Molecular Mechanisms for Lipopolysaccharide-induced Biphasic Activation of Nuclear Factor- (cid:1) B (NF- (cid:1) B)*

The nuclear factor- (cid:1) B (NF- (cid:1) B) is an important transcription factor necessary for initiating and sustaining inflammatory and immune reactions. The inducers of NF- (cid:1) B are well characterized, but the molecular mechanisms underlying multiple in vivo NF- (cid:1) B activation processes are poorly understood. The injection of lipopolysaccharide resulted in a biphasic activation of NF- (cid:1) B during the 18-h observation period in various organs of mice. The early and late phases of NF- (cid:1) B activation occurred at 0.5–2 h and 8–12 h, respectively. Platelet-activating factor, which is released in response to lipopolysaccharide injection, was responsible for the activation of the early phase of NF- (cid:1) B. The early NF- (cid:1) B activity led to the induction of proinflammatory cytokines, tumor necrosis factor (TNF), and interleukin (IL)-1 (cid:2) , which are known to be efficient inducers of NF- (cid:1) B. Using the TNF knockout and IL-1 receptor found in the activation of NF- cytokines the of platelet-activating the secondary activation of NF- (cid:1) B. These observations describe a novel autoregulatory molecular mechanism for the biphasic of

Gel Shift Assay-The nuclear extracts were prepared from the RAW 264.7 cells and various organs as described previously (9 -11). To inhibit endogenous protease activity, 1 mM phenylmethylsulfonyl fluoride was added. As a probe for the gel retardation assay, an oligonucleotide containing the Ig-chain-binding site (B, 5-CCGGTTAACAGAGGGG-GCTTTCCGAG-3) was synthesized. The two complementary strands were annealed and labeled with [␣-32 P]dCTP. Labeled oligonucleotides (10,000 cpm), 10 g of nuclear extracts, and binding buffer (10 mM Tris-HCl (pH 7.6), 500 mM KCl, 10 mM EDTA, 50% glycerol, 100 ng of poly (dIdC), and 1 mM dithiothreitol) were incubated for 30 min at room temperature in a final volume of 20 l. The reaction mixture was analyzed by electrophoresis on a 5% polyacrylamide gel in 0.5ϫ trisborate/EDTA buffer. Specific binding was controlled by competition with a 50-fold excess of cold B or cAMP response element (CRE) oligonucleotide. For supershift/inhibition assay, 1-2 g of specific supershifting antibodies against p50, p52, p65, RelB, or c-Rel components of NF-B were incubated with the nuclear extract on ice for 1 h before the addition of labeled oligonucleotide to the binding reaction. Signal intensity of specific bands was analyzed quantitatively using Fluor-STM Imager (Bio-Rad) and plotted as relative intensity.
Measurement of Plasma PAF-Blood was collected from the heart of anesthetized mice and placed into an Eppendorf tube containing sodium citrate (final concentration 0.38%). Plasma PAF was quantified as described previously (9, 10) by octadecyl column chromatography (Amersham Biosciences), thin-layer chromatography, and a scintillation proximity RIA kit (Amersham Biosciences).
Measurement of Serum TNF and IL-1␤-Serum TNF and IL-1␤ were quantified by using an ELISA kit from Endogen as instructed by the manufacturer.
Statistical Analysis-The data are represented as the mean Ϯ S.E. Statistical significance was determined by the Student's t test when two data sets were analyzed or, alternatively, by ANOVA followed by the appropriate post-hoc test for multiple data sets with the statistical software Staview (version 4.5). All experiments were conducted two or more times. Reproducible results were obtained and representative data are therefore shown in the figures.

RESULTS
PAF Is Involved in the Early Activation of NF-B-We first examined the kinetics of NF-B activation from 0.5 to 18 h after LPS injection in the lung. As shown in Fig. 1A, biphasic activation of NF-B was evident. The early activation of NF-B occurred within 30 min of LPS administration, peaked at 1 h, and subsequently declined to basal level by 6 h. A second phase with relatively lower levels of activity appeared at 8 -12 h post-LPS injection. Pretreatment with BN 50739 or CV 6209 prior to LPS injection resulted in abrogation of the early peak of NF-B as well as the subsequent second phase of NF-B activation. Similar effects of the PAF antagonists were seen in the spleen, liver, and kidneys (data not shown). These observations confirmed our previous data showing that PAF is responsible for the LPS-induced early activation of NF-B (9, 10) and suggests that the overall process of NF-B activation in response to an inflammatory stimulus is multi-phasic in which the first phase of NF-B activity acts as an inducer for the next.
