Interferon-α-inducible Dendritic Cells Matured with OK-432 Exhibit TRAIL and Fas Ligand Pathway-mediated Killer Activity

Active human dendritic cells (DCs), which efficiently induce immune responses through their functions as antigen-presenting cells, exhibit direct anti-tumour killing activity in response to some pathogens and cytokines. These antigen-presenting and tumour killing abilities may provide a breakthrough in cancer immunotherapy. However, the mechanisms underlying this killer DC activity have not been fully proven, despite the establishment of interferon-α (IFN-α)-generated killer DCs (IFN-DCs). Here mature IFN-DCs (mIFN-DCs), generated from IFN-DCs primed with OK-432 (streptococcal preparation), exhibited elevated expression of CD86 and human leukocyte antigen-DR (minimum criteria for DC vaccine clinical trials) as well as antigen-presenting abilities comparable with those of mature IL-4-DCs (mIL-4-DCs). Interestingly, the killing activity of mIFN-DCs, which correlated with the expression of CD56 (natural killer cell marker) and was activated via the tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) and Fas ligand pathway, was stronger than that of IFN-DCs and remarkably stronger than that of mIL-4-DCs. Therefore, mIFN-DCs exhibit great potential as an anti-cancer vaccine that would promote both acquired immunity and direct tumour killing.

Significant advances have been made in the surgical and radiotherapeutic techniques and chemotherapeutic agents (e.g. immune checkpoint inhibitors) that comprise the field of cancer therapeutics [1][2][3][4][5][6] . However, the treatment of advanced cancers, which are characterized by organ involvement and distant metastasis, remains extremely difficult. Accordingly, oncological researchers have sought immunologic therapies, such as anti-tumour vaccination.
Thus far, antigen-presenting cell (APC)-based vaccination with active dendritic cells (DCs) has been evaluated as a method of efficient immunity against cancer antigens 7 . In this process, autologous monocyte-derived mature DCs are conventionally manufactured using granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4) and are principally targeted against a specific cancer antigen [8][9][10] . However, human DC-based vaccine technology will require a breakthrough to achieve effective cancer treatment. We note that a population of cytotoxic DCs can directly kill tumour cells, in addition to inducing tumour antigen-specific cytotoxic T lymphocytes (CTLs); these 'killer DCs' , therefore, appear to exhibit strong potential in the field of cancer immunotherapy.
DCs are generally divided into two major subtypes, myeloid DCs and plasmacytoid DCs (pDCs); however, in mice, interferon-producing killer DCs (IKDCs) have been described as a third subtype 11,12 . These latter cells produce substantial amounts of type I interferon (IFN) and IL-12 or IFN-γ , depending on the activating stimuli and can kill typical targets of natural killer (NK) cells via NK-activating receptors in response to stimulation with CpG 11 . In particular, B220 + NK1.1 + DCs have been reported to secrete large amounts of IFN-α and to promote the tumour necrosis factor-related apoptosis-inducing ligand (TRAIL)-dependent lysis of tumour cells 12 . In addition to IKDCs in mice, several subsets of killer DCs that exhibit anti-tumour cytotoxicity have been reported in humans 13 . These human killer DCs can be generated from monocytes in vitro in response to some pathogens and cytokines 14 ; for example, immature IL-4-DCs (imIL-4-DCs) generated using GM-CSF and IL-4 exhibit killing Scientific RepoRts | 7:42145 | DOI: 10.1038/srep42145 activity following stimulation with bacterial lipopolysaccharide (LPS) 15 , CD40 ligand 16 or OK-432, a streptococcal preparation 17 .
In Japan, clinical-grade OK-432 has been used as anti-tumour agent for more than 20 years, and its safety is well established [18][19][20][21][22][23] . Mechanistically, OK-432 promotes the functional maturation of imIL-4-DC through ligation of TLR4 24 and TLR9 25 , and this maturation correlates with the upregulated expression of CD80, CD83 and CD86 17,26,27 , thus promoting the effective induction of antigen-specific T cells 26 . Combined treatment of mature IL-4 DCs (mIL-4-DCs) with OK-432 and prostaglandin E2 (PGE2) upregulates the expression of CD197 (CCR7), which is associated with migration to lymph nodes 27 . Katy et al., reported that following OK-432 activation, human mIL-4-DCs could specifically kill tumour cells via a novel CD40/CD40 ligand-mediated mechanism, without affecting normal cells 17 . OK-432 also induces the production of IL-12 from matured DCs without increasing the production of immuno-suppressive cytokines such as IL-10 17 .
