Palmitoylation of Inducible Nitric-oxide Synthase at Cys-3 Is Required for Proper Intracellular Traffic and Nitric Oxide Synthesis*

A number of cell types express inducible nitric-oxide synthase (NOS2) in response to exogenous insults such as bacterial lipopolysaccharide or proinflammatory cytokines. Although it has been known for some time that the N-terminal end of NOS2 suffers a post-translational modification, its exact identification has remained elusive. Using radioactive fatty acids, we show herein that NOS2 becomes thioacylated at Cys-3 with palmitic acid. Site-directed mutagenesis of this single residue results in the absence of the radiolabel incorporation. Acylation of NOS2 is completely indispensable for intracellular sorting and ·NO synthesis. In fact, a C3S mutant of NOS2 is completely inactive and accumulates to intracellular membranes that almost totally co-localize with the Golgi marker β-cop. Likewise, low concentrations of the palmitoylation blocking agents 2-Br-palmitate or 8-Br-palmitate severely affected the ·NO synthesis of both NOS2 induced in muscular myotubes and transfected NOS2. However, unlike endothelial NOS, palmitoylation of inducible NOS is not involved in its targeting to caveolae. We have created 16 NOS2-GFP chimeras to inspect the effect of the neighboring residues of Cys-3 on the degree of palmitoylation. In this regard, the hydrophobic residue Pro-4 and the basic residue Lys-6 seem to be indispensable for palmitoylation. In addition, agents that block the endoplasmic reticulum to Golgi transit such as brefeldin A and monensin drastically reduced NOS2 activity leading to its accumulation in perinuclear areas. In summary, palmitoylation of NOS2 at Cys-3 is required for both its activity and proper intracellular localization.

The gaseous radical nitric oxide (⅐NO) 1 modulates biological function in a wide range of tissue types, acting either as a signaling molecule or as a toxin. Three human NOS isoforms have been cloned and characterized. Among them, NOS2 (sometimes referred to as inducible NOS or iNOS) is mostly involved in the synthesis of the large amounts of ⅐NO that appear in inflammatory and immunologic processes (1,2).
Both crystallographic and enzymatic studies performed with recombinant proteins expressed in Escherichia coli have shown that the N terminus end of the three mammalian NOSs is not involved in ⅐NO synthesis but rather in subcellular targeting of the mature polypeptide chain (2,3). For instance, the PDZ domain of NOS1 (residues 1-90) interacts with dystrophin and becomes localized to the sarcolemma of fast twitch fibers (4). In fact, deletion of the first 226 amino acids of NOS1 results in a catalytic protein that synthesizes ⅐NO at a similar rate than the full-length protein (5). Likewise, the N-terminal end of NOS3 is covalently and irreversibly myristoylated at Gly-2 and reversibly palmitoylated at Cys-15 and Cys-26 in a well described process responsible for its targeting to caveolae (6,7). In addition, deletion of the first 52 amino acids of NOS3 does not affect catalytic activity, reflecting that this sequence stretch is not part of the enzymatic machinery but is involved in intracellular traffic (8).
As in the case of NOS1 and NOS3, the N-terminal end of NOS2 is very likely involved in subcellular targeting. In fact, deletion of the first 65 amino acids of a recombinant NOS2 expressed in E. coli results in a mutant protein that behaves like the wild-type counterpart (9). However, much less is known about the post-translational modifications of NOS2 in vivo within eukaryotic cells, such as acylation, phosphorylation, or its binding to other cellular proteins. Interestingly, when a peptide corresponding to the first 17 amino acids of NOS2 is used to elicit a rabbit antiserum the resulting antibodies recognize a soluble form of the enzyme, but not its membrane-bound counterpart (10). In contrast, both the particulate and the soluble NOS2 were recognized by a serum elicited against the C terminus end of the protein (10). Likewise, the presence of a post-translational modification in NOS2 resulting in membrane attachment with an increased molecular mass has also been reported (11,12). Although this membrane association of NOS2 has been well characterized in a number of tissues (11)(12)(13)(14), the mechanisms underlying the translocation of NOS2 to the particulate fraction and its subsequent dissociation is poorly characterized.
We demonstrate in this article that the palmitoylation of NOS2 at Cys-3 is a process necessary for its intracellular transit toward subcellular domains where ⅐NO synthesis is required. In fact, the site-directed mutant where this Cys-3 has been converted into a Ser residue did not incorporate the radioactive fatty acid and irreversibly aggregated in the Golgi compartment, becoming completely devoid of activity.
