Arsenic trioxide synergizes with B7H3‐mediated immunotherapy to eradicate hepatocellular carcinomas

Arsenic trioxide (As2O3), a valuable anticancer drug for the treatment of acute promyelocytic leukemia, may also have therapeutic potential for the treatment of solid tumors. However, its therapeutic efficacy against solid tumors is lacking even at high dosages. Other therapeutic strategies are required to enhance the efficacy of As2O3 against solid tumors such as hepatocellular carcinoma (HCC), which is refractory to chemotherapy. B7H3, a new member of the B7 family, has been shown to induce antitumor immunity. Intratumoral injection of B7H3 plasmids eradicates small EL‐4 lymphomas, but monotherapy is ineffective against large tumors. Here we investigated whether As2O3 would synergize with B7H3 immunotherapy to combat HCC. Large subcutaneous H22 HCCs (0.7–0.8 cm in diameter) established in BALB/c mice were rapidly and completely eradicated when intratumoral administration of As2O3 was preceded by in situ gene transfer of B7H3. In contrast, neither As2O3 nor B7H3 monotherapy was effective. The antitumor activity of As2O3 was attributed to increased tumor‐cell apoptosis, perhaps as a result of direct cytotoxicity as well as decreased tumor angiogenesis. Combination therapy generated potent systemic antitumor immunity mediated by CD8+ and NK cells that was effective in combating a systemic challenge of 1 × 107 parental H22 cells. It led to the simultaneous and complete regression of multiple distant tumor nodules, concomitant with increased levels of serum IFN‐γ and cytotoxic T lymphocyte (CTL) activity. In conclusion, combining B7H3‐mediated immunotherapy with As2O3 warrants investigation as a therapeutic strategy to combat HCC, and other malignancies. © 2005 Wiley‐Liss, Inc.

Arsenic agents have long been used as anticancer drugs in traditional Chinese medicine. 1 In the early 1970s the first clinical trial with an arsenic-containing remedy (main ingredient: arsenic trioxide, As 2 O 3 ) was carried out at Harbin Medical University in China for the treatment of acute promyelocytic leukemia (APL). The trial was very encouraging with two-thirds of patients experiencing complete remission. Nine of 32 patients survived for more than 10 years. 2 As 2 O 3 has subsequently been widely employed to treat APL. It even has therapeutic benefits for patients who are resistant to conventional chemotherapies. 3,4 Mechanisms that might explain the antitumor cytotoxicity of As 2 O 3 include its ability to induce cellular differentiation, tumor apoptosis, the degradation of specific APL transcripts and inhibition of tumor-cell growth by modulating redox balance and/or mitochondrial membrane potential. [5][6][7][8] The therapeutic potential of As 2 O 3 extends to non-APL acute myeloid leukemia, myeloma and chronic myeloid leukemia, and a variety of solid tumors. [9][10][11][12][13][14][15] However, the antitumor effects of As 2 O 3 against solid tumors were not as effective as against APL.
We previously reported a newly described member of the B7 family, B7H3, which could be employed to induce antitumor immunity. 16 Intratumoral gene transfer of mouse B7H3 was very effective against small EL-4 tumors (<0.3 cm in diameter), causing their complete regression in 50% of cases, but was not effective against larger tumors (>0.3 cm in diameter). In searching for ways to more effectively harness and strengthen the antitumor activity of cell adhesion molecule (CAM)-mediated immunotherapy, we previously demonstrated that immunogene therapy employing T-cell costimulators could be vastly improved by simultaneously targeting the tumor vasculature. CAM-mediated immunotherapy was combined with intratumoral injection of angiostatin 17 and antisense hypoxia-inducible-factor 1a 18 plasmids, and systemic delivery of the anticancer drug, 5,6-dimethylxanthenone-4acetic acid (DMXAA). 19 Combination therapy overcame tumor immune-resistance, and led to the complete and rapid eradication of large tumor burdens. More recently, we demonstrated that adeno-associated viral (AAV)-mediated delivery of angiostatin in combination with vaccination with tumor cells engineered to express B7.1 eradicated tumors disseminated to the liver. 20 Hepatocellular carcinoma (HCC) has been ranked as the second most common cause of cancer mortality in China since the 1990s. 21 Liver cancer has a very poor prognosis with less than a 6% five-year survival rate, and there are no effective drugs for treatment. A 5-year clinical trial carried out by the Department of General Surgery, The First Clinical College of Harbin Medical University, Harbin, revealed that As 2 O 3 is one of the most effective chemotherapeutic drugs ever used alone to treat HCC patients. 22 A recent study suggests that immunotherapy can reduce the risk of recurrence of HCC by 41%. 23 Here we test the premise that combining As 2 O 3 with B7H3-mediated immunotherapy can combat HCC and generate effective antitumor immunity. The rationale is that As 2 O 3 may aid B7H3-mediated antitumor immunity by inhibiting tumor growth, and render the tumor more susceptible to immunotherapy. In turn, B7H3 immunotherapy might enhance the efficacy of As 2 O 3, and allow for a reduction of the dosage, thereby decreasing the toxicity of treatment.
of Heilongjiang Provincial Tumor Hospital. The H22 cells were maintained by successive transplantation into the abdominal cavity of BALB/c mice. The cancerous ascites were collected, washed with PBS, stained with Trypan blue and counted by microscopy. An As 2 O 3 solution was purchased from Yida Pharmaceutical Co. Ltd, Harbin Medical University, China. The rabbit anti-FLAG and anti-CD4 (clone Gk1.5) antibodies were purchased from Sigma (MO) and Chemicon International, Inc., respectively. The anti-CD8 (clone 53-6.7), anti-NK (clone PK136) and anti-CD31 (MEC13.3) antibodies were purchased from Pharmingen, CA.

