Means and methods for treating angiogenesis-related diseases

Abstract

The present invention is concerned with a protein oligomer comprising at least two NC-1 monomers of human collagen 18 or fragments of an NC-1 monomer of human collagen 18 for use in the treatment or prevention of an angiogenesis-related disease. The invention further pertains to a fusion protein comprising a NC-1 monomer of human collagen 18 and a Fc domain of an immunoglobulin. The invention also relates to a fusion protein comprising: a) an endostatin peptide or endostatin-derived peptide and b) the RGD motif and/or PHSRN motif of Fibronectin. The invention further relates to a kit comprising the protein oligomer or fusion proteins of the invention.

Claims

1. A peptide comprising an amino acid sequence shown in SEQ ID NO. 7 or 13.

2. The peptide of claim 1, further comprising an Fc domain of an immunoglobulin.

3. The peptide of claim 1, wherein the amino acid sequence is shown in SEQ ID NO. 7.

4. The peptide of claim 1, wherein the amino acid sequence is shown in SEQ ID NO. 13.

Description

FIGURES

(1) FIG. 1 (A): Immunoprecipitation of human Platelets.

(2) Platelets from 15 ml of freshly collected plasma was lysed employing 25 mM Tris, 0.15 M NaCl, 1% NP-40, pH 7.5 and a cocktail of protease inhibitors (total volume was made 3 mls). The lysate was centrifuged and filtered. Proteins Fc, Fc-VEGF and Fc-endostatin (10 g of each) were individually added to 1 ml of lysate. Protein A was employed. Incubation time was 18 h at 4 C. After three washes, the eluted samples were applied to PAGE and were stained with coomassie stain. The control lanes refer to the samples devoid of lysate. The candidate proteins were sliced out of the gel and sent out for Mass Spectra analysis.

(3) FIG. 1 (B): Elisa.

(4) Coating of proteins were done in PBS at concentration of 10 g/ml. Fibronectin served as a ligand at the same concentration. All buffers contained 2% BSA, 0.1% Tween-20.

(5) FIG. 1 (C): Biacore measurements of equilibrium constants for binding of endostatin monomer, dimer and NC-1 trimer to fibronectin.

(6) Fibronectin samples were prepared by serial dilution into 0.01 M Hepes, pH 7.4, 0.15 M NaCl, 0.05% Surfactant containing 1 mg/ml BSA in the range 0.78 nM-100 nM and flowed over control and derivatized surfaces for three minutes at a flow-rate of 60 l/min. Dissociation phases were monitored for 5 minutes. Zero concentration blank buffer cycles were included as negative control samples. Sensor surfaces were regenerated using a 1 minute injection of 1 M ethanolamine, pH 8.5 following each interaction analysis cycle. Non-specific binding effects to sensor surface CM4 were not observed. Calculated ka, kd and KD are shown. Different curves correspond to different concentrations of fibronectin.

(7) FIGS. 2A-2G: Endostatin binding activity and co-localization with fibronectin in blood vessels.

(8) Immunohistochemistry was used to verify the distribution of Fc-endostatin in ASPC-1 xenograft mice after treatments. (A) The exogenous endostatin was detected by Alexa 488-labeled antibody (green) and vessel marker, CD31 was used and detected by Alexa 594-labeled antibody (red) in tumor, heart and kidney. (B) -SMA, pericyte marker (red) was used to confirm the endostatin binding to tumor vessels. (C) Exogenous endostatin (green) distribution is similar to fibronectin (red) in tumor. Human Fc (hFc) control shows no binding. (D) Endogenous collagen 18 (green) and fibronectin (red) were detected by polyclonal antibodies against endostatin and fibronectin in non-treated animal tumor. (E) Staining of Fc-endostatin and integrin 51. (F) Staining of vW and integrin 51. (G) E14.5 mouse brain embryonic sections were used to verify the distribution of fibronectin, VEGF and Fc-endostatin (Bar, 100 m).

(9) FIG. 3 (A): Binding of Fc-endostatin and collagen hNC-1 to endothelial cells.

(10) HUVECs were incubated with Fc-endostatin and detected with Alexa 488 IgG or human NC-1 (hNC-1) trimer and detected by anti-His tag monoclonal antibody. The fibronectin staining is showed in red color (Bar, 20 m).

(11) FIGS. 3 (B)(a)-(c): hNC-1 inhibits endothelial cell migration.

(12) HUVECs were plated into 24-well inserts of transwell plate in duplication. The lower chamber was filled with serum free medium containing 100 ng/ml rhesus VEGF (rhVEGF) plus different concentrations of hNC-1. After incubation for 16 h at 37 C., the cells were fixed and stained. Endothelial cells show less migration under 100 and 200 ng/ml of hNC-1 treated (a). The effect of hNC-1 in inhibition of endothelial cell migration shows a U-shape curve (b). Endostatin monomer, dimer and NC-1 were used for comparison (c).

(13) FIG. 4: Immunoprecipitation of human Sera and Platelets followed by Western analysis.

(14) (A) M1 and M2 refer to recombinant endostatin (187 amino acids) and NC-1, respectively. 1) pre-immune and human serum. 2) endostatin antibody and human serum. (B) M contains hNC1 and endostatin markers. Lanes 1 and 2 are the same as in (A). Lanes 3 (pre-immune serum) and 4 (anti-endostatin antibody) correspond to the serum of an individual not represented in (A). (C) Affinity purification of human serum obtained from a different PRP sample than the one employed in (A) followed by Western analysis without IP step. (D) Immunoprecipitation of human platelets. hNC-1 and endostatin markers are in lane M. 1) pre-immune serum and platelets lysate. 2) anti-endostatin antibody and platelets lysate. IPs were carried out in the presence of Protein A. After subjecting the samples to PAGE and transfer, the membranes were treated with anti-endostatin monoclonal antibody PDM.

(15) FIG. 5: Treatment of mouse bearing human melanoma cancer cells (A2058) with hNC-1.

