Immunosuppressant FK506 Activates NF-κB through the Proteasome-mediated Degradation of IκBα REQUIREMENT FOR IκBα N-TERMINAL PHOSPHORYLATION BUT NOT UBIQUITINATION SITES

The immunosuppressant FK506 activates NF-κB through IκBα degradation in nonlymphoid cells. In the present study, we analyzed mechanisms by which FK506 induces IκBα degradation. We found that FK506 induces the degradation of both IκBα and IκBβ and that the time courses of the FK506-induced degradation are quite different from degradation induced by interleukin 1 (IL-1). Despite this difference, FK506-induced IκBα degradation was dependent on the N-terminal Ser-32 and Ser-36 phosphorylation sites and was mediated by proteasomes, as is the case for IL-1-induced IκBα degradation. We further showed that FK506 induces weak and slow phosphorylation of IκBα at Ser-32. However, unlike IL-1-induced degradation, IKK-1 and IKK-2 were not activated significantly nor was FK506-induced IκBα degradation dependent on the N-terminal ubiquitination sites (Lys-21 and Lys-22). These results therefore indicate that FK506 and IL-1 utilize similar but distinct mechanisms to induce the phosphorylation and degradation of IκBα.

The immunosuppressant FK506 activates NF-B through IB␣ degradation in nonlymphoid cells. In the present study, we analyzed mechanisms by which FK506 induces IB␣ degradation. We found that FK506 induces the degradation of both IB␣ and IB␤ and that the time courses of the FK506-induced degradation are quite different from degradation induced by interleukin 1 (IL-1). Despite this difference, FK506-induced IB␣ degradation was dependent on the N-terminal Ser-32 and Ser-36 phosphorylation sites and was mediated by proteasomes, as is the case for IL-1-induced IB␣ degradation. We further showed that FK506 induces weak and slow phosphorylation of IB␣ at Ser-32. However, unlike IL-1-induced degradation, IKK-1 and IKK-2 were not activated significantly nor was FK506-induced IB␣ degradation dependent on the N-terminal ubiquitination sites (Lys-21 and Lys-22). These results therefore indicate that FK506 and IL-1 utilize similar but distinct mechanisms to induce the phosphorylation and degradation of IB␣.
Nuclear factor B (NF-B) 1 is a transcription factor that plays an important role in inducing the expression of diverse cellular genes, such as for various cytokines, cell surface receptors, and acute-phase proteins. It is a heterodimer mainly composed of the p50 and RelA proteins, but there might be a considerable heterogeneity in its composition in various cell types, because of the presence of p50/RelA-related proteins (p52, c-Rel, and RelB), which share extensive homology in their N-terminal DNA-binding/dimerization regions. These proteins are now known as the NF-B/Rel/Dorsal transcription factor family, as they are also related to the Drosophila maternal morphogen gene, dorsal. An unusual feature of this family is that they exist in the cytoplasm in an inactive form complexed with a family of inhibitor proteins termed IB (IB␣, IB␤, and IB⑀). A variety of stimuli, including virus infection, bacterial lipopolysaccharides, double-stranded RNA, phorbol esters, UV radiation, oxidative stress, and inflammatory cytokines such as interleukin-1 (IL-1) and tumor necrosis factor-␣ (TNF-␣) activate NF-B through the proteolytic degradation of IB and the subsequent translocation of NF-B to the nucleus, where it activates target genes (1)(2)(3).
The prototypic and best-studied of the IBs is IB␣, which is phosphorylated at its N-terminal two serine residues (Ser-32 and Ser-36) prior to degradation, when cells are exposed to appropriate NF-B activators (4 -6). This phosphorylation triggers the ligation of multiple ubiquitin molecules to nearby lysine residues (Lys-21 and Lys-22), leading to the subsequent degradation of the protein by proteasomes (6 -9). The signalinduced phosphorylation of IB␣ is therefore a critical step in NF-B activation and has been investigated intensively. Recently, two closely related IB kinases (IKKs), termed IKK-1 and IKK-2, have been identified and cloned. Both kinases directly phosphorylate Ser-32 and Ser-36 of IB␣ and their activities are stimulated by IL-1 and TNF-␣ treatment (10 -14). In addition, pp90rsk kinase (15) and a kinase related to IKK-1 and IKK-2 (termed IKK-3) (16) have also been shown to phosphorylate Ser-32 and Ser-36. Thus, it remains to be established how these various IB kinases are specifically activated in response to diverse NF-B activators.