Plasmid/PVP formulation
The B7H3-pcDNA3.1 plasmid containing a 951 bp cDNA encoding full-length mouse B7H3, and Flag-B7H3-pcDNA3.1 containing cDNA encoding a FLAG tag (DYKDDDDK) fused to the N-terminal sequence of mouse B7H3 which was used to detect expression in situ, have been described previously. 16 PVP (plasdone C-30, M r 50,000) was kindly supplied by Alchemy Chemicals Ltd., Auckland, New Zealand. Purified plasmids were formulated with 5% PVP to generate a plasmid/PVP formulation with a concentration of 1 mg DNA/ml, as described previously. 24,25 Tumor implantation and treatment All surgical procedures and care administered to the animals were in accordance with institutional guidelines. All experiments included 6 mice per treatment group, and each experiment was repeated at least once.
Single tumor model. Tumors were established by subcutaneous injection of 1 3 10 6 H22 HCC cells into a site in the left flank of BALB/c mice. The growth of tumors was determined by measuring 2 perpendicular diameters. Animals were randomly assigned to treatment when tumors reached around 0.7-0.8 cm in diameter after 14 days. The B7H3 plasmid/PVP formulation (100 ll) containing 100 lg of B7H3 plasmid was intratumorally injected at multiple sites. As 2 O 3 (5 lg) in 100 ll of PBS 10 was injected into the tumor at multiple sites every 2 days, and mice were euthanized when tumors reached 2 cm in diameter. For combination treatment, the B7H3 plasmid/PVP formulation solution was intratumorally injected, and 48 hr later, As 2 O 3 was injected into the tumor as above, every 2 days until the tumors disappeared. Empty vector (pcDNA3.1)/PVP formulation solution served as a control for the B7H3 plasmid/PVP formulation, whereas PBS served as a control for As 2 O 3 . The tumors in the B7H3 monotherapy group were injected with B7H3 plasmid followed by an equivalent volume of PBS 48 hr later; tumors in the As 2 O 3 monotherapy group were injected with empty vector/PVP formulation solution followed by As 2 O 3 and tumors in the control group received empty vector/PVP complex followed by PBS. Mice whose tumors completely regressed were rechallenged 4 weeks after the disappearance of tumors by injecting 1 3 10 6 H22 cells subcutaneously into the opposing flank (right flank). Mice that resisted this challenge were given a second challenge of 1 3 10 7 H22 cells 2 weeks later.
Multiple tumor model. A primary tumor was established by subcutaneous injection of 1 3 10 6 H22 tumor cells into a site in the left flank of BALB/c mice, and the same number of H22 cells were injected at 4 different sites in the right flank, 5 days later, to represent secondary tumors. Primary tumors reached around 0.5 cm in diameter after about 12 days, at which stage the 4 tumors in the right flank were around 0.15-0.2 cm in diameter. B7H3 plasmid/ PVP formulation (100 ll) was injected into the primary tumor in the left flank, and 48 hr later, As 2 O 3 was given every 2 days as above, until tumors disappeared. The growth of all the tumors was monitored and tumor size determined by measuring 2 perpendicular diameters.

