the Shwartzman reaction. Cytotoxic activities of NK1⁺ T cells and NK cells against hepatocytes were also induced by IL-12 and subsequent LPS injection.

It was recently reported that LPS can activate macrophages to release IL-12 (2, 4), a cytokine that stimulates T and NK cells to produce IFN-γ (4, 6, 21, 22). IFN-γ induced by IL-12 is important in the priming phase of the Shwartzman reaction (2, 3). On the other hand, it has been recently reported that NK1⁺ T cells also produce IFN-γ and IL-4 by the stimulation via their TCR (23, 24). However, NK cells as well as NK1⁺ T cells produce IFN-γ but not IL-4 by the stimulation via their NK1.1 Ag or by IL-12 (25). It also has been reported recently that liver NK1⁺ T cells produce IL-4 by stimulation with Con A in vivo (26). The present study confirmed that NK1⁺ T cells are potent producers of IFN-γ by stimulation of IL-12. Therefore, we propose that this novel population of T cells can trigger either Th1 or Th2 immune responses by producing IFN-γ or IL-4, depending upon Ags or factors.

IFN-γ produced by NK1⁺ T cells may further activate monocytes or macrophages in a positive feedback loop and augment their functions (27), including phagocytosis and Ag presenting capacity. It is known that a majority of bacteria and their components, including LPS and peptidoglycan polysaccharide, accumulates in the liver after entering the blood stream and is removed there by Kupffer cells (12). Thus, the liver seems to be an organ that is prepared to protect against acute bacterial infection by inducing a Th1 response. However, we do not deny that NK cells may produce larger amounts of IFN-γ than NK1⁺ T cells in response to other cytokines or factors. Further study of this phenomenon is now in progress.

IL-12 injection and subsequent LPS administration, but not two consecutive injections of IL-12, killed mice. It is suggested that LPS might induce a lethal reaction by inducing greater amounts of cytokines or perhaps even different inflammatory cytokines (e.g., IL-1, TNF-α) than those induced by IL-12. In fact, IL-12 administration to mice does not induce a significant elevation of serum TNF-α (our unpublished observation). TNF-α and IL-1 reportedly are effector molecules in the second phase of a lethal Shwartzman reaction. LPS also activates granulocytes to release enzymes and radicals (2). However, it was recently revealed that IFN-γ production in the priming phase, or presensitization of mice by this cytokine, is essential to cause mortality in the Shwartzman reaction (2) because neither IL-1, TNF-α, nor a combination of the two alone can cause mortality. Our finding is consistent with this observation. It is notable that TNF-α production in the second phase per se is independent of IFN-γ production in the priming phase. IFN-γ may prime the Shwartzman reaction through up-regulation of MHC class I or class II expression of tissues and cause the susceptibility of mice to LPS challenge (28). Consistent with this speculation, β₂m^−/− mice, which produced more IFN-γ than NK cell- and NK1⁺ T cell-depleted mice, are resistant to the Shwartzman reaction, supporting the possibility that class I molecules expressed on the tissues are involved in the mortality evolved in this phenomenon.

Although IL-12 and subsequent LPS administration induced cytotoxicity of NK1⁺ T cells and NK cells against hepatocytes in vitro and induced focal necroses in the liver in vivo, elevation of transaminase was not so evident. This is probably partly due to the fact that, in contrast to in vitro experiments, hepatocytes in vivo are surrounded by sinusoidal endothelial cells and are largely protected from direct contact with these activated cells. In addition, a new factor that causes more severe hepatocyte damage was recently reported in mice treated with Propionibacterium acnes and subsequent LPS challenge (29), indicating that additional factors are needed to cause more severe liver damage in vivo. Nevertheless, while infiltration by NK1⁺ T cells and NK cells per se may not be a major direct cause of mouse mortality in the Shwartzman reaction, it is indicated that these cells sometimes attack hepatocytes under certain conditions.

NK1⁺ αβ T cells mainly use an invariant Vα14 gene product (coupled with Vβ8 or Vβ7) for their TCR (19, 30, 31, 32). This finding indicates that these cells recognize limited sets of Ags. We recently reported that while human PBL contains several percentages of αβ T cells with a NK cell marker, CD56, human liver contains a large population of these cells (33). Nevertheless, CD56⁺ αβ T cells or CD56⁺ γδ T cells selectively expand and acquire strong MHC-unrestricted cytotoxicity against tumors when monocyte-depleted PBL are cultured with a combination of IL-12 and IL-2 (33). It can be speculated that these CD56⁺ αβ T cells in humans are a functional counterpart of mouse NK1⁺ αβ T cells (19, 33). In fact, CD4⁺CD56⁺ αβ T cells selectively expand in vitro from the liver MNC of hepatitis B patients and recognize hepatitis virus Ag with self-MHC molecules (34), and NK1⁺ αβ T cells are also reportedly responsible for the elimination of adenovirus-infected hepatocytes in mice (35). These findings and the present results raise the possibility that these T cells with an NK cell marker can recognize virus Ags and self-molecules of hepatocytes or of liver APC (Kupffer cells or dendritic cells) and sometimes can be autoreactive effectors.