© The Author 2006. Published by Oxford University Press.
ARTICLE |
Endorepellin In Vivo: Targeting the Tumor Vasculature and Retarding Cancer Growth and Metabolism
Affiliations of authors: Department of Pathology, Anatomy and Cell Biology and the Cellular Biology and Signaling Program, Kimmel Cancer Center (GB, RC, MB, JJZ, MW, RAI, RVI) and Radiopharmaceutical Research Center, Department of Radiation (CC, MLT), Thomas Jefferson University, Philadelphia, PA; Imaging and Molecular Therapeutics Section, Radiation Oncology Branch, National Cancer Institute, Bethesda, MD (CAB, KC)
Correspondence to: Renato V. Iozzo, MD, Department of Pathology, Anatomy and Cell Biology, Rm. 249 JAH, Thomas Jefferson University, Philadelphia, PA 19107 (e-mail: iozzo{at}mail.jci.tju.edu).
 |
ABSTRACT |
|---|
 TopNotes Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Background: The antiangiogenic approach to controlling cancer requires a better understanding of angiogenesis and the discovery of new compounds that modulate this key biological process. Here we investigated the role of endorepellin, an angiostatic protein fragment that is derived from the C-terminus of perlecan, a heparan sulfate proteoglycan, in controlling tumor angiogenesis in vivo. Methods: We administered human recombinant endorepellin systemically to mice bearing orthotopic squamous carcinoma xenografts or syngeneic Lewis lung carcinoma tumors. We monitored tumor growth, angiogenesis, metabolism, hypoxia, and mitotic index by using quantitative immunohistochemistry and positron emission tomography scan imaging. In addition, we determined the localization of injected endorepellin using near-infrared labeling and immunohistochemistry of frozen tumor sections. Finally, we isolated tumor-derived endothelial cells and tested whether endorepellin could interact with these cells and disrupt in vitro capillary morphogenesis. All statistical tests were two-sided. Results: Endorepellin specifically targeted the tumor vasculature as determined by immunohistochemical analysis and accumulated in the tumor perivascular zones where it persisted for several days as discrete deposits. This led to inhibition of tumor angiogenesis (as measured by decreased CD31-positive cells, mean control = 1902 CD31-positive pixels, mean endorepellin treated = 343.9, difference between means = 1558, 95% confidence interval [CI] = 1296 to 1820, P<.001), enhanced tumor hypoxia, and a statistically significant decrease in tumor metabolism and mitotic index (as measured by decreased Ki67-positive cells, mean control Ki67 pixels = 5970, mean endorepellin-treated Ki67 pixels = 3644, difference between means = 2326, 95% CI = 1904 to 2749, P<.001) compared to untreated controls. Endorepellin was actively internalized by tumor-derived endothelial cells causing a redistribution of
2
1 integrin such that both proteins colocalized to punctate deposits in the perivascular region. Endorepellin treatment inhibited in vitro capillary morphogenesis of both normal and tumor-derived endothelia. Conclusions: Our results provide support for the hypothesis that endorepellin is an effective antitumor vasculature agent that could be used as a therapeutic modality to combat cancer.
|
INTRODUCTION |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The antiangiogenic approach to controlling cancer can succeed only if we deepen our understanding of angiogenic mechanisms and discover new compounds that modulate this key biological process. The tumor endothelial network represents an interesting target for cancer therapy that has recently been exploited either for delivery of anticancer drugs that modulate the tumor's microenvironment or for hindering the tumor's development and biological function (1–3). Tumor angiogenesis is regulated by a balance of pro- and antiangiogenic factors (4–6), often involving heparan sulfate proteoglycans (7). When this balance is altered in favor of angiogenesis, a biochemical switch triggers the recruitment of new blood vessels from the adjacent quiescent vasculature (8), a process that fuels the growth of tumor cells (9). Recent findings have suggested that the extracellular matrix is involved in the control of vascular patterning, morphogenesis, and neovessel stabilization (10). Indeed, the extracellular matrix performs not only structural, but also instructive functions by actively signaling to cells (11,12).
Several biologically active fragments with intrinsic angiostatic activity have been isolated from basement membranes (13). An emerging theme is that these fragments—including endostatin, derived from collagen XVIII (14); tumstatin and arresten, derived from the 3 and 1 chains of collagen IV, respectively (15,16); and endorepellin, derived from the heparan sulfate proteoglycan perlecan (17)—act preferentially on endothelial cells to block angiogenesis (18). In general, these bioactive protein fragments are generated by C-terminal proteolytic cleavage of their parent molecules and have no apparent structural homology with each other. They frequently act as dominant negative ligands for specific integrin receptors and thereby affect cytoskeletal dynamics. This activity leads ultimately to inhibition of endothelial cell migration and capillary morphogenesis (19).
Perlecan plays a fundamental role in vascular biology as scaffolding for blood vessel formation during development (20–26) and in various experimental settings (27–30). In cancer, perlecan acts not only as a modulator of growth factor activity, but also as a physical and bioactive barrier to invading neoplastic cells (19). We previously found that endorepellin, the C-terminal domain V of the perlecan protein core, is a potent inhibitor of angiogenesis in several independent assays, including endothelial cell migration, collagen-induced capillary morphogenesis, and growth of blood vessels in Matrigel and in the chorioallantoic membrane (17,31,32). Thus, endorepellin affects three key steps in angiogenesis: adhesion, migration, and morphogenesis. It exerts these effects by binding to a distinct surface receptor, 21 integrin, an event that leads to disruption of the cytoskeleton and adhesion (31). Both 21 and 11 are key collagen receptors of endothelial cells (10), and they provide critical support for vascular endothelial growth factor (VEGF) signaling, endothelial cell migration, and tumor angiogenesis (33–36).
