Human Endothelial Precursor Cells Express Tumor Endothelial Marker 1/Endosialin/CD248

2536

Human endothelial precursor cells express tumor endothelial marker 1/endosialin/CD248

Rebecca G. Bagley,1 Cecile Rouleau,1 that the angiogenesis process in cancer can involve EPC. Thia St. Martin,1 Paula Boutin,1 William Weber,1 [Mol Cancer Ther 2008;7(8):2536–46] Melanie Ruzek,1 Nakayuki Honma,2 Mariana Nacht,1 Srinivas Shankara,1 Introduction Shiro Kataoka,2 Isao Ishida,2 Bruce L. Roberts,1 Angiogenesis is required for tumor growth and physiologic and BeverlyA. Teicher 1 studies showed that tumor vasculature is distinct from normal vasculature. For example, blood flow in tumors is 1Genzyme Corporation, Framingham, Massachusetts and 2 irregular with circulation occurring at lower than normal Kirin Pharma Co., Ltd., Gunma, Japan rates in vessels that can be corkscrew-like or tortuous in shape resulting in high interstitial pressure (1). Mosaicism Abstract in tumor vasculature results in greater permeability or ‘‘leakiness’’ than normal blood vessels (2). Angiogenesis occurs during normal physiologic processes Serial analysis of gene expression (SAGE)analysis as well as under pathologic conditions such as tumor comparing the gene expression profiles between endothe- growth. Serial analysis of gene expression profiling lial cells isolated from normal tissues and from tumors revealed genes [tumor endothelial markers (TEM)] that enables the identification of angiogenesis-related genes up- are overexpressed in tumor endothelial cells compared regulated in malignant disease. St. Croix et al. reported a with normal adult endothelial cells. Because blood vessel series of up-regulated genes, tumor endothelial markers development of malignant tumors under certain condi- (TEM), differentially expressed z10-fold by tumor endo- tions may include endothelial precursor cells (EPC) thelial cells versus normal endothelial cells (3). Endosialin, recruited from bone marrow, we investigated TEM or TEM1/CD248, expression was confined to the smaller expression in EPC. The expression of TEM1 or endosialin blood vessels in many human tumors including sarcomas, (CD248) and other TEM has been discovered in a carcinomas, and neuroectodermal tumors (4), suggesting population of vascular endothelial growth factor receptor À involvement in the earlier stages of angiogenesis. 2+/CD31+/CD45 /VE-cadherin+ EPC derived from hu- Endosialin is a type I cell surface glycoprotein of 757 man CD133+/CD34+ cells. EPC share some properties amino acids with a predicted molecular mass of 80.9 kDa with fully differentiated endothelial cells from normal and one transmembrane domain (5). Endosialin interacts tissue, yet reverse transcription-PCR and flowcytometry with extracellular matrix proteins to mediate cell adhesion reveal that EPC express higher levels of endosialin at the and migration (6). Endosialin has also been detected in molecular and protein levels. The elevated expression of human melanomas, squamous cell carcinomas, brain endosialin in EPC versus mature endothelial cells sug- tumors, and colorectal cancer (7–9). An endosialin knock- gests that endosialin is involved in the earlier stages of out mouse was fertile with normal body weight, vascula- tumor angiogenesis. Anti-endosialin antibodies inhibited ture, and wound healing capability (10). However, when EPC migration and tube formation in vitro. In vivo, human HCT116 colon carcinoma was implanted orthotopi- immunohistochemistry indicated that human EPC contin- cally into nude endosialin knockout mice, the tumor take ued to express endosialin protein in a Matrigel plug rate was 33% versus 90% in normal nude mice. HCT116 angiogenesis assay established in nude mice. Anti- tumors grew more slowly in the knockout mice with fewer endosialin antibodies delivered systemically at 25 mg/kg liver metastases, indicating that endosialin can influence were also able to inhibit circulating murine EPC in nude tumor vascular development. mice bearing s.c. SKNAS tumors. EPC and bone Vasculature formation within tumors can arise from the marrow–derived cells have been shown previously to cooption of existing blood vessels in close proximity to the incorporate into malignant blood vessels in some instan- site of malignancy. However, cells derived from the bone ces, yet they remain controversial in the field. The data marrow can also home to tumorigenic sites in response to presented here on endothelial genes that are up-regulated proangiogenic factors released by cancer cells and stroma. in tumor vasculature and in EPC support the hypothesis Bone marrow–derived endothelial precursor cells (EPC) have been identified in the tumor vasculature of some preclinical models (11–13). Importantly, EPC have recently been shown to control the angiogenic switch in mouse lung Received 1/15/08; revised 4/23/08; accepted 4/23/08. metastasis (14). In humans, EPC have been identified in Requests for reprints: Rebecca G. Bagley, Genzyme Corporation, 49 New York Avenue, Framingham, MA 01701-9322. Phone: 508-270-2455; patients with multiple myeloma, astrocytomas, inflamma- Fax: 508-271-4796. E-mail: [email protected] tory breast cancer, and non-small cell lung cancer (15–18). Copyright C 2008 American Association for Cancer Research. Because endosialin has been implicated in tumor angio- doi:10.1158/1535-7163.MCT-08-0050 genesis and EPC can incorporate into tumor vasculature,

