NK cells are a relatively rare cell population in peripheral lymphoid organs but are abundant in the liver, raising questions as to their function in immune responses to infections of this organ. To investigate this, cell-mediated immunity to viral liver infection induced by a type 5, replication-defective, adenovirus was examined. It is shown that NK cells in the absence of T cells cause hepatocyte apoptosis in virus-infected livers associated with an increase in liver enzymes in the serum. Concomitantly, NK cells induce production of IFN-γ, inhibitable by their elimination before infection. NK cells are shown to be necessary for optimal priming of virus-specific T cells, assessed by delayed-type hypersensitivity response and CTL activity, consistent with their ability to secrete IFN-γ. The conclusion is drawn that NK cells mediate two important functions in the liver: they induce cell death in the infected organ and concomitantly stimulate the induction of T cell-mediated immunity by release of IFN-γ.
The role of NK cells in immune responses has long been an area of intense interest as these cells are thought to provide a first line of defense against invading infectious microbes by exerting effector functions without the necessity for priming (1). The ability of these cells not only to lyse cells but also to secrete cytokines endows them with the ability to regulate responses of the adaptive immune system (2, 3). One limitation in elucidating the role of these cells has been their relative paucity in the peripheral lymphoid organs, and only a few examples have been reported in which elimination of NK cells results in impairment of immune responses to infectious agents. In contrast to the relatively low frequency of NK in the peripheral lymphatics, both conventional NK cells and NKT cells, which express NK as well as TCRs, are quite abundant in the liver (4, 5, 6, 7, 8). This has raised intriguing questions as to the underlying reason and has led to the speculation that optimal immune responses in the liver require the presence of NK and NKT cells (2, 7, 8, 9, 10, 11). Interestingly, the expression of MHC class I Ags, well known to inhibit NK cell functions via binding to NK inhibitory receptors (12), is almost undetectable in hepatocytes, yet is up-regulated by IFN-γ (13). One could therefore hypothesize that the liver is an organ in which NK cells play an important role in the early response to infections. To test this hypothesis, a viral infection model was used that consists of a replication-defective type 5 adenovirus with deletions in the E1 and E3 regions. Intravenous injection of this virus causes a high efficiency infection in the liver, reflected in the expression of virus-coded genes (14). Using this model we show that NK cells exert a dual function: they induce apoptosis in hepatocytes and concomitantly stimulate the induction of a T cell response. It is also shown that elimination of NK cells before the infection leads to inhibition of liver injury, secretion of IFN-γ, and induction of virus-specific T cell responses.
Materials and Methods
Animal strains, immunizations, Ab treatments, and assays for delayed-type hypersensitivity (DTH)3 and serum alanine aminotransferase (ALT)
Pathogen-free female C57BL/6 (H-2b) and C57BL/6 nude mice, 6–12 wk of age, were obtained from The Jackson Laboratory (Bar Harbor, ME). BALB/C SCID mice were purchased from the National Institutes of Health breeding colony (Frederick Cancer Research Center, Frederick, MD). For immunization type 5 adenovirus with deletions in the E1 and E3 regions and carrying the lacZ gene was purchased from Microbix Biosystems (Toronto, Canada). Virus was propagated in 293 cells as provided by the supplier. In all experiments animals were injected with 2–3 × 109 PFU virus into the tail vein. Experiments were conducted in accordance with guidelines from the University of Southern California institutional animal care and use committee.
To deplete NK or NKT cells, mice were injected i.p. daily with either 15 μl of anti-AsGM1 (Wako Pure Chemicals Industries, Osaka, Japan) or 250 μg/mouse of anti-NK1.1 mAb PK136 starting on day −1 until termination of the experiment. To neutralize IFN-γ, mice were injected with 250 μg/mouse anti-IFN-γ R4–6A2 (American Type Culture Collection, Manassas, VA) on days −1, 2, 4, 6, and 8. Control rat IgG1 (for anti-IFN-γ treatment) and control mouse IgG2a Ab (for anti-NK1.1 treatment) were purchased from PharMingen (San Diego, CA) and control rabbit Ig (for anti-AsGM1 treatment) was purchased from Calbiochem (La Jolla, CA). Control Abs were injected at equivalent doses and schedules. To activate NK cells in nude mice, animals were injected once i.p. with 150 μg/mouse poly(I:C) (Sigma, St. Louis, MO) on the day of virus injection (day 0). For assay of DTH reactivity mice were challenged into the right footpad with 109 PFU of virus in 25–40 μl of PBS and in the left footpad with PBS 9 days after virus infection. After 18 h swelling was measured with a caliper gauge (Mitsutoyo, Tokyo, Japan), and the difference in footpad diameters between left and right footpads was determined.
