Received date: July 25, 2013; Accepted date: October 08, 2013; Published date: October 10,2013
Citation: Machado C, Nicot ME, Stella CN, Vaz S, Prado C, et al. (2013) Digital Image Processing Assessment of the Differential in vitro Antiangiogenic Effects of Dimeric and Monomeric Beta2-Glycoprotein I. J Cytol Histol 4:187. doi:10.4172/2157-7099.1000187
Copyright: © 2013 Machado C, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Cytology & Histology
The β2 -glycoprotein I (β2GPI) is an endothelial cell ligand, accessible for systemic, autocrine and paracrine signaling. In vivo, β2GPI is immobilized by binding, mainly to the endothelial cell membrane heparan sulfate but also to anionic phospholipids, and functional receptors. The β2GPI was attributed antiangiogenic properties both in vitro and in vivo. This work was designed to evaluate the antiangiogenic effects of native, monomeric, and dimeric β2GPI. Monomeric as well as dimeric forms were purified from human plasma, and the native protein was obtained as a balanced mixture of both components present in human plasma. The proliferation, migration, and differentiation of Human Umbilical Vascular Endothelial Cells (HUVEC) were considered in an in vitro angiogenesis model based on tridimensional cultures and quantitative digital image processing techniques. The early events of the in vitro HUVEC growth and differentiation in the tridimensional cultures microenvironment were addressed by the morphological analysis. The morphological aspects were correlated to the cell growth, oxidative balance outcome and mitochondrial toxicity assays, leading to the evidence that non-confluent HUVEC cultures temporarily stop growing in the presence of the native protein, but remain competent to proliferate. The β2GPI monomer allowed the in vitro differentiation of the HUVECs into typical trabeculæ and incomplete capillary-like tubes, along with lowering the available proliferation fraction. In opposition, the dimer rich purification fraction exposure halted cell elongation and migration, and prevented the organization of the tubular structures in tridimensional cultures, maintaining cell growth. The morphological approach was useful to attribute to β2GPI dimerization the cell migration inhibition modulation, which potentially leads to overcome the diminished sprouting antiangiogenic effect of the monomer fraction of the native protein.
β2–Glycoprotein I; Angiogenesis; Digital image processing; Tridimesional cell cultures; HUVEC
β2–Glycoprotein I (β2GPI), or apolipoprotein H, was recently attributed anti-angiogenic properties both in vitro and in vivo [1-5]. These reports add new reasonable β2GPI importance on tissue regeneration and tissue proliferation to the already identified complexity of the signaling activities of β2GPI upon haemostasia and inflammation.
The plasma protein β2GPI was first reported in 1961  as a protein from the beta-globulin fraction of plasma. β2GPI is single a polypeptide chain of 326 aminoacids organized in five domains bearing 11 disulfide bridges, and 5 asparagine glicosylation sites . Plasma concentrations between 142 ± 30 and 271 ± 122 μg/ml were reported for healthy individuals, measured through ELISA tests by several research groups [8-12]. In plasma around 60% of the protein circulates in a free form and 40% bound to lipoproteins, mainly rich in triglycerides fractions, quilomicrons and VLDL, a minor portion in HDL . The gene responsible for β2GPI transcription is placed in 17q23-241 chromosome [14,15] and the main site of its synthesis is the liver [16,17]. Other synthesis sites are endothelial cells, intestines, trophoblasts and monocytes [18,19]. β2GPI undergoes renal reabsortion attached to megalin .
The β2GPI molecule represents a multifunctional endothelial cell ligand, accessible for systemic and local autocrine and paracrine signaling. In vivo β2GPI molecules are immobilized by binding to heparan sulfate molecules expressed in the endothelial cell membranes, but β2GPI can also bind to the anionic phospholipids enriched membranes of activated and senescent cells, and to functional receptors like ApoER2, annexin A2, TRL (Toll-like receptors) 2 e 4 and proteoglycans [21-24]. The short consensus repeat domains of β2GPI are typical of the complement control proteins superfamily (CCP), although the C4 convertase activity was not documented for this protein . Monomeric as well as dimeric forms can be found after conventional purification, and higher aggregates were observed in nonreductive conditions, by SDS-PAGE . Monomer and dimer differ in their redox abilities and also in the antinflammatory properties. Dimerization is increased after protein severing and oxidation.
