κB-Ras Binds to the Unique Insert within the Ankyrin Repeat Domain of IκBβ and Regulates Cytoplasmic Retention of IκBβ·NF-κB Complexes

The IκBα and IκBβ proteins inhibit the transcriptional potential of active NF-κB dimers through stable complex formation. It has been shown that inactive IκBα·NF-κB complexes shuttle in and out of the nucleus, whereas IκBβ·NF-κB complexes are retained exclusively in the cytoplasm of resting cells. The biochemical mechanism underlying this functional difference and its consequences are unknown. Although the two IκB proteins are significantly homologous, IκBβ contains a unique 47-amino acid insertion of unknown function within its ankyrin repeat domain. In this study, we assess the role of the IκBβ insert in regulating cytoplasmic retention of IκBβ·NF-κB complexes. Deletion of the IκBβ insert renders IκBβ·NF-κB complexes capable of shuttling between the nucleus and cytoplasm, similar to IκBα·NF-κB complexes. A small Ras-like G-protein, κB-Ras, participates with the IκBβ insert to effectively mask the NF-κB nuclear localization potential. Similarly, a complex between NF-κB and a mutant IκBβ protein containing four serine to alanine mutations within its C-terminal proline, glutamic acid, serine, and threonine-rich sequence exhibits nucleocytoplasmic shuttling. This suggests a phosphorylation state-dependent role for the C-terminal proline, glutamic acid, serine, and threonine-rich sequence of IκBβ in proper localization of IκBβ·NF-κB complexes. These results are consistent with structural studies, which predicted that binary IκBβ·NF-κB complexes should be capable of nuclear translocation, and with previous observations that hypophosphorylated IκBβ·NF-κB complexes can reside in the nucleus.

Although present in most cell types, NF-B dimers with transactivation potential are maintained inactive through their stable association with the inhibitor proteins IB␣ and IB␤ (4,5). These IB proteins share many properties, including domain structure and selectivity toward NF-B binding partners. However, significant differences exist between them (6 -9). Degradation of IB␣ leads to rapid but transient activation of NF-B (10 -12). However, many cellular activities, such as lymphoid cell development, endothelial and brain cell function, as well as various pathological conditions and viral infections, require sustained activation of NF-B (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). In each case, IB␤ has been shown to be the critical mediator of persistent NF-B activity. The mechanism underlying this persistently active NF-B remains unclear.
In an effort to understand the source of differential NF-B activation kinetics, several groups have begun investigating the molecular mechanisms of action of the IB␣ and IB␤ inhibitor proteins. One clear difference has emerged involving the manner by which these two proteins regulate NF-B subcellular localization (26). Immunofluorescence studies have revealed that IB␣⅐NF-B complexes shuttle between the cytoplasm and nucleus in quiescent cells (27)(28)(29). In contrast, inactive IB␤⅐NF-B complexes remain exclusively in the cell cytoplasm (30 -32).
A mechanistic explanation for the dynamic shuttling behavior of IB␣⅐NF-B complexes has been described recently. X-ray structures of the IB␣⅐NF-B p50/p65 heterodimer complex revealed that IB␣ masks the nuclear localization signal (NLS) 1 of the NF-B p65 subunit but fails to mask the p50 subunit NLS (33,34). Protease protection assays revealed that one NLS in the IB␣⅐NF-B p65 homodimer and IB␣⅐NF-B c-Rel homodimer complexes is also exposed to solvent and sensitive to protease cleavage (32). 2 Nuclear import of the IB␣⅐NF-B p50/p65 heterodimer complex is imparted by the free p50 NLS, whereas the one free p65 subunit NLS is sufficient to convey similar shuttling properties upon the IB␣⅐NF-B p65 homodimer complex (32). Active export of the complex from the nucleus relies on nuclear export signals located in the N-terminal signal response region of IB␣ and the transactivation domain of the p65 subunit (27)(28)(29)35).
