Hepatocellular carcinoma (HCC) develops with high risk from chronic liver diseases and recurs frequently after regional therapy such as surgical resection or percutaneous ethanol injection.1 Prevention of HCC development as well as its recurrence is one of the most challenging aspects for reducing the morbility and mortality of HCC. Many natural killer (NK) cells reside in the liver,2, 3 but NK cell activity has been demonstrated to decrease in patients with chronic liver diseases as well as HCC.4, 5 Because NK cells are known to play a distinct role in first-line immunologic defense against tumor cells,6, 7 augmentation of the NK-mediated innate immune response should be an attractive strategy for preventing HCC. To this end, it is important to clarify the mechanisms by which NK cells recognize and destroy HCC cells.
Recent studies have demonstrated that NK-mediated effector functions are regulated by a balance between inhibitory and stimulatory signals. In general, tumor cells expressing little or no MHC Class I molecules are highly susceptible to NK-mediated cytolysis. The molecular basis of this “missing self” hypothesis is that NK cells recognize MHC Class I on target cells via surface receptors that deliver signals to suppress NK cell function.8, 9 On the other hand, there are several receptors on NK cells engaged in activating the signal transduction that leads to augmented NK-mediated cytolysis.10, 11 NKG2D, one of these activating receptors, is expressed on virtually all NK cells and recognizes MHC Class I chain-related A (MICA) and a closely related molecule MHC Class I chain-related B (MICB).12, 13 MICA and MICB (MICA/B) are cell surface glycoproteins that are not associated with antigen processing and not induced by interferon (IFN)γ, indicating differences in their properties from MHC Class I.14, 15 Although normal tissue distribution of MICA/B is confined to thymus and gastrointestinal epithelium, MICA/B were shown to be detected in several carcinoma cells such as from lung, breast, ovary, prostate and colon cancer.16 This raises the possibility that MICA/B may be specific targets on tumor cells for NK-mediated cytolytic activity. If so, upregulation of endogenous MICA/B on tumor cells may lead to enhanced natural immunity by NK cells.
Although MICA/B are induced by heat shock or viral and bacterial infections,12, 13, 14, 17, 18, 19 there has been no report of a pharmacological agent that can modulate MICA/B expression. Whereas NKG2D is conserved across species, MICA/B orthologs have not been found in rodents. Recently, retinoic acid early transcript 1 (Rae-1) and H60 minor histocompatibility antigen have been shown to act as ligands for NKG2D in mice, and the transcriptional activity of Rae-1 has been reported to significantly increase with retinoic acid treatment.20, 21 These findings led us to examine whether retinoic acid can induce the expression of MICA/B molecules on human tumor cells, although human MICA/B exhibit few similarities to murine Rae-1.
We demonstrate that MICA/B is expressed in a subset of human HCCs and show that it plays an important role in the susceptibility of hepatoma cells to NK cytolysis. Furthermore, retinoic acid could upregulate MICA/B expression on hepatoma cells, and thereby renders them more susceptible to NK cytolysis. These findings suggest that retinoic acids may act as anti-tumor agents, not only by regulating apoptosis and differentiation of tumor cells as have been demonstrated, but also by modulating their sensitivity to the innate immune response through upregulation of MICA/B.
MATERIAL AND METHODS
Cell lines and tissues
Ten pairs of HCCs and adjacent non-tumor counterparts were obtained at the time of surgical resection. The tissues were frozen in liquid nitrogen and stored at −80°C until use. HepG2 and Hep3B human hepatoma cell lines were purchased from the American Type Culture Collection (Rockville, MD). Huh7 human hepatoma cell line was a generous gift from Dr. Masayoshi Nanba (Institute for Molecular and Cellular Biology, Okayama University, Okayama, Japan). These hepatoma cells were propagated in DMEM supplemented with 10% heat-inactivated FBS, antibiotics and antimycotics (Gibco-BRL Life Technologies Inc., Gaithersburg, MD) in a humidified atmosphere of 5% CO2 at 37°C. We also used human Burkitt's lymphoma cell line Raji and the chronic myelogenous leukemia cell line K562, both obtained from the American Type Culture Collection. Both cell lines were propagated in RPMI 1640 supplemented with 10% heat-inactivated FBS, antibiotics and antimycotics.
