Tumor survival depends in part on the ability of tumor cells to transform the surrounding extracellular matrix (ECM) into an environment conducive to tumor progression. Matricellular proteins are secreted into the ECM and impact signaling pathways that are required for pro-tumorigenic activities such as angiogenesis. Fibulin-5 (Fbln5) is a matricellular protein that was recently shown to regulate angiogenesis; however, its effect on tumor angiogenesis and thus tumor growth is currently unknown. We report that the growth of pancreatic tumors and tumor angiogenesis is suppressed in Fbln5-null (Fbln5−/−) mice compared with wild-type (WT) littermates. Furthermore, we observed an increase in the level of reactive oxygen species (ROS) in tumors grown in Fbln5−/− animals. Increased ROS resulted in elevated DNA damage, increased apoptosis of endothelial cells within the tumor, and represented the underlying cause for the reduction in angiogenesis and tumor growth. In vitro, we identified a novel pathway by which Fbln5 controls ROS production through a mechanism that is dependent on β1 integrins. These results were validated in Fbln5RGE/RGE mice, which harbor a point mutation in the integrin-binding RGD motif of Fbln5, preventing its interaction with integrins. Tumor growth and angiogenesis was reduced in Fbln5RGE/RGE mice, however treatment with an antioxidant rescued angiogenesis and elevated tumor growth to WT levels. These findings introduce a novel function for Fbln5 in the regulation of integrin-induced ROS production and establish a rationale for future studies to examine whether blocking Fbln5 function could be an effective anti-tumor strategy, alone or in combination with other therapies.

The tumor microenvironment contains many components that influence tumor survival and progression; namely, activated fibroblasts, infiltrating macrophages, neutrophils and endothelial cells that are recruited into the tumor mass and actively remodel the extracellular matrix (ECM) (Farrow et al., 2008; Mantovani et al., 2008). The acellular component of the tumor microenvironment, the ECM, consists of a three-dimensional network of structural proteins such as fibronectin and collagen that function as an architectural scaffold for cells, and provides the context for cell function and tissue development. The organization of, and the interaction between, these components within the tumor microenvironment are crucial determinants of tumor survival. Interactions between cells and the ECM stimulate signaling pathways that regulate cellular activities such as survival, proliferation and migration – processes that are crucial to tumor expansion (Albig and Schiemann, 2005; Li et al., 2007). Matricellular proteins, such as fibulin-5 (Fbln5), associate with the matrix and regulate the crosstalk between cells and their surrounding ECM. Numerous studies have shown that matricellular proteins can have strong effects on tumor growth. For example, tumor progression is generally enhanced in mice lacking the matricellular proteins SPARC (secreted protein acidic and rich in cysteine) or thrombospondin-1 or -2 (Brekken et al., 2003; Lawler and Detmar, 2004; Puolakkainen et al., 2004). By contrast, papilloma development is delayed in osteopontin-null mice (Hsieh et al., 2006). It is clear from these studies that, with further investigation, matricellular proteins could be valuable targets for controlling cancer growth and spread (Pozzi et al., 2000; Proia and Kuperwasser, 2005). Therefore, we are interested in determining the effect of the matricellular protein Fbln5 on tumor growth.

Fbln5 is a member of the fibulin family of proteins and consists of six calcium-binding epidermal growth factor-like repeats, one of which contains an integrin-binding RGD motif and a COOH-terminal fibulin-type module (Timpl et al., 2003; Zheng et al., 2007). Fbln5 is crucial for proper assembly of elastic fibers as it tethers tropoelastin to microfibrils where it is crosslinked by lysyl oxidase to form mature elastic fibers (Liu et al., 2004; Zheng et al., 2007). Mice that are deficient in Fbln5 have fragmented and disorganized elastic fibers leading to loose skin, vascular anomalies and alveolar defects (Yanagisawa et al., 2002). Fbln5 protein is secreted into the ECM by fibroblasts, vascular smooth muscle cells (VSMCs) and endothelial cells, and associates with cells through integrins αVβ3, α4β1 and α5β1 (Lomas et al., 2007; Timpl et al., 2003). Recent studies have suggested that Fbln5 can regulate angiogenesis. In vitro experiments showed that Fbln5 antagonized vascular endothelial growth factor (VEGF) signaling thereby abrogating VEGF-induced proliferation and migration of endothelial cells (Albig et al., 2006; Albig and Schiemann, 2004). In addition, in vivo studies indicated an increase in vascular invasion into polyvinyl sponges that were implanted into Fbln5−/− mice compared with wild-type (WT) animals (Sullivan et al., 2007). The exact mechanism by which Fbln5 exerts its effect on angiogenesis is unknown. Given these results and the aberrant vessel formation in Fbln5-null animals, we sought to determine whether tumor growth is altered in Fbln5−/− mice owing to effects on angiogenesis.

Herein, we report the first results from tumor studies performed in Fbln5−/− mice. Interestingly, the growth of an implanted murine pancreatic cancer cell line (Pan02) was decreased in mice lacking endogenous Fbln5. Moreover, blood vessel density in the tumor microenvironment of Fbln5−/− mice was reduced significantly compared with tumors grown in WT littermates. This effect was the result of elevated reactive oxygen species (ROS) production within tumors from Fbln5−/− mice, which increased oxidative damage and augmented endothelial cell death. This led us to the discovery that, in vitro, Fbln5 regulates ROS production in both fibroblasts and endothelial cells through a mechanism that is dependent on β1 integrin binding to fibronectin. We reproduced these findings in vivo by performing tumor studies in mice expressing a form of Fbln5 that is deficient in the ability to bind to integrins (Fbln5RGE/RGE mice). Likewise, we observed a reduction in tumor growth and angiogenesis in these Fbln5RGE/RGE mice, and again observed increased levels of ROS. Finally, when Fbln5RGE/RGE mice were treated with an antioxidant, blood vessel density increased and tumor growth was restored. These results identify a novel function for Fbln5 in the regulation of integrin-induced ROS production and give us insight in to how Fbln5 affects angiogenesis.

Pancreatic tumor growth is reduced in Fbln5−/− mice

To evaluate tumor growth in Fbln5−/− mice, Pan02 cells, a murine pancreatic carcinoma cell line, were injected subcutaneously into WT and Fbln5−/− mice. The subcutaneous tumor volume was recorded weekly throughout the experiment. Tumors grew significantly slower in Fbln5−/− mice than tumors in WT littermates (Fig. 1A). After four weeks, the mice were sacrificed and the tumors weighed. The tumors in Fbln5−/− mice were smaller than the tumors from WT animals (Fig. 1B). This result was recapitulated in our pancreatic orthotopic model where Pan02 cells were injected directly into the tail of the pancreas (Fig. 1C). Tumors were first palpable two weeks after tumor cell injection into the pancreas. At four weeks, animals were sacrificed and the pancreas with the tumor was harvested and weighed. The final orthotopic tumor weights in Fbln5−/− mice were again smaller when compared with tumors grown in WT littermates (Fig. 1C). We also observed a decrease in the number of peritoneal and lymph node metastases in Fbln5−/− mice (Fig. 1D). Histological analysis of orthotopic tumors from WT and Fbln5−/− animals revealed no striking differences in morphology, tissue architecture or fibrillar collagen deposition (supplementary material Fig. S1). As a control, we evaluated the expression of Fbln5 in Pan02 cells by reverse transcription (RT)-PCR and western blot analysis, and determined that Pan02 cells do not express Fbln5. This was confirmed by immunohistochemistry staining showing that tumors from WT animals express high levels of Fbln5, whereas tumors from Fbln5−/− mice were devoid of any Fbln5 immunoreactivity. These results indicated that only host cells found within the tumor expressed Fbln5 and not the tumor cells (supplementary material Fig. S1). Importantly, a significant portion of the Fbln5 present in tumors from WT mice was localized to blood vessels and perivascular areas (supplementary material Fig. S1).

