Cancer Cell Biology
Addition of integrin binding sequence to a mutant human endostatin improves inhibition of tumor growth
Article first published online: 20 MAY 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 111, Issue 6, pages 839–848, 10 October 2004
How to Cite
Yokoyama, Y. and Ramakrishnan, S. (2004), Addition of integrin binding sequence to a mutant human endostatin improves inhibition of tumor growth. Int. J. Cancer, 111: 839–848. doi: 10.1002/ijc.20336
- Issue published online: 3 AUG 2004
- Article first published online: 20 MAY 2004
- Manuscript Accepted: 2 MAR 2004
- Manuscript Revised: 27 FEB 2004
- Manuscript Received: 1 DEC 2003
- USARMY. Grant Number: DAM17-99-1-9564
- Minnesota Ovarian Cancer Alliance (MOCA)
- Gynecologic Oncology Group
- Sparboe Endowment. Grant Numbers: DA12104, CA 089652
- vascular targeting;
- colon cancer;
- ovarian cancer
Tumor vasculatures express high levels of αVβ3/αVβ5 and α5β1 integrins. Consequently, peptides containing the RGD (Arg-Gly-Asp) sequence, which is present in ligands of integrins, is effective in targeting therapeutic reagents to tumor vascular endothelium. In our study, we investigated whether the biologic activity of endostatin can be enhanced by the addition of an integrin targeting sequence. RGD sequence was added to either the amino or carboxyl terminus of endostatin containing a point mutation, P125A-endostatin. Earlier we have shown that the P125A mutation did not affect the biologic activity of endostatin but in fact had better antiangiogenic activity when compared to the native molecule. Further modification of P125A-endostatin with the RGD motif showed specific and increased binding to endothelial cells, and the increased binding coincided with improved antiangiogenic properties. Both amino and carboxyl terminal RGD-modification of P125A-endostatin resulted in greater inhibition of endothelial cell migration and proliferation. RGD modification increased tumor localization without affecting the circulatory half-life of P125A-endostatin, and RGD-modified P125A-endostatin was found to be more effective when compared to the P125A-endostatin in inhibiting ovarian and colon cancer growth in athymic mice. Complete inhibition of ovarian tumor growth was observed when P125A-endostatin-RGD was encapsulated into alginate beads. These studies demonstrate that addition of a vascular targeting sequence can enhance the biologic activity of an antiangiogenic molecule. © 2004 Wiley-Liss, Inc.
Establishing a new blood supply, or neovascularization, is important for tumor growth and metastasis.1 Formation of new blood vessels is a complex process involving endothelial cell proliferation, matrix degradation, migration, tube formation and maturation. Tumor cells along with stromal and inflammatory cells collectively create a proangiogenic microenvironment.2, 3 Angiogenic growth factors such as vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF) and angiopoietins stimulate endothelial cells to proliferate and migrate to form new blood vessels. In addition to growth factor receptor-mediated signaling, interaction between cell surface-anchored integrins and extracellular matrix components constitutes an additional pathway necessary for angiogenesis. In fact, 2 cytokine-mediated, integrin-dependent angiogenic pathways have been described. One of these pathways is associated with αvβ3 integrin, which selectively influences bFGF-mediated angiogenic signals.4 A second, nonoverlapping pathway is represented by crosstalk between αvβ5 integrin and PKC-dependent growth factor-mediated signaling (VEGF, IGF, TNF-α).4, 5 Tumor angiogenesis can therefore be inhibited either by blocking the interaction between αvβ3/β5 and RGD containing extracellular matrix or by interfering with angiogenic growth factors.
Angiogenesis is regulated by a delicate balance between pro- and antiangiogenic factors present in the microenvironment of tumor tissues. A number of proteolytic fragments of extracellular matrix6 and coagulation factors are capable of inhibiting angiogenesis. Endostatin is a proteolytic fragment of a collagen type XVIII.7 Endostatin is believed to be sequestered by glypican and presented to integrins.8 Rehn et al.9 recently showed that endostatin interacts with α5β1 and αvβ3 integrins on the surface of endothelial cells in an RGD-independent manner. Another antiangiogenic protein, tumstatin, derived from the NC-1 domain of collagen IV α-3 chain, also binds to αvβ3 integrins in an RGD-independent manner.10
These studies suggest that endostatin and tumstatin can transduce antiangiogenic signals by binding with integrins present on endothelial cells. However, a number of independent studies have reported that RGD-dependent interactions on the endothelial cell surface can also inhibit angiogenesis. This is consistent with the observation that RGD peptides and cyclic peptides containing the RGD motif are potent inhibitors of tumor angiogenesis.11 Therefore, we hypothesized that by introducing the RGD sequence into human endostatin, we might promote both RGD-dependent and -independent signaling via integrins, resulting in potent antiangiogenic activity.
