Peptides derived from human insulin‐like growth factor‐II mRNA binding protein 3 can induce human leukocyte antigen‐A2‐restricted cytotoxic T lymphocytes reactive to cancer cells

Abstract

Insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3) is an oncofetal protein expressed in various malignancies including lung cancer. This study aimed to identify immunogenic peptides derived from IMP‐3 that can induce tumor‐reactive and human leukocyte antigen (HLA)‐A2 (A*02:01)‐restricted cytotoxic T lymphocytes (CTL) for lung cancer immunotherapy. Forty human IMP‐3‐derived peptides predicted to bind to HLA‐A2 were analyzed to determine their capacity to induce HLA‐A2‐restricted T cells in HLA‐A2.1 (HHD) transgenic mice (Tgm). We found that three IMP‐3 peptides primed HLA‐A2‐restricted CTL in the HLA‐A2.1 Tgm. Among them, human CTL lines reactive to IMP‐3 ⁵¹⁵NLSSAEVVV⁵²³ were reproducibly established from HLA‐A2‐positive healthy donors and lung cancer patients. On the other hand, IMP‐3 ¹⁹⁹RLLVPTQFV²⁰⁷ reproducibly induced IMP‐3‐specific and HLA‐A2‐restricted CTL from healthy donors, but did not sensitize CTL in the HLA‐A2.1 Tgm. Importantly, these two IMP‐3 peptide‐specific CTL generated from healthy donors and cancer patients effectively killed the cancer cells naturally expressing both IMP‐3 and HLA‐A2. Cytotoxicity was significantly inhibited by anti‐HLA class I and anti‐HLA‐A2 monoclonal antibodies, but not by the anti‐HLA‐class II monoclonal antibody. In addition, natural processing of these two epitopes derived from the IMP‐3 protein was confirmed by specific killing of HLA‐A2‐positive IMP‐3‐transfectants but not the parental IMP‐negative cell line by peptide‐induced CTL. This suggests that these two IMP‐3‐derived peptides represent highly immunogenic CTL epitopes that may be attractive targets for lung cancer immunotherapy. (Cancer Sci 2011; 102: 71–80)

Lung cancer is currently reported as the most common cause of cancer death.^{( 1 )} Despite recent improvements in systemic therapy, the prognosis for patients with advanced‐stage lung cancer remains very poor.^{( 2 )} More effective treatment modalities are urgently required, and immunotherapy represents one promising approach for future lung cancer therapies.^{( 3} , ⁴ , ^{5 )} In this study, we focus on insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3) as a target for lung cancer immunotherapy.

IMP‐3 is an oncofetal protein that is expressed in various malignancies including lung cancer.^{( 6} , ⁷ , ⁸ , ^{9 )} IMP‐3 promotes tumor cell proliferation via an insulin‐like growth factor II‐dependent pathway^{( 10 )} and has a major influence on tumor cell invasion.^{( 11 )} It is known that patients with tumors overexpressing IMP‐3 show poor prognosis,^{( 7} , ¹² , ^{13 )} and it is also known that the expression of IMP‐3 is a useful marker in identifying lung tumors that are likely to have increased biological aggressiveness.^{( 6 )} A recent study reported that immunological tolerance to IMP‐3 at the humoral level is naturally overcome in a significant proportion of lung cancer patients.^{( 14 )} In addition, Suda et al. have shown that an IMP‐3 ⁵⁰⁸KTVNDLQNL⁶⁰⁶ peptide can induce IMP‐3‐specific and human leukocyte antigen (HLA)‐A24‐restricted CTL in vitro,^{( 15 )} and subsequently, a phase I clinical trial of HLA‐A24‐restricted IMP‐3 peptide‐based immunotherapy of esophageal cancer has been conducted.^{( 16 )} Importantly, the cancer vaccination therapy used in that trial was well tolerated and IMP‐3‐specific T‐cell immune responses were observed in the HLA‐A24‐positive esophageal cancer patients.^{( 16} , ^{17 )} These results indicate that IMP‐3 may be a valuable addition to the repertoire of cancer‐specific targets for the development of new immunotherapeutic approaches.

