Introduction

It is now widely accepted that tumors require de novo formation of new blood vessels, or angiogenesis, to provide oxygen and essential nutrients to support their growth and dissemination^5,11,12,15. As a tumor expands, a number of proangiogenic growth factors are elaborated, which stimulate quiescent endothelial cells, or their precursors, to proliferate, migrate, differentiate, and ultimately form a vascular network that permeates into the tumor^{5,11,12,14,15,46}. Tumors may also modulate their growth by liberating certain antiangiogenic factors that impede tumor-associated neovascularization, often resulting in dormancy. The growth and spread of many solid tumors appear to be tightly controlled by the balance of pro- and antiangiogenic factors. Intensive efforts are currently under way in preclinical and clinical studies for evaluating the effective use of antiangiogenic agents as a novel cancer treatment modality^8,16,43.

A number of physiological antiangiogenic factors are derived from proteolytic cleavage of larger native proteins. One such endogenous angiostatic factor is angiostatin. Angiostatin is a proteolytic fragment consisting of four triple-looped kringle domains of plasminogen^30,33,37,42. Unlike the parental plasminogen molecule, angiostatin is a potent inhibitor of angiogenesis, tumor growth, and metastasis^{4,32,33,38,44}. Angiostatin exerts its biological activity by inhibiting the growth and function of endothelial cells; however, its precise receptor and biochemical mechanism of action currently remain ill-defined. Although the individual kringle domains of angiostatin demonstrate distinct activities for modulating endothelial cell function, the kringles 1–3 fragment appears to be sufficient for tumor inhibition²³.

Systemic administration of recombinant angiostatin protein has been shown to suppress tumor growth and metastasis in a variety of different murine tumor models^2,18,33,47. In addition, angiostatin treatment does not appear to cause toxicity in mice or humans. Although the efficacy and toxicity profiles make angiostatin a promising cancer therapeutic, the optimal dosage and delivery of recombinant angiostatin protein required for effective antiangiogenic chronic therapy may limit its clinical feasibility. Human angiostatin is an unstable protein in vitro and is rapidly cleared from the circulation in vivo³². Repeated administration of high doses of recombinant angiostatin protein is usually required to achieve tumor eradication in vivo^18,32. However, recent studies have demonstrated that chronic low-dose administration of antiangiogenic proteins may have improved therapeutic activity in comparison to using high dosage cyclical delivery of the same agents. For example, continuous administration of angiostatin protein by a subcutaneous osmotic pump was found to be superior to conventional twice-daily bolus injections in achieving tumor suppression⁶. To overcome the limitations of chronic protein administration required for optimal antiangiogenic therapy, we explored the potential of continuously expressing angiostatin in vivo by systemic gene delivery using recombinant adeno-associated virus (AAV).

AAV is a small, nonpathogenic human parvovirus^27,28,31. The virus is replication-defective and contains a linear, single-stranded DNA capable of integrating into chromosome 19 of the host genome or persisting as a stable episome⁷. AAV is an attractive vector platform for antiangiogenic gene therapy since it allows for efficient long-term expression of transgenes in vivo without inducing an effective cellular immune response²⁷. AAV vectors have been successfully employed for achieving prolonged transgene expression to treat cancer and a number of genetic diseases using systemic gene therapy^1,13,28. Here we describe the construction and generation of recombinant AAV vectors expressing kringles 1–3 of human angiostatin. A single administration of AAV-angiostatin into the liver resulted in continuous systemic expression of human angiostatin in mice for extended periods without any evidence of liver-mediated toxicity. Continuous systemic expression of human angiostatin by AAV gene therapy resulted in significant inhibition of lung tumor burden and prolonged survival in B16 melanoma and Lewis lung carcinoma tumor-bearing mice. When AAV-angiostatin treatment was combined with certain cytotoxic agents, tumor-bearing mice had an increased survival advantage compared to chemotherapy alone; however, the effect was observed only within a narrow serum level of circulating angiostatin in vivo. These preclinical studies demonstrate the feasibility of using AAV antiangiogenic gene therapy as a novel cancer treatment and suggest that systemic angiostatin expression following gene transfer may have a narrow window of efficacy for blocking tumor growth in vivo.

Results

Expression of Human Angiostatin from AAV-Angiostatin-Transduced Cells

To evaluate the therapeutic effects of human angiostatin in preclinical tumor models using systemic gene therapy, we constructed recombinant AAV vectors that encoded a secreted truncated form of human angiostatin (encompassing kringle domains 1–3 of plasminogen with a tissue plasminogen activator signal sequence) under the control of the ubiquitous EF1 promoter. We also constructed a corresponding recombinant AAV vector encoding a green fluorescent protein (GFP) cDNA to serve as a vector control. We infected HuH7 liver cells with recombinant AAV vectors encoding human angiostatin or a control GFP construct at a multiplicity of infection of 1 10⁵ virus particles/cell. After 48 h, we harvested the conditioned medium and analyzed it for expression of soluble angiostatin using an anti-human plasminogen antibody that has been demonstrated previously to cross-react with the angiostatin-containing kringle domains 1–3 of plasminogen²³. As shown in Fig. 1A, we detected an immunoreactive band against human plasminogen with an apparent molecular weight of 35 kDa from the supernatants of AAV-angiostatin-transduced cells, but not from AAV-GFP control cells. Under SDS–PAGE conditions, angiostatin produced by the recombinant AAV vector appeared to migrate slower in comparison with recombinant human kringles 1–3 angiostatin protein produced from Pichia pastoris (Fig. 1A). This discrepancy in molecular weight is likely due to differential glycosylation since the recombinant protein produced from Pichia lacks an intact N-linked glycosylation site (N306) normally present in the third kringle subdomain of plasminogen³⁶.

Figure 1.

Expression of human angiostatin following AAV-angiostatin gene transfer in vitro and in vivo. (A) Western blotting analysis of AAV-produced human angiostatin. HuH7 cells were transduced with recombinant AAV vectors expressing GFP (AAV-GFP) or human angiostatin (AAV-Angiostatin) at a multiplicity of infection of 1 10⁵ for 48 h. The conditioned medium from AAV-GFP- (lane 1) or AAV-angiostatin- (lane 3) transduced cells, or 100 ng of recombinant angiostatin protein (lane 2), was resolved by SDS–PAGE and analyzed for human angiostatin expression by immunoblotting using HRP-conjugated anti-human plasminogen antisera. Molecular weight markers are shown on the right. (B) Long-term sustained expression of human angiostatin in vivo following AAV-angiostatin gene transfer in immune-deficient mice. Female Balb/c nude mice (n = 6) were injected with a single dose of 2 10¹¹ virus particles of AAV-angiostatin into the portal vein (squares) or tail vein (circles). Mice were bled by retro-orbital puncture on scheduled intervals up to 4 months and assayed for mean circulating angiostatin levels (+SEM) by using a capture sandwich ELISA.

