Role of ADAMTS-1 in Atherosclerosis
Remodeling of Carotid Artery, Immunohistochemistry, and Proteolysis of Versican
Objective— We investigated the potential role of ADAMTS-1 (a disintegrin and metalloprotease with thrombospondin motif type I) in atherogenesis.
Methods and Results— ADAMTS-1 is expressed at the highest levels in the aorta when compared with other human tissues examined. Immunolocalization studies in human aorta and coronary artery indicate that ADAMTS-1 expression is mainly seen at low levels in the medial layer, but upregulated in the intima when plaque is present. We found that ADAMTS-1 mRNA levels are significantly higher in proliferating/migrating cultured primary aortic vascular smooth muscle cells (VSMCs) compared with resting/confluent cells. Using the mouse carotid artery flow cessation model, we show that there are differences in vessel remodeling in ADAMTS-1 transgenic/apoE-deficient mice compared with apoE deficiency alone, particularly a significant increase in intimal hyperplasia. We show that ADAMTS-1 can cleave the large versican containing proteoglycan population purified from cultured human aortic VSMCs. Finally, using versican peptide substrates, we show data suggesting that ADAMTS-1 cleaves versican at multiple sites.
Conclusion— We hypothesize that ADAMTS-1 may promote atherogenesis by cleaving extracellular matrix proteins such as versican and promoting VSMC migration.
Complications from atherosclerosis are the most common cause of death in Western societies.1 It is a form of chronic inflammation involving cells and proteins that normally reside in the artery and/or infiltrate the artery from the lumen. They include lipoproteins, macrophages, vascular smooth muscle cells (VSMCs), endothelial cells, and extracellular matrix proteins, such as proteoglycans, collagens, and elastins.2
See page 12
Early in atherogenesis, VSMCs from the media are thought to migrate into the intima and contribute to the development of atherosclerotic lesions. Although what initially triggers these events is not known, it is thought that proteases released by VSMCs degrade the matrix proteins in the intima, particularly the main proteoglycan of the arterial intima versican, making the intima more permissive for invasion by VSMCs. One recently discovered family of metalloproteases, the ADAMTS family, might play a key role in atherogenesis by modulating the degradation of versican and possibly other proteoglycans.
The first member of this family to be identified is ADAMTS-1.3,4 It has been observed that ADAMTS-1 mRNA is upregulated substantially in human umbilical vein endothelial cells and cardiac microvascular endothelial cells under shear stress, suggesting regulation during flow-dependent vascular remodeling.5 ADAMTS-1 has been shown to cleave the proteoglycan versican, which is expressed by VSMCs.6–8 Versican can exist in 4 isoforms (V0, V1, V2, and V3), depending on alternative splicing of the chondroitin sulfate containing glycosaminoglycan domains. V0 versican contains all possible domains, whereas the glycosaminoglycan-alpha and glycosaminoglycan-beta domains are spliced out in the V1 and V2 versican isoforms, respectively.7 ADAMTS-1 and ADAMTS-4 have been shown to cleave V1/V0 versican at the Glu441-Ala442/Glu1428-Ala1429 bond and the product of this cleavage was shown to be present in human atherosclerotic plaques by immunohistochemistry using neo-epitope antibodies.8 ADAMTS-1 has also been shown to have a role in matrix remodeling during ovulation in mice, which involves dissolution of connective matrix and cellular layers. ADAMTS-1–deficient mice displayed impaired ovulation, and the authors proposed that this was at least partially caused by lack of versican degradation.9
In addition to its potential role in VSMC migration, versican is also thought to contribute to atherosclerosis by binding and protecting growth factors and cytokines from degradation. Versican can also bind apoB-containing lipoproteins, promote their modification, and enhance uptake by macrophages.10,11 Versican can interact directly with a number of matrix proteins, adhesion molecules, and membrane proteins such as CD44.12,13 Together, they can form a network of densely packed matrix proteins, which is expandable but resilient. Decreased versican levels are correlated with human abdominal aortic aneurysm, consistent with its role in maintaining visco-elasticity in the vessel wall.14
In the present study, we look at the potential contribution by ADAMTS-1 in atherogenesis by examining the expression pattern of ADAMTS-1 in human lesions, looking at changes in expression levels of ADAMTS-1 in primary human VSMCs under different proliferating conditions, investigating the effect of overexpression of ADAMTS-1 on intimal hyperplasia using the murine carotid artery ligation model, and characterizing possible cleavage sites using versican peptide substrates. Our results demonstrate that ADAMTS-1 may contribute to the development of atherosclerosis through degradation of versican and other potential proteins involved in regulating VSMC migration.
