Endorepellin-evoked Autophagy Contributes to Angiostasis *

Endorepellin, the C-terminal domain of perlecan, is an angiostatic molecule that acts as a potent inducer of autophagy via its interaction with VEGFR2. In this study, we examined the effect of endorepellin on endothelial cells using atomic force microscopy. Soluble endorepellin caused morphological and biophysical changes such as an increase in cell surface roughness and cell height. Surprisingly, these changes were not accompanied by alterations in the endothelial cell elastic modulus. We discovered that endorepellin-induced autophagic flux led to co-localization of mammalian target of rapamycin with LC3-positive autophagosomes. Endorepellin functioned upstream of AMP-activated kinase α, as compound C, an inhibitor of AMP-activated kinase α, abrogated endorepellin-mediated activation and co-localization of Beclin 1 and LC3, thereby reducing autophagic progression. Functionally, we discovered that both endorepellin and Torin 1, a canonical autophagic inducer, blunted ex vivo angiogenesis. We conclude that autophagy is a novel mechanism by which endorepellin promotes angiostasis independent of nutrient deprivation.

The network of heparan sulfate/growth factor interactions is a key regulator of angiogenesis (27). Perlecan sequesters VEGFA and FGFs via its N-terminal heparan sulfate side chains, which are then released by heparanases and subsequently bind to their cognate receptors, resulting in the induction of angiogenesis (9,28,29). In addition, there is a feedforward loop in that VEGFA induces perlecan synthesis via the activation of VEGFR2, leading to increased angiogenesis (30,31). Indeed, antisense targeting of perlecan inhibits in vivo tumor angiogenesis (32). During development, perlecan acts as a scaffold for blood vessel formation, and a restriction of Hspg2 expression in early embryogenesis results in cardiovascular defects (7,33).
In contrast, the C-terminal domain V of perlecan, known as endorepellin, exhibits angiostatic properties (34). Endorepellin is found in vivo, where it is proteolytically processed from perlecan (14) via matrix metalloproteinases, a family of enzymes involved in a multitude of biological processes (35)(36)(37)(38)(39). Although perlecan mRNA can undergo alternative splicing, no evidence exists for endorepellin production in vivo via this mechanism (14). This domain of perlecan is an 85-kDa protein comprised of four EGF-like repeats and three laminin-like globular domains (LG1-3). Structurally, LG2/LG3 domains of endorepellin are separated by two EGF-like repeats that can be cleaved by BMP1/Tolloid-like proteases (40,41) to release the LG3 domain (42).
As its name implies, endorepellin is an inhibitor of endothelial cell migration and capillary morphogenesis, thus preventing the formation of new blood vessels (34). These functional properties result from a "dual receptor antagonism" through its binding to VEGFR2 and ␣2␤1 integrin (43): LG1/2 bind to the IgG 3-5 repeats in the VEGFR2 ectodomain, whereas LG3 binds to ␣2␤1 integrin (44). This biological interaction leads to rapid internalization of both receptors and, ultimately, attenuation of the PI3K/phosphoinositide-dependent kinase/Akt/mTOR 6 and PKC/JNK/AP1 pathways and a decrease in expression of VEGFA, thus contributing to the anti-angiogenic activity of endorepellin (45).
In vivo studies have shown that endorepellin specifically targets the tumor vasculature and inhibits tumor angiogenesis (46). This bioactivity leads to inhibition of tumor growth without inducing apoptosis. Recently, we have discovered that sol-* This work was supported in part by National Institutes of Health Grants RO1 CA39481, RO1 CA47282, and RO1 CA164462 (to R. V. I.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 1 Both authors contributed equally to this work. 2  uble endorepellin induces autophagy in endothelial cells via the binding of its LG1/2 domains to VEGFR2 (47). This process occurs independently of the ␣2␤1 integrin and induces several autophagic markers (Beclin 1, LC3, and p62) under nutrientenriched conditions (47). In this study, we examined in detail the physical properties of endothelial cells treated with endorepellin via atomic force microscopy (AFM) imaging and nanoindentation. We further elucidated the mechanism behind endorepellin-evoked autophagy. Specifically, we found that endorepellin evoked phosphorylation of AMPK␣ at Thr 172 , a key residue necessary for autophagic progression. Moreover, endorepellin blunted vessel sprouting in ex vivo angiogenesis assays, and this bioactivity was blocked by halting AMPK␣ activation. Thus, we propose a new mechanism by which a fragment of an extracellular proteoglycan links angiostasis to autophagy.