(16) 4-6 tumor-bearing nude mice in each group were subcutaneously (s.c.) treated with hNC-1 (100 g/mouse once a day), clinical-grade endostatin (100 and 500 g/mouse once a day) or PBS. Treatment was stopped prior to the development of necrosis. Sites of injection were away from tumors. Tumor sizes and the ratio of treated/control (T/C) is shown. The hNC-1 treated group shows 67% inhibition of tumor growth in the end of experiment, whereas the group treated with endostatin shows only 48% inhibition.

(17) FIG. 6: Schematic model of interactions among fibronectin, integrin 51, VEGF-A and hNC-1.

(18) FIG. 7: SDS gel of a murine Fc-NC-1 fusion protein as a function of the amount of plasmid used in transfection of 293 kidney cells under reduced and non-reduced conditions.

(19) Under reduced conditions, the product is a single chain consisting of Fc and NC-1. Under non-reduced conditions, the product is a dimer because Fc is disulfide bonded. The single band in the middle of the gel is due to endostatin marker which has a molecular weight of 20 Kd.

(20) FIGS. 8.1-8.4: Generation of resistant Lewis Lung Cancer tumors in-vivo after prolonged exposure to mouse Fc-angiostatin and Fc-endostatin.

(21) Dimerization of the NC-1 endostatin domains were achieved using IgG-Fc. After strong initial inhibition of tumor growth (p1) tumors were exposed for a prolonged period of time to Fc-endostatin anti-angiogenic therapy (up to four consecutive passages, p1-4) by re-implanting the tumors in new animals once they reached a tumor size of >1000-1500 mm.sup.3. Sequential in-vivo tumor passaging was performed to achieve prolonged and continuous exposure to anti-angiogenic therapy to enrich for the resistant tumor cell population. P4 tumors, lower panel, were growing even faster than non-treated control p4 tumors despite continues exposure to oligomeric NC-1-fragment (Fc-endostatin). These data clearly show that tumors could develop acquired drug resistance to oligomeric fragments of NC-1.

(22) FIGS. 9.1-9.5: Genome-wide expression profiling of oligomeric-NC-1 fragments (mFc-Endostatin) resistant Lewis Lung Cancer (LLC) tumors.

(23) Expression profiling revealed that fibronectin (FN1) is markedly down regulated in murine (m)Fc-Endostatin (FcEndo) resistant tumors (heatmap, green box). The regulation of candidate genes was confirmed by q-RT-PCR and Fold-expression ratios relative to p4 control tumors are presented (diagrams). This finding supports the inventor's data demonstrating that anti-angiogenic effects of oligomeric NC-1-fragments are exerted via binding to FN1. Therefore, down regulation of FN1 in tumors render them resistant to oligomeric NC-1. Moreover, the inventors identified several key pathways such as IGF1R and CCL2 to be up regulated in Fc-endostatin resistant tumors (red box). Hence, down-regulation of the key binding partner and compensatory up regulation of alternative angiogenic pathways constitute a coordinated mechanism by which tumors may evade treatment with oligomeric NC-1 fragments. Therefore, FN1 level as well as IGF1R/CCL2 regulation might be instrumental in predicting tumor response to cancer therapies consisting of oligomeric-NC-1 fragments.

(24) FIG. 10: A schematic overview of critical motifs within the endostatin (ED)-domain and fibronectin (FN).

(25) The integrin binding domain within FN consists of two motifs, the RGD and PHSRN (SEQ ID NO: 20) motif. In analogy to NC1-ED, heparin binding sites such as HepII are available that mediate binding to other heparin binding factors such as VEGF. The amino-terminal Zinc binding motif of NC1-ED contains two critical histidines (H) that once mutated with Alanine (A) abrogate its activity; adapted from Tjin Tham Sjin et al. Cancer Res 2005, 65, 3656-63 and Wijelath et al. 2006, Circ. Res. 99, 853-860. The sequence of the active motif within NC1-ED containing the two critical histidines (mP1) is HTHQDFQPVLHLVALNTPLSGGMRGIR (SEQ ID NO: 18), and the sequence with the histidines mutated to Alanine (mP1-H1/3A) is ATAQDFQPVLHLVALNTPLSGGMRGIR (SEQ ID NO: 19).

(26) FIG. 11: Superstatin potently inhibits tumor growth.

(27) Wild type LLC tumors (10.000 cells) were implanted s.c. in C57B16 mice. Tumors were sham treated (PBS, control), with the reference FN-mimetic-peptide (FN-Motif) containing only the LYAVTGRGDSPASSK sequence (SEQ ID NO: 8) or with murine Superstatin (SEQ ID NO: 7) at the dose of 50 g peptide in 100 l PBS every 12 h s.c. (n: 5 in each group). Treatment was started 4 days after tumor implantation (prevention trial) and continued for 24 days. Of note, during the treatment period only a single tumor grew in the Superstatin group. Two additional tumors appeared only after cessation of Superstatin therapy indicating that these hard-to-treat tumors were controlled by this therapy. Tumor size reaching 1000 mm.sup.2 was considered as death event in the Kaplan-Meier analysis. Superstatin significantly prolonged survival (p<0.03 by log-rank test) as compared to control. In contrast, the FN-Motif alone (SEQ ID NO: 8) showed no significant improvement in prevention of tumor growth.

(28) FIG. 12: Knockdown of Fibronectin (FN) rendered tumor resistant to oligomeric NC1 substrates, as exemplified for ED-Dimer (Fc-Endostatin).

(29) In contrast, the Superstatin peptide (SEQ ID NO: 7) exerts potent tumor growth inhibition in FN / LLC tumors growing s.c. in C57B16 mice.

(30) FIG. 13: Sequential treatment with IGF1R-Inhibitor is effective in treatment of murine Fc-Endostatin (muFcEndo)-resistant LLC tumors.

