Introduction

Hepatocytes are more susceptible to death receptor-mediated apoptosis than other normal cells [6, 9]. For example, previous studies have demonstrated that severe apoptotic liver damage occurs in mice after administrating agonistic anti-Fas monoclonal antibodies, and that the apoptotic cell death mediated by Fas-Fas ligand (FasL) interactions plays an essential role in the development of hepatitis [8, 19, 33]. Furthermore, recent studies have shown that the treatment with recombinant tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) induces apoptosis in cultured normal hepatocytes [15]. This is despite the fact that TRAIL has been previously reported to induce apoptosis in various tumor cells but not in normal cells [36, 53]. In contrast to normal hepatocytes, most hepatocellular carcinomas (HCCs) are quite resistant to death receptor-mediated apoptosis when the cell surface death receptor is cross linked in vitro with either agonistic antibodies or soluble death ligand proteins [11, 19, 29, 31, 44, 45, 60]. The resistance of HCCs against death receptor-mediated apoptosis might be an essential step in the escape from the host immune surveillance; however, it has not been directly demonstrated that HCCs are actually resistant to cytotoxic T-cell- or natural killer (NK) cell-mediated apoptotic death.

Natural killer cells are a distinct subset of large granular lymphocytes, which possess the ability to lyse certain primary tumor cells, tumor cell lines, transplanted allogenic cells, and virus-infected cells without prior sensitization [50]. In particular, it is well known that NK cells effectively kill many leukemia cells, and that the process is primarily mediated by the secretory granules which contain perforin and granzymes [37, 42, 50, 59]. Perforin can induce the necrosis of the target cells and granzymes can induce the apoptosis of the target cells [24, 37, 46]. The cytotoxicity of NK cells against tumor cells has been commonly assessed by a standard 4-h 51Cr release assay [37, 50, 61]; however, many tumor cells, particularly solid tissue-derived malignant cells, appear to be resistant against NK cells when the cytotoxicity was measured using the classical 51Cr release assay. Moreover, the “NK-resistant” tumor cells were rapidly eliminated in vivo, suggesting that NK cells can also kill tumor cells through other mechanism that cannot be assessed using the classical NK cytotoxicity method [10, 38, 50, 55, 57]. Recently, Vujanovic et al. showed, using a [3H]-thymidine release assay which measures apoptotic target cell death, that NK cells can also effectively kill many solid tissue-derived tumor cells [51]. These results suggest that the nonsecretory/apoptotic pathway appear to play more important roles when NK cells eliminate solid tumor cells. Indeed, recent studies have shown that the nonsecretory/apoptotic pathway is mediated by the FasL/Fas, TNF/TNF receptor, and TRAIL/TRAIL receptor interactions, which trigger a relatively well-known death pathway involving caspase activation [1, 17, 23, 28, 41, 62]. The death receptor-mediated apoptotic pathway is also involved in the cytotoxicity of NK cells against many leukemia cells [34, 35].

Recent studies have demonstrated that NK cells can destroy many solid tissue-derived malignant cells, such as melanoma, breast cancer, lung cancer, gastric cancer, colon cancer, renal cancer, and ovarian cancer cell lines, and that this process is mediated primarily by death receptor/ligand interactions [17, 23, 51]; however, it is not clear whether or not human NK cells directly influence the destruction of human HCCs.

This study investigated whether NK cells can destroy HCCs like other solid tumors, and characterized how they kill those target cells. For this purpose, two distinct cytotoxic assays, a 4-h 51Cr-release assay and a 2-h [3H]-thymidine release assay, which selectively measures the extent of necrotic or apoptotic target cell death, respectively, were used independently. In order to determine if the NK cell-induced apoptotic target cell death is dependent on the secretory granules and/or the cell surface molecules, the NK cells were fixed with 0.5% paraformaldehyde and the cytotoxicity was then measured. In this study, NK cells appeared to use two distinct killing mechanisms in different ways depending on the HCC target cell types.