Reverse transcription and PCR
The presence of MICA/B mRNA was examined by reverse transcription and PCR (RT-PCR) analysis. Total RNA was extracted using ISOGEN (Nippon Gene Co., Toyama, Japan). One microgram of extracted RNA was added to 80 pmol of random primers (Takara Shuzo, Shiga, Japan) and 10 mmol/L of each deoxynucleotide triphosphate, incubated at 65°C for 5 min, and quickly chilled on ice. The mixture was combined with 50 mmol/L Tris-HCl, 75 mmol/L KCl, 10 mmol/L DTT, 3 mmol/L MgCl2 and 100 U of Molony murine leukemia virus reverse transcriptase (Gibco-BRL) and incubated at 37°C for 50 min. The resulting cDNA was used in the PCR reaction mixture containing 10 pmol of each upstream MIC sense primer (5′-ACACCCAGCAGTGGGGGGAT-3′) and downstream MICA antisense primer (5′-GCAGGGAATTGAATCCCAGCT-3′), or MICB antisense primer (5′-AGCAGTCGTGAGTTTGCCCAC-3′], 10 mmol/L Tris-HCl, 50 mmol/L KCl, 1.5 mmol/L MgCl2, 2.5 mmol/L each deoxynucleotide triphosphate, and 2.5 U of Taq DNA polymerase (Takara Shuzo). Each primer pair described above was designed to span at least 1 intron of the corresponding gene. The amplification protocol included 28 cycles of 95°C for 60 sec for denaturation, 56°C for 60 sec for annealing, and 90 sec by 72°C for extension. Preliminary experiments confirmed that MICA/B RNA can be detected semiquantitatively under these conditions. As a control for the integrity of total RNA, primers specific for glyceraldehydes-3-phosphate dehydrogenase (G3PDH), sense primer, (5′-GCCACCCAGAAGACTGTGGATGGC-3′) and antisense primer, (5′-CATGTAGGCCATGAGGTCCACCAC-3′) were used. The PCR products were analyzed by ethidium bromide-stained 1.5% agarose gel electrophoresis.
The surface expression of HLA Class I and MICA/B was evaluated by flow cytometry. Cells (1 × 106) were incubated at 4°C for 30 min with anti-HLA Class I MAb (W6/32) (Serotec Co., Oxford, UK) or anti-MICA/B MAb (6D4; IgG1 isotype).12, 14, 16 MICA/B MAb was used in the dilution of 1:10,000 for 1 × 106 cells staining. The cells were then washed with PBS containing 2% BSA and 0.05% NaN3 twice and incubated at 4°C for 30 min in phycoerythrin-labeled rat anti-mouse kappa (Becton Dickinson, San Jose, CA) or FITC-labeled goat anti-mouse IgG (Coulter-Immunotech, Marseille, France) as second-step antibodies (Abs) for HLA Class I and MICA/B, respectively. The cells were then washed with PBS containing 2% BSA and 0.05% NaN3 twice and fixed with 2% paraformaldehyde solution. Flow cytometry was carried out using a FACScan system (Becton Dickinson).
Immunohistochemical detection of MICA/B was done as described previously.16 Briefly, 8 μm cryostat sections of HCC tissues and their non-cancerous counterparts were fixed with 4% paraformaldehyde for 30 min and then incubated with anti-MICA/B MAb (6D4) overnight at 4°C. Bound primary Ab was visualized with streptavidin-biotin peroxidase and 0.05% 3,3′-diaminobenzidine-0.02% H2O2 solution (Vectastain ABC kit, Vector, Burlingame, CA) according to the manufacturer's instruction. The sections were counterstained with hematoxylin.
