Glypican-3 (GPC3) is useful not only as a novel tumor marker, but also as an oncofetal antigen for immunotherapy. We recently established HLA-A2-restricted GPC3144-152 peptide-specific CTL clones from hepatocellular carcinoma patients after GPC3144-152 peptide vaccination. The present study was designed to evaluate the tumor reactivity of a HLA-A2-restricted GPC3144-152 peptide-specific CTL clone against ovarian clear cell carcinoma (CCC) cell lines. The GPC3144-152 peptide-specific CTL clone could recognize HLA-A2-positive and GPC3-positive ovarian CCC cell lines on interferon (IFN)-γ enzyme-linked immunospot assay and showed cytotoxicity against KOC-7c cells. The CTL clone recognized naturally processed GPC3-derived peptide on ovarian CCC cells in a HLA class I-restricted manner. Moreover, we confirmed that the level of GPC3 expression was responsible for CTL recognition and that subtoxic-dose chemotherapy made tumor cells more susceptible to the cytotoxic effect of CTL. Thus, it might be possible to treat ovarian CCC patients by combining chemotherapy with immunotherapy. Our data suggest that GPC3 could be an effective target for immunotherapy against ovarian CCC. (Cancer Sci 2011; 102: 1622–1629)
Epithelial ovarian carcinoma (EOC) is the leading cause of death from gynecological malignancy. Cytoreductive surgery and systemic combination chemotherapy with a platinum drug and a taxane represent the standard of care for EOC patients. Ovarian clear cell carcinoma (CCC) is the second most frequent subtype of EOC in Japan, although CCC represents 8–10% of all EOC in the United States.(1,2) Compared with other EOC subtypes, ovarian CCC is associated with a poorer prognosis and increased chemoresistance.(1,3) More efficient conventional therapies and novel strategies for effectively treating ovarian CCC are required.
Glypican-3 (GPC3) is a member of the glypican family of heparan sulfate proteoglycans that are attached to the cell surface via the glycosylphosphatidylinositol (GPI) anchor.(4) It is known as an oncofetal antigen specifically overexpressed in hepatocellular carcinoma (HCC).(5) Previous studies have shown that GPC3 was also overexpressed in other malignant tumors, such as melanoma, Wilms’ tumor, hepatoblastoma, yolk sac tumor, ovarian CCC and lung squamous cell carcinoma.(6–10)
We previously identified the HLA-A24-restricted GPC3298-306 (EYILSLEEL) and HLA-A2-restricted GPC3144-152 (FVGEFFTDV) peptides, both of which can induce GPC3-reactive cytotoxic T cells (CTL).(11) Recently, HLA-A2-restricted GPC3144-152 peptide-specific CTL clones were established from HCC patients after GPC3144-152 peptide vaccination in our laboratory.(12) Although CTL reactivity against HCC cell lines was analyzed using these CTL clones, other GPC3-positive tumor cell lines have not been studied. Therefore, we examined the reactivity of a HLA-A2-restricted GPC3144-152 peptide-specific CTL clone against ovarian CCC cell lines, and whether subtoxic-dose chemotherapy sensitizes ovarian CCC cells to lysis of GPC3144-152 peptide-specific CTL.
