A gene transfer comparative study of HSA-conjugated antiangiogenic factors in a transgenic mouse model of metastatic ocular cancer

A Corrigendum to this article was published on 14 March 2007


Different antiangiogenic and antimetastatic recombinant adenoviruses were tested in a transgenic mouse model of metastatic ocular cancer (TRP1/SV40 Tag transgenic mice), which is a highly aggressive tumor, developed from the pigmented epithelium of the retina. These vectors, encoding amino-terminal fragments of urokinase plasminogen activator (ATF), angiostatin Kringles (K1–3), endostatin (ES) and canstatin (Can) coupled to human serum albumin (HSA) were injected to assess their metastatic and antiangiogenic activities in our model. Compared to AdCO1 control group, AdATF-HSA did not significantly reduce metastatic growth. In contrast, mice treated with AdK1–3-HSA, AdES-HSA and AdCan-HSA displayed significantly smaller metastases (1.19±1.19, 0.87±1.5, 0.43±0.56 vs controls 4.04±5.12 mm3). Moreover, a stronger inhibition of metastatic growth was obtained with AdCan-HSA than with AdK1–3-HSA (P=0.04). Median survival was improved by 4 weeks. A close correlation was observed between the effects of these viruses on metastatic growth and their capacity to inhibit tumor angiogenesis. Our study indicates that systemic antiangiogenic factors production by recombinant adenoviruses, particularly Can, might represent an effective way of delaying metastatic growth via inhibition of angiogenesis.


In recent years, it has become clear that angiogenesis is not only important in physiological processes such as embryonic development, wound healing and organ and tissue regeneration, but also plays a crucial role in the development of numerous pathologic conditions including diabetic retinopathy, rheumatoid arthritis, tumor progression and metastasis. 1, 2 Substantial evidence now confirms that a solid tumor cannot grow beyond a volume of 1–2 mm3 without the formation of new capillaries to supply tumor cells with oxygen and nutrients. Whereas the first stages of tumor development may rely on existing capillaries,3, 4 an angiogenic switch has been identified that comes into play when the tumor volume reaches a certain threshold.2 This switch corresponds to a local predominance of positive over negative angiogenic factors. Theoretically, tumor growth should be restricted if this imbalance is neutralized through an increase in the local concentration of antiangiogenic factors. Proangiogenic factors typically include vascular endothelial growth factors (VEGF) and fibroblast growth factors (FGFs). Many tumor- and non-tumor-related antiangiogenic factors have been described. The proteolytic cleavage of larger precursor molecules associated with the vascular system, such as the proteins of the coagulation cascade and the basement membrane,5, 6 is thought to play an important role in the generation of several of these antiangiogenic proteins.

Angiostatin has been characterized as a 38-kDa internal fragment of plasminogen that encompasses the first four disulfide-linked domains of the molecule (Kringles (K)1–4).7 Angiostatin has been described to exert potent antiangiogenic and antitumor activities in a variety of murine tumor models.7, 8, 9, 10 Similarly, a 20-kDa (184 amino acid (aa)) C-terminal fragment of collagen XVIII, termed endostatin (ES), is reported to be a potent inhibitor of tumor angiogenesis and micrometastases development.5, 11 The urokinase plasminogen activator (uPA) binds to its receptor uPAR via its light chain fragments,12 known as amino-terminal fragment ATF (residues 1–135). This interaction involves the EGF-like domain of ATF and its derivatives. The ATF, devoid of the uPA B-chain, does not display enzymatic activity and has been previously described to prevent lung carcinoma metastasis13 and to protect mice in a liver metastasis model of human colon carcinoma.14 Canstatin (Can), an endogenously produced fragment of the non-collagenous 1 (NC1) domain of the α2 chain of type IV collagen (24 kDa), was recently shown to inhibit endothelial cell proliferation and tube formation in vitro and to suppress tumor growth in several human xenograft mouse models.6

However, K1–3, ES, mouse ATF and Can exhibit a relatively low molecular weight (respectively 38, 20, 14 and 24 kDa), which favors their efficient clearance by the kidneys. In a previous study, we showed that genetic coupling of mATF or K1—3 of angiostatin to human serum albumin (HSA): (i) preserved their antiangiogenic properties and (ii) dramatically improved their serum half-lives.15, 10

Protein production, long-term storage of bioreactive proteins and cumbersome daily administration are serious hurdles in this context. In contrast, the antitumor effects that follow antiangiogenic gene delivery from a variety of viral and non-viral vectors in vivo highlight the potential of antiangiogenic gene therapy as an additional strategy for treatment against cancer and metastatic dissemination.10, 16, 17 We chose the highly efficient adenovirus-based gene delivery system to evaluate the antitumor potency of several conjugates between HSA and antiangiogenic factors (K1–3, ES, ATF and Can) in TRP-1/SV40Tag transgenic mice, which spontaneously develop eye tumors with brain metastases.18 Here, we present, for the first time, a comparative study of the antimetastatic and antiangiogenic activities of these fused factors in a metastatic ocular tumor model.

Materials and methods

Construction and propagation of recombinant adenoviruses

AdATF-HSA, AdK1–3-HSA, AdES-HSA and AdCan-HSA are ΔE1–ΔE3 recombinant adenoviruses that express, respectively, a genetic conjugate generated by polymerase chain reaction between human serum albumin (HSA) and murine ATF, angiostatin K1–3, ES or Can, under the control of the cytomegalovirus (CMV) promoter.