To test this hypothesis, human P125A-endostatin12, 13 was genetically modified to incorporate an RGD sequence and expressed in yeast. P125A-endostatin has a point mutation and is biologically active in inhibiting tumor angiogenesis. In vitro, cell-binding studies showed that endothelial cells bound to RGD-modified P125A-endostatin coated plates significantly higher than to plates coated with unmodified P125A-endostatin. The increased binding was completely blocked by anti αvβ3 antibody or RGD peptide. RGD-modified-endostatins were more potent in inhibiting bFGF-induced endothelial cell proliferation and migration when compared to P125A-endostatin. RGD-containing P125A-endostatin was also more effective in inhibiting tumor growth in athymic nude mice. Our studies further show that slow release of RGD-modified endostatin microencapsulated with alginate beads completely inhibited established ovarian cancers in athymic mice. These studies suggest that antiangiogenic activity of P125A-endostatin can be further improved by adding an RGD sequence.
MATERIAL AND METHODS
Cells and animals
Bovine adrenal gland capillary endothelial cells (BCE) were obtained from Clonetics (San Diego, CA). Human umbilical vein endothelial cells (HUVEC), passage 2, were kindly provided by Dr. Vercelotti (University of Minnesota). Human colon carcinoma cells, LS174T, were obtained from American Type Culture Collection (ATCC, Rockville, MD). MA148, a human epithelial ovarian carcinoma cell line was established locally at the University of Minnesota from a patient with stage III epithelial ovarian cystadenocarcinoma.14 Human primary melanoma cell line WM35, which expresses αvβ3 integrin, was provided by Dr. J. Iida and Dr. J.B. McCarthy (University of Minnesota). Culture conditions of these cell lines have been previously described.15 WM35 cell line was cultured under the same conditions as MA148.
Cloning and expression of human endostatin
The yeast expression system, Pichia pastoris, was purchased from Invitrogen (San Diego, CA). Restriction enzymes and Taq DNA polymerase were purchased from Boehringer Mannheim (Indianapolis, IN).
We have earlier described a mutant endostatin with a single amino acid substitution at the position 125, P125A-endostatin.12, 13 This clone was further modified to incorporate the RGD sequence either at the amino terminus or at the carboxyl terminus. The following sets of primers were used to modify P125A-endostatin by PCR. (i) RGD-human P125A-endostatin: Up [5′]: GGGGAATTCAGAGGAGATCACAGCCACCGCGACTTCCAG; Down [3′]: GGGGCGGCCGCCTACTTGGAGGCAGTCATGAAGCT; (ii) Human P125A-endostatin-RGD: Up [5′]: GGGGAATTCCACAGCCACCGCGACTTCCAG; Down [3′]: GGGGCGGCCGCCTAATCTCCTCTCTTGGAGGCAGTCATGAAGCT.
Amplified fragments were purified by using a DNA extraction kit (Amicon, Beverly, MA), digested with EcoRI and NotI and cloned into pPICZα-A vector. DNA sequencing was carried out by Applied Biosystems sequencer (ABI 377 at the Advanced Genetic Analysis Center of the University of Minnesota) to verify the identity. Plasmid DNA was then linearized at the SacI site and electroporated into the yeast host strain X-33 (Invitrogen). Recombinants were selected on Zeocin containing plates and characterized for expression of mutant endostatins. All endostatin constructs used in this study had the P125A substitution.
Purification of recombinant proteins
Pichia clones were cultured in baffled shaker flasks and induced by methanol as previously described.15 For large-scale preparations, a fermentation procedure was used. A mouse angiostatin expressing Pichia pastoris clone (kindly provided by Dr. V.P. Sukhatme) was cultured under similar conditions. P125A-endostatin and angiostatin were purified according to published methods.15
Cell attachment assay
One nmole/well endostatin preparations or RGD peptide [(H)4-(G)3-R-G-D-(G)3-C] or 0.2% gelatin were used to coat 96-well polystyrene plates. The plates were incubated at 4°C overnight and then blocked with 2% BSA in PBS at 37°C for 2 hr. HUVEC, MA148 (negative control) or WM35 (positive control for αvβ3 integrin-expressing cell line) were harvested by PBS containing 1 mM EDTA and prelabeled for 10 min at 37°C with 5 μM 5-(and-6)-carboxyfluorescein diacetate, succinimidyl ester (5(6)-CFDA), a vital, fluorescence dye (Molecular Probes, Eugene, OR). After washing with Hank's balanced salt solution (HBSS), fluorescence-labeled cells were resuspended in EGM medium (HUVEC) or RPMI 1640 medium supplemented with 10% FBS (MA148, WM35). The cells were incubated with or without competitors (1 μg anti-αvβ3 integrin monoclonal antibody (LM609, Chemicon, Temecula, CA), 1 μl anti-α5β1 integrin monoclonal antibody (JBS5, Chemicon) or anti-HLA monoclonal antibody (negative control; G46-2.6, BD PharMingen, San Diego, CA) or 25 nmole/well RGDS or RGES peptides (Sigma Chemicals, St. Louis, MO) for 1 hr at 37°C. Cells were then added to wells at a density of 40,000 cells/well (HUVEC and MA148) or 30,000 cells/well (WM35). After a 1 hr incubation at 37°C, plates were washed twice with HBSS to remove unbound cells. Cells bound to the wells were assayed by a fluorescence plate reader (Cyto Fluor II; PerSeptive Biosystems, Framingham, MA) (excitation 485 nm; emission 530 nm).
