Systemic Induction of the Angiogenesis Switch by the Tetraspanin D6.1A/CO-029

Research Article

Sabine Gesierich,1 Igor Berezovskiy,1 Eduard Ryschich,2 and Margot Zo¨ller1,3

1Department of Tumor Progression and Immune Defence, German Cancer Research Centre; 2Department of Surgery, University of Heidelberg, Heidelberg; and 3Department of Applied Genetics, Faculty of Chemistry and Bioscience, University of Karlsruhe, Karlsruhe, Germany

Abstract transcription of angiogenic factors (6, 7). Recent studies in Expression of the tetraspanin CO-029 is associated with poor knockout and transgenic mouse models have provided further prognosis in patients with gastrointestinal cancer. In a evidence that tumor angiogenesis is not only guided by the pancreatic tumor line, overexpression of the rat homologue, tumor cell itself, but is also closely tied to the tumor microenvironment (8). D6.1A, induces lethally disseminated intravascular coagulation, suggesting D6.1A engagement in angiogenesis. D6.1A- Tetraspanins are a large family of proteins grouped according to overexpressing tumor cells induce the greatest amount of structural relatedness. The key feature of tetraspanins is their angiogenesis in vivo, and tumor cells as well as exosomes potential to associate with each other and with a multitude of derived thereof strikingly increase endothelial cell branching molecules from other protein families (9–11), the most prominent in vitro. Tumor cell–derived D6.1A stimulates angiogenic partners being integrins (12). The tetraspanin, D6.1A (rat)/CO-029 a h a h factor transcription, which includes increased matrix metal- (human), associates with 3 1 and 6 1 and, after disassembly of a h loproteinase and urokinase-type plasminogen activator se- hemidesmosomes, with 6 4. It also associates with the tetraspa- cretion, pronounced vascular endothelial growth factor nins, CD9 and CD81, and the immunoglobulin superfamily a expression in fibroblasts, vascular endothelial growth factor member, prostaglandin F2 receptor regulatory protein, a type II receptor expression, and strong D6.1A up-regulation in phosphoinositide-4-kinase, EpCAM, and CD44v4-v7 (13–15). sprouting endothelium. Thus, D6.1A initiates an angiogenic According to their association with different molecules, tetra- loop that, probably due to the abundance of D6.1A in tumor- spanins are assumed to function as adaptors that assemble protein derived exosomes, reaches organs distant from the tumor. complexes in defined membrane microdomains, called tetraspanin- Most importantly, because of the strong D6.1A up-regulation enriched microdomains, that provide a link to specific signal- on sprouting capillaries, angiogenesis could be completely transducing molecules (16). This mode of activity might explain inhibited by a D6.1A-specific antibody, irrespective of whether why tetraspanins are said to take part in a wide range of diverse or not the tumor expresses D6.1A. Tetraspanins have been functions (9, 11), such as B and T cell activation, platelet aggre- suggested to be involved in morphogenesis. This is the first gation, migration, proliferation (10, 17, 18), and tumor cell prog- report that a tetraspanin, CO-029/D6.1A, promotes tumor ression (18, 19). With respect to the latter, high expression of CD9 growth by its capacity to induce systemic angiogenesis that (20, 21) and CD82 (22) is mostly associated with a favorable can effectively, and with high selectivity for sprouting prognosis. CD151, D6.1A, and its human homologue, CO-029, are endothelium, be blocked by a D6.1A-specific antibody. (Cancer supposed to exert tumor-promoting activities (23–26). A possible Res 2006; 66(14): 7083-94) mechanistic basis for the prometastatic functions of tetraspanins could be their association with certain integrins and the accom- Introduction panying increase in cell motility (22, 23). Furthermore, tetraspanin interactions with platelets and leukocytes were suggested to The term angiogenesis defines the process of new capillary provide tumor cells with a survival advantage in the hostile formation from preexisting vasculature that occurs in physiologic environment which they encounter during metastatic spread (27). as well as pathologic conditions (1). Tumor cells essentially depend CD151 has also been reputed to be involved in cellular morpho- on angiogenesis to grow beyond a threshold size of a few cubic genesis, promoting the formation of cord-like structures, consid- millimeters (2). Angiogenesis is the result of an intricate balance ered as a preform of sprouting vessels (28). Finally, tumor cells can between proangiogenic and antiangiogenic factors, and the grow in mosaics with endothelial cells, and mosaic growth depends dominance of proangiogenic factors is called the angiogenic switch on tetraspanin expression (29). Taking into account that coagul- (3). One of the main angiogenic switch inducers is hypoxia (4), opathy and thrombosis are frequent complications, particularly in whereas the most well-known proangiogenic factor is vascular patients with lung and pancreatic cancer (30), it is important to endothelial growth factor (VEGF; ref. 5), and matrix metal- note that CD9, CD63, and CD151 are expressed by platelets. CD9 loproteinases (MMP) are essentially required to create space for and CD63 associate with aIIbh3, which supports activated platelet sprouting capillaries by degradation of the basal membrane as well adhesion to neutrophils and platelet aggregation (31). In CD151 as for the liberation of angiogenic factors and for inducing the knockout mice, platelet spreading is impaired, aggregation is reduced, and the bleeding time is prolonged. This is likely due to the absence of an outside-in signal provided by CD151, leading to Requests for reprints: Margot Zo¨ller, Department of Tumor Progression and Tumor Defence, German Cancer Research Centre, Im Neuenheimer Feld, 280 D-69120 the activation of aIIbh3 (32). Heidelberg, Germany. Phone: 49-6221-422454; Fax: 49-6221-424760; E-mail: m.zoeller@ D6.1A overexpression on a pancreatic rat adenocarcinoma line is dkfz.de. I2006 American Association for Cancer Research. associated with the formation of hemorrhagic ascites (15) and can doi:10.1158/0008-5472.CAN-06-0391 induce a severe consumption coagulopathy, such that rats become www.aacrjournals.org 7083 Cancer Res 2006; 66: (14). July 15, 2006