The C2C12 myoblasts were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine in a 5% CO 2 atmosphere at 37°C. Differentiation of ϳ70% confluent myoblasts into myotubes was performed in Dulbecco's modified Eagle's medium supplemented with antibiotics and glutamine plus 50 nM insulin in the absence of serum for 3 days (12). Stimulation of myotubes, transfection, immunoprecipitation, immunofluorescence, and determination of the ⅐NO release was performed essentially as previously reported (12,15).
Molecular Cloning and Construction of the 16 NOS2 Chimeras-We have previously described the cloning and expression of the full-length wild-type NOS2 (12), which possessed the initial sequence MACPWK-FLFKVKSYQSD . . . We have created the following mutant NOS2-GFP constructs: wild-type, the single mutants A2C, C3S, P4K, K6E, and the double mutants K6E/K10E, P4E/K6E, and P4K/W5K. Every construct was obtained as a full-length NOS2-GFP chimera and as a NOS2-(1-94)-GFP chimera. We PCR amplified the NOS2 cDNA introducing the desired mutation together with a novel NcoI site at the 5Ј end and a novel BssHII site at the 3Ј end. The general design of the NOS chimeras consisted of NOS2 cDNA fused to the enhanced GFP using a BssHII site in the pCDNA3 plasmid (Invitrogen) under the control of the cytomegalovirus promoter as previously described (12,15). The PCR-amplified bands were then ligated into the pGEM-T vector (Invitrogen) and subsequently sequenced. We then double-digested the NOS2 in pGEM-T with NcoI plus BssHII and ligated the bands into the corresponding sites of a pUC-linker-GFP construct (12,15). In this construct we had a BssHII at the 5Ј end of the GFP coding region, as well as a XhoI at the 3Ј end followed by a His 6 ending with a XbaI site. This NOS2-GFP pUC construct was digested with XbaI for 2 h and EcoRI was then added for 5 min to induce a partial digestion (there are internal EcoRI sites within the NOS2 cDNA). Finally, the larger band obtained from this partial digestion (ϳ4.2 kb in the case of the large chimeras and 1 kb in the case of the short chimeras) was ligated into pCDNA3 that had been previously double digested with EcoRI plus XbaI. This NOS2-GFP pCDNA3 plasmid was used to transfect myoblasts and myocytes in culture using Escort-III and Escort-IV (Sigma) following the manufacturer's instructions.
Metabolic Labeling-Dishes of COS7 cells transfected with the desired NOS2 construct were metabolically labeled with [ 125 I]iodopalmitate (16) in the case of the NOS2-(1-94)-GFP chimeras described in Fig.  4 and with 3 H-tritiated palmitic acid in all the other cases as previously described (15). Both methods of radioactive labeling are equally valid and rendered similar results. After a 4-h incubation with 100 Ci per 100-mm dish, the radiolabeling media containing the isotope were removed and cells were washed twice with culture media. The cells were lysed in RIPA buffer and the radiolabeled proteins were immunoprecipitated with goat anti-GFP antibodies (Eusera.com) and Protein A-Sepharose as previously described (15,17). A control dish transfected under identical conditions but in the absence of radioactivity was also immunoprecipitated to determine the total amount of NOS2-GFP chimera (palmitoylated and non-palmitoylated) that was present in each experiment. These pilot Coomassie-stained gels were separated by SDS-PAGE and electroblotted onto Immobilon-P polyvinylidene difluoride membranes (Millipore, Bedford, MA). Western blot analysis of various GFP chimeras was performed with the use of the goat polyclonal anti-GFP serum followed by ECL detection (Amersham Biosciences). Incorporation of [ 125 I]iodopalmitate or 3 H-tritiated palmitic acid into the proteins was visualized by phosphorimaging and autora-diography of the polyvinylidene difluoride membrane. A control labeling experiment was equally performed using a chimeric Fyn kinase GFP and its mutant chimera Fyn(G2A) fused to GFP that included the first 15 amino acids of the N-terminal end of the sequence (17).
