INTRODUCTION

Nitric oxide (NO)¹ is an important signaling molecule that mediates a variety of essential physiological processes including neurotransmission, vasodilatation, and host cell defense (1, 2). NO is synthesized from L-arginine by NO synthase (NOS), a family of enzymes with distinct functional, biochemical, and regulatory properties (1, 2). The constitutive NOS isoforms, originally described in endothelial cells and in neurons, produce small amounts of NO in response to an intracellular calcium increase (1, 3). Cytokine-inducible NO synthase, whose expression requires protein synthesis, has been demonstrated in a wide variety of mammalian cells and tissues (2, 4), and was cloned in different cell types such as murine macrophages (5-7), human hepatocytes (8), and human chondrocytes (9). The role of sustained NO production by the inducible NOS (NOS-2) is well described in murine macrophages, where NO is responsible for their cytostatic and cytolytic activities toward invading organisms (2, 10). On the other hand, NO generated by NOS-2 is also involved in some pathophysiological states, generally related to local and systemic inflammation (2, 4).

In the retina, we have demonstrated that Müller glial cells can express the macrophage-type NOS, NOS-2, after endotoxin and cytokine stimulation (11). Retinal pigmented epithelial (RPE) cells from bovine (12), human (13), and murine (14, 15) species also contain an inducible isoform of NOS. In rat and bovine RPE cells, NOS-2 mRNA activity is induced only by cooperation between IFN- and LPS and can be potentiated in the rat by the addition of TNF- (15). RPE cells resemble other cell types with respect to the combination of mediators capable of inducing NO generation (4). Several lines of evidence indicate that in murine macrophages all three species of interferon, IFN-, IFN-, and IFN-, are able to aid LPS in the release of nitrite (16, 17), a stable end product of NO, and for the expression of NOS-2 mRNA (18). However, only IFN- induced NO release alone, and the combination of LPS with IFN- was the more potent for NOS-2 induction (16-18).

The purpose of the present study was to investigate whether IFN- and IFN-, which have been previously described to be able to interact with RPE cells (19, 20), could regulate the production of NO in bovine RPE cells. Our results indicate that, instead of enhancing LPS-induced NOS-2, IFN- and IFN- markedly inhibit NO production stimulated by the LPS/IFN- combination in bovine RPE cells. We have found that the inhibitory effect of IFNs on NOS-2 activity could be correlated with a decrease of NOS-2 protein and mRNA accumulation. This phenomenon is not due to a competition of IFN- and IFN- for IFN- receptors. In addition, there was no inhibition of the activation of the transcription factor NF-B. Analysis of the effects of IFN- on the induction of the two transcription factors, IRF-1 and IRF-2, revealed that IFN- inhibits the LPS/IFN--induced IRF-1 mRNA accumulation and increased IRF-2 mRNA accumulation in RPE cells.

MATERIALS AND METHODS

Bovine RPE cells were prepared, subcultured, and characterized as previously reported (21). Cells of passages 1-5 were used for the experiments.

N^G-Monomethyl-L-arginine was purchased from Calbiochem (Meudon, France). LPS from Salmonella typhymurium and NADPH were obtained from Sigma France. (6R)-5,6,7,8-Tetrahydro-L-biopterin dihydrochloride was obtained from B. Schircks Laboratories (Jona, Switzerland). L-[guanido-¹⁴C]Arginine (2 GBq/mmol) and L-[2,3,4,5-³H]arginine monohydrochloride (10.7 TBq/mmol) were obtained from Amersham France SA. Bovine recombinant IFN- was generously provided by Dr. T. Ramp (Ciba-Geigy Limited, Basel, Switzerland). Bovine recombinant IFN- was from Ciba-Geigy, through Dr. Schustermann (INSERM, U135, Paris). Human IFN- and IFN- were generous gifts from Dr. Duc-Goiran (INSERM, U361). The rabbit anti-liver NOS antibody (22) was a generous gift of Dr Ohshima (C.I.R.C., Lyon, France).

Confluent RPE cells were treated with LPS and IFN-, with or without different IFN- or -, in fresh Dulbecco's modified Eagle's medium, 10% fetal calf serum. After 72 h of incubation, the nitrite concentration was determined in cell-free culture supernatants using the spectrophotometric method based on the Griess reaction, as described previously (12).