Real-Time Reverse-Transcription Polymerase Chain Reaction
Distribution of ADAMTS-1 in human tissues and parts of the cardiovascular system was analyzed using cDNA made from pooled human poly A+ RNA (Cat. #1420 to 1, 1421 to 1, and 1427 to 1; Clontech Laboratories, Inc). For both human and mouse samples, ribosomal protein 36B4 was used as an internal control.
For more information about RNA preparations, reaction conditions, and primer sequences, please see http://atvb.ahajournals.org.
The following paraffin-embedded human materials were used: fatty streak aortic lesion from an 18-year-old man (Addenbrooke’s Histopathology Tissue Bank, UK), a type III–IV/VII aortic lesion (Pathology Department, Sahlgrenska Hospital, Sweden), and a type III–IV/VII lesion from coronary artery of a 53-year-old man (Ullevål Hospital, Norway). For more information on staining conditions, please see online supplement.
Four different polyclonal anti-peptide antibodies against ADAMTS-1 were generated for immunohistochemistry and Western blotting (Agrisera). Please see online supplement for description of peptides used to generate the ADAMTS-1 antibodies/purification and of antibodies used to identify macrophages and VSMCs.
Primary human aortic VSMCs (CC-2571; Clonetics) were cultured in SmGM-2 Bullet Kit (CC-3182; Clonetics) on ordinary or collagen-coated flasks and cultured according to supplier. Cells from passage 6 were seeded at a density of 5000 cells/cm2 for each condition, migrating/proliferating (M/P), or resting/confluent (R/C) (n=7). M/P cells were harvested at ≈50% confluence (3 to 4 days after plating) and the R/C cells were harvested 2 to 3 days after the cells had reached confluence (days 13 to 14).
For information about the culture conditions for DON cells overexpressing ADAMTS-1, please see online supplement.
ADAMTS-1 transgenic mice were generated using standard techniques at AstraZeneca. ADAMTS-1 transgenic (C57/BL6) and wild-type mice (C57/BL6) (Bommice M&B, Denmark) were euthanized at 7 weeks and tissues were taken out for real-time reverse-transcription polymerase chain reaction analysis (n=5).
Twenty-one male ADAMTS-1 transgenic/apoE-deficient mice (C57BL/6J) and 21 male apoE-deficient mice (C57BL/6J) for the carotid artery ligation experiments were fed a Western diet from age 10 to 14 weeks. Ligation of the left common carotid artery near the bifurcation was performed at age 14 weeks as described before.15 At 3, 7, and 14 days after ligation, the animals were perfusion-fixed and the carotid arteries were taken out for further study (n=7).
Each carotid artery was divided in 2 segments and marked with tissue touche at the distal part. Paraffin sections were collected at every 200-μm level throughout the whole length of the carotid artery for morphometric analysis, and additional sections were collected at every other 200-μm level for histological analysis.
Paraffin sections were counterstained with hematoxylin–eosin. Perimeters of the lumen, internal elastic lamina, and external elastic lamina were obtained by tracing the contours with eyepiece equipment Lucivid (BioMetricSystems GmbH) mounted to a Leica DM RBE microscope. Intima area (area between internal elastic lamina and lumen) and media area (area between external elastic lamina and internal elastic lamina) were provided by Microvid Image Access Analytic v1.0c (Bildanalyssystem AB) software program. These measurements were performed on all sectioned levels of the carotid to obtain a mean value for the media area, intima area, and lumen area along the whole length of the artery for each animal. All measurements were performed blind.
