Combination Angiostatin and Endostatin Gene Transfer Induces Synergistic Antiangiogenic Activity in Vitro and Antitumor Efficacy in Leukemia and Solid Tumors in Mice

ARTICLE

doi:10.1006/mthe.2000.0243, available online at http://www.idealibrary.com on IDEAL

Combination Angiostatin and Endostatin Gene Transfer Induces Synergistic Antiangiogenic Activity in Vitro and Antitumor Efficacy in Leukemia and Solid Tumors in Mice Frank A. Scappaticci,* Richard Smith,† Anjali Pathak,‡ Derrick Schloss,‡ Bert Lum,‡ Yihai Cao,§ Frances Johnson,‡ Edgar G. Engleman,* and Garry P. Nolan†,1 *Department of Pathology, †Department of Molecular Pharmacology, and ‡Department of Medicine, Stanford University Medical Center, Stanford, California 94305 § Laboratory of Angiogenesis Research, Microbiology and Tumor Biology Center, Karolinska Institute, S-171 77 Stockholm, Sweden Received for publication November 1, 2000; accepted in revised form December 15, 2000

Angiostatin and endostatin are potent endothelial cell growth inhibitors that have been shown to inhibit angiogenesis in vivo and tumor growth in mice. However, tumor shrinkage requires chronic delivery of large doses of these proteins. Here we report synergistic antitumor activity and survival of animals when these factors are delivered in combination to tumors by retroviral gene transfer. We have demonstrated this efficacy in both murine leukemia and melanoma models. Complete loss of tumorigenicity was seen in 40% of the animals receiving tumors transduced by the combination of angiostatin and endostatin in the leukemia model. The synergy was also demonstrated in vitro on human umbilical vein endothelial cell differentiation and this antiangiogenic activity may suggest a mechanism for the antitumor activity in vivo. These findings imply separate pathways by which angiostatin and endostatin mediate their antiangiogenic effects. Together, these data suggest that a combination of antiangiogenic factors delivered by retroviral gene transfer may produce synergistic antitumor effects in both leukemia and solid tumors, thus avoiding long-term administration of recombinant proteins. The data also suggest that novel combinations of antiangiogenic factors delivered into tumors require further investigation as therapeutic modalities. Key Words: angiogenesis inhibitor; gene therapy; leukemia; melanoma; endothelial cells.

INTRODUCTION Antiangiogenic approaches to cancer treatment represent a promising new strategy that may not allow for drug resistance of tumors (1). Drug resistance has been one of the major factors in the failure of standard chemotherapeutic agents to eradicate tumors (2). Since tumors rely on angiogenesis for growth and metastasis, inhibition of this process may indirectly inhibit tumor cell growth and lead to tumor dormancy (3). Angiostatin and endostatin are fragments of larger precursor molecules (plasminogen and collagen XVIII, respectively) and are generated via proteases secreted by tumor cells (3, 4). When administered to mice with a variety of tumors (B16F10 melanoma,

1 To whom correspondence and reprint requests should be addressed at Department of Molecular Pharmacology, Stanford University School of Medicine, 269 Campus Drive, CCSR 3220, Stanford, CA 94305-5332. Fax: (650) 725-2952. E-mail: [email protected].

186

T241 fibrosarcoma, and Lewis lung carcinoma), these proteins can induce tumor regression (2). These angiogenesis inhibitors do not appear to have any direct cytotoxic effects on tumor cells since growth is not affected by exposure to these proteins in culture (3). The mechanism responsible for the antiangiogenic activity of these proteins in vivo is not known but may be related to inhibition of cell cycle progression of endothelial cells or induction of apoptosis (5, 6). Although angiostatin and endostatin have been produced as recombinant proteins, their longterm delivery to tumors in the host may pose difficult pharmacologic problems (7). Delivery of the corresponding genes to tissues of the host may overcome these problems (7–9). The role of angiogenesis in the pathophysiology of solid tumors such as breast cancer, melanoma, glioblastoma, and others is becoming increasingly understood (10). Recent reports have suggested a role for angiogenesis in the survival and progression of malignant cells in various MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy 1525-0016/01 $35.00

ARTICLE leukemias (11). In particular, an increase in microvessel density in bone marrow specimens has been reported in patients with acute lymphocytic leukemia, acute myelogenous leukemia, chronic lymphocytic leukemia (CLL), chronic myelogenous leukemia, myelodysplastic syndromes, and multiple myeloma (11–19). In addition, increased levels of angiogenic factors such as basic fibroblast growth factor (bFGF) have been found in the urine of patients with CLL. In one study, increased levels of serum vascular endothelial growth factor (VEGF) correlated with increased risk for disease progression in CLL (20). These observations provide a rationale for evaluating antiangiogenic drugs in the treatment of leukemia (11). Synergy between angiostatin and endostatin was suggested by a recent report that showed potent antitumor activity of the combination in an ovarian cancer model in nude mice (21). In this report, we describe the synergistic effects on angiogenesis in vitro mediated by combination gene transfer of angiostatin and endostatin. Delivery of this combination by retroviral transfer also resulted in a synergistic effect on tumor growth and survival of animals with leukemia and solid tumors. This report demonstrates the potential role of these angiogenesis inhibitors on the progression of leukemia. The results suggest that gene transfer is an effective means of delivering angiostatin and endostatin and that the combination of these proteins may be synergistic.

tant with 2 ml of tumor cells (1 ⫻ 106) in the presence of 6 ␮g/ml Polybrene. The medium was changed 24 h postinfection. The cells were cultured in DMEM ⫹10% FBS ⫹ 1% PSG for 4 to 7 days prior to flow cytometry. Prior to in vivo studies, the infected cells were sorted by flow cytometry using a FACStar Plus cell sorter (Becton–Dickinson) set at 488-nm excitation wavelength. Gates were set to sort the highest 25% of fluorescent cells and the cells were collected in PBS.

