Exceptional antineoplastic activity of a dendritic-cell-targeted vaccine loaded with a Listeria peptide proposed against metastatic melanoma

Vaccination with dendritic cells (DCs) is proposed to induce lasting responses against melanoma but its survival benefit in patients needs to be demonstrated. We propose a DC-targeted vaccine loaded with a Listeria peptide with exceptional anti-tumour activity to prevent metastasis of melanoma. Mice vaccinated with vaccines based on DCs loaded with listeriolysin O peptide (91–99) (LLO91–99) showed clear reduction of metastatic B16OVA melanoma size and adhesion, prevention of lung metastasis, enhanced survival, and reversion of immune tolerance. Robust innate and specific immune responses explained the efficiency of DC-LLO91–99 vaccines against B16OVA melanoma. The noTable features of this vaccine related to melanoma reduction were: expansion of immune-dominant LLO91–99-specific CD8 T cells that helped to expand melanoma-specific CD8+ T cells; high numbers of tumour-infiltrating lymphocytes with a cytotoxic phenotype; and a decrease in CD4+CD25high regulatory T cells. This vaccine might be a useful alternative treatment for advanced melanoma, alone or in combination with other therapies.


INTRODUCTION
Human melanoma is a malignant tumour of melanocytes and an aggressive skin cancer that has registered a 3% increase in annual incidence in Northern Spain [1]. Melanoma is one of the most rapidly growing cancers worldwide but yet there is no satisfactory treatment except for surgery, either in the early stages or when it has advanced to metastatic disease. Pharmacological treatment with small molecule inhibitors such as vemurafenib leads to resistance and has major cutaneous effects, but fails to induce lasting responses, which has turned the focus to immunotherapy. Classical immunotherapy agents such as interleukin (IL)-2 [2] or interferon (IFN)-α induce durable responses but the survival benefit is low. New immunomodulatory antibodies such as ipilimumab and nivolumab that block T-cell-negative regulators cytotoxic T-lymphocyte-associated antigen-4 and programmed death-1, respectively, cause some side effects, such as high toxicity and activation of autoreactive T cells [3]. In this context, dendritic cells (DCs) are pivotal cells of the immune system with high capacity to induce T-cell immunity, and are used as vaccines to increase host resistance to melanoma [4,5]. However, the lack of immunodominant melanoma antigens triggering potent cytotoxic T-cell responses, induction of immunesuppressor T cells known as regulatory T (Treg) cells in melanoma patients, and a limited beneficial effect on survival, has dampened widespread use of DC-based vaccines as melanoma immunotherapy [6,7]. www.impactjournals.com/oncotarget Activation of DCs as cancer vaccines can be performed in vitro [6,7] or in vivo as in the case of different Listeria monocytogenes strains [8][9][10][11]. The exceptional adjuvant properties of the main Listeria antigen, listeriolysin O (LLO), such as activation of DCs, stimulation of potent cytotoxic T cells, disabling of the immune unresponsiveness against tumours, and enhancement of T helper (Th)1-dominant immune responses, explains the success of attenuated Listeria as an anti-tumour vaccine [12,13]. Another adjuvant property of LLO that is useful for cancer therapy is its ability to target to murine and human melanoma cells and transform them into DCs, causing melanoma regression [10]. Finally, the immune-dominant response of LLO peptide 91-99 (LLO 91-99 ) when presented to cytotoxic T cells, affecting the immune response to other antigens, is relevant for cancer and prophylactic vaccines [14][15][16]. LLO 91-99 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), GAPDH peptide 1-22 (GAPDH  ) and GAPDH [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] were used with success in DC vaccines for listeriosis as they induce strong cytotoxic-T-cell responses and DC activation [17][18][19], which are both useful properties for cancer vaccines.
Adjuvant properties of bacterial antigens to improve cancer therapy is not a new phenomenon, and Coley observed that a patient with neck cancer recovered after infection with erysipelas, which initiated the use of bacteria and their toxins to treat end-stage cancer [20]. However, pathogenicity and toxicity are important concerns limiting the broad clinical application of bacteria and their toxins as anti-cancer agents for immunecompromised patients.
In our search for safe immunotherapy for advanced melanoma, we used experimental DC vaccines loaded with different peptides of LLO or GAPDH virulence factors of Listeria, to examine their anti-tumour potential against metastatic B16OVA melanoma. The use of peptides of LLO or GAPDH, instead of attenuated pathogens or recombinant LLO, a pore-forming cytolysin, such as other groups have proposed [12][13][14], supposes that our therapy is safe and non-toxic.