In an accompanying study, we describe the x-ray crystal structure of an IB␤⅐NF-B p65 homodimer complex (see Ref. 43). The structure reveals that, like IB␣⅐NF-B complexes, the NLS of one NF-B p65 subunit (subunit A) is effectively masked by IB␤, whereas the second p65 subunit (subunit B) NLS is largely solvent-exposed. This observation is somewhat unexpected because the second NF-B subunit NLS is significantly less sensitive to cleavage by proteases when bound to IB␤ (32). Therefore, it remains unclear as to how IB␤⅐NF-B complexes are retained within the cytoplasm of resting cells.
Recently, a small Ras-like protein was identified from a yeast two-hybrid screen as an IB␤ C-terminal PEST sequence-interacting factor and inhibitor of NF-B activation (36). It was suggested that this protein, named B-Ras, might function by inhibiting IB␤ degradation. In the present study, we show that B-Ras is a critical regulator of IB␤⅐NF-B complex sub-cellular distribution. B-Ras binds through a unique 47-amino acid insert between ankyrin repeats 3 and 4 of IB␤ and partially masks one p65 subunit NLS in vitro and completely masks it in vivo. The conversion of serines in the IB␤ PEST sequence to non-phosphorylatable residues further alters the subcellular localization properties of IB⅐NF-B complexes. These data suggest that additional factors may bind B-Ras and IB␤⅐NF-B complexes and sequester them to the cytoplasm of quiescent cells. We propose that regulated association and dissociation of B-Ras determine the subcellular localization of IB␤⅐NF-B complexes.

FIG. 1. The IB␤ insert and the NF-B NLS affect subcellular localization of IB␤⅐NF-B complexes.
A, localization of free p65, IB␤⅐p65 complex, and IB␤⌬-(152-192)⅐p65 complexes shown with and without LMB treatment. Overexpressed p65 is localized to both the cytoplasm and the nucleus. Wild type IB␤⅐p65 complex is mostly (Ͼ80%) cytoplasmic (middle two panels). IB␤⌬-(152-192)⅐p65 complex is capable of localizing to the nucleus as indicated by the nuclear staining of both p65 and IB␤ in the LMB-treated cells (right two panels on the bottom). B, localization of IB␤⅐p50/p65 and IB␤⌬-(152-192)⅐p50/p65 complexes. Cells are co-transfected with expression vectors containing FLAG-tagged p50, p65, and the wild type or mutant IB␤. The cells are stained with anti-FLAG and anti-IB␤ antibodies. Left panel shows the control expressing only p50 and p65. The principally nuclear staining by anti-FLAG indicates that free p50 homodimer and the p50/p65 heterodimer are nuclear. Middle two panels show the staining of the IB␤⅐p50/p65 complex. IB␤ localizes to the cytoplasm in both LMB-treated cells and untreated cells (Ͼ80%). A significant amount of anti-FLAG is also observed in the nucleus, which suggests that excess p50 homodimer localizes to the nucleus. In the IB␤⌬-(152-192)⅐p50/p65 complex, both FLAG-p50 and IB␤ stain in the cytoplasm in the absence of LMB, but they are nuclear (Ͼ70%) in the presence of LMB, suggesting that the complex is capable of entering the nucleus (right two panels). C, co-immunoprecipitation of wild type and mutant IB␤ with p65 homodimer and p50/p65 heterodimer. Cytoplasmic extracts were prepared from IB␤-and NF-B-co-transfected HeLa cells and immunoprecipitated with anti-IB␤ antibody followed by immunoblotting with anti-p65 antibody. Lane 1 shows p65 from wild type IB␤-co-transfected cells. Lane 2 corresponds to p65 from the IB␤⌬-(152-192)-co-transfected cells. Lane 3 indicates immunoblotted p65 from cells transfected with p50, p65, and wild type IB␤. Lane 4 contains immunoblotted p65 from cells transfected with p50, p65, and IB␤⌬-(152-192).