Here we investigated whether endorepellin could retard tumor growth and angiogenesis in vivo. We also investigated the in vivo distribution of endorepellin after intraperitoneal administration of recombinant protein and its effects on tumor blood vessel complexity, tumor hypoxia, metabolism, and tumor cell growth.
|
MATERIALS AND METHODS |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Cell Lines and Purification of Recombinant Endorepellin Proteins
A431 human squamous cell carcinoma and Lewis lung carcinoma (LLC) cells (American Type Culture Collection, Manassas, VA) utilized for in vivo tumor experiments were routinely cultured in Dulbecco's modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) from Invitrogen (Carlsbad, CA). Human recombinant endorepellin (E3687-S4391) harboring a pentahistidine tag at its C-terminus was purified on an Ni-NTA resin as previously described (17). Briefly, the pCEP-Pu vector containing the sequence of the BM40 signal peptide and full-length endorepellin was electroporated into human embryonic kidney cells (293-EBNA) expressing the Epstein–Barr virus nuclear antigen (EBNA)-1. The GC-rich polymerase chain reaction system (Roche Diagnostics, Alameda, CA) was used for amplification of sequences needed in vector construction. Stable transfectants were obtained by selection with G418 and puromycin (500 µg/mL). The expressed endorepellin was then purified from serum-free conditioned media using Ni-NTA resin and eluted with 250 mM imidazole. The resultant eluate was dialyzed to remove the imidazole and concentrated in dialysis chambers using polyethylene glycol. The purity of each endorepellin batch was confirmed by immunoblotting with antiendorepellin (31) or anti-His (Calbiochem, San Diego, CA) antibodies as previously described (37). To further assess the purity of all preparations of endorepellin, we performed Colloidal Coomassie (Invitrogen) staining of endorepellin on 8% sodium dodecyl sulfate–polyacrylamide gels (SDS–PAGE), a procedure that can detect less than 10 ng of protein (38). When as much as 5 µg of endorepellin was analyzed, no peptide bands besides endorepellin and LG3 (a proteolytically cleaved portion of endorepellin) were present (Supplementary Fig. 1, A; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). These data indicated that our endorepellin preparations were approximately 99% pure. Finally, endotoxin detection assays (E-TOXATE Endotoxin Kit, Sigma, St Louis, MO) were performed on several batches of purified endorepellin. These endotoxin assays, which could detect 0.125 endotoxin units/mL in endotoxin controls provided with the kit, were unable to detect any endotoxin (data not shown) in our endorepellin preparations at endorepellin concentrations as high as 140 µg/mL (1.6 µM). This indicated that endotoxin is either absent or present at concentrations too low to affect the experiments reported here.
In some experiments, endorepellin was labeled with Texas Red via sulfonyl chloride conjugation (Pierce, Rockford, IL). For creation of endorepellin-enhanced yellow fluorescent protein (EYFP) chimera, the cDNA encoding the EYFP (pIRES-EYFP, Clontech, Mountain View, CA) was fused to the C-terminus of endorepellin using the following oligonucleotides: forward, 5'-TTGGCGCGCCATGGTGAGCAAGGGCGAGGA-3' containing an AscI site, and reverse, 5'-CCGCTCGAGCTTGTACAGCTCGTCCATGC-3' containing an XhoI site. The amplified fragment was ligated into the pCEP-Pu mammalian expression vector using the AscI and XhoI restriction sites to create the pCEP-Pu-EYFP, and then the cDNA encoding the full-length endorepellin (17) was inserted upstream of the EYFP moiety. The construct was confirmed to be correct by sequencing and used to produce stable expressing clones in 293-EBNA cells as described above. Positive clones expressing high levels of endorepellin-EYFP were confirmed by detection of endogenous fluorescence and by immunoblotting. All the various endorepellin preparations used in this study were determined to be biologically active by their ability to cause the collapse of endothelial cell actin cytoskeleton in vitro as previously published (31). In some experiments, endorepellin was rendered inactive (for use as a control) by boiling for 5 minutes.
In Vivo Tumor Assays
Male or female nu/nu mice (Charles River Laboratories, Wilmington, MA) or C57BL/6 mice (Taconic, Hudson, NY), housed in a veterinarian-staffed, climate-controlled animal facility at Thomas Jefferson University in clean-bedded cages with access to food and water ad libitum, were subcutaneously injected with 1–2 x 106 A431 cells or LLC cells, respectively, as previously described (39).
The set of mouse experiments performed in this study were as follows: For A431 in vivo tumor assays, 90 nu/nu mice were inoculated with A431 in nine separate experiments. Thirty mice received endorepellin as explained below after tumors became visible and 30 served as vehicle (phosphate-buffered saline [PBS]) control for endorepellin administration that was after tumors became visible. Tumors from these 60 mice as well as wound tissue, liver, and spleen from 20 of these mice (10 from each treatment group) were also analyzed for the presence of endorepellin. Fifteen mice received endorepellin before tumor visibility, and 15 mice served as controls for endorepellin administration before tumors became visible. For investigations of tumor metabolism, 12 nu/nu mice were imaged in three separate experiments by positron emission tomography (PET) and computed tomography (CT) scan (six controls and six endorepellin treated). Twelve A431 tumor–bearing nu/nu mice were utilized for two separate endorepellin tumor localization experiments, six having received IR800-labeled endorepellin as explained below and six having received free dye (control). For tumor hypoxia experiments, nine A431 cell–bearing nu/nu mice (four control and five endorepellin-treated mice) received injections of Hypoxiprobe-1 in two separate experiments as described below. For detection of endorepellin in tumors after acute injections, six A431 tumor–bearing nu/nu control mice were administered endorepellin 6 hours before sacrifice in two separate experiments. For tumor blood vessel quantification, 10 A431 tumors from nu/nu mice were analyzed from two separate experiments, five from control mice and five from endorepellin-treated (after the initial appearance of the tumor) mice. For the heat-inactivated endorepellin treatment experiment, six mice were inoculated with A431 cells, and three received heat-inactivated endorepellin, while three received PBS vehicle control in two separate experiments. For tumor endothelial cell isolation, A431 tumors from five control nu/nu mice from three separate experiments were processed for endothelial cells. For LLC in vivo tumor assays, 15 C57BL/6 mice were inoculated with LLC cells, five received endorepellin before visible tumors, five received endorepellin after tumors became visible, and five served as vehicle control (one single experiment). For tumor blood vessel quantification, 10 LLC tumors from C57BL/6 mice were analyzed, five control and five endorepellin treated (after tumors became visible). In a further LLC experiment, 10 nu/nu mice were inoculated with LLC cells, five served as vehicle control and five received endorepellin after tumors became visible.
