Regulation by cAMP of post-translational processing and subcellular targeting of endothelial nitric-oxide synthase (type 3) in cardiac myocytes.

Cardiac myocytes express the nitric-oxide synthase isoform originally identified in endothelial cells, termed eNOS or NOS3, where it plays a role in regulating myocyte responsiveness to both adrenergic and muscarinic cholinergic autonomic nervous system agonists. eNOS in endothelial cells has been shown to undergo extensive post-translational processing, and in cardiac myocytes as well as endothelial cells, eNOS has been shown to be targeted to plasmalemmal caveolae, a process that is dependent on myristoylation and palmitoylation. Other post-translational modifications essential for the correct subcellular targeting of eNOS have not been described previously. We demonstrate, using [35S]methionine pulse-chase experiments, that native eNOS in adult rat ventricular myocytes is initially translated as a nonpalmitoylated 150-kDa isoform, which is associated with cytosolic and intracellular membrane-enriched fractions. This is subsequently processed to a palmitoylated 135-kDa isoform, which is found only in a sarcolemma-enriched membrane fraction. Forskolin, an agent that elevates intracellular cAMP, rapidly inhibited processing of the 150-kDa isoform to the 135-kDa isoform and transport of eNOS to the sarcolemma, effects paralleled by protein kinase A-dependent phosphorylation of the larger eNOS isoform. Forskolin also decreased palmitoylation of the 135-kDa isoform, although it did not accelerate depalmitoylation of sarcolemmal eNOS, as determined by pulse-chase experiments with [3H]palmitate. Thus, post-translational processing of a 150-kDa isoform of myocyte eNOS appears to be necessary for intracellular trafficking of the enzyme to sarcolemmal caveolae. Both the post-translational processing and subcellular targeting of eNOS appear to be modified by changes in intracellular cAMP, an effect that may have important implications for cardiac myocyte responsiveness to autonomic agonists in vivo.

The constitutively expressed, calcium-sensitive isoform of nitric-oxide synthase originally described in large vessel endo-thelial cells, termed eNOS 1 or NOS3, is now known to be expressed in a number of cell types. In the heart, eNOS is present within the endothelium of the epicardial and microvascular coronary vessels and the endocardium, but also in cardiac myocytes and in specialized cardiac pacemaker and conduction tissue, such as sinoatrial and atrioventricular nodal cells (Refs. 1-4; for a review, see Ref. 5). eNOS is activated at increased pacing frequencies in cardiac myocytes and also appears to regulate cardiac myocyte contractile responsiveness to both adrenergic and muscarinic cholinergic autonomic nervous system agonists (1)(2)(3)(4)(5)(6)(7). Although originally presumed to be constitutively expressed in most cell types, it is now known that eNOS expression is regulated. Inflammatory cytokines decrease endothelial cell eNOS mRNA abundance (8), and, as we have demonstrated previously, agonists that increase intracellular cAMP in cardiac myocytes in vitro or in myocytes in situ in intact hearts, gradually decrease eNOS mRNA, protein, and activity levels (9).
All NOS isoforms undergo some degree of subcellular compartmentation. Neuronal NOS (nNOS or NOS1) is associated with post-synaptic density protein-95 and -93 in neurons and with ␣ 1 -syntrophin in skeletal muscle at an amino-terminal binding domain (PDZ domain) unique to this isoform (10,11). The cytokine-inducible NOS (iNOS or NOS2) appears also to undergo subcellular compartmentation, although neither the specific compartment nor the targeting sequence and/or relevant post-translational modification(s) are known (12). As in endothelial cells, eNOS in cardiac myocytes is targeted to specialized plasmalemmal and internal membrane-associated glycosphingolipid-and cholesterol-enriched microdomains that, in association with the integral transmembrane protein caveolin, form caveolae (13)(14)(15). In addition to plasmalemmal membranes, both these specialized lipid microdomains and caveolins are associated with intracellular compartments including a non-clathrin-coated "light vesicular fraction," the Golgi apparatus, and the endoplasmic reticulum-Golgi intermediate compartment (ERGIC) (16 -18). Efficient targeting of eNOS to these lipid microdomains, including plasmalemmal caveolae, requires co-translational myristoylation on an N-terminal glycine residue and post-translational palmitoylation of cysteines 15 and 26 (19 -22). Palmitoylation appears to be dynamically regulated in endothelial cells, since agents such as bradykinin induce depalmitoylation and translocation of eNOS from plas-malemmal caveolae to an intracellular compartment (13,20). This may have important consequences for eNOS-dependent signaling in endothelial cells and cardiac myocytes, given the increasing evidence that these glycosphingolipid-and cholesterol-enriched microdomains and caveolae act to spatially constrain a number of transmembrane and intracellular signal transduction pathways (23)(24)(25)(26). Finally, eNOS has been shown to undergo additional post-translational modifications, including phosphorylation by a calmodulin kinase (21). The importance of these post-translational modifications to subcellular targeting or to regulation of enzyme activity is not yet understood.