For assay of serum ALT, mice were anesthetized with methoxyflurane (Pittman Moore, Mundelein, IL) and bled via the retro-orbital venous plexus. Serum (100 μl) was mixed with 1 ml of ALT assay solution (Sigma) and incubated for 90 s at 30°C. OD340 was measured in a spectrophotometer following the supplier’s protocol.
Preparation of liver lymphocytes and fluorometric analysis
Mononuclear cells were isolated from livers by passing tissue through a 200-gauge stainless steel mesh in serum-free HBSS (4, 6). The cell suspension was centrifuged 500 × g for 5 min, and the supernatant was discarded. The cell pellet from one liver was resuspended by vigorous vortexing in 20 ml of Percoll-HBSS containing 150 U/ml heparin. A Percoll working solution of 100 ml was prepared by mixing 92.5 ml of Percoll stock (Pharmacia Biotech, Uppsala, Sweden), with 7.2 ml of 10× PBS and 1.2 ml of 7.5% sodium bicarbonate, pH 7.2–7.4. Four parts of the Percoll working solution were mixed with 6 parts of HBSS, and the Percoll-HBSS solution was used for resuspension of cell pellets. The cell suspension was centrifuged at 800 × g for 20 min at room temperature, and the cell pellet was resuspended in 10 ml of RBC lysis solution. The lysis solution consists of 155 mM NH4Cl, 10 mM KHCO3, 1 mM EDTA, and 170 mM Tris, pH 7.3. After incubation for 10 min at room temperature (15), cells were harvested by centrifugation and washed twice in HBSS 5% FCS before use.
For FACS analysis 106 cells were stained with mAbs at a concentration of 1 μg/100 μl of PBS containing 0.2% BSA (Roche, Indianapolis, IN) and 0.05% sodium azide for 30 min on ice (4). The following Abs were used: PE- or FITC-conjugated anti-CD3 (145-2C11), PE- or FITC-conjugated anti-CD4 (GK1.5), PE- or FITC-conjugated anti-CD8 (53-6.7), and PE-conjugated anti-NK1.1 (PK136), all purchased from PharMingen (San Diego, CA). FACS analysis was performed on a FACStarPlus (Becton Dickinson, Mountain View, CA). The numbers of CD3+, CD4+, CD8+, NK, and NKT cells per liver were calculated by multiplying the percentage of each population with the total number of mononuclear cells per liver.
Induction of in vitro CTL responses and IFN-γ-specific ELISPOT assays
To assay CTL priming, spleen cells were harvested 10 days after virus infection and cultured at 5 × 106 cells/ml in complete RPMI 1640 medium containing 10% FCS for 5 days in 24-well plates (14, 16). Cultures received 2 PFU of virus/input cell as immunogen. To prepare targets, 107 C57SV (H-2b) cells were incubated with 50 PFU of virus/cell in 2 ml of DMEM/10% FCS for 2 h at 37°C. Ten milliliters of complete DMEM/10% FCS were added, and the incubation was continued for 24 h at 37°C. Virus-infected C57SV cells (2 × 106) were labeled with 100 μCi Na2[51Cr]O4 (DuPont, Boston, MA) in 5% FCS/RPMI 1640 for 120 min at 37°C. To demonstrate that cell lysis is due to CTL, effector cells were treated with anti-CD8 (AD4) or anti-CD4 (GK1.5) Ab and Low Tox-M rabbit complement (Accurate Chemical, Westbury, NY).