MATERIALS

Matrigel assays. Growth factor-reduced Matrigel was obtained from Becton–Dickinson. The Matrigel was allowed to solidify for 1 h in serum-free medium (EBM from Clonetics) in the wells of six-well culture plates. Primary HUVEC were obtained using purification methods previously described (25). HUVEC were cultured in EBM supplemented with EGM. The endothelial cells (2 ⫻ 105 cells/well) were plated on Matrigel with 2.5 volumes of 24-h filtered viral supernatants harvested from B16F10 melanoma cells expressing angiostatin or endostatin. There was no exogenous supplementation with VEGF or bFGF. Approximately 24 h later, the cells were analyzed under a light microscope for endothelial cell tubes that form junctions (26). A dose–response effect of the combination of angiostatin and endostatin was assessed using dilutions of the supernatants in medium and counting the number of endothelial cell tubes 24 h after plating.

AND

METHODS

Construction of retroviral vectors. The Lazarus (LZRS)-based Moloney murine leukemia viral (MMuLV) vectors were constructed by cloning the cDNA fragment of mouse angiostatin (7) or mouse endostatin (1) (generous gifts from Y. Cao and T. Boehm, respectively) into the BglII/XhoI or BamHI sites, respectively, upstream of the internal ribosomal entry site (IRES) sequences of the LZRS IRES-GFP plasmid (22). The 5⬘ signal sequences that allow for secretion of angiostatin and endostatin were derived from plasminogen or collagen XVIII, respectively. In addition, a hemagglutinin tag (HA) was cloned into the 3⬘ end of the angiostatin gene. These vectors contain IRES– green fluorescence protein (GFP) sequences and have the advantage of (a) having a GFP marker for selection of transduced cells with the highest GFP expression and consequently the highest expression of angiostatin or endostatin, (b) determining the efficiency of transduction, and (c) the presence of Epstein–Barr viral nuclear antigen sequences that allow these plasmids to remain as stable episomes and produce high-titer virus for several months. Human embryonic kidney cells transformed with the SV40 T antigen and expressing all the viral components necessary for assembly and packaging (i.e., Phoenix cells, derived from 293T cells) were used as packaging cells (23). When transfected into Phoenix amphotropic cells, these vectors produce replication-defective virus with a titer of 1–5 ⫻106 as determined by GFP expression of infected NIH 3T3 fibroblasts (data not shown). Transfection/transduction of tumor cells and FACS. Phoenix cells were grown in DMEM with 10% FBS ⫹ 1% penicillin/streptomycin/glutamine (PSG) and transfected with 10 ␮g of the above retroviral vectors using standard calcium phosphate precipitation methods (22) in 6-cm petri dishes (3 ml total volume). The medium was discarded 24 h later and replaced with fresh medium. Viral supernatant was recovered 24 h later, filtered through a 0.4-␮m filter (Millipore), and immediately used for infection of B16F10 melanoma cells and L1210 leukemia cells (these cell lines were obtained from American Type Culture Collection). Alternatively, the viral supernatant was used for transduction of NIH 3T3 cells for standard titer determination. The titer was determined to be 1–5 ⫻ 106 viral particles per milliliter by FACS analysis for GFP expression. Transduction of tumor cells was performed by incubation of 1 ml of viral supernaMOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

In vitro cell growth assays. Retrovirally transduced L1210 leukemia or B16F10 melanoma cells were selected for GFP expression by flow cytometry. These cell lines were grown in DMEM ⫹ 10% FBS ⫹ 1% PSG. The L1210 cells were grown over a period of 4 days and counted using a hemacytometer on days 0, 2, and 4. Alternatively, cells were assessed for proliferation using a standard 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; thiazolyl blue (MTT) assay in B16F10 melanoma studies (24). Immunoprecipitation and Western blotting. Western blots were performed on protein that was immunoprecipitated from cell lysates or supernatants of retrovirally transduced B16F10 cells. Immunoprecipitation was performed overnight at 4°C with protein A/G Sepharose beads (Santa Cruz Biotechnology). Protein was loaded onto an SDS–PAGE gel and transferred onto PVDF membrane (Millipore). To detect endostatin protein, Western blotting was performed using a polyclonal rabbit anti-murine endostatin antibody (1:1000) as the primary antibody (gift from T. Boehm) and horseradish peroxidase (HRP)-labeled goat anti-rabbit antibody (Santa Cruz Biotechnology) as the secondary antibody (1:5000). The anti-murine endostatin antibody was used for immunoprecipitation of endostatin from B16F10 supernatants prior to SDS–PAGE/Western blotting. For detection of angiostatin protein, a murine anti-HA monoclonal antibody (1:1000; Berkeley Antibody Co.) which was used for immunoprecipitation served as the first antibody and an HRP-linked polyclonal rabbit antimouse antibody (1:10,000; Santa Cruz Biotechnology) was the second antibody. Bound protein was detected using chemiluminescence with emission wavelength 428 nm (Amersham’s ECL system).