Similarly, GM-CSF and IFN-α-generated DCs (IFN-DCs) also exhibit killing activity 28,29 . These cells express higher mRNA levels of NK cell markers, such as TRAIL, granzymes and killer cell lectin-like receptors (KLRs) 28 , and kill tumour cells via soluble TRAIL 29 , a factor that targets various tumours but does not affect normal cells [30][31][32] . According to previous reports, IFN-DCs exhibit a semi-mature phenotype, and full maturation can be achieved using CD40 ligand, LPS or polyribocytidylic acid (poly I-C) 33,34 . However, the effects of OK-432 on the maturation or killing activity of IFN-DCs have not previously been evaluated.
In the present study, we investigated the role of OK-432 in this setting and found that this agent not only induced the maturation, but also facilitated the killing activity of IFN-DCs. The mature IFN-DCs (mIFN-DCs) derived from OK-432-primed immature IFN-DCs exhibited similarly strong antigen-presenting abilities as mIL-4-DCs. Further study indicated that the killing activity of mIFN-DCs was stronger than that of IFN-DCs and remarkably superior to that of mIL-4-DCs. These characterizations suggest the strong clinical potential of a mIFN-DC-based vaccine for clinical use, as such an agent would be expected to induce acquired immunity and direct tumour killing.

Results
OK-432 induces a mature IFN-DC phenotype. We first examined the phenotypic effects of OK-432 on mIFN-DCs and mIL-4-DCs. After OK-432 stimulation, similar remarkable clusters of cells bearing the morphologic features of mature DCs were observed equally among mIFN-DCs and mIL-4-DCs (Fig. 1). A phenotypic assessment of surface markers revealed the increased expression of costimulatory molecules such as CD80 and CD86 on mIFN-DCs, whereas the levels of these markers on mIL-4-DCs had not significantly changed (Fig. 2). Similarly, increases in the levels of CD14 and human leukocyte antigen (HLA)-DR were more evident on mIFN-DCs than on mIL-4-DCs. In contrast, mIL-4-DCs featured more strongly elevated expression of CD40, CD83 and CD197. induce CTLs in an in vitro system and thus confirm the antigen presentation abilities of our DC populations. As shown in Fig. 3a, the frequency of MART-1 specific CTLs among total CD8 + T cells was 2.4-fold higher following mIFN-DC induction, compared to IFN-DC induction (mIFN-DCs vs. IFN-DCs = 4.9% vs. 1.9%). In contrast, the percentages of CTLs induced by mIFN-DCs and mIL-4-DCs did not differ significantly (Fig. 3b).
FITC-dextran uptake by DCs was determined to assess pinocytotic activity. The change in mean fluorescence intensity (MFI) of FITC-dextran was slightly lower in mIFN-DCs than in IFN-DCs, similar to mIL-4-DCs (Fig. 4a, left). Next, DQ-ovalbumin, a self-quenched albumin that fluoresces upon proteolytic degradation, was used to demonstrate an obvious reduction in the phagocytotic activity of mIFN-DCs upon OK-432 stimulation; this was also observed in mIL-4-DCs (Fig. 4a, right).

mIFN-DCs possess stronger cytotoxic activity than mIL-4-DCs.
The killing activity of DCs against K562 cells was investigated using flow cytometry. In a representative experiment, IFN-DCs exhibited a killing activity of 17.1%, compared with 40.3% by mIFN-DCs (Fig. 5a), indicating a significant increase only in mIFN-DCs relative to IFN-DCs (Fig. 5b). Moreover, mIFN-DCs showed remarkably enhanced killing activity against K562 cells when compared with mIL-4-DCs. Unexpectedly, we did not observe a significant increase in tumour killing activity by mIL-4-DCs following priming with OK-432 (Fig. 5b). Accordingly, we next focused on the mechanisms underlying this enhanced tumour killing by mIFN-DCs.