The gene frequency of HLA‐A24 (A*24:02) is relatively high in Asian populations, especially in the Japanese, whereas it is low in Caucasians. On the other hand, HLA‐A2 (A*02:01) is one of the most common HLA alleles in various ethnic groups, including Asians, Africans, Afro‐Americans and Caucasians.^{( 18 )} Therefore, it is suggested that the HLA‐A2‐restricted and IMP‐3‐derived CTL epitopes might be very useful for immunotherapy of many patients with lung cancer and various malignancies all over the world.

In this study, we identified highly immunogenic human IMP‐3‐derived and HLA‐A2‐restricted CTL epitopes. We reproducibly established CTL lines from the peripheral blood mononuclear cells (PBMC) of healthy donors and lung cancer patients that were reactive to these epitopes and cancer cells.

Materials and Methods

cDNA microarray analysis.  Gene expression profiles were generated by cDNA microarray analysis, as described previously.^{( 19} , ²⁰ , ^{21 )} The raw data from the microarray analysis is available upon request (direct requests to Professor Y. Nakamura, Human Genome Center, Institute of Medical Science, University of Tokyo). The tissue samples from cancers and adjacent noncancerous normal tissues were obtained from surgical specimens, and all patients provided written informed consent to participate in this study.

Mice.  HLA‐A2.1 (HHD) Tgm; H‐2D^b−/−β2m^−/− double knockout mice introduced with a human β2m‐HLA‐A2.1 (α1, α2)‐H‐2D^b (α3 transmembrane cytoplasmic; HHD) monochain gene construct were generated in the Department SIDA‐Retrovirus, Unite d’ Immunite Cellulaire Antivirale, Institut Pasteur, France^{( 22} , ^{23 )} and kindly provided by Dr. F.A. Lemonnier. The mice were maintained at the Center for Animal Resources and Development of Kumamoto University, and were handled in accordance with the animal care guidelines of Kumamoto University.

Cell lines and HLA expression.  The IMP‐3 and HLA‐A2‐positive human pancreatic cancer cell line PANC1, IMP‐3‐negative and HLA‐A2‐positive human breast cancer cell line MCF7, and a transporter associated with antigen processing (TAP)‐deficient and HLA‐A2‐positive cell line T2 were purchased from Riken Cell Bank (Tsukuba, Japan). The expression of HLA‐A2 was examined by flow cytometry using an anti‐HLA‐A2 monoclonal antibody (mAb), BB7.2 (One Lambda, Inc.), in order to select HLA‐A2‐positive blood donors for the assays.

Patients and blood samples.  The Institutional Review Board of Kumamoto University approved the research protocol for collecting and using PBMC from donors. Blood samples were obtained from lung cancer patients at Kumamoto University Hospital during routine diagnostic procedures after written informed consent was obtained. We also obtained blood samples from healthy donors after receiving their written informed consent. All samples were randomly coded to mask patient identities.

Reverse transcription‐PCR.  The reverse transcription‐PCR (RT‐PCR) analysis of cell lines was performed as described previously.^{( 24 )} The IMP‐3 primer sequences were: 5′‐GTGGGA‐GGTGCTGGATAGTT‐3′ (sense) and 5′‐TGGTTTCTGCTTGG‐ATACGGAT‐3′ (antisense).

Generation of the IMP‐3 transfectant.  Lentiviral vector‐mediated gene transfer was done as previously described.^{( 25} , ^{26 )} Briefly, 17 μg of CSII‐CMV‐RfA and CSIIEFRfA self‐inactivating vectors^{( 27 )} carrying IMP‐3 cDNA and 10 μg of pCMV‐VSV‐G‐RSV‐Rev and pCAG‐HIVgp were transfected into 293T cells grown in 10‐cm culture dishes using Lipofectamine 2000 reagent (Invitrogen Corporation, Carlsbad, CA, USA). After 60 h of transfection, the medium was recovered and the viral particles were pelleted by ultracentrifugation (50 000g, 2 h). The pellet was suspended in 50 μL of RPMI 1640, and 10 μL of viral suspension was added to 5 × 10⁴ MCF7 cells per well in a flat‐bottomed 96‐well plate. Expression of the transfected IMP‐3 gene was confirmed by western blot analysis.