Binding of P125A-endostatin and RGD-modified P125A-endostatins to HUVEC was also determined by using [125I]-labeled proteins. [125I]-labeling was carried out by the Iodogen method. HUVEC was harvested in 1 mM EDTA/PBS and was resuspended in 0.1% BSA/PBS. Five microliters of [125I]-endostatin (27380 cpm/μg) or [125I]-RGD-P125A-endostatin (32908 cpm/μg) were added to 40,000 HUVEC cells in 100 μl of 0.1% BSA/PBS. Cells were incubated at room temperature for 1 hr and then washed with 0.1% BSA/PBS twice. [125I]-P125A-endostatin and [125I]-RGD-P125A-endostatin bound to HUVEC were detected by a γ-counter (Packard, Meriden, CT).
Endothelial cell-proliferation assay
Essentially, the method described by O'Reilly et al. was used.7 Proliferating cells were quantified by 5′-bromo-2′-deoxyuridine (BrdU) incorporation (Roche, Indianapolis, IN) according to the manufacturer's instructions.
Endothelial cell-migration assay
The migration of endothelial cells was determined by using Boyden chambers (Neuro Probe, Gaithersburg, MD). Polycarbonate filters (pore size 12 μm) were coated with 0.2% gelatin for 1 hr at 37°C. HUVEC were harvested in 2 mM EDTA in PBS and prelabeled with 5 μM 5(6)-CFDA for 10 min at 37°C. Cells were resuspended in 0.5% FBS, M199 medium and then preincubated with P125A-endostatin, RGD-endostatin or endostatin-RGD for 60 min at 37°C. Basic FGF (25 μl of 25 ng/ml 0.5% FBS, M199 medium) was added to the lower chambers. HUVEC (200,000 cells/ml, control and treated) were added to upper chambers. After 4 hr of incubation at 37°C, endothelial cells that had migrated to the bottom side of the membrane were counted in a fluorescence microscope (Olympus) using FITC filters (magnification 200×). Two independent experiments were carried out.
Detection of bcl2 and bax mRNA
HUVEC were treated with P125A-endostatin or RGD-modified endostatins at a concentration of 50 nM. After 48 hr, the cells were trypsinized, and total RNA was extracted by using the RNeasy Mini kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. Reverse transcription was carried out with the SuperScript II kit (Invitrogen) using 1 μg of total RNA. DNA of bcl2 or bax was determined by PCR using primers of the Amplifluor Universal Amplification and Detection System (Intergen, Purchase, NY). As a control, the following human GAPDH primers were used: Up [5′]: CCACCCATGGCAAATTCCATGGCA; Down [3′]: TCTACACGGCAGGTCAGGTCCACC.
LS174T cells were injected subcutaneously in both sides of the flanks of female, athymic nude mice (8 weeks old). Tumor size reached about 500 mm3 on day 10. Tumor-bearing mice were randomized into 3 groups of 3 to 4 mice in each group. P125A-endostatin, RGD-endostatin or endostatin-RGD was injected at a dose of 20 mg/kg subcutaneously. Nineteen hours after injection, tumor tissues (6 to 8 tumor samples per group) and representative normal tissues were surgically removed. This time point was chosen to minimize overwhelming serum levels from obscuring the tissue-bound endostatin. For comparison, serum samples were also collected from the mice at the time of sacrifice. Tissues were snap-frozen and homogenized in RIPA buffer containing proteinase inhibitors (PBS, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 10 μg/ml PMSF), maintained at 4°C for 45 min and cleared by centrifugation. Human endostatin concentrations in serum and tissue lysates were measured using an enzyme-linked immunoassay (Cytimmune, College Park, MD) according to the manufacturer's instructions. Endostatin standards were also prepared in RIPA buffer. Statistical significance was determined by Student's t-test.
Determination of circulatory half-life
About 1 million cpm of [125I]-labeled P125A-endostatin or RGD-endostatin was injected intravenously. Mice were bled at different time points from the tail vein, and total radioactivity in aliquots of serum samples was determined.
