Hyperglycemia induces vascular smooth muscle cell dedifferentiation by suppressing insulin receptor substrate-1–mediated p53/KLF4 complex stabilization

Hyperglycemia and insulin resistance accelerate atherosclerosis by an unclear mechanism. The two factors down-regulate insulin receptor substrate-1 (IRS-1), an intermediary of the insulin/IGF-I signaling system. We previously reported that IRS-1 down-regulation leads to vascular smooth muscle cell (VSMC) dedifferentiation and that IRS-1 deletion from VSMCs in normoglycemic mice replicates this response. However, we did not determine IRS-1's role in mediating differentiation. Here, we sought to define the mechanism by which IRS-1 maintains VSMC differentiation. High glucose or IRS-1 knockdown decreased p53 levels by enhancing MDM2 proto-oncogene (MDM2)-mediated ubiquitination, resulting in decreased binding of p53 to Krüppel-like factor 4 (KLF4). Exposure to nutlin-3, which dissociates MDM2/p53, decreased p53 ubiquitination and enhanced the p53/KLF4 association and differentiation marker protein expression. IRS-1 overexpression in high glucose inhibited the MDM2/p53 association, leading to increased p53 and p53/KLF4 levels, thereby increasing differentiation. Nutlin-3 treatment of diabetic or Irs1−/− mice enhanced p53/KLF4 and the expression of p21, smooth muscle protein 22 (SM22), and myocardin and inhibited aortic VSMC proliferation. Injecting normoglycemic mice with a peptide disrupting the IRS-1/p53 association reduced p53, p53/KLF4, and differentiation. Analyzing atherosclerotic lesions in hypercholesterolemic, diabetic pigs, we found that p53, IRS-1, SM22, myocardin, and KLF4/p53 levels are significantly decreased compared with their expression in nondiabetic pigs. We conclude that IRS-1 is critical for maintaining VSMC differentiation. Hyperglycemia- or insulin resistance–induced IRS-1 down-regulation decreases the p53/KLF4 association and enhances dedifferentiation and proliferation. Our results suggest that enhancing IRS-1–dependent p53 stabilization could attenuate the progression of atherosclerotic lesions in hyperglycemia and insulin-resistance states.

Insulin-like growth factor-I (IGF-I) 3 and insulin coordinately regulate cellular growth and differentiation in response to changes in nutritional status and intermediary metabolism (1). However, the mechanisms by which both hormones signal cells to grow or differentiate are not well defined. Both hormones stimulate these processes through their receptors, which directly phosphorylate insulin receptor substrate-1 (IRS-1) and Src homology 2 domain-containing-transforming protein C (Shc) to transmit their signals to downstream signaling components of the PI3K and mitogen-activated protein kinase pathways (2). During normal physiologic conditions, the PI3K pathway promotes glucose influx; glycogen, lipid and protein synthesis; and changes in gene expression that help to maintain cellular differentiation (3,4). However, in response to hyperglycemia, IRS-1 is down-regulated in multiple cell types, and insulin or IGF-I signaling through IRS-1 is impaired (5,6). In cell types that are capable of undergoing dedifferentiation, such as vascular smooth muscle cells (VSMC), IRS-1 down-regulation is associated with up-regulation of a cell surface-associated scaffolding protein termed SHPS-1 (7). Under these conditions, stimulation of the IGF-I receptor leads to recruitment of the tyrosine kinase CTK to the plasma membrane, and CTK directly phosphorylates SHPS-1 (8). SHPS-1 functions as a scaffold and recruits kinases that activate both the PI3K and mitogen-activated protein kinase pathways (6,9). This signaling switch occurs in vivo in VSMCs of diabetic mice in response to hyperglycemia. This change is accompanied by VSMC dedifferentiation (10) and an enhanced cellular proliferative response to injury (10). To determine whether loss of IRS-1 expression in response to hyperglycemia was mediating these changes, we deleted IRS-1 in VSMC in mice. Aortic VSMC in which IRS-1 expression had been deleted underwent dedifferentiation under normoglycemic conditions and had a hyperproliferative response to vascular injury that was similar to the response of diabetic mice. Therefore, the loss of IRS-1 was sufficient to induce these changes.
Previous work shows that VSMC dedifferentiation is associated with enhanced expression of the transcription factor KLF4 (11) and that enhanced KLF4 expression, which occurs in response to cytokine-induced stress, suppresses transcription of the primary determinant of VSMC differentiation, myocardin (11,12). We demonstrated that KLF4 expression increased in arteries of diabetic or normoglycemic IRS-1 Ϫ/Ϫ mice, and myocardin was suppressed (10). Although the results showed that IRS-1 was required for myocardin expression and differentiation, they did not delineate the mechanism by which maintenance of IRS-1 expression enhances differentiation and inhibits dedifferentiation. Some studies show that increased KLF4 expression in VSMC leads to dedifferentiation, but other studies suggest that KLF4 expression increases expression of p21, which inhibits cell cycle progression and increases myocardin to promote VSMC differentiation. This discrepancy was resolved by Yoshida et al. (13), who showed that the variable that accounts for this difference is the balance between p53 and KLF4. In the presence of adequate p53, there is increased nuclear p53/KLF4 association, which enhances myocardin and p21 expression. In the absence of p53 association, KLF4 inhibits myocardin and p21. Because dedifferentiation is an important component of the atherosclerotic process (14) and hyperglycemia and insulin resistance, which down-regulate IRS-1, are known to accelerate the development of atherosclerosis (15,16), we undertook these studies to determine the mechanism by which IRS-1 functions to maintain VSMC differentiation.
To determine how p53 was down-regulated in VSMC following exposure to high glucose, we investigated the role of the ubiquitin ligase MDM2, which ubiquitinates p53 and targets it for proteasomal degradation. MDM2 increased 3.8 Ϯ 0.8-fold (p Ͻ 0.05) (n ϭ 3) during a 6 -8 h exposure to high glucose, which also increased p53/MDM2 complexes (Fig. 3A). To determine how the glucose-induced increase in MDM2 mediated the reduction in p53, we utilized nutilin-3, a compound that disrupts p53/MDM2 association (17). Exposure to nutlin-3 inhibited p53/MDM2 (Fig. 3B) and inhibited p53 ubiquitination ( Fig. 3C), confirming that MDM2 was functioning to lower p53 levels primarily by this mechanism. Exposure of cells maintained in high glucose to nutlin-3 increased total p53 and p53/ KLF4 to levels that were similar to VSMC maintained in 5 mM glucose (Fig. 3D). Importantly, nutlin-3 increased nuclear p53 and p53/KLF4 (7.2 Ϯ 1.4-fold, p Ͻ 0.01, n ϭ 3 and 9.1 Ϯ 2.6fold, p Ͻ 0.01, n ϭ 3, respectively) (Fig. 3E). The downstream factors myocardin, SM22, and p21 were increased (Fig. 3F). Total KLF4 was unchanged, suggesting that changes in myocardin and p21 expression were dependent upon increased KLF4/ p53 association (Fig. 3D). To confirm these results and assess their importance for regulating differentiation, we utilized a cell-permeable peptide that was designed to disrupt p53/ MDM2 association. Exposure of cells maintained in 25 mM glucose to the disrupting peptide was associated with a decrease in p53/MDM2 association and a decrease in p53 ubiquitination (Fig. 3G). Nuclear p53 concentrations as well as p53/KLF4 increased 2.6 Ϯ 0.3-fold (p Ͻ 0.05) (n ϭ 3) and 6.6 Ϯ 0.6-fold (p Ͻ 0.01) (n ϭ 3), respectively (Fig. 3H). Furthermore, p21, myocardin, and SM22 expression increased significantly even in the presence of high glucose to a concentration similar to that in cells maintained in normal glucose (Fig. 3I). In contrast, the addition of the peptide to VSMC maintained in 5 mM glucose resulted in no change in p21, SM22, or myocardin (Fig.  6D).

IRS-1 maintains VSMC differentiation
cose to the peptide reduced p53/KLF4 and decreased the expression myocardin, SM22, and p21 significantly (Fig. 5G). The peptide had no effect on the expression of these proteins when added to cells maintained in high glucose (Fig. 6D). To further examine the role of IRS-1, we prepared a synthetic peptide that disrupted p53/IRS-1 association. In the presence of normal glucose, this peptide increased MDM2/p53 association (2.7 Ϯ 0.2-fold, p Ͻ 0.01 (n ϭ 3)) and enhanced p53 ubiquitination (5.2 Ϯ 1.1-fold, p Ͻ 0.01 (n ϭ 3) (Fig. 6A). These changes were accompanied by a decrease in nuclear p53 and nuclear p53/KLF4 association (Fig. 6B) and significant reductions in differentiation marker proteins (Fig. 6C). Differentiation marker protein expression was unchanged when the peptide was added to cells maintained in 25 mM glucose (Fig. 6D).

IRS-1 maintains VSMC differentiation
aorta, which was increased in diabetic and IRS-1 Ϫ/Ϫ mice compared with control mice (Fig. 8, D and E).
Although VSMCs from diabetic mice undergo several changes in signaling that are similar to changes that occur in atherosclerotic lesions, they do not develop typical subintimal plaques unless they are also hyperlipidemic. To determine if the changes in p53, IRS-1, and KLF4 noted in our mice were present in an animal model that does develop subintimal lesions, we analyzed femoral arteries obtained from diabetic pigs that had been fed a high fat diet and had been shown to have extensive atherosclerosis (18). The results show that arterial extracts from the diabetic animals had a 2.6 Ϯ 0.1-fold reduction (p Ͻ 0.001) (n ϭ 5) in IRS-1 and a 2.8 Ϯ 0.1-fold decrease (p Ͻ 0.01) (n ϭ 5) in p53 compared with nondiabetic pigs (Fig. 9A). Myocardin and SM22 were also significantly decreased. Most importantly the extracts from the diabetic animals showed a marked reduction in p53/KLF4 association (3.5 Ϯ 0.2-fold (n ϭ 8), p Ͻ 0.001) (Fig. 9B). These changes are consistent with the changes that occur in diabetic and IRS-1 Ϫ/Ϫ mouse aorta.

Discussion
VSMCs possess the unusual characteristic that fully differentiated cells can dedifferentiate, leading to both accelerated proliferation and dysfunctional expression of proteins that are nec-

Figure 3. Prevention of p53 down-regulation via blocking ubiquitin proteasomal degradation rescues mSMC differentiation during hyperglycemia.
A, cell lysates from mSMCs exposed to HG for the indicated times were immunoprecipitated (IP) with anti-MDM2 and immunoblotted (IB) with anti-p53. They were also directly immunoblotted with anti-MDM2 or ␤-actin. B-F, cell lysates or nuclear fraction extracts from mSMCs cultured in NG or HG in the presence or absence of nutlin-3 (10 M). B, cell lysates (n ϭ 3) were immunoprecipitated with anti-MDM2 and immunoblotted with anti-p53 or directly immunoblotted with anti-MDM2. Scanning densitometry showed that nutlin-3 treatment significantly impaired high glucose-induced p53/MDM2 association (e.g. 2.4 Ϯ 0.6-fold reduction, n ϭ 4, p Ͻ 0.05). C, cell lysates were immunoprecipitated with anti-p53 and immunoblotted with an anti-ubiquitin (Ub) or only immunoblotted with an anti-p53. Scanning densitometry showed that high-glucose exposure increased p53 ubiquitination 3.2 Ϯ 0.6-fold (p Ͻ 0.01) (n ϭ 3), and nutlin-3 treatment prevented this increase (a 1.1 Ϯ 0.3-fold increase, p ϭ NS compared with cells exposed to NG). D, cell lysates were immunoprecipitated with an anti-KLF4 and immunoblotted with an anti-p53 or directly immunoblotted with anti-p53, anti-␤-actin, or anti-KLF4. The bar graph shows the ratios Ϯ S.D. (error bars) (n ϭ 4) of scanning densitometry values of p53 divided by KLF4. E, nuclear fraction extracts were immunoprecipitated with anti-KLF4 and immunoblotted with anti-p53 or directly immunoblotted for p53. PARP-1 was the loading control. F, cell lysates were immunoblotted with anti-MyoC, anti-p21, anti-SM22, or anti-␤-actin. The bar graph shows the ratios Ϯ S.D. (n ϭ 4) of scanning densitometry values of MyoC or p21 or SM22 divided by ␤-actin. G-I, whole-cell lysates or nuclear fraction extracts were obtained from mSMCs exposed to NG, NHG, or HG in the presence of a control peptide (Ctrl) or a p53/MDM2-disrupting peptide (DP). G, cell lysates were immunoprecipitated with anti-MDM2 or p53 and immunoblotted with anti-p53 or ubiquitin. Cell lysates were also directly immunoblotted with anti-MDM2 or p53 as loading controls. Scanning densitometry showed that the p53/MDM2-disrupting peptide reduced MDM2/p53 association by 3.7 Ϯ 0.1-fold (p Ͻ 0.05) (n ϭ 3) and p53 ubiquitination by 3.5 Ϯ 0.1-fold (n ϭ 3) (p Ͻ 0.01). H, nuclear fraction extracts were immunoprecipitated with anti-KLF4 and immunoblotted with anti-p53. They were directly immunoblotted for PARP-1 as a loading control. I, cell lysates were immunoblotted with an anti-MyoC, p21, SM22, or ␤-actin. The bar graph shows the ratios Ϯ S.D. (n ϭ 3) of scanning densitometry values of MyoC, p21, or SM22 divided by the ␤-actin. p Ͻ 0.05 and p Ͻ 0.01 indicate the significant differences.  Figure 5. Overexpression of IRS-1 up-regulates p53 and its association with KLF4, which maintains mSMC differentiation during hyperglycemia. A-E, whole-cell lysates or nuclear fraction extracts were obtained from mSMCs expressing LacZ or IRS-1 exposed to HG or NHG. A, cell lysates (n ϭ 3) were immunoprecipitated (IP) with anti-KLF4 and immunoblotted (IB) with anti-p53. They were also directly immunoblotted with anti-IRS-1, p53, ␤-actin, and KLF4. Scanning densitometry showed that overexpression of IRS-1 resulted in a 3.0 Ϯ 0.8-fold increase in p53 (p Ͻ 0.01) and 3.3 Ϯ 0.8-fold increase of KLF4-associated p53 (p Ͻ 0.001) under hyperglycemic conditions. B, nuclear extracts were immunoprecipitated with anti-KLF4 and immunoblotted with anti-p53. PARP-1 was used as a loading control. The bar graph shows the ratios Ϯ S.D. (error bars) (n ϭ 3) of scanning densitometry values of p53 divided by PARP-1. C, cell lysates were immunoprecipitated with anti-MDM2 or IRS-1 and immunoblotted with anti-p53 or IRS-1. They were also immunoblotted with anti-MDM2. D, cell lysates were immunoprecipitated with anti-ubiquitin (Ub) and immunoblotted with anti-p53. E, cell lysates were immunoblotted with anti-MyoC, p21, SM22, or ␤-actin. The bar graph shows the ratios Ϯ S.D. (n ϭ 3) of scanning densitometry values of MyoC, p21, or SM22 divided by ␤-actin. F and G, nuclear fraction extracts and cell lysates (n ϭ 3) were obtained from mSMCs exposed to NG treated with control peptide (Ctrl P) or p53/KLF4 disrupting peptide (DP). F, nuclear fraction extracts were immunoprecipitated with anti-KLF4 and immunoblotted with anti-p53. They were also directly immunoblotted with anti-KLF4 or PARP-1. G, cell lysates were immunoblotted with an anti-MyoC, anti-p21, anti-SM22, or anti-␤-actin. The bar graph shows the ratios Ϯ S.D. (n ϭ 3) of scanning densitometry values of MyoC or p21 or SM22 divided by ␤-actin. p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001 indicate the significant differences.

IRS-1 maintains VSMC differentiation
essary for normal contractile function (19). Hyperglycemia and insulin resistance predispose to increased VSMC dedifferentiation and atherosclerotic lesion development (10,15,20). Therefore, identification of the factors that regulate the ability of VSMC to maintain the differentiated state as well as those that lead to dedifferentiation will further our understanding of how changes in metabolism lead to arterial dysfunction and atherosclerosis. Both hyperglycemia and increased insulin resistance down-regulate IRS-1 in VSMCs and in blood vessels of diabetic animals (10,21). This reduction is due to enhanced IRS-1 serine phosphorylation and increased ubiquitination that targets it to a proteasome (22). Recently, we reported that down-regulation of IRS-1 in the arteries of diabetic mice was accompanied by the enhanced expression of KLF4, a transcription factor that enhances VSMC dedifferentiation under certain conditions (10). KLF4 suppressed expression of myocardin, a transcription factor that is required to maintain VSMC in the differentiated state. The importance of IRS-1 in regulating this response was confirmed by deleting IRS-1 expression in normoglycemic mice and then demonstrating increased KLF4, reduced myocardin expression, and enhanced the proliferative response to injury. We concluded that down-regulation of IRS-1 in response to hyperglycemia led to dedifferentiation, but the mechanism by which IRS-1 maintained differentiation was not defined.
Our results show that IRS-1 stabilizes p53 concentrations, which enhances nuclear p53/KLF4 and drives expression of myocardin and other proteins required for VSMC differentiation (Fig. 10). Additionally, expression of p21, a cell cycle inhibitor, increased significantly. Knocking down IRS-1 in VSMC in vitro and in vivo in the presence of normal glucose replicated the effects of hyperglycemia, suggesting that hyperglycemia functions by decreasing IRS-1. The importance of maintaining p53 was confirmed by demonstrating that hyperglycemia down-regulates p53 and overexpression of p53 in VSMC during hyperglycemia restores p53/KLF4 as well as differentiation marker protein expression. High glucose enhances MDM2-mediated ubiquitination of p53, which down-regulates p53, but IRS-1 overexpression inhibits the p53/MDM2 interaction and p53 ubiquitination, thereby stabilizing p53 and promoting p53/ KLF4-mediated induction of differentiation during chronic hyperglycemia. Although diabetic mice have signaling abnormalities that are present in atherosclerotic vessels, they do not develop lesions in the absence of hyperlipidemia; therefore, we analyzed arterial tissue obtained from diabetic pigs that develop extensive lesions (18). The results showed that p53 and IRS-1 Figure 6. Disruption of p53/IRS-1 interaction increases p53 ubiquitination and degradation and decreases p53/KLF4 association, leading to attenuated differentiation during normoglycemia. Whole-cell lysates or nuclear fraction extracts were obtained from mSMCs exposed to normal glucose (NG) treated with control peptide (Ctrl P) or p53/IRS-1-disrupting peptide (DP). A, cell lysates (n ϭ 3) were immunoprecipitated with anti-p53, ubiquitin (Ub), or MDM2 and immunoblotted with anti-IRS-1 or p53. They were also directly immunoblotted with anti-p53 or MDM2. B, nuclear extracts were immunoprecipitated with anti-KLF4 and immunoblotted with anti-p53. PARP-1 was used as a loading control. The bar graph shows the ratios Ϯ S.D. (n ϭ 3) of scanning densitometry values of p53-or KLF4-associated p53 divided by PARP-1. C, cell lysates were immunoblotted with an anti-MyoC, anti-p21, anti-SM22, or anti-␤-actin. The bar graph shows the ratios Ϯ S.D. (n ϭ 3) of scanning densitometry values of MyoC, p21, or SM22 divided by ␤-actin. p Ͻ 0.001 and p Ͻ 0.01 indicate the significant differences. D, whole-cell lysates (n ϭ 3) were obtained from mSMCs exposed to NG (5 mM) or HG (25 mM) in the presence of a control peptide (Ctrl) or a p53/MDM2-disrupting peptide (DP), a p53/KLF4-disrupting peptide, or a p53/IRS-1-disrupting peptide. Cell lysates were immunoblotted with an anti-MyoC, anti-p21, anti-SM22, or anti-␤-actin antibody.

