An Inducible Nitric-oxide Synthase (NOS)-associated Protein Inhibits NOS Dimerization and Activity*

A variety of transcriptional and post-transcriptional mechanisms regulate the expression of the inducible nitric-oxide synthase (iNOS, or NOS2). Although neurons and endothelial cells express proteins that interact with and inhibit neuronal NOS and endothelial NOS, macrophage proteins that inhibit NOS2 have not been identified. We show that murine macrophages express a 110-kDa protein that interacts with NOS2, which we call NOS-associated protein-110 kDa (NAP110). NAP110 directly interacts with the amino terminus of NOS2, and inhibits NOS catalytic activity by preventing formation of NOS2 homodimers. Expression of NAP110 may be a mechanism by which macrophages expressing NOS2 protect themselves from cytotoxic levels of nitric oxide.

Although NOS2 activity is independent of calcium concentrations, many other mechanisms regulate NOS2. NOS2 expression in macrophages is regulated in part at the transcriptional level; resting cells do not express NOS2, but a variety of external stimuli can trigger NOS2 transcription (reviewed in Ref. 5). Transcription factors that have been shown to interact with the NOS2 5Ј-flanking region and to activate NOS2 mRNA transcription include NF-B, IRF-1, Stat1-␣, and HIF-1 (6 -12). Transcription of NOS2 can also be repressed by unknown mechanisms following exposure to transforming growth factor-␤ or iron (13,14). NOS2 expression can also regulated at the post-transcriptional level by unknown mechanisms that influence NOS2 mRNA stability (15). Post-translational control of NOS2 activity occurs by several distinct mechanisms. transforming growth factor-␤ regulates NOS2 steady state protein levels (15). NOS2 is phosphorylated, although its biological significance is unclear (16). NOS2 activity can also be regulated by the availability of its substrates and co-factors in vitro, including arginine and tetrahydrobiopterin; however, it is unclear whether or not intracellular concentrations of arginine and tetrahydrobiopterin actually limit NOS2 activity in vivo (17)(18)(19)(20)(21). NO itself inhibits the activity of NOS2 as well (22)(23)(24). Although several proteins have been shown to interact with NOS1 and NOS3, regulating their location and activity, it is unknown whether similar proteins regulate NOS2 in macrophages.
Since NOS2 is a high output isoform of NOS that can produce cytotoxic levels of NO, we hypothesized that macrophages express proteins that interact with NOS2 and inhibit its activity. Screening a murine macrophage cDNA library, we found a previously identified protein of unknown function that directly interacts with NOS2 and inhibits NOS2 activity by preventing the formation of NOS2 homodimers.

EXPERIMENTAL PROCEDURES
Antibodies, cDNAs, Vectors, and Reagents-The following antibodies were used: a polyclonal antibody raised against the carboxyl-terminal amino acid residues of murine NOS2 (AKKGSALEEPKATRL) generated by us (25); and a polyclonal antibody raised against the carboxylterminal residues (CKDDGDSKDKKDDDEDM-SLD) of the NOS2associated protein-110 (NAP110), which was generated by us. Monoclonal antibody to mouse NOS2 and monoclonal antibody to endothelial NOS (eNOS) were purchased from Transduction Laboratories (Lexington, KY). The eukaryotic expression constructs, pCI-mu-NOS2 and pcDNA3.1-NAP110 encoding the full-length NOS2 and NAP110 proteins, respectively, were generated by us. The mouse NOS2 cDNA was cloned by us and introduced into pCI-neo expression vector (Promega, Madison, WI) (26). The cDNA for the human 110-kDa glycoprotein (566), which is highly homologous to mouse NAP110, was obtained as a gift from Dr. S. Shimada (Department of Surgery, Kumamoto University Medical School, Kumamoto, Japan). Standard cloning methods were used to introduce the EcoRI fragment (1,375 base pairs (bp)) of the 110-kDa glycoprotein cDNA into the multiple cloning site of the pcDNA3.1 vector (Invitrogen), generating the pcDNA3.1-NAP110 expression cassette. LPS was obtained from Sigma. Interferon-␥ (IFN-␥) was a generous gift from the American Cancer Society.
