The prognosis of pancreatic cancer, the fifth-leading cause of death by cancer in both Japan and the United States, is extremely poor; the median period of survival after diagnosis is 3–4 months, with a 5-year survival rate of approximately 3%.1, 2 Therefore, the development of new treatment modalities, including specific immunotherapy, is of great importance for the treatment of pancreatic cancer. Recent advances in molecular biology and tumor immunology have enabled us to identify a large number of genes and antigenic peptides recognized by cytotoxic T-lymphocytes (CTLs) reactive to melanomas3–7 and epithelial cancers,8–15 thereby opening the door to peptide-based specific immunotherapies for cancer patients. However, clinical trials of peptide-based specific immunotherapies for advanced melanoma patients have rarely resulted in tumor regression.16–19 This failure could be partly due to an insufficient induction of CTLs in the current regimes, so we should make more effort to induce peptide-specific CTL activity in each patient. Peptide-pulsed Langerhans cells in regional lymph nodes must come into contact with peptide-specific CTL precursor cells within 2 days in order to activate CTLs.20 Therefore, determining the frequency of CTL precursors reacting to vaccinated peptides in peripheral blood mononuclear cells (PBMCs) will be an important feature of achieving antitumor immunity. We have recently established a new and simple method to detect peptide-specific CTL precursors in circulation (Hida et al., unpublished data). With this new tool, our study has investigated whether CTL precursor cells reacting to peptides with vaccine candidates are detectable in PBMCs of pancreatic cancer patients. The new aspect of our study is the use of a new method to determine CTL precursor as a screening tool to determine those peptides to which the patient will most likely respond because reactive CTL precursor exist.
The prognosis of pancreatic cancer is extremely poor with a 5-year survival of approximately 3%. Thus, the development of new treatment modalities, including a specific immunotherapy, is required. Our study investigated whether cytotoxic T-lymphocyte (CTL) precursors reacting to peptides with vaccine candidates (13 peptides for HLA-A2+ or -A24+ patients, respectively) were detectable in the prevaccination peripheral blood mononuclear cells (PBMCs) of 15 pancreatic cancer patients. Peptide-specific CTL precursors were detectable in the majority (11 of 15, 73%) of patients, with a mean positive number of 1.5 peptides (ranging from 0–5 peptides) per patient. Positive peptide profiles varied among patients. These results may provide a scientific basis for a new kind of cancer immunotherapy, namely, a CTL precursor-oriented peptide vaccine, for pancreatic cancer patients. © 2001 Wiley-Liss, Inc.
Material and Methods
Patients and Cell Lines
Ten HLA-A24+ and 5 HLA-A2+ pancreatic cancer patients were enrolled in our study after informed consent was obtained. None of these patients were infected with human immunodeficiency virus (HIV) or human T-lymphotropic virus type 1 (HTLV-1). Twenty milliliters of peripheral blood was obtained, and PBMCs were prepared by Ficoll-Conray density gradient centrifugation. PBMCs were also obtained from healthy volunteers (5 HLA-A24+ and 5 HLA-A2+). HLA class I typing was performed on PBMCs by classical serologic methods that have been reported previously.9 Pancreatic cancer cell lines used for our study were Panc-1 (HLA-A0201/1101), paca-2 (HLA-A2402), YPK-1 (HLA-A2402), YPK-2 (HLA-A2402), YPK-3 (HLA-A0201), YPK-4 (HLA-A2601), HPAKII, SUIT2 and BxPC3. The KE4 esophageal SCC line (HLA-A2402/A2601), from which the SART1259, SART2 and SART3 genes were cloned, was used as a positive control.8 The HT1376 bladder carcinoma cell line (HLA-A2402), from which the ART4 gene was cloned, was used as a positive control.12 The MKN-45 gastric adenocarcinoma cell line (HLA-A2402) was used as a negative control for SART2 expression. The SW620 colon cancer cell line (HLA-A2402/0201) and phytohemagglutinin (PHA)-blastoid T cells were also used as target cells, as reported previously.13
Detection of Tumor Antigens
Expression of SART1, SART2, SART3 and ART4 tumor antigens at the protein levels in pancreatic cancer cell lines, pancreatic cancer tissues and nontumorous pancreatic tissues were investigated by Western blot analyses using previously reported polyclonal antibodies.8–10, 12, 22
The synthesized peptides used in our study are listed in Tables II and III. These peptides were purchased from Sawady Laboratory (Tokyo, Japan), and their purity was >95%. All peptides, except for Epstein-Barr virus (EBV)-derived peptide, were encoded by tumor-rejection antigens and have the ability to induce HLA-A24 or -A2-restricted CTLs specific to tumor cells in the PBMCs of cancer patients, as reported previously.8, 9–15 HIV-derived peptide with an HLA-A24-binding motif (RYLRQQLLGI)13 and HTLV-1-derived peptide with an HLA-A2-binding motif (SLYNTYATL)14 were used as a negative control.