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2006 American Association for Cancer Research. Cancer Research moribund due to disseminated intravascular coagulation rather supernatant after 2 days of culture by differential centrifugation. Super- than the tumor burden (13). These features prompted us to search natants were centrifuged twice for 10 minutes at 500 Â g, once for Â Â for the possible involvement of D6.1A in angiogenesis. 20 minutes at 2,000 g, and once for 30 minutes at 10,000 g to eliminate cell debris. The purified supernatant was centrifuged for 90 minutes at 100,000 Â g using a SW41 rotor. The pellet was resuspended in PBS and Materials and Methods centrifuged again for 90 minutes at 100,000 Â g. Rats and tumors. BDX rats, bred at the animal facilities of the German Western blotting. Cell lysates (1% Triton X-100) were resolved on 12% or Cancer Research Centre, were kept under specific pathogen–free con- 15% SDS-PAGE under nonreducing conditions. Proteins were transferred ditions, fed sterilized food and water ad libitum. Rats were used for to Hybond enhanced chemiluminescence at 30 V overnight. After blocking experiments at the age of 6 to 10 weeks. BSp73AS (AS) cells, a pancreatic (5% fat-free milk powder), immunoblotting was done with the indicated adenocarcinoma line of the BDX rat strain (33), were transfected with D6.1A antibodies, followed by horseradish peroxidase–labeled secondary anti- cDNA (AS-D6.1A; ref. 13). For intravital microscopy, AS cells were bodies. Blots were developed with the enhanced chemiluminescence transfected with the enhanced green fluorescent protein (EGFP) cDNA– detection system. containing pEGFP-N1 vector (Becton Dickinson, Heidelberg, Germany). AS- Zymography. Cells (106) were seeded in 24-well plates (35). After D6.1A cells were transfected with the EGFP cDNA inserted in the overnight culture, cells were washed and starved in serum-free medium. pcDNA3.1Hygro vector (Invitrogen, Karlsruhe, Germany). Transfected lines Conditioned medium was collected after 24 hours and was centrifuged were selected in RPMI 1640, 10% FCS, and 500 Ag/mL of G418. The selection (15 minutes, 15,000 Â g) to remove cell debris. Aliquots were incubated with medium for double-transfected lines contained an additional 50 Ag/mL of Laemmli buffer (15 minutes, 37jC) and separated in a 10% acrylamide gel hygromycin. A BDX fibroblast line, generated by NiSO4 treatment, and containing 1 mg/mL of gelatin. The gel was washed for 30 minutes each in RAEC, a Wistar rat–derived aortic endothelial cell line (Cell-lining, Berlin, 2.5% Triton X-100, 2.5% Triton X-100/50 mmol/L Tris (pH7.5), and 2.5% A Germany) was maintained in RPMI 1640 and 10% FCS. Confluent cultures Triton X-100/50 mmol/L Tris (pH7.5), 5 mmol/L CaCl 2,1 mol/L ZnCl2. were trypsinized and split. Gels were incubated (24-48 hours, 37jC) in 50 mmol/L Tris (pH7.5), A Antibodies. The following monoclonal and polyclonal antibodies were 5 mmol/L CaCl2,1 mol/L ZnCl2, and stained with Coomassie blue. used: mouse anti-D6.1A (D6.1; ref. 34), mouse anti-a6h4 (B5.5; ref. 34), rabbit Matrigel assay. Tumor cells (105) were seeded on matrigel-coated anti-CD151 (14); anti-CD9 and anti-CD44 (Ox50; European Collection of 24-well plates. Where indicated, antibodies (10 Ag/mL) were added to the Animal Cell Cultures); anti-a1, anti-a2, anti-a3, anti-a4, anti-a5, anti-a6, culture medium. Cable formation was evaluated after 7 and 48 hours and anti-h1, anti-h2, anti-h3, anti-h4, anti-CD31, anti–tumor necrosis factor-a was documented by light microscopy. (TNFa), anti-IL4 (BD PharMingen, Heidelberg, Germany); anti-TIMP-1, anti- In vitro angiogenesis. The mesentery was collected from 6-week-old TIMP-2 (Biozol, Munich, Germany); anti-MMP-2, anti-MMP-9, anti-MMP-13 rats and 1 cm2 pieces were placed in six-well plates in RPMI 1640 and 10% (Dianova, Hamburg, Germany); anti-VEGFR1, anti-VEGFR2 (Biotrend, Ko¨ln, FCS. Tumor cells (5 Â 105), supernatant of confluent tumor cell cultures Germany); anti–urokinase-type plasminogen activator receptor (uPAR), (1:1 diluted with fresh medium), exosome-depleted supernatant (1:1 diluted anti-uPA (American Diagnostica, Stanford, CT); anti-VEGF, anti–basic with fresh medium), or exosomes (equivalent to supernatant in fresh fibroblast growth factor (bFGF; R&D, Eschborn, Germany); biotinylated, medium and equilibrated for comparable amounts of CD9, CD81, and horseradish peroxidase– and fluorescence dye (FITC, APC, rhodamine, CD151) were placed on top of the mesentery. Antibodies (10 Ag/mL) were Cy2)–labeled secondary antibodies, and FITC-labeled phalloidin (BD added as indicated. Endothelial cell sprouting was microscopically PharMingen and Biotrend). evaluated. Flow cytometry. Flow cytometry followed routine procedures using 1 to In vivo angiogenesis and intravital microscopy. BDX rats received an 3 Â 105 cells per sample. Trypsinized cells were allowed to recover for i.p. injection of 5 Â 106 EGFP-transfected tumor cells and, where indicated, 2 hours at 37jC in RPMI 1640 and 10% FCS. For intracellular staining, i.p. antibody injections (200 Ag/injection, every 3rd day; ref. 36). At the cells were fixed and permeabilized in advance. Samples were analyzed with indicated time points, rats were sacrificed, the mesentery was excised and FACSCalibur (Becton Dickinson). shock-frozen for immunohistology. Alternatively, rats were anesthetized Immunohistochemistry. Cryostat sections (5 Am) of snap-frozen tissue (xylazine and ketamine) and the abdomen was opened by a midline were fixed in chloroform/acetone (1:1, 4 minutes). For intracellular staining, incision. Rats were fixed on a special plate automatically maintained at sections were fixed in paraformaldehyde (4%) and were permeabilized (0.1% 37jC. The mesentery with tumors was immobilized in 50 mL of Ringer’s Triton X-100, 4 minutes, 4jC). Tissues were incubated for 1 hour with the solution at 37jC in an immersion chamber. Microscopic images were taken first antibody, washed and exposed to the biotinylated secondary antibodies with a specialized microscope and a 40Â magnification objective using two (30 minutes), and alkaline phosphatase–conjugated avidin-biotin complex fluorescence filter blocks (I3 and N2.1; Leica GmbH, Wetzlar, Germany). (Vector Laboratories, Burlingame, CA) solutions (5-20 minutes). Sections Images were transmitted by a video camera (Kappa GmbH, Gleichen, were counterstained with Mayer’s hematoxylin. The primary antibody was Germany) to a monitor (PVM-1440M, Sony, Tokyo, Japan), and recorded replaced by normal mouse, rat, or rabbit IgG for negative controls. (sVHS, AG-7350-E, Panasonic, Osaka, Japan) for subsequent off-line analysis. For immunofluorescence microscopy, cells were seeded on matrigel- Evaluation of vessel diameter and vessel density was done with special coated cover slides. After 8 hours, cells were fixed in 4% paraformaldehyde software (CapImage, Zeintl GmbH, Heidelberg, Germany). Relative vessel (w/v in PBS) and, where indicated, were permeabilized (0.1% Triton X-100, width was defined as the ratio between vessel area to vessel-tumor contact 4 minutes). After washing and blocking (PBS/0.2% gelatin/0.5% bovine serum length. The surface of tumor blood vessels and the relative vessel width was albumin), cells were incubated with the primary antibody (2-10 Ag/mL, evaluated on 20 randomly chosen fields per rat. In vivo experiments were 60 minutes, 4jC). Slides were rinsed and incubated for 60 minutes at 4jC approved by the local animal care committee (Regierungspra¨sidium with a fluorochrome-conjugated secondary antibody. Washed cells were Karlsruhe, Karlsruhe, Germany). incubated with an excess of unlabeled mouse or rabbit IgG to block free Statistical analysis. Significance of differences was evaluated by the binding sites of the secondary antibody. Unlabeled IgG was also added Wilcoxon rank test or the two tailed Student’s t test. P values were adjusted during incubation with the second, dye-labeled antibody (60 minutes, 4jC). for multiple comparisons by the step-down Bonferroni method of Holm. Phalloidin-FITC (0.5 Ag/mL) was used for filamentous actin staining (60 minutes, 4jC). After washing thrice in PBS and once in water, slides were Results mounted in Elvanol. Digitized images were generated using a Leica DMRBE Microscope equipped with a SPOT CCD camera from Diagnostic Instru- Involvement of D6.1A in cellular morphogenesis. D6.1A- ments, Inc., and Software SPOT2.1.2 (Visitron Systems, Puchheim, Germany). overexpressing tumor cells can induce disseminated intravascular Preparation of exosomes. AS and AS-D6.1A cells were cultured in RPMI coagulation (13), and i.p.-injected D6.1A-overexpressing tumor cells 1640 and 10% exosome-depleted FCS. Exosomes were prepared from induce hemorrhagic ascites (15). Because the tetraspanin CD151