The gene encoding the ATF-HSA genetic conjugate (745 residues) can be described from its 5′ to 3′ end as the succession of: (i) a fragment encoding the native secretion signal of murine urokinase (uPA) and the first 135 residues of murine uPA (ATF), (ii) genetically linked to the 586 residues of mature HSA.15 The fused gene K1–3-HSA contains the native signal peptide followed by the first 333 residues of mature human plasminogen (K1–3), which is genetically linked to the 586 aa of mature HSA.10 The fused gene ES-HSA contains the murine-uPA signal peptide followed by the last 183 C-terminal residues of murine collagen XVIII, genetically linked to the 586 aa of HSA. The fused gene Can-HSA expresses (i) the Pre-Pro signal sequence of HSA (24 residues), (ii) the C-terminal NC1 domain of the human α2 chain of type IV collagen also named Can (Can) and (iii) the 586 residues of mature HSA.19

The ATF-HSA, K1–3-HSA and ES-HSA genetic conjugates were first inserted into the pCEP4 vector (Invitrogen, Cergy Pontoise, France) between the CMV promoter and the SV40 polyA. The CMV-conjugate-polyA expression cassettes were then cloned in the shuttle plasmid pXL3048. pXL3048 is a KmR-SacB-ColE1 plasmid containing the left end of the Ad5 genome (nucleotides 1–386), and a part of the Ad5 pIX gene (nucleotides 3446–4296). The recombinant adenoviral genomes encoding the HSA conjugates were generated by homologous recombination between shuttle plasmids and pXL3215 in Escherichia coli, as previously described.37 pXL3215 contains the Ad5 genome deleted for E1 (nucleotides 386–3446) and E3 (nucleotides 28592–30470) bordered by two PacI sites. The Can-HSA expression cassette was cloned in the pGB1, a KmR-oriR6K shuttle vector containing the left end of the Ad5 genome (nucleotides 1–382), flanked on the left by a PacI site, and part of the Ad5pIX gene (nucleotides 3513–4296). The recombinant adenoviral genome Can-HSA was obtained in one step by homologous recombination in E. coli between the shuttle plasmid and pOSE1000, a TetR IncP replicon.38

Recombinant adenovirus expressing no transgene (AdCO1) was used as a negative control. AdES is a ΔE1–ΔE3 adenovirus that expresses ES.30 In the same manner, AdCan expresses Can.

Recombinant viruses were expanded in 293 cells and purified by two-step CsCl gradient ultracentrifugation. Viral stocks were desalted using Pharmacia G50 columns (Orsay, France) and frozen at −80°C in phosphate-buffered saline (PBS) containing 10% glycerol. Viral titers were determined following infection of 911 cells and expressed as PFU/ml.

In vitro assays

For proliferation assay, subconfluent human mammary epithelial cel (HMEC)-1 cells were infected at different multiplicity of infection (MOIs) with AdES-HSA, AdES or AdCO1 in 24-well culture plates for 96 h. Supernatants were then discarded and cells were incubated with 250 μl of PBS and 25 μl of 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) (Sigma, Saint Quentin Fallarier, France) at 5 mg/ml. After 2 h incubation at 37°C, 250 μl of lysis buffer (20% sodium dodecyl sulfate (SDS)–33% DiMethylFormamide–2% acetic acid–0.025 N HCl–0.05 N NaOH) were added and incubation continued overnight. Two hundred microliters of each sample were distributed in 96-well plates for optical density reading at 570 nm.

To determine cytotoxicity of the HSA conjugates, subconfluent HUVEC cells (Cambrex, Verviers, Belgium) were infected at MOI 600 for 96 h in 0.2% fetal bovine serum – medium supplemented with bFGF (medium kit provided by Cambrex). After infection, living and dead cells were revealed by the Annexin V – PI kit (PN IM3546, Immunotech, Beckman Coulter, Villepinte, France), and analyzed by flow cytometry.

Transgenic mouse model

We worked with transgenic mice, which spontaneously developed metastatic eye tumors.18 In this line, 1.4 kb of the tyrosine-related protein 1 (TRP-1) promoter was fused to the SV40 Tag transforming sequence, and inserted on the Y chromosome leading to 100% transgenic males which develop tumors and brain metastases.

For the study of metastases size, TRP1/SV40 Tag transgenic mice were injected with 109 PFU of either recombinant Ad (AdATF-HSA, n=15; AdK1–3-HSA, n=14; AdES-HSA, n=14; AdCan-HSA, n=16) or AdCO1 (n=14) (in a volume of 50 μl) 3 days after birth in the temporal vein. A second injection was performed 15 days after birth in the retro-orbital vein. In addition, 13 mice were injected in the same manner with PBS. Blood was collected weekly from day 15 and until animals were killed, that is, 62 days after birth. Sera were collected from blood samples and stored at −20°C until tested. Circulating antiangiogenic chimeric proteins were measured by sandwich enzyme-linked immunosorbent assay (ELISA).

For the experiment of survival probability, transgenic mice were i.v. injected as above with recombinant adenovirus coding for antiangiogenic factors (AdATF-HSA, n=14; AdK1–3-HSA, n=13; AdES-HSA, n=13; AdCan-HSA, n=15) or AdCO1 (n=14). Blood samples were not collected.

Immunoassay of K1–3-HSA

The serum level of K1–3-HSA was quantified by a sandwich ELISA. Ninety-six-well microtiter plates (Nunc Maxisorb, Roskilde, Denmark) were coated with 100 ng/ml of a goat anti-human angiostatin antibody (AF226, R&D) diluted in PBS. After overnight incubation at 4°C, the wells were washed six times with TBST (Tris-buffered saline TBS, 0.02% Tween 20) and then incubated with TBST–5% non-fat dry milk (TBSTM) for 3 h at room temperature under shaking conditions. Once washed, as described above, wells were incubated in TBST–5% milk with serial dilutions of mice sera or standard human angiostatin K1–3 solution (ref 176 705, Calbiochem VWR International Strasbourg, France). After a 2 h incubation with shaking, wells were washed and incubated over 2 h with 50 μl of a mouse anti-human angiostatin antibody diluted at 500 ng/ml in TBST–5% milk. After washing, as described above, wells were incubated with 50 μl of an alkaline phosphatase-conjugated goat anti-mouse IgG (H+L) antibody (S3721, Promega, Madison, WI, USA) diluted to 1:5000 in TBST–5% milk. After 1 h incubation and washing, wells were then incubated for 30 min with the developing solution (alkaline phosphatase kit, Biorad Laboratories, Richmond, CA). The A405 was determined by using a microplate reader (BioRad, Richmond, CA, USA).