We report here that, in cardiac myocytes, eNOS is initially translated as a 150-kDa precursor that is subsequently processed to a 135-kDa protein. Subcellular fractionation studies revealed that the 150-kDa isoform of eNOS is found in intracellular compartments but not in a sarcolemmal membrane fraction. Although both isoforms associate with caveolin-3, a muscle-specific caveolin, and therefore are likely present in glycosphingolipid-and cholesterol-enriched microdomains and in caveolae, only the 135-kDa isoform appears to be palmitoylated. Agents that elevate intracellular cAMP rapidly decrease post-translational processing of the 150-kDa isoform to the 135-kDa isoform, and diminish palmitoylation of the 135-kDa isoform and its translocation to the sarcolemma.

MATERIALS AND METHODS
Cell Culture and Labeling-Adult rat ventricular myocytes (ARVM) were isolated as described previously (27). Myocytes were cultured in a defined medium, termed ACCITT (described in Ref. 28). Cells were exposed to forskolin were diluted in dimethyl sulfoxide (Me 2 SO) (Sigma) as specified in the text. Control cells received Me 2 SO only at an equivalent dilution.
[ 35 S]Methionine pulse-chase experiments were performed using 10 6 ARVM/condition incubated for 1 h in methionine-free ACCITT containing 100 Ci/ml Translabel (ICN) and then washed in ACCITT containing 20 mg/ml unlabeled methionine. Studies of protein phosphorylation in intact myocytes were performed by using 3 ϫ 10 5 cells/condition incubated for 2 h in phosphate-free ACCITT containing 75 Ci/ml 32 P-labeled orthophosphate (DuPont NEN). Drugs were applied at the end of this 2-h period. Control and treated plates were harvested at the same time. In vivo studies of eNOS palmitoylation were performed as described previously (13,21). Briefly, 10 6 ARVM were incubated for 2 h in ACCITT containing 1 mCi/ml of [ 3 H]palmitate (DuPont NEN). The labile nature of the thiopalmitate bond in the presence of hydroxylamine was used to verify the specificity of palmitate labeling as described previously (13,21). To study eNOS depalmitoylation, ARVM were labeled for 2 h as described above, washed twice in ACCITT containing 100 M unlabeled palmitate, and incubated in this medium either with or without 10 M forskolin.
Subcellular Fractionation-Two subcellular fractionation protocols were employed. In some experiments, the cellular fractionation method employed was a modification of that described by Busconi and Michel (19) in which two fractions were prepared (i.e. "particulate" and "soluble" fractions). Briefly, ARVM were scraped into buffer F and sonicated (Branson Sonifier 450; Branson Ultrasonic Corp., Danbury, CT; three 10-s bursts; output power ϭ 1), and then ultracentrifuged (Beckman L8 M) at 100,000 ϫ g for 1 h, washed once in buffer F, and ultracentrifuged again at 100,000 ϫ g for 30 min. The pellet was resuspended in buffer F with 1% Triton X-100 ("particulate fraction"). 1% Triton X-100 (final concentration) was also added to the supernatant ("soluble fraction").
In other experiments, the protocol described by Mery et al. (29) was used, with modifications. After sonication, ARVM lysates were centrifuged at low speed (8,000 ϫ g) for 15 min. The pellet was washed once in buffer F and then resuspended in buffer F with 1% Triton X-100. This fraction is referred as the "plasmalemma-enriched fraction." The supernatant was then processed according to the protocol described above by Busconi and Michel (19). The pellet and the supernatant obtained are referred to as the "internal membrane-enriched" (IM) and the "cytosolic" (C) fractions. Antibodies to Na,K-ATPase (ABR Reagents) and to mannosidase II (Babco) were used as markers for sarcolemmal and internal membranes (i.e. Golgi apparatus), respectively, by Western blot.