For IFN-γ ELISPOT assay (17) liver mononuclear cells were isolated on day 6 after virus infection and seeded with YAC-1 targets into 96-well ELISPOT plates. To prepare the plates, 100 μl of 10 μg/ml anti-IFN-γ (R4-6A2; PharMingen) in PBS was pipetted per well into Multiscreen 96-well filtration plates (Millipore, Bedford, MA), followed by incubation overnight at 4°C. Plates were washed three times with PBS, and a suspension of 5 × 105 YAC-1 cells and 1 × 105 liver mononuclear cells in 200 μl of RPMI 1640/10% FCS were added to each well and incubated for 24 h at 37°C. Medium was aspirated, and plates were washed three times in PBS containing 0.05% Tween-20. Into each well 100 μl of a solution containing 0.05% Tween-20, 1% BSA, and 5 μg/ml biotin-conjugated anti-IFN-γ Ab (XMG1.2, PharMingen) in PBS were pipetted, and plates were incubated overnight at 4°C. A solution containing a 1/400 dilution of 1 mg/ml avidin peroxidase (Sigma) in PBS containing 0.05% Tween-20 and 1% BSA was prepared, and 100 μl was pipetted into each well. After incubation for 2 h at room temperature, plates were washed four times in PBS/0.05% Tween-20. Into each well 200 μl of ABE solution (Zymed, South San Francisco, CA) were pipetted, followed by incubation for 15 min in the dark. Plates were washed in double-distilled H2O and dried for 2 h, and spots were counted under microscope. To demonstrate that cytokine secretion is due to NK cells, effector cells were treated before assay with anti-NK1.1 Ab PK 136 and Low Tox complement (16).
RT-PCR assay for cytokines, histology, and TUNEL staining
Total RNA was extracted from liver tissue by the phenol/chloroform method using the RNAzol B kit (Tel-Test, Friendswood, TX). Five micrograms of RNA was reverse transcribed to cDNA in a 50-μl reaction mixture using Superscript II RNase H− reverse transcriptase and random primers (Life Technologies, Grand Island, NY) according to the manufacturer’s instructions. For PCR, the equivalent amount of cDNA product (5 μl) was amplified in a 50-μl reaction mixture containing 10 mM Tris (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.2 mM dNTP, 2.5 U Taq DNA polymerase (Perkin-Elmer, Norwalk, CT), and 1 μM of each specific primer. The amplification was performed in a Thermoline Gene E thermocycler (Techne, Cambridge, U.K.) set at 1 min each at 94, 58, and 72°C for 35 cycles, followed by an extension at 72°C for 10 min. After amplification, PCR products were electrophoresed on a 2% agarose gel and visualized by ethidium bromide staining under UV illumination. Primers for IFN-γ and β-actin were obtained from Stratagene (La Jolla, CA).
For histology, liver tissue was fixed in 10% neutral buffered formalin and embedded in paraffin. Five-micron sections were affixed to slides, deparaffinized, and stained with hematoxylin and eosin to determine morphologic changes. Apoptotic cells were visualized by TUNEL staining in deparaffinized sections using an in situ cell death detection kit purchased from Roche.
NK1.1+ cells are required for induction of liver injury in virus-infected mice
Whereas NK and NKT cells constitute minor cell populations in the peripheral lymphoid organs, they are relatively abundant in the liver (5, 7). Fig. 1 shows that mononuclear cells isolated from livers of normal C57BL/6 mice contain 7.1% NK1.1+ CD3− cells, 16.2% NK1.1+ CD3+ cells, and 33.6% CD3+ NK1.1− cells. In contrast spleens contain 3.6% NK1.1+ CD3− cells, 1.4% NK1.1+ CD3+ cells, and 31.8% CD3+ NK1.1− cells. Specific Abs are able to selectively eliminate the two NK1.1+ cell populations in vivo. Thus injection of anti-NK1.1 Ab PK136 eliminates both NK1.1+ CD3+ and NK1.1+ CD3− cells (Fig. 1,C), whereas anti-AsGM1 Ab causes almost complete disappearance of NK1.1+ CD3− cells while sparing most NK1.1+ CD3+ cells (Fig. 1 D). The two Abs can therefore be used to assess the relative contributions of the two NK cell populations during immune responses.
