Intralesional rose bengal in melanoma elicits tumor immunity via activation of dendritic cells by the release of high mobility group box 1

Intralesional (IL) therapy is under investigation to treat dermal and subcutaneous metastatic cancer. Rose bengal (RB) is a staining agent that was originally used by ophthalmologists and in liver function studies. IL injection of RB has been shown to induce regression of injected and uninjected tumors in murine models and clinical trials. In this study, we have shown a mechanism of tumor-specific immune response induced by IL RB. In melanoma-bearing mice, IL RB induced regression of injected tumor and inhibited the growth of bystander lesions mediated by CD8+ T cells. IL RB resulted in necrosis of tumor cells and the release of High Mobility Group Box 1 (HMGB1), with increased dendritic cell (DC) infiltration into draining lymph nodes and the activation of tumor-specific T cells. Treatment of DC with tumor supernatants increased the ability of DCs to stimulate T cell proliferation, and blockade of HMGB1 in the supernatants suppressed DC activity. Additionally, increased HMGB1 levels were measured in the sera of melanoma patients treated with IL RB. These results support the role of IL RB to activate dendritic cells at the site of tumor necrosis for the induction of a systemic anti-tumor immune response.

Rose bengal (RB) was originally used as a staining agent by ophthalmologists and in liver function studies [7,8]. RB has a direct cytotoxic effect on microorganisms and cancer cells [9][10][11][12][13][14][15]. In tumor cells, RB selectively passes through the cell membrane and accumulates in the lysosomes resulting in autolysis [10]. IL injection of RB (also referred to as PV10, an investigational drug formulation of 10% RB in normal saline) into tumors has been shown to elicit regression of both treated and untreated tumors [12,16]. In murine models, IL PV10 induced T cell mediated tumor-specific immune responses in MT901 breast cancer and in B16 melanoma [16]. In a phase I study in metastatic melanoma patients, IL PV10 was well tolerated and led to a 48% objective response (OR) in treated lesions and a 27% OR in untreated lesions Research Paper www.impactjournals.com/oncotarget [12]. A recent phase II clinical trial of PV10 demonstrated a similar OR in treated and untreated lesions [17]. Additional evidence of systemic response was supported by the regression of untreated visceral lesions in a few patients. However, the underlying mechanism of induction of tumor immunity remains unknown.
In this study, we investigated whether IL PV 10 leads to the release of DAMPs contributing to the induction of an antitumor immune response.