After randomization, half of the mice received intraperitoneal injections of recombinant-filtered endorepellin (2–6 mg/kg) approximately 24 hours before visible tumor formation while in other experiments endorepellin was administered within 24 hours after tumors became visible. Endorepellin was subsequently administered every 2 days for 2–3 weeks. Animals were individually identified by 1-mm-diameter ear piercing that also served as wound epithelium for subsequent endorepellin detection experiments. Control animals received equivolume intraperitoneal injections of vehicle (PBS). The tumor diameter was measured every 2 days using calipers and the diameter converted to volume using the equation a(b2/2) where a and b represent the larger and smaller dimensions, respectively. For all in vivo experiments, animals were anesthetized with isoflurane before injections, in accordance with Thomas Jefferson University and National Institutes of Health (NIH) guidelines. At the conclusion of the experiment, usually after 2–3 weeks of treatment, the mice were euthanized by either cervical dislocation or CO2 asphyxia as per Thomas Jefferson University Institutional Review Board approval, photographed (both externally and internally), and the tumor, lung, liver, spleen, and wound epithelium were dissected and snap frozen in liquid nitrogen for immunohistochemical and biochemical analyses. Under the relatively short period of treatment with endorepellin, the mice did not show any signs of illness or loose any weight.
To determine whether a control, heat-inactivated endorepellin harboring no biochemical activity (31), but similar mobility in SDS–PAGE gels (Supplementary Fig. 1, B; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22), would have any effect on A431 tumor growth and/or localize to the tumor/tumor vasculature, nude mice (n = 6, three for each treatment group) were injected with 2 x 106 A431 cells subcutaneously. Twenty-four hours after tumors were visible, mice received either 3 mg/kg of heat-inactivated endorepellin in PBS or equivolume PBS intraperitoneal injections every other day for 2 weeks (six injections total).
PET and CT Scanning and Image Analysis
PET and CT studies of control and endorepellin-treated animals (carried out 2 days before the scheduled conclusion of the in vivo experiment to allow for maximally visible tumors) were performed using the MOSAIC PET scanner (Philips Medical Systems, Bothell, WA) and the MicroCATII CT scanner (Imtek, Inc, Bridgeport, NJ) at Thomas Jefferson University. Mice (n = 12, six control and six endorepellin-treated mice, two each from three separate in vivo experiments) were injected with 0.4–0.5 mCi of [18F]-flourodeoxyglucose (18FDG), and 2 hours were allowed for tracer distribution. Immediately before imaging, mice were anesthetized with an intraperitoneal injection of ketamine, xylazine, and acetopromazine (200, 10, and 2 mg/kg, respectively). Mice were then placed in a 50-mL specimen tube to facilitate multimodality stereotactic positioning. PET data were acquired in a single position for 15 minutes followed by CT data acquisition for approximately 5 minutes. Following PET and CT scanning, the images were registered with an internally developed automated mutual information rigid registration algorithm. For image analysis, volumes of interest were defined by drawing multislice regions of interest on the PET images using 50% of the full-width half maximum of the tumor. When necessary, PET tumor regions were compared to the CT images. Tumor counts of PET signal were normalized to average abdominal counts because defining small subcutaneous regions makes contralateral analysis difficult.
In Vivo Tumor Imaging With Near-Infrared–Labeled Endorepellin
Endorepellin was labeled with the near-infrared dye, IR800CW (Li-cor, Lincoln, NE), according to the manufacturer's guidelines. Briefly, 260 µg of endorepellin was mixed with 26 µg of IR800CW at 4 °C during constant inversion in the dark for 2 hours, to a 6.6 : 1 incorporated dye : peptide ratio. An equivalent amount of free dye was administered as a control for nonspecific dye binding and fluorescent signal emission. Endorepellin–IR800CW was administered in the form of a 100-µL intraperitoneal injection corresponding to 1.3 nmol of endorepellin. An equivalent amount of free dye was administered to mice as a control for nonspecific dye binding and fluorescent signal emission. Animals were imaged using the eXplore Optix small animal molecular imaging unit (Advanced Research Technologies, Saint-Laurent, Canada) at 1, 6, 12, 18, 24, 42, 48, 66, 72, and 90 hours after conjugate administration. This automated imaging system uses a laser diode emitting photons at 780 nm, with absorption and emission filters at 782 and 850 nm, respectively. Images were analyzed using the eXplore Optix software, by Advanced Research Technologies. The 25-mm2 region of maximum signal intensity on the topographic image was selected for fluorescence emission quantification. Relative signal intensities were plotted for kinetic analysis.
Immunohistochemistry, Quantification of Angiogenesis, and Isolation and Characterization of Tumor-Derived Endothelial Cells
Frozen sections of A431 or LLC tumors, obtained with a cryostat (Thermo Electron Corporation, Pittsburgh, PA), control organs (lung, liver, and spleen), and wound epithelium were dried in a vacuum chamber, briefly fixed with ice-cold acetone, washed in PBS, incubated with 5% bovine serum albumin, and then incubated with primary antibodies recognizing the following epitopes: CD31 (PECAM-1, BD Biosciences, San Jose, CA), His6 (Calbiochem), endorepellin (31), 2 integrin (BioLegend, San Diego, CA), Ki67 (Dako, Carpinteria, CA), caveolin-1 (BD Transduction Laboratories, Franklin Lakes, NJ), and VEGF-A (Santa Cruz Biotechnology, Inc, Santa Cruz, CA), all at 1 : 100 dilution. The sections were then incubated with the appropriate fluorescently conjugated secondary antibody, 1 : 200 dilution, including anti-mouse rhodamine-conjugated (sc-2092) or fluorescein isothiocyanate (FITC)–conjugated (sc-2010) antibodies and anti-rabbit rhodamine-conjugated (sc-2091) or FITC-conjugated (sc-2012) antibodies, counterstained with 4'-6-Diamindino-2-phenylindole (DAPI), and visualized by fluorescent microscopy as previously described (31). Images were acquired using an Olympus BX51 microscope driven by SPOT Advanced version 4.0.9 imaging software (Diagnostic Instruments, Inc, Sterling Heights, MI) and a Zeiss Axiovert 200M microscope (Zeiss LSM 5, Carl Zeiss Inc, Oberkochen, Germany). Metamorph 5.0 imaging software (Universal Imaging Corporation, Downingtown, PA) was used for image acquisition and analysis (40).