Immunoprecipitation-For eNOS immunoprecipitation as well as for Western blots, a monoclonal antibody against human eNOS was used (Transduction Labs), according to the protocol recommended by the manufacturer (i.e. 2 g of eNOS antibody was combined with 500 g of ARVM lysate). Immunoprecipitation experiments were performed as described previously (14) with minor modifications. Briefly, after subcellular fractionation, eNOS was immunoprecipitated with anti-eNOS antibody in buffer F plus detergents, as noted above, containing 4 M tetrahydrobiopterin and 1 mM L-arginine. All samples were run on 12% SDS-PAGE gels, and membranes were subsequently incubated with an anti-caveolin-3 monoclonal antibody (Transduction Labs) and processed as described below.
SDS-PAGE and Immunoblotting-Unlabeled or 32 P-labeled denatured proteins in Laemmli sample buffer (30) were loaded on 7.5% SDS-PAGE gels (Ready-Gels, Bio-Rad) and transferred onto polyvinylidene difluoride membranes (Bio-Rad). As described previously (9,14), the membranes were blocked, incubated for 1 h with the primary antibody (i.e. anti-eNOS, anti-iNOS, or anti-nNOS (Transduction Laboratories) or anti-nNOS-(a gift of Dr. David S. Bredt)), washed six times in TBST (Tris-buffered saline containing 0.1% Tween 20; Sigma), and incubated for an additional 30 min with the secondary antibody. Membranes were stripped in a buffer containing 50 mM Tris, 2% SDS, and 100 mM ␤-mercaptoethanol at 55°C for 30 min, washed six times in TBST, and immunoblotted with another antibody. For 32 P-labeled-proteins, this procedure was used to verify equivalent loading among lanes. The activity of 32 P-labeled proteins was obtained by exposing the polyvinylidene difluoride membranes to x-ray film for 24 h before immunoblotting. 35 S-or 3 H-labeled denatured proteins in Laemmli sample buffer were loaded onto 7.5% SDS-PAGE gels (Ready-Gels, Bio-Rad). Gels were incubated in En 3 Hance (DuPont NEN), washed in water, and dried for 2 weeks at Ϫ70°C.
Measurement of eNOS Enzymatic Activity-NOS activity was assessed as described previously using anion exchange chromatography to measure the conversion of L-[ 3 H]arginine into L-[ 3 H]citrulline (1). Assays were always performed in parallel with and without 1 mM CaCl 2 as a screen for "calcium-insensitive" (i.e. inducible) nitric-oxide synthase activity in ARVM lysates. Parallel reactions were performed also in buffer containing 1 mM nitro-L-arginine, an inhibitor of NOS activity. The specific signal was calculated by subtracting the background signal, determined in the presence of a NOS inhibitor. These results were normalized to protein content, as determined by a modified Lowry assay (Bio-Rad). The amount of eNOS activity is also presented in some experiments normalized to the amount of eNOS protein in each fraction for each experiment, as determined by densitometric analyses of Western blot autoradiograms of equal amounts of ARVM lysates (i.e. 25 l) processed in parallel with lysates used to determine eNOS enzymatic activity.
Statistical Methods-Data are presented as means Ϯ S.E. Comparisons among groups were performed by Student's t tests where data were normally distributed or by a Mann-Whitney test for nonparametric data. The null hypothesis was rejected at p Ͻ 0.05.

