Expression of ER stress markers (GRP78/BiP and PERK) in patients with tongue cancer.

The glucose-regulated protein (GRP78/BiP) and PKR-like endoplasmic reticulum kinase (PERK) plays a crucial role in the endoplasmic reticulum (ER) stress response. GRP78/BiP is highly elevated in various human cancers. Our study is to examine the clinicopathological significance of GRP78/BiP and PERK expression in patients with tongue cancer. A total of 85 tongue cancer patients were analyzed, and tumor specimens were stained by immunohistochemistry for GRP78/BiP, PERK, GLUT1, Ki-67 and microvessel density (MVD) determined by CD34.GRP78/BiP and PERK were highly expressed in 47% and 35% of all patients, respectively. GRP78/BiP disclosed a significant relationship with PERK expression, lymphatic permeation, vascular invasion, glucose metabolism and cell proliferation. The expression of GRP78/BiP was significantly higher in metastatic sites than in primary sites (79% vs. 47%, p=0.003). We found that the high expression of GRP78/BiP was proven to be an independent prognostic factor for predicting poor outcome in patients with tongue cancer. In the analysis of PFS, PERK was identified as an independent predictor. The increased GRP78/BiP expression was clarified as an independent prognostic marker for predicting worse outcome. Our study suggests that the expression of GRP78/BiP as ER stress marker is important in the pathogenesis and development of tongue cancer.

Prognostic significance of GRP78/BiP expression in patients with Stage III/IV hypopharyngeal squamous cell carcinoma.

The immunoglobulin heavy chain binding protein (BiP)/glucose-regulated protein 78 (GRP78) plays an essential role in the endoplasmic reticulum (ER) stress, and GRP78/BiP is known to be highly expressed in various human neoplasms. The clinicopathological features of GRP78/BiP expression in patients with advanced hypopharyngeal squamous cell carcinoma (HSCC) remain unclear. The aim of this study is to elucidate the prognostic significance of GRP78/BiP for HSCC.A total of 68 patients with advanced HSCC (stage III/IV) were analyzed, and tumor specimens were stained with immunohistochemistry for GRP78/BiP, Ki-67, and microvessel density (MVD), as determined through CD34 and p53 levels. GRP78/BiP was highly expressed in 80.8% (55/68) of all patients. The expression level of GRP78/BiP disclosed no significant relationship with any variables. Multivariate analysis confirmed that low expression of GRP78/BiP was an independent prognostic factor for predicting poor overall survival and progression-free survival in patients with advanced HSCC. The decreasing expression of GRP78/BiP was identified as a significant predictor related to shorter survival duration after surgery for advanced HSCC. Our study suggests that the reduced expression of GRP78/BiP contributes to worse survival for patients with advanced head and neck cancer.

Prognostic significance of amino-acid transporter expression (LAT1, ASCT2, and xCT) in surgically resected tongue cancer.

BACKGROUND: Amino-acid transporters are necessary for the tumour cell growth and survival, and have a crucial role in the development and invasiveness of cancer cells. But, it remains unclear about the prognostic significance of L-type amino-acid transporter 1 (LAT1), system ASC amino-acid transporter-2 (ASCT2), and xCT expression in patients with tongue cancer. We conducted the clinicopathological study to investigate the protein expression of these amino-acid transporters in tongue cancer. METHODS: Eighty-five patients with surgically resected tongue cancer were evaluated. Tumour sections were stained by immunohistochemistry for LAT1, ASCT2, xCT, 4F2hc/CD98hc (4F2hc), Ki-67, and microvessel density (MVD) determined by CD34, and p53. RESULTS: L-type amino-acid transporter 1 and 4F2hc were highly expressed in 61% (52 out of 85) and 45% (38 out of 47), respectively. ASC amino-acid transporter-2 and xCT were positively expressed in 59% (50 out of 85) and 21% (18 out of 85), respectively. The expression of both LAT1 and ASCT2 was significantly associated with disease staging, lymph-node metastasis, lymphatic permeation, 4F2hc expression and cell proliferation (Ki-67). xCT expression indicated a significant association with advanced stage and tumour factor. By univariate analysis, disease staging, lymphatic permeation, vascular invasion, LAT1, ASCT2, 4F2hc, and Ki-67 had a significant relationship with overall survival. Multivariate analysis confirmed that LAT1 was an independent prognostic factor for predicting poor prognosis. CONCLUSIONS: L-type amino-acid transporter 1 and ASCT2 can serve as a significant prognostic factor for predicting worse outcome after surgical treatment and may have an important role in the development and aggressiveness of tongue cancer.

