Targeted chemotherapy against intraperitoneally disseminated colon carcinoma using a cationized gelatin–conjugated HVJ envelope vector Free

The hemagglutinating virus of Japan envelope (HVJ-E; Sendai virus) vector derived from inactivated HVJ particles can be used to deliver DNA, proteins, and drugs into cells both in vitro and in vivo. HVJ-E is capable of delivering bleomycin, an anticancer drug, to various cancer cell lines, thereby producing 300-fold greater cytotoxicity than administration of bleomycin alone. In a mouse model of peritoneally disseminated colon cancer, we injected HVJ-E containing the luciferase gene into the peritoneum. Unexpectedly, luciferase gene expression was not observed within the tumor deposits or any organs. However, when combined with cationized gelatin (CG), CG-HVJ-E produced a high level of luciferase gene expression primarily within the tumor deposits. Forty-eight hours after introducing colon cancer cells into the peritoneum of experimental mice, CG-HVJ-E with or without bleomycin was injected into the abdominal cavity. Following six injections of bleomycin-incorporated CG-HVJ-E, complete responses were observed in 40% of the mice examined. All of the mice that received either empty CG-HVJ-E or bleomycin alone died within 40 days of having cancer cells introduced into the peritoneum. When the mice with complete responses were rechallenged with colon cancer cells from the same cell line, no tumors developed. Thus, CG-HVJ-E may suppress peritoneal dissemination of cancer. [Mol Cancer Ther 2006;5(4):1021–8]

Although improved surgical, chemotherapy, and radiotherapy methods have been developed to treat patients with cancer, it remains difficult to eradicate cancer completely. In particular, metastatic cancer involving multiple foci and microscopic cancers are hard to treat, and prevention of recurrence remains a difficult problem in cancer therapy (1). Antitumor immunotherapy holds great promise; however, a number of obstacles have been encountered in the course of its development (2, 3). Although immunotherapy can reduce, delay, and sometimes prevent tumor recurrence, a number of tumors progress after developing mechanisms to avoid recognition and elimination by the immune system (4, 5). A number of studies suggest that both tumor regression and induction of antitumor immunity are indispensable for complete tumor eradication. For this reason, immunotherapy is often combined with other therapies, such as surgical resection, radiotherapy, and chemotherapy (6, 7). Drug delivery systems have the potential to overcome the problem of escape from immune recognition and to increase the efficiency of killing of unresectable tumors (8).

One important issue in drug delivery vectors is how to cross the cell membrane to introduce therapeutic molecules (9). There are several ways to bypass the cell membrane. Liposome-mediated delivery results in drug uptake by endocytosis or phagocytosis. However, this requires rapid penetration of the endosome or phagosome membrane by the foreign molecules before degradation (10). Viruses can enter cells; thus, viral vectors are capable of penetrating cell membranes (11). Adenovirus can escape from the endosome by disrupting the endosomal membrane with penton fibers (12). This capability has been used to enhance the efficiency of gene transfer using transferrin-polylysine-DNA complexes (13). Other viruses fuse with the cell membrane, thereby introducing their genomes into the cytoplasm. There are two different mechanisms of virus-cell fusion: pH dependent and pH independent. Influenza virus (14), Semliki Forest virus (15), and vesicular stomatitis virus (16) exhibit pH-dependent fusion, whereas hemagglutinating virus of Japan (HVJ, Sendai virus; ref. 17) and retrovirus (18) fuse with the cell membrane at both acidic and neutral pH. Viral fusion proteins have been identified, and synthetic vectors expressing viral fusion proteins can transfer foreign genes efficiently into the cytoplasm (19). We attempted to use the fusion capabilities of Sendai virus for drug delivery. First, we developed HVJ-liposomes, in which drug-incorporated liposomes were fused with inactivated Sendai virus (20). HVJ-liposomes are capable of delivering genes and synthetic oligodeoxynucleotides into cells in various animal models (21). To further simplify drug delivery using HVJ-liposomes, we tried to develop a nonviral delivery system. Finally, we developed an HVJ envelope (HVJ-E) vector (22, 23). Using this system, macromolecules, such as plasmid DNA, RNA, synthetic oligonucleotides, proteins, and peptides, get incorporated into inactivated HVJ particles by treatment with a mild detergent and centrifugation, after which they can be delivered to cells in vitro and in vivo. To enhance drug delivery, we combined HVJ-E with cationized gelatin (CG; ref. 24). CG-conjugated HVJ-E (CG-HVJ-E) was still capable of fusion. In CT-26 cells, CG-HVJ-E–mediated luciferase gene expression was ∼10 to 20 times greater than that achieved using HVJ-E without conjugation to a polymer. Furthermore, the stability of HVJ-E in fresh mouse serum was greatly enhanced by conjugation with CG.