Full figure and legend (89K)

We analyzed the systemic expression of angiostatin in vivo following liver-directed AAV-mediated gene transfer in immunodeficient mice. We inoculated Balb/c nude mice with a single administration of 2 10¹¹ virus particles of AAV-angiostatin using tail-vein or portal-vein routes of administration. Both routes of administration result in AAV transduction of the liver and hepatocytes (T. Harding et al., submitted for publication). We detected low levels of human angiostatin in the sera of mice as early as 1 week following AAV-mediated gene transfer in vivo (Fig. 1B). Angiostatin transgene expression increased over time until steady-state levels were reached approximately 4 weeks post-vector inoculation (Fig. 1B). Delayed onset of transgene expression is common for this vector platform and is believed to be a result of AAV converting its single-stranded DNA into a double-stranded template for transgene expression to take place²⁴. Compared to portal-vein administration, tail-vein injections of AAV-angiostatin resulted in approximately 50% lower levels of circulating angiostatin following gene transfer. We detected continuous and sustained expression of angiostatin in the sera of mice for more than 6 months following a single injection of AAV-angiostatin (Fig. 1B). Animals treated with AAV-angiostatin did not display any adverse reactions or abnormal health conditions for up to half a year following gene transfer (data not shown). Moreover, no evidence of liver-mediated toxicity resulting from AAV-angiostatin gene transfer was noted as serum transaminase (sGPT/sALT) levels in mice showed normal ranges (<30 ng/ml) for more than 6 months following vector administration (data not shown).

Angiostatin Expressed from AAV-Transduced Cells Blocks Angiogenesis

To verify the biological activity of angiostatin produced from our recombinant AAV vectors, we tested the antiangiogenic properties of AAV-produced angiostatin in standard angiogenesis assays. Given that the precise stage by which angiostatin exerts its biological activity remains unclear, we tested the effects of AAV-produced angiostatin in the context of a true physiological microenvironment in which angiogenesis normally occurs. The chick chorioallantoic membrane (CAM) assay is a widely used in vivo model that recapitulates the multiple physiological, biochemical and cellular processes that occur during the angiogenic cascade and is relatively rapid for systematically evaluating antiangiogenic agents³. Incubation of basic fibroblast growth factor (bFGF)-soaked filter discs on chorioallantoic membranes of 11-day-old chick embryos stimulated an angiogenic response that could be readily quantified after 72 h by counting the number of new converging microvessel branch points beneath the excised filter disc by stereomicroscopy (Fig. 2). bFGF-induced angiogenesis in the CAMs was significantly attenuated when the filter discs contained either supernatants containing AAV-produced angiostatin or recombinant angiostatin protein, and this inhibition occurred in a dose-dependent manner (Figs. 2A and 2B). The addition of either recombinant angiostatin protein or AAV-produced angiostatin at a dose of 100 ng per disc appeared to abolish bFGF-stimulated angiogenesis completely (Fig. 2B). However, when applied at a lower dose of 10 ng per disc, recombinant angiostatin had no observable inhibitory antiangiogenic effects in the CAMs, while the same concentration of AAV-produced angiostatin still retained the ability to block new blood vessel sprouts potently. Coating the bFGF-filter discs with supernatants from mock or control AAV-transduced cells had no apparent inhibitory effects on growth factor-induced angiogenesis in the CAMs, suggesting that the antiangiogenic effects observed were specifically due to angiostatin (Figs. 2A and 2B). Taken together, these results demonstrate that angiostatin can be efficiently produced from AAV-transduced cells and the expressed protein retains potent antiangiogenic activity in vitro and in vivo. We also observed similar inhibitory effects by AAV-produced angiostatin on VEGF-stimulated angiogenesis when tested in the CAM assay (data not shown).

Figure 2.

Angiostatin expressed from AAV-transduced cells blocks bFGF-induced neovascularization in the chick chorioallantoic membrane (CAM) assay. Hydrocortisone-containing filter discs were soaked with 25 ng bFGF together with recombinant human angiostatin (10 or 100 ng) or conditioned medium from mock- or AAV-angiostatin- (10, 50, or 100 ng) transduced cells as described in the legend to Fig. 1 and placed on chorioallantoic membranes of 11-day-old chick embryos. (A) After 72 h incubation, the CAMs were excised and discs were harvested for analysis of bFGF-induced angiogenesis in situ. Shown are representative micrographs. (B) The mean number (SEM) of branch points occurring in the new sprouting blood vessels from treated CAMs was enumerated and considered significant between the control vs angiostatin-treated samples (^**P < 0.01).

Full figure and legend (132K)

AAV-Angiostatin Gene Therapy Inhibits B16F10 Tumor Growth

Preclinical observations suggest that antiangiogenic therapy may have the greatest clinical impact on blocking minimal residual disease or preventing micrometastasis¹¹. After verifying that AAV-produced angiostatin is potently antiangiogenic, and having established sustained expression of angiostatin following liver-directed AAV-mediated gene transfer in vivo, we evaluated the anti-tumor efficacy of AAV-angiostatin gene therapy in various murine tumor models, using first the B16F10 murine melanoma tumor model of experimental lung metastasis. The B16 melanoma model is a highly aggressive tumor model that is ideal for rigorous screening of anti-cancer agents and that has previously been shown to be modestly sensitive for tumor inhibition by recombinant angiostatin protein therapy²³. To monitor the growth of tumors in mice over the course of therapy, we used a stable luciferase-expressing B16F10 clone, B16F10-luc, which could be visualized by bioluminescence imaging. Since mice succumb to B16F10 tumor challenge rapidly (3 weeks) and given the delayed onset of angiostatin expression by AAV vectors in mice (3 weeks), it was necessary to administer the vector first to establish sustained transgene expression prior to challenging with tumor. We injected Balb/c nude mice with 5 10¹⁰ virus particles of AAV-angiostatin or AAV-GFP via portal-vein delivery and monitored circulating angiostatin levels on a weekly basis. Four weeks after liver-directed gene transfer, serum angiostatin levels in the AAV-angiostatin-treated mice averaged 15 ng/ml and were sustained at this level for the duration of the experiment. In contrast, mice treated with AAV-GFP control vector showed no measurable human angiostatin in their blood (data not shown). Following AAV treatment, mice were monitored for B16F10-luc tumor burden by bioluminescence imaging. Mice treated with AAV-GFP showed a progressive increase in B16F10-luc lung metastasis from day 3 to 18 post-tumor inoculation (Fig. 3A). The tumor burden detected by luciferase imaging in this group correlated with lung tumor metastasis by postmortem analysis (data not shown). In contrast, AAV-angiostatin-treated mice showed a delayed onset of tumor burden as luciferase-expressing B16F10 lung metastasis began to develop only at day 18 compared to day 3 in the AAV-GFP control group (Fig. 3A). Balb/c nude mice pretreated with 5 10¹⁰ virus particles of AAV-angiostatin demonstrated a threefold survival advantage in comparison to mice receiving the control vector after tumor challenge, with a median survival time of 94 days compared to 28 days (P = 0.007) in the AAV-GFP control group (Fig. 3B). Thus, a single systemic administration of AAV-angiostatin had a profound effect in delaying the onset of B16F10 lung tumors and prolonged the survival of tumor-bearing mice by threefold. Interestingly, mice pretreated with either higher (1 10¹¹ vp) or lower (1 10¹⁰ vp) doses of AAV-angiostatin, which resulted in circulating angiostatin levels of 75 and <5 ng/ml, respectively (data not shown), did not show any statistical increase in survival advantage compared to mice receiving corresponding control vectors (Fig. 3C), suggesting that AAV-angiostatin gene treatment has a narrow window of efficacy in the B16F10 model of experimental metastasis.

Figure 3.