Evaluation of testing face-mask filter samples with LAMP shows high rates of detection in pulmonary TB.

Detection of Mycobacterium tuberculosis (MTB) in bioaerosols derived from patients with active pulmonary TB is a potential alternative diagnostic method for patients with presumed TB who cannot expectorate sputum.To assess the efficacy of a bioaerosol particle collection method to capture MTB and diagnose TB.A mask-like filter holder (3D mask) with a water-soluble gelatine filter (GF) and one containing a water-insoluble polypropylene filter (PPF) were prepared. Eligible patients wore the 3D mask with GF or PPF within 3 days of starting anti-TB drugs. The GF and PPF filters were collected after 2 and 8 h. DNA was extracted from the filter samples and tested using loop-mediated isothermal amplification (LAMP).Filter samples were collected from 57 and 20 patients with and without active pulmonary TB, respectively. The GF and PPF sensitivity was 76.2% and 83.3%, respectively. The specificity of both methods was 100%. Of the 57 patients diagnosed with non-expectorated sputum samples, including suction phlegm, gastric lavage, and bronchial lavage fluid, 55.6% and 50.0% were positive by GF and PPF, respectively.We present a 3D mask filter sampling method for exhaled bioaerosol particles that can be used in clinical practice to diagnose patients with presumed TB..

Tongue surface model can predict radiation tongue mucositis due to intensity-modulated radiation therapy for head and neck cancer.

Acute radiation tongue mucositis has a profound effect on talking and eating. We examined whether the dose-volume histogram obtained from the tongue surface model correlates with mucositis severity, and whether it is useful for predicting acute radiation tongue mucositis in patients with head and neck cancer treated with intensity-modulated radiation therapy. Thirty-six patients who received intensity-modulated radiation therapy for head and neck cancer were analysed for acute radiation tongue mucositis according to the Common Terminology Criteria for Adverse Events, version 4.0, as well as the Radiation Therapy Oncology Group scoring systems. The corresponding high-dose locations in anatomical sub-regions in the tongue surface model and the development of high-grade acute radiation tongue mucositis were compared. The mucositis sites coincided with the high-dose anatomical sub-regions in the tongue surface model. There was a clear dose-response relationship between the mean dose to the tongue and the acute radiation tongue mucositis Radiation Therapy Oncology Group grade. According to the dose-volume histogram, patients receiving 16.0-73.0 Gy to the tongue were susceptible to grade 2-3 toxicity. The tongue surface model can predict the site and severity of acute radiation tongue mucositis. In future, radiation treatment plans ccould be optimized using this model.

Human tumor antigen-specific T lymphocytes and interleukin-2-activated natural killer cells: comparisons of antitumor effects in vitro and in vivo.