Tumour killing activity of mIFN-DCs correlates with the expression of CD56 and depends on soluble factor(s) and cell contact. Relative to the expression of IFN-DCs, CD56 was significantly increased on mIFN-DCs following OK-432 treatment (Fig. 6a), and a weak correlation was observed between the expression of CD56 and the killing activity of mIFN-DCs (r = 0.477; Fig. 6b). In contrast, the expression of CD56 on imIL-4-DCs and mIL-4-DCs was extremely low, consistent with the observed weak killing activity (Fig. 6a).
Killing activity of mIFN-DCs against K562 cells was observed using fluorescence microscopy to confirm the cell contact dependency of tumour killing. IFN-DCs or mIFN-DCs were labelled with the red fluorescence dye PKH26, followed by incubation with K562 cells labelled with the green fluorescence dye CFSE. Remarkably, mIFN-DCs were found to form larger clusters with K562 cells, compared with IFN-DCs (Fig. 6c). Furthermore, large numbers of dead K562 cells ( Fig. 6c: white arrows = DAPI staining) in clusters with mIFN-DCs, whereas only a few dead cells were observed in clusters with IFN-DCs. These results suggest that the formation of clusters between mIFN-DCs and target cells is important for tumour killing by mIFN-DCs. The involvement of cell contact was additionally investigated using transwell membrane assays. The killing activity of mIFN-DCs against K562 or MiaPaCa2 cells was reduced significantly in the presence of a transwell membrane insert, whereas no effect was observed for IFN-DCs (Fig. 6d). A cytotoxic assay in the presence of transwell inserts revealed that the killing activity of mIFN-DCs was reduced to a level similar to that of IFN-DCs (against K562 cells: median, IFN-DC = 9.5%, mIFN-DCs = 10.8%; against MiaPaCa2 cells: IFN-DCs = 17.0%, mIFN-DCs = 19.7%).

mIFN-DCs exert killing activity through the TRAIL and Fas ligand pathway. TNF family members
(TNF-α, Fas ligand, TRAIL) [35][36][37][38][39][40] , perforin/granzyme B 41 , NO 42,43 and peroxynitrites 44 have been reported to be involved in the tumour killing mechanisms of killer DCs. The above results suggested that the increased killing activity of mIFN-DCs in response to OK-432 depended on a cell contact mechanism, leading to the speculation of involvement by cell surface-localized cytotoxic factors in mIFN-DC-mediated tumour killing. Notably, increased cell surface expression of TRAIL and Fas ligand was observed on both mIFN-DCs and mIL-4-DCs primed with OK-432 (Fig. 7a). In particular, the expression of TRAIL was significantly higher on mIFN-DCs than on mIL-4-DCs, whereas both cell populations exhibited similar increases in the expression of Fas ligand.
We subsequently examined whether the tumour killing activity of mIFN-DCs was dependent on TRAIL and Fas ligand. K562 (TRAIL sensitive) 17 and MiaPaCa2 (Fas sensitive) cell lines 45 were treated with antibodies specific for TRAIL or Fas ligand, respectively, to test the blocking of each ligand. TRAIL antibody blockade significantly suppressed the killing of K562 cells by mIFN-DCs (Fig. 7b, left panel), whereas this treatment had no significant effect on the killing of K562 cells by IFN-DCs. Similarly, Fas ligand antibody blockade reduced the killing of MiaPaCa2 cells by mIFN-DCs (Fig. 7b, right panel), whereas this treatment did not influence IFN-DCs-mediated killing. These results suggested that, in contrast to IFN-DCs, the killing activity of mIFN-DCs is mediated through the TRAIL and Fas ligand pathway.

Discussion
OK-432 has been well established as a clinical agent for DC maturation [18][19][20][21][22][23] . Accordingly, a 4-day non-adherent culture system in which IFN-DCs were primed with OK-432 was developed to generate mIFN-DCs. In contrast to mIL-4-DCs, the short-term generation of mIFN-DCs is also clinically beneficial because although mIFN-DCs express lower levels of CD40 and CD83, the increased expression levels of CD86 and HLA-DR, which were similar to those of mIL-4-DCs, met the minimum release criteria for DC vaccines 46 . Based on the pinocytotic and phagocytotic activity, cytokine production and antigen presenting profiles of mIFN-DCs were found to correlate with those of mature DCs, similar to mIL-4-DCs. Interestingly, OK-432 also facilitated an increase in the killing activity of IFN-DCs. However, the killing activity of mIFN-DCs, which correlated with the expression of CD56, was stronger than that of both IFN-DCs and mIL-4-DCs. OK-432 upregulated the expression of CD56 and cell surface TRAIL and Fas ligand on mIFN-DCs, and the independent involvement of TRAIL and Fas ligand in mIFN-DC-mediated tumour killing was proven via specific antibody blockade.