IL PV-10 elicits a tumor-specific immune response
To investigate the underlying mechanism of the tumor-specific immune response elicited by PV-10, C57BL/6 mice were injected subcutaneously (s.c.) with M05 tumor cells expressing the ovalbumin (OVA) protein [36]. Similar to our previous findings in the B16 model, IL injection of PV10 directly inhibited tumor growth ( Figure 1A) [16]. Moreover, IL PV10 led to increased OVA-specific CD8 + T cells in the draining lymph nodes (DLNs) of PV10treated mice, compared to the PBS treated group ( Figure 1B). To determine whether IL PV10 induced T cells with memory characteristics, splenocytes from mice treated twice with IL PV10 were cultured in vitro in the presence of OVA peptide and media supplemented with the cytokines IL15 and IL21, which are required for maintaining CD8 + T memory cells [37]. T cells from PV10treated mice demonstrated a ca. 2 fold increase in secretion of IFN-γ in response to M05 cells, compared to T cells isolated from PBS-treated mice ( Figure 1C). To further confirm the induction of memory T cells, spleens, lymph nodes (LNs), and tumors were collected from mice 10 days after IL PBS or PV10 injection. Memory T cells (CD44 hi CD62L hi and CD44 hi CD62 low ) were increased in the LNs and spleens of mice treated with PV10 compared to mice treated with PBS ( Figure 1D). In contrast, there were decreased T memory cells in bystander tumors of treated mice. These results suggest that IL PV10 can induce tumor-specific T cells with memory characteristics in M05 melanomabearing mice.
We next investigated whether CD8 + T cells mediate the tumor-specific immune response elicited by IL injection of PV10. Using a lung metastasis model described previously, mice received B16F10 cells s.c. on day 0 to establish a solitary tumor on the flank and B16F10luc cells intravenously (i.v.) to establish multiple lung lesions [16]. On day 7, the palpable flank tumors were treated IL with 50 μl PBS or PV-10. Tumors in the lungs were monitored by bioluminescent imaging on day 21 ( Figure 1E). All mice that received IL PBS displayed growth of tumor in the flank and multiple lung lesions. In contrast, mice that received IL PV10 developed fewer lung lesions ( Figure 1F, p<0.05 compared to PBStreated mice). To examine the role of CD8 + T cells, mice received a CD8 depleting antibody (clone 2.43) or the isotype antibody (Rat IgG). After depletion of CD8 + T cells, mice receiving IL PV10 developed equivalent numbers of lung lesions compared to mice receiving PBS treatment ( Figure 1F). These results show that CD8 + T cells are crucial for the tumor-specific immune response induced by IL injection of PV10.
To monitor CD8 + T cell responses in vivo, PV10 or PBS was IL injected into s.c. M05 tumors on day 13, followed by injection of violetdye labeled OT1 cells. The average tumor size was 50 mm 2 on day 13 yielding sufficient tumor material to dissect for examination of T cell proliferation after IL PV10. OT1 T cells are CD8 + T cells that specifically recognize the SIINFEKL peptide derived from the OVA protein. Adoptive transfer of OT1 cells alone or treatment with IL PV-10 alone was not sufficient to prevent tumor growth when treatment began at day 13. The combination of IL injection of PV10 and transfer of OT-1 cells significantly reduced tumor progression and increased survival (Figure 2A-2B). We next measured the proliferation of OT1 T cells in tumors, LNs, and spleens after IL PV10 injection. In the spleens of PV10 treated mice, on day 4 after transfer, more than 60% of OT1 T cells demonstrated at least one division, hereafter called "divided" T cells ( Figure 2C, left panel). In contrast, there were fewer divided T cells in the spleens of PBStreated mice ( Figure 2C, right panel). IL PV10treated mice exhibited increased proliferation of OT1 T cells in the spleen compared to mice treated with IL PBS ( Figure 2D). However, when PV10 was injected s.c. into the opposite flank where there was no tumor, proliferation of OT-1 T cells was equivalent to proliferation of OT1 T cells in mice The percentage of CD8 + , OVA tetramer + T cells was measured in the DLNs after 8 days by flow cytometry. Data are representative from two independent experiments and are shown the mean number ± SEM. C. Mice were rechallenged with 3e5 M05 cells s.c. on the opposite flank on day 7 and 50 μl PV-10 or PBS were injected IL into the initial tumor lesion on days 7 and 17 (n=4). On day 23, splenocytes were expanded with 20 ng/ml IL-15 and IL-21 and 1 μg/ml SIINFEKL for 7 days and then co-cultured with M05 cells. IFN-γ production was measured after 48 hours. Data are presented as mean ± SEM from three independent experiments. D. Mice were inoculated with 3e5 M05 cells on both flanks (n=4 mice/group). On day 7, PV-10 or PBS were injected IL into the left flanks, and 10 days later, CD8 + T memory cells (CD44 hi CD62L hi and CD44 hi CD62L lo cells) were measured in LNs, spleens and right flank tumor. Data are the representative from two independent experiments and presented as mean ± SEM. E-F. Mice were injected with 1e5 B16 cells s.c. and 4e5 luciferasetagged B16F10luc cells i.v. (n=6-9 mice / group). On day 7, 50 μl PV-10 or PBS were injected IL. Mice were treated with 100 μg of purified CD8 depleting antibody (2.43) or control rat IgG antibody 2 and 3 days prior to PV10 treatment and twice per week until day 21. E. Representative Xenogen imaging results are shown. F. Data are the representative from two independent experiments and are shown the mean number ± SEM of luciferase intensity in the lung region. * p< 0.05, statistically significant versus control; ** p<0.01. P values were determined by an unpaired student ttest. www.impactjournals.com/oncotarget that received PBS in either the opposite flank or IL into tumor. At the tumor site, OT1 cells robustly proliferated in PV10 or PBStreated mice ( Figure 2E). This may be due to the relatively high OVA protein expression by the M05 tumor. A robust OT1 T cell proliferation was also observed in the tumor draining LNs and no difference was seen between PBS and PV10treated mice ( Figure 2F). Differences in OT1 T cell proliferation were measured in the nondraining LNs of mice treated with IL PV10 compared to mice treated with IL PBS or treated on the opposite flank ( Figure 2G). These data suggest that IL PV-10 can enhance T cell proliferation.

