• Open Access

Age-related changes in endothelial nitric oxide synthase phosphorylation and nitric oxide dependent vasodilation: evidence for a novel mechanism involving sphingomyelinase and ceramide-activated phosphatase 2A

Authors



Tory M. Hagen, PhD, Linus Pauling Institute/Oregon State University, 571 Weniger Hall, Corvallis, OR 97331, USA. Tel.: 541 7375083; fax: 541 7375077; e-mail: tory.hagen@orst.edu

Summary

Aging is the single most important risk factor for cardiovascular diseases (CVD), which are the leading cause of morbidity and mortality in the elderly. The underlying etiologies that elevate CVD risk are unknown, but increased vessel rigidity appears to be a major hallmark of cardiovascular aging. We hypothesized that post-translational signaling pathways become disrupted with age and adversely affect endothelial nitric oxide synthase (eNOS) activity and endothelial-derived nitric oxide (NO) production. Using arterial vessels and isolated endothelia from old (33-month) vs. young (3-month) F344XBrN rats, we show a loss of vasomotor function with age that is attributable to a decline in eNOS activity and NO bioavailability. An altered eNOS phosphorylation pattern consistent with its inactivation was observed: phosphorylation at the inhibitory threonine 494 site increased while phosphorylation at the activating serine 1176 site declined by 50%. Loss of phosphorylation on serine 1176 was related to higher ceramide-activated protein phosphatase 2 A activity, which was driven by a 125% increase in ceramide in aged endothelia. Elevated ceramide levels were attributable to chronic activation of neutral sphingomyelinases without a concomitant increase in ceramidase activity. This imbalance may stem from an observed 33% decline in endothelial glutathione (GSH) levels, a loss known to differentially induce neutral sphingomyelinases. Pretreating aged vessel rings with the neutral sphingomyelinase inhibitor, GW4869, significantly reversed the age-dependent loss of vasomotor function. Taken together, these results suggest a novel mechanism that at least partly explains the persistent loss of eNOS activity and endothelial-derived NO availability in aging conduit arteries.

Introduction

Cardiovascular diseases (CVD) and related complications are the leading cause of hospitalization and death in Western societies, affecting nearly 80% of persons over the age of 65 (Anderson, 2002). Age is the leading risk factor for most cardiovascular pathologies (Lakatta, 2003). While the mechanisms underlying this risk are largely unknown, deficits in vascular endothelial cell function appear to be a critical factor, leading to impaired vasomotor activity, hypertension and progression of nearly all types of CVD (Lakatta & Levy, 2003; Lakatta, 2003). This age-dependent impairment of endothelium-dependent vasorelaxation is not unique to humans having also been observed in numerous animal models (Hongo et al., 1988; Hatake et al., 1990; Tschudi et al., 1996).

In the large conduit arteries such as the aorta, the endothelium governs vasoresponsiveness by synthesizing and releasing factors that act on the vascular smooth muscle layer. Perivascular nerves also control vasomotor tone by stimulation of the endothelium to produce vasoactive compounds. In the aorta, the principal vasodilatory agent is nitric oxide (NO) (Ignarro et al., 1987). As age-related decline of vascular compliance is particularly prominent in major conduit arteries, impaired bioactivity of NO may be an important contributing factor to this decline. Indeed, lower NO availability presents an attractive explanation for diminished vasomotor function as NO is subject to oxidative inactivation, which would be enhanced by an increasingly pro-oxidant milieu in the aging vasculature (Beckman & Koppenol, 1996).

An additional yet underexplored route leading to lower NO bioactivity would be a decline in NO synthetic capacity with age. NO is produced by endothelial nitric oxide synthase (eNOS), a homodimeric hemoprotein that produces NO by the 5 electron oxidation of the guanidinium group of arginine (Alderton et al., 2001). eNOS activity, and hence NO production, is regulated in a complex and multilayered process. This includes regulation at the transcriptional and post-translational levels, availability of cosubstrates/cofactors, subcellular location, and protein–protein interactions (Alderton et al., 2001; Fulton et al., 2004). Thus, eNOS in the aging vasculature may be vulnerable to dysregulation at multiple points that would limit NO synthesis and consequently, NO availability. A preponderance of the evidence shows that both cosubstrate/cofactor availability and eNOS message and protein levels change little in the aging endothelium (Chou et al., 1998; Huang, 2003). Despite this, there is equal consensus that eNOS activity declines significantly with age (van der Loo et al., 2000; Strosznajder et al., 2004), which suggests that post-translational alterations to the enzyme may reconcile these two seemingly disparate observations.