Physiological injury-limiting mechanisms are being increasingly implicated in the redox regulation of microcirculation homeostasis, and may interfere with multiple cell functions [27-29]. The β2GPI antiaterogenic effects were attributed to the protein reductive faculty to inhibit lipid oxidation in vivo. Dimerization extent is probably low in physiologic situations. Patients with isquemic stroke have higher levels of cβ2GPI than healthy individuals . In disseminated intravascular coagulation, there is a reduction in plasma concentration of β2GPI, which suggests the consumption of the protein . In these cases, the proteolytic cleavage of β2GPI may not be the only mechanism of clearance, since the reduction in β2GPI levels is greater than the increase of cβ2GPI . Dimers, besides higher order aggregates, and closed conformation protein structures probably account for a further reduction in β2GPI plasma levels .
The fifth β2GPI domain, the relevant modification site within β2GPI structure, binds to membrane anionic phospholipids (Kd = 330 nM), DNA and heparin [12,32]. Heparin stimulates cleavage of the fifth β2GPI domain by plasmin, producing a redox labile form of the β2GPI nicked form (cβ2GPI) . Thrombin, tPA, tissue factor/ VIIa factor and uroquinase are not capable of cleaving β2GPI. The β2GPI cleavage by factor Xa, is very slow  or not observable . The cβ2GPI is generated in vivo in pathological states of increased fibrinolysis and is unable to bind phopholipids [10,35,36]. Redox mechanisms were investigated in addition to the enzymatic cleavage [37-39]. Recently, endothelial cells and platelet secretion of thiol oxyreductases were associated to the β2GPI reduction [40,41] and to the functional activation of its dependent vascular targets [40,42]. Ioannou et al.  reported the increased in vitro survival of endothelial cells acted by reduced β2GPI, by using thyorredoxin and NADPH.
Binding of β2GPI through the lysine rich portion of domain V to negatively charged phospholipids on the surface of endothelial cells and its later cross-linkage by anti- β2GPI antibodies was described to activate these cells .
The self amplification nature of the redox dimerization signaling was hypothesized as the switch event for vascular destabilization and sprouting. Therefore, this work was designed to evaluate the antiangiogenic effects of monomeric and dimeric β2GPI, along with their mixture, represented by the purified native protein. The proliferation, migration, and differentiation of endothelial cells were considered in an in vitro angiogenesis model based on 3D cultures and quantitative digital image processing techniques . This approach was considered convenient due to the low protein amount requirements for cell distribution assessment, even if localized effects were to be observed. The morphological aspects were correlated to the cell growth, oxidative balance outcome and mitochondrial toxicity assays.
β2- glicoprotein I (β2GPI) purification fractions
Human β2GPI was affinity-purified from long term stored human plasma so as to take advantage of the spontaneous β2GPI dimerization. The individual plasma bags from at least three different donors were kept under -80°C for three to five years, defrosted overnight at 4°C, pooled, and precipitated with perchloric acid [13,44], then centrifugated at 1100 g for 30 minutes at 4°C. The supernatant was collected, neutralized to pH 7.0 with saturated sodium carbonate solution, dialyzed twice against ten volumes of distilled water, then against ten volumes of 20 mM Tris/pH 7.0, 30 mM NaCl (equilibration buffer) for 24 hours at 4°C. The affinity chromatography was carried out with a Heparin- Sepharose column (Heparin Sepharose 6 Fast Flow, GE Healthcare, Sao Paulo). Unspecific bonds were eliminated by flushing the column with Tris 20 mM 50 mM NaCl pH 7.0 buffer. The β2GPI fractions were then eluted with 20 mM Tris 200 mM NaCl pH 8.0 buffer and collected in sequential 2 ml fractions. All the chromatographic procedure was performed at 4°C. After elution, the column was washed with 20 mM Tris 500 mM NaCl pH 8.0 buffer.
