β-arrestins Regulate Atherosclerosis and Neointimal Hyperplasia by Controlling Smooth Muscle Cell Proliferation and Migration

Abstract

Atherosclerosis and arterial injury-provoked neointimal hyperplasia involve medial smooth muscle cell (SMC) proliferation and migration into the arterial intima. Because many 7-transmembrane and growth factor receptors promote atherosclerosis, we hypothesized that the multifunctional adaptor proteins β-arrestin1 and -2 might regulate this pathologic process. Deficiency of β-arrestin2 in ldlr^-/- mice reduced aortic atherosclerosis by 40%, and decreased the prevalence of atheroma SMCs by 35%—suggesting that β-arrestin2 promotes atherosclerosis through effects on SMCs. To test this potential atherogenic mechanism more specifically, we performed carotid endothelial denudation in congenic WT, β-arrestin1^-/-, and β-arrestin2^-/- mice. Neointimal hyperplasia was enhanced in β-arrestin1^-/- mice, and diminished in β-arrestin2^-/- mice. Neointimal cells expressed SMC markers and did not derive from bone marrow progenitors, as demonstrated by bone marrow transplantation with GFP-transgenic cells. Moreover, the reduction in neointimal hyperplasia seen in β-arrestin2^-/- mice was not altered by transplantation with either WT or β-arrestin2^-/- bone marrow cells. After carotid injury, medial SMC ERK activation and proliferation were increased in β-arrestin1^-/- and decreased in β-arrestin2^-/- mice. Concordantly, thymidine incorporation, ERK activation and migration evoked by 7-transmembrane receptors were greater than WT in β-arrestin1^-/- SMCs, and less in β-arrestin2^-/- SMCs. Proliferation was less than WT in β-arrestin2^-/- SMCs, but not in β-arrestin2^-/- endothelial cells. We conclude that β-arrestin2 aggravates atherosclerosis through mechanisms involving SMC proliferation and migration, and that these SMC activities are regulated reciprocally by β-arrestin2 and β-arrestin1. These findings identify inhibition of β-arrestin2 as a novel therapeutic strategy for combating atherosclerosis and arterial restenosis after angioplasty.

Introduction

In the context of atherosclerosis,¹ or in response to arterial endothelial denudation,² the normally quiescent smooth muscle cells (SMCs) of the artery's tunica media proliferate, and migrate into the subendothelial intima. This SMC proliferation and migration into the intima constitute a critical component of atherosclerosis: SMC-specific genetic changes that enhance these processes also enhance atherosclerosis considerably in low-density lipoprotein receptor (LDLR) –deficient mice.³ Collectively referred to as neointimal hyperplasia, SMC proliferation and migration into the intima continues to complicate angioplasty and stenting used to treat human atherosclerosis.⁴ The myriad stimuli that contribute to neointimal hyperplasia include agonists secreted from activated platelets, macrophages, and SMCs themselves; consequently, candidate molecular strategies to limit neointimal hyperplasia must target multiple receptor signaling systems, and thereby affect SMC proliferation and migration, among other processes.

One such molecular strategy could involve proteins known as β-arrestin1 and β-arrestin2, ubiquitously expressed multifunctional scaffolding proteins that were originally discovered because of their ability to quench 7-transmembrane receptor signaling through heterotrimeric G proteins.⁵ In addition to this role in attenuating G protein signaling, however, β-arrestins also play multiple roles in diminishing as well as activating signaling downstream of not only 7-transmembrane receptors but also receptor protein tyrosine kinases, cytokine receptors, and ion channel receptors.⁵ Signal transduction pathways triggered or positively modulated by β-arrestins include the activation of Src family tyrosine kinases, phosphatidylinositol-3-kinase/Akt, IRS-1, RhoA, and the mitogen-activated protein kinase families comprising ERK, JNK, and p38 isoforms.⁵ Findings obtained in physiologic systems like striatal neurons and lymphocytes have corroborated anti-apoptotic and chemotactic roles discovered for β-arrestin2 in immortalized cell lines.⁵

β-arrestin1 and β-arrestin2 share 78% amino acid identity.⁵ It is therefore not surprising that many proteins interact with both β-arrestins (albeit with differing affinities), while several proteins interact preferentially, or even exclusively, with a single β-arrestin isoform.^5,6 At the level of signal transduction, this β-arrestin binding specificity manifests in two distinctive patterns. First, both β-arrestin isoforms may be required for full activation of signaling, as exemplified by G protein-independent activation of ERK1/2 signaling evoked by the β₂-adrenergic receptor.⁷ Second, signaling may be promoted by one β-arrestin isoform, and inhibited by the other, as exemplified by the angiotensin II AT₁ receptor⁸ and the protease-activated receptor-1.⁵

Since β-arrestins affect the survival, anti-apoptotic signaling, and migration of immune as well as other cells in response to diverse stimuli,⁵ we asked whether β-arrestins affect the neointimal hyperplasia associated with atherosclerosis or provoked by endothelial denudation, and whether β-arrestins affect the in vitro correlates of neointimal hyperplasia: migration, pro-mitogenic signaling, and proliferation in primary SMCs. The relevance of these questions to human disease is supported by the finding that β-arrestin2 mRNA is up-regulated in atherosclerotic, as compared with normal human coronary arteries.⁹ To address these questions, we used congenic low-density lipoprotein receptor (LDLR) -deficient, β-arrestin-deficient and WT mice.