Immunoassay of ATF-HSA

A 96-well plate (Nunc Maxisorb, Roskilde, Denmark) was coated with 500 ng/well of a rabbit anti-human albumin antibody (A0433, Sigma) diluted in PBS. After overnight incubation at 4°C, wells were washed six times with TBST and blocked with TBSTM for 3 h at room temperature on a rocking platform shaker. Once washed, as described above, wells were incubated for 1 h at room temperature with 50 μl of serial dilutions of mice sera prepared in TBSTM. After a 2-h incubation with shaking and washing as mentioned above, wells were incubated for 1 h at room temperature with 50 μl of antimurine ATF antibody (H77A10, Molecular Innovations, South Field, MI, USA) diluted to 300 ng/ml in TBSTM. After washing, wells were incubated with 50 μl of alkaline phosphatase-conjugated goat anti-mouse IgG (H+L) antibody (S3721, Promega) diluted to 1:5000 in TBSTM. After incubation and washing as described above, mATF-HSA was detected after a 20-min incubation in the dark with 50 μl of developing solution (alkaline phosphatase substrate kit; BioRad Laboratories, Richmond, CA). Absorbance was measured at 405 nm. Recombinant mATF-HSA (lab purification from infection supernatants) was used as a standard.

Immunoassay of ES-HSA

The serum ES-HSA level was determined with the ChemiKineTM mouse endostatin kit (Chemicon International Inc.). Briefly, precoated anti-rabbit antibody plates were used to capture a specific ES complex in each sample consisting of ES antibody, standard or samples, and biotinylated ES. The biotinylated ES conjugate (competitive ligand) and the sample or standard compete for ES-specific antibody binding sites. Therefore, as the concentration of ES in the samples increases, the amount of biotinylated ES captured by the antibody decreases. The assay is visualized using a streptavidin alkaline phosphatase conjugate and it is followed by a chromatographic substrate reaction.

Immunoassay of Can-HSA

The serum level of Can-HSA was quantified by a sandwich ELISA directed against the HSA part of the conjugate. Ninety-six-well microtiter plates (Nunc Maxisorb) were coated with 1 μg/ml of a goat anti-human albumin monoclonal antibody (Biovalley A101 ab 7793-100, Marne la Vallee, France) diluted in PBS. After overnight incubation at 4°C, the wells were washed six times with TBST and then incubated with TBST–5% gelatin for 3 h at room temperature under shaking conditions. Once washed, as described above, wells were incubated in TBST–5% gelatin with serial dilutions of mice sera or standard HSA solution (ref A-8763, Sigma). After a 2 h incubation with shaking, wells were washed and incubated for 2 h with 50 μl of a rabbit anti-HSA antibody diluted (ref A0433, sigma) to 250 ng/ml in TBST–5% gelatin. After washing, as described above, wells were incubated with 50 μl of an alkaline phosphatase-conjugated goat anti-mouse IgG (H+L) antibody (S3721, Promega) diluted to 1:5000 in TBST–5% milk. After 1 h incubation and washing, wells were then incubated for 30 min with the developing solution (alkaline phosphatase kit, Biorad Lab.). The A405 was determined by using a microplate reader (BioRad).

Immunoassay of HSA antibodies

The presence of HSA-specific antibodies in the sera was determined by ELISA. Ninety-six-well microplates (Nunc Maxisorb) were coated overnight with 100 ng of human albumin (A-8763, Sigma), washed with TBST and blocked with TBSTM. Serial dilutions of the sera were then added. Bound antibodies were detected with peroxidase-conjugated anti-mouse IgG (Jackson Immunoresearch Laboratories, West Grove, PA) goat antibodies diluted to 1:5000. The plate was revealed by incubation with the Sigma Fast OPD peroxidase substrate tablet (P9187, Sigma) for 30 min. The reaction was stopped by adding 3 N HCl, and a spectrophotometric reading was obtained at 490 nm.

Immunohistochemistry and image analysis

Skulls from TRP1/SV40 Tag transgenic mice were removed 62 days after birth and fixed in Glyofixx (ThermoShandon, Pittsburg, PA), decalcified in TDE30 Sakura solution (Japan) and embedded in paraffin for brain metastases detection. Five-μm sections were prepared for each animal and routinely stained with HES (hematoxylin-eosin-saffranin) for histological analysis. The sections were examined and relevant fields at × 50 magnification were digitized with a color Sensicam digital camera (PCO, Germany, resolution 1280 × 1024 pixels). Brain metastatic nodule sizes were determined using the calibration tool of the Matrox Inspector 2.2 software (Matrox, Dorval, Canada) and volumes were extrapolated.

Immunohistochemistry was used to assess the extent of intratumor vascularization within the different experimental groups. Sections for immunochemistry were pretreated for 10 min in 3% H2O2 and incubated for 1 h with a monoclonal mouse anti-human smooth muscle actin (Dako, clone HHF 35; dilution 1:100). This primary antibody was revealed with Histomouse-Max Kit (Zymed Inc., San Francisco, CA) as a secondary step. Sections were incubated in the presence of streptavidin-peroxydase for 15 min. The sections were finally incubated with a red substrate, aminoethylcarbazole (AEC) for 10 min, washed, counterstained with Meyer's hematoxylin and mounted in aqueous medium (Pertex, Histolab, Microm, Francheville, France).

Four fields per tumor were randomly digitized at magnification × 200. An image processing algorithm using the Matrox Inspector 2.2 Software was specially developed to quantify the surface of intravascular lumens. A mean ratio of vascular lumen surfaces vs carcinomatous surface was established for each field, and a mean value was obtained, taking into account four fields per animal. Reproducibility was ensured by keeping the same settings throughout the procedure.

In addition, number of microvessels in treated- vs untreated-mice was measured. Four fields per tumor ( × 200 magnification) were randomly selected and analyzed, microvessels were counted, and a mean value was obtained as above.


Construction and characterization of the recombinant adenoviruses encoding the ATF-HSA, K1–3-HSA, ES-HSA and Can-HSA conjugates.

We inserted a chimeric gene coding for a conjugate between ATF, angiostatin (K1–3), ES, or Can and HSA (HSA) under the control of the immediate early promoter of CMV into an E1/E3-deleted recombinant adenovirus. The ES-HSA chimeric protein contains the murine uPA signal peptide followed by the last 183 C-terminal residues of murine collagen XVIII, genetically fused to mature HSA. The K1–3-HSA chimeric protein contains the native secretion signal and the first three Ks of human plasminogen, genetically fused to HSA.10 The ATF-HSA chimeric protein contains the native secretion signal and the first 135 residues of murine urokinase (uPA) linked to 586 residues of mature HSA.15 Can-HSA describes the C-terminal NC1 domain of the human α2 chain of type IV collagen (Can, 24 kDa) fused with HSA.