Immunohistochemical analysis
Tumor cryosections (10 lm) prepared 2 days after intratumoral injection of plasmids were treated with acetone, rinsed with PBS, blocked with 2% BSA for 2 hr and incubated overnight with the primary antibody. They were subsequently incubated for 30 min with appropriate secondary antibodies using the SABC kit (Boster Biological Technology Ltd., Wuhan, China), and developed with Sigma FAST DAB (3,3 0 -diaminobenzidine tetrahydrochloride) and CoCl 2 enhancer tablets (Sigma). Sections were counterstained with Mayer's hematoxylin, mounted and examined by microscopy.

Western blotting
The method for detecting the expression of proteins in tumors has been previously described. 17,18 Briefly, tissues were excised, minced and homogenized in protein lysate buffer. Debris was removed by centrifugation at 10,000g for 10 min at 4 o C. Protein samples (50 lg) were resolved on 10% polyacrylamide SDS gels, and electrophoretically transferred to nitrocellulose Hybond C extra membranes. The membranes were incubated with primary antibodies, and subsequently with horseradish peroxidase-conjugated secondary antibodies. They were developed by enhanced chemiluminescence (Amersham International, Buckingham, England), and exposure to x-ray film.

Assessment of tumor vascularity
The methodology to determine tumor vascularity has been described previously. 17,18 Briefly, 10 lm frozen tumor sections prepared from tumors 3 weeks after treatment were immunostained with the anti-CD31 antibody, as described above. Stained blood vessels were counted in 5 blindly chosen random fields (0.155 mm 2 ) at 403 magnification, and the mean of the highest 3 counts was calculated.

In situ detection of apoptotic cells
This method has been described previously. 17,18 Briefly, serial sections of 6 lm thickness were prepared from tumors 3 weeks following treatment, stained with the TUNEL agent (Boehringer Mannheim, Germany) and examined by fluorescence microscopy. Adjacent sections were counterstained with haematoxylin and eosin. The total number of apoptotic cells in 10 randomly selected fields was counted. The apoptosis index (AI) was calculated as the percentage of positive staining cells, namely apoptosis index (AI) 5 number of apoptotic cells 3 100/total number of nucleated cells.

Detection of IFN-c by ELISA
Serum samples and splenocytes were harvested from mice, 4 days after treatment. The splenocytes were cultured in RPMI1640 supplemented with 10% FCS at 37 o C and 5% CO 2 for 48 hr, and the culture supernatant was collected. The levels of IFN-g in both serum and supernatant were measured with an ELISA kit (Boster Biological Technology Ltd., Wuhan, China).

Cytotoxicity assays
Splenocytes were harvested from mice, 10 days after treatment, and incubated at 37 o C with H22 parental target cells in graded E: T ratios in 96-well round-bottom plates, supplemented with recombinant interleukin-2 at a final concentration of 1,000 U/ml. After a 4 hr incubation, 50 ll of supernatant was collected, and lysis was measured using a Cyto Tox 96 Assay kit (Promega, Madison, WI). Background controls for nonspecific target and effector cell lysis were included. After background subtraction, the percentage of cell lysis was calculated using the formula: 100 3 (experimental2spontaneous effector2spontaneous target/ maximum target2spontaneous target). For antibody-mediated depletion of leukocyte subsets, splenocytes were incubated with specific antibody at a concentration of 2 lg of antibody per 10 6 cells for 60 min before they were mixed with H22 parental target cells in 100:1 ratios, and splenocyte cytotoxicity was measured as above.

Statistical analysis
Results were expressed as mean values 6 standard deviation (SD), and a Student's t test was used for evaluating statistical significance. A value of less than 0.05 (p < 0.05) was used for statistical significance.