Materials and methods

NK cell preparation

Natural killer cells were purified from the whole blood of a healthy volunteer by negative selection, using a RosetteSep NK enrichment antibody cocktail (StemCell Technologies, Vancouver, B.C.), which contained anti-CD3, anti-CD4, anti-CD19, anti-CD36, and anti-CD66b antibodies. Briefly, 1 ml of the whole blood from a single donor was mixed with 50 μl of the RosetteSep NK enrichment cocktail, and incubated for 20 min at room temperature. The blood sample was then diluted with the same volume of phosphate buffered saline (PBS, pH 7.4) containing 2% of fetal bovine serum (FBS, Gibco BRL, Grand Island, N.Y.). The diluted sample was layered on the top of a Ficoll-paque (Amersham Pharmacia Biotech, Uppsala, Sweden) and centrifuged for 20 min at 1600 rpm at room temperature. The NK cell layer was then collected and the enriched NK cells were washed three times with PBS containing 2% FBS. The purified NK cells were >90% CD56+ CD16+/CD3-, and the NK cells were maintained for a few days prior to use in RPMI 1640 media containing 10% FBS and 1 ng/ml of recombinant IL-2 (Endogen, Woburn, Mass.).

Cell lines and culture

Two HCC cell lines, HepG2 (ATCC HB8065) and Hep3B (ATCC HB8064), were used as the target cell lines. HepG2 and Hep3B cells were chosen since they represent the Fas+ and Fas− hepatoma cell lines, respectively. A chronic myelogenous leukemia cell line, K562 (ATCC CCL243), and an acute lymphoblastic leukemia cell line, MOLT-4 (ATCC CRL1582), were used as the NK-sensitive target cell lines, and a B-cell lymphoma cell line Daudi (ATCC CCL213) was used as the NK-resistant target cell lines. The HepG2 and Hep3B cell lines were maintained in MEM containing 10% FBS, the K562 was maintained in DMEM containing 10% FBS, and the Daudi and MOLT-4 cell lines were maintained in RPMI 1640 medium containing 10% FBS.

51Cr release assay

The NK cell-mediated target cell death was assessed using the standard 51Cr release assay [37, 50, 61]. For labeling, 3×103 cells were incubated with 10 μCi of 51Cr (NEN, Boston, Mass.) in a 96-well microtiter plate for 60 min at 37°C. The cells were then washed three times with the culture media without 10% FBS. The 51Cr-labeled target and NK cells were mixed at the indicated effector to target (E:T) ratios. After 4 h co-culturing with the NK cells, the cell free supernatant was collected and its radioactivity was measured with a gamma counter (Beckman, Calif.). The percentage of specific 51Cr release was calculated by the following formula: % cytotoxicity=[(experimental 51Cr release−spontaneous 51Cr release)/(maximum 51Cr release−spontaneous 51Cr release)]×100. For the control experiments, the target cells were incubated either in the culture medium alone to determine the spontaneous release or in a mixture of 2% Triton X-100 to define the maximum 51Cr release. The spontaneous release was always <10% of the maximum release. All the assays were performed in triplicate.

[3H]-thymidine release assay

The level of NK cell-mediated apoptotic target cell death was measured using a 2-h [3H]-thymidine release assay (the JAM test) [4, 26]. For labeling, 2×104 cells were incubated with 20 μCi of [3H]-thymidine (37 MBq/ml, NEN) for 20 h at 37°C in a 96-well microtiter plate. The cells were then washed three times with the culture media without 10% FBS. The [3H]-thymidine-labeled target cells and NK cells were mixed at the indicated E:T ratios. After 2 h incubation, the cells and their medium were aspirated onto glass fiber filters (size 90×120 mm; Wallac, Turkr, Finland) using a semi-automated 96-well harvester (Tomtec, Hamden, Conn.). The filters were washed, dried, and sealed with melt-on scintillator sheets (Meltilex, Wallac, Turkr, Finland), and their radioactivity was measured using a beta counter (Wallac, Turkr, Finland). The percentage of apoptotic cell death was calculated using the following formula: % DNA fragmentation=[1−(experimental value/control value)]×100. The control value was determined by incubating the target cells in the culture medium alone.

NK cell fixation

In order to block the release of granules, the NK cells were fixed with 0.5% paraformaldehyde, as previously described [21, 51]. Briefly, the NK cells were incubated with RPMI 1640 containing 0.5% paraformaldehyde (Sigma Chemical Co., St. Louis, Mo.) for 20 min and washed twice with PBS. The paraformaldehyde concentration and incubation time was minimized to minimize the chance of any adverse effects.