Confluent cells were incubated for 72 h with different combinations of LPS, IFN-, IFN-, and IFN-. Cells were harvested by trypsinization, washed once in phosphate-buffered saline (PBS), and resuspended in 50 mM Tris/HCl, pH 7.8, 10 mM dithiothreitol. After sonication, the homogenate was centrifuged for 20 min at 4 °C at 100,000 × g. The supernatant was passed through a Dowex AG 50W-X8 column to remove endogenous arginine, and NOS activity was evaluated by the conversion of L-[³H]arginine to L-[³H]citrulline, as described previously (23, 24), after the determination of protein content (Bio-Rad assay kit).

After treatment with LPS and IFN- with or without IFN- and IFN- for different periods, cells were washed with PBS and scraped into lysis buffer containing protease inhibitors. Samples were centrifuged, and after one freeze/thaw cycle, 100 µg of supernatant proteins were subjected to SDS-polyacrylamide gel electrophoresis. Proteins were then transferred to an Immobilon membrane (Millipore, Saint Quentin en Yvelines, France) by electroblotting. Western blot analysis using a polyclonal antibody specific for liver-inducible NOS (22) was performed as described previously (25). The intensity of the bands was quantified using densitometric measurements with One Descan densitometric software (Scanalytics, Billerica, MA).

Total RNA was extracted from cultured cells treated by cells lysis in guanidinium isothiocyanate followed by phenol acid extraction. The RNA was denatured, electrophoresed (25 µg/lane) in 1% formaldehyde-agarose gel, and then transferred to a nylon membrane. Blots were hybridized with a randomly primed ³²P-labeled NOS-2 cDNA probe (SmaI and EcoRI digestion of pGEM plasmid containing the cloned murine macrophage NOS cDNA) as described previously (11, 25). The hybridized blots were then washed and autoradiographed. To correct for differences in RNA loading, membranes were stripped and rehybridized with a full-length glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe as a control. A high resolution camera coupled to an image processor (Ultra-Lum Inc., Carson, CA) driven by One Descan densitometric software (Scanalytics) was used to perform densitometric measurements. Results are expressed in arbitrary units as the ratio of NOS-2/GAPDH expression. For mRNA stability analysis, RPE cells were stimulated with LPS (1 µg/ml) and IFN- (100 units/ml) in the presence or absence of either IFN- or IFN- for 18 h. Actinomycin D (5 µg/ml) was added, and total RNA was prepared at the time indicated and further processed for Northern hybridization as described above.

¹²⁵I-IFN- Binding Studies

IFN- was iodinated according to the chloramine T method as described previously (26). The specific activity of the labeled IFN- obtained was usually 25,000 cpm/ng. RPE cells were grown to confluence in 24-well dishes in the presence of Dulbecco's modified Eagle's medium. Cultures were transferred at 4 °C, washed with PBS, and then incubated 30 min at 4 °C in binding buffer (serum-free Eagle's modified medium, 25 mM Hepes, pH 7.4, and 0.1% bovine serum albumin). Cells were then washed with ice-cold PBS and incubated at 4 °C with ¹²⁵I-IFN- at 20 units/ml in the absence or in the presence of increasing amounts of IFN-, IFN-, or IFN-. After 3 h, the cells were washed with ice-cold PBS and then solubilized for 15 min in 0.1 N NaOH (26). Cell-bound radioactivity was analyzed using a  scintillation counter.

One µg of RNA was reverse-transcribed for 90 min at 42 °C with 200 units of superscript Moloney murine leukemia virus reverse transcriptase (Life Technologies SARL, Eragny, France), using random hexamers, and 2 µl of cDNA were added to each PCR reaction, as described previously (27). Amplification was performed as follows: 94 °C for 2 min; 24 cycles for IRF-1 and GAPDH and 30 cycles for IRF-2 (number of cycles that were below saturating conditions) at 94 °C for 30 s, 55 °C for 30 s, 72 °C for 45 s; and then 72 °C for 2 min. The amplified fragments were separated on a 1.2% agarose gel and transferred onto Nylon membrane (Amersham, Les Ulis, France). Specificity of the amplification process was verified by hybridization of blots with ³²P-labeled specific internal oligonucleotide probe, washed three times in 1 × SSC, 0.1% SDS at 50 °C, and exposed to x-ray films. The intensity of the bands was quantified using densitometric measurements as describe above and expressed in arbitrary units as the ratio of IRF-1/GAPDH or IRF-2/GAPDH expression.