Figure 2: IL injection of PV-10 facilitates the proliferation of tumor-specific CD8+ T cells in M05-bearing mice. A-G.
M05 cells (3e5) were injected into one flank of C57BL/6 mice on day 0. On day 13, 50 μl PV-10 or PBS were injected IL or s.c. in the opposite flank without tumor. Four hours later, 2e6 violet labeled CD45.1 + OT1 T cells were i.v. injected. Tumor growth (A) and survival (B) of mice were monitored (n=5 mice / group). After 4 days, cells from spleens (C, D), tumors (E), draining LNs (F) and non-draining LNs (G) were stained with CD45.1 and CD45.2 antibodies. Representative histograms of violet dye dilution show the progenies of CD45.1 + T cells, which have at least one division, after discrimination of dead cells (C). Data are presented as mean ± SEM from three independent experiments (n=5 mice / group). H. 3e5 M05 cells were injected s.c. into both flanks of C57BL/6 mice on day 0 (n=7). On day 7, 50 μl of PV-10 or PBS were injected IL in the left flank and OT-1 cells were transferred into mice. After an additional 4 days, CD3 + cells were measured in the bystander (right flank) tumor. Data are presented as mean ± SEM from two independent experiments. P values were determined by an unpaired student ttest (A, DH) or a logrank test (B). *p< 0.05 versus control; **p<0.01. www.impactjournals.com/oncotarget Next, we examined whether IL PV10 increased the infiltration of T cells into tumors. As IL PV-10 can ablate injected tumor, a bilateral tumor model was used to monitor the T cell infiltration. Bilateral M05 tumors were established in C57BL/6 mice. After 7 days, mice received IL PV10 in the left flank tumor, and four hours later were i.v. injected with OT1 cells. Four days later, bystander tumors were collected and T cell infiltration was measured. As shown in Figure 2H, there was a significant increase in T cells infiltrating the bystander tumor. Together, these data support that IL injection of PV-10 can boost T cell infiltration in tumors.

IL PV-10 leads to DC activation
Because DCs are capable of priming T cells, we next examined DCs in the spleens and LNs of mice treated with IL PV10. At 24 hours after IL PV10 injection, the number of infiltrating DCs increased in both draining LNs (DLNs) and nondraining LNs (NDLNs) ( Figure 3A). The overall total number of cells in the LNs was not significantly changed after treatment. After 72 hours, the number of infiltrating DCs in DLN of PV-10-treated mice decreased to the baseline level and was equivalent to the level in the PBStreated mice (data not shown), suggesting that the infiltration of DCs is transient. To examine whether DCs migrated from the site of tumor, OVA protein labeled with FITC (FITCOVA) was directly injected into tumor 4 hours after IL injection of PV10 or PBS. After 18 hours, lymph nodes were collected. We measured increased FITC + DCs in the DLNs but not in the NDLNs of the PV10treated mice ( Figure 3B). To assess DC activation, DLNs were collected from M05 bearing mice treated with IL PV10 or PBS. DCs were enriched and stained with costimulatory markers. DCs isolated from the DLNs of IL PV10 treated mice expressed higher levels of activation markers, including CD40, CD86 and CD80, compared to PBStreated mice ( Figure 3C-3D). Together, these studies support a role for IL PV10 to induce DCs to take up antigens at the tumor site, infiltrate into the DLN, and become functionally mature.
We next tested whether IL injection of PV10 led to DC activation. Bone marrow cell (BM)derived DCs were cocultured with supernatants of M05 tumors that were previously treated with IL PV10 or PBS. DCs were then pulsed with OVA protein and cocultured with OT1 T cells. DCs cultured with supernatants derived from M05 tumors isolated from PV10treated mice induced increased proliferation of OT1 T cells, compared to DCs cultured with supernatants derived from M05 tumors from PBS treated mice ( Figure 3E). These results suggest that PV10 treated tumors may release factors that activate DCs.