Assay of NOS2 Activity-Because NOS2-derived nitric oxide decomposes in nitrites and nitrates, we performed the Griess assay according to our published protocol (12). A volume of 0.5 ml of sample was incubated with 50 l of a 100 mM sulfanilic acid solution and 50 l of a 10 mM N-(1-naphthyl)ethylenediamine dihydrochloride solution. The mixture was allowed to react for 15 min, and the absorbance value at 540 nm was determined. Every sample was analyzed in triplicate. Fresh solutions of sodium nitrite were regularly prepared as standards. In addition, in COS7 cells transfected with the various NOS2 constructs we also performed the L-[ 14 C]arginine to L-[ 14 C]citrulline assay as previously described (18).
Recombinant Expression of NOS in E. coli and Purification-NOS2 cloned in the expression vector pCWori was used to transform competent BL21 cells (Novagen) where the coexpression vector for calmodulin was already inserted (19). Four liters of 2ϫ YT media were used for protein expression at 22°C. The protein was purified using two affinity columns as previously described (20).
Incubation of Live Cells with BODIPY-Texas Red-Ceramide-The live COS7 cells transfected with the various GFP chimeras were washed twice with phosphate-buffered saline and examined for their subcellular distribution after addition of 1,5 M BODIPY-Texas Redceramide. Fixing was avoided because the methanol distorts significantly the BODIPY-Texas Red-ceramide signal. The cells were kept at 37°C using a Peltier system.
Neutral Hydroxylamine Treatment-The wild-type NOS2-GFP chimera was immunoprecipitated using anti-GFP antibodies and divided into two fractions that were loaded in separate gels. One of the SDS-PAGE gels was treated with 1 M Tris, pH 7.0, for 24 h, whereas the other gel was incubated with a buffered solution of 1 M hydroxylamine.

Palmitoylation of Both Cytokine-induced and Transfected
NOS2-When C2C12 muscular myotubes are challenged with a mixture of LPS and IFN-␥ a clear NOS2 induction can be observed both by immunoblotting and immunofluorescence (12). Incubation of these cells with low concentrations of the palmitoylation inhibitors 2-Br-and 8-Br-palmitate lead to a dose-dependent diminution in ⅐NO synthesis (Fig. 1A, left panel). Likewise, when NOS2 was transfected in COS7 cells, the dose-dependent addition of 8-Br-palmitate produced a drastic decrease in NOS2 activity (Fig. 1A, right panel), although in both cases the total amount of NOS2 remained unchanged. In agreement with our previous data (12), when we used the wild-type NOS2-GFP construct to transfect COS7 cells a clear particulate distribution throughout the entire cytosol could be observed (Fig. 1B, left panel). This staining changed into a perinuclear distribution upon incubation of the cells with 10 M 8-Br-palmitate (Fig. 1B, right panel). In consequence, inhibition of NOS2 palmitoylation using both 2-Br-and 8-Br-palmitate resulted in profound changes both in ⅐NO synthesis and the subcellular distribution of NOS2.
NOS2 Is Palmitoylated on Cys-3-Next, we performed sitedirected mutagenesis of the N-terminal end of NOS2, changing the Cys residue at position 3 into a Ser. This mutation created a recombinant chimera that, when transfected in COS-7 cells, was devoid of activity and displayed a large degradation pattern according to the immunoblot data. In addition, the C3S NOS2 mutant accumulated to perinuclear areas (Fig. 1C), in clear resemblance with the phenotype shown by the wild-type protein when treated with 8-Br-palmitate. Whereas the wildtype NOS2 transfected in COS7 cells was found both in the soluble and particulate fractions (in an approximate ratio of 40:60), the C3S mutant was completely found in the particulate fraction (Fig. 1C). This finding reveals that either the inactive C3S mutant associates with membrane fractions or the mutation results in an improper folded aggregated protein. These results may also suggest a critical role for the palmitate moiety attached to Cys-3 in the wild-type protein that cannot be re-placed by a non-acylated Ser side chain. Hence the proper subcellular sorting of the C3S NOS2 mutant becomes disrupted. To investigate whether Cys-3 of NOS2 was palmitoylated within mammalian cells, we labeled COS7 cells transfected with the constructs of wild-type and C3S mutant with the 125 I-fatty acid analogue of palmitic acid as previously described (16). As depicted in Fig. 1D the full-length wild-type construct clearly incorporated the radioactive label, unlike the C3S mutant, that failed to show any significant labeling, despite similar amounts of total protein in both cases. When mouse C2C12 myotubes were incubated with radiolabeled palmitic acid and NOS2 expression was induced using a mixture of LPS and IFN-␥ (12), a clear band appeared at 135 kDa, which could be observed in the total lysate of cells subjected to the proinflammatory insult (Fig. 1E). This band was almost absent in the total lysate of both myoblasts and C2C12 myotubes that had not been treated with the mixture of cytokines (Fig. 1E).