Detection of Peptide-Specific CTL Precursor Cells
A simple method was used to detect peptide-specific CTLs in PBMCs. PBMCs (1 × 105 cells/well) were incubated with 10 μM of each peptide in U-bottom-type 96-well microculture plates (Nunc, Roskilde, Denmark) in 200 μl of culture medium. The culture medium consisted of 45% RPMI-1640 medium, 45% AIM-V® medium (GIBCO BRL, Walkersville, MA), 10% FCS, 100 U/ml of interleukin-2 (IL-2) and 0.1 mM MEM nonessential amino acid solution (GIBCO BRL). Half of the medium was removed and replaced with the new medium containing a corresponding peptide (20 μM) every 3 days for up 12 days. On the 12th day of the culture, 24 hr after the last stimulation, these cells were harvested, washed 3 times and then tested for their ability to produce IFN-γ in response to CIR-A2402 cells preloaded with either a corresponding peptide or HIV peptide (RYLRQQLLGI) as a negative control in HLA-A24+ PBMCs. In HLA-A2+ cases, T2 cells preloaded with a corresponding or HIV peptide (SLYNTYATL) were used as target cells. The target cells (CIR-A2402 or T2, 1 × 104/well) were pulsed with each peptide (10 μM) for 2 hr, and then effector cells (1 × 105/well) were added to each well with the final volume of 200 μl. After incubation for 18 hr, the supernatants (100 μl) were collected, and amounts of IFN-γ were measured by an ELISA (limit of sensitivity: 10 pg/ml). All experiments were performed in quadruplicate assays. A 2-tailed Student t-test was employed for the statistical analyses. Detectable levels of CTL precursors were judged as positive if the mean value of IFN-γ production by the peptide-stimulated PBMCs in response to a corresponding peptide was significantly (p < 0.05) higher than that in response to an HIV peptide.