Induction and response of IMP‐3‐reactive mouse CTL.  Human IMP‐3‐derived peptides carrying binding motifs for HLA‐A2‐encoded molecules were selected using the BIMAS software program (BioInformatics and Molecular Analysis Section, Center for Information Technology, NIH, Bethesda, MD, USA), and 40 peptides (20 nonamers and 20 decamers, purity >90%) were synthesized (American Peptide Company; nonamer peptide sequences are indicated in Table 1). Immunizations of the mice with peptides were performed as described previously.^{( 25} , ^{28 )} The frequency of cells producing mouse interferon‐γ (IFN‐γ)/1 ×10⁵ CD4⁻ spleen cells upon stimulation with syngeneic bone‐marrow‐derived dendritic cells (BM‐DC; 1 × 10⁴/well) pulsed with or without each peptide was analyzed by an enzyme‐linked immunospot (ELISPOT) assay as previously described.^{( 29 )}

Table 1.

 Candidate peptides derived from human IMP‐3 predicted to be bound to HLA‐A2 (A*02:01)

IMP‐3‐A2‐peptide	Position	Subsequence residue listing†	HLA‐A2 binding score‡
IMP‐3‐A2‐1	199–207	RLLVPTQFV	1415
IMP‐3‐A2‐2	280–288	KILAHNNFV	681
IMP‐3‐A2‐3	552–560	KIQEILTQV	316
IMP‐3‐A2‐4	92–100	LQWEVLDSL	141
IMP‐3‐A2‐5	26–34	KIPVSGPFL	57
IMP‐3‐A2‐6	515–523	NLSSAEVVV	29
IMP‐3‐A2‐7	223–231	KQTQSKIDV	25
IMP‐3‐A2‐8	367–375	GLNLNALGL	21
IMP‐3‐A2‐9	99–107	SLLVQYGVV	21
IMP‐3‐A2‐10	374–382	GLFPPTSGM	18
IMP‐3‐A2‐11	423–431	KQGQHIKQL	17
IMP‐3‐A2‐12	143–151	QLENFTLKV	17
IMP‐3‐A2‐13	407–415	TVHLFIPAL	16
IMP‐3‐A2‐14	502–510	VIGKGGKTV	13
IMP‐3‐A2‐15	263–271	IMHKEAQDI	12
IMP‐3‐A2‐16	429–437	KQLSRFAGA	12
IMP‐3‐A2‐17	105–113	GVVESCEQV	12
IMP‐3‐A2‐18	513–521	LQNLSSAEV	12
IMP‐3‐A2‐19	409–417	HLFIPALSV	9
IMP‐3‐A2‐20	321–329	YNPERTITV	9

HLA, human leukocyte antigen; IMP‐3, insulin‐like growth factor‐II mRNA binding protein 3. †Underlined amino acids are not conserved between human and mouse IMP‐3. ‡Binding scores were estimated by using BIMAS (BioInformatics and Molecular Analysis Section) software (http://www‐bimas.cit.nih.gov/molbio/hla_bind/).

Generation of IMP‐3‐reactive human CTL and assays of CTL responses.  We isolated PBMC from HLA‐A2 (A*02:01)‐positive Japanese healthy donors and lung cancer patients, and the peripheral monocyte‐derived dendritic cells (DC) were generated as previously described.^{( 28} , ^{30 )} The DC were pulsed with 20 μg/mL of the candidate peptides in the presence of 4 μg/mL β2‐microglobulin (Sigma‐Aldrich, Tokyo, Japan) for 2 h. The cells were then irradiated (40 Gy) and incubated with the isolated CD8⁺ T cells on day 0 as previously described.^{( 25} , ^{28 )} Two additional stimulations with peptide‐loaded autologous phytohemagglutinin(PHA)‐blasts were performed on days 7 and 14 as described previously.^{( 31 )} Six days after the last stimulation, the antigen‐specific responses of the induced CTL were investigated.