RESULTS AND DISCUSSION
The objective of this study was to identify a highly stimulatory DC vaccine with a survival benefit that prevented advanced melanoma. Since 2001, patients with advanced melanoma enrolled in phase II clinical studies have been administered autologous DC loaded with melanoma lysates or a cocktail of melanoma peptides, and showed some clinical benefit [4,5,7]. However, efforts should be made to implement these promising therapies since a robust CD8 + T-cell response that correlates with tumour regression is not always obtained and, more significantly, the number of patients with increased survival is small. According to the American Joint Committee on Cancer in 2009, advanced melanoma corresponds to any size of tumour, spreading to lymph nodes and other organs and distant metastasis [21]. Advanced melanoma is characterized also by severe immune tolerance, explained in part by low percentages of tumour peptides or poor immunogenicity of melanoma antigens.

B16OVA model of metastatic melanoma and DC vaccination
Here, we report the exceptional antineoplastic activity of an immune-dominant peptide of the human pathogen L. monocytogenes, LLO 91-99 , incorporated in a DC vaccine against experimental B16OVA melanoma. We provide evidence that this immunotherapy prevents adhesion, dissemination and metastasis of B16OVA murine melanoma, and induces robust innate and specific immune responses to Listeria and melanoma. To avoid the low immunogenicity of melanoma antigens, we used murine melanoma B16OVA, a cell line of B16-F10 melanoma cells expressing chicken ovalbumin, an antigen that induces robust CD4 + and CD8 + T-cell responses [22,23]. We inoculated B16OVA melanoma into the peritoneum of mice, because they induce carcinomatous peritonitis that allows analysis of tumour growth as well as dissemination and metastasis to the liver and lungs. Subcutaneous and intravenous models of B16OVA do not allow analysis of melanoma dissemination [10,24,25]. B16OVA growth in the peritoneum initiates as a single tumour of 12-15 mm at 7 days ( Figure 1A), disseminating to several tumours at 14 days of 30-35 mm size in half of the mice ( Figure 1A) or colonizing the whole peritoneum in the other half ( Figure 1A). These results show that at 14 days, melanoma has already disseminated in the peritoneum. B16OVA melanoma also metastasizes to the liver and lungs at 14 days and strongly infiltrates the peritoneal white adipose tissue ( Figure 1A).
Similar to advanced melanoma [3,21], we found that only 25% of mice survived at 14 days after inoculation with B16OVA cells and only 3% survived at 23 days ( Figure 1B). We noticed the deterioration of innate immune responses in the spleen by a decreased percentage of CD23 + DCs, from 42% to 6% ( Figure 1B) and reductions of CD56 + natural killer (NK) cells and CD68 + macrophages, from 22%-26% to 8.5%-10% (data not shown). The significant decrease in the percentage of CD4 + (from 33% to 9%) and CD8 + T cells (from 24% to 9%) ( Figure 1B) also suggested a decline in specific immune responses. We confirmed the specific immune tolerance by the strong reduction in Th1 cytokines, monocyte chemoattractant protein-1, tumour necrosis factor (TNF)-α and IFN-γ (Table 1), low 0.37% percentages of B16OVA extract-specific CD8 + T cells ( Figure 1C) and low (1.4%) frequencies of melanoma OVA 257-264 -specific CD8 + T cells using dimers of MHC-I molecules and peptides ( Figure 1C) [23]. Therefore, we proposed B16OVA peritoneal growth at 14 days as a model of advanced melanoma.

