Inflammation and atypical cell differentiation are hallmarks of atherosclerotic lesion development and are regulated by systemic and local factors. Bone morphogenetic proteins (BMP)-2 and -4, multifunctional growth factors and morphogens, have been implicated as mediators of endothelial inflammation in response to proatherogenic oscillatory shear stress, oxidative stress and proinflammatory cytokines.^1,2 Several BMPs have also been detected in calcified atherosclerotic plaques,^3,4 and BMP-2 promotes calcification of aortic myofibroblasts,⁵ consistent with the ability of BMPs to induce bone differentiation.⁶ Thus, the morphogenetic activities of BMP-2 and -4 might directly influence lesion progression through endothelial inflammation and cell differentiation.

Gene deletion in mice previously identified matrix Gla protein (MGP) as an essential inhibitor of arterial calcification,⁷ which prevents osteochondrogenic lineage reprogramming of smooth muscle cells (SMCs).⁸ Excessive MGP expression, on the other hand, has no discernable effect on the aortic vascular wall, but leads to growth inhibition of the pulmonary vascular tree through interference with BMP-4 signaling.⁹ MGP antagonizes BMP signaling^10,11 through direct protein-protein interaction involving essential proline and γ-carboxylated glutamate residues in MGP.¹² It acts as a negative feedback regulator in a BMP-activated pathway that is triggered by interaction between BMP-2 or -4 and the activin-like kinase receptor (ALK)2, first stimulating expression of the activin-like kinase receptor (ALK)1, and then expression of the vascular endothelial growth factor (VEGF) and MGP itself.^9,11,13 This regulatory pathway is likely to play an important role in embryonic development because ALK1 and VEGF are essential for angiogenesis,^13,14 BMP and MGP affects pattern formation in vascular mesenchymal cells,¹⁵ and MGP promotes SMC differentiation.⁸ Here, we hypothesize that perturbations in BMP-signaling play important roles in the development of vascular disease.

Methods

Animals and Diets

Apoe^−/− mice on a C57BL/6J background were purchased from The Jackson Laboratory (Bar Harbor, Me). MGP^tg/wt mice, generated in our laboratory on a C57BL/6J background,⁹ were crossed with Apoe^−/− mice to generate MGP^tg/wt;Apoe^−/− mice. We used the MGP^tg/wt mice because the phenotype was apparent in MGP^tg/wt mice, and a low birth rate of MGP^tg/tg mice made it difficult to obtain MGP^tg/tg mice.⁹ MGP^+/− mice on C57BL/6J background⁷ were obtained from Dr Gerard Karsenty (Columbia University, New York) and crossed with the Apoe^−/− mice to generate MGP^−/−;Apoe^−/− mice. All mice were maintained on standard chow (Diet 8604, Harlan Teklad, Laboratory, Madison, Wis). At 8 to 10 weeks of age, littermate Apoe^−/− and MGP^tg/wt;Apoe^−/− mice were either continued on standard chow or switched to a high-fat/high-cholesterol diet (Western diet) (Research Diets, New Brunswick, NJ, diet #D12108, containing 21% fat [wt/wt], 1.25% cholesterol [wt/wt]) for 16 weeks. The studies were reviewed by the Institutional Review Board and conducted in accordance with the animal care guidelines set by the University of California, Los Angeles.

See the expanded Methods section, available in the Online Data Supplement at http://circres.ahajournals.org, for details regarding animals, lipid and phosphate analyses, lesion and calcification quantification, validation of antibodies, immunoblotting, RNA analysis, immunohistochemistry and immunofluorescence, ELISA, complete blood count, and statistical analysis.

Results

Reduced Aortic Atherosclerotic Lesion Formation in MGP^tg/wt;Apoe^−/− Mice

To explore the ability of BMP signaling to influence vascular disease, we took advantage of the availability of transgenic mice expressing human MGP (MGP^tg/wt) and mice with MGP gene deletion (MGP^−/−).^7,9 These were crossbred with Apolipoprotein E null (Apoe^−/−) mice, a common model of atherosclerosis and lesion calcification.