Purification of β2GPI by affinity chromatography was monitored by 280 nm absorbance change (NanoDrop ND-1000, Uniscience, Sao Paulo), SDS-PAGE 12.5% and immunoblot with the anti-human β2GPI specific monoclonal antibodies 5F7 (ICN, Irvine, California) and K22 . The purity and stability of the produced batches were analyzed by SDS-PAGE 12.5%, immunoblotting and indirect enzyme immunoassay (ELISA). Aliquots of eluted protein were classified according to its content on dimer rich or monomer rich purification fractions, kept in a refrigerator at 4°C. Dimer and monomer rich aliquots were assayed as independent purified fractions or pooled to represent the plasma blend of β2GPI moieties and used without further processing. The average yield was 38.7 ± 15 mg purified β2GPI / L of human plasma (n=10 independent purifications, 500 mL each). Monomer and dimer contents can vary considerably among different donors, but there were considered the situation in which both β2GPI fractions produced densitometrically equivalent bands in the SDS-PAGE 12.5% .
HUVEC growth and differentiation studies
Human umbilical vein endothelial cells (HUVEC, ATCC CRL- 1730) were obtained from ATCC Rockville MD USA and cultivated in sterile polystyrene six-well plates 1x103 cells/well in 1mL RPMI 1640 (Invitrogen, São Paulo) pH 7.4 supplemented with 10% (v/v) fetal calf serum and 1% (v/v) antibiotic mixture, under 5% CO2 atmosphere, 37ºC and 70% humidity. Non-toxic β2GPI concentrations were tested in 2D HUVEC cultures, the purified protein was added to the culture medium in the 0.2 μM to 2 μM range. The HUVEC growth and the peroxidative response were evaluated in 24h intervals, in a 3 days period. Tubule formation was studied in 3D cultures grown in sterile polystyrene sixwell plates. The experimental procedure consisted in placing a 30 μL Matrigel® (BD Biosciences, São Paulo) in the middle area of the well and then dropping around it 1 mL of a 1x104 cells suspension in RPMI 1640 pH 7.4 supplemented with 1 % (v/v) antibiotic mixture, under 5% CO2 atmosphere, 37ºC and 70% humidity. The distinct purification fractions of β2GPI were added to the culture medium after cell adhesion in the same concentration 0,2 μM and the growth was evaluated after 24h culture.
Metiltetrazolium reduction assay for citotoxicity (MTT)
Cells were cultivated for 24h in 96-well plates, 3x105 cells/well in 100 μL RPMI 1640 pH 7.4 supplemented with 10% (v/v) fetal calf serum and 1% (v/v) antibiotic mixture, under 5% CO2 atmosphere, 37ºC and 70% humidity. Then, medium was replaced and cells were exposed to the distinct purification fractions of β2GPI in the same volume of RPMI 1640 pH 7.4 supplemented 1% (v/v) antibiotic mixture for 24h. Untreated cells were cultivated in the same conditions, only without the protein. Blank reaction controls were run without cells. Next day, the medium was carefully removed and 20 μL of a 5 mg/ml MTT solution was added to each well and the plate was horizontally shaken for 3h, 37ºC. After addition of 150 μL of DMSO to each well, the plate was delicately shaken and the 490 nm absorbance was measured in an ELISA reader, within 10 min.
Thiobarbituric acid assay of in vitro lipid peroxidation (TBARS)
The lipid peroxidation was measured in the cell cultures supernatants by the classical TBARS procedure . Briefly, 40 μL of each supernatant were added in triplicate tubes with 200 μL of TBA reagent, 0.86% thiobarbituric acid. There were also run triplicate tubes added with the reagent vehicle, 20% trichloroacetic acid. The reaction mixtures were kept at 100°C for 20 min, centrifuged at 8000 rpm for 4 min in a minifuge equipment and the supernatants absorbances were measured at 535 nm. Lipopoperoxidation was measured as aldehyde formation by the formula:
Colored photomicrographies from the cell cultures were captured with a Samsung ES80 camera, in 4000 x 3000 pixels resolution in the inverted microscope equipped with a plan achromatic 20x objective.
The Image processing and statistics algorithms were written in the Matlab language. The acquired images were converted to grayscale. These images have the non-homogeneous illumination, which was corrected by an a posteriori automatic background adjustment procedure [47-52]. The background surface was simulated as a Savitzky-Golay iterative procedure , appropriate for polynomial interpolation of a data set with smooth variations.
