Interaction of Neuronal Nitric-oxide Synthase with α1-Syntrophin in Rat Brain*

Neuronal nitric-oxide synthase (nNOS) has a PSD-95/Dlg/ZO-1 (PDZ) domain that can interact with multiple proteins. nNOS has been known to interact with PSD-95 and a related protein, PSD-93, in brain and with α1-syntrophin in skeletal muscle in mammals. In this study, we have purified an nNOS-interacting protein from bovine brain using an affinity column made of Sepharose conjugated with glutathione S-transferase-rat nNOS fusion protein and identified it as α1-syntrophin by microsequencing. Immunostaining of primary cultures of rat embryonic brain neuronal cells with antibodies against these proteins showed that nNOS and α1-syntrophin were colocalized in neuronal cell bodies and neurites. Immunohistochemical analysis indicated that the nNOS- and α1-syntrophin-like immunoreactive substances were highly expressed in the rat hypothalamic suprachiasmatic nucleus (SCN) and paraventricular nucleus. In the SCN, nNOS- and α1-syntrophin-like immunoreactive substances were colocalized in the same neurons as detected by confocal microscopy. These results indicate that nNOS in brain interacts with α1-syntrophin in specific neurons of the SCN and paraventricular nucleus and that this interaction might play a physiological role in functions of these neurons.

Nitric oxide is a major endogenous mediator involved in many physiological and pathological functions such as vasodilation, neurotransmission, and cytotoxicity (1)(2)(3). In brain, NO is synthesized mainly by nNOS 1 (4), which is expressed in various brain regions including the cerebellum, olfactory bulb, and several hypothalamic nuclei (5). One of the nNOS-positive nuclei in the hypothalamus is the suprachiasmatic nucleus (SCN) (6), which has a circadian oscillator to create circadian rhythms in hormonal secretions, enzyme activities, and behaviors. The SCN also controls energy metabolism through the regulation of the autonomic nervous system (7). We have previously shown that N G -methylarginine, an inhibitor of NOS, disturbs the circadian rhythm of drinking behavior in rats, suggesting that NO is involved in the generation and/or synchronization of the circadian rhythm (8).
nNOS is one of three known isoforms of nitric-oxide synthase. Although nNOS does not have a transmembrane domain, subcellular fractionation experiments showed that ϳ60% of the total NOS activity in brain was found in the particulate fraction, suggesting that nNOS is associated with membranes by interacting with some other membrane proteins (9). The N-terminal domain of nNOS is unique to this isoform, having a PDZ motif, which is found in various structural proteins (10). This domain of nNOS is reported to interact with PDZ motifs in PSD (postsynaptic density)-95 and PSD-93 (11) to form macromolecular signaling complexes at postsynaptic sites and possibly to modulate synaptic transmission.
nNOS is expressed not only in neuronal cells, but also in several other tissues such as the fast-twitch fibers of skeletal muscle (12). In skeletal muscle, nNOS is targeted to sarcolemmal membranes by association with another PDZ-containing protein, ␣1-syntrophin, through PDZ-PDZ interactions (11). The syntrophins are a multigene family of proteins including ␣1, ␤1, and ␤2 isoforms, each of which has one PDZ domain and three pleckstrin domains (13). In mammalian skeletal muscle, syntrophins are components of the dystrophin complex at sarcolemmal membranes, and are thought to function as adaptors that recruit signaling proteins to the membranes (14). ␣1-Syntrophin is also expressed in brain (15). However, the interaction of nNOS with ␣1-syntrophin in brain has not been precisely investigated yet.
To examine whether the PDZ domain of nNOS interacts with proteins other than PSD-95 in brain, we purified nNOS-interacting proteins from bovine brain lysate. In this report, we show that one of the nNOS-interacting proteins in brain is ␣1-syntrophin. We also investigated the localization of nNOS and ␣1-syntrophin in primary cultured neurons from rat brain and in neurons from the hypothalamus to gain insight into the physiological functions of nNOS and nNOS-associated proteins in the central regulation of metabolism.