We first determined the effect of increased MGP expression on atherosclerotic lesions and arterial calcification by comparing Apoe^−/− and MGP^tg/wt;Apoe^−/− mice after 16 weeks on Western diet. The hypercholesterolemia of the Apoe^−/− and MGP^tg/wt;Apoe^−/− mice was exacerbated by the Western diet, but there were no statistically significant differences in body mass or lipid levels in male or female mice, respectively (Online Table I). Cross-sectional analysis of the atherosclerotic lesions showed statistically significant differences between the two strains. The Apoe^−/− mice had an average aortic sinus lesion area of 540,117 μm² and 589,159 μm² in males (n=15) and females (n=11), respectively, whereas MGP^tg/wt;Apoe^−/− mice had an average area of 338 312 μm² and 216 890 μm² in males (n=12, P=0.0023) and females (n=7, P=0.0013), respectively (Figure 1A and 1B). The lower number of females was attributable to difficulties in obtaining sufficient females carrying the MGP transgene.⁹ En face analysis of whole aortas showed similar differences between the two strains. The Apoe^−/− mice had an average aortic lesion area of 8.5% and 9.1% in males (n=5) and females (n=4), respectively, whereas MGP^tg/wt;Apoe^−/− mice had an area of 4.2% and 4.1% in males (n=6, P=0.038) and females (n=4, P=0.018), respectively (Figure 1C and 1D). Thus, enhanced MGP expression limited formation of atherosclerotic lesions.

• Download figure
• Download PowerPoint

Reduced Aortic Atherosclerotic Lesion Calcification in MGP^tg/wt;Apoe^−/− Mice

Because of the well-defined role of MGP as an inhibitor of vascular calcification, we examined the effect of enhanced MGP expression on calcification. We first compared total calcium in aortic tissue in Apoe^−/− and MGP^tg/wt;Apoe^−/− mice after 16 weeks of regular chow diet (CH) or Western diet (high fat diet, HF). The results showed no significant effect of the Western diet in the respective strain (Figure 2A). However, total calcium was significantly decreased in both chow and fat-fed MGP^tg/wt;Apoe^−/− mice compared to the similarly fed Apoe^−/− mice (Figure 2A). We next compared lesion calcification by cross-sectional analysis (schematically illustrated in Figure 2B) in the Apoe^−/− and MGP^tg/wt;Apoe^−/− mice that were fed a Western diet. This showed that both the rate and amount of lesion calcification per animal were significantly lower in MGP^tg/wt;Apoe^−/− mice than in Apoe^−/− mice. The rate of calcification was 61.5% and 31.6% in the Apoe^−/− and the MGP^tg/wt;Apoe^−/− mice, respectively (Figure 2C), and the average amount of calcification was 16.7 μm³ (n=13) in Apoe^−/− mice but only 1.2 μm³ in MGP^tg/wt;Apoe^−/− (n=6) (Figure 2D and 2E). Only animals that exhibited calcification were analyzed, and males and females were analyzed together because of the low calcification rate in the MGP^tg/wt;Apoe^−/− mice. Thus, enhanced MGP expression limited total aortic calcium and lesion calcification. Similar to previous reports,¹⁶ we observed that the Apoe^−/− mice developed cartilage metaplasia and medial calcification in parts of the media without lesions. We found that 61.5% of the Apoe^−/− mice, but none of the MGP^tg/wt;Apoe^−/− mice, had medial calcification (Figure 2F), suggesting that MGP also limits medial calcification. Because of the important role of phosphate in stimulating vascular calcification, we compared serum phosphate levels in the different mice, but there were no statistically significant differences (Online Table II).