To induce a viral liver infection, a type 5 adenovirus with deletions in the E1 and E3 regions, containing the Escherichia coli β-galactosidase gene was chosen (14). Intravenous injection of 1–2 × 109 PFU virus had been reported to cause infection and viral gene expression in the liver (18). This, in turn, causes heavy mononuclear cell infiltration, associated with liver injury, indicated by an increase in liver enzyme ALT in serum. To investigate the role of the two NK cell populations in virus-induced liver injury, mice were treated with either anti-NK1.1 or anti-AsGM1 and infected with virus. The results presented in Fig. 2 show that serum ALT levels are suppressed in animals treated with either anti-NK1.1 or anti-AsGM1 Ab. In contrast, injection of mouse or rabbit control Abs had no demonstrable effect. These results indicate that cells expressing Ags NK1.1 or AsGM1 are required for induction of liver injury. To investigate the mechanism by which liver injury is induced, livers were sectioned and assayed for apoptotic hepatocytes by TUNEL staining. The results presented in Table I show that livers of normal mice injected with virus contain a substantial number of apoptotic hepatocytes. Treatment with anti-NK1.1 results in suppression of the number of apoptotic cells to baseline levels in control mice. These results suggest that hepatocyte death following the viral infection requires the presence of NK1.1+ cells.
|Mouse .||Virusa .||Abb .||Apoptotic Cellsc .||.|
|.||.||.||Per ×20 field ± SE .||p Value .|
|C57BL/6||−||2 ± 1.4|
|C57BL/6||+||15 ± 3.1|
|C57BL/6||+||α-NK1.1||1.8 ± 0.84||<0.001d|
|SCID||−||1.5 ± 1.2|
|SCID||+||16 ± 2.2|
|SCID||+||α-AsGM1||6 ± 1.5||<0.01e|
|Mouse .||Virusa .||Abb .||Apoptotic Cellsc .||.|
|.||.||.||Per ×20 field ± SE .||p Value .|
|C57BL/6||−||2 ± 1.4|
|C57BL/6||+||15 ± 3.1|
|C57BL/6||+||α-NK1.1||1.8 ± 0.84||<0.001d|
|SCID||−||1.5 ± 1.2|
|SCID||+||16 ± 2.2|
|SCID||+||α-AsGM1||6 ± 1.5||<0.01e|
Animals received 2 × 109 pfu i.v. on day 0 and livers were harvested on day 6.
Ab was injected daily, starting 1 day before virus infection.
Average number of apoptotic cells on a ×20 microscopic field from several sections made from two mice.
Compared to virus-infected C57BL/6 mice using Student’s t test.
Compared to virus-infected SCID mice using Student’s t test.
NK cells mediate liver injury in the absence of conventional T cells
The finding that injection of anti-NK1.1 and anti-AsGM1 Abs inhibits virus-induced liver injury raises the possibility that NK cells, rather than T cells, are the principal effectors causing initial liver injury. To examine this, use was made of mouse strains devoid of T cells but possessing NK cells. Fig. 3,A shows that injection of virus into nude mice induces serum ALT values. However, to reproducibly generate this effect, animals were primed with poly(I:C) to activate NK cells, as different batches of mice, not primed with poly(I:C), gave variable results. Treatment of poly(I:C)-primed mice with either anti-NK1.1 or anti-AsGM1 Ab suppressed the virus-induced increase in serum ALT (Fig. 3 A), consistent with the hypothesis that in nude mice effector cells expressing NK1.1 and AsGM1 cause injury in the infected liver.
Nude mice, while lacking conventional T cells, contain double-negative, CD3dim-staining T cells as well as B cells (5). It therefore cannot be excluded that respective cell populations participate in the NK cell-mediated response to the viral infection. To examine this possibility, SCID mice, lacking CD3dim-staining cells as well as B cells, were infected with virus and assayed for serum ALT values. It is shown in Fig. 3,B that SCID mice respond with a substantial increase in serum ALT after the viral infection. Interestingly, this response does not require activation of NK cells by poly(I:C) and is completely suppressed by injection of anti-AsGM1. TUNEL staining of liver sections reveals an abundance of apoptotic hepatocytes in the liver of infected SCID mice that is significantly suppressed in mice treated with anti-AsGM1 (Table I). Therefore, neither T cells, B cells, nor NKT cells are required for NK-mediated liver injury. Moreover, the finding that in SCID mice NK cells do not have to be activated by poly(I:C) suggests that in nude mice regulatory mechanisms can cause suppression of NK activity. It is interesting that in SCID mice viral infection initiates a slow increase in ALT values with kinetics similar to those in normal mice (Figs. 2 and 3). It therefore appears that even in SCID mice NK cells are activated by the infection before they induce liver injury.