Methods

C57BL/6 ldlr^-/- mice were crossed with C57BL/6-congenic β-arrestin2^-/- mice;⁶ progeny were interbred to obtain littermates used in atherosclerosis studies. For carotid endothelial denudation, we used congenic age- and gender-matched C57BL/6 WT, β-arrestin1^-/-, β-arrestin2^-/-, and GFP-transgenic mice. Bone marrow transplantation, SMC migration,¹⁰ RNA interference,¹¹ and cell proliferation assays¹¹ are described in the detailed Methods section, in the online supplement to this manuscript.

Results

β-arrestin2 promotes atherogenesis

Because β-arrestins affect proliferative signaling and cellular migration, and because β-arrestin2 mRNA levels are higher in atherosclerotic than in normal human arteries,⁹ we hypothesized that β-arrestin2 activity contributes to the neointimal hyperplasia of atherosclerosis. To test this hypothesis, we examined aortic atherosclerosis in congenic female ldlr^-/- mice that were either (+/+) or (-/-) at the β-arrestin2 locus. The β-arrestin2^+/+ and β-arrestin2^-/ mice had equivalent systolic blood pressures (120 ± 10 and 115 ± 6 mm Hg, respectively) and equivalent serum cholesterol values on a Western diet (28 ± 3 and 27 ± 2 mmol/L, respectively). After 12 weeks on a Western diet, however, ldlr^-/-/β-arrestin2^-/- mice developed aortic atherosclerosis that was only 59% of that observed in ldlr^-/-/β-arrestin2^+/+ mice, as judged by en face analysis (p < 0.02; Figures 1A, C).

Figure 1.

β-arrestin2 activity exacerbates atherosclerosis, and increases the prevalence of SMCs in atherosclerotic lesions. Eight-wk-old female ldlr^-/- mice of the indicated β-arrestin2 genotype were fed a Western diet for 12 weeks before sacrifice. A, Sudan IV-perfused aortas were excised from the aortic root (top) to the iliac bifurcation (bottom), and pinned flat. B, Aortic root cross sections were stained for α-SMC actin and DNA. The lumen is oriented upward, and the tunica media is indicated. Shown are 2 specimens stained in tandem, representative of 4 such pairs; scale bar=200 μm. C, Sudan IV-positive lesion area is plotted as a percent of total aortic area (mean±S.E.), from ≥8 aortas of each genotype. SMC actin immunofluorescence was normalized to DNA fluorescence within each microscopic field, to obtain values for “SMC prevalence,” plotted as mean ± S.E. of 4 specimens of each genotype. D, Aortic root cross sections from B were stained for DNA and immunostained in tandem for both SMC actin and β-arrestin1, using an antiserum that that detects β-arrestin1 ∼5 times more sensitively than β-arrestin2 (data not shown). Individual sections were imaged with three narrow band-pass filters; red and green images were each merged with blue DNA fluorescence images. Images shown are representative of 6 β-arrestin2^+/+ and (-/-) pairs. Specific β-arrestin1 immunofluorescence was visible in β-arrestin2^-/- samples, but only at higher magnification (not shown). Scale bar=200 μm.

β-arrestin2 expression increased not only the extent of atherosclerosis, but also the prevalence of atheroma SMCs. Measured as a fraction of all atheroma cells, aortic root SMCs in ldlr^-/-/β-arrestin2^-/- mice were only 65 ± 10% as abundant as they were in ldlr^-/-/β-arrestin2^+/+ mice (p < 0.01; Figures 1B, C). Strikingly, SMCs in the fibrous cap of aortic root atheromata expressed considerable levels of β-arrestins relative to the macrophage-rich components of the neointima (Figure 1D). Moreover, neointimal β-arrestin staining was much more prominent in mice that were (+/+) at the β-arrestin2 locus, even though we stained specimens with an antibody that binds β-arrestin1 more avidly than β-arrestin2 (Figure 1D). These data accord well with human transcriptional profiling data from human atherosclerotic coronary arteries, in which β-arrestin2 mRNA levels are ∼2-fold higher than levels in non-atherosclerotic control coronaries.⁹ Thus, β-arrestin2 contributes to atherosclerosis through mechanisms that affect SMC recruitment to the neointima, and β-arrestin2 appears to be the predominant β-arrestin isoform in the atherosclerotic lesion.