FIG. 2. The NLS of p50 is responsible for shuttling IB␤⅐p50/p65 complexes. The left panels show the localization of FLAG-p50⌬NLS/p65 complex. Wild type IB␤ in complex with FLAG-p50⌬NLS/p65 heterodimers is unable to enter the nucleus in both LMB-treated and untreated cells (Ͼ80%; middle two panels). The IB␤⌬-(152-192)⅐p50⌬NLS/ p65 heterodimer complex is cytoplasmic in both the presence and absence of LMB (Ͼ90%).

EXPERIMENTAL PROCEDURES
Mammalian Cell Transfection-HeLa cell transfection was performed by the LipofectAMINE method (Invitrogen). Leptomycin B (LMB) (5 ng/ml) was added 3 h before harvesting cells. LMB is a generous gift from Prof. M. Yoshida (University of Tokyo, Tokyo, Japan).
Immunofluorescence-Cells (with or without LMB treatment) were fixed in 3% paraformaldehyde for 20 min at room temperature and then permeabilized with phosphate-buffered saline buffer containing 0.5% Nonidet P-40 and 0.01% sodium azide (ISB). Blocking was done using 5 mg/ml bovine serum albumin followed by incubation for 30 min with primary antibodies in ISB. Cells were then washed three times with ISB. Fluorescent-tagged secondary antibody was added in ISB at room temperature.
Immunoprecipitation and Western Analysis-Cells were washed three times in phosphate-buffered saline buffer. Cytoplasmic extracts were made by lysing cells in 1% Triton X-100, 20 mM Tris-HCl (pH 7.6), 200 mM NaCl, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride (lysis buffer). Fifty g of extract was mixed with protein Aagarose and primary antibodies and incubated at 4°C overnight. The immunoprecipitates were washed three times in lysis buffer and eluted with SDS-PAGE buffer by heating at 100°C for 5 min. The supernatant was separated by10% SDS-PAGE. The separated proteins in the gel were transferred to Hybond nitrocellulose membrane (Amersham Biosciences). The membrane was blocked with 5% milk in phosphatebuffered saline with 0.2% Tween and incubated with anti-p65 polyclonal antibody (H-286; Santa Cruz Biotechnology) for 1 h at room temperature. The membrane was washed and incubated with horseradish peroxidase-conjugated anti-rabbit Ig (Santa Cruz Biotechnology). Blots were visualized by use of the ECL reagent kit (Amersham Biosciences).
For in vitro immunoprecipitation experiments, 0.5 g of NF-B was mixed with 4 g of IB in the presence or absence of 4 g of B-Ras in a 15 l binding reaction. The mixture was incubated on ice for 2 h followed by dilution to 100 l in lysis buffer. Four l of this diluted complex was used for the immunoprecipitation reaction using 0.1 g of the anti-p65 NLS monoclonal antibody (a generous gift from Roche Diagnostics). The immunoprecipitates were then loaded on a Western blot, as described above.

RESULTS
The Unique IB␤ Insert Inhibits Nuclear Import of IB␤⅐NF-B Complexes-The finding that one p65 subunit (subunit B) NLS is at most weakly bound by IB␤ suggests that this NLS may also be capable of translocating the IB␤⅐NF-B complex to the nucleus. It has been shown previously that IB␤⅐NF-B complexes are cytoplasmic in quiescent cells (30 -32). Therefore, it follows that in quiescent cells the p65 subunit B NLS polypeptide must bind to IB␤ more stably than the binding mode revealed in IB␤⅐NF-B p65 homodimer complex crystal structure (see Ref. 43).