Cancer Res 2006; 66: (14). July 15, 2006 7084 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2006 American Association for Cancer Research. D6.1A and Angiogenesis has been reported to be involved in cellular morphogenesis (37) CD151. AS/AS-D6.1A cells express the a3h1 and a6h1 integrins, with anti-CD151–inhibiting cord-like structure formation of tumor where the latter has been described to account for anti-CD151– cells and fibroblasts when grown on matrigel, it became interesting inhibitable cord formation of NIH3T3 cells (37). Instead, cord to see whether D6.1A would exert similar features. formation of AS-D6.1A cells was inhibited by anti-a3 and anti-h1, When D6.1A-overexpressing AS cells were grown on matrigel, but not by anti-a6(Fig.1A). AS-D6.1A, but not AS cells, formed cord-like, anastomosing Tetraspanins exert functional activities via their associating structures. Although both lines express CD9, CD81, and CD151 at molecules, this is in line with the observed inhibition of cord a high level, cord formation could only be inhibited by the D6.1A- formation by integrin-specific as well as tetraspanin-specific specific antibody, D6.1, but not by anti-CD9, anti-CD81, or anti- antibodies. To control for this assumption, the localization of a3,

Figure 1. Cord formation of AS-D6.1A cells on matrigel. A, AS and AS-D6.1A cells were cultured for 7 and 48 hours at 37jC on matrigel-coated 24-well plates. Only AS-D6.1A cells formed cord-like structures. To control for the effect of D6.1A on cord formation, AS-D6.1A cells were cultured for 7 hours on matrigel in the presence of D6.1, anti-CD9, anti-CD81, anti-CD151, anti-a3, anti-a6, or anti-h1 (10 Ag/mL). Phase contrast microscopy (bar, 100 Am). B, AS and AS-D6.1A cells which were cultured for 7 hours on matrigel-coated cover slides were stained with anti-h1, anti-a3, anti-a6, anti-VEGF, and secondary Cy2-labeled antibody or with Phalloidin-FITC. AS-D6.1A cells were counterstained with rhodamine-labeled D6.1. Single fluorescence and merged overlays (bar, 10 Am). www.aacrjournals.org 7085 Cancer Res 2006; 66: (14). July 15, 2006

Figure 2. Angiogenesis (in vivo) and endothelial cell growth (in vitro) induction by AS-D6.1A tumor cells. A, AS and AS-D6.1A cells were injected i.p. Where indicated, rats received two i.p. injections (days 0 and 3) of 200 Ag control IgG or D6.1 or anti-CD151. After 3 and 6 days, the size of the tumor nodules, vessel density, and vessel width were evaluated by intravital microscopy. The relative size of the tumor area is presented, with the mean tumor area of AS cells taken arbitrarily as 1. Relative vessel width was calculated after 6 days as the ratio of vessel area to vessel-tumor contact length. Columns, mean values for eight rats (except for anti-CD151 treatment with two rats per group) evaluating 20 fields per rat; bars, FSD. B, tumor nodules were collected 2 weeks after i.p. injection of AS and AS-D6.1A cells. Sections of shock-frozen tissue were stained with control IgG or anti-CD31 (bar, 50 Am). C, the mesenteric sheet of untreated rats was cultured in the absence of tumor cells or was cocultured with AS or AS-D6.1A cells or supernatants. Cultures contained 10 Ag/mL control IgG, D6.1, anti-CD9, anti-CD151, or anti-a3.

a6, and h1, and colocalization with D6.1A was evaluated, Thus, D6.1A exhibited morphogenic features, which differed respectively, in AS and AS-D6.1A cells that had been grown from those described for CD151 inasmuch as cord formation was on matrigel for 7 hours (Fig. 1B). D6.1A colocalized with a3 and inhibited by D6.1 and anti-a3, but not by anti-CD151 and anti-a6. h1, but hardly with a6. There has been no pronounced coloca- Although these unexpected differences between the two tetraspa- lization of D6.1A with actin bundles. Instead, D6.1A colocalized nins are interesting and require further elucidation, we were with VEGF. primarily concerned as to whether the morphogenic features of

Cancer Res 2006; 66: (14). July 15, 2006 7086 www.aacrjournals.org

Figure 2 Continued. D, cell lysate, exosomes (100,000 Â g pellet of supernatant), cleared supernatant (10,000 Â g), and exosome-depleted supernatant (100,000 Â g)of AS and AS-D6.1A cultures were separated by SDS-PAGE and blotted with D6.1, anti-CD9, anti-CD81, anti-CD151, and anti-VEGF. Gels were loaded with 2.5 Agof exosome preparation (corresponding to 2 mL of supernatant) and 40 AL of supernatants, Western blot of supernatant preparations being exposed to X-ray films overnight. D6.1A, CD9, CD81, CD151, and VEGF bands (arrow). The mesentery of untreated rats was cultured in the presence of AS cell–derived or AS-D6.1A cell–derived unfractionated or exosome-depleted supernatants or exosomes. AS-D6.1-derived exosomes corresponded to 2 mL of supernatant. Because AS-D6.1A supernatant-derived exosomes contained a higher amount of CD9, CD81, and CD151 than AS supernatant–derived exosomes, the relative protein content was estimated by densitometry. AS supernatant–derived exosomes added to the mesentery were adjusted to contain amounts of the tetraspanins CD9, CD81, and CD151 equivalent to that in AS-D6.1 supernatant–derived exosome preparations. To control for the D6.1A specificity of endothelial cell branching, endothelial cell–exosome cocultures contained, in addition, control IgG, D6.1, anti-CD9, or anti-CD151 (10 Ag/mL). Phase contrast microscopy is shown after 72 hours of coculture (bar, 50 Am).