Statistical analysis between groups was performed by 1-way-ANOVA test using Astute version 2.
Construction and Transfection of ADAMTS-1 Expression Vector
Please see online supplement.
Purification of Human ADAMTS-1
DON cells expressing hADAMTS-1 were cultured on plastic until confluence. Media was collected for purification procedure.
Purification of Total Proteoglycan Secreted by Primary Aortic VSMCs
35S and 3H labeling and purification of total proteoglycan secreted from VSMCs were performed as described previously.16
Cleavage of Proteoglycan Isolated From Cultured Primary Human VSMCs and Versican Peptides
One hundred microliters of total proteoglycan preparation (10 000 to 15 000 cpm/35S) was cleaved with 3.8 μg ADAMTS-1 at 37°C overnight and analyzed by size-exclusion chromatography.
Freeze-dried peptide was dissolved and mixed with ADAMTS-1 at a 10:1 molar ratio, incubated overnight at 37°C, and analyzed by high-performance liquid chromatography and mass spectrometry/mass spectrometry.
Real-Time Reverse-Transcription Polymerase Chain Reaction Analysis of ADAMTS-1 Expression in Human Tissues, Cultured Human Primary Aortic VSMCs, and Tissues From ADAMTS-1 Transgenic Mice
To determine the distribution of ADAMTS-1 mRNA in humans, real-time reverse-transcription polymerase chain reaction was performed on pooled cDNAs from human tissues/organs and from different parts of the cardiovascular system. ADAMTS-1 was expressed at the highest levels in the heart, lung, adipose tissues, and brain (Figure 1A), and in the cardiovascular system in the aorta (Figure 1B).
To determine whether there is any correlation between the level of ADAMTS-1 mRNA and cell M/P, we compared the level of ADAMTS-1 mRNA between M/P and R/C primary aortic human VSMCs. In 3 separate experiments with a total of 7 samples for each condition, we observed significantly higher levels of ADAMTS-1 in M/P cells compared with R/C cells (P<0.001) (Figure 1C).
In a separate experiment, BrdU incorporation and cell protein concentration were measured over time to measure cell proliferation and protein expression. Peak BrdU incorporation corresponded to when cells were ≈50% confluent (days 3 to 4), with BrdU absorbance levels approaching 0 after cells reach confluence (after day 11), confirming that the cells were indeed M/P or R/C (Figure IA, available online at http://atvb.ahajournals.org). Total cell protein content in each well, however, continued to increase, indicating that cells were healthy and were continuing to express proteins at high levels (Figure IA). Similar pattern of BrdU incorporation and protein expression were also observed in uterine and coronary artery smooth muscle cells (Figure IB and IC, respectively).
To confirm that ADAMTS-1 transgenic mice overexpressed ADAMTS-1 message, real-time reverse-transcription polymerase chain reaction was performed on various tissues from mice euthanized at 7 weeks of age (Figure 1D). Results show higher ADAMTS-1 expression in the transgenic mice compared with wild-type mice.
ADAMTS-1 Localization in Atherosclerotic Lesions
Immunolocalization studies were performed on a human aortic fatty streak lesion (type I) and on more advanced type III–IV/VII lesions from the coronary artery and aorta (Figure 2). Figure 2B shows the size and location of the fatty streak in relation to the medial layer, which is stained using antibodies against smooth muscle α-actin. In the fatty streak lesion, antibodies against ADAMTS-1 stained both the VSMCs and foam-like cells in the intima (Figure 2C, 2E, and 2F). ADAMTS-1 staining is also observed in the medial layer, but at much lower intensity compared with the intima (Figure 2C).