To test whether the unique 47-amino acid insertion within the ankyrin repeat domain of IB␤ plays any role in determining the subcellular localization of IB␤⅐NF-B complexes, we deleted this insert from IB␤ (residues 152-192) and observed the co-localization of this mutant (IB␤⌬-(152-192)) with NF-B p65 homodimer in HeLa cells (Fig. 1A). We observe that both complexes are cytoplasmic in resting cells. However, in the presence of LMB, an inhibitor of nuclear export receptor CRM1, the mutant IB␤⌬-(152-192)⅐NF-B p65 homodimer complex, but not the wild type complex, is predominantly nuclear (37,38) (Fig. 1A). Co-immunoprecipitation experiments show that, like the wild type IB␤, mutant IB␤⌬-(152-192) is also associated with p65 (Fig. 1C). Our results thus demonstrate that the insert of IB␤ plays a role in vivo in IB␤⅐NF-B complex cytoplasmic retention.
We have also tested whether the insert of IB␤ is responsible for cytosolic retention of the IB␤⅐NF-B p50/p65 heterodimer complex (Fig. 1B). As in the case of the IB␤⌬-(152-192)⅐NF-B p65 homodimer complex, we observe that the IB␤⌬-(152-192)  1-8). Samples were immunoprecipitated with a monoclonal anti-p65 NLS monoclonal antibody and then immunoblotted with an anti-p65 polyclonal antibody. As a negative control, wild type IB␤⅐NF-B p50/p65 heterodimer complex was also immunoprecipitated with the anti-p65 NLS monoclonal antibody (lanes 9 and  10). B, co-transfection and immunoprecipitation of HA-tagged p65 with IB␣, IB␤, and insert-deleted IB␤⌬ in 293 cells. Complexes were immunoprecipitated with anti-p65 antibody and anti-p65 NLS monoclonal antibody and detected by immunoblot with anti-HA and anti-IB␤ polyclonal antibodies. protein in complex with NF-B p50/p65 heterodimer can also enter the nucleus of resting cells. A likely explanation for this phenomenon is that when the insert of IB␤ is removed, the second NLS becomes free (or more loosely bound to IB␤), enabling the complex to be actively imported into the nucleus. These results suggest that the p50 subunit in the p50/p65 heterodimer acts similarly to p65 subunit B in the NF-B p65 homodimer, which possesses an NLS polypeptide that is primarily solvent-exposed.
The p50 Subunit NLS Regulates Nuclear Translocation of IB␤⅐NF-B p50/p65 Heterodimer Complexes-To further test whether nuclear translocation of the IB␤⌬-(152-192)⅐NF-B p50/p65 heterodimer complex is mediated by the free p50 NLS, we performed co-transfection experiments with a p50 subunit that lacks its NLS polypeptide (p50⌬NLS), p65, and IB␤⌬-(152-192) (Fig. 2). We observe that this complex is cytoplasmic in both the absence and presence of LMB, suggesting that only in the IB␤⌬-(152-192)⅐NF-B p50/p65 heterodimer complex is the p50 NLS free, whereas it remains masked in the wild type IB␤⅐NF-B p50/p65 heterodimer complex. These results indicate that one NF-B NLS and the IB␤ insert antagonize one another in directing subcellular localization of all IB␤⅐NF-B complexes.

B-Ras Regulates the p65 Subunit B NLS of the IB␤⅐NF-B p65 Homodimer Complex in Vitro and in Cells-
The above experiments suggest that the IB␤ insert could mask the second (subunit B) NLS of IB␤⅐NF-B complexes in resting cells. Direct contact likely requires modification of one or both of these interacting elements. The small GTPase B-Ras was recently shown to be involved in regulation of IB degradation (36). To determine whether B-Ras might be involved in masking the NF-B subunit B NLS, we have performed immunoprecipitation experiments with a monoclonal antibody directed against the p65 subunit NLS used to precipitate free p65, IB␤⅐NF-B p65 homodimer binary complexes, and B-Ras⅐IB␤⅐NF-B p65 homodimer ternary complexes. In parallel experiments, IB␣⅐NF-B p65 homodimer complexes were probed in the presence and absence of B-Ras. The p65 from the immunoprecipitated complexes was then visualized by anti-p65 polyclonal antibody (Fig. 3A).