Subcellular Localization of eNOS in ARVM-
To evaluate the subcellular distribution of eNOS within cardiac myocytes, two separate cell fractionation techniques were used: a two-fraction technique that we have employed previously that yields "particulate" and "soluble" fractions, and a three-fraction technique, yielding a plasmalemma-enriched membrane (PM) fraction, an IM fraction, and a C fraction, as described under "Materials and Methods." We used specific markers for the plasma membrane (Na/K-ATPase) and internal membranes (mannosidase II) to verify the specificity for subcellular components obtained with these differential centrifugation techniques. Fig. 1 shows immunoblots of ARVM lysates after fractionation using either the two-fraction technique (Fig. 1A) or the three-fraction technique (Fig. 1B) reprobed with mono-clonal antibodies against eNOS and caveolin-3, and then stripped and reprobed successively with monoclonal antibodies against Na,K-ATPase and mannosidase II. As expected, Na,K-ATPase localized to the particulate fraction, using the twofraction protocol, or to the plasmalemma-enriched fraction, using the three-fraction protocol, and mannosidase II localized to the particulate and the internal membrane-enriched fractions, respectively. As we have shown previously (1), about 80% of eNOS was in the particulate and 20% in the cytosolic fraction, using the two-fraction technique (Fig. 1A). When equal amounts of protein were loaded, however, there was no significant difference in the eNOS concentration between the plasmalemma-enriched and internal membrane-enriched fractions. Therefore, since the total amount of protein recovered from the sarcolemmal fraction is 8 times greater than in the IM fraction from the same cell lysate, the large majority of eNOS in cardiac myocytes must be localized to sarcolemmal membranes.
We have shown previously that eNOS is found in caveolae both in bovine aortic endothelial cells and in cardiac myocytes, in association with caveolin-1 and caveolin-3, respectively (14). Therefore, we also investigated the localization of caveolin-3, a muscle-specific caveolin, using both cell fractionation techniques. As shown in Fig. 1, caveolin-3 was present mainly in the particulate fractions using the two-fraction technique, and predominantly in the plasmalemma-enriched and internal membrane-enriched fractions using the three-fraction method, although a faint band could be observed in the cytosol as well. This observation was also consistent with data from other cell types on caveolin cycling between plasmalemmal and internal compartments (18).
Interestingly, we identified two bands in protein extracts from ARVM running at 135 and 150 kDa on 7.5% SDS-PAGE that were specific for eNOS by immunoblotting. After subcellular fractionation using the three-fraction technique, the low molecular mass band (135 kDa) was exclusively localized to the plasmalemma-enriched membranes, whereas the higher molecular mass band (150 kDa) could be detected only in the cytosolic and IM fractions. As a control, we verified that doublet bands were recovered only after mixing of sarcolemma-enriched and IM fractions, or sarcolemma-enriched and cytosolic fractions, but not from mixing IM and cytosolic fractions (data not shown). These two bands were also recovered after immunoprecipitation with the eNOS monoclonal antibody (Transduction Lab) and were not detected after immunoblotting using either monoclonal iNOS (NOS2) or nNOS (NOS1)-specific antibodies, or by a polyclonal antibody directed against nNOS-(data not shown). Although the reason(s) for these differences in the migration pattern of the two bands on 7.5% SDS-PAGE gel is still unclear, this provides a useful tool to determine eNOS subcellular localization (i.e. cytosol, internal membrane, or sarcolemma) as a function of its apparent molecular mass.
Using pulse-chase experiments with [ 35 S]methionine labeling in cultured ARVM (see "Materials and Methods") followed by immunoprecipitation with eNOS-specific antibodies, we determined that the 150-kDa eNOS band, localized as shown in Fig. 1B to the IM fraction, was processed and translocated to the plasma membrane (where the 135-kDa isoform localizes specifically). After a 1-h pulse labeling, ARVM were washed and incubated in buffer containing unlabeled methionine. As shown in Fig. 2 (A and B), radioactivity in eNOS immunoprecipitates prepared from whole myocyte lysates was first incorporated into a 150-kDa band, while minimal labeling of the 135-kDa band could be detected (the 1-h exposure to [ 35 S]methionine is sufficient time for some newly synthesized eNOS to be processed to the 135-kDa isoform). After the chase, the 135-kDa band became labeled as the labeling of the 150-kDa isoform rapidly declined with a half-life of the larger M r isoform of approximately 1 h. These data suggest that the 135-kDa band is derived from the 150-kDa band and that, given the specific subcellular localization of each band as indicated in Fig. 1, eNOS appears to be synthesized with an apparent molecular mass of 150 kDa and then processed and transported to the sarcolemma in cardiac myocytes.