Mol Cancer Ther 2008;7(8). August 2008

we investigated the expression of endosialin, TEM, and Reverse transcription was done using the cDNA archive anti-endosialin antibodies in EPC. The molecular markers kit (Applied Biosystems). Real-time PCR analysis was done that denote an EPC population remain controversial in the using the fluorogenic 5¶-nuclease assay in the ABI 7700 field. Progenitor cells in bone marrow or in circulation that sequence detector (Taqman). Final concentrations of each express the markers CD34 and CD133 identify a subpop- duplicate multiplex reaction were 900 nmol/L primer, ulation of stem cells that are capable of differentiation into 200 nm of probe, 50 nmol/L 18S primers, and 200 nmol/L more mature endothelial cells on stimulation with angio- 18S probe. Cycle thresholds were converted to transcript genic factors (19). The CD133, or AC133/Prominin-1 copy numbers using a cDNA standard curve. Even loading epitope, is a distinctive marker for EPC (20). The EPC that of the samples was verified by measurement of 18S levels. are presented here are derived from CD34+/CD133+ Reverse transcription-PCR for TEM6 was done with SYBR human bone marrow progenitor cells and express vascular Green PCR Master Mix. PCR were repeated at least once. endothelial growth factor receptor 2 (VEGFR2), CD31, and endosialin forward primer: GCAAGTGGCGAG- VE-cadherin. EPC express higher levels of endosialin CACCGCTGGCT, endosialin reverse primer: GGCAGGC- mRNA and protein compared with mature endothelial GCCCTCGAAGCCA, and endosialin probe: CGCTGGCT- cells such as human microvascular endothelial cells GTCGACGGCTACCTGTGCCAGTT. (HMVEC)and human umbilical vein endothelial cells TEM3 forward primer: TACCAACATTTCGGCTGTGG, (HUVEC). Endosialin continues to be expressed in vivo in TEM3 reverse primer: GATACACAGGGGCCACATCT, EPC in a Matrigel angiogenesis assay. EPC exposed to an and TEM3 probe: TGACCCCATTACCCACATGCCTC- antibody to endosialin were inhibited in vitro in migration CAGTTT. and tube formation assays. An antibody against endosialin TEM5 forward primer: AAGGCCTGCAGCCGCATCGT, was also effective at reducing circulating murine EPC in the TEM5 reverse primer: GCAGGTCAGGCCCACGTAGCTG, SKNAS neuroblastoma xenograft model in nude mice. and TEM5 probe: CACGTTCCTCGCATTCACTGA- GATGTGCTGG. Materials and Methods TEM6 forward primer: ACCCGTGACGTCATTTTC and Cell Culture TEM6 reverse primer: TGTACTTGCTTCGAGCATC. + + TEM7 forward primer: CACCATGACGGCCGCATTGT, CD34 /CD133 progenitor cells from human bone TEM7 reverse primer: AGGCCGGTTTTGACAGGATGC, marrow of healthy volunteers, HMVEC, and HUVEC + + and TEM7 probe: CTTTGCCTATAAAGAGATCCC- were obtained (Lonza). The CD34 /CD133 cells TATGTCTGTCCCGG. (1 Â 105-2 Â 105 cells/mL)were grown in Iscove’s modified ANTXR1/TEM8 forward primer: CCGGAGCAGGAA- Dulbecco’s medium (IMDM; Lonza), 15% fetal bovine TATGAATT, ANTXR1/TEM8 reverse primer: GACCCA- serum (FBS; Invitrogen), 50 ng/mL VEGF (R&D Sys- 165 CAAGGCATCGA, and ANTXR1/TEM8 probe: tems), 25 ng/mL basic fibroblast growth factor (bFGF; R&D TCCCCCCGGAAGTGGTACTC. Systems), and 5 units/mL heparin (Sigma) on fibronectin or TEM9 forward primer: GCAAGGACAAGAAGTG- native collagen I–coated flasks (BD Biosciences)to generate CGTGT, TEM9 reverse primer: GATGTAGGGCAA- adherent EPC (21, 22). After two passages, the EPC were GGCTGTCA, and TEM9 probe: CAGTCCTCTCGGCTG- maintained in IMDM/10% FBS without growth factors. GCTGTC. EPC were exposed to 10 Ag/mL Ac-Dil-LDL (Biomedical Technologies)in IMDM/10% FBS for 4 h. EPC were Generation of Endosialin Antibodies washed twice and imaged live by fluorescent microscopy. The rabbit polyclonal antibody was generated from HMVEC and HUVEC were grown in endothelial cell New Zealand white rabbits immunized with pDNA basal medium/2% FBS and supplements (Lonza). SKOV3 encoding the extracellular portion of human endosialin. and SKNAS cells (American Type Culture Collection)were Rabbits were immunized using the gene gun method and grown in RPMI 1640/10% FBS. Cultures were propagated blood was collected before and after immunization. Anti- j at 37 C with humidified 95% air/5% CO2. endosialin rabbit IgG were purified from serum by SAGE Libraries affinity chromatography using HiTrap Protein G HP SAGE libraries were constructed as described previously columns (GE Healthcare Life Sciences)as described by (3, 21, 23). EPC were cultured in the absence or presence of the manufacturer. Immune serum (10 mL)was mixed VEGF (150 ng/mL)for 72 h before RNA extraction. with 20 mL binding buffer (20 mmol/L sodium phos- mRNA Expression by Real-time PCR phate, pH 7.0)and passed through a 0.45 Am filter. Serum EPC were harvested in Trizol reagent (1 mL)plus was applied over an equilibrated HiTrap Protein G HP chloroform (200 AL). Samples were incubated at room column. The bound antibody was eluted with 2 to 5 temperature for 3 min and centrifuged at 3,000 rpm for column volumes of elution buffer (0.1 mol/L glycine-HCl, 15 min at 4jC. The aqueous layer was extracted and mixed pH 2.7). Fractions (1 mL) were collected and neutralized 1:1 with 70% ethanol. Samples were passed through a with 60 AL of 1 mol/l Tris-HCl (pH 9.0). The fractions A Qiagen column and prepared according to the manufac- with purified antibody (as measured by 260)were turer’s instructions. RNA integrity and purity was checked pooled and buffer exchanged using PD-10 desalting A A by gel electrophoresis and 260/ 280 measurements. columns into PBS.