To quantify tumor angiogenesis, one hundred x 10 magnification images of vascular hot spots, i.e., regions of high vascular density within the tumor, as detailed previously (41) in A431 or LLC tumor sections (five tumors each from control and endorepellin-treated groups, 20 images from each tumor) were analyzed, two separate in vivo experiments examined for CD31 quantification) were obtained with a charge-coupled device camera and the number of CD31-positive pixels per microscopic field were obtained using Adobe Photoshop 9.0 software as previously described (40). Endothelial cells from mouse A431 tumor xenografts were isolated using magnetic beads (Invitrogen) coated with anti-CD31 antibodies. Briefly, tumor tissue was removed aseptically, rinsed in Hanks' balanced salt solution containing penicillin–streptomycin, minced into small pieces using sterilized razor blades, and digested in 20 mL of collagenase A (1 mg/mL, Boehringer Mannheim) and deoxyribonuclease (200 U/mL, Boehringer Mannheim) for 1 hour at 37 °C in a water bath with occasional agitation. The cloudy supernatant containing single cells was removed and centrifuged at 400g for 10 minutes. The pellet was washed twice with DMEM 10% FBS and resuspended in 1 mL of DMEM 10% FBS. Subsequently, the cells underwent two rounds of purification using the CD31-coated magnetic beads (42,43) and were characterized by their ability to internalize acetylated low-density lipoprotein (LDL) labeled with the fluorescent dye Dil (Biomedical Technologies, Stoughton, MA) (44), expression of von Willebrand factor (vWF), and their ability to form capillary-like structures on Matrigel (BD Biosciences). For the capillary morphogenesis assay, endothelial cells were seeded for 5 or 10 hours on growth factor–deprived Matrigel-coated wells in the presence or absence of endorepellin and serum (31,32). At the end of the experiment, the wells were fixed and the tubes were visualized with an inverted microscope.
Apoptosis, Hypoxia, Mitotic Index, and Growth Assays
Frozen tissue sections of tumors were analyzed for the presence and extent of apoptosis with the BD ApoAlert DNA Fragmentation Assay Kit (BD Biosciences) following the manufacturer's instructions. This terminal deoxynucleotidyl transferase nick end labeling (TUNEL) assay utilizes the incorporation of fluorescein–deoxyuridine triphosphate at the free 3' hydroxyl ends of fragmented DNA which can be detected by fluorescence microscopy. For hypoxia assays, the Hypoxyprobe-1 Plus Kit (Chemicon, Temecula, CA) was utilized according to manufacturer's instructions. The method takes advantage of the preferential binding of pimonidazole hydrochloride to tissues with oxygen concentrations less than 14 µM (pO2 of 10 mm Hg, at 37 °C). For these assays, nine mice (four control and five endorepellin treated) received hypoxiprobe-1 probe (60 mg/kg) via intraperitoneal injection 90 minutes before euthanization. This time period allowed for the hypoxia probe to distribute to all tissues in the body. Hypoxia levels were quantified by using a FITC-conjugated anti–Hypoxyprobe-1 antibody and determining the FITC-positive pixels in at least five sections per tumor and 10 images per section as previously described for CD31 quantification. To determine the proliferative (mitotic) index, frozen sections of tumor xenografts were immunostained for Ki67, a proliferation-associated nuclear antigen uniquely expressed in all the phases of the cell cycle, but absent in resting (G0) cells (45). This was followed by quantification of the total pixel density of 20 individual, randomly selected fields from nine mice. Growth assays of A431 and LLC cells (n = 3 separate experiments per cell type) were performed using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI), which measures the bioreduction of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS), a water soluble tetrazolium compound whose levels are directly proportional to the number of living cells.
Statistical Analysis
Statistical analysis was carried out with SigmaStat for Windows version 3.10 (Systat Software, Inc, Point Richmond, CA). Results were compared by using the two-sided Student's t test and were expressed as the mean values with 95% confidence interval (CI). Differences were considered statistically significant at P<.05. All statistical tests were two-sided. The influence of endorepellin treatment on tumor growth was statistically evaluated by two-way repeated measures (RM) analyses of variance (ANOVA), followed by Student's t test.
|
RESULTS |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Slowing of In Vivo Tumor Growth and Reduction of Tumor Metabolism by Endorepellin
We hypothesized that endorepellin could inhibit in vivo tumor growth. To test this hypothesis, we utilized an A431 squamous cell carcinoma orthotopic tumor xenograft model in nu/nu mice. Human recombinant endorepellin (2–6 mg/kg) or vehicle was administered via intraperitoneal injection either just before (Fig. 1, A) or within 24 hours from the appearance of visible tumors (Fig. 1, B). Endorepellin was subsequently administered every 2 days. Each experiment was performed with five mice in the treatment and control arms. Regardless of the therapeutic regimen, i.e., time of injection post-tumor induction, endorepellin had a statistically significant inhibitory effect on tumor volume without any overt side effects (two-way RM ANOVA for tumor growth: for injection before tumor appearance, treatment effect P = .024, sampling time effect P<.001, treatment x sampling time effect P<.001; when first injection was after tumor appearance, treatment effect P = .008, sampling time effect P<.001, treatment x sampling time effect P<.001; when first injection was before tumor appearance, the difference in mean tumor volume on the last experimental day was 531.6 mm3, mean control tumor volume = 603.5 mm3, mean endorepellin-treated tumor volume = 71.9 mm3, 95% CI = 73.6 to 1323, P = .005; and when injection was after tumor appearance, the difference in mean tumor volume on last experimental day was 946.1 mm3, mean control tumor volume = 1143.1 mm3, mean endorepellin-treated tumor volume = 197.0 mm3, 95% CI = 21.1 to 1871.1, P = .047.
|
Control tumors contained several blood vessels directly on the tumor surface as well as blood vessels feeding into the tumor visible on the underside of the skin (Fig. 1, C). By contrast, endorepellin-treated tumors had few or no blood vessels directly on the tumor surface, and visible blood vessels on the skin ran adjacent to the tumor rather than directly into the tumor (Fig. 1, C).