D6.1A might be of functional relevance in vivo. To control the significant level in either AS or AS-D6.1A tumors. The antiangio- hypothesis, we first evaluated whether D6.1A-expressing tumor genic effect of D6.1 was not a consequence of tumor growth cells support angiogenesis in vivo. inhibition. Neither D6.1 nor anti-CD151 inhibited tumor growth D6.1A induces angiogenesis that is inhibited by D6.1. EGFP- in vitro (data not shown) or in vivo (Fig. 2A and B). transfected AS and AS-D6.1A cells were injected i.p. into To confirm the involvement of D6.1A in angiogenesis, AS-D6.1A syngeneic BDX rats. After 3 and 6 days, rats were anesthetized, cells were seeded in vitro on small pieces of the mesentery. After 72 the peritoneal cavity was opened and vessel formation was eva- hours of coculture, the mesentery was filled with uncountable luated by intravital microscopy. Vessel density and vessel width small, branching endothelial strings. Significantly less branching were evaluated in at least 20 tumor fields per rat in different endothelial cells were seen in mesenteries cocultured with AS cells. parts of the mesentery. In addition, individual fields were sur- Morphogenic changes were largely absent in mesenteries cultured veyed for up to 2 hours to see whether tumor cells would infil- in the absence of tumor cells. In line with the in vitro studies on trate preexisting vessels. cable formation by the tumor cells and the in vivo analysis of There was no evidence for immigration of AS or AS-D6.1A cells angiogenesis induction, in vitro–induced endothelial cell branching into preexisting vessels (data not shown). However, both tumor was inhibited by D6.1 and anti-a3, but not by anti-CD9 and anti- lines induced angiogenesis. Three days after tumor cell injection, CD151. This accounted for the strong morphogenic alterations capillary formation was seen in small AS-D6.1A tumor nodules. induced by AS-D6.1A tumor cells and the weaker changes induced After 6 days, the relative vessel density was increased by more than by AS cells. Notably, endothelial cell branching was also induced, five times in AS-D6.1A as compared with AS tumor nodules. Unlike albeit very weakly, by AS cell supernatant, but strongly by AS-D6.1A the vessel density, the vessel width, defined by the ratio of vessel supernatant (Fig. 2C). The latter finding might be explained by the area to contact length between vessels and tumor, was only slightly fact that tumor cells shed D6.1A, which was recovered in tumor- increased in AS-D6.1A as compared with AS tumors. When rats derived exosomes. AS and AS-D6.1A supernatant contain VEGF, received the D6.1 antibody concomitantly with the tumor cells, which was recovered in the soluble fraction following ultra- angiogenesis was largely abolished in both AS-D6.1A and AS tumor centrifugation. Instead, the tetraspanins, CD9, CD81, CD151 (AS nodules. Anti-CD151, expressed in both tumor lines, served as a supernatant) plus D6.1A (AS-D6.1A supernatant), are exclusively control. Vessel density was slightly reduced by anti-CD151 treat- recovered in the exosome-containing pellet. Notably, although the ment. However, reduction in vessel density did not reach a amount of CD9, CD81, and CD151 is roughly comparable in AS and www.aacrjournals.org 7087 Cancer Res 2006; 66: (14). July 15, 2006

AS-D6.1A lysates, fewer tetraspanins are recovered in AS-derived exosome-depleted supernatant, branching was exclusively induced than in AS-D6.1A-derived exosomes. To exclude the idea that the by AS-D6.1A-derived exosomes. In line with antibody-mediated amount of tetraspanins is decisive for functional activities, the inhibition of angiogenesis in vivo, exosome-induced endothelial amount of CD9, CD81, and CD151 in AS- and AS-D6.1A-derived cell branching was inhibited by D6.1, but not by anti-CD9 and exosome preparations was estimated by densitometry of Western anti-CD151, which instead supported endothelial cell sprouting blotting and exosome preparations, which were cocultured with (Fig. 2D). the mesentery and adjusted to contain comparable amounts of the Thus, D6.1A-expressing tumor cells as well as D6.1A-containing, tetraspanins CD9, CD81, and CD151. When mesenteries were tumor-derived exosomes, strongly induced angiogenesis in vivo cultured with AS-derived or AS-D6.1A-derived exosomes or with and endothelial cell branching in vitro. Surprisingly, D6.1

Figure 3. Capillary endothelium expresses D6.1A. One week after i.p. application of AS or AS-D6.1A tumor cells, the mesentery (A) and the pancreatic gland (B) were excised, tissue was shock-frozen and 5-Am slices were stained with (A) anti-CD9, anti-CD81, anti-CD151, D6.1 (tetraspanins), anti-CD31 (endothelial cells), anti-a3, anti-a6, and anti-h3 (integrins) or (B) control IgG, anti-CD31, D6.1, and anti-a3. C, shock-frozen (5-Am slices) tissue of a human pancreatic gland, of two sections from chronic pancreatitis, and of four sections from pancreatic adenocarcinoma were stained with a control IgG or anti-CD31 and anti-CO-029 (bar, 50 Am).

Cancer Res 2006; 66: (14). July 15, 2006 7088 www.aacrjournals.org

Figure 4. VEGF and MMP expression in AS and AS-D6.1A cells. A, AS and AS-D6.1A cells were stained with integrin-, tetraspanin-, angiogenic factor-, antiangiogenic factor-, and matrix- degrading enzyme-specific antibodies. The mean (three to five separate experiments) percentage of stained cells and intensity of staining are shown. *, significant differences in the percentage of stained cells; s, significant differences in the intensity of staining. B, an example of staining with D6.1, anti-bFGF, and anti-VEGF. Single fluorescence overlays of the negative control (nonbinding control antibody; light gray area) and the stained samples (black line). C and D, AS and AS-D6.1A cells were cultured overnight in the absence of FCS. C, lysates of AS and AS-D6.1A cells as well as medium and supernatants (concentrated 20-fold) were separated by SDS-PAGE and blotted with anti-VEGF. Arrowhead, the VEGF band. D, zymography of medium and supernatants of AS and AS-D6.1A cells (concentrated 20-fold) were run on gelatin gels. Arrow, gelatin digest by MMP-2 (72 kDa) and MMP-9 (92 kDa).

completely inhibited not only AS-D6.1A-induced, but also low- esis irrespective of whether or not the tumor cells express the level AS-induced angiogenesis. The latter observation required D6.1A tetraspanin. On the other hand, D6.1A expression in further exploration. endothelial cells may further sustain angiogenesis. Angiogenesis is accompanied by D6.1A expression in the There remains the question regarding the mechanism whereby capillary endothelium. We first explored how D6.1 could inhibit AS-D6.1A cells initiate the overshooting of angiogenesis. The low-level angiogenesis induced by a D6.1A-negative tumor, and findings that few i.p. seeded AS-D6.1A cells suffice for angiogenesis- considered it likely that endothelial cells might express D6.1A induction in an adjacent, although tumor-free organ (pancreatic because low-level D6.1A expression is seen on small capillaries, gland), that AS-D6.1A-derived exosomes induce endothelial cell although not on larger veins and arteries (38). When mesenteries sprouting, and that D6.1A transcription is initiated in sprouting were collected 5 days after i.p. tumor cell application, it became endothelial cells, point towards the activation of cells in the tumor obvious that newly formed capillary endothelium strongly expres- environment by tumor-derived D6.1A, rather than suggesting that ses D6.1A, i.e., CD31-positive cells were also stained by D6.1. In AS-D6.1A cells themselves provide all the requirements for addition, the expression of a3 was up-regulated in endothelial cells angiogenesis induction. and the expression of h3 integrin was augmented (Fig. 3A). Notably, D6.1A expression in AS cells is accompanied by minor up- capillary D6.1A expression was independent of D6.1A expression by regulation of VEGF and MMP-13. The integrin expression profile the tumor cell, i.e., after i.p. application of AS cells, newly formed of AS and AS-D6.1A cells has already been partly described (13, 15). capillaries also expressed D6.1A, but far fewer capillaries were AS/AS-D6.1A cells do not express a1, a2, h2, and h4(datanot recovered. Furthermore, AS-D6.1A-induced angiogenesis was not shown). They express a3, a4, a5, a6, and h1, with up-regulation of restricted to the tumor and the mesentery, but even in the a4 expression on AS-D6.1A as compared with AS cells. Both lines pancreatic gland, angiogenesis was significantly augmented, as express the tetraspanins CD9, CD81, and CD151, CD151 expression shown by CD31 staining. The pancreatic gland was apparently being slightly increased in AS-D6.1A cells. The angiogenic factors tumor-free. No GFP-expressing tumor cells were detected (data not VEGF, bFGF, and TNFa are expressed in both lines at low to inter- shown) and only endothelial cells were stained by D6.1 (Fig. 3B). mediate levels, with a moderate up-regulation of VEGF expression in The finding that sprouting endothelial cells of the rat up- AS-D6.1A cells. Both lines express the antiangiogenic factors IFNg, regulated D6.1A expression prompted us to see whether this also TIMP-1, TIMP-2, and IL4 at low to intermediate levels (Fig. 4A-C). accounts for the human homologue CO-029, high CO-029 The two lines do not express uPA or uPAR (also evaluated by Wes- expression being associated with poor prognosis in gastrointestinal tern blotting; data not shown). AS cells express MMP-2, MMP-9, and tumors (24, 25). Expression of CO-029 in endothelial cells of the MMP-13 at an intermediate level. MMP-13 is expressed at a higher normal pancreatic gland is hardly detectable. However, endothelial level, and MMP-9 at a lower level, in AS-D6.1A than in AS cells. Also, cell CO-029 expression is up-regulated in inflamed pancreatic MMP-9 secretion is reduced in AS-D6.1A cells (Fig. 4A and D). tissue and in pancreatic cancer including the tumor stroma Although by no means were all proangiogenic and antiangio- (Fig. 3C, third row). Capillary CO-029 expression in the tumor tissue genic factors evaluated, with respect to the major components, could be most clearly visualized in a tumor not expressing this we only observed a moderate up-regulation of VEGF and MMP-13 tetraspanin (Fig. 3C, fourth row, right). expression in AS-D6.1A cells. Therefore, it became likely that D6.1A expression in sprouting capillary endothelium could well D6.1A-expressing tumor cells may, at least in part, induce angio- explain that the D6.1 antibody inhibited tumor-induced angiogen- genesis by transactivation. www.aacrjournals.org 7089 Cancer Res 2006; 66: (14). July 15, 2006