We observe that both IB␤⅐NF-B p65 homodimer and IB␣⅐NF-B p65 homodimer complexes can be precipitated by the anti-p65 NLS antibody (Fig. 3A, lanes 2 and 7, respectively). Moreover, we observe that the addition of B-Ras inhibits the ability of this antibody to bind to IB␤⅐NF-B p65 homodimer complexes but not IB␣⅐NF-B p65 homodimer (Fig.  3A, lanes 3 and 8). To determine whether the insert of IB␤ plays a role in NLS masking by B-Ras, we performed similar immunoprecipitation experiments with IB␤⌬-(152-192)⅐NF-B p65 homodimer complexes in the presence and absence of B-Ras. We observe that the addition of B-Ras fails to enhance NLS masking of IB␤⌬-(152-192)⅐NF-B p65 homodimer complexes (Fig. 3A, lanes 4 and 5). To verify specificity of the anti-p65 NLS monoclonal antibody, we tested it against the IB␤⅐NF-B p50/p65 complex. The anti-p65 NLS antibody fails to recognize and precipitate p65 in these complexes. This results from the nearly complete masking of the p65 subunit A NLS polypeptide by IB␤ as observed in the IB␤⅐NF-B p65 complex crystal structure (Fig. 3A, lanes 9 and 10).
To further investigate whether B-Ras mediates blockade of the p65 NLS in cells, we next co-transfected 293 cells with HA-tagged p65 together with IB␣, IB␤, or IB␤⌬-(152-192). We first confirmed that p65 associates with all three IB proteins by co-immunoprecipitation using an antibody against the transcriptional activation domain of p65 (Fig. 3B). When the anti-p65 NLS antibody was used for immunoprecipitation, we observed that only IB␣ and IB␤⌬-(152-192) were pulled down, but not wild type IB␤. These experiments suggest that in IB␣⅐NF-B p65 homodimer and IB␤⌬-(152-192) complexes, at least one p65 NLS is free, whereas both p65 subunit NLS polypeptides are blocked in the IB␤⅐NF-B p65 homodimer complex. We conclude that the IB␤ insert is required for masking of the NF-B subunit B NLS and cytoplasmic retention of IB␤⅐NF-B complexes.
B-Ras Directly Interacts with the IB␤ Insert-To identify the role of the IB␤ insert in B-Ras binding, we have tested FIG. 5. Phosphorylation of the IB␤ PEST is important for subcellular localization of the IB␤⅐NF-B p65 homodimer complex. A, localization of IB␤-PEST-Ala⅐p65 complex shown with and without LMB treatment. The complex is mostly (Ͼ80%) cytoplasmic (top two panels) in the absence of LMB. However, the IB␤-PEST-Ala⅐p65 complex is capable of localizing to the nucleus as indicated by the nuclear staining of both p65 and IB␤ in the LMB-treated cells (bottom two panels). Over 70% of co-transfected LMB-treated cells show that this complex is nuclear. B, localization of IB␤-PEST-Asp complex. Cells were co-transfected with expression vectors containing p65 and mutant HA-IB␤. The cells were stained with anti-p65 and anti-HA antibodies. Similar to the wild type IB␤⅐p65 complex, this complex also localized most in the cytoplasm in the absence (top two panels) and presence of LMB (bottom two panels). In both cases, 80% of co-transfected cells showed co-localization in the cytoplasm. C, localization of IB␤-PEST-Asp⌬-(152-192)⅐p65 complex. Both mutant IB␤ (IB␤-PEST-Asp⌬-(152-192)) and p65 stain in the cytoplasm in over 65% of cells expressing both proteins (top two panels). In the presence of LMB, both proteins stain in the nucleus (bottom two panels). This observation is similar to that observed for IB␤⌬-(152-192)⅐p65 complex. the binding of co-transfected B-Ras and IB␤ in cells. COS cells were co-transfected with plasmids expressing both wild type IB␤ and B-Ras. We show that B-Ras can be precipitated only when wild type IB␤ is present (Fig. 4). However, the association is not observed in cells expressing IB␤⌬-(152-192) and B-Ras. Therefore, the IB␤ insert is required for association with B-Ras. Taken together, these experiments suggest that the small GTPase B-Ras may play an in vivo role in blocking the subunit B NLS of dimeric NF-B by binding directly to the unique insert region of the IB␤ inhibitor protein.