To determine mechanisms by which β-arrestin2 may enhance the prevalence of SMCs in atherosclerotic lesions, we focused on pathologic arterial SMC proliferation and migration using a simpler, non-atherosclerotic model system: neointimal hyperplasia induced by endothelial denudation, in the absence of hyperlipidemia.^2,12

β-arrestins regulate neointimal hyperplasia through effects in arterial SMCs

We used 0.36-mm angioplasty guidewires to denude the endothelium of the mouse carotid artery.² Arteries harvested within 2 h of wire-mediated injury demonstrated scattered small luminal thrombi, no endothelial cells, and intact internal elastic laminae (data not shown). By 4 weeks after endothelial denudation, considerable neointimal hyperplasia developed (online Figure I). Neointimal lesions comprised SMCs, as judged by the high prevalence (>90%) of neointimal cell SMC myosin heavy chain expression (online Figure I). As in our atherosclerotic lesions, the level of β-arrestin expression in these neointimal SMCs was consistently higher than that of the medial SMCs in WT mice (online Figure I). We next studied neointimal hyperplasia in congenic WT and β-arrestin2^-/- mice. Because β-arrestin1 can serve to antagonize β-arrestin2 activity in some signaling systems,⁵ we also included congenic β-arrestin1^-/- mice in these studies.

Before endothelial denudation, arteries from our three congenic mouse strains appeared histologically identical (data not shown). However, arteries harvested 2-4 weeks after endothelial denudation demonstrated clear β-arrestin isoform-specific differences (Figure 2). Two weeks after endothelial denudation, the average external carotid diameter was equivalent among all vessels (data not shown), and the extent of carotid re-endothelialization was equivalent (∼35%) in all three genotypes (Figure II). However, compared with WT carotids, medial and neointimal area were 50% greater in β-arrestin1^-/- carotids, and 45-60% less in β-arrestin2^-/- carotids (p < 0.01, Figure 2). By four weeks after endothelial denudation, neointimal area was 37% larger than WT in β-arrestin 1^-/- carotids, and 47% smaller than WT in β-arrestin 2^-/- carotids. Moreover, compared with WT, the β-arrestin 2^-/- carotid media was 33% smaller (p < 0.01, Figure 2) and the lumen area 44% larger (p < 0.05), even though external carotid diameter remained equivalent across genotypes (data not shown). Thus, it appeared that while β-arrestin2 expression exacerbated neointimal hyperplasia, β-arrestin1 expression attenuated it.

Figure 2.

Neointimal hyperplasia induced by endothelial injury is exacerbated by β-arrestin2 and mitigated by β-arrestin1. Congenic mice of the indicated genotype were sacrificed 2 or 4 weeks after carotid artery de-endothelialization. A, Perfusion-fixed carotids were stained with a modified connective tissue stain (“Masson”), or with SMC α-actin immunofluorescence and DNA counterstain. SMC actin-expressing cells constituted ≥95% of medial and neointimal cells. Samples shown represent ≥5 of each genotype (scale bar=50 μm). B, Neointimal and medial areas are presented as mean ± S.E. (n≥5 of each genotype) for specimens harvested 2 (top) or 4 (bottom) weeks post-injury. βarr, β-arrestin. Compared with WT: *, p<0.01 (βarr1^-/-, “KO”) or <0.001 (βarr2^-/-). Compared with βarr1^-/-: #, p<0.001.

This reciprocal regulation of neointimal hyperplasia by β-arrestin1 and β-arrestin2 could result from effects in two relatively distinct cellular compartments: (i) arterial SMCs, which proliferate and migrate into the neointima;¹² (ii) bone marrow-derived platelets, neutrophils, and monocyte/macrophages, which adhere to the basement membrane of endothelium-denuded arteries, and release chemo/cytokines that provoke arterial SMC proliferation and migration.¹³ Certain models of vascular injury have implicated bone marrow-derived vascular precursor cells as a source of SMC-like cells in non-atherosclerotic neointimal hyperplasia,¹³ and β-arrestin2 is known to affect migration of hematopoietic-lineage cells.⁵ Could β-arrestins affect neointimal hyperplasia by regulating possible recruitment and differentiation of bone marrow-derived cells into SMC-like cells of the arterial neointima? Alternatively (or in addition), could β-arrestins alter neointimal hyperplasia through mechanisms that affect platelet and leukocyte function?