Results
Endorepellin and Torin 1 Evoke Nanoscale Molecular Bumps in Endothelial Cells-To determine the nanoscale structural changes in porcine (PAER2) cells and human endothelial cells (HUVEC) evoked by endorepellin or Torin 1, a selective inhibitor of mTOR (48), we utilized tapping mode AFM imaging, which quantifies cell surface topography at nanoscale spatial resolution. We discovered that, although the vehicle-treated PAER2 cell surface was relatively smooth (Fig. 1A), that of endorepellin-treated cells (Fig. 1B) revealed increased surface roughness with the formation of discrete bumps. Identical bumps were detected in the cells treated with Torin 1 (Fig. 1C).
Next we determined the height of these bumps from the three-dimensional images (Fig. 1, D-F). Line scanning profiles of the three-dimensional bumps showed marked elevation in both endorepellin-treated (Fig. 1H) and Torin 1-treated cells (Fig. 1I), suggesting that these elevations may represent autophagosomes. In comparison, vehicle-treated cells exhibited a uniform height (Fig. 1G).
As the PAER2 cells, the human counterparts showed identical formation of nanomolecular bumps on their cell surface (Fig. 2, E-G) compared with vehicle-treated HUVEC (Fig. 2, A-C). Line scanning profiles of the three-dimensional bumps showed marked elevation in HUVEC treated with endorepellin ( Fig. 2H) vis à vis cells treated with vehicle (Fig. 2D). We also used low-magnification AFM images (Fig. 2, I-K) to quantify the number of these bumps, which we interpret as autophagosomes. There was a significant increase in the number of autophagosome-like structures evoked by either endorepellin or Torin 1 vis à vis vehicle-treated cells (p Ͻ 0.001, Fig. 2L). To quantify the changes in cell morphology associated with autophagy, we analyzed cell surface roughness. The two surface roughness parameters, Ra (arithmetic mean height) and Rq (root mean square height), were determined in an area of ϳ36 m 2 . We observed that, upon induction of autophagosome-like structures, the cell surface roughness (Ra) was markedly increased in both endorepellin-treated (p Ͻ 0.05, Fig. 2M) and Torin 1-treated HUVEC (p Ͻ 0.001, Fig. 2M). Rq values were similar to the Ra values (data not shown). We interpret these findings as representative of autophagosome generation within the cytoplasm of endothelial cells derived from both porcine and human large vessels.
Despite Cell Surface Nanostructural Changes, Endorepellin Does Not Alter the Elastic Modulus of Endothelial Cells-Following this observation of nanostructural changes in the endorepellin-treated endothelial cells, we hypothesized that autophagosome formation, induced both by endorepellin and nutrient deprivation, would result in changes in the cellular elastic modulus. To this end, we utilized a special type of AFM where nanoindentation was performed by a microspherical tip that directly indents the cell surface while the cells are alive and grown in appropriate culture medium (49,50). This configuration provides information regarding the dynamic biomechanical properties of the target cells in response to different cellular milieus, allowing for an accurate representation of the changes in cellular elasticity under varying conditions. Although the focus of this study was on endorepellin-induced biomechanical changes, we also utilized Hanks' balanced salt solution (HBSS), a nutrient deprivation method used to induce autophagy through inhibition of mTOR signaling (51). This acted as a positive control for autophagic induction to determine whether autophagy itself could alter the elastic modulus of endothelial cells.
We surmised that the presence of autophagosomes would increase the stiffness of the cells. To our great surprise, despite the presence of structural differences visible by differential interference contrast microscopy in both the endorepellin-and HBSS-treated cells vis à vis vehicle-treated cells (Fig. 3, A-D), the elastic modulus of these HUVEC was not significantly altered (Fig. 3E). Intriguingly, the positive control cells treated with HBSS also did not show any significant change in elastic modulus, suggesting that the formation of autophagosomes may not result in changes in stiffness as measured at the cell surface, at least in endothelial cells.
Endorepellin Evokes Autophagic Flux and Co-localization of mTOR and LC3 into Puncta-Following this biophysical analysis, we sought to delve deeper into the mechanism by which endorepellin contributes to the induction and/or progression of the autophagic process. First, we validated our cell system by performing autophagic flux experiments. To this end, HUVEC were treated with or without chloroquine, an inhibitor of the fusion of the lysosome with the autophagosome, which allows for the build-up of autophagic intermediates (51). Relative levels of LC3-II were measured via immunoblotting to ascertain autophagic activity. We found an increase in LC3-II upon addition of endorepellin with chloroquine in comparison with chloroquine alone (Fig. 4, A and B), thereby validating induction of autophagy by soluble endorepellin.