Determination of vessel density and apoptosis
To determine the effect of antiangiogenic treatments on vessel density and apoptosis, residual tumors from experimental (total of 10 mice) and control (5 mice) groups were surgically resected and snap-frozen. Cryostat sections (10 μm) of tumors were fixed in cold acetone, air dried and then treated with PBS containing 0.1% BSA and 5% human serum to block nonspecific binding (background). Sections were then incubated with 1:50 dilution of an anti-CD31 (mouse) monoclonal antibody conjugated to phycoerythrin (MEC 13.3, BD PharMingen, San Diego, CA). After 1 hr incubation at room temperature, sections were washed thoroughly with PBS containing 0.1% BSA and 5% human serum followed by washing with PBS and then examined by an Olympus BX-60 fluorescence microscope. Images (7–10) were captured by the Metamorph program for analysis. Detection of apoptosis was carried out by using an In Situ Cell Death Detection Kit (Boehringer Mannheim, Indianapolis, IN) following the manufacturer's protocol. Parts of the tumor samples were also fixed in 10% neutral buffered formalin and processed for histochemistry.
Tumor growth inhibition studies (bolus injection)
Female athymic nude mice (6–8 weeks old) were obtained from NCI and acclimatized to local conditions for 1 week. Logarithmically growing human colon carcinoma cells (LS174T) were harvested by trypsinization and suspended in fresh medium at a density of 1 × 107 cells/ml. One hundred microliters of the single-cell suspension were then subcutaneously injected into the flanks of mice. When the tumors became visible (3 days after inoculation), mice were randomized into groups of 5 mice each. The mice were treated with P125A-endostatins, RGD-endostatin s.c. at a dose of 20 mg/kg/day for 11 days. A control group of mice (n = 5) were treated with sterile PBS under similar conditions. All injections were given subcutaneously near the neck, about 2 cm away from the growing tumor mass. Tumor growth was monitored by periodic caliper measurements. Tumor volume was calculated by the following formula:
Statistical significance between control and treated groups was determined by repeated measurement analysis of variance.
Tumor growth inhibition studies: slow release from alginate beads
Alginic acid extracted from Macrocystis pyrifera was purchased from Sigma Chemicals (St. Louis, MO). A 4% solution of alginic acid in water was sterilized by autoclave. P125A-endostatin preparations made in 2% alginic acid were dropped gently into 0.1 M CaCl2 solution using a fine needle under aseptic conditions. The beads were kept at 4°C overnight and washed with sterile water before the subcutaneous implantation into tumor-bearing mice (MA148 ovarian cancer cell line). A BCA kit was used to determine concentration of protein encapsulated into alginic beads. P125A-endostatins encapsulated in alginate beads were administered 4 times at a dose of 20 mg/kg/mouse once a week. Control and the experimental arm of the study had 5 mice in each group.
First, we evaluated whether human P125A-endostatin could be modified at either of its termini without compromising its biologic activity. Addition of His-Tag or Myc epitope Tag did not affect the ability of P125A-endostatin to inhibit endothelial cell proliferation or migration in vitro (data not shown). Based on these results, P125A-endostatin was modified to incorporate an RGD motif at the amino (designated as RGD-endostatin) or carboxyl (endostatin-RGD) terminus.
RGD-endostatin increases endothelial cell attachment in vitro
RGD-peptide is well known for its binding to integrins on the surface of endothelial cells. To determine whether addition of RGD-motif to P125A-endostatin could enhance its binding to endothelial cells, cell attachment assays were performed. As a positive control, 0.2% gelatin-coated wells were used. Number of cells attached to gelatin-coated wells was considered as 100% to calculate relative binding. HUVEC attached to endostatin-coated wells (Fig. 1a). BSA-blocked wells were used as a negative control. In this assay system, P125A-endostatin-coated wells showed about 60% cell attachment, which was further increased by RGD-modification (Fig. 1a). RGD-endostatin (p < 0.05) and endostatin-RGD (p < 0.01) showed about 80% cell attachment. Parallel experiments with RGD-containing synthetic peptide showed similar binding of HUVEC. Under the experimental conditions used, a preparation of recombinant murine angiostatin (kringle 1-4, expressed in yeast) did not increase endothelial cell attachment over the control values (Fig. 1a).
Cell attachment studies were repeated using human microvascular endothelial cells (MVEC) and bovine adrenal gland capillary endothelial cells (BCE). These studies showed a profile similar to results obtained with HUVEC (data not shown). Then experiments were carried out to determine the specificity of RGD-mediated enhanced binding to endothelial cells (Fig. 1c). Presence of RGD sequence (COOH or NH2 terminus) increased endothelial cell attachment by 40% over unmodified endostatin. This was completely blocked by RGD containing synthetic peptide. As an additional control, an epithelial ovarian tumor cell line, MA148, was used in cell attachment assays. Results shown in Figure 1d suggest that RGD-containing peptide did not facilitate MA148 cell attachment. Human endostatin (native and RGD-modified)-coated wells again did not show any significant increase in tumor cell attachment when compared to control wells blocked with BSA alone. However, if an RGD motif were to be introduced into BSA, HUVEC attached to the wells very efficiently (Fig. 1b). A synthetic peptide containing RGD sequence was used to chemically link it to BSA containing activated thiol groups. CGGGRGD peptide has a free thiol group at the aminoterminus. Using a heterobifunctional cross-linking reagent, N-succinimidyl 3-[2-pyridyldithio] propionate (SPDP; Pierce Chemicals, Rockford, IL), CGGGRGD peptide was linked to bovine serum albumin (RGD-BSA). Based on molecular weight shift in SDS-PAGE, an average of 5 RGD moieties were introduced into each BSA molecule.