Gel supershift analysis was performed to determine the sub-

FIG. 1. Occurrence of LPS-induced biphasic activation of NF-B and its inhibition by pretreatment with a PAF antagonist.
A, mice were injected with LPS (50 g, intravenously), and the lungs were removed at the time indicated. BN 50739 (400 g) was administered intraperitoneally 30 min prior to LPS injection. CV 6209 (400 g) was administered intraperitoneally 1 h and 10 min prior to LPS injection. Nuclear extracts from the organs were incubated with an ␣-32 P-labeled B oligonucleotide and electrophoresed on a 5% polyacrylamide gel. Lane p contained probe incubated without extract. A 50-fold excess of cold B or cAMP response element oligonucleotide was added as competitor. B, supershift/inhibition assay was conducted with 1-2 g of anti-p50, p52, p65, RelB, c-Rel, or control rabbit antibody. C, total RNA was prepared from the lungs, and TNF, IL-1␤, IL-6 mRNA expression was measured by RT-PCR as described under "Experimental Procedures." Values are expressed as mean Ϯ S.E. n ϭ 3 animals for each time point. A representative of two to seven independent experiments is shown. unit composition of the gel-shifted complexes. Anti-p50, anti-p52, anti-p65, anti-RelB, or anti-c-Rel antibodies were added to the reaction mixtures prepared from the lungs of LPS-injected mice prior to the addition of the oligonucleotide probe. Addition of antibody specific for the p50 subunit to the assay reduced the intensity of the DNA-binding activity but with no visible shift in the mobility of the complex. The addition of anti-c-Rel antibody resulted in reduced intensity and a weaker shift in the mobility of the complex in the lungs at 1 h (Fig. 1B). In addition, a slight reduction (10 -30%) in the intensity of anti-p65, anti-p52, or anti-RelB suggests the possible involvement of these proteins also in complex formation. At 4 h after LPS injection, in gel shift assay, the complex migrated at a slower rate as compared with that of 1 h. In supershift assay, the band of NF-B complex was mainly reacted with p50 antibody and partially reacted with antibodies against other NF-B species (40 -45% reduction of the intensity in the presence of anti-p52 and anti-RelB antibodies, 30% reduction of intensity in the presence of anti-c-Rel antibody, and 20% reduction of the intensity in the presence of anti-p65 antibody, Fig. 1B). These results indicate activation of multiple NF-B species in vivo in response to LPS in which p50 is the predominant subunit, and the other subunits of NF-B, such as p52, p65, RelB, and c-Rel, may occur to a lesser degree.
We examined the transcription of several known NF-B-dependent genes, such as TNF, IL-1␤, and IL-6. We have observed that all of these genes are expressed in a biphasic fashion following LPS injection, indicating that mRNA expression of some NF-B-dependent genes parallels NF-B activation (Fig. 1C).
TNF and IL-1 Are Involved in the Late Phase Activation of NF-B-The early activation of NF-B results in the early expression of inflammatory cytokine genes such as TNF (3,9,10) and IL-1 (3). We have previously reported that blocking of the early NF-B activity by PAF antagonist resulted in the inhibition of TNF production (9) and mRNA expression (11). TNF and IL-1 are known to be potent inducers of NF-B activation (15)(16)(17). The kinetics of serum TNF and IL-1␤ levels after intravenous administration of LPS showed peak values at 1 h (TNF) or 2 h (IL-1␤) and then rapidly declined thereafter (Fig. 2). These observations suggest that TNF and IL-1 may be responsible for the second phase activation of NF-B. To assess this possibility, the effect of TNF or IL-1 inhibition was examined using TNF Ϫ/Ϫ and IL-1R Ϫ/Ϫ mice, respectively. No significant impairment of NF-B activation in either the first or second phase of activity was observed in these mice (Fig. 3A). These data suggest that either TNF or IL-1 alone is sufficient for the second phase of NF-B activation. In supershift analysis, a similar supershift pattern was also seen in both TNF Ϫ/Ϫ and IL-1R Ϫ/Ϫ mice, suggesting that TNF or IL-1 deficiency does not affect LPS-induced activation of the different NF-B subunits (Fig. 3B). We next examined the effect of inhibiting both cytokines simultaneously on the second phase of NF-B activation. Administration of anti-TNF in IL-1R Ϫ/Ϫ mice 10 min prior to LPS injection resulted in a significant inhibition of the second phase activity of NF-B (Fig. 4A). This was confirmed in in vitro experiments. Stimulation of the macrophage cell line (RAW 264.7) with LPS resulted in the biphasic activation of NF-B. Peak values were observed at 30 min and 14 -20 h, respectively (Fig. 4B). Pretreatment with BN 50739 or CV 6209 30 min prior to LPS treatment resulted in abrogation of the early peak of NF-B as well as the subsequent second phase of NF-B activation (Fig. 4B). Neither anti-TNF Ab nor anti-IL-1␤ Ab alone inhibited the LPS-induced second phase activation of NF-B in the cells. Using both antibodies did, however, inhibit the late activation of NF-B (Fig. 4C). Likewise, the second phase of LPS-induced NF-B activation in peritoneal macrophages obtained from TNF Ϫ/Ϫ and IL-1R Ϫ/Ϫ mice was blocked by pretreatment with IL-1Ra, anti-IL-1␤ Ab, and anti-TNF Ab, respectively (Fig. 4D). These data indicate that either TNF or IL-1 alone is sufficient for the second phase activation of NF-B.