To address these questions, we used WT and congenic transgenic mice that express GFP ubiquitously (WT^GFP) as donors and recipients for bone marrow transplantation, and then performed carotid endothelial denudation 4 weeks later. In these experiments, GFP⁺ cells appeared in the neointima only when the artery wall (and thus the bone marrow recipient mouse) was WT^GFP, and not when the bone marrow compartment alone was WT^GFP (Figure 3A). Thus, neointimal cells in our model of endothelial denudation did not derive from bone marrow precursors. Further support for this inference emerges from bone marrow transplantation with either WT or β-arrestin2^-/- cells, also depicted in Figure 3. Neointimal size correlated with the genotype of the bone marrow recipient only, and not with the genotype of the bone marrow cells themselves. Consequently, neointimal hyperplasia was reduced by ∼75% in β-arrestin2^-/- mice as compared with WT mice, whether these mice were transplanted with WT or β-arrestin2^-/- bone marrow cells (Figure 3B). Since β-arrestin2 expression in bone marrow-derived cells did not alter neointimal hyperplasia, we can also infer that β-arrestin2 augments neointimal hyperplasia through mechanisms fundamentally involving SMCs, and not bone marrow-derived cells.

Figure 3.

Neointimal hyperplasia triggered by endothelial denudation does not involve bone marrow-derived progenitor cells. WT, WT^GFP and β-arrestin2^-/- (βarr2^-/-, “KO”) mice underwent bone marrow transplantation (BMT) at 8 weeks of age with BM from WT, WT^GFP, or βarr2^-/- mice. Carotid injury was performed 4 weeks later. Carotid arteries were harvested 4 weeks post-injury, and processed for GFP immunofluorescence (original magnification ×1100) (A) and morphometry (B and C). A, Photomicrographs from the indicated BMT donors/recipients represent 5 independent specimens of each type. B, Data shown represent means±S.E. of ≥5 specimens of each BMT group. Compared with WT mice receiving WT^GFP BMT: *, p < 0.05.

β-arrestins regulate SMC proliferation and apoptosis in injured arteries

Since neointimal hyperplasia in our model system involved principally arterial cells, we reasoned that arterial SMC proliferation should be diminished by β-arrestin1 expression, and enhanced by β-arrestin2 expression. Moreover, the effect of β-arrestins on arterial SMC apoptosis, if any, should be the inverse of the β-arrestin effect on proliferation. To test these hypotheses, we examined the prevalence of proliferating cell nuclear antigen (PCNA) and cleaved caspase-3 (for apoptosis) in SMCs of the carotid artery media 2 weeks after endothelial denudation (while neointimal hyperplasia was still developing). Deficiency of β-arrestin2 reduced medial SMC proliferation by ∼2.4-fold, and augmented SMC apoptosis ∼2-fold (p < 0.01, Figure 4). By contrast, deficiency of β-arrestin1 augmented medial SMC proliferation by ∼1.8-fold, and reduced SMC apoptosis ∼2.5-fold (p < 0.01, Figure 4). Thus, βarrestin isoforms appeared to affect arterial SMC proliferation, and apoptosis, in a reciprocal manner.

Figure 4.

Arterial SMC proliferation and apoptosis are reciprocally regulated by β-arrestin1 (βarr1) and β-arrestin2 (βarr2) after endothelial denudation. Two-week-old specimens from Figure 2 were immunostained for proliferating cell nuclear antigen (PCNA) or cleaved caspase-3, and counter-stained for DNA. Within the tunica media (delimited by arrows), all nuclei were counted, as well as all PCNA-positive nuclei. The number of PCNA-positive nuclei was divided by the total number of nuclei to obtain “% of total,” plotted for each mouse genotype as the mean ± SE of 3 independent samples stained batch-wise. Cleaved caspase-3 immunofluorescence obtained for the entire arterial cross-section was normalized to nuclear fluorescence in the same cross section; this ratio (“arbitrary units”) was used to compare samples of each genotype, stained batch-wise. Photomicrographs represent ≥ 3 specimens of each genotype (scale bar = 50 μm). Compared with WT samples: *, p<0.01.

To determine mechanisms by which β-arrestin isoforms could effect this reciprocal regulation of net SMC proliferation, we examined the activation of ERK1/2 and Akt in SMCs of the arterial media. Signaling by ERK and Akt isoforms is critical for SMC survival and proliferation in response to diverse stimuli.¹⁴ Moreover, β-arrestin isoforms serve as scaffolds for the ERK cascade members c-raf, MEK, and ERK1/2, in addition to Akt, and β-arrestins affect ERK and Akt activation downstream of 7-transmembrane receptors as well as receptor tyrosine kinases.⁵ Two weeks after activation by endothelial denudation, carotid SMCs of the tunica media indeed demonstrated a β-arrestin isoform-dependent pattern of ERK and Akt activation: phosphorylated (activated) Akt and ERK1/2 were ∼1.5-to-2-fold more abundant in β-arrestin1^-/-, and 33-67% less abundant in β-arrestin2^-/- than in WT arteries (p < 0.05, Figure 5). Thus, β-arrestin isoform-dependent ERK and Akt activation in vivo correlated with β-arrestin isoform-dependent SMC proliferation and neointimal hyperplasia—in that SMC proliferative signaling and proliferation were enhanced by β-arrestin2 activity, and diminished by β-arrestin1 activity.