DC-LLO 91-99 vaccines showed exceptional antineoplastic activity
LLO is an immune-dominant antigen from Listeria widely used in cancer therapy, because it induces effector CD8 + T cells that are localized within the tumour and shows efficient adjuvant properties [26,27]. Previous observations revealed that LLO is responsible for effective therapy against melanoma, inducing immune-dominant CD8 + T-cell responses to melanoma antigens [10]. DC-LLO 91-99 anti-Listeria vaccines induce strong cytotoxic T-cell responses [15,[17][18][19], therefore, we tested DC-LLO 91-99 vaccines as melanoma immunotherapy. We also included in our study other immunodominant Listeria peptides that induced strong CD8 + T-cell responses such as GAPDH 1-15 , GAPDH 1-22 and LLO 296-304 [15,16,18,19]. We vaccinated mice in the peritoneum with a single dose of DC-LLO 91-99 , DC-LLO 296-304 , DC-GAPDH 1-15 or DC-GAPDH 1-22 formulations and 7 days later, inoculated B16OVA melanoma into the peritoneum. Mice vaccinated with DC-LLO 91-99 presented 10-fold and 30-fold lower tumour sizes at 7 and 14 days, respectively, than non-vaccinated (NV) mice ( Figure 1D). We observed that vaccination with a single dose of DC-LLO 91-99 decreased the tumour size with time, 10-30-fold, suggesting impairment in melanoma dissemination. Dissemination depends on the ability of the tumours to adhere to the peritoneal mesothelium [24,25]. Adhesion also affects the ability of tumours to migrate to distant organs and generate metastases. DC-LLO 91-99 vaccines prevented lung metastasis and adhesion ( Figure 1E). The strong antiadhesive activity of DC-LLO 91-99 vaccines correlated with 2-3-fold reductions in the surface expression of two b2-integrins, CD11b and CD11c, involved in adherence and cell migration ( Table 2, MEL-markers). In contrast, DC-GAPDH 1-22 vaccines showed three-and twofold lower melanoma sizes at 7 and 14 days, respectively ( Figure 1D) and fivefold reduction in the number of metastases ( Figure 1E). DC-GAPDH 1-22 vaccines did not show clear antiadhesive activity since adhesion was only 35% decreased ( Figure 1E) and expression of CD11b and CD11c was similar to that in NV mice (Table 2). We conclude that DC-GAPDH 1-22 vaccines showed some antimelanoma properties. DC-LLO 296-304 and DC-GAPDH 1-15 vaccines did not affect melanoma size, dissemination, metastasis or adhesion ( Figure 1D, 1E); therefore, we discarded them as melanoma therapies.
A major prognostic factor of melanoma survival is the mitotic index of melanoma cells because therapy causing melanoma apoptosis also causes tumour regression [28,29]. DC-LLO 91-99 vaccinations showed a 20-fold reduction in the mitotic indexes of recovered melanoma (Table 2), suggesting major alterations to their cell cycle because of apoptosis or necrotic death. An increase in late apoptosis corresponds to necrotic death, while increases in early apoptotic ratios indicates programmed cell death by different stimuli such as cytokines or the presence of tumour-infiltrating lymphocytes (TILs). DC-LLO 91-99 vaccines eradicated melanoma by programmed cell death and not by necrotic death, as they induced 2.6-fold increases in early apoptosis ( Figure 2A). We observed no variation in mitotic indexes or apoptosis of melanoma recovered from mice vaccinated with DC-GAPDH 1-22 vaccines (Table 2, Figure 2A).