FK506 is a powerful immunosuppressive drug that is currently in clinical use. It exerts its major immunosuppressive effect by inhibiting transcriptional events, including the activation of several cytokine genes, particularly the interleukin-2 gene, that lead to T-cell activation (17). We previously showed that FK506 induces IB␣ degradation and NF-B activation in nonlymphoid cells such as renal mesangial cells and fibroblasts. We further showed that, as a result of NF-B activation by FK506, interleukin-6 production is induced in the kidney, suggesting the possibility of a causal relationship between the FK506-induced NF-B activation/IL-6 production and some FK506-induced renal abnormalities (18). However, little is known about how FK506 induces IB␣ degradation in nonlymphoid cells. In the present study, we analyzed the mechanisms by which FK506 induce IB␣ degradation. We found that, as in the case of IL-1-induced IB␣ degradation, FK506-induced IB␣ degradation is dependent on the N-terminal serine phosphorylation sites and is mediated by proteasomes. However, the N-terminal ubiquitination sites were not essential for FK506-induced IB␣ degradation, and FK506 induced weak and slow phosphorylation of IB␣ at Ser-32, in the absence of significant IKK activation. Thus, these results suggest that FK506 and IL-1 induce the phosphorylation and degradation of IB␣ through similar but distinct mechanisms.

EXPERIMENTAL PROCEDURES
Chemicals-PSI (Z-Ile-Glu(OBuϩ)-Ala-Leu-H aldehyde), MG132, and MG115 were from the Peptide Institute, Inc., Japan. ICE inhibitor II (Ac-Tyr-Val-Ala-Asp-chloromethyl ketone) and clasto-lactacystin ␤-lactone (C 10 H 15 NO 4 ) were from Sigma and Boston Biochemical, respectively. E64d was kindly provided by Dr. K. Tanaka. PSI (19), MG132 (9), MG112 (9), and ␤-lactone (21) were protease inhibitors specific for proteasomes. E64d (22) and ICE inhibitor II (23) were specific inhibitors for calpain and ICE, respectively. Stock solutions were prepared in dimethyl sulfoxide (Me 2 SO) (Sigma) at 10 mg/ml (MG132, MG115, E64-d) or 100 mM (PSI, lactone). ICE inhibitor II was prepared in methanol at 50 mg/ml. All of these inhibitors were stored at Ϫ20°C. In every experiment presented, the amount of Me 2 SO was corrected in each sample such that the effect of Me 2 SO was controlled. FK506 (from Fujizawa Pharmaceutical Co., Japan) was prepared in ethanol at 1 mM and diluted in growth medium when used.
Plasmid Constructions-The cDNA encoding full-length wild-type human IB␣ (24) was used as a template to generate a cDNA encoding the N-terminal deletion mutant of IB␣ ( Fig. 2A) by PCR amplification. Various mutations of IB␣ as shown in Fig. 2A were introduced by overlap PCR mutagenesis. PCR products were purified, digested with EcoRI and BamHI, and were subcloned in frame into Bluescript KS downstream of the HA epitope sequence. cDNAs encoding various mutant forms of IB␣ with the HA tag sequence were excised by XbaI and inserted into the XbaI site of a mammalian expression vector (pEF-BOS) (25). The construction of mammalian expression vectors encoding IKK-2 (pFlag-IKK-2), JNK3 (pFlag-JNK3), and the truncated and constitutively active form of MEKK (pHA-⌬MEKK) will be described in detail elsewhere. Briefly, the entire IKK-2 coding sequence, the entire JNK3 coding sequence, and the MEKK1 coding sequence (residues 1169 -1488) with the HA tag sequence were amplified by PCR and subcloned into either the pFlag-CMV2 vector (Kodak) or the pEF-BOS vector. For the B-luciferase reporter gene construction, a synthetic NF-B binding motif was inserted into the pGBL3 basic vector (Promega). To construct a plasmid encoding the glutathione S-transferase (GST)-wild-type IB␣ (1-54 amino acid residues) fusion protein, a PCR fragment encoding the N-terminal part of IB␣ (1-54) was inserted into the BamHI-EcoRI fragment of the pGEX-4T3 vector, in frame.
