Activation of Vascular Protein Kinase C-β Inhibits Akt-Dependent Endothelial Nitric Oxide Synthase Function in Obesity-Associated Insulin Resistance

All protocols for animal use and euthanasia were reviewed and approved by the animal care committee of the Joslin Diabetes Center and were in accordance with National Institutes of Health guidelines. Male Zucker fatty rats and control Zucker lean rats (Harlan Sprague Dawley, Indianapolis, IN) at 14 weeks of age were allocated to either of two treatment regimens for 2 weeks: normal chow ad libitum or chow containing 0.062% (wt/wt) of the PKCβ-specific inhibitor ruboxistaurin (LY333531) (Lilly Research Laboratories, Indianapolis, IN), corresponding to a dose of 30 mg/kg body wt. Physiological parameters of the Zucker obese and lean rats are described in Table 1.

The creation of FVB mice with a PKCβ2 transgene under control of the prepro-endothelin promoter (PKCβ2Tg), resulting in overexpression of PKCβ2 in endothelial cells and vascular smooth muscle cells, has been described previously (16). The mice used in the current study were bred from the same colony. Incorporation of the transgene was demonstrated by Southern blotting, and transgene expression was confirmed by Northern blotting and immunoblotting from preparations of heart, aorta, and retina. Male mice were studied at 20 weeks of age using wild-type nonlittermate FVB mice as controls.

Isolation of microvessels from epididymal fat pads by collagenase digestion and centrifugation was performed with a previously validated method (5,11). Fat pads were excised from transgenic mice overexpressing vascular PKCβ2 or from control wild-type FVB mice at 20 weeks of age. Isolated microvessels were resuspended in Dulbecco’s modified Eagle’s medium (DMEM) with 0.1% BSA and left for 30 min at 37°C or stimulated for the same duration by adding insulin to the suspension.

Measurement of insulin-stimulated vascular cGMP production was used to assess NO bioavailability in vivo (see below). Because cGMP produced as a result of insulin stimulation would not be stable during isolation of microvessels as described above, we chose to measure cGMP concentration before and after insulin stimulation as part of a euglycemic-hyperinsulinemic clamp in Zucker rats, performed as described previously (11).

Immediately after killing each rat, the aorta and heart were isolated, frozen in liquid nitrogen, and stored at −80°C for subsequent analysis. Frozen tissue was crushed in liquid nitrogen, and the powdered aorta was divided into two parts for Western blot analysis and cGMP assay, respectively.

As described previously, DAG in the aorta was determined by a DAG kinase assay (17), and PKC activity was measured in cytosol and membrane fractions of tissue lysate as ³²P-incorporation into the peptide substrate RKRTLRRL (single-letter amino acid abbreviations) (17).

Vascular cGMP concentration was measured using a commercially available enzyme immunoassay (Amersham Biosciences). Aortas were homogenized in 6% (vol/vol) trichloroacetic acid at 4°C, thus precipitating sample protein. After centrifugation (2,000 × g for 15 min at 4°C), the pellet was stored for subsequent measurement of protein content, and the supernatant was used for the determination of cGMP concentration by enzyme immunoassay according to the manufacturer’s instructions.

Cytosol and membrane fractions of lysate from aorta were fractionated by a method involving ultracentrifugation steps as described in detail previously (17).

PI 3-kinase activity was assayed in insulin receptor substrate 2 immunoprecipitates followed by separation of ³²P-labeled phospholipids with thin-layer chromatography as described previously (5).

Bovine aortic endothelial cells (BAECs) were isolated by scraping the luminal side of calf aorta with a cover glass. Cells were cultured in DMEM containing 5.5 mmol/l glucose and 10% platelet-poor horse serum (Sigma, St. Louis, MO) and used between passages 2 and 7. Subconfluent cells were starved in DMEM containing 0.1% BSA for 24 h before stimulation with insulin or VEGF. In experiments with adenovirus infection, serum starvation was begun 24 h after the start of virus incubation.

