Inhibition of nitric-oxide synthase-I (NOS-I)-dependent nitric oxide production by lipopolysaccharide plus interferon-gamma is mediated by arachidonic acid. Effects on NFkappaB activation and late inducible NOS expression.

Previous results have indicated that lipopolysaccharide (LPS) plus interferon-gamma (IFNgamma) inhibits nitric-oxide synthase (NOS)-I activity in glial cells. We report here that arachidonic acid (AA) plays a pivotal role in this response, which was consistently reproduced in different glial cell lines and in primary rat astrocytes. This notion was established using pharmacological inhibitors of phospholipase A2 (PLA2), cytosolic PLA2 (cPLA2) antisense oligonucleotides, and AA add-back experiments. This approach not only allowed the demonstration that AA promotes inhibition of NOS-I activity but also produced novel experimental evidence that LPS/IFNgamma itself is a potential stimulus for NOS-I. Indeed, LPS/IFNgamma fails to generate nitric oxide (NO) via NOS-I activation simply because it activates the AA-dependent signal that impedes NOS-I activity. Otherwise, LPS/IFNgamma promotes NO formation, sensitive to exogenous AA, in cells in which cPLA2 is pharmacologically inhibited or genetically depleted. Because NO suppresses the NFkappaB-dependent NOS-II expression, inactivation of NOS-I by the LPS/IFNgamma-induced AA pathway provides optimal conditions for NFkappaB activation and subsequent NOS-II expression. Inhibition of cPLA2 activity, while reducing the availability of AA, consistently inhibited NFkappaB activation and NOS-II mRNA induction and delayed NO formation. These responses were promptly reestablished by addition of exogenous AA. Finally, we have demonstrated that the LPS/IFNgamma-dependent tyrosine phosphorylation of NOS-I and inhibition of its activity are mediated by endogenous AA.

Nitric oxide (NO) 1 is generated in different cell types by the conversion of L-arginine into citrulline mediated by at least three distinct isoforms of the enzyme NO synthase (NOS). Two enzymes, the neuronal (NOS-I) and the endothelial (NOS-III) isoforms, are Ca 2ϩ -dependent and constitutively expressed. The third enzyme is an inducible Ca 2ϩ -independent isoform (NOS-II), expressed after stimulation with Escherichia coli lipopolysaccharide (LPS) and/or different cytokines, such as interferon-␥ (IFN␥), interleukin-1␤, or tumor necrosis factor-␣ (1). NOS-II induction occurs at the transcriptional level and is mediated by the early activation of some nuclear transcriptional factors, including NFB (2). A large body of experimental evidence suggests that physiological levels of NO, similar to those produced by the basal activity of NOS-I or NOS-III, prevent induction of NOS-II mRNA expression through the suppression of NFB activation (3,4). As a consequence, NOS-II gene expression takes place after LPS/cytokine stimulation, provided that the NOS-I-or NOS-III-generated NO is reduced below a threshold value in a short time (5). We have recently reported (6,7) that NOS-II inducers (e.g. LPS and IFN␥) consistently elicit a rapid inactivation of NOS-I by tyrosine phosphorylation, an event leading to a decrease of basal NO levels.
In a recent study (8), we reported that inhibition of NOS-I can be achieved via activation of cytosolic phospholipase A 2 (cPLA 2 ), a large molecular mass member of the family of PLA 2 enzymes. The activities of NOS-I and cPLA 2 are both regulated by increases in the intracellular concentration of free Ca 2ϩ ([Ca 2ϩ ] i ) (9); not surprisingly, enhancing the [Ca 2ϩ ] i was found to cause a parallel increase in both activities and accumulation of respective products, NO and arachidonic acid (AA). Interestingly, however, critical levels of AA were eventually reached that reduced or even suppressed formation of NO (8). Thus, sustained activation of cPLA 2 is expected to promote a rapid inactivation of NOS-I via the AA-dependent inhibitory signaling. It is intriguing that both LPS and IFN␥ stimulate the activity of cPLA 2 and promote the release of AA (10,11).