DC-LLO 91-99 vaccines induced melanoma-specific immune responses and reverted melanoma immune tolerance
DC-LLO 91-99 vaccines showed exceptional antineoplastic activity, thus, we also investigated the immune responses that they induced. A single dose of DC-LLO 91-99 vaccine for 7 days followed by B16OVA transplantation for 14 days induced robust innate immune responses in the spleen, with high percentages of NK cells of the tumourigenic phenotype CD3 − CD49b + , DC phenotypes involved in tumour elimination CD8α + CD11c + CD83 + CD86 + iNOS + MHC-II + , and CD11b + macrophages ( Figure 2C). Consequently, DC-LLO 91-99 immunotherapy also produced high levels of IFN-γ and IL-12 Th1 cytokines ( Figure 2C) and increased the percentages of CD8 + T cells, while CD4 + T cells were not affected ( Figure 2C). In contrast, DC-GAPDH 1-22 vaccination followed by B16OVA transplantation failed to stimulate the immune responses ( Figure 2C). DC-LLO 91-99 vaccination followed by B16OVA transplantation resulted in expansion of LLO 91-99specific CD8 + T cells ( Figure 2D), verified by high 8.1% frequencies of LLO 91-99 -specific CD8 + T cells producing IFN-γ ( Figure 2D). However, DC-LLO 91-99 immune-therapy for B16OVA melanoma also induced expansion of melanoma OVA 257-264 -specific CD8 + T cells ( Figure 2D), confirmed by sixfold increased frequency of melanoma-specific CD8 + T cells producing IFN-γ ( Figure 2D). Therefore, LLO 91-99 immune-dominant CD8 + T-cell responses enhanced melanoma-specific CD8 + T cells, measured as OVA 257-264 specific CD8+ T cells since B16OVA are transfected with OVA. This immune-dominance of CD8 + T-cell responses elicited by DC-LLO 91-99 vaccination bypassed the unresponsiveness of metastatic melanoma.    5). b Cell surface markers analysed by FACS in recovered melanoma from a. Results are expressed as percentages of positive cells compared to total melanoma cells (P < 0.05). c Cell surface markers in TILs of recovered melanoma from a. Results are expressed as percentages of positive cells compared to total TILs (mean ± SD of triplicate samples, P < 0.005). d Mitotic index calculated as the ratio of the number of B16OVA in culture at time 0 h versus the number of cells at 16 h (mean ± SD) (P < 0.05).
How can DC-LLO 91-99 vaccinations break melanoma immune tolerance? The populations of TILs obtained from the recovered melanoma after DC-LLO 91-99 vaccinations provided insights into this question because we detected eightfold higher percentage of CD8 + T cells in melanoma TILs from DC-LLO 91-99 -vaccinated mice than from NV mice (Table 2). We detected that all CD4 + CD25 + T cells in TILs were also positive for the classical T reg marker FoxP3 (data not shown) and considered this phenotype as Treg. We observed fourfold higher percentage of CD4 + CD25 − T cells and 2.4-fold reductions in the percentages of CD4 + CD25 high Treg cells in melanoma TILs from DC-LLO 91-99vaccinated mice (Table 2). Other cells involved in tumour immunity also showed enhanced numbers such as CD11b + CD14 + monocytes/macrophages or CD49b + NK cells (Table 2). We conclude that the ability of LLO 91-99 peptides to suppress T reg cell function, [30] is maintained in DC-LLO 91-99 vaccines. Therefore, these immune-therapies induced the migration of lymphocytes to TILs with tumouricidal activity, such as CD8 + T cells, CD49b + NK cells or CD11b + CD14 + monocytes, while reducing the number of immune inhibitory cells such as CD4 + CD25 high cells, thus helping to revert melanoma immune unresponsiveness [31]. Vaccinations with DC-GAPDH 1-22 showed similar immune cell populations in TILs compared with NV mice (Table 2). Therefore, we discarded DC-GAPDH 1-22 vaccinations as melanoma immune-therapies.
