D2001 Nature Publishing Group 0929-1903/01/$17.00/+0 www.nature.com/cgt

Evaluation of antiangiogenesis gene therapy in vitro and in vivo Xiaomei Jin,1 Rob Bookstein,1 Ken Wills,1 Jenny Avanzini,1 Van Tsai,1 Drake LaFace,1 Gaby Terracina,2 Bin Shi,2 and Loretta L. Nielsen2,y 1Canji Inc., San Diego, California 92121; and 2Tumor Biology, Schering-Plough Research Institute, Kenilworth, New Jersey 07033.

Progressive growth and of solid tumors require , or the formation of new blood vessels. Endostatin is a 20-kDa carboxy-terminal fragment of XVIII that has been shown to inhibit endothelial cell proliferation and tumor angiogenesis. Replication-deficient recombinant adenovirus (rAd) vectors were constructed, which encoded secreted forms of human and mouse endostatin (HECB and MECB, respectively), and, as a control, human alkaline phosphatase (APCB). Accumulation of endostatin was demonstrated in supernatants of cultured cells infected with the endostatin rAds. These supernatants disrupted tubule formation, inhibited migration and proliferation, and induced in human dermal vascular endothelial cells or human vascular endothelial cells. Endostatin-containing supernatants had no effect on the proliferation of MidT2-1 mouse mammary tumor cells in vitro. A pharmacokinetic study of MECB in immunocompetent FVB mice demonstrated a 10-fold increase of serum endostatin concentrations 3 days after intravenous administration of 1Â1010 particles of this rAd (215–257 ng/mL compared to 12–38 ng/mL in control rAd-treated mice). Intravenous administration of MECB reduced b-FGF stimulated angiogenesis into Matrigel plugs by 38%. Intratumoral MECB inhibited growth of MidT2-1 syngeneic mammary tumors in FVB mice, but had minimal impact on the growth of MDA-MB-231 human breast tumors in SCID mice. Intravenous therapy with MECB also initially inhibited growth of MidT2-1 tumors, but this activity was subsequently blocked by induced anti-rAd . In summary, endostatin gene therapy effectively suppressed angiogenic processes in vitro and in vivo in several model systems. Gene Therapy (2001) 8, 982–989

Key words: Gene therapy; antiangiogenesis; endostatin; recombinant adenovirus.

ngiogenesis is a complex process, which includes the has been shown to suppress tumor-induced angio- Aactivation, proliferation, and migration of endothelial genesis and tumor growth.3,4 Notably, these studies cells; disruption of vascular basement membranes; for- required daily parenteral injections of large amounts of mation of vascular tubes and networks; and their linkage to recombinant protein, illustrating a common difficulty in the pre-existing vascular networks.1,2 It is involved in a pharmacologic use of therapeutic . number of normal and pathological biological processes, Gene therapy approaches may have advantageous proper- such as , several ocular disorders, and ties as methods for delivery of systemic protein , neoplasia.2 Specific inhibitors of angiogenesis may have especially wherein long-term production and secretion from therapeutic roles in these diseases. In the last few years, a a conveniently transduced tissue can be effected. The goal number of protein or peptide molecules have been of this study was to demonstrate, as a proof of principle, identified with antiangiogenic activity in vitro and in vivo. transgene-derived secreted local or systemic antiangiogenic Some of them are subfragments of other larger proteins, protein therapy. A recombinant adenovirus (rAd) vector such as plasminogen or collagen XVIII in the cases of was selected because of its transduction efficiency and and endostatin, respectively. Endostatin was versatility by several dosage routes (e.g., intratumoral and first isolated from conditioned media (CM) of cultured intravenous), despite its anticipated immune-mediated EOMA (hemangioendothelioma) cells and was found to restrictions on redosing. Endostatin was chosen as the be identical to a 20-kDa C-terminal fragment of collagen encoded protein based on its well-characterized activity as XVIII.3 Endostatin is a specific inhibitor of endothelial cell an antiangiogenic protein. Our results demonstrate both the proliferation in vitro, and systemic therapy with endostatin potential therapeutic efficacy as well as the problems of this approach.