Next, we investigated the possibility that endorepellin inhibits in vivo tumor growth through effects on tumor metabolism. Twelve mice from separate experiments as described above (six controls and six treated mice) were analyzed by CT and PET scan 2–3 weeks into each in vivo experiment (just before each experiment's conclusion). PET scanning allows for direct visualization and quantification of metabolic activity in animals that are administered a radioactive sugar, 18FDG (46,47). The amount of 18FDG taken up by an anatomic region is proportional to the corresponding tissue's metabolic activity. These studies demonstrated markedly lower 18FDG uptake in tumors of endorepellin-treated mice compared to that in controls (Fig. 1, D and E). For all the PET–CT experiments performed, the difference in means of normalized (tumor counts to abdominal counts) signals from tumors of control (mean = 2.902) and endorepellin-treated (mean = 1.476) mice was 1.426 (95% CI = 0.512 to 2.298, P = .01). These findings were not due to differences in tumor size because we found a statistically significant decrease in tumor metabolism even when comparing similar-sized tumors (similarly sized tumors are compared in Fig. 1, D). Furthermore, all imaged endorepellin-treated tumors were more than twice as large as the camera's volumetric resolution (
2 mm3), thereby allowing for valid uptake comparison between tumors. Therefore, PET scan image analysis demonstrated that endorepellin treatment reduced the metabolic activity of tumors.
Localization to Tumor Xenografts, Inhibition of Tumor Cell Proliferation, and Induction of Tumor Hypoxia by Endorepellin
Next, we investigated the in vivo localization of endorepellin in tumor-bearing mice, with the hypothesis that endorepellin targets the tumors. Endorepellin labeled with the near-infrared dye IR800 (48,49) or an equivalent amount of free dye (control) was injected intraperitoneally into mice (n = 3 each, in two independent experiments). Near-infrared optical imaging revealed that the endorepellin–IR800 fluorescent signal was localized at the site of the tumor (Fig. 2, A). Tomographic rendering of the volume reconstruction of endorepellin-treated tumors revealed that IR800-labeled endorepellin localized preferentially to the central areas of the tumors (Fig. 2, B). A major peak of intensity was observed at 48 hours after injection of labeled endorepellin (Fig. 2, C), and statistically significant fluorescence was observed for up to 72 hours (P<.05).
|
The above results suggested that endorepellin could affect tumor growth via direct effects on tumor cells or through its effects on tumor endothelial cells. Endorepellin treatment neither increased apoptosis (Fig. 2, D) detected by the TUNEL assay for DNA fragmentation nor inhibited in vitro tumor cell growth using various concentrations (10–20 µg/mL) of recombinant protein (Fig. 2, E).
A possible consequence of a diminished tumor blood supply that could also account for the changes in tumor cell proliferation caused by endorepellin could be an increase in tumor hypoxia. Endorepellin caused a statistically significant increase in total tumor hypoxia (Fig. 2, F and G) (mean control = 16 575.6 pixels, mean endorepellin treated = 7870.7 pixels, difference between mean pixel number = 61 495.1, 95% CI = 52 088.2 to 70 902, P<.001). Furthermore, endorepellin-treated tumors had many regions of hypoxia throughout the tumor, while in control tumors, hypoxia was restricted to the tumor margins. No induction of hypoxia was observed in other organs (data not shown), suggesting a tumor-specific effect of endorepellin.
In contrast to the effects of endorepellin on cultured A431 cells, systemic treatment with endorepellin of A431 tumor xenograft–bearing mice markedly inhibited the proliferation of the tumor as determined by quantitatively measuring the mitotic index with an antibody against human Ki67 (Fig. 2, H and I), a proliferation-associated nuclear antigen uniquely expressed in all the phases of the cell cycle, but absent in resting (G0) cells (45). Control tumors exhibited diffuse labeling for Ki67, whereas both acute (6 hours) and chronic treatment with endorepellin showed only pockets of Ki67-positive cells, especially around vascular channels (Fig. 2, H), identified by anti-CD31 immunohistochemistry (not shown). Quantification of Ki67-positive cells in vehicle- and endorepellin-treated tumor xenografts showed that endorepellin caused an approximately 40% and 70% inhibition of the mitotic index after acute (mean control Ki67 pixels = 5970, mean endorepellin-treated Ki67 pixels = 3644, difference between mean pixel number = 2326, 95% CI = 1904 to 2749, P<.002) and chronic (mean control pixels = 6010, mean endorepellin-treated pixels = 1510, difference between means = 4500, P<.001) treatment, respectively (Fig. 2, I).
Collectively, these results indicate that systemic delivery of recombinant endorepellin specifically targets tumor xenografts and suggest that endorepellin could affect tumor growth by increasing tumor hypoxia and decreasing the proliferative rate of the tumor cells. Both effects are consistent with an indirect effect on tumor metabolism via restriction of blood supply.
Endorepellin Targeting the Tumor Vasculature and Inhibition of Angiogenesis
Given endorepellin's angiostatic effects, we hypothesized that it might be absorbed by tumor endothelial cells in vivo. To test this hypothesis, we performed experiments in which vehicle-treated A431 tumor–bearing mice with a mean tumor volume of approximately 1100 mm3 (n = 6, two each from three separate experiments) received a single intraperitoneal injection of endorepellin 6 hours before sacrifice. Probing tumor sections for endorepellin with the anti-His antibody, we observed a localization of endorepellin at the luminal surface of all tumor blood vessels (a representative image of endorepellin distribution is shown in Fig. 3, B), suggesting that this is the major entry way of endorepellin. No signal was observed in all the tumor-bearing mice (Fig. 3, A) that did not receive endorepellin (n = 20, four each from five separate experiments). To further verify this finding, we labeled endorepellin with Texas Red and found a similar localization to the tumor vasculature (not shown). No detectable signal was found in several other organs with either anti-His immunohistochemistry or Texas Red–labeled endorepellin, suggesting that within the 6 hours after administration, endorepellin uptake was specific to the tumor vasculature.