Human antitumor effector cells include class I major histocompatibility complex (MHC)-restricted T cells and non-MHC-restricted natural killer (NK) cells. These two types of effector cells have not been directly compared for the ability to eliminate tumor cell targets. Here, we compare in vitro and in vivo antitumor functions of two human T-cell lines specific for a shared tumor antigen to the antitumor functions of A-NK cells, a subset of IL-2-activated NK cells. Human squamous cell carcinoma of the head and neck cell lines cultured in suspensions or as spheroids or tumor xenografts established in nude mice were used to evaluate antitumor functions of IL-2-activated and expanded T and NK effector cells in various assays, both in vitro and in vivo. Both tumor cell targets, PCI-13 and OSC-19, expressed class I and II MHC antigens after IFN-gamma pretreatment, gave rise to tumors upon injection into immunosuppressed nude mice, and were resistant to lysis by resting NK cells but sensitive to lysis mediated by A-NK cells or HLA-A2-restricted T-cell lines specific for a shared squamous cell carcinoma of the head and neck antigen. No significant differences were observed in the ability of A-NK cells or tumor-specific T cells to bind to tumor cell monolayers or to enter into spheroids. However, A-NK cells mediated significantly higher killing than tumor-specific CD8+ T cells in 4-h 51Cr-release assays (a measure of cell membrane damage and necrosis), 1-h [3H]thymidine-release assays (a measure of DNA fragmentation and apoptosis), and in terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assays (a measure of apoptosis). In contrast, CD8+ T cells were consistently more effective than A-NK cells in inducing growth inhibition of tumor cells in 24-h MTT assays. In the presence of tumor-specific antibodies, A-NK cell binding, entry into spheroids, and infiltration into tumor in vivo were significantly increased. In vivo perilesional delivery of effector cells to mice with established tumors indicated that human A-NK cells exert antitumor effects as potent as those of tumor-specific T cells. However, in contrast to tumor-specific T cells, A-NK cells are readily available for cancer therapy, expand rapidly in culture without prior sensitization, and can be armed with antitumor antibodies to increase localization of effector cells to the tumor.

A wild-type sequence p53 peptide presented by HLA-A24 induces cytotoxic T lymphocytes that recognize squamous cell carcinomas of the head and neck.

Evidence has accumulated indicating that HLA-A2-restricted CTLs specific for human wild-type sequence p53 epitopes lyse tumor cells expressing mutant p53. To explore the possibility that wild-type sequence p53 peptides could also be used in vaccines for patients expressing HLA-A24 antigen, another frequent HLA class I allele, we investigated the induction of HLA-A24-restricted p53-specific CTLs from the peripheral blood lymphocytes of normal donors. Of six p53-derived peptides possessing an HLA-A24 binding motif, the p53 peptide 125-134 (p53(125-134)) was found to have a high binding capacity and induced peptide-specific CTLs from peripheral blood mononuclear cells, using peptide-pulsed autologous dendritic cells and subsequent cultivation with cytokines interleukin 2 and interleukin 7. Bulk CTL populations lysed peptide-pulsed HLA-A24+ targets as well as HLA-A24+ squamous cell carcinoma of the head and neck (SCCHN) cell lines. However, IFN-gamma pretreatment of HLA-A24+ SCCHN cell lines was necessary for lysis, suggesting that a ligand density higher than that normally expressed by tumor cells is required for these CTLs to mediate lysis. Moreover, a cloned CTL, designated TH#99, isolated from the bulk population by limiting dilution, lysed HLA-A24+ SCCHN targets more efficiently than the bulk CTL population. Lysis was inhibited by anti-HLA class I monoclonal antibody but not by anti-HLA-DR monoclonal antibody. These results indicate that HLA-A24-restricted CTLs recognizing the wild-type sequence p53(125-134) can be generated using autologous dendritic cells from precursors present in peripheral blood lymphocytes obtained from normal HLA-A24+ donors. This finding suggests that vaccine strategies targeting wild-type sequence p53 epitopes can be extended to a wider range of cancer patients.

Generation of anti-p53 cytotoxic T lymphocytes from human peripheral blood using autologous dendritic cells.