Received August 20, 2001. Address correspondence and reprint requests to Dr. Xiaomei Jin, Canji, MATERIALS AND METHODS 3525 John J. Hopkins Street, San Diego, CA 92121. E-mail address: Cell lines [email protected] y Current address: Targeted Molecules, 6605 Nancy Ridge Drive, San The 293 cell line was obtained from Microbix and cultured Diego, CA 92121. E-mail address: [email protected] in Dulbecco’s modified Eagle’s medium (DMEM; Gibco,

982 Cancer Gene Therapy, Vol 8, No 12, 2001: pp 982–989 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS 983

Grand Island, NY) with 10% fetal bovine serum (FBS; 100000 Gibco). Human primary endothelial cells, HUVEC (human APCB MECB umbilical vein endothelial cells) and HDEC (human dermal 75000 endothelial cells), were purchased from Clonetics (San Diego, CA) and cultured in manufacturer’s growth medium. ARPE-19 is an immortalized cell line derived from human 50000 retinal pigment epithelial cell and was purchased from ATCC (Rockville, MD). The MidT2-1 clonal cell line was 25000 derived from female MMTV Polyomavirus Middle T (PyV

MidT) transgenic mouse mammary tumors as previously 0 described5 and are syngeneic for the immunocompetent (fluorescence) Index Cell Proliferation HDEC MidT2-1 FVB mouse strain. MDA-MB-231 human breast adeno- Figure 2. Inhibition of endothelial cell proliferation in vitro. HDEC or carcinoma cells were purchased from ATCC and cultured in MidT2-1 mouse mammary tumor cells were seeded into six-well DMEM with 10% FBS. All cell lines were cultured at 378C plates at a density of 5Â104 cells/well, allowed to attach for 8 hours, and 5% CO2. and starved with serum-free medium overnight. Cells were then treated with CM for 8 hours, followed by complete medium for an Construction and purification of rAd carrying endostatin additional 5 days before cell quantitation. cDNA cDNA containing the coding sequences of human and mouse digestion and transgene sequencing. DNA was linearized endostatin were obtained from the IMAGE Consortium. A with PacI and used to transfect 293 cells for generation of secretion signal peptide sequence from preprotrypsin (ACC rAd carrying the human or mouse endostatin (HECB or ATG TCT GCA CTT CTG ATC CTA GCT CTT GTT GGA MECB, respectively). A control vector, APCB, constructed GCT GCA GTT GCT) was added to the 50 end of each by the same procedure carried the secreted form of human alkaline phosphatase in place of endostatin. Viral vectors cDNA by polymerase chain reaction. They were then 7 subcloned into an adenoviral shuttle vector6 containing a were propagated and purified as previously described. Viral particle number (PN) was determined using anion exchange cytomegalovirus (CMV) promoter for transgene expression. 8 high-performance liquid chromatography and A260 nm Shuttle plasmids were cotransformed into Escherichia coli 9 (BJ5183) with a plasmid carrying type 5 adenoviral genomic measurement in 0.1% SDS (wt/vol). Viral construct DNA with E3 region deletion. DNA from recombinant infectivity was confirmed using a flow cytometry–based colonies was purified and verified by restriction enzyme adenovirus infection assay and expressed as cellular infectious units (ciu).10

Production of CM HECB MECB APCB PURIFIED A. ENDOSTATIN ARPE-19 cells were seeded in 10-cm plates and infected a ba b a b with HECB, MECB, or APCB, respectively, at dosage of 1Â109 PN/mL for 2 hours. Viruses were washed off with phosphate-buffered saline (PBS). Cells were then incubated in serum-free medium containing 0.1% bovine serum B. 2500 1000 NO ATRACTANT

HECB 2000 800 MECB

1500 600 APCB

400 1000

200 500 migration index (fluorecsence) migration

0

Protein Concentration (nglml) 0 0 1 2 3 HECB MECB APCB Figure 1. Endostatin expression from rAd vectors carrying the migration time (hour) secreted human or mouse endostatin cDNA. ARPE-19 cells were Figure 3. Inhibition of endothelial in vitro. HUVEC were infected by rAd-HEndo, rAd-MEndo, or rAd-AP at 1Â109 PN/mL labeled with Calcein-AM before being aliquoted onto 3-mm FALCON for 2 hours and incubated in serum-free medium for 24 hours. A: HTS FluoroBlok cell culture inserts. CM from adenovirus-transduced Endostatin expression shown by Western blot. Lane a: CM. Lane ARPE-19 cells was added to the top chamber. FBS and b-FGF were b: Cell lysate. B: Endostatin concentrations in CM measured by used as chemoattractants in the lower chamber. Cell migration into ELISA. the lower chamber was quantitated using cell fluorescence.

