NON-ENZYMATIC GLYCATION INTERFERES WITH FIBRONECTIN-INTEGRIN INTERACTIONS IN VASCULAR SMOOTH MUSCLE CELLS

Abstract

Objective

We aimed to investigate whether advanced non-enzymatic glycation of the extracellular matrix (ECM) protein, fibronectin, impacts its normal integrin-mediated interaction with arteriolar vascular smooth muscle cells (VSMC).

Methods

Atomic force microscopy (AFM) was performed on cultured VSMC from rat cremaster arterioles to study native and glycated fibronectin (FN and gFN) interactions with cellular integrins. AFM probes were functionalized with FN or gFN or with native or glycated albumin (gAlb) as controls.

Results

VSMC showed increased adhesion probability to gFN (72.9 ± 3.5 %) compared to native FN (63.0 ± 1.6 %). VSMCs similarly showed increased probability of adhesion (63.8 ± 1.7 %) to gAlb compared to native Alb (40.1 ± 4.7 %). Adhesion of native FN to VSMC was α5 and β1 integrin-dependent whereas adhesion of gFN to VSMC was integrin-independent. The RAGE-selective inhibitor, FPS-ZM1, blocked gFN (and gAlb) adhesion suggesting that adhesion of glycated proteins was RAGE-dependent. Interaction of FN with VSMC was not altered by soluble gFN while soluble native FN did not inhibit adhesion of gFN to VSMC. In contrast, gAlb inhibited adhesion of gFN to VSMC in a concentration-dependent manner.

Conclusions

Glycation of FN shifts the nature of cellular adhesion from integrin- to RAGE-dependent mechanisms.

Introduction

Diabetes mellitus is characterized by metabolic disturbances principally of which is hyperglycemia and if sustained over a period of time can lead to vascular disorders [1,13,14]. Importantly, the mortality associated with this metabolic disorder is largely due to the prolonged state of hyperglycemia [2,23]. While a number of mechanisms likely contribute to pathophysiological effects of hyperglycemia, considerable attention has been paid to the ability of glucose to non-enzymatically modify macromolecules, in particular, proteins. The carbonyl group of the reducing sugars including fructose, ribose, galactose, glucose and mannose can interact with amino groups of ECM proteins such as FN to form advanced glycation end products [4,27]. The epsilon amino group of free lysine [68] and the delta guanido group of free arginine residues [17] in ECM, blood and cellular proteins can be modified under hyperglycemic conditions to ultimately form Advanced Glycation Endproducts (AGEs) through a “Maillard reaction”. The Maillard reaction is a multistep process yielding a substantial number of different by-products and end-products including irreversible AGEs [46,56]. AGEs increase the stiffness of the vasculature by enhanced production and accumulation of ECM proteins in the vascular wall [21]. Glycation of ECM also increases cross-linking of ECM proteins and activation of the Receptor for Advanced Glycation Endproducts (RAGE) [15,16].

Interactions between AGEs and RAGE are known to induce a number of changes in VSMC by activating downstream signaling pathways. For example, exposure to glycated ECM can cause increased FN production in VSMC [43], increased calcification [54,61] and proliferation [57,70] leading to neointimal expansion as observed during the development of restenosis and atherosclerosis [24]. Glycated ECM also increase expression of matrix metalloproteinases such as MMP-9, MT1-MMP and RAGEs [29] that can lead to aberrant matrix degradation during hyperglycemia. A key molecule in the progression of these aforementioned RAGE-dependent processes is the pro-inflammatory transcription factor NFKB [16,29,45,53].

We hypothesized that glycation of ECM proteins would modify or interfere with physiologically normal integrin-mediated interactions between ECM and VSMC integrins and lead to preferential activation of RAGE. To test this hypothesis, we examined the effect of glycation of FN on its interactions with α5, β1 and β3 integrins and with RAGE. Further, we performed our studies using microvascular smooth muscle cells as there is increasing evidence for the complexity of ECM in small arteries [7] and its role in cellular signaling underlying mechanotransduction and vasomotor function.

Materials and Methods

Preparation and characterization of glycated proteins

Glycation of human plasma FN (Gibco, Waltham, MA) and bovine serum Alb (BSA) was performed by incubation with 50 mM methylglyoxal for 5 and 7 days, respectively, at 37°C under sterile conditions. Dialysis was performed against sterile sodium phosphate buffer for 24 hours to remove excess glycating agent. Control proteins underwent similar treatment but without the glycating agent. Endotoxin levels were determined using a limulus amoebocyte assay (Lonza, Walkersville, MD) and were measured to be <0.02 ng/ml. The final protein content was analyzed using a BCA protein assay kit (Pierce, Rockford, IL). The formation of AGE was established by the brown color of gAlb and gFN. Further, the production of AGEs formed via the Maillard reaction [68] was confirmed by an increase in intrinsic fluorescence and absorbance using a micro plate reader (Synergy HT, Bio-Tek Inc.). Fluorescence was measured at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. The increase in intrinsic absorbance was measured at 280 nm.