Mol Cancer Ther 2008;7(8). August 2008

The monoclonal anti-human endosialin antibody is a (25). The cDNA of murine endosialin was cloned into an fully human monoclonal antibody isolated after inoculation adenovirus shuttle vector DC67 by way of EcoRI and XhoI. of the KM mouse (24). The control antibody is a fully The DC67 construct was digested with I-CEU to isolate the human antibody raised against dinitrophenol in the KM murine endosialin cDNA, cytomegalovirus promoter re- mouse. CHO cells were transfected with plasmid constructs gion, and BGH poly(A)region fragment. This fragment was expressing anti-dinitrophenol or anti-endosialin. The high- cloned into the I-CEU site of a DC71 vector. The DC71 est producing CHO clones or hybridoma cells were construct encoding murine endosialin was digested with selected. CHO cells were grown in CD-CHO medium SnbaI to generate linear DNA, which was used to transfect (Invitrogen)supplemented with 32 mmol/L arginine, 293 cells for virus production. SKOV3 cells were cultured in 64 mmol/L asparagine, 8 mmol/L cysteine, 100 g/L RPMI 1640/10% FBS for 72 h with virus. glucose, 32 mmol/L glutamine, 64 mmol/L serine, and In vitro Tu b e F o r m a t i o n 16 mmol/L tyrosine. Hybridomas were seeded in PFHM Matrigel (BD Biosciences)was added to a 48-well plate medium with 5% low IgG FBS (Invitrogen)and expanded (150 AL/well). Hybridoma-produced monoclonal anti- in serum-free PFHM. The antibodies from conditioned bodies in PBS were mixed with EPC in IMDM with medium were captured over a protein A-Sepharose column 10% FBS (final concentrations: 2% FBS, 80% PBS, and (GE Healthcare)equilibrated with 50 mmol/L sodium 18% IMDM). EPC (2.5 Â 103)and antibodies were phosphate, 25 mmol/L NaCl (pH 7.1), loaded, and washed preincubated for 45 min before seeding on Matrigel in first with 50 mmol/L sodium phosphate, 25 mmol/L NaCl 300 AL. The antibody concentration was 120 Ag/mL. (pH 7.1)and then with 10 mmol/L sodium succinate Inhibition of tube formation was observed in six (pH 6.0). The column was eluted with 10 mmol/L sodium independent experiments. HMVEC and HUVEC were succinate (pH 3.75). The eluates with antibodies were conducted in endothelial cell basal medium/2% FBS. The sterile filtered through a 0.22 Am PES membrane (What- tube networks were stained with calcein (4 Ag/mL)for man), formulated in 20 mmol/L histidine buffer (pH 6.0) 30 min at 37jC and quantified by image analysis using and 0.005% PS80, and concentrated. Scion image (NIH)as fluorescent pixel area. Samples Immunostaining were assayed in triplicate. EPC were cultured in IMDM/10% FBS. EPC were Migration Assay washed twice with PBS, incubated with rabbit postimmune EPC (5 Â 104 cells)were placed into the upper chamber or preimmune serum on ice for 1 h, and washed twice. of a FluoroBlok insert with 8 Am pores (BD Biosciences)in Secondary donkey anti-rabbit FITC antibody (1:200; Jack- serum-free IMDM (300 AL)for 48 h. The inserts were son Immunochemicals)was added and cells were incubat- placed into the lower chamber wells of a 24-well plate ed for 30 min on ice, washed twice, and imaged by containing IMDM/0.5% FBS as chemoattractant. The fluorescent microscopy. rabbit polyclonal anti-endosialin antibody or rabbit anti- Flow Cytometry body control was added to both the insert and lower well EPC, HMVEC, and HUVEC were collected by brief to a final concentration of 800 Ag/mL. Because the rabbit exposure to 0.25% tryspin/1 mmol/L EDTA (Invitrogen), anti-endosialin is purified over a column that binds IgG washed twice with cold PBS/5% FBS buffer, resuspended and is not specific for anti-endosialin, this concentration in buffer (1 Â 105-2 Â 105 cells/100 AL), incubated with reflects the total amount of IgG in the prep and not only the primary antibody for 1 h on ice, washed twice, anti-endosialin. After 48 h, EPC in the lower chamber incubated with the secondary antibody for 45 min on ice, were stained with calcein (4 Ag/mL; Molecular Probes)in washed twice, and resuspended (500 AL). For endosialin PBS for 30 min at 37jC and fluorescence was quantified staining or control, the rabbit polyclonal IgG (20 Ag/mL) using a fluorometer set at 485 nm excitation and 530 nm was used. Two micrograms of the following antibodies emission. Data are expressed as number of cells that were applied: anti-TEM2, anti-TEM4, anti-TEM5 (ProSci), migrated F SD (n = 3). Before the migration assay, EPC anti-TEM7, and anti-TEM-8 (Abcam). Two micrograms of were preincubated with the antibodies (800 Ag/mL) antibodies were used for EPC characterization: anti-CD31, overnight in serum-free medium. Migration assays have anti-CD45, anti-VE-cadherin (Abcam), and anti-VEGFR2 been done six times with inhibitory effects observed. (Santa Cruz Biotechnology). The antibody isotype con- In vivo Matrigel Assay trols were mouse IgG (BD Biosciences)and rabbit IgG EPC were labeled with 4¶,6-diamidino-2-phenylindole (Abcam). The secondary antibody was PE-labeled goat (20 Ag/mL; Sigma)at 37j C for 20 min and washed anti-rabbit IgG (1:50; Jackson Immunoresearch Labs). twice with PBS. EPC (5 Â 105)in PBS (100 AL)were mixed Experiments were repeated at least twice to confirm with Matrigel (500 AL; BD Biosciences)containing results. 40 units/mL heparin and 150 ng/mL bFGF. The Matrigel SKOV3 human ovarian carcinoma cells (American Type containing EPC (500 AL)were implanted s.c. into the mid- Culture Collection)were infected with 100 multiplicities of dorsal region of female BALB/c nude mice (CAnN.Cg- nu infection of an adenovirus vector expressing murine endo- Foxn1 /Crl, Charles River Laboratories). The Matrigel sialin. Others have found 77.5% identity between human plugs containing EPC were collected after 7 days and snap and murine endosialin and 100% identity between the frozen in OCT medium. For detection of human endo- transmembrane portion of the human and murine proteins sialin, sections were fixed for 10 min in zinc-formaldehyde