To date, there has been virtually no work examining post-translational dysregulation of eNOS or its effects on NO availability. This is surprising given that eNOS is under significant regulatory control. The enzyme is subject to acylation, nitrosylation, and phosphorylation that, depending on the modification or site, either positively or negatively regulates NO synthetic capacity (Alderton et al., 2001). Of these, phosphorylation/dephosphorylation of the threonine 494 (T494) and serine 1176 (S1176) residues in the rat sequence of the enzyme appear to principally govern eNOS activity. Phosphorylation of T494 prevents calmodulin binding and electron flow from the reductase to the oxidase domains of the homodimer (Alderton et al., 2001). In contrast, a variety of stimuli initiate the Akt-driven phosphorylation of S1176, which markedly enhances eNOS activity by stabilizing its association with calmodulin and also maintaining eNOS activity even after intracellular Ca+2 is re-sequestered (Alderton et al., 2001). Removal of the phosphate group from S1176 is catalyzed by the ceramide-activated protein phosphatase 2 A (PP2A), which returns eNOS to a Ca+2 sensitive state. Thus, sustained endothelial NO production is controlled by the relative activities of several kinases and phosphatases and loss of this coordination would markedly affect eNOS activity.

The hypothesis that dysregulation of steady-state eNOS phosphorylation contributes to lower vascular tone with age is compatible with the observed loss in eNOS activity without a concomitant decline in its expression. Thus, the goals of the present study were to determine the extent of age-related alterations in eNOS activity relative to its phosphorylation status. Our results show a significant and persistently altered eNOS phosphorylation state occurs in aging rat aortic endothelia that is consistent with its inactivation. Moreover, we show that age-related changes in eNOS phosphorylation stem from chronically elevated sphingomyelinases (SMases) and resultant ceramide-driven PP2A activity.

Results

Vasomotor function and eNOS activity

To determine the extent of age-related changes in vessel reactivity, aortas from young and old rats were isolated and endothelium-dependent vasomotor function was determined. Acetylcholine-induced relaxation of aortas from old rats was significantly reduced compared with those from young animals (Fig. 1a), confirming previous findings by van der Loo et al. (2000). As a control, endothelial denudation led to complete loss of acetylcholine-induced vasorelaxation in both young and aged rats (data not shown). Conversely, addition of the NO donor, nitroprusside, to intact vessels mediated a dose-dependent relaxation (Fig. 1b), regardless of age. This suggests that NO-independent vasoresponsiveness remains intact in old rats and impaired relaxation is due to diminished production or bioactivity of NO.

Figure 1.

Age-associated decline of endothelium-dependent vasodilation is due to diminished NO production, not decline of eNOS protein. (a) Aortic rings from old rats exhibited a significantly lower response to acetylcholine-induced vasorelaxation, as compared with young animals (n = 3). (b) Addition of the NO analog, sodium nitroprusside, to the medium induced a marked vasorelaxation indicating maintenance of proper smooth muscle cell function with age (n = 3). (c) eNOS protein levels are unchanged with age in the aortic endothelium (n = 4). (d) Cyclic GMP is the second messenger signal induced by NO and a reliable marker for NO bioactivity. In aged aorta, resting levels of cGMP decline by approximately 60% compared with the levels seen in the young aorta (n = 6). (e) Maximal NO synthetic capacity of endothelial cells was determined by stimulation with the calcium ionophore A23187. The NOS inhibitor L-NAME was used to show specificity of eNOS activity. Significantly decreased maximal eNOS activity is characteristic endothelium from old animals, demonstrating a loss of NO synthesis. (n = 4). An asterisk (*) denotes statistical significance (P ≤ 0.05).

A potential explanation for the diminished age-dependent vasomotor response is a decline in eNOS protein levels. However, Western blot analysis showed no significant age-associated differences in eNOS protein content (Fig. 1c). Despite this, endothelium-derived NO bioactivity was lost with age as indicated by a significant reduction in whole-vessel cGMP content (P ≤ 0.05; Fig. 1d). The lower NO bioavailability appeared to be from a general decline in eNOS activity, as measurement of A23187-mediated NO production declined significantly in cells from old vs. young rats (Fig. 1e). Overall, these data show that the age-related loss in endothelium-derived NO bioactivity is partly attributable to lower eNOS activity, but this deficit cannot be from a lower enzyme content in the endothelia from aged animals.

Age-associated phosphorylation patterns of eNOS and Akt shift towards an inhibitory motif by chronic activation of PP2A

To determine whether phosphorylation status of eNOS was affecting enzyme activity, endothelia from young and old rats were isolated and subjected to Western blot analysis using antibodies to specific phosphorylation sites on the enzyme. Results revealed that eNOS from aortic endothelial membranes exhibited significantly less phosphorylation at the stimulatory S1176 site but a sustained phosphorylation at the inhibitory T494 site with age (Fig. 2a,b). A comparison of the S1176:T494 phosphorylation ratio, an indicator of overall eNOS activation and NO synthetic capacity, declined by 71% in an age-dependent manner.