Western blot analysis

Caspase-3 activation in target cells after co-incubation with NK cells was detected by western blot analysis. The target cells were lysed with a lysis buffer (10 mM Tris-HCl, pH7.4 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 1 mM PMSF, 15 μg/ml leupeptin, 2 mM NaF, 2 mM NaVO4), and the lysates were separated on SDS-polyacrylamide gels. The protein bands were transferred onto PVDF membranes (Pierce, Rockford, Ill.). The membranes were then blocked with 5% BSA in PBS containing 0.1% Tween-20 (PBST) for 2 h, incubated with either anti-caspase-3 or anti-cleaved caspase-3 antibodies (Cell Signaling Technology, Beverly, Mass.) for 4 h at room temperature [7, 32], and washed three times with PBST. Subsequently, the membranes were incubated with peroxidase-conjugated affinipure goat anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, Pa.) for 2 h, and washed with PBST. The blots were finally visualized using a Supersignal WestDico chemiluminescent substrate.

Flow cytometry

The cell surface Fas antigen was quantified by flow cytometric analysis. Five times 105 of the HepG2 and Hep3B cells were washed twice with ice-cold PBS containing 0.05% BSA. The cells were incubated with anti-Fas monoclonal antibodies (DX2; Calbiochem, La Jolla, Calif.) for 30 min at 4°C. After twice washing with 0.05% BSA-PBS, the cells were incubated with FITC-conjugated goat anti-mouse IgG (Becton Dickinson Bioscience, Lincoln Park, N.J.) for 30 min and again washed twice. The level of NK cell-induced caspase-3 activation in the target cells was quantified similarly by flow cytometric analysis. After the HepG2, Hep3B, MOLT-4, and K562 cells were co-cultured with the NK cells for 2 h, the target cells were washed twice with ice-cold PBS containing 0.05% BSA. The cells were then incubated with FITC-conjugated anti-caspase-3 mAb (BD Pharmingen) and PE-conjugated anti-CD56 mAB (Becton Dickinson Bioscience) for 30 min at 4°C and washed twice with 0.05% BSA-PBS. The cells were analyzed using a FACScan flow cytometer.

RT-PCR

Reverse transcriptase polymerase chain reaction (RT-PCR) was performed to study the TRAIL system expression with primers specific for the TRAIL receptor mRNAs. The total RNAs were extracted from the HepG2 and Hep3B cells using an RNAeasy Kit (Qiagen, Santa Clara, Calif.). The integrity of the total RNA isolated was confirmed using 1.5% agarose gel electrophoresis. In order to synthesize the cDNA, 1 μg of each RNA sample was mixed with 100 ng of a random hexamer (Amersham Pharmacia Biotech), 6 μl of a 5X first-strand buffer, 12 μl of 2.5 mM dNTPs, and 200 units of murine leukemia virus-reverse transcriptase (MMLV-RT; Gibco BRL), and incubated at 42°C for 80 min. The reaction mixture was boiled at 95°C for 5 min, chilled quickly on ice, and then used for PCR without any further manipulation. The cDNA was amplified using a GeneAmp PCR System (Perkin-Elmer, Norwalk, Conn.) in 25 μl of a reaction mixture (2.5 μl of cDNA, 2 μl of 2.5 mM dNTPs, 20 pmol primer, 2.5 μl of 10X PCR buffer, 13.8 μl of distilled water, and 1 unit of Taq polymerase (Takara, Shiga, Japan). The amplification protocol involved 5 min denaturation at 94°C, followed by 30~35 cycles at 94°C for 30 s, 59°C for 1 min, 72°C for 1 min, and finally a 10-min extension at 72°C. The PCR primers used were based on the published sequences as follows: DR4 sense primer, 5′-TTACACCAATGCTTCCAACAAT-3′, and antisense primers, 5′-AGGAGTCAAAGGGCACGATGTT-3′; DR5 sense primer, 5′-GCCTCAT GGTCAATGAGATAAAGGTGGCT-3′, and antisense primer, 5′-CCAAATCTCAAAGTACGC ACAAACGG-3′; DcR1 sense primer, 5′-GGTTTCCACAGTGGCATTGGC-3′, and antisense primer, 5′-GATCCC CAAGACCCTAAAGTT-3′; and DcR2 sense primer, 5′-AGGGATGGTCA AGGTCAGTAAT-3′, and antisense primer, 5′-GATGTCAGCGGAGTCAGCGTCA-3′. The expected sizes of the amplified cDNAs were 680, 502, 475, and 538 bp for DR4, DR5, DcR1, and DcR2, respectively. As an internal control, the β-actin cDNA was similarly amplified using the β–actin sense primer, 5′-CGTGGGCCGCCCTAGGCACCA-3′, and the antisense primer, 5′-TTGGCCTTAGGGTTCAGGGGGG-3′. The size of the amplified β–actin cDNA was expected to be 246 bp.