TNF or IL-1 Induces the Second Phase of NF-B Activation Indirectly via the Action of PAF-Although TNF and IL-1 are involved in the secondary activation of NF-B, it is unclear whether these cytokines are directly involved or they exert their effects through the releases and/or synthesis of the secondary mediator(s). The fact that the long lag period (more than 6 -7 h) between the peak plasma levels of TNF and IL-1␤ (Fig. 2) and the second peak of NF-B activation (Fig. 1) suggests that the latter possibility seems likely. PAF and TNF and/or IL-1 stimulate the release of each other via a positive feedback loop (13, 18 -21). TNF and IL-1 are released in response to PAF (18), and in turn, these cytokines stimulate the release of PAF (19). In fact, many in vivo biological functions of TNF and IL-1 are mediated through the action of PAF (13,20,21). We therefore investigated whether TNF and IL-1 indirectly induce the second phase of NF-B activation via the action of PAF. First, we determined the effect of PAF antagonist treatment on the LPS-induced second phase of NF-B activity. In vivo administration of BN 50739 or CV 6209 (Fig.  5A) 4 h after LPS injection resulted in an abrogation of the secondary activation of NF-B. Furthermore, Fig. 5B shows that in vitro treatment with BN 50739 or CV 6209 4 h after LPS stimulation in RAW 264.7 cells abrogated the secondary NF-B activation. The kinetics of plasma level of PAF following LPS injection paralleled those of NF-B activation; the major early peak was seen at around 10 -15 min and a small second peak accompanied at around 6 h (Fig. 5C). These data indicate that PAF is involved in the second phase of NF-B activation and support the hypothesis that TNF and IL-1 induce the late synthesis of PAF.
Finally, it was determined whether TNF and IL-1 induce NF-B activation through PAF. Injection of mice with either TNF or IL-1␤ alone resulted in NF-B activation. The ability of TNF and IL-1␤ to activate NF-B either alone or in combination was inhibited by pretreatment with CV 6209 (Fig. 6A). We performed supershift assays to investigate whether the PAF antagonist has qualitatively different effects on the appearance of different NF-B forms that are being activated by TNF or/and IL-1. The supershift assay revealed that PAF antagonist CV 6209 did not mediate a qualitatively different effect on the forms of NF-B but inhibited all NF-B species (data not shown). TNF-or IL-1␤-induced NF-B activation in RAW 264.7 cells and the effect of TNF or IL-1␤ was inhibited by pretreatment with BN 50739 or CV 6209 (Fig. 6B). These findings indicated that TNF and IL-1 play an important role in the second phase of NF-B activation via the action of PAF.

