Ceramide-Initiated Protein Phosphatase 2A Activation Contributes to Arterial Dysfunction In Vivo

Abstract

Prior studies have implicated accumulation of ceramide in blood vessels as a basis for vascular dysfunction in diet-induced obesity via a mechanism involving type 2 protein phosphatase (PP2A) dephosphorylation of endothelial nitric oxide synthase (eNOS). The current study sought to elucidate the mechanisms linking ceramide accumulation with PP2A activation and determine whether pharmacological inhibition of PP2A in vivo normalizes obesity-associated vascular dysfunction and limits the severity of hypertension. We show in endothelial cells that ceramide associates with the inhibitor 2 of PP2A (I2PP2A) in the cytosol, which disrupts the association of I2PP2A with PP2A leading to its translocation to the plasma membrane. The increased association between PP2A and eNOS at the plasma membrane promotes dissociation of an Akt-Hsp90-eNOS complex that is required for eNOS phosphorylation and activation. A novel small-molecule inhibitor of PP2A attenuated PP2A activation, prevented disruption of the Akt-Hsp90-eNOS complex in the vasculature, preserved arterial function, and maintained normal blood pressure in obese mice. These findings reveal a novel mechanism whereby ceramide initiates PP2A colocalization with eNOS and demonstrate that PP2A activation precipitates vascular dysfunction in diet-induced obesity. Therapeutic strategies targeted to reducing PP2A activation might be beneficial in attenuating vascular complications that exist in the context of type 2 diabetes, obesity, and conditions associated with insulin resistance.

Figure 1

Palmitate-induced ceramide activates PP2A, disrupts PP2A colocalization with I2PP2A, and impairs eNOS^Ser1177 phosphorylation. Representative images (A) and mean data (B) indicating that palmitate (pal)-induced ceramide accumulation is prevented by the Ser palmitoyl transferase inhibitor myriocin (myr) (n = 8 per treatment). veh, vehicle. Palmitate-induced increases in PP2A activity (C) and reductions in p-PP2A^Tyr307:PP2A (D) are prevented by myriocin (n = 6–8 per treatment). Palmitate reduced p-eNOS^Ser1177:total eNOS after PP2A immunoprecipitation from wt but not mut HA-tag PP2A (E). ECs transfected with I2PP2A-targeted siRNA displayed 60% less total protein content (F) (n = 6 per treatment), decreased I2PP2A in PP2A immunoprecipitates (G) (n = 6 per treatment), increased PP2A after eNOS immunoprecipitation (H) (n = 6 per treatment), and decreased p-eNOS^Ser1177:total eNOS (I) (n = 6 per treatment) vs. ECs transfected with scrambled (scr) siRNA. ECs treated with FTY720 or palmitate exhibited decreased PP2A activity vs. vehicle-treated cells (J) (n = 6). For C, D, F, and J, each n refers to 1 × 10 cm petri dish. For E and G–I, each n includes 4 × 10 cm petri dishes. Data are mean ± SE. *P < 0.05 vs. vehicle-treated ECs, i.e., (-) pal (-) myr (A–D), (-) pal wt ECs (E), scr ECs (F, G, H, and I), (-) pal (-) FTY720 (J). For D–J, two representative images are shown above the respective histogram. Additional data pertaining to the above are shown in Supplementary Figs. 1 and 2. IB, immunoblot; ins, insulin; IP, immunoprecipitation; S, serine.

Figure 2

PP2A inhibition prevents palmitate-induced increases in PP2A activity and reductions in NO generation without altering ceramide accumulation. Insulin-stimulated increases in p-eNOS^Ser1177:total eNOS are prevented by palmitate (pal) and restored by 4 μmol/L LB1 but not 0.4 or 1 μmol/L LB1 (A) (n = 6 per treatment). Palmitate-induced increases in PP2A activity (B) (n = 10 per treatment) and reductions in p-PP2A^Tyr307:PP2A (C) (n = 6 per treatment) were negated by 4 μmol/L LB1. Insulin-induced increases in p-eNOS^Ser1177:eNOS were suppressed by palmitate treatment but restored by LB1 treatment (D) (n = 6 per treatment). Likewise, palmitate-induced reductions in NO generation assessed via EPR spectroscopy (E) (n = 12 per treatment) or ELISA (F) (n = 8 per treatment) were restored by LB1 (each n is 3 wells of a 6-well plate). Additional data pertaining to the above are shown in Supplementary Figs. 3–5. Palmitate-induced ceramide accumulation was robust regardless of LB1 treatment (G and H). Mean data (G) and a representative image (H) are shown (n = 8). Data are mean ± SE. *P < 0.05 vs. all treatments. ins, insulin; S, serine; Tyr, tyrosine; veh, vehicle.