The nucleotide sequences of the oligonucleotide primers specific for mouse IRF-1 used for RT-PCR and that of hybridization probes re as follows: IRF-1 antisense (CTGGCAGGGAGTTCATGGCAC), IRF-1 sense (CTGGCTAGAGATGCAGATTAATTC), IRF-1 hybridization probe (TGTTCCGGAGCTGGGCCATTCACACAGG), IRF-2 antisense (AGATGACTCAACTGGTTCTTGC), IRF-2 sense (AACTGACGGGCTTTCATTTCCA), IRF-2 hybridization probe (ACCTTGCGGGATTGTATTGGTAGCGTG), GAPDH antisense (ATGGCATGGACTGTGGTCAT), GAPDH sense (ATGCCCCATGTTTGTGATG), and GAPDH hybridization probe (GCTGACAATCTTGAGGGAGTTGTCATATTT).

Whole cell extracts were prepared from cultured bovine RPE cells treated with various agents for 30 min or for 4 h, for NF-B analysis or for IRF-1 analysis, respectively. Cells were washed three times in cold PBS and lysed in Hepes (10 mM, pH 7.9) at 4 °C containing 0.1 mM EDTA, 5% glycerol, 0.4 M NaCl, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 µg/ml leupeptin, 4 µg/ml aprotinin, 1.5 µg/ml pepstatin, 1 µg/ml chymostatin, and 2 µg/ml antipain. After centrifugation for 15 min at 37,000 rpm, the fractions were stored at 80 °C. The protein concentrations of the extracts were determined by the Bradford procedure. Double-stranded consensus oligonucleotides NF-B (CTAGACAGAGGGGATTTCCGATTCCGAGAGGT) or IRF-1 (GGAAGCGAAAATGAAATTGACT) were end-labeled by Klenow polymerase (Appligene) and -³²P-labeled cytidine triphosphate. The binding reactions were carried out by incubating extracts (30 µg) with -³²P-labeled NF-B or IRF-1 consensus oligonucleotides (10,000 cpm) in a buffer containing 40% Ficoll, 200 mM Hepes, pH 7.5, 600 mM KCl, 20 mM dithiothreitol, 0.1% Nonidet P-40, 1 mg/ml bovine serum albumin, and 10 µg of salmon sperm DNA for 20 min at room temperature. The reaction mix was then loaded onto a 0.5 × Tris borate-EDTA (25 mM Tris, 44 mM borate, 0.5 mM EDTA), 5% polyacrylamide gel and electrophoresed at 200 V at room temperature. The gel was dried and autoradiographied. Autoradiograms were analyzed by densitometry as described above.

Results were expressed as mean ± S.E. They were analyzed statistically by Mann Whitney U test. p values less than 0.05 were considered as significant.

RESULTS

Effects of IFN- and IFN- on NO Production in Bovine RPE Cells

We previously reported that co-addition of LPS with IFN- to bovine RPE cells induced NO production, while LPS or IFN- alone had no effect (12, 24). To determine whether IFN- and IFN- could regulate the production of NO, RPE cells were incubated with human IFN- (hIFN-), human IFN- (hIFN-), or bovine IFN- (bIFN-) alone or combined with LPS. No nitrite release was detected with the different combinations, LPS/hIFN- (<1 µM), LPS/hIFN- (1 ± 0.3 µM), and LPS/bIFN- (1.5 ± 0.9 µM), in contrast to the co-stimulation LPS and IFN-, which gave 21 ± 2.8 µM. More interestingly, when cells were coincubated with hIFN-, hIFN-, or bIFN- and NOS inducers (LPS/IFN-), the production of nitrite was markedly reduced (Fig. 1A). Inhibition of LPS/IFN--induced nitrite release by hIFN-, hIFN-, and bIFN- was dose-dependent. The interferon concentrations causing 50% inhibition of nitrite release over 72 h averaged 7 units/ml for hIFN-, 300 units/ml for hIFN-, and 0.3 unit/ml for bIFN-.

Effect of IFN- and IFN- on nitrite release by LPS- and IFN--stimulated RPE cells.