Given n+1 points (x,y),with i = 0,1,,.,n, it was determined a (n) degree polynomium y(x), as described by:
With the data points, it was build a linear system with n+1 equations and the polynomial coefficients were determined thorough a least square estimation calculus. Then, the illumination was corrected by subtracting the background from the original image. This procedure assume that the input image I(x,y) is formed by a shadow component S(x,y), that corrupts the original image
The corrected image Uc(x,y), corresponds to an estimate of the original image U(x,y):
The corrected images are converted to binary black and white images, applying a threshold obtained by the Otsu method. From the global image data, the threshold value was selected to result in the lower interclass variation among black and white pixels, normalized to the (0,1) interval. The noise level that appear in the image was controlled by a morphology transformation of the opening and closing type. The binary regions of the resulting image were labeled and the contours of this were obtained . There were calculated the mean density, mean area, elongation coefficient and circularity of each region, as described below.
• Mean region area
n _ reg : Number of labeled regions
A : Area of the labeled region
A : Total image area
• Mean region density
• Circularity coefficient
preg : Perimeter of the labeled region
• Elongation coefficient
The structures HUVEC formed in the 3D culture substrates after 24h growth were automatically measured and classified. The region morphologies were classified as circular, elongated or deformed according to the relationships among their elongation and circularity coefficients. The elongation coefficient <0.5 was the classification criterion for elongated regions, while regions with elongation coefficient >0.5 were considered circular if their circularity coefficient >0.8. The regions with elongation coefficient >0.5 and circularity coefficient <0.8 were classified as deformed. Same procedure was applied to the 2D cultures, after 24 and 72h growth.
The groups were compared through two way ANOVA followed by Tukey test of the means, significance levels were set to 95%.
HUVEC growth and differentiation
β2GPI was largely nontoxic to the cultures in the tested concentrations. A concentration dependent 24h lag was observed in the cell proliferation, as reported by cell counting results. This growth interruption occurred without modification of the mitochondrial function as measured by the MTT reduction assay (Figure 1). Despite TBARs evaluation of lipid peroxide production were equivalent after 24h between β2GPI treated and untreated HUVECs (Figure 1), lipid peroxidation products increasingly accumulated in the HUVEC cultures treated with β2GPI after 48h and 72h, in comparison with untreated cultures. Altogether, results were ascribed to stabilization signaling by β2GPI upon non-confluent HUVEC cultures, which however remained competent to proliferate.
Figure 1: Purified β2GPI concentration effect upon of cell growth and differentiation. (A) HUVEC growth in 6-well NUNC plates, initial cell concentration 103cells/mL; (B) mitochondrial activity in the MTT reduction test, and (C) Peroxidative activity as TBARS concentration in the HUVEC cultures supernatants in the presence of purified β2GPI 0.2 μM. The asterisks indicate the significant differences between treated and untreated groups.
Each obtained photomicrography processing resulted in a set of sequentially modified images, as exemplified in Figure 2. The original image (Figure 2A) was first converted to the grayscale image (Figure 2B), then to the corrected background image (Figure 2C) and finally to the binary image (Figure 2D). The binary images were employed to calculate the model parameters for the 3D HUVEC cultures. The results are shown in Table 1.
|Mean region area values, %||Mean region density values, (pixel)-2||Circularity coefficient||Elongation coefficient||Other, deformed|
a HUVEC cultures were initiated with 104cells and treated with purified β2GPI and its purification fractions in the final concentration of 0.2 μM. Dilution buffer was employed in the untreated cultures as a control. Results from a typical assay
Table 1: Experimental angiogenesis quantification results in 3D HUVEC culturesa .
A differential anti-angiogenic effect of purified β2GPI monomer and dimer rich fractions was demonstrated. The morphological evaluation was correlated to the biochemistry assay, leading to the evidence that non-confluent HUVEC cultures temporarily stop growing in the presence of the native protein, while they remain competent to proliferate (Figure 1). Purified β2GPI was recently attributed antiangiogenic properties both in vitro and in vivo [1-5]. These reports added new reasonable β2GPI importance on tissue regeneration and tissue proliferation to the already identified complexity of the signaling activities of β2GPI upon haemostasia and inflammation.