Isolation of VSMC from rat cremaster artery

VSMC were enzymatically isolated from cremaster artery of male Sprague-Dawley rats (175–250 g body weight). Rats were anesthetized using 120 mg/kg of sodium pentobarbital injected intraperitoneally according to Institutional Guidelines and a protocol approved by the Animal Care and Use Committee of the University of Missouri. First and second order arteriolar segments were dissected from cremaster muscle in Ca²⁺⁺ free buffer comprised of 5.0 mM D-glucose, 2.0 mM pyruvate, 0.02 mM EDTA, and 3 mM MOPS, 147 mM NaCl, 8.6 mM KCl, 1.17 mM MgSO₄, 1.2 mM NaH₂PO₄ (pH adjusted to 7.4 with NaOH), with BSA (0.1 mg/ml; Amersham Life Science, Arlington Heights, IL). Vessel segments were then digested with 26 U/ml papain (Sigma Chemical Co., St. Louis, MO) and 1 mg/ml dithioerythritol (Sigma Chemical Co.) in low Ca²⁺⁺ buffer (0.44 mM NaH₂PO₄, 10 mM Hepes, 4.17 mM NaHCO₃, 144 mM NaCl, 5.6 mM KCl, 0.1 mM CaCl₂, 1.0 mM MgCl₂, 0.42 mM Na₂HPO₄ and 1 mg/ml BSA (pH adjusted to 7.4 with NaOH)) for 30 min at 37°C. The vessels were then transferred to low Ca²⁺ buffer with 1.95 U/ml collagenase (FALGPA; Sigma Chemical Co.), 75 U/ml elastase (Calbiochem, La Jolla, CA) and 1 mg/ml soybean trypsin inhibitor (Sigma Chemical Co.) for a further 15 min at 37°C. After washing (x3) cells were plated in DMEM/F12/20% FBS until attached and then switched to DMEM/F12media (TFS-Gibco, Waltham, MA)/10% FBS, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 2 mM L-glutamine, 1 mM sodium, 100 μg/ml streptomycin and pyruvate, 100U/ml penicillin. VSMC were used for atomic force microscopy (AFM) experiments on the second and third day after isolation. VSMC specific α-smooth muscle actin was used to characterize the isolated VSMC.

Measurement of protein adhesion to VSMC using AFM

Sharp tip MLCT-C silicon nitride probes (Veeco, Plainview, NY) with a spring constant of approximately 20 to 40 pN/nm and diameter of 50 nm were functionalized with glycated and native FN [71]. Firstly, the tip was coated with polyethyleneglycol (PEG) (Sigma-Aldrich, St Louis, MO) for 6 min and then washed with PBS (x3) followed by incubation with glycated or native FN (0.3 mg/ml) for 6 min and washed again with PBS (x3). The AFM tip coated with PEG adsorbed native and glycated protein to form a non-covalent link. An integrated Fluoview confocal microscope system (Olympus, Waltham, MA) with a Bioscope II atomic force microscope system (Veeco, Plainview, NY) was used to assess adhesion by measuring the force of rupture between the native or glycated proteins and VSMC. AFM was used in contact mode and operated at 0.5 Hz (tip speed 800nm/s) to cyclically approach, touch and retract from the cell membrane at a midpoint between the nucleus and the cell periphery. For one given experiment, force curves were collected from 5 cells with approximately 180 force curves collected for each cell. Force of indentation was determined as per Hooke’s law:

$$\mathrm{F}=\mathrm{\kappa }\mathrm{\Delta }\mathrm{\chi }$$

Where F = force of indentation, κ = spring constant of the cantilever and Δχ = cantilever deflection.

Approach force was approximately 1000 pN with a Z-scale or piezo movement of 800 nm (for both approach and retraction). To calibrate individual cantilevers used for our experiments, the method of thermal noise amplitude was adopted [5,26]. Probability and density of adhesion was determined by analyzing the number of ruptures (adhesion breaks) that were recorded in individual force curves during AFM probe retraction. The force curves were analyzed using proprietary software NForceR [52]. The probability of adhesion was calculated by determining the percentage of force curves with one or more adhesion events to the total number force curves. To analyze distribution of the rupture forces or adhesion events we plotted, the number of adhesion events vs. adhesion forces ranging from 0 to 300 pN, as histograms. We also examined the number of adhesion events per retraction trace, which was quantified as the number of total adhesion events divided by the total number of force curves.

To determine binding specificity and the role for specific integrins in adhesive events between native and glycated proteins and the cell surface, passage 2 VSMC were serum starved for one hr and then incubated with 30 μg/ml of anti-α5 integrin monoclonal antibody (HMα5–1), anti-β1 integrin monoclonal antibody (Ha2/5), anti-β3 integrin monoclonal antibody (F11) or isotype control antibodies for 45 min before performing AFM experiments. Antibodies were purchased from BD Pharmigen (San Jose, CA) as previous studies in our laboratory and others have used these antibodies to specifically block the interaction between α5, β1 and β3 integrins with their ligands [36,48,62]. To determine the role of RAGE in adhesion of glycated protein to VSMC we performed similar AFM experiments but incubated the cells with a RAGE inhibitor 50 nM FPS-ZM1 (EMD Millipore, Billerica, MA) or 0.02% control vehicle DMSO. For competitive inhibition studies, VSMC were incubated with 0.3 mg/ml of soluble native FN, gFN or with 0.3 mg/ml and 0.9 mg/ml of soluble native Alb and gAlb for 45 min. AFM adhesion studies were then performed with soluble protein in the media using native or glycated FN-coated tips.

Western blot analysis

As an indicator of RAGE binding and signaling, the effects of glycated proteins on the transcription factor, NFKB, were examined. Cultured VSMC were incubated with 100 μg/ml of soluble native or glycated Alb in order to look at the effect of AGE on NFKB protein and activity levels. Passage 2 VSMC were treated with either native or gAlb and were harvested 30 min, 2 hr and 4 hr after exposure. The cells were processed and western blot was performed as described below. To study the change in total NFKB (p65) levels in VSMC activated with glycated proteins NFKB was probed using 1:500 dilution of rabbit polyclonal antibodies against p65 (Abcam, MA) and then detected with 1:5000 goat anti-rabbit peroxidase conjugated secondary antibody (Invitrogen, Waltham, MA). Activation of NFKB was studied with 1:1000 dilution of rabbit polyclonal antibody against phospho-p65 (Cell Signaling Tech. Inc) primary antibody and 1:5000 goat anti-rabbit peroxidase conjugated secondary antibody (Invitrogen, Waltham, MA). The primary antibodies used to detect total NFKB and phosphorylated NFKB have been described in prior studies [12,18,19,30,49,69]. Also, in our studies we observed a single band of the expected molecular weight, which supported the conclusion that the antibodies used were specific for NFKB. To detect α-SMA a 1:2000 dilution of mouse monoclonal anti-α smooth muscle actin primary antibody and a 1:5000 dilution of anti-mouse rabbit IgG peroxidase-conjugated secondary antibody (Sigma, St. Louis, MO) was used, Total and phosphorylated levels of NFKB was normalized to α-SMA.

