Temporal Control of NF- (cid:1) B Activation by ERK Differentially Regulates Interleukin-1 (cid:2) -induced Gene Expression*

In cultured rat vascular smooth muscle cells, sustained activation of ERK is required for interleukin-1 (cid:2) to persistently activate NF- (cid:1) B. Without ERK activation, interleukin-1 (cid:2) induces only acute and transient NF- (cid:1) B activation. The present study examined whether the temporal control of NF- (cid:1) B activation by ERK could differentially regulate the expression of NF- (cid:1) B-dependent genes, including inducible nitric oxide synthase (iNOS), cyclooxygenase-2 (COX-2), vascular cell adhesion mole-cule-1 (VCAM-1), and manganese-containing superoxide dismutase (Mn-SOD). Treatment of vascular smooth muscle cells with interleukin-1 (cid:2) induced the expression of iNOS, COX-2, VCAM-1, and Mn-SOD in a time-depend-ent manner, but with different patterns. Either PD98059 or U0126, selective inhibitors of MEK, or overexpression of a dominant negative MEK-1 inhibited interleukin-1 (cid:2) induced ERK activation and the expression of iNOS and COX-2 but had essentially no effect on the expression of VCAM-1 and Mn-SOD. The expression of these genes was inhibited when NF- (cid:1) B activation was down-regulated by MG132,

leads to the phosphorylation and ubiquitination-dependent degradation of I-B, accompanied by NF-B translocation to the nucleus, where it binds to consensus sequences in the promoter region of target genes and activates transcription (1)(2)(3)(4)(5)(6)(7). I-B␣ is rapidly degraded after cytokine stimulation and rapidly resynthesized because of the presence of NF-B-binding motifs in the I-B␣ gene promoter (8). This autoregulatory loop has been suggested to limit NF-B activation in a transient manner (8,9). In contrast, I-B␤ slowly decreases after cytokine stimulation, a process that has been suggested to contribute to persistent activation of NF-B (9). Recently, the temporal control of NF-B activation by the coordinated degradation and synthesis of I-B proteins has been clearly revealed by using gene knock-out cell lines where NF-B activation was controlled by a single I-B isoform (10). TNF␣ stimulation of fibroblasts that contained only the I-B␣ isoform resulted in an oscillatory pattern of NF-B activation, whereas stimulation of cells harboring only I-B␤ resulted in sustained NF-B activation. Importantly, it was shown that the expression of NF-B-dependent genes may be differentially regulated by controlling the duration of NF-B activation (10). However, the mechanisms of differential control by which I-B␣ and I-B␤ mediate NF-B activation following cytokine treatment remain unclear.
We have recently demonstrated in cultured rat vascular smooth muscle cells (VSMCs) that activation of extracellular signal-regulated kinases (ERK) was required for IL-1␤ to induce persistent activation of NF-B that in turn was required for the subsequent expression of inducible nitric oxide synthase (iNOS) (11,12). Inhibition of ERK activation, either by the selective chemical inhibitors, PD98059 or U0126, or by antisense phosphorothioate-modified oligodeoxynucleotides that down-regulated ERK synthesis, selectively inhibited IL-1␤-induced prolonged activation of NF-B and iNOS expression. Interestingly, inhibition of ERK activation had no effect on early transient NF-B activation induced by IL-1␤. This suggested that an important ERK-dependent signaling mechanism may temporally control the activation of NF-B and NF-B-dependent gene expression. The present study was performed to test this hypothesis, using iNOS, cyclooxygenase-2 (COX-2), vascular cell adhesion molecule-1 (VCAM-1), and manganese-containing superoxide dismutase (Mn-SOD) as gene products representative of those controlled by either persistent or acute activation of NF-B. The results indicate that inhibition of ERK activation differentially regulates the expres-FIG. 1. ERK activation is required for iNOS and COX-2 induction, but not for VCAM-1 and Mn-SOD induction. In A and B, rat VSMCs were treated with IL-1␤ (3 ng/ml) for the indicated periods in the absence or in the presence of either PD98059 (20 M) or U0126 (20 M) that was added 1 h before IL-1␤. Whole cell lysates (20 g of proteins/lane) were used for Western blot analysis of indicated proteins. p-ERK, phosphorylated ERK. In C, VSMCs were treated with IL-1␤ (3 ng/ml) for the indicated periods in the absence or in the presence of PD98059 (20 M). Total RNA was extracted and used for RT-PCR to determine mRNA levels of iNOS, COX-2, VCAM-1, Mn-SOD, and GAPDH. Relative intensity ratios to GAPDH mRNA are shown in the bottom graph. The results shown in A-C are representative of three separate experiments each. sion of these NF-B-dependent genes in response to IL-1␤ stimulation by specifically down-regulating the persistent activation of NF-B.
