Angiopoietin-2 Stimulates Breast Cancer Metastasis Through the A5B1 Integrin-Mediated Pathway Yorihisha Imanishi,1,2 Bo Hu,1,3 Michael J

Research Article

Angiopoietin-2 Stimulates Breast Cancer Metastasis through the A5B1 Integrin-Mediated Pathway Yorihisha Imanishi,1,2 Bo Hu,1,3 Michael J. Jarzynka,1,2 Ping Guo,1,2 Esther Elishaev,2,4 Ifat Bar-Joseph,1,2 and Shi-Yuan Cheng1,2

1Cancer Institute and Departments of 2Pathology and 3Medicine, Research Pavilion at the Hillman Cancer Center, University of Pittsburgh and 4Magee-Womens Hospital, Pittsburgh, Pennsylvania

Abstract of breast cancers (2). Several transcription factors, such as Snail, a zinc finger transcriptional repressor, are involved in invasive Acquisition of a metastatic phenotype by breast cancer cells phenotype acquisition through direct repression of E-cadherin includes alternations of multigenic programs that permit expression (3). Snail expression significantly correlates with the tumor cells to metastasize to distant organs. Here, we report metastatic potential of primary breast cancers and established that angiopoietin-2 (Ang2), a known growth factor, is capable cancer cell lines (4). Integrin-linked kinase (ILK) has been shown to of promoting breast cancer cell invasion leading to metastasis. promote the transcription of Snail-enhancing cell motility (5). ILK Analysis of 185 primary human breast cancer specimens that is a serine/threonine protein kinase that interacts with h and h include 97 tumors showing lymph node and/or distant meta- 1 3 integrins and functions as an intracellular adaptor that links stasis reveals a significant correlation between the expression integrins to a range of signaling pathways (6). Overexpression of of Ang2 and E-cadherin, Snail, metastatic potential, tumor ILK in epithelial cells results in loss of E-cadherin expression, grade, and lymph-vascular invasion during breast cancer acquisition of an invasive phenotype, and cell transformation (7). progression. Using a xenograft model, we show that over- ILK stimulates transcription of Snail by phosphorylation of Akt (6). expression of Ang2 in poorly metastatic MCF-7 breast cancer ILK also negatively regulates the activity of glycogen synthase cells suppresses expression of E-cadherin and induces Snail kinase-3h (GSK-3h) through phosphorylation that suppresses Snail expression and phosphorylation of Akt and glycogen synthase phosphorylation and induces nuclear localization and protein kinase-3B (GSK-3B) promoting metastasis to the lymph nodes stabilization of Snail, thus stimulating Snail activities (4). and lung. In cell culture, Ang2 promotes cell migration and It has been shown that ECM proteins promote cell motility invasion in Tie2-deficient breast cancer cells through the through ILK activation of Snail transcription and suppression of A B integrin/integrin-linked kinase (ILK)/Akt, GSK-3B/Snail/ 5 1 E-cadherin (6). Interaction between integrin and ECM proteins also E-cadherin signaling pathway. Inhibition of ILK and the A B 5 1 modulates Snail phosphorylation (3). Integrins are a diverse family integrin abrogates Ang2 modulation of Akt, GSK-3B, Snail, and of glycoproteins that form at least 25 heterodimeric receptors E-cadherin and Ang2-stimulated breast cancer cell migration (one a subunit and one h subunit; ref. 8). Integrins are critical for and invasion. Together, these results underscore the signifi- cell invasion and migration, not only for physically mediating cant contribution of Ang2 in cancer progression, not only by the adhesion of invading tumor cells to the ECM but also for stimulating angiogenesis but also by promoting metastasis, transmitting signals that regulate these processes (9). Among the and provide a mechanism by which breast cancer cells acquire integrin receptors, a h , a h , a h ,anda h integrins are an enhanced invasive phenotype contributing to metastasis. 3 1 5 1 v 1 6 4 associated with enhanced cell motility and cancer metastasis [Cancer Res 2007;67(9):4254–63] (10). h1 integrin is involved in E-cadherin–dependent cell adhesion (11) and Src-induced down-regulation of E-cadherin (12). In breast Introduction cancers, h1 integrin activates ILK-mediated signaling pathways Metastasis, the spread of cancer cells from primary tumor sites promoting cancer cell motility and metastasis (6). to distant organs, is a complex process that involves induction of Angiopoietins (Ang1 and Ang2) are ligands for an endothelial cell motility, activation of extracellular matrix (ECM) proteases, cell–specific tyrosine kinase receptor, Tie2, and are important intravasation to vessels, travel via the circulatory system, and regulators of angiogenesis and tumor progression (13). Moreover, survival and establishment of secondary tumors in new microen- accumulating evidence shows that expression of Ang2 by tumor vironment (1). A hallmark of acquisition of cell motility by tumor cells is linked with invasive and metastatic phenotypes of gastric, cells is the functional loss of E-cadherin and expression of colon, brain, prostate, and breast cancers (14–21). Although Ang2 mesenchymal proteins, such as a-smooth muscle actin and modulates angiogenesis through interaction with the Tie2 receptor vimentin, which defines the key steps toward the invasive phase (13), the highly conserved COOH-terminal fibrinogen-like receptor- binding domain of Ang2 as well as other members of the angiopoietin family implies a functional association with the integrin receptors (22, 23). In fact, Ang2 not only enhances cell adhesion in Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). endothelial cell and Tie2-deficient fibroblasts and myocytes but Requests for reprints: Shi-Yuan Cheng, Cancer Institute and Department of also triggers integrin signaling cascades in these cells (24–26). Pathology, University of Pittsburgh, HCCLB, 2.26f, 5117 Centre Avenue, Pittsburgh, PA In this study, we explore the roles of Ang2 in promoting breast 15213. Phone: 412-623-3261; Fax: 412-623-4840; E-mail: [email protected] or Bo Hu, Cancer Institute and Department of Medicine, University of Pittsburgh, HCCLB, 2.19, tumor metastasis through a Tie2-independent pathway. We identify 5117 Centre Avenue, Pittsburgh, PA 15213. Phone: 412-623-7791; Fax: 412-623-4840; a significant correlation between expression of Ang2 and Snail, E-mail: [email protected]. I2007 American Association for Cancer Research. E-cadherin, and the metastatic potential of clinical breast cancers. doi:10.1158/0008-5472.CAN-06-4100 We also find that Ang2 induces breast cancer metastasis in vitro

