Development of effective tumor immunotherapy using a novel dendritic cell–targeting Toll-like receptor ligand

Although dendritic cell (DC)-based immunotherapy shows little toxicity, improvements should be necessary to obtain satisfactory clinical outcome. Using interferon-gamma injection along with DCs, we previously obtained significant clinical responses against small or early stage malignant tumors in dogs. However, improvement was necessary to be effective to largely developed or metastatic tumors. To obtain effective methods applicable to those tumors, we herein used a DC-targeting Toll-like receptor ligand, h11c, and examined the therapeutic effects in murine subcutaneous and visceral tumor models and also in the clinical treatment of canine cancers. In murine experiments, most and significant inhibition of tumor growth and extended survival was observed in the group treated with the combination of h11c-activated DCs in combination with interferon-gamma and a cyclooxygenase2 inhibitor. Both monocytic and granulocytic myeloid-derived suppressor cells were significantly reduced by the combined treatment. Following the successful results in mice, the combined treatment was examined against canine cancers, which spontaneously generated like as those in human. The combined treatment elicited significant clinical responses against a nonepithelial malignant tumor and a malignant fibrous histiocytoma. The treatment was also successful against a bone-metastasis of squamous cell carcinoma. In the successful cases, the marked increase of tumor-responding T cells and decrease of myeloid-derived suppressor cells and regulatory T cells was observed in their peripheral blood. Although the combined treatment permitted the growth of lung cancer of renal carcinoma-metastasis, the marked elevated and long-term maintaining of the tumor-responding T cells was observed in the patient dog. Overall, the combined treatment gave rise to emphatic amelioration in DC-based cancer therapy.


Introduction
The immune system has the potential to eliminate tumor cells by the interplay between innate and adaptive immunity. Dendritic cells (DC) are considered as the most potent antigen presenting cells to provide an essential link between innate and adaptive immunity. DC vaccination plays a major role in cancer immunotherapy, priming immune responses against cancer. Vaccination of DCs loaded with cancer antigen has given rise to some therapeutic effect in murine tumor models [1], and has been used in patients with differing types of cancer [2][3][4]. The treatment has almost no toxicity, but the immune responses were transient and the clinical outcome is not particularly successful. This may be partly due to degradation of DCs after injection, or inhibition of DC function by certain tumors [5,6], and various suppressor cells in the tumor environment [7]. Three improvements are required to enhance DC-based cancer immunotherapy. These are to strengthen the immune function of DCs, to improve the immune environment in cancer tissue so as to prevent degradation of DCs and facilitate the function of effector cells, and to control the generation of suppressor cells so as to maintain anti-cancer immune responses originally generated by the DCs.
Signals from Toll-like receptor (TLR) are an important link between innate and adaptive immunity. TLR 2 signals enhance the activation and maturation of DCs so as to induce antitumor cytotoxic activity [8] . Post-surgery treatment with Bacillus Calmette-Guerin cell wall skeleton, an agonist of TLR 2, improved the prognosis of patients with lung cancer [9]. TLR 2 is expressed not only by DCs but also macrophages and some epithelial cells [10]. Also, some agonists of TLR bind nonspecifically to membranes of various cells by means of cationic charge. These properties together result in severe inflammation at the injection site. A novel synthetic lipopeptide, h11c has both a TLR2 ligand (PC2: a modified bacterial lipopeptide with two palmitate) and a DC-targeting peptide (ATPEDNGRSFS), which selectively bind to human CD11c on DCs [11]. We therefore expect h11c to give rise selectively to a potent immune response against tumor antigens presented by DCs while averting nonspecific inflammation.
We recently found that interferon-gamma (IFNγ), which is a typical activator of T helper type 1 responses, induces maturation and activation of DC, and found satisfactory clinical outcomes in the treatment of dog tumors by intratumoral injection of IFNγ along with DCs [12]. Unfortunately, this treatment is difficult to use in cases of visceral metastasis. Furthermore, IFNγ induces myeloid-derived suppressor cells (MDSC) [13]. MDSC, as well as regulatory T cell (Treg), is an undesirable inhibitor of tumor immunity, and induced by prostaglandin E2, which is produced from tumor-infiltrated macrophages and is synthesized by cyclooxygenase2 (COX-2) [14]. It therefore, follows that COX-2 inhibitor (COX2-I) have a critical role in preventing generation of MDSCs.
In this study, we show that effective therapeutic responses obtained in DC-based therapy using a combination of h11c, IFNγ and a COX2-I in mouse models of visceral tumor and in clinical treatment against canine tumors. These results propose a promising method for human cancer therapy.