Cell Culture and Transfections-The RAW 264.7 mouse macrophage tumor cells and HeLa cells were obtained from the American Tissue and Cell Collection (ATCC, Rockville, MD). RAW 264.7 cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Mouse bone marrow-derived primary macrophages were isolated from female C57BL/6 mice (F2 wild type, from Jackson Laboratories, Bar Harbor, ME). After harvest, cells were cultured for 5 days in DMEM supplemented with 10% heat-inactivated low LPS FBS, 15% filter-sterilized L-cell conditioned medium, and 1% penicillin-streptomycin/1% glutamine (Biofluids, Rockville, MD) at 37°C under 8% CO 2 . Thioglycolate-elicited peritoneal primary macrophages were lavaged with cold phosphatebuffered saline from female C57BL6 mice 4 days after intraperitoneal injection of 2 ml of sterile 0.3 mM thioglycolate (Sigma). Peritoneal cells were resuspended in RPMI, 10% FBS and plated into 100-mm Petri dishes. The cells adhered overnight in RPMI 1640 medium supplemented with 10% low LPS FBS and 1% penicillin-streptomycin/1% glutamine before use. To exclude effects of contaminating LPS on experimental conditions, resting cell culture was carried out in the presence of 10 g/ml polymyxin B (Calbiochem).
Induction of NOS2 in cultured cells (mouse RAW 264.7 tumor macrophages or mouse primary bone marrow macrophages or mouse thioglycolate-elicited peritoneal macrophages) was performed using IFN-␥ (20 units/ml) and LPS (1 g/ml) for 16 h.
We also used human HeLa cells that were transiently transfected with pCI-mu-NOS2 or pcDNA3.1-NAP110 or combinations thereof. Cells were maintained in minimal essential medium supplemented with 10% FBS (Gemini, Calabasas, CA), penicillin, and streptomycin. The recombinant DNA was introduced into HeLa cells with aid of LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol. Briefly, 2.5 ϫ 10 6 cells were plated in 100-mm tissue culture dishes 1 day prior to transfection. The cells were washed twice with Opti-MEM (Life Technologies, Inc.) and incubated for 4 h with 1 ml of Opti-MEM containing 5 g of DNA and 1 l of LipofectAMINE. The DNA/LipofectAMINE solution was then aspirated, 3 ml of Opti-MEM was added, and the cells were incubated overnight at 37°C, then collected and analyzed.
For the detection of NOS2 and NAP110 complexes in vivo, female C57BL6 mice were injected with IFN-␥ (500 units/mice) and LPS (500 g/mice) intraperitoneally, and after 18 h tissue specimens (spleen and peritoneal primary macrophages) were collected, homogenized, and subsequently analyzed.
NOS Assay-The NOS assay was performed as described previously (28). Cells were sonicated in 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.1 mM tetrahydrobiopterin, 2 mM dithiothreitol, 10% (v/v) glycerol, aprotinin (25 g/ml), leupeptin (25 g/ml), 100 M phenylmethylsulfonyl fluoride, 10 M FMN, and 10 M FAD. A reaction mix was made of cell lysate containing 50 g of protein, 50 l of [U-14 C]arginine (50,000 cpm), 5 l of 5 mM FAD, 5 l of 100 M tetrahydrobiopterin, 1 l of 30 M calmodulin, and the volume was brought up to 250 l with 50 mM Tris-HCl, pH 7.4, 1 mM EDTA. The reaction was initiated by addition of 50 l of 10 mM NADPH, and incubated for 60 min at 37°C, applied to a Dowex-50W resin (Hϩ form) column, and the flow-through was counted in a liquid scintillation counter.
Yeast Two-hybrid Screen-We employed a yeast two-hybrid system to screen for potential binding partners of NOS2 (29). First, we constructed a cDNA library from mouse bone marrow-derived primary macrophages, which were stimulated with LPS and IFN-␥. Mouse bone marrow-derived primary macrophages were isolated from female C57BL/6 mice (F2 wild type, from Jackson Laboratories) and cultured (11 ϫ 10 6 cells/dish) for 5 days in DMEM supplemented with 10% low LPS FBS, 15% filter-sterilized L-cell conditioned medium, and 1% penicillin-streptomycin, 1% glutamine at 37°C under 8% CO 2 . Cells then were stimulated with IFN-␥ (20 units/ml) and LPS (1 g/ml) for 16 h. RNA was prepared from stimulated macrophages. cDNA prepared using the Moloney murine leukemia virus reverse transcriptase and oligo(dT) primer was inserted into the Hybri-Zap-Gal4-AD vector (Stratagene, La Jolla, CA). The resultant Gal4-AD library plasmids encoded fusion proteins consisting of the Gal4 activation domain and random macrophage polypeptides. The titer of the primary unamplified twohybrid cDNA library was estimated as of 1.6 ϫ 10 6 plaque-forming units. The two-hybrid cDNA library was first excised from phage in presence of ExAssist filamentous phage (Stratagene). Then the library was amplified by introduction into XLOLR Escherichia coli cells to a final titer of 3.3 ϫ 10 9 colony-forming units. The insert size range of cDNA in the library was 400 -1,300 bp. The final amplified DNA from the two-hybrid cDNA library was stored at Ϫ20°C, and the original phage two-hybrid cDNA library was stored at Ϫ86°C in the presence of 7% dimethyl sulfoxide and 0.3% chloroform.