The peptide-stimulated PBMCs were further expanded in the presence of feeder cells for 21–25 days in order to obtain a relatively large number of cells as reported previously.9, 10 Cells were tested for cytotoxicity against various target cells by a 6 hr 51Cr-release assay using a previously described method.13 For an inhibition assay, the CTL activity was measured in the presence of 20 μg/ml of anti-CD8, -CD4, -HLA class I (W6/32), or -HLA class II (DR) monoclonal antibody (MAb). The surface phenotypes of the cells were measured by an immunofluorescence assay with FITC-conjugated anti-CD3, anti-CD4 or anti-CD8 MAb as reported previously.9
We initially investigated the expression of SART1259, SART2, SART3 and ART4 tumor-rejection antigens at the protein levels in pancreatic cancers. Western blot analyses were carried out with polyclonal antibodies. Representative results are shown in Figure 1, and a summary is given in Table I. SART1259 antigen was expressed in the cytosolic fraction of half of the pancreatic cancer cell lines and pancreatic cancer tissues, but it was not expressed in the nontumorous pancreatic tissues. It was not expressed in the nuclear fraction from any of the tested samples. These results are consistent with our previous results regarding other cancers reported elsewhere.8 SART2, SART3 and ART4 tumor-rejection antigens were expressed in both the cytosolic and nuclear fractions from the majority of pancreatic cancer cells and cancer tissues tested. None of these antigens were expressed in either the cytosol or the nucleus of nontumorous pancreatic tissues. These results are also consistent with our previous results from studies of other cancers.9, 10, 12 Expression of the other 6 tumor-rejection antigens, from which the peptides of vaccine candidates in our study were derived (Tables II, III), was not investigated in our study. These 6 antigens are nonmutated self-antigens, the biologic functions of which are already known; thus expression of these latter antigens in pancreatic cancer cells was highly expected.11, 13, 14 For example, cyclophilinB (CypB) is a family of immunophilins involved in the cellular proliferation of both normal and malignant cells,11 whereas Lck is a family of src proteins also involved in cellular proliferation; the expression of metastatic pancreatic cancers was been reported elsewhere.13 The other 4 antigens (ppMAPkkk, WHSC2, UBE2V and HNRPL) were cloned from a pancreatic cancer cDNA.14 All of the 26 peptides used in our study were derived from these 10 tumor-rejection antigens, and all of them were able to induce HLA class I-restricted and tumor-specific CTLs in the PBMCs of cancer patients as reported previously.9–15
|Tumor antigens||Pancreatic cancer cell lines||Pancreatic cancer tissues||Nontumorous pancreatic tissues|
|Peptide||Sequence||Pt. 12||Pt. 2||Pt. 3||Pt. 4||Pt. 5||Pt. 6||Pt. 7||Pt. 8||Pt. 9||Pt. 10||HD1||HD2||HD3||HD4||HD5|
|Peptide||Sequence||Pt. 11||Pt. 12||Pt. 13||Pt. 14||Pt. 15||HD6||HD7||HD8||HD9||HD10|
PBMCs of pancreatic cancer patients (10 HLA-A24+ and 5 HLA-A2+) and healthy donors (5 HLA-A24+ and 5 HLA-A2+) were tested for their reactivity to the 13 peptides for HLA-A24+ patients or the 13 peptides for HLA-A2+ patients, respectively. Representative results of HLA-A24+ and -A2+ subjects are shown in Tables II and III, respectively. The values given in the tables represent the means of quadruplicate assays of IFN-γ production by the peptide-stimulated PBMCs in response to a corresponding peptide. The background IFN-γ production (10–200 pg/ml) in response to an HIV or HTLV-1 peptide was subtracted from the values given in the tables. CTL precursors were judged as positive (the score is double-underlined in the tables) if the mean value of IFN-γ production by the peptide-stimulated PBMCs in response to a corresponding peptide was significantly (p < 0.05) higher than that produced in response to an HIV or HTLV-1 peptide. Surface markers of these peptide-induced CTLs were mostly (>70%) CD3+CD4−CD8+ (data not shown). Among 15 patients, 1, 2, 3, 5 and 4 patients had detectable levels of CTL precursors to 5, 3, 2, 1 and 0 peptides of vaccine candidates (Tables II, III). Collectively, peptide-specific CTL precursors were detectable in 11 of 15 (73.3%) of patients, with the mean positive number of 1.5 peptides per patient (ranging from 0–5 peptides). The profile of positive peptides entirely varied among patients. In contrast, CTL precursors reactive to EBV-derived peptides, taken as control peptides to measure patients' immunocompetency, were detectable in only 1 (patient [Pt.] 13) of 15 (6.7%) patients. PBMCs from 4 patients (Pts. 2, 6, 10 and 15) had no detectable levels of CTL precursors reacting to any of the peptides tested.