Herein, we show that CG-HVJ-E is a very effective vehicle for delivering bleomycin to tumor cells after peritoneal dissemination. Multiple injections of CG-HVJ-E/bleomycin produced complete responses in 40% of the mice examined, all of which were further resistant to tumor rechallenge.

HVJ was amplified in chorioallantoic fluid from 10- to 14-day-old chick eggs, after which it was purified by centrifugation and inactivated by UV irradiation (99 mJ/cm²), as previously described (22). Inactivated virus cannot replicate, but its capacity for viral fusion remains intact.

Human cancer cells and mouse colon cancer CT-26 cells were maintained in DMEM supplemented with 10% fetal bovine serum and antibiotics.

Inactivated HVJ (1.8 × 10¹⁰ particles) was mixed with 60 μL of 5 or 40 mg/mL bleomycin and 2 μL of 3% Triton X-100, as previously described (22). After 15 minutes of incubation at 4°C, the suspension was washed with 500 μL of PBS (pH 7.4) and centrifuged (18,500 × g) for 15 minutes at 4°C. After this, the suspension was again washed twice with 500 μL of PBS to completely remove the detergent and any unincorporated bleomycin. After centrifugation, the HVJ-E/bleomycin was suspended in 180 μL of PBS.

Cationization of gelatin was done by introducing ethylene diamine into the carboxyl groups of low molecular weight gelatin (MW = 5,000), as previously described (24), and the mole-to-mole ratio of amino groups to carboxyl groups within the gelatin was 48.7. A 5-mg amount of CG was added to 300 μL of PBS containing 3 × 10¹⁰ particles of HVJ-E vector containing bleomycin. The solution was mixed by tapping several times. After this, the solution was incubated on ice for 30 minutes, during which it formed CG-conjugated HVJ-E vector containing bleomycin, which was purified by centrifugation. CG-conjugated (MW = 100,000), dextran-conjugated, and pullulan-conjugated HVJ-E vectors were prepared by mixing 5 mg of these polymers (25–27) with 3 × 10¹⁰ particles of HVJ-E vector containing the luciferase gene. For in vivo use, 1.5 × 10¹⁰ particles of polymer-conjugated HVJ-E vector was suspended in 500 μL of PBS. CG/bleomycin without HVJ-E was prepared by mixing 5 mg CG with 100 μL of 5 or 40 mg/mL bleomycin followed by centrifugation. CG-luciferase gene without HVJ-E vector was also prepared according to the previous method (25).

A total of eight HVJ-E/bleomycin suspensions were prepared using eight different concentrations of bleomycin (0, 5, 10, 20, 30, 40, 50, and 100 mg/mL). Each HVJ-E/bleomycin suspension (200 μL) was dissolved in an equal volume of chloroform. After being vortexed and centrifuged (18,500 × g for 15 minutes at 4°C), the aqueous layer was recovered and added to 800 μL of PBS. The bleomycin content was assessed by high-performance liquid chromatography. Peaks representing bleomycin A2 and B2 at 245 nm were added, after which the concentration of bleomycin was determined from the standard curve. The efficacy of bleomycin inclusion into this vector was quantitatively measured with high-performance liquid chromatography. The amount of bleomycin in 1,000 HAU of HVJ-E/bleomycin was 0.18 μg when 5 mg/mL bleomycin was used, 0.41 μg with 10 mg/mL bleomycin, 0.84 μg with 20 mg/mL bleomycin, 1.12 μg with 30 mg/mL bleomycin, 1.32 μg with 40 mg/mL bleomycin, 1.34 μg with 50 mg/mL bleomycin, and 1.52 μg with 100 mg/mL bleomycin. Thus, the amount of bleomycin incorporated into the HVJ-E vector increased in proportion to the bleomycin concentration up to 40 mg/mL of bleomycin, at which a plateau was reached. Therefore, 40 mg/mL of bleomycin was used to prepare the HVJ-E/bleomycin used for the in vivo experiments.