Results
Intratumoral gene transfer of a plasmid encoding mouse B7H3 results in intense in situ transgene expression A Flag tag was fused to the N-terminus of mouse B7H3 so as to detect expression of mouse B7H3 plasmids injected directly into H22 tumors in situ. H22 tumors that had been established 14 days earlier by subcutaneous injection of H22 cells were injected with 100 lg of Flag-B7H3/pcDNA3.1 expression plasmid and sectioned 2 days following gene transfer. Representative photographs reveal expression of Flag-tagged B7H3 throughout tumors, whereas control sections from vector-only-treated tumors were not stained with the anti-Flag antibody (Fig. 1a). Western blot analysis of homogenates of tumors prepared 2 days after gene transfer confirmed that the 45 kDa Flag-tagged B7H3 protein was expressed in situ (Fig. 1b, right lane). As expected, control homogenates of vectoronly-treated tumors were not stained with the anti-Flag tag antibody (Fig. 1b, left lane).
Arsenic trioxide synergizes with B7H3 immunotherapy to eradicate hepatomas It was possible that As 2 O 3 might impair B7H3-mediated antitumor immunity, as tumor cells dying in response to As 2 O 3 would not be able to adequately express the transgene. Hence established HCCs were first injected with B7H3 plasmid to stimulate antitumor immunity, and then, As 2 O 3 was intratumorally injected 48 hr later. Remarkably, tumors rapidly diminished in response to the combination of B7H3 and As 2 O 3 , accompanied by massive necrosis, such that by the third week of treatment, tumors had completely disappeared leaving perfectly healed skin. In contrast, monotherapy with either B7H3 plasmid or As 2 O 3 failed to eradicate tumors, but nevertheless each agent inhibited the growth of tumors for 1-3 weeks. As 2 O 3 monotherapy was more effective than B7H3 monotherapy in inhibiting tumor growth. The tumors of control mice treated with either empty vector or PBS grew rapidly, reaching 2 cm in diameter within 2 weeks (Fig. 2).

Combinational therapy generates potent antitumor immunity
Animals cured of their tumors by combination therapy completely rejected the challenge of 1 3 10 6 parental H22 tumor cells, and a subsequent more substantial burden of 1 3 10 7 H22 cells (Fig. 2). The level of IFN-g detected in the sera of mice 4 days after treatment of tumors with the As 2 O 3 and B7H3 combination was significantly increased compared with the levels in the sera of mice treated by As 2 O 3 monotherapy, or with the control agents PBS and empty pcDNA3.1 vector. There was no significant difference in the levels of IFN-g in the sera of mice treated with combinational therapy vs. those treated by B7H3 monotherapy, indicating that the production of IFN-g can be attributed to B7H3 gene transfer (Fig. 3a). The level of IFN-g detected in supernatants from cultured splenocytes harvested from mice 4 days after treatment showed a similar pattern to that of sera (Fig. 3b). Moreover, the antitumor cytotoxic activity of splenocytes obtained from mice 10 days after treatment with the combination of As 2 O 3 and B7H3 was significantly augmented, compared with animals treated with either As 2 O 3 monotherapy or PBS or empty pcDNA3.1 vector. Once again, there was no significant difference in the antitumor cytotoxic activity of splenocytes generated by mice treated with combinational therapy vs. those with B7H3 monotherapy, indicating that B7H3 gene transfer was responsible for the generation of cytotoxicity (Fig. 3c). These results suggested that As 2 O 3 did not participate in the generation of antitumor immunity, but neither did it impair antitumor immunity. CD4 1 and CD8 1 lymphocytes and NK cells were individually depleted by specific antibodies in the assay, which measured the cytotoxic activity of splenocytes in order to define the effector-cell types responsible for antitumor activity. As shown in Figure 3d, the antitumor cytotoxic activity of splenocytes was significantly reduced by 43% (p < 0.01) upon depletion of CD81 T cells, and by 37% (p < 0.01) upon deletion of NK cells, compared to controls in respect of splenocytes derived from mice treated by B7H3 monotherapy and the combi-  For combination therapy they were injected with a B7H3 expression plasmid, followed by intratumoral delivery of As 2 O 3 48 hr later and subsequently every 2 days. Empty plasmid (pcDNA3.1) and PBS served as controls. Cured mice were rechallenged by subcutaneous injection of 1 3 10 6 H22 cells 4 weeks after the disappearance of tumors (denoted by 1 vertical arrow). This challenge was subsequently followed 2 weeks later by an even larger challenge of 1 3 10 7 H22 cells (denoted by 2 vertical arrows). Rechallenged mice developed no tumors for the 8 weeks they were monitored (denoted by a star). Mice were euthanized when tumors reached more than 2 cm in diameter.
nation of B7H3 plus As 2 O 3 . In contrast, depletion of CD41 cells only slightly affected the cytotoxic activity of splenocytes (p > 0.05) (Fig. 3d). The results are consistent with our previous report that B7H3 mediated antitumor immunity largely depends on CD8 1 T cells and NK cells. 16 Combination therapy targeted to a single tumor generates antitumor immunity capable of simultaneously eradicating multiple distant tumor nodules The ultimate objective of cancer immunotherapy is to generate potent systemic antitumor immunity capable of simultaneously eradicating multiple tumor foci wherever they are located in the body and not to simply cause the destruction of a single treated tumor nodule. As reported above, mice cured by combination therapy completely rejected a substantial rechallenge of parental tumor cells, indicating that potent systemic antitumor immunity had been induced. A multiple tumor model was developed, in which a large tumor was established in 1 flank, and 4 small tumors in the opposing flank so as to provide a clinically relevant model for testing the therapeutic efficacy of As 2 O 3 and B7H3 combination therapy. The B7H3 expression plasmid was injected into the large tumor, followed 48 hr later and every 2 days by injection of As 2 O 3 until mice were cured of the primary tumor. The 4 smaller tumors were left untreated. Surprisingly, all 5 tumors completely regressed. The large tumor completely disappeared by the four-teenth day and the 4 smaller tumors were completely eradicated 1 week later (Fig. 4a). In contrast, all 5 tumors grew largely unchecked in mice treated by either B7H3, or As 2 O 3 monotherapy, and in control mice.
The distant tumor nodules were immunostained with antibodies against specific leukocyte subsets so as to determine whether they had been subjected to an immune attack. The multiple distant tumor nodules from the control group contained few CD8 1 and CD4 1 T cells, and sparse numbers of NK cells (Fig. 4b). In contrast, highly elevated numbers of CD8 1 T cells and NK cells, and conversely small numbers of CD4 1 T cells infiltrated the tumors of mice treated with the combination of B7H3 and As 2 O 3 (Fig.  4b).
The results indicate that B7H3-mediated antitumor immunity, which relies largely on CD8 1 T cells and NK cells, 16 is systemic since it is able to eradicate tumors wherever they are located in the body.