Peptide-specific CTL precursors were also detectable in 9 of 10 (90%) healthy donors with the mean positive number of 2.0 peptides per donor (range 0–5 peptides) (Tables II and III). CTL precursors reactive to EBV-derived peptides were detectable in 8 of 10 donors (80%). There were no significant differences between cancer patients and healthy donors in regard to reactivity to any of the vaccine candidate peptides. However, the percentages of cases with CTL precursors reactive to EBV peptides in cancer patients were significantly (p < 0.01) lower than those in healthy donors.
The peptide-induced CTL activity was confirmed by a 6 hr 51Cr-release assay, and representative results of HLA-A24+ (Pts. 1 and 9) and HLA-A2+ (Pt. 13) patients are shown in Figure 2a, b and c, respectively. The PBMCs of Pt. 1 were stimulated with no peptide, EBV, SART1690, SART2161 and SART3315 peptides and were used as effector cells. The PBMCs contained CTL precursors reactive to the latter 3 peptides as determined by an IFN-γ release assay (Table II). The PBMCs stimulated with SART1690, SART2161 and SART3315 peptides exhibited significant levels of cytotoxicity against HLA class I-matched tumor cells but not against either HLA class I-mismatched tumor cells or HLA class I-matched phytohemagglutinin (PHA)-blastoid T cells (Fig. 2a). In contrast, the PBMCs stimulated with no peptide and EBV peptide exhibited only modest levels of cytotoxicity to HLA class I-matched tumor cells and no cytotoxicity to either HLA class I-mismatched or PHA-blastoid T cells. The PBMCs of Pt. 9 were stimulated with no peptide, EBV, SART2161 and ART475 peptides and were used as effector cells. The PBMCs contained CTL precursors reactive to cyclophilinB91 and ART475 peptide, but not SART2161 peptide, as determined by an IFN-γ release assay (Table II). The PBMCs stimulated with ART475 peptide showed significant levels of cytotoxicity against HLA class I-matched tumor cells but not against either HLA class I-mismatched tumor cells or HLA class I-matched PHA-blastoid T cells (Fig. 2b). In contrast, those PBMCs stimulated with no, EBV or SART2161 peptide exhibited only modest levels of cytotoxicity to HLA class I-matched tumor cells and no cytotoxicity to either HLA class I-mismatched tumor cells or HLA class I-matched PHA-blastoid T cells. Similarly, the PBMCs of Pt. 13 have contained CTL precursors reactive to EBV and SART3302 peptides as determined by an IFN-γ release assay (Table III). The PBMCs stimulated with SART3302 peptide exhibited significantly higher levels of cytotoxicity against HLA class I-matched tumor cells than those stimulated with no peptide or EBV peptide used as a negative control (Fig. 2c). None of them exhibited cytotoxicity against HLA class I-mismatched tumor cells or HLA class I-matched PHA-blastoid T cells. These CTL activities were inhibited by anti-class I and -CD8 MAb but not by the other MAbs tested (data not shown). These peptide-specific CTLs showed HLA-A24- or -A2-restricted IFN-γ production against cancer cell lines, and these CTL activities were blocked by anti-HLA class I MAb and anti-CD8 MAb but not by the other MAbs tested (Fig. 3). These results indicate that the peptide-stimulated PBMCs possess MHC class I-restricted and tumor-specific CTL activity. Subsequently, peptides selected by this new culture method could be applicable in use for peptide-based specific immunotherapy for pancreatic cancers.