Materials and methods
Animals and cell lines and reagents C57BL/6 (B6), C3H/He (C3H) and BALB/c mice were purchased from Japan SLC Inc. (Hamamatsu, Japan). The mice were maintained under specific pathogen-free conditions. After starting experiments mice were monitored daily for weight loss, labored respiration and any sign of discomfort. There were no any unexpected deaths. The experimental endpoint was 60 days after injecting tumor cells. In the surface tumor models, mice were euthanized if the tumor mass grew greater than 20 mm. In the visceral tumor models, mice were euthanized if loss of appetite, suffering from pain or labored breathing was observed. The mice were humanely euthanized by anesthesia with sodium pentobarbital (200 mg/Kg, intraperitoneal injection) followed by cervical dislocation. Laboratory-bred beagle bitches, 4 to 8 years old, were housed and maintained according to NIH guidelines. All animals were kept in the animal facility of Osaka Prefecture University. The study protocols were approved by the animal experiment committee of Osaka Prefecture University. The clinical study was performed on dogs with malignant tumors, which had been admitted to the Veterinary Medical Center of Osaka Prefecture University (Izumisano, Japan) and Kakogawa Animal Hospital (Kakogawa, Japan). Written consent was obtained from all dog owners prior to the start of the study.
The h11c and P2CSK4 lipopeptides and FITC-labeled versions of these lipopeptides were prepared as described previously [11]. OVA was purchased from Sigma-Aldrich Chemical (St. Louis, MO). It was confirmed as a preliminary that h11c was not toxic to dogs at less than 10 μg/kg.

Preparation of DCs and tumor antigen
Mouse DCs were prepared from bone marrow cells, as described previously [16]. Dog DCs were prepared from peripheral blood monocytes (PBMCs) as described previously [12].
The tumor lysate was used for tumor antigens, and was prepared by the freeze-thaw method. OVA was used as the antigen of EG.7-OVA. The antigens were adjusted at 1 mg/ml in OPTI-MEM (Invitrogen).

Assay of immune activity
Activation of mouse DCs was evaluated with expression of costimulatory molecules and cytokines by flow cytometry (FCM) or enzyme-linked immunosorbent assay (ELISA) as described previously [11].
Activity of mouse DCs to induce antigen-specific cytotoxic T lymphocytes (CTL) was evaluated in a 51 Cr released assay as described by Koizumi et al [17]. where 51 Cr-labeled E.G7-OVA or EL4 cells were incubated with cells from the popliteal lymph node of B6 mice after inoculation of the OVA-loaded DCs (2×10 5 ) into the footpad.
To estimate the immune status of tumor tissues, immunohistochemistry (IHC) staining was preformed to detect macrophages/DCs, CTL/natural killer (NK) cells and Tregs in the tumor tissues.
Activation of dog DCs by h11c was evaluated by measuring interleukin (IL)-12 production as described previously [11]. As described by Ning et al [18], these tumor-responding T cells (TRTC) in dog PB mononuclear cells were quantified as the IFNγ-producing CD4 + or CD8 + cells by FCM after incubating with the tumor antigens (0.05-0.1 mg protein in 1 ml) at 37˚C overnight.
For evaluation of secreted cytokines in cultures, the culture supernatants were analyzed as described previously [11], using ELISA kits for mouse IL-12p40, mouse IL-6, mouse IL-10, mouse IFN-beta and dog IL-12p40 (R&D systems, Minneapolis, MN, USA).
IHC staining was performed on paraffin embedded section of tumor tissues as described previously [23], using rabbit polyclonal antibodies against ionized calcium-binding adapter molecule 1 (Iba1: Wako, Osaka, Japan), which is a marker of macrophage and dendritic cells, and enhanced in expression by activation [24,25], human granzyme B (Spring Bioscience, Pleasanton, CA, USA), which is a marker of CTLs and NK cells [26], and mAb against mouse FoxP3 (clone FJK-16s, eBioscience). Cross reactivity of the polyclonal antibodies to the corresponding mouse antigens was confirmed by the manufacturers. The numbers of positive cells were counted in five high-power (×400) fields selected randomly, and the positive cells in a total of 1000 cells were evaluated.