AAV-angiostatin delays onset of B16F10 melanoma tumor burden and significantly prolongs survival in tumor-bearing mice at a narrow dose range. Female Balb/c nude mice were inoculated with indicated doses of AAV-angiostatin (n = 9) or AAV-GFP vectors (n = 9) by portal-vein injections. Four weeks after AAV administration, mice were challenged with 5 10⁴ B16F10 melanoma cells stably expressing a luciferase reporter gene (B16F10-luc), by tail-vein injections, to induce experimental lung metastasis. Mice were (A) monitored at interim time points for lung tumor burden or (B and C) monitored for survival. (A) Bioluminescence imaging showing the growth of B16F10-luc tumor burden over time in representative Balb/c nude mice that were preinjected with 5 10¹⁰ virus particles of AAV-GFP (top) or AAV-angiostatin (bottom). (B) Kaplan–Meier survival curves showing increased median survival times (MST) in B16F10-luc tumor-bearing mice that were pretreated with a medium dose (5 10¹⁰ virus particles) of AAV-angiostatin (MST = 94 days) vs AAV-GFP (MST = 28 days) (P = 0.007). (C) Kaplan–Meier curves show no statistical survival advantage observed in mice receiving high dose (1 10¹¹ virus particles) or low dose (1 10¹⁰ virus particles) virus particles of AAV-angiostatin in comparison to mice pretreated with the corresponding high dose of AAV-GFP or low dose of AAV-GFP control vectors (P = 0.4457). Mice were sacrificed when labored breathing occurred or general health conditions deteriorated due to excessive tumor burden.

Full figure and legend (281K)

AAV-Angiostatin Gene Therapy Inhibits Lewis Lung Carcinoma (LLC) Tumor Growth within a Narrow Cose Range

To test the broad applicability of AAV-angiostatin as a therapeutic agent and to explore its effective dose range further, we investigated tumor models in addition to B16F10. We evaluated the anti-tumor effects of systemic AAV-angiostatin gene therapy against murine LLC. We made a stable luciferase-expressing LLC (LLC-luc) clone and injected it into Balb/c nude mice via the tail vein to induce experimental lung metastases. Similar to the B16F10-luc tumor model described above, mice were pretreated with AAV-angiostatin to establish transgene expression 4 weeks prior to tumor challenge. We evaluated three different dose levels of AAV-angiostatin to determine an effective dose of circulating angiostatin required for anti-tumor activity in this model. Administration of 1.5 10¹¹, 7.5 10¹⁰, and 1.5 10¹⁰ virus particles of AAV-angiostatin by tail-vein injection resulted in approximately 85, 10, and <0.5 ng/ml circulating angiostatin levels by week 4, and these levels appeared to be sustained during the course of tumor challenge (data not shown). After verifying transgene expression, we challenged mice with 1 10⁶ LLC-luc cells by tail-vein injection and analyzed them for LLC lung metastasis by bioluminescence imaging (Figs. 4A and 4B). At day 21 post-tumor inoculation, all of the mice pretreated with AAV-GFP showed visible luciferase-expressing LLC tumors in their lungs (Fig. 4A). In contrast, only a small percentage of mice pretreated with 7.5 10¹⁰ virus particles of AAV-angiostatin showed visible LLC-luc lung tumor burden. Luciferase imaging at days 15, 17, and 21 showed that the overall LLC-luc tumor burden progressed rapidly in the AAV-GFP control-treated mice, but was significantly suppressed by AAV-angiostatin for the first 3 weeks after tumor challenge (Fig. 4A). Mice pretreated with the highest and lowest doses of AAV-angiostatin had a similar median survival time of 27 days in comparison to mice treated with the AAV-GFP control vector (Fig. 4B). In contrast, the median survival time for mice treated with the medium dose (7.5 10¹⁰ vp) of AAV-angiostatin was 34 days (P = 0.037) (Fig. 4B). These studies demonstrate that continuous expression of angiostatin by liver-directed AAV gene transfer had a significant effect in delaying the onset of tumors and prolonging the survival of tumor-bearing mice; however, the anti-tumor effect appears to be limited to a narrow dose range of angiostatin.

Figure 4.

AAV-angiostatin inhibits Lewis lung carcinoma tumor development and prolongs survival in tumor-bearing mice in a dose-dependent manner. Female Balb/c nude mice (n = 10/group) were inoculated with a high dose (2 10¹¹ virus particles), a medium dose (7.5 10¹⁰ virus particles), or a low dose (1.5 10¹⁰ virus particles) of AAV-angiostatin by tail-vein injection. Control mice were inoculated with 7.5 10¹⁰ virus particles of AAV-GFP. Four weeks after AAV administration, mice were challenged with 1 10⁶ Lewis lung carcinoma cells stably expressing a luciferase reporter gene (LLC-luc), by tail-vein injection, to induce experimental lung metastasis. Mice were (A) monitored at interim time points for lung tumor burden or (B) monitored for survival. (A) Bioluminescence images of mice treated with 7.5 10¹⁰ virus particles of AAV-GFP (top) and AAV-angiostatin (bottom) taken at day 21 post-tumor inoculation to assess for luciferase-expressing LLC tumors. (B) Kaplan–Meier survival curves showing an increase in median survival time (MST) in the medium-dose group of AAV-angiostatin (MST = 34 days)-treated mice vs AAV-GFP control (MST = 27 days)-treated mice after being challenged with LLC-luc tumor cells as described above. Mice were sacrificed when labored breathing occurred or general health conditions deteriorated due to excessive tumor burden. A log-rank test performed on the Kaplan–Meier curves showed that the medium dose of AAV-angiostatin treatment is significant (P = 0.0376) compared to the other groups.

Full figure and legend (139K)

Anti-Tumor Effects of AAV-Angiostatin Treatment in Combination with Chemotherapy

Although the anti-tumor effects of AAV-angiostatin observed in the B16F10 and LLC models were encouraging, the use of AAV-angiostatin as a monotherapy was not effective at completely eradicating tumor growth. To improve the anti-tumor efficacy of AAV-angiostatin treatment, we investigated combination treatment strategies using chemotherapy in several xenograft models using preestablished human tumors grown in immunodeficient mice. We evaluated the effect of AAV-angiostatin as an adjuvant therapy to chemotherapy in a breast carcinoma xenograft model. The model was designed to assess the effects of AAV-angiostatin in a setting of minimal tumor burden, which best mimics a clinically relevant setting. In this model, mice bearing preestablished 100-mm³ MX-1 breast xenografts were treated with a single maximal tolerated dose (120 mg/kg) of Cytoxan (CTX) to debulk the tumor. One week following chemotherapy, mice received a single administration of AAV-angiostatin by tail-vein injections. We evaluated three different doses of AAV-angiostatin vector particles, 3 10¹¹ (high dose), 1.5 10¹¹ (medium dose), and 8 10¹⁰ vp (low dose), which resulted in circulating angiostatin expression profiles similar to those observed in the LLC and B16 studies described earlier. The vector particle doses used for these combination studies resulted in sustained circulating angiostatin expression levels of 75, 25, and 5 ng/ml, respectively (data not shown). Untreated mice had a median survival time of 28 days (Fig. 5A). Treatment with a single 120 mg/kg ip dose of CTX extended the overall survival of mice by approximately 30 days (median survival time (MST) = 57.5 days). When used in the adjuvant setting, only the middle (1.5 10¹¹ virus particles) dose of AAV-angiostatin appeared to increase the survival time of MX-1 tumor-bearing nude mice compared to the CTX monotherapy (66 days) (Fig. 5A).