In the more advanced lesions with acellular scar tissue, ADAMTS-1 staining was mostly localized to the matrix-like core at the base of the lesions in both the coronary artery and the aorta, although weak staining of VSMCs is still present (see arrow in Figure 2H; 2J to 2L). Again, staining pattern for ADAMTS-1 and VSMCs overlap (see arrow in Figure 2H; 2I). Although not all sections contained macrophages and foam cells, we see weak staining pattern in many sections where they are present (Figure 2G; see arrow in Figure 2L). In sections in which an intact endothelial cell layer was observed, ADAMTS-1 staining was also observed in these cells (Figure II, available online at http://atvb.ahajournals.org). We have examined ADAMTS-1 staining using aortic and coronary samples from a total of 13 individuals, including 2 normal arteries, 4 with fatty streak lesions, and 6 with advanced lesions; we have observed consistent results in staining pattern.
Preabsorbed antibodies did not show staining, demonstrating specificity (Figure 2D). The ADAMTS-1 antibodies were pre-absorbed or pre-incubated with peptides that rabbits were immunized with to eliminate specific recognition of ADAMTS-1. Foam cells/macrophages and VSMCs were identified using antibodies against HAM-56 and α-actin, respectively (Figure 2B, 2E, 2F, 2G, and 2I).
Vascular Remodeling in Murine Carotid Artery Flow Cessation Model
To determine whether ADAMTS-1 has a role in smooth muscle migration and proliferation in an in vivo model, we used the mouse carotid artery ligation model to compare neointima formation in mice overexpressing ADAMTS-1 with apoE-deficient background and control apoE-deficient mice at days 3, 7, and 14 after ligation (n=7) (Figure 3).15,17 Because remodeling can vary throughout the ligated area, we took measurements of the neointima, media, and lumen from sections taken every 200 μm throughout the entire length of the ligated artery and averaged these values for each carotid artery. The data from each mouse were then combined with the data from the other mice in the same group and averaged to calculate the mean value. In the control group at day 14, one sample was excluded because the neointima formed was so complex that we could not distinguish different tissue boundaries. Total area was calculated as the sum of the area for the media, lumen, and neointima.
We found that in the ADAMTS-1 transgenic/apoE-deficient group there was a significant increase in the mean area of the neointima compared with the control mice at day 14 (21550±3733 versus 9905±1019 μm2; P<0.02) (Figure 3A). Figure 3A and 3B graphically show the changes occurring over time for the mean medial, neointimal, lumenal, and total areas. In both the transgenic and control groups, lumen size was maintained even with neointimal growth by positive/expansive remodeling of the vessels at days 3 and 7. By day 14 in the control group, however, positive remodeling was followed by negative remodeling to return the vessel to its original size; because of the neointima, the mean area of the lumen is significantly smaller compared with that on day 3 (P<0.01). In the transgenic group, however, the total area remained significantly larger (P<0.03), and the size of the lumen was not significantly different between days 3 and 14, despite the larger neointima.
Figure 3C shows representative sections from a coronary artery taken from an ADAMTS-1 transgenic/apoE-deficient and ADAMTS-1 wild-type/apoE-deficient control mice at days 3, 7, and 21. Although the sections cannot be compared directly because they are only one of many sections used to calculate the average values for each time point, Figure 3C shows the kind of changes occurring during the growth of the neointima.
ADAMTS-1 Can Cleave Versican Secreted by Primary Aortic VSMCs
To test whether ADAMTS-1 has a role in the degradation of versican, we tested whether proteoglycans secreted by primary aortic VSMCs are cleaved by ADAMTS-1 (Figure 4). Proteoglycans purified from these cells can be separated by size-exclusion chromatography into 2 populations: the large proteoglycan population that is composed primarily of versican and the small population that is composed of smaller proteoglycans.16 Incubation of total proteoglycan with ADAMTS-1 decreased the size of the larger peak and increased the size of the smaller peak (Figure 4A and 4B). In addition, incubation of ADAMTS-1 with only purified large proteoglycan population shows the disappearance of the large peak and appearance of a smaller peak (unpublished data, 2000).