CTLs recognizing the HLA-A2.1-restricted, wild-type sequence p53 epitopes p53(149-157) and p53(264-272) were generated from CD8-enriched populations of nonadherent peripheral blood lymphocytes (PBLs) obtained from healthy donors. The PBLs were restimulated in vitro with peptide-pulsed granulocyte macrophage colony-stimulating factor- and interleukin (IL)-4-induced autologous dendritic cells in the presence of IL-6 and IL-12 and subsequently cultivated with IL-1alpha, IL-2, IL-4, IL-6, and IL-7. Bulk anti-p53(264-272) CTL populations were generated from PBLs obtained from two of five donors. Both CTL populations were cytotoxic against peptide-pulsed HLA-A2+ target cells, but not against untreated target cells. A CD8+ anti-p53 CTL clone designated p264#2 was isolated from one of the bulk populations. It was found to have an intermediate affinity of approximately 10(-9) M for the epitope and to mediate cytotoxicity against several human tumor cell lines, including the squamous cell carcinoma of the head and neck cell line SCC-9, which is known to present the wild-type sequence p53(264-272) epitope. In addition, CTLs reactive against p53(149-157)-pulsed targets as well as a HLA-A2+ tumor cell line were cloned from a bulk population of antitumor CTLs obtained from one of the five normal PBLs restimulated with this epitope. The results indicate that CTLs recognizing wild-type sequence epitopes can be generated from precursors present in PBLs obtained from some normal individuals using autologous dendritic cells as antigen-presenting cells and suggest that vaccine strategies targeting these epitopes can lead to antitumor CTL generation, thereby emphasizing the therapeutic potential of p53-based cancer vaccines.

In vitro and in vivo characteristics of human squamous cell carcinoma of the head and neck cells engineered to secrete interleukin-2.

Two human squamous cell carcinoma of the head and neck (SCCHN) cell lines, PCI-13 and PCI-52, were transduced with the retroviral construct containing human interleukin-2 (IL-2) cDNA and selected for neomycin resistance in G418 medium. Stably transduced SCCHN cells produced and secreted IL-2, which was shown to have biologic activity in a bioassay, using an IL-2-dependent CTLL-2 cell line. By immunohistochemistry, IL-2 gene-transduced PCI-13 cells were strongly positive for IL-2, and by flow cytometry showed both cell surface and intracytoplasmic expression of IL-2 protein. Expression of IL-2 mRNA was measured by quantitative RT-PCR and found to be considerably increased in transduced SCCHN relative to that in parental cells. There was no difference in expression of IL-2R between the parental and IL-2 gene-transduced cells. In vitro proliferation of IL-2 gene-transduced tumor cells was consistently more rapid than that of parental cells. Sensitivity of the parental and IL-2 gene-transduced targets to lysis or apoptosis mediated by purified human natural killer (NK) cells or IL-2-activated NK (A-NK) cells was comparable as measured in 4-hour 51Cr-release and 1-hour [3H]thymidine-release assays, respectively. However, transduced cells were significantly more sensitive than parental cells to these effectors in 24-hour MTT assays, most likely due to IL-2 production by the transduced targets. PCI-52 cells selected for in vivo experiments formed large subcutaneous tumors in immunosuppressed nude mice. Tumors established by subcutaneous injections of 1 x 10(7) IL-2 gene-transduced cells regressed completely by day 25, while those formed by parental or LacZ gene-transduced tumor cells grew progressively. Tumor regression was mediated by numerous mononuclear cells, identified as murine NK cells and macrophages by immunohistochemistry, which accumulated around the IL-2-secreting, but not parental, tumors within 5-6 days after tumor cell injections. Thus, IL-2 gene-transduced SCCHN cells produce functional IL-2 in vivo in amounts sufficient to support the recruitment to the tumor site and antitumor activity of cytotoxic effector cells. IL-2-secreting SCCHN cells may be a useful component of vaccines designed to induce and sustain effector cell activation at the tumor site.

Antitumor effects of cytolytic T lymphocytes (CTL) and natural killer (NK) cells in head and neck cancer.