As 2 O 3 suppresses tumor growth by inhibiting tumor angiogenesis and inducing tumor apoptosis, but does not induce antitumor immunity
Nodular tumors established in the left flank of mice were removed 3 weeks after intratumoral injection of As 2 O 3 or PBS, sectioned and stained with an anti-CD31 antibody (Fig. 5a vs. 5b). As 2 O 3 therapy resulted in a statistically significant (p < 0.01) 40% FIGURE 3 -B7H3 immunotherapy increases the cytotoxic activity of splenocytes and secretion of IFN-g. B7H3 gene transfer increases the levels of IFN-g in both serum (a) and supernatant from splenocyte cultures (b). Serum samples and splenocytes were harvested 4 days after mice were treated as indicated. The splenocytes were cultured for 48 hr, and the culture supernatant collected. The levels of IFN-g were measured by ELISA. (c) B7H3 immunotherapy increases the antitumor cytotoxic activity of splenocytes. Splenocytes were prepared 10 days after the mice were treated as indicated, and tested for cytotoxic activity against parental H22 cells. The percentage cytotoxicity is plotted against various effectors to target (E:T) ratios. Significant differences (p < 0.01) between the As 2 O 3 1 B7H3, and B7H3 treatment groups, and the As 2 O 3 , pcDNA3.1 and PBS control groups are denoted by stars. (d) B7H3-stimulated antitumor immunity is largely mediated by CD8 1 T cells and NK cells. Splenocytes were prepared 10 days after mice were treated as indicated, and the contribution of specific leukocyte subsets to the cytotoxic activity displayed by splenocytes in response to each treatment was examined by antibody blockade. Specific antibodies against CD4 1 and CD8 1 lymphocytes and NK cells, or PBS as a control, were preincubated with splenocytes before the splenocytes were mixed with target cancer cells at the effector to target ratio of 100:1. Significant differences (p < 0.01) between antibody blockade of CD8 1 T cells and NK cells vs. the PBS control group are denoted by stars. reduction in tumor vessel density compared with mock treatment with PBS (Fig. 5e). As 2 O 3 has previously been demonstrated to induce cell apoptosis in many different types of tumors. Therefore, we examined whether HCCs underwent programmed cell death in response to As 2 O 3 as measured by in situ labeling of fragmented DNA using the TUNEL method. Small numbers of apoptotic cells were de-   a and b, 340 magnification), and by TUNEL analysis of apoptotic cells (c and d; green fluorescent cells showing condensed fragmented nuclei). Tumor blood vessels stained with the anti-CD31 mAb were counted in blindly chosen random fields to record mean vessel density per high power field (0.155 mm 2 ). n, number of tumors assessed. A significant difference in mean vessel counts between tumors treated with As 2 O 3 vs. PBS (p < 0.01) is denoted by an asterisk (e). TUNEL-positive cells were counted to record the AI, which was significantly (p < 0.01, denoted by asterisk) different between tumors treated with As 2 O 3 , and PBS. n, numbers of tumors assessed ( f ). tected in tumors treated with PBS (Fig. 5c), whereas tumor cell apoptosis was tripled following As 2 O 3 treatment (Fig. 5d). Adjacent sections were stained with haematoxylin/eosin, and the AI was calculated. The AI for As 2 O 3 -treated tumors was significantly (p < 0.01) higher than that for PBS-treated tumors (Fig. 5f).
Sections prepared from As 2 O 3 -treated tumors were immunostained with specific antibodies against CD4 1 and CD8 1 lymphocytes and NK cells in order to determine whether repetitive injection of As 2 O 3 might have an effect on the recruitment and penetration of immune cells within the tumor. The results demonstrated that there was no increase in the infiltration of immune cells into tumors treated by intratumoral injection of As 2 O 3 , compared with control tumors (data not shown).