Mouse tumorigenicity studies. For B16F10 melanoma studies, C57BL/6 female mice (6 – 8 weeks of age; The Jackson Laboratory, Bar Harbor, ME) were injected subcutaneously on the left flank with 200 ␮l volume of cells that had been selected for high expression by flow cytometry. Approximately 5 ⫻ 104 cells in PBS were injected per mouse. Each mouse in the combination group received 2.5 ⫻ 104 cells expressing angiostatin and 2.5 ⫻ 104 cells expressing endostatin (i.e., angiostatin/endostatin ratio 1:1 based on GFP comparison after FACS). Tumors were measured with calipers in three dimensions approximately every 3– 4 days and the product was recorded as volume (cm3). For L1210 leukemia studies, Balb/C nude female mice (6 – 8 weeks of age, The Jackson Laboratory) were inoculated ip with 1 ⫻ 103 L1210 cells (200 ␮l volume) after resuspension in PBS immediately after FACS of transduced cells. The combination group received 5 ⫻ 102 angiostatin-transduced cells ⫹ 5 ⫻ 102 endostatin-transduced cells. Immunohistochemistry and analysis of GFP expression. Expression of angiostatin and endostatin was determined in situ using tumors taken from C57BL/6 mice 4 weeks after subcutaneous inoculation. The tumors were paraffin embedded and 5-␮m-thick sections were made for immunohistochemistry. Prior to use, the sections were deparaffinized using various gradients of ethanol from 100 to 30%. For these studies, the primary antibodies used were a rabbit polyclonal anti-HA-tag antibody at 1:1000

187

ARTICLE dilution and a rabbit anti-endostatin serum (1:1000) for analysis of angiostatin and endostatin expression, respectively. The secondary antibody was an HRP-linked goat anti-rabbit polyclonal serum (1:500) with DAB as the substrate. The sections were counterstained with hematoxylin/eosin (VWR) and then viewed under a light microscope. GFP expression was also determined in B16F10 melanoma tumors taken from C57BL/6 mice 4 weeks after subcutaneous inoculation of single suspension cells. The tumors were dissected from the subcutaneous injection site and immediately frozen with OCT cryopreservative by submerging in a cold 2-methylbutane bath made cold by liquid nitrogen. Sections measuring 5 ␮m in thickness were made in a cryostat, mounted directly on glass slides, and then analyzed for fluorescence. Assessment of tumor growth rates and survival. Tumor growth rates in vitro and in vivo were compared using exponential regression analysis (27). Since a Student t test was not used for analysis, error bars for the growth curves in vitro and in vivo are not shown. Convex curvilinear tumor cell growth curves were fit to the equation tumor assessment ⫽ A ⫻ exp (K ⫻ T). In this equation, tumor assessment is the measured tumor volume or tumor cell number, A is the initial baseline value, K is the rate of tumor growth, and T is time. In these analyses, the estimated parameter is K and its associated 95% confidence interval. The values for K with each treatment (LZRS-GFP, angiostatin, endostatin, and the combination of angiostatin and endostatin) were compared. For concave curvilinear tumor cell growth curves, the exponential equation used was tumor assessment ⫽ A ⫻ [1 ⫺ exp(⫺K ⫻ T)]. In this equation, tumor assessment is the measured tumor cell number, A is the mean asymptotic (maximal) value of cell number, K is the rate of tumor growth, and T is time. In these analyses, the estimated parameter is K and its associated 95% confidence interval. The values for K with each treatment were compared. Nonoverlapping 95% confidence intervals of the regression parameter (K) are statistically significant at P ⬍ 0.05. Survival of C57BL/6 mice was analyzed by the Fischer exact test.

RESULTS Expression of Angiostatin and Endostatin in Tumor Cells MMuLV vectors containing the genes for angiostatin and endostatin were constructed as described under Materials and Methods and depicted in Fig. 1a. As shown in Fig. 1b, the secretory forms of angiostatin and endostatin are present in the supernatants of B16F10 melanoma cells infected with retrovirus. The protein sizes are similar to what we have seen when these proteins were expressed in NIH 3T3 cells, transformed human embryonic kidney cells (293T), and L1210 leukemia cells, indicating that there are no significant differences in posttranslational processing between these divergent cell types. Prior to analysis of expression of these inhibitors, the cells were enriched for expression of the surrogate indicator, GFP, by flow cytometry. On the Western blots shown here, a 58-kDa band is seen directly above the immunoglobulin heavy chain band in the supernatants and cell lysates of B16F10 melanoma cells transduced with LZRS-angiostatin retrovirus. Although the molecular weight of angiostatin is approximately 38 kDa, a higher molecular weight has been reported and may reflect the presence of the preactivation peptide and the HA tag or posttranslational modification (28). A 20-kDa band corresponding to the molecular weight of endostatin is also seen in the supernatants of B16F10 cells transduced with endostatincontaining retrovirus. A final concentration of approximately 50 nM angiostatin and endostatin was estimated

188

to be present in the supernatants of B16F10 cells under the conditions described. In Fig. 2, two populations of cells are revealed (an untransduced population shown on the left and a population of cells expressing GFP shown on the right) and confirmed by fluorescence microscopy. Mock-transduced cells revealed only one peak similar to the peak corresponding to untransduced cells in the other groups. Cells were sorted for the highest (25%) GFP expression for subsequent studies in vitro and in vivo. Expression of GFP could be maintained long term in sorted populations (⬎8 weeks).

Inhibition of Endothelial Cell Differentiation by Angiostatin and Endostatin A Matrigel assay was used to explore the possible antiangiogenic effects of these inhibitors. In this assay tube formation is a function of differentiation of endothelial cells. Prior to being plated on Matrigel, primary HUVECS were incubated with supernatants from conditioned medium of retrovirally transduced B16F10 melanoma cells and kept in this medium throughout the remainder of the assay. The results demonstrate no effect on tube formation by angiostatin or endostatin alone compared with mock or viral control supernatants (Fig. 3a). Incubation of HUVECS in EGM also induced tube formation similar to mock and viral controls (data not shown). The combination of angiostatin and endostatin supernatants (1:1) led to inhibition of tube formation. The cells appeared clustered and methods to recover these cells from Matrigel to determine apoptotic and cell cycle effects have not been successful (data not shown). These results are consistent with a strong synergistic effect of angiostatin and endostatin with regard to an in vitro antiangiogenic effect. Previous reports using endostatin in collagen gels showed support of tubular morphogenesis of murine brain endothelial cells (29). We performed dose–response experiments to assess the potency of the antiangiogenic effect of the combination of angiostatin and endostatin (Fig. 3b). The results demonstrate that restoration of tube formation is achieved using one-fourth dilution of the combination of angiostatin and endostatin but not one-half dilution. Compared with control, the inhibition of endothelial cell tube formation was significant at 1:2 and 1:1, P ⬍ 0.0006 and P ⬍ 0.0012, respectively.