Cell Culture-Rat VSMCs were isolated from the thoracic aorta and cultured as described previously (13). Cells were used between passages 5 and 9. When confluent, the cells were washed with serum-free medium and then maintained in DMEM/F12 with 0.1% BSA for 24 -48 h. The medium was refreshed before treatment. The cells were then incubated with or without additions (cytokines, inhibitors, or vehicle) for designated times as indicated under "Results." Adenoviral Constructs-To generate adenoviral I-B␣ mutant (AdvIB␣M), an I-B␣ mutant (S32A/S36A) cDNA (HindIII/DraI) from pCMV-IB␣M (Clontech) was inserted into pShuttle-CMV vector between HindIII and EcoRV sites and then transferred into pAdEasy-1 by homologous recombination in BJ5183 bacterial cells. The recombinant adenoviral construct was linearized with PacI and transfected into Hek293 cells to produce viral particles. To generate adenoviral dominant negative MEK-1 (AdvMEK1dn), pCMV-MEK1 (Clontech) was mutated by PCR with QuikChange site-directed mutagenesis kit (Strat-agene) with the primer pairs 5Ј-GCT CAT CGA CGC CAT GGC CAA CGC CTT CGT GGG-3Ј and 5Ј-CCC ACG AAG GCG TTG GCC ATG GCG TCG ATG AGC-3Ј (the underlining indicates the mutation from T to G or from A to C), which made the mutation of MEK-1 Ser-218 and Ser-222 to Ala-218 and Ala-222. MEK1dn cDNA (XhoI/HindIII) from the pCMV-MEK1dn was then inserted into pShuttle-CMV vector. The adenovirus was generated by the same method as mentioned above. Adenoviruses were purified through two steps of ultracentrifugation, with discontinuous and continuous CsCl gradients, respectively. After dialysis against 10 mM Tris-HCl, pH 8.0, containing 2 mM MgCl 2 and 4% sucrose, viral particles in solution were evaluated based on DNA content. DNA was determined by absorbance at 260 nm assuming an extinction coefficient of 1.1 ϫ 10 12 viral particles/OD unit.
Western Blot Analysis-Whole cell lysates were prepared, and Western blot analyses were performed as described previously (14). Protein content of the cell lysates was determined with BCA protein assay reagent (Pierce), with BSA used as a standard. The images were obtained and analyzed by using Model GS-700 imaging densitometer (Bio-Rad).
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared, and DNA binding activity was assessed by EMSA using NF-B consensus oligonucleotide as described previously (14).
Immunofluorescence Staining-VSMCs were cultured on 4-well Lab-Tek II chamber slides (Nalge Nunc International) under the same conditions described above. After treatment, the cells were washed with cold PBS, fixed for 8 min in methanol at Ϫ20°C, and air-dried at room temperature. The staining was performed by incubating with 10% normal goat serum in PBS for 20 min followed by incubating with primary antibody (5 g/ml NF-B p65 polyclonal antibody (C-20), sc-372, Santa Cruz Biotechnology) for 2 h in PBS with 1.0% BSA, washing three times with PBS, incubating for 1 h with fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (1/100 dilution, Jackson ImmunoResearch Laboratories) in PBS with 1.0% BSA, washing three times with PBS, and finally mounting with aqueous mounting medium. The images observed under a fluorescence microscope were recorded on a linked computer using Openlab software (version 2.2.5, Improvision).