Cancer Res 2007; 67: (9). May 1, 2007 4254 www.aacrjournals.org

h and in vivo through a signaling pathway that involves a5 1 integrin, Orthotopic tumor growth was monitored and mice were sacrificed when 3 ILK, Akt, GSK-3h, Snail, and E-cadherin. Thus, our results provide a the tumor size approached 1,000 mm (14–18 weeks postinjection). The functional link between Ang2 expression and enhanced tumor cell lung, liver, spleen, kidney, brain, lymph nodes (axillary, inguinal, and motility and invasion in breast cancer metastasis. cervical), and orthotopic tumors were removed and processed (28). Metastasis in the organs was examined by direct epifluorescent examina- tion of GFP-positive cells using a stereomicroscope (SZX12, Olympus) Materials and Methods before embedding and an upright microscope (Olympus BX51) after Primary human breast cancer specimens and immunohistochem- sectioning followed by H&E staining. Images were captured with SPOT ical and statistical analyses. A total of 185 human breast cancer digital cameras (Diagnostic Instrument) equipped on these microscopes. specimens and the clinicopathologic data were obtained from the Health Immunofluorescent staining. LacZ, Ang#1, and Ang#52 cells were Science Tissue Bank at the University of Pittsburgh Medical Center seeded in serum-containing medium in slide chambers without coating (Pittsburgh, PA). Among these surgically resected samples, 97 specimens overnight. In separate experiments, MCF-7 and T47D cells (ATCC) were were from patients with cancer metastasis to peripheral lymph nodes with seeded in serum-containing medium in slide chambers that were precoated or without distant organ metastasis. The other 88 samples were from with 10 Ag/mL of heat-inactivated bovine serum albumin (BSA), poly-lysine, patients without detectable cancer cell spread. In addition, 10 metastatic or purified Ang2 overnight. After washing the cells extensively with PBS, the and 10 nonmetastatic frozen biopsies from the same patients were also Ang2-stimulated parental MCF-7 and T47D cells were incubated with used for real-time PCR analyses (Supplementary Table S3). The clinico- control or purified Ang2 for 2 h, whereas LacZ, Ang#1, and Ang#52 cells pathologic characteristics of these 185 tumor specimens were examined and were left untreated. After washing cells with PBS, the cells were fixed with verified by a pathologist specialized in breast cancers (E.E.). Importantly, methanol and separately stained with a mouse anti-E-cadherin antibody H&E staining of two separate sections of the same tissues from all 20 frozen (1:200; BD Biosciences) followed by secondary antibodies conjugated with specimens contained >90% tumor cells within these biopsies. Immunohis- Alexa Fluor 488 (Molecular Probes). The nuclei were visualized by Hoechst tochemical staining of tumor specimens and statistics were done as (Sigma) staining. Images were then captured with a microscope (Olympus described previously (19). Results of immunohistochemical staining were BX51) equipped with a SPOT digital camera. categorized independently by two experienced researchers according to the RNA interference. The sequence of ILK small interfering RNA (siRNA; number of immunopositive cells in a blinded fashion. The intensities of 5¶-AAGGACACAUUCUGGAAGGGG-3¶) was reported previously (29). The immunohistochemical staining with the antibody were defined as follows: 21-nucleotide synthetic siRNA duplex was prepared by Dharmacon. Breast negative, the reaction was indistinguishable from the background or <5% of cancer cells were transfected with the ILK siRNA or a 21-nucleotide tumor cells were stained; low, 5% to 30% of tumor cells were positively scrambled RNA duplex as a control using Oligofectamine (Invitrogen). The stained; and high, >30% of tumor cells were positively stained. transfected cells were analyzed 48 h after siRNA transfection by Western RNA isolation, cDNA synthesis, and quantitative real-time PCR. blotting with the anti-ILK antibody and subjected to different assays as Total RNA was isolated from snap-frozen nonmetastatic and metastatic specified. human breast cancer samples using the RNeasy Mini kit (Qiagen). The Western blot. Whole-cell lysates of various cells were analyzed by isolated RNA was DNase treated using the RNase-Free DNase Set (Qiagen), Western blot assays as described previously (30). For inhibition of integrin- and reverse transcription was done with total RNA (500 ng) from each meditated signaling pathway, the cells were preincubated with 25 Ag/mL of sample in a 100 AL reaction volume with random hexamer priming and various neutralizing integrin antibodies and the corresponding IgG as con- SuperScript II reverse transcriptase (Invitrogen). Quantitative real-time PCR trols for 30 min before cell lysis. For detection of phosphorylation of Akt and for Ang2 expression was done using the fluorogenic 5¶-nuclease assay on the GSK-3h and expression of Akt, GSK-3h, and Snail, breast cancer cells were ABI Prism 7700 Sequence Detection System (Applied Biosystems, Inc.). The treated with Ang2 or poly-lysine for 4 h before cell lysis. For detection of Ang2 primers and probe were designed using ABI Primer Express version expression of E-cadherin, breast cancer cells were treated with Ang2 or 2.0 software to span exon junctions as follows: 5¶-CTGAGCAAACGCG- poly-lysine for 48 h before cells were lysed (30). The information of antibodies GAAGTTAA-3¶ (forward primer), 5¶-TGTCGAGAGGGAGTGTTCCAA-3¶ used in Western blot analyses is described in the Supplementary Data. (reverse primer), and 5¶-(FAM)-TGGAAGCCCAAGTATTAAATCAGACCAC- In vitro migration and invasion assays. In vitro migration and invasion GAGA-(BHQ1)-3¶ (probe). The relative expression of Ang2 in the metastatic assays were done as described previously (30). Briefly, serum-starved LacZ, versus the nonmetastatic samples was calculated using a difference in Ang2#1, Ang2#52 cells, and parental MCF-7 and T47D cells were separately b Â 6 threshold cycle (DCt) method with -glucuronidase (control gene) as the suspended at 1 10 cells/mL in serum-free DMEM containing 0.5% BSA A normalizing control gene as described previously (27). The DCt was and preincubated with or without 25 g/mL of various neutralizing anti- calculated by subtracting the b-glucuronidase Ct from the Ang2 Ct, and the integrin antibodies for 30 min on ice. Fifty microliters of each cell relative Ang2 mRNA expression was determined. The PCRs were done in suspension were seeded into upper compartment of the Boyden chambers. triplicate two independent times per sample using a probe concentration of For migration assays, the cells were allowed to migrate through the 12-Am 200 nmol/L and primer concentrations of 300 and 100 nmol/L for Ang2 and pore size membranes precoated with 10 Ag/mL Ang2 or poly-lysine for b-glucuronidase, respectively. The PCR conditions were as follows: a 12-min 12 h at 37jC. For invasion assays, the cells were allowed to invade through denaturation phase at 95jC for 1 cycle followed by 40 cycles of a 15-s the growth factor–reduced Matrigel-coated (0.78 mg/mL; BD Biosciences) denaturation phase at 95jC and a 60-s annealing/extension phase at 60jC. membranes with additional precoating of 10 Ag/mL Ang2 or poly-lysine j The resultant Ct values for each sample and each gene were averaged for for 48 h at 37 C. To evaluate the effects of inhibition of ILK on Ang2 subsequent calculation of the DCt. stimulation, MCF-7 and T47D cells were separately transfected with siRNA Xenograft assays in nude mice. MCF-7 cells were from American Type of ILK and subjected to the migration and invasion assays as described Culture Collection (ATCC). We generated MCF-7 cell lines that stably above. After the membrane was fixed and stained, nonmigrating and express Ang2 and green fluorescent protein (GFP) and LacZ and GFP noninvading cells were removed. The number of migrating cells and (control) and observed no alteration in various cell properties of the derived invading cells was quantified under a microscope (30, 31). cell lines (see Supplementary Data). In an orthotopic model, 1 Â 107 of Statistical analysis. The correlation between expression of Ang2 in LacZ/GFP [a fluorescence-activated cell sorting (FACS) cell population of tumor cells and that of Snail, E-cadherin, and pathologic features in human MCF-7/LacZ cells expressing GFP], Ang2#1/GFP or Ang2#52/GFP cells breast cancer specimens was analyzed using m2 test (Mann-Whitney U test were separately injected into the mammary fat pads of 8-week old ovari- was also applied only for a data set composed of 2 rows Â 3 columns but ectomized female nude mice (Taconic). Two days before the injection, each not for a table of 3 rows Â 3 columns, and identical P values as those of m2 mouse was supplemented with 17h-estradiol (E2) pellets (0.72 mg/pellet) test were obtained in each category). The statistical difference in the that were implanted into the right skin on the lateral side on the neck of the incidence of metastasis between the cell types in the mouse xenograft animal with a trocar (both from Innovative Research of America). model was assessed using Fisher’s exact test. All analyses were done using www.aacrjournals.org 4255 Cancer Res 2007; 67: (9). May 1, 2007