In the present study, glial cell lines as well as rat primary astrocytes were utilized as cellular models to demonstrate that the mechanism whereby LPS and IFN␥ impair NOS-I activity indeed involves an AA-dependent tyrosine phosphorylation of the enzyme. The inhibitory signaling triggered by AA therefore allows downstream events leading to NFB activation, NOS-II mRNA expression, and NOS-II-dependent NO production. However, when the release of AA was prevented by pharmacological inhibition or genetic depletion of cPLA 2 , the mixture LPS/IFN␥ stimulated NOS-I activity and failed to elicit the above NFB-dependent delayed responses.

Materials
AA, A23187, LPS, N -nitro-L-arginine methylester (L-NAME), and other reagent grade biochemicals were from Sigma. IFN␥ (specific ac-* This work was supported by grants from Ministero dell'Università e della Ricerca Scientifica e Tecnologica, Progetti di Ricerca di Interesse Nazionale (to O. C. and M. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Methods
Cell Culture and Treatment Conditions-Primary cultures of cortical astrocytes derived from neonatal 1-or 2-day-old Sprague-Dawley rats (Charles River, Calco, Italy) were prepared as described (12). At this age, the progenitor cells have lost the ability to differentiate into neuronal cells (13), giving rise to a pure glial cell preparation that was also confirmed by immunocytochemistry using antibodies directed to glial fibrillary acidic protein. Primary astrocytes, human A172 astrocytoma, or C6 glioma cells were cultured in Eagle's minimal essential medium (primary astrocytes) or Dulbecco's modified Eagle's medium (A172 or C6 cells) supplemented with either 10% horse serum and 10% fetal bovine serum (HyClone Laboratories, Logan, UT) (primary astrocytes) or 10% fetal bovine serum (A172 or C6 cells), penicillin (50 units/ml), and streptomycin (50 g/ml) (HyClone) at 37°C in 35-mm Primaria dishes (primary astrocytes) or T-75 tissue culture flasks (A172 or C6 cells) gassed with an atmosphere of 95% air-5% CO 2 . For experiments, cells were used 2 weeks (primary astrocytes) or 18 -24 h (A172 or C6 cells) after plating into 35-mm culture dishes.
Stock solutions of IFN␥, LPS, L-NAME, and AA were freshly prepared in distilled water. AACOCF 3 and A23187 were dissolved in Me 2 SO. At the treatment stage, the final concentration of Me 2 SO was never higher than 0.05% (v/v).
NO Detection System-The production of NO was assayed using the DAF-2DA detection system, as described elsewhere (8). Briefly, the cells were loaded with 10 M DAF-2DA for 10 min in saline A (8.182 g/liter NaCl, 0.372 g/liter KCl, 0.336 g/liter NaHCO 3 , and 0.9 g/liter glucose) at 37°C. After washings, the cells were treated (15 min at 37°C), and cellular fluorescence was imaged using a confocal laser microscope (Bio-Rad DVC 250) equipped with a Hamamatsu 5985 (Milan, Italy) chilled CCD camera. Confocal images were digitally acquired and processed for fluorescence determination at the single cell level on a Macintosh G4 computer using the public domain NIH Image 1.61 program (developed at the U.S. National Institutes of Health and available on the Internet at rsb.info.nih.gov/nih-image/).
Immunocytochemical Detection of cPLA 2 -The cPLA 2 protein was detected using an immunocytochemical technique following the manufacturer's instructions, as previously described (8). Briefly, the cells were fixed with ethanol/acetic acid and blocked in phosphate-buffered saline containing bovine serum albumin (2%, w/v). Mouse monoclonal cPLA 2 antibodies (5 g/ml) were used as primary antibodies. After 18 h at 4°C, the cells were washed and exposed to fluorescein isothiocyanate-conjugated secondary antibodies for 2 h in the dark. Stained cells were analyzed using a confocal microscope, and the resulting images were processed for fluorescence determination as described above.