DC-based immunotherapy
In the mouse model, six treatments were prepared for examination of the therapeutic effects: (i) [DC], mouse DCs (2×10 6 ) were suspended in 0.2 ml tumor antigen solution and incubated at 37˚C for 2 hr. After incubation, the DCs were washed and suspended with 0.2 mL PBS (-) prior to inoculation. The DC inoculation was performed four times at 7 day-intervals. treatments were performed for the following combinations of tumor and mouse strain: E. G7-OVA-B6, LM8-C3H and CT26.WT-BALB/c. As described previously [11], E.G7-OVA cells (2×10 6 ) were s.c. injected on the back of B6 mice. Treatments began 10-12 days after the injection when the diameter of the tumors was 5-6 mm. DCs were injected directly into the tumors. The treatments were performed four times at 7-day intervals. The size of the tumors was measured 3 times per week, and the volume was calculated as described previously [27]. In the experiment using LM8 or CT26.WT, tumor cells (2×10 5 ) were i.v. injected and DCs were s.c. inoculated. Treatments started on the same day as the injection of tumor cells, and were performed four times at 7-day intervals. The therapeutic effect was evaluated according to the survival of the mice after the injection to the tumor. The endpoint was 60 days after the treatments.
In the clinical treatments in dogs, DCs were incubated with h11c (1.2-2.4 μg/kg) in 0.5 ml solution of tumor antigens (0.5-1.0 mg/ml) at 37˚C for 2 hr. After incubation the DCs were injected directly in to surface tumors or s.c. for visceral tumors, along with recombinant canine IFNγ (10 4 unit/kg, Interdog1, Toray, Tokyo, Japan). The dose of recombinant canine IFNγ was that recommended by the manufacturer. The injection was repeated at weekly intervals. In addition, the dogs were orally administrated daily with a COX2-I, firocoxib (Previcox1, Nippon Zenyaku Kogyo, Fukushima, Japan) at the manufacture's recommended dose. The therapeutic effect was evaluated by measuring the tumor size, and by FCM analysis for the TRTCs, Tregs (CD4 + FoxP3 + ) and MDSCs (CD11b + CD14 -MHC II -).

Statistical analysis
Survival of different groups was analyzed by the Kaplan-Meier method and compared using the log rank test. In the other experiments, groups were compared using Student's t (two parameters) or Tukey-Kramer multiple comparison test (more than three parameters). cells. In addition to these molecules, the incubated DCs significantly enhanced expression of IL-12 and IL-6 from DCs (Fig 2), which activate cellular and humoral immunity respectively. However, the DCs also enhanced the expression of IL-10, which suppresses immune responses. DCs expressed little amount of IFN-beta did not show any increase with h11c (data not shown). As same as the in vitro results, DCs collected from lymph node (LN) of mice injected with h11c significantly enhanced the expression of the costimulatory molecules and cytokines (Fig 3, S2 Fig and S3 Fig). However, the treatment with h11c did not effected on the expression of those molecules and cytokines of DCs in the spleens (data not shown).