StatView version 5.0 software (SAS Institute, Inc.). A P value <0.05 was plementary Table S3) but not Snail (Supplementary Table S2). considered statistically significant. Finally, using quantitative real-time PCR, we analyzed the expres- Other methods. Cell lines, reagents, antibodies, generation of MCF-7 cell sion of Ang2 mRNA in 20 frozen tumor specimens, including 10 lines that stably express LacZ, Ang2, and GFP, protein expression, tumors showing lymph node and/or distant metastasis (immuno- purification, reverse transcription-PCR (RT-PCR), FACS, immunohistochem- histochemical staining high intensities for Ang2 and Snail and ical analyses of MCF-7 cell-derived tumors, coimmunoprecipitation, and pull-down assays for integrin and ligand interaction are described in the negative for E-cadherin; Supplementary Tables S1 and S4) and 10 Supplementary Data (31, 32). nonmetastatic samples (immunohistochemical staining negative for Ang2 and Snail and positive for E-cadherin; Supplementary Tables S1 and S4). Ang2 mRNA expression was significantly higher Results in the metastatic tumors versus the nonmetastatic samples by at Ang2 expression is correlated with Snail, E-cadherin, and least 10-fold (Fig. 1B). pathologic features of human breast cancer specimens. We Overexpression of Ang2 promotes metastasis and primary first compared expression levels of Ang2 and two important tumor growth of MCF-7 breast cancers in mice. To show the modulators of cell invasion, Snail and E-cadherin (3), in a total of role of Ang2 expression in breast cancer progression, we 185 human breast tumors by immunohistochemical staining using determined whether overexpression of Ang2 by MCF-7 cells that antibodies specific for Ang2, Snail, and E-cadherin. As shown in are poorly metastatic and estrogen dependent is able to stimulate Fig. 1, up-regulation of Ang2 (Fig. 1A, d) significantly correlated metastasis of breast tumor xenografts in mice. We stably trans- with increased expression of nuclear localized Snail (Fig. 1A, e) and fected MCF-7 cells with cDNAs encoding LacZ or Ang2 and GFP down-regulation of E-cadherin (Fig. 1A, f ). Interestingly, Ang2 sequentially (designated as LacZ, Ang2#1, or Ang2#52 hereafter; expression significantly correlated with lymph node metastasis, Supplementary Fig. S1A) and could not detect Tie2 or Tie1 tumor grade, and lymph-vascular invasion (summarized in Supple- expression by RT-PCR analyses in the various MCF-7 and T47D mentary Table S1). In addition, both increased expression of Snail cells (Supplementary Fig. S1B; data not shown). Then, we injected and decreased expression of E-cadherin also significantly corre- various MCF-7 cells into mammary fat pads of ovariectomized lated with lymph node metastasis and lymph-vascular invasion mice supplemented with E2 pellets (an orthotopic model). As (Supplementary Tables S2 and S3), whereas tumor grade signifi- shown in Fig. 2, 14 to 18 weeks postinjection, mice that received cantly correlated with decreased expression of E-cadherin (Sup- control LacZ/GFP cells showed no detectable metastasis in the

Figure 1. Expression of Ang2 correlates with Snail and E-cadherin expression and metastatic potential in primary human breast cancer specimens. A, human breast cancer specimens were stained for Ang2 (a and b), Snail (b and e), and E-cadherin (c and f). Arrows, positive staining of Ang2 (a and d), Snail (b and e, especially in the nuclei), or E-cadherin (c, dominantly on the cell membrane; f, barely positive in the cytoplasm) in the tumor cells. Ang2 proteins were detected mostly in tumor cell cytoplasm. Representative stains from a total of 185 tumor samples. Each specimen was independently analyzed at least twice with similar results. Bar,20Am. B, expression of Ang2 mRNA correlates with the metastatic potential of primary human breast cancer specimens. Quantitative real-time PCR was done on cDNA derived from total RNA isolated from nonmetastatic and metastatic snap-frozen human breast cancer specimens. The relative expression of Ang2 in the metastatic versus the nonmetastatic samples was calculated using a difference in threshold cycle (DC t) method with b-glucuronidase as the normalizing control gene. PCRs were done in triplicate and each specimen was independently analyzed twice with similar results.