Figure 2.

Aging leads to inactivation of eNOS and Akt in the endothelium. The phosphorylation status of eNOS in membrane fractions of young and old rat aortic endothelium was analyzed by Western blotting. (a) Phosphorylation of eNOS at S1176, which up-regulates eNOS activity is significantly decreased by age (n = 4). (b) Conversely, phosphorylation of T494, which prevents NO synthesis is significantly increased (n = 4). The S1176:T494 ratio, a measure of eNOS synthetic capacity, significantly (P ≤ 0.02) declines by 71% in the aged endothelium (data not shown). (c) Akt activity was impaired with a significant decline of 27% as assessed by the ability of Akt to phosphorylate a synthetic GSK-3β substrate (n = 4). (d) Similarly, S473 phosphorylation, which is required for Akt activity, declined significantly by 37% in endothelial cells of aged animals (n = 6). An asterisk (*) denotes statistical significance (P ≤ 0.01) (**P ≤ 0.02).

As Akt is the upstream kinase that directly phosphorylates and activates eNOS at S1176 (Fleming & Busse, 2003), we hypothesized that there is an age-associated decline in Akt activity. Indeed, Akt activity declined by 27% with age in the vascular endothelium (Fig. 2c). The causes for this loss appear to be a lower phosphorylation state of Akt: phospho-specific antibodies and Western blot analysis showed constitutively diminished phosphorylation at the stimulatory serine 473 (S473) residue with age (Fig. 2d).

Because PP2A dephosphorylates both S1176 of eNOS and S473 of Akt (Shanley et al., 2001), we hypothesized that an age-associated change in PP2A activity was responsible for the decline in S1176 phosphorylation status of eNOS. As there is no direct assay specifically for PP2A, general serine/threonine phosphatase activities were measured in the absence or presence of increasing concentrations of okadaic acid (OA). OA specifically inhibits PP2A at very low concentrations (IC50: 0.5 × 10−9 m) but only inhibits activities of other phosphatases at relatively high levels of the inhibitor (IC50: ≥ 1.0 × 10−8 m). In the absence of OA, general phosphatase activities increased 2.3-fold with age (Fig. 3a). Only very low OA levels (≤ 10−10 m) were needed to completely attenuate this age-associated elevation, indicating that PP2A accounted for most, if not all, of the heightened general phosphatase activity found in aged endothelia (Fig. 3b).

Figure 3.

PP2A phosphatase activity is significantly elevated in the aged endothelium. General serine/threonine phosphatase activity was estimated in freshly isolated, unstimulated endothelium. (a) With age, general phosphatase activity significantly increases by a factor of 2.3 compared with young controls (n = 4). (b) Increasing concentrations of okadaic acid were added to endothelium from young and old rats. Results show that the age-associated difference in activity is abolished at 10−10 M suggesting that the age-associated difference in phosphatase activity is due to PP2A. An asterisk (*) denotes statistical significance (P ≤ 0.01).

Ceramides accumulate in the aging endothelia because of increased neutral sphingomyelinase activities

Because PP2A is a ceramide-activated phosphatase, we hypothesized that stimulated PP2A activity was mediated in a ceramide-dependent manner. Analysis of ceramide levels in lipid extracts from vascular endothelium showed a doubling with age (Fig. 4a).

Figure 4.

The role of ceramide and sphingomyelinase dysregulation in the aged endothelium. (a) Free ceramide levels were measured in endothelium from young and old animals. In the aged endothelium, ceramides were more than double the level seen in young animals (n = 4). (b) Sphingomyelinase (SMase) activity reported in Table 1 is graphically represented here in order to illustrate the significant elevation observed in the aged endothelium. Cellular ceramides are generated in part by the action of sphingomyelinases on membrane sphingolipids. Enzymatic activity of both the neutral and acidic isoforms of SMase was measured in endothelium isolated from young and old animals. For the neutral isoform, enzymatic activity in the old endothelium was nearly triple the activity seen in the young (n = 7). Although the age-related differences were not as great, a significant increase in acidic -SMase was also noted (n = 4). Asterisks (* and **) denote statistical significance (P ≤ 0.01 and P ≤ 0.05, respectively).

As steady-state ceramide concentrations are governed by both its release from sphingomyelin and its catabolism to other sphingolipid metabolites, experiments were undertaken to discern whether heightened ceramide levels could account for the observed age-related increase in PP2A activity. Both activities of sphingomyelinases (SMase) and ceramidases (CMase) with neutral or acidic pH optima were measured in endothelial cell membranes from young and old rats, as these enzymes regulate the overall ceramide balance. Neutral sphingomyelinase (nSMase) activity nearly tripled with age, while acidic sphingomyelinase increased only modestly (Table 1, Fig. 4b). Conversely, no significant changes in ceramidase activities were detected (Table 1). These results indicate sphingolipid metabolism in aging endothelia is tipped towards sphingomyelin breakdown and ceramide accumulation.