Materials and Methods
GPC3144-152 peptide-specific CTL clone and cell lines. We established the HLA-A2-restricted GPC3144-152 peptide-specific CTL clone from the PBMC of HCC patients vaccinated with GPC3144-152 (FVGEFFTDV) peptide by single-cell sorting using CD107a antibody. The established CTL clone was tested for avidity by using GPC3144-152 peptide-pulsed T2 targets with a range of peptide concentrations, starting at 10−6 M and decreasing by log steps to 10−14 M. The peptide concentration at which the curve crossed 50% cytotoxicity was defined as the avidity of the CTL clone and was rounded to the nearest log. This CTL clone had high avidity CTL (10−11 M) and could recognize HCC cell lines expressing GPC3 in a HLA-class-I-restricted manner.(12) Two human ovarian CCC cell lines, KOC-7c (HLA-A*0201/A*3101) and TOV-21G (HLA-A*1101/A*2601), and two human HCC cell lines, HepG2 (HLA-A*0201/A*2402) and SK-Hep-1 (HLA-A*0201/A*2402), were used in the present study. They were conserved in our laboratory. TOV-21G.A2 acquires expression of HLA-A2 following transfection with an HLA-A2 expression plasmid.(13) TOV-21G.A24 was similarly transfected with an HLA-A24 expression plasmid. SK-Hep-1.hG acquires expression of human GPC3 following transfection with a human GPC3 expression plasmid. SK-Hep-1.vec cell line transfected with an empty vector was used as a control. To study the effect of silencing GPC3, KOC-7c GPC3-shRNA and Neg-shRNA (control shRNA) were established by short hairpin RNA knockdown technology as described previously.(14) These cells were maintained in RPMI 1640 or DMEM medium (Sigma, St Louis, MO, USA) supplemented with 10% FCS, penicillin (100 U/ml) and streptomycin (100 μg/mL) at 37°C in a humidified atmosphere containing 5% CO2.
RNA preparation and quantitative real-time PCR (qRT-PCR). Total RNA was isolated using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. GPC3 gene expression levels were analyzed by qRT-PCR assays using the following primers generated according to the indicated reference sequences: sense, 5′-GAGCCAGTGGTCAGTCAAAT-3′ and antisense, 5′-CTTCATCATCACCGCAGTC-3′. Amplification reactions were carried out in 96-well plates in 25 μL reaction volume using the Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). All reactions were performed in technical triplicate using an ABI 7500 Fast Real-Time PCR System. Relative expression of the GPC3 gene to the endogenous control gene, β-actin, was calculated using the comparative CT method. β-actin qRT-PCR primer sequences were: sense, 5′-TCCATCATGAAGTGTGACGT-3′ and antisense, 5′-GAGCAATGATCTTGATCTTCAT-3′.
Flow cytometry analysis and cell sorting. Flow cytometry (FCM) was performed to quantify the expression of GPC3 and Fas on the cell surface using the following antibodies: primary anti-GPC3 (clone 1G12; BioMosaics, Burlington, VT, USA); Alexa Fluor 488 conjugated second Ab (Invitrogen); phycoerythrin (PE)-conjugated anti-Fas (clone DX2; BioLegend, San Diego, CA, USA); FITC-conjugated anti-HLA-A2 (clone BB7.2; MBL, Nagoya, Japan); and FITC-conjugated mouse IgG2b isotype control (clone 3D12; MBL).
The FCM data was acquired using the FACSCanto II system (BD Biosciences, San Jose, CA, USA) and analyzed using FlowJo software (Tree Star, Ashland, OR, USA). Mean fluorescence intensity (MFI) of GPC3 staining was calculated as follows: MFI ratio = MFI with the anti-GPC3 Ab/MFI with the secondary Ab. MFI of HLA-A2 staining was similarly calculated (MFI ratio = MFI with the anti-HLA-A2 Ab/MFI with isotype control Ab).
Cell sorting was performed using the FACSAria II cell sorter (BD Biosciences) to isolate GPC3+ and GPC3− cells from KOC-7c cells. We purified KOC-7c GPC3 high or low cells with the top or bottom 10% of GPC3 expression, respectively.
Response of GPC3144-152 peptide-specific CTL clone against cancer cell lines. GPC3144-152 peptide-specific CTL clone cells were co-cultured with each cancer cell line as target cells at the indicated effector/target (E/T) ratio and cytotoxicity assay or IFN-γ enzyme-linked immunospot (ELISPOT) assay was performed. Blocking of HLA class I was done as follows. Before coculturing the CTL clone with a cancer cell line in an assay, the target cancer cells were incubated for 1 h with anti-HLA class I mAb (clone W6/32; BioLegend), or isotype control IgG2a mAb, and then the effects of Ab on CTL clone activity was examined.