Cancer Res 2007; 67: (9). May 1, 2007 4256 www.aacrjournals.org

Figure 2. Overexpression of Ang2 by MCF-7 breast cancer cells promotes tumor metastasis in mice. MCF-7 parental cells were obtained from ATCC. Various MCF-7 cell lines were generated to stably express Ang2 or LacZ and GFP with no observable alterations in cell properties (see Materials and Methods in the Supplementary Data). A and B, tumor metastasis was examined by gross observation (a, e, and i) and epifluorescent observation of GFP (b, f, and j). Cryosections were further examined by epifluorescence (d, h, and l) followed by H&E staining (c, g, and k). Mice that received MCF-7/Ang2/GFP cells (clone 1 or 52) showed metastasis in the lung and lymph nodes (e–l in A and B) with high incidence. Grossly visible metastases (blue arrowheads) in the lung (e and i in A) and lymph nodes (e and i in B) were identified as green foci (f and j, white arrowheads) and micrometastatic cells expressing GFP (h and l, red arrowheads) and by H&E staining (g and k, red arrowheads). Bars, 120 Am (b, c, d, f, g, h, j, k, and l in A), 600 Am (b, f, and j in B), and 30 Am(c, d, g, h, k, and l in B). The experiments in (A) and (B) were done three independent times and similar results were obtained. Additionally, these experiments were also done twice using Ang2#4 (low expression of Ang2) and Ang2#5 (high expression of Ang2; Supplementary Fig. S1A) in mice. High incidence of lymph node and lung metastasis was found in mice that received Ang2#5 cells, whereas no tumor metastasis was found in mice that received Ang2#4 cells.

lung (Fig. 2A, a–d) and all other examined organs (data not shown). metastasis was found to be of similar frequency to that of Ang2#1 Lymph node metastasis was also absent in all mice (Fig. 2B, a–d) cell clone. In addition, we also found that Ang2 overexpression except for one mouse with a small cluster of LacZ/GFP cells in the moderately affects the growth, angiogenesis, but not lymphangio- lymph nodes (Supplementary Table S5). In contrast, multiple lung genesis of orthotopic MCF-7 tumors in the mice that displayed a metastases were detected in 12 of 13 (92%) mice that received metastatic phenotype (data not shown). Ang2#1 cells (Fig. 2A, e–h; Supplementary Table S5) and 5 of Ang2 modulates E-cadherin expression in breast cancer 8 (62.5%) mice that received Ang2#52 cells (Fig. 2A, i–l; cells. Because loss of E-cadherin expression is linked with Supplementary Table S5). Metastatic lesions on the surface of the acquisition of breast cancer cell invasion (2), we sought to lung were grossly visible in more than half of the cases with determine whether Ang2-stimulated MCF-7 tumor metastasis is metastasis (Fig. 2A, e and i, arrowheads). Metastasis in the lymph associated with modulation of E-cadherin expression in breast nodes was found in 10 of 13 (77%) mice that received Ang2#1 cells cancer cell lines that were stimulated by Ang2. As shown in Fig. 3A, (Fig. 2B, e–h; Supplementary Table S5) and 4 of 8 (50%) mice that LacZ control cells formed compact cell clusters with strong E- received Ang2#52 cells (Fig. 2B, i–l; Supplementary Table S5). Liver cadherin expression along the cell-cell contacts. However, Ang2#1 metastasis was also found in two (15%) mice that received Ang2#1 and Ang2#52 cells had diminished E-cadherin expression. In cells (Supplementary Table S5). It is noteworthy that the metastatic various orthotopic MCF-7 tumors, Ang2 was expressed at low potential of Ang2-expressing cells correlates with Ang2 expression levels in LacZ tumors (Fig. 3B, b), whereas high levels of E-cadherin levels in these cells (Supplementary Fig. S1A). Ang2#1 displayed were found in the cytoplasm and the cell-cell contacts (Fig. 3B, c). high incidence of metastasis, Ang2#52 had lower metastatic In contrast, Ang2-overexpressing tumors (Fig. 3B, e) showed a potential (Supplementary Table S5), whereas Ang2#4 did not cause significant decrease in E-cadherin expression (Fig. 3B, f ). metastasis in mice (data not shown). Experiments were also done We then assessed whether stimulation by exogenous Ang2 on using Ang2#5 cell clone in mice and Ang2 stimulation of tumor MCF-7 and T47D parental breast cancer cells modulates www.aacrjournals.org 4257 Cancer Res 2007; 67: (9). May 1, 2007

Figure 3. Ang2 stimulation induces loss of E-cadherin expression in breast cancer cells. MCF-7 and T47D cells used in Figs. 3 to 6 were obtained from ATCC. A, immunofluorescent staining of MCF-7/LacZ control and MCF-7/Ang2 cells (clones 1 and 52) using an anti-E-cadherin antibody (green) and Hoechst (blue for nuclear staining). MCF-7/Ang2 cells showed decreased expression of E-cadherin in the cell membrane leading to loss of cell-cell contact (arrows in MCF-7/Ang2 cells). Bar, 15 Am. B, immunohistochemical analyses of orthotopic tumors established by LacZ/GFP (a–c) or Ang2#1 cells (d and f) using H&E (a and c), anti-Ang2 (b and e), or anti- E-cadherin antibodies (c and f). Arrows, Ang2 (b and e) or E-cadherin (c and f) staining. Three to five serial-cut paraffin sections from five or more individual tumor samples of each group were independently analyzed. Bars, 20 Am. C, immunofluorescent staining of MCF-7 and T47D cells adhered to exogenous Ang2, heat-inactivated BSA, or poly-lysine (Poly-L). Down-regulation of E-cadherin in the cell membrane and loss of cell-cell contact (arrows) were observed in Ang2-treated cells. Bar, 15 Am. D, Western blot analyses of E-cadherin expression in exogenous Ang2-treated breast cancer cells. MCF-7 and T47D cells were treated with heat-inactivated BSA, Ang2, or poly-lysine for 48 h. h-Actin was used as a loading control. The experiments in (A)to(D) were done two independent times with similar results. expression of E-cadherin in vitro. As shown in Fig. 3C, when MCF-7 activation; ref. 33) had strong E-cadherin staining along the cell- and T47D cells were treated with exogenous Ang2, loss of cell border. Next, loss of E-cadherin expression in Ang2-stimulated E-cadherin expression along the cell-cell contacts, as well as in cells was further examined by Western blot analyses. A decreased the cytoplasm, was evident. The levels of decreased E-cadherin expression of E-cadherin was detected in MCF-7 and T47D cells expression in cells treated with Ang2 were similar to that of Ang2- stimulated with Ang2 compared with stimulation of heat- expressing cells. In contrast, cells treated with heat-inactivated inactivated BSA or poly-lysine (Fig. 3D). Interestingly, when we BSA or poly-lysine (an inert substrate for integrin receptor reseeded these Ang2-stimulated cells onto poly-lysine–coated