[View Larger Version of this Image (25K GIF file)]

We therefore determined how long RPE cells had to be exposed to IFN- and IFN- to inhibit LPS/IFN--stimulated nitrite formation. The inhibitory effect of hIFN- and hIFN- could also be observed (Fig. 1B) when cells were pretreated for 12 h with these IFNs before the addition of the inducers (LPS and IFN-), but not as well as that observed when IFN- or IFN- were continuously present with LPS/IFN-. The addition of hIFN- and hIFN- 24 h after the treatment of RPE cells with LPS and IFN- decreased the inhibitory effect of these interferons, and when hIFN- or hIFN- were added 36 h after LPS/IFN-, no inhibitory effect was observed (Fig. 1B). These results suggested that the inhibitory response of hIFN- and hIFN- on the release of NO caused by LPS and IFN- required the presence of the inhibitors (hIFN- and hIFN-) during the early period of the induction.

Absence of NOS Activity in Extracts from Bovine RPE Cells Stimulated with IFN- and IFN-

The effect of IFNs on NOS activity in RPE cells was further characterized using cytosolic extracts isolated from cells cultured for 72 h with LPS and IFN- with or without hIFN- and hIFN-. As described previously (24), incubation of cytosols from LPS/IFN--stimulated RPE cells with L-[³H]arginine resulted in the synthesis of L-[³H]citrulline (Table I), which was inhibited by addition of L-NMMA in the cytosol extract. More interestingly, the results in Table I demonstrate that enzyme activity was largely reduced in cells previously co-incubated with hIFN- or hIFN-, at concentrations that largely prevented nitrite and citrulline release in the culture medium (see above).

Table I. Reduction of NOS-2 activity in cytosolic extracts of RPE-stimulated cells by IFN- and IFN- Confluent cells were incubated for 72 h in fresh Dulbecco's modified Eagle's medium DMEM alone (control) or in the presence of LPS (1 µg/ml) plus IFN- (100 units/ml) with or without hIFN- (100 units/ml) or hIFN- (103 units/ml). After this time cells were harvested by trypsinization, and formation of L-[3H]citrulline was determined in different cell extracts in the absence or in the presence of L-NMMA (0.1 mM) as described under "Materials and Methods." Values are means ± S.E. for three different experiments, each done in duplicate. L-[3H]Citrulline   L-NMMA + L-NMMA % of total radioactivity Control 0.3  ± 0.1 0.2  ± 0.15 LPS + IFN- 8.6  ± 0.1 1.1  ± 0.20a LPS + IFN- + hIFN- 2.2  ± 0.3b 0.8  ± 0.10a LPS + IFN- + hIFN- 1.5  ± 0.4b 0.8  ± 0.05a a p < 0.01, significantly different from corresponding incubation in the absence of L-NMMA. b p < 0.005, significantly different from the stimulation with LPS/IFN- alone.

Decrease of NOS-2 Protein by IFN- and IFN-

To determine whether the inhibitory effect of IFN- and IFN- on LPS/IFN--induced NOS activity is due to the direct inhibition of NOS expression or to an indirect effect involving the synthesis of a coenzyme necessary for enzymatic activity, the expression of inducible RPE NOS protein was investigated by Western blot analysis after 72 h of treatment with LPS and IFN- with or without hIFN- and hIFN- (Fig. 2). In LPS/IFN--treated cells, but not in untreated cells, a band at 130 kDa, corresponding to the size described for NOS-2 protein (2, 22, 25) was observed. Furthermore, densitometric analysis revealed that a simultaneous treatment with hIFN- or with hIFN- greatly decreased the 130-kDa protein signal by 92 and 71.5%, respectively, compared with LPS/IFN- (Fig. 2), demonstrating that IFN- and - markedly prevent NOS protein expression.

Effect of IFN- and IFN- on NOS-2 protein.