ERK Activation Is Required for iNOS and COX-2 Induction, but Not for VCAM-1 and Mn-SOD Induction-In cultured rat
VSMCs, there was basal expression of VCAM-1 and Mn-SOD that could be detected by Western blot analysis, whereas iNOS and COX-2 were not detectable without cytokine treatment (Fig. 1, A and B). IL-1␤ increased the expression of these genes, but temporal differences were observed. After IL-1␤ stimulation, VCAM-1 was clearly increased by 3 h, and high levels were maintained between 6 and 24 h (4.9-, 12.4-, and 13.9-fold greater than basal levels, respectively). In contrast, iNOS protein was undetectable at 3 and 6 h but readily detectable 16 h after IL-1␤ addition. The induction of COX-2 and Mn-SOD by IL-1␤ occurred in a temporal pattern similar to that of iNOS, expression being delayed as compared with that of VCAM-1. Treatment of the cells with IL-1␤ also caused a sustained increase of ERK phosphorylation. Notably, inhibition of ERK phosphorylation by either PD98059 or U0126, specific inhibitors of ERK kinases, dramatically reduced the expression of both iNOS and COX-2 but did not prevent the induction of VCAM-1 as well as Mn-SOD by IL-1␤ (Fig. 1, A and B). RT-PCR ( Fig. 1C) showed that the inhibition of ERK phosphorylation by PD98059 influenced IL-1␤ induction of iNOS and COX-2 at the transcriptional level. At 6 and 9 h after IL-1␤ addition, iNOS mRNA was detectable in the absence of PD98059 but was not detectable in the presence of PD98059. COX-2 mRNA showed a biphasic increase during IL-1␤ treatment, with an increase at 1 h followed by a reduction at 3 h and then another increase that reached the maximum between 6 and 9 h. The reduction of COX-2 mRNA at 3 h was not due to artifactual RNA degradation during RNA preparation or the reverse transcription reaction because the result was reproducible, and there was no decrease in GAPDH mRNA levels. In the presence of PD98059, IL-1␤ still induced an increase in COX-2 mRNA at 1 h but failed to induce the second phase increase at 6 and 9 h. Both VCAM-1 mRNA and Mn-SOD mRNA gradually increased during the IL-1␤ treatment and remained at high levels for up to 6 and 9 h. PD98059 had no effect on the enhancement of VCAM-1 and Mn-SOD mRNA levels induced by IL-1␤.
To further confirm that the inhibition of iNOS and COX-2 expression by the chemical inhibitors was due to the specific inhibition of ERK signaling, the effect of adenoviral mediated expression of MEK1dn was determined. Consistent with the results obtained using chemical inhibitors, Western blot analysis showed that overexpression of MEK1dn reduced IL-1␤- induced ERK phosphorylation and selectively reduced the induction of iNOS and COX-2 without significantly influencing the induction of VCAM-1 or Mn-SOD by IL-1␤ (Fig. 2). The control adenovirus expressing LacZ showed no effect on these processes.

The Duration of NF-B Activation Is Responsible for the Differential Regulation of IL-1␤-induced Gene Expression-As
shown using EMSA (Fig. 3), at 30 min after IL-1␤ addition, the major DNA-bound form of NF-B was the p65/p50 heterodimer, whereas the p50/p50 homodimer showed a relatively weak signal. However, at 6 h, when p65/p50 DNA binding was still obvious, the DNA binding of the p50/p50 homodimer was increased. Inhibition of ERK by PD98059 reduced the prolonged activation of NF-B as evidenced by reduced binding of both p65/p50 and p50/p50 at 6 h, without having a significant effect on the early activation as evidenced by p65/p50 binding at 30 min ( Fig. 3) (11, 12). Because inhibition of ERK activation selectively inhibits prolonged NF-B activation, we tested the possibility that the expression of VCAM-1 and Mn-SOD induced by IL-1␤ might be dependent solely on the earlier and transient NF-B activation. We therefore treated the cells with MG132, a proteasome inhibitor that inhibits I-B degradation. In contrast to PD98059, pretreatment of the cells with MG132 reduced both early and prolonged NF-B activation induced by IL-1␤, decreasing p65/p50 DNA binding to 56 Ϯ 5% at 30 min and to 9 Ϯ 1% at 6 h, as compared with those without MG132 treatment. In addition, MG132 reduced p50/p50 DNA binding at 6 h.