Next we determined the effects on the endorepellin-evoked redistribution of mTOR, a previously reported target of endorepellin (45). Under basal conditions, most of the LC3positive puncta were not associated with mTOR ( Fig. 4C). However, upon endorepellin treatment, the majority of LC3 immunoreactivity co-distributed with mTOR into large, autophagosome-like structures (Fig. 4D) in a fashion similar to Torin 1-treated HUVEC (Fig. 4E). Quantification of three independent experiments showed a significant increase in the number of mTOR/LC3-positive autophagosomes per cell upon treatment with endorepellin (p Ͻ 0.05, Fig. 4F) or Torin 1 (p Ͻ 0.05, Fig. 4F) vis à vis vehicle-treated cells. Collectively, these findings corroborate the nanoscale molecular changes described above and suggest that the proautophagic activity of endorepellin can further evoke downregulation of mTOR via autophagic clearance.
Endorepellin Promotes Autophagy via AMPK␣ Activation-It is well established that AMPK plays a key function in regulating autophagy and mTOR activity following phosphorylation of its catalytic ␣ subunit at Thr 172 (52,53). We hypothesized that endorepellin could induce autophagy through a canonical activation of AMPK␣. Under nutrient-rich conditions, endorepellin increased the phosphorylation of AMPK␣ at Thr 172 over time (Fig. 5A), reaching a peak at 4 h, with levels almost three times greater than those of untreated cells (Fig. 5B).
To further verify the role of AMPK␣ in endorepellin-induced autophagy, we treated HUVEC with either endorepellin and/or compound C, a potent inhibitor of AMPK (53). Upon addition of this inhibitor, the phosphorylation of AMPK␣ was no longer detected in endorepellin-treated cells (Fig. 5, C and D). Compound C also attenuated the effect of endorepellin on LC3-II and Beclin 1 levels (Fig. 5, C, E, and F). These results indicate that the activation of AMPK␣ is involved in the endorepellinmediated increase in the autophagic markers Beclin 1 and LC3-II.
Next we investigated the role of AMPK␣ in endorepellinmediated autophagy by imaging the intracellular movement of LC3 and Beclin 1 after treatment with compound C, endorepellin, or both. Endorepellin alone promoted the co-localization of Beclin 1 and LC3 into large autophagosomes compared with vehicle-treated cells (Fig. 6, A and B). In contrast, compound C significantly decreased the formation of endorepellin-induced Beclin 1/LC3-positive autophagosomes ( Fig. 6D) relative to endorepellin alone (Fig. 6E).
VEGFR2 Is Required for Endorepellin-evoked Phosphorylation of AMPK␣ Thr 172 -Next we investigated the relationship between AMPK␣ and VEGFR2 following treatment of endothelial cells with endorepellin by genetically targeting this receptor via RNAi. Following verification of successful knockdown of VEGFR2 (Fig. 6, F and G), we found that an ϳ50% reduction in VEGFR2 protein levels prevented the endorepellin-evoked phosphorylation of AMPK␣ at Thr 172 (Fig. 6F), and these data were significant vis à vis scrambled siRNA in three independent experiments (p Ͻ 0.01, Fig. 6H). We conclude that VEGFR2 is required for the proper activation of AMPK␣ following interaction at the cell surface with the LG1/2 domains of endorepellin (44), as its knockdown leads to a significant attenuation of endorepellin-induced autophagy.
Endorepellin and Torin 1 Inhibit ex Vivo Angiogenesis-Given its well established angiostatic activity (34,43,47), we hypothesized that endorepellin could evoke autophagy and inhibit angiogenesis via a common pathway. To this end, we utilized an ex vivo model, the aortic ring assay (54). This assay closely mimics the in vivo environment required for angiogenesis, as it includes both endothelial and supporting cells, which surround the aorta in vivo (55). In rings grown in a nutrient-rich environment, we observed a well structured microvessel network with clearly defined tubules and regular branching (Fig. 7,  A and B). In contrast, the samples treated with endorepellin, using the same concentration (200 nM) as in the biochemical assays, showed a significant decrease in the number of microvessels growing from the rings (Fig. 7, C and D).