Table 1.  Endothelial sphingomyelinase and ceramidase activities from the young and old rat aorta
 YoungOld
  1. Data expressed as means ± SD; n = 4; *P ≤ 0.05 by Student's t-test.

Enzyme activity: (pmol mg−1 protein h−1)
 Neutral sphingomyelinase0.068 ± 0.0090.203 ± 0.011*
 Acidic sphingomyelinase0.088 ± 0.0030.102 ± 0.01*
 Neutral ceramidase0.780 ± 0.0210.847 ± 0.043
 Acidic Ceramidase0.898 ± 0.1510.903 ± 0.100

Because both proinflammatory cytokines (e.g. TNFα) and loss of glutathione (GSH) status are known to activate nSMase activity relative to ceramidases, we measured circulating TNFα levels, aortic cell TNFα expression, and any age-related changes in GSH status and its redox state in young and old rats. As shown in Fig. 5(a), circulating TNFα levels were at or below (≤ 3 pg mL−1) the detection limit of the assay in plasma from young rats but increased significantly (P = 0.01) to 51.4 ± 3.2 pg mL−1 in plasma of old animals. Measurement of aortic wall expression of TNFα, as quantified by reverse transcriptase–polymerase chain reaction (RT–PCR), demonstrated a significant age-associated increase in TNFα mRNA relative to β-Actin (P ≤ 0.01, Fig. 5b). Additional studies revealed that endothelial GSH concentrations were also over 30% lower on an age basis (Fig. 5c). The reduced to oxidized glutathione (GSH:GSSG) ratio, a general assessment of cellular thiol redox balance, also fell by 36% in old vs. young rat endothelium (41.1 ± 6.4 vs. 26.1 ± 2.6, respectively; n = 4; P ≤ 0.05, Fig. 5c), indicating a pro-oxidant state exists in endothelial cells with age. Together with elevated TNFα, these results suggest that conditions in the aging vessel are amenable to chronic induction of nSMase activities and resultant ceramide-driven loss of eNOS phosphorylation at its stimulatory site.

Figure 5.

Age results in exposure of the endothelium to inflammatory and oxidative stress. (a) In order to confirm the exposure of the endothelium to inflammatory cytokines, plasma levels of TNF-α were assayed. TNF-α was not detectable (< 2–3 pg mL−1; lower limit of detection for the assay) in the young rats; however, average TNF-α concentrations in the plasma of aged animals were 51.4 ± 3.19 pg mL−1 (n = 6). (b) Comparative RT–PCR from mRNA collected from whole aortic homogenates confirms that cells of old vessels produce significantly more TNFα message (n = 4). (c) Reduced GSH levels were measured in young and old endothelium. Age results in a 31% decline in endothelial GSH (n = 4). The decline in GSH also results in a significant decrease in the thiol redox ratio (GSH:GSSH). An asterisk (*) denotes statistical significance (P ≤ 0.01) compared to young.

Neutral-SMase inhibition significantly improves whole vessel vasocompliance

A summary of results suggests that a persistent induction of nSMase activity leads to ceramide-activated PP2A, which in turn limits eNOS activity and NO availability. To validate this hypothesis, we determined whether inhibition of nSMase activity would improve overall vascular function in whole vessels. Isolated contracted rings were pre-incubated with the specific nSMase inhibitor, GW4869, then dose-dependently relaxed with acetylcholine. Results showed that GW4869 pre-incubation significantly, but not fully, restored endothelium-dependent vasodilation of the aged vessels (Fig. 6). In contrast, inhibition of nSMase had little effect on relaxation characteristics in vessels from young animals (Fig. 6). These results indicate that the NO-dependent decline in vasomotor function is at least partly due to nSMase-induced ceramide accumulation and subsequently lower eNOS phosphorylation and activity.

Figure 6.

Pharmacologic inhibition of nSMase improves vasorelaxation. The specific neutral sphingomyelinase (nSMase) inhibitor GW4869 was pre-incubated with vessel rings from young and old animals for 1 h prior to myography (n = 3). Inhibition of nSMase partly restored the age-associated decline of acetylcholine (ACh)-dependent vasorelaxation in the old vessel rings while it had no effect on vasomotor action of the young vessels (solid lines). ACh-dependent dilation of matched rings from untreated vessels is shown for comparison (dotted lines). An asterisk (*) denotes statistical significance (P ≤ 0.01) compared to old untreated.