Transfection of Sense or Antisense Oligonucleotides into C6 Cells-The antisense and sense oligonucleotides for rat cPLA 2 were transfected into the cells using the LipofectAMINE PLUS method as detailed elsewhere (8). Treatments were performed 24 h after transfection.

Measurement of Extracellular Release of [ 3 H]AA-Cells
were subcultured in 6-well plates at 2 ϫ 10 5 cells/well with [ 3 H]AA (0.5 Ci/ml) and grown for 18 h. Before treatments, the cells were washed twice with saline A supplemented with 1 mg/ml fatty acid-free bovine serum albumin and then exposed to LPS/IFN␥, in the absence or presence of a PLA 2 inhibitor, in a final volume of 2 ml of saline A. The solution was then separated and centrifuged at 12,000 rpm for 3 min. 500 l of the resulting supernatant were removed, and radioactivity was determined in a Wallac 1409 liquid scintillation counter (Turku, Finland).
NFB Activation Assay-NFB activation was quantified using the TransAM NFB kit (Active Motif, Rixensart, Belgium). Briefly, wholecell extracts were prepared from 5 ϫ 10 5 C6 cells according to the manufacturer's instructions. Protein concentrations of cell extracts were determined according to Bradford (14). Ten micrograms/well of cell extracts were incubated in a 96-well plate on which have been immobilized double-stranded oligonucleotides containing the consensus NFB DNA binding site (5Ј-AGTTGAGGGGACTTTCCCAGGC-3Ј). The primary antibody used to detect NFB recognized an epitope on p65 subunit that is accessible only when NFB is activated and bound to its target DNA. After incubation with a horseradish peroxidase-conjugated secondary antibody and the developing solution, absorbance was read at 450 nm with a reference wavelength of 655 nm.
Reverse Transcriptase-PCR-Reverse transcriptase-PCR was carried out on total cellular RNA purified from 1 ϫ 10 6 rat C6 glioma cells and reverse-transcribed into cDNA as previously reported (15). cDNA was amplified for the NOS-II gene (450 bp) using rat NOS-II-specific primers as described elsewhere (16). The mRNA for the constitutive GAPDH enzyme was examined as the reference cellular transcript. GAPDH mRNA amplification products (195 bp) were present at equivalent levels in all cell lysates. Estimates of the relative NOS-II mRNA amounts were obtained dividing the area of the NOS-II band by the area of the GAPDH band (Bio-Rad Multianalyst).
Statistical Analysis-Statistical analysis of the data for single comparisons was performed by Student's t test.

RESULTS AND DISCUSSION
We recently reported that exposure of C6 cells to concentrations of A23187 in the 0.5-2.5 M range leads to a progressive increase in NO formation and that this response is reduced, or even abolished, when the ionophore is utilized at 5 or 7.5 M (8). These results were readily explained by the observation that high A23187 concentrations cause a cPLA 2dependent release of AA, which then promotes a NOS-I inhibitory signaling pathway. On the other hand, exogenous AA suppressed formation of NO mediated by lower concentrations of A23187 (8).
The results illustrated in Fig. 1A confirm and extend these findings by showing that the DAF-2 fluorescence response elicited by 2.5 M A23187 (15 min), although sensitive to both L-NAME (1 mM) and exogenous AA (30 nM), is also inhibited by a pretreatment (5 min) of the cells with the mixture LPS (1 g/ml)/IFN␥ (1000 units/ml). At these concentrations, LPS and IFN␥ promote maximal activation of NFB and NOS-II expression (data not shown), and as we previously showed (6, 7) also cause suppression of basal NO production, an effect readily detected by measuring conversion of L-arginine to L-citrulline in cell homogenates supplemented with cofactors for NOS-I activity. Under our experimental conditions, however, LPS/ IFN␥ failed to affect the basal DAF-2 fluorescence, and similar results were obtained using AA or L-NAME (Fig. 1A). This represents a limitation of the assay, most likely associated with the significant NO-independent DAF-2 fluorescence signal, which does not allow detection of small changes mediated by basal NO levels generated by NOS-I in the absence of stimuli. The time of DAF-2 exposure is indeed too short to allow formation of NO levels promoting detectable changes in the DAF-2 fluorescence signal. However, as indicated above, such variations are readily detected in cells stimulated with A23187, and LPS/IFN␥ was found to suppress this response, thus confirming the LPS/IFN␥-mediated NOS-I inactivation.