• Download figure
• Download PowerPoint

Diminished Vascular BMP Activity in MGP^tg/wt;Apoe^−/− Mice

To examine BMP signaling in atherosclerotic vessels and its response to enhanced MGP expression, we first compared expression of BMP-2, -4, and MGP in Apoe^−/− and MGP^tg/wt;Apoe^−/− mice after 16 weeks of chow or Western diet. Aortic expression of BMP-4 increased ≈3-fold in both groups of mice on the Western diet compared to chow, whereas expression of BMP-2 remained the same (Figure 3A). Glyceraldehyde-3-phosphate dehydrogenase was used as reference gene for the PCR reactions. BMP-4 protein in serum and aorta also increased, as determined by ELISA and immunoblotting with quantification by densitometry (Figure 3B and 3C). β-Actin was used as loading control. Aortic expression of MGP was ≈3-fold higher in chow-fed MGP^tg/wt;Apoe^−/− mice than in Apoe^−/− mice. It increased ≈3-fold in the Apoe^−/− mice and 15-fold in the MGP^tg/wt;Apoe^−/− mice on the Western diet (Figure 3A), suggesting a synergistic effect of the transgene and the Western diet.

• Download figure
• Download PowerPoint

We next examined MGP in the vascular wall by immunochemistry. In aortic segments without obvious atherosclerotic lesions, MGP was detected mostly in proximity to the endothelium and the adventitia in chow-fed Apoe^−/− mice (Figure 3D). Because Western diet and the transgene increased MGP expression, the staining in these areas intensified and MGP staining also appeared in the media (Figure 3D). Control experiments using MGP^−/− aorta and the same MGP-antibodies showed no staining (Figure 3E). In atherosclerotic lesions, MGP and BMP-4 were detected in the same areas (Figure 3F), but the MGP expression was higher relative to BMP-4 in the MGP^tg/wt;Apoe^−/− mice than in the Apoe^−/− mice (Figure 3F).

We predicted that BMP signaling activity would be less in the aortic walls of the MGP^tg/wt;Apoe^−/− mice because of the enhanced MGP expression. To test this, we determined the levels of activated, phosphorylated (p)SMAD1/5/8 in aortas from Apoe^−/− and MGP^tg/wt;Apoe^−/− mice fed chow or a Western diet. The level of pSMAD1/5/8 was lower in the MGP^tg/wt;Apoe^−/− aortas than in the Apoe^−/− aortas as determined by immunoblotting with densitometry (Figure 3G) and immunohistochemistry (Online Figure II). Total SMAD and pSMAD2/3, which mediate TGF-β signaling, were included for comparison and showed no change. Furthermore, pSMAD1/5/8 levels were diminished in atherosclerotic lesions from MGP^tg/wt;Apoe^−/− mice compared to Apoe^−/− mice as determined by immunofluorescence (Figure 3H). Thus, the enhanced MGP expression in MGP^tg/wt;Apoe^−/− mice diminished BMP signaling.

Increased MGP Expression Suppresses Expression of ALK1 and VEGF

Expression of ALK1 is stimulated by BMP,¹¹ an effect that is mediated by the ALK2 receptor.¹⁷ ALK1 is essential for normal angiogenesis, and has a dual role in VEGF regulation. ALK1 suppresses VEGF expression when stimulated by BMP-9, a circulating BMP that has been referred to as a vascular “quiescence” factor.¹⁸ Yet, ALK1 also enhances VEGF levels by inducing the ALK5 receptor, which reacts with TGF-β1 to increase VEGF expression.¹³ To examine this pathway, we determined aortic expression of ALK2, ALK1 and VEGF. The aortic expression of ALK1 and VEGF was increased in Apoe^−/− mice on Western diet compared to chow, and suppressed in MGP^tg/wt;Apoe^−/− mice on both chow and Western diet as determined by real-time PCR (Figure 4A), immunoblotting with densitometry (Figure 4B), and immunohistochemistry of aortas without atherosclerotic lesions (Figure 4C). There was no significant change in ALK2 expression. In atherosclerotic lesions, expression of both ALK1 and VEGF was strongly enhanced in Apoe^−/− mice on Western diet compared to MGP^tg/wt;Apoe^−/− mice (Figure 4D). ALK1 and VEGF colocalized to a large extent with the von Willebrand (vWF), a marker for vascular endothelium, in the MGP^tg/wt;Apoe^−/− mice (Figure 4D). Thus, diminished BMP signaling led to suppression of aortic ALK1 and VEGF, suggesting that the same signaling pathway is active in vascular disease and in angiogenesis. To determine whether other BMP inhibitors contributed to the decreased BMP activity, we also examined the expression of crossveinless 2 (CV2), Noggin, and Chordin.¹⁹ The results showed a stimulation of Noggin and Chordin in both types of mice on the Western diet, whereas the expression of CV2 correlated with BMP activity and was inhibited in MGP^tg/wt;Apoe^−/− mice (Online Figure III). Consequently, none of these inhibitors selectively contributed to the decreased BMP activity in the MGP^tg/wt;Apoe^−/− mice.

