IκB Family Members Function by Different Mechanisms*

The IκB family of proteins regulates NF-κB-dependent transcription by inhibiting DNA binding and localizing these factors to the cell cytoplasm. IκBα does this by shifting the balance between nuclear import of Rel proteins and their export from the nucleus. Here we show that, unlike IκBα, IκBβ and IκBε appear to sequester p65 or c-Rel in the cytoplasm by inhibiting nuclear import. Furthermore, because IκBβ does not undergo nucleocytoplasmic shuttling, it cannot remove nuclear proteins like IκBα does. We conclude that the mechanism of action differs among IκB family members.

preceding the first ankyrin domain (13)(14)(15). A second nuclear export sequence has been identified at the C terminus of IB␣, but its functional significance is unclear at present (16,17). The observation that cytoplasmic sequestration of p65⅐RelA also required nuclear export was unexpected and led to a reassessment of the existing sequestration model. We and others (13)(14)(15) have proposed that the cytoplasmic location of Rel proteins by IB␣ is a dynamic process that depends on the active export of Rel⅐IB␣ complexes out of the nucleus.
In our earlier studies we also showed that IB␤ and IB⑀, unlike IB␣, do not shuttle via the CRM1 pathway. Specifically, the subcellular distribution of GFP-IB␤ or GFP-IB⑀ in yeast was not affected by a mutation in the CRM1 gene, and leptomycin B (LMB) treatment did not alter the location of these proteins in transiently transfected mammalian cells. In addition, we did not detect an association between CRM1 protein and either IB␤ or IB⑀ in a yeast two-hybrid assay (15). In this paper we demonstrate that cytoplasmic retention of Rel proteins by IB␤ and IB⑀ involves sequestration rather than tilting the balance of nuclear import and export as is the case with IB␣. Furthermore, although newly synthesized IB␤ can enter the nucleus, it cannot restore nuclear Rel proteins to the cytoplasm. These observations suggest that IB␤ and IB⑀ function differently from IB␣.

MATERIALS AND METHODS
Cell Lines and Strains-D5 h3 T hybridoma cells and A20 mature B cells were grown in Dulbecco's modified Eagle's medium and RPMI 1640 medium, respectively, with 10% heat-inactivated fetal bovine serum, 50 M ␤-mercaptoethanol and antibiotics. COS cells were cultured in Dulbecco's modified Eagle's medium with 10% newborn calf serum and antibiotics. Yeast strain W303 and its transformants were generally grown in synthetic medium with the appropriate amino acid and nitrogen base supplement.
Plasmids-pGFP-p65 and pCDNA3-HA-IB␣ have been described previously (15). pGFP-cRel contains full-length murine cRel in frame after GFP. pCDNA3-Myc-IB␤ and pCDNA3-Myc-IB⑀ were made by inserting full-length murine IB␤ and IB⑀ cDNA, respectively, in frame behind a c-Myc tag (MEQKLISEEDL). Yeast galactose-inducible plasmid encoding GFP-p65 and copper-inducible HA-IB␣ (pCuIB␣) have been described previously (15). The copper-inducible HA-IB␤ was made by replacing the IB␣ gene with a murine IB␤ full-length gene in the same vector. All plasmids used in this study were confirmed by sequencing, and expression of proteins was verified by immunoblotting.
Immunostaining-The procedures for immunostaining adherence cells were the same as described previously (15). For staining suspension cells (T and B cells), the procedures were also as described previously (18).
Protease Digestion-The proteases Asp-N and Lys-C were purchased from Roche Molecular Biochemicals. Proteases were used according to the manufacturers' specifications.
Fluorescence Microscopy-The subcellular localization of GFP and the immunofluorescence signals were observed by fluorescence microscopy (Axiophot II, Zeiss) with a GFP generic filter, fluorescein isothiocyanate, rhodamine, and DAPI filter.

RESULTS
The nuclear export property of IB␣ is essential for cytoplasmic location of Rel proteins. However, IB␤ and IB⑀, which are not nucleo-cytoplasmic shuttling proteins, can also effectively localize Rel proteins to the cytoplasm. One possibility was that cytoplasmic retention by IB␤/⑀ may be mediated by export determinants in the Rel proteins. To test this possibility, we coexpressed green fluorescent protein (GFP)-tagged Rel proteins with IB␤ or IB⑀ in COS cells and assayed the loca-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Rosenstiel Basic Medical Sciences Research Ctr., Brandeis University, 415 South St., Waltham, MA 02454. E-mail: sen@brandeis.edu. tion of Rel proteins by GFP fluorescence. Both p65 and c-Rel (data not shown) were located in the cytoplasm in the presence of IB␤ (Fig. 1A, left panel). However, these complexes did not translocate to the nucleus when the cells were treated with LMB, an inhibitor of CRM1-mediated nuclear export (Fig. 1A, right panel). Therefore, CRM1 was not involved in determining the subcellular location of these complexes. Similar results were obtained with IB⑀. As expected, IB␣-associated p65, or c-Rel (data not shown), was predominantly nuclear in LMBtreated cells (Fig. 1A, top row). Thus, cytoplasmic retention by IB␤ and IB⑀ may involve true sequestration rather than a balance between import and export as is the case with IB␣.