IRS-1 maintains VSMC differentiation
were down-regulated in porcine lesions as well as the markers of differentiation. Importantly, there was a 3.5 Ϯ 0.2-fold reduction in p53/KLF4 association. These results are consistent with our findings in diabetic and IRS-1 knockout mice and suggest that these changes may be an important component of diabetic animal atherosclerotic lesion development.
P53 regulates VSMC proliferation, and its deletion enhances VSMC proliferation in response to hypercholesterolemia (23). Bone marrow transplant studies comparing p53-expressing and non-p53-expressing VSMC showed that endogenous p53 reduces atherosclerosis in APOE knockout mice (24). Similarly, overexpression of the p55␥ subunit of PI3K blocked MDM2/ p53 interaction, thereby increasing p53 and attenuating the proliferative response to rat carotid artery injury (25). Additionally, a long noncoding RNA, link RNA p21, that inhibited p53 ubiquitination in APOE Ϫ/Ϫ mice enhanced p53-mediated repression of VSMC proliferation (26). Although p53 knockdown in VSMC in APOE Ϫ/Ϫ mice did not alter lesion size, it did increase cell number in lesions, and it enhanced invasiveness (27). Because lesion size is regulated by multiple factors, such as blood pressure, hyperlipidemia, and cytokines, it is possible that they increased lesion size in these mice independently of

IRS-1 maintains VSMC differentiation
changes in p53. Diabetes is known to accelerate the rate of lesion progression, but it is not believed to be the cause of lesion initiation. Therefore, glucose-induced changes in p53 would be expected to contribute more to histologic changes that are related to lesion progression.
p53 is down-regulated during hyperglycemia (28); however, induction of AMP kinase stimulated p53 and p21 and down-regulated cyclin D1 in high glucose-exposed VSMC, leading to inhibition of cell proliferation in vitro (29). However, AMPK was equally effective in inhibiting cyclin D1 in normal and high glucose, making it difficult to conclude that it specifically counteracts the effect of high glucose. Our studies show that inhibition of p53/MDM2 with nutlin-3 or a specific peptide that disrupted their interaction reduced ubiquitinated p53, increased . C, aortic thickness were measured from each group (n ϭ 5) following the procedure described under "Experimental procedures." D and E, total aortic extracts were prepared from NM, nondiabetic WT (WT), DM, and nondiabetic IRS-1SMC knockout mice (IRS-1 Ϫ/Ϫ) that had been injected with nutlin-3 or PBS (Ctrl) as described under "Experimental procedures." The aortic extracts were immunoblotted with an anti-macrophage scavenger receptor (MSR) (1:500; Trans Genic Inc.) or ␤-actin antibody. Scanning densitometry values showed that MSR expression was significantly increased in DM (e.g. 3.0 Ϯ 0.4-fold, n ϭ 3, p Ͻ 0.001) or in IRS-1 Ϫ/Ϫ mice (e.g. 2.4 Ϯ 0.7-fold, n ϭ 3, p Ͻ 0.001) compared with NM or WT mice, respectively. Nutlin-3 treatment prevented this increase in both types of mice (e.g. 3.1 Ϯ 1.0-fold reduction in DM (n ϭ 3, p Ͻ 0.001) or 2.7 Ϯ 0.6-fold reduction in IRS-1 Ϫ/Ϫ mice (n ϭ 3, p Ͻ 0.001)). p Ͻ 0.001, p Ͻ 0.01, and p Ͻ 0.05 indicate significant differences when the two treatments were compared, and NS indicates no significant difference.

