α5β1 integrin signaling mediates oxidized low-density lipoprotein-induced inflammation and early atherosclerosis

Abstract

Objective: Endothelial cell activation drives early atherosclerotic plaque formation. Both fibronectin deposition and accumulation of oxidized low-density lipoprotein (oxLDL) occur early during atherogenesis, and both are implicated in enhanced endothelial cell activation. However, interplay between these responses has not been established. The objective of our study was to determine whether endothelial matrix composition modulates the inflammatory properties of oxLDL.

Approach and results: We now show that oxLDL-induced nuclear factor-κB activation, proinflammatory gene expression, and monocyte binding are significantly enhanced when endothelial cells are attached to fibronectin compared with basement membrane proteins. This enhanced response does not result from altered oxLDL receptor expression, oxLDL uptake, or reactive oxygen species production, but results from oxLDL-induced activation of the fibronectin-binding integrin α5β1. Preventing α5β1 signaling (blocking antibodies, knockout cells) inhibits oxLDL-induced nuclear factor-κB activation and vascular cell adhesion molecule-1 expression. Furthermore, oxLDL drives α5β1-dependent integrin signaling through the focal adhesion kinase pathway, and focal adhesion kinase inhibition (PF-573228, small interfering RNA) blunts oxLDL-induced nuclear factor-κB activation, vascular cell adhesion molecule-1 expression, and monocyte adhesion. Last, treatment with the α5β1 signaling inhibitor, ATN-161, significantly blunts atherosclerotic plaque development in apolipoprotein E-deficient mice, characterized by reduced vascular cell adhesion molecule-1 expression and macrophage accumulation without affecting fibrous cap size.

Conclusions: Our data suggest that α5β1-mediated cross-talk between fibronectin and oxLDL regulates inflammation in early atherogenesis and that therapeutics that inhibit α5 integrins may reduce inflammation without adversely affecting plaque structure.

Keywords: atherosclerosis; fibronectins; inflammation; integrin alpha5beta1; lipoproteins, LDL.

Figure 1. Matrix controls oxLDL-induced inflammation

(A) HAECS on either BM or FN were treated with oxLDL (100 μg/mL) for 6 hours and human primary monocyte adhesion was analyzed. Results are expressed as percent adherent to the endothelial monolayer. Representative images are shown, n=4. (B and C) HAECs on different matrices were treated with oxLDL for the indicated times. mRNA was analyzed by qRT-PCR for VCAM-1, ICAM-1, and GAPDH, n=4. (D, E, and F) VCAM-1, ICAM-1, and GAPDH protein expression was determined by Western blotting. Representative Western blots are shown, n=7. Values are means ± SE, *p < 0.05 compared with no treatment condition, #p < 0.05 comparing matrices.

Figure 2. Matrix regulates VCAM-1 expression through NF-KB

(A) HAECs plated on BM or FN were treated with oxLDL (100 μg/mL) for the indicated times. Immunoblotting was performed for P-NF-κB (p65, Ser536) and GAPDH. Representative Western blots are shown, n=5. (B) HAECs on different matrices were treated with oxLDL for 1 hour and NF-κB nuclear translocation determined by immunocytochemistry. Representative images are shown, n=3. (C) HAECs were treated with Bay11-7082 (10 μM, 1 hr) and oxLDL-induced VCAM-1 expression analyzed. Representative Western blots are shown, n=3. (D). oxLDL-induced VCAM-1 expression was determined in HAECs infected with either a CMV or SR-IκBα expressing adenovirus. Representative Western blots are shown, n=3. Values are means ± SE, *p < 0.05 compared with no treatment condition, #p < 0.05 comparing matrices.