Further analyses revealed that the LPS/IFN␥-induced inhibition of the A23187-stimulated NO formation is because of AA-dependent NOS inhibition. Three separate lines of evidence support this inference. First, as shown in Fig. 1B, the effects mediated by LPS/IFN␥ were prevented by the PLA 2 inhibitor AACOCF 3 (75 M) (or 5,8,11,14-eicosatetraynoic acid, 50 M, not shown), which was otherwise ineffective in the absence of additional treatments (not shown). Second, addition of the PLA 2 product AA abolished the effects of AACOCF 3 on the LPS/IFN␥-mediated inhibition of the A23187-dependent NO formation. An identical outcome was obtained from experiments in which AA was replaced with L-NAME. The third line of evidence derives from experiments using cells transfected with cPLA 2 antisense oligonucleotides. These cells, although showing a remarkable decrease in cPLA 2 protein immunoreactivity ( Fig. 2A), responded to 2.5 M A23187 with formation of NO insensitive to LPS/IFN␥, albeit sensitive to either AA or L-NAME (Fig. 2B). In marked contrast, the experiments performed with cPLA 2 sense oligonucleotide-transfected cells (Fig.  2C) led to outcomes identical to those obtained using non- a ͓ 3 H͔AA-labeled cells were exposed for 5 min to 75 M AACOCF 3 and then treated for a further 15 min with LPS/IFN␥. After treatments, the ͓ 3 H͔AA release was quantified as described under "Experimental Procedures." b The cells were labeled with ͓ 3 H͔AA after a 6-h incubation with the cPLA 2 sense or antisense oligonucleotides. After an additional 18 h, the cells were treated as described above.

FIG. 2. LPS/IFN␥ fails to inhibit formation of NO promoted by A23187 in cPLA 2 antisense oligonucleotide-transfected cells. A, cells
were transfected with cPLA 2 sense or antisense oligonucleotides as described under "Experimental Procedures." After 24 h, levels of the cPLA 2 protein were measured using an immunocytochemical detection assay. B and C, cells transfected with cPLA 2 antisense (B) or sense (C) oligonucleotides were loaded with DAF-2DA, incubated for 5 min in the absence or presence of LPS/IFN␥, L-NAME, or AA and treated for an additional 15 min with 2.5 M A23187. Cells were analyzed as described in the legend to Fig. 1. Data points are the means Ϯ S.E. from three to five separate experiments, each performed in duplicate. *, p Ͻ 0.05, **, p Ͻ 0.001 versus untreated cells (unpaired t test). transfected cells (Fig. 1, i.e. AA, L-NAME as well as LPS/IFN␥ abolishing formation of NO).
These results confirm that AA promotes inhibition of NOS-I (8) and provide experimental evidence indicating that LPS/ IFN␥ inhibits A23187-induced NOS-I activation by triggering the AA-dependent mechanism. The antisense oligonucleotide experiments imply that the cPLA 2 isoform is specifically involved in the LPS/IFN␥-induced formation of AA. Consistently, LPS/IFN␥ caused an extensive release of AA, sensitive to AA-COCF 3 and remarkably lower in cells transfected with cPLA 2 antisense oligonucleotides (Table I). Experiments using cells transfected with cPLA 2 sense oligonucleotides produced results identical to those obtained with non-transfected cells. It should be noted that AACOCF 3 is not efficiently taken up in some cell types. By monitoring its intrinsic fluorescence, we found that this was the case for C6 cells (not shown), which is the reason a high concentration of AACOCF 3 was employed in this study. Thus, although a minor contribution of an AACOCF 3 -insensitive (e.g. group II) PLA 2 in the LPS/IFN␥-induced AA release cannot be ruled out, the inability of the inhibitor to completely suppress the release of AA might be dependent on its low cellular uptake.