Cancer Res 2007; 67: (9). May 1, 2007 4258 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Ang2 Stimulates Breast Cancer Metastasis surfaces, E-cadherin expression was restored (Supplementary phorylation at Ser473 stimulates transcription of Snail (6), Fig. S2). whereas inactivation of GSK-3h by phosphorylation at Ser9 leads Ang2 activates the ILK/Akt and GSK-3B/Snail pathway in to enhanced nuclear localization and protein stabilization of Snail breast cancer cells. Next, we examined Snail expression in (34). Therefore, we assessed whether phosphorylation of Akt various MCF-7 orthotopic tumors by immunohistochemistry. As (pSer473-Akt, activated form) and GSK-3h (pSer9-GSK-3h, inacti- shown in Fig. 4A, up-regulated Snail expression was found mostly vated form) were modulated in various MCF-7 orthotopic tumors. in the nuclei of Ang2-expressing tumors (Fig. 4A, d) but not in the As shown in Fig. 4A, the phosphorylation of Akt and GSK-3h was LacZ controls (Fig. 4A, a), suggesting that nuclear localized Snail enhanced in the Ang2-expressing tumors compared with the may be responsible for decreased E-cadherin expression in these controls (Fig. 4A, compare e and f with b and c). tumors. GSK-3h and Akt are two major regulators modulating Next, we determined whether Ang2 stimulation of MCF-7 cells Snail expression in tumor cells (4). Activation of Akt by phos- activates Akt and GSK-3h/Snail signaling pathway regulators

Figure 4. Ang2 stimulates the ILK/Akt, GSK-3h/Snail/E-cadherin pathway. A, immunohistochemical analyses of identical orthotopic tumors established by LacZ/GFP (a–c) or Ang2#1 (d–f) using anti-Snail antibody (a and d, paraffin-embedded tissue sections), anti-p-Akt (Ser473) antibody (b and e, frozen tissue sections), and anti-p-GSK-3h (Ser9) antibody (c and f, frozen tissue sections). Arrows, positive staining of Snail especially in nuclei of tumor cells (a and d), p-Akt in cytoplasm (b and e), or p-GSK-3h in cytoplasm (c and f). Three to five serial-cut sections from five or more individual tumor samples of each group were independently analyzed. Bar, 20 Am. B, Western blot analyses of Ang2-expressing cells or LacZ control cells treated with Ang2 using anti-p-Akt (S473), anti-p-GSK-3h (S9), anti-Snail, anti-E-cadherin, and anti-vimentin antibodies. The membranes were reprobed with anti-Akt, anti-GSK-3h, or anti-h-actin antibodies as loading controls. C and D, Western blot analyses of parental MCF-7 (C) and T47D (D) cells transfected with siRNA for ILK (I), control siRNA (C), or no transfection (À) followed by stimulation with Ang2 or poly-lysine. The cell lysates were then analyzed as described in (B) except an anti-ILK antibody was used. The results from (A)to(D) are representative of three independently repeated experiments. www.aacrjournals.org 4259 Cancer Res 2007; 67: (9). May 1, 2007

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Cancer Research critical for breast cancer cell invasion (2). As shown in Fig. 4B, both Ang2-expression and exogenous Ang2 stimulation increased phosphorylation of Akt at Ser473 and GSK-3h at Ser9, enhanced intracellular Snail protein level, and attenuated E-cadherin expression in MCF-7 cells. Vimentin, a mesenchymal marker, was also induced in cells stimulated by Ang2 but was undetectable in nonstimulated MCF-7 cells. ILK activation promotes phosphorylation of Akt on Ser473 and GSK-3h on Ser9 and the activation of integrin and ILK stimulates cell invasion (6). Therefore, we assessed whether activation of the ILK/Akt and GSK-3h pathway is essential to modulate Ang2- regulated expression of Snail and E-cadherin in breast cancer cells. As shown in Fig. 4C and D, siRNA suppression of ILK (29) attenuated Ang2-stimulated phosphorylation of Akt and GSK-3h in MCF-7 (Fig. 4C) and T47D (Fig. 4D) cells and resulted in restoration of E-cadherin expression and inhibition of Snail and vimentin expression. In these experiments, we only used the inert ligand, poly-lysine, as a negative control for integrin/ILK stimulation (33) because BSA has minimal effects on integrin signaling (31). Ang2 associates with B1 integrin in Tie2-deficient MCF-7 cells. ILK exerts its function as an intracellular signaling adaptor through interaction with the cytoplasmic domains of h1 and h3 integrin subunits (6). Moreover, integrin mediates cellular signaling events by forming ah heterodimeric integrin receptors (8). Therefore, we examined the expression profile of various a and h Figure 5. Ang2 associates