Chiu et al. [56,57] reported an inhibitory effect of β2GPI upon human aortic endothelial cells migration in vitro, without affecting cell proliferation and survival. They couldn’t attribute the observed effects to the β2GPI purification fractions. The biochemical evidence on the mitochondrial functionality is in accordance with the effects observed by Chiu and collaborators (Figure 1). The viability and growth profile of the unexposed HUVEC cultures and the effect of native β2GPI upon the cells agree well with the literature. Even though mitochondrial toxicity was not produced by β2GPI in the 0.2 μM to 2 μM range (Figure 1), it was observed a delay in the cell proliferation and an increase in lipoperoxide production after 48h and 72h in the presence of the purified protein at 0.2 μM concentration (Figure 1). Comparable concentrations of β2GPI were previously observed to modify ex-vivo liver uptake of colloidal carbon particles .
HUVEC proliferation and migration was feebly sensitive to native β2GPI monomer. The morphological approach allowed to unequivocally attributing the inhibition of tube organization by the β2GPI dimer (Figure 3). The β2GPI dimerization, and further molecule-molecule association is determined by concentration and surface association is reversible . It can be fixed by subsequent linkage of enzymatically or oxidatively nicked protein molecules by the initial biochemical pathways of the regenerative cascade at vascular wall, producing an autocrine signaling looping for antiangiogenesis [5,22].
Figure 3: Typical morphology of 3D HUVEC cultures. Untreated (A), or in the presence of purified β2GPI (B) and the (C) Dimer; and (D) Monomer subfractions. Cultures were initiated with 1x104 cells seeded in 1mL RPMI 1640 medium supplemented with an antibiotic mixture around a 30 μL Matrigel® button, as described under materials and methods section. The β2GPI moieties (0.2 μM) were added into medium after cell adhesion and the growth was evaluated after 24h culture. Photomicrographies were taken in an inverted microscope under 10x magnification.
The conventional morphological analysis of the in vitro HUVEC growth and differentiation, which simulates the vessel sprouting mechanism of angiogenesis in the tridimensional cultures microenvironment, was applied to the study of the early steps of tube formation. The in vitro modeling of the angiogenesis related phenomena of cell growth and differentiation into tubular structures was feasible after background correction (Figure 2). The conventional counting of the number and node regions of the trabeculae formed in the measured field was substituted by individual cell functional descriptors. There were considered the mean area and number densities of the individual regions produced in the binary images, as well as its morphology. The mean region area stands for the increase in the cell numbers in the field, while the mean region density combines the cell number increase with the cell to cell bridge formation, a differentiation event. Cell size modification and formats interferes with both parameters, and then they were always analyzed as a profile, rather than an independent representation of the cellular behavior. The time distribution of the region formats counts revealed the differential increase of elongated and deformed regions, corresponding to the several cell to cell interactions arranges which contribute to the trabeculae formation. Circular regions were kept in lower numbers and probably correspond to the proliferative and degenerative fraction of cellular elements in the observation field. Elongation received special attention, because elongation precedes the successful trabeculae formation. Circular regions were associated with proliferative behavior. Other shapes were referred to completed trabeculae, superimposed cell regions, and probably also non identifiable fragments and cell debris. Number increases of all parameters were indicative of cell proliferation, while the proportions could be employed to analyze cell migration and differentiation.
The protein exposure could be associated with mean region area values equivalent to those obtained for the unexposed HUVEC cultures in the cultivation time tested, but the dimer and monomer fractions behaved slightly different. Dimer and monomer rich fractions presented very divergent aspects in the 3D cultures, both for the area and density parameters and for the distribution of the selected region format counts (Figure 3 and Table 1). In the first 24h, the HUVEC cultures exposed to purified protein pool reached mean region area values equivalent to those observed in unexposed conditions. Equivalent values were obtained for the dimer rich fraction, while the monomer rich fraction induced 35% lower values. The mean region densities were higher in the cultures exposed to the dimer rich than those obtained after exposure to the monomer rich fractions; while the purified pool exposure were associated to region density values close to those promoted by the dimer rich fraction. Taken together with the region format counts, it seems reasonable to associate the higher cell expansions with the dimer rich fractions. Circular regions represent either the proliferative and senescent cell morphology, while elongated regions derive from cell migration and distribution in the trabecular structures, rather corresponding to cell differentiation and sprout stabilization. True trabeculae correspond ideally to only one region, because of the close proximity of the cell membranes and the interdigitating structures within their interfaces, which are not discriminable in the binary images. Noteworthy, the pool apparently resulted in a nonhomogeneous distribution of the cells in a pattern composed by morphological domains of both types (Figure 3).
