Adeno-Associated Virus–Mediated Antiangiogenic Gene Therapy with Thrombospondin-1 Type 1 Repeats and Endostatin Free

Purpose: Recombinant adeno-associated virus (rAAV)-mediated antiangiogenic gene therapy offers a powerful strategy for cancer treatment, maintaining sustained levels of antiangiogenic factors with coincident enhanced therapeutic efficacy. We aimed to develop rAAV-mediated antiangiogenic gene therapy delivering endostatin and 3TSR, the antiangiogenic domain of thrombospondin-1.

Experimental Design: rAAV vectors were constructed to express endostatin (rAAV-endostatin) or 3TSR (rAAV-3TSR). The antiangiogenic efficacy of the vectors was characterized using a vascular endothelial growth factor (VEGF)-induced mouse ear angiogenesis model. To evaluate the antitumor effects of the vectors, immunodeficient mice were pretreated with rAAV-3TSR or rAAV-endostatin and received orthotopic implantation of cancer cells into the pancreas. To mimic clinical situations, mice bearing pancreatic tumors were treated with intratumoral injection of rAAV-3TSR or rAAV-endostatin.

Results: rAAV-mediated i.m. gene delivery resulted in expression of the transgene in skeletal muscle with inhibition of VEGF-induced angiogenesis at a distant site (the ear). Local delivery of the vectors into the mouse ear also inhibited VEGF-induced ear angiogenesis. Pretreatment of mice with i.m. or intrasplenic injection of rAAV-endostatin or rAAV-3TSR significantly inhibited tumor growth. A single intratumoral injection of each vector also significantly decreased the volume of large established pancreatic tumors. Tumor microvessel density was significantly decreased in each treatment group and was well correlated with tumor volume reduction. Greater antiangiogenic and antitumor effects were achieved when rAAV-3TSR and rAAV-endostatin were combined.

Conclusions: rAAV-mediated 3TSR and endostatin gene therapy showed both localized and systemic therapeutic effects against angiogenesis and tumor growth and may provide promise for patients with pancreatic cancer.

Tumor growth beyond 1 to 2 mm depends on angiogenesis, the formation of new blood vessels (1, 2). The homeostatic processes regulating tumor angiogenesis have become more thoroughly understood in recent years. In the tumor microenvironment, angiogenic stimulators outbalance angiogenic inhibitors, generating a proangiogenic response and an increased blood vessel density (3, 4), which in turn is related to decreased survival in cancer patients (5, 6). To restore the tightly regulated angiogenesis balance in the tumor microenvironment, antiangiogenic therapy can be aimed at up-regulation of antiangiogenic factors or down-regulation of proangiogenic factors or both.

Among endogenous angiogenic inhibitors, endostatin, angiostatin, and thrombospondin-1 (TSP-1) are the most important and have been studied extensively (1, 7–9). TSP-1 is a multifunctional extracellular matrix protein with pivotal roles in the regulation of vascular development and angiogenesis (7, 8). The discrete antiangiogenic domain of TSP-1, the TSP-1 type 1 repeats, designated 3TSR, is an attractive candidate for antiangiogenic treatment. We previously have shown the antiangiogenic and antitumor efficacy of recombinant 3TSR in several preclinical studies (10–13). A TSR-derived mimetic peptide, ABT-510, has been tested in clinical trials for patients with advanced cancers (14–16). Recombinant endostatin, a COOH-terminal proteolytic fragment of collagen XVIII, has also entered clinical trials for patients with solid tumors (17, 18).

Evidence suggests that a consistent level of angiogenic inhibitors might improve the therapeutic potency and efficacy of cancer treatment (19–23). We and others have shown that continuous delivery of angiogenic inhibitors, including 3TSR, endostatin, and angiostatin, was more effective than daily bolus injections (19–23), indicating the therapeutic benefits of sustained levels of angiogenic inhibitors. Gene therapy provides a way to achieve sustained delivery of antiangiogenic factors to the tumor site from a single or small number of treatments and to maintain the continuous production of angiogenic inhibitors for a set period of time. Gene therapy strategies directly targeting tumor cells with genes encoding prodrug-converting enzymes or cytokines/chemokines for oncolysis require high-efficiency transduction of cancer cells by gene vectors. In contrast, antiangiogenic gene therapy does not require high-efficient cancer cell transduction, and may target nontumor cells as well as tumor cells, using both tissues to provide stable platforms for expression of secreted proteins.