Cancer Gene Therapy, Vol 8, No 12, 2001 984 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS

MEDIUM APCB HECB MECB

Figure 4. Inhibition of endothelial cell differentiation in vitro.HDECwere seeded into 24-well plates coated with ECM at a density of 40,000 cells/well. CM was added into the wells at 200 mL/ well. Cell growth and differentiation were documented as observed 24 hours later. albumin (BSA) for 48 hours. Supernatants (CM) were in (ECM)–coated 24-well plates harvested and concentrated with Centricon-3 units (Ami- (Biocoat Cellware; Becton Dickinson Biosciences) at con, Beverly, MA). 40,000 cells/well. CM was added into the wells at 200 mL/well. Cell growth and differentiation was documented Endostatin expression from HECB and MECB 24 hours later. To detect endostatin expression, CM was analyzed by Western blot using a polyclonal generated against a In vitro endothelial cell apoptosis assays conjugated peptide (N-CWPQKSVWHGSDPNGRRL-C) HUVEC were infected in six-well plates with different (Multiple Peptide Systems, San Diego, CA). Endostatin vectors at 1Â109 PN/mL for 4 hours. Viruses were then concentration in CM was measured by the Accucyte human washed off with PBS and cells were incubated with complete or murine endostatin enzyme-linked immunosorbent assay medium for 48 hours. Cells were trypsinized, stained with (ELISA) kit (Cytimmune Sciences, College Park, MD) annexin V-FITC (Roche Molecular Biochemicals, Indian- according to the manufacturer’s instructions. apolis, IN), and analyzed by FACS.

In vitro endothelial cell proliferation assays Animal care Human dermal endothelial cells or MidT2-1 cells were Mice were obtained from SPRI breeding colonies. All mice 4 seeded in six-well plates at 5Â10 cells/well for 8 hours and were maintained in a VAF barrier facility. Animal proce- then starved with serum-free medium overnight. Cells were dures were performed in accordance with the rules set forth then treated with CM at 1 mL/well for 8 hours before in the NIH Guide for the Care and Use of Laboratory complete medium was added to each well at 1 mL/well. Animals and approved by the SPRI Animal Care and Use Cells were incubated in 378C with 10% CO2 for 5 days Committee. before they were counted with a hemacytometer. Pharmacokinetic study In vitro endothelial cell migration assays Each female FVB mouse was injected with 1Â1010 PN HUVEC were grown to 80–90% confluency in a BIOCOAT MECB or APCB via the tail vein on day 0. Mice were Collagen I T-75 flask (Becton Dickinson Biosciences, Franklin Lakes, NJ) and then labeled in situ with 5 mM Calcein-AM in DMEM containing 2% FBS for 2 hours at 100 378C. After washing, cell monolayers were briefly trypsi- nized, and cell pellets were resuspended in DMEM supplemented with 0.1% BSA. Cells were placed in 3-mm 75 FALCON HTS FluoroBlok Cell Culture Inserts (Becton Dickinson Biosciences) at a density of 1Â106 cells/insert. CM from different vector-infected cells (see above) were 50 added into the inserts and incubated with cells for 30 minutes before DMEM containing 1% FBS and basic FGF (b-FGF, 20 ng/mL; R&D Systems, Minneapolis, MN) was added to 25

the lower wells as a chemoattractant. Cells were allowed to Apoptosis (%) migrate for 3 hours. Fluorescence of cells that had migrated through the FALCON HTS FluoroBlok Inserts was meas- ured on a Cytofluor 4000 plate reader (PerSeptive 0 Biosystems, Framingham, MA) using excitation/emission APCB MECB wavelengths of 485/530 nm. HECB Figure 5. Endostatin induces apoptosis in vitro. HUVEC were 9 In vitro endothelial cell differentiation assays transduced with 1Â10 PN rAd/mL for 4 hours. Viruses were then washed off and cells incubated for a further 48 hours in complete Endothelial cell differentiation assays were performed as medium. Apoptosis was measured using FACS on cells labelled with 11 described. Human dermal endothelial cells were seeded annexin V-FITC.