These observations were confirmed in mammalian cells by investigating the shuttling dynamics of endogenous Rel⅐IB complexes. Endogenous proteins in mature B (A20) and mature T (D5 h3) cell lines (data not shown) were visualized by staining fixed, permeabilized cells with anti-IB␣, or anti-IB␤, antibodies in the presence or absence of LMB to block nuclear export. In untreated cells both IBs were predominantly cytoplasmic (Fig. 1B, left panel). A 1-h LMB treatment induced considerable nuclear translocation of IB␣ but not IB␤ (Fig.  1B, right panel). Because most of the cellular IB is associated with Rel proteins, we concluded that Rel⅐IB␣ complexes shuttled continuously, but Rel⅐IB␤ complexes did not. Lack of Rel⅐IB␤ shuttling is consistent with sequestration being the major mechanism of cytoplasmic retention by IB␤.
We found more direct evidence for differences in interaction between IB␣ or IB␤ and p65 through partial proteolysis assays. p65 protein was expressed by transient transfection in BOSC 23 cells in the presence of HA-IB␣ or Myc-IB␤. The p65⅐IB complex was immunoprecipitated from whole cell extract with anti-IB␣ antibody or anti-IB␤ antibody and digested with different proteases. The p65 fragments were detected using antibodies directed against the N or C terminus of p65 to estimate the cut site from one or the other end of p65. Only two of seven proteases showed significant differences in the pattern of p65 fragments generated in the presence of IB␣ or IB␤. p65 alone generated one major fragment when treated with Asp-N of ϳ28 kDa when assayed from the C terminus (Fig. 2,  lanes 5 and 6); this corresponds to a cut site located 293 amino acids from the N terminus (Fig. 2, top). In the p65⅐IB␣ complex, two bands of approximately equal intensity were seen (Fig. 2, lanes 1 and 2), whereas in the p65⅐IB␤ complex the faster mobility (23 kDa) band was enhanced. Therefore, cutting at residue 293 was reduced in the p65⅐IB␣ complex compared with p65 alone, allowing the detection of the cut site at residue 360 (which was not evident with p65 alone). This is presumably because of the protection of the p65 NLS by IB␣, which lies close to residue 293 between residues 301 and 304. Cutting at 293 was further inhibited in the p65⅐IB␤ complex as shown by a relative increase in the intensity of the 23-kDa compared with the 28-kDa band. These observation suggest that the region around residue 293, including the NLS, is more protected in the p65⅐IB␤ complex.
p65⅐IB␣ and p65⅐IB␤ complexes were also probed using the protease Lys-C and p65 antibodies directed against the N terminus. Increased cutting at the residue 425 site was evident in the IB␤ complex compared with the IB␣ complex (Fig. 2,  lanes 8 and 11). These observations also support the interpretation that IB␣ and IB␤ interact differently with p65. We suggest that the p65 NLS is better hidden by IB␤ than by IB␣. The simplest interpretation of the experiments described above was that Rel⅐IB␤ complexes did not enter the nucleus because the nuclear localization sequences in both proteins were very effectively hidden in the complex. Therefore, the question of nuclear export did not arise. However, the question remained that if any Rel⅐IB␤ complexes formed in the nucleus, would IB␤ be able to bring the complex out to the cytoplasm? Such a situation may occur at the end of cell stimulation when Rel proteins are already nuclear and new IBs are synthesized to terminate NF-B-dependent gene expression. We addressed this question in a yeast model.
