Eliciting cytotoxic T lymphocytes against human laryngeal cancer-derived antigens: evaluation of dendritic cells pulsed with a heat-treated tumor lysate and other antigen-loading strategies for dendritic-cell-based vaccination

Each year, more than 500,000 cases of head and neck squamous cell carcinoma (HNSCC) are diagnosed worldwide [1]. Laryngeal squamous cell carcinoma (LSCC) accounts for a large proportion of HNSCCs [2]. Despite the rapid development of diagnosis and treatment of LSCC, including surgery, chemotherapy, and radiotherapy, the 5-year overall survival rate of LSCC remains relatively low [3]. The high rate of local recurrence and considerable mortality that accompany this malignancy mandate the development of novel therapeutic alternatives.

Immunotherapy is regarded as a promising strategy that complements current cancer therapies [4]. Dendritic cell (DC)-based vaccination trials have been performed for some advanced cancers, such as melanoma and colon, renal, and prostate cancers [5‐7]. They have proven to be feasible and safe, and to elicit immunological responses [8]. Current vaccination strategies include antigen (Ag)-defined vaccines (peptide, DNA, and recombinant tumor proteins) and polyvalent undefined vaccines (tumor lysate, heat shock proteins (HSPs), whole tumor cells, and whole tumor RNA) [5,6]. The former type of vaccine is prepared using a single tumor-associated antigen (TAA). However, its application using specific peptides is limited to patients expressing tumor-associated MHC class I restricted peptides that fail to elicit MHC class II and CD40L-mediated CD4+ T cell helper responses. In contrast, the latter type of vaccine, generated using whole tumor cells or their derivatives, are advantageous in that they provide a full complement of TAAs, including both MHC class I and class II-restricted epitopes, thus reducing the risk of immune escape by antigen loss variants.

Whole tumor cell Ag-loading strategies for DC-based vaccination have commonly used freeze-thawed tumor cell lysates as an immunogenic source of TAAs [9, 10]. However, DCs loaded with such lysates have provided only modest protection in various animal models [9, 11, 12] and stimulated limited responses in a range of clinical trials [10, 13, 14]. Some studies have even shown that freeze-thawed lysates suppress DC maturation and functions in vitro, and are an ineffective loading strategy for DC therapy in vivo [15, 16].

To enhance antitumor immunity, studies have attempted to optimize antigen loading. Compared with ordinary lysates obtained by subjecting tumor cells to several freeze (liquid nitrogen) and thaw (37 °C water bath) cycles, stressed lysates prepared from heat-treated (≥42 °C) tumor cells are a more effective source of TAAs for pulsing DCs. Heat-treated tumor lysate-pulsed DCs generate a stronger and broader T cell response against pancreatic cancer [17]. Irradiated tumor-primed DCs provide more potent protection against solid tumors, such as human melanoma [15]. Even retinoblastoma cell supernatants exert an immunostimulatory effect on DCs [18]. Because different tumor strains have various immunogenicities, it remains uncertain which Ag-loading strategy has superior efficacy for treating human laryngeal carcinoma.

In this study, we performed a detailed examination of DC phenotypes loaded with TAAs, including a tumor supernatant (TCS), tumor-stressed lysate (TSL), and irradiated tumor cells (ITC). Subsequently, we assessed TAA-specific T cell proliferation and Th1 cytokine production induced by these Ag-loaded DCs. Lastly, we compared the in vitro cytotoxic T lymphocyte (CTL) responses. Our study identified the human LSCC cell line SNU899-derived TSL as an efficacious source of TAAs for pulsing DCs to induce a more potent CTL response against laryngeal cancer.

DC vaccines take advantage of the potent antigen-presenting capacity of DCs to stimulate primary and secondary T and B cell immune responses. Clinical trials have been performed for several advanced cancers such as melanoma and renal, prostate, and colon cancers, which have shown encouraging results [5‐7]. DC vaccines targeting mutated proteins in the p53 pathway [22] and human papilloma virus peptides have been produced for the treatment of HNSCC [22, 23]. These TAA-loaded DCs stimulate immunological and clinical responses in certain patients. However, a significant proportion of patients remain unresponsive to immunomodulatory therapy.

Unlike specific TAAs, whole tumor cells or their derivatives may better apply to various tumors for which few or no defined tumor-specific antigens are available, such as for laryngeal cancer. Furthermore, whole tumor cell immunizations result in polyvalent stimulation of both CD4+ Th cells and CD8+ CTLs against a broad range of Ags. Therefore, DCs loaded with Ags in this manner may be used to treat laryngeal cancer.

Among the current methods for pulsing DCs with whole tumor-derived materials, tumor cell lysates, which are prepared by repeated cycles of freezing and thawing with solid debris removed by centrifugation, have been widely applied in early stage clinical trials. However, an increasing number of trials have shown that the immune response elicited by DC-TSL does not generate a sufficient antitumor potency [10, 13, 14]. Recently, several studies have demonstrated that freeze-thawed lysates suppress DC maturation and function in vitro, and are ineffective for loading DCs for therapies in vivo [15, 16]. Our preliminary study also suggests that DCs cultured in the presence of human laryngeal cancer-derived lysates decreased surface expression of B7–1, B7–2, and MHC molecules, and inhibit T cell proliferation (data not shown). Potential explanations for these findings include the use of different tumor stains, alternate vaccination schedules, and the use of DCs at various stages of maturity [10].