In addition, we purified recombinant NOS2 obtained expressed in E. coli as previously described (20). As expected, the purified enzyme appeared as a single band of ϳ135 kDa in a SDS-PAGE gel ( Fig. 2A). Using this recombinant NOS2 (in which Cys-3 is not acylated), we tested if increasing concentrations of 8-Br-palmitate altered its enzymatic activity. In contrast with the results obtained with palmitoylation inhibitors of NOS2 tested on mammalian cells, purified NOS2 does not significantly change its ⅐NO synthesizing activity in the 10 -100 M range of 8-Br-palmitate ( Fig. 2A, right panel). Hence, the diminution in activity observed in both transfected and cytokine-induced NOS2 cells when treated with 2-and 8-Br-palmitate must be because of alterations in the normal sorting routes of NOS2 when the incorporation of palmitic acid is abrogated. It is consequently logical to speculate that there is a tight interconnection between palmitoylation of NOS2 and its proper ⅐NO synthesis.
The changes observed in the activity of NOS2 transfected in COS7 cells were also determined using the L-[ 14 C]arginine to L-[ 14 C]citrulline assay (Fig. 2B). As expected, incubation of wild-type NOS2 with 20 M 2-and 8-Br-palmitate lead to decreased levels of citrulline, as previously observed when the levels of nitrites were determined using the Griess assay. Likewise, the C3S mutant of NOS2 displayed less than 10% of the activity of its wild-type counterpart.
Indeed, a further proof of the existence of a palmitic acid linked to NOS2 through a thioester linkage was attained using the potent nucleophile hydroxylamine (Fig. 2C). When COS7 cells transfected with the wild-type NOS2-GFP chimera and incubated with [ 3 H]palmitic acid were immunoprecipitated followed by treatment with 1 M Tris or 1 M hydroxylamine, cleavage of the S-acyl bond could only be observed in the presence of the latter. These results establish that the linkage between NOS2 and the palmitate moiety occur through a labile thioester bond.
Colocalization of NOS2 with the Golgi Markers BODIPY-Texas Red Ceramide and ␤-Cop-Next, we analyzed the intracellular traffic of both the NOS2 and C3S GFP chimeras when transfected in COS7 cells. We performed a double immunofluorescence of both GFP constructs with a Golgi marker in red (Cy3 or Texas Red) and the cell nuclei in blue (Hoechst) (Fig. 3). BODIPY-Texas Red-ceramide is commonly used as a marker of the late Golgi apparatus and trans-Golgi network together with certain endosomes (12,15,17,21). On the other hand, coatomer proteins are involved in regulating the transport between the endoplasmic reticulum and the Golgi complex as well as in intra-Golgi transport. Hence, antibodies against ␤-cop are frequently used as ER to cis-Golgi markers (22,23). When the subcellular distribution of the full-length WT and C3S mutant of NOS2 along the sorting pathways were compared, consistent differences could be observed (Fig. 3). Whereas, the non-palmitoylated mutant clearly accumulated in perinuclear areas that colocalized with both BODIPY-Texas Red-ceramide and ␤-cop, the wild-type phenotype showed a more clear distribution throughout the cytosol. When we quantified the colocalization percentages for the BODIPY-Texas Red-ceramide and ␤-cop we obtained the data shown in Fig. 3. Although the wild-type NOS2 partially colocalized with the Golgi markers in certain perinuclear locations, the C3S mutant displayed an almost complete overlap with ␤-cop and BODIPY-Texas Red-ceramide, indicative that the correct sorting of the protein has been interrupted in the Golgi apparatus. Only 17% of the total WT GFP fluorescence colocalizes with BODIPY labeling, whereas 95% of the C3S GFP fluorescence colocalizes with BODIPY staining, reflecting that the wild-type chimera has partially passed through the Golgi in transit to areas closed to the plasma membrane. On the other hand, 12% of WT GFP fluorescence colocalizes with ␤-cop labeling, in contrast with the C3S mutant, where 63% of its GFP fluorescence overlaps with ␤-cop staining. When the red-green colocalization is considered, WT GFP fluorescence displays an almost complete colocalization with both the BODIPY and ␤-cop fluorescences (100 and 99%), whereas the GFP fluorescence of the C3S chimera only "covers" 24% of the total BODIPY fluorescence and 81% of the total ␤-cop fluorescence. These results not only indicate that the full-length NOS2-GFP chimera moves along the traffic machinery of the COS7 cells in transit through the ER-Golgi-TGN compartments, but also provide direct evidence that the palmitoylation of the cysteine residue located at position 3 of its sequence is required for this process.