|
We next investigated the fate of endorepellin after chronic (2–3 weeks) treatment. As expected, with anti-His immunohistochemistry, no endorepellin was detected in tumor blood vessels from control mice (Fig. 3, C), whereas the endorepellin-treated animals (n = 30, five tumors from six separate experiments) showed a prominent deposition of the protein in perivascular regions (Fig. 3, D). At higher magnification, multiple deposits of endorepellin were evident within the endothelium, vascular basement membranes, and the immediate subvascular regions, with little or no punctate deposits in the tumor proper (Fig. 3, E). Localization of endorepellin in the tumor was corroborated by immunoblotting of tumor extracts. The anti-His antibody detected both endorepellin and the terminal LG3 fragment of endorepellin (Fig. 3, F, lane 2), which is generated by bone morphegentic protein (BMP)-1/Tolloid-like metalloproteases (32). When protein extracts from tumors were probed with the anti-endorepellin antibody, we observed not only endorepellin and LG3 but also other intermediate forms, from which the C-terminal His tag had presumably been removed (data not shown). We next examined whether endorepellin might localize to wound endothelium. We did not detect endorepellin in the proliferating endothelium of 1-mm ear hole piercings made contemporaneously with the injection of tumor cells in mice to which endorepellin was chronically administered (Fig. 3, G). This was another indication that endorepellin specifically localizes to tumor endothelium, unlike endostatin (48,49). Endorepellin did not appear to inhibit the healing/closure of the piercing (data not shown) as has been reported for endostatin (50), further underscoring the tumor vascular specificity of endorepellin. Normal tissues (including liver and spleen) did not demonstrate any measurable endorepellin accumulation (Fig. 3, G).
We quantified tumor blood vessels (n = 5 tumors from control and endorepellin-treated animals, with 20 images per tumor) by probing tissue sections with antibody to CD31. Endorepellin treatment caused a statistically significant decrease (mean control = 1902, mean endorepellin treated = 344, difference between means = 1558, 95% CI = 1296 to 1820, P<.001) in the number of CD31-positive pixel number per microscopic field, reflecting the number of tumor blood vessels (Fig. 3, H and I). Not only was the tumor vasculature decreased in the endorepellin-treated xenografts, but the overall morphology was also altered; endorepellin-treated tumors contained numerous keratin pearls and showed a more cohesive morphology (not shown) suggesting some degree of cytodifferentiation toward a less aggressive phenotype.
We tested whether heat-inactivated endorepellin, possessing no biochemical activity but with similar electrophoertic mobility (31), would affect A431 tumor growth and/or localize at the tumor vasculature. At the conclusion of the experiment (24 hours after the last intraperitoneal injection), the difference (44.2 mm3) between the mean tumor volume of the PBS control tumors (735.8 mm3) and the heat-inactivated endorepellin–treated tumors (780.0 mm3) was not statistically significant (P = .87, 95% CI = 660.3 to 748.7). Immunohistochemical analysis of frozen tumor sections with anti-His antibody, which is capable of detecting heat-inactivated endorepellin by Western blot (data not shown), failed to demonstrate any detectable administered endorepellin within these chronically treated tumors (Supplementary Fig. 1, D; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). Further analysis with anti-CD31 antibody demonstrated no qualitative difference in blood vessel density or appearance (Supplementary Fig. 1, D; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). Therefore, a heat-inactivated form of endorepellin lost its antitumor, antiangiogenic, and tumor vascular homing activity.
Inhibition of the Growth of Syngeneic LLC Tumors by Endorepellin
To test whether endorepellin was active in a nonimmunocompromised animal and against other tumor cell types, we utilized an established syngeneic tumor animal model, the LLC model in C57BL/6 mice (14,51). Systemic delivery of endorepellin using the same protocol as was used for the A431 tumor xenograft (delivery was initiated after the tumors became visible) caused more than 90% inhibition of tumor growth (n = 15, five for each group, control or endorepellin given before or after the appearance of the tumor xenografts; two-way RM ANOVA for tumor growth: treatment effect P<.001; sampling time effect P<.001; treatment x sampling time effect P<.001; mean control = 2664 mm3, mean endorepellin treated = 503 mm3, difference between means on last experimental day = 2161 mm3, 95% CI = 636 to 3688, P = .01, Fig. 4, A). Furthermore, endorepellin nearly completely blocked the growth of LLC tumors when administered just before the appearance of the tumors (difference between mean tumor volumes on last experimental day = 2530 mm3, 95% CI = 1218 to 3843, P = .006, Fig. 4, A). As in the case of A431 tumor xenografts, endorepellin did not cause any detectable apoptosis as detected by the TUNEL assay on tumor sections (Fig. 4, B) but caused a statistically significant decrease in CD31-positive blood vessels (mean control pixels per microscopic field = 3.31, mean endorepellion-treated pixels per microscopic field = 1.45, difference in mean number of CD31-positive pixels = 1.860, 95% CI = 1.706 to 2.015, P<.001, Fig. 4, C and D) as well as a marked reduction of the mean blood vessel length (not shown). Endorepellin antibodies stained the tumor vasculature with a linear rather than punctate pattern, in close association with CD31-positive vessels (Fig. 4, E), but it did not cause any growth inhibition of LLC cells in vitro (not shown). Finally, endorepellin inhibited the growth of LLC tumor xenografts in nu/nu mice (not shown). Thus, endorepellin is active in squamous carcinoma xenografts and syngeneic LLC tumors, further suggesting that its activity is independent of tumor type and that this activity could be mediated primarily by targeting of the vasculature rather than the tumor cells.
|
Colocalization In Vivo of Endorepellin and Tumor Endothelial Cell
21 Integrin
In a previous work, we demonstrated that the endothelial cell 21 integrin is a receptor for endorepellin in vitro (31) and we hypothesized that this integrin could also serve as an endorepellin receptor in vivo. Specifically, our previous work has demonstrated that: 1) endorepellin binds to 21 integrin–positive cells, but not otherwise identical to 21 integrin–negative cells; 2) endorepellin binds to the ligand binding I domain of 2 integrin in surface plasmon resonance experiments; 3) functional blockade of endothelial cell 21 integrin, but not blockade of other endothelial cell integrins, inhibits endorepellin bioactivity and prevents adhesion of endothelial cells to endorepellin; and 4) soluble 2 I domain–glutathione S-transferase (GST) protein, but not other integrin soluble I domain–GST proteins, inhibits endorepellin bioactivity. Therefore, we analyzed A431 tumors after acute and chronic endorepellin administration using an anti-mouse 2 integrin antibody that allowed us to assess the amount and distribution of 21 in stromal cells of murine origin and the human grafted A431 cells. The 2 integrin signal could be detected in both control and endorepellin-treated tumor endothelial cells (Fig. 5, A). The distribution of the 2 signal after acute or chronic treatment (Fig. 5, B and C) resembled that of the punctate endorepellin deposits shown in Fig. 3, B, and indicates that 2 integrin colocalized with endorepellin on the tumor endothelial cells (Fig. 5, D–F). These data indicate that endorepellin binds to and causes a redistribution of tumor endothelial cell 21 integrin in vivo, and they are further evidence that this integrin is a functional receptor for endorepellin on the surface of tumor endothelia.