Cancer Gene Therapy, Vol 8, No 12, 2001 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS 985 sacrificed and serum collected on days 3, 4, 5, 6, and 7. rocked for 2 days at 48C. Absorption of the supernatants In addition, serum was collected from three untreated containing total Hgb was measured at 540 nm and the Hgb mice to establish baseline levels of endogenous endo- levels calculated using a Hgb standard curve run in parallel statin (day 0 value). Samples were assayed using the with the unknowns. Accucyte murine endostatin ELISA kit (see above) according to the manufacturer’s instructions. Endostatin Intratumoral administration levels for any two treatment groups on each day were MidT2-1 model. Each female FVB mouse was injected with compared using Students’ t test (Statview Software; SAS 5Â106 MidT2-1 syngeneic mouse mammary tumor cells Institute, Cary, NC). into the mammary fat pad. MDA-MB-231 model. Each female SCID mouse was injected Matrigel assay with 5Â106 MDA-MB-231 human breast tumor cells into the mammary fat pad. Treatment was started 16 days later Growth factor–reduced Matrigel matrix without phenol red 10 was purchased from Becton Dickinson Biosciences. Human when large tumors had formed. Adenovirus (1Â10 PN) or b-FGF was from R&D Systems. b-FGF was dissolved in vehicle was administered intratumorally on days 0 and 5 in a sterile PBS buffer containing 0.1–1% BSA and 1 mM DTT volume of 0.1 mL. Tumor measurements were recorded to make 25 mg/mL stock solution. Matrigel was liquidized during the course of the study and serum samples were slowly at 48C. Each female FVB mouse was injected with collected on day 8 for determination of circulating endostatin 0.5 mL of unpolymerized Matrigel containing b-FGF at a levels as described above. Tumor volumes were calculated as concentration of 700 ng/mL subcutaneously along the lengthÂwidthÂdepth. Tumor volumes for any two treatment abdominal midline. Mice were given one intravenous groups on each day were compared using Students’ t test injection of 1Â1010 PN MECB or APCB via the tail vein (Statview Software; SAS Institute). 3 days before Matrigel injection. Matrigel plugs were harvested 7 days later and put into Drabkin’s solution for Intravenous administration hemoglobin (Hgb) assay (Sigma, St. Louis, MO). Briefly, Each female FVB mouse was injected with 1Â106 MidT2-1 Matrigel plugs removed from surrounding tissues were syngeneic mouse mammary tumor cells into the mammary resuspend in 2 mL of Drabkin’s solution, mixed well, and fat pad. Treatment was started 10 days later when palpable

(3) Alkaline Phosphatase Ad 300 Mouse Endostatin Ad (3)

) (2) (3) 250

200

(3) 150

100 Serum Endostatin (ng/ml

50 (2) (3) (3) (3) (3) (3) Figure 6. Pharmacokinetic study in immune-competent FVB mice. Mice 0 were given one intravenous dose of 0 1 2 3 4 5 6 7 10 1Â10 PN MECB on day 0. Mean± Day SEM. n=3 mice per group.

