Regulation of Prostaglandin Biosynthesis by Nitric Oxide Is Revealed by Targeted Deletion of Inducible Nitric-oxide Synthase*

We investigated the effects of targeted deletion of the inducible NO synthase (iNOS) gene on the formation of prostaglandins in vivo and ex vivo . Peritoneal macrophages were obtained from control and iNOS-deficient mice, and prostaglandin E 2 (PGE 2 ) was quantified after stimulation with g -interferon and lipopolysaccharide to induce COX-2. Total nitrate and nitrite production was completely abolished in cells from iNOS-deficient animals compared with control cells. PGE 2 formation by cells from iNOS-deficient animals was decreased compared with cells from control animals 80% at 12 h (0.85 6 0.90 ng/10 6 cells versus 15.4 6 2.1 ng/10 6 cells, p < 0.01) and 74% at 24 h (9.4 6 4.3 ng/10 6 cells versus 36.8 6 4.1 ng/10 6 cells, p < 0.01). COX-2 protein expression was not significantly different in cells from control or knockout animals. Levels of PGE 2 in the urine of iNOS-deficient mice were decreased 78% (0.24 6 0.14 ng/mg of creatinine versus 1.09 6 0.66 ng/mg of creatinine, p < 0.01) compared with control animals. In addition, the levels of urinary F 2 -isoprostanes, an index of endogenous oxi- dant stress, were significantly decreased in iNOS-defi-cient

A significant body of experimental evidence suggests a relationship between nitric oxide biosynthesis and prostaglandin generation (1,2). NO donors have been reported to stimulate or inhibit prostaglandin biosynthesis in a variety of cellular or broken cellular systems (3)(4)(5)(6)(7)(8)(9). Inhibitors of nitric oxide formation or depletion of the nitric oxide precursor arginine are also reported to stimulate or inhibit biosynthesis in cellular or intact animal experiments (10 -13). These effects may be mediated by altered transcription of cyclooxygenase (COX) 1 genes (14,15), by inhibition or stimulation of cyclooxygenase activity (9,16), or by inactivation of downstream metabolizing enzymes that convert prostaglandin endoperoxides to stable eicosanoid products (17). It is likely that NO does not mediate all of these actions but that NO-derived species such as peroxynitrite or nitrosothiols may be responsible (17,18). For example, we have demonstrated that peroxynitrite activates the cyclooxygenase activity of COX-1 and COX-2 by acting as a substrate for the peroxidase activity of each enzyme (18).
The complexity of possible interactions between NO or NOderived species and expression or activation of enzymes of arachidonic acid metabolism makes it difficult to extrapolate results from in vitro experiments to in vivo settings. Although inhibitors of NO biosynthesis and prostaglandin biosynthesis are available, specificity issues frequently complicate the interpretation of pharmacological investigations of complex metabolic interactions. Therefore, we have used a genetic approach to probe for linkage between NO biosynthesis and prostaglandin formation. We have determined the effect of targeted deletions of the inducible NO synthase gene on prostaglandin biosynthesis in vivo and ex vivo. Three different contexts in which prostaglandins are biosynthesized were probed. First, peritoneal macrophages from iNOS knockout animals were stimulated with LPS and ␥-interferon to induce COX-2 activity, and the levels of PGE 2 released were determined. Second, urine samples from iNOS knockout animals were analyzed for prostaglandin content as an indication of biosynthesis in the kidney. Third, blood samples from iNOS knockout animals were analyzed for levels of thromboxane, an indicator of COX-1 activity in the platelet. The results of these ex vivo and in vivo experiments mirror the complexity of interactions established from in vitro experiments but clearly establish a role for NO or NO-derived species in the control of prostaglandin biosynthesis.

MATERIALS AND METHODS
iNOS-deficient Mice-iNOS-deficient mice were generated as described (19) and were kindly provided by John MacMicking and Carl Nathan (Cornell University School of Medicine) and John Mudgett (Merck Research Laboratories). Control mice were C57BL/6J (Jackson Laboratories, Bar Harbor, ME) ϫ 129Sv/Ev (Taconic Laboratories, Germantown, NY). All mice used for studies were 5-week-old males. The mice were placed on a low nitrate diet of AIN-76 (Bio-Serve, Frenchtown, NJ) for 2 days prior to and during the urine collection.
