Tyrosine Phosphatase PTPRJ/DEP-1 Is an Essential Promoter of Vascular Permeability, Angiogenesis, and Tumor Progression

Angiogenesis and increased vessel permeability play key roles during the development of a number of pathologies including tumor growth and cancer progression (1). Therefore, much effort aims at understanding the molecular mechanisms underlying the formation and remodeling of blood vessels as a way to identify new approaches to harness these processes. The receptor-like protein tyrosine phosphatase DEP-1, also named PTPRJ or CD148, is expressed in various cell types including epithelial, hematopoietic, and endothelial cells. It encompasses an extracellular domain containing eight fibronectin-type III-like motifs, a transmembrane domain, a single intracellular catalytic domain, and a short C-terminal tail (2). DEP-1 expression is found to increase with cell density, and contributes to cell contact–mediated growth inhibition (2, 3). Similarly, overexpression of DEP-1 in many tumor cells is associated with inhibition of cell proliferation and migration (2, 4–7). Consistent with this role, some of its known substrates include growth factor receptors such as the PDGFR, EGFR, MET, and ERK1/2 (8–11).

In confluent and quiescent endothelial cells, DEP-1 colocalizes at cell–cell junctions with VEGFR2 and VE-cadherin, and negatively regulates VEGFR2 phosphorylation and cell proliferation (3, 12, 13). However, in addition to its role as an attenuator of VEGFR2 activity, DEP-1 is also a positive regulator of Src activation and of important Src kinase–dependent biological functions in VEGF-stimulated cells, including cell survival, permeability, invasion, and capillary formation (14–19). This is induced through the VEGF-mediated tyrosine-phosphorylation of the C-terminal tail of DEP-1 on Y1311 and Y1320, which allows DEP-1 to associate with Src and dephosphorylate its inhibitory Y529 (15). Interestingly, inactivation of DEP-1 in mice, via the swapping of its catalytic domain with GFP, resulted in mid-gestation death, characterized by enlarged primitive vessels, increased endothelial cell proliferation, and defective vascular remodeling and branching (20). Surprisingly, DEP-1 knockout (KO) mice are viable and fertile, with no apparent phenotype (21–23). However, further characterization of these mice models revealed that DEP-1 positively regulates B cell and macrophage immunoreceptor signaling, platelet activation, and airway smooth muscle contractility (22, 24, 25). As the underlying mechanism, DEP-1 was proposed to promote these biological responses via activation of Src family kinases (SFK), similarly to what was observed in endothelial cells in vitro (14). Therefore, to determine the role of DEP-1 during angiogenesis and permeability in vivo, the biological consequences of its lost expression were investigated in DEP-1 KO mice. We demonstrate here that DEP-1 is essential for Src activation and the phosphorylation of its substrate VE-cadherin in response to VEGF stimulation in vivo. Consequently, impaired vascular permeability, capillary formation, and recovery from hindlimb ischemia are observed in DEP-1 KO mice. Also, tumor growth and the formation of spontaneous and experimental metastases were greatly reduced in mice with no expression of DEP-1, consistent with an important reduction in tumor-associated microvessel density and defective vascular permeability in these mice. These results thus demonstrate for the first time the positive and essential role played by DEP-1 in postnatal and tumor-associated angiogenesis, and identify stroma-derived DEP-1 as an important promoter of tumor progression.

Antibodies against Src, non-p^Y529Src, p^Y1175VEGFR2, cleaved caspase-3, and horseradish peroxidase (HRP)-conjugated secondary IgGs were purchased from Cell Signaling Technology, New England Biolabs. p^Y418Src, Alexa Fluor 488–coupled donkey anti-goat, and Alexa Fluor 594-coupled donkey anti-rabbit antibodies were purchased from Invitrogen. DEP-1 goat antibody was from R&D Systems. CD31 (M-20), Ki67, PY99, and VEGFR2 antibodies were purchased from Santa Cruz Biotechnology. VE-Cadherin (11D4.1) and F4/80 (A3-1) antibodies were from BD Pharmingen (BD Biosciences) and Thermo Scientific, respectively. Recombinant human VEGF-A was obtained from the Biological Resources Branch Preclinical Repository of the National Cancer Institute - Frederick Cancer Research and Development Center. Evans blue dye and formamide were purchased from Sigma.