We also measured eNOS enzymatic activity in each subcellular fraction to determine whether these differences in electrophoretic mobility would affect its activity. Following subcellular fractionation, equal amounts of cell lysates from each fraction were used for determination of enzymatic activity and for a parallel eNOS immunoblot. All enzymatic assays were performed in the presence and absence of 1 mM CaCl 2 and in the presence and absence of 1 mM nitro-L-arginine to measure FIG. 1. Two isoforms of native eNOS can be resolved in cellular lysates from ARVM. ARVM lysates were fractionated, and equal amounts of protein were loaded onto 7.5% SDS-PAGE and then immunoblotted with eNOS and caveolin-3 monoclonal antibodies. Membranes were then stripped and successively hybridized with monoclonal antibodies against mannosidase II (a Golgi marker) and then against Na,K-ATPase (a plasmalemmal marker). A, using a subcellular (twofraction) fractionation technique that yields a particulate (P) and a soluble (S) fraction (see "Materials and Methods"), both a 135-and 150-kDa isoform of eNOS could be resolved. As expected, Na,K-ATPase and mannosidase II localized to the particulate fraction. Caveolin-3 also localized to the particulate fraction, although longer exposures revealed its presence also in the soluble fraction. B, the same procedure of sequential immunoblotting with first antibodies to eNOS and caveolin-3, followed by an antibody to mannosidase II, after which Na,K-ATPase was applied to ARVM lysates after processing using a threefraction technique. eNOS could be detected in PM, C, and IM fractions. Na,K-ATPase localized almost exclusively to plasmalemma-enriched membranes, and mannosidase II localized to the internal membraneenriched fraction. Caveolin-3 was found both in plasmalemmal and internal membrane-enriched fractions and after longer exposure times in the cytosolic fraction as well. Subcellular fractionations and immunoblots were performed three times with similar results. Ca 2ϩ -sensitive, NOS-specific activity. There was approximately a 2-fold higher maximal NOS activity in the cytosolic and internal membrane-enriched fractions (85 Ϯ 21 pmol/min, 178 Ϯ 34 pmol/min, and 187 Ϯ 97 pmol/min in PM, C, and IM fractions, respectively) compared with maximal NOS activity in the sarcolemma-enriched fraction, measured with saturating concentrations of substrate and co-factors. Calcium-sensitive NOS activity in each fraction has been corrected for the relative intensity of bands of the appropriate size identified on eNOS immunoblots performed in parallel on each subcellular fraction used for the enzymatic activity assay. The reason(s) for this modest, although consistent, increase in the apparent V max of eNOS in the intracellular fractions is unknown, but could be due to differences in the lipid composition of each fraction, in the access of co-factors or substrate to eNOS within vesicles, or to other factors in addition to a change in intrinsic enzyme activity.
Effect of Elevated Intracellular cAMP on Subcellular Distribution of eNOS in ARVM-We have shown previously that agents that elevate cAMP result in a gradual but substantial decline in eNOS mRNA abundance, protein, and enzymatic activity in ARVM, both in primary culture and in situ in intact rat hearts (9). We sought to determine whether short-term elevations of cAMP would modify the subcellular distribution of eNOS in cardiac myocytes. Using the "two fraction" subcellular fractionation technique, 10 M forskolin increased the abundance of eNOS in the nonparticulate, soluble fraction in a time-dependent manner, whereas its abundance remained unchanged in the particulate fraction over a 2-h period (data not shown). Exposure to forskolin for longer than 2 h did not further increase eNOS localized to the soluble fraction. The three-fraction cell fractionation protocol yielded complementary data, as shown in Fig. 3 (A and B). After a 2-h exposure to 10 M forskolin, the more slowly migrating eNOS band was increased in both the cytosolic and IM fractions, with no change over this time period in the eNOS signal in the plasmalemmal fraction and no change in the distribution of caveolin-3 among those three fractions over this time period.
These data suggest that increased intracellular cAMP levels inhibited translocation of eNOS from internal membranes to the sarcolemma. To verify this, ARVM were incubated in buffer containing [ 35 S]methionine for 1 h and then washed and cultured in buffer containing unlabeled methionine in the continuous presence of 10 M forskolin. Under these conditions, as shown in Fig. 3 (C and D), the 150-kDa form was [ 35 S]methionine-labeled and remained labeled up to 24 h after the chase, whereas little labeled 135-kDa eNOS band appeared, in contrast to what was observed in the absence of forskolin, as shown in Fig. 2.
We verified that the changes we observed in Western blot experiments were accompanied by redistribution of eNOS activity in ARVM. We performed citrulline assays after exposure of intact cells to forskolin, followed by subcellular fractionation in the presence of phosphatase inhibitors using the three-

FIG. 3. Forskolin regulates post-translational processing of native eNOS in ARVM.