Figure 3

Palmitate-induced PP2A colocalization with eNOS disrupts the physical interaction among eNOS, Akt, and Hsp90. Palmitate (pal) decreased PP2A that coimmunoprecipitated with I2PP2A (A) and increased PP2A that associated with eNOS (B), but neither effect of pal was altered by PP2A inhibition. Palmitate suppressed insulin (ins)-stimulated p-Akt^Ser473, Hsp90, and p-eNOS^Ser1177 that coimmunoprecipitated with eNOS, and all responses were normalized by LB1 (C–E). For experiments A–E, n = 6–9 and each n includes 4 × 10 cm petri dishes. For A, one representative image is shown above each respective histogram. For B–E, three representative images are shown above each respective histogram. Data are mean ± SE. *P < 0.05 vs. vehicle, i.e., (-) pal (-) ins. Additional data pertaining to the above are shown in Supplementary Fig. 6. IB, immunoblot; IP, immunoprecipitation; S, serine.

Figure 4

PP2A inhibition prevents palmitate-induced disruption of the Akt-Hsp90-eNOS complex in the membrane fraction. Successful fractionation was verified in a subset of ECs exposed to each treatment (A). Compared with vehicle, i.e., (-) pal (-) ins, palmitate (pal) increased PP2A:eNOS at the membrane, a response that was negated by myriocin (myr) but not by LB1 (B). Insulin (ins) increased Akt:eNOS (C), p-Akt^Ser473/T308:eNOS (D and E), Hsp90:eNOS (F), and p-eNOS^Ser1177:eNOS (G). This response was inhibited by palmitate but restored by myriocin and by LB1. For A–G, n = 10; each n represents 10 × 10 cm petri dishes. Data are mean ± SE. *P < 0.05 vs. vehicle-treated cells. Additional data pertaining to the above are shown in Supplementary Fig. 7. CAV-1, caveolin-1; IR, insulin receptor; S, serine; T, threonine.

Figure 5

Lard oil increases PP2A activity; disrupts interactions among Akt, Hsp90, and eNOS in the vasculature; and precipitates arterial dysfunction, each of which is prevented by ceramide synthesis inhibition and PP2A inhibition. Arteries from C57Bl6 mice treated for 3 days with 1.5 mg/kg LB1 exhibit reduced PP2A activity (A) and increased p-PP2A^Tyr307:PP2A (B) vs. arteries from mice treated with vehicle (V). Compared with arteries from des1^+/+ (WT) and des1^+/− (HET) mice pretreated for 3 days with vehicle followed by 6 h of glycerol (G) infusion (collectively referred to as CON), p-PP2A^Tyr307:PP2A was lower in liver from WT mice pretreated with vehicle and subsequently infused for 6 h with lard oil (LO) (C). The lard oil–induced reduction in vascular PP2A^Tyr307:PP2A is attenuated in HET mice pretreated with vehicle and in WT mice pretreated with LB1. Aorta/iliac homogenates were prepared as appropriate to immunoblot for PP2A, Akt, and Hsp90 after eNOS immunoprecipitation (D–F). Compared with arteries from control mice, LO infusion to WT mice pretreated with vehicle increased PP2A association with eNOS (D), while Akt (E) and Hsp90 (F) that coimmunoprecipitated with eNOS were reduced. The LO-induced disruption of the Akt-Hsp90-eNOS complex was prevented in HET mice pretreated with vehicle (implicating ceramide) and in WT mice pretreated with LB1 (implicating PP2A). Vascular function was assessed in femoral artery segments from the same animals. Compared with arteries from control mice, endothelium-dependent vasorelaxation was impaired, and receptor- and non–receptor-mediated vasocontraction was exaggerated in vessels from WT mice pretreated with vehicle and infused for 6 h with LO (G–I). Indices of vascular dysfunction were attenuated in HET mice pretreated with vehicle (implicating ceramide) and in WT mice pretreated with LB1 (implicating PP2A). Because no differences were observed among groups for endothelium-independent vasorelaxation (J), the LO-induced defect observed in WT mice pretreated with vehicle likely was specific to the endothelium. For A–C, n = 2–3 mice per group. For D–F, n = 5–7 mice per group. For G–J, n = 5–7 mice per group, 2–4 vessels per mouse. Data are mean ± SE. *P < 0.05 vs. all. d, days; IB, immunoblot; IP, immunoprecipitation.