Discussion
This present study has demonstrated for the first time that the activity of As 2 O 3 against solid tumors can be dramatically enhanced by combining it with B7H3 immunotherapy. Intratumoral injection of As 2 O 3 inhibited tumor angiogenesis and led to increased tumor-cell apoptosis. The combination of As 2 O 3 and B7H3 immunotherapy completely eradicated large subcutaneous HCCs, and generated potent and memorized antitumor immunity as evidenced by protection against subsequent challenge of cured mice with a heavy burden of parental tumor cells. The systemic antitumor immunity generated by treating a single primary tumor was able to simultaneously eradicate multiple distant tumor nodules that were left untreated. The antitumor immunity directed against distant tumor nodules was largely dependent on the presence of CD8 1 T and NK cells. The outcome was more successful than that achieved by treating HCC with As 2 O 3 in combination with the chemotherapeutic agents cisplatin and doxorubucin. 26 Following the discovery of As 2 O 3 as a new and promising treatment for various types of leukemia, particularly APL, a large number of studies have investigated the use of As 2 O 3 in the treatment of solid cancers, including neuroblastoma, 27 head and neck, 28 gastric, 14 transitional cell, 15 prostate 9 and renal cell carcinoma, 29 esophageal, 10 prostate, 11 colorectal 12 and hepatocellular 13,26 cancers. However, the results achieved were not comparable with the previous success obtained with APL. As 2 O 3 was found to induce the apoptosis of various cancer cell lines, including human hepatocarcinoma cell lines, 13 and inhibited their growth in vitro, but the concentrations of As 2 O 3 required were higher than those used against hematologic malignancies and not clinically achievable without the risk of As 2 O 3 -mediated side effects. Consequently, a search for agents suited to increase the efficacy of As 2 O 3 against less sensitive solid tumors was initiated, 30-32 based on the results achieved with APL. 33 Strategies to reduce its toxicity, including local delivery, were also devised. 22 Cellular apoptosis induced by As 2 O 3 was shown to be caspaseindependent, and involved reactive oxygen species (ROS)-mediated activation of poly-(ADP-ribose)-polymerase-1 and translocation of apoptosis-inducing factor from mitochondria to the nucleus. 34 However, the As 2 O 3 -induced apoptosis of a neuroblastoma cell line was shown to be mediated via the activation of caspase 3 in vitro. 27 Paradoxically, the expression of the 2 apoptosisrelated proteins Bcl-2 and Bax was upregulated in the human HCC cell line BEL-7402 cultured in vitro with As 2 O 3 . 35 Further, Bcl-2 was upregulated in an established rat hepatocarcinoma model in response to As 2 O 3 . 36 In another report, the apoptosis of leukemia U937 cells induced by As 2 O 3 was dependent on the activation of p38, inactivation of ERK and the Ca 21 -dependent production of superoxide. 37 The anticancer activity of As 2 O 3 is in part related to its effects on mitochondria, 38,39 and the generation of nitric oxide and ROS. 40,41 Accordingly, As 2 O 3 exerts antitumor effects as a result of its highly cytotoxic nature. Perhaps not surprising therefore, systemically administered As 2 O 3 has undesirable side-effects. 42 Intraperitoneally administered As 2 O 3 has binding affinity for liver and kidneys, and causes damage to both organs. 43 Thus, systemically delivered As 2 O 3 has high toxicity and low therapeutic efficacy against solid tumors. Intratumoral injection strategies have been developed in animal models to increase the localization of drugs in the tumor, and to reduce their uptake by other organs, and thus improve the tumor/normal tissue uptake ratios. This strategy permits higher and more frequent doses of an anticancer drug for use in tumor therapy. 