(31) Previously generated passage four Fc-Endostatin-resistant LLC cells (Endo P4, as described in the following examples) were subcutaneously injected to C57B16 mice. In passage 5, tumors grew faster if endostatin selection pressure was maintained as compared to sham treated tumors (Ctrl.). Sequential inhibition of IGF1R signaling (20 mg/kg cyclolignan picropodophyllin, PPP, IP injection) inhibited tumor growth. However, concurrent administration only partially reversed the enhanced growth kinetic induced by mFc-Endo selection pressure.

(32) FIG. 14: Differential expression of proteins in passage 5 (P5) Fc-Endostatin-(Endo)-resistant LLC tumor cells. Protein analysis by Western blot further confirmed enhanced IGF1R expression and phosphorylation (p-IGF1R), down-regulation of Fibronectin and up-regulation of CCL2 as the function of therapy with murine Fc-Endostatin (Endo) in passage 5 LLC tumors. Sequential treatment with IGF1R inhibitor partially reversed this phenotype.

EXAMPLES

(33) The invention will now be illustrated by examples which shall, however, not be construed as limiting the scope of the invention.

Example 1: Materials and Methods

(34) 1.1 Cell lines and cell culture. Human tumor cell lines A2058 (melanoma) and human pancreatic cancer cell line ASPC-1 were cultured in DMEM with L-glutamine and supplemented with 10% FCS and antibiotics. HUVEC (Lonza, Switzerland) was maintained in EBM endothelial growth media and EGM Bullet Kit (Lonza, Switzerland) with antibiotics.

(35) 1.2 Expression and Purification. Construction, expression and purification of human Fc-endostatin (hFc-endostatin), artificial endostatin dimer and NC-1 have been described previously (Bergers et al., 1999, Science 284, 808, Lo et al. 1998, Protein Eng 11, 495; Kuo et al. 2001, J Cell Biol 152, 1233; Wen et al. 1999. Cancer Res 59, 1233). The recombinant constructs were prepared by placing the Fc domains at the N-terminus of endostatin Stable cell lines of these constructs were produced in NS/0 murine myeloma cells. The proteins were expressed and secreted into the media. Protein A was used for purification of the recombinant proteins (at least 90% purity). Approximately, 50 mg/liter of Fc-endostatin has been obtained by employing fermentors of 10-18 liter capacity. Human collagen 18 NC-1 preparation was described previously (Wen et al., loc. cit.) The protein was expressed and secreted into the media and purified on Ni-Agarose (Invitrogen).

(36) 1.3 Surface Plasmon Resonance (SPR) binding assays. This analysis provides for a unique method for measuring equilibrium constants between two binding partners. It is able to evaluate the kinetics of an interaction by recording the rates of complex formation (ka) and dissociation (kd) followed by employing a software which determines the values of these two parameters. Equilibrium constant (KD) is obtained by calculating the ration of kd/ka Human VEGF (R&D), endostatin, endostatin dimer and human NC-1 (hNC-1) were diluted to 50 g/ml in 10 mM Sodium Acetate, pH 5.5 and immobilized onto series S sensor chip(s) CM4 via a standard N-ethyl-N-(dimethyl-aminopropyl)carbodiimide/N-hydroxysuccinimide (EDC/NHS) amine coupling procedures. Control surfaces were prepared similarly without protein derivatization and utilized as a reference surface for compound binding experiments.

(37) Binding measurements were performed using a Biacore (GE Healthcare, Uppsala, Sweden) instrument which employs surface plasmon resonance to detect and monitor molecular interactions.

(38) Data analysis was carried out using Biacore T100 evaluation software v1.1.1. Data were prepared by subtraction of reference surface data and blank buffer sample data, a procedure commonly referred to as double referencing and fitted to a 1:1 langmuir binding model.

(39) 1.4 Animal and tumor models. All animal procedures were carried out in compliance with Children's Hospital Boston guidelines Protocols were approved by the Institutional Animal Care and Use Committee. Eight-week-old male (24-27 g) nude/nude mice (Massachusetts General Hospital, Boston, MA) were used. Mice were acclimated, caged in groups of five in a barrier care facility, and fed animal chow and water ad libitum. Animals were euthanized by CO.sub.2 inhalation. Human melanoma cell line A2058 was used for animal studies. A suspension of 210.sup.6 tumor cells in 0.1 ml of PBS was injected subcutaneously (s.c.) into the dorsa of mice at the proximal midline. Mice were weighed and tumors were monitored twice a week in two diameters with digital calipers. Tumor volumes were determined using a.sup.2b0.52 (where a is the shortest and b is the longest diameter). Tumors were allowed to grow to 100 mm.sup.3 and mice were randomized. Treatment was by bolus s.c. injections. After experiments were completed, tumors and organs were excised and fixed in either 4% paraformaldehyde or were snap frozen. Four to six mice were treated with each group.

(40) 1.5 Immunocytochemistry. HUVECs were plated and grown on cover slips. Cells were incubated with 10 g/ml hFc-endostatin, hFc, hNC-1 or control IgG for 120 min at 37 C. and then fixed. The slips were incubated in the blocking buffer (2% BSA PBS) for 30 min. For hFc-endostatin or hFc groups, the slides were incubated with Alexa 488 antihuman IgG for imaging. For hNC-1 or IgG groups, the slides were incubated with mouse anti-His-tag monoclonal antibody, then probed by Alexa 488 anti-mouse IgG. Anti-fibronectin antibody (R&D) was used for secondary staining for all slides and probed by Alexa 594 anti-goat IgG and imaged by confocal-microscopy. DAPI counterstaining of nuclei is shown in blue.

(41) 1.6 Immunohistochemistry. Tumors sections were rinsed by cold PBS and fixed with 4% paraformaldehyde for 10 min with before staining. Human Fc-endostatin was detected by Alexa 488 anti-human IgG. Antibodies to collagen 18 (R&D), fibronectin (R&D), integrin 5 (R&D) CD31 (BD Pharmingen, San Jose, CA) and von Willebrand Factor (Dako, Carpinteria, CA), -SMA (Dako, Carpinteria, CA) were used for staining. The primary antibodies were detected by Alexa 488 or 594-labeled secondary antibodies (Molecular Probes, Eugene, OR). The sections were imaged by confocal-microscopy (model DM IRE2: Leica).