DEP-1 KO and wild-type (WT) mice were derived from DEP-1^+/− mice, after backcrossing the previously described DEP^+/− mice (C57/BL6J background) for at least 10 generations with FVB/N mice (21). All animal experiments were conducted in accordance with the guidelines set out by the CRCHUM Animal Experimentation Ethics Committee. The Mvt-1/Pei-1 and MET-1 mouse mammary tumor cell lines were generously provided in 2013 by Kent Hunter (NCI, Bethesda, MD) and in 2014 by Robert Cardiff (UC Davis, Davis, CA), respectively (26, 27). We did not authenticate the cells. Cells were cultured in DMEM containing 10% FBS and 50 μg/mL gentamycin (Wisent). For tumor studies, both cell lines were injected at passage 17.

Anesthetized mice were sacrificed by exsanguination 10 minutes following tail-vein injection of 2 μg of VEGF or PBS (100 μL). Lungs were collected, frozen in liquid nitrogen, crushed, and lysed in RIPA buffer pH 7.4 containing 50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1% (v/v) Triton X-100, 1% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, 1 mmol/L EDTA, 10 μg/mL aprotinin, 10 μg/mL leupeptin (Roche Applied Science), 1 mmol/L phenylmethylsulfonyl fluoride, 5 mmol/L sodium fluoride, and 1 mmol/L sodium vanadate. Protein concentration was determined using the BCA protein assay reagent (Pierce). VE-cadherin was immunoprecipitated from 500 μg of lung lysates with 3 μg of antibody overnight, and incubated for 2 hours with Protein G–conjugated Sepharose beads (GE Healthcare). Lung lysates and immunoprecipitated proteins were resolved by SDS-PAGE and transferred on nitrocellulose membranes (0.45 μm; Bio-Rad). Western blotting and ECL detection (Amersham Biosciences/GE Healthcare) were performed according to the manufacturer's recommendations. Alternatively, lungs were frozen in OCT and 5-μm cryosections were processed for immunofluorescence. Formalin-fixed sections were incubated for 1 hour with CD31 antibody (1/50), washed twice with PBS 5 minutes, and then incubated 45 minutes with Alexa Fluor 488–coupled anti-mouse antibody (1/800). After PBS washes, the same steps were followed with p^Y418Src (1/100) and Alexa Fluor 594–coupled anti-rabbit (1/800) antibodies. Slides were next washed twice with PBS and water, and mounted in ProLong Gold antifade reagent (Invitrogen/Molecular Probes). Fluorescence was visualized using a Nikon Eclipse E600 microscope and photographed with an ×20 objective lens, using a Photometrics CoolSNAP camera (Roper Scientific). Images were captured with NIS-Elements AR 3.0 software (Nikon).

Anesthetized WT or DEP-1 KO mice (8/10-week-old) were injected with Evans blue (50 mg/kg) via the tail vein. Immediately after, mice were injected intradermally on each side of the ventral midline with an equal volume of PBS, VEGF (200 ng), or histamine (625 ng). Thirty minutes later, mice were perfused with PBS and the skin was photographed. For basal permeability assays, mice were perfused with PBS 1 hour after injection of Evans blue. For the quantification of intratumor vascular permeability, mice were injected with Evans blue 21 days after injection of tumor cells and perfused with PBS after 1 hour. Sites of injection, organs, and tumors were collected, weighed, and incubated for 48 hours at 56°C in formamide to extract Evans blue. Absorbance at 600 nm was determined in triplicate and normalized on tissue weight, and on the relative concentration of Evans blue present in the blood of the corresponding mice.