Subcellular Distribution and Radioactive Labeling of NOS2-GFP Chimeras Mutated in the N-Terminal End-A detailed analysis revealed that the N-terminal end of NOS2 includes hydrophobic amino acids (i.e. Pro, Trp, Phe, Leu, and Val) together with three basic Lys residues at positions 6, 10, and 12 (Fig. 4A). To reveal the requirements for NOS2 palmitoylation, we designed mutations in amino acids located at the N-termi-

FIG. 2. Inhibitors of protein palmitoylation do not affect the activity of recombinant NOS2 activity per se. Panel A shows a
Coomassie Blue-stained gel of recombinant NOS2 expressed in E. coli (ϳ2 g of purified protein were loaded). The effect of increasing doses of 8-Br-palmitate on the activity of recombinant NOS2 protein is shown on the right. COS7 cells were also transfected with the full-length wildtype NOS2 and its activity was determined using the 14 C-labeled L-Arg to citrulline conversion assay (B). The thioester bond that links the side chain of Cys-3 of NOS2 with palmitate can be cleaved using a strong nucleophile such as hydroxylamine (C). Immunoprecipitated NOS2 obtained from [ 3 H]palmitate-labeled COS7 cells was incubated with either 1 M Tris, pH 7.0, or 1 M hydroxylamine, pH 7.0, for 24 h. The gels were dried and residual radiolabel was detected by autoradiography. The position of the NOS2-GFP chimera is indicated by the arrow.

FIG. 3. Colocalization of full-length wild-type NOS2-GFP and its palmitoylation-defective mutant C3S-GFP with the cis-and trans-Golgi makers BODIPY-Texas Red-ceramide and ␤-cop.
COS7 cells were transfected with the full-length wild-type NOS2-GFP and C3S NOS2-GFP constructs and incubated in vivo with the Golgi/ TGN marker BODIPY-Texas Red-ceramide (1.5 M in Dulbecco's modified Eagle's medium) (A). COS7 cells transfected with these same constructs were fixed with paraformaldehyde and methanol and incubated with the cis-Golgi marker ␤-cop (B). In both treatments the merge signal is depicted on the right. The BODIPY-Texas Red-ceramide as well as ␤-cop fluorescence were visualized by confocal microscopy at an excitation wavelength of 543 nm and are shown in red, whereas the GFP fluorescence of the constructs was obtained after excitation at 488 nm and is shown in green. The position of the cell nuclei (blue) was obtained after staining with Hoechst and excitation at 405 nm. The percentages of red-green and green-red colocalization are shown. nal end of full-length NOS2-GFP (boxed in Fig. 4A) that included an additional Cys residue at position 2 as well as mutants P4K, P4K/W5K, K6E, K6E/K10E, and P4E/K6E (Fig. 4B). Immunofluorescence data indicated that all the full-length chimeras were excluded from the cell nucleus and associated with perinuclear areas to a different extent (Fig. 4B). Interestingly, introduction of an additional palmitoylable Cys residue rendered a mutant protein that partially associated with the plasma membrane (Fig. 4B). Although the substitution of hydrophobic residues for basic residues only seem to result in limited changes when compared with the wild-type phenotype, the introduction of acidic residues at the N-terminal end of NOS2 yielded a slight increase in perinuclear labeling, especially in the K6E NOS2 chimera (Fig. 4B). Because NOS2 activity has been shown to be inhibited by synthetic caveolin peptides (24) and by caveolin-1 within muscular cells (12) we considered that the full-length NOS2 might possess various targeting motifs. To elucidate the role of the N-terminal end in the absence of interference from the caveolin binding motif (FPGCPFNGW) that is found spanning residues 325-333, we created deletional chimeras of NOS2-GFP containing only the first 94 amino acids of NOS2 fused to GFP. These constructs would presumably localize to certain subcellular locations dictated by the N-terminal acylation, hence bypassing a possible targeting to caveolae. Short chimeras were constructed of wildtype NOS2-GFP and also of mutants C3S, P4K, P4K/W5K, K6E, K6E/K10E, P4E/K6E, and A2C (Fig. 4C). Remarkably, unlike the full-length chimeras, most