Figure 5.

ERK1/2 and Akt activation in SMCs of injured arteries is potentiated by β-arrestin1 (βarr1) deficiency, and reduced by β-arrestin2 (βarr2) deficiency. Two-week-old specimens of each genotype from Figure 2 were immunostained for phospho-ERK1/2 (red) or phospho-Akt (not shown) and counter-stained for DNA (blue) (original magnification ×1100). Within the tunica media (delimited by arrows), phospho-ERK1/2 or phospho-Akt immunofluorescence was normalized to nuclear fluorescence, to obtain “phospho-ERK (or phospho-Akt) per cell.” These quotients were obtained from 4 randomly selected high-power fields throughout each specimen, and normalized to those obtained for WT specimens within the same staining group, to obtain “% of WT.” Plotted are the mean±S.E. of 3 specimens of each genotype. Compared with WT: *, p<0.05.

β-arrestin2 affects SMC migration and proliferation in vitro

To investigate further the role of β-arrestin isoforms in the SMC proliferation and migration that are critical to neointimal hyperplasia, we used aortic SMCs derived from our congenic WT, β-arrestin1^-/- and β-arrestin2^-/- mice (2-3 distinct lines of each genotype). These congenic SMC lines demonstrated the expected, genotype-specific expression patterns of β-arrestins, assessed by immunoblotting (Figure IV.A). However, these congenic SMC lines demonstrated equivalent expression of G protein-coupled receptor kinase-2 (GRK2) and GRK5—kinases whose activities can determine the extent and functional consequences of β-arrestin isoform binding to 7-transmembrane receptors¹⁵ (immunoblot data not shown). In addition, as assessed by semi-quantitative PCR, the SMC lines expressed equivalent levels of the 7-transmembrane receptors for lysophosphatidic acid (LPA), LPA1 and LPA2 (Figure IV.A), protease-activated (thrombin) receptors (PAR1-4), and sphingosine-1-phosphate receptors (S1P_1-5) (data not shown). These receptors are believed to be important for SMC responsiveness to platelet-derived products in the context of vascular injury.^16,17 In response to LPA, calcium signaling was comparable in β-arrestin-expressing and –deficient SMC lines (Figure IV.B). Thus, to the extent that β-arrestins regulate LPA-induced calcium signaling, it appeared that expression of either β-arrestin1 or β-arrestin2 suffices; that β-arrestin1 may compensate for deficiency of β-arrestin2, and vice versa.

Could expression of either β-arrestin1 or β-arrestin2 also suffice for regulating SMC migration, which involves signaling not only through heterotrimeric G proteins but also small G proteins,¹⁸ phosphoinositide 3-kinase, and ERKs?¹⁴ To address this question, we evoked SMC migration through 7-transmembrane receptors and receptor tyrosine kinases, respectively, for LPA and PDGF—both abundantly derived from the activated platelets involved in neointimal hyperplasia.¹⁸ PDGF-promoted SMC migration appeared to be independent of β-arrestin1 and β-arrestin2 (Figure 6A). However, LPA-promoted migration was 30% less in β-arrestin2^-/- than in WT SMCs, even though it was equivalent in β-arrestin1^-/- and WT SMCs (Figure 6A). Similarly, thrombin-promoted migration (2-fold/basal) was 35% less in β-arrestin2^-/- than in WT SMCs (p < 0.05, data not shown). Thus, the ability of β-arrestin2 to promote SMC migration in response to some platelet-derived products accords well with the ability of β-arrestin2 to promote neointimal hyperplasia (Figure 2) and atherosclerosis (Figure 1).

Figure 6.

β-arrestin isoform-specific effects on SMC migration and proliferation. SMCs of the indicated genotypes were subjected to (A) migration, (B) thymidine incorporation, or (C) proliferation assay without (basal) or with 10 μmol/L LPA, 100 nmol/L S1P, 10 units/ml of α-thrombin, 0.4 nmol/L PDGF-BB, as indicated (n=5/group), or 10% FBS (C). A, Migration values for stimulated SMCs were normalized to cognate unstimulated SMCs, to obtain “fold/basal.” Shown are means ± S.E. (n=5/SMC line). Basal SMC migration was equivalent in all SMC lines. Compared with LPA-stimulated WT SMCs: *, p<0.01. B, CPM from agonist-stimulated SMCs were divided by those obtained in cognate unstimulated SMCs, to obtain “fold/basal.” Basal values were equivalent in all SMC lines. Compared with WT SMCs: *, p<0.05. C, SMCs harvested at the indicated time points were quantitated by lamin ELISA (see Methods). The means±S.E. of 4 independent experiments are shown. Compared with WT SMCs: *, p<0.05 (2-way ANOVA).