TRAIL blocking experiment

The NK cells were fixed with 0.5% paraformaldehyde (Sigma) for 20 min, and the fixed NK cells were pretreated with a recombinant TRAIL receptor-Fc fusion protein for 1 h. The labeled Hep3B target cells were then added at different E:T ratios, and the released [3H]-thymidine was measured using a β-counter.

Results

NK cells effectively kill HepG2 and Hep3B cells

The morphologic changes in the HepG2 and Hep3B cells were first observed after co-incubation with purified human primary NK cells. As shown in Fig. 1, most of the target cells exhibited marked morphologic changes that were characteristics of dead cells when they were co-incubated with the NK cells for 4 h at an E:T ratio of 3:1 (Fig. 1). Interestingly, the HepG2 and Hep3B cells exhibited distinct morphologic changes after co-incubation with the NK cells.

Fig. 1
figure 1

Morphologic changes in the HepG2 and Hep3B cells after co-incubation with the natural killer (NK) cells. The target cells were seeded in a 6-well plate and cultured for 1 day in minimum essential medium (MEM) containing 10% fetal bovine serum (FBS). The NK cells were then added to the cultured target cells to an E:T ratio of 3:1. The cells were co-incubated for 4 h in RPMI 1640 media containing 10% FBS. The morphologic changes of the target cells were observed using an inverted microscope

Full size image

A standard 4-h 51Cr release assay was subsequently performed to quantify the cytotoxicity of the purified NK cells against HepG2 and Hep3B. A chronic myelogenous leukemia cell line, K562, was used as the NK-sensitive control cells, and a B-cell lymphoma cell line, Daudi, was used as the NK-resistant control cells. As previously reported [35, 51], the NK cells induced a high level of 51Cr release in the K562 targets in a E:T ratio dependent manner, whereas they induced almost no 51Cr release in the Daudi targets (Fig. 2). The NK cells also induced a comparatively high level of 51Cr release in the HepG2 and Hep3B targets, although less so than the K562 cells. The NK cytotoxicity against the HepG2 and Hep3B targets was also proportional to the E:T ratios. At an E:T ratio of 10:1, approximately 60~70% of the 51Cr was released in both the HepG2 and Hep3B targets. Furthermore, the NK cells from five different donors all appeared to have a similar cytotoxicity against the HepG2 and Hep3B cells (data not shown). These results suggest that human primary NK cells can effectively kill HCCs.

Fig. 2
figure 2

Cytotoxicity of the NK cells assessed by the 4-h 51Cr release assay. The HepG2 and Hep3B cells were cultured for 1 day in a 96-well plate in MEM containing 10% FBS, but the K562 and Daudi cells were immediately seeded in a 96-well plate in RPMI 1640 media prior to the labeling experiments. The target cells were labeled with 51Cr for 1 h, and the NK cells in the RPMI 1640 media containing 10% FBS were added to the 51Cr labeled target cells with different E:T ratios. The cells were co-incubated for 4 h, and the radioactivity was measured with a γ-counter. The percentage cytotoxicity was calculated, as described in “Materials and methods.” The data are presented as a mean of at least three independent experiments (mean±SD)

Full size image

NK cell-mediated apoptotic killing of HepG2 and Hep3B cells

Microscopic observations indicated that most of the HepG2 and Hep3B cells were killed by NK cells within 4 h at an E:T ratio of 3:1 (Fig. 1); however, the 4-h 51Cr release assay results demonstrate that less than 40% of the target cells were lysed by the NK cells under the same conditions (Fig. 2). Since it has previously been reported that the 4-h 51Cr release assay primarily detects necrotic cell death induced by the cytotoxic granules [51, 52], NK cells were next examined to determine if they could induce the apoptosis of the HCC target cells. In order to observe the apoptotic target cell death mediated by NK cells, the extent of DNA fragmentation was first measured using a [3H]-thymidine release assay (JAM test), as previously described [4, 26]. After 2 h co-incubation with the NK cells, the Hep3B and MOLT-4 cells (an acute lymphoblastic leukemia cell line) exhibited a significant amount of DNA fragmentation, as assessed by the release of [3H]-thymidine, whereas the HepG2 and K562 cells showed a small amount of DNA fragmentation (Fig. 3). More DNA fragmentation was induced as the E:T ratio was increased in the Hep3B and MOLT-4 targets. Previous studies have demonstrated that NK cells kill the K562 targets primarily by the necrotic pathway, whereas they kill the MOLT-4 targets by the apoptotic pathway [35]; therefore, the results suggest that the NK cells can induce apoptotic cell death in the Hep3B targets as in the case of MOLT-4, but not in the HepG2 targets.