Purification and Microsequencing of an nNOS-interacting Protein-Purifications of nNOS-interacting proteins were carried out at 0 -4°C. Bovine brain (500 g) was homogenized in 5 volumes (w/v) of TNE buffer (25 mM Tris-HCl, 1% Nonidet P-40, and 1 mM EDTA, pH 7.4) containing 0.5 M NaCl and centrifuged at 15,000 ϫ g for 30 min. The supernatant was loaded on a GST or GST-nNOS fusion protein column. Each column was washed with 20 volumes of TNE buffer containing 0.5 M NaCl, and bound proteins were eluted by addition of 10 mM glutathione. Eluted proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE), electroblotted onto a polyvinylidene difluoride membrane, and stained with Ponceau S. An nNOS-interacting protein was digested with a lysylendopeptidase, Achromobacter protease I, and the fragments yielded were separated by reversed-phase HPLC (16). Sequences of three of the fragments (AP-1, AP-2, and AP-3) were determined by peptide microsequencing (17).
Tissue Extraction and Western Blot Analysis-Rat brains (3.0 g) were homogenized in 10 volumes (w/v) of TNE buffer containing 150 mM NaCl and centrifuged at 15,000 ϫ g for 30 min. The supernatant was resolved by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were blocked overnight with Tween/TBS and incubated with primary antibody and then with horseradish peroxidaseconjugated anti-rabbit IgG (Zymed Laboratories, Inc., South San Francisco, CA). The resultant immunoreactive bands were visualized by enhanced chemiluminescence (Renaissance ® , NEN Life Science Products) according to the specifications of the manufacturer.
Protein Overlay Assays-Rat brain extracts were separated by SDS-PAGE using 7.5% SDS-polyacrylamide gels and transferred to nitrocellulose membranes, which were then blocked with Tween/TBS and incubated with purified GST-␣1-syntrophin-(69 -201) fusion protein in Tween/TBS for 1 h at 25°C. After washing with Tween/TBS, the blots were incubated with anti-GST antibody and then with horseradish peroxidase-conjugated anti-rabbit IgG. Bands were visualized by enhanced chemiluminescence (Renaissance ® ).
Immunoprecipitations-Rat brain extracts were pretreated with protein G-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4°C. After centrifugation, the supernatants were incubated for 1 h at 4°C with protein G-Sepharose that was preincubated with anti-nNOS monoclonal antibody. The Sepharose beads were washed five times with TNE buffer containing 150 mM NaCl. The resultant immunoprecipitated proteins were resolved by SDS-PAGE.
Pull-down Assays-Rat brain extracts (2.5 mg) were incubated with glutathione-Sepharose beads coupled with GST-nNOS-(1-230) or GST. The beads were washed three times with TNE buffer containing 0.15 M NaCl and eluted with glutathione (10 mM). Eluted proteins were analyzed by Western blotting.
Cell Culture and Immunocytochemistry-Neuronal cultures were prepared from embryonic day 18 Wistar rats by the method of Brewer et al. (18) with slight modifications. Briefly, neurons were isolated by trypsin treatment and plated on poly-D-lysine-coated glass coverslips in Dulbecco's modified minimal essential medium with 10% calf serum at a density of 2000 cells/cm 2 . After attachment of cells, the coverslips were transferred to serum-free Neurobasal medium (Life Technologies, Inc.) with N2 supplements. For immunocytochemical staining, the neurons were fixed with 4% paraformaldehyde 10 days after plating, blocked with 10% bovine serum albumin in Tween/TBS, and exposed to primary antibodies. Primary antibodies were visualized with fluorescein isothiocyanate-conjugated anti-rabbit IgG (ICN Pharmaceuticals, Inc., Costa Mesa, CA) or Texas Red-conjugated anti-mouse IgG (Amersham Pharmacia Biotech), which were incubated together for doublelabeling experiments. Fluorescent photomicroscopy was performed on a Nikon Diaphot-TMD microscope with Fuji Provia 1600 film.