• Download figure
• Download PowerPoint

Increased MGP Expression Suppresses Gene Expression Associated With Inflammation

Inflammation is a hallmark of atherosclerotic lesions and is partly mediated by BMP-2 and -4.^1,2 To examine the effect of MGP overexpression on inflammatory activity, we compared aortic expression of CD68, a macrophage marker,²⁰ interleukin (IL)-6,²¹ heat shock protein (HSP)70, a stress-induced protein,²² and the endothelial adhesion molecules intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1, and E-selectin in Apoe^−/− and MGP^tg/wt;Apoe^−/− mice on regular chow or Western diet. The results showed that aortic expression of all proteins increased with Western diet in the Apoe^−/− mice, but not in the MGP^tg/wt;Apoe^−/− mice as determined by real-time PCR and immunoblotting with densitometry (Figure 5A through 5C). The real-time PCR results in the mice on Western diet were also normalized to lesion area, which did not significantly alter the results (Online Figure IV). The serum levels of IL-6 and HSP70 reflected the changes in the aortic expression (Figure 5D). Furthermore, expression of all proteins decreased in lesions of MGP^tg/wt;Apoe^−/− mice compared to Apoe^−/− mice as determined by immunofluorescence (Figure 5E). Thus, the results suggested that inflammation is reduced by MGP overexpression.

• Download figure
• Download PowerPoint

Reduced Formation of Atherosclerotic Lesions in MGP^−/−;Apoe^−/− Mice

MGP^−/− mice are known to develop arterial calcification,^7,8 and our results showed increased aortic pSMAD1/5/8 levels (Online Figure V), suggesting increased BMP activity. As previously reported, the calcification also caused structural changes in the aortic wall (see H&E staining; Online Figure VA, top).^7,8

To determine whether lesion formation occurred in MGP deficiency, we compared MGP^−/−;Apoe^−/− mice to Apoe^−/− and MGP^+/−;Apoe^−/− mice. The MGP^−/−;Apoe^−/− mice were unable to tolerate a Western diet because of its severe phenotype, which was indistinguishable from that of MGP^−/− mice.⁷ All mice were therefore maintained on a chow diet for 18 weeks. As expected, the MGP^−/−;Apoe^−/− mice were smaller than the Apoe^−/− mice,⁷ but there were no significant differences in lipid levels (Online Table I). Cross-sectional analysis showed that the average aortic sinus lesion area was 99,981 μm² in chow-fed Apoe^−/− mice (n=13), whereas lesion formation was minimal, 3,857 μm² in MGP^−/−;Apoe^−/− mice (n=5, P=0.0231) (Figure 6A and 6B). Males and the females were combined in the analysis because of the limited number of MGP^−/−;Apoe^−/− mice that survived 18 weeks. There were no significant differences between the Apoe^−/− and the MGP^+/−;Apoe^−/− mice (Figure 6A). The MGP^−/−;Apoe^−/− mice demonstrated the same altered aortic wall structure as the MGP^−/− mice, which prevented en face analysis.