The 3D cultures shown in Figure 3 and Table 1 well demonstrate how the morphological evidence could help to describe the complexity of the outcome possibilities for biological systems in which there putatively coexist biochemical conversion of bioactive molecules and emergence of interdependent effects. The selected parameters were suited to differentially demonstrate the proliferative and maturative components of the trabeculae formation.
The focal proliferation association with the migration inhibition constrained the proportions of the considered parameters, and then the resultant tube loss could be discriminated from pure nonproliferative or survival inhibition effects (Table 1). Endothelial cell activation and death were previously ascribed to dimeric and nicked protein forms . The β2GPI monomer and dimer were previously reported to differentially modify platelet activation [22,23,59]. Native β2GPI inhibits the respiratory burst of resident macrophages  and also the foam cell formation . Physiologically, this later phenomenon was related to lipid-carrier proteins intake [61,62]. Both platelet and leukocyte activation are essentially dependent on surface characteristics and there may occur topographical signaling by β2GPI structures, an underexplored source of biological effects. The morphological approach, and the assay of β2GPI solutions enriched with either monomer or dimer fractions was mostly useful to reveal the motility interference pathogenesis of the paradoxical association of increased cell growth and diminished sprouting associated to the protein and to specifically attribute this antiangiogenic activity to the β2GPI dimerization, while dimeric and monomeric forms apparently associated with opposite effects. Despite vascular tree architecture and dynamic organization should be sensitive to the protein monomer, an extensive early antiangiogenic outcome by excess β2GPI monomer is indeed not expected to occur, because both lower proliferation and preserved elongation are essentially self-limited in time by β2GPI consumption and dimerization . The cell migration resulted less susceptible to the dimeric fraction, even though viability and growth were preserved. This work demonstrated that cross-linked dimeric protein is able to perform the whole endothelial cell activation, without the requirement of the anti- β2GPI antibodies.
The morphological modeling of 3D cultures then allowed identifying dimerization interference for the inhibition of the endothelial cell migration by β2GPI, an antiangiogenic and self-limited in vitro effect of the protein monomer that precedes the peroxidative response associated with the untied cell proliferation in endothelial cell cultures.
Purified β2GPI preserved HUVEC mitochondrial viability in vitro, but cell proliferation was reduced or arrested during the first 24 h. This lag-inducing effect was not concentration-dependent in the 0.2 μM to 2 μM range. Peroxidation was not increased before 72h in culture treated with 0.2 μM _2GPI. There were observed a differential effect of monomer and dimer rich β2GPI purification fractions upon the cell cultures, either 2S and 3D. The native β2GPI monomer allowed the in vitro differentiation of the HUVECs into typical trabeculae and incomplete capillary-like tubes, but reduced apparent cell densities. In opposition, the dimer rich fraction exposure halted cell elongation and migration, and prevented the organization of the tubular structures in 3D cultures, without disrupting cell proliferation. The pooled fractions were able to both proliferate and organize tubular structures.
The morphological approach was useful to attribute to the in situ β2GPI dimerization the recovery of untied cell proliferation in endothelial cell cultures and the overcome of the diminished sprouting antiangiogenic effect of the monomer fraction of the native protein.
This work received financial support from National Research Council (CNPq), regular project protocol number 474654/2004-4 and from São Paulo Research Foundation (FAPESP), protocol number 2007/57728-5. Sara Vaz was in receipt of a Santander Mobility Fellowship for undergraduate student. Prof. Francisco Palacios Fernández and Prof. Miriela Escobedo Nicot were in receipt of Mobility and Ph.D. Fellowships from Brazilian Government under the Education Department CAPES/ MES Brazil-Cuba Cooperation Program. The Blood Bank of the Public Servants´ Hospital of the State of São Paulo (HSPE-IAMSPE) furnished the human plasma used in this work.
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals