Endothelial cell migration and adhesion depend on cytoskeletal remodeling mediated by actin dynamics 3435. Cell migration has been associated with β-actin mRNA localization and increased actin synthesis 3637. Angiogenesis inhibitors, such as endostatin and endorepellin, target endothelial cell migration by interfering with actin organization24383940.

Endorepellin was found to decrease the protein levels of endothelial cell actin (Figures 2 and 3A). In additional experiments, the continuous exposure to endorepellin (200 nM) resulted in a decline of endothelial cell β-actin levels with a T_1/2 of ~45 minutes (Figure 3C and 3D). The observed changes in actin may reflect endorepellin action through the α2β1 integrin 24. Previously, we identified α2β1 integrin as the major receptor for endorepellin on endothelial cells, and associated the interaction with promoting endothelial cell disassembly of the actin cytoskeleton and focal adhesion complexes. The observed decrease in actin may be directly related to the breakdown of the cellular actin network which would interfere with endothelial cell migration and capillary morphogenesis.

One of the key proteins identified in our proteomics analysis and validated by immunoblotting, was calreticulin (Figure 3A and 3B), a multifunctional and widespread calcium-binding protein found in the endoplasmic reticulum, but also at the cell surface and as a secreted form 41. Additionally, vasostatin, a fragment of calreticulin, inhibits angiogenesis and tumor growth 4243. The localization of calreticulin in other cellular compartments has prompted reconsideration of this protein as a mediator of a broader array of cellular functions.

Notably, calreticulin is essential for integrin-mediated calcium signalling and adhesion 44, and interacts with α2β1 on the surface of platelets 45, Jurkat and PC3 cells 4647. Calreticulin can regulate adhesion by a number of mechanisms. For example, calreticulin modulates the affinity of α2β1 integrin for its ligands via transient interactions with the cytoplasmic domain of the integrin α2 subunit 48. The transient interaction is believed to function during the formation of a primary pro-adhesive cellular complex but may not be required during the later establishment of a full focal adhesion 4649. Consistent with this, calreticulin-deficient cells have impaired cell adhesion 44, whereas cells overexpressing calreticulin have increased levels of vinculin and are consequently more adhesive 50. Thus, a possible contributing factor to the anti-adhesive properties of endorepellin would be reducing intracellular calreticulin levels.

Calreticulin regulation by endorepellin must also be considered beyond the subcellular levels and extend to the cell surface. Cell surface calreticulin, in contrast to intracellular calreticulin, actually counters cell adhesion. The extracellular matrix thrombospondin proteins function through cell surface calreticulin to promote focal adhesion disassembly and ultimately favor cell migration 515253. Our proteomics study was not capable of distinguishing between cell surface and intracellular calreticulin. Additionally an endorepellin—cell surface-calreticulin functional and cooperative interaction has not been investigated. The potential for the observed endorepellin-induced downregulation of calreticulin to represent changes in either cell surface or intracellular calreticulin levels alone versus overall total levels is intriguing to consider within the context of endothelial cell adhesion, migration and endorepellin function.

We found that endorepellin treatment reduced endothelial expression levels of chaperonin/Hsp60. The heat shock protein family member chaperonin (heat shock protein 60-kDa or mitochondrial matrix protein P1) functions primarily in cooperation with Hsp10 to aide in mitochondrial protein folding, unfolding and degradation events 54. Numerous other functions have been attributed to chaperonin/Hsp60 outside of its role as a classical heat shock protein, including α3β1-specific integrin activation and influencing apoptotic cell death 5455.

While the observed down-regulation could presumably disturb the endothelial cell mitochondrial stress response, perhaps the decreased expression may promote apoptosis in these cells. Indeed, attenuated expression of Hsp60 by antisense mediated knockdown in cardiac myocytes has been associated with promoting apoptosis 56. Additionally the PPARγ agonist, PGJ₂(15-deoxy-Δ^12–14-prostaglandin J₂) has pro-apoptotic effects on human vascular endothelial cells that has been associated with a decrease in Hsp60 expression 57. Intriguingly, endothelial cells during apoptotic death release into their media a fragment of endorepellin which could act in a paracrine fashion on adjacent cells 58. The exact apoptotic role of chaperonin/Hsp60 may be a complex and variable one 565960, but it is intriguing to consider in light of the anti-angiogenic influence of endorepellin on the endothelium.