To perform in vitro transfection of HVJ-E/bleomycin, 5 × 10⁴ CT-26 cells were seeded into six-well plates 1 day before transfection. Cells were maintained in DMEM (Nakalai Tesque, Kyoto, Japan) supplemented with 10% fetal bovine serum, penicillin (50 units/mL), and streptomycin (50 μg/mL) and incubated at 37°C in 5% CO₂. Tissue culture medium and supplements were purchased from Nakarai Tesque. HVJ-E/bleomycin was prepared with 5 or 40 mg/mL bleomycin. A 5-μL aliquot of 5 mg/mL protamine sulfate (Nakarai Tesque) and 500 μL of medium were added to 30 μL of HVJ-E/bleomycin (3 × 10⁹ particles). This HVJ-E/bleomycin solution was added to each well after removal of the cell culture medium. HVJ-E alone and five different concentrations of bleomycin alone (0, 1, 10, 25, and 100 μg/mL) were then added to some of the wells. After 30 minutes of incubation, the medium was changed to fresh medium. After 2 days of incubation, the cells were counted with a Coulter particle counter (Beckman Coulter, Fullerton, CA).

For the evaluation of fusion-mediated delivery, we used antiserum against the F protein of HVJ prepared in our laboratory by immunizing a rabbit with purified F protein (24). The concentration of anti-F antibodies in the antiserum was ∼30 μg/mL. Aliquots of antiserum were stored at −80°C. The antiserum was diluted with saline. HVJ-E (3 × 10⁹ particles) and CG-HVJ-E containing the luciferase gene were preincubated with diluted or undiluted antiserum (20 μL) for 30 minutes at 37°C. This mixture was then added to cultured cells. Preimmune rabbit serum was used as the control. Luciferase activity was measured 24 hours after transfection.

To evaluate the effect of endocytosis-mediated delivery, wortmannin (Sigma Chemical Co., St. Louis, MO), an inhibitor of phosphatidylinositol-3-kinase that inhibits endocytosis (28), was used. The reagent was dissolved in DMSO to a final concentration of 10 mmol/L, dispensed into 5-μL aliquots, and stored at −80°C. Before use, the aliquots were thawed and diluted in serum-free DMEM. Care was taken to shield the aliquots from light. Before transfection, cells were washed with serum-free DMEM and incubated with various concentrations of wortmannin for 15 minutes. The cells were then subjected to in vitro transfection, as described above.

All animal experiments were approved by the Animal Committee of Osaka University and conducted in a humane fashion according to their guidelines. Male BALB/c mice, 6 to 7 weeks of age, were obtained from Charles River Japan, Inc. (Yokohama, Japan). Mice were housed for 7 to 14 days and allowed ad libitum access to food and water. For tumor cell implantation, CT-26 cells were enzymatically detached from their culture flasks and counted. Viable cells (1.5 × 10⁶) were resuspended in 500 μL of PBS and injected into the peritoneal cavity of each mouse. Two days after tumor inoculation, 500 μL of CG-HVJ-E/bleomycin (1.5 × 10¹⁰ particles) or CG-bleomycin were injected six times every third day, and animal survival was monitored.

On day 115 after tumor inoculation, two surviving mice and three age-matched naive mice were rechallenged by i.d. injection of 5 × 10⁶ parental cells into both sides of their trunk. In addition, two other surviving mice from the same experiment were rechallenged by i.p. injection of 1.5 × 10⁶ parental cells.