To prepare bait plasmids, murine NOS2 cDNA sequences (bp 1-549 or 1-210) were prepared by the PCR and inserted 3Ј to the cDNA for the Gal4 binding domain of plasmid pGal4-BD (Stratagene). The resultant pGal4-BD-NOS2 bait plasmids encoded a fusion protein comprised of the Gal4 binding domain and an amino terminus fragment of NOS2 (amino acid residues 1-183 or 1-70). The identity of the cDNA encoding the fusion protein was determined by sequencing.
YRG-2 yeast cells were co-transformed with pGal4-BD-NOS2 and the HybriZap-Gal4-AD cDNA library, and grown on a selective medium lacking tryptophan or leucine or histidine or combinations thereof. Colonies that contained cDNA encoding target library proteins interacting with the bait fusion protein were identified by transcription of the HIS3 and lacZ genes. A total of 1.4 ϫ 10 6 yeast transformants were placed under selection. Plasmids from ␤-galactosidase-positive yeast TABLE I Specific interaction between NOS2 and NAP110 in the yeast two-hybrid assay S. cerevisiae (strain SFY526) were co-transformed with various combinations of bait plasmids (pGal4-BD expressing a fusion protein of the Gal4 binding domain and portions of NOS2) and target plasmids (pGal4-AD expressing a fusion protein of the Gal4 activation domain and portions of NAP110). Yeast were grown on agar plates lacking tryptophan (Trp); leucine (Leu); tryptophan and leucine; or tryptophan, leucine, and histidine (His); and then were assayed for ␤-galactosidase activity by liquid culture assay method using ONPG as a substrate (nmol ONPG cleaved/min/mg protein measured at OD 420 ) (3). Assays were performed in triplicate Ϯ S.D.

Bait
Target colonies were isolated and retransformed into competent E. coli DH5␣ cells.
For the quantitative liquid ␤-galactosidase assay, Saccharomyces cerevisiae (strain SFY526) were co-transformed by the lithium acetate method with various combinations of bait plasmid (pGal4-BD expressing a fusion protein of the Gal4 binding domain and portions of NOS2) and target plasmids (pGal4-AD expressing a fusion protein of the Gal4 activation domain and portions of NAP110). Transformants were plated on selective medium lacking tryptophan and leucine. After 3 days at 30°C, cells were assayed for ␤-galactosidase activity using o-nitrophenyl-␤-D-galactopyranoside (ONPG) as a substrate (nmol of ONPG cleaved/min/mg of protein measured at OD 420 ).
Tunicamycin Treatment-To evaluate the effect of glycosylation on NOS2/NAP110 complex formation, RAW 264.7 mouse tumor macrophages or HeLa cells were grown as described above for 20 h in the absence or presence of 2 g/ml tunicamycin, an inhibitor of N-glycosylation (30). RAW 264.7 macrophages were stimulated with LPS and IFN-␥ as indicated above. Cell lysates were then resolved by a denaturing gel electrophoresis following by immunoblotting with antibody to NOS2 or antibody to NAP110. For co-precipitation, precleared cell lysates from stimulated RAW 264.7 macrophages were precipitated with a polyclonal antibody to NAP110 followed by immunoblotting with monoclonal antibody to NOS2 or with a monoclonal antibody to NOS2 followed by immunoblotting with a polyclonal antibody to NAP110.