Isolation and propagation of NK cells
NK cells used in our study were isolated and propagated from human peripheral blood as described previously.22 Briefly, human mononuclear cells were isolated from heparinized venous blood from healthy volunteers by Ficoll Hypaque density centrifugation and were incubated in a complete medium for 2 hr at 37°C to remove adherent cells. The recovered cells were resuspended at a concentration of 1 × 106 cells/mL in RPMI 1640 supplemented with 200 IU/mL recombinant human interleukin 2 (Stratmann Biotech GBMH, Hannover, Germany) and cultured at 37°C for 5 days. Following these procedures, the CD3-positive cells were removed using magnetic beads coated with anti-CD3 monoclonal antibody (MAb) (Dynal, Oslo, Norway). Flow cytometric analysis showed that more than 85% of these cells expressed CD56, but not CD3, CD4, CD8, CD14 and CD19 (data not shown). We applied 5 different healthy donors as a source of NK cells. The isolated CD56+CD3− cells expressed NK cell receptors, such as KIR2DL1, KIR2DL2-3, KIR3DL1, CD94 and NKG2A, at various rates (data not shown).
Target cells (HepG2, Hep3B, Huh7 and Raji) were labeled with 51Cr and incubated with NK cells for 4 hr at various effector/target ratios. The supernatants were obtained after the incubation and subjected to γ-counting. The maximum or spontaneous release was defined as counts from samples incubated with 5% Triton-X or medium alone, respectively. Cytolytic activity was calculated with the following formula: % lysis = (experimental release − spontaneous release) × 100/(maximum release − spontaneous release). The spontaneous release in all assays was <25% of the maximum release. To analyze the involvement of HLA Class I or MICA/B in cytolytic activity of NK cells, anti-HLA Class I MAb (W6/32), anti-MICA/B MAb (6D4) or isotype-matched control Ab was added during the cytolytic assay, as previously reported.12, 16, 23
Retinoic acid and heat shock treatment
All-trans retinoic acid (ATRA) was purchased from Sigma (St. Louis, MO). The synthetic retinoid Ro41-5253 was a generous gift from F. Hoffmann-La Roche Ltd. (Basel, Switzerland). This retinoid is known to selectively bind retinoic acid receptor (RAR) α without generating a downstream effect, thus acting as a competitive RARα antagonist. It was shown previously that 10-fold excess of Ro 41-5253 is appropriate to show RARα-specific antagonism.24
Cells were incubated with or without 1 μmol/L of ATRA for 72 hr, unless otherwise noted. For each experiment, ATRA was directly diluted into the culture medium from a stock solution prepared as 1,000-fold concentrated stocks in dimethylsulfoxide. The same amount of vehicle was added for the control experiment. All treatments were carried out under subdued light. For the blocking experiment, the cells were treated with 1 μmol/L of ATRA in the presence of 10 μmol/L of Ro41-5253.
Heat shock treatment was carried out as described previously.14 Briefly, hepatoma cells grown in 6-cm plastic flasks were immersed in a 42.5°C water bath for 90 min, followed by incubation at 37°C for 60 min and then subjected to flow cytometry or PCR analysis.
Intracellular IFNγ staining of NK cells cultured with hepatoma cells
Hepatoma cells were treated with or without 1 μM of ATRA for 72 hr, washed briefly, and then cultured with NK cells. In some experiments, anti-MICA/B MAb or isotype-matched IgG were added at the beginning of the coculture of hepatoma cells with NK cells. After 24 hr of coculture, NK cells were collected and subjected to intracellular flow cytometric analysis carried out as reported previously.25 Briefly, NK cells (5 × 105) in 24-well plates were stimulated with 10 ng/ml PMA plus 1 μM ionomysin in the presence of 1 μl/ml GolgiPlug™ (Becton Dickinson) for 4 hr at 37°C. At the end of the incubation period, the NK cells were stained with PE or APC-labeled CD56 MAb for 30 min at 4°C. These cells were then fixed and permeabilized with Cytefix/Cytoperm™ buffer (Becton Dickinson) for 15 min at room temperature. Permeabilized cells were stained with FITC-labeled anti-IFNγ MAb (BD-Pharmingen). The stained cells were analyzed by FACscan systems.