PV-10 treatment increases DC activation via HMGB1
To examine whether tumor death induced by IL PV10 is linked to the activation of DCs, we first investigated how PV10 induces cell death. As shown in Figure 4A, PV10 resulted in a dosedependent cytotoxicity in B16 melanoma cells, with an IC50 value of 60 μM after 48 hours of treatment. There was less cytotoxicity in mouse NIH3T3 fibroblasts, with an IC50 value of 110 μM after 48 hours of treatment with PV10 ( Figure 4A). The IC50 value of PV10 on B16 and 3T3 cells was similar at 6, 12, and 24 hours (data not shown). There was a significant increase in necrosis (DAPI + ) of B16 cells, human 888 melanoma cells and human primary (P13) melanoma cells after 48 hours of treatment with 50 μM of PV-10 ( Figure 4B). However, after 48 hours, a relatively small proportion of cells were in early apoptosis (Annexin V + DAPI , Figure 4C), with no difference after 6, 12, and 24 hours of treatment (data not shown). This indicates that treatment with 50 μM of PV-10 leads to tumor cell death through necrosis rather than apoptosis. Little necrosis or apoptosis was measured in 3T3 fibroblasts or human embryonic kidney 293T cells in the presence of the same dose of PV10 ( Figure 4B and 4C). These studies suggest that PV10 can kill tumor cells at a dose that is not toxic to nontumor cells.
We next examined the factors that are released by tumor cells after treatment with PV10. It has been shown that necrosis is associated with the disruption of the integrity of the cell membrane and the uncontrolled release of cytosolic contents into extracellular space, including DAMPs such as HMGB1, IL1a, and HSP proteins. Murine B16 melanoma, human 888 melanoma cells and human primary melanoma cells were treated with 0, 100 or 200 μM PV-10 for 48 hours. The release of DAMPs including HMGB1, HSP70, HSP90 and IL 1a were measured by ELISA or western blot. HSP70 and IL1a were not detected and HSP90 was unchanged after treatment with PV10 (data not shown). HMGB1 was measured in the supernatants of B16, primary melanomas, and 888 cells in a dosedependent manner ( Figure 5A).
To determine if secreted HMGB1 contributed to DC activation, BMderived DCs were incubated for 2 days with 20% supernatant from B16 cells treated with PV10 in the presence of HMGB1 neutralizing antibody or isotype control antibody. Tumor supernatant from PV10treated cells led to DC maturation, with the up regulation of surface CD40. Neutralization of HMGB1 in supernatants significantly decreased CD40 expression ( Figure 5B). Treatment of DCs with PV10 directly did not change CD40 expression, suggesting that PV10 itself does not lead to DC maturation. Other costimulatory markers on DCs, including CD86 and CD80, were unchanged (data not shown).
To compare the antigen presentation capacity of DCs, BMderived DCs were incubated with the supernatant from M05 tumor treated with IL PV10 in the presence of HMGB1 neutralizing antibody or isotype antibody for 2 days. The pretreated DCs were pulsed with OVA protein and cocultured with OT1 T cells. T cell www.impactjournals.com/oncotarget proliferation was measured. The blockade of HMGB1 decreased the ability of DCs to stimulate OT1 T cell proliferation ( Figure 5C). This suggests that treatment of melanoma cells with PV10 leads to the release of HMGB1, and the activation of DCs.