Vimentin is a type-III intermediate filament of the cellular cytoskeleton. Intermediate filaments form a cage-like network throughout the entire cell, which links intermediate filaments to cytoskeletal crosstalk among microfilaments, microtubules, and mediates signaling events from one region of the cell to another 61. Vimentin intermediate filaments are capable of influencing cell shape, cell adhesion and migration and cell signaling events.

Notably, the α2β1 integrin interacts with vimentin and co-localizes with the integrin in endothelial cell focal adhesions and contributes to the vimentin-associated matrix adhesion structures identified in endothelial cells 6263. Given these roles, a decreased expression of vimentin protein levels might be predicted to support anti-angiogenic action. Unexpectedly, our studies found that endorepellin increased vimentin protein levels (Table I and Figure 3E). Increased expression of vimentin may negatively contribute to the dynamics of cytoskeletal rearrangement induced by endorepellin action or may represent an increase in vimentin cleavage products as a result of apoptosis. Indeed, vimentin is a substrate for caspase cleavage. Cleavage produces vimentin fragments that inhibit continued intermediate filament assembly, and results in the breakdown of the intermediate filament cytoskeletal network—hallmarks of apoptotic cell death 64.

Collagen biosynthesis requires the enzyme prolyl 4-hydroxylase, which catalyzes the formation of 4-hydroxyproline 65. Hydroxylation at these residues is a key factor in facilitating collagen triple helix assembly and stability. Prolyl 4-hydroxylase is a tetramer composed of two α and two β subunits, with the β subunit being identical to protein disulfide isomerase 6667. Protein disulfide isomerase possesses an enzymatic role, catalyzing protein disulfide bond formation and a chaperone role, promoting proper protein folding 686970. In the proteomics analysis, endothelial cells did not show any detectable levels of the enzyme by using Colloidal blue in contrast to the endorepellin-treated cells (Figure 2). However by immunoblotting using an antibody specific for the β subunit we were able to detect low levels of the enzyme and these levels were markedly upregulated by endorepellin (Figure 3E). The immunoblotting data further indicate a sustained induction of the enzyme with kinetics similar to those shown for vimentin (Figure 3E).

We believe the observed increase in the P4-H β subunit may be related to the protein's inherent dual roles. Enzymatically active prolyl 4-hydroxylase requires the β subunit for proper tetramer formation, folding and to prevent aggregation of the α subunits 71. The increase in β subunits could potentially promote the formation of the active enzyme, essentially favor collagen biosynthesis and eventual deposition into the extracellular matrix. Endorepellin may induce prolyl 4-hydroxylase levels as a means to support extracellular matrix assembly and counter angiogenesis since the matrix serves as a barrier to migrating endothelial cells.

The β subunit in the α₂β₂ tetramer as well as monomer form is equivalent to protein disulfide isomerase and both possess disulfide isomerase activity. The endorepellin-evoked increase in β subunit monomers may actually be an increase in the chaperone functioning protein disulfide isomerase. Since proper protein folding by chaperones is essential to protein synthesis, we believe that the endorepellin-evoked increase in β subunits/protein disulfide isomerase levels may link anti-angiogenic endorepellin to interfering with protein synthesis. Of note, tumstatin, the anti-angiogenic fragment of type IV collagen, inhibits endothelial cell protein synthesis and results in apoptotic cell death 7273.