Preparation of Peritoneal Macrophages-Peritoneal macrophages were elicited by intraperitoneal injection of 1-3 ml of 3% sterile thioglycolate medium (Sigma). After 3 days mice were sacrificed, and macrophages were harvested as described (20). Cells were plated on tissue culture plates for 2 h, and then nonadherent cells were removed by washing with sterile phosphate-buffered saline. The cells were treated with ␥-interferon (500 units/ml) (Genzyme, Cambridge, MA) * This work was supported by National Institutes of Health Grants CA47479, CA26731, DK48831, GM15431, GM42056, DK26637, and CA77839. 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.
§ To whom correspondence should be addressed. Quantification of Total Nitrate/Nitrite-Total nitrate and nitrite generation by peritoneal macrophages was quantified by colorimetric determination in medium aliquots using the Griess reagent as described (21).
Quantification of PGE 2 -PGE 2 was quantified in medium from cell incubations by gas chromatography/negative ion chemical ionization mass spectrometry (22).
Detection of COX-2 Protein by Western Blot Analysis-Frozen pellets of peritoneal macrophages (3 ϫ 10 6 ) were thawed prior to addition of 125 l of radioimmune precipitation buffer (phosphate-buffered saline, 1% Nonidet P-40, 0.1% SDS, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin A, and 1 g/ml chymostatin) (Sigma). Each sample was passed through a 22-gauge needle to shear the DNA, and samples were incubated on ice for 30 min before centrifugation at 15,000 ϫ g for 20 min at 4°C. Supernatants (5 g of total protein) were heated with Laemmli reducing sample buffer for 3 min at 95°C, run on a 7% Nu-Page Tris acetate polyacrylamide gel (NOVEX, San Diego, CA) for 45 min at 160 V, and transferred onto a polyvinylidene difluoride membrane (Millipore) for 2 h at 30 V. The blot was probed overnight with goat anti-COX-2 (C-20, 1:750, Santa Cruz Biologicals, Santa Cruz, CA) and then for 1 h with rabbit anti-goat horseradish peroxidase conjugate (1:3000, Pierce). Detection was carried out with the ECL Western blotting system (Amersham Pharmacia Biotech), and quantification was by densitometry (18).
Quantification of Eicosanoids in Vivo in Mice-PGE 2 and F 2 -isoprostanes (F 2 -isoPs) were measured in timed urine collections from either control or iNOS-deficient mice. F 2 -IsoPs were quantified by gas chromatography/mass spectrometry as described (23). Results are expressed per mg of creatinine. Urinary creatinine concentration was measured by the sodium picrate method with an AutoAnalyzer II (Technicon, Tarrytown, NY). TxB 2 was measured in mouse serum ex vivo after platelet activation by endogenous thrombin during whole blood clotting at 37°C as described using gas chromatography/negative ion chemical ionization mass spectrometry (24).
Statistical Analysis-Where appropriate, data were analyzed for statistical significance using Student's t test. Differences were considered significant if p Ͻ 0.05. Fig. 1 demonstrates the generation of PGE 2 in peritoneal macrophages from iNOS-deficient and control mice after treatment with ␥-interferon and LPS to induce COX-2. PGE 2 is a major eicosanoid produced by these cells and was quantified in cell medium either 12 (Fig. 1A) or 24 h (Fig. 1B) after LPS. As is evident, unstimulated peritoneal macrophages from either WT or iNOS-deficient animals produced only small amounts of PGE 2 . In contrast, activation with ␥-interferon and increasing concentrations of LPS significantly increased PGE 2 production in peritoneal macrophages from WT cells in a concentration-dependent manner. After 12 h, the maximum increase in PGE 2 formation was 84-fold, and after 24 h, it was 133-fold. Treatment of peritoneal macrophages from iNOSdeficient animals also resulted in increases in PGE 2 production (11-fold at 12 h and 19-fold at 24 h) although these increases were significantly less (p Ͻ 0.05 for all times and concentrations of LPS) than those observed in WT cells.

Generation of PGE 2 by Peritoneal Macrophages from iNOSdeficient Mice-
We then sought to determine whether the decrease in PGE 2 generation in peritoneal macrophages from iNOS-deficient mice compared with WT animals was because of a reduction in cyclooxygenase protein expression. Thus, we analyzed COX-2 protein in peritoneal macrophages from iNOS-deficient and WT animals. Fig. 2 displays the results after 24 h of treatment with varying concentrations of LPS. As is apparent, there is no significant difference in COX-2 expression between iNOS-deficient and control animals despite a marked difference in PGE 2 generation. Previous studies have shown that LPS treatment of macrophages does not alter COX-1 expression (25).