Although OK-432-activated human DCs kill tumour cells through CD40/CD40 ligand interactions 17 , other factors that may be involved in this cytotoxic process have not been fully examined. The association of CD56 with cell adhesion 47,48 suggested that tumour killing by mIFN-DCs with expression of CD56 would depend on cell contact. We further speculated that cell surface cytotoxic molecules, such as tumour necrosis factor (TNF) family members, might participate in the required cell-to-cell interactions. Although OK-432 stimulation was previously found to upregulate the expression of Fas ligand on mononuclear cells 49 , our report is the first to demonstrate the increased expression of CD56, cell surface TRAIL and Fas ligand on mIFN-DCs following OK-432 stimulation. The observed clustering of mIFN-DCs and tumour cells might facilitate tumour killing via either cell surface TRAIL or Fas ligand. The reduced killing activity of mIFN-DCs in the presence of anti-TRAIL antibody suggests that these cells use mechanisms other than CD40/CD40 ligand signalling to exert their cytotoxic effects 50 . In contrast to previous reports, however, OK-432 did not promote a significant increase in tumour killing by mIL-4-DCs, possibly because of differences in the maturation cocktail used in combination with OK-432 and PGE2, as well as the incubation time 17 ; both factors might have affected tumour killing by OK-432-induced DCs.
Clearly, OK-432-induced mIFN-DCs exhibited remarkably strong killing activity, which correlated with the expression of CD56 and depended on TRAIL and Fas ligand. Previously, CD56 + IFN-DCs exhibited stronger cytotoxicity against K562 cells, the majority of which are TRAIL positive, when compared with CD56 − IFN-DCs 29 . However, in our study, we did not observe significant suppression of tumour killing by IFN-DCs in and calculated by subtracting the MFI at 4 °C from that at 37 °C. (b) DCs were seeded into 24-well plates at a density of 2 × 10 5 cells/ml for 24 h, and the supernatants were subjected to enzyme-linked immunosorbent assay (ELISA) analysis to determine the levels of IL-12p70, IFN-γ , TNF-α, IL-10 and IL-6. A summary of the cytokine production data is shown (N = 8).
Scientific RepoRts | 7:42145 | DOI: 10.1038/srep42145 response to anti-TRAIL antibody treatment. The observed expression of granzyme B 28 suggests that IFN-DCs might use multiple soluble factors to kill tumour cells. However, IFN-DCs are known to share phenotypic features with pDCs (e.g. CD11c and CD123) 29 , and pDC tumour-killing also correlates with the expression of TRAIL and CD56. Furthermore, the expression of TRAIL and TRAIL-dependent cytotoxicity are upregulated in pDCs in response to HIV infection 51 . pDCs activated by the tick-borne encephalitis vaccine FSME specifically upregulate the expression of CD56, which coincides with elevated levels of cytotoxic molecules such as TRAIL; furthermore FSME-activated pDCs specifically lyse NK target cell lines in a cell-cell contact dependent manner 52 . Accordingly, OK-432-induced mIFN-DCs might exhibit pDC-like responses.
It remains to be determined whether mIFN-DCs can kill tumour cells and induce CTLs in vivo. As mentioned above, mIFN-DCs express CD86 and HLA-DR 46 and thus potentially possess antigen presentation ability comparable to that of mIL-4-DCs. However, the low level of expression of CD197 on mIFN-DCs might not be sufficient to promote migration to the lymph nodes and subsequent CTL induction. In addition to the ability to migrate to specific regions for direct tumour killing, the ability of antigen-pulsed mIFN-DCs to induce CTLs should be examined in tumour-bearing animal models. A report by Padovan et al., in which IFN-DCs produced IP10/ CXCL10 and MIG/CXCL9 to attract activated CD8 + T cells and thus expand CTLs efficiently 53 , led to speculation that mIFN-DCs could potentially induce CTLs directly in the tumour regions, rather than in the lymph nodes. We further note that we observed higher levels of IL-6 in cultures of mIFN-DCs, compared with mIL-4-DCs (Fig. 4b). IL-6 mediates febrile and acute-phase responses by upregulating chemokines and cell adhesion molecules that support the growth of B cells and Th17 cells, but suppress regulatory T cells (T regs) 54 . Accordingly, the observed high production of IL-6 might provide mIFN-DCs with an advantage with respect to the induction of CTLs in tumour regions, which tend to harbour T regs 55 .