Mouse aortic rings were prepared according to the method of Baker and colleagues (28). Briefly, thoracic aortas of 8-week-old WT and DEP-1 KO mice were dissected, cut into 0.5-mm rings, and incubated in Opti-MEM (Gibco, Thermo Fischer Scientific) without serum or antibiotics overnight. Rings were next embedded in Matrigel (BD #356231; BD Biosciences), and cultured in OptiMEM (2.5% FBS, 50 μg/mL gentamycin) containing PBS or VEGF (50 ng/mL). At day 7, rings were photographed. Number of sprouts and total length of sprouts were calculated using the AxioVision 4.8.2 software (Zeiss; at least in triplicate; 3 mice per group).

Experiments were performed as previously described (29). Briefly, growth factor–depleted Matrigel (400 μL) was premixed with heparin (15 U; Sandoz) and VEGF (200 ng) or PBS. The mixtures were injected subcutaneously on both sides of the ventral midline of 6-week-old WT or DEP-1 KO mice and collected 12 days later. Hemoglobin content was measured using the Quantichrom Hemoglobin Assay Kit (BioAssay Systems) and normalized to plug weight (10 animals/group). Neovascularization was determined on hematoxylin and eosin (H&E)–stained slices of paraffin-embedded plugs. Capillary structures were counted per region of interest (ROI; 5 ROI/plugs; 10 plugs/condition).

Unilateral hindlimb ischemia was surgically induced in mice as previously described (30). Hindlimb perfusion measurements were obtained with a Laser Doppler Perfusion Imager (LDPI) system (Moor Instruments Ltd.) immediately after surgery and 14 days later on anesthetized animals. The results are expressed as the ratio of perfusion in the ischemic versus nonischemic hindlimb. The mice were sacrificed at day 14 (6 animals/group). Immunohistochemistry of CD31 (Pharmigen) was performed on paraffin-embedded ischemic gastrocnemius muscle. The capillary density per muscle fiber was determined on three muscle slices taken from different parts of the muscle.

WT and DEP-1 KO female mice (6-week-old) were injected with 2 × 10⁵ Mvt-1/Pei-1 or 4 × 10⁵ MET-1 cells in the mammary fat pad and sacrificed at day 28 and 25, respectively. Tumors were fixed in formaldehyde (3.7%, v/v) and embedded in paraffin, or lysed in RIPA buffer, and then processed for IHC (two slices/tumor) or Western blotting, respectively (31, 32). For pY⁴¹⁸Src and CD31 immunofluorescence, paraffin-embedded tumor slices were dewaxed, subjected to antigen retrieval, and incubated with primary antibodies using the BenchMark XT automated stainer (Ventana Medical System Inc.; ref. 31). Afterward, on the bench, tumor slices were incubated with secondary antibodies as described above (“In vivo VEGF stimulation” section). Autofluorescence was quenched with Sudan Black (0.1%, m/v) for 15 minutes before rinsing and mounting the slides. Lung lobes were embedded in OCT, and 5-μm cryosections were H&E stained to determine the metastatic burden using Photoshop (two slices/lobe). For experimental metastasis, 8-week-old WT and KO female mice were tail-vein injected with 4 × 10⁵ Mvt-1/Pei-1 or 7 × 10⁵ MET-1 cells. On day 14, mice were sacrificed and lungs were processed as described above (at least 9 animals/group).

The GSE15363 dataset was downloaded from the NCBI Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo; ref. 33). Tumor samples from breast cancer patients were segregated in two groups based (i) on microvessel density (low and high), as determined by PECAM (CD31) immunostaining; or (ii) based on a distinct genetic signature defining vessels as either proliferative and undergoing active remodeling (group A), or as being more stable and mature (group B; ref. 33). DEP-1 expression (probe 25355) was evaluated in each group and plotted on box and whiskers graphs. Analyses were performed with the GraphPad Prism 6.0 software (GraphPad Software Inc.).