Cell Cultures and Transfection-Murine fibroblast L-TK cells (a thymidine kinase-deficient cell line derived from L929 cells) were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal calf serum (Life Technologies, Inc.), 50 units/ml of penicillin G, and 50 g/ml streptomycin sulfate (Life Technologies, Inc.) in a 5% CO 2 humidified incubator. 293 cells were grown in minimum essential medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal calf serum. L-TK and 293 cells were transfected with various plasmids using the DEAEdextran and calcium phosphate methods, respectively. Twenty-four or 48 h after transfection, cells were left untreated or were treated with IL-1 or FK506 for various periods of time prior to harvest. In some experiments, cells were pretreated with protease inhibitors before the addition of IL-1 or FK506. Human recombinant IL-1 (Otsuka Pharmaceutical Company) was prepared in Dulbecco's modified Eagle's medium at 100 g/ml and stored at Ϫ80°C.
Luciferase Assay-Twenty-four hours after transfection with the Bluciferase reporter gene and IB␣ expression vectors, L-TK cells were stimulated with IL-1 or FK506 for 24 h before harvesting them for the luciferase assay, which was carried out according to the manufacturer's instruction (Promega).

RESULTS
FK506 Induces the Degradation of IB␣ and IB␤-To analyze the effects of FK506 stimulation on IB degradation, HAtagged IB␣ expression vectors were transfected into L-TK cells, which were most efficient in FK506-mediated NF-B activation (18). Cells were then treated with IL-1 or FK506 for different periods of time. As shown in Fig. 1A and in agreement with the results of previous studies (27), treatment with IL-1 for only 2 min resulted in the appearance of a slow-migrating band, corresponding to the phosphorylated form of IB␣ (27), with almost complete disappearance of the IB␣ band at 10 min as a consequence of its proteolytic degradation, and then reappearance at 30 min because of resynthesis of IB␣ (Fig. 1A, upper panel). By contrast, FK506 treatment resulted in only a small induction of the high molecular band at 15 min, and much slower IB␣ degradation, which was detectable only after 30 min. Resynthesized IB␣ bands appeared only after 240 min ( Fig. 1A, middle panel). Essentially similar patterns of degradation were observed for endogenous IB␣ (data not shown). IB␤ degradation induced by FK506 was also slower than that induced by IL-1 (Fig. 1B).

FK506-induced IB␣ Degradation Requires N-terminal Phosphorylation Sites and Is Mediated by Proteasomes-The
prevailing model for IB␣ degradation is that IB␣ becomes phosphorylated at Ser-32 and Ser-36 prior to ubiquitination and subsequent degradation in proteasomes (4 -9). To determine whether FK506 also induces IB␣ degradation through the same or similar mechanisms, we first constructed the expression vector encoding a truncated form of IB␣ lacking the N-terminal region (amino acids 1-36) termed IB␣⌬N ( Fig.  2A). After transfection with this expression vector, L-TK cells were treated with IL-1 or FK506. As shown in Fig. 2B, whereas wild-type IB␣ was degraded both by IL-1 and FK506 stimulation (panel a), the degradation of IB␣⌬N by IL-1 and FK506 was completely blocked (panel d), indicating that the N-terminal 36 amino acids are essential for degradation. This is in agreement with previous studies showing that the N-terminal region is essential for IB␣ degradation (4 -6). To further determine the amino acid residues required for FK506-induced IB␣ degradation, various site-specific mutations were introduced into the N-terminal region of IB␣; Ser-32 and Ser-36 were replaced with alanine (S32A/S36A), and Tyr-42 was replaced with phenylalanine (Y42F) (Fig. 2A). The Tyr-42 phosphorylation was previously shown to be required for NF-B activation induced by some atypical activators (28). Expression vectors encoding these mutant forms of IB␣ were then transfected into L-TK cells, and cells were stimulated with IL-1 or FK506. As shown in Fig. 2B, although the Y42F mutant was degraded both by IL-1 and FK506 with time courses similar to the wild-type IB␣ (panel b), the S32A/S36A mutant was degraded by neither IL-1 nor FK506 (panel c). These results therefore indicate that either Ser-32 or Ser-36 are also essential for FK506-induced IB␣ degradation.