[View Larger Version of this Image (57K GIF file)]

Decrease of NOS-2 mRNA Accumulation by IFN- and IFN-

The expression of NOS-2 messenger was investigated by Northern blot to determine if the decrease of inducible RPE NOS protein could result from a decrease in NOS-2 mRNA accumulation. Total RNA was extracted from RPE cells after 24 h of treatment, corresponding to the maximal expression of NOS-2 mRNA (25). Only one detectable mRNA signal at 4.4 kilobase pairs was detected, while in the unstimulated RPE cells, NOS-2 mRNA was not detectable (Fig. 3), as we recently reported (25). Inclusion of hIFN- or - in the culture medium inhibited to a large extent the expression of mRNA of inducible NOS in RPE cells stimulated with LPS/IFN- in concordance with NOS activity evaluated by nitrite release (Fig. 3). A similar inhibition was obtained with bIFN- (data not shown). Hybridization with a probe for GAPDH revealed a single message at about 1.4 kilobase pairs, with equal intensity in all lanes, demonstrating a similar RNA loading throughout.

Down-regulation of NOS-2 mRNA by IFN- and IFN-.

[View Larger Version of this Image (27K GIF file)]

The effect of IFNs on NOS-2 mRNA stability was assessed by experiments using actinomycin D. Total RNA was isolated at various times after the addition of actinomycin D and examined by Northern blot successively for NOS-2 and GAPDH, to correct the loading differences. Fig. 4 shows that the addition of hIFN- did not decrease the stability of NOS-2 mRNA induced by LPS and IFN-. In experiments not reported here, similar results were obtained with hIFN- and bIFN-.

Effect of IFN- and IFN- on NOS-2 mRNA stability in RPE stimulated cells.

[View Larger Version of this Image (30K GIF file)]

Absence of Effect of hIFN- or hIFN- on IFN- Receptors

Even though IFN type I ( and ) and IFN- use different receptors, we tested whether IFN type I-induced modulation of IFN- receptors could be involved in the inhibition of NOS-2 mRNA accumulation, by evaluating the IFN- binding to RPE cells. Confluent cells were incubated at 4 °C for 3 h with ¹²⁵I-labeled IFN-, in the presence of increasing concentrations of cold hIFN-, human or bovine IFN-, or bovine IFN-. As depicted in Fig. 5, a large amount of unlabeled bovine IFN- displaced the binding for their receptors, while hIFN-, hIFN-, and bIFN- were unable to compete with IFN- for binding in their receptors.

Effect of IFN- and IFN- on the binding of IFN- on RPE cells.

[View Larger Version of this Image (14K GIF file)]

Absence of Effect of hIFN- on NF-B Activation

Because the NOS-2 gene promoter region has been shown to contain NF-B consensus sequences (28) and because NF-B was activated in RPE cells after LPS/IFN- treatment,² we examined the effect of IFN- on NF-B activation by electrophoretic mobility gel-shift assays. Fig. 6 shows a rapid activation of NF-B after LPS and LPS/IFN- stimulation. The addition of excess unlabeled consensus oligonucleotide completely prevented the complex formation, demonstrating the specificity of the DNA/protein interaction. Coincubation with hIFN- did not change the LPS/IFN--induced NF-B activation (Fig. 6, lane 3).

Effect of IFN- on the NF-B activation.

[View Larger Version of this Image (24K GIF file)]

Effect of hIFN- on IRF-1 Activation

We tested if stimulation of RPE cells with IFN- modify the activation of IRF-1, a transcriptional factor involved in NOS-2 induction (29, 30). EMSA analysis (Fig. 7) revealed the presence of an induced DNA-protein complex in extracts of RPE cells stimulated with IFN-. This complex was absent in control (lane 1) and in LPS-treated cells (lane 2). The presence of IFN- increased the formation of this complex (lane 3), while hIFN- alone had no effect (lane 4). The formation of this complex was potentiated by co-addition of LPS and IFN- (lane 6) and was prevented by the addition of excess unlabeled IRF-1 oligonucleotide (lane 8), demonstrating the specificity of the DNA-protein interaction. Densitometric analysis revealed that the amount of the complex observed after IFN- and LPS/IFN- stimulation is largely decreased in the presence of hIFN- (lanes 5 and 7, compared, respectively, to lanes 3 and 6), indicating that this interferon induces a decrease of IRF-1 binding to its specific DNA target sequence.

Effect of IFN- on the IRF-1 activation.

[View Larger Version of this Image (51K GIF file)]

Differential Regulation of IRF-1 and IRF-2 mRNA Accumulation by hIFN-

Total RNA was extracted from RPE cells after 3 h of treatment, corresponding to the maximal accumulation of IRF-1 mRNA.² RT-PCR analysis showed that IRF-1 mRNA was induced in RPE cells by IFN- alone and by LPS/IFN-, while hIFN- alone had no significant effect (Fig. 8). IRF-1 mRNA accumulation induced by IFN- or by the combination LPS/IFN- was decreased by hIFN- treatment at concentrations that completely abrogated NOS-2 expression (Fig. 8).