We have previously described construction of recombinant adenovirus (18), including those encoding a constitutively active Akt (gag fused to the NH₂-terminal of Akt); an Akt mutant acting as a dominant-negative active Akt (Akt T308A/S473A); enhanced green fluorescent protein (GFP) gene together with PKCα, -β2, -δ, or -ζ; GFP together with a PKCζ K273W dominant-negative mutant; and control adenovirus encoding either gag or GFP. An adenovirus encoding PKCβ1 was made with similar techniques. We have previously shown that BAECs have endogenous expression of PKCα, -β1, -β2, -δ, -ε, and -ζ (19).

BAECs were infected with one of the adenovirus constructs described above or with control adenovirus containing the same parental genome but being a vector only for gag or for GFP. Cell cultures were incubated with 0.5 × 10⁸ plaque-forming units for 1 h. The virus solution was then diluted fourfold in growth medium, and the cultures were incubated for an additional 24 h. Expression of each recombinant protein was confirmed by Western blotting. The PKC constructs caused expression of the isoform in question to increase ∼10-fold compared with cells infected with the control adenovirus.

Cells, isolated microvessels, or frozen, powdered aorta were solubilized in lysis buffer as described previously (5). In the case of aorta, lysis was promoted by treatment in a glass Dounce homogenizer on ice. Lysate was assayed with Western blotting as described previously (5), using antibodies to phospho-Akt (Ser473), Akt (both from Cell Signaling Technology, Beverly, MA), and eNOS (BD Biosciences Pharmingen, San Diego, CA).

Total RNA from BAECs was isolated, and eNOS mRNA was assayed by Northern blotting and normalized to 36B4 mRNA as described previously (5).

Data are presented as the means ± SE. Results were analyzed with the Kruskal-Wallis one-way ANOVA, using post hoc comparisons with paired or unpaired t test as appropriate. P values <0.05 were considered statistically significant.

To study obesity-associated changes in vascular insulin signaling, we used the model of nondiabetic insulin resistance offered by the Zucker fatty rat (20). As expected, Zucker fatty rats had elevated fasting plasma insulin, 10.1- ± 2.0-fold higher than Zucker lean rats (Table 1). Fasting blood glucose was 98 ± 5 mg/dl (5.4 ± 0.3 mmol/l) in Zucker lean rats and 124 ± 8 mg/dl (6.9 ± 0.4 mmol/l) in Zucker fatty rats (P = 0.04). During euglycemic-hyperinsulinemic clamp, insulin-stimulated glucose disposal rate was reduced by 78 ± 5% in Zucker fatty rats compared with lean rats (P < 0.001, Table 1).

In the aorta of Zucker fatty rats, total DAG concentration was 46% higher than in lean rats (Fig. 1A, P < 0.05). PKC activity was not different in the cytosol fraction in fatty and lean rats (Fig. 1B), but membrane-associated PKC activity was 63% higher in fatty rats (Fig. 1B, P < 0.05). We have previously found, by using antibodies against PKCα, -β1, -β2, and -γ, that only PKCα and -β2 were detectable in the aorta of rats with streptozotocin-induced diabetes and that only PKCβ2 expression was increased in the membrane fraction (17). To examine PKC isoform expression in the lean and fatty Zucker rats, we immunoblotted PKCα, -β2, -δ, and -ε in the cytosol and membrane fractions of lysate from aorta. The membrane-associated expression was increased for all of these isoforms (Fig. 2A), whereas cytosol-associated expression was unchanged (Fig. 2A; summarized data not shown). Furthermore, protein expression of eNOS was decreased by 64% in the heart of fatty rats compared with lean rats (P < 0.05, Fig. 2B), although it was unchanged in aorta (120 ± 36% of the levels in lean rats, NS). In microvessels isolated from fat pads, Akt protein expression was not significantly different between lean and fatty Zucker rats (94 ± 25% in lean rats relative to fatty rats, P = 0.6; results not shown).

In lean rats, hyperinsulinemic clamp stimulated Akt phosphorylation by 3.0- ± 1.0-fold compared with saline infusion, but hyperinsulinemia had no effect on Akt phosphorylation in fatty rats (Fig. 3A). NO production in the aorta was measured as cGMP production. Hyperinsulinemic clamp increased cGMP concentration in the aorta by 2.9- ± 0.7-fold in lean rats (P < 0.05) but had no effect in fatty rats (Fig. 3B).