(42) 1.7 Endothelial cell migration assay. HUVECs were washed by serum free EBM medium twice, re-suspended at 510.sup.4 cells/well in 0.6 ml of medium, were plated into 24-well inserts (Coring, 8 m pore size) in duplicates. The lower chamber was filled with 0.6 ml of serum free EBM medium containing 100 ng/ml rhesus (rh)VEGF (R&D) After incubation for 16 h at 37 C., the cells were fixed by methanol and stained with eosin and hemotoxlin. Cells on the upper side of the transwell membrane were removed by cotton swab. Cells migrating to the lower side of membrane were counted.

(43) 1.8 Flow-cytometry analysis of human Fc-endostatin (hFcES) binding on cell-surface. All operations were performed at 4 C. HUVECs were trypsinized and resuspended in PBS (2% BSA) for 30 min followed by 1 h incubation with 1 and 10 g/ml hFcES or hFc. Cells were centrifuged and washed by cold PBS, and then incubated with FITC-labeled secondary antibodies (Sigma, St. Louis, MO) against human Fc fragment and analyzed by BD Biosciences FACS Calibur flow cytometer.

(44) 1.9 Statistical methods. Data are expressed as means plus or minus SD. Statistical significance was assessed using the Student t test.

Example 2: Results

(45) 2.1 Dimeric Endostatin and NC-1 bind Fibronectin. Because of reports demonstrating the existence of endostatin in platelets, it has been proceeded with a platelets lysate to identify proteins binding to endostatin (Italiano et al. 2008, Blood 111, 1227). One of the advantages of Fc-endostatin is that it enables to use this construct for immunoprecipitation (IP) without introducing an additional antibody to form a complex. Three protein constructs human Fc (control), dimeric human (h)Fc-Endostatin and hFc-VEGF were employed. The data are shown in FIG. 1(A). Comparing IP results for the three above mentioned reagents, the major difference among the lanes of coomassie stained polyacrylamidegel was in the vicinity of 200 kDa. Mass spectra analysis of this region led to identification of fibronectin as the candidate binding protein for both endostatin and VEGF. To confirm binding of endostatin and VEGF to fibronectin, Elisa was performed (FIG. 1B). Endostatin dimer, NC-1, VEGF and integrin 51 (a natural receptor for fibronectin) bound fibronectin whereas endostatin monomer did not. Artificial endostatin dimer was generated by introducing a single mutation in amino acid position 7 of endostatin (changing glutamine to cysteine) in the recombinant Fc-endostatin. Upon digestion of this mutant protein by enterokinase, endostatin dimer is produced, linked by a disulfide bond (Kuo et al., loc. cit.).

(46) In order to measure equilibrium binding constants for the above proteins, a Biacore system was employed. High affinity constants were obtained for endostatin dimer, NC-1 and VEGF (FIG. 1C). No binding was detected between endostatin monomer and fibronectin. In support of this data, VEGF binding to fibronectin has been reported by other groups previously (Wijelath et al. 2006, Circ Res 99, 853). Fc-endostatin imposes a dimeric structure on the two molecules of endostatin at the C-terminus of Fc dimer. Binding constant of Fc-endostatin to fibronectin is similar to endostatin dimer (data not shown).

(47) 2.2 Immunohistochemistry studies demonstrate that dimeric Endostatin targets endothelial cells through fibronectin. Immunofluorescence (IF) analysis was used to verify the systemic distribution of hFc-endostatin in an ASPC-1 xenograft animal model. The distribution of exogenous oligomeric endostatin was detected, using antibody to Fc, and shown in FIG. 2A. The imaging results showed that injected oligomeric endostatin was found not only in tumor but also in heart and kidney endothelial cells. In addition to CD31 (endothelial cell) staining, the pericyte marker, -SMA was used to confirm the interaction of endostatin with blood vessels (FIG. 2B). To examine binding of endostatin to fibronectin, hFc-endostatin treated tumor sections of xenograft models were prepared. The exogenous dimeric hFc-endostatin (hFcES) shows co-localization with endogenous fibronection (FIG. 2C). Co-localization of endogenous collagen 18 and endogenous fibronectin has also been detected (FIG. 2D). Integrin 51 is a receptor of fibronectin (Hynes 1992, Cell 69, 11). The imaging data showed that hFcES was also co-localized with integrin 51 (FIG. 2E). Finally, we detected co-localization of VWF, integrin 51 on tumor sections (FIG. 2F). These data demonstrate close proximity of dimeric-hFcES, fibronectin, integrin 51 and blood vessels. In addition, ex vivo E14.5 mouse brain embryonic sections were used to confirm this phenomenon. Fibronectin, VEGF and hFcES show similar binding pattern in mouse embryonic brain (FIG. 2G), confirming that VEGF is a component of this assembly.

(48) 2.3 Binding of oligomeric hFc-endostatin and hNC-1 to HUVECs. Immunoflourescence technique was employed to detect endothelial cell surface binding of oligomeric endostatin and NC-1. Fc-endostatin and NC-1 showed a poor binding to HUVECS when the cells were cultured only for 24 hours (data not shown). Prolonging incubation time to 72 hours resulted in significant binding, possibly, as a consequence of fibronectin upregulation (FIG. 3A upper panel). These data suggest a critical role for fibronectin as mediator of dimeric endostatin binding to endothelial cells. 3D image showed the distributions of endostatin and fibronectin on endothelial cells.