RESULTS
We used the yeast two-hybrid system to search for proteins in murine macrophages that associate with NOS2. Yeast expressing a fusion protein consisting of the Gal4 DNA binding domain and the amino terminus of NOS2 (amino acid residues 1-183) were transformed with plasmids encoding fusion proteins consisting of the Gal4 activation domain and random polypeptides from mouse primary bone marrow macrophages stimulated with IFN-␥/LPS. The amino-terminal region of NOS2 was selected as a potential target of interacting proteins since it is not homologous to the other NOS isoforms. Screening of 1.4 ϫ 10 6 yeast transformants revealed 18 candidate proteins that might interact with NOS2, including: NOS2 (7 individual clones), cytochrome c oxidase (2 clones), GATA-3 tran-FIG. 1. Endogenous NOS2 and NAP110 interact in a macrophage cell line. RAW 264.7-transformed murine macrophages were incubated with media alone (Ϫ) or with LPS and IFN-␥ (ϩ). Cell lysates were fractionated by SDS-PAGE and immunoblotted with antibody to NOS2 or with antibody to NAP110 (left panels). For co-immunoprecipitation (IP), precleared cell lysates were precipitated with a monoclonal antibody to NOS2 followed by immunoblotting with a polyclonal antibody to NAP110, or lysates were precipitated with a polyclonal antibody to NAP110 followed by immunoblotting with a monoclonal antibody to NOS2 (right panels). Arrows indicate positions of NOS2 or NAP110.

FIG. 2. Endogenous NOS2 and NAP110 interact in primary macrophages.
Mouse peritoneal primary macrophages were incubated with media alone (Ϫ) or with LPS and IFN-␥ (ϩ). A, cell lysates were fractionated by SDS-PAGE and immunoblotted with antibody to NOS2 or with antibody to NAP110. B, for co-immunoprecipitation (IP), precleared cell lysates were precipitated with a monoclonal antibody to NOS2 followed by immunoblotting with a polyclonal antibody to NAP110, or lysates were precipitated with a polyclonal antibody to NAP110 followed by immunoblotting with a monoclonal antibody to NOS2. Arrows indicate positions of NOS2 or NAP110.
FIG. 3. Physiological complexes between NOS2 and NAP110 in vivo. C57BL/6 wild type mice were injected intraperitoneally with buffer alone (Ϫ) or with 500 g of IFN-␥ and 500 g of LPS (ϩ). After 24 h, spleens and peritoneal macrophages (PM) were harvested, and lysates were prepared, fractionated by SDS-PAGE, and analyzed with antibody to NOS2 or with antibody to NAP110 (left and center panels). For co-immunoprecipitation (IP), precleared cell lysates were precipitated with antibody to NAP110 followed by immunoblotting with antibody to NOS2 (right panel). scription factor, ␣ 2 -macroglobulin, ␥-actin, Tum-P91A antigen, and SPT-5 transcription initiation factor (one clone each) (Table I). Four positive yeast clones contained overlapping DNA fragments highly homologous to a human 110-kDa glycoprotein, which we now refer to as NOS2-associated protein-110 kDa (NAP110) (31).
In order to define more precisely the regions of NOS2 and NAP110 that interact, yeast expressing a NAP110 fusion protein (consisting of amino acid residues 349 -401 of NAP110 fused to the Gal4 activation domain) were transformed with various plasmids expressing different domains of NOS2 (residues 1-183, 1-70, and 71-220). Analysis of transformed yeast revealed that residues 1-70 of NOS2 are capable of interacting with NAP110 in yeast (Table I). Moreover, residues 363-401 of NAP110 are sufficient to interact with residues 1-70 of NOS2 (Table I).
A search of the GenBank data base revealed that the mouse NAP110 protein is highly homologous to an 110-kDa glycoprotein (70% identity at the amino acid level), which was cloned from an human gastric carcinoma cell line (31). NAP110 is a glycosylated acidic protein (pI ϭ 4.09), which consists of 407 amino acid residues. Its expression in some human gastric carcinoma cells is increased by IFN-␥ (31). Its function was previously unknown.
NAP110 Interacts with NOS2 in Mammalian Cells and Mice-We next examined the ability of NOS2 and NAP110 to interact in mammalian cells. We generated a polyclonal antibody raised against the carboxyl-terminal residues 381-401 of mouse NAP110 (CKDDGDSKDKKDDDEDMSLD), which were identical to the carboxyl-terminal residues of human 110-kDa glycoprotein (31). Next, we prepared cell lysates from RAW 264.7 macrophages that were stimulated or not with LPS/ IFN-␥. NOS2 is absent from resting cells but is expressed in RAW cells that are treated with LPS and IFN-␥, as detected by immunoblotting (Fig. 1). NAP110 is expressed in resting RAW cells at a low level, and its expression is significantly higher in stimulated cells (Fig. 1). NAP110 and NOS2 interact in RAW cells that are stimulated with LPS and IFN-␥ but not in resting ones, as shown by co-precipitation (Fig. 1).