Figure 5.

Increased survival of tumor-bearing mice following AAV-angiostatin treatment in combination with chemotherapy. (A) Kaplan–Meier survival curves showing the efficacy of AAV-angiostatin as an adjuvant therapy in nude mice bearing MX-1 breast carcinomas. A 1-mm³ MX-1 breast carcinoma fragment was implanted sc in the flank region of female athymic nude mice. Mice were pair-matched when the tumors reached 100 mm³ (considered day 1) and treated with saline (white squares) or a single ip injection of Cytoxan (CTX, cyclophosphamide) at a dose of 120 mg/kg to debulk the primary tumor. At day 7 following CTX reduction of tumor size, mice were treated adjuvantly with a high dose (3 10¹¹ virus particles), medium dose (1.5 10¹¹ virus particles), or low dose (8 10¹⁰ virus particles) of AAV-angiostatin by tail-vein injection. Mice were euthanized as a "cancer death" when the MX-1 carcinoma reached or exceeded 1500 mm³. The medium dose of AAV-angiostatin was compared to the other treatment arms and shown to be nonsignificant based on a log-rank test (P = 0.142). (B) Kaplan–Meier survival curves showing the efficacy of AAV-angiostatin in combination with irinotecan (CPT-11) chemotherapy in nude mice bearing LS174t colon carcinomas. A 1-mm³ LS174t colon carcinoma fragment was implanted sc in the flank region of female athymic nude mice. Mice were pair-matched when the tumors reached 100 mm³ (considered day 1) and treated with saline (white squares) or ip injections of irinotecan at a dose of 100 mg/kg, once a week for 3 weeks. At day 3, mice received saline (black squares) or a high dose (3 10¹¹ virus particles), medium dose (1.5 10¹¹ virus particles), or low dose (8 10¹⁰ virus particles) of AAV-angiostatin by tail-vein injection. Mice were euthanized as a "cancer death" when the LS174t colon carcinoma reached or exceeded 1500 mm³. The medium dose of AAV-angiostatin was compared to the other treatment arms and shown to be nonsignificant based on a log-rank test (P = 0.192).

Full figure and legend (138K)

We also evaluated the effect of AAV-angiostatin in combination with chemotherapy in a human colon carcinoma xenograft tumor model. We treated athymic mice bearing preestablished LS174t colon carcinoma tumors initially with irinotecan (CPT-11) at a standard dosing regimen of 100 mg/kg given once weekly for 3 weeks by ip injection. Subsequently we treated the mice with a single tail-vein injection of AAV-angiostatin at one of three doses: high dose (3 10¹¹ virus particles), medium dose (1.5 10¹¹ virus particles); or low dose (8 10¹⁰ virus particles). We saw sustained circulating angiostatin expression levels of 75, 25, and 5 ng/ml in mice receiving the high, middle, and low dose of AAV-angiostatin, respectively (data not shown). Control mice receiving no treatment had a median survival time of 28 days. Irinotecan treatment, given as a standard dose of 100 mg/kg once weekly, extended the median survival time to 54 days (Fig. 5B). Only the middle (1.5 10¹¹ virus particles) dose of AAV-angiostatin increased the survival time of LS174t-bearing nude mice compared to irinotecan therapy alone (71.5 days vs 54 days). Although angiostatin's anti-tumor effects appear to be marginal, both combination studies from the two different tumor models examined demonstrated a trend in efficacy when the middle dosage of AAV-angiostatin was used. Thus, AAV-angiostatin treatment afforded a marginal increase in overall survival in tumor-bearing mice receiving chemotherapy; however, the effect appears to be limited to a very narrow dose range of angiostatin.

Discussion

The use of angiogenic inhibitors, such as angiostatin, has been postulated to help reverse the angiogenic switch and impede the growth of tumor-associated vascularization^{8,11,12,16,17,37,43}. Indeed, preclinical studies have shown that bolus injections of recombinant angiostatin protein curtails tumor growth and shrinks tumor blood vessels; however, continuous infusion or large doses of proteins are often necessary to achieve complete tumor regression^{18,32,33,38,44}. The feasibility of using recombinant angiostatin protein as an effective pharmaceutical agent for the chronic treatment of cancer may be limited since the protein is highly labile and displays poor pharmacokinetic properties in vivo. An alternative to frequent bolus injections of recombinant angiostatin protein is to continuously express the therapeutic transgene systemically by using gene therapy vectors. AAV possesses several advantages for chronic antiangiogenic therapy over other viral vector delivery systems. For example, a single systemic administration of AAV leads to efficient and stable transgene expression in vivo without stimulating an effective cellular immune response or causing vector related toxicities^27,28. In this report we demonstrate the feasibility of using liver-directed recombinant AAV gene transfer of angiostatin as a potential antiangiogenic therapeutic strategy for blocking tumor growth and metastasis.

Recombinant AAV vectors encoding human angiostatin (kringles 1–3) under the control of a ubiquitous promoter were constructed. Kringles 1–3 of angiostatin were expressed rather than the four-kringle version since the three-kringle version appears to be sufficient for angiostatin's tumor suppressor activity²³. Recombinant AAV vectors were able to express human angiostatin efficiently from transduced hepatocytes in culture or following gene delivery in mice. In addition, vector-produced angiostatin possessed potent antiangiogenic properties, suggesting that the expressed kringle domains correctly adopted the native conformational folding necessary for its biological activity. Although a systematic pharmacokinetic comparison of recombinant angiostatin and vector-produced angiostatin proteins in vivo has not been conducted, studies in the CAM assay suggest that AAV-produced angiostatin is biologically more potent at blocking angiogenesis than an equivalent dose of the recombinant angiostatin protein produced from P. pastoris.

A single administration of AAV-angiostatin into the tail or portal vein of mice resulted in high-level expression of the therapeutic protein in the circulation that was sustained at continuous levels for over half a year. Mice with continuous expression of systemic angiostatin showed no adverse health effects or evidence of liver toxicity during the entire experimental time course (6 months). Future studies aimed at incorporating inducible promoter systems into recombinant AAV vectors may offer an additional safety feature in regulating the long-term expression of angiostatin in vivo.

In this report we demonstrate that sustained low-dose expression of angiostatin following AAV gene transfer has a significant impact on delaying tumor growth and increasing the survival of tumor-bearing mice in several preclinical models of experimental metastasis in a preventive setting. Mice pretreated with AAV-angiostatin had a significant delay of LLC or B16 tumor onset in comparison to control-treated mice. Anti-tumor effects following AAV-angiostatin gene delivery were consistently observed when serum levels of angiostatin measured between 5 and 50 ng/ml. Circulating angiostatin levels above and below this range had minimal effects on tumor growth, suggesting that the effective dose required for angiostatin's anti-tumor efficacy in the context of vector-mediated delivery occurs within a narrow range. This is in contrast to preclinical studies demonstrating efficacy using high doses of recombinant angiostatin protein. Prolonged expression of human angiostatin at the doses used here did not result in complete tumor regression or dormancy. However, B16-tumor-bearing mice continuously expressing angiostatin had a threefold survival advantage in comparison to AAV control-treated mice. Preadministration of AAV-angiostatin also prolonged survival in mice, albeit to a lesser degree, in a LLC model of experimental metastasis.