Clinically, mIFN-DCs may directly kill tumour cells that retain sensitivity to TRAIL and Fas ligand signalling. Unlike normal cells, various types of tumours are susceptible to TRAIL-induced apoptosis, and the Fas/Fas ligand-mediated killing activity of mIFN-DCs might be less advantageous for cancer treatment. Fas/Fas ligand interactions are known to induce apoptosis in hepatocytes 56 and to have immunomodulatory effects on immune homeostasis and tolerance 57 . Although TRAIL-or Fas ligand-mediated apoptosis-resistant cancer cells have been reported, one option for cancer therapy would combine the anti-tumour effects of TRAIL and Fas ligand with standard chemotherapy 58,59 .
In conclusion, this study confirms that OK-432 induced the maturation of IFN-DCs with strong antigen-presenting abilities (comparable to mIL-4-DCs), and facilitated the TRAIL-and Fas ligand-mediated killing activity of these mIFN-DCs. Further clarification of the characteristics of mIFN-DCs is expected to contribute to the development of adaptive immunotherapies and the achievement of acquired anti-cancer immunity.

Phenotyping of DCs.
Pinocytotic and phagocytotic assay. To evaluate pinocytotic activity, DCs were incubated with 250 μ g/ml FITC-dextran (Molecular Probes, Eugene OR, USA) in AIM-V medium supplemented with 10% FBS at 37 °С for 3 h and subsequently washed twice with FACS buffer. DCs were then incubated with 10 μ g/ml DQ-ovalbumin (Molecular Probes, Eugene OR, USA) at 37 °C for 30 min to examine phagocytotic activity. After two washes, cells were analysed using flow cytometry.
Cytokine production. DCs were seeded at a density of 2 × 10 5 cells/ml into a 24-well plate. After a 24-h incubation in AIM-V medium, cytokines were measured in the collected supernatants using ELISA kits for IL-12p70, IFN-γ (R&D Systems), TNF-α, IL-6 and IL-10 (BD Biosciences) according to the manufacturers' protocols.
Cytotoxicity assay. To examine the cytotoxic activity of DCs, we used a previously reported method 28 with slight modifications. Initially, 1 × 10 6 K562 (ATCC, Manassas, VA, USA) or MiaPaCa2 cells (RIKEN BRC, Tsukuba, Japan) were labelled with carboxyfluorescein succinimidyl ester (CFSE; 5 μ M; Molecular Probes) in PBS supplemented with 0.1% FBS at 37 °C for 10 min, followed by two washes with AIM-V medium. Different numbers of DCs were used as effectors and were incubated with 1 × 10 4 CFSE-labelled K562 or MiaPaCa2 cells (targets) in AIM-V medium containing 10% FBS at 37 °C for 18 h. To evaluate intrinsic controls, an aliquot of K562 or MiaPaCa2 cells equivalent to the number of effector cells was incubated with labelled target cells. The cells were then washed in FACS buffer, stained with propidium iodide (PI; 2 μ g/ml; Sigma-Aldrich Co. LLC., Tokyo, Japan) for 5 min and analysed using flow cytometry. Tumour-specific killing activity was determined as follows: % killing = experimental % of PI + CFSE-labelled cells − intrinsic % of PI + CFSE-labelled cells. To assess contact-dependent killing, CFSE-labelled target cells and effector cells were separated using transwell inserts fitted with 0.4-μ m pore polyester membranes (Corning Inc., Corning, NY, USA). For the blocking experiment, effector cells were pre-treated with 10 μ g/ml anti-TRAIL or anti-Fas ligand mAbs (Biolegend, San Diego, CA, USA) at 37 °C for 2 h and were incubated with labelled target cells. Statistical analysis. The Wilcoxon signed-rank test with Bonferroni correction was used to compare differences among groups. Spearman's rank-order correlation was used to calculate the correlation of the expression of CD56 with killing activity. All statistical analyses were performed using IBM SPSS Advanced Statistics software, version 23.0 (IBM Japan, Tokyo, Japan). Differences were considered statistically significant at a p value < 0.05.