Cancer Gene Therapy, Vol 8, No 12, 2001 986 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS tumors had formed. Adenovirus (1Â1010 PN/dose) or for mouse endostatin compared to control CM (Fig 3). vehicle was administered via the tail vein on days 0, 7, and ECM-coated plates were used to examine the effect of 14 in a volume of 0.1 mL. Tumor sizes were analyzed as secreted endostatin on in vitro tubule formation, which above, and serum samples were collected on day 18 for anti- assesses migration and differentiation of human dermal Ad antibody testing. endothelial cells. As shown in Figure 4, tube-like structures formed by endothelial cells were abrogated by CM Neutralizing antibody assay containing the secreted human or mouse endostatin. By Modification of a previously described assay12 was used to contrast, CM from APCB-infected cells had no observable determine the level of neutralizing antibodies in serum. effect on tube structure. Serial dilution (1:2) of the serum sample of interest (starting at 1:20, 1:40, 1:80, etc.) was pipetted in a 96- HECB- and MECB-induced endothelial cell apoptosis well format. A replication-deficient (deleted for E1, E3, Although the mechanism of endostatin’s antiangiogenic pIX) type 5 human rAd vector13 expressing the Green activity is unknown, it has been shown that endostatin Fluorescent Protein under control of the CMV immediate specifically induces endothelial cell apoptosis.14 We early promoter (GFCB) was used. The appropriately tested the effects of HECB and MECB on the induction diluted serum sample was allowed to incubate with GFCB of endothelial cell apoptosis. HUVEC infected with at a final concentration of 4Â108 particles/mL for 1 hour at HECB or MECB demonstrated markedly enhanced 378C. The serum/virus mixture was transferred onto HeLa apoptosis 24 hours postinfection compared with the cells plated at 1Â104 cells/well in a 96-well format. The control vector (Fig 5). CM containing human or mouse virus was allowed to infect the cells overnight. The relative endostatin were also used to induce endothelial cell fluorescence was measured the following day on a apoptosis, but the effect was not as dramatic as direct CytoFluor 4000 Fluorescence Multiwell Plate Reader infection with Ad vector (data not shown). (PerSeptive Biosystems, Foster City, CA) to assay neutralizing capacity. Serum endostatin levels after intravenous MECB injection As a prelude to the use of rAd endostatin vectors in vivo,a pharmacokinetic study of intravenously injected MECB RESULTS was performed in immune-competent FVB mice. Endostatin protein expression from rAds Although the particular disposition of these rAds was not followed, more than 90% of one rAd dose administered by rAds encoding human and mouse endostatin (HECB and MECB, respectively) were constructed, produced, and purified as described in Materials and Methods. Endo- 3 statin gene expression from HECB or MECB was examined by Western blot analysis (Fig 1A) and ELISA (Fig 1B). Human or mouse endostatin was detected in 2.5 both CM and cell lysates from rAd-infected ARPE-19 cells 24 hours postinfection. The concentrations of human 2 or mouse endostatin in the CM, measured by ELISA, were 2.1 g/mL for HECB and 1.8 g/mL for MECB. No 1.5 endostatin was detected from the CM or cell lysate of control APCB-infected cells. 1 Hemoglobin (g/dL)

Biological activity of HECB- and MECB-derived endostatin 0.5 We sought to verify the biological activity of rAd-expressed endostatin in several in vitro assays representing various 0 important aspects of the angiogenic process, including endothelial cell proliferation, migration, and differentiation APCB

in response to ECM, which plays a dynamic role in MECB endothelial cell biology. To determine whether secreted endostatin from HECB or MECB can inhibit endothelial cell

proliferation, human dermal vascular endothelial cells were APCB + bFGF treated with CM containing human or mouse endostatin MECB + bFGF followed by stimulation with complete medium. As shown in Figure 2, CM with endostatin inhibited HDEC proliferation Figure 7. Endostatin gene therapy suppresses bFGF- induced angiogenesis into Matrigel plugs in vivo. Female FVB mice were by approximately 50% compared to CM without endostatin, 10 7 given one intravenous dose of 1Â10 PN rAd (3.59Â10 ciu MECB whereas the growth of MidT-2-1 tumor cells was not or 2.35Â108 ciu APCB) via the tail vein on day 0 — 3 days before affected. Similarly, CM containing human or mouse endo- Matrigel injection. Mice were injected with Matrigel with or without statin inhibited HUVEC migration through porous mem- bFGF on day 3. Hgb content of Matrigel plugs was assessed on day branes by approximately 40% for human endostatin and 50% 10. Mean±SD. n=5 mice per group.

Cancer Gene Therapy, Vol 8, No 12, 2001 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS 987 this route has been shown to be retained in the liver. After 10 A Alkaline Phosphatase Ad (n=4) intravenous injection of 1Â10 PN MECB, serum endo- 5 MouseEndostatin Ad (n=5) statin concentrations rose to 215–257 ng/mL during days 3–6 (Fig 6), which was a 10-fold increase over e pretreatment or APCB-treated levels (range 12–38 ng/ 4 mL). By day 7, serum endostatin levels had fallen to 131±30 ng/mL ( ±SEM). Ad