Recently it has been speculated that certain antiangiogenic agents, such as endostatin, may be more efficacious when circulating levels are maintained at lower doses or at dose ranges that follow a bell-shaped curve¹⁶. Several recent in vivo studies have documented that high expression of intravascular endostatin following gene transduction of hematopoietic stem cells failed to demonstrate any antiangiogenic or anti-tumor activity^10,34. In addition, Kuo et al. demonstrated that high expression of angiostatin or endostatin by adenovirus-mediated gene delivery had poor efficacy in several tumor models, including LLC²⁰. One difference that may explain the discrepancy in the results obtained by Kuo et al. and by the current study may be related to the expression profiles of circulating angiostatin characteristic of the different vector platforms used. Recombinant E1-deleted adenoviral vectors, used in the study by Kuo et al., lead to levels of serum angiostatin in the microgram per milliliter range, which decline over time due to immune responses directed against adenoviral vectors. In contrast, angiostatin serum levels are generally much lower following recombinant AAV-mediated gene transfer but are sustained for much longer periods. We have also observed that high expression of serum angiostatin following E1-deleted adenovirus gene transfer did not have any significant effects in prolonging survival or delaying tumor growth in the B16 experimental lung metastasis model (data not shown). At present it is unknown as to why sustained serum angiostatin levels that occur within a very narrow dose range at lower levels result in better tumor efficacy than higher dosages by systemic gene therapy delivery. High expression of endogenous antiangiogenic agents such as endostatin and angiostatin may lead to misfolding or aggregation of proteins, rendering them inactive, although the precise conformational requirements of both proteins are unknown at present²⁵. In addition, higher constitutive concentrations of circulating angiostatin or endostatin may induce desensitization in their respective tumor-endothelial receptors. Unfortunately, the molecular mechanism by which angiostatin exerts its anti-tumor properties in vivo has yet to be clearly defined, and a precise angiostatin receptor has yet to be clearly established^29,30,37.

Ma et al. have also demonstrated anti-tumor efficacy using recombinant AAV-angiostatin gene therapy in two different preclinical glioma tumor models; however, unlike our study, a systematic dose response of AAV-angiostatin was not evaluated^21,22. Intramuscular administration of AAV vectors expressing angiostatin was shown to suppress the growth of human U87MG and rat C6 glioma tumors and resulted in 40% of mice surviving for more than 6 months^21,22. In addition, Xu et al. have recently shown that liver-directed AAV gene transfer of murine angiostatin resulted in significant suppression of liver metastasis and improved survival in a murine EL-4 lymphoma tumor model⁴⁵.

Administration of AAV-angiostatin as a monotherapy was not effective at completely eradicating B16 and LLC tumor growth and only a fraction of mice receiving AAV-angiostatin treatment had a significant survival advantage in a prevention-type setting. Moreover, no therapeutic effects were observed after treating preestablished human tumors using a range of AAV-angiostatin doses in a prostate PC-3 or breast MDA-231 xenograft tumor model in nude mice (data not shown). Combination of cytotoxic therapy with antiangiogenic treatment modalities such as angiostatin should target both the tumor cell and the tumor-associated cell compartments, which may lead to a synergistic anti-tumor response. AAV-angiostatin, in combination with chemotherapy, extended the survival of mice bearing preestablished human tumors in comparison to chemotherapy treatment alone. However, the anti-tumor efficacy of AAV-angiostatin was evident only within a narrow dose range of circulating angiostatin levels. AAV-angiostatin gene delivery resulting in sustained serum levels of human angiostatin of 25 ng/ml resulted in an increase in median survival when combined, or administered as an adjuvant therapy, with chemotherapy in human LS174t colon or MX-1 breast carcinoma-bearing mice. No therapeutic benefit was observed with AAV-angiostatin when sera transgene levels were measured below 5 or above 75 ng/ml in these models. Thus, the use of AAV-angiostatin gene transfer as an effective anti-cancer strategy may be limited given that the window of efficacy of this antiangiogenic agent may occur with a narrow dose range in vivo.

The results presented in this report, as well as recent studies⁴⁵, support the rationale for the use of AAV-mediated gene transfer into the liver as a potentially safe and effective strategy for continuously expressing antiangiogenic agents systemically for cancer therapy. A single systemic injection of AAV-angiostatin resulted in prolonged and continuous expression of circulating angiostatin, which had marked anti-tumor effects in multiple tumor models. However, anti-tumor efficacy was limited to a narrow dose range of sustained circulating angiostatin levels following AAV gene delivery, which appeared to follow a bell-shaped curve. These results may have important implications for optimizing the dose and schedule of angiostatin in future preclinical and clinical studies.

Materials and Methods

Plasmid construction

A 1599-bp fragment containing the kringle domains 1–3 of the human angiostatin gene with the tissue plasminogen activator signal sequence was derived from pcDNA3.1/hAS1-3²³ and cloned together with the woodchuck hepatitis virus posttranscriptional regulatory element (WPRE)⁴⁸ into the EcoRI site of plasmid pSSV9-MD2⁴¹ to create pSSV9-MD2-Angiostatin-Pre. The pSSV9-MD2 expression cassette consists of the cytomegalovirus (CMV) immediate early promoter/enhancer and the intervening sequence 2 and polyadenylation signals from the human -globin gene. The CMV promoter was subsequently replaced with an EF1 promoter²⁶ to create the plasmid pSSV9-EF1a-Angiostatin-Pre. The resulting AAV-angiostatin plasmid, pSSV9-EF1a-Angio-pre, is 4615 bp from both inverted terminal repeats and was used for generating functional recombinant AAV-angiostatin virus particles as described below. Plasmid pTR-EF1-GFP was made from plasmid ptet-EF1-GFP in which the enhanced GFP gene (BamHI to NotI fragment) from plasmid pEGFP (Clontech, Palo Alto, CA) was placed downstream of the EF1 promoter derived from a HindIII/EcoRI digest of pEF-BOS²⁶. To generate plasmid pTR-EF1-GFP, plasmid pTRUF5¹⁹ was digested with KpnI and NotI to remove the CMV-GFP cassette and replace it with a KpnI/NotI fragment of ptetEF1-GFP encompassing the EF1-GFP expression cassette. Plasmid pSSV9-EF1-mAP was generated by removing the angiostatin gene and WPRE from pAAV-EF1-angio by a digestion with SpeI and SalI and replacing it with the human placental alkaline phosphatase gene and the bovine growth hormone polyadenylation signal from pSSV9-MFG-mAP by digestion with XbaI and SalI. The WPRE and -globin poly(A) signal were then removed by a partial XbaI digest followed by religation.