Figure 6

Metabolic phenotype of fat-fed mice with or without PP2A inhibition. Arteries from mice treated (intraperitoneally) with 1.0 mg/kg/day LB1 ×21 days vs. 0.5 mg/kg/day exhibited increased p-PP2A^Tyr307:PP2A vs. arteries from mice treated with vehicle (V) (A). Mice that consumed standard (Con) or high-fat (HF) chow for 14 weeks were treated with vehicle or LB1 for the final 14 days. p-PP2A^Tyr307:PP2A was lower in arteries from HF-V mice vs. all other groups (B). Increases in body mass (C), gonadal fat pad mass (D), whole-body fat mass (E), area under the curve (AUC) during the glucose tolerance test (GTT) (F), triglycerides (G), and FFAs (H) and reductions in circulating glucose upon insulin stimulation (I) were similar in HF-fed mice regardless of LB1 treatment. Glucose levels measured at 20 min of the insulin tolerance test (ITT) were lower in Con-fed mice treated with LB1 vs. vehicle. Additional data pertaining to the above are shown in Supplementary Fig. 8. For A and B, n = 3 mice per group. For C–I, n = 8–10 mice per group. Data are mean ± SE. *P < 0.05 vs. all. AU, arbitrary units; d, days; Tyr, tyrosine.

Figure 7

Systemic blood pressure is elevated and vascular Akt-Hsp90-eNOS association is disrupted in obese vs. lean mice, each of which is prevented by PP2A inhibition. Elevations in arterial blood pressure evoked by fat feeding were negated in obese mice treated with LB1 (A). Heart rate (bpm) during conscious blood pressure determinations was not different among groups, i.e., CON-V, 739 ± 13; HF-V, 765 ± 12; CON-LB1, 734 ± 18; HF-LB1, 742 ± 12. p-eNOS^Ser1177:eNOS was lower in arterial homogenates from HF-V mice vs. all other groups (B). Using the same samples, an IP for eNOS was performed, followed by IB for PP2A, Akt, and Hsp90 (C–E). Consistent with reduced p-eNOS^Ser1177:eNOS in arteries from HF-V mice vs. all groups, PP2A was elevated in the eNOS complex (C), while Akt (D) and Hsp90 (E) were reduced. These changes were normalized in obese mice treated with LB1 for the final 14 days of fat feeding, except that PP2A remained in the eNOS complex. n = 6 mice per group. Data are mean ± SE. *P < 0.05 vs. all. DBP, diastolic blood pressure; IB, immunoblot; IP, immunoprecipitation; MAP, mean arterial blood pressure; SBP, systolic blood pressure; S, serine; V, treatment with saline for final 14 days.

Figure 8

Vascular dysfunction exists in obese vs. lean mice, which is prevented by PP2A inhibition. Vascular function was assessed in femoral artery segments from the same mice used in Fig. 7. Endothelium-dependent vasorelaxation was impaired in arteries from HF-V mice vs. all groups (A), while endothelium-independent vasorelaxation (B) was similar among animals. Non–receptor-mediated vasocontraction (C) and receptor-mediated vasocontraction (D) were greater in arteries from HF-V mice vs. all groups but normalized by PP2A inhibition. Impaired responses to acetylcholine (A) and increased responsiveness to phenylephrine (D) observed in arteries from HF-V mice were equalized among groups by prior treatment with N^G-monomethyl-l-arginine salt (L-NMMA) (E and F, respectively). For A–F, n = 8–10 mice per group, 2–4 vessel segments per mouse. Data are mean ± SE. *P < 0.05 vs. Con-fed mice. A schematic summary of our findings is presented in G. Palmitate, lard oil infusion, and fat feeding elevate cellular ceramide biosynthesis. Ceramide binds I2PP2A, the restraint that I2PP2A confers upon PP2A is inhibited, and PP2A that translocates to the membrane disrupts interactions among Akt, Hsp90, and eNOS to an extent that NO generation and vascular function are impaired. All effects can be prevented by pharmacological (myriocin) and genetic (Sptlc1 siRNA; des1^+/− mice) manipulations that inhibit ceramide biosynthesis, and all effects (except for PP2A translocation to the eNOS complex) can be prevented by mutation of the catalytic site on PP2A and by PP2A inhibition using LB1. These results indicate that PP2A activation contributes to vascular dysfunction in vivo. P, phosphorylation; S, serine; T, threonine; V, treatment with saline for final 14 days.