We also quantified nitrate and nitrite production in peritoneal macrophages from iNOS-deficient and WT mice. The ability of the cells from iNOS-deficient mice to secrete nitrate and nitrite into the medium was completely abolished relative to cells from WT mice. Medium alone contained approximately 8 -14 M nitrate/nitrite, and this was not increased in medium containing cells from iNOS-deficient mice.
Taken together, these data provide evidence that PGE 2 production is significantly decreased in activated peritoneal macrophages from iNOS-deficient mice compared with macrophages from WT animals. Further, this decrease is not because of a reduction in COX protein expression.
Formation of Eicosanoids in Vivo in iNOS-deficient and WT Mice-Because ex vivo formation of PGE 2 is significantly decreased in peritoneal macrophages from iNOS-deficient animals compared with WT mice, we undertook studies to determine whether eicosanoid formation was altered in vivo in these animals. Therefore, we quantified urinary PGE 2 production in either iNOS-deficient or WT mice. The major source of PGE 2 in the urine is the kidney (26). The results are summarized in Fig.  3. Collections were timed, and PGE 2 values were expressed based on creatinine clearance to account for any differences in renal function between individual animals. As is evident, iNOS-deficient mice excreted significantly lower amounts of PGE 2 (mean decrease of 78%, p Ͻ 0.01) than did WT animals.
In addition, we also quantified urinary excretion of F 2 -isoPs in iNOS-deficient and WT mice. We have previously shown that excessive peroxynitrite formation in vitro and in vivo is associated with increased F 2 -IsoP generation (27,28). F 2 -IsoPs are PGF 2 -like compounds derived from the free radical-catalyzed peroxidation of arachidonic acid independent of the cyclooxygenase (29, 30). They are highly accurate indices of oxidant stress in vivo in animals and humans. As shown in Fig. 4, urinary levels of F 2 -IsoPs are significantly decreased in iNOSdeficient mice compared with WT animals, although the decrease is less than that of PGE 2 in the same mice. These data suggest that NO and/or peroxynitrite may regulate oxidant tone in particular tissues to a significant extent in vivo.
We also measured the effect of iNOS deficiency on platelet TxB 2 formation. TxB 2 is the major PG produced by platelets and is derived entirely from COX-1 in this cell type (31). It has been postulated that endogenous NO formation is important in vascular homeostasis, and indeed NO inhibits platelet aggregation (32)(33)(34)(35). Excessive TxB 2 formation is associated with platelet aggregation. Therefore, we assessed the formation of TxB 2 in platelets from iNOS-deficient and WT mice allowed to aggregate ex vivo. The results are shown in Fig. 5. Interestingly, unlike urinary excretion of PGE 2 , levels of TxB 2 derived from platelet aggregation ex vivo are not decreased in iNOSdeficient animals but are significantly increased. These data suggest that NO may regulate platelet thromboxane formation but by a mechanism different from that observed in the kidney of mice in vivo.

DISCUSSION
The current study provides strong genetic evidence for a functional link between NO biosynthesis and prostaglandin biosynthesis. Mice bearing targeted deletions in the iNOS gene demonstrated a dramatically reduced ability to synthesize prostaglandins in stimulated peritoneal macrophages and in kidney but a significantly enhanced ability to synthesize thromboxane in platelets. The decreases in PGE 2 production by macrophages and kidney are consistent with previous reports 1) that administration of NO synthase inhibitors to rats reduces the production of prostaglandins in inflammatory lesions (4,36), 2) that inflamed brain tissue from iNOS knockout mice displays reduced PGE 2 synthesis relative to wild-type mice (12), and 3) that arginine depletion decreases prostaglandin synthesis in mouse macrophages stimulated with LPS (3, 10). These observations point to a stimulatory effect of NO or an NO-derived species on prostaglandin biosynthetic capacity in inflammatory cells and, with the present report, in the kidney.
The biochemical mechanism by which NO stimulates prostaglandin synthesis has been the subject of much investigation. Despite evidence suggesting a direct effect of NO on COX enzymes (3), incubation of purified protein with NO sources does not stimulate activity in most reported studies (9,16). NO alters the expression of COX genes in some cell types either at the transcriptional or posttranscriptional level, but a consistent pattern of stimulation or inhibition is not observed (15,37). The effects of NO on COX transcription appear to be dependent on the cell type and the signal transduction pathway that NO modulates. In the present study, analysis of COX-2 protein levels in stimulated peritoneal macrophages suggests that no significant reduction in expression occurs in iNOS-deficient mice despite an approximate 80% reduction in PGE 2 biosynthesis in the stimulated macrophages.
We have demonstrated that peroxynitrite, the coupling product of NO and superoxide, is an efficient substrate for the peroxidase activity of COX enzymes and activates cyclooxygenase activity even in the presence of concentrations of glutathione and glutathione peroxidase that are high enough to completely prevent COX activation by fatty acid hydroperoxides and H 2 O 2 (18). Activation can be achieved by bolus addition of peroxynitrite or by in situ generation from NO and superoxide sources. Activation by NO and superoxide is completely inhibited by superoxide dismutase. Furthermore, membrane-permeant superoxide dismutase-mimetic agents reduce prostaglandin biosynthesis by 85% in mouse macrophages stimulated with LPS and ␥-interferon (18). This extent of inhibition is comparable with the reduction in prostaglandin biosynthesis observed in stimulated peritoneal macrophages from iNOS-deficient animals. Thus, it is conceivable that the reduction in PGE 2 synthesis observed in iNOS knockout mice is because of a reduced ability to generate peroxynitrite in inflammatory cells. A similar explanation may account for the reduction of PGE 2 secretion into urine, which presumably reflects a reduction in renal prostaglandin biosynthetic capacity. However, it is noteworthy that superoxide dismutase does not inhibit prostaglandin biosynthesis in the carrageenan model of inflammation despite the fact that it exhibits antiinflammatory activity (38). The lack of inhibition of prostaglandin production by superoxide dismutase may reflect the latter's inabilty to penetrate cell membranes.
In addition to a decrease in the urinary excretion of prostaglandins in iNOS knockout animals, we found a significant reduction in the levels of F 2 -IsoPs, products of the free radicalcatalyzed peroxidation of arachidonic acid. Urinary IsoPs, like prostaglandins, may originate in the kidney, and previous studies by us have revealed that peroxynitrite is a potent inducer of IsoP formation both in vitro in low density lipoprotein or in vivo in the plasma of rats deficient in superoxide dismutase (27,28,39). Therefore, our data suggest that iNOS may regulate, to an extent, IsoP formation in the kidney. These findings contrast with a previous study in which we reported that feeding of aminoguanidine to aged rats did not significantly alter renal tissue IsoP levels (40). However, the reduction of urinary nitrate/nitrite in that study was only ϳ50%, compared with the complete inhibition of production of nitrate/ nitrite in macrophages from iNOS-deficient mice in the present study. Thus, it may be the case that the decreases in NO/ peroxynitrite formation in the kidney of iNOS-deficient mice result in a decreased excretion into the urine of F 2 -IsoPs.
The stimulation of TxB 2 production during ex vivo aggregation of platelets from iNOS-deficient animals relative to WT may appear paradoxical but is consistent with several reports demonstrating that NO inhibits platelet aggregation. In fact, administration of the NO source, isosorbide dinitrate, to humans inhibits platelet aggregation and the ability of platelets to generate TxB 2 . Likewise, S-nitrosocysteine reduces TxB 2 biosynthesis by intact platelets in a cGMP-independent fashion. The precise locus of NO action is unknown but is unlikely to be platelet COX-1 because several studies indicate that NO does not inhibit this enzyme in vitro (9,16,18). An intriguing possibility is that NO inhibits thromboxane synthase, a P450 type hemeprotein. P450 hemeproteins are quite sensitive to peroxide inactivation (41)(42)(43). Thus, deletion of iNOS would lower the synthesis of an endogenous inhibitor of TxB 2 synthesis, as is observed. The inability of platelets to synthesize new protein to replace inactivated thromboxane synthase would account for the fact that differences in the levels of TxB 2 between WT and iNOS-deficient mice could be detected following ex vivo platelet aggregation. The major determinant of TxB 2 levels generated ex vivo is the amount of active thromboxane synthase in the platelet at the time of removal of blood from circulation, and this would be elevated in iNOS-deficient mice.
In summary, our study provides evidence that a targeted deletion of iNOS with a resultant reduction in NO or an NOderived molecular species, leads to significant alterations in prostaglandin and IsoP formation in vivo. This strengthens the tie between NO biosynthesis and eicosanoid biosynthesis but indicates that multiple pathways exist by which NO or NO derivatives interact with endoperoxide-generating or -metabolizing enzymes. This provides great flexibility for regulation of arachidonic acid-dependent signal transduction and multiple opportunities for pharmacologic manipulation of the NOarachidonate interrelationship.