Recombinant AAV vector preparation

Recombinant AAV vectors were prepared according to the methods of Snyder et al. with a few modifications⁴⁰. Briefly, subconfluent human embryonic kidney 293 cells were cotransfected with the angiostatin or GFP vector plasmid and the helper plasmid, pUC-ACG³⁹, using the calcium phosphate method. Eight hours following transfection, cells were infected with adenovirus Addl312 (an E1A^- deletion mutant) at a multiplicity of infection (m.o.i.) of 2, and the infection was allowed to proceed for 72 h. Cells were subsequently lysed by three freeze/thaw cycles, treated with benzonase for 15 min at 37°C, and then centrifuged to remove cellular debris. The clarified cellular lysates were fractionated by ammonium sulfate precipitation and the recombinant AAV virions were isolated on two sequential cesium chloride gradients. Gradient fractions containing AAV were dialyzed against sterile PBS containing 0.9 mM CaCl₂ and 0.5 mM MgCl₂, heated for 10 min at 56°C to inactivate any residual adenovirus, and stored at -80°C. To determine viral titers by dot-blot analysis, the recombinant AAV preps were treated with DNase I to degrade unencapsidated DNA and then treated with proteinase K (0.25 mg/ml) in the presence of 0.5% SDS and 10 mM EDTA to liberate virus genomes. Viral DNA was then denatured in alkali and applied to a nylon membrane. A radioactive probe specific for the human angiostatin transgene was hybridized to DNA on the filter and exposed to film, and the radioactive-bound probe was quantified using a counter (1450 Micobeta Trilux, Perkin–Elmer, Inc., Wellesley, MA). Dilutions of the corresponding vector plasmid were used as standards to determine the recombinant AAV virion copy number.

Cell culture and angiostatin-conditioned medium

The cell lines LLC, HuH7, and 293 were obtained from ATCC (Manassas, VA) and the luciferase-expressing B16F10 (B16F10-luc) melanoma cells were from Xenogen Corp. (Alameda, CA). All cell lines were cultured at 37°C with 5% CO₂ in DMEM high-glucose medium (JRH Biosciences, Lenexa, KS) supplemented with 10% irradiated FCS (JRH Biosciences), 2 mM glutamine (JRH Biosciences), 100 U/ml penicillin, and 100 g/ml streptomycin (Gibco BRL, Rockville, MD).

Generation of a stable luciferase-expressing LLC (LLC-luc) cell line was established by overnight infection using an HIV-1-based lentiviral vector containing the firefly luciferase gene under the control of the human CMV immediate early promoter/enhancer at an m.o.i. of 10. After expansion of the transduced cells in vitro, 1 10⁶ LLC-luc cells were injected into the tail vein of female Balb/c nude mice (Taconic, Germantown, NY) for in vivo selection of luciferase-expressing LLC tumors that seeded in the lungs. Mice were sacrificed 21 days after tumor inoculation, and luciferase-positive lung tumor nodules, as determined by bioluminescence, were aseptically harvested and expanded in vitro to yield a pure population of LLC-luc cells that were used for subsequent tumor studies.

To generate conditioned medium containing human angiostatin, six-well dishes containing subconfluent HuH7 cells were infected with recombinant AAV vectors at an m.o.i. of 1 10⁵ virus particles/cell in 10% FCS containing DMEM. Forty-eight hours after infection, the supernatants were harvested and clarified of cellular debris by centrifugation, and aliquots were frozen at -70°C until use. HuH7 cells were transduced with AAV-GFP or AAV-alkaline phosphatase vectors in a similar fashion for generating control supernatants.

Immunoblotting detection of angiostatin

Recombinant human angiostatin (encompassing kringles 1–3) produced from P. pastoris was provided by EntreMed, Inc. (Rockville, MD). An aliquot of conditioned medium harvested from recombinant AAV-transduced HuH7 cells, or recombinant angiostatin protein, was resolved using NuPage Bis-Tris gels and Mops buffer by 4–12% SDS–PAGE (Invitrogen, Carlsbad, CA). Resolved proteins were transferred onto nitrocellulose for 1 h in 20% methanol-containing transfer buffer (Invitrogen). Membranes were blocked for 1 h in Tris-buffered saline (TBS) containing 3% bovine serum albumin and 0.2% Tween 20 (ICN Pharmaceuticals, Inc., Costa Mesa, CA) and then probed with 0.5 g/ml HRP-conjugated goat anti-human plasminogen antibody for 1 h (Cedarlane Laboratories, Hornby, Ontario, Canada). The blots were washed extensively with TBS–3% BSA and subsequently visualized by enhanced chemiluminescence using the Supersignal substrate (Pierce, Rockford, IL) as per the manufacturer's protocol.

Quantitation of human angiostatin by ELISA

The angiostatin sandwich ELISA was performed using a monoclonal anti-human plasminogen capture antibody (MAb) from Biodesign (Saco, ME) and an HRP-conjugated affinity-purified goat anti-human plasminogen detection antibody from Accurate Chemical (Westbury, NY). Both antibodies to human plasminogen did not cross-react with murine plasminogen. Briefly, 96-well microtiter plates were coated with 1 g/well anti-plasminogen MAb in 0.1 M carbonate pH 9.6 buffer and incubated overnight at 4°C. The plates were washed extensively with PBS–0.05% Tween 20 and blocked with PBS–0.5% BSA–0.1% Tween 20 buffer at room temperature for 1 h. Recombinant human angiostatin protein (kindly provided by EntreMed, Inc.) was used for standard curves after serial dilutions in 5% normal mouse serum (Calbiochem, San Diego, CA). Samples and the standard were incubated for 2 h at room temperature and washed, and the HRP-conjugated plasminogen antibody was then applied at 1.0 g/ml for 45 min at room temperature. Following extensive washing, 100 l of Sure Blue TMB substrate (KPL, Gaithersburg, MD) was added and the reaction was allowed to proceed until the 450/650 nm optical density for the 50 ng/ml angiostatin standard read between 1.5 and 3.0. The sensitivity of angiostatin detection in this assay is approximately 100 pg/ml.

CAM assays

CAM assays were performed as previously described³. Briefly, hydrocortisone-containing filter discs were soaked with 25 ng bFGF (R&D Systems, Minneapolis, MN) together with the indicated doses of recombinant human angiostatin protein or conditioned medium from control or AAV-angiostatin-transduced cells and aseptically placed on top of the chorioallantoic membranes of 11-day-old chick embryos. After 72 h incubation, the CAMs were surgically removed and fixed and the discs were analyzed for bFGF-induced angiogenesis by stereomicroscopy. The mean number (SEM) of branch points occurring in the new blood vessel sprouts beneath the applied filter disc from the treated CAMs was enumerated from multiple (n = 8–10) samples.

Long-term in vivo vector expression study

Six- to eight-week-old female Balb/c nude mice were obtained from Taconic. All mice were housed under SPF conditions and treated according to the ILAR Guide for the Care and Use of Laboratory Animals. Mice (n = 6) were injected with a single dose of 2 10¹¹ virus particles of AAV-angiostatin (100–300 l total volume) diluted in sterile PBS through a portal vein catheter, in the case of hepatic portal vein administration, or via direct tail vein injection for intravenous administration. Mice were bled by alternate retro-orbital puncture on scheduled intervals up to 4 months to measure the serum levels of human angiostatin.