(49) 2.4 hNC-1 inhibits endothelial cells migration. Endothelial cell migration is an important step in new blood vessel formation and tumor angiogenesis. To evaluate the effect of hNC-1 on endothelial cell migration, rhVEGF has been used to induce HUVECs migration in a transwell assay. The migration of cells has been monitored and quantified Cells migrating across the membrane were stained with blue-purple stain (FIG. 3B(a)) and counted. The hNC-1 inhibited VEGF-induced endothelial cell migration as a function of concentration and the dose effect followed a U-shaped curve (FIG. 3B(b)) (Lee et al., loc. cit.). NC-1 was the most potent anti-migratory agent among different endostatin molecules tested (FIG. 3B(c)).

(50) 2.5 Absence of Endostatin in human circulation. Endostatin was initially isolated from condition media of a mouse tumor cell line (EOMA). Purification of the proteins involved a number of steps. Identification of its N-terminus sequence resulted in an endostatin molecule starting at the N-terminus sequence HTH. Both recombinant mouse and human endostatins were constructed on the basis of this N-terminus sequence (O'Reilly et al. 1997, Cell 88, 277).

(51) The size of endostatin in the sera of several individuals has been investigated. To this end, a half-liter bag of PRP (platelet rich plasma) collected 4 days earlier from 1-4 individuals was obtained. Following a low RPM centrifugation to remove platelets, a high RPM was applied to obtain the serum. The serum was passed through Protein A to remove IgGs. It was then subjected to immunoprecipitation by an endostatin polyclonal antibody followed by Western analysis. A monoclonal antibody directed to endostatin was used for treating the membrane (FIG. 4). The protein markers are NC-1 and endostatin. The size of identified protein is between the two markers with several protein bands present (FIG. 4A). An identical procedure was applied to a serum from a different individual (FIG. 4B) To confirm these data, following Protein A step, affinity purification of an earlier PRP sample (different from the referred PRP bag) yielded similar size molecules (FIG. 4C). Finally, immunoprecipitation of the platelets by the same polyclonal antibody showed that the identified protein is a larger size molecule than endostatin (FIG. 4D).

(52) These results prompted the inventors to reanalyze the earlier data with endostatin (O'Reilly et al., loc. cit). The earliest condition media employed for endostatin isolation was continuously kept at 4 C. for a long period of time (at least a month) prior to final purification step. Presence of fetal calf serum in the media can cause degradation of NC-1. Confirmation of this hypothesis was established by another group in the inventor's laboratory a year later, employing the same mouse tumor cell line (Wen et al., loc. cit.). Upon shortening the duration of protein isolation and employing a single step of purification, the major proteins were found to be mNC-1 and endostatin (Wen et al., loc. cit.). Consequently, the earlier observation of an endostatin size protein being the only product following HPLC purification, strongly supports the inventor's hypothesis that NC-1 was completely digested to an endostatin size molecule after prolonged incubation at 4 C. Additional evidence for this hypothesis comes from published data by a different group of investigators (Sasaki et al 1998, EMBO J 17, 4249).

(53) Most probably, the first reported endostatin was a result of degradation taking place in the laboratory. NC-1 is the precursor of endostatin. The data of the inventors on human sera were based on the experiments performed days after collection and purification and the inventors could not avoid digestion of NC-1 to smaller fragments. The inventors conclude from these data that in contrast to mouse EOMA, 187 amino acid endostatin is absent in human serum and platelet. Some of these larger than endostatin degradation products of NC-1 may correspond to dimers which are possible candidates to be present in human circulation or results of NC-1 post collection degradation. Consequently, the present data strongly suggest that NC-1 is the most physiologically relevant molecule in human circulation.

(54) 2.6 hNC-1 suppresses tumor growth in vivo. In order to compare anti-tumor activities of NC-1 and endostatin, nude mice bearing human melanoma cell line A2058 were employed. The data are shown in FIG. 5. Two doses of clinical grade endostatin differing by 5-fold in concentration of protein (low and high doses) were used. A corresponding low dose of NC-1 showed anti-tumor activity similar to high dose of endostatin. The results demonstrate a better anti-tumor efficacy by NC-1 compared with endostatin. It cannot be ruled out the possibility that injected NC-1 becomes degraded upon entering the circulation.

(55) 2.7 Generation and Expression of a mFc-NC-1 Fusion Protein.

(56) A murine Fc-NC-1 fusion construct comprising the non-triple helical trimerization domain (association domain), the hinge region and the endostatin domain (comprising the zinc binding site) of the NC-1 domain of mouse collagen 18 shown in SEQ ID NO: 3 has been generated and expressed in 293 kidney cells. In this construct, the Fc domain shown in SEQ ID NO: 5 is located at the N-terminus and the murine NC-1 domain (shown in SEQ ID NO. 3) at the C-terminus of said fusion protein. In addition, a linker carrying an enterokinase cleavage site has been interposed between Fc and NC-1, in order to allow for cleaving the fusion construct. It is evident to those skilled in the art that a corresponding therapeutic product for administering in patients will not have such a cleavage site. FIG. 7 shows the picture of an SDS gel of said murine Fc-NC-1 fusion construct as a function of the amount of plasmid used in transfection of 293 kidney cells under reduced and non-reduced conditions. Under reduced conditions, the product is a single chain consisting of Fc and NC-1. Under non-reduced conditions, the product is a dimer because Fc is disulfide bonded. Dimeric Fc can accommodate only two chains of NC-1 by covalent linkage. It is therefore presumed that the third chain of NC-1 which is non-covalently associated gets dissociated in the presence of SDS and, thus, is in loading sample and the gel itself. The single band in the middle of the gel is due to the endostatin marker which has a molecular weight of 20 Kd. Subsequently, said construct will be tested for anti-tumor activity and longer half-life of this protein.