To detect RAGE expression in VSMC, passage 2 cells were harvested and probed for RAGE. Lung tissue from SD rats and RAGE-KO mice were used as positive and negative controls for RAGE expression, respectively. The positive and negative controls were also used to demonstrate specificity of the RAGE antibody. The lung tissue was harvested, homogenized and used for immunoblotting using the protocol described below. Expression of RAGE on VSMC was studied by using 1:1000 dilution of rabbit monoclonal 1° Ab (Abcam, MA) against RAGE and 1:5000 dilution of goat anti-rabbit peroxidase conjugated secondary antibody.

In preparation for Western blotting, the harvested cells and tissue homogenates were incubated for 30 min on ice in RIPA buffer plus 1% protease inhibitor cocktail (Sigma, St Louis, MO) and sonicated for 45 sec. Homogenates were then centrifuged (6000 g for 1 min) and the supernatants collected. Total protein concentration of the supernatants was determined using the BCA protein assay kit (Pierce, Rockpoint, IL). Desired amounts of protein were mixed with 1× Laemmli sample buffer with 5% β-mercaptoethanol and heated for 5 min at 95°C. Pre-determined volume of protein was loaded in a 4–20% gradient sodium-dodecyl-sulphate (SDS) polyacrylamide gel and run at 80 V for 15 min and then at 100 V for 2 hr. The gel separated protein was then transferred onto a polyvinylidene fluoride (PVDF) membrane at room temperature for 90 min or at 4°C for 17 hr. The membranes were blocked with non-fat dry milk for 2 hr at room temperature or overnight at 4°C then incubated with primary antibody overnight at 4°C followed by secondary antibody for 2 hr at room temperature. Immunoblots were imaged using a Bio-Rad ChemiDoc XRS+ system after which Image Lab software by Bio-Rad was used to quantify the protein bands.

Statistical analysis

Data are shown as mean ± SE for n sets of experiments. “n” represents the number of experiments performed on VSMC derived from cell cultures from different cremaster arterioles. For the AFM experiments each “n” number represents data collected from 3 to 5 cells from a single plated culture of cells. Differences in statistical significance were analyzed using paired Student’s t-test or unpaired Student’s t-test as appropriate. Multiple comparisons were performed using analysis of variance (ANOVA) followed by Tukey’s comparison. A P value of < 0.05 was considered statistically significant unless otherwise stated.

Results

Studies were conducted on primary cultures of VSMC. Immunofluorescence staining for α smooth muscle actin confirmed the cell preparations to be predominantly VSMC (Figure 1). Example AFM height and deflection images (Figure 1) show morphological and cytoskeletal features consistent with VSMC.

Figure 1. Representative Images of VSMC.

Panel A shows immunofluorescent staining for α smooth muscle actin (left) and a secondary antibody only control (right). Antibody staining shows the cell population to be predominantly VSMC. Panel B shows example AFM height (left) and deflection (right) images showing morphological and cytoskeletal characteristics consistent with VSMC.

Glycation of FN increases adhesion to VSMC

Force curves were analyzed to determine adhesive interactions between the cell surface and protein coated AFM probes. Figure 2A and B demonstrates representative force curves illustrating the interaction between native and glycated FN, respectively. The force curve pair consists of an approach curve (blue line) and retraction curve (red line). The number of adhesion events, as represented by the encircled “abrupt breaks” in the retraction curve, which indicate the location of an adhesion rupture between the AFM probe and the VSMC. gFN showed a significant (p<0.05) increase in the probability of adhesion to VSMC (72.9 ± 3.5 %) compared to native FN (63.0 ± 1.6 %) (Figure 2C and D). The increased probability of adhesion also occurred with a significant (p<0.05) increase in the number of adhesion events per retraction curve of gFN (1.19 ± 0.14) compared to native FN (0.84 ± 0.04) (Figure 2E). Plotting histograms of the normalized density of adhesion events (Y axis) relative to adhesion forces (X-axis over a range of 0 to 300 pN) showed that gFN exhibited increased density of adhesion events when compared to native FN (Figure 2F). Collectively, these analyses are consistent with glycation of FN increasing adhesive interactions with VSMC as seen by an increase in both probability and density of adhesion events between gFN and VSMC relative to the native protein.

Figure 2. Glycated FN demonstrates increased adhesion probability and density of adhesion.

Cells were serum starved for an hour and AFM adhesion studies performed (see Methods). Panels A and B show representative force curves between 0.3 mg/ml protein coated silicon nitride AFM probes and VSMC. The blue line represents the tip approaching the cell surface while the red line shows retraction of the tip: A. Force curve for interaction between native FN and VSMC. The retraction curve shows a single “rupture” indicated by the green circle or a single binding event. B. Force curve for interaction between gFN and VSMC. The retraction curve shows two “ruptures” indicated by the green circles. Panels C and D illustrate relative binding probabilities between 0.3 mg/ml protein coated silicon nitride probes and VSMC. C. Mean data showing increased adhesion of gFN to VSMC compared with native FN (*p<0.05, 2 tailed paired t-test; n=5). D. Data from panel C. shown as line graphs depicting the increased probability of adhesion between gFN and VSMC. Each line represents individual primary cultures with gFN and FN studied on the same day. Panel E. gFN shows a significant increase in the number of adhesion events per force curve compared to native FN (p<0.05, 2 tailed paired t-test) (n=5). Panels F: histograms showing the distribution of rupture force-adhesion events for gFN and native FN with VSMC (n=5). gFN shows an increase in density of adhesion with VSMC.