The above results indicate that LPS/IFN␥ potently stimulates the activity of cPLA 2 and that the ensuing release of AA promotes inhibition of NOS-I activity. Interestingly, we found that incubation of LPS/IFN␥-stimulated cells with the PLA 2 inhibitor AACOCF 3 , although reducing AA release (Table I), caused a remarkable increase of DAF-2 fluorescence sensitive to either L-NAME or AA (Fig. 3A). Consistently, LPS/IFN␥, in the absence of additional treatments, caused a similar DAF-2 fluorescence signal in cells transfected with cPLA 2 antisense oligonucleotides (Fig. 3B). This response, although insensitive to AACOCF 3 , was suppressed by L-NAME or AA. The results obtained using cells transfected with cPLA 2 sense oligonucleotides were identical to those generated with non-transfected cells (Fig. 3C). The above findings are consistent with the possibility that LPS/IFN␥ is paradoxically a potential activator of NOS-I and that it fails to do so simply because the prevailing AA-dependent inhibitory signaling suppresses NOS-I activity.
In other words, LPS/IFN␥ may produce effects superimposable on those previously observed using a high A23187 concentration (7.5 M) (8). This notion was established in additional glial

FIG. 3. LPS/IFN␥ is a potential stimulus for NOS-I activity that is, however, prevented by parallel AA-dependent NOS-I inhibitory signaling.
A, DAF-2DA-preloaded cells were incubated for 5 min in the absence or presence of AACOCF 3 alone or associated with either L-NAME or AA. Cells were then treated for an additional 15 min with LPS/IFN␥ and analyzed as described in the legend to Fig. 1. B and C, cells were transfected with cPLA 2 antisense (B) or sense (C) oligonucleotides and loaded with DAF-2DA. cPLA 2 antisense oligonucleotide-transfected cells were incubated for 5 min in the absence or presence of AACOCF 3 , L-NAME, or AA. Cells were then treated for an additional 15 min with LPS/IFN␥. cPLA 2 sense oligonucleotide-transfected cells were treated as detailed in panel A. Cells were analyzed as described in the legend to cell lines (not shown), including human A172 astrocytoma cells (see below) as well as primary rat astrocytes (Fig. 4).
Thus LPS/IFN␥, under conditions not permissive for AA release, promotes formation of NO. Under physiological condi-tions, however, the LPS/IFN␥-mediated NOS-I activation is prevented by the parallel NOS-I inhibitory signaling pathway(s) triggered by AA.
Because it is well established that suppression of NOS-I activ- ity is an early, necessary event for cytokine-induced NFB activation and NOS-II expression, the role of AA in these responses was investigated next. As expected, exposure to LPS/IFN␥ caused activation of NFB (Fig. 5A), expression of NOS-II (Fig.  5B), and increased formation of NOS-II-dependent NO (Fig. 5C) after 0.5, 4, or 8 h, respectively. Interestingly, AACOCF 3 prevented these effects via an AA-sensitive mechanism. AACOCF 3 , AA, or a combination of the two agents did not produce effects (not shown). The specificity of the effects mediated by the PLA 2 inhibitor is once again emphasized by the results obtained using C6 cells transfected with cPLA 2 antisense oligonucleotides. Under these conditions, LPS/IFN␥ indeed fails to induce NOS-II mRNA expression (4 h) that, however, becomes readily apparent upon supplementation of exogenous AA (Fig. 6A). Consistently, the DAF-2 fluorescence response elicited by NOS-II-dependent formation of NO (8 h) was not observed in cells exposed to LPS/ IFN␥ in the absence of additional treatments (Fig. 6B). Exogenous AA, however, promptly re-established this response in an L-NAME-sensitive fashion.