[View Larger Version of this Image (28K GIF file)]

Since IRF-2 has been demonstrated to function as a repressor of interferon-stimulated genes (31), we have investigated the effect of hIFN- on its mRNA level. RT-PCR analysis, depicted in Fig. 9, demonstrated that IRF-2 mRNA expression was constitutive in RPE cells and that hIFN- treatment increased its accumulation, while IFN- largely decreased the level of IRF-2 mRNA. The combination LPS/IFN-, responsible for NOS-2 induction, also decreased the IRF-2 mRNA level. Furthermore, hIFN-, which inhibited NOS-2 induction, counteracted the decrease of IRF-2 mRNA induced by IFN- or by LPS/IFN- (Fig. 9).

Differential regulation of IRF-2 mRNA accumulation by IFN- and hIFN- in RPE cells.

[View Larger Version of this Image (30K GIF file)]

DISCUSSION

In this work, we demonstrate that type I IFN (IFN- and IFN-) can suppress the LPS/IFN--dependent expression of NOS activity in bovine RPE cells as assessed by nitrite accumulation in the culture medium. Measurements of L-citrulline synthesis by cytoplasmic extracts further support the conclusion that levels of NOS activity are largely decreased. This loss of activity reflected the absence of NOS protein in the IFN-- and IFN--treated RPE cells, as revealed by Western blot. Finally, the marked inhibition by IFN- and IFN- of the LPS/IFN--induced increase in NOS mRNA suggests that IFN- and IFN- affect the regulation of the inducible form of NOS at the mRNA level. The detected levels correlated with the levels of NOS protein activity. Experiments with actinomycin D revealed that hIFN- and hIFN- do not decrease the half-life of NOS-2 mRNA in RPE cells stimulated with LPS/IFN-, demonstrating that the decreased expression of NOS-2 observed after hIFN- or hIFN- treatment is not attributable to decreased mRNA stability. Therefore, inhibition of the transcription rate of the NOS-2 gene seems to be the most likely mechanism involved. The negative regulation of NO synthase induction was previously reported (2) for different cytokines and growth factors, such as interleukin-4 and -10, fibroblast growth factor, and transforming growth factor  (2, 24, 25, 32).

Our results contrast with those of previous studies (16-18, 33, 34), in which IFN- and IFN- were generally considered to be activators of NOS induction. Indeed, exogenous addition of IFN- or IFN- on murine peritoneal macrophages could induce nitrite release (16, 17), and blockade of endogenous IFN- or IFN- production could reduce LPS- or IFN--induced NOS-2 mRNA accumulation in macrophages (33, 34). In preliminary experiments not reported herein, we have tested the ability of IFN- to regulate the nitrite release in peritoneal macrophages from thioglycollate-treated mice. We have found that IFN- is unable to potentiate nitrite release due to LPS or IFN- and had no effect on LPS/IFN--induced nitrite release, suggesting that the inhibitory effect of IFN- on NOS-2 induction in RPE cells is specific from the species and/or from the cell type. However, this phenomenon is not exclusive to RPE cells since recent reports demonstrated the ability of IFN- and IFN- to largely reduce the IFN--mediated nitrite release in rat peritoneal macrophages (35) and in human thyrocytes (36). In rat macrophages this decrease of nitrite production was due to a down-regulation of NOS-2 mRNA (37), as in bovine RPE cells.

Concerning the molecular mechanism which operates for the suppressive action of NOS-2 by IFN- and -, we have postulated that they may interrupt some steps in the signaling pathway by which LPS/IFN- induces NOS-2 gene expression. First, on the basis of our binding experiments the inhibition of NOS induction by type I IFNs was not due to a competition between IFN- or IFN- with IFN- at the receptor level. We have also looked for the activation of two transcription factors, NF-B and IRF-1, involved in NOS-2 induction (28-30). We demonstrated, by EMSA, that IFN- did not inhibit LPS/IFN--induced NF-B activation, indicating that the inhibitory effect of this compound on LPS/IFN--induced NOS-2 appeared to be independent of NF-B activation. The down-regulation of NF-B activity by other compounds could be an inhibitory regulatory pathway for NOS induction in RPE cells, as it has been recently reported for the inhibition of NOS-2 induction by the glucocorticoids in human alveolar epithelium A549/8 cells (38).