Table 1. Effect of D6.1A-expressing tumor cells on surrounding cells and tissue

Marker Control AS, i.p. AS-D6.1A, i.p. % stained (intensity) % Stained P value control, % Stained P value control, P value AS, (intensity) % (intensity)* (intensity) % (intensity) % (intensity)

Peritoneal exudate macrophages Integrins

aM 97 (243) 97 (1,254) — (<0.0001) 98 (1,841) — (0.0002) — (0.0100) a3 12 (30) 25 (138) 0.0334 (0.0011) 26 (177) 0.0360 (0.0003) — a4 92 (90) 97 (598) — (<0.0001) 97 (580) — (<0.0001) — a5 80 (122) 81 (582) — (0.0019) 82 (633) — (0.0019) — a6 24 (25) 33 (35) — 36 (25) — — h2 72 (347) 99 (1,155) — (0.0024) 99 (1,864) — (0.0002) — (0.0122) h3 38 (17) 55 (48) 0.0276 (0.0047) 73 (72) 0.0026 (0.0005) 0.0407 (0.0213) Tetraspanins CD9 87 (69) 98 (635) — (<0.0001) 97 (658) — (<0.0001) — CD81 61 (134) 91 (696) 0.0312 (<0.0001) 89 (816) 0.0354 (<0.0001) — CD151 36 (18) 54 (92) 0.0313 (0.0012) 56 (98) 0.0344 (0.0020) — D6.1A Not detectable 16 (503) 0.0080 (<0.0001) 32 (563) 0.0010 (<0.0001) 0.0238 (—) Angiogenic factors VEGF 12 (14) 27 (31) 0.0154 (0.0384) 41 (30) 0.0034 (0.0422) 0.0479 (—) bFGF 13 (15) 46 (69) 0.0045 (0.0063) 53 (63) 0.0024 (0.0052) — TNFa 37 (96) 59 (118) 0.0288 (—) 72 (132) 0.0077 (—) — Matrix-degrading enzymes and inhibitors MMP-2 28 (19) 38 (52) — (0.0101) 52 (48) 0.0213 (0.0126) — MMP-9 14 (13) 47 (81) 0.0011 (0.0212) 47 (71) 0.0019 (0.0256) — MMP-13 22 (15) 31 (43) — (0.0103) 38 (46) 0.0425 (0.0090) — TIMP-1 6 (10) 18 (23) 0.0315 (0.0445) 17 (26) 0.0409 (0.0389) — TIMP-2 4 (11) 12 (15) 0.0363 (—) 19 (15) 0.0085 (—) — uPA 2 (11) 8 (15) — 15 (40) 0.0070 (0.0090) — (0.0084) uPAR 9 (13) 17 (25) — (0.0424) 30 (56) 0.0066 (0.0046) 0.0334 (0.0161)

Mesenteric cells Integrins

aM 9 (129) 76 (601) 0.0003 (0.0002) 68 (789) 0.0004 (0.0004) — (0.0436) a1 4 (37) 44 (201) 0.0027 (0.0009) 30 (183) 0.0062 (0.0013) — a2 22 (20) 36 (79) 0.0361 (0.0013) 37 (92) 0.0292 (0.0012) — a3 50 (31) 87 (182) 0.0060 (0.0035) 78 (281) 0.0068 (0.0005) — (0.0184) a4 36 (147) 39 (127) — 64 (122) 0.0123 (—) 0.0206 (—) a5 74 (91) 70 (193) — (0.0080) 74 (217) — (0.0038) — a6 59 (114) 68 (128) — 66 (134) — — h1 45 (23) 72 (55) 0.0154 (0.0105) 74 (147) 0.0151 (0.0022) — (0.0073) h2 86 (135) 42 (258) 0.0025 (0.0275) 55 (230) 0.0138 (0.0406) — h3 35 (90) 30 (74) — 33 (104) — — Tetraspanins CD9 89 (312) 87 (198) — (0.0276) 86 (186) — (0.0271) — CD81 95 (412) 77 (556) — (0.0241) 75 (760) — (0.0042) — (0.0265) CD151 94 (115) 89 (272) — (0.0132) 78 (165) — — D6.1A 21 (67) 33 (73) — 41 (172) — (0.0345) — (0.0284) Angiogenic factors VEGF 37 (35) 42 (61) — (0.0388) 70 (217) 0.0008 (0.0016) 0.0009 (0.0029) bFGF 46 (81) 58 (134) — (0.0225) 76 (322) — (0.0015) 0.0128 (0.0050) TNFa 65 (105) 69 (121) — 65 (112) — — Antiangiogenic factors IFNg 39 (50) 8 (37) 0.0055 (—) 11 (31) 0.0092 (—) — IL4 47 (81) 29 (35) 0.0277 (0.0098) 25 (31) 0.0309 (0.0073) — TIMP-1 59 (50) 27 (49) 0.0168 (—) 32 (39) 0.0222 (—) — TIMP-2 49 (64) 42 (46) — 46 (52) — — Matrix-degrading enzymes MMP-2 15 (31) 27 (67) 0.0472 (0.0116) 48 (212) 0.0090 (0.0006) 0.0021 (0.0014)

(Continued on the following page)

Cancer Res 2006; 66: (14). July 15, 2006 7090 www.aacrjournals.org

Table 1. Effect of D6.1A-expressing tumor cells on surrounding cells and tissue (Cont’d)

Marker Control AS, i.p. AS-D6.1A, i.p. % stained (intensity) % Stained P value control, % Stained P value control, P value AS, (intensity) % (intensity)* (intensity) % (intensity) % (intensity)

MMP-9 13 (30) 27 (40) 0.0235 (—) 47 (140) 0.0198 (0.0043) 0.0032 (0.0067) MMP-13 13 (52) 17 (134) — (0.0032) 29 (178) 0.0500 (0.0023) 0.0113 (—) uPA 15 (68) 20 (57) — 39 (294) 0.0267 (0.0001) 0.0281 (0.0001) uPAR 39 (90) 32 (93) — 38 (205) — (0.0101) — (0.0094) Endothelial cell markers CD31 43 (137) 48 (113) — 65 (298) 0.0211 (0.0066) 0.0234 (0.0053) VEGFR1 23 (44) 34 (47) — 54 (703) 0.0213 (0.0003) 0.0311 (0.0004) VEGFR2 31 (53) 45 (47) — 58 (102) 0.0381 (0.0192) 0.0431 (0.0109)

*Nonsignificant differences (P > 0.05) are indicated by a dash.