Mol Cancer Ther 2008;7(8). August 2008

(EMS), blocked with 10% goat serum, and incubated at CD45, a marker of hematopoetic cells (Fig. 1A). EPC were room temperature with 10 Ag/mL biotinylated monoclonal capable of acetylated LDL uptake, an endothelial cell trait. anti-endosialin from hybridomas. Samples were washed After confirmation that EPC presented the expected with TBS, incubated for 30 min at room temperature with endothelial cell phenotype, EPC were subsequently evalu- streptavidin-Cy3 (Jackson Immunoresearch), washed, ated for the expression of markers up-regulated by tumor mounted with Antifade mounting medium, and imaged endothelial cells, TEM. using Metamorph acquisition software (Molecular Devi- Preliminary data linking EPC and endosialin became ces). Experiments were done twice with consistent evident on SAGE analysis. Two SAGE libraries from EPC results. maintained under either nonstimulatory or stimulatory In vivo SKNAS/c-EPC Model (VEGF)conditions resulted in the identification of 18,710 The SKNAS cells were collected when in exponential and 20,277 unique tags, respectively (Fig. 1B). Thirty of the growth phase and 5 Â 106 cells were injected in 1:1 RPMI 46 TEM identified in colon tumor endothelium were 1640/Matrigel (100 AL)in the flank region of female nude present in the EPC libraries. Endosialin (TEM1)mRNA mice age 6 to 8 weeks (CD1 nu/nu,CharlesRiver was one of 23 of 46 TEM present in both libraries (3). Of the Laboratories)on day 0. In a single experiment, CHO- remaining SAGE tags, approximately one-third were found derived monoclonal anti-endosialin or control antibodies in both VEGF-stimulated and unstimulated libraries, (25 mg/kg)in physiologic saline were administered i.v. via highlighting the significant effect of VEGF exposure on the tail vein on days 7 and 9 when the tumors were 80 to gene expression in EPC. Sixteen of the 46 TEM identified in 100 mm3. On day 11, blood was collected by cardiac colon tumor endothelium were not found in the EPC puncture into EDTA-containing tubes from three mice per libraries. These markers may reflect either the differences group. Cells per milliliter of blood were counted using a Vi- between cell culture and in vivo gene expression of EPC. Cell automated cell counter (Beckman Coulter). For blood Alternatively, TEM not reflected in EPC but present in leukocyte counting, 20 AL whole mouse blood was tumor vasculature may reflect coisolation of pericytes or combined with 480 AL Versalyse (Beckman Coulter)and represent endothelial cells derived from nearby host blood live blood leukocytes were enumerated using a Vi-Cell vessels in the original study. automated cell counter (Beckman Coulter). Lysis of RBC in The more primitive starting population of CD133+/ the remainder of the whole blood sample was done by CD34+ bone marrow progenitor cells express nondetectable incubating the sample with at least a 5-fold excess RBC levels of endosialin mRNA. Nonadherent bone marrow lysing solution (Sigma)for 20 min at room temperature. progenitor cells exposed to VEGF/bFGF in fibronectin or Remaining blood cells were centrifuged and washed before collagen-coated plates also did not express endosialin addition of antibodies. Antibodies used for endothelial cell mRNA. However, the adherent subpopulation of cells from detection included FITC anti-CD13, PE anti-VEGFR2, these cultures after 3 to 4 weeks in culture with continuous PerCP or PerCP-Cy7 anti-CD45, and APC anti-CD117 (BD exposure to VEGF/bFGF on either substratum express Biosciences). The anti-CD117 antibody was used to distin- endosialin mRNA (Fig. 2A). Studies looking into the effects guish EPC from endothelial cells. Additional markers of VEGF/bFGF/heparin on endosialin mRNA expression reported to be expressed by endothelial cells were also in propagated cultures EPC indicated that the presence or examined including PE anti-CD202 (Tie-2; EBiosciences), absence of these growth factors did not have a significant anti-CD144 (VE-cadherin)with secondary PE anti-rat IgG effect once the CD133+/CD34+ bone marrow cells had (BD Biosciences), FITC anti-CD201 (Stem Cell Technolo- differentiated into EPC (data not shown). gies), and PE anti-CD61 (BD Biosciences). Propidium The levels of endosialin mRNA in EPC were compared iodide (BD Biosciences)excluded dead cells. Cell staining with the expression of several other TEM by these cells for specific markers was determined by flow cytometric (Fig. 2B). The mRNA for ANTXR1/TEM8 and TEM9 were analysis (FACSCalibur and FACSCanto, BD Biosciences). expressed at levels higher than endosialin mRNA, whereas Given the low volumes of blood available and the high TEM3, TEM5, and TEM6 were expressed at lower levels number of cells required for this analysis, isotype controls and TEM7 mRNA was barely detectable. Expression of for were done for VEGFR2 and CD117 but not for CD45 endosialin by EPC and HMVEC were compared by SAGE and CD13 on samples from this experiment. These controls analysis. Levels of endosialin mRNA were found to be have been done in previous experiments and showed 10-fold higher in EPC (10-14 copies)than in HMVEC minimal background staining. (1 copy), consistent with the notion that a mature endothelial cell from normal vasculature would not express a TEM (data not shown). By reverse transcription-PCR Results analysis, although there is some heterogeneity, endosialin There is no single molecular marker that denotes an EPC, mRNA levels in EPC from three individual donors are but rather a panel of markers serves to distinguish EPC and significantly higher than endosialin mRNA levels in confirm differentiation along an endothelial cell lineage. HMVEC (P < 0.05; Fig. 2C). The comparison suggests that EPC were derived from CD34+/CD133+ bone marrow cells EPC are more representative of tumor endothelium than and on exposure to VEGF/bFGF/heparin up-regulated the HMVEC and that EPC may contribute to early events of expression of CD31, VEGFR2, and VE-cadherin but not tumor angiogenesis.

Mol Cancer Ther 2008;7(8). August 2008

Figure 1. A, flow cytometry profiles reveal that EPCexpress CD31, VEGFR2, and VE-cadherin but not CD45. Solid profiles represent isotype antibody controls. A light microscopy image shows EPCmorphology (magnification, Â10). EPCare capable of Ac- Dil-LDL uptake (magnification, Â20). B, Venn diagram showing overlaps in gene expression by SAGE. EPCwere maintained in basal medium or stimulated with VEGF (150 ng/mL) for 72 h. A total of 32,043 unique tags were identified in the EPCpopulations. Of the 46 TEM identified from colon tumor endothelium (3), 30 were expressed by EPC.