In addition to cell attachment studies, we also examined the direct binding of endostatin to HUVEC by using [125I]-labeled endostatins. Data in Figure 1e show that about 0.65 pmole of radioiodinated endostatin bound to 106 cells, about 1 × 104 molecules per cell. Under similar conditions, about 1.2 pmole of [125I]-RGD-endostatin bound to HUVEC (a 2-fold increase in binding). These studies suggest that endostatin binding to endothelial cells can be specifically increased by RGD modification. Neither amino nor carboxyl terminal modifications caused any steric hindrance to RGD-mediated binding to endothelial cells.
Because endostatin has been known to bind α5β1 and αvβ3 integrins, blocking antibodies to integrins were used to study the interaction between HUVEC and RGD-modified endostatins. Binding of HUVEC to P125A-endostatin was blocked by antibodies to α5β1 (27.0% inhibition) and αvβ3 (51.4%; Fig. 1f). Binding of HUVEC to endostatin-RGD was blocked by these 2 antibodies more evenly (α5β1, 31.3%; αvβ3, 34.8%) under similar conditions. A recent study showed that activity of endostatin was impaired in fibronectin-deficient mice.16 Since α5β1 and αvβ3 bind to fibronectin, we investigated whether endostatin and endostatin-RGD can block the binding of HUVEC to fibronectin (Fig. 1g). Native endostatin inhibited the binding of HUVEC to fibronectin by 24.6%, and P125A-endostatin and endostatin-RGD inhibited slightly better than the native molecule, 36.1% and 40.0%, respectively.
To differentiate the interaction between RGD/integrins and endostatin/integrin (α5β1), a nonendothelial cell line overexpressing αvβ3 was used. Human melanoma cell line WM35 expresses higher levels of αvβ3 integrin but no detectable levels of α5β1 and αvβ5 based on flowcytometric analysis (data not shown). Representative photomicrographs of cell attachment to coated wells are shown in Figure 2b–g. Unlike HUVEC, WM35 cells did not attach to endostatin-coated wells (18%, which is similar to BSA-blocked control wells) (Fig. 2a,b). However, WM35 cells specifically attached to RGD-endostatin-coated wells (60%) (Fig. 2a,c). To determine whether the increased binding of RGD-endostatin was indeed specific, 2 methods were used. In 1 experiment, a monoclonal antibody to anti-αvβ3 integrin was used to block the interaction. As a control, isotype-matched mouse IgG was used at a similar concentration. Preincubation of WM35 cells with the anti-αvβ3 integrin antibody completely blocked cell attachment to RGD-endostatin (Fig. 2a,d). The control antibody did not prevent WM35 cells from binding to RGD-endostatin-coated wells (Fig. 2a,e). In a second series of experiments, synthetic peptides were used as competitive inhibitors. Inclusion of RGDS peptide in the medium completely prevented attachment of WM35 cells (Fig. 2a,f), whereas a control peptide, RGES, did not affect WM35 cells from attaching to RGD-endostatin-coated wells (Fig. 2a,g). Endostatin-RGD was also tested in WM35 cell attachment and showed similar profile as RGD-endostatin (data not shown).
Inhibition of endothelial cell proliferation
To determine whether the modification of human endostatin affects the biologic activity, endothelial cell proliferation assays were carried out by BrdU incorporation. Inhibition of bFGF-induced proliferation was determined in the presence of different concentrations of endostatin preparations. Data in Figure 3a show that at 1 μM, endostatin inhibited bFGF-induced BCE cell proliferation by 42.1%, which is similar to findings in our previous study.12 Addition of RGD motif at the amino or carboxyl terminal end of endostatin further enhanced the basal antiproliferative activity. Amino terminal modification of endostatin with RGD sequence showed 57.7% inhibition of bFGF-induced proliferation. Interestingly, RGD-modification at the C-terminal end was much more effective and completely inhibited bFGF-induced BrdU incorporation into DNA (100%). To determine whether the introduction of the RGD sequence leads to nonspecific inhibition, we determined the effect of RGD-P125A-endostatin on nonendothelial cells. Human aortic smooth muscle cell (AoSMC, Clonetics), human ovarian cancer cell (MA148) and human melanoma cell (WM35) were not affected by either P125A-endostatin or RGD-modified P125A-endostatins.