Since β-arrestins significantly influenced SMC proliferation in vivo, we sought to determine which SMC signaling pathways could confer β-arrestin isoform-specific proliferative responses. To that end, we tested SMC thymidine incorporation elicited by receptors for platelet-derived serum constituents implicated in neointimal hyperplasia: the 7-transmembrane receptors for LPA, thrombin, and sphingosine-1-phosphate, as well as by the receptor protein tyrosine kinases for PDGF and EGF.^16,19-21 As with SMC migration, β-arrestin1 and β-arrestin2 activity appeared to have no effect on SMC thymidine incorporation stimulated by PDGF (Figure 6B) or EGF (data not shown). In contrast, LPA evoked 33% more thymidine incorporation in β-arrestin1^-/- than in WT SMCs, while LPA, S1P, and thrombin evoked 20-40% less thymidine incorporation in β-arrestin2^-/- than in WT SMCs (Figure 6B, p<0.05). Concordantly, serum evoked ∼40% less proliferation from β-arrestin2^-/- than from WT SMCs (Figure 6C), and serum deprivation engendered 50 ± 10% more cell death in β-arrestin2^-/- than in WT SMCs (n=7, p < 0.01, data not shown). It thus appears that β-arrestin1 activity attenuates, while β-arrestin2 activity potentiates mitogenic signaling evoked through 7-transmembrane and perhaps other growth factor receptors in SMCs. Consequently, the activities of the β-arrestin isoforms in vitro appear concordant with those observed with neointimal hyperplasia in vivo, and we can infer that β-arrestin isoform-mediated signaling downstream of key 7-transmembrane receptors may be responsible for β-arrestin isoform-specific effects on neointimal hyperplasia.

Unlike SMCs, β-arrestin2^-/- and WT endothelial cells demonstrated equivalent proliferation and apoptosis, in response to serum and serum deprivation, respectively (Figure V and data not shown)—just as β-arrestin2^-/- and WT mice demonstrated equivalent carotid re-endothelialization (Figure II). This failure of β-arrestin2 deficiency to affect endothelial cell proliferation likely results from the undetectably low β-arestin2 expression levels in mouse (Figure V.A) and human umbilical vein endothelial cells (data not shown).

β-arrestins affect SMC signaling through ERK

Proliferative and chemotactic signaling downstream of β-arrestins include both Raf/MEK/ERK and phosphoinositide 3-kinase/Akt pathways.⁵ In light of the β-arrestin isoform-specific effects on migration and proliferation observed in Figure 6, we asked whether ERK, Akt, or both pathways confer β-arrestin isoform-specific proliferation or migration in SMCs. To address this question in the context of a mixture of agonists that promote neointimal hyperplasia, we challenged our congenic SMCs with fetal bovine serum, which contains all of the agonists used in Figure 6.^16,18 Serum evoked equivalent Akt activation in WT and β-arrestin-deficient SMCs, but evoked 32% less ERK activation in β-arrestin2^-/- than in WT SMCs (p < 0.01, Figure 7). Serum similarly evoked 38 ± 10% less ERK activation in WT SMCs when β-arrestin2 expression was reduced (79 ± 8%) by acute treatment with siRNA (p < 0.05, Figure 7C). In contrast, serum activated Akt and ERK equivalently in β-arrestin2^-/- and WT endothelial cells (Figure V.C). Thus, SMCs and endothelial cells demonstrate cell type-specific roles of β-arrestin2, and serum-stimulated SMCs recapitulate only a subset of the β-arrestin isoform-dependent ERK and Akt findings we obtained in neointimal hyperplasia (Figure 5). Differences between in vivo and in vitro findings likely relate to greater inflammatory cytokine levels present in injured arteries as compared with purified serum, as well as the time course of the assays (weeks versus minutes). Moreover, while SMC Akt activity promotes the “contractile” SMC phenotype when ERK is not concomitantly activated, SMC ERK activity engenders the “activated” SMC phenotype associated with migration, proliferation, and neointimal hyperplasia.^14,18 We therefore sought to characterize further the effects of β-arrestins on SMC ERK activation induced by the components of serum utilized in Figure 6.

Figure 7.