IL PV-10 leads to HMGB1 increase in the sera of melanoma patients
To determine whether HMGB1 release is relevant in melanoma patients treated with IL PV10, we compared the level of HMGB1 in the serum of patients obtained before and after treatment with IL PV10 (see "Materials and Methods" and Supplementary Figure S1). IL PV10 led to tumor regression in both an injected and an uninjected bystander lesion, shown by IHC staining for the melanoma antigen MelanA/MART1 (melA) in the biopsy specimens (Supplementary Figure S1BS1C). Notably, the concentration of HMGB1 was significantly increased in serum collected 714 days after treatment with IL PV10 (Figure 6). Therefore, HMGB1 may contribute to the bystander effect induced by IL PV10 in patients with metastatic melanoma.
The percentage of infiltrating immune cells in PV-10 treated and bystander lesions were compared before and after treatment with IL PV10. However, very few infiltrates were detected in the lesions that completely regressed, and no significant changes were measured. Thus   an alternative method was used to compare the presence of immune subsets in peripheral blood mononuclear cells (PBMCs) before and after treatment. There was a statistically significant increase in circulating CD8 + T cells, CD4 + T cells, and NKT cells after PV-10 treatment (Supplementary Figure S2). There was no difference in circulating NK cells, MDSC, CD4 + FOXP3 + regulatory T cells or plasmacytoid DCs before and after treatment (data not shown).
To determine whether we could measure tumor specific T cell responses after PV-10 treatment, CD8 + T cells were purified from PBMC collected from 7 patients before and after treatment. T cells were cocultured with autologous or HLAmatched melanoma cell lines in vitro for 24 hours and supernatants were collected. A significant increase in IFN-γ production was measured in the CD8 + T cells isolated after treatment with IL PV10 in 5 patients out of 7 patients that were tested. No change was measured when CD8 + T cells were cocultured with HLAmismatched cell lines (Supplementary Figure S3). These preliminary results support the role of IL PV10 treatment to induce a systemic antitumor immune response in patients with metastatic melanoma.

DISCUSSION
Melanoma incidence rates have increased rapidly in the United States over the past 30 years and is the fifth most common cancer in men and the seventh most common cancer in women [38]. IL therapy is a promising treatment modality for patients with dermal and/or subcutaneous metastatic melanoma. Importantly, it may induce not only local tumor regression but also a systemic antitumor immune response. In a recent clinical trial in metastatic melanoma patients, IL PV10 led to a 50% objective response rate with mild to moderate side effects [17]. In treated patients, 8% had no evidence of disease after 52 weeks and 26% experienced complete regression in bystander lesions. However, the mechanism by which IL PV10 leads to systemic antitumor immunity is unknown.
In this study, we showed that IL PV10 led to the necrosis of melanoma cells and the release of HMGB1. These data are consistent with the observation that HMGB1 was passively released from photosensitized HeLa cells treated with a rose bengal analog [39]. Pretreatment with rose bengal acetate led to apoptosis and autophagy and the secretion of HSP70, HSP90 and HMGB1. In contrast, our results showed that PV10 treatment induced necrosis in melanoma cells and the secretion of HMGB1, but not HSP70, while the amount of HSP90 was unchanged. This discrepancy may be explained by differences in response to RB and its acetate analog, dose of test article, differences in the cell lines used, or mechanisms of ablative and photodynamic therapies. Moreover, HMGB1 levels in the sera of patients were increased after IL PV10. This is in line with another study that showed increased HMGB1 levels in the serum of cancer patients after chemoradiation; notably, HMGB1 levels were increased in patients with antigen-specific T cell responses and higher expression of HMGB1 in resected tumor samples was correlated with better survival [40].
Maturation of DCs is crucial for priming CD8 + T cells [41]. HMGB1 has been shown to be important for activation of myeloid and plasmacytoid DCs [25,31,[42][43][44][45][46]. In our model, DC maturation with upregulation of CD40, CD80 and CD86 was measured in tumor draining LN after IL PV10. Furthermore, our study showed that HMGB1 in the supernatant of tumor cells treated with PV10 was responsible for the upregulation of CD40 expression on BMderived DCs and for the increased ability of DC to stimulate T cell activation. It has been shown that shortterm CD40 signaling augments DC migration to tumordraining LNs and induced protective immunity. Moreover, HMGB1 has been shown to enhance DC responses to CCL9 and CXCL12 [47]. Interactions between HMGB1 and RAGE can induce the migration of s.c. injected DCs into DLNs [48]. In our study, IL PV10 increased the number of DCs migrating from the tumor site into the draining LNs.
In this study, we have shown a mechanism of tumor-specific immune response induced by IL PV-10. In melanomabearing mice, IL PV10 induced necrosis of tumor cells leading to the release of HMGB1, which is crucial for DC activation. This resulted in DC maturation and infiltration into draining LNs for the activation of tumor-specific T cells. Additionally, increased HMGB1 levels measured in sera of patients treated with IL PV 10 suggests that HMGB1 may be involved in eliciting a systemic immune response in patients. We have shown that circulating T cell populations and tumor-specific CD8 + T cells are increased in melanoma patients after IL PV10 therapy. Together these results support the design of additional clinical studies to measure antitumor immune responses after IL injection of PV10 in patients with melanoma.