B16F10-luc and LLC-luc models of experimental metastasis

A single dose of AAV-angiostatin or AAV-GFP vectors was injected into the tail vein or portal vein of 6- to 8-week-old Balb/c nude mice (Taconic). Blood samples were harvested by alternate retro-orbital puncture to measure the serum levels of angiostatin on a weekly basis for the duration of the experiment. Approximately 4 weeks after AAV administration, mice were injected with either 1 10⁶ LLC-luc or 5 10⁴ B16F10-luc tumor cells into the tail vein. Mice were monitored at interim time points for luciferase-expressing tumors by in vivo luminescence imaging. Briefly, LLC-luc and B16F10-luc tumor-bearing mice were injected with luciferin substrate (Xenogen Corp., Alameda, CA) at a dose of 1.5 mg/g mouse body weight by intraperitoneal injection. Twenty minutes after substrate injection, mice were anesthetized and the ventral sides of the mice were imaged using the Xenogen IVIS Imaging System luminescence-sensitive CCD camera (Xenogen Corp.). Data were collected and analyzed using Living Image 2.11 software^9,35. For the survival studies, following tumor cell injections, mice were evaluated for tumor burden twice a week by in vivo luminescence and health condition was monitored until sacrifice due to extensive tumor burden. Serum was collected by retro-orbital puncture on scheduled intervals to measure the serum levels of angiostatin. Survival advantage and statistics were quantitated using a Kaplan–Meier survival curve and utilizing GraphPad Prism software. All animal protocols were reviewed and approved by the Cell Genesys Institution Animal Care and Use Committee.

MX-1 breast and LS174t colon carcinoma xenograft models

A 1-mm³ MX-1 breast carcinoma or LS174t colon carcinoma fragment was implanted sc into the flank region of female athymic NCR/Nu nude mice (Charles River). Mice were pair-matched into treatment groups when the tumors reached 100 mm³ (considered day 0) and treated with the indicated chemotherapy regiments. For the MX-1 adjuvant study, mice were treated with a single ip injection of Cytoxan (Neosar, Pharmacia & Upjohn Co.) at a dose of 120 mg/kg at day 1 to cause tumor debulking. On day 7, mice received a single tail-vein injection of AAV-angiostatin vector as an adjuvant therapy. For the LS174t studies, pair-matched mice bearing 100-mm³ established tumors were treated with ip injections of irinotecan/CPT-11 (Camptosar, Pharmacia & Upjohn Co.) once a week for 3 weeks. On day 3, mice received a single iv injection of AAV-angiostatin vectors at the doses indicated. Animals were euthanized as a "cancer death" when the tumors reached or exceeded a tumor volume of 1500 mm³, and survival advantage is presented as a Kaplan–Meier survival curve using GraphPad Prism software. During the study, animals were weighed twice weekly and examined frequently for clinical signs of any adverse, drug-related side effects.

Statistical analysis and data presentation

The figures show data from a single representative experiment or the means SEM of data pooled from "n" independent experiments (data normalized as described in the figure legends). Data from the CAM assay are reported in the text as means SEM and statistical analysis was by two-tailed Student's t test, accepting P < 0.05 as statistically significant. Multiparameter statistics for the Kaplan–Meier survival curves were performed by a log-rank test using GraphPad Prism software.