Discussion

The observation that eNOS phosphorylation state is significantly altered with age represents the major finding of this study and provides new insights into the age-related loss in endothelium-derived NO bioactivity. Because the extent of eNOS phosphorylation markedly regulates its activity, our data suggest that the significant decline in S1176:T494 ratio may play a major role in reducing NO synthetic capacity in the aging endothelium. Previous studies have shown that a high state of T494 phosphorylation inhibits Ca+2/calmodulin binding to eNOS, rendering the enzyme refractory to agonist-induced stimuli. Additionally, when S1176 is phosphorylated, NO synthesis is induced by increased electron flux from the oxidase to reductase domains. High S1176 phosphorylation also maintains eNOS activity at a high level even following an agonist-induced Ca+2 transient. Thus, the significant decline in S1176:T494 phosphorylation ratio indicates that eNOS in the aging endothelium is both constitutively less active and less inducible by agonist stimulation or shear stress. This scenario fits with our current data and previous work of van der Loo and coworkers (2000) showing that eNOS activity is less stimulated by A23187 in endothelial cells taken from old vs. young rats.

Our present findings suggest a plausible mechanism, which may lead to lower eNOS phosphorylation and loss of NO synthetic capacity. We hypothesize that chronic nSMase activation induces PP2A activity in a ceramide-dependent manner, which ultimately dysregulates eNOS phosphorylation. PP2A is a serine/threonine phosphatase that is markedly stimulated when ceramide levels are high by aiding PP2A subunit recruitment and assembly. PP2A directly dephosphorylates eNOS at S1176 but not T494. This correlates with the age-dependent differences in the eNOS phosphorylation pattern observed in the present study when it is considered that T494 phosphorylation is constitutive in the absence of stimulatory signals. The concept that PP2A is directly responsible for eNOS S1176 phosphorylation in aged animals is also supported by a similar age-related loss in Akt phosphorylation at the T473 residue.

It is worth noting that only nSMase activity chronically increased with age, and without a concomitant elevation in ceramidase activities. This enzyme normally works in concert to govern a so-called sphingolipid ‘rheostat’, represented by the ceramide:sphingosine 1-phosphate ratio. The ceramide:sphingosine 1-phosphate ratio directly affects cell proliferative state and cell survival pathways via activation or inactivation of specific sets of cellular transduction networks (Alewijnse et al., 2004). Typically, this ratio is homeostatic; transient perturbations of the ratio elicit changes in membrane structure and function, yet the ratio quickly returns to its set-point. Our studies therefore suggest that this rheostat becomes imbalanced in the aging endothelium, leading to elevated ceramides and ultimately the altered eNOS phosphorylation pattern seen. Thus, we propose that chronically altered sphingolipid-dependent signaling leads to eNOS inactivation and lower NO synthetic capacity with age.

Our hypothesis suggesting that ceramides act as a negative effector of eNOS activity via inducing PP2A is seemingly inconsistent with reports showing ceramide-dependent eNOS activation. Work by Czarny and coworkers (2003) and by Mogami et al. (2005) reported shear stress-mediated mechanotransduction and nSMase-dependent eNOS-dependent vasodilation, in conjunction with transiently increased ceramide levels. Czarny's work also demonstrated that inhibition of sphingomyelinase activity prevented shear stress-mediated activation of eNOS. However, a subtle yet important difference lies between these results and our observations. Mechanotransduction leads to rapid and transient increases in membrane ceramides followed by their conversion to sphingolipid species that actually induce vasorelaxation (e.g. sphingosine-1 phosphate). A critical examination of Mogami's work illustrates rather significant differences in the model and methods employed by Mogami et al. compared with ours. That study involved exposure of the luminal surface of bovine coronary artery rings to recombinant (bacterial) neutral SMase and measured enhanced endothelium-dependent relaxation (to bradykinin) compared to untreated rings. This is not a comparable system with ours. Functional SMase in the endothelium resides on the inner leaflet of the plasma membrane, associated (along with eNOS, Akt, etc.) with the caveolae. Thus, the location of SMase-mediated ceramide production in their system does not mimic the endogenous one seen in ours. Furthermore, the transient exposure to exogenous nSMase would theoretically lead, once more, to a short burst of ceramide production in the membrane which would be rapidly converted to sphingosine-1-phosphate – a potent vasodilator. Our results indicate that ceramide chronically accumulates with age. Taken together with the finding of increased SMase without compensatory adjustment of ceramidase activity, this suggests a decreased likelihood for conversion to the vasodilatory compound, sphingosine-1-phosphate. The most current theory regarding the mechanism of ceramide-mediated cellular signal transduction partly centers around its effect on membrane fluid dynamics and the liquid-ordered domains of caveolae (Smart et al., 1999). Thus, the specific cellular locale, the time course of formation, the species of ceramides formed and the secondary metabolites formed from ceramide ultimately determine the cellular effects seen. In our study, we eliminated the possibility of supra-physiological artifacts by simply quantifying the steady-state levels of ceramides in the endothelial membrane and measurement of endogenous SMase and ceramidase activities. No age-related changes in ceramidases were noted, which indicates that sphingosine 1-phosphate levels were not elevated concomitantly with ceramides. Thus, conditions exist in the aging endothelium that would lead to a ceramide-dependent decline in eNOS activity while transient increases in cellular ceramide may actually stimulate eNOS.