Fig. 3
figure 3

The NK cell-induced apoptotic cell death of target cells assessed by the [3H]-thymidine release assay. The HepG2 and Hep3B cells were cultured for 1 day in a 96-well plate in MEM containing 10% FBS, but the K562 and MOLT-4 cells were immediately seeded in a 96-well plate in RPMI 1640 media prior to labeling experiments. The target cells were labeled with [3H]-thymidine for 20 h, and the NK cells in the RPMI 1640 media containing 10% FBS were added to the [3H]-thymidine labeled target cells with different E:T ratios. The cells were co-incubated for 2 h, and the radioactivity was measured with a β-counter. The percentage apoptosis was calculated as described in “Materials and methods.” The data are presented as a mean of at least three independent experiments (mean±SD)

Full size image

In order to confirm the apoptotic cell death induced by NK cells, procaspase-3 was next examined to determine if it was activated upon co-incubation of the target cells with the NK cells. Among the numerous caspase family members that have been identified thus far, caspase-3 is one of the most common and important effectors of the apoptotic process [5, 58]. After 4 h co-incubation, the cells were analyzed for their caspase-3 activation by flow cytometry. In order to exclude cross-contamination between the NK cells and target cells as much as possible, PE-conjugated anti-CD56 antibodies were used as an NK cell marker together with the FITC-conjugated anti-active caspase-3 antibodies. Caspase-3 activation was observed in the MOLT-4 and Hep3B targets among the CD56− cells, as shown in Fig. 4A. The active-caspase-3 positive cell population was increased from 1.4 to 18.0% and from 30.9 to 42.9% in the MOLT-4 and Hep3B targets, respectively, but there was no significant change in the K562 and HepG2 targets. Interestingly, in the K562-NK co-incubation system, the active-caspase-3 positive cell population increased to approximately one-third of the CD56+ cells, suggesting that caspase-3 was activated in the NK cells. This observation is consistent with a previous report showing that the K562 and endothelial cells can induce the apoptosis of NK cells [39]. The western blot analysis results shown in Fig. 4B also indicate that caspase-3 is activated in the MOLT-4 and Hep3B targets, but not in the HepG2 targets. These results suggest that NK cells induce the apoptotic cell death in the Hep3B cells, but not in the HepG2.

Fig. 4A, B
figure 4

Activation of caspase-3 in the target cells after co-incubation with NK cells. A Flow cytometric analysis of the active caspase-3 in the K562, MOLT-4, HepG2, and Hep3B cells after co-incubation with the NK cells. The target cells were seeded in a 12-well plate and co-cultured with the NK cells with an E:T ratio of 3:1 for 2 h. The cells were stained with FITC-conjugated anti-caspase-3 antibodies and PE-conjugated anti-CD56 antibodies and analyzed by flow cytometry. B Western blotting analysis of active caspase-3 in the K562, MOLT-4, HepG2, and Hep3B cells after co-incubation with the NK cells. The target cells were seeded in a 6-well plate and co-incubated with the NK cells with an E:T ratio of 3:1 for 2 h. The cells were lysed and Western blotting was performed with either anti-procaspase-3 or anti-cleaved active caspase-3 antibodies

Full size image

Effect of NK cell fixation on their cytotoxicity

Natural killer cells utilize two different pathways to eliminate specific target cells, a granule-dependent (perforin/granzyme-mediated) and a granule-independent (death receptor-mediated) process [1, 16, 25, 51, 54]. Previous studies have shown that the mild fixation of NK cells and macrophages eliminates their secretory activity; however, it preserves both the structure and function of the cell-surface receptors [21, 51]. In order to determine if the NK cell-induced apoptotic target cell death is dependent on the secretory granules and/or the cell surface molecules, the NK cells were fixed with 0.5% paraformaldehyde and the cytotoxicity was then measured. As shown in Fig. 5, the fixed NK cells had no detectable activity in the 4-h 51Cr release assay against both the HepG2 and Hep3B targets; however, the fixed NK cells were as effective as the untreated NK cells in the [3H]-thymidine release assay (Fig. 6), which suggests that the apoptotic cell death of the Hep3B targets is granule independent and possibly mediated by the death ligands on the NK cell surface.