To determine whether the enhanced inhibition of endothelial cell proliferation was due to RGD moiety and endostatin, the effect of synthetic peptides containing RGD sequence was tested under similar conditions. Unlike the RGD-modified endostatins, the synthetic peptides did not inhibit endothelial cell proliferation at 1.0 μM concentration. However, RGD peptide inhibited endothelial cell proliferation at higher concentrations (>10 μM). In a parallel set of cultures, RGDS peptide was admixed with unmodified endostatin to determine whether 2 independent bindings could lead to improved antiproliferative activity. Data shown in Figure 3b clearly suggest that at the concentrations tested, the free peptides either alone or in the presence of endostatin did not significantly change the basal level of inhibition seen with endostatin alone. These studies suggest that RGD moiety engineered into endostatin molecule acquires enhanced antiproliferative activity, perhaps by cross-linking α5β1 and αvβ3/αvβ5 integrins.
Increased inhibition of endothelial cell migration by endostatin containing RGD motif
To evaluate whether RGD motif can affect the ability of endostatin to inhibit endothelial cell migration, Boyden chamber-based migration assays were performed (Fig. 3c). Basic FGF was used to induce endothelial cell migration. Endostatin treatment at 50 nM inhibited bFGF-induced migration by 50.6%. Relative to this, RGD-endostatin and endostatin-RGD showed statistically significant improvement in the inhibition of cell migration. Cultures treated with RGD-endostatin showed 86.3% inhibition, and endostatin-RGD-treated cultures showed almost complete inhibition (101.4%) of bFGF-induced cell migration. In a parallel study, RGD containing synthetic peptide was also evaluated for its ability to inhibit cell migration. Free peptide did not show any inhibition at 50 nM. When a mixture of an equimolar concentration of endostatin and peptide was used, there was no enhancement in the basal activity of endostatin. As a negative control, RGES peptide was used, which did not affect cell migration.
These results suggested that the RGD moiety should be an integral part of endostatin in order to inhibit endothelial cell migration better. This conclusion was further validated by an additional control experiment using RGD containing peptides that were chemically linked to a control protein, BSA, via a thiol group. Accessibility of RGD in the BSA conjugate was validated in cell attachment assays (Fig. 1b). Endothelial cells attached to RGD-BSA coated wells very well (84% when compared to gelatin-mediated attachment). Cell attachment was similar to RGD-peptide and RGD-endostatin. In spite of high binding to endothelial cells, however, chemically linked RGD-BSA did not inhibit migration of endothelial cells at equimolar concentrations (10–50 nM concentration, data not shown). These studies suggest that the RGD sequence alone at the concentrations tested does not inhibit endothelial cell proliferation and migration, although it enhances the inhibitory activity of endostatin.
Downregulation of bcl2 mRNA
Endostatin and RGD peptides are independently known to affect apoptotic pathways.17, 18 As a preliminary study, we determined whether RGD modification changes the levels of a key antiapoptotic molecule, bcl-2. Relative levels of bcl-2 and bax transcripts were determined by RT-PCR in endothelial cells treated with endostatin preparations. Endothelial cells treated with bFGF showed an increase in bcl-2-specific transcripts when compared to control cultures (-bFGF). Treatment with P125A-endostatin at a concentration of 50 nM decreased mRNA levels of bcl2. RGD-modified endostatin also downregulated bcl2 levels significantly when compared to bFGF-treated control cultures (Fig. 3e) but to the same degree as P125A-endostatin. Parallel studies showed no significant change in bax transcript levels in treatment with either of the endostatin preparations (data not shown).
RGD modification increases tumor localization of endostatin
To assess whether the improved endothelial cell binding in vitro could translate into enhanced tumor homing in vivo, tumor localization studies were carried out. P125A-endostatin, RGD-endostatin and endostatin-RGD were injected subcutaneously into human colon cancer-bearing athymic mice. Relative levels of endostatin in the tumor tissue when compared to serum levels are shown in Figure 4a. P125A-endostatin accumulation in tumors was considered as 100%. RGD-modified endostatins localized more than 2-fold higher when compared to P125A-endostatin (Fig. 4a).
Another pharmacologic property that can affect biologic efficacy in vivo is clearance rate. Therefore, serum half-life was compared. Figure 4b shows the beta phase of clearance. Both unmodified and RGD-modified endostatin showed a similar half-life. The α phase of P125A-endostatin and RGD-endostatin was 5.3 ± 1.53 min, 2.3 ± 1.14 min, respectively, and the β phase of P125A-endostatin and RGD-endostatin was 9.1 ± 2.27 hr, 8.9 ± 2.40 hr, respectively (Fig. 4b). The differences seen in the α phase and β phase were not statistically significant. Collectively, these results suggest that RGD-modification does not alter serum clearance but facilitates tumor tissue accumulation.
Inhibition of colon cancer growth by bolus injection of RGD-modified endostatin
To test whether RGD modification of endostatin could improve antitumor activity, we used first an LS174T xenograft model system. LS174T tumors grow very fast and were allowed to establish for 3 days. At this time, small palpable tumor nodules could be easily seen. Mice were then randomized and divided into groups. RGD-endostatin and P125A-endostatin were administered subcutaneously at a dose of 20 mg/kg/day for a period of 11 days. As shown in Figure 5, RGD-endostatin inhibited tumor growth significantly better than unmodified endostatin. In this model system, the control tumors reached a size of about 500 mm3 by day 14. Endostatin treatment inhibited the tumor growth by about 30% under the conditions tested. In contrast, groups of animals treated with RGD-endostatin significantly decreased the tumor growth by 78% when compared to the control animals (p = 0.005).