To differentiate endothelial from non-endothelial cells, we immunostained the aortic ring explants with an antibody toward CD31/platelet endothelial cell adhesion molecule, a cell surface glycoprotein highly expressed by endothelial cells. The confocal images clearly showed that the sprouting vessel-like structures were indeed positive for CD31 (Fig. 7, E and F) and that endorepellin markedly reduced them (Fig. 7, G and H).   Furthermore, treatment of aortic rings with Torin 1 revealed a pattern similar to that evoked by endorepellin (Fig. 7, I-L). Quantification of three to four independent experiments revealed a significant suppression of the number of sprouts per aortic ring by both endorepellin and Torin 1 vis à vis vehicle treatment (p Ͻ 0.01, Fig. 7M).
Autophagic Inhibition Reverses the Angiostatic Response Triggered by Endorepellin-Having established that both endorepellin and Torin 1 had comparable effects on angiogenesis, we sought to elucidate whether induction of autophagy was the impetus for the endorepellin-mediated angiostasis. To this end, we utilized compound C as a means to prevent endorepellin-mediated autophagy and angiostasis. In agreement with the findings presented above, endorepellin inhibited angiogenic sprouting from mouse aortic rings compared with those treated with vehicle (Fig. 8, A and B). Importantly, compound C blocked the inhibitory activity of endorepellin, as angiogenic sprouts grew to levels comparable with those seen in vehicle-treated rings (Fig. 8D). This outgrowth of vessels was statistically significant compared with endorepellin treatment alone but not significantly different from either vehicle-treated (Fig. 8A) or compound C-treated (Fig. 8C) rings (Fig. 8E). Thus, we can conclude that some of the anti-angiogenic effects of endorepellin are directly intertwined with its pro-autophagic capabilities, and, hence, we provide autophagy as a novel mechanism for the inhibition of neovascularization.

Discussion
Autophagy is a catabolic process in which double or multiple membrane-delineated autophagosomes sequester cytoplasmic material and fuse with lysosomes, resulting in the degradation of their contents (56). This process allows for the recycling of long-lived proteins and damaged organelles, thus maintaining cellular homeostasis. In addition to its function in the maintenance of cellular housekeeping under normal conditions, autophagy is also highly up-regulated during nutrient deprivation, hypoxia, and other unfavorable conditions, where it promotes cell survival (57). Given both the cytoprotective and sometimes cytotoxic functions of autophagy, defects in this process can contribute to a number of human pathologies, including cancer and neurodegenerative diseases (51).
Despite recent progress, the functional role of autophagy remains unclear in many biological contexts, particularly in angiogenesis, where its role remains controversial (58,59). Endorepellin, the C-terminal domain of perlecan, induces autophagy in endothelial cells (47). Here we investigated in depth the mechanism through which endorepellin evokes autophagy and angiostasis and discovered that protracted autophagy under nutrient-rich conditions can negatively regulate angiogenesis.
First, we analyzed the effect of endorepellin on the morphology and microphysical properties of endothelial cells using AFM. The latter enables high-resolution topographical imaging of a single cell surface with minimal sample preparation (60). Notably, nanomolar concentrations of soluble endorepellin caused morphological and biophysical changes in endothelial cells derived from either porcine or human macrovessels, including an increase in cell surface roughness, cell height, and number of autophagosome-like structures. Similar changes were documented in cells treated with Torin 1, which induces autophagy through the inhibition of mTOR (48). Surprisingly, these changes were not accompanied by alterations in the endothelial cell elastic modulus. It is possible that endothelial cells do not incur any changes in elasticity following autophagic induction. However, it is also possible that, given the dynamic nature of autophagy, the lack of changes seen in the treated cells might be due to a high turnover rate of autophagosomes during this process. We note that there was slightly more variability in the modulus of endorepellin-treated cells compared with vehicle-treated samples, suggesting that there may be minor changes in modulus from transient formation of autophagosomes but that their rapid turnover makes any significant changes undetectable. Also, we must mention that, although the majority of cells undergo autophagy initiation following endorepellin treatment, it is possible that some cells do not. In these studies, the moduli were measured in cells chosen at random. Using AFM that can detect fluorescence-labeled autophagic markers could be a way to circumvent this limitation.
We also wanted to scrutinize more thoroughly the intracellular signaling events that accompanied these biophysical changes observed at the cell surface. We focused on AMPK␣ and mTOR, two opposing regulators of the autophagic pathway (61). When activated, AMPK␣, the master nutrient-sensing enzyme, binds and phosphorylates ULK1, a key kinase in the initiation of this process (53,62). Our working model (Fig. 9) delineates the necessity of phosphorylation of AMPK␣ at Thr 172 for endorepellin-evoked autophagy. Interestingly, the kinetics of this phosphorylation are rather slow, peaking 4 h after endorepellin treatment, especially when compared with another extracellular matrix constituent, decorin (63)(64)(65). Decorin treatment also induces autophagy via AMPK␣ phosphorylation, which increases rapidly within 30 min of treatment and is maintained for up to 2 h (66,67). However, the slow induction of AMPK␣ by endorepellin does correlate with increases in several autophagic proteins, including Beclin 1, LC3-II, Peg3, and p62 (47).