Effect of glycation on adhesive properties of Alb to VSMC

Albumin was used as both a non-ECM control protein and as a contrast for non-integrin binding interactions with the cell surface. Figure 3A and B shows a significantly (p<0.05) increased probability of adhesion between gAlb (63.8 ± 1.7 %) and VSMC compared to native Alb (40.0 ± 4.7; p<0.05). Glycation of Alb also significantly increased the number of adhesion events per retraction curve from 0.48 ± 0.23 in the native form to 0.89 ± 0.14 in the glycated form (p<0.05); Figure 3C. In Figure 3D we plotted the normalized density of adhesion events between protein and cells over a range of adhesion forces as histograms. gAlb exhibited an increased density of adhesion events, as represented by higher peak and area under the curve in the plotted histogram, when compared to native Alb. These data show that gAlb, similar to gFN, shows a higher probability and density of adhesion to VSMC relative to its native protein.

Figure 3. Glycated Alb demonstrates increased adhesion probability and density of adhesion.

Panels A and B show relative binding probabilities between 0.3 mg/ml protein coated silicon nitride probes and VSMC: A. Bar graph showing increased probability of adhesion of gAlb to VSMC compared to native Alb (p<0.05, 2 tailed paired t-test) (n=5). B. Data in panel A. represented as line graphs as described for Figure 1. C. Characteristics of adhesion events changes as depicted by significant increase in number of adhesion events per force curve of gAlb with VSMC. Bar graph represented as number of events per curve ± SEM. (p<0.05, 2 tailed paired t-test) (n=5). Panel D: Histogram analysis showing the distribution of rupture force-adhesion events for gAlb and native Alb binding to VSMC (n=5). gAlb showed a significant increase in the density of adhesions with VSMC.

Function of integrins in the adhesive interactions of FN and gFN with VSMC

To determine the VSMC cell surface receptors contributing to the increased adhesion of gFN, we first considered the major FN binding integrins on VSMC. Specifically, we examined binding to β1, α5 and β3 integrins. These integrin subunits were specifically targeted using anti-integrin function-blocking antibodies. IgG antibody was used as an isotype control for the adhesion studies. Figure 4A shows that anti β1 antibody significantly attenuated adhesion probability between native FN coated sharp tips and VSMC from 72.4 ± 4.0 % to 31.8 ± 3.7 % (p<0.05) but did not alter the adhesion significantly between gFN and VSMC (82.5 ± 0.5 %/84.4 ± 4.8 %) (Figure 4B). Figure 4C illustrates the marked difference in effectiveness of anti β1 antibody on the probability of adhesion of FN with VSMC compared to that of gFN. Histogram analysis of the force-adhesion events distribution shows reduced density of adhesion of FN to VSMC with anti β1 antibody treatment (Figure 4D) while having negligible effect on the density of adhesion between gFN and VSMC (Figure 4E). Similarly, anti α5 antibody treatment of VSMC reduced FN adhesion probability significantly from 81.8 ± 0.4 % to 67.6 ± 2.9 % (Figure 4F) (p<0.05) whereas inhibition of α5 integrin did not cause significant differences in adhesion of gFN to VSMC (Figure 4G). In Figure 3H the same data were represented as normalized to the probability of adhesion between gFN and control IgG antibody treated cells. Inhibition of α5 integrins also reduced the normalized density of adhesion of FN (Figure 4I) but not gFN (Figure 4J) as shown by histogram analysis. Thus, it was concluded that native FN interaction with VSMC was β1 and α5 integrin-dependent, whereas gFN interaction was integrin-independent.

Figure 4. Effect of anti-integrin function blocking antibody treatment on FN/gFN adhesion to VSMC.

Cells were serum starved for an hour and then incubated with 30 μg/ml of anti-β1, α5, β3 Ab or IgG (isotype control protein) for 45 minutes (for specific protocols see text). Group data (Panels A–C, F–H and K–M) are presented as relative binding probabilities ± SEM between 0.3 mg/ml protein coated silicon nitride probes and VSMC: A. In the presence of β1 antibody, adhesion of native FN to VSMC was significantly decreased. (p<0.05, t-test) (n=3). B. Blocking β1 (n=3) integrin with β1 antibody did not change adhesion to gFN to VSMC; and C. Function blocking β1 integrin antibody inhibits native FN adhesion but has no significant effect on adhesion probability of gFN (n=3). F. In the presence of α5 integrin antibody, adhesion of native FN to VSMC was significantly decreased. (p<0.05, t-test) (n=3). G. Blocking α5 (n=3) integrin with α5 Ab did not change adhesion to gFN to VSMC; and H. Function blocking α5 integrin antibody inhibits native FN adhesion but has no significant effect on adhesion probability of gFN (n=3). K. The presence of β3 antibody did not cause any significant difference in the adhesion of native FN to VSMC (p<0.05, t-test) (n=4). L. Blocking β3 (n=4) integrin with β3 antibody similarly did not change adhesion to gFN to VSMC; and M. Function blocking β3 integrin antibody has no significant effect on adhesion probability of native FN or gFN (n=4). Corresponding histograms (Panels D, E, I, J, N and O) depicting the distribution of rupture force-adhesion events with either anti-β1, anti-α5, anti-β3 antibody or IgG. D. Anti-β1 antibody inhibits FN-integrin interaction as depicted by the reduced adhesion density. E. Anti-β1 antibody does not inhibit the density of adhesion of gFN with VSMC. I. Anti-α5 antibody inhibits FN-integrin interaction as shown by the reduced adhesion density. J. Anti-α5 antibody does not impact density of adhesion of gFN with VSMC. N. Anti- β3 antibody does not inhibit FN-integrin interaction as shown by unchanged adhesion density. O. Anti-β3 antibody does not impact density of adhesion of gFN with VSMC.