We have previously shown that export-dependent cytoplasmic localization of p65 by IB␣ can be recapitulated in yeast (15). To test the properties of IB␤, we coexpressed IB␤ and GFP-p65 from galactose-inducible promoters in wild type or Crm1p-deficient (crm1-1) yeast strains. GFP-p65 was located in the cytoplasm under these conditions in both strains (data not shown), correlating closely with the observations in mammalian cells (Fig. 1). In contrast, when GFP-p65 and IB␣ were coexpressed in crm1-1 cells, the complex remained in the nucleus (15). To compare the ability of IB␣ and IB␤ to remove nuclear p65, we expressed GFP-p65 using a galactose-inducible promoter, followed by either IB␣, or IB␤, from a copperinducible promoter. A 3-h induction with galactose was followed by growth in glucose to suppress GFP-p65 transcription. In cells that did not contain IB expression vectors, GFP fluorescence was strictly nuclear. Even when cells contained either IB␣ or IB␤ expression plasmids, GFP fluorescence was largely restricted to the nucleus, although whole cell expression was observed in ϳ15% of the cells (Fig. 3, middle and  bottom rows, left panel). Cytoplasmic expression under these conditions was most likely due to basal IB␣, or IB␤, expression from the copper-inducible promoter. Induction of IB␣ with copper for 2 h resulted in a significant redistribution of GFP-p65 to the cytoplasm, indicating that the newly synthesized IB␣ exported nuclear GFP-p65 to the cytoplasm (Fig. 3, middle row, right panel). This was mediated by Crm1p because it did not occur in the crm1-1 strain that contains a mutated  5, 6, 11, and 12). Anti-IB␣ or anti-IB␤ antibodies were used to immunoprecipitate the p65⅐IB complex from whole cell extracts. Precipitated materials were digested with Asp -N (left panel, A) and Lys-C (right panel, L). Undigested samples are indicated by a "Ϫ" in the figures. Digested and undigested products were fractionated by SDS-polyacrylamide gel electrophoresis and detected using an antibody against the C terminus of p65 (left panel) or antibody against the N terminus of p65 (right panel). Relevant protease sites of p65 were predicted by MacVector version 6.0 (top panel). NLS represents the nuclear localization signal of p65, with critical residues between residues 301 and 304. Arrows (in the lower panels) indicate the relative degree of Lys-C or Asp-N cutting in the IB␤⅐p65 and IB␣⅐p65 complexes.
FIG. 3. Sequential induction of GFP-p65 and IB. GFP-p65 was cloned into an expression plasmid with a galactose-inducible promoter. HA-tagged IB␣, or IB␤, was cloned into an expression plasmid containing a copper-inducible promoter. Yeast strain W303 transformed with both GFP-p65 and HA-IB expression plasmids was treated with galactose for 3 h to induce GFP-p65 expression (left panel). Half of the cells were then treated with glucose to terminate the expression of GFP-p65 followed by 0.75 mM copper sulfate to induce IB expression (right panel). GFP fluorescence was visualized directly with fluorescence microscopy. Whole cell extracts were made from the cells to confirm the induction of GFP-p65, HA-IB␣, and HA-IB␤ proteins by immunoblotting (data not shown). Results shown are from one of three independent experiments. CRM1 gene (data not shown). In contrast, there was little redistribution of GFP-p65 after secondary induction of IB␤ (Fig. 3, bottom row, right panel). The small increase in whole cell GFP-p65 expression was probably because of residual GFP-p65 translation during IB␤ induction, which resulted in its cytoplasmic sequestration. These observations indicate that IB␤ cannot remove nuclear Rel proteins to the cytoplasm. DISCUSSION We found that p65 or c-Rel associated with IB␤ or IB⑀ were retained in the cytoplasm, although these IBs did not shuttle via the CRM1 pathway. We suggest that IBs, unlike IB␣, sequester rather than shuttle Rel proteins, which implies that there is no available NLS in the Rel⅐IB␤ (or IB⑀) complexes to induce nuclear entry. Conversely, Rel⅐IB␣ complexes must have an available NLS to shuttle. We hypothesize that the Rel and not the IB component provides the functional NLS of a Rel⅐IB complex. Thus, IB␤ or IB⑀ must hide the Rel NLS more effectively than IB␣. Evidence in favor of this idea was obtained from partial proteolytic studies of p65⅐IB complexes.
The sequestration mechanism is based on the lack of an effect of leptomycin B or a mutated CRM1 gene in Rel protein localization by IB␤. Alternatively, these results could indicate that Rel⅐IB␤ complexes shuttled by a CRM1-independent pathway. To test this theory, we generated nuclear Rel⅐IB␤ complexes and determined whether they could reach the cytoplasm by an unidentified pathway. As shown in Fig. 3, IB␤mediated GFP-p65 export was inefficient compared with IB␣. We conclude that IB␤ is not an export chaperone like IB␣. Consequently, IB␤ cannot efficiently down-regulate nuclear Rel proteins to restore the resting state of the cell. These results highlight the functional differences between IB␣ and IB␤.
Cheng et al. (19) showed that substituting IB␤ for the IB␣ gene compensated for the most obvious defects in IB␣ Ϫ/Ϫ mice. They concluded that IB␣ and IB␤ were functionally similar and that regulation of expression accounted for most of the phenotype of IB␣-deficient mice. Our contrasting conclu-sion regarding the mechanism of IB␣ and IB␤ function is not at odds with the biological results. Clearly, if sufficient IB␤ is synthesized in a cell, it can retain Rel proteins in the cytoplasm, albeit by a mechanism different from IB␣. The biological results show that retention of Rel proteins by either mechanism is good enough to rescue lethality. That IB␤ is a less efficient nuclear export chaperone than IB␣ may be manifest under conditions that were not directly assayed, such as during an immune response or chronic inflammation. We suggest that control of such situations may require the active export-dependent reduction of NF-B activity.