2 × 10⁶ human umbilical vein endothelial cells (VEC Technologies, Inc.) were cultured to confluence in M131 endothelial cell growth factor-supplemented media. Endothelial cells were either treated with 500 nM endorepellin-supplemented media (experimental, n = 5) or PBS-supplemented media (control, n = 5) for 2 hours. Cell cultures were gently washed in Hank's balanced salt solution (with Ca⁺²/Mg⁺²) and protein was extracted by total cell lysis in 4% CHAPS; 8 M urea lysis buffer. Cell lysates were collected via gentle scraping, centrifuged at 15,000 RPM for 10 minutes at 4°C, and followed by mild sonication over ice. Protein was quantified using the 2-D Quant Kit (GE Healthcare), collected and stored at -80°C.

Forty μg of total protein lysate from control (n = 5) and endorepellin-treated (n = 5) endothelial cells were each suspended in lysis buffer (4% CHAPS; 8 M urea), 1 M DTT, IPG buffer pH 4–7 and a trace of bromophenol blue at a final volume of 350 μl. Sample mixtures were centrifuged at 14,000 RPM for 15 minutes. Protein samples were separated in the first dimension on 18 cm IPG strips pH 4–7 using an IPG phor (GE Healthcare) with active rehydration at 30 V for 12 h followed by focusing at 500 V for 1 h, 1000 V for 1 h and 8000 V for 6 h. Prior to protein separation in the second dimension, the IPG strips were equilibrated for fifteen minutes with gentle agitation at room temperature in equilibration buffer (20% glycerol, 2% SDS, 0.375 M Tris/HCl pH 8.8, 6 mol/L urea) containing 13 mM DTT followed by equilibration buffer containing 2.5% iodoacetamide. IPG strips were then directly applied to 8–18% SDS Tris-Glycine gels (Jule Inc. Biotechnologies), and protein separation in the second dimension was performed using the DALT6 platform (GE Healthcare) at 2 W/gel for thirty minutes, followed by 20 W/gel until completion. 2-D gels were stained with the protein sensitive Colloidal Blue Staining Kit according to the manufacturer's protocol and stored in ultrapure water at 4°C (Invitrogen Life Technologies). Each gel image was recorded by the Typhoon fluorescent imager and the Image Quant software program (GE Healthcare).

To compare proteins differentially expressed in control and endorepellin-treated samples, each of the ten gel 2-D images was analyzed in the DIA program of the DeCyder v5.01 software package (GE Healthcare). The DIA program compares protein expression profiles by assessing and quantifying the intensity of the protein spots identified in each gel. Each of the ten DIA analysis files were collected and analyzed together in the DeCyder BVA program. The BVA program identified reproducible spots of interest in control and endorepellin-treated HUVEC proteomes. These protein spots of interest were extracted from the gel and trypsin digested by the Ettan Dalt spot handling work station (GE Healthcare) in preparation for protein identification by Electrospray LC-MS/MS analysis (Finnigan Proteome X LTQ). Sequence analysis was performed with the BioWorks Browser v3.2 software package using, XCorr, ΔCn (>0.1), RSp (<4) and probability values (<1), as filters for positive protein identification.

Statistical analysis of protein expression data was completed through missing value imputation via K-nearest neighbors (KNN, K = 10) analysis as previously described 3233, followed by log-transformation of the imputed data and comparison of control and endorepellin-treated values with statistical validation by Student's t test (p-value ~< 0.1, considered statistically significant). Fold changes were calculated by the formula 10^(log-scale mean endorepellin – log-scale mean control). Proteins were categorized according to protein expression, down-regulated (fold change ≤ 0.75) or up-regulated (fold change ≥ 2), in response to endorepellin treatment.

Endothelial cells were endorepellin treated as described above. Cell lysates were collected in RIPA buffer, separated by SDS-PAGE and transferred to Nitrocellulose (Biorad). Immunoblotting was performed with polyclonal anti-calreticulin (Stressgen, SPA-600); monoclonal anti-β-actin (Sigma, A 5316); monoclonal anti-vimentin (Sigma, V 6630) and monoclonal anti-prolyl-4-hydroxylase β (ImmunO, 63164). Primary antibodies were detected with donkey anti-rabbit HRP (GE Healthcare) or goat anti-mouse HRP (Pierce) secondary antibodies, and developed by ECL (Pierce).