CD107a mobilization assay.  To identify degranulating CD8⁺ T lymphocytes stimulated with epitope peptides, CD107a exposed on the cell surface was analyzed by flow cytometry.^{( 32} , ^{33 )} A CD107a mobilization assay was performed with an immunocyto CD107a detection kit (MBL, Nagoya, Japan) according to the manufacturer’s instructions. The induced CTL were suspended in a final concentration of 2 × 10⁶ cells/mL of AIM‐V (Gibco‐Invitrogen, Tokyo, Japan) supplemented with 2% heat‐inactivated autologous plasma, and 150 μL of the cell suspension was added to each well of a 96‐well, round‐bottomed microplate. The IMP‐3‐derived peptide or control HIV peptide (1 μg/mL) was added as a stimulant, and FITC‐labeled anti‐human CD107a mAb or FITC‐labeled isotype control mouse IgG1 and monensin were added to each well. Cells were cultured for 5 h at 37°C. After culture, the cells were stained with phycoerythrin (PE)‐conjugated anti‐human CD8a antibody (Biolegend, San Diego, CA, USA) and analyzed by flow cytometry (FACScan; BD Biosciences, San Jose, CA, USA).

Human CTL responses against cancer cell lines.  The frequency of cells producing IFN‐γ/1 × 10⁵ CTL upon stimulation with PANC1 cells (2 × 10⁴/well) or peptide‐pulsed T2 cells (2 × 10⁴/well) was analyzed by ELISPOT assay as previously described.^{( 29} , ^{34 )} The CTL were co‐cultured with cancer cells or peptide‐pulsed T2 cells as target cells (5 × 10³/well) at the indicated effector‐to‐target ratio, and a standard ⁵¹Cr‐release assay was performed as described previously.^{( 35} , ^{36 )} The HLA restriction of cytotoxic activity was tested by a blockade of the killing of HLA‐A2‐positive cancer cells using the anti‐human HLA‐class I mAb, W6/32 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and anti‐HLA‐A2 mAb, BB7.2 (Abcam, Tokyo, Japan) in comparison with the anti‐human HLA‐class II mAb (anti‐human HLA‐DR; BD Biosciences) as described previously.^{( 29} , ^{37 )}

Statistical analysis.  Two‐tailed Student’s t‐test was used to evaluate the statistical significance of differences in the ELISPOT data. P values <0.05 were considered to be statistically significant. Statistical analysis was performed with a commercial statistical software package (StatView 5.0, Abacus Concepts, Calabasas, CA, USA).

Results

Markedly enhanced expression of IMP‐3 mRNA in lung cancer tissues and other cancerous tissues.  Previously reported cDNA microarray data containing 27 648 genes of various kinds of cancerous tissues^{( 20} , ²¹ , ^{38 )} were used to analyze the expression of IMP‐3 in 13 different types of cancer tissues as described previously.^{( 25} , ^{28 )} Table 2 shows the overexpression of IMP‐3 in various cancer tissues compared with their normal counterparts. The cDNA microarray analysis revealed markedly enhanced expression of the IMP‐3 gene in lung cancer tissues in 11 of the 12 non‐small‐cell lung cancer patients (average relative expression ratio, 42 600; range, 0.7–154 000) and six of the 10 small‐cell lung cancer patients (average relative expression ratio, 30 600; range, 2.0–195 000). The expression level of the IMP‐3 gene was also enhanced in the majority of malignancies analyzed, including cholangiocellular carcinoma, malignant lymphoma and osteosarcoma. Next, the expression of IMP‐3 was analyzed in 29 kinds of normal tissues including four embryonic tissues using the cDNA microarray data (Fig. 1a). We found that the IMP‐3 gene was overexpressed only in the placenta, and faintly expressed in the testis and four embryonic tissues (fetal brain, fetal kidney, fetal liver and fetal lung).

Table 2.