Fig. 5A, B
figure 5

Cytotoxicity of the fixed NK cells against the HepG2 and Hep3B cells assessed by a 4-h 51Cr release assay. A HepG2 and B Hep3B cells were seeded in a 96-well plate and cultured for 1 day with MEM containing 10% FBS. The target cells were labeled with 51Cr for 1 h, and the fixed NK cells in RPMI 1640 media containing 10% FBS were then added to the 51Cr-labeled target cells with different E:T ratios. The fixed NK cells and target cells were co-incubated for 4 h, and the radioactivity was measured with a γ-counter. The percentage cytotoxicity was calculated as described in “Materials and methods.” The data are presented as a mean of at least three independent experiments (mean±SD)

Full size image
Fig. 6A, B
figure 6

Fixed NK cell-induced apoptotic cell death of the target cells assessed by the [3H]-thymidine release assay. A HepG2 and B Hep3B cells were seeded in a 96-well plate and cultured for 1 day in MEM containing 10% FBS. The target cells were labeled with [3H]-thymidine for 20 h, and the fixed NK cells in RPMI 1640 media containing 10% FBS were added to the [3H]-thymidine labeled target cells with different E:T ratios. The cells were co-incubated for 2 h, and the radioactivity was measured with a β-counter. The percentage apoptosis was calculated as described in “Materials and methods.” The data are presented as a mean of at least three independent experiments (mean±SD)

Full size image

Involvement of the death receptor/ligand interaction in NK-induced apoptosis of HCCs

In granule-independent, death receptor-mediated apoptosis, the death ligands on the effector cells, such as the Fas ligand (FasL/CD95/Apo-1), the TNF, or the TNF-related apoptosis-inducing ligand (TRAIL/Apo-2), bind to the death receptors on the target cell surface and initiate the apoptotic signaling pathway [1, 2, 17, 23, 28, 34, 62]. This study investigated which of the death receptor/ligand interaction is involved in the NK cell-mediated apoptotic killing of HCCs. The HepG2 cells expressed Fas on the cell surface but the Hep3B cells did not (Fig. 7A), and an agonistic antibody of Fas (CH11) did not, induce apoptosis in either of the target cells (data not shown), which is consistent with previous reports [43, 44]. Furthermore, the morphologic changes in the HepG2 targets, which were observed when they were co-incubated with the NK cells, were unaffected by the pretreatment with a neutralizing Fas antibody, ZB4 (Fig. 7B). These results suggest that a Fas/FasL interaction is not involved in the NK cell-mediated apoptotic death of HepG2 and Hep3B cells.

Fig. 7A, B
figure 7

Effects of Fas/FasL interaction in NK-induced apoptotic killing of HepG2 and Hep3B. A Expression of Fas on the target cells. The expression of Fas on the HepG2 and Hep3B surfaces was determined by flow cytometry. The target cells were stained with anti-Fas Ab (DX2) and fluorescein isothiocyanate (FITC)-conjugated secondary antibodies. B Morphologic change in the HepG2 after treating with neutralizing Fas antibodies. The HepG2 cells were seeded in a 6-well plate and cultured for one day in MEM containing 10% FBS, and ZB4 (2 μg/ml) was then added to HepG2 cells. After 1 h, the NK cells were added to the cultured target cells with an E:T ratio of 3:1, and the target cells were then cocultured with the NK cells for 4 h in the RPMI 1640 media containing 10% FBS. The morphologic changes in the target cells were observed using an inverted microscope