Immunohistochemistry-Rats were anesthetized with sodium pentobarbital and perfused with 4% paraformaldehyde-containing phosphate-buffered saline. The brains were removed, post-fixed at 4°C for 2 days, cryoprotected in 30% sucrose at 4°C for 5 days, and cut into 20-m sections with a microslicer. The sections were treated for 3 h in phosphate-buffered saline containing 3% bovine serum albumin and 0.3% Triton X-100; incubated overnight with primary antibodies to ␣1-syntrophin or nNOS, which were diluted into Tween/TBS; and then incubated for 1 h with horseradish peroxidase-conjugated anti-rabbit IgG). Signals were visualized with 3,3Ј-diaminobenzidine (Dojindo Laboratories, Kumamoto, Japan). For confocal microscopy, signals were visualized with fluorescein isothiocyanate-labeled anti-mouse IgG and rhodamine-labeled anti-rabbit IgG and were then observed on a Bio-Rad Micro Radiance confocal scanning system.

Purification and Identification of an nNOS-interacting Protein-
To examine whether the PDZ domain of nNOS in brain interacts with proteins other than PSD-95 and PSD-93, we purified nNOS-interacting proteins by affinity chromatography using glutathione-Sepharose beads coupled with GST-nNOS-(1-230) fusion protein. We also used glutathione-Sepharose coupled with GST as a control. Crude extracts from bovine brain were loaded on each column, and proteins were eluted with glutathione. We found a protein of ϳ60 kDa that was eluted from the GST-nNOS affinity column, but not from the GST column, indicating that this protein was selectively bound to the N-terminal region of nNOS (Fig. 1A). Because the eluate contained a large amount of GST-nNOS fusion protein (55 kDa) and its proteolytic products, proteins with molecular masses Ͻ55 kDa were not analyzed.
To identify the protein, it was subjected to amino acid sequencing (Fig. 1B). The protein was digested with Achromobacter protease I on a polyvinylidene difluoride membrane,

FIG. 1. Purification and sequence analysis of nNOS-interacting proteins.
A, purification of an nNOS-interacting protein by GST-nNOS affinity column chromatography. Crude brain extract was loaded onto a glutathione-Sepharose column previously coupled with GST or GST-nNOS-(1-230). Bound proteins together with GST or GST-nNOS were eluted with glutathione, subjected to SDS-PAGE, and detected by silver staining. The asterisk denotes an nNOS-interacting protein. B, sequence analysis of the nNOS-interacting protein. The nNOS-interacting protein was transferred to a polyvinylidene difluoride membrane and digested with Achromobacter protease I, and the fragments were obtained by reversed-phase HPLC. Sequences of three of the fragments(AP-1, AP-2, and AP-3) were determined by peptide microsequencing and aligned with those of ␣1-syntrophin deduced from its cDNA sequence. and three peptide fragments (AP-1, AP-2, and AP-3) were analyzed on a peptide sequencer. A homology search analysis showed that all these sequences were identical to the fragments of Mus musculus ␣1-syntrophin except for the fifth amino acid in the peptide AP-3 (Fig. 1B). The molecular mass of ␣1-syntrophin calculated from its cDNA sequence was 58 kDa, which was close to that of the nNOS-interacting protein estimated by SDS-PAGE. We therefore concluded that the nNOS-interacting protein was bovine ␣1-syntrophin.
Interaction of nNOS with ␣1-Syntrophin in Brain-To confirm whether nNOS can interact with ␣1-syntrophin, we raised an antibody against ␣1-syntrophin-(31-90). We chose this region as an antigen because it has Ͻ50% sequence homology to two other isoforms, ␤1 and ␤2. The antibody was affinitypurified with CH-Sepharose coupled with GST-␣1-syntrophin-(31-90). Immunoblotting of a crude rat brain extract with the antibody showed that it specifically reacted with a 60-kDa protein (Fig. 2A, lane 1). Some other bands under 45 kDa were also detected, but they seemed to be nonspecific signals of the secondary antibody used because they were detected even when the primary antibody was omitted (Fig. 2A, lane 2). FIG. 3. Analysis of the interaction between nNOS and ␣1-syntrophin by overlay assay. Rat brain extracts were immunoprecipitated with protein G-Sepharose coupled with or without anti-nNOS antibody. To confirm that the immunoprecipitation (IP) was successful, the immunoprecipitates were analyzed by Western blotting with anti-nNOS antibody (lanes 1 and 2). The same samples were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and overlaid with GST-␣1-syntrophin-(69 -201) fusion protein (lanes 3 and 4) or GST (lanes 5 and 6). The membranes were then incubated with anti-GST antibody followed by horseradish peroxidase-labeled anti-rabbit IgG and developed with chemiluminescence reagent.