 Overexpression of the IMP‐3 gene in lung cancer and various malignancies determined by cDNA microarray analysis

Malignant tumor	n	Positive rate† (%)	Average of relative expression ratio
Cholangiocellular carcinoma	13/3	100	28 900
Malignant lymphoma	6/6	100	56 500
Osteosarcoma	4/4	100	77 800
Non‐small‐cell lung cancer	11/12	95	42 600
Soft tissue tumor	7/9	78	13 300
Small‐cell lung cancer	6/10	60	30 600
Esophageal cancer	13/35	37	6 800
Gastric cancer	2/6	33	4 200
Chronic myelocytic leukemia	6/22	27	37 700
Urinary bladder cancer	2/8	25	3.3
Cervical cancer	1/7	14	2.9
Prostate cancer	0/5	0	1.3
Breast cancer	0/4	0	0.4

IMP‐3, insulin‐like growth factor‐II mRNA binding protein 3.

†Relative expression ratio (cancer/normal tissue) >5 was considered to be positive.

Figure 1.

 Markedly enhanced expression of insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3) mRNA in lung cancer tissues and various cancer cell lines. (a) Markedly enhanced expression of IMP‐3 mRNA in lung cancer tissues compared with various normal tissues revealed by cDNA microarray analysis. (b) RT‐PCR analysis of IMP‐3 mRNA expression in lung cancer cell lines and cancer cell lines derived from melanoma, breast, oral, pancreatic, colon, esophageal and liver cancers.

The expression of IMP‐3 mRNA in cancer cell lines was also analyzed by semiquantitative RT‐PCR (Fig. 1b). The strong expression of IMP‐3 was detected in various cancer cell lines including lung cancer, melanoma, oral cancer, pancreatic cancer, colon cancer, esophageal cancer and liver cancer cell lines by RT‐PCR. On the other hand, RT‐PCR analysis of cancer cell lines derived from breast cancers (MCF7 and CRL1500) revealed that the IMP‐3 gene was not expressed, and this is in accordance with the data from the cDNA microarray analysis (Table 2).

Screening for immunogenic IMP‐3 peptides using HLA‐A2.1 (HHD) Tgm.  To identify IMP‐3‐derived and HLA‐A2‐restricted CTL epitopes, we selected a total of 40 different candidate peptides that were nine or 10 amino acids long and expected to have a higher binding affinity to HLA‐A2, using the BIMAS software program. The scores of the selected 20 nonamer peptides calculated based on BIMAS algorithms predicting the HLA‐A2‐binding affinity are shown in Table 1. We also wanted to identify HLA‐A2‐restricted epitopes from the 40 selected peptides by using HLA‐A2.1 Tgm. We set up immunizations of the HLA‐A2.1 Tgm with these 40 peptides. Series of repeated experiments revealed that three nonamer peptides (IMP‐3‐A2‐3 ⁵⁵²KIQEILTQV⁵⁶⁰, IMP‐3‐A2‐5 ²⁶KIPVSGPFL³⁴ and IMP‐3‐A2‐6 ⁵¹⁵NLSSAEVVV⁵²³) primed CTL with a relatively higher IFN‐γ‐producing ability (Fig. 2). However, none of the selected 20 decamer peptides could prime the CTL in the HLA‐A2.1 Tgm (data not shown). These results suggest that the IMP‐3‐A2‐3, ‐5 and ‐6 peptides might be the HLA‐A2‐restricted CTL epitope peptides in the HLA‐A2.1 Tgm. Thus, these peptides were selected for analysis of their ability to stimulate human CTL.

Figure 2.

 Identification of human leukocyte antigen (HLA)‐A2‐restricted mouse cytotoxic T lymphocytes (CTL) epitopes of human insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3) using HLA‐A2.1 (HHD) Tgm. The frequency of CTL producing IFN‐γ was analyzed by enzyme‐linked immunospot (ELISPOT) assay. The asterisks indicate that IMP‐3‐A2‐3 ⁵⁵²KIQEILTQV⁵⁶⁰, IMP‐3‐A2‐5 ²⁶KIPVSGPFL³⁴ and IMP‐3‐A2‐6 ⁵¹⁵NLSSAEVVV⁵²³ were able to induce peptide‐reactive CTL with statistical significance (*P < 0.05).