IFN-γ ELISPOT analysis. ELISPOT assay for detecting antigen-specific IFN-γ-producing T cells was performed using the ELISPOT kit (BD Biosciences). The spots were automatically counted and analyzed with the Eliphoto system (Minerva Tech, Tokyo, Japan).
Cytotoxicity assay. The cytotoxic capacity was analyzed with the Terascan VPC system (Minerva Tech). The CTL clone was used for effector cells. Target cells were labeled in calcein-AM solution for 30 min at 37°C. The labeled cells were then co-cultured with effector cells for 4–6 h. Fluorescence intensity was measured before and after the 4–6 h culture, and specific cytotoxic activity was calculated as previously described.(12)
Cold inhibition assay. Calcein AM-labeled target cells were cultured with effector cells in a 96-well plate with cold target cells. T2 target cells, which were prepulsed with either HIV19–27 peptide or GPC3144-152 peptide, were used as cold target cells.
CD107a degranulation assay. GPC3144-152 peptide-specific CTL clone cells were incubated with cancer cell lines at a 2:1 ratio for 4 h at 37°C. APC-conjugated CD107a-specific mAb (clone H4A3; BD Biosciences) were present during the incubation period; after incubation, cells were stained with additional PE-conjugated anti-CD8 mAb (clone HIT8a; BioLegend) and analyzed by FCM.
Growth inhibition assay. Growth inhibition was evaluated by a 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt (WST-8) colorimetric assay using a Cell Counting Kit (Dojindo, Kumamoto, Japan). Cells (5 × 103) were seeded into 96-well plates in 100 μL of culture medium for 24 h prior to drug exposure, and then treated with various concentrations of paclitaxel (PTX) or cisplatin (CDDP) for 18 or 48 h. Cell viability was determined colorimetrically by optical density at 450 nm wavelength using a microplate reader (Bio-Rad, Hercules, CA, USA). The percentage of cell survival for each drug concentration was calculated as: (absorbance of test wells/absorbance of control wells) × 100.
Apoptosis analysis. The Annexin V-FITC Apoptosis Detection Kit (BioVision, Mountain View, CA, USA) was used to determine apoptosis after treatment with PTX or CDDP. After treatment with the chemodrug, floating and adhering cells were collected via trypsinization and centrifuged. The supernatant was removed and resuspended in 500 μL of binding buffer to which 5 μL of Annexin-V-FITC and propidium iodido (PI) was added. The cells were incubated at room temperature for 5 min in the dark and assessed by FCM.
Statistical analysis. Univariate regression analysis was used to evaluate the correlation between GPC3 expression and GPC3-specific CTL recognition. Mann–Whitney U-test and Kruskal–Wallis test followed by Scheffe’s post hoc test were used to detect differences between groups. For all statistical tests, differences were considered significant at P < 0.05.
HLA-A2-restricted GPC3144-152 peptide-specific CTL clone recognizes ovarian CCC cell lines. To ascertain whether the HLA-A2-restricted GPC3144-152 peptide-specific CTL clone recognizes ovarian CCC cell lines expressing HLA-A2 and GPC3, we first evaluated the expression of GPC3 on cancer cell lines. We used KOC-7c and HLA-A0201 gene stable transfectant TOV-21G.A2 and two human HCC cell lines for the target cells. As positive controls, we used two HCC cell lines. SK-Hep-1.hG cells were an established stable GPC3-expressing cell line. As we performed qRT-PCR and FCM of GPC3 in these cell lines, GPC3 expression in ovarian CCC cell lines was less than that in HCC cell lines. Representative data of relative mRNA expression (ratio to KOC-7c) and MFI ratio are shown (Fig. 1A). The CTL response generally correlates with the numbers and density of MHC/antigen peptide complex on the target cells. Accordingly, we also evaluated HLA-A2 expression on the cell surface in cancer cell lines with FCM analysis (Fig. 1B). IFN-γ production of the CTL clone was detected against two ovarian CCC cell lines (Fig. 1C). In Figure 1C, we used TOV-21G.A24 as a negative control. Furthermore, we determined whether efficient GPC3144-152 peptide-specific CTL clone recognition was correlated with GPC3 expression levels. We found that CTL clone recognition was correlated with the relative GPC3 mRNA expression and GPC3 MFI ratio in the cell lines (r2 = 0.995 and 0.935, respectively) (Fig. 1D,E). In addition, we also analyzed whether CTL reactivity is correlated with not only GPC3 expression but also the expression of HLA-A2. The correlation between HLA-A2 expression levels on FCM analysis and CTL clone recognition (IFN-γ production or CD107a degranulation) was insufficient in the cell lines (data not shown). Although HLA-A2 expression on the cell surface in TOV-21G.A2 was moderately low, that in three other cell lines was sufficient on FCM analysis. TOV-21G.A2 cells have low expression of not only HLA-A2 but also GPC3. Therefore the GPC3 expression level is more important than the HLA-A2 expression level on GPC3144-152 peptide-specific CTL clone reactivity.