The HVJ-E vector delivers molecules, such as anticancer drugs, into cells by membrane fusion, as illustrated in Fig. 1. We investigated the delivery mechanism using anti-F antibody or wortmannin. As shown in Supplementary Fig. S1A and B,⁴

undiluted anti-F antibody reduced the transfection efficiency of HVJ-E to 20% of that of the control group, whereas 100 nmol/L wortmannin did not significantly decrease transfection efficiency. These results suggest that drug delivery by HVJ-E vector occurs mainly by fusion. Therefore, agents delivered by the HVJ-E vector are released directly into the cytoplasm, bypassing endocytosis, the mechanism by which liposome-mediated and naked DNA-mediated delivery are achieved.

We attempted to incorporate anticancer drugs into the HVJ-E vector to enhance its cytotoxicity following fusion-mediated delivery. We used bleomycin as the anticancer agent to be incorporated into HVJ-E. Bleomycin has a marked antineoplastic effect but limited cell permeability (29). To evaluate the potential of the HVJ-E vector to effectively deliver bleomycin into cancer cells, we incorporated bleomycin into the vector and assessed its cytotoxicity against various cancer cell lines in vitro. We prepared HVJ-E/bleomycin using 5 or 40 mg/mL of bleomycin in the HVJ-E inclusion reaction and tested its cytotoxicity against CT-26 mouse colon adenocarcinoma cells. HVJ-E/bleomycin or bleomycin alone were incubated with cultured cells for 30 minutes, and the cells were further cultivated for 48 hours. Bleomycin alone was not particularly cytotoxic, killing only 60% of CT-26 cells at 100 μg/mL of BML. Although HVJ-E alone was not toxic to CT-26 cells, HVJ-E/bleomycin killed >90% of CT-26 cells in the culture (Fig. 2A). The bleomycin content of HVJ-E/bleomycin prepared using 5 and 40 mg/mL of bleomycin in the inclusion reaction was 0.36 and 2.54 μg/mL, respectively. Therefore, HVJ-E/bleomycin was 300-fold more cytotoxic than bleomycin alone. HVJ-E/bleomycin exhibited similar effects on a number of other human cancer cell lines (Fig. 2B). Although the cytotoxicity of 100 μg/mL bleomycin varied among different human cancer cell lines, HVJ-E/bleomycin caused a marked reduction in cell survival in all cell lines tested. In addition, treatment with HVJ-E/bleomycin reduced the proportion of cells in the G₀-G₁ phase of the cell cycle and increased the proportion of cells in the G₂-M phase (data not shown). These cell cycle changes are a characteristic of bleomycin treatment. Thus, it seems that the HVJ-E vector efficiently delivered bleomycin into cells by fusion with the plasma membrane.

A major challenge in cancer chemotherapy is finding a way to treat inoperable and invisible cancer lesions and hence the need for cancer-specific drug delivery vectors. Here, we attempted to treat peritoneal deposits of cancer cells using the HVJ-E vector system. First, we confirmed the presence of peritoneal tumor deposits 1 week after i.p. injection of CT-26 tumor cells (10⁷ cells). As previously reported (30), a number of tumor deposits are observed to develop within the i.p. cavity following the introduction of CT-26 tumor cells. In particular, metastasis to lymph nodes around the root of the mesentery are frequently detected in the i.p. cavity.

We used the luciferase gene as a reporter gene to measure gene expression. To evaluate the transfection efficiency of HVJ-E into peritoneally disseminated colon tumor cells, HVJ-E containing pcLuc plasmid was injected into the peritoneal cavities of mice 7 days after i.p. injection of CT-26 tumor cells. Twenty-four hours after injection of HVJ-E, the mice were killed, and all tumor deposits and nonaffected organs were removed and examined for luciferase activity. As shown in Fig. 3, luciferase gene expression was not detected in either the tumor deposits or other organs, such as the lung, liver, or spleen. No significant luciferase activity was detected in any tissue when either naked plasmid DNA encoding the luciferase gene or CG-conjugated luciferase gene was i.p. injected (data not shown).