The mechanism(s) involved in heightened nSMase activity in the aging endothelium is not presently known but under study in our laboratory. It is important to note that nSMase activity can be significantly induced when either cellular GSH status declines or when proinflammatory cytokine receptors become activated. Using MCF7 cells, Liu and coworkers (1998) showed that nSMase activities were markedly increased when GSH levels were diminished. Induced nSMase activity was directly attributable to GSH-mediated changes in the Vmax of the enzyme. Additionally, it is known that a protein, Factor Associated with nSMase (FAN), interacts with the TNFα receptor (TNFR1) upon ligand binding and positively regulates nSMase activity (Segui et al., 2001). As we report both marked age-related declines in endothelial GSH levels but significant increases in circulating TNFα and its endothelial expression, it is enticing to speculate that lower GSH status in this heightened inflammatory environment of the aging vessel may dysregulate normal eNOS phosphorylation and NO bioactivity. This scenario correlates well with the results of Kim et al. (2001) who demonstrated that TNFα prevented flow- and insulin-mediated endothelial Akt and eNOS phosphorylation and concomitantly, NO release. Significantly more work will be necessary to directly link the age-related increases in endothelial proinflammatory and pro-oxidant state with changes in eNOS phosphorylation.

The results presented here must be considered in the context of other routes that could also result in lost vasocompliance in the elderly. It is known that oxidative modification of NO to peroxynitrite increases in the aged cardiovasculature (van der Loo et al., 2000; Drew & Leeuwenburgh, 2002), which would also limit NO bioactivity. Our results showing an overall age-related loss in low molecular-weight antioxidants and a decline in GSH:GSSG ratio would be conducive to peroxynitrite formation (Maytin et al., 1999). However, peroxynitrite formation cannot be the sole means for impaired NO availability as this study and others clearly show that eNOS activity markedly declines with age. However, these other mechanisms do not apparently extend to changes in transcriptional regulation of the enzyme (Chou et al., 1998) or loss in cofactor and substrate availability (Hong et al., 2001; Katusic, 2001; Higashi et al., 2002). Thus, age-related changes in post-translational modification of eNOS represent a plausible means leading to the loss of NO synthesis and to lower NO bioavailability in aged vessels.

It is also well known that during aging, mechanical damage, apoptosis and necrosis contribute to loss of endothelial cells (Chang & Harley, 1995; Barton et al., 1997; Hoffmann et al., 2001). In order to maintain proper barrier and vasomotor function, regenerated endothelial cells grow to take the place of those lost. The physiological characteristics of these regenerated endothelial cells are significantly altered compared with ‘virgin’ endothelium. It has been reported that these regenerated endothelial cells suffer altered membrane transport, increased sensitivity to apoptosis, shortened telomeres, altered expression of critical endothelial enzymes and mitochondrial dysfunction (Chang & Harley, 1995; Barton et al., 1997; van der Loo et al., 2000; Hoffmann et al., 2001; Deshpande et al., 2003).

Additionally, age-associated changes in other post-translational regulatory pathways could also adversely affect eNOS activity. These pathways, with respect to the aging vessel, remain largely unexplored. Theoretically, aging may also influence eNOS palmitoylation, its association with positive (HSP90) and/or negative (caveolin-1) accessory proteins, and its subcellular location. We currently have evidence indicating that eNOS subcellular locale and interaction with effector proteins are also altered with age in a pattern consistent with an overall loss in enzyme activity (data not shown). While this evidence must be considered preliminary, it is interesting to note that the patterns we see in the aging vascular endothelium are also evident in chronic inflammatory diseases.

As the altered vasoreactivity observed in rodent models of aging approximate that of humans (Hongo et al., 1988; Hatake et al., 1990; Tschudi et al., 1996), the results currently presented potentially affect a large and growing segment of the population in Western societies – the elderly. The clinical relevance of the present work is heightened by our data showing that pharmacologic inhibition of nSMase with GW4869 significantly improves endothelium-dependent vasodilation in the aged aorta. Even though vasocompliance was not fully reverted to that of young animals, a significant improvement was demonstrated. This suggests that the novel mechanisms of eNOS inactivation reported in this paper might be pharmacologically addressed in clinical settings in the future. Further relevance is reinforced by accumulating data on the prognostic value of endothelium-dependent vasomotion toward overall health in the elderly (Landmesser et al., 2004). This may be especially important in elderly people, where development of atherosclerosis (which is exacerbated by decreased NO production) (Chen et al., 2001) strongly contributes to both morbidity and, in particular, cardiovascular mortality (Lakatta & Levy, 2003). In conclusion, the newly discovered mechanisms that underlie endothelial dysfunction and impaired vasoreactivity in the elderly build a new rationale for the development of pharmacological tools aimed at the restoration of proper eNOS signal transduction pathways.