Does DC-LLO 91-99 vaccinations provoke changes in melanoma that contribute to immune enhancement? We previously reported that Listeria therapy of melanoma transformed tumour cells into DCs [10]. When we analysed cell surface markers of melanoma recovered from DC-LLO 91-99 -vaccinated mice, we detected increased percentages of several DC markers such as CD14 + , Toll-like receptor (TLR)2 accessory molecule that binds to LLO [32], and activation markers such as double-positive CD83 + CD86 + cells. We also observed that the percentages of H-2K b MHC class I molecules increased, while the percentages of IA b MHC class II molecules were reduced by fivefold (Table 2). Therefore, DC-LLO 91-99 vaccinations mimicked most of the effects observed with Listeria on the transformation of melanoma cells into DCs. This transformation into DCs might also increase CD8 + and decrease CD4 + T-cell responses. However, it is also possible that DC-LLO 91-99 vaccination before B16OVA transplantation induces reprogramming of DC antigen presentation, increasing melanoma-specific as well as LLO-specific CD8 + T-cell interactions and decreasing DC interactions with Treg cells (CD4 + CD25 + FoxP3 + ), which is responsible for immune tolerance to melanoma antigens. DC presentation of melanoma antigens in NV mice and B16OVA transplantation triggered a high proportion of Treg cells (CD4 + CD25 + FoxP3 + ) that explains the low induction of melanoma-specific CD8 + T cells. In this way, DC-LLO 91-99 vaccinations increased the positive signals between DC and T cells, promoting stronger antimelanoma immune responses, and controlling tumour growth and dissemination.
In summary, we propose the following molecular model of action of DC-LLO 91-99 vaccines against advanced melanoma ( Figure 2E). In NV and B16 transplanted mice, there is no increase in the percentages of splenic CD83 + CD86 + activated DCs or any change in cell surface markers of melanoma. This scenario seems to expand TILs in melanoma with the CD4 + CD25 high immunosuppressor phenotype that blocks the immune response of melanoma-specific CD8 + T cells and abrogates any Th1 cytokine response, allowing the tumour to grow exponentially ( Figure 2E). These observations indicate that DC immune responses in NV and melanoma-transplanted mice deliver negative signals to T cells that explain the low percentages of melanomaspecific CD4 + and CD8 + T cells. DC-LLO 91-99 vaccination increases expression of co-signalling molecules such as CD83 or CD86 and other co-stimulatory molecules involved in TLR signalling such as CD14 in spleenactivated DCs and melanoma, which re-programmes DCs and melanoma antigen presentation. In this regard, melanoma and splenic DCs present higher expression of MHC class I molecules and significantly reduced expression of MHC class II molecules, driving a Th1 response with high levels of IFN-γ and IL-12. This re-programming causes an increase in the percentages of LLO-and melanoma-specific CD8 + T cells, and a decrease in the percentage of melanoma-specific CD4 + T cells ( Figure 2D, Table 2). Regression of melanoma is also helped by the decreased percentage of negative Treg (CD4 + CD25 + FoxP3 + ) cells (Table 2) [33]. These two important changes in the immune response caused by this immune-therapy appeared to induce melanoma apoptosis and promote survival of the mice.
We foresee several scenarios for the future use of DC-LLO 91-99 vaccinations for advanced melanoma. First, it might function to pre-condition the vaccine site with a potent recall antigen such as LLO 91-99 , similar to tetanus toxoid DC vaccines against glioblastoma [34]. Second, DC-LLO 91-99 vaccination can be used to increase the positive signals between DC and T cells, promoting stronger anti-melanoma immune responses alone or in combination with other immunotherapies [35,36]. We suggest that vaccines against pathogens such as Listeria, based DC-LLO 91-99 vaccines or the recently reported tetanus vaccine [34], might help to redesign anti-tumour therapies and prepare the immune system.