IRS-1 maintains VSMC differentiation
p53 and p53/KLF4 association, and enhanced differentiation in diabetic or IRS-1 Ϫ/Ϫ mice. p53 overexpression gave similar findings. Therefore, we conclude that MDM2 activation is an important mechanism by which hyperglycemia down-regulates p53 in VSMC. Overcoming the hyperglycemia-induced reduction in p53 allows VSMC to retain the ability to differentiate.

IRS-1 maintains VSMC differentiation
Hyperglycemia-induced p53 down-regulation resulted in decreased p53 nuclear content and decreased p53/KLF4 association. The role of nuclear p53/KLF4 in regulating VSMC differentiation has been analyzed (13). Pidkoka et al. (30) found that KLF4 was a potent transcriptional repressor of VSMC differentiation markers, such as myocardin, in the absence of p53 association. Other studies confirmed that KLF4 is induced during dedifferentiation, and this is consistent with our findings in both hyperglycemic and IRS-1 Ϫ/Ϫ mice (31,32). KLF4 associates with histone deacetylase 2 or 5 and the cofactors SRF and ELK-1, which results in inhibition of differentiation marker gene expression (31). Yoshida et al. (33) showed that KLF4stimulated dedifferentiation was induced by stimulation of NF-B association with KLF4 and that inhibition of NF-B following neointimal injury inhibited KLF4-induced dedifferentiation. Our prior studies showed that high glucose induced a signaling switch from IRS-1 to SHPS-1 (8). This resulted in activation of p65Rel A by PKC on the SHPS-1 scaffold and NF-B activation (34). Therefore, in high glucose, the increased activated nuclear NF-B could bind to KLF4, leading to suppression of myocardin expression.
In contrast to its role as a mediator of dedifferentiation, Wassmann et al. (35) demonstrated that KLF4 enhanced expression of the VSMC differentiation marker SM22A. Shi et al. (36) found that all transretinoic acid induced multiple VSMC differentiation marker genes in a KLF4-dependent manner. Several studies confirm that the transactivation function of KLF4 to induce specific genes depends upon post-translational modifications, which mediate cofactor association and determine the response to KLF4 induction (37). These discrepant findings regarding KLF4 function in VSMC have been addressed by Yoshida et al. (13). They demonstrated that conditional deletion of KLF4 delayed down-regulation of VSMC differentiation markers but also accelerated neointimal proliferation (38). Enhanced expression of KLF4 in VSMC was associated with induction of p21 and reduced cellular proliferation. Importantly, they and others (39) documented that increased binding of p53 and KLF4 to the p21 promoter led to increased p21 and inhibited proliferation. Wassmann et al. (35) confirmed these findings by demonstrating that in the presence of p53, KLF4 induced VSMC differentiation. They also reported that inhibitor of differentiation 3 (ID3) binding to KLF4 resulted in p53 repression and enhancement of VSMC proliferation and that in the absence of ID3 overexpression, the predominant function of p53/KLF4 was to inhibit VSMC proliferation (40). Thus, high-glucose suppression of p53 may be a predominant mechanism by which phenotypic switching from the quiescent, differentiated phenotype in the presence of low glucose and high p53 levels is reversed, allowing unrestrained KLF4 suppression of differentiation. Our finding that disruption of p53/KLF4 during normoglycemia resulted in loss of p21 expression and dedifferentiation as characterized by reduced myocardin and SM22 further strengthens the conclusion that p53/KLF4 association is required for maintaining the differentiated phenotype in VSMC.
Our results show that the major mechanism by which IRS-1 functions to retain VSMC in the differentiated phenotype is to protect p53 from degradation. Overexpression of IRS-1 in high glucose resulted in decreased MDM2/p53 association and p53 ubiquitination. This increased nuclear p53 and p53/KLF4 association. Conversely, knockdown of IRS-1 promoted p53 ubiquitination and decreased p53/KLF4 and the downstream differentiation marker proteins. Furthermore, in diabetic mice, nutlin-3, which increased nuclear p53 and p53/KLF4 association, stimulated VSMC differentiation, indicating that reversal of p53 ubiquitination would reverse hyperglycemia-induced phenotypic switching. Because IRS-1 overexpression results in similar findings, we conclude that a basal level of IRS-1 expression is necessary to retain sufficient p53 in the nucleus to induce the differentiation-promoting effects of KLF4. This conclusion was confirmed by disrupting p53/IRS-1 in normoglycemic mice, which led to p53 degradation, ubiquitination, loss of nuclear p53, and a major reduction in myocardin and p21. Taken together, our results show that IRS-1 prevents p53 ubiquitination, which promotes downstream signaling events that maintain the differentiated phenotype. Conversely, loss of this association during exposure to high glucose results in loss of p53, dedifferentiation, and increased VSMC proliferation. Because our prior study showed that loss of IRS-1 led to a hyperpoliferative response to injury, we conclude that IRS-1 is an important modulator of VSMC proliferation and that loss of IRS-1 in hyperglycemia could account for the acceleration in VSMC proliferation noted in diabetics with atherosclerosis.
There are minimal published data regarding IRS-1 and maintenance of the normal vascular phenotype or loss of IRS-1 in atherosclerosis. Sobue and co-workers (41) demonstrated that dephosphorylation of IRS-1 Tyr-895 by SHP-2 resulted in blockade of ERK activation in vitro. Therefore, maintenance of high IRS-1 concentrations in the presence of adequate SHP-2 would be expected to help to maintain a low rate of VSMC proliferation (41). Similarly, adiponectin induced VSMC differentiation and stabilized IRS-1 concentrations in vitro (3). Conversely, Thomas et al. reported that overexpression of SHPS-1 in skeletal muscle during normoglycemia resulted in loss of IRS-1 activation by the insulin receptor and inhibition of protein synthesis, suggesting that activation of SHPS-1 inhibited this differentiated function. T cadherin expression down-regulated IRS-1, and this was associated with VSMC dedifferentiation and reversed with rapamycin (42). Similarly, Taniyama et al. (43) showed that induction of reactive oxygen species downregulated IRS-1 in VSMC in response to angiotensin II, and this was associated with increased proliferation. Because hyperglycemia induces reactive oxygen species, it is likely that both mechanisms are operating in parallel. One study suggested that attenuating IRS-1 function may be relevant to the development of human atherosclerosis. Baroni et al., who demonstrated that the G792R mutation in IRS-1 impairs insulin signaling (44), reported that this mutation was associated with a 2.93-fold increase in the relative risk ratio for the presence of coronary artery disease (45). Therefore, these studies support the conclusion that maintenance of IRS-1 facilitates VSMC differentiation and may reduce the risk of developing atherosclerosis.
In summary, glucose-induced down-regulation of IRS-1 results in loss of P53/KLF4 association, thereby decreasing p53/ KLF4 complex induction of myocardin and p21. Reduced myocardin and p21 leads to enhanced VSMC dedifferentiation and