D6.1A-expressing tumor cells initiate angiogenesis-promot- Discussion ing activities in the tumor environment. The effect of D6.1A- Angiogenesis is a hallmark for tumor growth and progression (2) expressing tumor cells on surrounding tissues was evaluated for and, accordingly, interference with the angiogenic switch is peritoneal exudate cells and cells of the mesentery 1 week after considered an important therapeutic tool (39). Here, we reported i.p. application of AS or AS-D6.1A tumor cells. With few exceptions that the tetraspanin D6.1A strongly induces angiogenesis, which is (a6, uPA), integrin, tetraspanin, angiogenic factor, and matrix- strikingly inhibited by a D6.1A-specific antibody. There is evidence degrading enzyme expression was up-regulated in peritoneal macro- that tumor cell–associated D6.1A provides only the initial trigger, phages from rats receiving AS or AS-D6.1A cells. Major differences with the tumor microenvironment accounting for an angiogenic between peritoneal macrophages from AS-D6.1A-bearing versus AS- amplification loop. h bearing rats were only noted for 3anduPAR,thatexpressionwas D6.1A-expressing tumor cells form cable-like structures on more strongly up-regulated in peritoneal macrophages of AS-D6.1A- matrigel and induce pronounced angiogenesis when injected i.p. bearing rats (Table 1). The former has also been seen in NIH3T3 cells overexpressing the More striking differences were observed between mesenteric cells tetraspanin CD151 (28), the latter has, to our knowledge, not yet collected from AS-D6.1A-bearing versus AS-bearing rats. Notably, been explored for other tetraspanins. Cable formation by a6h1 has VEGF, bFGF, MMP-2, MMP-9, uPA, uPAR, CD31, VEGFR1, and been found to be due to its association with CD151, where the VEGFR2 expression were significantly more up-regulated in mesen- cytoplasmic tail is essentially required to promote outside-in teric cells of AS-D6.1A-bearing than AS-bearing rats. This was signaling by a6h1 (28). CD82, instead, interferes with cable evaluated quantitatively after digestion of the mesentery by flow formation of a prostate tumor line, likely by down-modulating cytometry (Table 1) and zymography (data not shown) as well as a6h1 expression. Anti-CD9, anti-CD81, and anti-CD151 did not semiquantitatively by immunohistochemistry (Fig. 5A). Immunohis- interfere with the cable-like growth of this tumor line (40). D6.1A- tochemistry provided evidence that capillary endothelial cells might induced cable formation is inhibited by D6.1 and anti-a3, but not be the major source of VEGFR1, VEGFR2, uPA/uPAR, and MMPs. by anti-a6, anti-CD9, anti-CD81, and anti-CD151. Thus, the To substantiate this assumption, a syngeneic fibroblast line and morphogenic features of tetraspanins obviously proceed different- a rat aortic endothelial cell line were cultured in the presence of AS ly, even if the cells have a similar tetraspanin expression profile. We and AS-D6.1A supernatant. AS-D6.1A supernatant induced strong do not yet know the underlying mechanisms, but we speculate that up-regulation of TNFa, uPA, and uPAR in the syngeneic fibroblast the cell type–specific integrin profile will be of importance. This line. Increased VEGF, bFGF, and MMP expression were also induced assumption is derived from our observation that overexpression of by AS supernatant. Coculture of the aortic endothelial cell line with a6h4 in AS-D6.1A cells inhibits cable formation and angiogenesis AS-D6.1A supernatant induced a selective up-regulation of CD31, induction in vivo, but neither cable formation nor angiogenesis can VEGFR1, MMP-9, and uPAR. D6.1A and VEGFR2 expression became be rescued by an anti-a6h4 antibody.4 up-regulated, but to a lesser degree (Fig. 5B and C). These in vitro AS-D6.1A-promoted angiogenesis was independent of cell-cell studies with histologically defined tissue lines confirmed that D6.1A contact, i.e., supernatant from AS-D6.1A cells sufficed for ‘‘angio- influences gene expression (particularly TNFa) in fibroblasts and genesis’’ induction, and addition of the D6.1 antibody inhibited (particularly D6.1A, VEGFR, and CD31) in endothelial cells. supernatant-induced angiogenesis. This points towards D6.1A Taken together, D6.1A is the strongest angiogenesis inducer and contained in the supernatant as being the responsible element. a D6.1A-specific antibody selectively prevents tumor angiogenesis. Tetraspanins are known to be released in vesicles, called exosomes Tumor cell–bound and tumor cell–released D6.1A obviously (41, 42). Indeed, D6.1A shed by AS-D6.1A cells is exclusively initiated an angiogenic loop by inducing D6.1A expression on sprouting endothelial cells. The mode of D6.1A-induced angiogenic gene transcription in endothelial cells and adjacent fibroblasts remains to be elucidated. 4 Unpublished observation. www.aacrjournals.org 7091 Cancer Res 2006; 66: (14). July 15, 2006

Figure 5. Angiogenic factor induction by AS-D6.1A cells and supernatant. Rats received AS or AS-D6.1A cells, i.p., and were sacrificed after 7 days. A, the mesentery was excised and shock-frozen. Sections (5 Am) were stained with anti-VEGF, anti-bFGF, anti-VEGFR1, anti-VEGFR2, anti-uPA, anti-uPAR, anti-MMP-2, anti-MMP-9, anti-MMP-13, and anti-TIMP-2 (bar, 50 Am). B and C, flow cytometry analysis of the aortic endothelial cell line RAEC (B) and a BDX fibroblast line (C) after 72 hours of coculture with supernatant of AS or AS-D6.1A. Cells were stained with the indicated antibodies. Columns, mean percentage of stained cells from three independently done assays; bars, FSD. *, significant differences between cells cultured in the absence of AS and AS-D6.1A supernatants; *, significant differences between cells cultured in the presence of AS-D6.1A as compared with AS supernatant.

recovered in exosomes, whereas VEGF is recovered in the soluble adjacent cells. Peritoneal exudate cells are obviously not a major fraction. The components of AS-D6.1A-derived exosomes have not target of D6.1A, only uPA and uPAR expression became selectively yet been fully resolved; however, we know that aside from up-regulated in AS-D6.1A-bearing compared with AS-bearing rats. tetraspanins, they also contain integrins.5 Importantly, D6.1A This was different for the mesentery. The expression of several enriched exosomes, which do not contain VEGF and induce angiogenesis-related genes, such as bFGF, VEGF, MMP-2, and endothelial cell branching. The enrichment of D6.1A in exosomes MMP-9 became up-regulated in the presence of AS as well as AS- also provides an explanation for the observation that the D6.1 D6.1A supernatant. However, the expression of some genes (CD31, antibody inhibited AS-D6.1A supernatant–induced and exosome- D6.1A, and VEGFR) was forced by AS-D6.1A cells/supernatants. induced endothelial cell branching. D6.1A-initiated angiogenesis Up-regulated expression of these endothelial cell markers was independent of cell-cell contact was also evidenced by strong confirmed in cocultures of a rat aortic endothelial cell line with AS angiogenesis in an adjacent organ, the pancreatic gland, a few days and AS-D6.1A supernatant. In fact, up-regulated expression of after i.p. tumor cell application. Thus, transactivation via D6.1A D6.1A/CO-029 is a general feature of newly formed capillaries in became a likely feature. rats and humans, which is independent of whether tumor cells In fact, AS-D6.1A cells/supernatant do not provide the full express D6.1A/CO-029, albeit in the absence of tumor-derived requirement for angiogenesis induction, but rather stimulate D6.1A, only weak angiogenesis was observed. Therefore, we suggest that the D6.1A-mediated and up-regulated expression of these and additional genes is a secondary phenomenon, i.e., D6.1A does not 5 Unpublished finding. directly promote multiple gene transcription, but initiates the