Sandy et al demonstrated that ADAMTS-1 can cleave V1/V0 versican at the Glu441-Ala442/Glu1428-Ala1429 bond.8 However, it is likely that other cleavage sites exist. ADAMTS-4 and ADAMTS-5/11, for example, can cleave aggrecan at 5 sites.18 Using different synthetic 40 to 42 amino acid versican peptides, we were able to detect 2 additional cleavage sites by high-performance liquid chromatography and sequence analysis using MS/MS: Glu950-Gly951 bond in V0/V2 versican and Tyr1410-Ile1411/Tyr423-Ile424 bond in V0/V1 versican (Figure 5A to 5F).
The findings in this article suggest a role for ADAMTS-1 in atherosclerosis and possibly in vascular thrombosis. Although the expression of ADAMTS-1 is generally low in normal tissues and organs, it is possible that expression is induced by stimuli, as has been demonstrated in vitro by the inflammatory cytokine IL-1 and tumor necrosis factor-α, in vivo by lipopolysaccharide and both in vitro and in vivo by parathyroid hormone.3,19–20 In human fatty streak lesions, we observe ADAMTS-1 staining with stronger intensity in the VSMCs and foam cells in the lesion compared with the medial layer by immunohistochemistry. This is similar to the staining pattern we observe in ADAMTS-1 transgenic/apoE-deficient mice administered a high-fat diet (unpublished results, 2003). This is also consistent with our in vitro cell experiments. We observe significantly more ADAMTS-1 expression in migrating and proliferating VSMCs, as those presumably found in lesions, compared with R/C cells, as those found in the medial layer.
In more advanced type III–IV/VII human lesions, ADAMTS-1 staining is observed with strong intensity at the base of the lipid core containing matrix-like elements and adjacent to the medial layer in the aorta and coronary artery. The latter observation is consistent with in vitro studies using COS-7 cells, which showed association with the extracellular matrix.21 ADAMTS-1 staining was observed in some of the macrophages/foam cells when they were present, although they were not present in all sections (Figure 2L). In addition, mRNA for ADAMTS-1 is detected in THP-1 cells by real-time reverse-transcription polymerase chain reaction analysis, and expression is induced by the inflammatory cytokine interferon-γ, providing additional support that macrophages may also express ADAMTS-1 (unpublished data, 2000). These results suggest that localization of ADAMTS-1 may change with progression of disease.
Using a mouse carotid artery ligation model to study the role of ADAMTS-1 in development of neointima in vivo, we found that overexpression of ADAMTS-1 in apoE-deficient mice significantly increased intimal hyperplasia compared with control apoE-deficient mice by day 14. Although the carotid artery ligation model is not a model for atherosclerosis, many of the changes taking place resemble those observed early in atherogenesis, such as VSMC migration and proliferation.15
The increased mass by the growth of the neointima is compensated by the expansion of the blood vessel to maintain lumen size and blood flow. Although the expansion of blood vessel may be a beneficial compensatory mechanism, this remodeling has been associated with vulnerable plaques.17 In the control group, the positive remodeling is followed by negative remodeling; these changes have been described for C57/BL/6J mice.17 In the transgenic group, however, the vessel remains significantly expanded. ADAMTS-1 transgenic mice appear to have reduced capacity for negative remodeling compared with controls, but they do not appear to be impaired in their capacity for positive remodeling. Although not significant, there appears to be even more positive remodeling in the ADAMTS-1 transgenic mice compared with control mice from days 3 to 7 (34% versus 20%, respectively).
Using the same carotid artery ligation model, Galis et al found that there was decreased intimal hyperplasia in mice deficient in matrix metalloproteinase-9 compared with control mice and less lumen reduction/negative remodeling.22 They attributed the latter to the decreased capacity for the VSMCs to contract and constrict vessels. Therefore, in terms of intimal hyperplasia, the results are consistent with metalloproteases promoting smooth muscle cell migration and proliferation. However, in terms of negative remodeling, matrix metalloproteinase-9 appears to contribute to negative remodeling, whereas ADAMTS-1 appears to inhibit it. This would suggest that different mechanisms are operating in formation of intimal hyperplasia and negative remodeling and that these 2 metalloproteases have overlapping and distinct functions.