Immune effector cells were generated in vitro and compared for activity against human head and neck cancer (HNC). A subset of human IL2-activated natural killer (A-NK) cells was able to eliminate tumor cells in vitro by necrosis, apoptosis or cytokine-mediated effects and to induce regression of established tumors in vivo, in a xenograft nude mouse model of HNC. Cytolytic T lymphocytes (CTL) with specificity for a human oral carcinoma were as effective as A-NK cells in mediating killing of autologous tumor cells or entering spheroids. Stable HNC transfectants with the human IL2 gene, which produced IL2 were implanted in nude mice. IL2-secreting tumors regressed spontaneously. Strategies available for use in immunotherapy of human HNC are being evaluated to provide a solid basis for their future therapeutic applications.

Expression of genes MAGE-1, -2, and -3 by human maxillary carcinoma cells.

We analysed the expression of melanoma antigen-encoding (MAGE) gene-1, -2, and -3 in 20 maxillary carcinomas consisting of two cell lines: freshly isolated cancer cells from specimens from 13 patients, and 5 biopsy specimens. The cells were subjected to reverse transcription by the polymerase chain reaction. Fourteen (70%) out of 20 maxillary carcinomas expressed at least one of the MAGE genes. In contrast, five control samples of inflammed mucosa from the maxillary sinus of patients with chronic sinusitis were all negative for the expression of these genes. Results indicated that patients with maxillary carcinoma may be good candidates for specific immunotherapy.

Low doses of anticancer drugs increase susceptibility of tumor cells to lysis by autologous killer cells.

Pretreatment of squamous cell carcinoma (SCC) cells from four patients with low doses of cisplatin, carboplatin or 5-fluorouracil increased the susceptibility to lysis by autologous killer cells in vitro. Exposure of two SCC cell lines to low doses of these drugs increased the cell surface expression of both HLA class I and intercellular adhesion molecule-1 (ICAM-1). HLA class II, neural cell adhesion molecule and B7 were not expressed on the cell surface before or after such treatment. The results suggest that these drugs increase the susceptibility of tumor cells to autologous cell-mediated cytotoxicity, at least in part, by enhancing the expression of HLA class I and ICAM-1.

Expression of the MAGE-1, -2, -3, -4, and -6 genes in non-squamous cell carcinoma lesions of the head and neck.

The messenger RNA level of several MAGE genes, some of which have been proven to encode tumor rejection antigens recognized by cytotoxic T lymphocytes, were examined in 41 benign and malignant lesions of the head and neck region. By a reverse transcription-polymerase chain reaction assay and Southern blot hybridization, MAGE-1, -2, -3, -4, and -6 genes were expressed in 25%, 41.7%, 33.3%, 8.3% and 33.3% of 12 non-squamous cell carcinomas, respectively. These tumors consisted of 6 papillary adenocarcinomas, 3 adenoid cystic carcinomas, 2 adenocarcinomas, and 1 mucoepidermoid tumor. Of 7 non-Hodgkin's lymphomas, one case from the oropharynx and 2 from the nasopharynx expressed for the MAGE-1 and MAGE-2 genes, respectively. In contrast, none of 12 benign tumors expressed any of these MAGE genes. Interestingly, of 10 other lesions including hyperplasia, keratosis, and ulcer, one histologically diagnosed as dysplasia expressed the MAGE-2, -3, -4, and -6 genes. These results suggest that the MAGE genes may be expressed in malignant tumors and precancerous lesions but not in benign tumors. In addition, non-squamous cell carcinomas may be suitable targets for specific immunotherapy against MAGE gene products.

Establishment of a human cell line from maxillary squamous cell carcinoma and its biological features as a stimulator for induction of cytotoxic T lymphocytes.