Identification of IMP‐3‐derived and HLA‐A2‐restricted CTL epitopes in healthy donors and lung cancer patients.  Next we attempted to generate IMP‐3‐specific CTL from the PBMC of healthy donors and lung cancer patients positive for HLA‐A2 by stimulating the cells with IMP‐3‐A2‐3, ‐5 and ‐6 peptides. In addition, we attempted to generate IMP‐3‐specific CTL from the PBMC of healthy donors by stimulating the cells with several peptides that were unable to induce mouse CTL in the HLA‐A2.1 Tgm. We did this because we considered the possibility of the existence of a CTL epitope that exhibited high immunogenicity in humans only.

After stimulations three or four times with autologous monocyte‐derived DC and PHA blasts that were pulsed with each peptide, the frequency of CD8⁺ T cells specific to the IMP‐3‐derived peptides in the resulting CTL lines was examined by IFN‐γ ELISPOT assay. For background controls, CTL were stimulated with T2 cells pulsed with an irrelevant HLA‐A2‐restricted HIV (HIV‐A2) peptide. Among the IMP‐3‐derived peptides tested, the CTL lines that were generated by stimulation with the IMP‐A2‐1 ¹⁹⁹RLLVPTQFV²⁰⁷ or IMP‐3‐A2‐6 ⁵¹⁵NLSSAEVVV⁵²³ peptides reproducibly produced large amounts of IFN‐γ upon stimulation with T2 cells that had been pulsed with their cognate peptides (Fig. 3a). To further analyze the CTL‐stimulating capacity of these two peptides, a CD107a mobilization assay was performed to evaluate the antigen‐specific secretion of the cytolytic granule content of the CTL.^{( 32} , ^{33 )} A significantly higher proportion of CD8⁺ T cells was stained with anti‐CD107a mAb in the CTL lines generated by stimulation with one of the two immunogenic peptides re‐stimulated with their cognate peptides, as compared with re‐stimulation with the irrelevant HIV‐A2 peptide (Fig. 3b).

Figure 3.

 Induction of insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3)‐specific human cytotoxic T lymphocytes (CTL) from CD8⁺ T cells of human leukocyte antigen (HLA)‐A2‐positive lung cancer patients and healthy donors. (a) Enzyme‐linked immunospot (ELISPOT) assay of IMP‐3 peptide‐reactive CTL generated from the peripheral blood mononuclear cells (PBMC) of HLA‐A2‐positive healthy donors and a lung cancer patient. Bars indicate the number of IFN‐γ spots when the generated CTL lines were re‐stimulated with T2 cells loaded with cognate IMP‐3 peptides (open bars) or irrelevant human immunodeficiency virus (HIV)‐A2 peptides (closed bars). The effector‐to‐target cell ratio was 10:1. Data are presented as the mean ± SD of triplicate assays. Statistically significant differences are indicated with asterisks. (b) Detection of CD107a exposed on the cell surface of CD8⁺ T cells after antigenic stimulation. All peptides were used at a final concentration of 1 μg/mL. Events shown are gated for CD8⁺ T cells. Upper and middle panels: cells stimulated with the cognate IMP‐3 peptide. Lower panels: cells stimulated with the irrelevant HIV‐A2 peptide. The numbers inside the plots indicate the percentage of the cell population with the quadrant characteristic (CD8⁺ CD107a⁺ T lymphocytes). A representative of two independent experiments with similar results is shown. (c) Cytotoxicity of IMP‐3‐specific CTL against T2 cells pulsed with the cognate IMP‐3‐derived peptides. Cytotoxicity of CTL against T2 cells pulsed with IMP‐3‐A2‐1 (Δ; left and middle panels) or IMP‐3‐A2‐6 (Δ; right panel), and T2 cells pulsed with irrelevant HIV‐A2 peptides () in a ⁵¹Cr‐release assay. Each value represents the percentage of specific lysis calculated from the mean values of a triplicate assay.