To further characterize the activity of the ES-HSA chimeric protein expressed by the virus, we infected human endothelial HMEC-1 cells at various MOI of virus to perform an in vitro proliferation assay. After 96 h of infection, the percentage of surviving cells was determined by addition of the MTT reagent. AdCO1, an ‘isogenic’ adenovirus vector lacking the expression cassette, was used as a negative control, and the AdES virus coding for non-fused endostatin was used as a positive one. Infection with AdES and AdES-HSA induced a dose-dependent inhibition of endothelial cell proliferation. Survival of 14.8 and 26.8% at MOI 600 was obtained after infection of HMEC-1 cells with AdES and AdES-HSA respectively, compared with a 100% survival following AdCO1 infection (P<0.001, Student's test) (Figure 1). AdCO1 did not affect HMEC proliferation, indicating that inhibition of proliferation was specifically linked to the expression of ES and ES-HSA proteins. These results show that the genetic coupling of ES to HSA does not significantly modify its ability to inhibit endothelial cell proliferation. Similar results have previously been reported for K1–3, K1–3-HSA: survival of 58.1 and 50.6% at MOI 600 was obtained after HMEC-1 infection with AdK3 and AdK1–3-HSA, respectively.10 mATF and mATF-HSA, are antagonists of the uPA/uPAR system involved in the degradation of the extracellular matrix, a key step for cell migration and angiogenesis, and induce inhibition of plasminogen activation. We have previously shown that ATF and ATF-HSA inhibit plasminogen activation and uPA activity: 30.9% plasminogen activation with AdATF-HSA vs 20.8% with AdATF at MOI 100.15

Figure 1

(a) Inhibition of endothelial cell proliferation. HMEC-1 cells were infected at different MOIs for 96 h with AdES-HSA, AdES and AdCO1. (P<0,001, Student's t-test). (b) Cytotoxic effect of the HSA conjugates. Endothelial HUVEC cells were infected at MOI 600 for 96 h. Dead cells were detected by flow cytometry as described in Materials and methods.

As shown in Figure 1b, K1–3-HSA, ES-HSA and Can-HSA conjugates appeared to be cytotoxic factors on HUVEC endothelial cells, as most of cells are destroyed 96 h after virus infection at MOI 600 comparing to AdCO1 control infection. Results following AdATF-HSA infection did not differ significantly from those obtained with AdCO1 infection. Furthermore, Magnon et al.19 have published a better induction of endothelial cell apoptosis with AdCan-HSA than with AdK1–3-HSA.

Growth of TRP1 brain metastases is differentially inhibited by the systemic expression of the conjugates

The ability of each recombinant adenovirus to provide systemic inhibition of metastasis was determined in TRP1/SV40 Tag transgenic mice which spontaneously develop eye tumors with brain metastases.18 These mice express the SV40 Tag under the control of the TRP-1 promoter. TRP-1 is an enzyme expressed during development in melanocytes/melanoblasts and in the retinal pigmented epithelium. This enzyme catalyzes steps following the action of tyrosinase, and modulates the type of melanin produced. At E12.5 during development, several preneoplastic foci were observed in the retinal pigmented epithelium of the TRP1/SV40 Tag transgenic littermates. Tumor cell proliferation was detected in the region of the optic nerve and near the anterior chamber, where tumor cell aggregates expanded into larger masses of neoplastic tissue. The tumor consisted of epithelioid cells arranged in a tubular fashion, similar to human adenocarcinoma of the retinal pigmented epithelium. At 1 month of age, transgenic mice exhibited visible enlargement of the eye that led to complete degeneration of the eyeball after 2 months. At this stage, tumor cells invaded the optic nerve. At 3 months, dissemination of metastases to the brain and lymph nodes was observed in all mice, and to the spleen in about 30% of the mice, leading to the death of the mice.

We injected 109 PFU of AdATF-HSA (n=15), AdK1–3-HSA (n=14), AdES-HSA (n=14), AdCan-HSA (n=16) or AdCO1 (n=14) in the temporal vein in 3-day-old TRP1/SV40 Tag transgenic male mice, before maturation of the immune system. A second i.v. injection (109 PFU) in the retro-orbital vein was performed at day 15, to compensate for gene dilution due to the growth of mice. In addition, 13 TRP1/SV40 Tag transgenic mice were injected with PBS in the same protocol.

At 2 months of age, mice exhibited primary ocular tumors with a similar evolution whatever the expressed conjugate. Mice were killed for histological analysis of the brain. In the control groups (PBS and AdCO1), metastases were huge with multiple intraparenchymal nodules. The surface of these metastases was then measured by numerical analysis on each histological slide and volume extrapolated. The volume of brain metastases in AdCO1- and PBS-injected mice was not statistically different. Injection of AdATF-HSA did not significantly reduce metastatic growth compared to that observed in the AdCO1 control group (mean brain tumor volume was, respectively, 3.93±3.33 and 4.04±5.12 mm3 (Figure 2). In contrast, mice treated with AdK1–3-HSA, AdES-HSA and AdCan-HSA had significantly smaller metastases than controls (the mean volume of the metastases was, respectively, 1.19±1.19, 0.87±1.5, 0.43±0.56 mm3) (respectively P=0.036, P=0.0054, P=0.0005). Moreover, further inhibition of metastatic growth was achieved more strongly by AdCan-HSA than by AdK1–3-HSA (P=0.04). The difference in the volume of brain metastases between mice treated with AdES-HSA and AdCan-HSA-treated mice did not reach statistical significance (P=0.66).

Figure 2

Inhibition of metastasis in TRP1/SV40 Tag transgenic mice. Newborn TRP1/SV40 Tag male mice received an i.v. injection at days 3 and 15 after birth of PBS (n=13), or 109 PFU of AdCO1 (n=14), AdATF-HSA (n=15), AdK1–3-HSA (n=14), AdES-HSA (n=14) or AdCan-HSA (n=16). Data represent the volume of metastases (mean volume±s.d.) for each group. Global statistical significance was determined with a Kruskall Wallis test (P=0.0001). Statistical significance between control AdCO1 and AdK1–3-HSA, AdES-HSA or AdCan-HSA was determined with a Wilcoxon test (respectively P=0.036, P=0.0054, P=0.0005).