To clarify whether FK506 induces IB␣ degradation through proteasome-dependent mechanisms, L-TK cells transfected with the wild-type IB␣ expression vector were pretreated with various protease inhibitors, including specific proteasome inhibitors before IL-1 or FK506 stimulation. As shown in Fig. 3, the IB␣ degradation induced by both IL-1 (upper panel) and FK506 (middle panel) was specifically blocked by proteasome inhibitors such as MG132, MG115, and lactone, indicating that the FK506-induced IB␣ degradation is mediated by proteasomes.
FK506 Induces Ser-32 Phosphorylation of IB␣ in the Absence of IKK Activation-The above results (Fig. 2B) indicated that Ser-32 and Ser-36 are essential for FK506-induced IB␣ degradation. To further examine whether FK506 induces the phosphorylation of IB␣ at Ser-32, L-TK cells transfected with the wild-type IB␣ expression vector were treated with IL-1 or FK506 in the presence or absence of proteasome inhibitor (PSI), and IB␣ phosphorylated at Ser-32 was detected with an anti-phosphor IB␣ (Ser 32) antibody. As shown in Fig. 4, IL-1 induced rapid Ser-32 phosphorylation at 2 min, as expected, but Ser-32 phosphorylation could not be detected when the cells were treated with FK506 in the absence of PSI. However, when the cells were pretreated with PSI and then stimulated with FK506, weak Ser-32 phosphorylation was detected at 30 min (Fig. 4, lower panel, lanes 8 -9).
Because FK506 induces the phosphorylation of IB␣ at Ser- E64-D, 10 g/ml; ICE inhibitor II, 10 g/ml. Cell lysates (100 g) were subjected to immunoblot analysis using an anti-HA antibody. Hyperphosphorylated IB␣ (IB␣-p) appears as a distinct, more slowly migrating protein band. Lane 3 shows that IB␣ was almost completely degraded with IL-1 or FK506 treatment alone, whereas the control panel shows that pretreatment with inhibitors alone did not affect the IB␣ level or its phosphorylation status. *NS, nonspecific bands. 32, we next examined the effects of FK506 on the activity of IKK, the recently cloned protein kinase that preferentially phosphorylates Ser-32 and Ser-36 of IB␣ (10 -14). 293 and L-TK cells transfected with an IKK-2 expression vector (pFlag-IKK-2) were stimulated with IL-1 or FK506, whereas cells cotransfected with an expression vector encoding truncated and constitutively active forms of MEKK (pHA-⌬MEKK) served as positive controls, as it is known that ⌬MEKK activates both IKK-1 and IKK-2 (29,30). Flag-IKK-2 proteins were immunoprecipitated with anti-Flag antibodies and then subjected to an in vitro kinase assay using GST-IB␣ (1-317) as a substrate. As shown in Fig. 5, A and B, although IL-1 stimulated IKK activity about 4-fold, no significant IKK activation by FK506 was detected either in the L-TK or 293 cells. However, FK506 was fully active in JNK activation in L-TK cells (Fig.  5C). Fig. 6, substituting the N-terminal lysine residues 21 and 22 with arginine blocked IL-1-induced IB␣ degradation without affecting IB␣ phosphorylation. This result agrees with recent studies that show Lys-21 and Lys-22 are primary ubiquitination sites necessary for Tax-and TNF-induced IB␣ degradation (6,8). However, in FK506-stimulated cells, this mutation did not block IB␣ degradation, and IB␣ was degraded with a similar time course to wild-type IB␣, although this mutation did not affect the IB␣ phosphorylation induced by FK506 (Fig. 6A, lanes 11-12 and Fig. 6B, lanes 6 -9). In sum, both the wild-type and the K21R/ K22R mutant