Alongside the activation of the pro-autophagic pathway is the inhibition of the anti-autophagic mTOR pathway. mTOR also regulates ULK1 at different phosphorylation sites from AMPK, which results in deactivating ULK1 and, thus, inhibiting the initiation of autophagy. We have established previously that endorepellin attenuates the mTOR pathway via dephosphorylation at Ser 2448 (45). Here we found that, along with these biochemical changes, mTOR is taken up by autophagosomes upon endorepellin stimulation, in a fashion similar to that evoked by starvation in renal epithelial cells (68). We hypothesize that autophagosome-mediated degradation of mTOR caused by endorepellin/VEGFR2 signaling in endothelial cells would further enhance the pro-autophagic role of endorepellin. Proteoglycan receptors may thus play a role in regulating this important catabolic process (69).
Perhaps the most pivotal discovery of our study is the demonstration that autophagy can curtail neovascularization, as both endorepellin and Torin 1 reduce sprouting in ex vivo aortic ring assays. For the first time we were able to restore angiogenic sprouting following endorepellin treatment by blocking AMPK␣-evoked autophagy using compound C. Interestingly, the synthesis of hyaluronan, a key component of the provisional angiogenic matrix under a complex metabolic control (70 -72), is also inhibited by AMPK (73). This is due to the specific AMPK-evoked phosphorylation of hyaluronan synthase 2 at Thr 110 , a modification that blocks its enzymatic activity (73). Hence, along with the induction of autophagy, endorepellin may also simultaneously alter the cellular microenvironment to favor angiostasis by reducing the expression of this pro-angiogenic glycosaminoglycan. Given this information along with previous in vivo data depicting endorepellin as a powerful means to curtail tumor growth and angiogenesis (46), we believe that our findings from this study may yield unique therapeutic targets for novel drug design.
We should point out, however, that, although compound C typically crosses the plasma membrane to inhibit AMPK␣, it also has affinity for VEGFR2. Because of this interaction, compound C can actually inhibit neovascularization through downregulation of VEGFR2 signaling in some models of angiogenesis (74). Indeed, we saw a non-significant reduction in vessel number in rings treated with compound C alone compared with vehicle. As endorepellin binds and signals through VEGFR2 (43,45), it is possible that compound C may interfere with the endorepellin/VEGFR2 axis.
Remarkably, our findings are corroborated by other studies that depict a number of angiostatic matrix proteins and their domains, including endostatin (75) and Kringle V (76), which can concurrently induce autophagy. Notably, a recent study also illustrated that the natural compound capsicodendrin exhibits angiostatic activity and autophagy induction via VEGFR2 inactivation (77). Conversely, impairing autophagy in retinal epithelial cells leads to enhanced angiogenesis (78). Thus, our study contributes new information in support of a paradigm shift whereby autophagy may act primarily as an antiangiogenic mechanism.
In summary, these studies have shown that both human and porcine endothelial cells undergo morphological and biochemical changes following treatment with endorepellin. In both cell types, we observed autophagosome formation in response to endorepellin treatment, accompanied by changes in cell surface roughness and height. Mechanistically, endorepellin-induced autophagy is dependent on AMPK␣ activation downstream of VEGFR2. This autophagic induction includes the activation of the pro-autophagic complex including Peg3, LC3, Beclin 1, and p62 ( Fig. 9). This process is ultimately associated with downregulation of mTOR signaling through autophagic clearance of this master inhibitor of autophagy as well as with reduced angiogenesis (Fig. 9). Future studies utilizing these properties of endorepellin may allow the development of new therapeutic modalities for treating devastating diseases, such as cancer, that involve aberrant angiogenesis.