Statistical significance was evaluated with the Student t test using GraphPad Prism 6.0 software. P values were considered significant when less than 0.05. Densitometry analyses were performed with the Quantity One 4.6.3 software (Bio-Rad).

Despite negative regulatory functions associated with high levels of DEP-1 expression in endothelial cells, our group previously demonstrated that VEGF-induced tyrosine phosphorylation of moderately expressed DEP-1 promotes Src activation (14, 15). This was shown to occur through a phospho-displacement mechanism allowing phosphorylated DEP-1 to bind the Src SH2 domain and dephosphorylate its inhibitory Y529, resulting in the phosphorylation of its activating Y418 residue (14, 15). Here, to determine whether DEP-1 was also involved in Src activation upon VEGF stimulation in vivo, WT and DEP-1 KO mice were systemically injected with VEGF or PBS, and lungs were collected and lysed after 10 minutes. VEGF stimulation induced the dephosphorylation of Src Y529 and the phosphorylation of Y418 in WT mice, but this was attenuated in lysates from DEP-1 KO mice (Fig. 1A and B). Consistent with this, the tyrosine phosphorylation of immunoprecipitated VE-cadherin, an endothelial cell–specific Src substrate, was inhibited upon VEGF stimulation in DEP-1 KO mice (Fig. 1C). Moreover, dephosphorylation of VE-cadherin–associated Src on Y529 was also defective, further emphasizing the inhibition of endothelial Src activity (Fig. 1C). In contrast, the relative phosphorylation of VEGFR2, a DEP-1 substrate (3, 14), was maintained or slightly increased in DEP-1 KO mice (Fig. 1A and B), demonstrating that reduced Src activation was not resulting from impaired VEGFR2 phosphorylation. Immunofluorescence staining of p^Y418Src on lung cryosections confirmed that VEGF-stimulated Src activation was reduced in CD31-positive endothelial cells of the DEP-1 KO mice (Fig. 1D and E). Together, these results demonstrate the essential role of DEP-1 in VEGF-dependent Src activation in vivo.

VEGF was originally described for its vascular permeability inducing capacity, which relies on its ability to induce the remodeling and loosening of intercellular adhesions (34). DEP-1 and Src mediate the tyrosine and PAK2-dependent serine phosphorylation of VE-cadherin, resulting in the weakening of its intracellular association with catenins and its increased internalization from cell–cell junctions (15, 35–39). As impaired phosphorylation of Src and VE-Cadherin was observed in DEP-1 KO mice, we characterized the role of DEP-1 in VEGF-induced vascular permeability in vivo using a modified Miles assay. In control mice, dermal injection of VEGF led to an important vascular leakage of Evans blue dye, which was blocked in DEP-1 KO animals (Fig. 2A and B). However, we observed no difference in histamine-induced vascular permeability, shown to be Src independent (Fig. 2C and D; refs. 16, 40). Basal vascular permeability in spleen, liver, kidney, and brain was also equivalent in DEP-1 KO and WT mice (Fig. 2E). These results importantly demonstrate that DEP-1 is essential for the promotion of VEGF-induced vascular permeability in vivo, and correlate with its ability to activate the Src-dependent phosphorylation of VE-cadherin.

Angiogenesis is a complex process requiring endothelial cell proliferation, remodeling of endothelial cell–cell junctions, and cell invasion (41). We have previously shown that DEP-1 expression is required for endothelial cell invasion and the organization of capillary-like structures when grown on Matrigel (15). Src has been shown to mediate the remodeling and loosening of cell–cell junctions, and to promote angiogenesis and vascular permeability in vivo (16, 18, 35). To characterize the consequences of DEP-1 loss on capillary formation, several approaches were used. First, in a mouse aortic ring assay, an important induction of capillary formation was observed in response to the stimulation of WT aortic rings embedded in Matrigel with VEGF (Fig. 3A). In contrast, a reduced number of shorter capillaries elongated from the DEP-1 KO aortic rings (Fig. 3A–C). VEGF-induced capillary formation in vivo was also studied using a Matrigel plug assay. Upon harvesting, Matrigel plugs were photographed (Fig. 3D), and the hemoglobin content was determined. Quantification indicated that VEGF induced the formation of functional capillaries in the WT mice, but not in the KO animals (Fig. 3E). Capillary density, as characterized by H&E staining of paraffin-embedded sections of the plugs (Fig. 3F), was increased in response to VEGF stimulation in the WT animals, whereas no induction was detected in the Matrigel plugs from the DEP-1 KO mice (Fig. 3G). Taken together, these experiments demonstrate that DEP-1 is positively regulating the formation of perfused capillaries in vivo.