h11c significantly enhanced DC functions for tumor immunity
As shown in Fig 4A, h11c significantly enhanced the presentation of OVA peptide on MHC class I molecules of DCs incubated with OVA. Moreover, when DCs treated with OVA and h11c (OVA+h11c) were immunized, significantly higher OVA-specific CTL activity was induced, relative to DCs treated with OVA alone (Fig 4B). The combination of h11c-treated DC and COX-2 inhibitor significantly inhibited tumor growth and enhanced immune activity against tumor in the mouse model Three combinations of the tumor transplantation models in mouse were chosen in order to investigate the therapeutic effect. To estimate the local inhibition of tumor growth, the OVA antigen-presenting DCs were injected into tumors of E.G7-OVA growing on the backs of syngeneic B6 mice. As shown in Fig 5A, (Fig 5C). There was a significantly better survival in the [h11c-DC]+IFN +COX2-I group than the other group, however.

h11c had an affinity for dog DCs and activated them
Since the combined treatment ([h11c-DC]+IFN+COX2-I) elicited successful results in murine tumor-models, we, as the next step, examined the enhancing effects in the clinical treatment for tumor patients of dog. The h11c was originally designed to bind to human CD11c. In a previous study it was demonstrated that h11c has high affinity for mouse bone marrow derived DCs [16]. We therefore first examined whether h11c has an affinity for dog DCs, and activates them. As shown in Fig 12A, the affinity of h11c for the dog DCs was greater than that of PC2SK4, which has the same TLR2 motif, but is a defect of the CD11c targeting molecule. In contrast, the affinity of h11c for PBMCs was less than that of PC2SK4. Moreover, as shown in Fig 12B, h11c significantly activated dog DCs in concentrations greater than 500 ng/ml. Because this dog had a heart disorder, the owner wanted a nonoperational, DC treatment. As shown in Fig 13B, the volume of the tumor had increased by 49 days after the start of the combined treatment, but began to decrease thereafter, and by day 91 was one third of the start size. The treatment interval was then changed to one in two weeks. The tumor volume continued to decrease by day 120, but then began to increase again. As a result, from day 148, the treatment interval reverted to weekly. The tumor again began to reduce, and had almost disappeared by day 198. However, the patient dog died on day 230, due to an unrelated heart disorder. FCM analysis was performed using PB collected on days 0, 14, 49, 91, 148 and 198. As shown in Fig 13C, TRTC responses against own- tumor antigens were evaluated by IFNγ production. TRTC were first increased as shown by day 49, then decreased as shown by day 91, then finally increased by remission. The response against unrelated tumor antigens was low and did not change throughout the treatment period. As shown in Fig 13D, a reduced level of Tregs was observed on day 49, but had increased by day 148, and was lowest at the remission. The level of MDSCs changed in a similar manner to Tregs.
Case no. 2 was affected by renal cell carcinoma (S5 Fig). The tumor was removed by operation, but as shown in Fig 14A, metastases were found in the lung. For this patient, the h11ctreated DCs were s.c. injected weekly, and IFNγ was s.c. injected three times per week. Growth of the metastatic tumors was not inhibited (Fig 14A), and the patient died on day 260 due to breathing problems. The TRTCs had increased by day 83, whereas decreased gradually thereafter (Fig 14B). In contrast, Tregs and MDSCs had decreased by the day 76, whereas gradually increased thereafter (Fig 14C). Case no. 3 had a tumor comprising squamous cell carcinoma in the left humeral bone ( Fig  15A and S6 Fig). For this patient, h11c-treated DCs and IFNγ were injected into subcutaneous tissue close to the affected site. After 10 cycles of the treatment, the tumor was somewhat smaller (Fig 15A). The TRTCs gradually increased with time (Fig 15B). In contrast, Tregs and MDSCs decreased with time ( Fig 15C). Unfortunately, this patient suffered unrelated renal disorder from day 84, and died on day 119. Case no. 4 had multiple tumors of the nodular type on the back (Fig 16A). Histologically, many histiocytic cells with significant atypical nucleus were observed, and diagnosed as indicating a malignant fibrous histiocytoma (S7 Fig). The h11c-treated DC and IFNγ were injected weekly into two of the tumors indicated by arrows. After eight cycles of this treatment, the treated tumors, and also the other tumors on the back had disappeared completely. No recurrence was subsequently observed, over 800 days. The TRTCs increased upon repeating the treatment, and was greatest at the end of the treatment (Fig 7B). Interestingly, a relatively high response was still observed 28 days after the treatment had stopped. The percentage of Tregs was unrelated to the clinical outcome. But MDSCs nevertheless decreased with time ( Fig 7C).