Next, we combined HVJ-E with CG (MW = 5,000 and 100,000), dextran, and pullulan. When CG (MW = 5,000)-HVJ-E containing the luciferase gene was i.p. injected, a high level of expression of the luciferase gene was preferentially detected in the tumor deposits compared with other polymers, such as CG (MW = 100,000), dextran, and pullulan, with limited expression in the spleen and liver and no expression in lung (Supplementary Fig. S2).⁴

Then, we combined HVJ-E containing bleomycin with CG (CG-HVJ-E/bleomycin) and compared its cytotoxicity against cultured CT-26 cells with that of HVJ-E/bleomycin. As shown in Fig. 4, CG-HVJ-E/bleomycin killed CT-26 cells as efficiently as HVJ-E/bleomycin. No cytotoxicity was observed when CG was combined with bleomycin alone (CG/bleomycin). We then investigated the mechanism of CG-HVJ-E–mediated delivery using anti-F antibody or wortmannin. As shown in Supplementary Fig. S3A and B,⁴ both undiluted anti-F antibody and 100 nmol/L wortmannin significantly reduced the transfection efficiency of CG-HVJ-E. Thus, the delivery mechanism of CG-HVJ-E vector seems to depend on both membrane fusion and endocytotic uptake.

Next, we injected CG-HVJ-E/bleomycin into the peritoneal cavity 48 hours after the introduction of CT-26 cells. Three injections were given every third day. Although mouse survival was prolonged by administration of either CG-HVJ-E or CG-HVJ-E/bleomycin, compared with bleomycin alone, all of the mice died within 40 days (data not shown). Next, we evaluated the effects of i.p. injection of the vectors six times every third day. As shown in Fig. 5, complete responses were observed in 40% of the mice injected with CG-HVJ-E/bleomycin. The mice that received CG-HVJ-E without bleomycin showed prolonged survival, compared with those administered bleomycin alone, but still died within 50 days. The mice showing complete responses survived >90 days and experienced complete remission.

Next, we injected CT-26 cells into the surviving mice i.d. or i.p. Approximately 5 × 10⁶ CT-26 cells were i.d. injected into both sides of the mouse trunk. Although the tumor masses were observed to grow up to 3 to 4 mm in diameter within 5 days of the rechallenge in both control mice and those showing complete responses, tumor masses among the mice previously showing complete responses then underwent a rapid reduction in size within 11 days of the rechallenge (Fig. 6). However, tumor masses among the control mice continued to grow up to 8 to 10 mm in diameter within 11 days of the rechallenge (Fig. 6). In addition, ∼1.5 × 10⁶ CT-26 cells were also injected directly into the i.p. cavities of mice showing complete responses, as well as control mice. Within 11 days of the rechallenge, control mice were emaciated and had diarrhea, and a number of tumor deposits were observed in the peritoneal cavity, whereas the mice previously showing complete responses remained healthy with no tumor deposits observed within the peritoneal cavity (data not shown).

However, when Meth-A cells, which are fibrosarcoma cells derived from BALB/c mouse, were i.d. injected into the surviving mice, tumors arose in those mice as well as in naive mice (data not shown).

Thus, multiple i.p. injection of CG-HVJ-E/bleomycin not only inhibits the tumorigenesis of CT-26 cells following i.p. dissemination but also induces an effective and long-lasting anti-CT-26 memory in mice.

In this study, we showed the feasibility of using the HVJ-E vector for cancer treatment. One of the advantages of the HVJ-E vector is fusion-mediated drug delivery and thereby enhancing drug efficacy. Bleomycin is ∼1,500 Da and thus has limited permeability into cells and low bioavailability following oral, i.m., or i.v. administration (29). In the present study, in vivo, cell culture experiments showed enhanced cytotoxicity of bleomycin delivered by the HVJ-E vector against a variety of cancer cells. Fusion between the HVJ-E vector and cancer cell membranes might overcome the problem of limited bleomycin cell permeability. As shown in Supplementary Fig. S3,⁴ the delivery by CG-HVJ-E was mediated by both fusion and endocytosis. CG-HVJ-E may be more effective for drug delivery than HVJ-E alone because HVJ can fuse with the cell membrane under neutral conditions and also with the endosomal membrane under acidic conditions (17).