SMC β-arrestin2 promotes serum-stimulated signaling through ERK, but not Akt. A, B, WT, β-arrestin1^-/- and β-arrestin2^-/- SMCs were serum-starved for 72 hr, exposed to serum-free medium without (basal) or with 10% (v/v) fetal bovine serum (FBS) for 5 min, and then lysed. Lysates were immunoblotted for either phospho-(p-)Akt or p-ERK, and then for actin. Band densities for p-Akt and p-ERK were divided by cognate actin band densities; these quotients were normalized to those of WT SMCs, to obtain “% of WT control.” Shown are immunoblots from a single experiment, and means±S.E. from 3 experiments performed with 2 independent SMC lines of each genotype. Basal ERK and Akt activation were equivalent in SMC lines of each genotype. Compared with WT SMCs: *, p<0.01. C, WT SMCs transfected with control or β-arrestin2-targeting siRNA were treated exactly as in (A). Shown are immunoblots from a single experiment, representing 3 performed.

Stimulated by individual agonist constituents of serum, SMCs demonstrated ERK activation patterns consistent with antagonistic roles of β-arrestin1 and β-arrestin2, as we observed with SMC thymidine incorporation in Figure 6. ERK activation triggered by the 7-transmembrane receptors for LPA, thrombin, and sphingosine-1-phosphate was diminished in β-arrestin2^-/- SMCs, but enhanced in β-arrestin1^-/- SMCs (Figure 8). In contrast, deficiency of β-arrestin isoforms did not affect ERK activation elicited by SMC receptor tyrosine kinase(s) for PDGF (Figure 8B). These agonist-specific results for β-arrestin-dependent ERK activation and thymidine incorporation are highly congruent with β-arrestin isoform-specific effects on neointimal hyperplasia. Consequently, these agonist-specific results suggest that β-arrestin-dependent effects on neointimal hyperplasia are mediated substantially, if not exclusively, by effects on signaling through 7-transmembrane receptors. Together with ERK and Akt activation data from Figures 5 and 7, these data in Figure 8 also support the inference that β-arrestins affect neointimal hyperplasia through ERK-mediated mechanisms.

Figure 8.

β-arrestin2 promotes and β-arrestin1 inhibits ERK activation in response to 7-transmembrane receptor agonists. SMCs were serum-starved for 72 hr, and then exposed (37 °C) to serum-free medium without (basal) or with agonists identical to those in Figure 6B, for the times indicated. SMCs were then lysed for SDS-PAGE and immunoblotting for phospho- and total ERK. A, Phospho- and total ERK1/2 immunoblots are presented from LPA-stimulated SMCs. B-D, Densities of phospho-ERK bands were normalized to those of cognate total ERK bands; these ratios were divided by those for unstimulated WT SMCs, to obtain “fold/basal.” Fold/basal values are plotted as mean±S.E. from ≥5 independent experiments. Compared with WT time course curves: *, p<0.05 (two-way ANOVA).

Discussion

Our studies demonstrate that the multifunctional scaffolding protein β-arrestin2 contributes to the pathogenesis of atherosclerosis. This novel finding in mice seems highly pertinent to human atherosclerosis, since atherosclerotic human coronary arteries show ∼2-fold higher β-arrestin2 mRNA levels than non-atherosclerotic coronary arteries.⁹ That β-arrestin2 activity augments atherosclerosis by augmenting SMC proliferation and migration derives credence from three concordant lines of evidence: atheroma SMC prevalence and β-arrestin2 expression, SMC-dependent neointimal hyperplasia triggered by endothelial denudation, and cultured primary SMCs. Moreover, we discovered that while β-arrestin2 promotes SMC proliferation, migration, and consequently neointimal hyperplasia, β-arrestin 1 does not. Indeed, β-arrestin1 inhibits these processes in vivo. Thus, in response to the innumerable stimuli associated with vascular injury, the activities of β-arrestin1 and β-arrestin2 in SMCs appear to be antagonistic to each other—as we found them to be with regard to SMC ERK signaling and proliferation induced in vitro by 7-transmembrane receptors for LPA, thrombin, and sphingosine-1-phosphate.