References

1. Athanasopoulos, T., Fabb, S. and Dickson, G. (2000). Gene therapy vectors based on adeno-associated virus: characteristics and applications to acquired and inherited diseases. Int. J. Mol. Med.. 6: 363–375. | PubMed | ChemPort |
2. Bergers, G., Javaherian, K., Lo, K. M., Folkman, J. and Hanahan, D. (1999). Effects of angiogenesis inhibitors on multistage carcinogenesis in mice. Science. 284: 808–812. | Article | PubMed | ISI | ChemPort |
3. Brooks, P. C., Montgomery, A. M. P. and Cheresh, D. A. (1999). Use of the 10-day old chick embryo model for studying angiogenesis. In: Howlett A. R. (Ed). Methods in Molecular Biology: Integrin Protocols. vol. 129. Human Press: Totowa, NJ. 257–269.
4. Cao, Y. (1999). Therapeutic potentials of angiostatin in the treatment of cancer. Haematologica. 84: 643–650. | PubMed | ChemPort |
5. Carmeliet, P. and Jain, R. K. (2000). Angiogenesis in cancer and other diseases. Nature. 407: 249–257. | Article | PubMed | ISI | ChemPort |
6. Drixler, T. A., Rinkes, I. H., Ritchie, E. D., van Vroonhoven, T. J., Gebbink, M. F. and Voest, E. E. (2000). Continuous administration of angiostatin inhibits accelerated growth of colorectal liver metastases after partial hepatectomy. Cancer Res.. 60: 1761–1765. | PubMed | ISI | ChemPort |
7. Dutheil, N., Shi, F., Dupressoir, T. and Linder, R. M. (2000). Adeno-associated virus site-specifically integrates into a muscle specific DNA region. Proc. Natl. Acad. Sci. USA. 97: 4862–4866. | Article | PubMed | ChemPort |
8. Eatock, M. M., Schatzlein, A. and Kaye, S. B. (2000). Tumour vasculature as a target for anticancer therapy. Cancer Treat. Rev.. 26: 191–204. | Article | PubMed | ISI | ChemPort |
9. Edinger, M., et al. (2002). Advancing animal models of neoplasia through in vivo bioluminescence imaging. Eur. J. Cancer. 38: 2128–2136. | Article | PubMed | ISI | ChemPort |
10. Eisterer, W., et al. (2002). Unfulfilled promise of endostatin in a gene therapy–xenotransplant model of human acute lymphocytic leukemia. Mol. Ther.. 5: 352–359. | Article | PubMed | ISI | ChemPort |
11. Folkman, J. (2002). Role of angiogenesis in tumor growth and metastasis. Semin. Oncol.. 29: 15–18. | Article | PubMed | ISI | ChemPort |
12. Folkman, J. (2000). Tumor angiogenesis. In: Holland J. F. (Ed). Cancer Medicine. 5th ed. Decker: Hamilton, Ontario, Canada. 132–152.
13. Hallek, M. and Wendtner, C. M. (1996). Recombinant adeno-associated virus (rAAV) vectors for somatic gene therapy: recent advances and potential clinical applications. Cytokines Mol. Ther.. 2: 69–79. | PubMed | ChemPort |
14. Hanahan, D. and Folkman, J. (1996). Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell. 86: 353–364. | Article | PubMed | ISI | ChemPort |
15. Hanahan, D. and Weinberg, R. A. (2000). The hallmarks of cancer. Cell. 100: 57–70. | Article | PubMed | ISI | ChemPort |
16. Kerbel, R. and Folkman, J. (2002). Clinical translation of angiogenesis inhibitors. Nat. Rev. Cancer. 2: 727–739. | Article | PubMed | ISI | ChemPort |
17. Kerbel, R. S. (1991). Inhibition of tumor angiogenesis as a strategy to circumvent acquired resistance to anti-cancer therapeutic agents. Bioessays. 13: 31–36. | Article | PubMed | ISI | ChemPort |
18. Kirsch, M., Strasser, J., Allende, R., Bello, L., Zhang, J. and Black, P. M. (1998). Angiostatin suppresses malignant glioma growth in vivo. Cancer Res.. 58: 4654–4659. | PubMed | ISI | ChemPort |
19. Klein, R. L., et al. (1998). Neuron-specific transduction in the rat septohippocampal or nigrostriatal pathway by recombinant adeno-associated virus vectors. Exp. Neurol.. 150: 183–194. | Article | PubMed | ISI | ChemPort |
20. Kuo, C. J., et al. (2001). Comparative evaluation of the antitumor activity of antiangiogenic proteins delivered by gene transfer. Proc. Natl. Acad. Sci. USA. 98: 4605–4610. | Article | PubMed | ChemPort |
21. Ma, H. I., et al. (2002). Suppression of intracranial human glioma growth after intramuscular administration of an adeno-associated viral vector expressing angiostatin. Cancer Res.. 62: 756–763. | PubMed | ISI | ChemPort |
22. Ma, H. I., et al. (2002). Intratumoral gene therapy of malignant brain tumor in a rat model with angiostatin delivered by adeno-associated viral (AAV) vector. Gene Ther.. 9: 2–11. | Article | PubMed | ChemPort |
23. MacDonald, N. J., Murad, A. C., Fogler, W. E., Lu, Y. and Sim, B. K. (1999). The tumor-suppressing activity of angiostatin protein resides within kringles 1 to 3. Biochem. Biophys. Res. Commun.. 264: 469–477. | Article | PubMed | ChemPort |
24. Malik, A. K., Monahan, P. E., Allen, D. L., Chen, B. G., Samulski, R. J. and Kurachi, K. (2000). Kinetics of recombinant adeno-associated virus-mediated gene transfer. J. Virol.. 74: 3555–3565. | Article | PubMed | ChemPort |
25. Marshall, E. (2002). Cancer therapy: setbacks for endostatin. Science. 295: 2198–2199. | Article | PubMed | ChemPort |
26. Mizushima, S. and Nagata, S. (1990). pEF-BOS, a powerful mammalian expression vector. Nucleic Acids Res.. 18: 5322. | PubMed | ISI | ChemPort |
27. Monahan, P. E., Jooss, K. and Sands, M. S. (2002). Safety of adeno-associated virus gene therapy vectors: a current evaluation. Expert Opin. Drug Saf.. 1: 79–91. | Article | PubMed | ChemPort |
28. Monahan, P. E. and Samulski, R. J. (2000). Adeno-associated virus vectors for gene therapy: more pros than cons? Mol. Med. Today. 6: 433–440. | Article | PubMed | ISI | ChemPort |
29. Moser, T. L., et al. (1999). Angiostatin binds ATP synthase on the surface of human endothelial cells. Proc. Natl. Acad. Sci. USA. 96: 2811–2816. | Article | PubMed | ChemPort |
30. Moser, T. L., Stack, M. S., Wahl, M. L. and Pizzo, S. V. (2002). The mechanism of action of angiostatin: can you teach an old dog new tricks? Thromb. Haemostasis. 87: 394–401. | ChemPort |
31. Muzyczka, N. and Berns, K. I. (2001). Parvoviridae: the viruses and their replication. In: Knipe D., Howley P., Griffin D., Lamb R., Martin M., Roizman B., and Straus S. (Eds). Fields Virology. vol. 2. Lippincott Williams & Wilkins: Philadelphia. 2327–2347.
32. O'Reilly, M. S., Holmgren, L., Chen, C. and Folkman, J. (1996). Angiostatin induces and sustains dormancy of human primary tumors in mice. Nat. Med.. 2: 689–692. | Article | PubMed | ChemPort |
33. O'Reilly, M. S., et al. (1994). Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell. 79: 315–328. | Article | PubMed | ChemPort |
34. Pawliuk, R., Bachelot, T., Zurkiya, O., Eriksson, A., Cao, Y. and Leboulch, P. (2002). Continuous intravascular secretion of endostatin in mice from transduced hematopoietic stem cells. Mol. Ther.. 5: 345–351. | Article | PubMed | ISI | ChemPort |
35. Rehemtulla, A., et al. (2000). Rapid and quantitative assessment of cancer treatment response using in vivo bioluminescence imaging. Neoplasia. 2: 491–495. | Article | PubMed | ISI | ChemPort |
36. Shepard, S. R., et al. (2000). Large-scale purification of recombinant human angiostatin. Protein Expression Purif.. 20: 216–227. | Article | ChemPort |
37. Sim, B. K., MacDonald, N. J. and Gubish, E. R. (2000). Angiostatin and endostatin: endogenous inhibitors of tumor growth. Cancer Metastasis Rev.. 19: 181–190. | Article | PubMed | ISI | ChemPort |
38. Sim, B. K., et al. (1997). A recombinant human angiostatin protein inhibits experimental primary and metastatic cancer. Cancer Res.. 57: 1329–1334. | PubMed | ISI | ChemPort |
39. Snyder, R. O. (1999). Adeno-associated virus-mediated gene delivery. J. Gene Med.. 1: 166–175. | Article | PubMed | ChemPort |
40. Snyder, R. O. and Flotte, T. R. (2002). Production of clinical-grade recombinant adeno-associated virus vectors. Curr. Opin. Biotechnol.. 13: 418–423. | Article | PubMed | ISI | ChemPort |
41. Snyder, R. O., et al. (1997). Persistent and therapeutic concentrations of human factor IX in mice after hepatic gene transfer of recombinant AAV vectors. Nat. Genet.. 16: 270–276. | Article | PubMed | ISI | ChemPort |
42. Soff, G. A. (2000). Angiostatin and angiostatin-related proteins. Cancer Metastasis Rev.. 19: 97–107. | Article | PubMed | ChemPort |
43. Tennnat, T. R., Rinker-Schaeffer, C. W. and Stadler, W. M. (2000). Angiogenesis inhibitors. Curr. Oncol. Rep.. 2: 11–16. | PubMed |
44. Wu, Z., O'Reilly, M. S., Folkman, J. and Shing, Y. (1997). Suppression of tumor growth with recombinant murine angiostatin. Biochem. Biophys. Res. Commun.. 236: 651–654. | Article | PubMed | ISI | ChemPort |
45. Xu, R., et al. (2003). Long-term expression of angiostatin suppresses metastatic liver cancer in mice. Hepatology. 37: 1451–1460. | Article | PubMed | ISI | ChemPort |
46. Yancopoulos, G. Y., Davis, S., Gale, N. W., Rudge, J. S., Wiegand, S. J. and Holash, J. (2000). Vascular-specific growth factors and blood vessel formation. Nature. 407: 242–248. | Article | PubMed | ISI | ChemPort |
47. Yokoyama, Y., Dhanabal, M., Griffioen, A. W., Sukhatme, V. P. and Ramakrishnan, S. (2000). Synergy between angiostatin and endostatin: inhibition of ovarian cancer growth. Cancer Res.. 60: 2190–2196. | PubMed | ISI | ChemPort |
48. Zufferey, R., Donello, J. E., Trono, D. and Hope, T. J. (1999). Woodchuck hepatitis virus posttranscriptional regulatory element enhances expression of transgenes delivered by retroviral vectors. J. Virol.. 73: 2886–2892. | PubMed | ISI | ChemPort |

Acknowledgements

We thank M. Gebreyesus, G. Huan-Tu, and A. Lam for AAV vector preparation and Y. Ge for generating the LLC-luc cell line. We also thank J. Brodsky and E. Rimmer for assistance in performing ELISAs and the Cell Genesys Animal Services Group for their excellent technical assistance in the animal studies. We also thank Piedmont Research Center for assistance with the combination studies in the xenograft tumor models. We also acknowledge W. Fogler and B. Kim Lee Sim (EntreMed, Inc.), for angiostatin reagents, and D. Ando, K. Hege, and P. K. Working for critically reading the manuscript.