of their short counterparts displayed a partial nucleus staining. Remarkably, both the wild-type and A2C short NOS2-GFP chimeras displayed a distinctive plasma membrane localization (see arrows in Fig.  4C). When we compared the subcellular distribution together with the radioactive fatty acid incorporation (Fig. 4D), we could conclude that palmitoylation can be correlated with a certain amount of the recombinant chimera in the completion of the subcellular transit leading to the plasma membrane. On the other hand, the short mutants that failed to become palmitoylated, such as C3S, K6E, K6E/K10E, P4E/K6E, and P4K never reached the plasma membrane, and in many cases showed a clear perinuclear localization (Fig. 4, C and D). Consequently, the presence of both the basic residues Lys-6 and Lys-10 together with the hydrophobic residue Pro-4 seem to be determinant in governing the proper palmitoylation of the NOS2-(1-94) chimeras as well as their subcellular traffic. In addition, at least in the case of WT and A2C chimeras (which become palmitoylated), the targeting information conferred by these first 94 amino acids of NOS2 allows a full transit past the trans-Golgi network and clear localization in the plasma membrane (shown by arrows in Fig. 4C). Nevertheless, as previously shown in our double immunofluorescence studies, 83% of the WT full-length GFP chimera has already exited the TGN and is in transit to the plasma membrane in a catalytically active state. On the other hand, to explain the divergence in subcellular distribution displayed by the short and long chimeras we propose that one of the multiple protein-protein interactions reported for NOS2 might be responsible for this differential targeting (caveolin binding, interaction with Nap110 or kalirin, etc).
Palmitoylation of NOS2 Is Not Involved in Caveolae Targeting-Because in the case of endothelial NOS, palmitoylation of Cys-15 and Cys-26 is necessary for its translocation to caveolae (6, 7), we isolated Triton X-100-insoluble domains enriched in caveolin-1 where acylated proteins are frequently found. Interestingly, in transfected COS7 cells neither the wild-type NOS2 nor the C3S mutant associated with Triton X-100-insoluble rafts (Fig. 5). A positive caveolin-1 immunodetection in the more buoyant fractions was indicative of caveolae, whereas ␤-cop appeared enriched in fractions with large non-caveolar membranes. We tested the putative caveolar targeting of our long and short wild-type NOS2-GFP chimeras. The majority of the wild-type NOS2-GFP appeared in the high density fractions (at the bottom of the tube) and a very limited amount of protein co-fractionates with caveolae (Fig. 5A, fractions 5-8), where a single degradation band was clearly observed. Fulllength mutant C3S was unable to associate with caveolae. The short GFP chimeras (amino acids 1-94) of both wild-type and C3S NOS2 were partially degraded, and only a very small fraction of the total wild-type NOS2 (but not the C3S mutant) could be observed in caveolae (Fig. 5B). This piece of data contrasts with the clear plasma membrane distribution of a significant population of the short wild-type NOS-GFP according to immunofluorescence data (see above) from which one might expect a significant caveolae enrichment. When live COS7 cells transfected with the full-length NOS2 construct were kept at 4°C and 1% cold Triton X-100 was added, the GFP fluorescence progressively vanished, reflecting that NOS2 localizes to Triton X-100-soluble regions. Hence, although NOS2 significantly associated with total intracellular membranes (Fig. 1C), the majority of these membranes are apparently not in rafts/caveolae enriched in cholesterol and sphingomyelin, which might be Triton X-100-insoluble (Fig. 5C), in agreement with the data obtained in the sucrose gradients (Fig. 5A). In fact, according to our immunofluorescence data, full-length and short NOS2-GFP chimeras colocalize with caveolin only partially in the plasma membrane and in intracellular areas of the Golgi apparatus (yellow staining in Fig. 5D).