Ischemia is the most common angiogenic signal emerging from wounded tissues. To study the physiologic angiogenic response of DEP-1 KO mice, we exploited the hindlimb ischemia model. In this system, angiogenesis is the primary response, leading to the improved collateral blood flow detected by LDPI. Fourteen days after the removal of the femoral artery from the left hindlimb, LDPI measurements revealed a reduction of limb perfusion in DEP-1 null mice compared with WT mice (Fig. 4A and B), whereas a similar reduction of perfusion was observed in all animals immediately after surgery (data not shown). In addition, staining of paraffin-embedded ischemic gastrocnemius muscles with CD31 detected a reduction of capillaries per muscle fiber (Fig. 4C and D). Consistent with their decreased limb perfusion recovery, DEP-1 KO mice were less mobile than their WT littermates (Fig. 4E), and had a greater number of necrotic or inflamed fingers (Fig. 4F). These data demonstrate that DEP-1 expression significantly enhances the promotion of physiologic neovascularization in the context of ischemia-induced tissue damage.

The recruitment of blood vessels is essential for tumor growth and contributes to metastasis. We studied the role of DEP-1 in cancer growth and progression using two highly metastatic and syngeneic tumor cell models, Mvt-1/Pei-1 and MET-1 mammary tumor cells, following mammary fat pad injection in WT and DEP-1 KO mice (26, 27). Tumor growth, as characterized by tumor volume and tumor weight, was significantly decreased in DEP-1 KO mice compared with WT animals (Fig. 5A, B, G, and H). Consistent with the reduced angiogenesis previously observed in the DEP-1 KO mice (Figs. 3 and 4), CD31 staining also revealed a decrease in tumor microvessel density (Fig. 5C, D, I and J). Accordingly, decreased proliferation and increased apoptosis of tumor cells were detected in DEP-1 KO mice, as shown by reduced Ki67 immunostaining and increased caspase-3 cleavage, respectively (Fig. 5C, E, F, I, K, and L). These results thus suggest that DEP-1 is a driver of tumor growth by regulating tumor-associated angiogenesis.

Tumor microvessels also facilitate the recruitment of immune cells and the dissemination of cancer cells. Tumor infiltration by macrophages was impaired in DEP-1 KO mice, as shown by the decreased F4/80 immunostaining (Fig. 5C and I). This may have resulted from the reduced permeability of tumors grown in DEP-1 KO mice (Fig. 6A), and/or to cell-autonomous effects of DEP-1 on macrophage function (22, 42). Importantly, in the absence of DEP-1, fewer Mvt-1 tumor cells formed spontaneous lung metastases (Fig. 6B). However, for MET-1 cells, only one WT mice presented metastatic lesions at the time of sacrifice (data not shown). Mvt-1 and MET-1 cells were also injected directly into the blood circulation to investigate experimental metastasis. Their ability to colonize the lungs of DEP-1 KO mice was greatly diminished (Fig. 6B). As a postulated mechanism, the reduced tumor angiogenesis and metastasis observed in the DEP-1 KO mice were shown to correlate with weaker Src activation levels in the residual tumor microvessels, reinforcing the key role of DEP-1 in Src activation during these processes (Fig. 7A and B). Interestingly, in breast cancer patients, higher DEP-1 expression in tumor-associated microvessels was associated with greater vascularization of the tumors, and with a genetically defined tumor vascular subtype (A) described as being actively remodeling (33), further supporting a proangiogenic role for DEP-1 in tumor-associated angiogenesis (Fig. 7C and D).