Although peptide-specific CTL precursors were detectable in 73% of the cancer patients studied here (mean positive number: 1.5 peptides), only 26 peptides, 13 for the HLA-A2+ and 13 for the HLA-A24+ cancer patients, were used in our study. Beside these peptides, there are many others with vaccine candidates that possess the ability to induce HLA class I-restricted CTLs reactive to cancer cells.3–15 Therefore, an increased number of peptides for the assay would be associated with an increased percentage of patients with positive peptides. In addition, an increased number of peptides for the assay would be associated with an increased number of positive peptides per patient. Regardless of these limited conditions, this study showed that peptide-specific CTL precursor cells were detectable in the majority (73%) of HLA-A2+ or -A24+ cancer patients prior to peptide vaccination. These 2 HLA class I alleles are observed in >70% of Caucasians, >80% of Asians and >40% of Blacks.23, 24 All of the 13 peptides for HLA-A2 patients that were used in our study were able to induce CTLs from different subtypes of HLA-A2 (HLA-A0201, 0206 and 0207).14, 15 Therefore, the new method mentioned in our study could be useful to detect peptide-specific CTL precursors in a large number of pancreatic cancer patients. Peptide-specific CTLs induced by the employed method showed HLA-A24- or A2-restricted cytotoxicity against pancreatic cancer cell lines, and these CTL activities were blocked by anti-HLA class I MAb and anti-CD8 MAb. Since the medium containing 10 μM of each peptide was replaced every 3 days in this method, higher concentrations of the peptides could be kept throughout the culture, which in turn might facilitate the generation of HLA class I-restricted and CD8+ CTLs. This assumption is in part due to the fact that large amounts of peptides in culture achieve a high density of peptides on the groove of HLA class I molecule of antigen-presenting cells and therefore tend to stimulate CD8+ CTL as well as T-helper 1 cell responses, whereas low-density presentation tends to elicit T-helper 2 cell responses.
Our study also showed that the profile of positive peptides varied greatly among patients, suggesting that peptides suitable for use in CTL precursor-oriented peptide vaccines are different from patient to patient. These variations would be partly due to 2 factors, namely, heterogeneity of tumor cells and immunologic diversity of T cells in each patient. Subsequently, this new immunotherapeutic approach may be characteristic of an order-made cancer immunotherapy. The same approach might be applicable in the case of malaria or HIV or in other infectious diseases for which no effective vaccine protocols have been established.21
CTL precursor frequency analysis is very accurate and sensitive, but it is very laborious, costly and time-consuming and can't handle many samples.11, 13 A recently developed HLA-tetramer assay is also sensitive and accurate.25 However, it needs both labeled tetramer and a relatively large number of cells per assay, and further, it can't measure CTL activity. The method employed in our study has several demerits from a point of sensitivity and accuracy but could have several merits for monitoring the frequency of the peptide-specific CTLs, including simplicity and the ability to handle relatively large samples. This method can also measure CTL activity.
The frequency of positive peptides with vaccine candidates is not significantly different among cancer patients and healthy donors. This is expected from our previous results showing that all of the tumor-rejection antigens and their peptides used in our study are self-antigens, which are preferentially expressed in proliferating cells, including malignant and normal cells.8–15 However, CTLs induced by these peptides showed cytotoxicity against cancer cells but not against normal proliferating cells (PHA-blastoid T cells), as demonstrated in our study and also in previous studies.8–14 Therefore, vaccinations using these peptides may not be associated with adverse effects on normal cells and normal tissues. Indeed, no severe adverse effects were observed in the phase I clinical studies carried out at Kurume University Hospital; these studies analyzed peptide vaccines using 13 different peptides. The same peptides were used in vitro in the present study involving HLA-A24+ cancer patients (Gouhara et al., unpublished data).
The frequency of positive cases in cancer patients (1 of 15, 6.7%) who had CTL precursors reactive to EBV peptide was significantly lower (p < 0.01) than that of healthy donors (8 of 10, 80%). Further, the PBMCs of 4 patients had no detectable levels of CTL precursors reacting to any of the vaccine candidate peptides, nor to any of the control peptides tested. Cellular immunity of these patients might be depressed, and therefore, the other supportive immunotherapies might be needed for these patients in order to increase general levels of immunity prior to peptide-based specific immunotherapy.
In conclusion, our study investigated CTL precursors in the PBMCs of pancreatic cancer patients prior to vaccination and provided a scientific basis for considering CTL precursor-oriented peptide vaccines as an order-made cancer immunotherapy for the majority of pancreatic cancer patients.