44 Injection of As 2 O 3 into transplanted esophageal carcinomas resulted in a higher concentration of As 2 O 3 in the tumors than in other organs, and efficiently suppressed tumor growth without systemic side-effects. 10 The strategy adopted in our study is a further advance in that it targets not only primary HCCs, but also targets secondary distant tumors due to the antitumor systemic immunity that develops. This study accords with a previous report in which the antiangiogenic drug DMXAA synergized with B7.1-mediated immunotherapy to combat EL-4 tumors by generating potent systemic antitumor immunity. 19 Conventional cytotoxic chemotherapeutic drugs affect the endothelium of the growing tumor vasculature, in addition to proliferating cancer cells and various types of normal cells, such as those of bone marrow. The antiangiogenic efficacy of some chemotherapeutic drugs seems to be optimized by administering comparatively low doses of drug-sometimes referred to as ''metronomic'' chemotherapy. 45 As 2 O 3 has been demonstrated to inhibit angiogenesis by interrupting the reciprocal stimulant loop between endothelial cells and leukemic cells, where endothelial cells release cytokines that stimulate leukemic cells to release factors such as VEGF. 46 It also directly inhibits the proliferation of endothelial cells. 47 In the present study, we also found that intratumorally injected As 2 O 3 exhibits antiangiogenic effects. The vascular status of the distant tumor nodules that had not directly received As 2 O 3 was not examined. Theoretically, intratumorally injected As 2 O 3 should leak into the circulation at a very low concentration. Roboz et al. reported that As 2 O 3 causes dose-dependent apoptosis of the endothelium, 46 indicating that As 2 O 3 needs to reach a certain level to exert an antiangiogenic effect on tumors not directly injected with As 2 O 3 .
B7H3 immunotherapy stimulates the generation of antitumor cytotoxic T cells, and selectively enhances IFN-g expression. IFN-g plays a critical role in the immune-mediated destruction of tumors when expressed within the tumor bed. 48 Its antitumor activity stems from its ability to control antigen processing and presentation, leukocyte trafficking and indirect tumor cytotoxicity. 48 B7H3 immunogene therapy activates both acquired and innate immunity, as it leads to NK cell and CD8 1 T-cell dependent killing of tumor cells, 16 where CD8 1 T cells are crucial to the eradication of tumors that express high levels of MHC Class I, and NK cells are critical to eradicating tumors that express low levels of MHC Class I. B7H3 was shown to costimulate a tumor-specific CD8 1 CTL response in a recent report. 49 It may also serve as an NK-cell receptor like other B7 family members, enabling NK-cell dependent killing. 50 It also induces modest amounts of TNF-a, which could be cytotoxic to tumor cells. As 2 O 3 may facilitate a B7H3-stimulated immune attack on large primary tumors by disrupting the integrity of the tumor so that it is more accessible to immune cells. Further, by preventing tumor cells from growing, it will inhibit the generation of immune escape variants. The B7H3mediated immune response appears sufficient to eradicate small distant tumor nodules that, unlike large primary tumors, have not developed sophisticated immune defense mechanisms to ward of an immune attack, and are more readily accessible to immune effector cells.
In conclusion, these results confirm that cancer therapy employing either B7H3 or As 2 O 3 alone is of limited therapeutic use against large HCCs. However, the efficacy of B7H3 immunotherapy and As 2 O 3 therapy can be harnessed by combining the 2 therapies, thereby obtaining a synergistic result that renders large tumors susceptible to immune attack. This approach warrants consideration for the treatment HCC.