(57) 2.8 Acquired Drug Resistance to Endogenous Angiogenesis Inhibitors: Endostatin

(58) Murine Lewis Lung Carcinoma (LLC) tumors were implanted s.c. into C57/B16 mice. Tumors were treated with a murine oligomeric-NC1 fragment, mFc-endostatin, at doses indicated. LLC tumors were exposed in-vivo for a prolonged period to mFc-endostatin by sequential transplantation of tumors (up to four passages, p4) once they evade therapy and reach a size of approximately 1000-2000 mm.sup.3. Control tumors were also sequentially transplanted into new mice without mFc-endostatin treatment. Tumor volume was measured by caliper measurement (FIG. 8). After four passages, both mFc-endostatin resistant as well as control tumors were excised, total RNA was isolated and after QC hybridized on genome-wide microarrays for transcriptional analysis. Microarray data were analyzed using SUMO software package. Quantitative real-time RT-PCR was performed using Taqman technology to confirm the expression of candidate genes identified by microarray analysis.

(59) As a result, this data shows that, in contrast to previous reports, the inventors were able to generate tumors resistant to mFc-endostatin that mimics the NC-1 effect described elsewhere herein. These tumors were generated by sequential implantation and treatment of tumors, in murine lung cancer (LLC) and human pancreatic adenocarcinoma (BxPC3) up to 4 passages. Genome-wide expression profiling revealed down regulation of fibronectin rendering tumors resistant to mFc-endostatin treatment (FIG. 9). This is in line with the inventor's observation of selective binding of oligomeric mFc-endostatin and oligomeric NC-1 to fibronectin. Therefore, analysis of fibronectin levels can be used as a prognostic marker for cancer therapy response. Further, the inventors identified a number of compensatory pathways being activated rendering tumors resistant to Fc-endostatin therapy, in particular sequential treatment with IGF1R inhibitors seems promising, according to preliminary animal data, and CCL2 seems to constitute another promising candidate target.

(60) 2.9 Novel hybrid peptides Superstatins with anti-tumor activity

(61) In the previous examples, the inventors hypothesized and provided evidence that the physiological substrate of Collagen 18 in human circulation consists of oligomers of the endostatin domain (ED) from the non-collagenous NC-1 region of collagen 18. Further, they confirmed that synthetic ED-dimer built based on fusion of endostatin to human immunoglobin Fc-region (Fc-Endostatin) binds Fibronectin (FN), whereas the monomer does not. High affinity binding of FN to VEGF was further confirmed. Taken together, the inventors proposed that oligomeric NC-1 may elicit their effects via FN binding via interference with at least two pivotal angiogenesis pathways, i.e., VEGF and integrin alpha 5 beta 1 (ITGA5B1) signaling. Moreover, they found that FN is significantly down-regulated in tumors that become resistant to oligomeric NC-1 (Fc-Endostatin) after prolonged exposure, i.e. four serial in-vivo passages. Therefore, they postulated that loss of FN might constitute a key mechanism of inherent and acquired resistance to oligomeric NC-1 substrates.

(62) The inventors followed two strategies to provide an ultimate proof of this concept.

(63) The first approach was to engineer a minimal peptide sequence that would mimic the key effects of the ED-FN complex. Towards this goal, the inventors selected the most active motif in the entire ED-domain consisting of a 27 amino acid-NH2-terminal region that was originally identified by Dr. Javaherian, one of the present inventors (Tjin Tham Sjin et al. 2005, Cancer Res. 65, 3656-63). Preliminary data by the present inventors indicate that this region itself may be capable of binding to VEGF and that the two histidines (Zinc binding domain) in this peptide sequence may be critical for VEGF binding. This is conceivable, because a mutated peptide in which Histidines were replaced by Alanine residues failed to compete with VEGF-ED-dimer (Fc-Endostatin) binding. On the other hand, fibronectin contains two active motifs that are critical for its binding to ITGA5B1, i.e. a PHSRN-(SEQ ID NO:20) and a RGD-dependent motif. FIG. 10 shows a schematic overview of critical motifs within the ED-domain and FN.

(64) In order to mimic the physiological complex of oligomeric NC-1 and FN that mediated integrin signaling and other properties of the NC-1-ED, the inventors aimed to fuse these two critical motifs, i.e. the above-mentioned most active motif in the NC-1-ED domain and the integrin-binding motif of fibronectin comprising RGD and surrounding amino acids, and generated a hybrid peptide called Superstatin. For each peptide sequence, a human and mouse equivalent was designed, as described in more detail in Example 2.10.

(65) In the following, data in a syngeneic murine lung cancer model (LLC) are presented using the murine Superstatin sequence:

(66) TABLE-US-00001 (SEQIDNO:7) HTHQDFQPVLHLVLYAVTGRGDSPASSK NC1-EDMotif-FN-Motifhybrid

(67) To proof that the effects are not mediated by the FN-Motif per se, a reference peptide was employed lacking the NC1-ED mimetic motif, i.e. consisting only of the FN-Motif:

(68) TABLE-US-00002 (SEQIDNO:8) LYAVTGRGDSPASSK.

(69) Additional constructs containing the PHSRN (SEQ ID NO:20) instead of the RGD motif of FN, as well as constructs facilitating dimerization of the Superstatin peptide via disulfide bounds or Fc regions are currently in preparation or already under in-vivo evaluation; see Example 2.10.

(70) Using the protypic LLC murine (C57BL6) lung cancer model, the inventors were able to show the efficacy of the murine Superstatin peptide to potently inhibit tumor growth.

(71) As shown in FIG. 11, wild type LLC tumors (10.000 cells) were implanted s.c. in C57B16 mice. Tumors were sham treated (PBS), with the reference FN-mimetic-peptide containing only the LYAVTGRGDSPASSK sequence (SEQ ID NO: 8; FN motif) or with murine Superstatin (SEQ ID NO: 7) at the dose of 50 g peptide in 100 l PBS every 12 h s.c. (n: 5 in each group). Treatment was started 4 days after tumor implantation (prevention trial) and continued for 24 days. Of note, during the treatment period only a single tumor grew in the Superstatin group. Two additional tumors appeared only after cessation of Superstatin therapy indicating that these hard-to-treat tumors were controlled by this therapy. Tumor size reaching 1000 mm.sup.2 was considered as death event in the Kaplan-Meier analysis. Superstatin significantly prolonged survival (p<0.03 by log-rank test) as compared to control. In contrast, the FN-Motif alone (SEQ ID NO: 8) showed no significant improvement in prevention of tumor growth.