GPC3144-152 peptide-specific CTL clone lyses ovarian CCC cell lines. We detected GPC3-specific CTL responses by a CD107a degranulation assay. GPC3-specific CTL responses against TOV-21G.A2 and KOC-7c cells exhibited 2.79% and 5.42% CD107a staining, respectively, approximately 1.8- and 3.4-fold increases compared with the SK-Hep-1.vec as a negative control (Fig. 2A). CD107a degranulation was also correlated with the relative GPC3 mRNA expression and GPC3 MFI ratio in the cell lines (r2 = 0.978 and 0.865, respectively) (Fig. 2B). The GPC3144-152 peptide-specific CTL clone was further tested for its capacity to kill ovarian CCC cell lines, by a calcein-AM-based cytotoxicity assay. SK-Hep-1.vec cells were used for a negative control. The CTL clone displayed mild, but clear, specific cytotoxicity against KOC-7c cells (Fig. 2C). However, GPC3-specific cytotoxicity was insufficient against TOV-21G.A2 cells compared with TOV-21G.A24 cells (data not shown). In both ovarian CCC cell lines, Fas expression on the cell surface was sufficiently similarly to that of the HCC cell lines on FCM analysis (Fig. 2D).
HLA class I specificity was confirmed by the blockade of reactivity against ovarian CCC cell line KOC-7c. HLA class I-restricted activity was demonstrated by blocking of IFN-γ release and lysis of the GPC3144-152 peptide-specific CTL clone against KOC-7c after pretreatment with a HLA class I-specific mAb (W6/32) or mouse IgG2a isotype control, respectively, for 1 h. This reactivity could be inhibited by anti-HLA class I mAb but not by isotype control (Fig. 3). These results clearly indicate that the CTL clone recognized KOC-7c in a HLA class I-restricted manner.
Effect of GPC3 silencing using shRNA on the response of GPC3144-152 peptide-specific CTL clone against KOC-7c cells. To verify the GPC3 antigen-specific response of the CTL clone against ovarian CCC cell lines, we examined GPC3 knockdown on the GPC3-positive cell line KOC-7c. KOC-7c GPC3-shRNA was established using shRNA knockdown technology. The GPC3 expression of KOC-7c was obviously decreased by GPC3 shRNA on qRT-PCR. We examined the IFN-γ production and lysis of the CTL clone against KOC-7c GPC3-shRNA and KOC-7c GPC3 Neg-shRNA cells. IFN-γ production was significantly decreased by GPC3 shRNA (P = 0.004) (Fig. 4A). GPC3-specific cytotoxicity was reduced against KOC-7c GPC3-shRNA cells compared with KOC-7c Neg-shRNA cells (Fig. 4B). These results indicate that HLA-A2-restricted GPC3144-152 peptide could be processed naturally by ovarian CCC cells, and the peptides in the context of HLA-A2 could be expressed on the surface of ovarian CCC cells.