Preparation of DCm-peptide vaccines
B16OVA transplantation into DCm-peptide vaccinated and NV mice C57BL/6 female mice (n = 5) received a single injection of DCm-peptide vaccines (10 6 cells) in the peritoneal cavity or saline (NV mice) for 7 days. Mice were B16F10 transplanted with a single injection of 5 × 10 5 B16OVA cells in the peritoneal cavity. On day 14 after B16OVA transplantation and before mice were killed, we collected and processed serum in the first 50 min and stored it at −80°C until cytokine analysis by FACS. On the same day, we collected spleens, transplanted melanoma, livers and lungs of DCm-peptide vaccinated and NV mice to photograph organs and metastases. Spleens and recovered melanoma were also processed for FACS analysis in the following 2 h after their isolation. Melanoma size was measured with a calliper and expressed in mm 2 by multiplication of diametrically perpendicular measurements. Results are expressed as the mean ± SD.

Adhesion of B16OVA melanoma to plates
Recovered melanoma after DCm-peptide vaccination or not (NV mice) and transplantation into mice was minced, disaggregated and passed through a 70-µm strainer to obtain a single cell suspension. Cells were counted and seeded into six-well plates at 5 × 10 6 cells/ml. After 2 h, culture medium was removed and replaced with fresh medium and cells were allowed to grow for 16 h. Cells were detached and viable cells quantified after staining with Trypan blue. B16OVA melanoma not transplanted into mice was also seed into six-well plates at the same concentration and served as a control with 100% adherence. Results were expressed as percentages of viable cells compared to total cells seeded. B16OVA recovered from NV mice had the same percentage of adherent cells as non-transplanted B16OVA melanoma.

FACS analysis of spleens, melanoma, intracellular IFN-γ staining and cytokine measurements
Cell surface markers of bone-marrow-derived DCs, spleens or recovered melanoma from mice vaccinated with DCm-peptides or NV mice, transplanted with B16OVA, were analysed by FACS using the following antibodies: anti-CD4-PE, anti-CD8α-PE, anti-CD49b-PE, anti-F4/80-PE, anti-CD11b-APC, anti-CD11c-PE, anti-MHC-II-APC, anti-CD40-PE, anti-CD83-APC and anti-CD86-V450 (Miltenyi Biotech). Recovered melanoma cells were also analysed for apoptosis in the following hour after their isolation by FACS using two reported products for apoptosis, Annexin-V conjugated with APC fluorochrome and 7-AAD (7-amino-actinomycin D) (BD-Biosciences, San Jose, CA, USA). Mice sera or supernatants of DCm-peptide vaccines were used to quantify cytokines using the CBA kit (Becton Dickinson, Palo Alto, CA, USA). Samples were analysed in triplicate and results were expressed as the mean ± SD of two separate experiments. For measuring of intracellular IFN-γ, spleen cells were cultured in 96-well plates (5 × 10 6 cells/ml) and stimulated with B16OVA extract (50 µg/ml) (B16OVA ext ), LLO 91-99 or OVA 257-264 peptides (50 µM) for 5 h in the presence of brefeldin A [38]. Next, cells were surface labelled for CD4 or CD8, fixed and permeabilised with cytofix/cytoperm kit to measure intracellular IFN-γ (BD Biosciences). After sample acquisition, data were gated for CD4 + or CD8 + events, and the percentages of these cells expressing IFN-γ were determined. Results were corrected according to the percentages of total CD4 + or CD8 + positive cells. Data were analysed using FlowJo software (Treestar, Ashland, OR, USA).

Isolation of TILs from transplanted melanoma
B16OVA melanoma was recovered from mice vaccinated with DCm-peptide or not (NV mice) with a single dose of vaccine for 7 days, followed by B16OVA transplantation for 14 days. At the end of melanoma transplantation, we processed melanoma in the first hour after isolation. Melanoma processing included mincing, disaggregating, passing through a 70-µm strainer and isolating TILs by centrifugation over a Ficoll gradient at a 1.077 g/ml density (Histopaque-1077; Sigma-Aldrich, St Louis, MO, USA). We recovered TILs in the interphase gradient, while collecting melanoma cells in pellets. In control samples, we performed an enzymatic digestion with 10% fetal calf serum and collagenase IV (200 U/ml) before Ficoll gradient centrifugation, compared with samples without enzymatic digestion. We observed no differences in the number of TILs recovered in samples digested or not digested with collagenase. Cells were stained with CD4-FITC, CD8-PE, CD25-V459, CD14-APC, CD11b-FITC or CD49b-PE antibodies and analysed by FACS.

Statistical analysis
For statistical analysis, the Student's t test was applied. P ≤ 0.05 was considered significant using GraphPad for graphic presentation.