MICA/B are expressed in human HCC tissues and hepatoma cell lines
Expression of MICA/B was examined by RT-PCR in 10 pairs of surgically resected human HCC tissues and adjacent non-tumor tissues (Fig. 1a). Five of the 10 HCC tissues expressed both MICA and MICB transcripts (numbers 1, 3, 6, 8, 10) and 1 expressed the MICB transcript (number 4), whereas neither transcript was detected in non-tumor liver tissues. Figure 1b depicts typical immunoperoxidase staining of MICA/B in HCC tissue that was positive for MICA/B transcripts and an adjacent non-tumor counterpart (number 6). Most, if not all, tumor cells were stained with anti-MICA/B MAb and showed a diffuse cellular staining pattern. In contrast, hepatocytes in non-tumor liver were negative to the staining. Expression of MICA/B was also examined in 3 human hepatoma cell lines, Huh7, HepG2 and Hep3B, by flow cytometry. As shown in Figure 2a, MICA/B were expressed on Huh7 and HepG2 at substantial levels but on Hep3B at only a marginal level. These results suggest that at least some of HCCs express MICA/B.
MICA/B expressed on hepatoma cells are involved in their susceptibility to NK cells
When NK cells propagated from human peripheral blood were subjected to effector cells, the three hepatoma cell lines tested were relatively sensitive to NK cytolysis compared to NK-resistant Raji cells (Fig. 2b). To examine whether MICA/B expressed on hepatoma cells are involved in their susceptibility to NK cells, we used MAb 6D4 to mask the cell-surface MICA/B as described previously (Fig. 2c).13, 16 Addition of MAb 6D4 during the cytolytic assay substantially inhibited cytolytic activity against Huh7 and HepG2 cells in comparison with addition of control mouse IgG, whereas the same treatment led to a minimum decrease of NK cytolytic activity against Hep3B cells as well as MICA/B-negative Raji cells (Fig. 2a). These results indicated that MICA/B expressed on Huh7 and HepG2 hepatoma cells play an important role in their susceptibility to NK cells.
Flow cytometric analysis showed that hepatoma cell lines expressed HLA Class I at high levels on their cell surface (Fig. 2a). To find whether HLA Class I on hepatoma cells has any role in the susceptibility to NK cells, we incubated anti-HLA Class I MAb during the cytolytic assay (Fig. 2c). This clearly increased the cytotoxicity of NK cells against Raji cells, suggesting that HLA Class I expressed on this cell line can transduce an inhibitory signal to NK cells. In contrast, the same treatment did not affect NK cytotoxicity against hepatoma cell lines, implying that surface expression of HLA Class I on hepatoma cells does not downmodulate NK cytotoxicity under these experimental conditions.
Retinoic acid upmodulates MICA/B expression in hepatoma cells
To investigate the effect of retinoic acid on the expression of MICA/B, hepatoma cells were treated with ATRA and then subjected to flow cytometric analysis. Whereas ATRA did not affect the expression of HLA Class I, this treatment clearly upregulated expression of MICA/B on Huh7 and HepG2 cells in a dose-dependent manner, but not on Hep3B (Fig. 3a). It was reported previously that heat shock treatment induces MICA/B expression.14, 16 Interestingly, heat shock treatment also induced MICA/B on Huh7 and HepG2, but not on Hep3B. The time course of MICA/B expression by retinoic acid differed from that by heat shock treatment; MICA/B was induced more than 48 hr after retinoic acid treatment (data not shown), but as early as 1 hr after heat shock treatment (Fig. 3a). RT-PCR analysis also showed that both transcripts of MICA and MICB were upregulated after ATRA treatment of Huh7 and HepG2 cells, but not Hep3B cells (Fig. 3b). To confirm that ATRA-induced MICA/B modulation is a downstream event of its specific receptors, Huh7 cells were treated with ATRA in the presence of a synthetic retinoid that competitively antagonizes RARα. The results showed that MICA/B expression induced by ATRA was partially inhibited by RARα antagonist, indicating that ATRA upregulates MICA/B on Huh7 cells, at least in part, by activating RARα (Fig. 3c).