Cell lines and cell culture
NIH3T3, 293T and melanoma B16 cells were obtained from American Type Culture Collection (ATCC). Human melanoma cells (526, 624 and 888) were obtained from the NIH. M05 tumor was generated by transfection of B16 melanoma with pAcneoOVA plasmid and was provided by Dr. Kenneth Rock (Dana-Farber Cancer Institute). M05 cells were maintained in media supplemented with 0.8 mg/ml G418 [36]. B16F10lucG5 cells stably expressing firefly luciferase were obtained from Caliper Life Sciences. All cell lines were passaged less than 10 times after initial revival from frozen stocks and tested negative for mycoplasma contamination. For tumor digestion, tumors were isolated from tumorbearing mice treated with PV10 or PBS and were digested with tumor dissociation buffer (Miltenyi Biotec) and GentleMACS (Miltenyi Biotec). After lysis of RBCs, singlecell suspensions were analyzed by FACS.

Flow cytometry, cell apoptosis assay and tetramer staining
Singlecell suspensions from the indicated tissues were prepared by pressing cells through a 70 μm cell strainer. After RBC lysis, cells were stained in FACS buffer with the following antibodies for flow cytometric analysis: antihuman CD3, CD4, CD8, CD25, CD11b, HLA-DR, CD14, FOXP3 and CD56; anti-mouse CD11c, IA b , CD45.1, CD45.2, CD8, CD4, CD3, CD86, CD80 and CD40 (all from BD Bioscience). Cell apoptosis assays were performed by using Annexin V apoptosis detection kits (ebioscience). For tetramer staining, cells were stained with H-2K b /SIINFEKL tetramer (MBL international) at room temperature for 20 minutes, followed by an additional 20 minute incubation with additional antibodies on ice. Live/dead fixable near-IR or aqua fluorescent reactive dyes (Thermo Fisher Scientific) were used to exclude dead cells before analysis. Cells were acquired by LSR II equipped with four lasers (BD Biosciences), and the data were analyzed with FlowJo (Tree Star).

ELISA
For detection of IFN-γ in mouse samples, splenocytes from PBS or PV10treated M05bearing mice were expanded with 1 μg/ml SIINFEKL peptide, 20 ng/ml IL-15 and 20 ng/ml IL21 (R&D Systems) and then mixed with irradiated M05 cells at a ratio of 10:1 [49]. For detection of IFN-γ from human samples, CD8 + T cells were isolated from peripheral blood mononuclear cells (PBMCs) with a human CD8 + T cell isolation kit (Miltenyi Biotec). Cells were cocultured with tumor cells in triplicate at a ratio of 1:1. IFN-γ production in the supernatants was measured after 48 hours with an IFN-γ ELISA kit (BD bioscience). For detection of DAMPs in cell supernatants or in patient serum, an ELISA kit for HMGB1 (IBL international), IL-1α or HSP70 ELISA kit (both from R&D Systems) was used.

Induction of lung and subcutaneous B16 lesions
C57BL/6 mice were injected i.v. with 5e5 viable B16-F10-luc-G5 and s.c. in the left flank with 1e5 B16-F10 cells. Seven days later, mice were treated by IL injection of the subcutaneous tumor with 50 μl PV-10 or PBS. Mice were treated with 100 μg of purified CD8 depleting antibody (2.43, BioXcell) or isotype (rat IgG 2b ) 2 and 3 days prior to PV10 treatment and twice per week thereafter. Mice were shaved, intraperitoneally injected with luciferin (150 mg/ kg of body weight) and imaged with Xenogen IVIS® 200 Image Series (PerkinElmer) on day 21.