The IB␤ PEST Functions in Cytoplasmic Retention of IB␤⅐NF-B Complexes-Because B-Ras was identified as an IB-interacting protein in a yeast two-hybrid screen with the IB␤ C-terminal PEST used as the bait, it is important to evaluate the role of the IB␤ PEST in NF-B cytoplasmic sequestration. To test whether phosphorylation of the IB␤ PEST plays any role in cytoplasmic retention of NF-B, we mutated four phosphorylatable serines (Ser 312 , Ser 313 , Ser 314 , and Ser 316 ) within this region to either alanine (PEST-Ala) or aspartic acid (PEST-Asp). Two serine residues in this region have been shown to be phosphorylated by casein kinase II, and phosphorylation at these sites is important for NF-B binding in cells (39,40). However, we have previously shown that phosphorylation of these serines or their conversion to phosphomimetic glutamic acid residues does not alter the stability of IB␤⅐NF-B binary complexes in vitro (32).
If constitutive phosphorylation of the PEST is important for stable complex formation, then substitution of these residues to alanine could prevent cytoplasmic retention of IB␤⅐NF-B complexes. To test this possibility, we co-transfected HeLa cells with IB␤-PEST-Ala and p65 and observed co-localization of both proteins in the cytoplasm. However, when these cells were treated with LMB followed by immunostaining, we observed that, in a large fraction of cells expressing both of these proteins, the complex localized to the nucleus (Fig. 5A). In contrast, when the cells were co-transfected with p65 and IB␤-PEST-Asp, the complex was cytoplasmic with or without LMB treatment (Fig. 5B). This property of the IB␤-PEST-Asp mutant complex is identical to that of the wild type complex. We conclude that IB␤ PEST phosphorylation is critical for cytoplasmic sequestration of IB␤⅐NF-B complexes.
We further created a mutant of IB␤ in which the insert is deleted within the background of phosphomimetic IB␤-PEST-Asp sequence ((IB␤⌬-(152-192)-PEST-Asp). When HeLa cells were co-transfected with this mutant and wild type p65, we observed that the complex shuttled between the cytoplasm and nucleus (Fig. 5C). This profile is identical to that observed in the IB␤⌬-(152-192)⅐NF-B p65 homodimer complex.
It is not immediately clear to us what role a phosphorylated IB␤ PEST plays in sequestering the IB␤⅐NF-B complexes to the cytosol. It is possible that B-Ras interacts with both the IB␤ insert and phosphorylated PEST. It is also possible that the interaction between the IB␤ PEST and B-Ras is indirect, mediated by another bridging factor. DISCUSSION IB␤⅐NF-B complexes are cytoplasmic in quiescent cells (30 -32). This suggests that the NF-B subunit B NLS can exhibit a binding mode, alternative to that observed in the IB␤⅐NF-B p65 homodimer complex crystal structure, in which it is completely masked (see Ref. 43). We have tested this hypothesis derived from our structural analyses of IB⅐NF-B complexes by both in vitro and cell-based studies. Transient transfection and immunostaining experiments reveal that the unique and structurally disordered insert between the third and fourth ankyrin repeats of IB␤ functions to regulate nuclear import of the IB␤⅐NF-B p65 homodimer complex in resting cells. These experiments suggest that the insert functions as a nuclear export signal or blocks a nuclear localization sequence. No recognizable export sequence is observed within the IB␤ insert. Interestingly, IB␤⌬-(152-192)⅐NF-B complexes shuttle between the nucleus and cytoplasm, despite the lack of a nuclear export sequence in IB␤. We suggest that the partially exposed NLS of NF-B subunit B and the export potential of the p65 activation domain drive the dynamic shuttling behavior of these complexes (32,35).