Matrigel angiogenesis assay 3 Angiogenic factors, such as b-FGF, induce the growth of new blood vessels into solid plugs of Matrigel injected 2 subcutaneously into mice. Quantitation of blood Hgb Ad content in these plugs serves as a surrogate marker for volum tumor Normalized angiogenesis. As shown in Figure 7, the Hgb content of Matrigel plugs containing b-FGF was 10-fold higher than 1 the Hgb content of Matrigel plugs containing no additional angiogenic factors, indicating successful measurement of proangiogenic activity of b-FGF in this assay. In mice with 0 0 1 2 3 4 5 6 7 8 Matrigel plugs containing b-FGF, MECB therapy sup- Day pressed angiogenesis by 38% compared to mice treated with 3.5 APCB (P=.018). Circulating endostatin levels were B Alkaline Phosphatase Ad (n=4) determined 10 days after intravenous dosing to confirm Mouse Endostatin Ad (n=5) and extend the earlier PK study. Mice treated with MECB 3 had serum endostatin level of 136 ng/mL (groups 2 and 4) Ad compared to those dosed with APCB with 36 ng/mL 2.5 (group 1) and 40 ng/mL (group 3), respectively.

Antitumor activity of MECB 2 The antitumor activity of MECB was tested in the syngeneic MidT2-1 mammary tumor model in FVB mice 1.5 Ad and in the MDA-MB-231 human breast tumor model in SCID mice. Two intratumoral doses of 1Â1010 PN MECB suppressed growth of MidT2-1 mammary fat pad tumors Normailzed tumor volume 1 (P.01; Fig 8A), but had minimal impact on the growth of MDA-MB-231 tumors (Fig 8B). The first intravenous .5 dose of MECB also retarded the growth of MidT2-1 tumors, but subsequent doses were minimally effective (Fig 0 8C) presumably because mice developed neutralizing 0 1 2 3 4 5 6 7 8 antibodies against rAd. Anti-rAd antibodies were undetect- Day able in mice dosed with vehicle (n=10); however, in mice dosed with APCB or MECB, 4 mice had titers of 1:1280, 5 C Vehicle (n=10) mice had titers of 1:2560, and 10 mice had titers greater 1200 Alkaline Phosphatase Ad (n=10) than 1:2560. Mouse Endostatin Ad (n=10) Ad 1000

)

3

Figure 8. Efficacy of endostatin gene therapy tumor models in vivo. m

m 800

( A: MidT2-1 model — Immunocompetent FVB mouse bearing well-

e

established MidT2-1 tumors were dosed with intratumoral rAd or m

u vehicle on days 0 and 5. Tumor volumes on day 0 ranged from 266 to l o 600 3 v Ad 920 mm . Tumor volumes on subsequent days were normalized to

r day 0 volume for each individual mouse. B: MDA-MB-231 model — o

m

Immunodeficient SCID mouse bearing well-established MDA-MB- u Ad

T 400 231 tumors were dosed with intratumoral rAd or vehicle on days 0 and 5. Tumor volumes on day 0 ranged from 136 to 550 mm3. Tumor volumes on subsequent days were normalized to day 0 volume for 200 each individual mouse. C: MidT2-1 model — FVB mice bearing well-established MidT2-1 tumors were dosed with intravenous rAd or vehicle on days 0, 7, and 14. For all experiments, each rAd dose 0 was 1Â1010 PN (3.59Â107 ciu MECB or 2.35Â108 ciu APCB). 02.5 5 7.5 10 12.5 15 17.5 20 Mean±SEM plotted. Day

Cancer Gene Therapy, Vol 8, No 12, 2001 988 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS

DISCUSSION growth of MidT2-1 tumors might be more dependent on continuing angiogenesis to support tumor growth than Both positive and negative regulators of the process of MDA-MB-231 tumors. Feldman et al25 used a mouse angiogenesis have been described. Tumor cells are known to tumor cell line, which was relatively resistant to adenovirus secrete many proangiogenic factors, such as fibroblast infection, to demonstrate a 40% reduction in tumor size in growth factor and vascular endothelial growth factor, nude mice after in vivo endostatin gene therapy. This study consistent with the requirement of angiogenesis for pro- confirmed the sensitivity of blood vessel endothelial cells in gressive tumor growth.15 Surprisingly, tumors may also nude mice to endostatin-mediated suppression of angio- produce angiogenesis inhibitors. Endostatin, a 20-kDa C- genesis. terminal fragment of collagen XVIII and a specific inhibitor In conclusion, gene therapy has now been employed for of endothelial cell proliferation, migration, and angio- the delivery of angiogenesis inhibitors to suppress tumor genesis3,16 – 18 was initially identified in supernatants of growth and metastasis. It represents a more cost-effective cultured tumor cells. The role of this peptide in natural and convenient way of delivering steady-state levels of human tumorigenesis is not yet known.14,19,20 antiangiogenic factors to the tumor. Our study provides Inhibition of angiogenesis by endostatin or other further evidence that antiangiogenic gene therapy can be proteins is thought to be a promising new strategy in used as an effective approach in cancer treatment, and cancer therapy.15 However, its widespread application has translation of antiangiogenic gene therapy from laboratory to been hampered by difficulties in the large-scale production clinic may soon become a reality. of the antiangiogenic proteins. This limitation may be resolved by in vivo delivery and expression of the antiangiogenic genes.21 – 26 In this study, rAds expressing REFERENCES secreted forms of human (HECB) and mouse (MECB) endostatin were shown to be efficient agents for endostatin 1. Folkman JDAP. Blood vessel formation: What is its molecular gene therapy. Immortalized human retinal pigment epithe- basis? Cell. 1996;87:1153–1155. lial (ARPE-19) cells secreted high levels of endostatin 2. Hanahan DFJ. Patterns and emerging mechanisms of the into the surrounding medium after rAd transduction. This angiogenic switch during tumorigenesis. Cell. 1996;86:353– CM inhibited the proliferation of human dermal endothe- 364. 3. O’Reilly MSBT, Shing Y, Fukai N, et al. Endostatin: An lial cells, but had no effect on the proliferation of ARPE- endogenous inhibitor of angiogenesis and tumor growth. Cell. 19 epithelial cells (data not shown) and MidT2-1 mouse 1997;88:277–285. mammary carcinoma cells. Transduction of human dermal 4. Boehm TFJ, Browder T, O’Reilly MS. Antiangiogenic therapy endothelial cells with HECB or MECB also suppressed the of experimental cancer does not induce acquired normal response of cellular differentiation into tubules in resistance. Nature. 1997;390:404–407. vitro. In addition, endostatin inhibited proliferation, sup- 5. Nielsen LL, M Gurnani, B Shi, et al. Derivation and initial pressed migration, and induced apoptosis of HUVEC. characterization of a mouse mammary tumor cell line carrying Intravenous gene therapy in immune-competent FVB mice the polyomavirus middle T antigen: Utility in the development produced a 10-fold increase of circulating endostatin of novel cancer therapeutics. Cancer Res. 2000;60:7066–7074. 6. Chartier CED, Gantzer M, Dieterle A, Pavirani A, Mehtali M. levels within 3 days after intravenous administration of 10 Efficient generation of recombinant adenovirus vectors by 1Â10 PN MECB. These elevated levels of circulating homologous recombination in Escherichia coli. JVirol. endostatin were maintained until day 6 and began to fall 1996;70:4805–4810. by day 7. The same single, intravenous dose of MECB 7. Huyghe BLX, Sutjipto S, Sugarman B, et al. Purification of a suppressed b-SSFGF–stimulated angiogenesis into Matri- type 5 recombinant adenovirus encoding human p53 by gel plugs by 38%. column chromatography. Hum Gene Ther. 1995;6:1403–1416. The antitumor activity of MECB was tested in two in vivo 8. Shabram PGD, Gouldreau A, Gregory R, et al. Analytical tumor models — the syngeneic MidT2-1 mammary tumor anion exchange HPLC of recombinant type-5 adenoviral model in FVB mice and the MDA-MB-231 human breast particles. Hum Gene Ther. 1997;8:453–465. tumor model in immunodeficient SCID mice. Two intra- 9. Maizel JWD, Scharff M. The polypeptides of adenovirus: 1. Evidence for multiple protein components in the virion and tumoral doses of MECB halted further growth of MidT2-1 a comparison of types 2, 7a, and 12. Virology. 1968;36: mammary fat pad tumors, but had minimal impact on the 115–125. growth of MDA-MB-231 tumors. The first intravenous 10. Musco MCS, Small D, Nodelman M, Sugarman B, Grace M. dose of MECB had a similar cytostatic effect on MidT2-1 Comparison of flow cytometry and laser scanning cytometry tumors. However, further dosing had little impact on tumor for the intracellular evaluation of adenoviral infectivity and growth rates that corresponded to the development of p53 protein expression in gene therapy. Cytometry. 1998;33: neutralizing antibodies against rAd. Because endostatin 290–296. had no effect on MidT2-1 cell proliferation in vitro,we 11. Schuetz JDSE. Extracellular matrix regulation of multidrug conclude that endostatin gene therapy halted MidT2-1 tumor resistance in primary monolayer cultures of adult rat hepato- cytes. Cell Growth Differ. 1993;4:31–40. growth in FVB mice by suppressing tumor angiogenesis in 12. Kozarsky KF, Mckinley DR, Austin LL, Raper SE, Stratford- vivo. Further investigation is required into the reason why Perricaudet LD, Wilson JM. In vivo correction of low density endostatin had little effect on MDA-MB-231 tumor growth lipoprotein receptor deficiency in the Watanabe heritable in SCID mice. SCID blood vessel endothelial cells may be hyperlipidemic rabbit with recombinant adenoviruses. J Biol less sensitive to endostatin than FVB endothelial cells, or the Chem. 1994;269:13695–13702.