Experimental procedures

The model animal

Throughout this study, male Fischer 344 × Brown Norway hybrid (F344 × BrN) rats were used as the experimental model. This is a well-characterized rat strain that is an approved rodent model for aging studies by the National Institute on Aging (NIA) of the National Institutes of Health (NIH). Young (2–4 months) and old (32–34 months) rats were used for all studies. Although F344 × BrN rats do not develop atherosclerosis, they do exhibit the same age-related decline in vasocompliance as humans and other mammalian species do (Hynes & Duckles, 1987).

Preparation of vascular endothelium samples

Freshly isolated aortae from male F344 × BrN rats were perfused with Hank's buffered saline solution (HBSS), pH 7.4. For some experiments, protease and phosphatase inhibitors (proprietary inhibitor cocktail; Sigma, St. Louis, MO, USA) were included. Endothelium was isolated by freezing and scraping of the aorta, an adaptation of a method described by Ryan & Maxwell (1986). Briefly, the washed and cleaned aortae were opened longitudinally and adhered to poly-l-lysine-coated glass, then rapidly frozen over liquid nitrogen. After freezing, the endothelial surface was carefully scraped from the luminal surface of the vessel segments with a surgical scalpel and collected into homogenization buffer. Immunostaining was employed to determine that the endothelial isolate was not significantly contaminated with smooth muscle cells (Simmons et al., 2004). Based on positive expression of factor VIII antigen (von Willebrand's factor) and negligible presence of α-smooth muscle actin (α-SMA), the isolated fraction was comprised of relatively pure aortic endothelium (data not shown).

Aortic ring myography

Segments of thoracic aorta were cleaned of adherent connective tissue, cut into 3- to 5-mm-long rings and suspended in an organ-bath chamber containing Krebs-Henselheit solution, pH 7.2, gassed with 95% O2 and 5% CO2, and maintained at 37 °C. Tissues were mounted on an isometric force-displacement transducer (Kent Scientific, Torrington, CT, USA) and changes in isometric forces were continuously recorded. Rings were gradually stretched to 1–1.5 g and allowed to equilibrate for 90 min. Maximal contractility was evaluated by the addition of KCl 60 mm. After washing and further equilibration, the rings were contracted with 3 × 10−7 m norepinephrine. After stabilization (10–15 min), relaxation was assessed by the cumulative addition of acetylcholine (10−10−10−4 m). Sodium nitroprusside (10−10−10−5 m) was used to evaluate endothelium-independent vasorelaxation. In some studies, the specificity of ACh-dependent stimulation of endothelial NO was determined by rubbing the vessel ring to denude the endothelium prior to myography.

Cyclic GMP measurement

The accumulation of cGMP was evaluated in homogenates of whole vessels. Briefly, aortas were perfused, excised and cleaned of adherent adventitial tissue and approximately 0.5 g (wet weight) of the upper aorta was homogenized at 4 °C in HBSS with a stainless steel tipped motorized tissue homogenizer. Ice-cold trichloroacetic acid (TCA; 6%, final concentration) was added to the homogenate, the acid extract was centrifuged at 12 000 g for 10 min and TCA was removed using three extractions with water-saturated ether. The cGMP content of the aqueous phase was determined with a commercially available enzyme-linked immunoassay kit (Cayman Chemical, Ann Arbor, MI, USA).

Immunochemical analysis of endothelial proteins

Vascular endothelium was isolated from young and old rats as described above. Protein homogenates were quantified by a modification of Lowry's method (Peterson, 1977). Samples were separated by SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) and analyzed by Western blot. The immunodetected proteins (anti-eNOS and anti-Akt, Cell Signaling Technology, Beverly, MA, USA; antiβ-actin, Sigma, St. Louis, MO, USA) were quantified by horseradish peroxidase-linked secondary antibodies and subsequent chemiluminescent detection.

Glutathione measurement

Endothelial preparations were lysed in 10% (w/v, final concentration) perchloric acid. The acid soluble fractions containing GSH and glutathione disulfide (GSSG) were derivatized with iodoacetic acid (40 mm). The resulting carboxymethyl derivatives were further derviatized with dansyl chloride (75 mm) and separated using the method of Jones et al. (1998). GSH and GSSG were separated by HPLC (high-performance liquid chromatography) using fluorescent detection [Hitachi L7000 (Hitachi, San Jose, CA, USA); Ex/Em: 330/515 nm]. Quantification was achieved relative to GSH and GSSG standards. γ-glutamyl-glutamate was used as an internal control to assess completeness of derivatization.