We have previously shown that activation of PKC with phorbol ester or overexpression of PKCβ1 inhibits insulin-stimulated eNOS expression in endothelial cell culture (5). We therefore evaluated whether vascular insulin resistance in vivo can be reversed by inhibition of PKCβ. A separate group of Zucker lean and fatty rats was treated for 2 weeks with the selective PKCβ inhibitor ruboxistaurin administered in chow before being studied by a euglycemic-hyperinsulinemic clamp. Glucose disposal rate was not statistically different between ruboxistaurin-treated and untreated rats in either lean or fatty rats (Table 1). Insulin-stimulated Akt phosphorylation in aorta was not different in ruboxistaurin-treated lean rats compared with untreated lean rats, but ruboxistaurin treatment partly restored insulin-stimulated Akt phosphorylation in fatty rats (no increase compared with saline infusion in untreated fatty rats; 2.1- ± 0.2-fold increase in ruboxistaurin-treated fatty rats; P < 0.01, Fig. 3A). Similar results were obtained for cGMP production. Thus, ruboxistaurin treatment did not change insulin-stimulated cGMP production in aorta of lean rats but completely normalized insulin-stimulated cGMP production in fatty rats (no increase in cGMP concentration compared with saline infusion in untreated fatty rats; 3.4- ± 0.8-fold increase in ruboxistaurin-treated fatty rats; P < 0.05, Fig. 3B).

Akt, activated by PI 3-kinase, directly phosphorylates eNOS on Ser1177 after stimulation of endothelial cells with insulin, growth factors, shear stress, or other factors (12). We have previously shown that insulin-stimulated induction of eNOS gene expression is dependent on PI 3-kinase (5). For the present study, we wished to evaluate whether insulin-stimulated eNOS gene induction is dependent on Akt or PKCζ, both of which are downstream of PI 3-kinase during insulin signaling. In BAECs, dominant-negative Akt expression significantly decreased eNOS mRNA compared with the control condition (Fig. 4). After infection with GFP or gag control adenovirus, 6 h of insulin stimulation increased eNOS mRNA expression (by 1.7- ± 0.3- and 1.7- ± 0.1-fold compared with GFP control and gag control, respectively; Fig. 4). Expression of dominant-negative Akt completely inhibited insulin-stimulated eNOS induction, whereas dominant-negative PKCζ had no effect (Fig. 4).

As mentioned above, we have previously shown that activation of PKC with phorbol ester or overexpression of PKCβ1 inhibit insulin-stimulated eNOS expression (5). We next evaluated whether effects on insulin-stimulated Akt-mediated eNOS induction are limited to the PKCβ isoforms. In BAECs, adenovirus-mediated overexpression of PKCα, -β1, -β2, -δ, or -ζ did not change basal Akt phosphorylation compared with control adenovirus infection. In cultures infected with control adenovirus, insulin stimulation for 30 min increased Akt phosphorylation by 3.6-fold, and this stimulation was inhibited by overexpression of PKCβ1 and -β2 (by 50 and 51%, respectively, P < 0.01, Fig. 5A). Overexpression of PKCα, -δ, and -ζ inhibited insulin-stimulated Akt phosphorylation to a lesser degree (by 27–37%), but these differences were not statistically significant.

A similar pattern was observed with regard to eNOS mRNA expression. Overexpression of PKCα, -β1, -β2, -δ, or -ζ did not change basal eNOS mRNA. Insulin stimulation for 6 h increased eNOS mRNA 2.2-fold, and this stimulation was significantly inhibited by overexpression of PKCβ1 and -β2. Inhibition was also observed for PKCα, -δ, and -ζ, although these changes were not statistically significant (Fig. 5B). Thus, the inhibitory effect of PKC on eNOS induction stimulated by insulin and mediated by Akt is preferentially due to activation of the PKCβ isoforms.