IRS-1 maintains VSMC differentiation
proliferation in diabetic mouse aorta. The results have important implications for understanding the mechanism by which hyperglycemia facilitates atherosclerotic lesion formation.

Cell culture
VSMCs were isolated from mouse (mSMCs) or pig (pSMCs) aortas using a method that had been described previously (46). The mSMCs were maintained in Dulbecco's modified Eagle's medium with glucose (1 g/liter normal glucose (NG) or 4.5 g/liter high glucose (HG)) supplemented with 10% fetal bovine serum (Atlanta Biologics, Flowery Branch, GA), streptomycin (100 ng/ml), and penicillin (100 units/ml). Cultures were grown to confluent density prior to the initiation of the experiment. In the case of normal-high glucose transient exposure (NHG), mSMCs were cultured in the NG until confluence before adding 20 mM glucose for the indicated times. For MDM2/p53disrupting experiments, the MDM2/p53-disrupting peptide (10 g/ml) was added to mSMC cultures when 20 mM glucose was added and incubated for 6 h. Nutlin-3 (10 M) was added when 20 mM glucose was added and incubated for 8 h. For differentiation marker studies, disrupting peptides (10 g/ml) or nutlin-3 (10 M) were added every 24 h, and cultures were incubated for 48 h before cells were harvested. The cells that were used in these experiments were used between passages 5 and 15.

Transient transfection with siRNA targeting IRS-1 and p53
siRNA targeting IRS-1 (sc-29377), siRNA targeting p53 (sc-29436), and a control siRNA (sc-37007) were purchased from Santa Cruz Biotechnology, Inc. mSMCs were transfected using a concentration of 30 pM for IRS-1 and 90 pM for p53 and using the PepMute Plus reagent (SignaGen Laboratories, MD) following the manufacturer's instructions. The experiments were initiated 24 -48 h after transfection.