A and B, myocyte cell lysates processed using the three-fraction subcellular fractionation technique were prepared from cells exposed to 10 M forskolin for 2 h, resolved with 7.5% SDS-PAGE, and immunoblotted with both anti-eNOS and anti-caveolin-3-specific antibodies. As shown in A, the higher M r isoform of eNOS was enriched in the C and IM fractions compared with the PM fraction in forskolin-pretreated cells, although there was no obvious shift in the distribution of caveolin-3 among these three fractions. Combined data from four experiments are shown in B for total eNOS abundance, expressed as a percentage of that present in plasmalemmal membranes from control, non-forskolin-treated cells (CON), following normalization for the protein content of each fraction (mean Ϯ S.E.; *, p Ͻ 0.05 compared with the same fraction in control). C and D, to verify that forskolin inhibited post-translational processing of the 150-kDa band and its translocation to the sarcolemma, eNOS immunoprecipitates were prepared from whole myocyte lysates and resolved by 7.5% SDS-PAGE, from ARVM previously exposed to fraction technique. Fig. 3E shows eNOS activity, expressed both per mg of protein in each subcellular fraction (open bars) and as a function of the amount of immunoreactive eNOS present in each fraction (solid bars). Maximal enzyme activity was significantly increased after forskolin exposure in both the IM and cytosolic fractions, whereas no significant change in activity was observed in the sarcolemma-enriched fraction. These results paralleled the changes in subcellular localization of eNOS after exposure to forskolin, as determined by Western blotting (Fig. 3, A and B). To determine whether this increase in apparent V max represented an increase in activity per mg of eNOS protein or an increase in the amount of eNOS protein, the measurement of enzymatic activity was normalized to the relative amount of eNOS in each fraction. As shown in Fig. 3E, the increase in eNOS activity per mg of protein in the cytosolic and internal membrane-enriched fractions following forskolin was due to the increased eNOS protein content in these fractions.
cAMP-dependent Phosphorylation of eNOS in Vivo-eNOS has a consensus sequence for PKA-dependent phosphorylation. In endothelial cells, phosphorylation of eNOS by bradykinin but not forskolin was associated with translocation of the enzyme from a particulate fraction to the cytosol (22). To determine whether forskolin would initiate protein kinase A-dependent phosphorylation of eNOS in intact ARVM, ARVM were incubated with [ 32 P]orthophosphate and exposed to 10 M forskolin for 20 min, with or without preincubation with 1 M KT5720, an inhibitor of protein kinase A. eNOS was then immunoprecipitated from cell homogenates and run on 7.5% SDS-PAGE. As shown in Fig. 4A, although no phosphorylation was observed in control conditions (i.e. in the absence of forskolin), both the 150-kDa band (corresponding to the cytosolic and IM localization) and to a lesser extent the 135-kDa band (sarcolemmal localization) were phosphorylated. This phosphorylation could be partially inhibited by 1 M KT5720. An immunoblot (Fig. 4B) of the same membrane shown in Fig. 4A verifies the identities of the two phosphorylated bands as the two isoforms of eNOS. Laser densitometry of autoradiograms of [ 32 P]orthophosphate-labeled myocyte lysates under control conditions and following forskolin treatment revealed that 93 Ϯ 3% of the total radioactivity was incorporated into the 150-kDa band with forskolin (mean Ϯ S.E. of data from four experiments). Since the ratio of the abundance of the 150-kDa to the 135-kDa eNOS species in intact cells is approximately 1:8, the majority of phosphorylated eNOS in these cells is the 150-kDa isoform.