Scanning Cell Morphology and Ultrastructure by Atomic Force Microscopy-A Dimension Icon atomic force microscope was used for nanostructural studies (BrukerNano, Santa Barbara, CA). A confluent monolayer of HUVEC or PAER2 cells was grown on a 0.2% gelatin-coated four-chamber slide (Thermo Fisher Scientific). After treatment, cells were washed in ice-cold PBS before fixing on ice for 2 h in glutaraldehyde diluted to 2% in HBSS. The fixed cells were washed again in PBS before drying in a desiccation chamber. Images were acquired in tapping mode using a nanosized silicon tip (NCHV-A; Bruk-erNano; tip radius R, ϳ10 nm; spring constant, ϳ42 N/m). To determine the elastic modulus of individual cells, cells were grown in either basal medium with growth factors Ϯ 200 nM endorepellin (6 h) or HBSS (4 h) and tested without fixing. Assisted by an optical microscope to locate individual cells, nanoindentation was performed at 7 m/s indentation depth rate using a microspherical tip (R Ϸ 2.5 m, k Ϸ 0.1 N/m) and the Dimension Icon AFM in the same medium. A Hertz model with finite cell height correction (79) was applied to each forcedepth loading curve to calculate the effective elastic indentation modulus. The Poisson's ratio of the cells was assumed to be 0.5 (80).
Immunofluorescence Microscopy-HUVEC, grown on 0.2% gelatin-coated four-chamber slides (Thermo Fisher Scientific), were treated for the respective analyses. Cells were subsequently washed with PBS and fixed on ice for 30 min in 4% paraformaldehyde at room temperature (82,83). Cells were blocked in PBS ϩ 5% bovine serum albumin, incubated with various antibodies for 1 h, washed in PBS, and incubated for 1 h with an appropriate secondary antibody. Nuclei were stained and visualized with DAPI (Vector Laboratories). Fluorescence images were acquired with a ϫ63, 1.3 oil immersion objective using a Leica DM5500B microscope programed with the Leica Application Suite, Advanced Fluorescence v1.8, from Leica Microsystems, Inc. All resulting immunofluorescence images were analyzed using ImageJ software (National Institutes of Health) and Adobe Photoshop CS5.1 (Adobe Systems).
Transient siRNA-mediated Knockdown-Transient knockdown of VEGFR2 in HUVEC was achieved via transfection with validated siRNAs specific for VEGFR2 (sc-29318) from Santa Cruz Biotechnology. Scrambled siRNA (siScr, sc-37007) served as a control for all siRNA experiments presented here. Six-well plates were seeded with 2 ϫ 10 5 HUVEC, followed by incubation at 37°C ϩ 5% CO 2 until cultures were ϳ70% confluent. Targeting or scrambled siRNA duplex was mixed with transfection medium and Lipofectamine RNAiMAX. After incubation at an ambient temperature (ϳ25°C), the ribonucleic acid/cationic complexes were applied directly to the cells. Following a 48-h transfection, the cells were treated and lysed in radioimmune precipitation assay buffer. Verification of RNAi-mediated knockdown of the target protein was determined via immunoblotting.
Aortic Ring Assays-All animal protocols were performed according to the Guide for Care and Use of Laboratory Animals and the Institutional Animal Care and Use Committee of Thomas Jefferson University. Thoracic aortae from 5-7-weekold C57/BL6 mice (The Jackson Laboratory) were surgically isolated, cleaned, and dissected into 0.5-mm rings. Rings were embedded in 1 mg/ml of type I collagen in a 96-well plate as described previously (84). When embedded, the rings were divided into respective groups: vehicle (PBS or DMSO), endorepellin (200 nM), Torin 1 (40 nM), compound C (1 M), and compound C ϩ endorepellin. Endothelial microvessel sprouts growing out from the rings were counted during the exponential growth phase to obtain angiogenic response data. Before the regression phase, rings were fixed for immunofluorescence staining of CD31 or BS-1 Lectin. Pictures were taken on day 12, and the total number of branches was counted using ImageJ.
Quantification and Statistical Analysis-Immunoblots were quantified by densitometry using ImageJ software. All experiments contained here were carried out with a minimum of three independent trials (85). Results are expressed as the mean Ϯ S.E. Statistical analysis was performed with SigmaStat for Windows version 3.10 (Systat Software, Inc., Port Richmond, CA). Significance of differences was determined by paired and unpaired Student's t test or one-way analysis of variance followed by Tukey-Kramer post-hoc multiple comparison. Data were considered significant with p Ͻ 0.05.
Author Contributions-R. V. I. conceived the study and coordinated the work. A. G., M. A. G., and R. V. I. performed experimental work and wrote the manuscript. D. R. C. performed the AFM measurements, data analysis, and interpretation. L. H. supervised the AFM work. All authors reviewed the results, contributed to data interpretation, and approved the final version of the manuscript.