We then examined the association of NOS2 and NAP110 in primary macrophages isolated from thioglycolate-stimulated mice. These macrophages only express NOS2 after stimulation with LPS/IFN-␥, but in contrast to RAW cells, they express NAP110 in both conditions, resting and stimulated ( Fig. 2A). Such NAP expression in resting primary macrophages may be due to the thioglycolate pretreatment of mice during macrophage isolation (see "Experimental Procedures"). As shown, NOS2 and NAP110 interact in primary macrophages (Fig. 2B).
Then we analyzed the expression of NOS2 and NAP110 in the spleen and in peritoneal macrophages obtained from untreated mice and mice which were injected with IFN-␥ and LPS. NOS2 protein is expressed in spleen and in peritoneal macrophages of mice treated with IFN-␥ and LPS; untreated mice normally do not express NOS2 (Fig. 3). NAP110 protein is expressed in resting mouse spleen and macrophages, and IFN-␥ and LPS increase NAP110 expression in vivo (Fig. 3). Physiological complexes of endogenous NAP110 and endogenous NOS2 were detected by co-precipitation in extracts obtained from the spleen and peritoneal macrophages of treated mice, but were not detected in extracts obtained from the spleen and peritoneal macrophages of untreated mice (Fig. 3). More than 20% of the endogenous NAP110 and NOS2 are involved in complex formation. Specificity of NAP110 Interaction with NOS Isoforms-To test the specificity of the NAP110 interaction with various NOS isoforms, we analyzed NAP110 interaction with eNOS in the spleen of untreated and IFN-␥/LPS-treated mice. eNOS is expressed in the spleen of untreated and treated mice (Fig. 4). These same lysates contain the NAP110 protein (Fig. 3). However, eNOS and NAP110 do not interact in the mouse spleen, as shown by co-immunoprecipitation (Fig. 4). Thus, NAP110 interacts only with NOS2 and not with the eNOS isoform.
Since the protein product of the NAP110 gene is heavily glycosylated (NAP without glycans has a molecular mass of 44 kDa versus 110 kDa for the glycoprotein), we evaluated whether or not the glycans of NAP110 play a role in its interaction with NOS2. Tunicamycin was added to stimulated RAW cells to block N-glycosylation (30). NAP protein produced under these circumstances has a molecular mass of 44 kDa, presumably because it is not glycosylated (Fig. 5A). Deglycosylated NAP is still capable of forming physiological complexes with NOS2 (Fig. 5B).
NAP110 Inhibits NOS2 Activity-We next examined the effect of NAP110 upon NOS2 activity. NOS activity was measured in HeLa cells transfected with expression vectors for NOS2, NAP110, or with combinations thereof. HeLa cells transfected with NOS2 alone synthesize NO (Fig. 6A, lane 2). However, co-expression of NAP110 reduces NOS2 activity by more than 90% (Fig. 6A, lane 4). Since expression of NAP110 does not reduce the amount of NOS2 protein expressed in response to IFN-␥/LPS treatment (Fig. 6B, lanes 2 and 4), we explored other mechanisms by which NAP110 could inhibit NOS2 activity.
NAP110 Associates with NOS2 and Inhibits NOS2 Homodimerization-We hypothesized that NAP110 inhibits NOS2 activity by interfering with NOS2 homodimerization, which is required for NOS2 activity (32)(33)(34)(35). To test this hypothesis, cell lysates were prepared from HeLa cells trans-FIG. 5. Tunicamycin, an inhibitor of N-glycosylation, does not affect the interaction between NOS2 and NAP110. RAW 264.7 macrophages were stimulated with LPS and IFN-␥, and grown in the absence (Ϫ) and in the presence (ϩ) of 2 g/ml tunicamycin for 20 h. A, cell lysates were resolved by SDS-PAGE followed by immunoblotting with antibody to NOS2 or with antibody to NAP110. B, for co-precipitation (IP), RAW cells were stimulated as above, and precleared cell lysates were precipitated with a polyclonal antibody to NAP110 followed by immunoblotting with monoclonal antibody to NOS2, or lysates were precipitated with a monoclonal antibody to NOS2 followed by immunoblotting with a polyclonal antibody to NAP110. Arrows indicate positions of NOS2 or NAP110 or non-glycosylated NAP. fected with expression vectors for NOS2, NAP110, or both, and these lysates were fractionated on a gel filtration column; aliquots were then electrophoresed on a denaturing gel and then immunoblotted. In cells expressing NOS2 alone, about half of the NOS2 exists as homodimers (eluting in fractions 11-18 at approximately 300 -400 kDa) and about half as monomers (eluting fractions 23-29 at approximately 140 kDa) (Fig. 7A). No NAP110 protein is detected in cells transfected only with the NOS2 expression vector (Fig. 7B). In contrast, expression of NAP110 together with NOS2 shifts the elution profile of NOS2 from two peaks at approximately 140 and 300 -400 kDa (Fig.  7A) to one peak at approximately 300 -400 kDa (Fig. 7C). Immunoblotting with antibody to NAP110 detected NAP110 in the same eluted fractions (Fig. 7D).