Western blot analysis (Fig. 4A) showed that phosphorylated I-B␣ was detected after IL-1␤ addition, and the abundance increased in the cells pretreated with MG132. Phosphorylated I-B␣ was not detectable in untreated cells but was observed in the cells treated with MG132 alone (lane 3), suggesting an accumulation of basal levels due to reduced degradation. The expression of iNOS, COX-2, VCAM-1, and Mn-SOD induced by IL-1␤ was obviously inhibited in cells pretreated with MG132, indicating the involvement of NF-B in the expression of all four proteins. MG132 had no effect on the sustained phosphorylation of ERK that was observed after IL-1␤ addition. The ability of MG132 to prevent iNOS and COX-2 induction was apparent if the inhibitor was added at any time from 1 h preceding IL-1␤ addition up to 6 h after the cytokine was added (Fig. 4B). However, MG132 did show a lesser effect on inhibition of VCAM-1 and Mn-SOD induction if the inhibitor was added at 6 h after IL-1␤ addition.
To confirm that the inhibitory effect of MG132 on IL-1␤induced gene expression was due to its influence on components of NF-B, the cells were infected with AdvIB␣M (S32A/ S36A). Similar to MG132, the adenoviral-mediated overexpression of the degradation-resistant IB␣ mutant (S32A/ S36A) prevented the IL-1␤-induced increase in iNOS, COX-2, VCAM-1, and Mn-SOD seen after treatment for 24 h (Fig. 5A). Immunofluorescent staining of the NF-B p65 subunit was performed to determine whether the I-B␣ mutant prevented the nuclear translocation of the predominant heterodimeric form of NF-B. The images (Fig. 5B) show clearly that IL-1␤ induced nuclear translocation of p65 by 1 h and that this persisted up to 16 h. Overexpression of the I-B␣ mutant inhibited IL-1␤-induced nuclear translocation of NF-B p65 at both time points, consistent with the inhibitory effect of the I-B␣ mutant on IL-1␤-induced gene expression shown in Fig. 5A.
Inhibition of ERK Activation Attenuates IL-1␤-induced I-B␤ Degradation-We further determined whether the ability of the ERK signaling cascade to regulate the persistent activation of NF-B might relate to temporal changes in I-B levels. As shown in Fig. 6, A and B, IL-1␤ induced I-B␣ phosphorylation during the entire period observed. The total I-B␣ levels were dramatically reduced at 30 min and then gradually approached basal levels thereafter. Inhibition of ERK activation by either PD98059 (Fig. 6A) or overexpression of MEK1dn (Fig. 6B) did not prevent IL-1␤-induced I-B␣ phosphorylation or degradation at 30 min. IL-1␤ stimulation also reduced I-B␤ levels (Fig. 6, C and D). The reduction of I-B␤ was not as pronounced as that of I-B␣ at 30 min, but the reduction of I-B␤ at 6 h was obvious. Inhibition of ERK activation by PD98059 attenuated the reduction in I-B␤ induced by IL-1␤, although the effect of PD98059 on I-B␤ degradation was less marked than that of MG132, suggesting that ERK might be involved in the regulation of I-B␤ degradation.