NK1.1+AsGM1+ cells stimulate virus-specific T cell responses
The demonstration that NK cells mediate virus-induced liver injury in nude and SCID mice raises the question as to the function of these cells in normal mice, in particular as it relates to the putative role of cytotoxic T cells in liver injury. Fluorometric analysis of stained cells from liver infiltrates of infected normal mice reveals that virus infection induces a dramatic increase in mononuclear cells in the liver (Fig. 4,A). Whereas the increase in NK1.1+ cells appears to be quite moderate, there is a dramatic increase in CD4+ and CD8+ T cells (Fig. 4,B). It was therefore important to test infected animals for priming of virus-specific T cells. To monitor priming of CD4+ cells, infected mice were challenged into the footpad with virus, followed by evaluation of footpad swelling to assess DTH reactivity. Fig. 5,A shows that virus-primed mice mount a robust DTH response, pointing to sensitization of Th1 cells. To assay for priming of cytotoxic T cells, spleens from virus-injected mice were restimulated in vitro, followed by assay of cytolytic activity. It is shown in Fig. 5 B that cultures from primed mice express high virus-specific cytolytic activity. Treatment of the effector cells with anti-CD8 and complement before the cytotoxicity assay inhibited cytolytic activity; in contrast, treatment with anti-CD4 Ab had no detectable effect. These results demonstrate that virus infection sensitizes virus-specific CD8+ CTL.
The accumulation of CD8+ cells in infected livers and priming of virus-specific CTL in the spleen are consistent with the idea that CTL play a role in virus-induced liver injury (14), but these results also raise the question of why elimination of cells expressing NK1.1 and AsGM1 inhibits liver injury (Fig. 2). A plausible explanation would be that NK cells mediate a dual function, namely the early lysis of hepatocytes in infected liver, followed by stimulation of virus-specific T cell responses. To examine this, mice were treated with anti-NK1.1 or anti-AsGM1, injected with virus, and tested for DTH and CTL responses. The results in Fig. 5,C show that Ab-treated mice respond with a severely suppressed DTH reaction. Similarly, spleen cells from Ab-treated mice mount a lower virus-specific CTL response than cells from control mice (Fig. 5D). These results suggest that cells expressing Ags NK1.1 and AsGM1 play a stimulatory role in the induction of virus-specific CD4+ and CD8+ cells.
IFN-γ is produced by NK cells and is required for optimal priming of virus-specific CTL
The finding that cells expressing NK1.1 and AsGM1 stimulate the priming of virus-specific CD4+ and CD8+ cells raises the question as to the mechanism by which this may occur. A plausible mode of action would be if NK cells secrete a Th1 cytokine, e.g., IFN-γ, which facilitates T cell priming. To examine this possibility, mononuclear cells isolated from the livers of infected mice were assayed for secretion of IFN-γ by ELISPOT assay. The results presented in Fig. 6 demonstrate that cells isolated from the livers of virus-infected mice secrete IFN-γ upon incubation with NK target YAC-1. In contrast, cells isolated from control mice contain very few cells able to secrete IFN-γ when incubated with YAC-1 targets. Treatment of liver cell infiltrate from infected mice with anti-NK1.1 and complement before incubation with YAC-1 targets significantly decreased the number of IFN-γ-secreting cells. Therefore, the majority of cytokine-secreting cells were NK1.1+ cells. Based on these results one would predict that virus infection should induce IFN-γ transcripts in the liver, and elimination of NK cells should inhibit this effect. It is shown in Fig. 7,A that in the liver of normal mice, IFN-γ mRNA was induced by day 2 and even more so on day 6 after infection. Treatment of mice with either anti-NK1.1 or anti-AsGM1 before infection strongly suppressed the induction of IFN-γ transcripts, consistent with the idea that NK cells are producers of IFN-γ at early times after infection. Data from T cell-deficient mice support this conclusion. In the livers of nude mice IFN-γ mRNA was increased by the infection well above the level present in poly(I:C)-primed mice, and this effect was suppressed by injection of anti-NK1.1 and anti-AsGM1 Ab (Fig. 7,B). Similar results were seen in SCID mice. Here again, virus injection induced IFN-γ mRNA in the liver, and this effect was suppressed in anti-AsGM1-treated animals (Fig. 7 C). Taken together these results suggest that NK cells constitute a source of IFN-γ secretion at early times after the infection.