with h1 integrin in Tie2-deficient MCF-7 cells. integrin subunits that have been implicated in cancer metastasis in Coimmunoprecipitation (IP) and pull-down assays. Exogenous Ang2 was encoded by a c-Myc and His-tagged pSecTagB expression vector that is breast cancer cells (10). We found that h1 was abundantly h described in the Supplementary Data and expressed in MCF-7 Ang2#1 cells. expressed, but 3 was absent in both MCF-7 and T47D cells. The The associated Ang2 with h1 integrin on various cells was assessed by h subunits, a , a , a , a , a but not a and a , were also expressed in immunoprecipitation separately with anti- 1 (A), anti-c-Myc (B), or anti-Ang2 (C) 2 3 5 6 v 1 4 h these breast cancer cell lines (Supplementary Table S6; refs. 35, 36). antibodies followed by Western blot using anti- 1 (A–C, top) or anti-Ang2 (A–C, bottom) antibodies. D, a pull-down assay. The Ang2-associated h1 integrin was + These data led us to focus on the involvement of h1 and a subunits pulled down by its His tag of Ang2 using Ni -NTA beads followed by Western blot using anti-h or anti-Ang2 antibodies. Similar results were also found when that have been shown to associate with h1.Next,wedid 1 the identical experiments were done using T47D cells (data not shown). Results coimmunoprecipitation assays and found that equal amounts of are representative of three independently repeated experiments. h1 integrin in immunoprecipitated complexes from both groups, but only immunoprecipitated complexes from the MCF-7 cells sti- inished Ang2-stimulated cell motility, whereas suppression of a mulated with exogenous Ang2, had Ang2 protein present (Fig. 5A). 2 had no effect on Ang2 stimulation (Supplementary Fig. S3A). In reciprocal experiments, only the immunoprecipitates from MCF- Interestingly, an anti-Ang2 antibody (30) also attenuated Ang2- 7 cells treated with c-Myc–tagged Ang2 contained h integrin 1 stimulated cell migration and invasion (Fig. 6B; Supplementary (Fig. 5B and C). In a pull-down assay, Ang2-associated h integrin 1 Fig. S3A), corroborating an autocrine effect by Ang2. Additionally, was pulled down with His-tagged Ang2 by Ni+-NTA beads, and both when ILK expression was inhibited by a specific siRNA for ILK h integrin and Ang2 were detected only in the cells incubated with 1 in MCF-7 and T47D cells, Ang2-stimulated cell migration and Ang2 but were undetectable in the cells treated with control invasion were completely abrogated (Fig. 6C; Supplementary (Fig. 5D). Fig. S3B). Ang2-stimulated breast cell motility is mediated through the A5B1 integrin/ILK pathway. We tested the hypotheses that Ang2 stimulation would increase breast cancer cell motility and Discussion inhibition of integrins will suppress Ang2 activation of the ILK/Akt In this study, we show that expression of Ang2 by Tie2-deficient and GSK-3h pathway, cell migration, and invasion. As shown in breast cancer cells induces tumor metastasis to distant organs in h h h Fig. 6A, inhibition of 1, a5 but not a2, 4 (data not shown), or animals by stimulating cell invasion potentially through the a5 1 IgG suppressed Ang2-induced phosphorylation of Akt and GSK-3h, integrin-mediated pathway. The association of Ang2 with a5h1 Ang2-enhanced expression of vimentin and Snail, and Ang2 integrin results in activation of ILK, Akt, and GSK-3h leading to the h repression of E-cadherin, suggesting that a5 1 could be the inte- stimulation of Snail and vimentin expression and down-regulation grin receptor involved in these Ang2-stimulated intracellular of E-cadherin. Our results provide a probable explanation for the signaling processes. Finally, we determined whether functional prometastatic ability Ang2 in animals and support our finding of a inhibition of h1 and a5 integrins and ILK attenuates Ang2 significant correlation of Ang2 up-regulation with metastatic stimulation of breast cancer cell motility. As shown in Fig. 6B potential in human breast cancers. These data also corroborate and Supplementary Fig. S3A, Ang2 expression and exogenous Ang2 recent studies showing that a5h1 integrin is centrally implicated in stimulation promoted MCF-7 cell invasion (Fig. 6B) and migration tumor cell invasion induction of Ha-Ras–transformed epithelial (Supplementary Fig. S3A). Consistent with the results shown in cells (37). However, we cannot rule out whether a3 and av integrins Fig. 6A, inhibition of h1 and a5 by neutralizing antibodies dim- are also involved in mediating Ang2 effects because inhibition of a3