Cytotoxic activity against peptide‐pulsed T2 cells was examined by ⁵¹Cr‐release assays (Fig. 3c). The CTL induced from the PBMC of healthy donors exhibited cytotoxic activity to T2 cells that were pulsed with the IMP‐3‐A2‐1 or IMP‐3‐A2‐6 peptides, but not to T2 cells pulsed with the irrelevant HIV‐A2 peptide. This indicates that these CTL have peptide‐specific cytotoxicity.

Among the 40 candidate peptides of nine or 10 amino acids in length that were expected to have higher binding affinity to HLA‐A2, the two IMP‐3‐derived peptides were able to elicit specific CTL in human cells, suggesting that these two peptides might be highly immunogenic human CTL epitopes and, therefore, attractive targets for lung cancer immunotherapy.

Natural processing of IMP‐3 CTL epitopes in cancer cells.  We next examined whether the two immunogenic peptides were produced from the IMP‐3 protein by natural intracellular processing. IMP‐3‐specific IFN‐γ production and killing activity against cancer cells by CTL induced from HLA‐A2‐positive healthy donors and lung cancer patients were analyzed by IFN‐γ ELISPOT assay and ⁵¹Cr‐release assays, respectively. As shown in Figure 4, the CTL lines generated from the healthy donors and a lung cancer patient by stimulation with either IMP‐3‐A2‐1 or IMP‐3‐A2‐6 exhibited cytotoxic activity against PANC1, SW620, SKHep1 and MCF7/IMP‐3 (MCF7 cells transfected with the IMP‐3 gene; HLA‐A2+, IMP‐3+) but not against A549 (HLA‐A2−, IMP‐3+) or MCF7 cells (HLA‐A2+, IMP‐3−).

Figure 4.

 Cytotoxic activities of insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3)‐specific cytotoxic T lymphocytes (CTL) against cancer cell lines. (a) Cytotoxic activities of IMP‐3‐specific CTL against PANC1 (○), SW620 (Δ), SKHep1 (◊), MCF7 (•) or A549 (♦) analyzed by ⁵¹Cr‐release assay. (b) Cytotoxic activities of IMP‐3‐specific CTL against MCF7/IMP3 (○; MCF7 cells transfected with IMP‐3 gene) or MCF7 (•) analyzed by ⁵¹Cr‐release assay.

To confirm that the CTL specific to these two IMP‐3‐derived peptides recognized target cells naturally expressing IMP‐3 in a HLA‐class‐I‐restricted manner, we used mAb specific to HLA‐class I (W6/32) and HLA‐A2 (BB7.2) to block recognition by CTL. The IFN‐γ production and cytotoxicity were significantly inhibited by the mAb against HLA‐class I and HLA‐A2, but not by control anti‐HLA‐class II mAb (Fig. 5). These results clearly indicate that these peptides were naturally processed from IMP‐3 protein in cancer cells and presented in the context of HLA‐A2 to be recognized by peptide‐induced CTL.

Figure 5.

 Inhibition of insulin‐like growth factor‐II mRNA binding protein 3 (IMP‐3)‐reactive cytotoxic T lymphocyte (CTL) responses by anti‐human leukocyte antigen (HLA) class I mAb or anti‐HLA‐A2 mAb. IFN‐γ production (a) and cytotoxicity (b and c) mediated by CTL are indicated. (○), PANC1; (•), PANC1+ W6/32; (□), PANC1+ control mAb. Bars indicate IFN‐γ production (a) or cytotoxicity (c) when the generated CTL lines were co‐cultured with PANC1 (open bars), PANC1+ control mAb (open bars) or PANC1+ blocking mAb (closed bars). Representative data from two independent experiments with similar results are shown. Statistically significant differences in Figure 5a are indicated with asterisks.

Finally, the rates of successful induction of IMP‐3‐specific CTL lines were high (IMP‐3‐A2‐1, 100% in the cases of two healthy donors, not tested in lung cancer patients; IMP‐3‐A2‐6, 75% in the cases of four healthy donors, 50% in the cases of four lung cancer patients).