Angiostatin and Endostatin Inhibit B16F10 Tumor Growth in Vivo but Not in Vitro Prior to assessing the tumorigenicity of transduced tumor cells in mice, it was critical to determine that expression of these inhibitors did not result in differences in the growth curves of cells in vitro. In particular, a slower growth rate of angiostatin- or endostatin-expressing cells in vitro could be misinterpreted as due to an antiangiogenic effect in vivo. As shown in Fig. 4, there was no significant growth advantage of the B16F10 cells expressing angiostatin, endostatin, or the combination compared with GFP alone. MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

ARTICLE

FIG. 1. MMuLV vectors used for secretion of angiostatin and endostatin. (a) Schematic representation of antiangiogenic retroviral constructs. Shown here are LZRS IRES-GFP-based retroviral vectors alone (top) or containing the cDNAs for murine angiostatin (middle) or murine endostatin (bottom). The secretory signals (SS) for these genes are from the native plasminogen or collagen XVIII genes for angiostatin and endostatin, respectively. A preactivation peptide sequence (PA) is located 3⬘ of the secretory signal and immediately preceding the angiostatin gene. A short HA tag sequence was added to the 3⬘ end of the angiostatin gene for immunologic detection of the secreted protein. Transfection of Phoenix amphotropic packaging cells with these plasmids leads to production of replication-defective retrovirus capable of infecting various murine tumor cell types. (b) Left: Expression of angiostatin in cell lysates and supernatants of transduced B16F10 cells by Western blotting analysis. For detection of angiostatin protein, a mouse anti-HA monoclonal antibody served as the first antibody and an HRP-linked polyclonal rabbit anti-mouse antibody was the second antibody. Western blotting analysis was performed on protein that was immunoprecipitated from cell lysates or supernatants of retrovirally transduced B16F10 melanoma cells. Right: To detect endostatin protein, Western blotting was performed using a polyclonal rabbit anti-mouse endostatin antibody as the primary antibody and an HRP-labeled goat anti-rabbit antibody as the secondary antibody. Left lane shows supernatant from endostatin-transduced B16F10 cells. Right lane represents supernatant taken from LZRS IRES-GFP-transduced cells. Chemiluminescence detection of bound protein is shown on these blots.

To determine in vivo effects of these inhibitors, C57BL/6 mice were injected subcutaneously with B16F10 melanoma cells in order to measure localized tumor growth (Fig. 5a) as well as median survival (Fig. 5b) as a reflection of distant metastases. At the end of 28 days, all surviving animals were sacrificed for immunohistochemistry studies at a synchronous time interval. All surviving animals at 28 days had tumors, but the groups receiving melanoma cells transduced with the combination of angiostatin and endostatin had considerably smaller tumors compared with the other groups. Compared with LZRS IRESGFP-transduced control tumors, there was a statistically significant difference in tumor growth in the angiostatin, endostatin, or combination groups. There was a synergistic benefit for the combination group of having smaller tumors than either the angiostatin- or the endostatinalone group and this was statistically significant (P ⬍ 0.05). MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

The most favorable effect on median survival was seen in the combination group (27 days versus 21 or 19 days in the angiostatin and endostatin groups, respectively). Since surviving animals were sacrificed for immunohistochemistry studies on day 28, average survival was not determined. However, there was an advantage for the combination group on day 27, when 60% of the animals were alive compared with 40% in the angiostatin- or endostatin-alone group and 0% in the control group. In the mock-transduced B16F10 melanoma group, the tumors never grew beyond a small size with the relatively rapid death of the animals within 22 days (data not shown). It has previously been reported that this tumor metastasizes to the lungs (30). The more aggressive nature of these untransduced tumors may be secondary to lack of an immune response against foreign transgenes such as GFP (31). In Fig. 6 expression of angiostatin, endostatin, and GFP

189

ARTICLE

FIG. 2. Flow cytometric analysis of fluorescent B16F10 cells. B16F10 cells that were mock infected (a) were compared with cells infected with retrovirus expressing (b) IRES-GFP alone, (c) angiostatin IRES-GFP, or (d) endostatin IRES-GFP by FACS analysis for GFP expression. A distinct peak representing GFP-expressing cells is seen on the right. Gates were set to sort the highest 25% of fluorescent cells.

is shown on the resected B16F10 tumors at 4 weeks. Expression is shown in situ using an HRP-linked immunoassay directed against the HA tag for angiostatin and the whole protein in the case of endostatin. Expression of angiostatin and endostatin in these tumors appeared patchy with a high degree of heterogeneity among tumor sections. This degree of patchiness may be related to expression of these factors by certain clones of cells within a tumor or, alternatively, these secreted factors could be bound by extracellular matrix differently in various sections of the tumor. Diffuse GFP expression with varying intensities of fluorescence was also demonstrated in these two groups by comparison with a group of animals inoculated with control retrovirally transduced tumors lacking the GFP gene. By both gross inspection of tumors and histologic analysis, there was considerable heterogeneity in the vascularity despite significant size differences of tumors between the two groups. This heterogeneity was evident not only among different groups but within individual tumors as well (data not shown). Because of this degree of heterogeneity, comparison between groups using quantitative methods such as microvessel density analysis was not performed. Similarly, determination of

190

VEGF and bFGF levels may not correlate with degree of angiogenesis within tumors (32).