Figure 3. Integrin α5 regulates oxLDL-induced VCAM-1 and NF-KB activation

(A) HAECs were treated with oxLDL (100 μg/mL) for 30 min, and α5β1 activation was determined by measuring GST-FNIII_9-11 retention by Western blotting. Representative images are shown. n=4. (B) α5 surface expression in HAECs treated with oxLDL was determined by FACS analysis and is expressed as average mean fluorescence intensity, n=3. (C) HAECs pretreated with the nonblocking (11E5, 40 μg/ml) or fibronectin-blocking antibody (16G3, 40 μg/ml) for 1 hour were treated with oxLDL for 6 hours and VCAM-1 expression was determined by Western blotting, n=4. (D and E) HAECs plated on FN were treated with blocking antibodies for α5 (P1D6, SNAKA52), αvα3 (LM609), or αvα5 (P1F6) at 10 μg/mL for 1 hour, and oxLDL-induced VCAM-1 expression (6 hr) and NF-κB activation (1 hr) were determined by Western blotting, n=4. (F) oxLDL-induced VCAM-1 expression was determined in conditionally immortalized MAE cells either wildtype (CMV) or deficient for α5 integrins (CRE). Representative Western blots for VCAM-1, integrin α5, integrin αV, and ERK are shown, n=3. Values are means ± SE, *p < 0.05 compared with no treatment condition. #p < 0.05 compared with α5 deficient (Cre) cells.

Figure 4. FAK regulates oxLDL-induced VCAM-1 and NF-KB activation

(A) HAECs were treated with oxLDL (100 μg/mL) for the indicated times. Immunoblotting was performed for phospho-FAK Y397, Y576, Y577, total FAK, and ERK. Representative images are shown, n=4. (B) HAECs pretreated with blocking antibodies to α5 (P1D6, SNAKA52), αvβ3 (LM609), or αvα5 (P1F6) were treated with oxLDL for 1 hour. Immunoblotting was performed for phospho-FAK Y397 and ERK. Representative images are shown, n=3. (C) HAECs were pretreated with PF-573228 (10 μM; 1 hour) and treated with oxLDL for indicated times. Immunoblotting was performed for phospho-FAK Y397, P-NF-κB, and ERK, n=4. (D) HAECs pretreated with PF-573228 were treated with oxLDL for 6 hours, and VCAM-1 expression was determined by Western blotting. Representative images are shown, n=3. (E) HAECs were transfected with anti-FAK siRNA and treated for the indicated times. Immunoblotting was performed for FAK, P-NF-κB, and β-tubulin. Representative images are shown, n=4. (F) HAECs were transfected with anti-FAK siRNA then treated with oxLDL for 6 hours. Immunoblotting was performed for FAK, VCAM-1, and ERK. Representative images are shown, n=4. Values are means ± SE, *p < 0.05 compared with no treatment condition. #p < 0.05 comparing treatment conditions.

Figure 5. Inhibiting integrin α5 signaling in vivo is sufficient to delay atherosclerosis

8 week old ApoE^−/− mice were fed a high fat, Western Diet for 8 weeks during which mice were treated with either saline or ATN-161 (5 mg/kg) by intraperitoneal injection. (A, B) Oil Red O staining of aortas was performed, and plaque area was analyzed as the percent Oil Red O positive area. (C, D) Plaque size in the aortic root was quantified following Russell-MOVAT pentachrome staining. Plaque area was calculated as the neointimal area inside the internal elastic lamina. (E, F) VCAM-1 expression in the carotid sinus and innominate artery was determined by immunofluorescence immunohistochemistry and expressed as the positive area in the vessel wall. Images taken at 20X with insets at 10X. Analysis was performed using NIS elements software. n=8-10 mice per group. Values are means ± SE.

Figure 6. Inhibiting integrin α5 signaling in vivo reduces macrophage content without altering fibrous cap formation

(A) Aortic roots from saline and ATN-161 treated mice were stained by immunohistochemistry for Mac2 (macrophage marker, green), smooth muscle actin (smooth muscle cell marker, red) and DAPI (blue). Representative 40X images are shown. (B) The number of individual plaques per aortic root were quantified for each mouse. (C) Macrophage area was analyzed by quantifying the Mac2-positive area for each aortic root. (D) Smooth muscle area was analyzed by quantifying the SMA-positive area for each aortic root. (E) The percentage of the plaque that was SMA positive was calculated by dividing the SMA-positive area by the total plaque area and averaged for each group. (F) Thickness of the individual fibrous caps (averaged from more than four regions per cap) was calculated and expressed as the average cap thickness within each group. (G) Plaques were scored for SMA-positive fibrous caps, and the percentage of fibrous cap-positive plaques were calculated by dividing by the total number of plaques in each group. All analysis was performed using NIS elements software, n=10 mice per group. Values are means ± SE.