Cancer Gene Therapy, Vol 8, No 12, 2001 JIN, BOOKSTEIN, WILLS, ET AL: ENDOSTATIN ANTIANGIOGENESIS 989

13. Wills KN, Maneval DC, Menzel P, et al. Development and 21. Kong HLHD, Song W, Kovesdi I, Hackett NR, Yayon A, characterization of recombinant adenoviruses encoding human Crystal RG. Regional suppression of tumor growth by p53 for gene therapy of cancer. Hum Gene Ther. 1994;5: in vivo transfer of a cDNA encoding a secreted form 1079–1088. of the extracellular domain of the flt-1 vascular 14. Mohanraj Dhanabal RR, Waterman MJF, Lu H, Knebelmann endothelial growth factor receptor. Hum Gene Ther. 1998; B, Segal M, Sukhatme VP. Endostatin induces endothelial cell 9:823–833. apoptosis. J Biol Chem. 1999;274:11721–11726. 22. Pengnian Lin JAB, Acheson A, Czeslaw, et al. Antiangiogenic 15. Gabriele Bergers KJ, Lo K-M, Folkman J, Hanahan D II. gene therapy targeting the -specific receptor Effects of angiogenesis inhibitors on multistage carcinogenesis tyrosine kinase Tie2. Proc Natl Acad Sci USA. 1998;95: in mice. Science. 1999;284:808–812. 8829–8834. 16. Erhard Hohenester TS, Olsen BR, Timp R. Crystal structure of 23. Sauter BVMO, Zhang WJ, Mandeli J, Woo SL. Adeno- the endostatin at 1.5 A˚ resolution. virus-mediated gene transfer of endostatin in vivo results in EMBO J. 1998;17:1656–1664. high level of transgene expression and inhibition of tumor 17. Dhanabal MRR, Volk R, Stillman IE, et al. Endostatin: Yeast growth and metastases. Pro Nat Acad Sci USA. 2000;97: production, mutants, and antitumor effect in renal cell 4802–4807. carcinoma. Cancer Res. 1999;59:189–197. 24. Chen QRKD, Stass SA, Mixson AJ. Liposomes complexed to 18. Dhanabal MVR, Ramchandran R, Simons M, Sukhatme VP. plasmids encoding angiostatin and endostatin inhibit breast Cloning, expression, and in vitro activity of human endostatin. cancer in nude mice. Cancer Res. 1999;59:3308–3312. Biochem Biophys Res Commun. 1999;258:345–352. 25. Feldman ALRN, Alexander HR, Bartlett DL, Hwu P, Seth P, 19. Noriko Yamaguchi BA-A, Lee M, Takako, et al. Endostatin Libutti SK. Antiangiogenic gene therapy of cancer utilizing a inhibits VEGF-induced endothelial cell migration and tumor recombinant adenovirus to elevate systemic endostatin levels in growth independently of zinc binding. EMBO J. 1999;18: mice. Cancer Res. 2000;60:1503–1506. 4414–4423. 26. Blezinger PWJ, Gondo M, Quezada A, et al. Systemic 20. Boehm TOrM, Keough K, Shiloach J, Shapiro R, Folkman J. inhibition of tumor growth and tumor metastases by intramus- Zinc-binding of endostatin is essential for its antiangiogenic cular administration of the endostatin gene. Nat Biotechnol. activity. Biochem Biophys Res Commun. 1998;252:190–194. 1999;17:343–348.

Cancer Gene Therapy, Vol 8, No 12, 2001