Enhanced Survival of Nude Mice Bearing Retrovirally Transduced L1210 Leukemia L1210 cells were transduced with retrovirus expressing GFP alone, angiostatin, and endostatin; sorted for GFP expression by flow cytometry; and subsequently tested for in vitro cell growth. From Fig. 7a it is apparent that L1210 cells transduced with LZRS-GFP, angiostatin, endostatin, or the combination grow in a robust manner without any statistically significant differences among the four groups. These cells were then tested for their ability to establish leukemia in Balb/C nude mice via ip injection. Figure 7b reveals survival differences among the groups and shows the synergistic effect on survival in the group receiving cells transduced with angiostatin in combination with cells transduced with endostatin. Leukemic growth was evidenced by distended abdomens, bulky palpable masses, and occasional tumors at the subcutaneous region of the injection site. Bloodstream infiltration of leukemic cells was not seen using this dose of 1 ⫻ 103 cells by ip route of MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

ARTICLE

FIG. 3. (a) Inhibition of endothelial cell differentiation in vitro by angiostatin- and endostatin-containing supernatants. One volume of supernatant (1 ml) from retrovirally transduced B16F10 melanoma cells (106) was incubated with primary HUVECs (2 ⫻ 105) and analyzed for differentiation via tube formation on Matrigel. (A) Mock-transduced cells, (B) GFP, (C) angiostatin, (D) endostatin, and (E) angiostatin ⫹ endostatin (0.5 volume of each). Similar results were seen from two other experiments. (b) Dose–response effect of angiostatin ⫹ endostatin. B16F10 supernatants containing angiostatin and endostatin were combined (1:1 ratio) and then diluted in DMEM at 1:4 or 1:2 or left undiluted (1:1). Endothelial cell tubes were counted at 24 h after plating and were scored as the average of 10 high-powered fields from duplicate wells. A Student t test showed statistical significance comparing control with 1:2 and 1:1 dilutions, P ⬍ 0.0006 and P ⬍ 0.0012, respectively.

delivery. This was determined at 2 and 4 weeks after inoculation by microscopic evaluation of the peripheral smear for blasts or GFP expression in the leukocytes. Angiostatin and endostatin could not be detected in the blood of any of the treatment groups, possibly due to low systemic levels of these inhibitors or to high background caused by cross-reacting epitopes. Forty percent of the mice in the combination group remained alive and disease free at the end of 90 days. There appeared to be little or no survival advantage for the groups receiving cells transduced by angiostatin or endostatin alone. Of the animals that remained alive at 90 days, there was MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

no evidence for leukemic cells in the peripheral blood or GFP-expressing leukocytes in the peritoneum (as determined by peritoneal lavage). There was also no evidence for L1210 leukemia cells using a PCR assay for amplification of endostatin sequences. One surviving animal in the combination group was sacrificed and no gross or histologic evidence for tumor was seen. The other surviving animals (one in the endostatin-alone and three in the combination groups) were subsequently rechallenged with untransduced L1210 leukemia cells (1 ⫻ 103) injected ip on day 106. Tumors grew in all retreated animals, suggesting that the original

191

ARTICLE DISCUSSION

FIG. 4. Growth of transduced tumor cells in vitro. B16F10 melanoma cells transduced with retrovirus expressing GFP alone, angiostatin IRES-GFP, or endostatin IRES-GFP were sorted for GFP expression and seeded at 2 ⫻ 103 cells per well of a 96-well plate on day 0. Growth was determined by MTT assay on days 1–5. Points shown are the averages of 16 wells. The exponential growth rate constants of all groups were similar (P ⬎ 0.05).

survival benefit was not due to an immune response such as from natural killer cells directed against the parent cell. These results are consistent with the prevention of tumor establishment occurring as a consequence of the expression of angiostatin and endostatin by L1210 cells.

Delivery of antiangiogenic recombinant proteins such as angiostatin and endostatin in preclinical models has been problematic (and the findings controversial) due to the stability and solubility of these agents (7). Since antiangiogenic therapies are expected to become chronic therapies for the prevention of recurrent cancer, large recombinant proteins are expected to be pharmacologically inferior compared with peptides or small molecules with regard to systemic delivery. Gene transfer represents one way to facilitate delivery of such recombinant proteins (7–9, 33–35). It also represents a potential technology by which antiangiogenic factors may be expressed constitutively in the host for long-term suppression of tumordependent blood vessel growth. It is envisioned that, if successful, this therapy could (i) be used in combination with chemotherapy for treatment of established tumors; (ii) be used as adjuvant treatment after eradication of tumors using conventional surgical, radiotherapeutic, or chemotherapeutic treatments; or (iii) be used in combination with biologic treatments such as immunotherapy, gene therapy, or other antiangiogenic agents (8). In preclinical models, several gene therapy studies using angiostatin-containing adenovirus or retrovirus have demonstrated tumor shrinkage although in some instances not complete eradication of established tumors (5, 7). Further progress will depend on a more in-depth understanding of angiogenesis as well as im-

FIG. 5. (a) Growth of angiostatin- and endostatin-transduced B16F10 melanoma cells in vivo. Mice were inoculated with 5 ⫻ 104 cells subcutaneously on day 0. Tumor growth was measured in volume (cm3) every 3– 4 days and represented by the mean of the group of mice alive at that time point. There were five animals per group. The exponential growth rate constants of all groups were significantly lower than that for LZRS-GFP, and the combination group was significantly lower than either angiostatin or endostatin alone (P ⬍ 0.05). (b) Median survival of C57BL/6 mice bearing angiostatin- and endostatin-transduced B16F10 melanoma tumors in the above study. At day 27 the proportion of mice surviving in the combination group was significantly greater than in the LZRS-GFP group (P ⬍ 0.035 by Fischer exact test).