Retinoic acid-induced increase in MICA/B enhanced activation of NK cells and NK cytolysis of hepatoma cells
To test the hypothesis that retinoic acid-induced increase of MICA/B on hepatoma cells is relevant to NK activation, NK cells were co-cultured with ATRA-treated hepatoma cells and then subjected to analysis of IFNγ production. Intracellular IFNγ was clearly increased by co-culture with ATRA-treated Huh7 cells compared to co-culture with untreated cells (Fig. 4a). In contrast, co-culture with ATRA-treated Hep3B had only minimum effect on IFNγ production of NK cells. We also investigated whether retinoic acid treatment renders hepatoma cells more susceptible to NK cytolysis. ATRA treatment of Huh7 as well as HepG2 cells enhanced their susceptibility to NK cells (Fig. 4b). In contrast, ATRA treatment showed little effect on cytolytic activity against Hep3B or Raji cells, both of which did not show upregulation of MICA/B by retinoic acid (Fig. 3a and data not shown).
To confirm the involvement of retinoic acid-induced increase of MICA/B in enhanced activation of NK cells, MAb against MICA/B was added during the co-culture of hepatoma cells with NK cells. When MICA/B MAb was added into the co-culture of Huh7 cells with NK cells, IFNγ production from NK cells was almost completely inhibited, indicating that MICA/B expressed on Huh7 is responsible for the production of IFNγ from NK cells. Again, retinoic acid treatment augmented the ability of Huh7 cells to induce IFNγ production from NK cells; this ability was completely abolished to a basal level by MAb-mediated masking of MICA/B (Fig. 5a). Furthermore, when MICA/B MAb was added during the cytolytic assay, increased susceptibility of retinoic acid-stimulated Huh7 cells and HepG2 cells was completely abolished (Fig. 5b). These data indicate that the increased IFNγ production and NK cytolysis conferred by retinoic acid treatment of hepatoma cells is dependent on upregulation of MICA/B on hepatoma cells.
Our present study showed that MICA/B were expressed in a subset of human HCC tissues as well as hepatoma cell lines, but not in surrounding non-tumor tissues. The preference of MICA/B expression in tumors has been previously demonstrated for other malignancies such as colon, prostate, and lung cancer.16 Involvement of MICA/B expression in NK sensitivity of hepatoma cells was demonstrated by the fact that MAb-mediated masking of MICA/B substantially suppressed the NK sensitivity of Huh7 and HepG2 cells. Intriguingly, Hep3B was also susceptible to NK cells despite its low level of MICA/B expression, suggesting that another stimulatory pathway might exist for initiating NK cytolysis of this cell type. In this context, it should be noted that other human NKG2D ligands, ULBP1-3, have been recently identified.23, 29 In addition, NK activating receptor family, including NKp44, NKp46 and NKp30, also plays an important role in activating NK cells although ligands of these receptors have not been identified.11 Therefore, it may be interesting to examine relevance of these molecules in tumor immunology in future study. Our present study did not directly address the in vivo relevance of MICA/B expression on HCC tissues. In this regard, it would be interesting to analyze a larger number of patient samples to examine the relationship between MICA/B expression and disease prognosis, because NK cell activity has been suggested to be inversely related to the HCC recurrence rate. The question arises of why tumor can expand in vivo despite MICA/B expression that may activate NK cells. Although the answer is obscure at present, it is possible that the levels of MICA/B expression may not be sufficient for effectively activating NK cells. If so, modulation of MICA/B expression may be an attractive strategy for controlling tumor development and progression.