Effects of ERK Signaling on NF-B-dependent Gene Expression Induced by a Combination of Cytokines-Although treatment of rat VSMCs with IFN␥ or TNF␣ alone does not induce iNOS, either IFN␥ or TNF␣ does enhance iNOS expression induced by IL-1␤ (14,15). We therefore examined whether or not ERK signaling could similarly regulate IL-1␤-induced NF-B-dependent gene expression in the presence of IFN␥ or TNF␣. As shown in Fig. 7A, both IFN␥ and TNF␣ enhanced IL-1␤-induced iNOS, COX-2, and VCAM-1, but not Mn-SOD expression, although the synergistic effect was more pronounced when TNF␣ was added with IL-1␤. Inhibition of ERK activation by PD98059 dramatically reduced iNOS and COX-2 expression but not that of VCAM-1 and Mn-SOD induced by IL-1␤ alone or in combination with IFN␥ or TNF␣. In addition, although TNF␣ alone did not induce iNOS and COX-2 expression, it did increase VCAM-1 and Mn-SOD levels, and these were by a mechanism not inhibited by PD98059. To examine whether TNF␣ could activate ERK and NF-B in a manner similar to IL-1␤, we determined the changes in both I-B levels and the phosphorylation of ERK at different time points after TNF␣ addition (Fig. 7B). Upon TNF␣ stimulation, I-B␣ levels rapidly decreased by 30 min and then returned to basal levels by 3 h, whereas I-B␤ decreased gradually and remained at a lower level at later time points (3 and 9 h). TNF␣ activated ERK with a strong phosphorylation observed at 15 min and then returned to basal levels at 1 h. There was a second phase of ERK phosphorylation persisting from 3 to 9 h after TNF␣ addition. These effects of TNF␣ on I-B degradation and ERK phosphorylation were similar to those observed with IL-1␤ (11).
We further tested whether ERK activation was still required for IL-1␤ to induce persistent activation of NF-B in the presence of TNF␣. As shown by EMSA (Fig. 7C), although PD98059 had essentially no effect on NF-B activation at 1 h after cytokine addition, it dramatically reduced the prolonged activation of NF-B induced by either IL-1␤ alone or IL-1␤ plus TNF␣ as documented by the data shown at the 16-h treatment time. The prolonged NF-B activation induced by TNF␣ alone, like that caused by IL-1␤ alone, was also reduced by inhibition of ERK with PD98059.

DISCUSSION
In many cell types, NF-B activation is essential for the expression of numerous genes such as iNOS, COX-2, VCAM-1, and Mn-SOD in response to inflammatory stimuli, including IL-1␤ (16 -20). The present studies clearly show that IL-1␤ induces all of the genes mentioned above in cultured rat VSMCs but that the expression of these genes occurs in a temporally distinct manner. This temporal control is dependent upon the temporal activation of both ERK and NF-B, and ERK phosphorylation is a requirement for the persistent activation of NF-B.
Neither iNOS nor COX-2 mRNA or protein was detectable in the cells without IL-1␤ stimulation. After IL-1␤ stimulation, iNOS and COX-2 expression was delayed in comparison with changes in VCAM-1 or Mn-SOD. Unlike VCAM-1 or Mn-SOD, iNOS and COX-2 accumulation was entirely dependent on the prolonged activation of both ERK and NF-B. The decrease in p65/p50 caused by ERK inhibition likely was I-B␣-independent as shown by persistent I-B␣ phosphorylation and early I-B␣ degradation that was unchanged by PD98059. Rather, as suggested previously, prolonged NF-B activation may be mediated by I-B␤ (9,21). This possibility is also supported by the fact that in this study, inhibition of ERK attenuated IL-1␤induced I-B␤ degradation. It is known that I-B kinases (IKKs) can phosphorylate I-B␤ at Ser-19 and Ser-23, two phosphorylation sites similar to those in I-B␣ and responsible for its subsequent degradation (22,23). Because activation of ERK alone does not induce NF-B activation, as we reported previously (12), and inhibition of ERK does not affect I-B␣ phosphorylation induced by IL-1␤, the influence of the ERK signaling cascade on IL-1␤-induced I-B␤ degradation may be mediated by mechanisms other than influencing IKK activity. Further studies are required to elucidate whether or not the ERK signaling cascade influences the susceptibility of I-B␤ to phosphorylation by IKKs or to degradation by the proteasome. Although the requirement of ERK activation for iNOS and COX-2 induction might relate to multiple mechanisms, one role for ERK signaling apparently is to maintain the persistent activation of NF-B.