Mortality of TRP1 mice is differentially delayed by the systemic expression of the conjugates

In order to determine if this inhibition of brain metastases development improves the survival of treated mice, we performed a long-term follow-up of mice i.v. injected at days 3 and 15 after birth with 109 PFU of either AdCO1 (n=14), AdATF-HSA (n=14), AdK1–3-HSA (n=13), AdES-HSA (n=13) or AdCan-HSA (n=15). Figure 3 showed a significant survival advantage for all the animals except for those injected with AdCO1 or AdATF-HSA. The maximum duration of survival of the AdCan-HSA animals was 138 days whereas the maximum survival of mice treated with AdK1–3-HSA and AdES-HSA was, respectively, 122 and 125 days. Median survival was thus improved by 4 weeks (78 days with AdCO1 vs 101, 97 and 110 days with, respectively, AdK1–3-HSA, AdES-HSA and AdCan-HSA P<0.0001).

Figure 3

Survival of male TRP1/SV40 Tag mice treated with recombinant adenovirus coding for antiangiogenic factors. Male transgenic mice received an i.v. injection (at 3 and 15 days) of 109 PFU of AdCO1 (n=14, ▪), AdATF-HSA (n=14, ♦), AdK1–3-HSA (n=13, ), AdES-HSA (n=13, •) and AdCan-HSA (n=15, ). Statistical significance was determined using a Kaplan–Meier test.

Taken together, these results demonstrate that early systemic injections of AdK1–3-HSA, AdES-HSA or AdCan-HSA in TRP1/SV40 Tag transgenic mice slow down the implantation and growth of brain metastases and prolong their survival. Moreover, Can-HSA appears to be the most effective of these antiangiogenic factors.

Vascularization of brain metastases is differentially inhibited by the systemic expression of the conjugates

To characterize the antitumor effect associated with the expression of the ATF-HSA, K1–3-HSA, ES-HSA and Can-HSA conjugates, we collected the skull of TRP1/SV40 Tag-treated mice (injected twice, at days 3 and 15) 60 days after birth and evaluated neoangiogenesis in brain tissue. Vascularization of brain metastases was determined on brain sections after α-smooth muscle actin immunostaining of blood vessels. This marker recognizes capillary constitutive elements, such as péricytes, even at an early stage of vascularization development. Very large and thick vessels with a mature lumen were seen in brain metastases from AdC01-treated mice but not in treated groups (all animals) (Figure 4a–e). However, AdATF-HSA-treated mice exhibited numerous microvessels (Figure 4b). The surface of optically empty spaces located within actin-positive vessel walls was quantified by image analysis, which provides a mean score for each sample. As shown in (Figure 4f), intratumor vessels surfaces vessels were markedly reduced within the AdATF-HSA-, AdK1–3-HSA-, AdES-HSA- and AdCan-HSA-treated groups compared to the control group (P=0.01, P=0.002; P=0.003, P=0.002, respectively).

Figure 4

Immunohistological analysis of vascularization of brain metastases after systemic treatment. 109 PFU of each adenovirus was injected i.v. into TRP1/SV40 Tag mice at days 3 and 15 after birth. Mice were killed at 2 months after birth. Paraffin-embedded brain sections from mice treated with AdCO1 (a), AdATF-HSA (b), AdK1–3-HSA (c), AdES-HSA (d) and AdCan-HSA (e) were submitted to α-smooth muscle actin immunostaining (original magnification × 50). Four fields (original magnification × 200) per tumor were digitized and the surfaces of optically empty spaces located inside vessel walls were quantified by image analysis which provides a mean score for each sample (f). The surface of vessels lumens in AdATF-HSA-, AdK1–3-HSA-, AdES-HSA- and AdCan-HSA-treated mice was significantly different from that of the AdCO1 group (Wilcoxon test, P=0.01; P=0.002; P=0.003; P=0.002, respectively). There was no significant difference between the four treated groups.

The number of microvessels per field has been used extensively as a marker for tumor angiogenesis, and its inhibition has been used as a measure of antiangiogenic activity.39 We failed to observe any significant decrease in the density of blood vessels in brain metastases of AdCO1 controls and AdATF-HSA-treated mice (9.44±4.24 vessels per field and 6.77±1.77 vessels per field, respectively, P=0.35). In sharp contrast, the density of blood vessels was drastically reduced in AdK1–3-HSA, AdES-HSA and especially in AdCan-HSA-treated mice (3.36±1.47, P=0.01; 2.7±1.6, P=0.009; 1.6±1.19 vessels per field, P=0.009, respectively).

Determination of in vivo expression of angiogenesis inhibitors

In order to evaluate the systemic expression of angiogenesis inhibitors, sera of TRP1/SV40 Tag-treated mice were collected weekly after the second injection of the ATF-HSA, K1–3-HSA, ES-HSA and Can-HSA recombinant adenoviruses and monitored by ELISA (Figure 5, Table 1).

Figure 5

Expression level of antiangiogenic factors in transgenic TRP1/SV40 Tag mice. Sera were collected at different times after the two injections of 109 PFU of AdATF-HSA (a), AdK1–3-HSA (b), AdCan-HSA (c) or AdES-HSA (d) and analyzed by ELISA (see Materials and methods). Each triangle represents individual mouse data, and the bar graph depicts the mean data values.

Table 1 Comparison, between the antiangiogenic molecules detected in the mice sera and the size of the brain metastasis (at day 42 after the first injection)

The concentration of ATF-HSA (A) peaked at 14.8 μ M, 21 days after the first injection and decreased to 6 μ M after 56 days. K1–3-HSA (B) reached a level of 114.6 nM in 15-day-old treated mice and decreased slowly to 61.7 nM after 56 days. Can-HSA (C) reached a level of 137 nM in 21-day-old treated mice and decreased slowly to 71 nM after 56 days, whereas the ES-HSA serum level (D) in AdES-HSA-treated mice remained fairly stable during the experiment, with a peak at 13.1 nM and remained at 12.5 nM at day 56. Following AdC01 injections, no ATF-HSA, K1–3-HSA, ES-HSA or Can-HSA was detectable in the sera of TRP1-Tag mice.