Discussion

IMP‐3 is highly expressed at both the mRNA and protein levels in most cancer cell lines and in the cancer tissues tested, but barely detectable in most normal adult tissues, except for the testis and placenta.^{( 8} , ^{14 )} Significantly higher expression levels of IMP‐3 proteins have been demonstrated in lung cancer and several malignancies than in normal tissues.^{( 6} , ⁷ , ¹³ , ^{39 )} It is interesting to note that IMP‐3 was also found to be overexpressed in cholangiocellular carcinoma, malignant lymphoma and osteosarcoma (Table 2). Thus, the therapeutic range of IMP‐3 might extend to other cancer types. In view of these findings that IMP‐3 is strongly and selectively expressed in various cancer tissues, IMP‐3 might be an ideal target molecule for the treatment of patients with various cancers.

In this study, we first predicted candidate epitopes on the IMP‐3 antigen using a professional software program (BIMAS). We obtained 40 candidate epitopes based on high immunogenicity scores (20 nonamers and 20 decamers), and screened for their immunogenicity using HLA‐A2.1 (HHD) Tgm. A series of repeated experiments revealed a reproducible CTL response to three nonamer peptides. We subsequently attempted to generate IMP‐3‐specific CTL from the PBMC of healthy donors and lung cancer patients positive for HLA‐A2 by stimulation with these three peptides. Additionally, we attempted to generate IMP‐3‐specific CTL from healthy donors by stimulation with several candidate peptides that could not induce HLA‐A2‐restricted CTL in the HLA‐A2.1 Tgm, as we considered the possibility of CTL epitopes exhibiting high immunogenicity only in humans. As a result, we identified two HLA‐A2‐restricted and highly immunogenic epitopes, one being immunogenic in both HLA‐A2.1 Tgm and humans, and the other being immunogenic only in humans. Thus, major HLA‐A2‐restricted CTL epitopes for humans are not always the same as those for HLA‐A2.1 Tgm, although Tgm are very useful mice as they avoid the use of precious human blood samples.

The amino acid sequence of the IMP‐3‐A2‐1 peptide is completely conserved between human and mouse IMP‐3. The difference in the CTL response to IMP‐3‐A2‐1 between HLA‐A2‐positive human and HLA‐A2 transgenic mice is most likely due to the difference in the peripheral T cell repertoire. Several factors affect the T cell receptor (TCR) repertoire. The structure and variation of the TCR gene is different between human and mice. There is also a difference in the peptide repertoire presented in the context of MHC molecules expressed on thymic epithelial cells that induce a positive selection of thymocytes.

In this study, the rates of successful induction of IMP‐3‐specific CTL lines from healthy donors were very high. These results suggest that the frequency of IMP‐3‐specific CD8⁺ T cells in PBMC of the healthy donors was sufficiently high for in vitro CTL induction. Molecular mimicry that causes cross‐reaction of microbial antigen‐reactive T cells to self‐antigens^{( 40 )} might account for such high frequency of IMP‐reactive T cells in healthy individuals.

CD8⁺ CTL have an important role in the immunodefense against cancers.^{( 41 )} The frequent expression of IMP‐3 in lung cancers and the finding that novel immunogenic peptides can elicit an effective immune response against IMP‐3‐expressing tumor cells from CD8⁺ T cells of lung‐cancer‐bearing patients suggests that IMP‐3 might be an antigen suitable for T‐cell‐mediated immunotherapy.

In conclusion, our results suggest that the IMP‐3‐A2‐1 and IMP‐3‐A2‐6 peptides might be capable of inducing HLA‐A2‐restricted CD8⁺ CTL, which kill tumor cells that express IMP‐3 and HLA‐A2 in lung cancer patients. We expect that the combination of IMP‐3‐derived peptide‐based immunotherapy with standard therapies, such as chemotherapy or radiation, will have enormous potential to improve the current outcomes of conventional lung cancer therapies.

Disclosure Statement

The authors have no financial or any other kind of personal conflicts with this article.