DISCUSSION
The transcription factor NF-B is important for initiating and sustaining inflammatory reactions. The sustained or biphasic activation of NF-B may be responsible for the prolonged inflammatory processes known to occur in various pathological situations (4, 5) and microbial infections (6 -8). However, the molecular mechanisms underlying NF-B activation in inflammatory reactions are poorly understood. In this study, we demonstrated an autoregulatory pathway of biphasic activation of NF-B in response to the inflammatory stimulus LPS. Injection of mice with LPS resulted in biphasic NF-B activation in various organs. We demonstrated that PAF is involved in the early phase activation of NF-B based on the findings that (i) the major early peak of plasma PAF was seen 10 -15 min after LPS injection and (ii) pretreatment with BN 50739 or CV 6209 prior to LPS injection resulted in abrogation of the early peak of NF-B. The presence of an early phase of NF-B activity in TNF Ϫ/Ϫ or IL-1R Ϫ/Ϫ mice suggests that TNF or IL-1 is not involved in the first wave of NF-B activation. It is well known that PAF is an inducer of NF-B activation in vitro (22,23). Regarding the in vivo role of PAF in NF-B activation, our observations also suggest that PAF is a critical mediator in the early activation of NF-B (10,11,15) in a reactive oxygen intermediator-dependent manner (14). This early NF-B activity induced by PAF results in the expression of a variety of early response genes. Thus, we believe that PAF acts as a triggering molecule in initiating a variety of early inflammatory reactions by activating early NF-B.
Importantly, we have also demonstrated that either TNF or IL-1 alone, which is induced by the PAF-dependent early NF-B activity, is indirectly responsible for the second phase of NF-B activation via the synthesis of PAF. Considering the potential role for TNF and IL-1 in NF-B activation (3,(15)(16)(17), we initially expected that TNF or IL-1␤ might play a direct role in the LPS-induced second phase of NF-B activation. However, the TNF and IL-1 inhibition studies using TNF Ϫ/Ϫ and IL-1R Ϫ/Ϫ mice revealed that little impairment of biphasic activation of NF-B in TNF Ϫ/Ϫ or IL-1R Ϫ/Ϫ mice and inhibition of the second phase of NF-B activation was achieved by inhibition of both TNF and IL-1␤ action in vivo as well as in vitro, indicating that these cytokines individually play a significant role in the late phase of NF-B activation. There was no significant impairment in late plasma PAF levels after LPS injection in TNF Ϫ/Ϫ or IL-1R Ϫ/Ϫ mice relative to the control wild type mice (data not shown). These findings indicate that either TNF or IL-1 alone is sufficient to produce PAF.
PAF is not stored in the cells but is present in the form of an inactive precursor (alkylacyl-glycero-3-phosphorylcholine) in the membrane (24). Upon an inflammatory stimulus, phospholipase A 2 (PLA 2 ) becomes activated and cleaves the phospholipid at the 2(R) position, leading to the formation of the lysophospholipid derivative, lyso-PAF (25). Lyso-PAF is in turn acetylated by acetyl-transferase into PAF (26). Recent studies (27,28) have demonstrated that MAP kinase is associated with PAF release through PLA 2 activation by phosphorylation. However, the upstream pathway leading to MAP kinase activation is poorly understood. Toll-like receptors (TLRs), which recognize conserved products of microbial metabolism, play a critical role in innate immunity. TLRs and members of the IL-1 receptor family share homologies in their cytoplasmic domains called Toll/IL-1R/plant R gene homology (TIR) domains (29,30). Intracellular signaling mechanisms mediated by TIRs are similar, with MyD88 (31, 32) and TRAF6 (33,34) having critical roles. The serine-threonine kinase IL-1 receptor-associated kinases (IRAKs) are known to involve the signal transduction between MyD88 and TRAF6 (35,36). Importantly, a marked impairment of MAP kinase activity has recently been reported in MyD88-or IRAK-4-deficient animals (32,37), suggesting that MyD88 and IRAK may be upstream mediators requiring MAP kinase activation for phosphorylation of PLA 2 . Further studies are required to address these issues.
The PAF produced in response to TNF and IL-1 in turn is responsible for the secondary wave of NF-B activation. Such a conclusion is supported by the following observations. (i) The plasma PAF levels paralleled those of NF-B activation following LPS injection. (ii) BN 50739 or CV 6209 prevented the LPS-induced second phase of NF-B activation in vivo as well as in vitro. (iii) TNF-or IL-1-induced activation of NF-B was inhibited by the PAF antagonist BN 50739 or CV 6209. These findings clearly indicate that together TNF and IL-1 play an important indirect role in the secondary activation of NF-B via the synthesis of PAF. PAF and proinflammatory cytokines, such as TNF and IL-1, stimulate the release of each other by a positive feedback loop. In fact, many in vivo biological functions of TNF and IL-1 are mediated through the synthesis of PAF (13,20,21). Thus, we provide another example of a positive feedback loop between PAF and TNF/IL-1 resulting in a second phase of NF-B activation.