To elicit an effective immune response, we used three strategies for loading DCs with tumor-derived Ags and compared their ability to stimulate CTL activity. The loading strategies included using DC-TCS, DC-TSL, and DC-ITC. Our results show that human laryngeal cancer-derived TSL was an effective TAA source for pulsing DCs. DC-TSL not only induced the most potent CTL activity, but also induced the strongest MHC class I TAA-specific T response (proliferation of TAA-specific CD8⁺ T cells), the highest expansion of TAA-specific T cells (based on the number of TAA-specific T cells), and the strongest Th1 cytokine response (elevated levels of IFN-γ and IL-2 production). Interestingly, DC-TSL induced both class I and class II TAA-specific responses, suggesting that DC-TSL are capable of both class I and II cross-presentation. These results emphasize that the choice of Ag-loading strategy is critical to the strength of the immune response.

We identified alterations in the DC phenotype after treatment with tumor-derived Ags. DC-TSL expressed the highest level of HLA-DR and CD86, whereas DC-ITC also had elevated HLA-DR expression but expressed the lowest levels of co-stimulatory molecules (CD80, CD86, and CD40). HLA-DR expression on the surface of DC-TCS remained at the same level as that in DC controls. A significant body of evidence has shown that stimulating a stress response in tumor cells increases the production of HSPs, which may expand the repertoire of TAAs and enhance TAA delivery to professional antigen-presenting cells (APCs) [17, 24, 25]. Additionally, stress-induced HSPs stimulate DCs and induce APC cytokine and chemokine secretion [26‐29]. Our findings support the idea that the stressed lysis strategy has advantages by providing a larger TAA repertoire. Moreover, elevation of co-stimulatory molecules such as CD86 may enhance DC capacity to prime T cell responses. In contrast, cell culture supernatants may not contain sufficient or adequate Ags secreted by laryngeal cancer cells. Although one study reported that DC antigen presenting function can be improved by supernatants from retinoblastoma cells [18], it is not surprising that laryngeal cancer has a distinct tumor milieu to interact with DCs. LSCC cells may evade the host immune system through manipulation of their own immunogenicity, production of immunosuppressive mediators, and promotion of immunomodulatory cell types [30‐32]. Consistent with some studies [33], there was a reduction of CD80, CD86, and CD40 in DC-ITC. Various mechanisms through which intact tumor cells suppress DC functions have been described, including cytokines (e.g., TGF-β, IL-10, and vascular endothelial growth factor), ceramide [34], and other tumor-derived lipids [35]. However, a complete understanding of the mechanisms by which DC suppression functions will require additional study.

Studies have shown that CD8⁺ and Th1 cells play an important role in controlling tumor growth. For example, CD8⁺ T cells mediate antitumor immunity [36]. Th1 cells, a subset of CD4⁺ T cells, constitutively express IFN-γ and TNF-α, and play a role in priming tumor-specific CTLs through the release of soluble IL-2 in the proximity of CTLs [37]. In addition, induction of MHC class I tumor-specific immunity requires epitope linkage between Th1 and CTL epitopes, which is important for CTL induction [38]. In the current study, DC-TSL pulsing was the only effective method to prime CD8⁺ and CD4⁺ T cells. This result indicates that a TSL is not only processed for MHC class I presentation, but also cross-presented by the MHC class II pathway, consistent with the elevation of class II molecules (HLA-DR) on DC-TSL. Further experimentation showed that both the percentage and number of Th1 cells (including IFN-γ- and IL-2-secreting cells) were increased in T cell and DC-TSL co-culture, which strongly supports the choice of TSL-pulsed as a promising CTL activator against laryngeal cancer. It is worth noting that the stronger and broader T cell response induced by DC-TSL may benefit from production of HSPs, such as HSP70 and HSP90 [39, 40], whose antitumor activity is exerted through various mechanisms. The inefficiency of DC-ITC to induce Ag-specific T cell responses can be explained by the inefficient cross-presentation of TAAs by DCs that express low levels of co-stimulatory molecules. Although apoptotic tumor cells induced by irradiation effectively prime APCs in vitro, they likely cannot stimulate DCs to generate an antitumor immune response in the absence of additional maturation [41].

The incompetence of DC-TCS, which express low levels of MHC molecules (e.g., HLA-DR) and maturation markers (e.g., CD83), may result from inhibition by the tumor milieu and fewer laryngeal cancer cell-secreted Ags. Although CTL activity of DC-TCS was lower than that of DC-TSL, which induced the highest as expected, there was no significant difference between them. The increased CTL activity of DC-TCS-stimulated PBLs may be due to more active natural killer and natural killer T cells, which comprise the majority of non-CD8 T lymphocyte effector cells [42]. Additionally, to ensure that the Ag-MHC complex was recognized by CTLs, we used Ag-pulsed DCs as the source of MHC-matched target cells, which may not accurately reflect tumor cell susceptibility to CTL lysis [43].

We demonstrate that a heat-treated TSL is an effective source of TAAs for pulsing DCs to treat human LSCC. DC-TSL induced the greatest expansion of TAA-specific T cells, the strongest Th1 cytokine response, and the most potent CTL activity, whereas DC-TCS or DC-ITC inhibited T cell activation and induced only a certain extent of CTL activity. Although the efficacy of this loading strategy needs to be tested further, such as in a setting whereby the CTL response could be tested using autologous laryngeal cancer cells, we have provided an encouraging alternative strategy for DC-based immunotherapy for laryngeal cancer.