The demonstration that NK1.1+ and AsGM1+ cells secrete IFN-γ and that elimination of these cells leads to an inefficient priming of virus-specific T cells supports the hypothesis that NK cells act via the secretion of IFN-γ. For this case, then, elimination of this cytokine should interfere with an efficient priming of CTL. To examine this, mice were injected with anti-IFN-γ Ab and infected with virus, and their spleens were restimulated with virus in vitro. The results presented in Fig. 8 show that spleen cells from anti-IFN-γ-treated mice express lower CTL activity than spleen cells from mice injected with control Ab. Therefore, the presence of IFN-γ facilitates the priming of virus-specific CTL.
The liver contains a significant number of NK and NKT cells (4, 5, 6, 7, 8), raising questions about the roles of the two cell populations in liver infections. Both cell types have been shown to mediate cytolytic activity (7, 8, 19, 20, 21) and to secrete cytokines (2, 3, 7, 15, 22), enabling them to perform distinct functions in response to infections. To elucidate the roles of the two cell types, a virus infection model was chosen in which i.v. injection of a replication-defective virus causes efficient infection and viral gene expression in the liver (14, 18). Numerous reports demonstrated that the T cell-mediated immune system responds to adenoviral gene expression in the liver, resulting in elimination of infected hepatocytes (14, 18, 23, 24, 25, 26, 27). This model was therefore suitable to answer questions about the relative contributions of NK cells, NKT cells, and T cells to the response to viral liver infection.
It is shown that NK cells in SCID mice can cause significant liver injury, reflected in high serum ALT and hepatocyte apoptosis, as demonstrable by TUNEL staining. These results clearly identify NK cells as a cell population capable of causing liver injury in the absence of T cells, B cells, and NKT cells. Given the demonstration that adenoviral gene expression in liver reaches a maximum between 36 and 48 h after infection (28), it is surprising that no elevated ALT values are demonstrable by day 3. This suggests that liver NK cells exist in a dormant state and have to be activated before they cause hepatocyte injury. It is likely that IL-12 and IL-18, produced by APC in the liver, are involved in this process (19, 29, 30). A surprising finding in this respect is that NK cells in nude mice require preactivation by poly(I:C) to reproducibly cause virus-induced liver injury. This may point to regulatory mechanisms, perhaps by NKT cells or CD3dim-staining cells, that modulate NK cell activity.
Our observation that serum ALT values increase in SCID mice with time kinetics similar to those in normal mice raises the possibility that even in the presence of T cells an early increase in ALT values is to a large extent due to the action of NK cells. In support it is shown that injection of Abs anti-NK1.1 and anti-AsGM1 into normal mice inhibits the increase in ALT. The demonstration that anti-AsGM1 leaves NKT cells relatively unaffected while efficiently eliminating NK cells suggests that NK cells constitute a major cell type responsible for early injury in the virus-infected liver of immunocompetent animals.
The conclusion that in normal mice NK cells are responsible for early liver injury is unexpected, as NKT cells constitute a far larger population of cells in liver with well-documented cytolytic activity (20, 21, 31, 32, 33). It is therefore important to stress that Ab ablation experiments do not completely exclude participation of NKT cells in cytotoxic liver injury. It is possible, for example, that activated NKT cells with high cytolytic activity increase cell surface expression of AsGM1, thereby making them sensitive to elimination by anti-AsGM1 in vivo. A most intriguing possibility regarding the function of NKT cells in liver is raised by the recent observation that NKT cells may stimulate NK cells by secretion of IFN-γ (34). Therefore, NKT cells could provide helper functions for NK cells by stimulating their activity in the liver. These functions could provide a rationale for the relative abundance of NKT cells in this organ.