Level of GPC3 expression on the cell surface is related to GPC3144-152 peptide-specific CTL clone recognition. To confirm that the level of GPC3 expression on the cell surface is responsible for CTL recognition, KOC-7c GPC3 high and low cells were sorted by FACSAria II (Fig. 5A). As shown in Figure 5B, KOC-7c GPC3 high cells expressed higher mRNA of GPC3 than GPC3 low cells. Figure 5C shows the IFN-γ release of GPC3144-152 peptide-specific CTL clone against KOC-7c wild type, GPC3 high and GPC3 low cells. There were significant differences in IFN-γ production between the three populations (P < 0.001). GPC3-specific cytotoxicity was increased against KOC-7c GPC3 high cells compared with GPC3 low cells in a cytotoxicity assay without cold target cells. In a cold target inhibition assay, cytotoxicity against KOC-7c GPC3 high cells was suppressed by the addition of GPC3144–152 peptide-pulsed T2 cells but not by the addition of HIV19–27 peptide-pulsed T2 cells, even though cytotoxicity against KOC-7c GPC3 low cells was not changed by T2 pulsed with either GPC3144–152 or HIV19–27 peptide (Fig. 5D).
Chemotherapy sensitizes KOC-7c cells to the cytotoxic effect of GPC3144-152 peptide-specific CTL clone. Taxane plus platinum combination chemotherapy is generally considered to be the “gold standard” regimen for treatment of EOC. As PTX and CDDP have different mechanisms of action, we chose these two agents to investigate whether they sensitize ovarian CCC cells to GPC3-specific lysis. To evaluate the subtoxic dose of each drug, we assessed growth inhibition and apoptosis assays by FCM using Annexin V and PI staining. Growth-inhibitory effects were observed for treatment with either PTX or CDDP alone in a time- and dose-dependent manner. We calculated the 25% inhibitory concentration (IC25) of each drug as the minimum cytotoxic condition and regarded lower values as the subtoxic dose. The IC25 values of PTX and CDDP for 18 h were 22.8 ng/mL and 6.2 μg/mL, respectively (Fig. 6A). Exposure of CTL clone or KOC-7c cells to PTX (10 ng/mL) or CDDP (1 μg/mL) for 18 h had no significant cytotoxic effect, as determined by apoptosis assay. In other words, cell viability in untreated and PTX- and CDDP-treated groups of CTL clone or KOC-7c cells exceeded 95% in all cases (Fig. 6B). These conditions excluded direct cytoxic effects of the compounds and effects as a subtoxic dose. In contrast, PTX (10 ng/mL) or CDDP (1 μg/mL) for 48 h showed mild cytotoxicity (basal levels of apoptosis >5%), and PTX (1 μg/mL) or CDDP (10 μg/mL) for 18 h induced substantial cell death (data not shown). KOC-7c cells were exposed to the subtoxic dose of each drug for 18 h and then examined by cytotoxicity assay. Pretreatment of KOC-7c cells with PTX (10 ng/mL) or CDDP (1 μg/mL) significantly increased CTL-mediated cytotoxicity of target cells (Fig. 6C). In all experiments, the level of spontaneous calcein release of target cells treated with chemotherapeutic agents was similar to that of untreated cells.
Ovarian CCC has a poor prognosis due to low sensitivity to conventional chemotherapy.(1,3) To improve the prognosis, strategies are needed to efficiently kill all cancer cells by surgery and chemotherapy, as well as to stimulate the immune response to keep residual tumor cells in check. Thus, effective novel treatment strategies combined with surgery and chemotherapy are needed for treating ovarian CCC. Cancer vaccines are an attractive approach because of their low toxicity.