Cancer Res 2006; 66: (14). July 15, 2006 7092 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2006 American Association for Cancer Research. D6.1A and Angiogenesis angiogenic switch by an as yet undefined mechanism. The finding contribute to transcription/activation of angiogenesis-promoting that AS-D6.1A supernatant selectively induced changes in angio- molecules. genesis-related molecule expression was less pronounced in tissue There remains the question regarding the primary interaction of culture lines than on the mesentery provides additional support D6.1A with endothelial cells. We hypothesize that such an for our interpretation and indicates the requirement for crosstalk interaction is mediated either by cell membrane D6.1A-associated not only between tumor cells/tumor-derived exosomes and molecules and/or by molecules enriched in D6.1A-containing endothelial cells, but also with the surrounding stroma. Nonethe- exosomes. Tetraspanin-mediated protein sorting as described for less, strong D6.1A expression on newly formed capillaries supports the immunologic and the so-called infectious synapse (10, 11, 17, 47) the creation of an angiogenic loop. could well facilitate such a mechanism of transactivation. Notably, this D6.1A-initiated angiogenic switch is hypoxia- Alternatively, one could speculate that D6.1A-containing micro- independent. The assumption is derived from the observations of domains or exosomes may be fusogenic, a feature well known, ‘‘angiogenesis’’ induction in vitro and in vivo at a very early stage of e.g., CD9, that is essential for egg-sperm and myoblast fusion (48, tumor growth as well as in a tumor-free organ. 49). We will explore these hypotheses by analyzing the content of We have not yet explored the biochemical events of D6.1A- AS-D6.1A-derived compared with AS-derived exosomes and by initiated angiogenesis. Because strong D6.1A expression was noted evaluating the mode of exosome-mediated intercellular communi- in sprouting endothelial cells and because D6.1 inhibited cation. angiogenesis, even if the tumor cells did not express D6.1A, we suggest that D6.1A expression on endothelial cells is an early event Conclusions in tumor-derived D6.1A-initiated angiogenesis, where D6.1A- The tumor growth–promoting activity of the tetraspanin, D6.1A/ expressing endothelial cells subsequently support the maintenance CO-029, is due to its capacity to induce the angiogenic switch. of angiogenesis. Tetraspanins, including D6.1A, are located in Notably, membrane-bound as well as exosomal D6.1A supports glycolipid-enriched membrane microdomains (37), which serve as angiogenic factor transcription in targeted cells. As newly formed a scaffold for signal-transducing molecules (10, 16–18). Some capillaries also strongly express D6.1A, an angiogenic loop is tetraspanins associate with G protein–coupled receptors (GPCR) created that further sustains capillary formation. Supporting the and mediate signal transduction via associated heterotrimeric G central role of the tetraspanin, the process can be inhibited by a proteins (43). Although a GPCR association is not yet known for D6.1A-specific antibody. Thus, antibody blockade of this tetraspa- D6.1A, D6.1A associates with prostaglandin F2a receptor regula- nin could well become a new and supposedly highly selective and tory protein (14), which regulates the binding of ligands to GPCRs efficient drug. (44). D6.1A also associates with PKC and a type II phosphoinositide-4-kinase, which initiates signal transduction via PIP2 phos- phorylation (14). Finally, tetraspanins could induce the activation Acknowledgments of associated integrins (10, 16, 18). D6.1A associates with a3 Received 1/31/2006; revised 4/23/2006; accepted 5/23/2006. (13, 14), which, similar to a4 expression, becomes strongly up- Grant support: Deutsche Forschungsgemeinschaft (Zo40-8/3, M. Zo¨ller) and the regulated in endothelial cells during D6.1A-initiated angiogenesis. Tumorzentrum Heidelberg Mannheim (M. Zo¨ller and E. Ryschich). The costs of publication of this article were defrayed in part by the payment of page Both a3 and a4 have been implicated in MMP transcription (45, charges. This article must therefore be hereby marked advertisement in accordance 46). Thus, there are several pathways whereby D6.1A can with 18 U.S.C. Section 1734 solely to indicate this fact.

References 10. Levy S, Shoham T. The tetraspanin web modulates 18. Wright MD, Moseley GW, van Spriel AB. Tetraspanin immune-signalling complexes. Nat Rev Immunol 2005;5: microdomains in immune cell signalling and malignant 1. Folkman J. Fundamental concepts of the angiogenic 136–48. disease. Tissue Antigens 2004;64:533–42. process. Curr Mol Med 2003;3:643–51. 11. Yunta M, Lazo PA. Tetraspanin proteins as organisers 19. Boucheix C, Rubinstein E. Tetraspanins. Cell Mol Life 2. Folkman J. Role of angiogenesis in tumor growth and of membrane microdomains and signalling complexes. Sci 2001;58:1189–205. metastasis. Semin Oncol 2002;29:15–8. Cell Signal 2003;15:559–64. 20. Funakoshi T, Tachibana I, Hoshida Y, et al. Expres- 3. Bergers G, Benjamin LE. Tumorigenesis and the 12. Berditchevski F. Complexes of tetraspanins with sion of tetraspanins in human lung cancer cells: angiogenic switch. Nat Rev Cancer 2003;3:401–10. integrins: more than meets the eye. J Cell Sci 2001;114: frequent downregulation of CD9 and its contribution 4. Giordano FJ, Johnson RS. Angiogenesis: the role of the 4143–51. to cell motility in small cell lung cancer. Oncogene 2003; microenvironment in flipping the switch. Curr Opin 13. Claas C, Seiter S, Claas A, Savelyeva L, Schwab M, 22:674–87. Genet Dev 2001;11:35–40. Zo¨ller M. Association between the rat homologue of 21. Furuya M, Kato H, Nishimura N, et al. Down- 5. Ribatti D. The crucial role of vascular permeability CO-029, a metastasis-associated tetraspanin molecule regulation of CD9 in human ovarian carcinoma cell factor/vascular endothelial growth factor in angio- and consumption coagulopathy. J Cell Biol 1998;141: might contribute to peritoneal dissemination: morphogenesis: a historical review. Br J Haematol 2005;128: 267–80. logic alteration and reduced expression of h1 integrin 303–9. 14. Claas C, Wahl J, Orlicky DJ, et al. The tetraspanin subsets. Cancer Res 2005;65:2617–25. 6. Kraling BM, Wiederschain DG, Boehm T, et al. The D6.1A and its molecular partners on rat carcinoma cells. 22. Zhou B, Liu L, Reddivari M, Zhang XA. The role of matrix metalloproteinase activity in the matu- Biochem J 2005;389:99–110. palmitoylation of metastasis suppressor KAI1/CD82 is ration of human capillary endothelial cells in vitro.JCell 15. Herlevsen M, Schmidt DS, Miyazaki K, Zo¨ller M. The important for its motility- and invasiveness-inhibitory Sci 1999;112:1599–609. association of the tetraspanin D6.1A with the a6h4 activity. Cancer Res 2004;64:7455–63. 7. Oh J, Takahashi R, Kondo S, et al. The membrane- integrin supports cell motility and liver metastasis 23. Gesierich S, Paret C, Hildebrand D, et al. Colocaliza- anchored MMP inhibitor RECK is a key regulator of formation. J Cell Sci 2003;116:4373–90. tion of the tetraspanins, CO-029 and CD151, with extracellular matrix integrity and angiogenesis. Cell 16. Hemler ME. Tetraspanin proteins mediate cellular integrins in human pancreatic adenocarcinoma: impact 2001;107:789–800. penetration, invasion, and fusion events and define a on cell motility. Clin Cancer Res 2005;11:2840–52. 8. Sato Y, Abe M, Tanaka K, et al. Signal transduction and novel type of membrane microdomain. Annu Rev Cell 24. Kanetaka K, Sakamoto M, Yamamoto M, Takamura transcriptional regulation of angiogenesis. Adv Exp Med Dev Biol 2003;19:397–422. M, Kanematsu T, Hirohashi S. Possible involvement of Biol 2000;476:109–15. 17. Tarrant JM, Robb L, van Spriel AB, Wright MD. tetraspanin CO-029 in hematogenous intrahepatic 9. Hemler ME. Specific tetraspanin functions. J Cell Biol Tetraspanins: molecular organisers of the leukocyte metastasis of liver cancer cells. J Gastroenterol Hepatol 2001;155:1103–7. surface. Trends Immunol 2003;24:610–7. 2003;18:1309–14. www.aacrjournals.org 7093 Cancer Res 2006; 66: (14). July 15, 2006