We hypothesized that this new elution profile of NOS2 was due to the formation of NAP110/NOS2 heterodimers. However, it is difficult to distinguish NAP110/NOS2 heterodimers (mass 110 ϩ 135 kDa) from NOS2 homodimers (mass135 ϩ 135 kDa) on the basis of relative mobility since their masses are similar. To solve this problem, we decided to remove the N-glycan component from NAP110 before the size exclusion chromatography. HeLa cells were transfected with both expression vectors for NAP110 and NOS2, and then were cultured for 20 h in presence of tunicamycin (2 g/ml), which inhibits N-glycosylation (30). Cell extracts were subjected to Superdex-200 size exclusion chromatography and immunoblotted. Deglycosylation of NAP shifts the elution profile of NOS2 from approximately 300 -400 kDa (Fig. 7C) to approximately 200 kDa in fractions 18 -26 (Fig. 7E). NAP co-elutes with the NOS2 in fractions 18 -25 (Fig. 7F). Thus, a reduction in the molecular weight of NAP changes the retention time for NOS2/NAP complexes. This shift in the NOS2 elution profile suggests that NOS2 forms heterodimers with NAP110.
A small amount of rapidly eluting NOS2 is present in fractions 10 -13, and may be NOS2 homodimers (Fig. 7E). This amount of NOS2 homodimers in cells expressing NOS2 and NAP110 is substantially less than in cells expressing NOS2 alone. Thus, NAP110 reduces the amount of NOS2 homodimer formation.
NAP110 Associates with NOS2 Monomers-To see if NAP110 only associates with NOS2 monomers, we mixed cell extracts containing NAP with NOS2 monomers or NOS2 homodimers (which had been isolated by column chromatography, as in Fig. 7A). Mixtures of NAP110 and NOS2 dimers, and mixtures of NAP110 and NOS2 monomers both contain NAP110 and NOS2, as verified by immunoblotting (Fig. 8).
Co-precipitation shows that NOS2 was detected in the NAP110 immunoprecipitates only when NAP110 containing extracts were mixed with NOS2 monomers (Fig. 8). Thus NAP110 interacts with NOS2 monomers, but does not interact with NOS2 homodimers.
To confirm that NAP only interacts with NOS2 monomers, we examined the effect of NAP upon NOS2 when these proteins are synthesized in separate cells. Lysates were prepared from HeLa cells expressing NAP110 or expressing NOS2 or expressing both. Pairs of these lysates were mixed together, and analyzed for NOS2 activity. NAP110 inhibits NOS2 activity when expressed in the same cell (Fig. 6A, lane 7) but not when NAP110 is expressed separately and then mixed with NOS2 (Fig. 6A, lane 6). NOS and NAP110 interact when co-expressed in the same cell (Fig. 6D, lanes 4 and 7). NOS and NAP110 also interact when expressed separately and then are mixed (Fig.  6D, lane 6), possibly because NAP110 is interacting with NOS2 monomers.