IB␣ were less effective for inhibiting the NF-B activation induced by FK506 than the S32A/S36A IB␣ mutant (Fig. 7B), which was not degraded by FK506 treatment (Fig. 2). On the other hand, the S32A/S36A and K21R/K22R mutants, neither of which were degraded by after treatment with IL-1, were equally effective in inhibiting the NF-B activation induced by IL-1 (Fig. 7A). DISCUSSION FK506 inhibits the activation of several transcription factors involved in cytokine gene expression in T cells, including NF-B. We previously showed that FK506 activates NF-B through IB␣ degradation in nonlymphoid cells, and this FK506-induced NF-B activation results in the efficient induction of IL-6 production in vitro and in vivo (18). However, little is known about how FK506 induces NF-B activation through IB␣ degradation in nonlymphoid cells. In the present study, we found that FK506 induced the degradation of both IB␣ and IB␤ and that the time courses of their degradation were completely different from those of the degradation mediated by IL-1 (Fig. 1). However, as in the case of IB␣ degradation induced by IL-1 (4 -9), FK506-induced IB␣ degradation was also dependent on the N-terminal Ser-32 and Ser-36 phosphorylation sites (Fig. 2) and was mediated by proteasomes (Fig.   3). We further demonstrated that FK506 induced the weak and slow phosphorylation of Ser-32 (Fig. 4). These results therefore indicate that, whereas the time course of the FK506-mediated IB␣ degradation is quite different from that induced by IL-1, FK506 and IL-1 utilize similar mechanisms for inducing IB␣ degradation and hence NF-B activation.

FK506-induced IB␣ Degradation Does Not Require N-terminal Ubiquitination Sites-As shown in
Inducing the phosphorylation of the N-terminal serines is a key step in IB degradation and the subsequent NF-B activation, induced by various NF-B activators, including IL-1. Because FK506-mediated IB␣ degradation is also dependent on N-terminal phosphorylation sites (Fig. 2) and FK506 induces Ser-32 phosphorylation (Fig. 4), it is of interest to study what IB kinases are activated by FK506 and how FK506 activates them. A protein kinase complex whose activity is stimulated by IL-1 and TNF-␣ and which mediates IB␣ phosphorylation at Ser-32 and Ser-36 was recently purified, and two of the subunits of this complex (IKK-1 and IKK-2) have now been cloned and sequenced (9 -14 and 29). The results of recent mouse knockout studies indicate that whereas IKK-2 is essential for IB␣ phosphorylation induced by inflammatory cytokines such as IL-1 and TNF-␣, IKK-1 is dispensable for IL-1/TNF-induced IB␣ phosphorylation and is involved in limb and skin morphogenesis (31)(32)(33). Although we detected IL-1-induced IKK-2 (Fig. 5), in agreement with the results of previous studies (9 -14 and 22), we have not so far detected significant IKK-1 2 or IKK-2 (Fig. 5) activation with FK506. These results suggest that other recently described IB kinases such as pp90rsk (15) and IKK-3 (16) or unidentified IB kinases are involved in the FK506-mediated IB␣ phosphorylation. However, whereas FK506 is very effective in JNK activation (Fig. 5), it did not induce a significant activation of Erk, 2 which lies immediately upstream of pp90rsk in the phorbol ester and growth factor signaling pathway (34,35). Therefore, the involvement of pp90rsk in IB␣ phosphorylation mediated by FK506 is unlikely, although a possible direct pp90rsk activation by FK506 cannot be excluded.