To confirm endosialin protein expression, EPC were endosialin and a fluorescently labeled secondary antibody analyzed using a rabbit polyclonal IgG to endosialin. For (Fig. 3C). Endosialin protein is localized to the cell surface comparison, endosialin protein expression was also as expected (4). The antibody generated to endosialin had investigated in HMVEC and HUVEC. Flow cytometry great affinity for EPC and therefore was tested in analysis indicated a higher level of expression of endo- functional assays. sialin in EPC than in HMVEC or HUVEC (Fig. 3A). Thus, EPC perform well in assays in culture used to model the protein expression findings were consistent with neovascularization including tube formation and migra- SAGE and reverse transcription-PCR results obtained with tion. These cellular functions can be quantified in culture as mRNA from EPC and HVMEC. Because EPC are a more a measure of potential angiogenic or antiangiogenic immature progenitor cell, endosialin expression may be activity. Migration of endothelial cells through tissues is necessary to guide differentiation and subsequent neo- necessary in homing to areas of angiogenic stimuli. vascularization under angiogenic conditions as would be Disruption of this process may retard blood vessel expected with tumor growth. Endosialin expression has development during tumor neovascularization. EPC were not been observed in resting or activated HUVEC but was placed into upper chambers with porous membranes that overexpressed in HUVEC that were immortalized (26). allow migration of cells toward a chemoattractant, serum, The protein expression of other TEM was investigated in in the lower chamber. The addition of the rabbit polyclonal EPC by flow cytometry (Fig. 3B). Cell surface expression anti-endosialin antibody to both upper and lower chambers of TEM2, TEM4, TEM5, and TEM8 were detected with ensured that EPC were continuously exposed to anti- greater levels of TEM8. There was no protein expression endosialin. Exposure to anti-endosialin inhibited EPC of TEM7. Immunostaining of endosialin in adherent EPC activity in the migration assay by f65% (P = 0.009; was carried out using a rabbit polyclonal antibody to Fig. 4A).

Mol Cancer Ther 2008;7(8). August 2008

EPC form networks or tubes similar to those of HMVEC von Willebrand factor and CD31 (21). The alignment of and HUVEC on Matrigel in culture (21). When exposed to a EPC under the proangiogenic conditions in the Matrigel hybridoma-produced monoclonal anti-endosialin antibody, plug is indicative of the potential of EPC to contribute to EPC were incapable of generating capillary-like structures neovessel formation. compared with control following an overnight incubation The ability of CHO-produced monoclonal anti-endosialin (Fig. 4B and C). EPC were preincubated with the antibodies to influence EPC in vivo was investigated in the SKNAS before being seeded on Matrigel to allow for sufficient neuroblastoma xenograft model. Binding of the anti- binding before initiation of the tube formation process and endosialin antibody to the murine homologue of endosialin adherence of the cells to the matrix. Image analysis was confirmed using SKOV3 ovarian carcinoma cells software was used to measure the length of the tubes and infected with an adenovirus expressing the murine homo- provide a quantitative readout. The area of EPC tube logue of endosialin (Fig. 6A). S.c. SKNAS xenograft tumors formation was significantly decreased by f85% when the were 200 to 300 mm3 in volume when blood samples were EPC were exposed to anti-endosialin (P < 0.05). By collected. The antigen CD117 was used to distinguish EPC comparison, the anti-endosialin did not inhibit HMVEC from endothelial cells (Fig. 6B). Circulating levels of host or HUVEC tube formation. endothelial cells and EPC were confirmed to be higher in When EPC were incorporated into a Matrigel plug tumor-bearing mice than in naive mice by four-color flow angiogenesis assay in immunodeficient mice, EPC contin- cytometry (Fig. 6C). I.v. administration of anti-endosialin to ued to express endosialin (Fig. 5). EPC were prelabeled the tumor-bearing mice significantly decreased the number with the nuclear stain 4¶,6-diamidino-2-phenylindole to of circulating EPC (c-EPC)compared with mice that distinguish between human EPC and invading host cells. In received a control antibody. Thus, targeting endosialin in this model system, EPC express the cell surface proteins, a model system of cancer can lead to a reduction of a

Figure 2. A, endosialin mRNA expression was determined by reverse transcription-PCR in unstimulated (basal medium + serum) or stimulated (medium + serum with VEGF, bFGF, and heparin) CD133+/CD34+ bone marrow progenitor cells and differentiated EPC. Endosialin expression increased as cells differentiated from bone marrow progenitor cells to adherent EPC( P < 0.05). B, when the mRNA expression of endosialin/TEM1 and other TEM were quantified TEM1, TEM8, and TEM9 were most highly expressed followed by TEM3, TEM5 and TEM6; TEM7 was detected at negligible levels. C, when EPC and HMVECwere analyzed for endosialin mRNA, endosialin was more highly expressed by EPCthan by HMVEC( P < 0.05). Data are expressed as number of transcripts per reaction, normalized to 18S.

Mol Cancer Ther 2008;7(8). August 2008

Figure 3. A, different endothelial cell types were immunofluorescently stained for endosialin protein expression and analyzed by flow cytometry using the rabbit polyclonal anti-endosialin antibody. Endosialin protein is more highly expressed in EPCthan in either HMVECor HUVEC. B, expression of other TEM by EPCrevealed that EPCalso express TEM2, TEM4, TEM5, and TEM8 but not TEM7. C, EPCexpressing endosialin were stained with a rabbit polyclonal anti-endosialin. Cell surface labeling of endosialin on EPCis evident compared with EPCincubated with preimmune rabbit serum.

precursor cell population, EPC, in circulation. Although the multiple myeloma and non-small cell lung cancer levels of circulating endothelial cells (c-EC)appeared to be (17, 18). Although endosialin mRNA and protein are diminished by the anti-endosialin as well, this result was detectable at low levels in mature endothelial cells not statistically significant. However, there was no reduc- such as HMVEC and HUVEC when grown in culture, tion in tumor volume in mice that received the anti- the induction of endosialin expression in EPC and the endosialin antibodies, suggesting that EPC are not critical absence of endosialin in the starting population of bone to the development of blood vessels in SKNAS s.c. tumors marrow progenitor cells suggest that endosialin is and do not substantially promote the growth of primary involved in early stages of endothelial cell differentiation tumors in this particular model. The lack of efficacy may in response to proangiogenic conditions. EPC have value also be attributed to little detection by immunohistochem- as a better model of tumor endothelium than are ical methods of endosialin in the established vasculature of HMVEC or HUVEC because EPC more readily express SKNAS tumors (data not shown)despite positive expres- endosialin that is overexpressed in the blood vessels of sion in the immature EPC. malignant tissues and EPC can be incorporated in assays for drug development. EPC also express other TEM, particularly TEM4 and TEM8. Discussion Endosialin expression that we have discovered in EPC EPC have emerged as an intriguing component of presents value for the field of oncology because endosialin malignant neovascularization of some cancers, such as has been readily detected in human tumors of several