192

MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

ARTICLE

FIG. 6. Expression of angiostatin, endostatin, and GFP in LZRS IRES-GFP- (A, C, E) or angiostatin ⫹ endostatin/IRES-GFP- (B, D, F) transduced melanoma tumors in situ by immunohistochemical analysis. The tumors were resected at 4 weeks following subcutaneous inoculation of 5 ⫻ 104 tumor cells in C57BL/6 mice. The top four images reveal high-magnification views of melanoma tumors stained with anti-HA antibody (A, B) or anti-endostatin serum (C, D) for detection of angiostatin and endostatin, respectively. Protein expression is visualized by the brown pigment generated by HRP-mediated catalysis of DAB substrate and is seen only in the combination groups (B, D). GFP expression is demonstrated in LZRS IRES-GFP- (E) or angiostatin ⫹ endostatin/IRES-GFP- (F) transduced tumors but not in tumors transduced with control vector (bottom).

MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

193

ARTICLE

FIG. 7. (a) Growth of L1210 leukemia cells in vitro. L1210 cells were transduced by retrovirus expressing GFP alone, angiostatin IRES-GFP, or endostatin IRES-GFP. GFP-sorted cells were seeded at 1 ⫻ 104 cells per well of a 12-well plate and then grown in culture for 4 days. The cells were counted in triplicate on days 0, 2, and 4. The exponential growth rate constants were similar for all groups (P ⬎ 0.05). (b) Effect of angiostatin and endostatin on survival of mice inoculated with L1210 leukemia. Balb/C nude mice were inoculated ip on day 0 with 1 ⫻ 103 retrovirally transduced leukemia cells. There were 10 animals per group. Using exponential analysis, treatment with the combination of angiostatin and endostatin improved survival versus the other groups (P ⫽ 0.03).

provement in gene therapy vector technology for both enhanced and persistent gene expression. In our studies, there appeared to be correlation between the antiangiogenic effects of angiostatin and endostatin in vitro and the in vivo antitumor effects seen in the B16F10 melanoma and L1210 leukemia models. A synergistic antitumor effect is strongly supported by the in vivo tumor growth kinetics of subcutaneously inoculated B16F10 melanoma tumors in mice. Similarly, it is also supported by the survival studies of mice inoculated with L1210 leukemia, in which the combination group of mice had a clear survival advantage over the other groups. There has been some suggestion that combining angiostatin and endostatin protein therapy potentiates the inhibitory effect on the growth of tumors (1, 21, 36). The

194

significance of a potential synergism of two antiangiogenic agents in combination will be better understood when the downstream effectors of these two agents are further described. Reduction of Bcl-2 and Bcl-XL has been observed with endostatin-mediated apoptosis (6). One report has implicated endostatin-induced tyrosine kinase activity via the Shb adaptor protein as critical to the regulation of endothelial cell apoptosis (29). Cell cycle regulatory proteins that are critical to endothelial cell proliferation such as M-phase phosphoproteins have recently been shown to be downregulated by angiostatin (5). Angiostatin and endostatin may thus be acting on separate pathways during induction of a synergistic antiangiogenic effect. Despite the delayed growth of angiostatin- or endostatin-transduced tumors in mice, all animals receiving B16F10 melanoma cells eventually developed large tumors. It is possible that high levels of expression of these inhibitors were not achieved compared with bolus delivery of these proteins in previous preclinical models. Alternatively, it is also possible that there was loss of expression of these proteins in transduced tumors over a period of 4 weeks as is supported by the patchy expression of these proteins in tumors by immunohistochemical analysis. This has also been observed in a recent report that showed loss of angiostatin expression in B16F10 melanoma after subcutaneous injection in vivo (37). It could be that the genetic instability of these tumors has facilitated loss of angiostatin- or endostatin-expressing cells followed by expansion of the negative clones. On the other hand, the mice that survived the L1210 leukemia inoculation with endostatin-transduced cells or the combination had no evidence of tumors. Two possibilities include that the cells underwent apoptosis after injection or perhaps the tumor foci remained in a microscopic state and were unable to grow beyond a small size because of expression of angiogenesis inhibitors. Rechallenge of these animals with L1210 cells led to subsequent tumor formation which did not support an immune- or natural killer cell-mediated effect against the parent cell as the reason for lack of establishment of the original tumors. Similarly, we also have preliminary evidence for loss of tumorigenicity in 100% of rats receiving RT-2 glioblastoma cells co-injected with angiostatin- and endostatin-producing retroviral packaging cells in brain. There was no evidence for microscopic tumors in these animals and it is possible that the tumors could not develop a blood supply and subsequently underwent apoptosis. However, an immune response could not be ruled out since 38% of animals receiving control packaging cells also had no evidence for microscopic tumors. On the other hand, there was no evidence for inflammation or lymphocytic infiltrates in the brains of surviving animals. An alternative explanation is the production of other antiangiogenic factors such as vasostatin in packaging cells expressing EBV sequences as were present in our retroviral vectors (38). The significance of angiostatin- and endostatin-mediated slowing of tumor growth in two different preclinical MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