We assessed the immune response to the HSA component of the chimeric molecules by analyzing the presence of IgG anti-HSA in mouse sera. We did not detect any antibodies directed against the transgene product in the sera of the treated mice (data not shown). This data could be explained by the schedule of injections. The first i.v. injection of the recombinant vectors was performed at 3 days of age, whereas the immune system was not yet mature. Consequently, we tolerize the mice against the adenovirus but also against the product of each transgene, allowing the second injection at 15 days of age. Thanks to this protocol, chimeric antiangiogenic factors were stably expressed during the experiment length.


Recently, angiostatic genes have been thrust into the limelight because it may be possible to deliver them by gene transfer methods. Angiogenesis seems to be an important prognostic factor in uveal melanomas.20, 21, 22, 23, 24, 25, 26 The above-mentioned studies provide important information on the relative potency of a number of antiangiogenic gene products previously shown to possess antitumor activity.10, 11, 13, 14, 27, 28, 29, 30, 31 However, there is a dearth of information on the antitumor activity of Can. Our objective was to determine the sensitivity of an intraocular tumor to four endogenous angiogenesis inhibitors stabilized by HSA genetic coupling and delivered by viral vectors in the TRP1/SV40 Tag transgenic mouse model, and to possibly identify the most effective agent in this setting.

Our findings indicate that K1–3-HSA, ES-HSA and Can-HSA possess significantly more potent antimetastatic activity than ATF-HSA when delivered via gene transfer. Moreover, mice treated with AdCan-HSA showed much slower tumor growth than AdK1–3-HSA-treated mice. The inability of AdATF-HSA to inhibit tumor growth could reflect two hurdles encountered in our study: the multiplicity of neoplastic foci and the fact that the soluble form of ATF-HSA we used may have been suboptimal. Although ATF-HSA did not delay the growth of metastases, it was highly expressed in the sera of TRP1/SV40 Tag-treated mice, as confirmed by the ELISA analysis of sera samples.

The heterogeneous antitumor effects observed in vivo might be related to the levels of angiostatin, ES, Can and the amino-terminal fragment of uPA. As expression levels can vary widely between different tumors following gene delivery, they should be considered as a potential limiting factor of such approaches. In fact, the ATF-HSA levels produced by transduced liver cells were approximately 100-fold higher than those produced in the sera of AdK3-HSA- and AdCan-HSA-treated mice. These ELISA quantification were obtained by sandwich ELISAs using a controlled standard solution for K1–3-HSA, ES-HSA and Can-HSA. However, for the ATF-HSA ELISA, we produced by ourselves the recombinant ATF-HSA protein used as a standard. The difference observed in the quantifications may be also explained by the different quality of the standards. The expression levels we attained, probably represent a theoretical ‘maximum’ reflecting the inherent pharmacokinetic properties governing the circulating levels of each protein that can be achieved via gene transfer. The serum half-life of antiangiogenic factors is rather limited.30 Such high clearance may be a direct consequence of their low molecular weight and efficient renal filtration. Indeed, continuous administration of these agents is required to maintain their therapeutic activity in vivo.8 Moreover, we already showed that HSA stabilization enhanced K1–3 and ATF bioavailability without altering their biological effect.10, 15 Furthermore, adenovirus-derived vectors are particularly interesting for the evaluation of new therapeutic genes for systemic cancer therapy because they can efficiently transduce the host liver and then use it as an endogenous factory for the production of high circulating levels of the therapeutic proteins.

The inhibitory effects on metastasis were closely correlated with markedly decreased intratumor vascularization. Although the four molecules exhibited antiangiogenic activity as compared with that observed in AdCO-1-treated control mice, microvessel density was significantly more limited in mice treated with AdK1–3-HSA, AdES-HSA and AdCan-HSA than in AdATF-HSA-treated mice. Antiangiogenic activity was also correlated with antimetastasis activity. However, intratumor vascularization did not differ significantly between AdK1–3-HSA-, AdES-HSA and AdCan-HSA-treated mice. This study indicates that the main antitumor mechanism was the antiangiogenic effect of these molecules. We cannot exclude an additional direct antiproliferative effect of ES-HSA, as demonstrated in other studies,30 and of Can-HSA19 on tumor cells but this was not substantiated in our in vivo experiments. In this study, the ATF-HSA conjugate appeared to be the least effective. Its antagonist effect of the uPA/uPAR system seems to not be a therapeutic target in the TRP1/SV40 Tag transgenic mouse model.

The studies by Kuo et al.17 and De Boüard et al.32 suggested that the soluble form of Flk-1, Flt-1 and INF-α possessed significantly more potent antitumor activity than angiostatin or ES when delivered via gene transfer. A likely explanation for this diminished effect is an insufficient concentration of the antiangiogenic factors at the tumor site because of their short half-life in the circulation. In contrast, in our study, it is highly likely that far greater levels of the chimeric proteins were obtained after adenoviral-mediated gene transfer.

A strong factor in this transgenic model is that cancers arise from normal cells in their natural tissue microenvironments and progress through multiple stages, just like human cancer. None of the agents tested completely blocked the growth of small metastases, or completely eradicated the lethal tumor burden. However, it is important to recognize that TRP-1/SV40 Tag oncogene expression occurred in all retinal pigment epithelial cells in these transgenic mice. This resulted in the development of numerous neoplastic foci, which makes this system particularly stringent for assaying pharmacological agents. Although multifocal disease may represent a tough standard in comparison to standard tumor transplant models, such stringency is likely to prove of considerable value, given the biological and clinical resilience of human cancers.

Our findings indicate that Can-HSA is the most active single agent against our intraocular brain metastatic tumor. As Can and ES could exert angiostatic effects at low and constant doses, which are readily obtained with current gene delivery systems as shown here, it is tempting to speculate that systemic antiangiogenic factor production by genetically modified adenovirus might represent an effective way of delaying metastatic growth through the inhibition of angiogenesis.