A cancer cell line named FS-1 was established from maxillary cancer of which histological diagnosis was well differentiated squamous cell carcinoma (SCC). The HLA class I typing showed that FS-1 expressed the same HLA class I antigens as those of the host lymphocytes. The chromosome analysis and nuclear DNA contents suggested that FS-1 was not a normal human cell. FS-1 is characteristic of SCC in that it grows adhesively on the surface of a culture flask. The structure of tumor tissue obtained from FS-1-transplanted nude mice is very much like the original tissue of SCC. The SCC-antigen was determined in the culture supernatant of FS-1. Autologous tumor killing activity was induced from peripheral blood lymphocytes of some patients suffering from head and neck SCC after mixed lymphocyte tumor culture with FS-1. Thus, FS-1 can serve as a useful allogeneic stimulator of SCC for induction of autologous tumor killing activity.

Angioimmunoblastic lymphadenopathy-like T-cell lymphoma. A case report and immunologic study.

BACKGROUND: Angioimmunoblastic lymphadenopathy (AILD) is rare in the head and neck and its definition remains controversial. METHOD: A case of AILD with an ulcer of the lateral pharyngeal wall was studied for viral infection, immunohistologic findings and T-cell receptor (TCR) V beta rearrangement. RESULTS: We observed elevation of antibodies against herpes simplex virus and herpes zoster virus as well as Epstein-Barr virus considered closely associated with AILD. The affected neck lymph node showed a preponderance of T-cells, predominantly CD4+ over CD8+ T-cells and all V beta gene families were expressed in the T-cells without enhancement of any particular TCR gene usage. CONCLUSION: Viral infection may occur easily in patients with AILD, possibly owing to immunodeficiency. Assessment of TCR V beta gene usage indicated T-cells to non-specifically become lymphomatous in AILD-like T-cell lymphoma.

Analysis of individual specific cytotoxic T lymphocytes for two MAGE-3-derived epitopes presented by HLA-A24.

BACKGROUND: The human MAGE-3 gene encodes tumor-specific antigens that are recognized by cytotoxic T lymphocytes (CTLs) and expressed in a high percentage of various malignant tumors. Of the five MAGE-3-derived CTL epitopes identified to date, two nonapeptides (TFPDLESEF and IMPKAGLLI, designated MAGE-3.A24a and MAGE-3.A24b, respectively) can be expressed on the tumor surface by binding to the HLA-A24 molecule, which is the most frequent HLA class I molecule in Asian populations. To compare the immunogenecities of the two peptides, individual specific CTL lines were generated for each peptide (MAGE-3.A24a and MAGE-3.A24b). METHODS: Peripheral blood mononuclear cells (PBMCs) from four HLA-A24+ healthy donors were stimulated in vitro with autologous dendritic cells pulsed with MAGE-3.A24a, MAGE-3.A24b or both and were subsequently cultivated with a cytokine combination including interleukin-2. RESULTS: We succeeded in generating peptide-specific CTL lines in two of the four donors. The two CTL lines showed similar cytolytic levels against three MAGE-3+/HLA-A24+ cancer cell lines and also target cells pulsed with the corresponding peptide. Cytolytic activities were blocked by either anti-CD8 or anti-HLA-A24 monoclonal antibodies. CONCLUSIONS: The results suggest that MAGE-3.A24a and MAGE-3.A24b peptides have equal potential in inducing MAGE-3-specific and HLA-A24-restricted CTLs.

Functional and T cell receptor gene usage analysis of cytotoxic T lymphocytes in fresh tumor-infiltrating lymphocytes from human head and neck cancer.