Additional studies showed that the short-term (i.e. 0.5 Hz approach-retraction cycle) adhesive interactions between FN and VSMC were not dependent on β3 integrins. The probability of adhesion of FN with VSMC incubated with β3 function blocking antibody (77.5 ± 6.5 %) was not significantly different to the probability of adhesion of FN with cells treated with the control IgG antibody (78.4 ± 4.0 %) (Figure 4K) similar to previous observations made in our laboratory [52]. Inhibition of β3 integrin also did not alter the probability of adhesion of gFN to VSMC (85.9 ± 4.5 %/89.9 ± 4.3 %) (Figure 4L). In Figure 4M the probability of adhesion was normalized to the adhesion interaction of gFN with control IgG treated VSMC. Further, there was no change in the density of adhesion between either native FN (Figure 4N) or gFN (Figure 4O) with VSMC. These data thus show that β3 integrin does not play a role in, at least, the short-term interaction of native or gFN with VSMC.

Increased adhesion of glycated proteins to VSMC is dependent on RAGE

The interaction of glycated proteins with RAGE may lead to increased adhesion of glycated proteins to VSMC. In order to study the effect of RAGE on the adhesion of glycated proteins to VSMC, RAGE was inhibited using the small molecular weight inhibitor, FPS-ZM1. The inhibitor appeared to not non-specifically affect cellular function/viability as assessed by trypan blue exclusion, morphology and subsequent ability to undergo passage. Figure 5A and 5C demonstrates that treatment of VSMC with FPS-ZM1 attenuated the adhesion of gAlb and gFN to VSMC as compared to vehicle treated cells. The probability of adhesion of gAlb to VSMC decreased significantly from 55.9 ± 1.3 % to 32.1 ± 2.3 % in the presence of FPS-ZM1 (Figure 5A) (P<0.05). Similarly, FPS-ZM1 (P<0.05) significantly decreased adhesion of gFN with VSMC, 89.3 ± 2.3 % to 60.7 ± 1.5 % (Figure 5C). However in the absence or presence of FPS-ZM1 there were no observed differences in adhesion of native Alb (38.4 ± 0.4 %/35.6 ± 2.4 %) (Figure 5B) or native FN (64.6 ± 3.7 %/59.2 ± 1.3 %) (Figure 5D) to VSMC. Collectively, these data indicate that FPS-ZM1 treatment of VSMC attenuates RAGE-dependent adhesion of either gAlb or gFN while having no effects on the interaction between native Alb or native FN with the cells.

Figure 5. RAGE inhibitor FPS-ZM1 (50 nM) treatment of VSMC inhibits Glycated protein-RAGE interaction.

VSMC were incubated (4 hrs) with the RAGE inhibitor, FPS-ZM1 (50 nM) or an equivalent concentration of DMSO (vehicle). For specific protocol see text. Data are presented as relative binding probabilities ± SEM between 0.3 mg/ml protein coated silicon nitride tips and VSMC: A. In the presence of FPS-ZM1, adhesion of gAlb to VSMC was significantly decreased (p<0.05, 2 tailed unpaired t-test) (n=3). B. Inhibition of RAGE (n=3) with FPS-ZM1 did not change adhesion of native Alb to VSMC. C. FPS-ZM1 significantly decreased adhesion of gFN to VSMC (p<0.05, 2 tailed unpaired t-test) (n=4). D. Inhibition of RAGE (n=3) with FPS-ZM1 did not alter adhesion of FN to VSMC.

gFN and Native FN interacts with VSMC via different cell surface receptors

Additional inhibition studies were performed to corroborate our findings and further confirm that when glycated, FN acts through different, non-integrin, cell surface receptors when compared to the native protein. Passage 2 VSMC were incubated with and without soluble gFN or native FN to perform competitive inhibition studies with adhesion to the functionalized AFM probes. AFM studies revealed that there was no difference in the probability of adhesion between native FN with (87.8 ± 1.5 %) and without (88.5 ± 2.6 %) gFN in the media (Figure 6A). The aforementioned adhesion data is represented in line graphs between passage 2 cells from the same primary culture in Figure 6B. The line graphs show no significant differences in the probability of adhesion in each set of experiments between FN and VSMC with and without native FN in the media. These data suggest that gFN in the soluble form does not hinder native FN interaction with the cell surface integrins. Also, we observed no significant difference in the probability of adhesion events of gFN with (85.1 ± 1.8 %) and without soluble FN (87.8 ± 2.7 %) in the media (Figure 6C). Line graph analysis shows no significant differences in the probability of adhesion for each set of data between passage 2 cells from the same primary culture and gFN with and without soluble FN (Figure 6D). Thus, this suggests that native FN in the soluble form does not interfere with gFN interaction with cell surface RAGE.

Figure 6. Soluble FN and gFN do not block AFM probe-bound native FN and gFN interaction with VSMC.

VSMC were first incubated with and without soluble gFN or native FN for 45 minutes (for specific protocol see text). Panels A and B show relative binding probabilities between 0.3 mg/ml FN coated silicon nitride tip and VSMC: A. In the presence of gFN adhesion of FN to VSMC was not affected. Bar graph represented as relative binding probabilities ± SEM. (p<0.05, 2 tailed paired t-test) (n=4). B. Data in panel A represented as line graphs depicting no change in adhesion probability between native FN and VSMC from the same primary culture with and without soluble gFN. Panels C and D show relative binding probabilities between 0.3 mg/ml FN coated silicon nitride tip and VSMC: C. In the presence of FN adhesion of gFN to VSMC was not affected. Bar graph represented as relative binding probabilities ± SEM. (p<0.05, 2 tailed paired t-test) (n=4). D. Data in panel C represented as line graphs depicting no change in adhesion probability between gFN and VSMC from the same primary culture with and without soluble FN. Panel E shows relative adhesion probability ± SEM between 0.3 mg/ml gFN coated silicon nitride probes and VSMC in the absence and presence of soluble gAlb. Pre-exposure of cells with gAlb significantly (* P < 0.05, ANOVA) decreased the adhesion of probe-bound gFN in an apparent concentration dependent manner.