• Download figure
• Download PowerPoint

Furthermore, total calcium in aortic tissue was highly increased in the MGP^−/−;Apoe^−/− mice (Figure 6C), and cross-sectional analysis of the aortas demonstrated the extensively calcified media in the MGP^−/−;Apoe^−/− mice (Figure 6D and 6E) compared to the Apoe^−/− mice that had minimal calcification in both types of analysis, also when compared to Apoe^−/− mice fed a Western diet (Figure 1). The rate of calcification was 100% in the MGP^−/−;Apoe^−/− mice compared to 0% in the Apoe^−/− mice (Figure 6D). The average amount of calcification in the aorta was 73,136 μm³ (n=7, 30 μm aortic segments) in MGP^−/−;Apoe^−/− mice, and 0 μm³ in the Apoe^−/− mice (n=13) (Figure 6C), consistent with previous reports.^7,8 There were no significant differences between the Apoe^−/− and the Apoe^−/−;MGP^+/− mice. In addition calcification without lesion formation was observed in the aortic valves of the MGP^−/−;Apoe^−/− mice, suggesting that the valvular BMP response was similar to that of the aortic wall. There were no significant differences in serum phosphate levels (Online Table II).

We then compared SMAD1/5/8 signaling and expression of ALK2, ALK1, VEGF, BMP-2 and -4 in the MGP^−/−;Apoe^−/− and the Apoe^−/− mice. Again, the MGP-deficiency resulted in increased levels of pSMAD1/5/8, as determined by immunoblotting with densitometry and immunofluorescence (Figure 7A and 7B). Expression of ALK2, ALK1, and VEGF, but not BMP-2 and -4, was also enhanced in the MGP^−/−;Apoe^−/− mice as determined by real-time PCR, immunoblotting with densitometry and immunofluorescence (Figure 7C through 7E). Furthermore, expression of CV2, but not Noggin and Chordin, was significantly increased (Online Figure VI), supporting that CV2 is responsive to changes in BMP activity.

• Download figure
• Download PowerPoint

Diminished Inflammatory Response in MGP^−/−;Apoe^−/− Mice

To examine whether changes in the inflammatory response are involved in reducing lesion formation, we examined the expression of CD68, IL-6, HSP70 and endothelial adhesion molecules in the MGP^−/−;Apoe^−/− and the Apoe^−/− mice. Expression of all proteins was dramatically reduced in the aortas of the MGP^−/−;Apoe^−/− mice compared to the Apoe^−/− mice, as determined by real-time PCR, immunoblotting with densitometry and immunofluorescence (Figure 8A through 8D). No differences were detected in levels of circulating monocytes and other white blood cells as determined by standard methodology (Online Table III). The results suggest that reduced expression of endothelial adhesion molecules explains, at least in part, the decrease in lesion formation.

• Download figure
• Download PowerPoint

Discussion

BMP signaling in vertebrate development is essential for regulation of gastrulation, patterning, and organogenesis,²³ whereas postnatal BMP signaling regulates injury repair, remodeling, and inflammation.^1,2,24,25 In this study, we show that BMP signaling is a connecting link between hyperlipidemia, an inflammatory stimulus, and the atherogenic response in the arterial wall. We also present evidence that a BMP-activated regulatory pathway, which was identified in vitro and includes ALK2, ALK1, MGP, and VEGF,^11,17 is active in vivo during the development of vascular disease.

The presence of the MGP transgene led to more efficient reduction in BMP activity and lesion formation in response to a high fat diet in Apoe^−/− mice. It also led to less expression of ALK1 and VEGF, the precise role of which is likely dependent on location and timing. Suppression of ALK1 and VEGF in lesions may limit neoangiogenesis and reduce intraplaque hemorrhages and calcification,²⁶ although a baseline expression of VEGF is required for maintenance and repair of a healthy endothelium.²⁷ It is also known that VEGF is required for normal ossification during endochondral bone formation,²⁸ and high VEGF levels might actually promote calcification. The beneficial effect of BMP inhibition, however, has to be weighed against the effect of limiting BMP in vascular growth considering that excess MGP inhibits angiogenesis and pulmonary vascular growth.^9,22 It illustrates the delicate balance in BMP signaling that directs growth and homeostasis of different vascular beds.

We detected enhanced MGP expression close to the endothelium, in the media and in atherosclerotic lesions. It is consistent with previous reports of increased MGP expression in calcified media and atherosclerotic lesions.^3,29–31 It suggests that MGP has a dual role in limiting BMP-induced events in both the endothelium and the media, including endothelial inflammation^1,2 and atypical differentiation of medial cells.⁸ We also observed changes in expression of CV2, Noggin, and Chordin, especially in the endothelium. Interestingly, CV2 expression was enhanced by increased BMP activity, whereas Noggin and Chordin expression was induced by high fat diet, suggesting distinct gene regulation.