Choroidal melanoma is a rare pathology, with an incidence of 6–7 per million annually in France and in the USA.33, 34, 35 Approximately, 35% of patients with uveal melanoma develop metastases.36 Histologic sections of aggressive choroidal melanomas and their metastases contain perfusion channels that are associated with extracellular matrix.20, 21 The histologic detection of loops, networks and crosslinking parallel matrix was strongly associated with death from metastatic melanoma.22 Our findings indicate that antiangiogenic therapy represents a promising approach to uveal melanoma treatment and prevention of metastases.


  1. 1

    Folkman J . Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med 1995; 1: 27–31.

    CAS  Article  Google Scholar 

  2. 2

    Hanahan D, Folkman J . Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996; 86: 353–364.

    CAS  Article  Google Scholar 

  3. 3

    Holash J, Maisonpierre P, Compton D, Boland P, Alexander CR, Zagzag D et al. Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF. Science 1999; 284: 1994–1998.

    CAS  Article  Google Scholar 

  4. 4

    Zagzag D, Amirnovin R, Greco MA, Yee H, Holash, Wiegand SJ et al. Vascular apoptosis and involutionin gliomas precede neovascularisation: a novel concept for glioma growth and angiogenesis. Lab Invest 2000; 80: 837–849.

    CAS  Article  Google Scholar 

  5. 5

    O'Reilly MS, Boehm T, Shin Y, Fukai N, Vasios G, Lane WS et al. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 1997; 88: 277–285.

    CAS  Article  Google Scholar 

  6. 6

    Kamphaus GD, Colorado PC, Panka DJ, Hopfer H, Ramchandran R, Torre A et al. Canstatin, a novel matrix-derived inhibitor of angiogenesis and tumor growth. J Biol Chem 2000; 275: 1209–1215.

    CAS  Article  Google Scholar 

  7. 7

    O'Reilly MS, Holmgren L, Shing Y, Fukai N, Vasios G, Lane WS et al. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 1994; 79: 315–328.

    CAS  Article  Google Scholar 

  8. 8

    O'Reilly MS, Holmgren L, Chen C, Folkman J . Angiostatin induces and sustains dormancy of human primary tumors in mice. Nat Med 1996; 2: 689–692.

    CAS  Article  Google Scholar 

  9. 9

    Griscelli F, Li H, Bennaceur-Gricelli A, Soria J, Opolon P, Soria C et al. 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 1998; 95: 6367–6372.

    CAS  Article  Google Scholar 

  10. 10

    Bouquet C, Frau E, Opolon P, Connault E, Abitbol M, Griscelli F et al. Systemic administration of a recombinant adenovirus encoding a HSA-angiostatin kringle 1–3 conjugate inhibits MDA-MB-231 tumor growth and metastasis in transgenic model of spontaneous eye cancer. Mol Ther 2003; 7: 174–183.

    CAS  Article  Google Scholar 

  11. 11

    Sauter BV, Martinet O, Zhang WJ, Mandeli J, Woo SL . Adenovirus-mediated gene trasnfer of endostatin in vivo results in high level of transgene expression and inhibition of tumor growth and metastases. PNAS 2000; 97: 4802–4807.

    CAS  Article  Google Scholar 

  12. 12

    Kobayashi H, Gotoh J, Hirashima Y, Fujie M, Sugino D, Terao T . Inhibitory effect of a conjugate between human urokinase and urinary trypsin inhibitor on tumor cell invasion in vitro. J Biol Chem 1995; 270: 8361–8366.

    CAS  Article  Google Scholar 

  13. 13

    Li H, Lu H, Griscelli F, Opolon P, Sun LQ, Ragot T et al. Adenovirus-mediated delivery of a uPA/uPAR antagonist suppresses angiogenesis-dependant tumor growth and dissemination. Gene Ther 1998; 5: 1105–1113.

    CAS  Article  Google Scholar 

  14. 14

    Li H, Griscelli F, Lindenmeyer F, Opolon P, Sun LQ, Connault E et al. Systemic delivery of antiangiogenic adenovirus AdmATF induces liver resistance to metastasis and prolongs survival of mice. Human Gene Ther 1999; 10: 3045–3053.

    CAS  Article  Google Scholar 

  15. 15

    Apparailly F, Bouquet C, Millet V, Noel D, Jacquet C, Opolon P et al. Adenovirus-mediated gene tranfer of urokinase plasminogen inhibitor inhibits angiogenesis in experimental arthritis. Gene Therapy 2002; 2002: 192–200.

    Article  Google Scholar 

  16. 16

    Chen QR, Kumar D, Stass SA, Mixon AJ . Liposomes complexed to plasmids encoding angiostatin and endostatin inhibit breast cancer in nude mice. Cancer Res 1999; 59: 3308–3312.

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17

    Kuo CJ, Farnebo F, Yu EI, Christofferson R, Swearingen RA, Carter R et al. Comparative evaluation of the antitumor activity of antiangiogenic proteins delivered by gene transfer. Proc Natl Acad Sci 2001; 98: 4605–4610.

    CAS  Article  Google Scholar 

  18. 18

    Penna D, Schmidt A, Beerman F . Tumors of the retinal pigment epithelium metastasise to inguinal lymph nodes and spleen in tyrosinase-related protein 1/SV40T antigen transgenic mice. Oncogene 1998; 17: 2601–2607.

    CAS  Article  Google Scholar 

  19. 19

    Magnon C, Galaup A, Mullan B, Rouffiac V, Bouquet C, Bidart J-M et al. Canstatin acts on endothelial and tumor cells via mitochondrial damage initiated through interaction with alpha beta 3 and alpha beta 5 integrins. Cancer Res 2005; 65: 4353–4361.

    CAS  Article  Google Scholar 

  20. 20

    Folberg R, Pe'er J, Gruman LM, Woolson RF, Jeng G, Montague PR et al. The morphologic characteristics of tumor blood vessels as a marker of tumor progression in primary uveal melanoma: a matched case–control study. Hum Pathol 1992; 23: 1298–1305.

    CAS  Article  Google Scholar 

  21. 21

    Folberg R, Rummelt V, Parys-Van ginderdeuren R, Hwang T, Woolson RF, Pe'er J et al. The pronostic value of tumor blood vessel morphology in primary uveal melanoma. Ophthalmology 1993; 100: 1389–1398.