As reported previously by Sandy et al, our data also show that ADAMTS-1 can cleave the proteoglycan versican, which is thought to have a role in the development of lesions by regulating aortic VSMC migration and in plaque stability by affecting the strength of the fibrous cap.8 To test potential cleavage sites, we used peptide substrates corresponding to the different sites, although it is not proof that versican is cleaved at these sites in vivo. In addition to the cleavage site described by Sandy et al, we also observed 2 additional potential cleavage sites, at the Glu950-Gly951 (V1) and Tyr423-Ile424 (V1) bonds, in a manner analogous to the multiple cleavage sites described for the proteoglycan aggrecan by ADAMTS-4 and ADAMTS-5/11.18 Sandy et al also predicted a second cleavage site in V0/V2 versican at the Glu405-Gln406 bond; we were able to cleave this bond using a synthetic versican peptide (unpublished data, 2000). For aggrecan, cleavage by ADAMTS-4 and ADAMTS-5/11 leads to the release of aggrecan degradation products out of cartilage; leakage of versican cleavage products out of the arterial wall has never been reported but may be possible in an analogous manner.18
Kunjathoor et al indicate that unlike human intima, there is very little versican present in the mouse intima by immunohistochemistry.23 This would suggest that other substrates may also be targets for ADAMTS-1 cleavage. It also suggests that in humans, the role of ADAMTS-1 on VSMC migration and neointima growth may be even more pronounced because versican is prominently expressed in human intima. However, versican is a difficult matrix protein to analyze by immunohistochemistry because of its many large chondroitin sulfate glycosaminoglycan side chains. It is possible that versican is present in mouse intima and cleaved by ADAMTS-1 because we do see an effect on smooth muscle migration using the mouse carotid artery ligation model.
Preliminary results suggest that our ADAMTS-1 preparation also cleaves tissue factor pathway inhibitor (tissue factor pathway inhibitor-1) but does not cleave a general matrix metalloproteinase peptide substrate, collagen IV, elastin, and decorin (Figure III, available online at http://atvb.ahajournals.org; unpublished results, 2001). Inhibitors of cysteine, serine, and aspartic proteases did not interfere with cleavage, indicating that our protein preparation was not contaminated with proteases from other families and that cleavage of versican peptide substrates was ADAMTS-1–specific (unpublished data, 2000).
Based on our results, we hypothesize that ADAMTS-1 may be involved in atherogenesis by modulating VSMC migration. The expression of ADAMTS-1 is regulated by cytokines and nuclear hormone receptor agonists such as peroxisome proliferator-activated receptor (PPARγ) agonists thought to be important in atherogenesis (Figure IV, available online at http://atvb.ahajournals.org).3,19 ADAMTS-1 is also expressed by the relevant cells and localized to the relevant areas of lesions. Further experiments with animal models of atherosclerosis are required to better-understand the physiological and pathological functions of ADAMTS-1 in the vascular wall.
Appreciation is extended to Sofia Lundin, Zsofia Berke, and Martin Kjerrulf for their help and advice.
A.-C.J.-R. and T.N. contributed equally to this work.
- Received January 21, 2004.
- Accepted September 17, 2004.
Kuno K, Kanada N, Nakashima E, Fujiki F, Ichimura F, Matsushima K. Molecular cloning of a gene encoding a new type of metalloproteinase-disintegrin protein with thrombospondin motifs as an inflammation associated gene. J Biol Chem. 1997; 272: 556–562.
Vázquez F, Hastings G, Ortega MA, Lane TF, Oikemus S, Lombardo M, Iruela-Arispe ML. METH-1, a human ortholog of ADAMTS-1, and METH-2 are members of a new family of proteins with angio-inhibitory activity. J Biol Chem. 1999; 274: 23349–23357.
Bongrazio M, Baumann C, Zakrzewicz A, Pries AR, Gaehtgens P. Evidence for modulation of genes involved in vascular adaptation by prolonged exposure of endothelial cells to shear stress. Cardiovasc Res. 2000; 47: 384–393.