Compounds That Alter the ER to Golgi Transit Interfere with the Activity of NOS2 and with Its Subcellular Localization-The possibility therefore exists that the proper sorting of NOS2 might be indispensable to reach complete activity. Hence, we tested next the effect of both brefeldin A and monensin on the NOS2 activity and subcellular distribution both in C2C12 mouse myotubes challenged with LPS/IFN-␥ and in COS7 cells transfected with a full-length NOS2-GFP chimera. Brefeldin A treatment, which inactivates Arf1, is known to lead to the dissociation of COPI and other peripheral proteins from Golgi membranes, resulting in Golgi enzymes redistributing to the ER as the Golgi structure disassembles (25). On the other hand, the ionophore monensin is commonly used to partially disrupt the integrity of the Golgi apparatus and to inhibit vesicular transport in eukaryotic cells (26). Induction of NOS2 expression in muscular myotubes followed by the addition of both brefeldin A and monensin drastically diminished the ⅐NO synthesis to 29 and 41% of the control levels, respectively (Fig.  6A). These alterations in the secretion pathways resulted in a dispersive phenotype of the Golgi apparatus accompanied by a clear vesicularization in the cytosol (Fig. 6A). Similarly, NOS2-GFP transfected in COS7 cells was also severely affected by treatment with brefeldin A and monensin (Fig. 6B), which lowered the amount of ⅐NO detected to 32 and 38%, respectively, of the control levels treated with the vehicle only. Likewise, these treatments affected the intracellular distribution of NOS2-GFP (Fig. 6B), with brefeldin A resulting in large perinuclear aggregates and monensin inducing the formation of dispersive "patches" of NOS2 throughout the cytosol.
The "Gain of Function" Mutant A2C/C3S That Restores a Potential Palmitoylation Site Partially Rescues the Wild-type Phenotype-Next, we created both a full-length and a NOS2- FIG. 5. Neither wild-type NOS2-GFP nor its C3S chimera appear significantly enriched in caveolae. A, the full-length wild-type and C3S NOS2-GFP chimeras were transfected in COS7 cells and caveolae were purified in a discontinuous sucrose gradient in the presence of 1% Triton X-100 at 4°C. After centrifugation, the tubes were equally divided into 12 fractions and each was immunoblotted against NOS2, caveolin-1, and ␤-cop (used here as a Triton X-100 soluble marker). B, the NOS2-(1-94)-GFP wild-type and its C3S mutant chimeras were also transfected in COS7 cells and caveolae were purified in a discontinuous sucrose gradient in the presence of 1% Triton X-100 at 4°C. C, live COS7 cells grown on coated glass coverslips transfected with the wild-type NOS2-GFP construct were submerged in medium at 4°C and then pre-chilled 1% Triton X-100 in phosphate-buffered saline was added. Photographs of the same field were taken every 10 s in a 10-min period. D, the subcellular distribution of both the full-length and NOS2-(1-94)-GFP constructs were analyzed by laser confocal microscopy in a triple staining with anti-caveolin-1 (red) and cell nuclei (blue).
(1-94) chimera that introduced a novel cysteine residue at position 2 (where an Ala residue is found in the original sequence), maintaining the Ser residue at position 3. Because the C3S mutant is completely devoid of activity and becomes jammed in the secretory pathway of the cell, we wanted to test if a novel Cys positioned at the N-terminal end of NOS2 could at least regain a wild-type phenotype to some extent. Both the long and short A2C/C3S GFP chimeras partially accumulated in perinuclear areas, although their phenotype is somehow between the WT and C3S mutant (Fig. 7A). In fact, the fulllength NOS2 A2C/C3S mutant displays 51% activity when compared with the WT polypeptide (Fig. 7B). When we added radiolabeled palmitic acid and immunoprecipitated the A2C/ C3S transfected in COS7 cells, a distinctive incorporation of the radioisotope could be observed (Fig. 7C). Hence, engineering a surrogate palmitoylable Cys residue at the N-terminal end of NOS2 renders a polypeptide that becomes palmitoylated and partially rescues the wild-type phenotype in terms of subcellular distribution and activity. DISCUSSION The major finding described in this work is the fact that inducible nitric-oxide synthase is S-acylated with palmitic acid in Cys-3 of its sequence. Palmitoylation occurs both in NOS2 transcriptionally induced in muscular myotubes insulted with proinflammatory stimuli as well as in transfected NOS2 and takes place through a hydroxylamine-sensitive thioester bond. However, when recombinant NOS2 was expressed in a bacterial expression system, the purified enzyme was not inhibited by levels of 8-Br-palmitate that had a profound effect when added to COS7 cells transfected with NOS2.