Full size image

Tumor necrosis factor (TNF)-related apoptosis-inducing ligand (also known as Apo-2L) is a member of the tumor necrosis factor (TNF) ligand family that rapidly induces apoptosis in a variety of transformed cell lines [2, 36, 56]. In order to explore the possible involvement of a TRAIL/TRAIL receptor interaction in the NK cell-induced apoptosis of HCCs, the expression of TRAIL receptors was investigated in the HepG2 and Hep3B cells. As shown in Fig. 8A, the mRNA transcripts of the TRAIL receptors could be detected in both target cells; therefore, the effect of a recombinant TRAIL-receptor 1-Fc fusion protein was investigated in order to verify the involvement of the TRAIL/TRAIL receptor interaction in NK cell-induced apoptotic killing of the Hep3B cells. The mildly fixed NK cells were pretreated with the recombinant TRAIL receptor 1-Fc fusion protein and co-incubated with the Hep3B target cells for 2 h. As shown in Fig. 8B, the level of [3H]-thymidine release was significantly inhibited when the NK cells were pretreated with the recombinant TRAIL receptor. Overall, this data suggests that a TRAIL/TRAIL receptor interaction plays an important role in the NK cell-induced apoptotic killing of Hep3B, although another death ligand/death receptor interaction might also be involved in this process. Interestingly, Hep3B and the other HCCs are resistant to the recombinant TRAIL-induced apoptosis in vitro [19, 45].

Fig. 8A, B
figure 8

Involvement of tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL)/death receptor interaction in NK-induced apoptosis of Hep3B. A Expression of the TRAIL receptor mRNAs (DR4, DR5, DcR, and DcR2) in the HepG2 and Hep3B cells determined by reverse transcriptase polymerase chain reaction. The amplified PCR products were analyzed on 1.5% agarose gels. B Effect of TRAIL blocking on the fixed NK cell-induced apoptotic cell death of the Hep3B cells. The percentage apoptosis was calculated as described in “Materials and methods.” The data are presented as a mean of at least three independent experiments (mean±SD)

Full size image

Discussion

The NK cells kill tumor cells using both secretory (perforin/granzyme-mediated) and nonsecretory (membrane bound death ligand-mediated) mechanisms [34, 35, 51]. The NK cell-mediated target cell death via the secretory pathway usually leads to necrosis of the target cells within a short period, as can be determined by a 4-h 51Cr release assay, whereas the death via the nonsecretory pathway usually leads to apoptosis within a short period as can be assessed by a 2-h [3H]-thymidine release assay [34, 35, 37, 50, 51, 61, 63]. In this study we investigated the characteristics and molecular mechanism of NK cell-mediated cytotoxicity against two HCCs: the HepG2 and Hep3B cell lines. The results demonstrated that the NK cell effectively killed the HepG2 and Hep3B cells. The NK cytotoxicity against these HCCs was comparable to that against K562, a NK-sensitive leukemia cell line, when the cytotoxicity was assessed by a 4-h 51Cr release assay (Figs. 2, 5). The NK cells also induced apoptotic cell death in the Hep3B targets, but not in the HepG2 targets, when the cytotoxicity was assessed by a 2-h [3H]-thymidine release assay (Figs. 3, 6); therefore, NK cells appear to use two distinct killing mechanisms in different ways depending on the HCC target cell types. The NK cells kill the HepG2 targets primarily via the necrotic pathway, which can be assessed by a 4-h 51Cr release assay. In contrast, they destroy the Hep3B targets via the apoptotic pathway, which can be assessed by a 2-h [3H]-thymidine release assay, in addition to the necrotic pathway. Furthermore, our data demonstrated that a TRAIL/TRAIL receptor interaction is involved in the NK cell-mediated apoptotic death of Hep3B targets, but the Fas/FasL interaction is not.

It has long been well known that NK cells effectively kill many leukemia cells, and that the process is primarily mediated by secretory granules [37, 42, 50, 59]. More recent studies have suggested that the nonsecretory pathway plays more important role when the NK cells eliminate solid tumor cells, although the secretory pathway is also involved in the elimination of some solid tumor cells [51]. Interestingly, however, our data demonstrated that the secretory death mechanism was primarily utilized when the NK cells destroyed the HepG2 and Hep3B targets (Fig. 3). In accordance with these results, previous studies have demonstrated that the HCCs are generally resistant to death receptor-mediated apoptosis when the death receptor is crosslinked with either agonistic monoclonal antibodies or soluble death ligands in vitro [11, 19, 29, 31, 44, 45, 60]. To the best of our knowledge, this is the first report that clearly demonstrates that NK cells effectively kill HCCs via the secretory/necrotic pathway, although the HCCs are comparatively resistant to the nonsecretory/apoptotic killing mechanism, which is unlike that observed in other solid tumor cell lines [51]. In addition, class-I MHC molecules, the natural ligands of inhibitory killer Ig-like receptors (KIRs), which are down-regulated in many tumor cells, are important regulators of NK cell killing [24, 50, 55]; however, both HepG2 and Hep3B cells appeared to express well the class-I MHC molecules (data not shown), suggesting that the inhibitory receptor/ligand interactions might not be sufficient for controlling the NK cytotoxicity against these HCCs.