Another important and unresolved question is how FK506 activates the putative IB kinase. Because we found that a nonimmunosuppressive FK506 analog (36) is inactive in NF-B activation and competitively inhibits FK506-mediated NF-B activation and IB␣ degradation, 2 it appears that cytosolic FK506-binding proteins (termed FKBP) are involved in FK506mediated NF-B activation and IB␣ degradation. However, it is unlikely that the inhibition of FKBP peptidyl-prolyl isomerase activity by FK506 (17) results in the accumulation of misfolded proteins in the endoplasmic reticulum, thus leading to NF-B activation (37), because this FK506 analog is effective in inhibiting FKBP peptidyl-prolyl isomerase activity (36). It is more likely that an FK506-FKBP complex interacts with kinases or phosphatases involved in an IB kinase activation pathway and that this interaction results in IB kinase activation and subsequent IB␣ degradation. In this context, it is interesting to note that an FK506-FKBP complex interacts with various cellular signaling factors such as calcineurin (17), ryanodine receptors (38), and type-I receptors for TGF-␤ (39) and can modulate the functions of these factors.
It is now clear that the phosphorylation of IB␣ at Ser-32 and Ser-36 results in a phosphorylation-dependent interaction with the IB ubiquitin ligase, leading to ubiquitination and subsequent degradation of IB␣ by proteasomes (40). In the present study, we found that both IL-1 and FK506 induce the phosphorylation of IB␣, at least at Ser-32 (Fig. 4), and these N-terminal serine residues are essential for both IL-1 and FK506-induced IB␣ degradation (Fig. 2). However, whereas the N-terminal ubiquitination sites (Lys-21 and Lys-22) are essential for IL-1-induced IB␣ degradation, these ubiquitination sites are dispensable for FK506-induced IB␣ degradation (Fig. 6). In agreement with these results, the K21R/K22R IB␣ mutant was less effective in inhibiting FK506-induced NF-B activation than was the S32A/S36A IB␣ mutant (Fig. 7). Although the possibility that IB␣ is ubiquitinated at other lysine residues in FK506-treated cells has not been completely ruled out, these results raise the possibility that IB␣ is degraded by proteasomes in an ubiquitin-independent manner in FK506treated cells. Several examples exist of proteins, including c-Jun and IB␣, being degraded in a ubiquitin-independent, proteasome-mediated manner (41)(42)(43). Thus, c-Jun and IB␣ appear to be degradable by proteasomes in both ubiquitin-dependent and -independent manners. Interestingly, it was recently reported that a 450-kDa complex, whose subunits show sequence homology to those of a proteasome regulatory complex, phosphorylates c-Jun as well as IB␣ (20). It is therefore possible that this regulatory complex not only phosphorylates but also presents IB␣ for degradation by proteasomes in a ubiquitin-independent manner.   1 and 8) or expression vectors encoding the wild-type (Wt) (lanes 6, 7, 13, and 14), S32A/S36A (S32/36A) mutant (lanes 2, 3, 9, and 10), or K21R/K22R (K21/22R) mutant (lanes 4, 5, 11, and 12) IB␣ proteins as indicated and were treated with IL-1 for 15 min or FK506 for 60 min; Ϫ and ϩ denote untreated cells or treatment with IL-1 or FK506, respectively. Cell lysates were subjected to immunoblot analysis with anti-HA antibodies. Note that K21R/K22R IB␣ was completely degraded in FK506-stimulated cells but appeared as a high molecular weight band in IL-1-treated cells. B, L-TK cells, transfected with 2 g of empty vectors (lane 1) or expression vectors encoding K21R/K22R mutant IB␣ (lanes 2-9), were treated with IL-1 and FK506 for the time periods indicated. 100-g lysates were analyzed by immunoblotting with anti-HA antibodies. Note the presence of high molecular weight bands corresponding to the phosphorylated forms of the K21R/K22R mutant IB␣ proteins in both IL-1-and FK506-treated cell lysates. *NS, nonspecific bands.