DEP-1 has mainly been depicted as having negative regulatory functions in a broad range of cell types. Despite reports highlighting inhibitory roles for DEP-1 in endothelial cell proliferation and capillary formation (3, 43, 44), our group previously showed that endogenously expressed DEP-1 in nonquiescent endothelial cells positively regulates Src activation, cell invasion, permeability, and capillary formation (14, 15, 19). We now show that VEGF-induced permeability (Fig. 2) and angiogenesis (Figs. 3 and 4), whether in response to VEGF or ischemia, are inhibited in a DEP-1 KO mouse model. In addition, tumor-associated angiogenesis, tumor growth, and metastasis are also impaired in these conditions (Figs. 5 and 6), establishing stroma-derived DEP-1 as a promoter of cancer progression. At the molecular level, this correlates with the inhibition of VEGF-induced Src activation in vivo, and consequently, with the abrogated phosphorylation of its endothelial cell–specific substrate VE-cadherin, both critical mediators of vascular permeability and angiogenesis (Fig. 1; refs. 16, 18, 34). These results thus demonstrate for the first time that DEP-1 expression has positive and essential regulatory functions in vivo that mediate increased vascular permeability and angiogenesis (14, 15).

One of the main functions promoted by DEP-1 is the activation of SFKs and downstream biological responses. This has now been reported in hematopoietic cells, platelets, smooth muscle cells, breast tumor cells, and endothelial cells in vitro, and relies on the ability of DEP-1 to dephosphorylate the inhibitory C-terminal tyrosine residue of SFKs (14, 15, 22, 24, 25, 32). Here, we identify DEP-1 as a key mediator of VEGF-dependent Src activation in vivo (Fig. 1). Although this was shown in lung lysates and tissues where several cell types are present, an important fraction represents endothelial cells. Thus, the observation that its endothelial cell–specific substrate, VE-cadherin, is no longer tyrosine phosphorylated in DEP-1 KO mice, and that decreased Src Y418 phosphorylation colocalizes with CD31-positive regions in lung tissues strongly suggest that endothelial Src activity was inhibited in the absence of DEP-1 expression. Moreover, VE-cadherin–associated Src, which represents the main pool of VEGF-activated Src in endothelial cells (12, 45), was also not dephosphorylated on Y529 (Fig. 1C; refs. 12, 14). These results thus clearly identify DEP-1 as a critical regulator of Src activation and VE-cadherin phosphorylation in VEGF-stimulated endothelial cells in vivo. They are also consistent with the reduced induction of vascular permeability mediated by VEGF in the DEP-1–deficient mice (Fig. 2). In contrast to our results, VEGF was recently reported to induce similar levels of vascular permeability in WT and DEP-1–deficient mice (46). The reason for this discrepancy remains unclear, but could possibly be explained by experimental differences or the use of a different mouse strain.

Because SFKs act as pivotal regulators of VEGF-dependent permeability and angiogenesis (16, 18, 35, 47), our results suggest that the dramatic block of these biological responses in the DEP-1 KO mice is due to the abrogation of the Src pathway in endothelial cells. The Src-dependent phosphorylation of VE-cadherin is critical for the reorganization of cell–cell junctions that occur during permeability (35–37, 39). Expression of a catalytically active DEP-1 mutant (Y1311F/Y1320F), defective in Src activation, was also unable to promote any of these biological activities in endothelial cells in vitro (15). This then strongly suggests that through its ability to activate Src, DEP-1 is a key regulator of postnatal angiogenesis and permeability. However, because in the Src KO mice, only permeability and metastasis were affected (16), but that the inhibition of all SFKs was required to block capillary formation in Matrigel and in tumors (16), we also conclude that other SFKs expressed in endothelial cells might be regulated by DEP-1 and contribute to the phenotypes observed. Notably, the Fyn kinase mediates the endothelial migratory response to VEGF stimulation and was shown to be regulated by DEP-1 in T cells (15, 48, 49), suggesting it could also be part of the pathways involved.