The findings in this and other studies^7,8,32 suggest that the response in developing arteries to loss of MGP and highly enhanced BMP activity is calcification of medial cells. Speer et al⁸ recently showed that the calcification of the MGP^−/− aorta involves an osteochondrogenic lineage reprogramming of SMCs, suggesting that such transdifferentiation of SMCs is driving calcification in the vascular wall. This effect appears to be dependent on locally expressed MGP given that only MGP expression targeted to SMCs, but not circulating MGP, is able to rescue the MGP^−/− mouse.³² Our results further suggest that it occurs in absence of inflammatory activation. A similar calcific response may be triggered by increased BMP signaling in the atherosclerotic lesions, supporting the general concept that developmental programs can be reactivated in vascular disease.

Only BMP-4 was found to increase in response to the high fat diet even though both BMP-2 and -4 have been implicated in proinflammatory effects in the endothelium and vascular calcification.^1,2,5,33 Although we cannot exclude local variations of BMP-2 not detected in the analysis, it suggests that BMP-4 more readily initiates atherosclerotic development in mice. The expression of endothelial adhesion molecules and proteins associated with inflammation and cellular stress correlated well with increased BMP activity and lesion formation in the Apoe^−/− and MGP^tg/wt;Apoe^−/− aortas. Despite a high BMP activity, expression of adhesion molecules and other inflammatory proteins was dramatically reduced in the MGP-deficient mice. The low expression of adhesion molecules likely prevented monocyte infiltration of the artery wall, thereby explaining, at least in part, the reduction in lesions. The reason for the lack of induction of endothelial adhesion molecules is still unclear. We speculate that the uncontrolled BMP activity affects endothelial cell differentiation, which in turn causes abnormal endothelial cell responses. However, further studies are required to clarify this. Other factors that may affect the inflammatory response include potential defects in monocyte function and the severe phenotype of the MGP^−/− mice.

Interestingly, HDL, a well-known antiatherogenic factor, is able to induced expression of ALK2, which in turn allows induction of ALK1, VEGF, and MGP.¹⁷ Indeed, the Apo-AI transgenic mice, which have high HDL and minimal atherosclerosis, showed enhanced aortic expression of all these factors,¹⁷ which suggests that HDL protects against inflammation and atherosclerosis in part by influencing BMP signaling. However, it is not clear whether the protective effect derives from ALK2 activating several protective pathways or from early induction of ALK1, MGP, and VEGF before lesion formation.

In this study, we have focused on the role MGP as an inhibitor of BMP signaling. We have not evaluated other aspects of its role as a matrix protein, including potential changes in extracellular location and binding of calcium and other proteins in the extracellular matrix. For example, MGP may attach to vitronectin through its C terminus,³⁴ which could alter its ability to bind BMP. The function of MGP is affected by factors that interfere with γ-carboxylation. For example, warfarin, a vitamin K antagonist, increases the levels of uncarboxylated, nonfunctioning MGP and may enhance calcification.^12,35,36 It is also known that the reduction of vitamin K is less efficient in vascular SMCs than in the liver.³⁵ Furthermore, MGP contains phosphorylated serines, which contribute to its function as a calcification inhibitor.³⁷ However, it is not clear whether there is a relationship between the phosphorylation status of MGP and its effect on BMP signaling.

In summary, our results support the concept that modulation of BMP signaling through alterations in MGP expression regulates vascular disease, as well as the development of a normal arterial wall. A regulatory pathway triggered by BMP and involving ALK2, ALK1, VEGF, and MGP is central for vascular development, disease, and homeostasis. Understanding the role of vascular BMP may reveal new approaches to treat and prevent vascular disease.

Sources of Funding

This work was funded, in part, by NIH grants HL30568 and HL81397 and the American Heart Association (Western Affiliate).

Disclosures

None.

Footnotes