|
Effect of Endorepellin on Tumor Endothelial Cell Function
Using anti-CD31–coated magnetic beads, we isolated endothelial cells from A431 tumor xenografts and confirmed their endothelial cell identity using several criteria: a cobblestone appearance when reaching confluence (Fig. 5, G), ability to internalize acetylated LDL labeled with the fluorescent dye Dil (Fig. 5, H), presence of an established endothelial cell marker (44), expression of vWF (Fig. 5, I), and ability to form capillary-like structures on Matrigel (Fig. 6, A). Furthermore, fluorescent microscopy using anti-CD31 and anti-vWF showed that more than 98% of the cells were positive for both antigens, whereas immunostaining with a mouse monoclonal antibody directed against domain III of human perlecan (52) (this antibody recognizes only human epitopes) was negative (not shown), indicating that there was no contamination by human-derived cells. These cells could be readily labeled with the endothelial cell–specific marker Bandeirea Simplicifolia Lectin I-B4 (Supplementary Fig. 1, C; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). Importantly, these cells also expressed high levels of 21 integrin (Fig. 5, I–K), with a cell surface distribution similar to that observed in human umbilical vein enthothelial cells (HUVECs) (not shown). In contrast, the human-derived A431 cells did not react with the mouse-specific 21 integrin antibody (Fig. 5, J).
|
We next investigated the interaction between early-passage tumor endothelial cells and endorepellin. Endorepellin (150–200 nM) blocked tube formation on Matrigel (Fig. 6, A), disrupted actin cytoskeleton (data not shown), and was efficiently internalized (Fig. 6, F) similar to what was observed when endorepellin was added to HUVEC cultures (Fig. 6, C and D). To further verify endorepellin binding and internalization in HUVECs and tumor-derived endothelial cells, we generated a functional chimera containing endorepellin fused to the fluorescent EYFP moiety. As predicted, the EYFP-endorepellin migrated as an approximately 110-kDa band in SDS–PAGE and could be detected by fluorescence microscopy (Fig. 6, G). EYFP-endorepellin was fully active on HUVECs and tumor endothelial cell, inducing rapid depolymerization of actin stress fibers like wild-type endorepellin (Fig. 6, H). Therefore, we tested EYFP-endorepellin for binding and internalization in tumor endothelial cells. EYFP-endorepellin was actively internalized by the tumor endothelial cells and a functional blocking antibody, directed toward the mouse
2 integrin I domain, blocked this process (Fig. 6, I and J). Finally, to control for the possibility that endocytosis might be reduced in general by the disruption of 21 interactions with the extracellular matrix, we determined that 2-blocking antibody had no statistically significant effect (P = .985) on the scavenger receptor–mediated endothelial cell endocytosis of Dil-labeled acetylated LDL (data not shown).
These findings indicate that human endorepellin is capable of interacting with and affecting the angiogenesis of mouse tumor endothelial cells in vitro via the 21 integrin and further support the in vivo data on the spatial targeting of human endorepellin to the tumor vasculature.
|
DISCUSSION |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
In previous studies, using the yeast two-hybrid system, we demonstrated that a C-terminal fragment of perlecan specifically interacted with the C-terminus of collagen type XVIII, which harbors endostatin (17). Soon it became evident that this fragment was antiangiogenic in itself, and it was named endorepellin to signify its antiendothelial activity. Subsequently, we showed that endorepellin exerts its angiostatic function by triggering an
21 integrin-dependent signaling pathway that leads to disruption of actin cytoskeleton and focal adhesions (31,32). In the present work, we provide evidence that recombinant human endorepellin, when administered intraperitoneally, targeted the tumor vasculature after acute systemic administration in two human tumor models in mice, an orthotopic squamous cell carcinoma xenograft and a syngeneic LLC tumor. Other routes of administration and their possible effect on endorepellin tumor targeting were not evaluated in this study. After acute or chronic treatment, endorepellin deposited in the perivascular zones and homes onto the tumor endothelia. To our knowledge, among antiangiogenic substances targeting angiogenic endothelia, endorepellin is the only one that has been shown to specifically and selectively localize to the tumor vasculature (48,49). Furthermore, we demonstrated that endorepellin inhibits in vivo tumor growth by inhibiting tumor angiogenesis, regardless of the tumor cells used or the immunocompetency of the host. As a direct result of a diminished blood supply, endorepellin-treated tumors exhibited hypoxia, decreased metabolism, and decreased cell proliferation, without any overt increase in tumor cell apoptosis. Its lack of effect on the extent of apoptosis distinguishes endorepellin from other endogenous angiogenesis inhibitors, including endostatin and angiostatin, which increase the apoptotic index of tumor cells in vivo (14,53). This difference suggests that endorepellin acts in a novel fashion to inhibit tumor growth. We also provided evidence that the tumor-targeting activity of endorepellin tumor may be mediated by the tumor endothelial cell 21 integrin. Our findings using near-infrared dye–labeled endorepellin indicate that this molelcule not only targets growing tumor xenografts, but also remains associated with the tumors for extended periods. This is likely due to the presence of high-affinity (Kd = 10–20 nM) endorepellin-binding sites on endothelial cells (17,31), which could rapidly concentrate endorepellin in the tumor endothelium. Our ability to detect both intact and processed forms of endorepellin in extracts of tumor xenografts after chronic systemic treatment further supports this concept. While the tumor blood vessels are leaky and contain large pores of 300–400 nm through which endorepellin could easily reach the perivascular regions, the substantial in vivo binding of endorepellin to the luminal surface of endothelial cells that we observed 6 hours after a single systemic injection suggests that endorepellin is actively taken up and internalized by tumor endothelial cells. Furthermore, the failure of heat-inactivated endorepellin (functionally inert but otherwise similar to active endorepellin) to deposit around tumor vasculature would seem to suggest that tumor vessel hyperpermeability cannot account for endorepellin homing.