The potential of antiangiogenic gene therapy in cancer is currently being evaluated using both viral and nonviral vectors (24–27). Antiangiogenic vectors need to be capable of sustained, long-term expression without vector-associated toxicity or immunity. Compared with other gene therapy vectors, adeno-associated virus (AAV) vectors are highly advantageous for antiangiogenic gene therapy. AAV-mediated transgene persists in host cells primarily as stable episomes, thus resulting in prolonged but not permanent transgene expression, with low risk of insertional mutagenesis. AAVs are nonpathogenic vectors with a limited host immune response and have been used in humans with no adverse effects (28). AAV-expressing endostatin has been used in several preclinical cancer models showing long-term endostatin expression and significant protective effects against tumor growth (26, 29, 30).

In the present study, we engineered vectors derived from AAV2, the most commonly used serotype, to express either 3TSR or endostatin and characterized the antiangiogenic efficacy of these vectors in a vascular endothelial growth factor (VEGF)-induced mouse ear angiogenesis model. The antitumor effects of the viral vectors were then studied in a murine orthotopic pancreatic cancer model with direct relevance to the disease seen in humans. We find that AAV-mediated antiangiogenic gene therapy can be used to express transgenes in normal tissue, such as skeletal muscle and liver, to inhibit tumor growth at distant sites. Direct intratumoral delivery of the recombinant AAV (rAAV) vectors also significantly inhibited the growth of established pancreatic tumors. Furthermore, more significant antiangiogenic and antitumor effects were achieved when the AAV vectors expressing 3TSR and endostatin were used in combination.

Cell culture. Human pancreatic cancer cells AsPC-1 and MiaPaCa-2 (American Type Culture Collection) were grown in RPMI 1640 (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and penicillin-streptomycin. Cells were grown in 5% CO₂/95% air at 37°C in a humidified incubator.

rAAV construction and production. rAAV vectors expressing endostatin (rAAV-endostatin) or 3TSR (rAAV-3TSR) were constructed using methods described previously (31). For rAAV-endostatin, a fragment containing a 250-bp segment encoding the endogenous murine collagen XVIII signal sequence fused to a 600-bp cDNA fragment encoding the COOH-terminal noncollagenous domain (NC1) of the protein, identical to endostatin, was inserted into an AAV-based plasmid. A cDNA encoding the three type 1 repeats of TSP-1 was cloned as described previously (10) and subcloned into the same AAV plasmid to generate rAAV-3TSR. In each case, transcription of the inserted gene was under the control of the human cytomegalovirus immediate early promoter. Isogenic rAAV encoding green fluorescent protein (GFP) or LacZ was used as control in these studies. Figure 1A shows the structure of the rAAV vectors. The rAAV vectors were packaged in 293 cells and functional viral titers were estimated by real-time PCR as described previously (31).

In vitro transduction and Western blot for 3TSR. Pancreatic cancer cells were transduced with rAAV-3TSR as described previously (31). Fresh medium was added to transduced cells 48 h after transduction. Twenty-four hours later, the culture medium was collected and precipitated with ammonium sulfate. The 3TSR protein was detected via Western blotting with a customized chicken anti-TSR antibody (Aves Labs, Inc.). To produce chicken antibody against TSR, recombinant human TSR2 (the second TSP-1 type 1 repeat) protein was used as immunogen. Hens were injected with TSR2 protein on days 3, 24, 38, and 52, and immune eggs were collected 1 week after the last injection. Chicken IgY was purified from the immune eggs.

Tumor models. All animal work was done in accordance with federal, local, and institutional guidelines as described previously (12). Male severe combined immunodeficient mice (Taconic), 4 to 6 weeks of age, were used. The orthotopic pancreatic cancer model was made via surgically implanting a suspension of 1 × 10⁶ AsPC-1 cells into the body of the pancreas (12).

rAAV-mediated gene therapy. rAAV vectors were delivered via i.m., intrasplenic, or intratumoral injections. In mice receiving i.m. or intrasplenic rAAV delivery, rAAV particles (1 × 10¹¹) in 100 μL were injected. Four weeks after vector administration, each mouse received surgical implantation of cancer cells into the pancreas. The following groups were included: I, control, mice received rAAV-GFP i.m. injection; II, rAAV-3TSR i.m. injection; III, rAAV-3TSR intrasplenic injection; IV, rAAV-endostatin i.m. injection; and V, rAAV-3TSR plus rAAV-endostatin i.m. injection. Intrasplenic injection was done using a similar regimen as described above. For i.m. injection, viral vectors were injected into the quadriceps muscle of the hind limbs.