Effect of Tunicamycin on NOS2 Activity in Cells-Since deglycosylated NAP110 interacts with NOS2 (Figs. 5 and 7), it is possible that deglycosylated NAP also inhibits NOS2 activity. We treated LPS-and IFN-stimulated RAW macrophages with tunicamycin, and measured NOS2 activity. Tunicamycin slightly decreases NOS2 activity in stimulated macrophages (Fig. 9). However, tunicamycin also slightly decreases NOS2 activity when added directly to stimulated macrophage extracts (Fig. 9). Perhaps tunicamycin has a mild inhibitory effect directly upon NOS2. Since tunicamycin inhibits glycosylation of NAP (Fig. 5), but tunicamycin does not affect the interaction between NOS2 and NAP (Figs. 5 and 7), and tunicamycin only slightly affects NOS2 activity, therefore these data suggest that deglycosylated NAP can bind to NOS2 and inhibit its activity. DISCUSSION We have shown that a glycoprotein expressed in murine macrophages inhibits NOS2 activity. This protein, which we  4). Lysates of these cells were assayed individually for NOS2 activity (lanes 1-4). Lysates of these cells were also mixed together and then assayed for NOS2 activity as follows: lysate of cells transfected with vector expressing NOS2 plus lysate of cells transfected with empty vector alone (lane 5); lysate of cells transfected with vector expressing NOS2 plus lysate of cells transfected with vector expressing NAP110 (lane 6); lysate of cells co-transfected both with vector expressing NAP110 and vector expressing NOS2 plus lysate of cells transfected with empty vector alone (lane 7). For individual cell lysates in lanes 1-4, 50 g of cell lysate was assayed for the conversion of arginine to citrulline. For mixed cell lysates in lanes 5-7, the reaction mixture contained 50 g of lysate from cells expressing NOS2 alone or NOS2 plus NAP110, mixed with 50 g of cell lysates transfected with vector expressing NAP110 or empty vector. Control samples had negligible amounts of NOS activity and are not visible on the chart. (n ϭ 3 Ϯ S.D.). B, immunoblot with antibody to NOS2 (same samples as above in A). C, immunoblot with antibody to NAP110 (same samples as above in A). D, co-precipitation of NOS2/NAP110 complexes with a polyclonal antibody to NAP110, followed by immunoblotting with a monoclonal antibody to NOS2 (same samples as above in A). Arrows indicate positions of NOS2 or NAP110.
have named NAP110, interacts with the amino terminus of NOS2, a region of NOS2 that is not homologous to the other NOS isoforms. NAP110 inhibits NOS2 activity by combining with NOS2 monomers to form NAP110/NOS2 heterodimers, thus preventing the formation of NOS2 homodimers. However, NAP110 is incapable of converting active NOS2 homodimers into inactive NOS2/NAP110 heterodimers.
Various domains of the NOS2 protein are involved in NOS2 homodimerization. Stuehr (34,38) has shown that the oxygenase domain of NOS2 contains all of the structural determinants necessary to form dimers, since the oxygenase domain can form homodimers and the reductase domain cannot. (In contrast, Venema (40) has shown that both the oxygenase and reductase domains of NOS1 and NOS3 are necessary for homodimerization.) Mutational analysis has emphasized the importance of various portions of the oxygenase domain in forming NOS2 homodimers, including amino acid residues 66 -114, 242-335, and 450 and 453 (41)(42)(43).
Analysis of crystal structures of residues 66 -498 of NOS2 complement the functional studies of NOS2 (44). Several hydrophobic domains of NOS2 refold to surround the arginine and tetrahydrobiopterin molecules in NOS2 homodimers, including the helical lariats (residues 454 -459 and 463-471) and helical T domains (residues 401-453). At least a portion of the amino terminus of NOS2, residues 77-100, interact with the helical lariat and helical T domain of the other subunit. Since the residues 1-76 were not included in the NOS2 oxygenase fragment that was crystallized, it is possible that these aminoterminal residues contribute to NOS2 dimer interactions as well. Our data show that NAP110 interacts with the aminoterminal region of NOS2 from residues 1-70. Perhaps NAP110 prevents NOS2 homodimerization by sterically hindering interactions of adjacent NOS2 domains such as residues 77-100. Although NAP may prevent NOS2 homodimerization by blocking residues 1-70 from participating in dimerization, this possibility is less likely, since NOS2 can form homodimers without these amino-terminal residues.
Proteins that interact with NOS isoforms can be divided into two groups, proteins that regulate NOS activity, and proteins that regulate NOS location, although some proteins serve both functions. The activity of NOS1 is regulated by direct interaction with dynein light chain (45). Dynein light chain interacts with residues 163-245 of NOS1 and converts NOS1 homodimers into monomers. The activity of NOS3 is regulated in part by its interaction with caveolin-1 in endothelial cells and caveolin-3 in myocytes (46 -48). Caveolin interacts with both the oxygenase and the reductase domains of NOS3 (49). (Caveolin can also inhibit NOS1 as well (46,50).) NOS3 also reversibly interacts with the bradykinin B2 receptor, which inhibits eNOS activity (51). We recently reported that a protein expressed in neurons, kalirin, can interact with NOS2, maintaining it in the inactive monomeric form as well (27). Kalirin interacts with the same amino terminus of NOS2 as NAP110. Thus, a heterogeneous group of proteins inhibits the activity of different NOS isoforms by a similar mechanism, maintaining NOS as inactive monomers.