Evanko SP, Angello JC, Wight TN. Formation of hyaluronan- and versican-rich pericellular matrix is required for proliferation and migration of vascular smooth muscle cells. Arterioscler Thromb Vasc Biol. 1998; 19: 1004–1013.
Lemire JM, Braun KR, Maurel P, Kaplan ED, Schwartz SM, Wight TN. Versican/PG-M isoforms in vascular smooth muscle cells. Arterioscler Thromb Vasc Biol. 1999; 19: 1630–1639.
Sandy JD, Westling J, Denegy RD, Iruela-Arispe ML, Verscharen C, Rodriguez-Mazaneque JC, Zimmermann DR, Lemire JM, Fischer JW, Wight TN, Clowes AW. Versican V1 proteolysis in human aorta in vivo occurs at the Glu441-Ala442 bond, a site that is cleaved by recombinant ADAMTS-1 and ADAMTS-4. J Biol Chem. 2001; 276: 13372–13378.
Russell DL, Doyle KMH, Ochsner SA, Sandy JD, Richards JS. Processing and localization of ADAMTS-1 and proteolytic cleavage of versican during cumulus matrix expansion and ovulation. J Biol Chem. 2003; 278: 42330–42339.
Hakala JK, Öörni K, Pentikäinen MO, Hurt-Camejo E, Kovanen PT. Lipolysis of LDL by human secretory phospholipase A2 induces particle fusion and enhances the retention of LDL to human aortic proteoglycans. Arterioscler Thromb VascBiol. 2001; 21: 1053–1058.
Hurt-Camejo E, Olsson U, Wiklund O, Bondgers G, Camejo G. Cellular consequences of the association of apoB lipoproteins with proteoglycans. Arterioscler Thromb Vasc Biol. 1997; 17: 1011–1017.
Olin AI, Mörgelin M, Sasaki T, Timpl R, Heinegård D, Asperg A. The proteoglycan aggrecan and versican form networks with fibulin-2 through their lectin domain binding. J Biol Chem. 2001; 276: 1253–1261.
Kawashima H, Hirose M, Hirose J, Nagkubo D, Plaas AHK, Miyasaka M. Binding of a large chondroitin sulfate/dermata sulfate proteoglycan, versican, to L-selectin, P-selectin, and CD44. J Biol Chem. 2000; 275: 35448–35456.
Bodin D, Ivan E, Johnson C, Magid R, Galis ZS. Remodeling of carotid artery is associated with increased expression of matrix metalloproteinases in mouse blood flow cessation model. Circulation. 2000; 102: 2861–2866.
Sartipy P, Johansen B, Camejo G, Rosengren B, Bondjers G, Hurt-Camejo E. Binding of human phospholipase A2 type II to proteoglycans. J Biol Chem. 1996; 271: 26307–26314.
Ivan E, Khatri JJ, Johnson C, Magid R, Godin D, Nandi S, Lessner S, Galis ZS. Expansive arterial remodeling is associated with increased neointimal macrophage foam cell content. Circulation. 2002; 105: 2686–2691.
Tortorella MD, Pratta M, Liu RQ, Austin J, Ross OH, Abbaszade I, Burn T, Arner E. Sites of aggrecan cleavage by recombinant human aggrecanase-1 (ADAMTS-4). J Biol Chem. 2000; 275: 18566–18573.
Kuno K, Terashima Y, Matsushima K. ADAMTS-1 is an active metalloprotease associated with the extracellular matrix. J Biol Chem. 1999; 274: 18821–18826.
Galis ZS, Johnson C, Godin D, Magid R, Shipley JM, Senior RM, Ivan E. Targeted disruption of the matrix metalloproteinase-9 gene impairs smooth muscle cell migration and geometrical arterial remodeling. Circ Res. 2002; 91: 852–859.
Kunjathoor VV, Chiu DS, O’Brien KD, LeBoeuf RC. Accumulation of biglycan and perlecan, but not versican, in lesions of murine models of atherosclerosis. Arterioscler Thromb Vasc Biol. 2002; 22: 462–468.