(72) Once the efficacy of the Superstatin peptide (SEQ ID NO: 7) was confirmed, the inventors moved toward the second strategy to validate the proposed concept that FN signaling is critical for oligomeric NC-1 to exert their anti-angiogenic and anti-tumoral effects. Using a lentiviral shRNA construct against murine Fibronectin, they down-regulated FN in LLCs mimiking the natural phenotype of previously generated p4 Fc-Endostatin resistant tumors. LLC FN / tumor cells (100.000) were implanted in C57B16 mice and treatment with dimeric murine Fc-Endostatin vs. Superstatin was started on day 4 post implantation. Knockdown of FN rendered LLC tumors resistant to oligomeric NC-1 (Fc-Endostatin). Moreover, in analogy to the originally reported data on naturally selected p4 resistant LLC tumor cells, LLC FN/ tumor showed a trend towards faster growth, as compared to control. In contrast, the Superstatin peptide containing both the ED-motif and the FN-motif (SEQ ID NO: 7) significantly inhibited tumor growth, as compared to control or Fc-Endostatin treated tumors (p<0.01); see FIG. 12. These data show that Superstatin activity is independent of FN availability in tumors, hence circumventing the resistance mechanism that renders tumors resistant to oligomeric NC-1.

(73) Based on the findings in the previous examples, the inventors proposed that compensatory up-regulation of pro-tumorigenic and pro-angiogenic pathways identified might constitute a promising target to circumvent acquired tumor resistance to ED derived agents or oligomeric NC-1. Inhibition of two particular pathways, i.e. IGF1R signaling and CCL2 was proposed to be most promising due to availability of pharmacological agents already entering advanced clinical trial stages. Here, the inventors show that sequential but not concurrent administration of an IGF1R inhibitor impaired the growth of tumors that are resistant to murine Fc-Endostatin (NC-1-ED-Dimer); see FIG. 13.

(74) Protein analysis by Western blot further confirmed enhanced IGF1R expression and phosphorylation, down-regulation of Fibronectin and up-regulation of CCL2 as the function of therapy with murine Fc-Endostatin (Endo) in passage 5 LLC tumors; see FIG. 14. Sequential treatment with IGF1R inhibitor partially reversed this phenotype.

Example 2.10: Designed Peptides

(75) Previous data by the present inventors has demonstrated that the active motif of NC-1-ED resides in the N-terminus of the protein and can be mimicked by a 25 amino acid peptide. SEQ ID NO: 9 shows the corresponding murine sequence, whereas SEQ ID NO: 10 shows the corresponding human sequence. Furthermore, in view of the fact that the zinc binding coordinates of ED are mediated by three histidines in this peptide, the inventors have shown previously that substitution of histidines by alanines resulted in a peptide which was inactive in inhibiting tumor growth, angiogenesis and vessel permeability (Tjin Tham Sjin et al., Cancer Res. 2005, 65, 3656-63).

(76) SEQ ID NO: 11 (murine) and SEQ ID NO: 12 (human) show the sequences of the RGD motif and surrounding amino acids in Fibronectin important for binding to integrin alpha 5 beta 1. In addition, the PHSRN motif (SEQ ID NO:20) in the integrin binding domain of Fibronectin has been found to be critical for binding of Fibronectin to ITGA5B1 (integrin alpha 5 beta 1).

(77) Based on preliminary data by the inventors that oligomeric endostatin (NC-1) binds Fibronectin, they have generated a series of peptides which combine motifs of the mentioned endostatin N-terminal peptide with an RGD containing domain of Fibronectin:

(78) SEQ ID NO: 7 shows the corresponding murine Superstatin sequence, whereas SEQ ID NO: 13 shows the corresponding human Superstatin sequence, i.e. the hybrid peptide sequences of the present invention.

(79) In order to examine the importance of the Zinc-binding region of endostatin in relation to the antitumor properties of the hybrid peptide, the inventors have modified the human Superstatin (SEQ ID NO: 13) by replacing the critical histidines at positions 1 and 3 by alanines (SEQ ID NO: 14).

(80) To test for relevance of dimerization properties of the hybrid peptides to antitumor activity, the inventors have employed an earlier reported method of endostatin dimerization (Kuo et al., 2001, JCB 152, 1233-46). By replacing glutamine at position 7 in the human Superstatin sequence (SEQ ID NO: 13) by cysteine, the hybrid peptide should form a dimer and can be tested for activity in tumor-bearing mice (SEQ ID NO: 15).

(81) As a control for RGD binding of fibronectin to the integrin, RGD in the human Superstatin sequence (SEQ ID NO: 13) has been changed to RAD, in a new hybrid peptide (SEQ ID NO: 16) to evaluate its antitumor activity, in comparison with the wild-type peptide.

(82) Presented are the human versions of these modified peptides (denoted by h). The mouse versions are similar sequences (denoted by m) and can be derived based on the sequence information provided herein.

(83) Additional larger control peptides are currently designed containing 25 amino acids of integrin binding regions of fibronectin (SEQ ID NO: 17) to be evaluated in antitumor studies.

(84) Finally, the human Superstatin peptide (SEQ ID NO: 13) is conjugated to the complexing agent 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (also known as DOTA) providing the ability to conjugate the peptide with, e.g., radionuclides such as Gallium (.sup.68Ga) for non-invasive imaging (Positron emission tomography, PET). The inventors check currently if DOTA conjugation is affecting the efficacy of the human Superstatin peptide in-vivo in a BxPC3 human pancreatic cancer model. In case this experiment confirms the activity of the Superstatin-DOTA constructs, in-vivo PET-Imaging evaluating the potential of Superstatin-DOTA as diagnostic agent is envisioned.