cAMP and eNOS Palmitoylation in ARVM-Palmitoylation appears to be required for efficient targeting of eNOS to caveolae in COS cells transfected with wild-type or palmitoylationdeficient eNOS mutants (13,20,21). It has also been shown that palmitoylation is a dynamic process and that depalmitoylation is regulated by bradykinin in endothelial cells. Thus, we examined whether increased intracellular cAMP would affect eNOS palmitoylation in cardiac myocytes, by examining cell lysates from ARVM that had been exposed to [ 3 H]palmitate for 2 h in the absence or presence of forskolin. As shown in Fig. 5A, the only [ 3 H]palmitate-labeled eNOS band is the 135-kDa band associated with the sarcolemma-enriched fraction. No 150-kDa band could be detected, suggesting that this eNOS isoform is not palmitoylated during the period of biosynthetic labeling. It is possible that the assay described here for the detection of eNOS palmitoylation was insufficiently sensitive to detect low, but biologically relevant levels of acylation of the less abundant 150-kDa isoform. If this were true, some degree of [ 3 H]palmitate labeling should have been apparent in the immunoprecipitated eNOS lane under control conditions (Fig. 5A, upper part,  lane 3). Moreover, after 2 h of incubation in forskolin when the ratio of the 135-kDa isoform to the 150-kDa isoform had declined to only 1:2 (see Fig. 3B), there was no detectable palmitoylation of the 150-kDa band. Two bands corresponding to the 150-and 135-kDa isoforms can be identified in a parallel immunoblot performed on the eNOS immunoprecipitate from this experiment (Fig. 5A, lower part).
When myocytes were exposed to forskolin during [ 3 H]palmitate labeling, palmitate labeling of the 135-kDa eNOS signal was markedly reduced. As shown in Fig. 3B, this cannot be explained solely by removal and degradation of the lower M r eNOS from the sarcolemma-enriched fraction caused by only a 2-h exposure to forskolin. This happened without a noticeable change in palmitoylation of nonimmunoprecipitated proteins remaining in the supernatant, suggesting that the forskolinmediated effect may be relatively specific for eNOS palmitoylation. To determine whether elevated intracellular cAMP would also affect eNOS depalmitoylation, pulse-chase experiments were performed in myocytes labeled for 2 h with skolin. As shown in Fig. 5B, there was no apparent effect of forskolin on the rate of eNOS depalmitoylation. The corresponding immunoblot showing the presence of both the 150and 135-kDa isoforms is shown in the lower part of Fig. 5B. DISCUSSION The recognition that there were two distinct eNOS M r species was evident only when ARVM eNOS whole cell lysate immunoprecipitates were resolved by 7.5% SDS-PAGE, rather than on the 12% gels employed by this laboratory in previous studies of this cell type (1,14). In addition, due to the approximately 8-fold greater absolute abundance of the 135-kDa isoform, a 150-kDa band was not observed until longer exposure times for Western blots were routinely used. Interestingly, in a report by Busconi and Michel (19), doublet bands corresponding to 150-and 135-kDa proteins could be resolved by 7% SDS-PAGE of cell lysates of COS-7 cells that had been transfected with either wild-type eNOS or myristoylation-deficient eNOS mutant cDNAs. Since this eNOS doublet was only observed in lysates from transfected COS-7 cells and not in lysates from bovine aortic endothelial cells that express a native eNOS, where only the 135-kDa isoform could be resolved with either 7 or 12% SDS-PAGE gels, it was presumed that the higher M r band represented either a post-translational modification of eNOS that was more prominent in COS-7 cells or an anomalous processing of the transfected eNOS cDNA in these cells. Neither band, however, was localized to the particulate fraction in COS-7 cells transfected with the myristoylationdeficient mutant.
Several explanations are possible for the slower migrating 150-kDa band on SDS-PAGE. Based on the eNOS mRNA sequence, the predicted size of the protein is 135 kDa. Nevertheless, as noted above, transfection of COS-7 cells with the fulllength eNOS cDNA yielded both 135-and 150-kDa bands (19). While differences in phosphorylation constitute one possible explanation, it is worth noting that increased phosphorylation by PKA in response to forskolin (Fig. 4) did not affect migration of either the 135-or 150-kDa band. Palmitoylation is also an unlikely explanation, since this post-translational modification has not been shown to affect eNOS mobility on 7% SDS-PAGE (19 -22). Incomplete reduction of disulfide bonds is also an unlikely explanation, since the loading conditions are identical for both eNOS bands, and both bands can be recovered when plasmalemmal and internal membrane samples are combined. Finally, a consensus sequence for glycosylation does exist on eNOS, although glycosylation of this enzyme has not yet been reported. The identification of the post-translational processing responsible for the presence of the two apparent eNOS M r species is being actively investigated.