Previous studies have shown that HepG2 cells express Fas on their surfaces but are resistant to Fas-mediated apoptosis when the Fas molecules are crosslinked with an agonistic monoclonal anti-Fas antibody, CH11 [43, 44]. In accordance with these results, our data demonstrated that the HepG2 cells appear to be resistant to NK cell-induced apoptotic cell death, although they are quite sensitive to NK cell-mediated granule-dependent necrotic death (Fig. 3). Little [3H] thymidine was released (Fig. 3) and no active caspase-3, which is a hallmark of apoptotic cell death, was generated (Fig. 6) when the HepG2 cells were co-incubated with the NK cells. These results suggest that HepG2 cells are resistant to the apoptotic death induced by FasL and the other death ligands of NK cells. It would be interesting to investigate how the HepG2 cells can evade the NK cell-induced apoptotic death when they express the death receptors. Many tumor cells are known to express proteins that inhibit apoptosis, such as FLIP, XIAP, FAP-1, and PI9 [13, 14, 27, 40, 41, 43]. We also observed that HepG2 cells express the mRNA transcripts of the inhibitory proteins of apoptosis (data not shown); therefore, it is tempting to speculate that these inhibitory proteins might protect the HepG2 cells from NK cell-induced apoptotic death.

Previous studies have demonstrated that Hep3B cells are also resistant to the Fas-mediated apoptosis induced by a crosslinking anti-Fas antibody (CH11). In contrast to HepG2, Hep3B cells are resistant to Fas-mediated apoptosis since they do not express sufficient amounts of Fas on their surfaces. Consequently, Hep3B cells become sensitive to Fas mediated apoptosis when the cells are transfected with the recombinant Fas gene [22]. This suggests that the death receptor-mediated apoptosis signaling pathway is functional in the Hep3B cells. In this study, the Hep3B cells appeared to be sensitive to NK cell-induced apoptotic death (Figs. 3, 6). The results demonstrated that a significant amount of [3H] thymidine is released and active caspase-3 is generated when the Hep3B cells are co-incubated with the NK cells. Furthermore, the NK cytotoxicity assessed by [3H]-thymidine release was not blocked by a mild fixation of the NK cells with paraformaldehyde, which suggests that surface molecules are involved in the apoptotic process. Indeed, the NK cytotoxicity assessed by [3H]-thymidine release was significantly blocked by a treatment with a recombinant TRAIL receptor-Fc fusion protein (Fig. 8B), although it was not completely blocked. Overall, these results indicate that NK cells induce apoptosis in the Hep3B via the TRAIL/TRAIL receptor interaction, and perhaps additionally through other death ligand/receptor interactions; however, the Fas/FasL interaction is not involved in this process. In agreement with our data, recent reports have demonstrated the importance of TRAIL in tumor surveillance. TRAIL is expressed on the activated NK cells [18], and appears to play a crucial role in the rejection of a tumor metastasis by the liver NK cells [48].

The resistance of the death receptor-mediated apoptosis commonly observed in HCCs either by down regulation of the death receptors or by expression of the inhibitory proteins of apoptosis is suggested to be an effective way of evading the immune surveillance function. Our data also demonstrated that the HCC targets, particularly HepG2 cells, are quite resistant to NK cell-induced apoptotic death. Interestingly, these HCCs are still quite sensitive to the granule-dependent necrotic death by NK cells. These results raise an interesting question as to how HCCs overcome the NK cell attack armed with cytotoxic granules. It is tempting to speculate that HCCs, and presumably other tumor cells as well, might utilize a special pathway to modulate the granule-dependent cytotoxic function of NK cells. In support of this hypothesis, previous studies have demonstrated that the NK cell activity is significantly lower in HCC patients [3, 12, 47, 49], and that low NK activity increases the risk of developing a HCC in liver cirrhosis patients and may cause the early development and invasion of HCC [3, 30]. Further studies are necessary to determine how the NK cell function, particularly the granule-mediated cytotoxicity, is modulated by the HCCs in detail, and to identify the molecules involved in this phenomenon.