Altogether, these results confirm our previous findings that NFB-dependent NOS-II expression evoked by inflammatory stimuli requires early inhibition of NOS-I (4,5,(17)(18)(19) and identify AA as an intermediate signaling molecule leading to NOS-I inhibition. A possible candidate of the downstream kinase leading to NOS-I phosphorylation and inactivation is a tyrosine kinase, because AA is a known activator of this pathway (20) and our previous work showed that LPS/IFN␥-induced inhibition of NOS-I activity is mediated by tyrosine phosphorylation (7). To address this question, experiments were performed using the human A172 astrocytoma cell line, which expresses higher levels of NOS-I than C6 cells. As illustrated in Fig. 7A, A172 cells responded to LPS/IFN␥ as C6 cells (e.g. NO being formed only in the presence of AACOCF 3 via an AA-or L-NAME-sensitive mechanism). Interestingly, however, formation of NO was also promoted by the tyrosine kinase inhibitor genistein (300 M), and this response, although prevented by L-NAME, was only slightly reduced by AA. Genistein produced identical effects in C6 cells (not shown); therefore, tyrosine kinase-dependent phosphorylation appears to be a likely mechanism of the AA-dependent NOS-I inhibitory signaling. To obtain more information in this direction, immunoprecipitation experiments were performed using A172 cells. We found under basal conditions NOS-I is partially tyrosine-phosphorylated and that LPS/IFN␥ (10 min) further enhances this response (Fig. 7B). Interestingly, however, tyrosine phosphorylation, although inhibited by genistein (not shown), was also markedly reduced by AACOCF 3 via an AA-sensitive mechanism. Taken together, these results strongly suggest that the downstream target of AA-dependent NOS-I inhibitory signaling is indeed a tyrosine kinase.
In conclusion, the present study underscores a novel mechanism whereby stimulation by LPS/IFN␥ elicits NFB activation and NOS-II expression. Our previous studies indicated that these events were preceded by, and causally linked to, removal of the NFB block mediated by NO generated under conditions of basal NOS-I activity (5). We have demonstrated here that inhibition of NOS-I is achieved via AAdependent inhibitory signaling and that the downstream kinase leading to NOS-I phosphorylation and inactivation is a tyrosine kinase.
As a final note, it is important to keep in mind that AA is extensively released during inflammation and that, under these conditions, autocrine or paracrine AA may cause profound effects on cytokine-induced NFB activation and NOS-II expression. Growing experimental evidence suggests that reduced availability of constitutive NO levels is detrimental in a variety of conditions, including neurodegenerative disorders associated with inflammation (21). One possible explanation for these observations is that physiological levels of NO might promote cytoprotection (22)(23)(24). In addition and/or as an alternative, the possibility exists that AA-dependent suppression of NOS-I activity might facilitate cytokine-mediated activation of astrocytes and/or other cell types with the concomitant generation of an array of toxic species.
Acknowledgments-We thank Prof. Valeria Bruno for help in setting up the cultures of rat primary astrocytes and Dr. Zulema Percario for helpful technical support. FIG. 7. LPS/IFN␥ induces tyrosine phosphorylation of NOS-I and inhibition of its activity via an AA-dependent mechanism. A, DAF-2DA-preloaded A172 cells were incubated for 5 min with AACOCF 3 or genistein (300 M) in the absence or presence of either L-NAME or AA. Cells were then treated for an additional 15 min with LPS/IFN␥. After treatments, the DAF-2 fluorescence response was determined as described above. Data points are the means Ϯ S.E. of three to five separate experiments, each performed in duplicate. *, p Ͻ 0.001 versus LPS/IFN␥-treated cells; (*), p Ͻ 0.001 versus LPS/IFN␥ϩAACOCF 3 -treated cells; §, p Ͻ 0.001 versus LPS/IFN␥ϩgenisteintreated cells (unpaired t test). B, NOS-I was immunoprecipitated from A172 cells using a polyclonal anti-NOS-I antibody. Immunoprecipitated NOS-I was Westernblotted using a specific antiserum against phosphotyrosine. NOS-I from untreated cells was weakly and partially tyrosine-