25. SelaBA,SteplewskiZ,KoprowskiH.Colon34. Matzku S, Wenzel A, Liu S, Zo¨ller M. Antigenic tetraspanin slows down cell migration and translocates carcinoma-associated glycoproteins recognized by differences between metastatic and nonmetastatic to the endosomal-lysosomal-MIICs route after extracel- monoclonal antibodies CO-029 and GA22-2. Hybrid- BSp73 rat tumor variants characterized by monoclonal lular stimuli in human immature dendritic cells. Blood oma 1989;8:481–91. antibodies. Cancer Res 1989;49:1294–9. 2004;104:1183–90. 26. Tokuhara T, Hasegawa H, Hattori N, et al. Clinical 35. Galis ZS, Sukhova GK, Lark MW, Libby P. Increased 43. Little KD, Hemler ME, Stipp CS. Dynamic regulation significance of CD151 gene expression in non-small cell expression of matrix metalloproteinases and matrix of a GPCR-tetraspanin-G protein complex on intact lung cancer. Clin Cancer Res 2001;7:4109–14. degrading activity in vulnerable regions of human cells: central role of CD81 in facilitating GPR56-Ga q/11 27. Yunta M, Lazo PA. Apoptosis protection and survival atherosclerotic plaques. J Clin Invest 1994;94:2493–503. association. Mol Biol Cell 2004;15:2375–87. signal by the CD53 tetraspanin antigen. Oncogene 2003; 36. Ryschich E, Schmidt J, Loeffler T, et al. Different 44. Orlicky DJ, Nordeen SK. Cloning, sequencing and 22:1219–24. radiogenic effects on microcirculation in healthy proposed structure for a prostaglandin F2 a receptor 28. Zhang XA, Kazarov AR, Yang X, Bontrager AL, Stipp pancreas and in pancreatic carcinoma of the rat. Ann regulatory protein. Prostaglandins Leukot Essent Fatty CS, Hemler ME. Function of the tetraspanin CD151-6h1 Surg 2003;237:515–21. Acids 1996;55:261–8. integrin complex during cellular morphogenesis. Mol 37. Zhang XA, Bontrager AL, Hemler ME. Transmem- 45. Clayton A, Turkes A, Dewitt S, Steadman R, Mason Biol Cell 2002;13:1–11. brane-4 superfamily proteins associate with activated MD, Hallett MB. Adhesion and signaling by B cell- 29. Longo N, Yanez-Mo M, Mittelbrunn M, et al. protein kinase C (PKC) and link PKC to specific h(1) derived exosomes: the role of integrins. FASEB J 2004;18: Regulatory role of tetraspanin CD9 in tumor-endothelial integrins. J Biol Chem 2001;276:25005–13. 977–9. cell interaction during transendothelial invasion of 38. Claas C, Herrmann K, Matzku S, Mo¨ller P, Zo¨ller M. 46. Ito H, Seyama Y, Kubota S. Calreticulin is directly melanoma cells. Blood 2001;98:3717–26. Developmentally regulated expression of metastasis- involved in anti-a3 integrin antibody-mediated secre- 30. De Cicco M. The prothrombotic state in cancer: associated antigens in the rat. Cell Growth Differ 1996;7: tion and activation of matrix metalloprotease-2. Bio- pathogenic mechanisms. Crit Rev Oncol Hematol 2004; 663–78. chem Biophys Res Commun 2001;283:297–302. 50:187–96. 39. Folkman J. Endogenous angiogenesis inhibitors. 47. Garcia E, Pion M, Pelchen-Matthews A, et al. HIV-1 31. Moseley GW. Tetraspanin-Fc receptor interactions. APMIS 2004;12:496–507. trafficking to the dendritic cell-T-cell infectious synapse Platelets 2005;16:3–12. 40. He B, Liu L, Cook GA, Grgurevich S, Jennings LK, uses a pathway of tetraspanin sorting to the immuno- 32. Lau LM, Wee JL, Wright MD, et al. The tetraspanin Zhang XA. Tetraspanin CD82 attenuates cellular mor- logical synapse. Traffic 2005;6:488–501. superfamily member CD151 regulates outside-in integ- phogenesis through down-regulating integrin a6-medi- 48. Schwander M, Leu M, Stumm M, et al. h1 integrins rin aIIbh3 signaling and platelet function. Blood 2004; ated cell adhesion. J Biol Chem 2005;280:3346–54. regulate myoblast fusion and sarcomere assembly. Dev 104:2368–75. 41. Bartosch B, Vitelli A, Granier C, et al. Cell entry of Cell 2003;4:673–85. 33. Matzku S, Komitowski D, Mildenberger M, Zo¨ller M. hepatitis C virus requires a set of co-receptors that 49. Zhu GZ, Miller BJ, Boucheix C, et al. Residues SFQ Characterization of BSp73, a spontaneous rat tumor and include the CD81 tetraspanin and the SR-B1 scavenger (173-175) in the large extracellular loop of CD9 are its in vivo selected variants showing different metasta- receptor. J Biol Chem 2003;278:41624–30. required for gamete fusion. Development 2002;129: sizing capacities. Invasion Metastasis 1983;3:109–30. 42. Mantegazza AR, Barrio MM, Moutel S, et al. CD63 1995–2002.

Cancer Res 2006; 66: (14). July 15, 2006 7094 www.aacrjournals.org

Sabine Gesierich, Igor Berezovskiy, Eduard Ryschich, et al.

Cancer Res 2006;66:7083-7094.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/14/7083

Cited articles This article cites 49 articles, 19 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/14/7083.full#ref-list-1

Citing articles This article has been cited by 14 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/14/7083.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://cancerres.aacrjournals.org/content/66/14/7083. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.