Mol Cancer Ther 2008;7(8). August 2008

pathologies. In primary brain tumors metastasis, endo- immunohistochemistry (4). Several recent reports have sialin colocalized with CD31+ endothelial cells and detected endosialin mRNA and/or endosialin protein in Thy-1+ or a-SMA+ cells associated with vasculature (9). various tumor settings. Examination of human breast The largest proportion of endosialin-expressing tumors cancer (n = 120)and normal breast tissues ( n = 33)found were glioblastoma multiforme, anaplastic astrocytomas, that elevated levels of endosialin were associated with and brain metastasis from carcinomas. Endosialin was either nodal involvement and/or disease progression, a expressed by melanoma, oligodendrogliomas, astrocyto- potentially prognostic value in breast cancer (29). Preclin- mas, meningiomas, and ependymomas. SAGE analysis on ical studies of breast cancer in mice have shown that bone endothelial cells derived from fresh surgical samples of marrow–derived endothelial cells can incorporate into the normal temporal lobe cortex or gliomas revealed 16 genes blood vessels (13). The EPC in this study were derived from highly up-regulated in tumor endothelial cells compared bone marrow and the link between endosialin expression in with normal endothelial cells and endosialin was among EPC and in breast cancer progression supports the these genes (23). Brain tumors are highly vascularized and hypothesis that cells from the bone marrow can contribute the overexpression of endosialin both in this disease and to tumor vasculature. in EPC support previous observations that EPC may It should be noted that the significance of EPC involve- contribute in part to the tumor vasculature due to a ment in tumor development can be a controversial topic. tropism for brain tumors (27, 28). Efforts to quantify the degree of EPC contribution to A similar correlation with disease progression and human or murine tumor endothelium have generated low endosialin expression (and TEM8, also highly expressed numbers ranging from 2% to 15% in clinical human by EPC in our study)was observed in breast cancer (29). biopsies and in some murine models (30–32). In the Most normal human tissues were negative for endosialin, experiment described here, a significant reduction in whereas 41 of 61 sarcomas, 26 of 37 carcinomas, and 18 of c-EPC did not inhibit the growth of SKNAS tumors and 25 neuroectodermal tumors were endosialin positive by this result could be attributed to several factors. For

Figure 4. A, EPCwere incorporated into a transwell migration assay with the rabbit polyclonal anti-endosialin or with the control antibody. After 48 h, EPCmigration was quantified by staining migrated cells with calcein and measuring fluorescence. EPCmigration was inhibited by the polyclonal anti - endosialin (P = 0.009). Mean F SD (n =3).B and C, effect of the hybridoma-produced monoclonal anti-endosialin antibody on EPCtube formation on Matrigel was evaluated. Tube formation was significantly decreased in the presence of the anti-endosialin compared with IgG control (P < 0.05). The final antibody concentration was 120 Ag/mL. The anti-endosialin antibody had no effect on HMVECor HUVECtube formation. Mean F SD (n = 3).

Mol Cancer Ther 2008;7(8). August 2008

example, the degree of EPC involvement of tumor antiangiogenic therapies. In concert with other molecular vasculature can vary depending on the grade of tumor, markers such as CD34, CD133, and VEGFR2, the detection tumor type, or vary between strains of mice (31, 33). of endosialin protein would be advantageous in distin- Alternatively, EPC may have greater influence on tumor guishing c-EPC from c-EC using multicolor flow cytometry. vasculature in preclinical models depending on whether In non-small cell lung cancer cases, c-EPC numbers the tumor is s.c. or orthotopic or are metastatic lesions. The were increased before treatment, correlated with poor elevated numbers of c-EPC, however, were apparent in overall survival, and were significantly reduced in tumor-bearing mice compared with naive animals. An patients who responded to therapy (18). The progression increase in c-EC was also observed in mice with SKNAS of lung metastasis by EPC in preclinical models has led to tumors that may have been shed from tumor vasculature, the proposal that selective targeting of EPC in lung cancer yet the decrease in numbers following anti-endosialin patients may be beneficial (14). Given that the data administration was not as meaningful. generated by Goa et al. indicate that the blockade of bone Several reports have described results indicating that marrow–derived EPC does not prevent the initial seeding EPC blockade can inhibit tumor growth. Continuous of metastatic lesions in the lung but rather inhibits the infusion of endostatin, a fragment of collagen XVIII, development of micrometastasis to macrometastasis, we reduced EPC mobilization from the bone marrow and predict that an anti-endosialin antibody would exert reduced or abrogated Namalwa and Granta 519 human similar effects and delay the progression of disease in lymphoma tumor growth in immunodeficient mice (34). In the lung until the malignant cells are able to compensate Id1+/-Id3-/- mutant mice, bone marrow–derived EPC by recruiting resident progenitor cells, for example. recruitment to B6RV2 lymphoma or Lewis lung tumors is The protein structure of endosialin revealed by EPC in impaired resulting in diminished tumor growth (35). A this study places it in the Group XIV thrombomodulin-like marked delay in the growth of Lewis lung tumors was also family of C-lectin domain proteins (40). However, the achieved with an anti-VE-cadherin antibody that selective- absence of expression of endosialin in mature, normal ly targeted EPC but not endothelial cells (36). Importantly, endothelium differentiates it from thrombomodulin. The this study also showed that the EPC were most crucial at function of endosialin has been implicated with attachment the earliest stages of angiogenesis. and adhesion to extracellular matrix proteins (6)and the Although opposing points of view have been expressed expression pattern of this protein in solid tumors may make on the relevance of and characteristics that define EPC, it a favorable target for cancer therapy with an antibody or c-EPC exist and were reported in several clinical settings. antibody-toxin conjugate. In an alternative strategy, a Elevated c-EPC levels correlated with increased serum single-chain antibody fragment directed toward the endo- VEGF, erythropoietin, and angiopoietin-2 in breast cancer sialin extracellular domain was used to guide a liposome- patients (37). Levels of c-EPC were increased in patients encapsulated cytotoxic agent to tumor vasculature with with gliomas versus healthy donors or those suffering favorable results (41). recent ischemic strokes (38). Higher levels of c-EPC were Using EPC with the expression of endosialin as a also detected in patients with hepatocellular carcinoma molecular marker in conjunction with other markers versus healthy controls (39). The number of endosialin- associated with EPC to distinguish these cells from other expressing cells in circulation may be a useful biomarker/ cells of the vasculature or in circulation would add surrogate marker for the development and/or use of relevance and possibly more predictable information that

Figure 5. Human EPCwere prestained with 4¶,6-diamidino-2- phenylindole, mixed with Matrigel, and injected s.c. into immunodeficient nude mice in a final volume of 0.5 mL. On day 7, Matrigel plugs were removed, flash frozen, and immunostained for endosialin/TEM1 expression. A, ahybridoma- produced anti-endosialin antibody shows that EPCcontinue to express endosialin in vivo under angiogenic conditions. EPCform linear, capillary-like structures. B, immunohistochemistry with negative control IgG primary antibody and Cy3 secondary antibody. Bar, 10 Am.