NF-B is not the only regulatory factor of NOS-2 gene expression, but could function in concert with several other transcription factors, such as IRF-1 (29, 30). We demonstrated in this study that treatment of RPE cells with IFN- alone or with IFN- plus LPS resulted in an increase of IRF-1 mRNA and in an appearance of a prominent IRF-1-specific DNA-protein complex, which were clearly inhibited in the presence of IFN-. Interestingly, a similar result has been reported by Politis et al. (39) with another member of interferon-induced transcription factor, the interferon consensus sequence binding protein. They demonstrated that IFN- could suppress the IFN--induced interferon consensus sequence binding protein mRNA accumulation in peritoneal macrophages. Besides these interferon-induced transcription factors, there is another transcription factor, termed IRF-2, that is structurally similar to IRF-1, but that represses the effect of IRF-1 (31, 40). We demonstrated that IRF-2 mRNA is differentially regulated by the IFNs, since IFN- largely decreased its accumulation, while IFN- significantly increased it. Furthermore, IFN- is able to prevent the IFN-- and the LPS/IFN--induced IRF-2 mRNA decrease. These results with IRFs demonstrated that inducers of NOS-2 (LPS/IFN-) increased IRF-1 and decreased IRF-2 mRNA, while the inhibitor of NOS-2 induction (IFN-) decreased IRF-1 and increased IRF-2 mRNA. A very similar regulation of IRF-1 has been recently reported in murine macrophages by NOS inducers and inhibitors (41, 42). It would be consistent to propose that IFN-, by increasing IRF-2 mRNA and decreasing IRF-1 mRNA accumulation, could favor the neosynthesis of the repressor IRF-2, which is more stable than IRF-1 (43). As previously suggested, IRF-1 and IRF-2 could compete for the specific ISRE sequence (31, 40). Then, the accumulation of IRF-2 and diminution of IRF-1 could favor the interaction of IRF-2 with the NOS-2 gene promoter in place of IRF-1 and could suppress the activation of NOS-2 gene by IRF-1, after IFN- treatment. It has been recently reported that closed tyrosine kinases and transcription factors in the JAK/STAT pathway could be involved in the transduction of the IFNs signals, leading to the transcription of early response genes (reviewed in Darnell et al. (44)). There are at least two distinct sets of genes that can be stimulated in response to IFNs: those containing GAS elements, which bind STAT-1 homodimeres, and those containing ISRE sequences, which bind STAT1 homodimeres or STAT1/STAT2 heterodimeres coupled to p48 (ISGF3), respectively, named STAT1-p48 complex and ISGF3 (45). IFN- activates both STAT1 and STAT2 by phosphorylation on tyrosine residues, while IFN- phosphorylated tyrosines only on STAT1, leading to a difference in complex assembly that results in a functional difference in gene expression (45). In this context Li et al. (46) recently reported that activation of IRF-1 transcription in response to IFN could be modulated by the level of p48. It is conceivable that IFN- and - might counteract the initial phase of the IFN- signaling cascade in bovine RPE cells, possibly by a phosphorylation-dephosphorylation process and/or by a competition between the different STAT complexes (44-46).

Although NO might act as an antimicrobial and antiviral effector molecule in the retina (47), the sustained NO release by NOS-2 may cause retinal cell damage as we previously observed in the case of retinal light damage (48) or during endotoxin-induced uveitis (27). Type I IFNs are produced by various cell types in the early phase of viral or bacterial infection. Thus, the production of IFN- and - during viral infection could decrease the potential antiviral NO activity (49) and might weaken the antiviral defense in the retina. This could represent a mechanism by which the virus protects itself against a large release of NO. This hypothesis is in agreement with recent results in macrophages (50), in which the viral infection alters the induction of NOS-2, by increasing the synthesis of endogenous IFN- and -. However, we cannot exclude that actions of IFN- and - on RPE cells might also prevent NO-mediated tissue damage induced by LPS and IFN- in addition to an important role in the regulation of immune response in the retina.