Tumor endothelial cells make a uniquely accessible target for cancer therapy, primarily because of their intimate association with growing neoplastic cells, and several peptides that specifically recognize endothelia of distinct dysplastic and cancerous tissues have been characterized (54). This and other studies have demonstrated that different tumor endothelia possess unique properties that distinguish them from each other as well as from physiologic or angiogenic endothelium. These properties can potentially be exploited to develop more selective anti-cancer drugs, which would better discriminate between transformed and normal cells within the tumor environment. One of the best characterized angiostatic peptides, endostatin, localizes to CD31-positive tumor endothelium in LLC tumor xenografts 42 hours after systemic administration, but also targets proliferating wound endothelium (48,49) with possible negative consequences for wound healing (50). In contrast, administered endorepellin neither targeted the proliferating wound endothelium nor inhibited wound healing, suggesting that it targets tumor endothelial cells in a novel, highly specific way. The specific targeting of the tumor endothelium could potentially be exploited therapeutically as an angiostatic therapy with possibly less impact on physiologic angiogenesis.
Although we have provided evidence that tumor endothelial cell 21 integrin could be an anti-tumor receptor for endorepellin, the ubiquitous expression of this integrin suggests that other as yet unidentified factors are involved in the targeting of endorepellin to the tumor vasculature. Alternatively, elevated tumor endothelial cell 21 integrin expression and/or increased affinity of this integrin for ligand due to differences in its activation state (55) or other as yet unidentified mechanisms could account for endorepellin targeting to the tumor vasculature. Experiments to address these possibilities are ongoing. Because VEGF is an important mediator of angiogenesis and is known to influence 21 integrin expression in that context (33), we examined A431 tumors that were chronically treated with endorepellin to determine whether VEGF could possibly be involved in endorepellin effects. Surprisingly, immunofluoresence analysis of endorepellin-treated tumors with an antibody to VEGF demonstrated a striking vesicular signal for VEGF (Supplementary Fig. 1, E; available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22) in CD31-negative cells that was different from that of endorepellin (Fig. 3, B) and 21 integrin (Fig. 5, C). By contrast, in control tumors, we observed faint, diffuse VEGF signal throughout the tumor stroma, without visible perivascular localization (data not shown). Whether this possible increase in perivascular VEGF signal is directly caused by endorepellin, results in increased 21 integrin expression, or is a result of the tumors attempting to escape suppression of angiogenesis need to be addressed in future experiments.
Once deposited in the tumor endothelial basement membrane, endorepellin might further disrupt this already improperly assembled structure. Consequences of this effect might include altered tumor endothelial cell adhesion, motility, and growth. Moreover, we observed that chronically administered endorepellin closely associated with endogenous tumor perlecan (data not shown), suggesting, among other possibilities, that tumor endothelial cells could similarly process endorepellin derived from both the tumor and external sources. Thus, it is likely that endorepellin acts in vivo in a dominant negative fashion, and a tumor vascular basement membrane with extra endorepellin deposited in it may not be able to support ongoing angiogenesis.
Our study has some limitations that will require further investigation. First, the data were obtained from intraperitoneal administration of endorepellin. Therefore, whether the in vivo tumor-targeting and antiangiogenic efficacy of endorepellin is dependent on the route of administration is unknown. Second, we have not performed dose–response studies which might determine the therapeutic concentration range of endorepellin. Third, our study provides supportive but not conclusive evidence that the tumor endothelial 21 integrin is the antitumor endorepellin receptor. Fourth, it is unclear from our 2- to 3-week in vivo experiments whether endorepellin might suppress tumor growth indefinitely or whether the tumors would eventually escape suppression of angiogenesis. This issue is complicated by the possibility that immunocompetent mice, such as C57BL/6, might mount an immune response to administered human recombinant endorepellin with prolonged exposure, making longer term experiments in these animals difficult or impossible to perform. For this reason, the generation and testing of murine endorepellin may be necessary. Finally, our study is limited to two different carcinoma cell lines; future studies will need to determine whether endorepellin is effective against other carcinoma and sarcoma cell lines.
Despite the need for further investigation for endorepellin's mechanism of action in vivo, endorepellin's ability to localize to the tumor vasculature and its seemingly selective interference with tumor capillarization represent a novel, additional tool that could be added to the anticancer therapeutic armamentarium.
|
NOTES |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
This work was supported in part by National Institutes of Health (NIH) grants RO1 CA39481 and RO1 CA47282 and a grant from the Commonwealth of Pennsylvania (to R. V. Iozzo). G. Bix and J. J. Zoeller were supported by NIH National Research Service Award grant T32 AA07463. R. Castello was supported by a Fellowship EX2005-0559 from the Spanish Ministry of Education and Science. M. Weech was supported by NIH NRSA training grant T32 CA09678. C. A. Barker was supported by the Howard Hughes Medical Institute–NIH Research Program. The funding agencies had no role in the design of this study, data collection, analysis and interpretation of the results, or the writing of the manuscript.
We would like to thank Angela McQuillan and Shelly Campbell for expert technical assistance and Dr Terry Hyslop for help with the statistical analysis.
|
REFERENCES |
|---|
|
TopNotesAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
(1) Ruoslahti E. Specialization of tumour vasculature. Nat Rev Cancer 2002;2:83–90.[CrossRef][ISI][Medline]
(2) Tozer GM, Kanthou C, Baguley BC. Disrupting tumour blood vessels. Nat Rev Cancer 2005;5:423–35.[CrossRef][ISI][Medline]
(3) Neri D, Bicknell R. Tumor vascular targeting. Nat Rev Cancer 2005;5:436–46.[CrossRef][ISI][Medline]
(4) Nyberg P, Xie L, Kalluri R. Endogenous inhibitors of angiogenesis. Cancer Res 2005;65:3967–79.
(5) Simons M. Angiogenesis: where do we stand now? Circulation 2005;111:1556–66.
(6) Sage EH. Pieces of eight-bioactive fragments of extracellular proteins as regulators of angiogenesis. Trends Cell Biol 1997;7:182–6.[CrossRef][Medline]
(7) Iozzo RV, San Antonio JD. Heparan sulfate proteoglycans: heavy hitters in the angiogenesis arena. J Clin Invest 2001;108:349–55.[CrossRef][ISI][Medline]
(8) Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996;86:353–64.[CrossRef][ISI][Medline]
(9) Folkman J. Role of angiogenesis in tumor growth and metastasis. Semin Oncol 2002;29:15–8.[ISI][Medline]