We have demonstrated for the first time that retinoic acid induces expression of MICA/B. This is not restricted to the hepatoma cell lines Huh7 and HepG2, because ATRA upregulates MICA/B in other cancer cell lines such as DLD-1 and HCT116 colon cancer cells as well as MCF7 breast cancer cells (our unpublished data). ATRA-enhanced MICA/B expression was associated with upregulation of both MICA and MICB transcripts. Previous reports have shown that heat shock treatment results in increased mRNA and protein expression of MICA/B, which is presumably due to the presence of putative heat shock-elements in the 5′-flanking region of the corresponding genes.14, 15 Retinoid-mediated induction of transcriptional activation as well as surface expression of MICA/B occurred much later than that by heat shock treatment, implying that ATRA may indirectly activate transcription of MICA/B genes by inducing other factors. Although the present study showed that RARα mediates ATRA-modulated MICA/B expression, further studies are needed to determine the precise mechanisms by which retinoic acid activates MICA/B.
Retinoic acid treatment rendered hepatoma cell more susceptible to NK cells and also augmented the ability of hepatoma cells to induce IFNγ production from NK cells. It was previously reported that retinoic acid can modulate expression of HLA Class I, ICAM-1 and CD95 in certain cell types and thereby affect their susceptibility to effector cell-mediated cytolysis and apoptotic cell death.26, 28 This is not likely to be the case with hepatoma cells, however, because ATRA treatment did not affect the expression of these molecules on Huh7, HepG2 and Hep3B hepatoma cells (Fig. 3a and unpublished data). Instead, the MAb-mediated masking experiment clearly indicated that upregulation of MICA/B on hepatoma cells is responsible for enhanced IFNγ production from NK cells and NK cytolytic ability. We also examined difference in NKG2D expression between IFNγ-producing CD56 positive cells and IFNγ-negative CD56 positive cells by flow cytometry. All of the IFNγ-producing CD56 positive cells ubiquitously expressed NKG2D on their cell surface, whereas approximately half of the IFNγ-negative cells expressed NKG2D but the others did not (our unpublished data). The observation suggests strongly that NKG2D expression on CD56 positive cells is a critical requirement for IFNγ production upon stimulation of ATRA-treated hepatoma cells. Further study is needed to examine why some of the NKG2D positive cells did not produce IFNγ upon the enhanced MICA/B expression. It has been well known that NK-mediated tumor immunity is critically dependent on direct cytolytic activity as well as IFNγ production. The present study raises the possibility that retinoic acid treatment may augment these NK-mediated anti-tumor immune responses via upregulation of MICA/B.
Retinoic acid has been shown to be effective against some forms of malignancy.30 Recently, a randomized controlled clinical trial conducted to test the efficacy of acyclic retinoid to prevent HCC showed that there was a much lower incidence rate of recurrent or new HCC in retinoid-treated groups compared to those in the placebo group, indicating that oral acyclic retinoids can reduce HCC recurrence.31, 32 As the mechanisms of the action, Muto et al. suggested that retinoids can bring hepatoma cells toward differentiated hepatocytes or induce apoptosis in hepatoma cells via downregulation of transforming growth factor α.33, 34 Our present data suggest a novel mechanism by which retinoic acid may function as an anti-tumor agent.
In conclusion, MICA/B were expressed on a subset of human HCCs and shown to play an important role in NK cytolysis against hepatoma cells. Retinoic acid can serve as an inducer of MICA/B expression and thereby further activate NK cells. These findings raise the possibility of retinoic acid may being an attractive candidate for augmenting innate immunity against HCC.