Mol Cancer Ther 2008;7(8). August 2008

Figure 6. A, ability of a CHO-produced monoclonal antibody against human endosialin/TEM1 was shown to cross-react in cells expressing the mouse homologue by flow cytometry. SKOV3 human ovarian cancer cells were infected with an adenovirus to express murine TEM1. B, examples of flow cytometric analysis of c-ECand c-EPCin mouse blood. c-ECwere defined as CD45 -/C13+/VEGFR2+. Positive expression of CD117 distinguishes c-EPC from c-EC. CD45low/-, CD13+ cells were first evaluated for VEGF-R2 expression and those positive for VEGF-R2 were considered c-EC. C-EPC were identified within this c-EC population as being CD117 (c-kit) positive. Isotype controls for VEGF-R2 and CD117 on the CD45low/-, CD13+ population are also shown. C, C-EPC and c-EC numbers were elevated in the peripheral blood of mice bearing SKNAS human neuroblastoma tumors compared with naive mice. Systemic administration of the CHO-produced anti-endosialin/TEM1 decreased c-EPC (P < 0.05) but not c-EC. Mean F SD (n = 3). would subsequently lead to a more optimal treatment endosialin and EPC hold potential value both as regime that has been lacking thus far (42). Bolus diagnostic markers and as putative targets for interven- administration of cytotoxics or vascular-disrupting agents tive therapies. can result in an acute mobilization of EPC that may be associated with disease relapse (43, 44). Close monitoring of patients’ responses to treatment by measuring c-EPC Disclosure of Potential Conflicts of Interest levels could improve patient outcome and guide the R.G. Bagley, C. Rouleau, T. St. Martin, P. Boutin, W. Weber, M. Ruzek, M. Nacht, S. Shankara, B.L. Roberts, and B.A. Teicher: Genzyme employees. development of more effective therapies. The results B.A. Teicher: Genzyme ownership interest. The other authors reported no presented in this article support the proposal that potential conflicts of interest.