Only two proteins have been shown to stimulate NOS activity. Calmodulin was the first protein shown to interact with all NOS isoforms, and this interaction is necessary for NOS activity (4,28,52). Recently, HSP-90 was shown to interact with NOS3 and increase NOS3 activity by unknown mechanisms (53).
Another set of proteins interacts with NOS isoforms and localizes them within specific subcellular compartments. For example, interaction with the dystrophin glycoprotein complex targets NOS1 to the plasma membrane of skeletal muscle (54). The post-synaptic density proteins 93 and 95 (PSD-93 and PSD-95) localize NOS1 to the synapse of neurons, by an interaction between PDZ domains of PSD and PDZ domains of the NOS1 amino terminus (55,56). CAPON interacts with the PDZ domain of NOS1 through the CAPON carboxyl-terminal domain (57). CAPON may affect the location of NOS1 by competing with PSD for binding to NOS1. Caveolin not only regulates the activity of NOS3, but also localizes NOS3 to caveolae, invaginations in the surface of endothelial cells and cardiac myocytes (48,58).
The amino-terminal domains of NOS isoforms are targets for interactions with other proteins. PSD, syntrophin, CAPON, and dynein light chain interact with the amino terminus of NOS1, caveolin interacts with the amino terminus of NOS3, and Kalirin interacts with the amino terminus of NOS2. Since NOS isoforms are highly homologous except in their amino termini, interactions with the unique amino-terminal domains of NOS isoforms permit regulation and localization of specific NOS isoforms.
Although our data suggest that NAP110 inhibits the activity of NOS2 in macrophages, many questions about the interaction of NAP and NOS2 are not unanswered. If NO production by NOS2 is necessary for the killing of pathogens, then it is unlikely that NAP110 always inhibits NOS2 at all subcellular locations. For example, NAP110 may inhibit NOS2 in specific FIG. 8. NAP110 interacts with NOS2 monomers but not NOS2 homodimers. HeLa cells were transfected separately with vector expressing NOS2 or vector expressing NAP110. Cells expressing NOS2 were lysed and fractionated by column chromatography as in Fig. 7. Fractions containing NOS2 monomers (M) or NOS2 homodimers (D) were mixed with lysate from cells expressing NAP110. Entire lysate mixtures were fractionated by a denaturing gel electrophoresis and probed with antibody to NAP110 or with antibody to NOS2 (left and center panels). Lysate mixtures were also precipitated (IP) with antibody to NAP110, fractionated by SDS-PAGE, and probed with antibody to NOS2 (right panel).
FIG. 9. Deglycosylated NAP110 inhibits NOS2. RAW 264.7 macrophages were stimulated or not with LPS and IFN-␥, and grown in the absence or in the presence of 2 g/ml tunicamycin for 20 h. Extracts of cells were assayed for NOS activity. To test for a direct effect of tunicamycin upon NOS2, tunicamycin was added to aliquots of stimulated macrophage extracts as well (gray bar) (n ϭ 3 Ϯ S.D.). subcellular compartments of the macrophage where NO might be cytotoxic, but NAP110 may be absent in locations where NO could be antimicrobial, such as the cell surface. Possibly NAP110 may serve as a chaperone that delivers inactive NOS2 to specific macrophage locations, and then dissociates from NOS2 and allows active NOS2 homodimers to form. Alternatively, NAP110 may limit the amount of time that NOS2 is active; perhaps NAP110 is expressed at specific times after NOS2 is expressed, in order to limit the time which active NOS2 is synthesized. Finally, the role of the glycans attached to NAP110 is unclear. Tunicamycin does not affect the interaction between NAP110 and NOS2, and tunicamycin also does not affect NOS2 activity. Therefore, the glycans attached to NAP110 do not contribute to the inhibitory effect of NAP110 upon NOS2.
In summary, we have shown that a novel protein interacts with NOS2 and inhibits its activity by preventing the formation of active NOS2 homodimers. The physiological role of this interaction has not yet been established.