(85) The corresponding sequences of the above-indicated peptides are shown in the following Table 1:

(86) TABLE-US-00003 mP1 HTHQDFQPVLHLVALNTPLSGGMRGI (SEQIDNO:9) hp1 HSHRDFQPVLHLVALNSPLSGGMRG (SEQIDNO:10) mPint LYAVTGRGDSPASSK (SEQIDNO:11) hPint VYAVTGRGDSPASSK (SEQIDNO:12) mSUPERSTATIN HTHQDFQPVLHLVLYAVTGRGDSPASSK (SEQIDNO:7) hSUPERSTATIN HSHRDFQPVLHLVYAVTGRGDSPASSK (SEQIDNO:13) hPint.Endo(twoH ASARDFQPVLHLVYAVTGRGDSPASSK (SEQIDNO:14) atN-terminusto A) hPint.Endo(Q7to HSHRDFCPVLHLVYAVTGRGDSPASSK (SEQIDNO:15) C) hPin.Endo(Gto HSHRDFQPVLHLVYAVTGRADSPASSK (SEQIDNO:16) A) mPint-L IKPGADYTITLYAVTGRGDSPASSK (SEQIDNO:17) Human DOTA- (SEQIDNO:13) Superstatin-DOTA HSHRDFQPVLHLVYAVTGRGDSPASSK- CONH2

3. Discussion

(87) In these examples, the inventors have presented strong evidence that the molecular size of endostatin reported in literature is likely a product of protease degradation following collection of mouse cell culture media or human serum. Evidence presented by Sasaki and collaborators (loc. cit) appear to support this hypothesis. Their analysis of a human serum following two purification steps resulted in a number of endostatin-like molecules in terms of their sizes. All three reported molecules had additional amino acids at the N-termini, pointing to absence of an endostatin molecule starting with the reported histidine at the N-terminus Amino acid sequence analysis of higher molecular weight bands was not reported. Furthermore, the authors' investigation of collagen 18 distribution in different organs of mice indicated absence of endostatin size molecules in different tissue extracts (Sasaki et al., loc. cit.). The most prominent identified protein corresponded to NC-1.

(88) The inventors have identified fibronectin as a binding protein for oligomeric endostatin (Fc-endostatin and artificial endostatin dimer) and not endostatin monomer. NC-1 and endostatin dimer have been shown previously to bind a number of ECM proteins, a property not shared by endostatin monomer (Sasaki et al., loc. cit.; Javaherian et al. 2002, J Biol Chem 277, 45211). However, fibronectin has distinct properties which make it unique among ECM proteins. Angiostatic peptides use plasma fibronectin to home to angiogenic vasculature (Yi et al. 2003, PNAS 100, 11435). Fibronectin contains the sequence of amino acids RGD which allows the protein to bind to a number of integrins. Integrin 51 is a receptor on the cell surface which binds fibronectin. This integrin is an important mediator of angiogenesis (Hynes, loc. cit.).

(89) Endostatin binding to two integrins v3 and 51 was first reported in 2001 (Rehn et al. 2001, PNAS 98, 1024). Presumably, such a binding inhibits interactions of fibronectin with these integrins. Later, another group of investigators presented data indicating that endostatin only binds 51 (Wickstrom et al. 2002, Cancer Res 62, 5580). Endostatin lacks the sequence RGD. Consequently, such a binding must be mediated by other amino acids on endostatin. The present inventors have attempted to demonstrate direct binding of endostatin to 51 employing Elisa, immunoprecipitation and cell adhesion assays without success.

(90) The data of the inventors point to the importance of oligomeric endostatin and NC-1 (Lee et al., loc. cit.). Starting with NC-1 trimer precursor, NC-1 is converted into a dimer following a size reduction as a result of degradation. Finally, endostatin monomer is formed upon further size reduction. The data of the inventors lend support to the hypothesis that hNC-1 and possibly dimers of molecules larger than endostatin are present in circulation. Existence of endostatin size monomer in circulation is questionable. It has been shown previously that the physiological termination of the angiogenesis process by endostatin is well coordinated (Abdollahi et al. 2004, Mol Cell 13, 649).

(91) In order to incorporate the data presented here, the inventors suggest a model where fibronectin serves as template (FIG. 6) Both VEGF and NC-1 bind fibronectin. Presence of VEGF and NC-1 modulate the biological activity via interactions with integrin 51. The interplay of a pro-angiogenic protein (VEGF) and an anti-angiogenic protein (NC-1) may be crucial for regulating angiogenesis at the surface of the cell.

(92) The inventor's utilization of Fc-endostatin (a dimer) has clearly demonstrated that it is far superior to ordinary endostatin (monomer) (Lee et al., loc. cit.). A much smaller dose of Fc-endostatin is required to achieve the same tumor reduction in comparison with endostatin alone (50-100 fold) In the past, the inventors have attributed this difference to longer half-life associated with any Fc conjugated molecules. The present data of the inventors suggest that the dimeric state of endostatin in Fc-endostatin may also contribute to exhibiting better efficacy. On the other hand, following injection of NC-1 into mice, it may undergo a rapid degradation and this may account for lack of a dramatic increase in its anti-tumor effects, in comparison with endostatin. In that case, a fusion construct of NC-1 linked to a Fc domain of an immunoglobulin may turn out to be an improved reagent for anti-cancer treatment among different type of endostatin. Experiments of this type are currently in progress, see Example 2.7.

(93) Together, these data indicate that FN is critical for anti-angiogenic and anti-tumor action of collagen 18 fragments. Superstatin, a hybrid peptide containing the critical FN motif and ED-motif was able to potently inhibit/prevent tumor growth and to reverse the resistant phenotype conferred by tumor specific down-regulation of Fibronectin. Alternatively, rational design of combination therapies and development of innovative scheduling schemes aiming to target compensatory pathways activated in Fc-Endostatin resistant tumors could pose a promising strategy to circumvent- or reverse, inherent or acquired tumor resistance.