That atherosclerosis can be affected primarily, or exclusively, by changes in SMC physiology was first clearly demonstrated with the SMC-specific knockout of the multifunctional low-density lipoprotein receptor-related protein-1 (LRP1).³ As a result of SMC LRP1 deficiency, SMCs express more PDGF receptor-β (PDGFRβ), and demonstrate enhanced proliferation and migration. Consequently, SMC LRP1 deficiency substantially aggravates atherosclerosis in ldlr^-/- mice, in a manner that can be largely eliminated by inhibiting PDGFRβ signaling.³ Although β-arrestin2 can exacerbate atherosclerosis by enhancing SMC proliferation and migration, our SMC findings certainly do not exclude possible atherogenic effects of β-arrestin2 on migration of CD4⁺ lymphocytes⁵ or monocytes, or on macrophage proliferation.¹

When 7-transmembrane receptors are activated in SMCs, β-arrestin1 and β-arrestin2 exert comparable effects on intracellular calcium signaling, but markedly different effects on ERK signaling and thymidine incorporation. These divergent, signal-specific effects of β-arrestins on 7-transmembrane receptor signaling characterize the LPA receptors studied in our SMCs as well as the atherogenic²² angiotensin II AT₁ receptor, and appear to result from divergent signaling mediated by heterotrimeric G proteins or other, distinct mechanisms. In mouse embryo fibroblasts and HEK cells, expression of either β-arrestin1 or β-arrestin2 suffices to achieve normal desensitization of angiotensin II-evoked phosphoinositide hydrolysis (just upstream of cytosolic calcium signaling).^5,8 However, like our SMC ERK signaling evoked by LPA, thrombin, and sphingosine-1-phosphate, angiotensin II-induced ERK signaling in HEK cells is promoted by β-arrestin2, and reduced by β-arrestin1 activity.⁸ While the association of either β-arrestin isoform with 7-transmembrane receptors terminates G protein-dependent signaling, this same association also triggers signaling through ERKs and other mechanisms—in a manner that depends not only on the specific β-arrestin isoform, but also on the phosphorylation state of the 7-transmembrane receptor.^5,15 In addition, the newly described role of β-arrestins in activating Wnt/Frizzled/β-catenin signaling may also contribute to β-arrestin2-promoted SMC proliferation.²³

Because PDGF plays such a prominent role in promoting neointimal hyperplasia provoked by arterial injury²⁰ and atherosclerosis,¹ it may seem surprising that β-arrestins affect neointimal hyperplasia significantly, and yet have no effect on primary SMC mitogenic signaling or migration in response to PDGF (Figures 6 and 8). This apparent paradox may be resolved, however, if we consider that PDGF and 7-transmembrane receptors signal synergistically through many mechanisms, including NAD(P)H oxidases.²² Mitogenic signaling in primary SMCs more than doubles when agonists for G_q/11- and G_i-coupled 7-transmembrane receptors are added to PDGF, even at receptor-saturating PDGF concentrations.^11,24 By altering SMC mitogenic signaling elicited by 7-transmembrane receptors, β-arrestin activity should augment (β-arrestin2) or reduce (β-arrestin1) mitogenic signaling elicited by the combination of PDGF and platelet-derived agonists for 7-transmembrane receptors. Furthermore, since PDGFRs can be transactivated by 7-transmembrane receptors,²² β-arrestin-mediated regulation of 7-transmembrane receptors may alter levels of SMC PDGFR activation. Although β-arrestins transactivate the EGFR in a manner that is 7-transmembrane receptor- and cell type-specific,^25,26 the role of β-arrestins in transactivating the PDGFR remains obscure.

Our bone marrow transplantation data demonstrate that β-arrestin2 expression in bone marrow-derived cells does not affect carotid injury-induced neointimal hyperplasia, and that neointimal cells do not derive from bone marrow precursors in our model of carotid de-endothelialization. These findings may reflect the possibility that although β-arrestin2 is important for lymphocyte chemotaxis to CXCL12,⁵ it is not important for the migration of leukocytes in response to the diversity of stimuli encountered in carotid injury. To reconcile the absence and presence of bone marrow-derived cells in the neointima of the injured carotid and femoral¹³ arteries, respectively, we must consider arterial dimensions. Relative to the injury-inducing guidewire diameter (0.36 mm), the mouse femoral artery diameter is only 50%,²⁷ while the carotid diameter is ∼90% (Figure I).¹⁰ Consequently, the abundant medial apoptosis and necrosis observed with femoral artery injury²⁷ is greatly diminished (or absent) with carotid artery injury. It is therefore quite conceivable that guidewire injury of the carotid artery provides insufficient stimulus to recruit bone marrow-derived cells that differentiate into neointimal SMCs.

In light of the newly discovered role of β-arrestin2 in atherogenesis and in regulating injury-induced neointimal hyperplasia, novel therapeutic approaches to atherosclerosis and arterial restenosis after angioplasty could conceivably involve inhibition of SMC β-arrestin2 activity, or enhancement of SMC β-arrestin1 activity. Our studies of SMCs and endothelial cells in vivo and in vitro suggest a salient potential advantage of anti-β-arrestin2 approaches: the possibility of inhibiting atherogenic SMC activity without inhibiting anti-atherogenic endothelial cell activity. Whether these therapeutic possibilities will prove practicable, of course, remains to be ascertained.