Cancer Res 2007; 67: (9). May 1, 2007 4260 www.aacrjournals.org

Figure 6. Ang2-stimulated breast cancer cell invasion is mediated through the a5h1 integrin-mediated pathway. A, Western blot analyses. Inhibition of h1 and a5, but not a2, integrins attenuates Ang2 modulation of Akt, GSK-3h, Snail, E-cadherin, and vimentin. The membranes were reprobed with anti-Akt, anti-GSK-3h, or anti-h-actin antibodies as loading controls. B and C, in vitro cell invasion assays. B, inhibition of h1 and a5, but not a2, integrins suppresses Ang2 stimulation of breast cancer cell invasion. Mouse IgG and an anti-Ang2 (a nonneutralizing) antibody are included as controls. C, inhibition of ILK using siRNA specific for ILK suppresses Ang2 stimulation of breast cancer cell invasion. Results in (A)to(C) are representative of three independent times. Columns, mean percentage increase in the number of invading cells compared with control; bars, SD.

www.aacrjournals.org 4261 Cancer Res 2007; 67: (9). May 1, 2007

and av integrins moderately attenuated Ang2-stimulated cell survival has been shown in various types of human cancers, motility as well as ILK signaling (data not shown). Because a5h1 including breast cancers (15–21, 38–41). In one study, tumor cell- and a3h1 integrins are involved in cancer cell motility (9), it is expressed Ang2 was detected in primary human glioma tissues by plausible that Ang2 promotes breast cancer cell metastasis through an anti-Ang2 antibody and expression was verified in identical association with a5h1 and to a lesser extent a3h1 and avh1 tumor cells in sister tissue sections by in situ hybridization analysis integrins. These results suggest a complex interplay between Ang2 using a human Ang2 DNA probe (14). Our analyses of 185 primary and integrin signaling, leading to stimulation of breast cancer cell breast cancer specimens corroborate these findings. We show that invasion and facilitation of metastasis. Ang2 expression significantly correlates with Snail up-regulation Emerging evidence has shown that angiopoietins can function as and E-cadherin down-regulation as well as with lymph node adhesive ligands to stimulate cell adhesion, migration, and survival metastasis and tumor grade. Similar to the aforementioned studies, of endothelial cells, fibroblasts, cardiac and skeletal myocytes, and we show that up-regulated Ang2 was primarily found in tumor cancer cells through integrin-mediated pathways independent of cells. Quantitative real-time PCR analysis indicates a 10-fold higher Tie2. Ang1 and Ang2 were first shown to stimulate Tie2- expression of Ang2 in the metastatic samples versus nonmetastatic independent cell adhesion of endothelial cells and fibroblasts to tumors. Although we cannot rule out the contribution of the tumor Ang1- or Ang2-coated surfaces through a5h1 and avh5 integrin- stroma to Ang2 expression, this is unlikely due to the fact that mediated activation of extracellular signal-regulated kinase (ERK) fresh-frozen breast cancer tissues analyzed by immunohistochem- and focal adhesion kinase (FAK) signaling (24). Skeletal myocytes istry contained >90% tumor cells, thus suggesting that the up- lacking Tie2 adhere to Ang1- and Ang2-coated surfaces in a similar regulation of Ang2 was primarily by cancer cells not stromal or manner as to laminin, fibronectin, and vitronectin. The angiopoie- endothelia compartments in the metastatic breast cancers. In tin-stimulated skeletal myocyte adhesion is mediated by integrin orthotopic MCF-7 tumors, although the anti-Ang2 antibody that h receptors, such as a5 1, activating ERK, FAK signaling, and was used in immunohistochemical analyses recognizes both promoting cell survival (25). It has been postulated that the human and mouse Ang2, Ang2 proteins were detected almost interaction of angiopoietin with integrins is likely through the exclusively in tumor cells in the Ang2-expressing tumors. fibrinogen-like receptor-binding domain present in the angiopoietin Additionally, we also found that Ang2 expression in MCF-7 cells protein structure (22). This theory has been examined recently promoted a moderate increase in angiogenesis and tumor growth showing that a monomeric Ang1 variant (DAng), composed only of in metastatic breast tumors. Because MCF-7 tumors express low the fibrinogen-like receptor-binding domain that is also present in levels of vascular endothelial growth factor (VEGF; ref. 28), Ang2, ligates Tie2 without activating the receptor. Moreover, DAng insufficient VEGF within the tumor microenvironment resulted binds to a5h1 integrin with similar affinity compared with Tie2. in a moderate increase in tumor vessel growth in Ang2-expressing When endothelial cells were plated on DAng-coated surfaces, DAng tumors (42), most likely limiting the contribution of Ang2- displays similar biological effects as full-length Ang1, such as stimulated angiogenesis to the acquisition of tumor metastasis. stimulation of cell adhesion, ERK signaling, and vascular matura- Together, these data show that in both human breast cancer tion (23). Even in endothelial cells that express the Tie2 receptor, specimens and tumor xenograft models, Ang2 was primarily h immobilized Ang1 is able to selectively mediate a5 1 integrin derived from tumor epithelial cells, thus supporting the hypothesis outside-in signaling leading to a cross-talk between Tie2 and a5h1 that Ang2 is involved in breast cancer metastasis through an and promotion of angiogenesis (26). In glioma cells that lack Tie2 autocrine pathway independent of Tie2. expression, Ang2 induces glioma cell invasion by stimulating matrix In summary, we have presented evidence showing that Ang2 metalloproteinase-2 expression through the avh1 integrin and FAK induces breast cancer cell metastasis through the a5h1 intergin/ pathway (31). In this study, we report that Ang2 associates with a5h1 ILK-Akt-GSK-3h-Snail–mediated pathway stimulating breast can- integrin in Tie2-deficient breast cancer cells. Ang2 activates an cer cell migration and invasion. Our results underscore the integrin-mediated signaling pathway leading to breast cancer cell contribution of Ang2 in cancer progression, not only by stimulating invasion and metastasis. Inhibition of h1 or a5, but not other angiogenesis but also by promoting metastasis. These data provide integrins, attenuates Ang2 modulation of ILK, Akt, GSK-3h, Snail, E- new insight into the mechanisms underlying cancer metastasis and cadherin, and vimentin and Ang2-stimulated breast cancer cell could establish Ang2 and its effectors as potential targets for breast motility. Furthermore, similar to the association of Ang2 cancer treatment. with integrins in skeletal myocytes (25) and PG-MV/vesican with h integrin in glioma cells (32), the association of Ang2 with a5 1 integrin was highly calcium and manganese dependent in our Acknowledgments system. Taken together, our results establish a critical role of Ang2 Received 11/7/2006; revised 2/6/2007; accepted 2/16/2007. in promoting breast cancer metastasis through stimulation of cell Grant support: Department of Defense grants DAMD17-01-1-0375 and DAMD-17- 02-1-0584 (S-Y. Cheng) and the Hillman Fellows Program at the University of motility and invasion mediated by the a5h1 integrin/ILK pathway Pittsburgh Cancer Institute (S-Y. Cheng and B. Hu). independent of Tie2. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance A significant association between Ang2 up-regulation in tumor with 18 U.S.C. Section 1734 solely to indicate this fact. cells and cancer invasion/metastasis, and decreased patient We thank C. Wu and T. El-Hefnawy for their advice and assistance.

References 2. Thiery JP. Epithelial-mesenchymal transitions in 4. Zhou BP, Deng J, Xia W, et al. Dual regulation of Snail tumour progression. Nat Rev Cancer 2002;2:442–54. byGSK-3h-mediated phosphorylation in control of 1. Chambers AF, Groom AC, MacDonald IC. Dissemina- 3. Nieto MA. The snail superfamily of zinc-finger epithelial-mesenchymal transition. Nat Cell Biol 2004;6: tion and growth of cancer cells in metastatic sites. Nat transcription factors. Nat Rev Mol Cell Biol 2002;3: 931–40. Rev Cancer 2002;2:563–72. 155–66. 5. Barbera MJ, Puig I, Dominguez D, et al. Regulation