Discussion
Because of their high efficiency in initiating and activating immune responses, DCs should be a potent adjuvant for tumor immunity. Nevertheless, treatments using DCs expressing the tumor antigens have not generally been successful, especially in cases of metastasis in the internal organs [28]. In humans, additional treatments have therefore recently been performed to enhance DCbased immunotherapy. These include the enhancement of DC function by TLR agonist [29], the amelioration of immune status by inoculation of cytokines to encourage the effector cells [30][31][32], and the control of suppressor cells such as MDSC [33,34] and Tregs [34,35]. In the present study, we studied all three types of treatment using h11c, IFNγ and COX2-I.
We recently found that IFNγ, together with DCs, induces satisfactory clinical outcomes against surface tumors of the dog in the small size or early stage [12]. However, it revealed thereafter that the DC plus IFNγ treatment still not enough to elicit effective clinical outcomes  against malignant tumor in the large size or progressive stage and metastatic tumors in which DCs and IFNγ is difficult to be directly injected. The h11c was designed to bind selectively to human CD11c and activate CD11-expressing DC, and was reported to selectively bind to mouse DC [11]. This property is important to prevent injurious responses caused by activation of macrophages or T cells, and to significantly elicit immune responses against antigens expressed by DCs. As well as the activation of DCs, we newly found that h11c significantly enhanced antigen presentation and antigen-specific CTL-induction by DCs. Moreover, h11c had a significantly higher affinity for canine DCs, and activated them as well as human and murine DCs. In the mouse models of either surface or visceral tumor, the h11c-treated DCs significantly suppressed tumor growth and elicited significantly higher survival. Moreover, the h11c-DCs significantly exerted to reduce monocytic MDSCs. Based on these results, h11c is expected to be a promising tool towards a successful clinical outcome. The COX2-I treatment, which was reported to inhibit the generation of MDSC [14], significantly enhanced the DC plus IFNγ treatment in growth suppression of the mouse surface tumor, and elicit significantly higher survival in the mouse visceral tumor model by adding to the h11c-DC plus IFNγ treatment. The combination of h11c-DC, IFNγ and COX2-I treatment significantly reduced not only monocytic MDSCs but also granulocytic MDSCs. In addition, the combination of treatments significantly enhanced activity of DCs for eliciting immune responses against tumors. Furthermore, the combination induced immune status in tumors to facilitate to remove tumor cells by immune cells. Taking these results together, the combination of h11c-DC, IFNγ and COX2-I treatment was expected to be highly effective in enhancing clinical outcome against both surface and visceral tumors.
In the dog clinical study, using the combination treatment, we challenged to treat surface tumors of larger size (the case no. 1) or systemically diffused (the case no. 4), or metastatic tumors (the cases no. 2 and 3). In three of the four cases the combined treatment significantly suppressed the growth of tumors. In these cases TRTCs in PB increased and inversely almost correlate with the tumor volume. It is noteworthy that not only MDSC but also Tregs decreased in correlation with the tumor volume, although the underlying mechanism was not clear. In case no. 1 the tumor volume had increased by day 49, regardless of the treatment, but decreased rapidly thereafter. The TRTCs increased during this period, but Tregs and MDSCs in the PB decreased. From these results it appears that the early growth of the tumor is due to partially inflammatory swelling caused by immune responses against the tumor. On the other hand, the combined the combined treatment permitted the growth of lung metastatic tumors (the case no, 2). However, the TRTCs level markedly increased in the beginning of the treatment. Although gradually decreased with time, it kept higher than that before treatment by the end period. The level of the Tregs and MDSCs decreased at the beginning, but increased gradually thereafter, and reached higher than that before treatment. Based on these results, it is therefore suggested that potent control of these suppressor cells is a key to succeed in treatment against visceral metastatic tumors. Hence, to control these suppressor cells in the dog, we are planning to use additional reagents such as sunitinib [34]. Overall, these results indicate