Immunoprecipitation and immunoblotting
The immunoprecipitation and immunoblotting procedures were performed as described (47). Immunoprecipitation was performed by incubating 0.5 mg of cell lysate protein with 1 g of anti-KLF4 or MDM2 or p53 or IRS-1 or ubiquitin antibodies at 4°C overnight. Immunoblotting was performed using a dilution of 1:1000 for anti-PARP-1, SM22, and ␤-actin antibodies and a dilution of 1:500 for anti-KLF4, p53, MDM2, IRS-1, p21, and myocardin antibodies. The proteins were visualized using enhanced chemiluminescence (Thermo Fisher Scientific).

Mice
All mouse experiments were approved by the institutional animal care and use committees of the University of North Carolina (Chapel Hill, NC). The floxed IRS-1 mice were provided by Dr. Morris White (Harvard Medical School). The generation of smooth muscle-specific IRS1 Ϫ/Ϫ mice has been described (10). Mice were maintained at 22°C with a 12-h light/dark cycle and given free access to regular chow (2018 Teklad global rodent diet containing 18.6% protein, 6.2% fat, and 3.5% crude fiber) and water. All groups of mice maintained normal nutrient intake and growth during the experiment.

Induction of hyperglycemia in mice and preparation of whole aortic lysates for biochemical analysis
Hyperglycemia was induced in WT male mice (C57BL/6) using low-dose streptozotocin (49). All mice had serum glucose concentrations that were Ͼ250 mg/dl, and the levels were maintained during the experiments. After acclimation, mice underwent conscious tail-cuff blood pressure measurement (10 times for each mouse), recording systole/diastole averaged over 40 cycles (Code 8, Kent Scientific, Inc., Torrington, CT), which showed no differences between diabetic mice and nondiabetic mice (103 Ϯ 15/141 Ϯ 12 versus 107 Ϯ 9/141 Ϯ 8 mm Hg, p ϭ NS) or WT mice and smooth muscle-specific IRS-1 Ϫ/Ϫ mice (117 Ϯ 12/151 Ϯ 5 versus 108 Ϯ 4/145 Ϯ 11 mm Hg, p ϭ NS). There were 16 mice per group (WT, WT with diabetes, and smooth muscle-specific IRS-1 Ϫ/Ϫ mice) that were used for biochemical analyses and Ki67 staining studies. IGF-I (1 mg/kg) (n ϭ 6) or PBS (n ϭ 6) was administered i.p. 24 h and 15 min before sacrifice for assessment of Ki67 labeling. For nutlin-3 treatment, 5 mg/kg was administered i.p. every day for 5 days. PBS was used as a control. For the aortic thickness experiments, IGF-I was administered i.p. every day for 5 days. For the p53/ IRS-1-disrupting experiment, the disrupting peptide (2 mg/kg) or control peptide (2 mg/kg) was injected every day for IRS-1 maintains VSMC differentiation preparation followed a procedure described previously (10).

Induction of diabetic pigs and preparation of femoral artery extracts
Diabetes was induced in pigs using strepozotocin, and femoral artery extracts were prepared as described previously (18). The animals were diabetic for 6 months prior to analysis.

Nuclear fraction extract preparation
Nuclear fraction extracts from cells or mice aortas were isolated using NE-PER nuclear extraction reagents (Thermo Scientific) following a procedure provided by the manufacturer. Protein concentrations of extracts were measured using a BCA assay (Thermo Scientific). Equal amounts of protein were used for direct immunoblotting. For protein/protein interaction studies, that a similar amount input of the protein of interest was added was determined based on the results obtained from direct immunoblotting.

Immunohistochemistry
The aortas from mice were fixed with 4% paraformaldehyde overnight, and paraffin-embedded sections were prepared by the University of North Carolina histology core facility. An immunohistochemistry paraffin protocol provided by Abcam was followed, and the procedures were described previously for Ki67 and DAPI staining (50). The Ki67-positive and total nuclei (DAPI-positive) in a whole aortic ring were quantified using ImageJ (version 1.50i, National Institutes of Health) and expressed as the percentage of the total nuclei. To study aortic thickness, eight adjacent 5-m sections were cut every 500 m and stained with hematoxylin and eosin. The thickness was measured following the procedures described previously (10).

Statistical analysis
The results that are shown in all experiments are representative of at least three independent experiments and expressed as the mean Ϯ S.D. Student's t test was used when two data points were compared. Analysis of variance was applied when multiple points were compared. p Ͻ 0.05 was considered statistically significant.
Author contributions-G. X. designed and performed many of the experiments. He planned the mouse breeding program as well as supervising technical work that was necessary to complete the manuscript. D. R. C. helped to design experiments and plan the studies. He reviewed the data extensively and prepared the manuscript. M. F. W. prepared and provided the floxed IRS-1 mice. X. S. performed some in vitro experiments. C. W. maintains the mouse breeding program and assisted in preparation of the tissues for biochemical and immunohistochemical analysis.