Adoptive transfer of T cells
CD45.1 + OT-1 T cells were purified with a T cell enrichment column (CD3 + T cell purity >90%) (R&D Systems) and were incubated with CellTracer™ Violet (Thermo Fisher Scientific) for 20 min at 37°C. After two washes in PBS, 3e5 labeled cells were resuspended in 100 μl of PBS and injected i.v. into MO5 tumorbearing mice. After 4 days, spleens, lymph nodes (LNs) and tumors were harvested and stained with CD45.1, CD45.2 and CD3 antibodies. Cells were gated on CD45.1 + CD45.2 CD3 + and cells with at least one division were considered "divided cells".

Dendritic cell enrichment
M05 melanoma tumorbearing mice were treated with IL PV10 or PBS on day 7. After 18 hours, mice were sacrificed and tumor draining LNs were dissected and pooled for single cell preparation. DCs were enriched from pooled samples with a Pan Dendritic cell isolation kit (Miltenyi Biotec) and stained with antibodies against CD11c, MHC II, CD40, CD86 and CD80. Cells were analyzed on live CD11c + MHC II + cells. www.impactjournals.com/oncotarget

T cell proliferation assay
Bone marrow derived DCs were purified by Optiprep gradient (AxisShield) after 5 days of BM culture with 10 ng/ml GMCSF and 20 ng/ml GMCSF and 10 ng/ml IL4 [50,51]. DCs were cultured for 2 days with GMCSF, IL4 and 20% tumor supernatants (as described above). Next, DCs were pulsed for 2 hours with 10 μg/ml of OVA protein (SigmaAldrich). After multiple washes, DCs were cocultured with 1e5 responder OT1 T cells in triplicate, in Ubottom 96well plates at different stimulatortoresponder ratios for 3 days. 3 H-thymidine (1 μCi) was added to each well 18 hours prior to cell harvesting. T cell proliferation was measured by 3 Hthymidine incorporation in a liquid scintillation counter Microbeta® Trilux (PerkinElmer).

HMGB-1 blockade
BMderived DCs and 20% tumor supernatants were separately incubated with an antagonistic antibody against HMGB1 (IBL international) or the relevant isotype for 4 hours, and then were mixed together for culture for 2 days. After multiple washes, DCs were pulsed with OVA protein and cocultured with OT1 T cells for T cell proliferation as described above.

Determination of IC50
Cells were incubated in the dark with 12.5, 25, 50, 100, or 200 μM PV-10 or PBS in a 12-well plate for 6, 12, 24 and 48 hours. All cells in wells were collected, mixed with counting beads and acquired by LSR II. DAPI was used to exclude dead cells before analysis. The absolute number of live cells was calculated by comparing the ratio of bead events to cell events. The half maximal inhibition of PV10 on cell growth was determined as IC50 using GraphPad Prism.

Human subjects
Fifteen patients with dermal and/or subcutaneous metastatic melanoma were enrolled in a pilot study (NCT01760499). Peripheral blood and serum were collected prior to biopsy, 714 days after IL PV10 injection into a single melamona tumor, and 2128 days after IL PV10 injection. PBMCs were isolated by Ficoll-Paque Plus (GE healthcare). Blood samples were sent for HLA typing to determinate HLAmatched tumor and HLA mismatched tumor for each patient. Serum was prepared by collecting the supernatant after incubation of blood at room temperature for 1 hour and centrifugation at 1,000 g. Two tumor lesions in each patient were sampled by biopsy pre-treatment; one of the two lesions was injected with IL PV10 7 days after biopsy, then both residual sites were completely excised 714 days later. Biopsy specimens were fixed in formalin and embed in paraffin. The specimens were stained with hematoxylin and eosin stains for determination of pathologic complete response. Immunohistochemistry for melanin A (mel A) was performed. Flow cytometry was performed to detect CD3, CD4, CD8, and CD56 staining on PBMC.

Statistical analysis
The data were analyzed with a twotailed Student ttest or Wilcoxon matched pairs test by GraphPad Prism. A p value of < 0.05 was considered statistically significant.