Twenty-one cytotoxic T lymphocyte (CTL) clones or lines that killed autologous tumor cells, but not allogeneic tumor, K562, or Daudi cells, were established from fresh tumor-infiltrating lymphocytes of two individuals (HP-1 and HP-2) with head and neck cancer by limiting dilution in the presence of recombinant interleukin-2. Sixteen (76%) of these 21 clones or lines comprised CD4+ CTLs and the other five comprised CD8+ CTLs. These observations suggest that autologous tumor cell-specific CD4+ CD8- and CD4- CD8+ CTLs are present in vivo at the tumor site in head and neck cancer. Analysis of T cell receptor (TCR) gene arrangements in 20 of the 21 CTL isolates with reverse transcriptase and the polymerase chain reaction revealed that five of 12 and five of eight isolates from HP-1 and HP-2, respectively, were clones, the other isolates being lines comprised of two or more clones. Each CTL clone showed a different combination of V alpha and V beta gene expression, suggesting that more than five different tumor-associated antigens may be expressed on head and neck cancer cells. In spite of the diversity of TCR alpha beta combinations, TCR V alpha 1, V alpha 3, V alpha 8, V alpha 10, V beta 8, V beta 9, and V beta 17 were also frequently expressed in both patients. These data suggest that specific CTLs proliferate oligoclonally and contribute to the specific immune response against head and neck cancer in vivo.

Analysis of T cell receptor variability in fresh tumor-infiltrating lymphocytes from human head and neck cancer.

In this study, we analyzed T cell receptor (TCR) gene rearrangements in tumor-infiltrating lymphocytes (TIL) freshly obtained from 15 patients with head and neck cancer using the reversely transcribed polymerase chain reaction (RT-PCR) method. These TILs showed preferential expression of V alpha 10, V alpha 8 and V alpha 1, detected in 13 (87%), 11 (73%), and 9 cases (60%), respectively. The TCRV beta gene revealed diversity without preferential usage. The head and neck region is exposed to bacteria and viruses, so it is possible that the tumor site can become infected and accumulate T cells involved in infection and inflammation. Therefore, we also investigated TCR gene usage in T cells infiltrating in chronic sinusitis mucosa to address the question of whether the V alpha 1, V alpha 8, and V alpha 10 subfamilies are characteristic in TIL from squamous cell carcinoma of head and neck. TCR V alpha 10 gene usage was also the most common in V alpha segment in T cells infiltrating the sinus mucosa, but V alpha 1 and V alpha 8 were not detected in the T cells in sinusitis. These results indicate that the V alpha 10 subfamily, the preferred T cell population in both TIL and T cells in inflammatory disease, might be involved mainly in inflammation or infection. On the other hand, V alpha 1 and V alpha 8 appear to be relatively specific populations for antitumor immunity in head and neck cancer.

The role of major histocompatibility complex expression on head and neck cancer cells in the induction of autologous cytotoxic T lymphocytes.

Using head and neck tumors, we studied the role of HLA class I and DR antigens on tumor cells in cytotoxic T lymphocyte (CTL) induction. Expression of major histocompatibility complex (MHC) antigens was investigated by two-color flow cytometry analysis and for this study we used the tumor cells, over 50% of which expressed both HLA class I and DR antigens on their surface. In seven cases, tumor cells were divided into three groups according to the specificity of monoclonal antibodies (mAb) to MHC to study the role of MHC antigens on tumor cells in CTL induction: one was not blocked (MHC double-positive tumor), a second was blocked by anti-class I mAb (class-I-negative DR-positive tumor) and third was blocked by anti-DR mAb (class-I-positive DR-negative tumor). Subsequently, these tumors were used to stimulate an autologous mixed lymphocyte/tumor cell culture for 5 days (MLTC) followed by further cultivation with interleukin-2 for 12 days. The induced autologous tumor killer cells were most cytotoxic when non-treated tumors, which consist mainly of cells that are both HLA-class I and DR-positive, were used as stimulator cells. When the tumor cells blocked by anti-DR mAb were used as stimulators, autologous tumor killer activity was lower than that induced by tumor cells blocked by anti-class-I mAb. Moreover, cytolysis by autologous tumor killer cells induced by stimulation of non-treated tumor cells was blocked during the effector phase, 26.6%-42.3% and 32.7%-53.8% by anti-class-I and anti-DR mAb respectively, suggesting that majority of the autologous tumor killer cells are MHC-restricted CD8+ or CD4+ CTL. These results suggest that both MHC class I and class II antigens on head and neck tumor cells play a critical role in inducing CTL.