Phosphatase activity

Total serine/threonine phosphatase activity in endothelial extracts was estimated using the IQ assay (Pierce, Rockford, IL, USA). Briefly, cellular protein extracts were incubated with a proprietary synthetic phosphopeptide (LRRApSLG), which functions as a substrate for endogenous serine/threonine phosphatases. A developing agent confers fluorescent activity upon nonphosphorylated peptide. Thus, fluorescence intensity of the sample, postdevelopment, is directly correlated with phosphatase activity over the time course of the assay. To ensure specificity and accuracy, control experiments were run in parallel with the experimental samples including: omission of endothelial protein, omission of the peptide substrate, addition of a phosphatase inhibitor cocktail (Sigma, P2850), 100% phosphorylated peptide substrate and 0% phosphorylated peptide substrate (Pierce).

In order to estimate the relative contribution of the phosphatase PP2A to overall phosphatase activity, phosphatase activity was measured in young and old endothelial samples in the presence of increasing concentrations of the phosphatase inhibitor okadaic acid (Sigma). The concentration of okadaic acid required to inhibit PP1 and other phosphatases is over 100 times that required to inhibit PP2A.

Ceramide measurement

Free ceramides in the endothelium were estimated using a modification of the diacylglycerol (DG) kinase assay (Bielawska et al., 2001). Briefly, lipids were extracted from a known amount of endothelial protein and dried under N2. The dried lipids were solubilized by bath sonication in a detergent solution (7.5% (w/v) n-octyl-β-d-glucopyranoside and 5 mm 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)]) and incubated with 5 U recombinant bacterial DG kinase and 4 µCi [γ-32P]ATP for 2 h at 25 °C. The reaction was quenched by addition of ice-cold methanol and the lipids were extracted and dried under N2. The resultant [γ-32P]-labelled phospholipids were separated by thin layer chromatography. [γ-32P]-labelled 1-phospho-ceramide bands were visualized by autoradiography and scraped from the TLC plates and quantified by scintillation counting. Since the DG kinase reaction is an entirely in vitro method, synthetic C6-ceramide was included in all reactions as an internal standard to control for completeness of the reaction. Endogenous ceramide levels were quantified according to external standards which consisted of using of synthetic C16- and C18-ceramide in DG kinase reactions in place of sample lipids.

Sphingomyelinase assay

Endogenous neutral and acidic sphingomyelinase and ceramidase activities were estimated by incubation of endothelial membrane fractions with fluorescent nitrobenzofuran (NBD)-derivatized substrates (NBD-sphingomyelin, NBD-ceramide; Molecular Probes, Eugene, OR, USA) in vitro. Briefly, endothelial membrane fractions are incubated with either NBD-sphingomyelin or NBD-ceramide in PBS pH 7.4 containing 5 mm MgCl for 1 h at 37 °C. The reaction was terminated by addition of ethanol and the resulting solution separated by HPLC with fluorescence detection (455/530 nm; Ex/Em). Liberated NBD-ceramide (from sphingomyelinase activity) and NBD-fatty acid (from ceramidase activity) were quantified according to NBD-ceramide NBD-fatty acid standards. In order to confirm specificity of the assay, an inhibitor of neutral SMase (GW4869, 50'M) and acidic SMase (desipramine, 50'M) were used as negative controls.

TNF-α measurement

TNF-α in rat plasma was determined with an ultrasensitive ELISA kit (Biosource, Camarillo, CA, USA). RNA was isolated from vascular or endothelial preparations using RNA mini-prep molecular biology kits (QIAGEN, Valencia, CA, USA). mRNA levels of TNF-α were measured by RT–PCR using Sybr-green® detection. Quantitative analysis of resting mRNA levels was made by comparing target gene mRNA expression with β-actin, which was unchanged by age.

Statistical methods

Experimental data are presented as the mean ± standard deviation of the mean. When applicable (comparison of young vs. old), experimental means were compared using the Student's t-test. For multiple comparisons (myographical data), analysis of variance with Bonferroni's post hoc analysis was used. Differences were considered significant if P-values were less than 0.05.

Acknowledgments

This research was supported by grants RO1 AG17141A and P01 AT002034-01 from the NIH (TMH) and in part by an American Heart Association predoctoral fellowship (0110213Z; ARS). We thank Jeffrey Monette for his expertise in development and adaptation of ceramide quantitation and the sphingomyelinase/ceramidase assay.

Ancillary