Mol Cancer Ther 2008;7(8). August 2008

References 24. Tomizuka K, Shinohara T, Yoshida H, et al. Double trans-chromoso- mic mice: maintenance of two individual human chromosome fragments 1. Yuan F, Salehi HA, Boucher Y, Vasthare US, Tuma RF, Jain RK. containing Ig heavy n loci and expression of fully human antibodies. Proc Vascular permeability and microcirculation of gliomas and mammary Natl Acad Sci U S A 2000;97:722 – 7. carcinomas transplanted in rat and mouse cranial windows. Cancer Res 2004;54:4564 – 8. 25. Opavsky R, Haviernik P, Jurkovicova D, et al. Molecular characterization of the mouse Tem1/endosialin gene regulated by cell 2. McDonald DM, Baluk P. Significance of blood vessel leakiness in density in vitro and expressed in normal tissues in vivo. J Biol Chem cancer. Cancer Res 2002;62:5381 – 5. 2001;276:38795 – 807. 3. St. Croix B, Rago C, Velculescu V, et al. Genes expressed in human 26. Tentori L, Vergati M, Muzi A, et al. Generation of an immortalized tumor endothelium. Science 2002;289:1197 – 202. human endothelial cell line as a model of neovascular proliferating 4. Rettig WJ, Garin-Chesa P, Healey JH, Su SL, Jaffe EA, Old LJ. endothelial cells to assess chemosensitivity to anticancer drugs. Int J Identification of endosialin, a cell surface glycoprotein of vascular Cancer 2005;27:525 – 35. endothelial cells in human cancer. Proc Natl Acad Sci U S A 1992;89: 27. Moore XL, Lu J, Sun L, Zhu CJ, Tan P, Wong MC. Endothelial 10832 – 6. progenitor cells’ ‘‘homing’’ specificity to brain tumors. Gene Ther 2004; 5. Christian S, Ahorn H, Koehler A, et al. Molecular cloning and 11:811 – 8. characterization of endosialin, a C-type lectin-like cell surface receptor of 28. Udani V, Santarelli J, Yung Y, et al. Differential expression of tumor endothelium. J Biol Chem 2001;276:7408 – 14. angiopoietin-1 and angiopoietin-2 may enhance recruitment of bone- 6. Tomkowicz B, Rybinski K, Foley B, et al. Interaction of endosialin/TEM marrow-derived endothelial precursor cells into brain tumors. Neurol Res 1 with extracellular matrix proteins mediates cell adhesion and migration. 2005;27:801 – 6. Proc Natl Acad Sci U S A 2007;104:17965 – 70. 29. Davies G, Cunnick GH, Mansel RE, Mason MD, Jiang WG. Levels of 7. Huber MA, Kraut N, Schweifer N, et al. Expression of stromal cell expression of endothelial markers specific to tumor-associated endothelial markers in distinct compartments of human skin cancers. J Cutan Pathol cells and their correlation with prognosis in patients with breast cancer. 2006;33:145 – 55. Clin Exp Metastasis 2004;21:31 – 7. 8. Rmali KA, Puntis MCA, Jiang WG. Prognostic values of tumor 30. Bolontrade MF, Zhou R, Kleinerman ES. Vasculogenesis plays a role in endothelial markers in patients with colorectal cancer. World J Gastro- the growth of Ewing’s sarcoma in vivo. Clin Cancer Res 2002;8:3622 – 7. enterol 2005;11:1283 – 6. 31. Li H, Gerald W, Benezra R. Utilization of bone marrow-derived 9. Brady J, Neal J, Sadakar N, Gasque P. Human endosialin (tumor endothelial cell precursors in spontaneous prostate tumors varies with endothelial marker 1) is abundantly expressed in highly malignant and tumor grade. Cancer Res 2004;64:6137 – 43. invasive brain tumors. J Neuropath Exp Neurology 2004;63:1274 – 83. 32. Peters B, Diaz L, Polyak K, et al. Contribution of bone marrow-derived 10. Nanda A, Karim B, Peng Z, et al. Tumor endothelial marker 1 (TEM1) endothelial cells to human tumor vasculature. Nat Med 2005;11:261 – 2. functions in the growth and progression of abdominal tumors. Proc Natl 33. Shaked Y, Bertolini F, Man S, et al. Genetic heterogeneity of the Acad Sci U S A 2006;103:3351 – 6. vasculogenic phenotype parallels angiogenesis: implications for cellular 11. Lyden D, Hattori K, Dias S, et al. Impaired recruitment of bone surrogate marker analysis of antiangiogenesis. Cancer Cell 2005;7: marrow-derived endothelial and hematopoietic precursor cells blocks 101 – 11. tumor angiogenesis and growth. Nat Med 2001;7:1194 – 201. 34. Capillo M, Mancuso P, Gobbi A, et al. Continuous infusion of 12. Asahara T, Masuda H, Takahashi T, et al. Bone marrow origin of endostatin inhibits differentiation, mobilization, and clonogenic potential endothelial progenitor cells responsible for postnatal vasculogenesis in of endothelial cell progenitors. Clin Cancer Res 2003;9:377 – 82. physiological and pathological neovascularization. Circ Res 1999;85: 35. Lyden D, Hattori K, Dias S, et al. Impaired recruitment of bone- 221 – 8. marrow-derived endothelial and hematopoietic precursor cells blocks 13. Duda D, Cohen K, Kozin S, et al. Evidence for incorporation of bone tumor angiogenesis and growth. Nat Med 2001;7:1194 – 201. marrow-derived endothelial cells into perfused blood vessels in tumors. 36. Nolan DJ, Ciarrocchi A, Mellick AS, et al. Bone marrow-derived Blood 2006;107:2774 – 6. endothelial progenitor cells are a major determinant of nascent tumor 14. Goa D, Nolan DJ, Mellick AS, Bambino K, McDonnell K, Mittal V. neovascularization. Genes Dev 2007;21:1546 – 58. Endothelial progenitor cells control the angiogenic switch in lung mouse 37. Furstenberger G, von Moos R, Lucas R, et al. Circulating endothelial metastasis. Science 2008;319:195 – 8. cells and angiogenic serum factors during neoadjuvant chemotherapy of 15. Shirakawa K, Shibuya M, Heike Y, et al. Tumor-infiltrating endothelial primary breast cancer. Br J Cancer 2006;94:524 – 31. cells and endothelial precursor cells in inflammatory breast cancer. Int J 38. Zheng PP, Hop WC, Luider TM, Sillevis Smitt PA, Kros JM. Cancer 2002;99:344 – 51. Increased levels of circulating endothelial progenitor cells and circulating 16. Bian X, Du L, Shi J, Cheng Y, Liu F. Correlation of bFGF, FGFR-1, and endothelial nitric oxide synthase in patients with gliomas. Ann Neurol VEGF expression with vascularity and malignancy of human astrocyto- 2007;62:40 – 8. mas. Anal Quant Cytol Histol 2000;22:267 – 74. 39. Yu D, Sun X, Qiu Y, et al. Identification and clinical significance of 17. Zhang H, Vakil V, Braunstein M, et al. Circulating endothelial mobilized endothelial progenitor cells in tumor vasculogenesis of hepato- progenitor cells in multiple myeloma: implications and significance. Blood cellular carcinoma. Clin Cancer Res 2007;13:3814 – 24. 2005;105:3286 – 94. 40. Dean YD, McGreal EP, Akatsu H, Gasque P. Molecular and cellular 18. Dome B, Timar J, Dobos J, et al. Identification and clinical properties of the rat AA4 antigen, a C-type lectin-like receptor with significance of circulating endothelial progenitor cells in human non-small structural homology to thrombomodulin. J Biol Chem 2000;275: cell lung cancer. Cancer Res 2006;66:7341 – 7. 34382 – 92. 19. Salven P, Mustjoki S, Alitalo R, Alitalo K, Rafii S. VEGFR-3 and CD133 41. Marty C, Langer-Machova Z, Sigrist S, Schott, Schwendener RA, + identify a population of CD34 lymphatic/vascular endothelial precursor Ballmer-Hofer K. Isolation and characterization of a scFv antibody specific cells. Blood 2003;101:168 – 72. for tumor endothelial marker 1 (TEM1), a new reagent for targeted tumor 20. Shmelkov SV, St. Clair R, Lyden D, Rafii S. AC133/CD133/Prominin- therapy. Cancer Lett 2005;235:1 – 11. 1. Int J Biochem Cell Biol 2005;37:715 – 9. 42. Rafii S, Lyden D, Benezra R, Hattori K, Heissig B. Vascular and 21. Bagley RG, Walter-Yohrling J, Cao X, et al. Endothelial precursor cells haematopoietic stem cells: novel targets for anti-angiogeneis therapy? as a model of tumor endothelium: characterization and comparison to Nature 2002;2:826 – 35. mature endothelial cells. Cancer Res 2003;63:5866 – 73. 43. Shaked Y, Ciarrocchi A, Franco M, et al. Therapy-induced acute 22. Bagley RG, Weber W, Rouleau C, Teicher BA. Pericytes and recruitment of circulating endothelial progenitor cells to tumors. Science endothelial precursor cells: cellular interactions and contributions to 2006;313:1785 – 7. malignancy. Cancer Res 2005;65:9741 – 50. 44. Shaked Y, Kerbel RS. Antiangiogenic strategies on defense: on the 23. Madden SL, Cook BP, Nacht M, et al. Vascular gene expression in possibility of blocking rebounds by the tumor vasculature after chemo- non-neoplastic and malignant brain. Am J Pathol 2004;165:601 – 8. therapy. Cancer Res 2007;67:7055 – 8.

Mol Cancer Ther 2008;7(8). August 2008

Rebecca G. Bagley, Cecile Rouleau, Thia St. Martin, et al.

Mol Cancer Ther 2008;7:2536-2546.

Updated version Access the most recent version of this article at: http://mct.aacrjournals.org/content/7/8/2536

Cited articles This article cites 44 articles, 23 of which you can access for free at: http://mct.aacrjournals.org/content/7/8/2536.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/7/8/2536.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/7/8/2536. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.