To further test the specificity of the gFN interaction with VSMC, cells were incubated with increasing concentration of gAlb (0.3 mg/ml and 0.9 mg/ml) before performing AFM adhesion studies using tips coated with gFN. gFN shows a significant inhibition of approximately 10% in the probability of adhesion with (82.7 ± 0.5) (p<0.05) (Figure 6E) and without (94.1 ± 0.9) (p<0.05) (Figure 6E) 0.3 mg/ml of gAlb in the media. Further, there was approximately 20% attenuation in the probability of adhesion of gFN in the presence of 0.9 mg/ml of gAlb decreasing the probability of adhesion further to 72.05 ± 2.4 (Fig 6E). Thus, gAlb inhibits the adhesion of gFN to VSMC in a concentration dependent manner. Together these data suggest that gFN and gAlb interact with RAGE and that native FN interacts with integrins.

Glycated protein-RAGE interactions can induce sustained activation of NFKB and increases the total protein content of NFKB in cultured small artery VSMC

Figure 7A shows western blotting experiments performed in passage 2 cultured VSMC. Rat lung tissue extract was used as a positive control. 10 μg total protein extract was loaded for the controls and two lanes with samples obtained from separate cell cultures. In Figure 7B the bar graph represents the expression of RAGE relative to total protein loaded as indicated. These data showed that VSMC have approximately 15% of the amount of RAGE expressed in lung tissue. The relative abundance of RAGE from the different VSMC samples was consistent (compare Sample 1 and Sample 2 in Figure 7 and Supplementary Figure 1). The amount of RAGE also increased with increasing amounts of total protein loaded. To demonstrate specificity of the RAGE antibody, we used lung tissue extract from RAGE-KO mice as a negative control. Figure 7A shows that there was no detectable band in the lanes loaded with the negative control when probed with the RAGE antibody. Thus, the data showed that early passage cultured VSMC express RAGE.

Figure 7. RAGE dependent activation of VSMC NFKB by glycated proteins.

Panel A. Immunoblot showing expression of VSMC RAGE protein. Lung tissue from RAGE-KO mice was used as negative control and lung tissue from SD rats was used as positive control. Panel B. Relative quantification of RAGE expression as determined by densitometric analysis. Protein extracts from two different VSMC samples (S1 and S2) were probed for RAGE expression. Panel C. Representative Western blot illustrating phosphorylated NFKB and α-SMA in VSMC following control, Alb (100 μg/ml) and gAlb (100 μg/ml) treatments with and without FPS-ZM1 (12.5 nM). Treatment exposure times of 30 mins, 2hrs and 4hrs were examined. VSMC show a sustained and increase in NFKB phosphorylation after 2hrs of incubation with gAlb while inhibition of RAGE with FPS-ZM1 attenuates the increase in NFKB phosphorylation. Panel D. Bar graphs showing densitometric analysis of phosphorylated NFKB levels as detected by Western blotting. Data are normalized to Phospho-NFKB/α-SMA levels after 2hrs incubation with control Alb and is represented as mean ± SEM. (★p<0.05, ANOVA; n=5, n=7 and n=6 for NFKBp65 with 30 mins, 2hrs and 4hrs exposure to native and gFN with and without FPS-ZM1 respectively). Panel E. Representative western blot showing total NFKB and α-SMA in VSMC treated as described above (C.). Although a trend towards increased total NFKB level was noted after 2hrs of incubation with gAlb this did not reach statistical significance (ANOVA, P = 0.15). Inhibition of RAGE with FPS-ZM1 attenuated the increase in NFKB protein levels. Panel F. Group data showing densitometric analysis of NFKB levels as described in E. Data are normalized to NFKB/α-SMA levels after 2hrs incubation with control Alb and is represented as mean ± SEM.

To extend the studies showing that the VSMC express RAGE, we examined whether the binding of glycated proteins initiated cell signaling by performing immunoblotting for phosphorylated NFKBp65. Passage 2 VSMC were incubated with 100 μg/ml of glycated or native Alb for 30 min, 2hr or 4hr and then harvested for protein extraction. Western blotting demonstrated that VSMC show a time-dependent increase in NFKB activity as shown by elevated NFKB phosphorylation at 2 hr (1.0 ± 0.3 vs 2.3 ± 0.5) and 4hr (0.9 ± 0.1 vs 2.3 ± 0.4; p<0.05) of exposure to gAlb as compared to native Alb. Data were normalized to phospho-NFKBp65 levels at 2hr exposure to native Alb (Figure 7C). Some of the cells were pretreated for 4 hr with 12.5 nM FPS-ZM1, a RAGE inhibitor, and then incubated with gAlb. Inhibition of AGE-RAGE interaction by FPS-ZM1 attenuated the increase in the level of phospho-NFKBp65 levels at both 2 hr (2.3 ± 0.5 vs 1.4 ± 0.3) and 4 hr (2.3 ± 0.4 vs 1.1 ± 0.1) of exposure to gAlb. Despite the changes in NFKB activity (phosphorylation) no statistically significant changes in newly synthesized (or total) NFKBp65 were observed when VSMC were incubated with gAlb (Figure 7E). Figure 7E however, appears to show a trend (ANOVA, P = 0.15) towards a transient increase in total NFKB p65 protein after 2 hr of exposure to gAlb (1.56 ± 0.4) when compared to cells exposed to native Alb (0.8 ± 0.2) and this was also attenuated by pretreating the cells with FPS-ZM1 (0.7 ± 0.2). Data were normalized to NFKBp65 levels at 2hr exposure to native Alb. Representative immunoblots for phospho-NFKBp65 (Figure 7D) and NFKBp65 (Figure 7F) are shown. Thus, glycated proteins can produce a sustained increase in NFKB activity via the AGE/RAGE pathway in small artery VSMC.