FIG. 2. Interaction of nNOS with ␣1syntrophin.
A, characterization of anti-␣1-syntrophin antibody. An antibody was raised against GST-␣1-syntrophin-(31-90) in a rabbit and applied to Western blotting of a crude brain extract (lane 1). As a control, the extract was treated in the same manner without the primary antibody (lane 2). The asterisk denotes ␣1syntrophin. B, analysis by pull-down assay. Crude brain extract was incubated with glutathione-Sepharose coupled with GST-nNOS-(1-230) fusion protein. After washing the beads, bound proteins were subjected to SDS-PAGE and analyzed by Western blotting with anti-␣1-syntrophin antibody (lane 3). In the same gel, purified GST (lane 2) and GST-␣1-syntrophin-(31-90) fusion protein (lane 1) were run to confirm the specificity of the antibody. The same series of samples was also subjected to Western blotting with anti-␣1syntrophin antibody preabsorbed with its antigen (lanes 4 -6).
The specificity of the antibody was further confirmed in Fig.  2B. The antibody reacted with GST-␣1-syntrophin (Fig. 2B,  lane 1), but not with GST (lane 2). In addition, binding of the antibody to GST-␣1-syntrophin was completely abolished when it was preincubated with the antigen (Fig. 2B, compare lanes 1  and 4). These results suggest that the antibody specifically reacts with ␣1-syntrophin.
Next, we analyzed the interaction of nNOS with ␣1-syntrophin in brain by pull-down assay. Glutathione-Sepharose beads coupled with GST-nNOS-(1-230) were incubated with a rat brain lysate and then precipitated. Western blotting with anti-␣1-syntrophin antibody showed that ␣1-syntrophin was retained by GST-nNOS-conjugated Sepharose (Fig. 2B, lane 3), but not by GST-Sepharose (data not shown). The 60-kDa band was not detected when the antibody was preabsorbed with antigen (Fig. 2B, lane 6). These results suggest that nNOS interacts with ␣1-syntrophin in rat brain.
Analysis of the Interaction between nNOS and ␣1-Syntrophin by Overlay Assay-To confirm the interaction of nNOS with ␣1-syntrophin and to examine whether binding of nNOS to ␣1-syntrophin is direct or indirect, we performed protein overlay assays using GST-␣1-syntrophin-(69 -201), which contains the PDZ domain, as a probe. Rat brain extracts were immunoprecipitated with or without anti-nNOS antibody (Fig. 3). Western blotting with anti-nNOS antibody confirmed that nNOS was precipitated with the antibody (Fig. 3, lane 1). The same samples were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and overlaid with GST-␣1-syntrophin (Fig. 3, lanes 3 and 4) or GST (lanes 5 and 6). The probes were detected with anti-GST antibody and visualized by enhanced chemiluminescence. The band corresponding to nNOS was detected when the immunoprecipitated materials were overlaid with GST-␣1-syntrophin-(69 -201) (Fig. 3, lane 3), but not with GST (lane 5). These results indicate that nNOS can directly interact with ␣1-syntrophin in rat brain.
Assessment of the Affinity of nNOS for ␣1-Syntrophin-We further examined the affinity of nNOS for ␣1-syntrophin in the presence of different kinds of detergents (Fig. 4). Nonidet P-40 (1%) was used as the detergent in other analyses, but 1% Nonidet P-40, 0.5% deoxycholate, and 0.1% SDS (buffer b) or 1% Triton X-100 (buffer c) also gave similar results as detected by silver staining (Fig. 4A) and Western blotting (Fig. 4B). Deoxycholate (1%) (buffer d) was the strongest condition in which interaction of nNOS with ␣1-syntrophin was partly perturbed, whereas the interaction was not perturbed when deoxycholate was used in combination with 1% Nonidet P-40 and 0.1% SDS (buffer b).