Effect on tumor blood vessels and apoptosis
To evaluate the consequence of antiangiogenic therapy, we examined the residual tumors histologically. Frozen tumor sections were stained with anti-CD31 PE conjugate. Both native and RGD-modified endostatin treatment resulted in reduced vessel density (Fig. 6a–c). The same frozen sections were also analyzed for changes in the apoptosis of tumor cells using a TUNEL assay (Fig. 6d–f). Serial sections of each tumor were also stained by H & E to assess histologic changes (Fig. 6g–i). H & E and TUNEL staining revealed that RGD-endostatin induced more apoptosis in tumor tissue (Fig. 6f,i) when compared to control (Fig. 6d,g) and P125A-endostatin treated tumors (Fig. 6e,h). A quantitative analysis of apoptotic index is shown in Figure 6j. RGD-endostatin-treated tumors showed an apoptotic index of 7.91 ± 2.40%. This value is about 45-fold higher than the control tumors (0.176 ± 0.048). Native endostatin-treated tumors showed an apoptotic index of 1.32 ± 0.774%, an increase of 7.5-fold over the control tumors.
Slow release of P125A-endostatin and endostatin-RGD by alginate beads: improved antitumor activity
In a preliminary study, we compared the relative effects of amino (RGD-endostatin) and carboxyl (endostatin-RGD) terminal modification by daily injection. Ovarian cancer cell line MA148 was subcutaneously injected into female, athymic mice. Groups of 5 mice were treated either with vehicle or with the respective RGD-modified proteins. Mice were injected daily (i.p.) with 20 mg/kg of P125A-endostatin preparations starting on day 7 until day 28. Bolus injection of P125A-endostatin alone did not inhibit tumor growth significantly. However, slow-release formulation of P125A-endostatin by alginate encapsulation inhibited tumor growth significantly.12 For the purpose of comparing amino vs. carboxyl modification, we used bolus injection. Amino terminal modification with RGD inhibited tumor growth by 28% when compared to the untreated control group. Carboxyl terminal modification of endostatin with RGD moiety showed a moderate improvement in tumor growth inhibition (57% inhibition) (data not shown).
Subsequently, we investigated whether the slow release of P125A-endostatins using alginate microspheres could improve the antitumor activity in the ovarian cancer model system. Alginic acid, which is a naturally occurring biopolymer, has been used as a matrix for entrapment and delivery of a variety of biologic agents. Ovarian cancer cell line MA 148 was injected s.c. into the flanks of athymic mice. After 7 days, alginate beads containing endostatin preparations were administered to groups of mice. Endostatins were given once a week, 4 times, at a dose of 20 mg/kg. Control group of mice received vehicle-encapsulated alginate beads on a similar schedule. Unlike bolus, daily injections, administration of P125A-endostatin in alginate formulation showed significant antitumor activity (Fig. 7). Tumor growth was inhibited throughout the experiment. For example, 40% of P125A-endostatin-treated animals did not show any tumor growth up to 42 days, at which time control animals had a mean tumor volume of 1,150 mm.3 Subsequently, the tumors began to grow and reached a size of about 1,300 mm3 by day 60. Interestingly, the endostatin modified with RGD motif when delivered in alginate beads showed a complete inhibition of tumor growth for 75 days (Fig. 7). Tumor growth was suppressed for a prolonged time even after the cessation of P125A-endostatin-RGD treatment. These studies demonstrate that RGD-modification of P125A-endostatin can improve the antitumor activity of endostatin and that a slow release formulation can be used to inhibit tumor growth very effectively.
Our previous studies identified human endostatin with a single amino acid substitution, P125A-endostatin. P125A mutation resulted in better antiangiogenic activity when compared to the native molecule.12 In our present study, we introduced a vascular-targeting, integrin-binding motif to P125A-endostatin and investigated its effect on biologic activity in vitro and in vivo. Endostatin binds to Glypican and integrins.8, 9, 19 Glypican is believed to sequester endostatin and present it to integrins on endothelial cells. Of the 2 integrins (α5β1,, αvβ3) known to interact with endostatin, binding to α5β1 appears to be biologically relevant. Blocking α5β1 integrin mediated interaction with endostatin (immobilized) and inhibited endothelial cell migration.9 While retaining this interaction, RGD modification of endostatin is likely to provide additional binding to integrins, such as αvβ3/α5β1. Both of these integrins play an important role in tumor neovascularization. Interestingly, a recent study using competitive inhibition by peptides showed that endostatin may bind to endothelial cells via an RGD-dependent manner.20 However, endostatin does not have an RGD sequence. Therefore, RGD-mediated direct binding of native endostatin is not possible. It is possible that endostatin could complex with a matrix protein, which may contain a surrogate RGD moiety that could participate in the interaction with integrins. Further studies are necessary to characterize this interaction. Independent binding to these integrins can affect the biologic activity. Indeed, when endostatin was modified with RGD sequence, there was increased binding to endothelial cells. Increased endothelial cell attachment to RGD-endostatin was highly specific, since it could be completely blocked by a synthetic peptide containing RGD sequence.