We propose the hypothesis that B-Ras may act in concert with the IB␤ insert to sequester IB␤⅐NF-B complexes to the cytoplasm. Indeed, we show that the small GTPase B-Ras is able to reduce access to the NF-B p65 homodimer subunit B NLS in complexes between p65 and IB␤. We further present evidence that binding of B-Ras to IB␤ requires the insert within the ankyrin repeat domain of IB␤. Because this insert is unique to IB␤, these results serve to explain the specificity of B-Ras for the IB␤⅐NF-B complex.
There are several published observations which suggest that phosphorylation of the IB␤ PEST might play an important role in regulation of NF-B by IB␤ (7,9,22). The most important of these is the identification of B-Ras through the utilization of the IB␤ PEST sequence as bait in a yeast two-hybrid screen (36). Also, it has been shown that PEST phosphorylation FIG. 6. Proposed models for the nucleocytoplasmic shuttling profile of IB⅐NF-B complexes and the ternary IB␤⅐p65⅐DNA complex. A, IB␣⅐NF-B complexes shuttle between the cytoplasm and nucleus in quiescent cells. The free p50 NLS of the p50/p65 heterodimer (or the second p65 NLS of p65 homodimer) is responsible for nuclear entry. The nuclear export signals in IB␣ and p65 activation domain (not depicted in this model) are responsible for nuclear export. Nuclear IB␣⅐NF-B complexes are unable to bind DNA. B, IB␤⅐NF-B complexes are cytoplasmic in resting cells. Both NLSs of NF-B dimers are masked by IB␤ in the presence of B-Ras. The IB␤ insert and B-Ras mask the p50 NLS in p50/p65 heterodimer and one p65 NLS from the p65 homodimer. Removal of B-Ras leads to release of the second NLS. One free NLS directs import of the IB␤ complex into the nucleus. Within the nucleus, the complex is capable of binding to DNA. is important for stable complex formation with NF-B (39). We show here that PEST phosphorylation of IB␤ functions in cytoplasmic retention of IB␤⅐NF-B complexes. Although we do not observe a clear need for PEST phosphorylation in the interaction between IB␤ and NF-B or B-Ras⅐IB␤⅐NF-B complex formation, our experiments do suggest that PEST phosphorylation cooperates with B-Ras binding and the IB␤ insert in regulating cytosolic retention of IB␤⅐NF-B complexes. We propose that this cooperativity might be mediated through other cellular factors or through further post-translational modification elsewhere in the complex. We favor the first possibility in light of our recent findings that IB␤ is present in the cytoplasm of quiescent cells as large complexes, which include NF-B and B-Ras as well as other as-yet-unknown factors. 3 Working together, the IB␤ insert and the PEST, one NF-B NLS, and B-Ras represent a molecular mechanism for switching between nuclear and cytoplasmic IB␤⅐NF-B complexes (Fig. 6).
Our experiments help to explain the observation that hypophosphorylated IB␤⅐NF-B complexes can localize into the nucleus (7). We suggest that the hypophosphorylated form of IB␤ is functionally equivalent to the IB␤-PEST-Ala protein construct used in this study.
Finally, our results serve to explain one long-standing puzzle. It has long been thought that IB␣ binds NF-B dimers with a significantly higher affinity than does IB␤. This conclusion was drawn based on the respective abilities of IB␣ and IB␤ to inhibit NF-B DNA binding. We have shown previously that compared with IB␣, IB␤ binds NF-B p50/p65 heterodimer and p65 homodimer with only slightly weaker affinity (32,41,42). In light of the IB␤⅐NF-B p65 homodimer complex crystal structure and the biochemical evidence surrounding B-Ras binding to IB␤⅐NF-B complexes, we suggest that the in vivo stabilities of IB␤⅐NF-B complexes are nearly equivalent to those of IB␣⅐NF-B complexes. This would explain why both complexes are present in almost all cells in almost equal amounts (5).