The mechanism underlying TNF-or IL-1-induced NF-B activation through PAF has not been fully understood. According to the literature published so far, TNF or IL-1 appear to activate NF-B through two separate pathways, PLA 2 -independent and -dependent pathways. The former involves the recruitment of at least three proteins (TRADD, RIP, and TRAF2) to the type 1 TNF receptor tail, leading to the sequential activation of the downstream NF-B-inducing kinase (NIK) and IB-specific kinases (IKK␣ and IKK␤), resulting in the phosphorylation of the two N-terminal regulatory serines within IB␣ (38,39). However, recent in vivo studies with the alymphoplasia (aly) mouse, a natural strain with a mutated NIK (40), and NIK-deficient mice (41) have revealed that NIK is not a critical element in the TNF signaling pathway to NF-B activation, indicating that the role of NIK in NF-B activation through TNF receptor signaling requires further investigation. Apart from this pathway, TNF and IL-1 are also capable of activating PLA 2 , resulting in biosynthesis of arachidonic acid metabolites, such as prostaglandins, leukotrienes, and PAF (42)(43)(44)(45). Several reports have demonstrated that PLA 2 inhibitors block TNF-induced NF-B activation (46,47). Given this information, our data suggests an inhibition of TNFor IL-1-induced NF-B activation by PAF antagonists may induce the late phase of NF-B activation via the pathway involving PLA 2 -induced PAF synthesis. However, it is not clear whether these two pathways operate separately or in cooperation to activate NF-B.
Although biphasic activation of NF-B is supposed to be associated with sustained inflammatory processes, the precise biological significance of prolonged NF-B activation is poorly understood. Recently, Lawrence et al. (48) reported that the late phase activation of NF-B is associated with the expres-sion of cyclooxygenase 2-derived anti-inflammatory prostaglandins and the anti-inflammatory cytokine, transforming growth factor-␤1. Thus, inhibition of the late phase activation of NF-B protracts inflammatory responses. This suggests that NF-B has an anti-inflammatory role involving the regulation of inflammatory resolution. If that is the case, given the critical role for PAF in the induction of the late phase activation of NF-B in this study, clinical studies showing no beneficial effect of PAF antagonists in patients with sepsis (49,50) supports the anti-inflammatory role of NF-B.
Whatever the biological significance of prolonged NF-B activation, this study is the first to provide the precise molecular mechanisms for biphasic activation of NF-B, which occurs in many inflammatory conditions. Such a biphasic or prolonged activation of NF-B may develop in pathological situations in which the early release of PAF is abnormally enhanced. FIG. 5. Evidence that PAF is involved in the secondary activation of NF-B. A, mice were treated with 400 g of BN 50739 or CV 6209 4 h after LPS (50 g, intravenously) injection, and the lungs were removed at the time indicated. B, RAW 264.7 cells (1 ϫ 10 6 ) were treated with 10 g/ml of BN 50739 or CV 6209 4 h after LPS (0.5 g/ml) treatment and collected at the times indicated. Gel shift assays were performed as described in the legend to Fig. 1. C, mice were injected with LPS (50 g, intravenously), and blood was collected at the time indicated. Plasma PAF was measured as described under "Experimental Procedures." n ϭ 3 animals for each time points. Values are expressed as mean Ϯ S.E. A representative of three to five independent experiments is shown.
FIG. 6. Inhibition of TNF and/or IL-1␤-induced NF-B activation by pretreatment with a PAF antagonist. A, mice were treated with 400 g CV 6209 (intraperitoneally) 30 min prior to TNF (0.1 g, intravenously) and/or IL-1␤ (0.2 g, intravenously) injection, and the lungs were removed at 1 h. n ϭ 3 animals for each groups. B, RAW 264.7 cells (1 ϫ 10 6 ) were treated with 10 g/ml BN 50739 or CV 6209 30 min prior to TNF (10 ng/ml) and IL-1␤ (20 ng/ml) treatment either alone or in combination and collected the cells at 40 min. Gel shift assays were performed on nuclear extracts as described in the legend to Fig. 1. The results of the gel shift experiments (upper panels) were used to quantitate the corresponding relative intensity by densitometry (lower panels). All values are compared against TNF-treated group and expressed as mean Ϯ S.E. A representative of three to five independent experiments is shown. *, p Ͻ 0.005 versus TNF or IL-1␤-treated group; **, p Ͻ 0.005 versus TNF and IL-1␤-treated group.