Because nNOS has been known to interact with PSD-95, we compared the affinity of nNOS for ␣1-syntrophin with that for PSD-95 (Fig. 4B). When detergent lysates from rat brain were precipitated with GST-nNOS-conjugated Sepharose, ␣1-syntrophin was highly concentrated in the precipitated fractions. PSD-95 was also solubilized with all the buffers tested and detected in the lysates by Western blotting. But, in contrast to ␣1-syntrophin, only a small amount of PSD-95 was precipitated with GST-nNOS. These results suggest that the affinity of nNOS for ␣1-syntrophin is higher than that for PSD-95.
Colocalization of nNOS with ␣1-Syntrophin in Primary Cultures of Neuronal Cells-We next examined whether nNOS and ␣1-syntrophin colocalized in neuronal cells by immunocytochemistry. Primary cultures of neuronal cells were prepared from fetal rat brain and maintained for 7-10 days in serum-free medium. Most neurons were double-labeled with anti-nNOS and anti-␣1-syntrophin antibodies with similar subcellular distribution. ␣1-Syntrophin-like immunoreactive substances were present in both neuronal cell bodies and neurites (Fig. 5, A and  C). nNOS-like immunoreactivity was also detected in both neuronal cell bodies and neurites (Fig. 5, B and E). All these immunoreactivities became very weak when the primary antibodies were preincubated with the respective antigens (data not shown).
We further examined the distribution of synaptotagmin, an essential component of the synaptic membranes, to determine the location of presynaptic structures in these cells. Synaptotagmin-like immunoreactivity was found as punctate signals along the neurites (Fig. 5, D and F) and was not detected in their cell bodies in most neurons. Double staining with antisynaptotagmin and anti-␣1-syntrophin antibodies confirmed that syntrophin was not restricted to synapses (Fig. 5, C and  D). Double staining with anti-synaptotagmin and anti-nNOS antibodies showed that a fraction of nNOS was colocalized in synapses, but the majority of nNOS seemed to be present outside of synapses (Fig. 5, E and F).
Distribution of nNOS and ␣1-Syntrophin in the Hypothalamus-Relative amounts of nNOS and ␣1-syntrophin localized in various brain regions were examined by Western blotting with the respective antibodies. nNOS was highly expressed in the cerebellum and olfactory bulb as reported previously (5), but low level expression was detected in all other regions tested, including the striatum, cerebral cortex, SCN, and hypothalamic paraventricular nucleus (PVN) (Fig. 6). ␣1-Syntrophin was also present in all regions tested, with the highest expression observed in the PVN (Fig. 6).
Immunohistochemical staining of rat hypothalamic sections using anti-nNOS and anti-␣1-syntrophin antibodies was done. In the hypothalamus, ␣1-syntrophin-like immunoreactive substance was observed in the SCN, PVN, and anterior hypothalamic area (Fig. 7, A and B). nNOS-positive neurons were also detected in the SCN and PVN (Fig. 7, D and E), consistent with previous studies (5,6). In the SCN, nNOS-and ␣1-syntrophinlike immunoreactive substances were most concentrated in the dorsomedial region, whereas weak signals were also detected in the ventrolateral region (Fig. 7, A, B, D, and E). In the PVN, ␣1-syntrophinand nNOS-like immunoreactive substances were detected mainly in the magnocellular part. No immunostaining was detected when primary antibodies were preincubated with their respective antigens, confirming the specificity of immunolabeling (Fig. 7, C and F).
Fine distributions of nNOS and ␣1-syntrophin in the dorsomedial region of the SCN were observed using a confocal microscope. In the SCN, nNOS-like immunoreactive substance was detected in the cell matrix of neuronal cell bodies, but not in the nuclei (Fig. 8A). ␣1-Syntrophin also showed a subcellular distribution similar to nNOS (Fig. 8B). Superimposing fluorescence images for nNOS and ␣1-syntrophin gave a yellow color, suggesting that nNOS and ␣1-syntrophin are colocalized in the same regions of the same SCN neurons (Fig. 8C). In the PVN, the colocalization of nNOS with ␣1-syntrophin was also observed (data not shown).