The addition of an initial low speed centrifugation step following preparation of ARVM lysates by sonication, followed by the same subcellular fractionation technique used in the report by Busconi and Michel (19) on the remaining supernatant, permitted the preparation of sarcolemmal membrane-enriched fractions identified by the presence of Na,K-ATPase only in this fraction. This technique does not permit quantitative separation of "plasma membranes" from "internal membranes" and "cytosolic" fractions of these cells. Nevertheless, it was in this fraction exclusively that the 135-kDa fraction was identified, as shown in Fig. 1B. The identity of the subcellular compartments containing the 150-kDa isoform is not resolved here, but these compartments are likely to include, based on the co-localization of mannosidase II in this fraction, portions of the Golgi apparatus or ERGIC complex as had been reported previously for eNOS in bovine aortic and bovine pulmonary microvascular endothelial cells (15,31).
Our finding that caveolin-3 can be detected in both the plas-malemmal membrane-and internal membrane-enriched fractions is consistent with previous reports by us and by others that eNOS (at least the 135-kDa isoform) is associated with plasmalemmal caveolae in endothelial cells and cardiac myocytes (13)(14)(15). This is because caveolae (and, therefore, caveolins and associated glycosphingolipid-and cholesterol-enriched microdomains) have been shown to participate in intracellular trafficking into and from the Golgi apparatus and the ERGIC as well as from subplasmalemmal light vesicular fractions (16 -18, 32, 33). However, the observation that only the 135-kDa isoform appears to be palmitoylated, as shown in Fig. 5, suggests that palmitoylation is necessary for eNOS association with sarcolemmal caveolae, since only the 135-kDa isoform, but not the higher M r eNOS species, could be identified in Na,K-ATPase-enriched myocyte subcellular fractions. This is consistent with a recent report from this laboratory, using a detergent-free technique for the isolation of plasmalemmal caveolae only, that showed that palmitoylation-deficient eNOS mutants expressed in COS-7 cells were not enriched in a caveolar fraction prepared from plasma membranes compared with COS-7 cells transfected with wild-type eNOS (13). The data reported in the present manuscript are also consistent with the observation by Garcia-Cardeñ a (15) that native, acylated eNOS in rat pulmonary microvascular endothelial cells is present within glycosphingolipid-, cholesterol-and caveolae-enriched plasmalemmal microdomains isolated using a method of in situ perfusion of rat lungs with a solution of positively charged colloidal silica particles (34). These authors also noted that a palmitoylation-deficient eNOS mutant expressed in NIH 3T3 cells co-localized with caveolins in internal membranes (with a perinuclear distribution consistent with Golgi or ERGIC localization) but not with caveolins in the plasma membrane (15). This was in contrast to wild-type eNOS, which co-localized with caveolin in both internal and plasmalemmal membranes. As shown in the present study, palmitoylation of the mature 135-kDa isoform is associated with trafficking of native eNOS to the sarcolemma in myocytes. The observation that forskolin, an agent that increases intracellular cAMP by directly activating adenylyl cyclase, could interrupt processing of the 150-kDa eNOS precursor to the lower M r isoform, has not previously been reported. Indeed, as shown in Fig. 3C, little [ 35 S]methionine-labeled 135-kDa band could be identified in 7.5% SDS-PAGE gels of lysates that had been prepared from forskolin-treated cardiac myocytes during chase periods as long as 24 h after pulse labeling. In contrast, in the absence of forskolin, most of the 150-kDa band is processed to the 135-kDa form within 4 h, following a [ 35 S]methionine pulse, as shown in Fig. 2. Therefore, increased intracellular cAMP appears to prevent the trafficking of mature, palmitoylated native eNOS to sarcolemmal caveolae.
Interestingly, Mineo et al. (35) have recently shown that the recruitment of Raf-1 to plasmalemmal caveolae in rat-1 cells, following their stimulation with EGF, could be blocked by treating these cells with either forskolin or with 8-bromo-cAMP, a nonhydrolyzable cell-permeant cAMP analogue. Thus, elevated cAMP may interrupt trafficking to plasmalemmal caveolae of a number of proteins involved in signal transduction, including both growth factor-dependent and NO-dependent signaling pathways. Given the emerging evidence in support of a role for endogenous production of NO in mediating muscarinic cholinergic regulation of cardiac myocyte responsiveness to ␤-adrenergic agonists (1)(2)(3)(4)(5), the data reported here suggest that even short term elevations in intracellular cAMP may rapidly uncouple NO-dependent signaling by inhibiting post-translational processing of eNOS and its translocation to plasmalemmal caveolae.