Discussion

Using AFM this study focused on determining how advanced non-enzymatic glycation influences cell surface binding/adhesion of the ECM protein, FN to integrins. Adhesion of gAlb was similarly studied as a non-ECM protein control. The results indicate that once glycated, both FN and Alb show increased adhesion to VSMC and that this occurs, at least partially, in a RAGE-dependent manner. Further, while native FN interacts with cell surface α5β1 integrins this mechanism of interaction is significantly attenuated when the protein undergoes glycation.

In the context of this study, AFM provided a method for quantifying adhesion and forces of interaction between native or glycated proteins and the cell plasma membrane. The adhesion forces are determined by the bending of a cantilever (of known spring constant), which is dependent on the strength of binding. Bio-functionalized (i.e. FN or Alb coated) sharp silicon nitride (SiN3) tips (attached to one end of the AFM cantilever) were used to analyze single molecule protein-receptor binding events as opposed to using fibronectin-functionalized beads and longer binding times which may lead to formation of focal adhesions as well as tethers [51,52].

FN interacts with α5β1 and αvβ3 integrins on the VSMC membrane forming focal adhesions and importantly mediates both outside-in and inside-out signaling. α5β1 and αvβ3 integrins are, for example, implicated in cell spreading, attachment [10], myogenic constriction [35], regulation of ion channels [8,37,38,41,63]. While a number of previous studies have used AFM to understand the mechanics of adhesive interactions between FN and integrins [51,52,55] the effect of glycation on these interactions has not been extensively considered. We hypothesized that non-enzymatic glycation may impact binding as this post-translational modification has been suggested to alter secondary and tertiary structures of proteins [25,34]. The data demonstrate an increased adhesive interactions of gFN with VSMC, as depicted by an approximate 10% increase in the probability of adhesion, number of binding events per curve and the increase in the adhesion events at the level of a single gFN-receptor interaction. As the increase in adhesive events of FN to VSMC with glycation could be due to several factors including i) stronger interaction with integrins ii) interaction with integrins and other glycated protein receptor such as RAGE iii) changes in interaction from integrins in the native form to RAGEs in the glycated form, we performed experiments to address these possibilities.

In the current studies, Alb was used as a non-ECM protein control on the basis that Alb does not interact with VSMC integrin receptors [51,52]. Native Alb did, however, show an adhesion probability of 40.1 ± 4.7 % which is assumed to represent non-specific interactions. Such non-specific interactions between the Alb functionalized AFM probe and VSMC could result from hydrogen bonds, electrostatic and hydrophobic interactions and depend on the loading force, retention time and unfolding of Alb [6,58].

Consistent with the gFN data, glycation of Alb significantly increased both the adhesion probability and the number of adhesion events per force curve to VSMC. Further, the adhesion forces were plotted as histograms with their corresponding number of rupture/adhesion events. We analyzed the primary force peak in the force-density of adhesion distribution as it is considered to represent the rupture of a single gAlb and VSMC surface binding partner. The other force peaks, or distributions at a higher or lower force of rupture, can be attributed to multiple protein-cell surface molecule bond ruptures; other ‘receptors’ with different binding affinity; local differences in binding interaction due to loading rate; binding of proteins on the tip in various conformational states; or unfolding of proteins on the cantilever tip. Although, we did not observe any significant changes in the force values for single Alb/gAlb-VSMC adhesion events there is an increase in the number of adhesions at the single molecule level between the glycated Alb and the VSMC (Figure 3D). This increase in adhesion was anticipated to be due to a binding interaction of gAlb with an alternate VSMC receptor such as RAGE.

Prior studies have shown that activation of integrins with histamine [55], LPA [48] or divalent metal ions [28] induces a stronger interaction between FN with integrins, which can be ruled out in the current studies as such activators were not used and our experiments were performed under identical conditions for the native and glycated proteins. For our studies, in addition to integrins, we focused on RAGE as likely candidate receptor for AGEs as the literature supports RAGE being the predominant cell surface receptor on VSMC that interacts with glycated proteins [31,44,60,66].

Additional support that our force measurements between FN and the VSMC surface are highly dependent on α5β1 integrin is provided by studies using specific function blocking antibodies. Published studies have shown that function-blocking antibodies can be used to prevent protein-receptor interactions and inhibit subsequent downstream signaling events [35,52,64]. Anti-β1 antibody reduced the adhesion probability of FN by approximately half whereas anti-α5 antibody reduced the adhesion probability of FN by approximately 20%. A possible explanation for the disparity in inhibition of adhesion probability by the anti-α5 and anti-β1 antibodies relates to the specificity/relative selectivity of these antibodies or perhaps differences in expression levels. Our VSMC were isolated from rat cremaster arterioles and the β1 function blocking antibody was anti-rat whereas α5 function blocking antibody was an anti-mouse polyclonal antibody. Thus the anti-β1 antibody could be more potent in blocking the cell surface α5β integrin dimer on rat VSMC. Both anti- α5 and β1 antibodies markedly decreased the single FN-integrin bond as depicted by the force-adhesion histograms. In contrast, treatment of the cells with a β3 function-blocking antibody did not have any significant effect on the adhesive interactions of FN with VSMC. Earlier studies of Meininger and colleagues [52] and Li et al. [33] yielded similar results and showed that there was decreased adhesion of FN to cultured cells with anti-α5 and anti-β1 antibody treatment but not with anti-β3 antibody. Thus, we can infer that short-term interactions of FN with VSMC are predominantly mediated by α5 and β1 integrins.

In contrast to the adhesion of FN to VSMC, gFN binding with VSMC appeared independent of integrins. Thus, blocking α5, β1 or β3 integrins did not result in significant differences in the enhanced density or probability of adhesion of gFN with VSMC. This suggests that the process of glycation hinders the interaction of FN with the cell surface integrins on VSMC. This is a novel idea suggesting that glycation blocks the integrin-mediated signaling functions of ECM proteins such as FN, which, otherwise helps in maintaining vascular homeostasis. A recent study in human umbilical vein endothelial cells (HUVECs) suggests that glycation of vitronectin inhibited the binding of vitronectin to αvβ3 integrin by obstructing the binding of the RGD domain [59]. Similarly, MGO glycated FN used in our studies could also interfere with the RGD binding domain thus negating the effect of anti-integrin antibodies in reducing adhesion with VSMC.