For intratumoral gene therapy, tumor cells were implanted into the pancreas and allowed to grow for 3 weeks. Tumor-bearing mice were then randomized and received single intratumoral gene transfer of 1 × 10¹¹ (in 100 μL) rAAV-GFP, rAAV-endostatin, or rAAV-3TSR. Five mice were used in each group. Mice in all groups were sacrificed and underwent necropsy 31 days after tumor cell implantation. Tumor volume was calculated as π/6 × length × width × height.

Mouse ear angiogenesis assay. The mouse ear angiogenesis assay was done using VEGF-expressing adenoviral vectors (Adeno-VEGF) as described by Nagy et al. (32). Briefly, Adeno-VEGF (1 × 10⁷ or 2 × 10⁷) in 10 μL was injected i.d. into the ear of male nu/nu mice (National Cancer Institute, Bethesda, MD) to induce angiogenesis. To test the antiangiogenic efficacy of the rAAV vectors, the ear angiogenesis assay was done in three experimental settings: I, a mixture of 2 × 10⁷ Adeno-VEGF and 1 × 10¹⁰ rAAV-3TSR (or rAAV-endostatin) in 10 μL was injected into the mouse ear; II, mice were pretreated with rAAV-3TSR or rAAV-endostatin (1 × 10¹⁰ in 10 μL) ear injection and received Adeno-VEGF ear injection in the same sites 2 weeks later; and III, mice were pretreated with rAAV-3TSR or rAAV-endostatin (1 × 10¹¹ in 100 μL) i.m. injection and received Adeno-VEGF ear injection 2, 4, 6, and 8 weeks later. rAAV-GFP was used as negative control. To study the combinatory effects of rAAV-3TSR and rAAV-endostatin, mice were pretreated with rAAV-3TSR and/or rAAV-endostatin (1 × 10⁹) ear injection, and Adeno-VEGF (1 × 10⁷) was injected into the same sites 2 weeks later. Five days after Adeno-VEGF injection, the ear pictures were captured with a dissection microscope, and the angiogenic area of the mouse ear was measured using IPLab software (Scanalytics, Inc.). Ear samples were collected, fixed in paraformaldehyde-glutaraldehyde, and processed for 1-μm Epon sections as described by Nagy et al. (33).

Immunohistochemistry. Immunohistochemistry was done as described previously (12) using anti-CD31 antibody (BD PharMingen), anti-endostatin antibody (kindly provided by Dr. Kashi Javaherian, Children's Hospital, Boston, MA), or anti-TSR antibody.

Image processing and quantification. Microscopic pictures were captured using a Spot digital camera mounted to a Nikon TE300 microscope. IPLab software was used for quantification of the images. Tumor microvessel density was quantified via calculating the percentage of total vascular area in a given ×20 visual field (0.584 × 0.438 mm²), 40 fields for each group. Endostatin staining was quantified by measuring the absorbance value of 10 visual fields (×20) for each group. To quantify angiogenesis and edema of the mouse ear 1-μm sections, the area of a 2-mm (from center) ear section was measured. The average thickness of the ear was calculated as the area (in mm²) divided by 2 mm.

Statistics. All tumor volumes and quantified variables were expressed as the mean ± SE. Statistics was done with GraphPad Prism software (GraphPad Software, Inc.). Student's t test was used to compare variables of treated tumors versus untreated control. Sample size and power of all the analysis were calculated with PS Power and Sample Size Program (34). Differences were considered statistically significant when P ≤ 0.05.

Expression of rAAV-3TSR in pancreatic cancer cells and mouse skeletal muscle. Our previous published data showed in vitro expression of endostatin in rAAV-endostatin–transduced cells (31). Increased endostatin levels were detected in the culture medium of rAAV-endostatin–transduced human tumor cells, and conditioned medium from those transduced cells significantly inhibited capillary endothelial cell proliferation (31). Here, we tested the expression of rAAV-3TSR both in vitro and in vivo. Human pancreatic cancer cells (AsPC-1 and MiaPaCa-2) were transduced with rAAV-GFP (control vector) or rAAV-3TSR. Forty-eight hours after transduction, GFP expression was observed in rAAV-GFP–transduced cells (Fig. 1B). Culture medium was changed 48 h after transduction and collected 24 h later for 3TSR protein detection. In this way, we verified that the detected 3TSR was secreted by the transduced cells following de novo synthesis and did not arise from preexisting protein within the virus preps. Figure 1C shows the secretion of 3TSR protein by the rAAV-3TSR–transduced MiaPaCa-2 and AsPC-1 pancreatic cancer cells. 3TSR production was consistent with the transduction efficiency observed in the control rAAV-GFP transduction. In vivo expression of rAAV-3TSR was detected via immunofluorescence staining in mouse skeletal muscle at 5 and 9 weeks after a single i.m. injection (Fig. 1D). In control mice receiving only a rAAV-LacZ injection, no positive 3TSR staining was observed in the injection sites (Fig. 1D). The expression of rAAV-endostatin was tested in wild-type FVB mice. Plasma endostatin level was elevated 2 weeks after i.m. injection and peaked at 4 weeks, and the elevated endostatin levels persisted at least until 8 weeks after rAAV-endostatin injection.

In vivo antiangiogenic efficacy of rAAV-3TSR and rAAV-endostatin. The antiangiogenic efficacy of rAAV-endostatin and rAAV-3TSR in vivo was studied using an Adeno-VEGF–induced mouse ear angiogenesis model in three experimental settings (Fig. 2). Injecting the ears simultaneously with a control rAAV-GFP vector did not alter the proangiogenic effects of Adeno-VEGF (Fig. 2A). In contrast, when Adeno-VEGF was injected simultaneously with rAAV-endostatin or rAAV-3TSR, VEGF-induced ear angiogenesis was significantly inhibited (Fig. 2A and B). Pretreatment of mouse ears with i.d. injection of rAAV-3TSR followed by induction of ear angiogenesis 2 weeks later by injection of Adeno-VEGF at the same sites also resulted in significant inhibition of ear angiogenesis in the rAAV-3TSR pretreated but not control ears (Fig. 2C).

To show the systemic effects of rAAV-3TSR treatment, we injected rAAV-3TSR i.m. and then induced angiogenesis in the mouse ear with the Adeno-VEGF. No significant inhibition of ear angiogenesis was seen if the Adeno-VEGF was given 2 weeks after i.m. injection of rAAV-3TSR. However, at 4, 6, and 8 weeks after rAAV-3TSR i.m. injection, Adeno-VEGF–induced ear angiogenesis was significantly inhibited in the rAAV-3TSR–treated mice compared with control. Figure 2D shows a typical result at 6 weeks. Quantification of the mouse ear angiogenic area showed a 61.0% reduction in mice treated with a single i.m. injection of rAAV-3TSR 6 weeks before induction of ear angiogenesis with Adeno-VEGF (Fig. 2D). Adeno-VEGF–induced mouse ear angiogenesis is characterized by enlarged, thin-walled, serpentine, pericyte-poor, hyperpermeable, and sinusoid-appearing mother vessels, which arise from the enlargement of preexisting microvessel, primarily venules (35). In mice pretreated with rAAV-3TSR i.m. injection, the VEGF-induced mother vessels were limited to the center of the angiogenic areas, and the average thickness of the ear was significantly decreased compared with the control animals (Fig. 2E).

Inhibition of orthotopic pancreatic tumor growth after rAAV-3TSR and/or rAAV-endostatin treatment. An orthotopic animal model of pancreatic cancer was chosen to evaluate both the antiangiogenic and antitumor efficacy of our novel gene therapy vectors. Figure 3A shows the average tumor volume and the volume of each individual tumor from control and different treatment groups. There was a significant protective effect against tumor growth in immunocompromised mice pretreated with a single i.m. injection of either rAAV-3TSR or rAAV-endostatin 4 weeks before implantation of human AsPC-1 pancreatic cancer cells into the mouse pancreas. The average tumor volume was decreased by 45.7% (rAAV-3TSR) and 36.2% (rAAV-endostatin), respectively (P < 0.01 versus control). The rAAV-3TSR and rAAV-endostatin pretreatments were equally effective; there was no significant difference between the average tumor volume of rAAV-3TSR–treated and rAAV-endostatin–treated mice. Of note, a greater inhibitory effect on tumor growth was observed when mice received injections of both rAAV-3TSR and rAAV-endostatin. The average tumor volume was significantly smaller in this combined treatment group (60.8% reduction) comparing with those in any single treatment group (P < 0.02).

Intrasplenic injection of rAAV-3TSR was also done to allow uniform hematogenous distribution and expression of the vectors in the liver. Intrasplenic injection with the rAAV-3TSR vector resulted in comparable tumor inhibitory effects as i.m. injection of the same vector.

Treatment of established pancreatic tumors with localized administration of rAAV-3TSR and rAAV-endostatin. To mimic the more clinically relevant situation, when tumors are already established, we also treated mice bearing sizable orthotopic pancreatic tumors with intratumoral injection of rAAV-endostatin or rAAV-3TSR and sacrificed the mice 10 days later. Mice in the control group received an intratumoral rAAV-GFP injection, which showed no effect on tumor volume compared with mice receiving rAAV-GFP i.m. As shown in Fig. 3B, a single intratumoral injection of either rAAV-3TSR or rAAV-endostatin significantly decreased the volume of established orthotopic pancreatic tumors (57.1% and 48.8% reduction, respectively).

Detection of endostatin in mouse plasma and tumor tissue after rAAV-endostatin gene therapy. Due to the high background in ELISA assay with mouse plasma and in immunostaining on human tumor tissue, we were unable to detect 3TSR protein in mouse plasma or the orthotopic human tumor xenograft from the mouse model. Detection of endostatin was used to examine the pattern of expression of secreted proteins following rAAV-mediated tumor gene therapy. Significantly enhanced endostatin immunohistochemical staining was observed in tumors after injection with rAAV-endostatin; the absorbance value of the staining was increased 2.6-fold (Fig. 4A). Plasma endostatin levels were also elevated in mice treated with rAAV-endostatin compared with control or rAAV-3TSR–treated mice (Fig. 4B).

rAAV-3TSR and rAAV-endostatin gene therapy decreased tumor microvessel density. We further analyzed change in pancreatic tumor microvessel density after treatment with rAAV-endostatin and/or rAAV-3TSR. The average microvessel density was between 7.2% and 7.7% in the i.m. and intratumoral injection control tumors. As shown in Fig. 5, the tumor microvessel density was significantly decreased in each treatment group. Consistent with the decreases noted in tumor volume, mice treated with the combination of rAAV-3TSR and rAAV-endostatin showed a significantly lower tumor microvessel density compared with mice treated with either vector alone, indicating a stronger antiangiogenic efficacy when the two vectors were combined.

A combination of rAAV-3TSR and rAAV-endostatin exhibited stronger antiangiogenic efficacy in the mouse ear angiogenesis model. To further study the combinatorial effects of rAAV-3TSR and rAAV-endostatin, we pretreated mouse ears with a lower dosage of the rAAV vectors and induced ear angiogenesis with Adeno-VEGF 2 weeks later in the same injection sites. The lower dose was given to make the detection of any combinatorial effects easier. Figure 6 shows representative pictures of ear angiogenesis from each group. At this lower dosage, neither rAAV-3TSR nor rAAV-endostatin alone showed significant inhibitory effects on VEGF-induced angiogenesis. However, the combination of rAAV-3TSR and rAAV-endostatin significantly inhibited ear angiogenesis induced by Adeno-VEGF (Fig. 6).

Antiangiogenic gene therapy offers a powerful strategy for cancer treatment with its capability to maintain sustained levels of antiangiogenic factors, thereby enhancing their antiangiogenic and antitumor efficacy (19–23, 27). Gene therapy is also highly flexible, offering the option of custom tailoring either highly localized or systemic treatments, and is more cost effective than expensive recombinant proteins.

rAAV offers highly efficient as well as long-term in vivo expression without progressive silencing over time (27) and provides distinct advantages for antiangiogenic gene therapies. AAV2-mediated in vivo gene transduction has been reported to be most efficient in skeletal muscle and brain followed by hepatocytes. Consequently, i.m. and intraportal vein injection have been the most widely used delivery methods for AAV-mediated systemic therapies, including systemic cancer gene therapy (27). After i.m. or intraportal vein delivery, it takes 4 to 6 weeks for rAAV-mediated transgenes to reach steady-state expression levels in the circulation (26, 27, 29), and transduced skeletal muscle cells or hepatocytes will persistently synthesize a secreted factor for a prolonged period and thus maintain a sustained plasma level of the protein.

In the present study, we show that rAAV-mediated delivery of 3TSR or endostatin significantly inhibited the growth of orthotopic pancreatic tumors. The direct antiangiogenic effect of these vectors was clearly shown by decreased ear angiogenesis in response to VEGF in a mouse ear model and by decreased tumor microvessel density in the tumor model. We delivered rAAV-3TSR via both i.m. and intrasplenic injection, which is used here as a substitute for intraportal vein delivery. Comparable antitumor efficacy was observed in the i.m. and intrasplenic treatment groups. From the clinical perspective, i.m. delivery of rAAV vectors is easier and less invasive, whereas the potential advantage of overexpressing antiangiogenic factors in hepatocytes may lie in preventing liver metastasis after surgical removal of a primary pancreatic tumor, for example. To our knowledge this is the first time that a TSR-based gene therapy approach has been used successfully to inhibit angiogenesis and tumor progression in a mouse pancreatic tumor model or that a combination antiangiogenic approach using TSP-1 and endostatin has been reported. It also represents the first occasion where we have seen this mouse ear angiogenesis model used to screen antiangiogenic gene vectors.

Increased antitumor and antiangiogenic efficacy were observed when mice were treated with the combination of rAAV-3TSR and rAAV-endostatin. When the two rAAV vectors were used in combination, more significant angiogenic inhibition was observed in the ear angiogenesis model. Mice treated with the combination of rAAV-3TSR and rAAV-endostatin also showed significantly decreased tumor volume and tumor microvessel density compared with mice that received either vector alone. This is consistent with their possessing distinct mechanisms of action. The antiangiogenic effects of TSP-1 are reportedly mediated by interaction of the TSRs with CD36 on the endothelial cell membrane (36). 3TSR binds CD36 and sequentially activates p59 fyn, caspase-3, and p38 mitogen-activated protein kinase, triggering apoptosis of microvessel endothelial cells (37). By contrast, endostatin acts by binding to tropomyosin, integrins, and matrix metalloproteinases (38–40). The half-lives of 3TSR⁴

and endostatin are different, complicating drug scheduling with recombinant proteins. Stable gene therapy approaches such as described here can overcome these limitations by maintaining stable systemic levels of both 3TSR and endostatin.

To mimic the clinical situation, we also treated established tumors using localized gene delivery and showed significant tumor volume reductions after intratumoral rAAV-3TSR or rAAV-endostatin treatment. Similarly, intratumoral delivery of rAAV vectors expressing angiostatin or tissue inhibitor of metalloproteinase-1 has been reported to have significant antitumor efficacy, inhibiting tumor angiogenesis in animals bearing either Kaposi's sarcoma or orthotopic glioma (41, 42).

Our data also indicate the therapeutic importance of local levels of antiangiogenic factors. Although it requires ∼4 weeks for AAV2-mediated gene expression to reach the steady-state level, expression of a transgene commences shortly after the single-stranded genome of AAV is converted into a double-stranded structure (27, 43). This is in line with our in vitro data that production of endostatin and 3TSR could be detected at 48 to 72 h after transduction (31). A potential advantage of intratumoral gene delivery is that, in a given cell type, expression from an AAV-delivered transgene may be higher in cells that are actively dividing because in actively dividing cells, such as tumor cells, the enhanced metabolic rate may promote DNA replication and gene expression (27). In our study, the increased levels of 3TSR and endostatin in the tumor microenvironment, although not reaching steady-state or peak levels, were sufficient to shift the angiogenic balance toward the antiangiogenic side, decrease tumor microvessel density, and consequently inhibit tumor growth. This notion is also supported by the data from the mouse ear angiogenesis model. Significant inhibition of angiogenesis was observed when rAAV-3TSR and Adeno-VEGF were injected together into the ear. However, when we injected rAAV-expressing antiangiogenic factors i.m., significant inhibition of VEGF-induced ear angiogenesis was observed only when rAAV was injected at least 4 weeks before the Adeno-VEGF ear injection.

In summary, rAAV vectors expressing 3TSR and endostatin showed significant antiangiogenic efficacy in vivo. Using experimental settings that closely mimic clinical situations, we conclude that rAAV-mediated 3TSR and endostatin antiangiogenic gene therapy could provide a promising regimen for patients with pancreatic cancer as well as other gastrointestinal tumors. The antiangiogenic vector can be delivered i.m. and intratumorally in patients with unresectable pancreatic tumors using radiologic guidance or injected into the portal vein after surgical removal of the primary tumors.

We thank Drs. Harold F. Dvorak and Janice A. Nagy for providing Adeno-VEGF and the mouse ear angiogenesis model and Eleanor Manseau for processing the 1-μm ear sections.