Unlike NOS3, where palmitoylation is a prerequisite for caveolar targeting, interaction with caveolin, and protein inactivation, NOS2 requires the acylation for its proper sorting along the secretory route after its initial synthesis and localization to the ER. Hence, the proper localization of the highly active NOS2 within the cells might be a tightly regulated process to avoid the toxic effects of ⅐NO. Alternatively, NOS2 might be targeted to certain subcellular areas with a large availability of its substrate L-Arg or its cofactor tetrahydrobiopterin.
Remarkably, substitution of Cys-3 for Ser resulted in the loss of fatty acid incorporation, apparent protein misfolding, and aggregation in subcellular membranes of the Golgi apparatus concomitant with a complete loss of NOS activity. Both basic amino acids and the hydrophobic Pro residue at position 4, which are in proximity of the palmitoylated Cys residue, seem to be determinant for the acylation process, because their mutation changed the subcellular distribution of NOS2 and eliminated its S-acylation as well.
Elegant studies performed by Felley-Bosco and co-workers (27,28) have proven that in Caco cells, only about 1% of the total NOS2 protein content was associated with caveolin-1 in caveolae, where it was targeted for proteasomal degradation. In fact, ectopic expression of caveolin-1 promoted NOS2 degradation at a single site on its N terminus (28). Our own studies have shown that, in C2C12 myotubes, NOS2 can also associate with caveolin-1, becoming inhibited, although the same stimuli that up-regulated NOS2 expression seemed to down-regulate caveolin-1 expression via the extracellular signal-regulated kinase pathway (12). The results described herein agree with a small amount of degraded NOS2 that associates with caveolin-1 in Triton X-100-insoluble fractions, although no significant differences could be observed when the wild-type and the C3S NOS2 constructs were compared. However, key differences can be observed when the wild-type protein and the C3S mutant were compared in terms of sorting along the ER to the cis-and eventually trans-Golgi network. In fact, 63% of the GFP of the C3S chimera colocalized with the cis-Golgi marker ␤-cop, and 95% with the Golgi-TGN marker BODIPY-Texas Red-ceramide. The behavior of this C3S chimera, together with its insolubility and perinuclear staining, strongly suggests that its proper subcellular sorting is impaired.
Precedents for the interconnection between palmitoylation and proper intracellular targeting of other non-myristoylated, N-terminal palmitoylated proteins such as GAP43, PSD-95, or G␣ S (29) is well established. For instance, brefeldin A treatment abrogates palmitoylation of GAP43, suggesting that palmitoylation occurs in the Golgi or trans-Golgi network (29).
It must be noted that the 70 N-terminal amino acids of NOS2 are known to interact with the membrane protein NAP110 (a protein of unknown function) as well as with kalirin (30,31). Because it is not established how NOS2 interacts with these two proteins during its sorting as well as the functional significance of these interactions, it remains to be confirmed how palmitoylation would affect binding to these two proteins.
When we created the "rescue mutant" A2C/C3S, we could restore, albeit partially, a wild-type phenotype. It is interesting, in this context, to remark that most of the non-myristoylated, N-terminal palmitoylated proteins possess the Cys residue at position 3, rather than 2. That is the case, for instance, of PSD95 (MDCLCIVT . . . ), PSD93␤ (MICHCKVA . . . ), or GAP43 (MLCCMRRT . . . ). Hence, it is tempting to speculate that, in proteins that become palmitoylated within the mammalian cell, the S-acylation at Cys-3 might be favored over the S-acylation at Cys-2.
Finally, when we characterized the phenotype of two of the short NOS2 chimeras, in particular the wild-type construct and the A2C mutant, the intense palmitoylation was accompanied by a significant plasma membrane localization. Remarkably, this localization was not associated with caveolae targeting, when inspected by Triton X-100 flotation experiments and dou-ble immunofluorescence studies. Consequently, our data indicate that the N-terminal end of NOS2 possesses in itself a potent palmitoylation and plasma membrane targeting sequence that allows the proper folding of the mature polypeptide. In consequence, the pharmacological inhibition of the palmitoyltransferases involved in NOS2 thioacylation might be a useful approach in the treatment of inflammatory diseases associated with increased NOS2 levels.