The RGD motif also improved the ability of endostatin to inhibit endothelial cell proliferation and migration. COOH-terminal addition of RGD resulted in enhanced inhibition when compared to NH2-terminal addition of RGD. These differences can be attributed to peptide folding and presentation of RGD sequence. Theoretical prediction using Swiss Protein-View software suggests 4 hydrogen-bonding possibilities when the RGD sequence was introduced at the carboxyl terminus compared to 3 hydrogen bonding with neighboring amino acid residues at the amino terminus (data not shown). The mere presence of the RGD sequence was not sufficient to enhance activities, since RGD-BSA or RGD mixed with endostatin did not lead to improvement at the concentrations tested. Thus, the RGD moiety in the context of endostatin is necessary for enhanced biologic activity against endothelial cells.
The mechanism underlying the enhanced activity of RGD-modified endostatin is yet to be fully understood. Endostatin has been reported to decrease bcl2 levels in endothelial cells,17, 18 and RGD peptide itself is known to interact with caspases and induce apoptosis when delivered to intracellular sites.21 Studies on the internalization rate and compartmentalization of RGD-endostatin may provide mechanistic insight into the enhanced biologic activity. It is possible that these 2 independent events in combination can account for the pronounced effects seen on endothelial cells. Our studies suggest that RGD-modified endostatin downregulated transcripts for the antiapoptotic protein bcl2 without changing the levels of bax, a constituent of proapoptotic signals. The extent of reduction seen in bcl2 levels was greater in RGD-endostatin-treated cultures when compared to P125A-endostatin, but the difference was not statistically significant. Therefore, it is likely that changes in other molecules such as bcl-x may be contributing to the increased apoptosis observed in RGD-endostatin-treated endothelial cells. Further studies are necessary to understand the mechanism by which increased apoptosis is brought about by RGD-modified endostatins.
Tumor vasculature is known to overexpress αvβ3 and αvβ5 integrins. Therefore, peptides containing RGD sequence are very effective in delivering cytotoxic drugs or antibodies to tumor vasculature.22, 23 Increased binding to endothelial cells combined with a potential for targeting by RGD moiety improved tumor localization and bioavailability of endostatin. RGD-endostatin, for example, accumulated 2-fold higher in tumor tissues when compared to endostatin alone. Increased accumulation was achieved without any changes in the serum clearance. Pharmacokinetic studies demonstrated that RGD modification did not alter the serum half-life. Increased accumulation of RGD-endostatin correlated with greater antiangiogenic effects. Morphometric analysis revealed that RGD-endostatin treatment reduced vessel density in tumor tissues. A reduced number of blood vessels coincided with increased apoptosis of tumor cells. These results confirm other recent studies showing improved therapeutic efficacy of many biologic response modifiers by genetic modification. TNF-α engineered to contain NGR sequence, for example, exhibited improved antitumor activities.24 RGD motif introduced into adenoviral coat proteins improved gene transfer efficacy into integrin αvβ3-positive cells.25
Enhanced biologic activity and tumor targeting can potentially improve the inhibition of tumor growth. Two different tumor models were used to assess the effect of RGD-endostatins. LS174T colon cancer cells form rapidly growing tumors in athymic mice. In this model system, administration of RGD-endostatin showed a significant improvement in tumor-growth inhibition when compared to unmodified P125A-endostatin given at similar doses.
Consistent with our earlier studies using mouse endostatin, ovarian cancer growth in athymic mice was not significantly affected by bolus injections of human endostatin.15 However, RGD modifications showed improvement in antitumor activity even when given as bolus injections. Delivery of P125A-endostatin and native endostatin as a slow-release formulation indeed showed significant antitumor activity.12 This observation is also confirmed by other recent studies showing improvement in efficacy of endostatin by slow-release administration.26, 27, 28 P125A-endostatin and RGD-endostatins were encapsulated into alginate beads. Microencapsulated P125A-endostatin treatment of animals showed significant inhibition of tumor growth when compared to control animals throughout the study. However, tumors began to grow after the discontinuation of treatment. In comparison, endostatin-RGD treatment under similar conditions produced long-term remissions. Tumor growth was completely inhibited for 75 days after tumor cell transplantation. These studies demonstrate that addition of a vascular targeting sequence to P125A-endostatin further improves its biologic activity. RGD sequence facilitates tumor localization. Enhanced biologic activity in combination with improved pharmacologic properties significantly potentiated the antitumor effect of P125A-endostatin.
The authors thank Dr. R.L. Bliss for statistical analysis.