A disadvantage of the HVJ-E vector is rapid degradation in the presence of fresh serum, although the in vitro transfection efficiency of HVJ-E was not inhibited by culture medium containing 10% fetal bovine serum (22). However, we previously showed that CG-HVJ-E is remarkably stable in 50% fresh mouse serum. Although it is unproven that HVJ is degraded by complement lysis in mouse serum, as known to be the case for retrovirus and HIV (31, 32), the interaction between serum proteins and HVJ-E may contribute to reduced transfection efficacy of HVJ-E. Conjugation with CG seems to protect the surface molecules of HVJ-E from the detrimental effects of serum proteins (24). In this report, CG-HVJ-E showed a proclivity toward i.p. tumor deposits. In a previous report, we showed that HVJ-cationic liposomes also have a proclivity for tumor deposits when injected into peritoneal cavity (30). We speculate that HVJ-cationic liposomes might be absorbed into lymph vessels and fuse with tumor cells when they reach lymph nodes enlarged with tumor cells. We observed enhanced fusion of HVJ with tumor cells than normal cells.⁵

Therefore, it is likely that positively charged CG-HVJ-E (ζ potential = 11.30 mV; ref. 24) results in preferential delivery of bleomycin to tumor cells within the peritoneum.

Another advantage of the HVJ-E vector is its capacity for repeated injection. Gene transfer to mouse muscle was not inhibited by repeated injections (23). Similar results were obtained when HVJ-liposomes was repeatedly injected into rat liver. After repeated injections, the anti-HVJ antibodies generated in mice were not sufficient to neutralize HVJ-liposomes (33). Presumably, this is because fusion occurs more rapidly than recognition by neutralizing antibodies. Indeed, fusion between the HVJ-E vector and the cell membrane occurs in 10 seconds (23).

In this experiment, complete responses were observed in 40% of tumor-inoculated mice. Furthermore, these mice were resistant against tumor rechallenge. This result indicates that antitumor immunity was achieved in mice treated with CG-HVJ-E/bleomycin. Some reports suggest that antitumor immunity might be induced in mice following a variety of forms of treatment, including chemotherapy, radiotherapy, or gene therapy (34–36). On the other hand, it is well known that viruses and bacteria can induce robust and long-lasting immune responses through Toll-like receptors (37, 38). Some reports suggest that live HVJ infection induces the secretion of various cytokines, including IFN-α, IFN-β, IFN-γ, interleukin-6, interleukin-12, and tumor necrosis factor in dendritic cells, while also inducing a maturation of dendritic cells (39, 40). Moreover, inactivated HVJ can induce dendritic cell maturation as effectively as live HVJ (39, 41). We observed a retained capability of dendritic cell maturation by the HVJ-E vector, and i.t. injection of the HVJ-E vector into mice mildly enhanced IFN-γ and interleukin-5 production by antigen stimulated splenocytes.⁶

This explains why CG-HVJ-E prolonged mouse survival even without bleomycin. When rechallenged with tumor cells, the mice treated with CG-HVJ-E/bleomycin rejected the same CT-26 tumor not Meth-A. This suggests that tumor-specific cytotoxic T cells, not natural killer cells, were generated in those mice by CG-HVJ-E/bleomycin treatment. Therefore, CG-HVJ-E/bleomycin might induce antitumor immunity and via direct toxicity to tumor cells by bleomycin delivery. Given its dual function, the HVJ-E vector might be especially useful for tumor eradication. Currently, no drug delivery system has shown similar capabilities specific to the treatment of cancer.

Thus, the HVJ-E vector system shows potential for the treatment of cancer in the future.