ARTICLE cancer models underscores the potential broad utility of antiangiogenic approaches to a variety of tumors. Inhibition of the growth of human tumors in immunocompromised mice by high levels of angiostatin or endostatin protein has been demonstrated (39, 40). On the other hand, the maturation state of vessels that arise from spontaneous tumors in patients may be different from that of immature blood vessels growing in these mouse models (41). It has been suggested from studies in vitro that the putative target of angiostatin may be the endothelial precursor cell as opposed to the mature endothelial cell (42). For this reason, these mouse models may have limitations in predicting the response of spontaneous tumors to antiangiogenic agents delivered by bolus protein or gene transfer. It has also been suggested that tumor blood vessels are inherently different from normal vessels as well as the neovasculature that arises during normal physiologic states of angiogenesis (43). Numerous investigators are now exploring new antiangiogenic factors as well as endothelial cell markers that may be upregulated in tumor blood vessels compared with normal endothelium, and these include ␣vb3 integrins, angiopoietin receptors such as Tie2, VEGF receptors, and others (33, 38, 44 – 48). These molecules may serve as new targets for antiangiogenic strategies in cancer therapy, diabetic retinopathy, and other inflammatory diseases or serve as cotherapeutic targets with inhibitors such as angiostatin and endostatin shown here. The potential clinical utility of gene therapy with angiostatin- and endostatin-containing retroviral vectors to transduce tumor cells can be envisioned in malignancies in which control of local tumor growth is the end goal. Human malignant glioblastoma tends to recur locally, resisting surgical, radiotherapeutic, and chemotherapeutic approaches. This tumor does not metastasize and it can serve as an appropriate model in which intervention with retrovirus via direct delivery of packaging cells into the tumor can be attempted. Intratumoral delivery of retroviral packaging cells expressing the herpes thymidine kinase suicide gene into human glioblastoma has been described (49). It will be important to determine if delivery of antiangiogenic genes using this strategy will be effective in control of tumor growth in such patients. The approach suggests that although individual agents have been controversial in the past with regard to their efficacy (most likely attributable to different preparations of angiostatin or endostatin), such protein combinations produced in situ by mammalian cells can have antiangiogenic effects against blood vessel growth in tumors. Thus, concerns about systemic downregulation of blood vessel growth could be minimized. Combination therapy of locally produced factors, therefore, deserves further attention. ACKNOWLEDGMENTS We acknowledge Todd Kinsella for advice regarding cloning strategies and Donna Bouley for assistance with tumor resections and paraffin sectioning. We thank Mark Kay for his critical review of the manuscript. This work was supported in part by NIH 1 R01 (AR/AI44565), NIH K08 (CA79695-01A1), the MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy

Lymphoma Research Foundation, the Pfizer Fellowship, and the Janssen-ECOG Young Investigator grants.

REFERENCES 1 Boehm, T., Folkman, J., Browder, T., and O’Reilly, M. S. (1997). Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance. Nature 390: 404 – 407. 2 Beck, W. T., and Dalton, W. S. (1997). Mechanisms of drug resistance. In Principles and Practice of Oncology (S. Hellman, V. T. Devita, and S. A. Rosenberg, Eds.), pp. 498 –512. Lippincott–Raven, Philadelphia. 3 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. 4 O’Reilly, M. S., et al. (1997). Endostatin: An endogenous inhibitor of angiogenesis and tumor growth. Cell 88: 277–285. 5 Griscelli, F., et al. (1998). Angiostatin gene transfer: Inhibition of tumor growth in vivo by blockage of endothelial cell proliferation associated with a mitosis arrest. Proc. Natl. Acad. Sci. USA 95: 6367– 6372. 6 Dhanabal, M., et al. (1999). Endostatin induces endothelial cell apoptosis. J. Biol. Chem. 274: 11721–11726. 7 Tanaka, T., Cao, Y., Folkman, J., and Fine, H. A. (1998). Viral vector-targeted antiangiogenic gene therapy utilizing an angiostatin complementary DNA. Cancer Res. 58: 3362– 3369. 8 Folkman, J. (1998). Antiangiogenic gene therapy. Proc. Natl. Acad. Sci. USA 95: 9064 – 9066. 9 Kong, H. L., and Crystal, R. G. (1998). Gene therapy strategies for tumor antiangiogenesis. J. Natl. Cancer Inst. 90: 273–286. 10 Folkman, J. (1997). Antiangiogenic therapy. In Principles and Practice of Oncology (S. Hellman, V. T. Devita, and S. A. Rosenberg, Eds.), pp. 3075–3085. Lippincott–Raven, Philadelphia. 11 Perez-Atayde, A. R., Sallan, S. E., Tedrow, U., Connors, S., Allred, E., and Folkman, J. (1997). Spectrum of tumor angiogenesis in the bone marrow of children with acute lymphoblastic leukemia. Am. J. Pathol. 150: 815– 821. 12 Hussong, J. W., Rodgers, G. M., and Shami, P. J. (2000). Evidence of increased angiogenesis in patients with acute myeloid leukemia. Blood 95: 309 –313. 13 Veiga, J. P., Afonso, H., Sallan, S. E., Nadler, L. M., and Cardoso, A. A. (1998). Acute lymphoblastic leukemia cells and their bone marrow stroma collaborate to induce endothelial cell proliferation thereby promoting leukemia cell survival. Blood 92: 511a. 14 Shami, P. J., Hussong, J. W., and Rodgers, G. M. (1998). Evidence of increased angiogenesis in the bone marrow of patients with acute nonlymphocytic leukemia. Blood 92: 512a. 15 Kini, A. R., Peterson, L. C., and Kay, N. E. (1998). Evidence for abnormal angiogenesis in the bone marrow of patients with B-cell chronic lymphocytic leukemia (CLL). Blood 92: 717a. 16 Aguayo, A., et al. (1998). Increased angiogenesis in chronic myeloid leukemia and myelodysplastic syndromes. Blood 92: 607a. 17 Moehler, T. M., et al. (1998). Monitoring of vascularity and blood vessel permeability in bone lesions of multiple myeloma by dynamic magnetic resonance imaging. Blood 92: 99a. 18 Pruneri, G., et al. (1998). Angiogenesis in myelodysplastic syndromes. Analysis of 59 cases. Blood 92: 715a. 19 Rajkumar, S. V., Fonseca, R., Ansell, S. M., Witzig, T. E., Gertz, M. A., and Greipp, P. R. (1998). Computerized image analysis in the assessment of bone marrow angiogenesis in multiple myeloma. Blood 92: 100a. 20 Molica, S., Vitelli, G., Levato, D., Gandolfo, G. M., and Liso, V. (1999). Increased serum levels of vascular endothelial growth factor predict risk of progression in early B-cell chronic lymphocytic leukaemia. Br. J. Haematol. 107: 605– 610. 21 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. 22 Kinsella, T. M., and Nolan, G. P. (1996). Episomal vectors rapidly and stably produce high-titer recombinant retrovirus. Hum. Gene Ther. 7: 1405–1413. 23 Kinoshita, S., Chen, B. K., Kaneshima, H., and Nolan, G. P. (1998). Host control of HIV-1 parasitism in T cells by the nuclear factor of activated T cells. Cell 95: 595– 604. 24 Carmichael, J., DeGraff, W. G., Gazdar, A. F., Minna, J. D., and Mitchell, J. B. (1987). Evaluation of a tetrazolium-based semiautomated colorimetric assay: Assessment of chemosensitivity testing. Cancer Res. 47: 936 –942. 25 Folkman, J., Haudenschild, C. C., and Zetter, B. R. (1979). Long-term culture of capillary endothelial cells. Proc. Natl. Acad. Sci. USA 76: 5217–5221. 26 Riccioni, T., Cirielli, C., Wang, X., Passaniti, A., and Capogrossi, M. C. (1998). Adenovirus-mediated wild-type p53 overexpression inhibits endothelial cell differentiation in vitro and angiogenesis in vivo. Gene Ther. 5: 747–754. 27 Schwartz, M. A. (1961). Biomathematical approach to clinical tumor growth. Cancer 14: 1272–1294. 28 Cao, Y., O’Reilly, M. S., Marshall, B., Flynn, E., Ji, R. W., and Folkman, J. (1998). Expression of angiostatin cDNA in a murine fibrosarcoma suppresses primary tumor growth and produces long-term dormancy of metastases. J. Clin. Invest. 101: 1055–1063. 29 Dixelius, J., et al. (2000). Endostatin-induced tyrosine kinase signaling through the Shb adaptor protein regulates endothelial cell apoptosis. Blood 95: 3403–3411. 30 Humphries, M. J., Yamada, K. M., and Olden, K. (1988). Investigation of the biological effects of anti-cell adhesive synthetic peptides that inhibit experimental metastasis of B16-F10 murine melanoma cells. J. Clin. Invest. 81: 782–790. 31 Stripecke, R., Carmen-Villacres, M., Skelton, D., Satake, N., Halene, S., and Kohn, D.