Cancer Res 2007; 67: (9). May 1, 2007 4262 www.aacrjournals.org

of Snail transcription during epithelial to mesenchy- the invasiveness of human glioma. Am J Pathol 2005;166: metalloprotease 2 expression through the avh1 integrin mal transition of tumor cells. Oncogene 2004;23: 877–90. and focal adhesion kinase signaling pathway. Cancer 7345–54. 20. Lind AJ, Wikstrom P, Granfors T, et al. Angiopoietin 2 Res 2006;66:775–83. h 6. Hannigan G, Troussard AA, Dedhar S. Integrin-linked expression is related to histological grade, vascular 32. Wu Y, Chen L, Zheng PS, Yang BB. 1-Integrin- kinase: a cancer therapeutic target unique among its density, metastases, and outcome in prostate cancer. mediated glioma cell adhesion and free radical-induced ILK. Nat Rev Cancer 2005;5:51–63. Prostate 2005;62:394–9. apoptosis are regulated by binding to a C-terminal domain 7. Wu C, Keightley SY, Leung-Hagesteijn C, et al. 21. Sfiligoi C, de Luca A, Cascone I, et al. Angiopoietin-2 of PG-M/versican. J Biol Chem 2002;277:12294–301. Integrin-linked protein kinase regulates fibronectin expression in breast cancer correlates with lymph 33. Huang J, Asawa T, Takato T, Sakai R. Cooperative matrix assembly, E-cadherin expression, and tumorige- node invasion and short survival. Int J Cancer 2003; roles of Fyn and cortactin in cell migration of metastatic nicity. J Biol Chem 1998;273:528–36. 103:466–74. murine melanoma. J Biol Chem 2003;278:48367–76. 8. Hynes RO. Integrins: bidirectional, allosteric signaling 22. Davis S, Papadopoulos N, Aldrich TH, et al. 34. Blanco MJ, Moreno-Bueno G, Sarrio D, et al. machines. Cell 2002;110:673–87. Angiopoietins have distinct modular domains essential Correlation of Snail expression with histological grade 9. Hood JD, Cheresh DA. Role of integrins in cell invasion for receptor binding, dimerization, and superclustering. and lymph node status in breast carcinomas. Oncogene and migration. Nat Rev Cancer 2002;2:91–100. Nat Struct Biol 2003;10:38–44. 2002;21:3241–6. 10. Felding-Habermann B. Integrin adhesion receptors in 23. Weber CC, Cai H, Ehrbar M, et al. Effects of protein 35. Deryugina EI, Bourdon MA, Jungwirth K, Smith JW, h tumor metastasis. Clin Exp Metastasis 2003;20:203–13. and gene transfer of the angiopoietin-1 fibrinogen-like Strongin AY. Functional activation of integrin aV 3 in 11. Schreider C, Peignon G, Thenet S, Chambaz J, receptor-binding domain on endothelial and vessel tumor cells expressing membrane-type 1 matrix metal- Pincon-Raymond M. Integrin-mediated functional po- organization. J Biol Chem 2005;280:22445–53. loproteinase. Int J Cancer 2000;86:15–23. larization of Caco-2 cells through E-cadherin-actin 24. Carlson TR, Feng Y, Maisonpierre PC, Mrksich M, 36. Saad S, Bendall LJ, James A, GottliebDJ, Bradstock complexes. J Cell Sci 2002;115:543–52. Morla AO. Direct cell adhesion to the angiopoietins KF. Induction of matrix metalloproteinases MMP-1 and 12. Avizienyte E, Wyke AW, Jones RJ, et al. Src- mediated by integrins. J Biol Chem 2001;276:26516–25. MMP-2 by co-culture of breast cancer cells and bone induced de-regulation of E-cadherin in colon cancer 25. Dallabrida SM, Ismail N, Oberle JR, Himes BE, marrow fibroblasts. Breast Cancer Res Treat 2000;63: cells requires integrin signalling. Nat Cell Biol 2002;4: Rupnick MA. Angiopoietin-1 promotes cardiac and 105–15. 632–8. skeletal myocyte survival through integrins. Circ Res 37. Maschler S, Wirl G, Spring H, et al. Tumor cell 13. Yancopoulos GD, Davis S, Gale NW, et al. Vascular- 2005;96:e8–24. invasiveness correlates with changes in integrin expres- specific growth factors and blood vessel formation. 26. Cascone I, Napione L, Maniero F, Serini G, Bussolino sion and localization. Oncogene 2005;24:2032–41. Nature 2000;407:242–8. F. Stable interaction between a5h1 integrin and Tie2 38. Ahmad SA, Liu W, Jung YD, et al. Differential 14. Koga K, Todaka T, Morioka M, et al. Expression of tyrosine kinase receptor regulates endothelial cell expression of angiopoietin-1 and angiopoietin-2 in colon angiopoietin-2 in human glioma cells and its role for response to Ang-1. J Cell Biol 2005;170:993–1004. carcinoma. A possible mechanism for the initiation of angiogenesis. Cancer Res 2001;61:6248–54. 27. Tassone F, Hagerman RJ, Taylor AK, et al. Elevated angiogenesis. Cancer 2001;92:1138–43. 15. Etoh T, Inoue H, Tanaka S, et al. Angiopoietin-2 is levels of FMR1 mRNA in carrier males: a new 39. Loges S, Heil G, Bruweleit M, et al. Analysis of related to tumor angiogenesis in gastric carcinoma: mechanism of involvement in the fragile-X syndrome. concerted expression of angiogenic growth factors in possible in vivo regulation via induction of proteases. Am J Hum Genet 2000;66:6–15. acute myeloid leukemia: expression of angiopoietin-2 Cancer Res 2001;61:2145–53. 28. Guo P, Fang Q, Tao HQ, et al. Overexpression of represents an independent prognostic factor for overall 16. Tanaka F, Ishikawa S, Yanagihara K, et al. Expression vascular endothelial growth factor by MCF-7 breast survival. J Clin Oncol 2005;23:1109–17. of angiopoietins and its clinical significance in non- cancer cells promotes estrogen-independent tumor 40. Mitsuhashi N, Shimizu H, Ohtsuka M, et al. small cell lung cancer. Cancer Res 2002;62:7124–9. growth in vivo. Cancer Res 2003;63:4684–91. Angiopoietins and Tie-2 expression in angiogenesis 17. Ogawa M, Yamamoto H, Nagano H, et al. Hepatic 29. Fukuda T, Chen K, Shi X, Wu C. PINCH-1 is an and proliferation of human hepatocellular carcinoma. expression of ANG2 RNA in metastatic colorectal obligate partner of integrin-linked kinase (ILK) func- Hepatology 2003;37:1105–13. cancer. Hepatology 2004;39:528–39. tioning in cell shape modulation, motility, and survival. 41. Wang J, Wu K, Zhang D, et al. Expressions and 18. Ochiumi T, Tanaka S, Oka S, et al. Clinical J Biol Chem 2003;278:51324–33. clinical significances of angiopoietin-1, -2 and Tie2 in significance of angiopoietin-2 expression at the deepest 30. Hu B, Guo P, Fang Q, et al. Angiopoietin-2 induces human gastric cancer. Biochem Biophys Res Commun invasive tumor site of advanced colorectal carcinoma. human glioma invasion through the activation of matrix 2005;337:386–93. Int J Oncol 2004;24:539–47. metalloprotease-2. Proc Natl Acad Sci U S A 2003;100: 42. Lobov IB, Brooks PC, Lang RA. Angiopoietin-2 19. Guo P, Imanishi Y, Cackowski FC, et al. Up-regulation 8904–9. displays VEGF-dependent modulation of capillary of angiopoietin-2, matrix metalloprotease-2, membrane 31. Hu B, Jarzynka MJ, Guo P, et al. Angiopoietin 2 structure and endothelial cell survival in vivo. Proc Natl type 1 metalloprotease, and laminin 5 g2 correlates with induces glioma cell invasion by stimulating matrix Acad Sci U S A 2002;99:11205–10.

www.aacrjournals.org 4263 Cancer Res 2007; 67: (9). May 1, 2007

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Angiopoietin-2 Stimulates Breast Cancer Metastasis through the α5β1 Integrin-Mediated Pathway

Yorihisha Imanishi, Bo Hu, Michael J. Jarzynka, et al.

Cancer Res 2007;67:4254-4263.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/9/4254

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2007/04/30/67.9.4254.DC1

Cited articles This article cites 42 articles, 17 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/9/4254.full#ref-list-1

Citing articles This article has been cited by 31 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/9/4254.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/67/9/4254. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.