As previously stated, several hormones and growth factors besides insulin are known to regulate eNOS activation through the PI 3-kinase/Akt pathway (12). To determine whether Akt activation caused by stimuli other than insulin is also inhibited by PKC, BAECs overexpressing PKCα, -β1, -β2, and -δ were stimulated with VEGF. Adenovirus-mediated overexpression of PKCβ1 and -β2 inhibited VEGF-stimulated Akt phosphorylation by 98 ± 21 and 95 ± 15%, respectively, compared with control adenovirus infection (P < 0.01, Fig. 6). In contrast, overexpression of PKCα, -δ, and -ζ did not significantly inhibit VEGF-stimulated Akt phosphorylation. Thus, the selective ability of the PKCβ isoforms to inhibit Akt signaling is not limited to insulin signaling.

To confirm whether the PKCβ isoforms can inhibit vascular insulin signaling in vivo, we studied mice overexpressing PKCβ2 only in vascular tissue (PKCβ2Tg). In microvessels isolated from epididymal fat from PKCβ2Tg mice, protein expression of PKCβ2 and PKC activity were both increased by twofold compared with wild-type mice (data not shown). There was no significant difference in body weight or fasting plasma glucose concentrations between transgenic and wild-type mice (data also not shown). mRNA expression of eNOS in microvessels was decreased by 42 ± 7% in PKCβ2Tg mice compared with wild-type mice (data not shown; P < 0.05). Akt phosphorylation in response to stimulation of microvessels with insulin ex vivo was decreased by 60 ± 12% (P < 0.05, Fig. 7).

This study has provided evidence that activation of PKC, in particular the β-isoforms, can induce insulin resistance in endothelial cells and vascular tissue and cause endothelial dysfunction both in cultured cells and in vivo. More specifically, overexpression of PKCα, -β1, -β2, -δ, and -ζ in endothelial cell culture showed that insulin-stimulated, Akt-dependent eNOS gene expression is preferentially inhibited by the PKCβ isoforms. Furthermore, studies in mice overexpressing PKCβ2 in the vasculature and in Zucker fatty rats showed that in vivo, PKCβ2 activation is sufficient to inhibit vascular insulin-stimulated Akt signaling and that the impaired insulin-stimulated vascular NO production in obesity-associated insulin resistance can be partially normalized with the PKCβ-selective inhibitor ruboxistaurin.

It is well-established that chronic activation of vascular PKC in diabetes is due to de novo synthesis of DAG from glucose (13). We provide the first evidence that DAG concentrations and PKC activity are increased in blood vessels in insulin resistance associated with obesity. In the Zucker fatty rat, vascular PKC activation may be the result of increased plasma concentrations of FFAs (21). Elevated FFAs have been reported to activate PKC in skeletal muscle (22–24). Decreased insulin action is associated with activation of PKCβ2 and -δ in humans (22), as opposed to the PKCθ isoform in rats (23) and mice (24), suggesting that the PKC isoform involved may be species dependent. A tissue-specific activation of PKC isoforms may also exist: because the PKCβ-specific inhibitor ruboxistaurin improved vascular insulin signaling in the Zucker fatty rat in our study, but had no effect on whole-body insulin sensitivity, it is possible that different PKC isoforms inhibit insulin signaling in blood vessels than in other tissues. We performed experiments in rats and bovine cell culture because we have extensive experience with insulin clamp and ruboxistaurin treatment in rats and with insulin signaling and adenovirus-mediated overexpression of PKC isoforms in BAECs. Furthermore, for in vivo overexpressing of PKCβ2, we used mice in which transgenic techniques are standard. Even though the conclusions regarding the isoform or isoforms activated in vascular tissue are limited by the use of several species in the current study, we have previously shown that the β2 isoform of PKC is selectively activated by diabetes and high glucose concentrations in vascular tissue of rats (17) and mice (25) as well as in bovine vascular cells (26). Furthermore, the observation that ruboxistaurin prevented impairment of endothelium-dependent vasodilation in humans during hyperglycemia is consistent with activation of PKCβ in human vasculature (15).

Insulin-stimulated cGMP production in the aorta was decreased in parallel with Akt phosphorylation in Zucker fatty rats compared with lean rats and increased after ruboxistaurin treatment compared with no treatment. cGMP is a second messenger for NO. A limitation of cGMP as a marker of NO production is that it only indicates NO bioavailability. Thus, differences in insulin-stimulated cGMP production may reflect differences in eNOS expression, eNOS phosphorylation, eNOS regulation by intracellular Ca²⁺ concentrations, or NO stability. Furthermore, cGMP may also be produced in response to stimulation of the transmembrane form of guanylate cyclase by hormones such as atrial naturitic factor.