The demonstration that NK cells play an important role in the early response to infection of the liver is in agreement with results from related models. Thus, NK cells have been shown to mediate the clearance of vaccinia/IL-2 infections in nude mice (35), and removal of NK cells from murine CMV-infected normal mice prompts an increase in the severity of viral hepatitis (2). In these experiments replication-competent virus had been employed, leaving open the question of how NK cells eliminate the infectious pathogen, i.e., by inhibiting viral replication via cytokine secretion or by cytotoxicity. Our approach of using a replication-defective virus clearly shows that hepatocyte apoptosis is induced by NK cells. The question, however, that remains to be resolved is whether hepatocyte lysis by NK cells is specifically directed against virus-infected cells. Our attempts at demonstrating that ex vivo NK cells from the liver of infected mice are virus specific have not been successful (unpublished observations). It is therefore quite possible that liver injury induced by NK cells is not specifically directed against virus-infected cells.
Demonstration of NK-mediated cytotoxicity in infected liver raises the question of whether this constitutes the principal function of these cells. We show that elimination of NK1.1+ and AsGM1+ cells interferes with efficient priming of virus-specific CTL and DTH responses, suggesting that NK cells mediate additional functions. Using an ELISPOT assay we demonstrate that NK cells from virus-infected livers, when stimulated with NK target YAC-1, secrete IFN-γ. Moreover, infection of normal or T cell-deficient mice with virus induces IFN-γ mRNA, which is inhibited in NK-depleted mice. Therefore, NK cells produce IFN-γ in response to the infection, raising the question of whether it is this process that causes stimulation of T cell priming. Support for this mechanism is provided by the finding that injection of anti-IFN-γ Ab inhibits CTL priming. This effect, however, is not complete, leaving open the possibility that additional cytokines, released by NK cells, are also involved.
Another unresolved question is the possible role of NKT cells in cytokine secretion. The observation that NKT cells, when stimulated via their TCR, very rapidly engage in the secretion of IFN-γ, which, in turn, stimulates NK cells to secrete this cytokine, could provide a mechanism for activation of NK cells in the stimulation of T cell responses (34, 36, 37). There are a limited number of reports supporting this attractive hypothesis. NKT cells have been shown to stimulate the induction of CD8+ effector cells in Toxoplasma infection (36), and priming of influenza virus-specific CTL was found to be inhibited in NK cell-depleted mice (38). The in vivo induction of mouse CTL, specific for Plasmodium, involving infection of hepatocytes was also reported to depend on the presence of NK cells (39). These findings together with the ones reported here provide strong support for the idea that NK cells in conjunction with NKT cells may play an important stimulatory role in the induction of Th1 and CTL responses.
Our finding that NK cells secrete IFN-γ following adenovirus injection is in line with previous reports showing that NK cells secrete IFN-γ in response to viral infections, and this causes suppression of virus replication (2, 40, 41). This, then, raises the question of the mechanism by which IFN-γ stimulates priming of T cell responses. Several possibilities exist. It is well documented that MHC expression on hepatocytes is almost undetectable and is stimulated by IFN-γ (13). Therefore, NK cell-derived IFN-γ could increase MHC expression on hepatocytes, which, in turn, could inhibit NK function while stimulating T cell responses. In support of this, it has been shown that isolated hepatocytes are able to induce CTL responses in vitro; hence, hepatocytes can act as APC (42). An alternative action of IFN-γ could be direct stimulation of CD4+ or CD8+ cells during priming or restimulation in the liver. Yet another mechanism is that IFN-γ can stimulate immature liver dendritic cells to migrate to the peripheral lymphatics and to induce a Th1 response (43).
In summary, we show here that NK cells play a dual role in the cell-mediated immune response to adenoviral infection in the liver. They lyse hepatocytes in the virus-infected liver and stimulate, probably by their ability to secrete IFN-γ, the induction of a virus-specific, T cell-mediated immune response. It is therefore suggested that NK cells constitute a very important component required for optimal responses to viral liver infections and thereby may hold the key to a successful cell-mediated immune response in this organ.
This work was supported by U.S. Public Health Service Grants CA59318, AI43954, and AI40038.
Abbreviations used in this paper: DTH, delayed-type hypersensitivity; AsGM1, asialo-GM1; ALT, alanine aminotransferase; ELISPOT, enzyme-linked immunospot; lacZ, β-galactosidase.