195

ARTICLE (1999). Immune response to green fluorescent protein: Implications for gene therapy. Gene Ther. 6: 1305–1312. 32 Eatock, M. M., Schatzlein, A., and Kaye, S. B. (2000). Tumour vasculature as a target for anticancer therapy. Cancer Treat Rev. 26: 191–204. 33 Lin, P., et al. (1998). Antiangiogenic gene therapy targeting the endothelium-specific receptor tyrosine kinase Tie2. Proc. Natl. Acad. Sci. USA 95: 8829 – 8834. 34 Goldman, C. K., et al. (1998). Paracrine expression of a native soluble vascular endothelial growth factor receptor inhibits tumor growth, metastasis, and mortality rate. Proc. Natl. Acad. Sci. USA 95: 8795– 8800. 35 Chen, Q. R., Kumar, D., Stass, S. A., and Mixson, A. J. (1999). Liposomes complexed to plasmids encoding angiostatin and endostatin inhibit breast cancer in nude mice. Cancer Res. 59: 3308 –3312. 36 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. 37 Ambs, S., Dennis, S., Fairman, J., Wright, M., and Papkoff, J. (1999). Inhibition of tumor growth correlates with the expression level of a human angiostatin transgene in transfected B16F10 melanoma cells. Cancer Res. 59: 5773–5777. 38 Pike, S. E., et al. (1998). Vasostatin, a calreticulin fragment, inhibits angiogenesis and suppresses tumor growth. J. Exp. Med. 188: 2349 –2356. 39 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. 40 Dhanabal, M., et al. (1999). Endostatin: Yeast production, mutants, and antitumor effect in renal cell carcinoma. Cancer Res. 59: 189 –197.

196

41 Benjamin, L. E., Golijanin, D., Itin, A., Pode, D., and Keshet, E. (1999). Selective ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal. J. Clin. Invest. 103: 159 –165. 42 Ito, H., et al. (1999). Endothelial progenitor cells as putative targets for angiostatin. Cancer Res. 59: 5875–5877. 43 Arap, W., Pasqualini, R., and Ruoslahti, E. (1998). Cancer treatment by targeted drug delivery to tumor vasculature in a mouse model. Science 279: 377–380. 44 Brooks, P. C., Clark, R. A., and Cheresh, D. A. (1994). Requirement of vascular integrin alpha v beta 3 for angiogenesis. Science 264: 569 –571. 45 Gerber, H. P., Dixit, V., and Ferrara, N. (1998). Vascular endothelial growth factor induces expression of the antiapoptotic proteins Bcl-2 and A1 in vascular endothelial cells. J. Biol. Chem. 273: 13313–13316. 46 Presta, L. G., et al. (1997). Humanization of an anti-vascular endothelial growth factor monoclonal antibody for the therapy of solid tumors and other disorders. Cancer Res. 57: 4593– 4599. 47 Zhai, Y., et al. (1999). VEGI, a novel cytokine of the tumor necrosis factor family, is an angiogenesis inhibitor that suppresses the growth of colon carcinomas in vivo. FASEB J. 13: 181–189. 48 Kamphaus, G. D., et al. (2000). Canstatin, a novel matrix-derived inhibitor of angiogenesis and tumor growth. J. Biol. Chem. 275: 1209 –1215. 49 Ram, Z., et al. (1997). Therapy of malignant brain tumors by intratumoral implantation of retroviral vector-producing cells. Nat. Med. 3: 1354 –1361.

MOLECULAR THERAPY Vol. 3, No. 2, February 2001 Copyright © The American Society of Gene Therapy