In our studies we used FPS-ZM1, a multimodal RAGE inhibitory molecule, to prevent the binding of RAGE to its ligands. Specifically the molecule exerts its inhibitory action by blocking the V domain of RAGE [65]. Although this is a recently described RAGE inhibitor several in-vivo and in-vitro studies have validated the binding specificity of FPS-ZM1 with RAGE and the blocking of RAGE-mediated downstream signaling cascades [9,39,50,67]. The concentration of FPS-ZM1 used in our experiments was also well within the effective and non-toxic levels as described in the literature [11]. Our studies suggest that FPS-ZM1 treatment inhibits the adhesion of gFN and gAlb by almost 30% and 20% respectively with VSMC whereas there was no significant change in the binding interaction of native FN or native Alb. This suggests that both FN and Alb in their glycated forms interacts with RAGEs on VSMC surface. However, such interaction with RAGE is not seen with the native proteins.

To further establish a role for RAGE, we examined binding interactions of glycated vs. native proteins for VSMC incubated with soluble gFN or FN and then performed adhesion studies with tips coated with native FN or gFN respectively. This approach was used to determine possible competitive inhibition. There were, however, no apparent differences in the binding of FN with α5β1 integrins with and without gFN in the media. We also did not observe any significant difference in the adhesion characteristics of gFN with RAGE on cells incubated with native FN. Thus, this suggests that native and glycated FN interact with disparate VSM cell surface receptors. We can also infer that soluble gFN and FN does not inhibit the binding of FN with α5β1 integrins or the binding of gFN with RAGE, respectively. Physiologically under hyperglycemic conditions there is increased production of FN [47] and reduced matrix protein turnover [20]. Thus, under a prolonged state of hyperglycemia there is more opportunity for FN glycation. This increased level of glycated FN in the vasculature can be a ‘double edged sword’ by not only activating RAGE but also by diminishing integrin-mediated downstream signaling processes which are critical in maintaining arteriolar functions. Further evidence regarding the specificity of interaction of glycated proteins with RAGE on VSMC comes from the observation that gAlb inhibited the adhesion between gFN and VSMC in a concentration dependent manner. Thus, gFN interaction with VSMC is dependent on RAGE and glycated proteins, per se, will interact with this receptor.

In order to confirm glycated protein-RAGE interactions in our VSMC we first performed immunoblotting to confirm RAGE expression in these cells. We observed that cultured VSMC cells isolated from rat cremaster arterioles grown in DMEM/F-12 culture media with 15.2 mM glucose express RAGE (Figure 7A). As mentioned, glycated proteins can interact with RAGE to increase NFKB activity. We noted RAGE-mediated increases in phosphorylated NFKB levels in the cells incubated with gAlb. These data are consistent with other studies performed in cultured VSMC from large conduit vessels stimulated with glycated proteins [29,40]. Sustained increases in NFKB activity are associated not only with increases in the phosphorylation levels of NFKB but also in the de novo synthesis of NFKB protein [22]. Long-term and sustained expression of NFKB have been observed under hyperglycemic conditions such as diabetes [3]. In the context of the present study we did not find a significant change in levels of total NFKB protein. Taken together the data suggest that glycated protein interacts with RAGE on the VSMC to activate the AGE/RAGE signaling pathway leading to an increase in NFKB activity.

Additional experiments were performed to determine the effects of gFN, on NFKB activation (phospho-NFKBp65) similarly to those described for gAlb (data not shown). While there was a trend towards a time-dependent increase in NFKB activation greater variability was observed. Interestingly data from other groups have shown that NFKB activity is significantly increased in fibroblasts and VSMC plated on native FN coated dishes over a 2 to 4 hr period [42]. Furthermore, NFKB itself, transcriptionally upregulates FN expression as shown by an increase in FN expression at both the mRNA and protein levels in rat hepatocytes when stimulated with interleukin-β [32]. This positive feedback conceivably causes an increase NFKB activation in VSMC exposed to the native form of FN which may contribute to the increased variability of the FN/gFN data in terms of NFKB phosphorylation. On the basis of this we have chosen the phospho-NFKBp65 response to gAlb as a down-stream indicator of RAGE activation.

Collectively, our data indicate that the process of advanced non-enzymatic glycation of FN not only promotes RAGE mediated downstream signaling pathway but also inhibits its interaction with cell surface α5β1 integrins. As a result, this changes the very nature of interaction of FN with VSMC from cell surface integrins in the native form to RAGE in the glycated form.

PERSPECTIVES

Binding of extracellular matrix proteins through integrins regulates vascular tone via multiple mechanisms including regulation of ion channels; modulation of cellular adhesion and acute remodeling of the cortical cytoskeleton. Advanced glycation of the ECM protein FN alters its interaction with VSMCs from apparent binding to cell surface integrins to RAGE. Under pathological conditions such as in diabetes both attenuated integrin activation and augmented RAGE activation (together with associated signaling events) in VSMC may lead to arteriolar dysfunction. Thus, this study provides us with a novel mechanism that adds to the pathophysiology of vascular dysfunction as seen in diabetes.

List of abbreviations

α-SMA

α-smooth muscle actin

AFM

Atomic force microscopy

AGE

advanced glycation endproducts

Alb

Albumin

ECM

Extracellular matrix proteins

FN

Fibronectin

gAlb

Glycated albumin

gFN

Glycated fibronectin

MMP

Matrix metalloproteinases

RAGE

receptor for advanced glycation endproducts

RAGE-KO

Receptor for advanced glycation endproducts knockout

VSMC

Vascular smooth muscle cells