Proliferation decreased and apoptosis increased in tumors from Fbln5−/− mice

A disruption in tumor growth is often caused by a decrease in the proliferation rate of cells that make up a tumor, along with an increase in the death of these cells. To further confirm that tumor growth in Fbln5−/− mice was attenuated, we examined these characteristics in Pan02 tumors from WT and Fbln5−/− animals. Immunohistochemistry with an antibody that is specific for phosphorylated histone H3 revealed a significant decrease in the number of proliferating cells in tumors grown in the absence of Fbln5 (Fig. 2). By contrast, TUNEL (TdT-mediated dUTP nick-end labeling) analysis showed a significant increase in the number of apoptotic cells in tumors grown in Fbln5−/− animals compared with WT mice (Fig. 2). These results correlate with the anticipated characteristics of smaller, slower growing tumors such as those seen in Fbln5-deficient animals.

Angiogenesis is decreased in tumors from Fbln5−/− mice

Sustained tumor growth requires angiogenesis to facilitate nutrient delivery to the tumor mass. Given this knowledge and the recent studies suggesting that Fbln5 is a regulator of angiogenesis, we evaluated vascular density in tumors from Fbln5−/− mice to determine whether the decrease in tumor growth was because of an inadequate blood supply (Sullivan et al., 2007). We examined the blood vessel density in tumors by immunohistochemistry using the endothelial cell marker MECA-32 (Fig. 3A). The number of blood vessels per mm2 was significantly lower in subcutaneous (Fig. 3A, top panels) and orthotopic (bottom panels) tumors grown in Fbln5−/− mice compared with WT animals. These results suggested that the reduced tumor growth observed in Fbln5−/− mice was caused, in part, by the inability of tumors to develop and maintain a sufficient blood vessel network.

Fig. 1.

Pancreatic tumor growth is reduced inFbln5−/−mice. Murine pancreatic carcinoma cells (Pan02) were injected subcutaneously (3×106 cells) (A,B) or orthotopically (5×105 cells) (C,D) into WT and Fbln5−/− mice. (A)Mean subcutaneous tumor volume ± standard error of the mean (s.e.m.) of a representative experiment (n=5 for WT and Fbln5−/−). (B)Mean final subcutaneous tumor weight ± s.e.m. for three individual experiments (WT, n=18; Fbln5−/−, n=19). (C) The mean tumor/pancreas weight ± s.e.m. and (D) the number of observed metastatic events (Farrow et al., 2008) from WT (n=7) and Fbln5−/− (n=6) animals injected orthotopically with Pan02 cells. *P<0.05, **P<0.01.

Fig. 1.

Pancreatic tumor growth is reduced inFbln5−/−mice. Murine pancreatic carcinoma cells (Pan02) were injected subcutaneously (3×106 cells) (A,B) or orthotopically (5×105 cells) (C,D) into WT and Fbln5−/− mice. (A)Mean subcutaneous tumor volume ± standard error of the mean (s.e.m.) of a representative experiment (n=5 for WT and Fbln5−/−). (B)Mean final subcutaneous tumor weight ± s.e.m. for three individual experiments (WT, n=18; Fbln5−/−, n=19). (C) The mean tumor/pancreas weight ± s.e.m. and (D) the number of observed metastatic events (Farrow et al., 2008) from WT (n=7) and Fbln5−/− (n=6) animals injected orthotopically with Pan02 cells. *P<0.05, **P<0.01.

Fig. 2.

The absence of host Fbln5 results in increased apoptosis and decreased proliferation in Pan02 tumors. Formalin-fixed sections of subcutaneous tumors from WT (n=4) and Fbln5−/− (n=3) mice were evaluated for the levels of apoptosis by TUNEL staining (TUNEL, top panels) and cell proliferation by immunohistochemistry for phosphorylated histone H3 (Phospho-H3, bottom panels). The number of TUNEL- or phospho-H3-positive cells was determined by fluorescence microscopy using tumor slides counterstained with DAPI. Metamorph imaging software was used to quantify the percentage signal above a threshold, which was set by evaluating sections that were treated with secondary antibody alone. Ten pictures (total magnification, 200×) were taken from each tumor for quantification. The results shown are mean±s.e.m. *P<0.01, **P<0.001, Student’s t-test. Bars, 50 μm.

Fig. 2.

The absence of host Fbln5 results in increased apoptosis and decreased proliferation in Pan02 tumors. Formalin-fixed sections of subcutaneous tumors from WT (n=4) and Fbln5−/− (n=3) mice were evaluated for the levels of apoptosis by TUNEL staining (TUNEL, top panels) and cell proliferation by immunohistochemistry for phosphorylated histone H3 (Phospho-H3, bottom panels). The number of TUNEL- or phospho-H3-positive cells was determined by fluorescence microscopy using tumor slides counterstained with DAPI. Metamorph imaging software was used to quantify the percentage signal above a threshold, which was set by evaluating sections that were treated with secondary antibody alone. Ten pictures (total magnification, 200×) were taken from each tumor for quantification. The results shown are mean±s.e.m. *P<0.01, **P<0.001, Student’s t-test. Bars, 50 μm.

Previous results have reported an increase in vessel sprouting in Fbln5−/− animals (Sullivan et al., 2007). Therefore, we next asked whether the reduction in angiogenesis that we observed was specific to the tumor microenvironment. To answer this, we analyzed blood vessel density in the pancreas of non-tumor-bearing WT and Fbln5−/− animals. Consistent with results described previously, we observed an increase in the number of vessels in the pancreas of non-tumor-bearing Fbln5−/− animals compared with WT animals (Fig. 3B). These results indicated that the reduction in blood vessels observed in tumors grown in Fbln5−/− mice was due to a characteristic solely of the tumor microenvironment and not present under non-pathological conditions.

Increased ROS level in tumors grown in Fbln5−/− mice

We next sought a mechanism for the specific reduction of angiogenesis in the tumor microenvironment. Recently, Fbln5 has been shown to interact with superoxide dismutase 3 (SOD3), a protein that is important in the breakdown of superoxides, a common ROS (Nguyen et al., 2004). It was reported that, in aortas of Fbln5−/− mice, the loss of Fbln5 reduced the amount of available SOD3 and caused a build-up of ROS within the aorta. Therefore, we hypothesized that tumors grown in Fbln5−/− mice would also have an increase in the level of ROS. To test this, we examined ROS levels in tumors from WT and Fbln5−/− mice. Dihydroethidium (DHE)-based analysis revealed a significant increase in ROS in tumors grown in Fbln5−/− mice compared with tumors from WT mice (Fig. 4A). However, the expression and activity of SOD3 was not decreased in tumors from Fbln5−/− mice compared with WT mice (data not shown), suggesting that the accumulation of ROS is not because of a loss of SOD3 expression.

Fig. 3.

Vessel density is decreased in tumors grown inFbln5−/−mice. (A) Sections from subcutaneous and orthotopic tumors were stained with MECA-32 (green), and counterstained with DAPI (blue), to analyze blood vessel density. The numbers of MECA-32-positive structures were counted in a minimum of six microscopic fields (total magnification, 200×) from each tumor (at least three tumors per genotype) in both the subcutaneous (top panels) and orthotopic (bottom panels) settings. *P<0.05, ***P<0.0001, Student’s t-test. Mean blood vessel density ± standard deviation (s.d.). (B) MECA-32-positive blood vessel density in non-tumor-bearing normal pancreas from WT and Fbln5−/− mice was determined by immunohistochemistry. Mean blood vessel density ± s.e.m. is shown. *P<0.05. Bars, 50 μm.

Fig. 3.

Vessel density is decreased in tumors grown inFbln5−/−mice. (A) Sections from subcutaneous and orthotopic tumors were stained with MECA-32 (green), and counterstained with DAPI (blue), to analyze blood vessel density. The numbers of MECA-32-positive structures were counted in a minimum of six microscopic fields (total magnification, 200×) from each tumor (at least three tumors per genotype) in both the subcutaneous (top panels) and orthotopic (bottom panels) settings. *P<0.05, ***P<0.0001, Student’s t-test. Mean blood vessel density ± standard deviation (s.d.). (B) MECA-32-positive blood vessel density in non-tumor-bearing normal pancreas from WT and Fbln5−/− mice was determined by immunohistochemistry. Mean blood vessel density ± s.e.m. is shown. *P<0.05. Bars, 50 μm.

It is well known that, owing to the rapid metabolic rate of tumor cells, the tumor microenvironment exists in a high oxidative state containing elevated levels of ROS compared with normal tissue (Das, 2002). Knowing this, we hypothesized that the additional increase in ROS due to the loss of Fbln5 would lead to a state of chronic oxidative stress within the tumor microenvironment. Chronic oxidative stress would result in more oxidative damage; therefore, we analyzed the level of two markers of oxidative damage, 8-hydroxydeoxyguanosine (8-OHdG) and phosphorylated histone H2A (γH2AX) (Tanaka et al., 2006; Valavanidis et al., 2009). Correlating with the increase in ROS, we observed a significant increase in the level of 8-OHdG and γH2AX in tumors grown in Fbln5−/− mice compared with tumors from WT animals (Fig. 4B,C). Low levels of ROS are well tolerated by endothelial cells; however, chronic oxidative stress induces endothelial cell dysfunction and death (Touyz and Schiffrin, 2004). To determine whether the elevated production of ROS in tumors from Fbln5−/− mice was affecting endothelial cell survival, we examined the amount of endothelial cell death in tumors from WT and Fbln5−/− animals by analysis of cleaved caspase 3. We observed a significant increase in the number of apoptotic endothelial cells in tumors from Fbln5−/− mice compared with WT animals (Fig. 4D). Therefore, we proposed that the decrease in blood vessel density observed in tumors from Fbln5−/− mice was due, in part, to increased oxidative stress and damage on endothelial cells making them unable to survive in the tumor microenvironment.

Fig. 4.

Oxidative stress is elevated in tumors grown inFbln5−/−mice. (A) Fresh frozen sections of subcutaneous Pan02 tumors from WT (n=3) and Fbln5−/− (n=3) animals were incubated with dihydroethidium (DHE) for in situ detection of ROS. The resulting DHE-mediated fluorescence was quantified by Metamorph software. A representative image from tumors grown in WT and Fbln5−/− animals is shown. The results shown are the mean fluorescence intensity (±s.e.m.) from at least five 200×fields from each tumor. (B,C) The presence of DNA damage in frozen sections of subcutaneous Pan02 tumors from WT (n=5) and Fbln5−/− (n=7) animals was detected by immunohistochemistry with an antibody specific for 8-OHdG (B) and γH2AX (C) followed by development with a fluorescently conjugated secondary antibody. A representative image from tumors grown in WT and Fbln5−/− animals is shown. The results shown are the mean fluorescence intensity (±s.e.m.) from at least five 400×fields from each tumor stained. (D) Pan02 tumors from WT (n=3) and Fbln5−/− (n=3) mice were sectioned and stained with MECA-32 (green) for blood vessel identification and co-localized with anti-cleaved caspase 3 antibody (red) to determine apoptotic endothelial cells. The total number of dying endothelial cells was determined by hand counting and graphed as a percentage of the total number of apoptotic cells. Arrowheads indicate double positive cells. Ten pictures were taken from each tumor section (total magnification, 200×) for analysis. Results shown are ±s.e.m. *P<0.05, **P<0.01. Bars, 50 μm.

Fig. 4.

Oxidative stress is elevated in tumors grown inFbln5−/−mice. (A) Fresh frozen sections of subcutaneous Pan02 tumors from WT (n=3) and Fbln5−/− (n=3) animals were incubated with dihydroethidium (DHE) for in situ detection of ROS. The resulting DHE-mediated fluorescence was quantified by Metamorph software. A representative image from tumors grown in WT and Fbln5−/− animals is shown. The results shown are the mean fluorescence intensity (±s.e.m.) from at least five 200×fields from each tumor. (B,C) The presence of DNA damage in frozen sections of subcutaneous Pan02 tumors from WT (n=5) and Fbln5−/− (n=7) animals was detected by immunohistochemistry with an antibody specific for 8-OHdG (B) and γH2AX (C) followed by development with a fluorescently conjugated secondary antibody. A representative image from tumors grown in WT and Fbln5−/− animals is shown. The results shown are the mean fluorescence intensity (±s.e.m.) from at least five 400×fields from each tumor stained. (D) Pan02 tumors from WT (n=3) and Fbln5−/− (n=3) mice were sectioned and stained with MECA-32 (green) for blood vessel identification and co-localized with anti-cleaved caspase 3 antibody (red) to determine apoptotic endothelial cells. The total number of dying endothelial cells was determined by hand counting and graphed as a percentage of the total number of apoptotic cells. Arrowheads indicate double positive cells. Ten pictures were taken from each tumor section (total magnification, 200×) for analysis. Results shown are ±s.e.m. *P<0.05, **P<0.01. Bars, 50 μm.

Fig. 5.

Fbln5−/−mouse embryonic fibroblasts produced higher levels of ROS in a fibronectin- and β1 integrin-dependent fashion. MEFs harvested from WT and Fbln5−/− mice were grown on fibronectin or collagen. (A,B) After 24 hours, the cells were stained with DCF-DA (green) to detect ROS. Fbln5−/− MEFs plated on fibronectin expressed higher levels of ROS than WT MEFs grown on fibronectin or Fbln5−/− MEFs grown on collagen. ROS production by Fbln5−/− MEFs was inhibited by the addition of recombinant full-length Fbln5 (rFbln5), but not by mutant Fbln5 where the RGD domain had been deleted. DCF-DA staining was quantified by normalizing to cell number by DAPI staining (blue). Ten pictures (total magnification, 200×) were taken from each condition and used for analysis. The assay was performed in triplicate. *P<0.05. (C) ROS production was reduced by blocking β1 activation, but was not affected by a control antibody (IgG). The level of ROS was determined by incubating with DCF-DA. Ten pictures (total magnification, 200×) were taken of each condition and used for analysis. DCF-DA staining was normalized to cell number by DAPI staining. The assay was performed in triplicate. The results shown are mean±s.e.m. *P<0.05. Bars, 50 μm.

Fig. 5.

Fbln5−/−mouse embryonic fibroblasts produced higher levels of ROS in a fibronectin- and β1 integrin-dependent fashion. MEFs harvested from WT and Fbln5−/− mice were grown on fibronectin or collagen. (A,B) After 24 hours, the cells were stained with DCF-DA (green) to detect ROS. Fbln5−/− MEFs plated on fibronectin expressed higher levels of ROS than WT MEFs grown on fibronectin or Fbln5−/− MEFs grown on collagen. ROS production by Fbln5−/− MEFs was inhibited by the addition of recombinant full-length Fbln5 (rFbln5), but not by mutant Fbln5 where the RGD domain had been deleted. DCF-DA staining was quantified by normalizing to cell number by DAPI staining (blue). Ten pictures (total magnification, 200×) were taken from each condition and used for analysis. The assay was performed in triplicate. *P<0.05. (C) ROS production was reduced by blocking β1 activation, but was not affected by a control antibody (IgG). The level of ROS was determined by incubating with DCF-DA. Ten pictures (total magnification, 200×) were taken of each condition and used for analysis. DCF-DA staining was normalized to cell number by DAPI staining. The assay was performed in triplicate. The results shown are mean±s.e.m. *P<0.05. Bars, 50 μm.

Fbln5 controls fibronectin-mediated integrin-induced ROS production

To further validate the effect of Fbln5 on ROS production, we performed in vitro studies using mouse embryonic fibroblasts (MEFs) harvested from Fbln5−/− and WT embryos. Infiltrating fibroblasts make up a large percentage of the cells found in the tumor microenvironment, therefore MEFs provided us with a means of initially investigating the effect of Fbln5 on ROS production in a tumor-relevant cell population (Kunz-Schughart and Knuechel, 2002). MEFs were grown on fibronectin- or collagen-coated chamber slides or 96-well plates, and the level of ROS was evaluated by DCF-DA staining. Fbln5−/− MEFs produced significantly higher levels of ROS than WT MEFs when plated on fibronectin but not on collagen or gelatin (Fig. 5A,B; data not shown). Thus, cell association with fibronectin in the absence of Fbln5 was necessary for elevated ROS formation. To determine whether the difference in ROS production was due directly to a loss of Fbln5 expression, Fbln5−/− MEFs were treated with full-length recombinant Fbln5 protein prior to plating on fibronectin. Treatment with full-length Fbln5, but not with an NH2-terminal truncation mutant that lacks the RGD-integrin binding motif, reduced ROS production by Fbln5−/− MEFs (Fig. 5B). This result indicated that Fbln5 binding to integrins was necessary to block ROS generation.

The α5β1 integrin is the primary fibronectin receptor and it has been shown that binding of this integrin to fibronectin can stimulate downstream signaling, leading to ROS generation (Chiarugi et al., 2003). To rule out an elevation in β1 integrin expression as a cause of the increased ROS seen in null MEFs, we examined the expression of β1 integrins by RT-PCR and observed equal expression between WT and Fbln5−/− MEFs (data not shown). Lomas et al. (Lomas et al., 2007) recently provided evidence that Fbln5 competed with fibronectin for β1 integrin binding. Furthermore, it was also shown that binding of Fbln5 to β1 integrins does not cause activation of the integrin or of downstream signaling pathways (Lomas et al., 2007). To determine whether the effect on ROS formation was dependent on activation of β1 integrins, we treated Fbln5−/− and WT MEFs, plated on fibronectin, with a β1 function-blocking antibody. Addition of the antibody reduced the level of ROS produced by Fbln5−/− MEFs, whereas the control IgG antibody had no effect (Fig. 5C). These results suggested that Fbln5 competes with fibronectin for integrin binding and, in the absence of Fbln5, β1 integrin activation by fibronectin is increased leading to higher levels of ROS.

Results from our tumor studies suggested that the higher level of ROS present within tumors from Fbln5−/− mice affected endothelial cell survival. Therefore, we next wanted to determine whether Fbln5 controlled ROS generation in endothelial cells by a mechanism similar to that observed in fibroblasts. We treated bEnd.3 cells, a mouse endothelial cell line shown to express β1 integrins (Sheibani and Frazier, 1998), with small interfering RNA (siRNA) to knockdown the expression of Fbln5. We confirmed, by western blot, that two siRNAs directed against Fbln5 had successfully reduced the expression in bEnd.3 cells (Fig. 6A). Fbln5 expression was not affected by transfection of a scrambled siRNA. After transfection, cells were plated on either fibronectin- or collagen-coated 96-well plates and stained with DCF-DA. The loss of Fbln5 expression significantly increased ROS formation in endothelial cells plated on fibronectin but not on collagen; this was reversed with the addition of the antioxidant N-acetyl cysteine (NAC) (Fig. 6B,C). Treatment with the β1 integrin function-blocking antibody inhibited the increase in ROS production within these cells (Fig. 6C). These results confirmed that Fbln5 regulates integrin-induced ROS production in endothelial cells, and provided further support that one possible mechanism for the reduced angiogenesis in tumors from Fbln5−/− mice was increased ROS generation by endothelial cells. This increase, combined with the high level of ROS already present within the tumor, would apply chronic oxidative stress to endothelial cells resulting in cell death.

Finally, to determine whether the loss of Fbln5 caused tumor cells to increase production of ROS, we analyzed the effect of Fbln5 on Pan02 ROS production. As stated previously, we showed that Pan02 cells do not express Fbln5; therefore, we determined the effect of exogenous Fbln5 on Pan02 cells. Before plating on fibronectin-coated 96-well plates, tumor cells were treated with a full-length recombinant Fbln5 protein. After one hour, cells were treated with DCF-DA to measure the level of ROS. Although plating on fibronectin induced an increase in ROS, Fbln5 did not inhibit production as seen in the MEFs and endothelial cells, indicating that ROS production by tumor cells was not affected by the loss of host Fbln5 (supplementary material Fig. S2).

Fig. 6.

Fbln5 controls integrin-induced ROS production in endothelial cells. bEnd.3 cells were transfected with siRNAs to reduce expression of Fbln5. (A) As determined by western blot, transfection with two siRNAs directed against Fbln5 reduced its expression, whereas transfection with a siRNA designed from a scrambled sequence (SCR) did not reduce Fbln5 expression. UNT, untreated. β-actin was used as a loading control. (B) Twenty-four hours after transfection, cells were plated on fibronectin-coated 96-well plates and stained with DCF-DA for ROS detection. Fluorescence was measured on a plate reader. Treatment with both Fbln5 siRNAs increased ROS production, whereas antioxidant (NAC) treatment reduced ROS expression. (C) The assay was repeated by plating cells on either fibronectin or collagen. Only cells plated on fibronectin had a significant increase in ROS production, whereas treatment with an anti-β1 integrin function-blocking antibody inhibited this increase. Results shown are mean±s.e.m. **P<0.001, *P<0.05.

Fig. 6.

Fbln5 controls integrin-induced ROS production in endothelial cells. bEnd.3 cells were transfected with siRNAs to reduce expression of Fbln5. (A) As determined by western blot, transfection with two siRNAs directed against Fbln5 reduced its expression, whereas transfection with a siRNA designed from a scrambled sequence (SCR) did not reduce Fbln5 expression. UNT, untreated. β-actin was used as a loading control. (B) Twenty-four hours after transfection, cells were plated on fibronectin-coated 96-well plates and stained with DCF-DA for ROS detection. Fluorescence was measured on a plate reader. Treatment with both Fbln5 siRNAs increased ROS production, whereas antioxidant (NAC) treatment reduced ROS expression. (C) The assay was repeated by plating cells on either fibronectin or collagen. Only cells plated on fibronectin had a significant increase in ROS production, whereas treatment with an anti-β1 integrin function-blocking antibody inhibited this increase. Results shown are mean±s.e.m. **P<0.001, *P<0.05.

Tumor growth is reduced and ROS levels are increased in Fbln5RGE/RGE mice

We determined that the binding of Fbln5 to integrins is crucial for its regulation of ROS production. To confirm this in vivo, we repeated our tumor studies in transgenic knock-in mice containing an altered RGD motif. In these mice, the aspartic acid of the integrin-binding RGD sequence of Fbln5 was changed to glutamic acid, rendering the protein unable to bind to integrins (Yang et al., 2007). These mice appear normal and do not exhibit any of the defects observed in Fbln5−/− mice (H. Yanagisawa, unpublished). Pan02 cells were injected subcutaneously into WT, Fbln5−/− and Fbln5RGE/RGE mice, and tumor growth was monitored. As observed in the Fbln5−/− mice, tumors from Fbln5RGE/RGE mice grew slower and were smaller than tumors from WT animals (Fig. 7A,D). At the time of sacrifice, tumors were harvested and the level of ROS was determined by DHE staining. Consistent with the results from tumors grown in Fbln5−/− animals, tumors from the Fbln5RGE/RGE mice had significantly higher levels of ROS than tumors grown in WT animals, further confirming the function of Fbln5 in controlling integrin-induced ROS production (Fig. 7B). We next evaluated blood vessel density and observed a significant decrease in angiogenesis in tumors grown in Fbln5RGE/RGE mice compared with tumors from their WT counterparts (Fig. 7C). Thus, we show, in two models (Fbln5−/− and Fbln5RGE/RGE mice), that the loss of Fbln5 function in vivo reduces tumor growth and angiogenesis, and leads to increased ROS production. Results from tumor studies in the Fbln5RGE/RGE mice also further validate that integrin binding is crucial for Fbln5-mediated control of ROS generation.

Fig. 7.

Tumor growth and angiogenesis is reduced inFbln5RGE/RGEmice, but restored with antioxidant treatment. (A–C) Pan02 cells were injected subcutaneously into WT (n=8), Fbln5−/− (n=7) and Fbln5RGE/RGE mice (n=8). The final mean (±s.e.m.) tumor weight at the time of harvest, 4 weeks post-tumor cell injection, is displayed (A) (Farrow et al., 2008). *P<0.05, **P<0.01, analysis of variance (ANOVA). (B,C) Sections of tumor tissue from mice of each genotype were stained with DHE to assess the level of ROS (Farrow et al., 2008), and with MECA-32 to assess vessel density. For ROS and blood vessel density analysis, a minimum of ten pictures (total magnification, 200×; high-power field) of three tumors from each group was analyzed. *P<0.05, **P<0.01 versus WT, ANOVA. (D–F) Pan02 cells were injected subcutaneously into WT (n=20) and Fbln5RGE/RGE mice (n=21). NAC (7 mg/ml) was added to the drinking water of tumor-bearing WT (n=10) and Fbln5RGE/RGE (n=11) mice for 3 weeks. (D) Tumor volumes were recorded over 3 weeks. ***P<0.001 versus Fbln5RGE/RGE, ANOVA. (E) Sections of tumors from untreated and NAC-treated WT and Fbln5RGE/RGE mice were stained with DHE to determine the level of ROS. Ten pictures (total magnification, 200×) taken of each tumor from WT (n=5), WT+NAC (n=4), Fbln5RGE/RGE (n=5) and Fbln5RGE/RGE+NAC (n=5) animals were used for analysis. (F) Tumor sections were also stained with MECA-32 to analyze vessel density (n=5 for each condition). The number of blood vessels/mm2 is displayed. The results shown are mean±s.e.m. *P<0.01, **P<0.005, ***P<0.001.

Fig. 7.

Tumor growth and angiogenesis is reduced inFbln5RGE/RGEmice, but restored with antioxidant treatment. (A–C) Pan02 cells were injected subcutaneously into WT (n=8), Fbln5−/− (n=7) and Fbln5RGE/RGE mice (n=8). The final mean (±s.e.m.) tumor weight at the time of harvest, 4 weeks post-tumor cell injection, is displayed (A) (Farrow et al., 2008). *P<0.05, **P<0.01, analysis of variance (ANOVA). (B,C) Sections of tumor tissue from mice of each genotype were stained with DHE to assess the level of ROS (Farrow et al., 2008), and with MECA-32 to assess vessel density. For ROS and blood vessel density analysis, a minimum of ten pictures (total magnification, 200×; high-power field) of three tumors from each group was analyzed. *P<0.05, **P<0.01 versus WT, ANOVA. (D–F) Pan02 cells were injected subcutaneously into WT (n=20) and Fbln5RGE/RGE mice (n=21). NAC (7 mg/ml) was added to the drinking water of tumor-bearing WT (n=10) and Fbln5RGE/RGE (n=11) mice for 3 weeks. (D) Tumor volumes were recorded over 3 weeks. ***P<0.001 versus Fbln5RGE/RGE, ANOVA. (E) Sections of tumors from untreated and NAC-treated WT and Fbln5RGE/RGE mice were stained with DHE to determine the level of ROS. Ten pictures (total magnification, 200×) taken of each tumor from WT (n=5), WT+NAC (n=4), Fbln5RGE/RGE (n=5) and Fbln5RGE/RGE+NAC (n=5) animals were used for analysis. (F) Tumor sections were also stained with MECA-32 to analyze vessel density (n=5 for each condition). The number of blood vessels/mm2 is displayed. The results shown are mean±s.e.m. *P<0.01, **P<0.005, ***P<0.001.

Treatment with an antioxidant rescues angiogenesis and restores tumor growth in Fbln5RGE/RGE mice

To determine whether increased ROS production in tumors from Fbln5RGE/RGE mice was a major antagonist of blood vessel development and tumor growth, we performed tumor implantation studies in the presence of the antioxidant NAC. Pan02 cells were injected subcutaneously into the flank of WT and Fbln5RGE/RGE mice, and NAC was added to the drinking water. Tumor volume was monitored over the course of the experiment. As shown previously, tumors grew slower in Fbln5RGE/RGE mice than in WT mice. However, the tumor growth rate was restored to WT levels in Fbln5RGE/RGE mice treated with NAC (Fig. 7D). Tumors stained with DHE demonstrated that treatment with NAC reduced the level of ROS in Fbln5RGE/RGE mice (Fig. 7E).

To determine whether increased tumor growth in Fbln5RGE/RGE mice treated with NAC was due to alleviation of the angiogenic defect, we analyzed blood vessel density in tumors from NAC-treated and untreated Fbln5RGE/RGE mice. The number of MECA-32-positive blood vessels in tumors from Fbln5RGE/RGE mice was increased to WT levels in the presence of NAC (Fig. 7F). These findings further support our hypothesis that the loss of Fbln5 led to chronic ROS exposure within tumors, causing endothelial cell death and reduced tumor growth.

A growing tumor interacts with and co-opts the surrounding ECM. To ensure continued growth, tumor cells modify the nature of the ECM to create a microenvironment that is conducive to tumor progression; this appears to be particularly true for highly desmoplastic tumors such as pancreatic cancer. Matricellular proteins are a crucial component of that process. Until now, the effect of the matricellular protein Fbln5 on tumor growth has been unclear. Previous reports have shown that fibrosarcoma cells designed to constitutively express Fbln5 had reduced tumor growth when injected subcutaneously into genetically normal mice (Albig et al., 2006). Herein, we report results from the first tumor studies performed in Fbln5−/− mice. Pancreatic tumor growth was attenuated in both Fbln5−/− and Fbln5RGE/RGE mice compared with WT animals. These findings highlight the functional importance of host-derived Fbln5 and the value of engineered mouse models to dissect protein function. We also observed a decrease in angiogenesis in tumors grown in Fbln5−/− mice. This observation was surprising given the recent report that vascular invasion into implanted polyvinyl sponges was increased in Fbln5−/− mice compared with WT animals (Albig and Schiemann, 2004; Sullivan et al., 2007). We found the effect on angiogenesis to be context-specific such that microvessel density was actually elevated in normal non-tumor-bearing pancreas in Fbln5−/− animals compared with the pancreas of WT mice. Given that minor increases in ROS can stimulate angiogenesis, we speculate that the increased angiogenesis observed in Fbln5−/− mice under non-tumor conditions, where ROS levels are normally low (e.g. normal pancreas or polyvinyl sponges), might also be a result of increased ROS production (Sullivan et al., 2007).

Chronic increases in ROS production have been shown to contribute to the onset of certain diseases that are also connected to the loss of Fbln5 expression, such as macular degeneration and cardiovascular disease (Bonomini et al., 2008; Lotery et al., 2006; Spencer et al., 2005; Takahashi et al., 2004). Our data would suggest that the loss of Fbln5 promotes the progression of these diseases by causing an increase in ROS production. Also, it is interesting to note that, although the level of Fbln5 expression is reduced in adult blood vessels compared with neonatal blood vessels, Fbln5 expression is induced in adult mice during times of trauma, such as in response to vascular injury, in atherosclerotic plaques, and in interstitial fibroblasts during lung injury repair (Kowal et al., 1999; Kuang et al., 2003; Nakamura et al., 1999). These are also instances when ROS production is increased; therefore, Fbln5 may be required to regulate ROS levels during the remodeling process that occurs after these events.

It has been reported that Fbln5 inhibits fibronectin-mediated cell spreading, migration and proliferation by competing with fibronectin for β1 integrin binding. Interestingly, binding of Fbln5 to β1 integrins does not induce integrin activation, suggesting that Fbln5 functions as a blocking protein to control intracellular signaling (Lomas et al., 2007). In support of this idea, it was shown that VSMCs harvested from Fbln5−/− mice exhibit no difference in proliferation rates compared with WT VSMCs, until treated with platelet-derived growth factor (PDGF). After treatment with PDGF, Fbln5−/− VSMCs exhibited a twofold increase in proliferation compared with PDGF-treated WT cells. This effect was curtailed by treatment with recombinant Fbln5 protein, indicating that Fbln5 blocks PDGF binding (Spencer et al., 2005). Our data provides further support for this idea by showing that Fbln5 prevents ROS production by blocking the interaction between fibronectin and β1 integrins. ROS are prominent signaling molecules that induce the expression of multiple proteins including matrix metalloproteinases (MMPs) (Wu, 2006). Recently, it was shown that Fbln5 inhibits MMP-7 expression by an unknown mechanism (Yue et al., 2009). Given our data, it is possible that Fbln5 inhibits MMP-7 expression by limiting ROS production. In addition to this, Fbln5 has also been shown to inhibit VEGF expression, a pro-angiogenic molecule whose expression can be induced by ROS (Albig and Schiemann, 2004; Ushio-Fukai and Alexander, 2004). We speculate that Fbln5 controls VEGF expression indirectly by blocking ROS production.

Our tumor studies performed in the presence of NAC indicate that increased ROS levels are the main determinant of decreased tumor angiogenesis and growth in Fbln5−/− and Fbln5RGE/RGE mice. Increasing the level of ROS within tumors to control tumor growth is the mechanism of action behind many successful chemotherapeutics and is the basis for a new type of therapy coined ‘oxidation therapy’ (Engel and Evens, 2006; Fang et al., 2007). In addition, recent studies have shown that increasing oxidative stress on the blood vasculature, by specifically inhibiting the breakdown of ROS within endothelial cells, inhibited tumor growth by decreasing tumor angiogenesis while having no effect on non-tumor, pre-established blood vessel networks (Marikovsky, 2002). Our data support this idea by showing that increased ROS production in the tumor microenvironment effectively reduces tumor growth by diminishing tumor angiogenesis. A remaining question that is under active investigation is: what are the cell type(s) that are responsible for ROS production in the absence of Fbln5? Given the striking effect on the vascular network in Fbln5−/− animals, it is likely that endothelial cells participate in integrin-induced ROS generation. However, we also showed that Fbln5 functions in fibroblasts to control ROS production. Cancer-associated fibroblasts are crucial to tumor progression, and decreasing their level within tumors can slow tumor growth (Pietras et al., 2008). It is therefore possible that increased production of ROS by fibroblasts within tumors from Fbln5−/− mice resulted in fibroblast apoptosis and contributed to the decrease in tumor growth.

Our results provide further support for the development of drugs that alter the oxidative environment within tumors to control tumor growth. Although much work needs to be carried out to fully understand the function of Fbln5 in controlling ROS production, the validation of its relevance in vivo provides an intriguing target for cancer therapy.

Mice

Fbln5−/− mice and WT littermates (C57BL/6 × 129/SvEv hybrid background) aged 4–6 months old were used for tumor studies. The phenotype of the mice was identical to that described previously (Yanagisawa et al., 2002). The generation and phenotypic analysis of Fbln5RGE/RGE mice will be described elsewhere. Investigators interested in obtaining Fbln5−/− or Fbln5RGE/RGE mice should contact Dr Hiromi Yanagisawa. Mice were housed in a pathogen-free facility and all experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center (Dallas, TX).

Cell culture

Mouse pancreatic carcinoma cells (Pan02, also known as Panc02), obtained from the Developmental Therapeutics Program, NCI (Frederick, MD), murine endothelial cells (bEnd.3), obtained from Dr Werner Risau, and mouse embryonic fibroblasts (MEFs) (harvested as described previously) (Yanagisawa et al., 2002) were all maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS) (Gemini Bio-Products, Woodland, CA). All cells were grown in a 37°C humidified 5% CO2 atmosphere. MEF passages 1–6 were used for experiments.

Tumor growth in vivo

Pan02 cells were injected subcutaneously or orthotopically into WT and Fbln5−/− mice. Cells were harvested from subconfluent cultures by treatment with 0.25% trypsin and 5 mM EDTA in PBS. After trypsinization, the cells were washed once in serum-free medium. Cells were pelleted, washed with PBS, counted, and resuspended in PBS. Only single-cell suspensions with greater than 90% viability (as determined by trypan blue exclusion) were used for injections.

For subcutaneous injections, Pan02 cells (3×106/100 μl) were injected into the flank region of WT and Fbln5−/− mice (Puolakkainen et al., 2004). After tumor cell injection, the mice were monitored for weight, signs of discomfort or morbidity, and tumor size. Three to four weeks after tumor cell injection, mice were euthanized. Subcutaneous tumors were excised, weighed, and the tumor samples were frozen in liquid nitrogen or fixed in methyl Carnoy’s fixative for histological analysis. N-acetyl cysteine (NAC) (7 mg/ml) was given in the drinking water for the length of the experiment.

Orthotopic tumor cell injection was carried out as described previously (Arnold et al., 2008). Briefly, the mice were anesthetized with isoflurane, which was maintained throughout the time of surgery. A small left abdominal incision was made and the spleen was exposed to access the pancreas. Tumor cells (5×105/50 μl) in PBS were injected into the tail of the pancreas. A successful subcapsular intrapancreatic injection of tumor cells was identified by the appearance of a fluid bleb without intraperitoneal leakage. The animals tolerated the surgical procedure well, and no anesthesia-related deaths occurred. Four weeks after injection, mice were euthanized and the peritoneal cavity (including the liver and spleen), inguinal and axillary regions, kidneys, thoracic cavity (including the lungs and heart), and brain were screened for metastases by visual inspection under a dissecting microscope and then confirmed by histology. The entire pancreas containing the tumor lesion was harvested and weighed.

Immunohistological analysis

Immunohistochemistry was performed on frozen or fixed tissue sections. For immunohistochemistry on fixed tissues, tissues were deparaffinized in xylene, followed by a graded series of ethanol exchanges, and rehydrated in PBS containing 0.2% Tween (PBST). The primary antibodies used were: rabbit anti-phosphorylated histone H3 (phospho-H3) (Upstate Biotechnology, Inc., Lake Placid, NY); rat anti-mouse endothelial cell marker, MECA-32 (Hallmann et al., 1995) (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA); rabbit anti-mouse smooth muscle actin (SMA) (RB-9010-P, LabVision, Fremont, CA); goat anti-8-hydroxydeoxyguanosine (8-OHdG) (AB5830, Chemicon, Billerica, MA); rabbit anti-γH2AX (NB100-2280, Novus Biologicals, Littleton, CO); and rabbit anti-mouse Fbln5 (BSYN1923) (Zheng et al., 2007). All antibodies were used at a concentration of 5–10 μg/ml. Fluorescent sections were mounted with ProLong Gold antifade reagent with DAPI (Invitrogen). Primary antibody binding was detected with peroxidase-conjugated or fluorophore-conjugated (fluorescein FITC or Cy3) secondary antibodies (Jackson Immunoresearch, West Grove, PA).

TUNEL staining was performed using a DeadEnd Fluorometric TUNEL System (Promega, Madison, WI), according to the manufacturer’s protocol.

Quantification of immunohistochemistry

Multiple fluorescent images per sample were captured under identical conditions (i.e. room temperature, exposure). The percentage of pixels exceeding the threshold value (background) was calculated automatically by the software (% average intensity). Mean blood vessel counts and area were measured either by blind hand counting or by using Metamorph’s ‘Integrated Morphometry Analysis’ tool. For analysis of endothelial cell death, tumor sections were stained for cleaved caspase 3 and co-localized with MECA-32. Double positive cells were counted and taken as a percentage of the total number of positive cleaved caspase 3 cells. All samples were visualized with a Nikon Eclipse E600 microscope (Nikon, Lewisville, TX). All fluorescent images were captured with a Photometric Coolsnap HQ camera and analyzed with either Metamorph software (Universal Imaging Corporation, Downington, PA) or the Nikon Imaging software Elements.

ROS detection

Frozen tumors were sectioned (10 μm) and treated with dihydroethidium (DHE; Molecular Probes, Eugene, OR). 50 μl of 5 μM DHE was added to each tumor section, coverslipped, and incubated at 37°C in a humidified incubator for 30 minutes. Nuclear Red fluorescence was visualized and images captured using the Photometric Coolsnap HQ camera. Fluorescence intensity was quantified using Metamorph software, as described above. To visualize ROS in MEFs, cells were grown on slides that were precoated with fibronectin (F1141, Sigma-Aldrich) or collagen type 1 (354236, BD Biosciences) at 5 μg/ml. After 24 hours the media was removed; cells were washed once with warm PBS and then incubated with 10 μM of 2′-7′-dicholordihydrofluorescein diacetate (DCF-DA) (D399, Molecular Probes) in warm PBS with 5.5 mM glucose for 10 minutes at 37°C. The DCF-DA was removed and replaced with DMEM with 10% FCS for an additional 10 minutes at 37°C. Cells were again washed with warm PBS, fixed with 10% formalin for 5 minutes, and mounted with ProLong Gold antifade reagent with DAPI (Invitrogen). Images and fluorescence were quantified and normalized to the cell number, as described above. ROS levels were also detected in cells by use of a fluorometer. MEFs or bEnd.3 cells (2×104) were plated on 96-well plates that were pre-coated with either fibronectin or collagen (5 μg/ml). Cells were allowed to adhere for 1 hour and then media was removed and DCF-DA (20 μM) added for 1 hour at 37°C. DCF-DA was removed and the cells were washed once with PBS. Fluorescence was measured with an excitation wavelength of 480 nm and emission wavelength of 520 nm. Cells were treated with either full-length recombinant Fbln5, NH2-terminal truncation mutant Fbln5 (a generous gift from Elaine C. Davis, McGill University) or anti-β1 integrin blocking antibody (CD29 #555002, BD Biosciences), at 10 μg/ml, for 15 minutes at 37°C before plating in wells. To block the ROS signal, N-acetyl cysteine (NAC) (10 mM) was added to the DCF-DA reagent prior to adding to cells. Each experiment was performed in triplicate.

siRNA knockdown of Fbln5

bEnd.3 endothelial cells were plated in six-well plates and allowed to adhere overnight. Cells were transfected with either pre-designed siRNAs directed against Fbln5 or a scrambled sequence (Sigma-Aldrich) using the N-TER Nanoparticle siRNA Transfection System (Sigma-Aldrich). Twenty-four hours after transfection, cells were trypsinized and a portion was plated in 96-well plates for ROS detection. The remaining portion of cells was resuspended in sample buffer (Bio-Rad) for western blot analysis. Western blots were performed using standard protocols.

Statistical analysis

Statistical analyses between genotypes and various conditions were performed using Student’s t-test or ANOVA. P values less than 0.05 were considered statistically significant.

This work was supported in part by grants from the NIH (R01 HL07115 to H.Y., R01 HL070187 to T.F. and R01 CA118240 to R.A.B.), The Welch Foundation (to H.Y.), the Effie Marie Cain Scholarship in Angiogenesis Research (to R.A.B.) and The Department of Surgery, UT-Southwestern Medical Center (to R.A.B.). We gratefully acknowledge the support and critical evaluation of this work by members of the Brekken and Yanagisawa laboratories. Deposited in PMC for release after 12 months.

AUTHOR CONTRIBUTIONS

M.K.S. planned and performed experiments, analyzed data and wrote the paper; S.L.C. maintained the mouse colony and performed experiments; G.K. performed experiments and analyzed data; K.O. and T.F. performed experiments and analyzed data; H.Y. planned the project and analyzed data; R.A.B. planned the project, analyzed data and wrote the paper.

Albig
AR
,
Schiemann
WP
(
2004
).
Fibulin-5 antagonizes vascular endothelial growth factor (VEGF) signaling and angiogenic sprouting by endothelial cells
.
DNA Cell Biol
.
23
,
367
379
.
Albig
AR
,
Schiemann
WP
(
2005
).
Fibulin-5 function during tumorigenesis
.
Future Oncol
.
1
,
23
35
.
Albig
AR
,
Neil
JR
,
Schiemann
WP
(
2006
).
Fibulins 3 and 5 antagonize tumor angiogenesis in vivo
.
Cancer Res
.
66
,
2621
2629
.
Arnold
SA
,
Mira
E
,
Muneer
S
,
Korpanty
G
,
Beck
AW
,
Holloway
SE
,
Manes
S
,
Brekken
RA
(
2008
).
Forced expression of MMP9 rescues the loss of angiogenesis and abrogates metastasis of pancreatic tumors triggered by the absence of host SPARC
.
Exp Biol Med (Maywood)
233
,
860
873
.
Bonomini
F
,
Tengattini
S
,
Fabiano
A
,
Bianchi
R
,
Rezzani
R
(
2008
).
Atherosclerosis and oxidative stress
.
Histol Histopathol
.
23
,
381
390
.
Brekken
RA
,
Puolakkainen
P
,
Graves
DC
,
Workman
G
,
Lubkin
SR
,
Sage
EH
(
2003
).
Enhanced growth of tumors in SPARC null mice is associated with changes in the ECM
.
J Clin Invest
.
111
,
487
495
.
Chiarugi
P
,
Pani
G
,
Giannoni
E
,
Taddei
L
,
Colavitti
R
,
Raugei
G
,
Symons
M
,
Borrello
S
,
Galeotti
T
,
Ramponi
G
(
2003
).
Reactive oxygen species as essential mediators of cell adhesion: the oxidative inhibition of a FAK tyrosine phosphatase is required for cell adhesion
.
J Cell Biol
.
161
,
933
944
.
Das
U
(
2002
).
A radical approach to cancer
.
Med Sci Monit
.
8
,
RA79
RA92
.
Engel
RH
,
Evens
AM
(
2006
).
Oxidative stress and apoptosis: a new treatment paradigm in cancer
.
Front Biosci
.
11
,
300
312
.
Fang
J
,
Nakamura
H
,
Iyer
AK
(
2007
).
Tumor-targeted induction of oxystress for cancer therapy
.
J Drug Target
.
15
,
475
486
.
Farrow
B
,
Albo
D
,
Berger
DH
(
2008
).
The role of the tumor microenvironment in the progression of pancreatic cancer
.
J Surg Res
.
149
,
319
328
.
Hallmann
R
,
Mayer
DN
,
Berg
EL
,
Broermann
R
,
Butcher
EC
(
1995
).
Novel mouse endothelial cell surface marker is suppressed during differentiation of the blood brain barrier
.
Dev Dyn
.
202
,
325
332
.
Hsieh
YH
,
Juliana
MM
,
Hicks
PH
,
Feng
G
,
Elmets
C
,
Liaw
L
,
Chang
PL
(
2006
).
Papilloma development is delayed in osteopontin-null mice: implicating an antiapoptosis role for osteopontin
.
Cancer Res
.
66
,
7119
7127
.
Kowal
RC
,
Richardson
JA
,
Miano
JM
,
Olson
EN
(
1999
).
EVEC, a novel epidermal growth factor-like repeat-containing protein upregulated in embryonic and diseased adult vasculature
.
Circ Res
.
84
,
1166
1176
.
Kuang
PP
,
Goldstein
RH
,
Liu
Y
,
Rishikof
DC
,
Jean
JC
,
Joyce-Brady
M
(
2003
).
Coordinate expression of fibulin-5/DANCE and elastin during lung injury repair
.
Am J Physiol Lung Cell Mol Physiol
.
285
,
L1147
L1152
.
Kunz-Schughart
LA
,
Knuechel
R
(
2002
).
Tumor-associated fibroblasts (part II): functional impact on tumor tissue
.
Histol Histopathol
.
17
,
623
637
.
Lawler
J
,
Detmar
M
(
2004
).
Tumor progression: the effects of thrombospondin-1 and -2
.
Int J Biochem Cell Biol
.
36
,
1038
1045
.
Li
H
,
Fan
X
,
Houghton
J
(
2007
).
Tumor microenvironment: the role of the tumor stroma in cancer
.
J Cell Biochem
.
101
,
805
815
.
Liu
X
,
Zhao
Y
,
Gao
J
,
Pawlyk
B
,
Starcher
B
,
Spencer
JA
,
Yanagisawa
H
,
Zuo
J
,
Li
T
(
2004
).
Elastic fiber homeostasis requires lysyl oxidase-like 1 protein
.
Nat Genet
.
36
,
178
182
.
Lomas
AC
,
Mellody
KT
,
Freeman
LJ
,
Bax
DV
,
Shuttleworth
CA
,
Kielty
CM
(
2007
).
Fibulin-5 binds human smooth-muscle cells through alpha5beta1 and alpha4beta1 integrins, but does not support receptor activation
.
Biochem J
.
405
,
417
428
.
Lotery
AJ
,
Baas
D
,
Ridley
C
,
Jones
RP
,
Klaver
CC
,
Stone
E
,
Nakamura
T
,
Luff
A
,
Griffiths
H
,
Wang
T
, et al. 
(
2006
).
Reduced secretion of fibulin 5 in age-related macular degeneration and cutis laxa
.
Hum Mutat
.
27
,
568
574
.
Mantovani
A
,
Allavena
P
,
Sica
A
,
Balkwill
F
(
2008
).
Cancer-related inflammation
.
Nature
454
,
436
444
.
Marikovsky
M
(
2002
).
Thiram inhibits angiogenesis and slows the development of experimental tumours in mice
.
Br J Cancer
86
,
779
787
.
Nakamura
T
,
Ruiz-Lozano
P
,
Lindner
V
,
Yabe
D
,
Taniwaki
M
,
Furukawa
Y
,
Kobuke
K
,
Tashiro
K
,
Lu
Z
,
Andon
NL
, et al. 
(
1999
).
DANCE, a novel secreted RGD protein expressed in developing, atherosclerotic, and balloon-injured arteries
.
J Biol Chem
.
274
,
22476
22483
.
Nguyen
AD
,
Itoh
S
,
Jeney
V
,
Yanagisawa
H
,
Fujimoto
M
,
Ushio-Fukai
M
,
Fukai
T
(
2004
).
Fibulin-5 is a novel binding protein for extracellular superoxide dismutase
.
Circ Res
.
95
,
1067
1074
.
Pietras
K
,
Pahler
J
,
Bergers
G
,
Hanahan
D
(
2008
).
Functions of paracrine PDGF signaling in the proangiogenic tumor stroma revealed by pharmacological targeting
.
PLoS Med
.
5
,
e19
.
Pozzi
A
,
Moberg
PE
,
Miles
LA
,
Wagner
S
,
Soloway
P
,
Gardner
HA
(
2000
).
Elevated matrix metalloprotease and angiostatin levels in integrin alpha 1 knockout mice cause reduced tumor vascularization
.
Proc Natl Acad Sci USA
97
,
2202
2207
.
Proia
DA
,
Kuperwasser
C
(
2005
).
Stroma: tumor agonist or antagonist
.
Cell Cycle
4
,
1022
1025
.
Puolakkainen
PA
,
Brekken
RA
,
Muneer
S
,
Sage
EH
(
2004
).
Enhanced growth of pancreatic tumors in SPARC-null mice is associated with decreased deposition of extracellular matrix and reduced tumor cell apoptosis
.
Mol Cancer Res
.
2
,
215
224
.
Sheibani
N
,
Frazier
WA
(
1998
).
Down-regulation of platelet endothelial cell adhesion molecule-1 results in thrombospondin-1 expression and concerted regulation of endothelial cell phenotype
.
Mol Biol Cell
9
,
701
713
.
Spencer
JA
,
Hacker
SL
,
Davis
EC
,
Mecham
RP
,
Knutsen
RH
,
Li
DY
,
Gerard
RD
,
Richardson
JA
,
Olson
EN
,
Yanagisawa
H
(
2005
).
Altered vascular remodeling in fibulin-5-deficient mice reveals a role of fibulin-5 in smooth muscle cell proliferation and migration
.
Proc Natl Acad Sci USA
102
,
2946
2951
.
Sullivan
KM
,
Bissonnette
R
,
Yanagisawa
H
,
Hussain
SN
,
Davis
EC
(
2007
).
Fibulin-5 functions as an endogenous angiogenesis inhibitor
.
Lab Invest
.
87
,
818
827
.
Takahashi
A
,
Masuda
A
,
Sun
M
,
Centonze
VE
,
Herman
B
(
2004
).
Oxidative stress-induced apoptosis is associated with alterations in mitochondrial caspase activity and Bcl-2-dependent alterations in mitochondrial pH (pHm)
.
Brain Res Bull
.
62
,
497
504
.
Tanaka
T
,
Halicka
HD
,
Huang
X
,
Traganos
F
,
Darzynkiewicz
Z
(
2006
).
Constitutive histone H2AX phosphorylation and ATM activation, the reporters of DNA damage by endogenous oxidants
.
Cell Cycle
5
,
1940
1945
.
Timpl
R
,
Sasaki
T
,
Kostka
G
,
Chu
ML
(
2003
).
Fibulins: a versatile family of extracellular matrix proteins
.
Nat Rev Mol Cell Biol
.
4
,
479
489
.
Touyz
RM
,
Schiffrin
EL
(
2004
).
Reactive oxygen species in vascular biology: implications in hypertension
.
Histochem Cell Biol
.
122
,
339
352
.
Ushio-Fukai
M
,
Alexander
RW
(
2004
).
Reactive oxygen species as mediators of angiogenesis signaling: role of NAD(P)H oxidase
.
Mol Cell Biochem
.
264
,
85
97
.
Valavanidis
A
,
Vlachogianni
T
,
Fiotakis
C
(
2009
).
8-hydroxy-2′-deoxyguanosine (8-OHdG): a critical biomarker of oxidative stress and carcinogenesis
.
J Environ Sci Health C Environ Carcinog Ecotoxicol Rev
.
27
,
120
139
.
Wu
WS
(
2006
).
The signaling mechanism of ROS in tumor progression
.
Cancer Metastasis Rev
.
25
,
695
705
.
Yanagisawa
H
,
Davis
EC
,
Starcher
BC
,
Ouchi
T
,
Yanagisawa
M
,
Richardson
JA
,
Olson
EN
(
2002
).
Fibulin-5 is an elastin-binding protein essential for elastic fibre development in vivo
.
Nature
415
,
168
171
.
Yang
Z
,
Mu
Z
,
Dabovic
B
,
Jurukovski
V
,
Yu
D
,
Sung
J
,
Xiong
X
,
Munger
JS
(
2007
).
Absence of integrin-mediated TGFbeta1 activation in vivo recapitulates the phenotype of TGFbeta1-null mice
.
J Cell Biol
.
176
,
787
793
.
Yue
W
,
Sun
Q
,
Landreneau
R
,
Wu
C
,
Siegfried
JM
,
Yu
J
,
Zhang
L
(
2009
).
Fibulin-5 suppresses lung cancer invasion by inhibiting matrix metalloproteinase-7 expression
.
Cancer Res
.
69
,
6339
6346
.
Zheng
Q
,
Davis
EC
,
Richardson
JA
,
Starcher
BC
,
Li
T
,
Gerard
RD
,
Yanagisawa
H
(
2007
).
Molecular analysis of fibulin-5 function during de novo synthesis of elastic fibers
.
Mol Cell Biol
.
27
,
1083
1095
.

COMPETING INTERESTS

The authors declare no competing financial interests.

Supplementary information