    CAS  Article  Google Scholar 

  22. 22

    Folberg R, Mehaffey M, Gardner LM, Meyer M, Rummelt V, Pe'er J . The microcirculation of choroidal and ciliary body melanomas. Eye 1997; 11: 227–238.

    Article  Google Scholar 

  23. 23

    Rummelt V, Folberg R, Rummelt C, Gruman LM, Hwang T, Woolson RF et al. Microcirculation architecture of melanocytic naevi and malignant melanoma of the ciliary body and choroid:A comparative histopathologic and ultrastructural study. Ophthalmology 1994; 101: 718–727.

    CAS  Article  Google Scholar 

  24. 24

    Rummelt V, Folberg R, Woolson RF, Hwang T, Pe'er J . Relation between the microcirculation architecture and the aggressive behavior of ciliary body melanomas. Ophthalmology 1995; 102: 844–851.

    CAS  Article  Google Scholar 

  25. 25

    Apte RS, Niederkorn JY, Mayhew E, Alizadeh H . Angiostatin produced by certain primary uveal melanoma cell lines impedes the development of liver metastases. 2001; 119: 1805–1809.

  26. 26

    Mueller AJ, Freeman WR, Schaller UC, Kampik A, Folberg R . Complex microcirculation patterns detected by confocal indocyanine green angiography predict time to growth of small choridal melanocytic tumors: MuSIC Report II. Ophthalmology 2002; 109: 22204–22214.

    Article  Google Scholar 

  27. 27

    Blezinger P, Wang J, Gondo M, Quesada A, Mehrens D, French M et al. Sytemic inhibition of tumor growth and tumor metastases by intramuscular administration of endostatin gene. Nat Biotechnol 1999; 17: 343–348.

    CAS  Article  Google Scholar 

  28. 28

    Chen CT, Lin J, Li Q, Phipps SS, Jacubczak JL, Stewart DA et al. Antiangiogenic gene therapy for cancer via systemic administration of adenoviral vectors expressing secretable endostatin. Hum Gene Ther 2000; 11: 1983–1996.

    CAS  Article  Google Scholar 

  29. 29

    Libutti SK . Antiangiogenic gene therapy of cancer utilizing a recombinant adenovirus to elevate systemic endostatin levels in mice. Cancer Res 2000; 60: 1503–1506.

    PubMed  PubMed Central  Google Scholar 

  30. 30

    Dkhissi F, Lu H, Soria C, Opolon P, Griscelli F, Liu H et al. Endostatin exhibits a direct antitumor effect in addition to its antiangiogenic activity in colon cancer cells. Hum Gene Ther 2003; 14: 997–1008.

    CAS  Article  Google Scholar 

  31. 31

    MacDonald NJ, Murad AC, Fogler WE, Lu Y, Sim BK . The tumor suppressing activity of angiostatin protein resides within kringles 1 to 3. Biochem Biophys Res Commun 1999; 264: 469–477.

    CAS  Article  Google Scholar 

  32. 32

    De Boüard S, Guillamo JS, Christov C, Lefevre N, Brugieres P, Gola E et al. Antiangiogenic therapy against experimental glioblastoma using genetically engineered cells producing interferon-alpha, angiostatin, or endostatin. Hum Gene Ther 2003; 14: 883–895.

    Article  Google Scholar 

  33. 33

    Murphy SF, Mahl CF, Bloom SM . Choroidal melanoma. Optomet Clin 1993; 3: 63–77.

    CAS  Google Scholar 

  34. 34

    Egan K, Seddon J, Glynn R, Gragoudas ES, Albert DM . Epidemiologic aspect of uveal melanoma. Surv Ophthalmol 1988; 32: 239–251.

    CAS  Article  Google Scholar 

  35. 35

    Vidal JL, Bacin F, Albuisson E, Rozan R, Desjardins L, D'Hermies F et al. ‘Melanome 92’ Etude épidémiologique des mélanomes uvéaux en France. J Fr Ophtalmol 1995; 18: 520–528.

    CAS  PubMed  Google Scholar 

  36. 36

    Diener-West M, Hawkins BS, Markowitz JA, Schatchat AP . A review of mortality from choroidal melanoma. II. A meta-analysis of 5-years mortality rates following enucleation, 1966 through 1988. Arch Ophthalmol 1992; 110: 245–250.

    CAS  Article  Google Scholar 

  37. 37

    Crouzet J, Naudin L, Orsini C, Vigne E, Ferrero L, Le Roux A et al. Recombinational construction in E. coli of infectious adenoviral genomes. Proc Natl Acad Sci 1997; 94: 1414–1419.

    CAS  Article  Google Scholar 

  38. 38

    Mullan B, Dugué C, Moutard V, Raoux D, Tremp G, Denefle P et al. Towards robust functional gene validation by adenoviral vectors: One Step E-coli-derived Recombinant Adenoviral Genome Construction. Gene Ther 2004; 11: 1599–1605.

    CAS  Article  Google Scholar 

  39. 39

    Weidner N, Folkman J, Pozza F, Bevilacqua P, Allred EN, Moore DH et al. Tumor anglogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma. J NaH Cancer Inst 1992; 84: 1875–1887.

    CAS  Article  Google Scholar 

Download references


We are very grateful to all the staff (in particular C Chianale and P Ardouin) in the animal facilities at the Institut Gustave Roussy for their help during the in vivo experiments. We thank Faroudy Boufassa (INSERM CHU de Bicêtre) for his help in statistical analysis. We also thank L St Ange for editing. La Ligue National Contre le Cancer, l'Association pour la Recherche Contre le Cancer (ARC), and le Centre National de la Recherche scientifique (CNRS) are acknowledged for financial support.

Author information



Corresponding author

Correspondence to C Bouquet.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Frau, E., Magnon, C., Opolon, P. et al. A gene transfer comparative study of HSA-conjugated antiangiogenic factors in a transgenic mouse model of metastatic ocular cancer. Cancer Gene Ther 14, 251–261 (2007). https://doi.org/10.1038/sj.cgt.7701005

Download citation


  • antiangiogenic factors
  • HSA conjugates
  • metastasis
  • transgenic mice

Further reading


Quick links