TNF␣ alone was unable to induce iNOS expression (14,15), although TNF␣ induced ERK activation and NF-B activation in a pattern similar to that induced by IL-1␤, suggesting the involvement of other signaling pathway(s) that may either induce or suppress iNOS expression. IFN␥ alone did not activate NF-B and was unable to induce iNOS expression in VSMCs (14,15). However, it is known that both TNF␣ and IFN␥ potentiate IL-1␤-induced gene expression, probably due to the involvement of other signaling pathways. For example, signal transducer and activator of transcription (STAT)-1 is activated by IFN␥ (14,15) and this transcription factor has been shown to promote iNOS expression (24). Although IL-1␤induced iNOS and COX-2 expression was enhanced by either TNF␣ or IFN␥, the expression of these two genes was dramatically suppressed by inhibition of ERK activation, whereas the induction of VCAM-1 and Mn-SOD by the combination of cytokines was not affected. These data further demonstrated that the ERK signaling pathway has a major role in regulating the expression of certain NF-B-dependent genes even when multiple cytokines coexist.
VCAM-1 and Mn-SOD were constitutively expressed at relatively low levels but were detectable by either Western blot for the proteins or RT-PCR for mRNA in rat VSMCs under basal conditions. A low basal turnover of I-B proteins is supported by evidence obtained from treating cells for 25 h with MG132, a proteasome inhibitor, which inhibited I-B␣ degradation in the absence of added cytokine and led to an accumulation of phosphorylated I-B␣. The I-B␣-mediated NF-B activation was acutely and dramatically enhanced by IL-1␤ stimulation, which was accompanied by increased expression of VCAM-1. Consistent with previous reports by others, one of the NF-Bdependent genes is I-B␣ itself (8), which is activated in a similar time frame to VCAM-1 (protein levels increased within 3 h after IL-1␤ addition). Interestingly, the newly synthesized I-B␣ could not prevent the activation of NF-B by IL-1␤, possibly due to the fact that newly synthesized I-B␣ underwent constitutive and rapid phosphorylation and degradation during IL-1␤ stimulation. Because inhibition of I-B␣ degradation by MG132 reduced NF-B activation without influencing I-B␣ phosphorylation, the activation of VCAM-1 gene transcription may be dependent on the turnover rate of I-B␣. Neither PD98059 nor MEK1dn affected IL-1␤-induced I-B␣ phosphorylation at Ser-32/Ser-36. Considering that inhibition of ERK showed no effect on I-B␣ phosphorylation and degradation, and also had no obvious effect on VCAM-1 expression, it is apparent that I-B␣-mediated NF-B activation in response to IL-1␤ is ERK-independent and responsible for triggering VCAM-1 gene expression. The regulation of Mn-SOD gene expression appears to be similar to that of VCAM-1, with respect to not requiring the persistent activation of both ERK and NF-B.
The differential regulation of NF-B-dependent gene expression could be important in determining the nature of the inflammatory response. The duration of NF-B activation might determine the course of the inflammatory response with regard to the local reactions and resulting morphologic changes, the destruction or removal of the injurious material, and the responses that lead to repair and healing (1)(2)(3)25). A potential role for NF-B in the resolution of inflammation has been reported recently (26). Because activation of NF-B may regulate the expression of numerous genes encoding either proinflammatory or anti-inflammatory mediators, including cytokines, adhesion molecules, chemokines, growth factors, and inducible enzymes such as iNOS, COX-2, and Mn-SOD, each of which may play critical roles during different stages of the inflammation, our findings demonstrating a role for ERK in the temporal control of NF-B activation and NF-B-dependent gene expression provides new insight into the mechanisms regulating NF-B activation.