DISCUSSION
In this study, we have purified an nNOS-binding protein from bovine brain and identified it as ␣1-syntrophin. We further demonstrated that nNOS-and ␣1-syntrophin-like immunoreactive substances showed similar subcellular distribution in primary cultures from fetal rat brain. Finally, the two proteins were expressed at relatively high levels and colocalized in the PVN and SCN in hypothalamic sections of adult rat brains. These results suggest that nNOS interacts with ␣1-syntrophin in specific neurons in brain.
nNOS was originally found in mammalian brain, but was later shown to be present also in skeletal muscle, lung epithelial cells, and certain endocrine glands (12,19). In skeletal muscle, nNOS is localized at the sarcolemmal membranes by association with syntrophins, a component of the dystrophin complex (11,20). The dystrophin complex is a membrane cytoskeletal structure that links the sarcolemmal membranes to extracellular matrix proteins and intracellular actin fibers. In brain, on the other hand, nNOS has been shown to be associated with PSD-95 and PSD-93 (11), which are localized at the synapses as a component of the postsynaptic density structures (10). But the distribution of nNOS is found not only in the synapses, but also in the entire surfaces of cell bodies and neurites in several types of neurons such as those in the PVN of the hypothalamus (5). These findings indicate that there are different mechanisms that define the intracellular localization of nNOS depending on cell types.
Our present data showed that ␣1-syntrophin was associated with nNOS in vitro even in the presence of strong detergents such as Nonidet P-40, deoxycholate, and SDS. The binding affinity of nNOS for ␣1-syntrophin seemed to be much higher than that for PSD-95 as estimated by pull-down assay. In addition, nNOS and ␣1-syntrophin were colocalized in cell bodies, neuronal processes, and synapses in cultured neurons from fetal rat brains. These two proteins were also colocalized in neuronal cells in the SCN and PVN. From these results, we propose that ␣1-syntrophin also contributes to determining the subcellular localization of nNOS in certain regions of the brain.
␣1-Syntrophin was expressed in most regions in adult rat brain judging from Western blotting with anti-␣1-syntrophin antibody. The present immunohistochemical data showed that the level of ␣1-syntrophin was relatively high in several neurons, including the PVN and SCN in the hypothalamus (Fig. 7). These two nuclei have been known to contain nNOS. These results support the possibility that nNOS and ␣1-syntrophin interact with each other in certain brain regions and imply that they have functional relationships. The immunohistochemical distribution of ␣1-syntrophin is not completely consistent with its mRNA distribution previously shown by in situ hybridization (21), but the discrepancy might be elicited by the stability of mRNA and protein.
In skeletal muscle, syntrophins have been shown to be a component of the dystrophin complex and to function as molecular adaptors that recruit signaling proteins to the membrane (14). Dystrophin is also present in brain and has been reported to localize at postsynaptic densities (22). However, nNOS is unlikely to interact with dystrophin in brain because nNOS is membrane-associated even in the brains of mdx mice that lack dystrophin (20). Rather, nNOS might be associated with membrane proteins such as other dystrophin family proteins via ␣1-syntrophin.
The SCN and PVN have key roles in the regulation of metabolism and behavior through controlling the autonomic nervous system and endocrine functions in mammals. The PVN regulates various neuroendocrine hormones through the hypothalamohypophysial system, and NO has been suggested to be implicated in its function (23). The SCN is the nucleus containing the circadian oscillator responsible for circadian rhythms (24) and a mechanism controlling the autonomic nervous system (7). We have previously shown that infusion of N G -methylarginine into the third ventricle in rats disrupts the circadian rhythm of drinking behavior (8), suggesting that NO might be involved in the generation and/or synchronization of the circadian oscillator. nNOS-and ␣1-syntrophin-like immunoreactive substances were detected at relatively high levels in the dorsomedial region of the SCN, where vasopressin-containing neurons exist, and it was shown that these vasopressin neurons are involved in the regulation of secretion of adrenal glucocorticoid (25). Therefore, it will be interesting to examine the coexistence of nNOS and ␣1-syntrophin with vasopressin in the SCN to obtain further information on the physiological functions of nNOS and ␣1-syntrophin in the SCN.