In previous studies, GPC3 was overexpressed in several malignant tumors, including ovarian CCC.(6–10) GPC3 is useful as a novel biomarker and oncofetal antigen for immunotherapy.(15–22) However, association of ovarian CCC with CTL recognition has not been performed, hindering the selection of appropriate candidates for GPC3-specific immunotherapy. We recently established HLA-A2-restricted GPC3144-152 peptide-specific CTL clones.(12) In the present study, we analyzed the IFN-γ production and cytotoxicity of an established CTL clone against ovarian CCC cell lines expressing HLA-A0201 and GPC3. The GPC3144-152 peptide-specific CTL clone could recognize HLA-A2-positive and GPC3-positive ovarian CCC cell lines, suggesting that ovarian CCC present endogenously processed GPC3144-152 peptide. Even though the CTL clones recognized two ovarian CCC cell lines on the IFN-γ ELISPOT assay, they showed inefficient lysis against TOV-21G.A2 cells. This was not due to a low expression level of HLA-A2 molecules on the cell surface, because the tumor cells were lysed after being pulsed with the antigenic peptide (data not shown). We also confirmed that the level of antigen expression is important in GPC3-specific CTL recognition of malignant cells. Therefore, low-level expression of GPC3 on tumor cells might be insufficient for triggering CTL-mediated killing.
Recent clinical studies have reported high rates of objective clinical response when cancer vaccines are combined with chemotherapy in patients with various cancers.(23–27) To evaluate the feasibility of chemoimmunotherapy for ovarian CCC, we investigated the cytotoxic effect of subtoxic-dose PTX or CDDP combined with GPC3144-152 peptide-specific CTL clone in the human ovarian CCC cell line KOC-7c. We found that chemotherapy made ovarian CCC cells more susceptible to the cytotoxic effect of the GPC3144-152 peptide-specific CTL clone. Chemotherapeutic drugs generally suppress the immune function, and each drug has a different level of immune suppression. Therefore, combination therapy requires an optimal dose that does not suppress peptide-induced immune activation. Importantly, the synergistic cytotoxic effect remained when both CTL and tumor cells were pretreated with PTX or CDDP under identical conditions (data not shown). However, high-dose chemotherapy has been shown to be toxic and the synergistic effect increased slightly more compared with the subtoxic dose, therefore limiting its potential therapeutic usefulness in vitro. The mechanism of improvement in immunotherapy with chemotherapy remains unclear, but the two possible types of mechanism are: systemic factors and local tumor microenvironment factors. For example, possible systemic effects include the elimination of cells with immunosuppressive activity such as regulatory T cells(28) and myeloid-derived suppressor cells,(29) or improved cross-presentation of tumor antigens. Examples of possible local effects include the disruption of tumor stroma that results in improved penetration of CTL into the tumor site, increased permeability of tumor cells to CTL-derived granzymes via upregulation of mannose-6-phosphate (M6P) receptors on the surface of tumor cells,(30) increased expression of tumor-associated antigens by tumor cells or upregulation of Fas (and other death receptors) on tumor cells, or FasL on CTL, etc.(31,32) We performed experiments to address the change in permeability for GrzB and the expression of M6P receptors in KOC-7c cells pretreated with PTX or CDDP. However, both drugs had no significant effect on the expression of M6P receptors. Moreover, we could not confirm the mechanism through an increase in permeability to GrzB in CCC cell line KOC-7c cells. Paclitaxel is known to upregulate the expression of Fas on the surface of tumor cells, resulting in an increase in Fas–FasL interaction.(33) However, Fas expression was sufficient in ovarian CCC cell lines without chemotherapy, and both drugs had no significant effect on Fas expression. The threshold for Fas-induced apoptosis in ovarian CCC is high and/or Fas signaling in CCC is altered through unknown mechanisms. In addition, both drugs had no significant effect on GPC3 expression under subtoxic-dose conditions (data not shown).
In conclusion, the present study suggests that GPC3 could become an effective target for HLA-A2-restricted peptide vaccine therapy against ovarian CCC. Moreover, our data suggest the possibility of treating ovarian CCC patients by combining standard chemotherapy with relatively non-toxic and highly specific immunotherapy. We will clarify the mechanisms of this phenomenon in our next study.
This work was supported in part by Grants-in-Aid for Research on Hepatitis and for Clinical Research from the Ministry of Health, Labour and Welfare, Japan.