In endothelial cell culture, we observed that insulin-stimulated induction of eNOS expression is dependent on Akt. Thus, insulin stimulation of acute activation of eNOS (4) as well as upregulation of eNOS expression is mediated by Akt. Furthermore, we observed that PKCβ1 and -β2, but not PKCα, -δ, or -ζ, decreased insulin-stimulated Akt-dependent induction of eNOS gene expression. The comparison of different PKC isoforms overexpressed in BAECs is limited by the fact that similar overexpression levels of PKC isoforms in whole-cell lysate may not result in the same extent of PKC activation for each isoform. Therefore, we cannot exclude the possibility that isoforms other than PKCβ2 may also interfere with Akt/eNOS signaling. We did not attempt to measure activity of individual overexpressed PKC isoforms, because immunoprecipitated activity from the membrane fraction of cell lysate would not only be technically challenging but also inherently flawed because it would depend on variable affinity in antibody association with each PKC isoform. Nonetheless, the importance of PKCβ in modulating this pathway was further supported by the finding that insulin-stimulated Akt activation was blunted in PKCβ2Tg mice and that insulin-stimulated Akt activation and NO bioavailability could be partly normalized by pharmacological PKCβ inhibition in Zucker fatty rats. Importantly, in PKCβ2Tg mice, in which the prepro-endothelin promoter was used to target PKCβ2 overexpression to vascular cells, we demonstrated a reduction of eNOS expression by >40% caused by vascular PKCβ2 activation without systemic signs of metabolic abnormalities. These findings are in agreement with findings in mice with knockout of endothelial insulin receptors, in which loss of endothelial insulin signaling and reduced vascular eNOS expression did not result in abnormal glucose tolerance (27).

In cultured endothelial cells, we also demonstrated that activation of PKC can inhibit Akt activation stimulated not only by insulin but also by VEGF. This finding has important implications. First, impaired Akt signaling in insulin resistance may compromise enzyme activation and gene expression of eNOS stimulated by several factors, not only insulin. Second, decreased insulin-stimulated eNOS gene expression may limit eNOS enzyme activity stimulated by several factors besides insulin, such as VEGF.

In the present study, eNOS protein expression was decreased in the heart of Zucker fatty rats compared with lean rats but not in the aorta. We have previously published that in microvessels isolated from Zucker fatty rats, vascular eNOS mRNA expression is reduced to 29% of the level in control Zucker lean rats (5). Biological effects in microvascular cells are much more sensitive to insulin than in cells from the aorta, as we have previously published (28). To support this idea, we have recently studied the effect of subcutaneous insulin implants on eNOS mRNA expression in microvessels from lean and fatty rats after a treatment duration of 2 weeks. The results showed that the effect of insulin on eNOS expression was reduced in Zucker fatty rats compared with lean rats (data not shown). The concept that eNOS expression decreases as a result of impaired insulin signaling in vascular endothelium was supported in the current study by demonstrating that eNOS expression is decreased in PKCβ2Tg mice. These findings are consistent with the 62% reduction of eNOS mRNA in the aorta found in mice with knockout of endothelial insulin receptors (27).

In summary, the current study shows that activation of PKC, in particular the β-isoforms, is present in vascular tissue in insulin resistance associated with obesity and causes inhibition of Akt-dependent eNOS regulation. We propose that these mechanisms may be part of the explanation for why the atherosclerotic process is accelerated in people with obesity-associated insulin resistance and in patients with diabetes. Further studies will be needed to show whether PKCβ activation enhances atherosclerosis or restenosis and not just endothelial cell dysfunction. Clinical trials have shown that the PKCβ isoform inhibitor ruboxistaurin is safe during long-term treatment in humans, and this will allow testing of the hypothesis that PKCβ activation causes endothelial dysfunction and accelerates atherosclerosis in insulin resistance and diabetes.

G.L.K. has received National Institutes of Health grant NIDDK R01-DK-53105. C.R-M. has received Danish Medical Research Council Grant 22-01-0498 and Danish Heart Foundation Grant 01-2-2-79-22946.