Our results also interestingly revealed that expression of DEP-1 in tumor blood vessels correlated with Src activation (Fig. 7), and with the development of highly vascularized tumors (Fig. 5). Thus, in DEP-1 KO mice, defective angiogenesis impaired tumor growth, consistent with the decreased tumor cell proliferation and increased apoptosis observed (Fig. 5). As these tumors had fewer vessels and were less permeable, spontaneous metastasis was greatly reduced (Fig. 6). However, injection of Mvt-1 and MET-1 mammary tumor cells into the blood circulation of DEP-1 KO mice similarly resulted in their impaired capacity to metastasize to the lungs, suggesting that defective VEGF-induced vascular permeability in these animals impaired their ability to extravasate in the lungs. Consistent with this new role for DEP-1 in the promotion of tumor-associated angiogenesis, DEP-1 gene expression levels in the tumor vasculature of a cohort of breast cancer patients (33) was found to be elevated in highly vascularized tumors, and in a genetically distinct tumor vascular subtype defined as being actively remodeling (Fig. 7). These observations thus further support our conclusions that DEP-1 is an important mediator of angiogenesis in vivo.

We also observed that the recruitment of macrophages to the tumors grown in the DEP-1 KO mice was impaired. This could possibly result from the decreased vessel density and permeability of tumors grown in these mice (Figs. 5 and 6). Also, this could be due to a cell autonomous effect as CSF-1–induced chemotaxis of macrophages in vitro is partially mediated by DEP-1 (42). The infiltration of tumors by macrophages can contribute to the promotion of angiogenesis and cancer progression, in part via the secretion of VEGF (50). However, as Mvt-1 tumor cells secrete high levels of VEGF (26), no decrease in VEGF tumor levels were observed in DEP-1 KO mice (Supplementary Fig. S1). Thus, although the impaired recruitment of macrophages may have contributed to the tumor phenotype observed in DEP-1 KO mice, the impaired capillary formation observed during Matrigel and aortic ring assays is most likely the primary reason for defective tumor-associated angiogenesis.

The work presented here thus brings novel insight into our comprehension of the molecular machinery regulating permeability and angiogenesis in vivo, and importantly establish stroma-derived DEP-1 as a critical promoter of tumor cell growth and metastasis. Recently, breast tumor-expressed DEP-1 was shown to mediate a Src proinvasive signaling pathway associated with the promotion of metastasis (32). These findings together with those presented here therefore strongly suggest that inhibiting DEP-1 functions could provide a new therapeutic strategy to counteract cancer progression.

No potential conflicts of interest were disclosed.

Conception and design: P. Fournier, I. Royal

Development of methodology: P. Fournier

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Fournier, S. Dussault, A. Fusco, A. Rivard

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Fournier, A. Rivard

Writing, review, and/or revision of the manuscript: P. Fournier, A. Rivard, I. Royal

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Rivard

Study supervision: I. Royal

We thank Jean-Philippe Gratton and Sara Weis for providing advice on Matrigel and permeability assays, Kent Hunter and Robert Cardiff for providing cell lines, and Véronique Barrès of the Molecular Pathology core facility of the CRCHUM for performing the IHC experiments.

This work was supported by the Cancer Research Society (I. Royal) and the Canadian Institutes of Health Research (MOP-93681 to I. Royal and MOP-123490 to A. Rivard). P. Fournier was supported by student scholarships from Fonds de Recherche en Santé – Québec (25988), the Canadian Institutes of Health Research (292353), and Institut du cancer de Montréal.

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 with 18 U.S.C. Section 1734 solely to indicate this fact.